summaryrefslogtreecommitdiffstats
path: root/src/gallium/auxiliary
diff options
context:
space:
mode:
Diffstat (limited to 'src/gallium/auxiliary')
-rw-r--r--src/gallium/auxiliary/Makefile181
-rw-r--r--src/gallium/auxiliary/SConscript185
-rw-r--r--src/gallium/auxiliary/cso_cache/Makefile11
-rw-r--r--src/gallium/auxiliary/cso_cache/SConscript11
-rw-r--r--src/gallium/auxiliary/cso_cache/cso_context.c212
-rw-r--r--src/gallium/auxiliary/cso_cache/cso_context.h32
-rw-r--r--src/gallium/auxiliary/draw/Makefile46
-rw-r--r--src/gallium/auxiliary/draw/SConscript46
-rw-r--r--src/gallium/auxiliary/draw/draw_context.c84
-rw-r--r--src/gallium/auxiliary/draw/draw_context.h30
-rw-r--r--src/gallium/auxiliary/draw/draw_gs.c338
-rw-r--r--src/gallium/auxiliary/draw/draw_gs.h76
-rw-r--r--src/gallium/auxiliary/draw/draw_pipe_aaline.c126
-rw-r--r--src/gallium/auxiliary/draw/draw_pipe_aapoint.c294
-rw-r--r--src/gallium/auxiliary/draw/draw_pipe_clip.c42
-rw-r--r--src/gallium/auxiliary/draw/draw_pipe_cull.c2
-rw-r--r--src/gallium/auxiliary/draw/draw_pipe_offset.c2
-rw-r--r--src/gallium/auxiliary/draw/draw_pipe_pstipple.c79
-rw-r--r--src/gallium/auxiliary/draw/draw_pipe_stipple.c5
-rw-r--r--src/gallium/auxiliary/draw/draw_pipe_vbuf.c6
-rw-r--r--src/gallium/auxiliary/draw/draw_pipe_wide_line.c2
-rw-r--r--src/gallium/auxiliary/draw/draw_pipe_wide_point.c14
-rw-r--r--src/gallium/auxiliary/draw/draw_private.h37
-rw-r--r--src/gallium/auxiliary/draw/draw_pt.c38
-rw-r--r--src/gallium/auxiliary/draw/draw_pt.h11
-rw-r--r--src/gallium/auxiliary/draw/draw_pt_emit.c6
-rw-r--r--src/gallium/auxiliary/draw/draw_pt_fetch.c80
-rw-r--r--src/gallium/auxiliary/draw/draw_pt_fetch_emit.c5
-rw-r--r--src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline.c70
-rw-r--r--src/gallium/auxiliary/draw/draw_pt_post_vs.c65
-rw-r--r--src/gallium/auxiliary/draw/draw_pt_util.c16
-rw-r--r--src/gallium/auxiliary/draw/draw_pt_varray_tmp_linear.h4
-rw-r--r--src/gallium/auxiliary/draw/draw_pt_vcache.c3
-rw-r--r--src/gallium/auxiliary/draw/draw_vs.c4
-rw-r--r--src/gallium/auxiliary/draw/draw_vs.h2
-rw-r--r--src/gallium/auxiliary/draw/draw_vs_aos.c220
-rw-r--r--src/gallium/auxiliary/draw/draw_vs_ppc.c6
-rw-r--r--src/gallium/auxiliary/draw/draw_vs_varient.c15
-rw-r--r--src/gallium/auxiliary/gallivm/Makefile92
-rw-r--r--src/gallium/auxiliary/gallivm/SConscript16
-rw-r--r--src/gallium/auxiliary/gallivm/tgsitollvm.cpp70
-rw-r--r--src/gallium/auxiliary/indices/Makefile16
-rw-r--r--src/gallium/auxiliary/indices/SConscript28
-rw-r--r--src/gallium/auxiliary/pipebuffer/Makefile19
-rw-r--r--src/gallium/auxiliary/pipebuffer/SConscript19
-rw-r--r--src/gallium/auxiliary/pipebuffer/pb_buffer.h3
-rw-r--r--src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c5
-rw-r--r--src/gallium/auxiliary/pipebuffer/pb_bufmgr_cache.c2
-rw-r--r--src/gallium/auxiliary/rbug/Makefile14
-rw-r--r--src/gallium/auxiliary/rbug/SConscript14
-rw-r--r--src/gallium/auxiliary/rtasm/Makefile13
-rw-r--r--src/gallium/auxiliary/rtasm/SConscript13
-rw-r--r--src/gallium/auxiliary/rtasm/rtasm_x86sse.c7
-rw-r--r--src/gallium/auxiliary/rtasm/rtasm_x86sse.h1
-rw-r--r--src/gallium/auxiliary/sct/Makefile9
-rw-r--r--src/gallium/auxiliary/sct/SConscript9
-rw-r--r--src/gallium/auxiliary/sct/sct.c453
-rw-r--r--src/gallium/auxiliary/sct/sct.h123
-rw-r--r--src/gallium/auxiliary/sct/usage.c61
-rw-r--r--src/gallium/auxiliary/tgsi/Makefile22
-rw-r--r--src/gallium/auxiliary/tgsi/SConscript23
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi-instruction-set.txt32
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_build.c657
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_build.h123
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_dump.c250
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_dump.h5
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_dump_c.c228
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_exec.c2785
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_exec.h62
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_info.c30
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_iterate.c7
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_iterate.h6
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h23
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_parse.c198
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_parse.h48
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_ppc.c58
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_sanity.c318
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_scan.c94
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_scan.h9
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_sse2.c76
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_text.c702
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_transform.c35
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_transform.h4
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_ureg.c325
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_ureg.h107
-rw-r--r--src/gallium/auxiliary/tgsi/tgsi_util.c38
-rw-r--r--src/gallium/auxiliary/translate/Makefile12
-rw-r--r--src/gallium/auxiliary/translate/SConscript12
-rw-r--r--src/gallium/auxiliary/translate/translate.h18
-rw-r--r--src/gallium/auxiliary/translate/translate_generic.c40
-rw-r--r--src/gallium/auxiliary/translate/translate_sse.c186
-rw-r--r--src/gallium/auxiliary/util/Makefile46
-rw-r--r--src/gallium/auxiliary/util/SConscript58
-rw-r--r--src/gallium/auxiliary/util/u_blit.c35
-rw-r--r--src/gallium/auxiliary/util/u_blitter.c820
-rw-r--r--src/gallium/auxiliary/util/u_blitter.h230
-rw-r--r--src/gallium/auxiliary/util/u_clear.h8
-rw-r--r--src/gallium/auxiliary/util/u_debug.c36
-rw-r--r--src/gallium/auxiliary/util/u_debug_dump.c6
-rw-r--r--src/gallium/auxiliary/util/u_dl.c79
-rw-r--r--src/gallium/auxiliary/util/u_dl.h73
-rw-r--r--src/gallium/auxiliary/util/u_format.c15
-rw-r--r--src/gallium/auxiliary/util/u_format.csv32
-rw-r--r--src/gallium/auxiliary/util/u_format.h290
-rw-r--r--src/gallium/auxiliary/util/u_format_access.py4
-rwxr-xr-xsrc/gallium/auxiliary/util/u_format_table.py37
-rw-r--r--src/gallium/auxiliary/util/u_gen_mipmap.c112
-rw-r--r--src/gallium/auxiliary/util/u_linear.c2
-rw-r--r--src/gallium/auxiliary/util/u_linear.h19
-rw-r--r--src/gallium/auxiliary/util/u_math.h60
-rw-r--r--src/gallium/auxiliary/util/u_pack_color.h184
-rw-r--r--src/gallium/auxiliary/util/u_prim.h35
-rw-r--r--src/gallium/auxiliary/util/u_rect.c90
-rw-r--r--src/gallium/auxiliary/util/u_rect.h6
-rw-r--r--src/gallium/auxiliary/util/u_simple_shaders.c88
-rw-r--r--src/gallium/auxiliary/util/u_simple_shaders.h13
-rw-r--r--src/gallium/auxiliary/util/u_surface.c8
-rw-r--r--src/gallium/auxiliary/util/u_texture.c102
-rw-r--r--src/gallium/auxiliary/util/u_texture.h54
-rw-r--r--src/gallium/auxiliary/util/u_tile.c113
-rw-r--r--src/gallium/auxiliary/util/u_upload_mgr.h2
-rw-r--r--src/gallium/auxiliary/vl/Makefile13
-rw-r--r--src/gallium/auxiliary/vl/SConscript13
-rw-r--r--src/gallium/auxiliary/vl/vl_compositor.c32
-rw-r--r--src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c130
-rw-r--r--src/gallium/auxiliary/vl/vl_shader_build.c91
126 files changed, 7976 insertions, 4886 deletions
diff --git a/src/gallium/auxiliary/Makefile b/src/gallium/auxiliary/Makefile
index 5446eb68a98..e3af41c6e04 100644
--- a/src/gallium/auxiliary/Makefile
+++ b/src/gallium/auxiliary/Makefile
@@ -1,12 +1,177 @@
-# src/gallium/auxiliary/Makefile
TOP = ../../..
include $(TOP)/configs/current
-SUBDIRS = $(GALLIUM_AUXILIARY_DIRS)
+LIBNAME = gallium
-default install clean:
- @for dir in $(SUBDIRS) ; do \
- if [ -d $$dir ] ; then \
- (cd $$dir && $(MAKE) $@) || exit 1; \
- fi \
- done
+C_SOURCES = \
+ cso_cache/cso_context.c \
+ cso_cache/cso_cache.c \
+ cso_cache/cso_hash.c \
+ draw/draw_context.c \
+ draw/draw_gs.c \
+ draw/draw_pipe.c \
+ draw/draw_pipe_aaline.c \
+ draw/draw_pipe_aapoint.c \
+ draw/draw_pipe_clip.c \
+ draw/draw_pipe_cull.c \
+ draw/draw_pipe_flatshade.c \
+ draw/draw_pipe_offset.c \
+ draw/draw_pipe_pstipple.c \
+ draw/draw_pipe_stipple.c \
+ draw/draw_pipe_twoside.c \
+ draw/draw_pipe_unfilled.c \
+ draw/draw_pipe_util.c \
+ draw/draw_pipe_validate.c \
+ draw/draw_pipe_vbuf.c \
+ draw/draw_pipe_wide_line.c \
+ draw/draw_pipe_wide_point.c \
+ draw/draw_pt.c \
+ draw/draw_pt_elts.c \
+ draw/draw_pt_emit.c \
+ draw/draw_pt_fetch.c \
+ draw/draw_pt_fetch_emit.c \
+ draw/draw_pt_fetch_shade_emit.c \
+ draw/draw_pt_fetch_shade_pipeline.c \
+ draw/draw_pt_post_vs.c \
+ draw/draw_pt_util.c \
+ draw/draw_pt_varray.c \
+ draw/draw_pt_vcache.c \
+ draw/draw_vertex.c \
+ draw/draw_vs.c \
+ draw/draw_vs_varient.c \
+ draw/draw_vs_aos.c \
+ draw/draw_vs_aos_io.c \
+ draw/draw_vs_aos_machine.c \
+ draw/draw_vs_exec.c \
+ draw/draw_vs_llvm.c \
+ draw/draw_vs_ppc.c \
+ draw/draw_vs_sse.c \
+ indices/u_indices_gen.c \
+ indices/u_unfilled_gen.c \
+ pipebuffer/pb_buffer_fenced.c \
+ pipebuffer/pb_buffer_malloc.c \
+ pipebuffer/pb_bufmgr_alt.c \
+ pipebuffer/pb_bufmgr_cache.c \
+ pipebuffer/pb_bufmgr_debug.c \
+ pipebuffer/pb_bufmgr_fenced.c \
+ pipebuffer/pb_bufmgr_mm.c \
+ pipebuffer/pb_bufmgr_ondemand.c \
+ pipebuffer/pb_bufmgr_pool.c \
+ pipebuffer/pb_bufmgr_slab.c \
+ pipebuffer/pb_validate.c \
+ rbug/rbug_connection.c \
+ rbug/rbug_core.c \
+ rbug/rbug_texture.c \
+ rbug/rbug_context.c \
+ rbug/rbug_shader.c \
+ rbug/rbug_demarshal.c \
+ rtasm/rtasm_cpu.c \
+ rtasm/rtasm_execmem.c \
+ rtasm/rtasm_x86sse.c \
+ rtasm/rtasm_ppc.c \
+ rtasm/rtasm_ppc_spe.c \
+ tgsi/tgsi_sanity.c \
+ tgsi/tgsi_build.c \
+ tgsi/tgsi_dump.c \
+ tgsi/tgsi_exec.c \
+ tgsi/tgsi_info.c \
+ tgsi/tgsi_iterate.c \
+ tgsi/tgsi_parse.c \
+ tgsi/tgsi_ppc.c \
+ tgsi/tgsi_scan.c \
+ tgsi/tgsi_sse2.c \
+ tgsi/tgsi_text.c \
+ tgsi/tgsi_transform.c \
+ tgsi/tgsi_ureg.c \
+ tgsi/tgsi_util.c \
+ translate/translate_generic.c \
+ translate/translate_sse.c \
+ translate/translate.c \
+ translate/translate_cache.c \
+ util/u_debug.c \
+ util/u_debug_dump.c \
+ util/u_debug_symbol.c \
+ util/u_debug_stack.c \
+ util/u_blit.c \
+ util/u_blitter.c \
+ util/u_cache.c \
+ util/u_cpu_detect.c \
+ util/u_dl.c \
+ util/u_draw_quad.c \
+ util/u_format.c \
+ util/u_format_access.c \
+ util/u_format_table.c \
+ util/u_gen_mipmap.c \
+ util/u_handle_table.c \
+ util/u_hash_table.c \
+ util/u_hash.c \
+ util/u_keymap.c \
+ util/u_linear.c \
+ util/u_network.c \
+ util/u_math.c \
+ util/u_mm.c \
+ util/u_rect.c \
+ util/u_simple_shaders.c \
+ util/u_snprintf.c \
+ util/u_stream_stdc.c \
+ util/u_stream_wd.c \
+ util/u_surface.c \
+ util/u_texture.c \
+ util/u_tile.c \
+ util/u_time.c \
+ util/u_timed_winsys.c \
+ util/u_upload_mgr.c \
+ util/u_simple_screen.c \
+ vl/vl_bitstream_parser.c \
+ vl/vl_mpeg12_mc_renderer.c \
+ vl/vl_compositor.c \
+ vl/vl_csc.c \
+ vl/vl_shader_build.c
+
+GALLIVM_SOURCES = \
+ gallivm/gallivm.cpp \
+ gallivm/gallivm_cpu.cpp \
+ gallivm/instructions.cpp \
+ gallivm/loweringpass.cpp \
+ gallivm/tgsitollvm.cpp \
+ gallivm/storage.cpp \
+ gallivm/storagesoa.cpp \
+ gallivm/instructionssoa.cpp
+
+INC_SOURCES = \
+ gallivm/gallivm_builtins.cpp \
+ gallivm/gallivmsoabuiltins.cpp
+
+# XXX: gallivm doesn't build correctly so disable for now
+#ifeq ($(MESA_LLVM),1)
+#DEFINES += -D__STDC_CONSTANT_MACROS -D__STDC_LIMIT_MACROS
+#CPP_SOURCES += \
+# $(GALLIVM_SOURCES)
+#endif
+
+
+include ../Makefile.template
+
+
+gallivm/gallivm_builtins.cpp: gallivm/llvm_builtins.c
+ clang --emit-llvm < $< |llvm-as|opt -std-compile-opts > temp1.bin
+ (echo "static const unsigned char llvm_builtins_data[] = {"; od -txC temp1.bin | sed -e "s/^[0-9]*//" -e s"/ \([0-9a-f][0-9a-f]\)/0x\1,/g" -e"\$$d" | sed -e"\$$s/,$$/,0x00};/") >$@
+ rm temp1.bin
+
+gallivm/gallivmsoabuiltins.cpp: gallivm/soabuiltins.c
+ clang --emit-llvm < $< |llvm-as|opt -std-compile-opts > temp2.bin
+ (echo "static const unsigned char soabuiltins_data[] = {"; od -txC temp2.bin | sed -e "s/^[0-9]*//" -e s"/ \([0-9a-f][0-9a-f]\)/0x\1,/g" -e"\$$d" | sed -e"\$$s/,$$/,0x00};/") >$@
+ rm temp2.bin
+
+
+indices/u_indices_gen.c: indices/u_indices_gen.py
+ python $< > $@
+
+indices/u_unfilled_gen.c: indices/u_unfilled_gen.py
+ python $< > $@
+
+util/u_format_table.c: util/u_format_table.py util/u_format_parse.py util/u_format.csv
+ python util/u_format_table.py util/u_format.csv > $@
+
+util/u_format_access.c: util/u_format_access.py util/u_format_parse.py util/u_format.csv
+ python util/u_format_access.py util/u_format.csv > $@
diff --git a/src/gallium/auxiliary/SConscript b/src/gallium/auxiliary/SConscript
new file mode 100644
index 00000000000..782eb533863
--- /dev/null
+++ b/src/gallium/auxiliary/SConscript
@@ -0,0 +1,185 @@
+Import('*')
+
+from sys import executable as python_cmd
+
+env.Append(CPPPATH = [
+ 'indices',
+ 'util',
+])
+
+env.CodeGenerate(
+ target = 'indices/u_indices_gen.c',
+ script = 'indices/u_indices_gen.py',
+ source = [],
+ command = python_cmd + ' $SCRIPT > $TARGET'
+)
+
+env.CodeGenerate(
+ target = 'indices/u_unfilled_gen.c',
+ script = 'indices/u_unfilled_gen.py',
+ source = [],
+ command = python_cmd + ' $SCRIPT > $TARGET'
+)
+
+env.CodeGenerate(
+ target = 'util/u_format_table.c',
+ script = 'util/u_format_table.py',
+ source = ['util/u_format.csv'],
+ command = 'python $SCRIPT $SOURCE > $TARGET'
+)
+
+env.CodeGenerate(
+ target = 'util/u_format_access.c',
+ script = 'util/u_format_access.py',
+ source = ['util/u_format.csv'],
+ command = 'python $SCRIPT $SOURCE > $TARGET'
+)
+
+source = [
+ 'cso_cache/cso_context.c',
+ 'cso_cache/cso_cache.c',
+ 'cso_cache/cso_hash.c',
+ 'draw/draw_context.c',
+ 'draw/draw_pipe.c',
+ 'draw/draw_pipe_aaline.c',
+ 'draw/draw_pipe_aapoint.c',
+ 'draw/draw_pipe_clip.c',
+ 'draw/draw_pipe_cull.c',
+ 'draw/draw_pipe_flatshade.c',
+ 'draw/draw_pipe_offset.c',
+ 'draw/draw_pipe_pstipple.c',
+ 'draw/draw_pipe_stipple.c',
+ 'draw/draw_pipe_twoside.c',
+ 'draw/draw_pipe_unfilled.c',
+ 'draw/draw_pipe_util.c',
+ 'draw/draw_pipe_validate.c',
+ 'draw/draw_pipe_vbuf.c',
+ 'draw/draw_pipe_wide_line.c',
+ 'draw/draw_pipe_wide_point.c',
+ 'draw/draw_pt.c',
+ 'draw/draw_pt_elts.c',
+ 'draw/draw_pt_emit.c',
+ 'draw/draw_pt_fetch.c',
+ 'draw/draw_pt_fetch_emit.c',
+ 'draw/draw_pt_fetch_shade_emit.c',
+ 'draw/draw_pt_fetch_shade_pipeline.c',
+ 'draw/draw_pt_post_vs.c',
+ 'draw/draw_pt_util.c',
+ 'draw/draw_pt_varray.c',
+ 'draw/draw_pt_vcache.c',
+ 'draw/draw_vertex.c',
+ 'draw/draw_vs.c',
+ 'draw/draw_vs_aos.c',
+ 'draw/draw_vs_aos_io.c',
+ 'draw/draw_vs_aos_machine.c',
+ 'draw/draw_vs_exec.c',
+ 'draw/draw_vs_llvm.c',
+ 'draw/draw_vs_ppc.c',
+ 'draw/draw_vs_sse.c',
+ 'draw/draw_vs_varient.c',
+ 'draw/draw_gs.c',
+ #'indices/u_indices.c',
+ #'indices/u_unfilled_indices.c',
+ 'indices/u_indices_gen.c',
+ 'indices/u_unfilled_gen.c',
+ 'pipebuffer/pb_buffer_fenced.c',
+ 'pipebuffer/pb_buffer_malloc.c',
+ 'pipebuffer/pb_bufmgr_alt.c',
+ 'pipebuffer/pb_bufmgr_cache.c',
+ 'pipebuffer/pb_bufmgr_debug.c',
+ 'pipebuffer/pb_bufmgr_fenced.c',
+ 'pipebuffer/pb_bufmgr_mm.c',
+ 'pipebuffer/pb_bufmgr_ondemand.c',
+ 'pipebuffer/pb_bufmgr_pool.c',
+ 'pipebuffer/pb_bufmgr_slab.c',
+ 'pipebuffer/pb_validate.c',
+ 'rbug/rbug_core.c',
+ 'rbug/rbug_shader.c',
+ 'rbug/rbug_context.c',
+ 'rbug/rbug_texture.c',
+ 'rbug/rbug_demarshal.c',
+ 'rbug/rbug_connection.c',
+ 'rtasm/rtasm_cpu.c',
+ 'rtasm/rtasm_execmem.c',
+ 'rtasm/rtasm_x86sse.c',
+ 'rtasm/rtasm_ppc.c',
+ 'rtasm/rtasm_ppc_spe.c',
+ 'tgsi/tgsi_build.c',
+ 'tgsi/tgsi_dump.c',
+ 'tgsi/tgsi_dump_c.c',
+ 'tgsi/tgsi_exec.c',
+ 'tgsi/tgsi_info.c',
+ 'tgsi/tgsi_iterate.c',
+ 'tgsi/tgsi_parse.c',
+ 'tgsi/tgsi_sanity.c',
+ 'tgsi/tgsi_scan.c',
+ 'tgsi/tgsi_ppc.c',
+ 'tgsi/tgsi_sse2.c',
+ 'tgsi/tgsi_text.c',
+ 'tgsi/tgsi_transform.c',
+ 'tgsi/tgsi_ureg.c',
+ 'tgsi/tgsi_util.c',
+ 'translate/translate_generic.c',
+ 'translate/translate_sse.c',
+ 'translate/translate.c',
+ 'translate/translate_cache.c',
+ 'util/u_bitmask.c',
+ 'util/u_blit.c',
+ 'util/u_blitter.c',
+ 'util/u_cache.c',
+ 'util/u_cpu_detect.c',
+ 'util/u_debug.c',
+ 'util/u_debug_dump.c',
+ 'util/u_debug_memory.c',
+ 'util/u_debug_stack.c',
+ 'util/u_debug_symbol.c',
+ 'util/u_dl.c',
+ 'util/u_draw_quad.c',
+ 'util/u_format.c',
+ 'util/u_format_access.c',
+ 'util/u_format_table.c',
+ 'util/u_gen_mipmap.c',
+ 'util/u_handle_table.c',
+ 'util/u_hash.c',
+ 'util/u_hash_table.c',
+ 'util/u_keymap.c',
+ 'util/u_network.c',
+ 'util/u_math.c',
+ 'util/u_mm.c',
+ 'util/u_rect.c',
+ 'util/u_simple_shaders.c',
+ 'util/u_snprintf.c',
+ 'util/u_stream_stdc.c',
+ 'util/u_stream_wd.c',
+ 'util/u_surface.c',
+ 'util/u_texture.c',
+ 'util/u_tile.c',
+ 'util/u_time.c',
+ 'util/u_timed_winsys.c',
+ 'util/u_upload_mgr.c',
+ 'util/u_simple_screen.c',
+ 'vl/vl_bitstream_parser.c',
+ 'vl/vl_mpeg12_mc_renderer.c',
+ 'vl/vl_compositor.c',
+ 'vl/vl_csc.c',
+ 'vl/vl_shader_build.c',
+]
+
+if env['llvm']:
+ source += [
+ 'gallivm/gallivm.cpp',
+ 'gallivm/gallivm_cpu.cpp',
+ 'gallivm/instructions.cpp',
+ 'gallivm/loweringpass.cpp',
+ 'gallivm/tgsitollvm.cpp',
+ 'gallivm/storage.cpp',
+ 'gallivm/storagesoa.cpp',
+ 'gallivm/instructionssoa.cpp',
+ ]
+
+gallium = env.ConvenienceLibrary(
+ target = 'gallium',
+ source = source,
+)
+
+Export('gallium')
diff --git a/src/gallium/auxiliary/cso_cache/Makefile b/src/gallium/auxiliary/cso_cache/Makefile
deleted file mode 100644
index 8726afcd949..00000000000
--- a/src/gallium/auxiliary/cso_cache/Makefile
+++ /dev/null
@@ -1,11 +0,0 @@
-TOP = ../../../..
-include $(TOP)/configs/current
-
-LIBNAME = cso_cache
-
-C_SOURCES = \
- cso_context.c \
- cso_cache.c \
- cso_hash.c
-
-include ../../Makefile.template
diff --git a/src/gallium/auxiliary/cso_cache/SConscript b/src/gallium/auxiliary/cso_cache/SConscript
deleted file mode 100644
index 651e68a191a..00000000000
--- a/src/gallium/auxiliary/cso_cache/SConscript
+++ /dev/null
@@ -1,11 +0,0 @@
-Import('*')
-
-cso_cache = env.ConvenienceLibrary(
- target = 'cso_cache',
- source = [
- 'cso_context.c',
- 'cso_cache.c',
- 'cso_hash.c',
- ])
-
-auxiliaries.insert(0, cso_cache)
diff --git a/src/gallium/auxiliary/cso_cache/cso_context.c b/src/gallium/auxiliary/cso_cache/cso_context.c
index 4f13b3e2bad..2b16332e143 100644
--- a/src/gallium/auxiliary/cso_cache/cso_context.c
+++ b/src/gallium/auxiliary/cso_cache/cso_context.c
@@ -42,6 +42,7 @@
#include "cso_cache/cso_context.h"
#include "cso_cache/cso_cache.h"
#include "cso_cache/cso_hash.h"
+#include "cso_context.h"
struct cso_context {
struct pipe_context *pipe;
@@ -50,28 +51,43 @@ struct cso_context {
struct {
void *samplers[PIPE_MAX_SAMPLERS];
unsigned nr_samplers;
+
+ void *vertex_samplers[PIPE_MAX_VERTEX_SAMPLERS];
+ unsigned nr_vertex_samplers;
} hw;
void *samplers[PIPE_MAX_SAMPLERS];
unsigned nr_samplers;
+ void *vertex_samplers[PIPE_MAX_VERTEX_SAMPLERS];
+ unsigned nr_vertex_samplers;
+
unsigned nr_samplers_saved;
void *samplers_saved[PIPE_MAX_SAMPLERS];
+ unsigned nr_vertex_samplers_saved;
+ void *vertex_samplers_saved[PIPE_MAX_VERTEX_SAMPLERS];
+
struct pipe_texture *textures[PIPE_MAX_SAMPLERS];
uint nr_textures;
+ struct pipe_texture *vertex_textures[PIPE_MAX_VERTEX_SAMPLERS];
+ uint nr_vertex_textures;
+
uint nr_textures_saved;
struct pipe_texture *textures_saved[PIPE_MAX_SAMPLERS];
+ uint nr_vertex_textures_saved;
+ struct pipe_texture *vertex_textures_saved[PIPE_MAX_SAMPLERS];
+
/** Current and saved state.
* The saved state is used as a 1-deep stack.
*/
void *blend, *blend_saved;
void *depth_stencil, *depth_stencil_saved;
void *rasterizer, *rasterizer_saved;
- void *fragment_shader, *fragment_shader_saved;
- void *vertex_shader, *vertex_shader_saved;
+ void *fragment_shader, *fragment_shader_saved, *geometry_shader;
+ void *vertex_shader, *vertex_shader_saved, *geometry_shader_saved;
struct pipe_framebuffer_state fb, fb_saved;
struct pipe_viewport_state vp, vp_saved;
@@ -244,7 +260,9 @@ void cso_release_all( struct cso_context *ctx )
if (ctx->pipe) {
ctx->pipe->bind_blend_state( ctx->pipe, NULL );
ctx->pipe->bind_rasterizer_state( ctx->pipe, NULL );
- ctx->pipe->bind_sampler_states( ctx->pipe, 0, NULL );
+ ctx->pipe->bind_fragment_sampler_states( ctx->pipe, 0, NULL );
+ if (ctx->pipe->bind_vertex_sampler_states)
+ ctx->pipe->bind_vertex_sampler_states(ctx->pipe, 0, NULL);
ctx->pipe->bind_depth_stencil_alpha_state( ctx->pipe, NULL );
ctx->pipe->bind_fs_state( ctx->pipe, NULL );
ctx->pipe->bind_vs_state( ctx->pipe, NULL );
@@ -255,6 +273,11 @@ void cso_release_all( struct cso_context *ctx )
pipe_texture_reference(&ctx->textures_saved[i], NULL);
}
+ for (i = 0; i < PIPE_MAX_VERTEX_SAMPLERS; i++) {
+ pipe_texture_reference(&ctx->vertex_textures[i], NULL);
+ pipe_texture_reference(&ctx->vertex_textures_saved[i], NULL);
+ }
+
free_framebuffer_state(&ctx->fb);
free_framebuffer_state(&ctx->fb_saved);
@@ -378,6 +401,46 @@ enum pipe_error cso_single_sampler(struct cso_context *ctx,
return PIPE_OK;
}
+enum pipe_error
+cso_single_vertex_sampler(struct cso_context *ctx,
+ unsigned idx,
+ const struct pipe_sampler_state *templ)
+{
+ void *handle = NULL;
+
+ if (templ != NULL) {
+ unsigned hash_key = cso_construct_key((void*)templ, sizeof(struct pipe_sampler_state));
+ struct cso_hash_iter iter = cso_find_state_template(ctx->cache,
+ hash_key, CSO_SAMPLER,
+ (void*)templ);
+
+ if (cso_hash_iter_is_null(iter)) {
+ struct cso_sampler *cso = MALLOC(sizeof(struct cso_sampler));
+ if (!cso)
+ return PIPE_ERROR_OUT_OF_MEMORY;
+
+ memcpy(&cso->state, templ, sizeof(*templ));
+ cso->data = ctx->pipe->create_sampler_state(ctx->pipe, &cso->state);
+ cso->delete_state = (cso_state_callback)ctx->pipe->delete_sampler_state;
+ cso->context = ctx->pipe;
+
+ iter = cso_insert_state(ctx->cache, hash_key, CSO_SAMPLER, cso);
+ if (cso_hash_iter_is_null(iter)) {
+ FREE(cso);
+ return PIPE_ERROR_OUT_OF_MEMORY;
+ }
+
+ handle = cso->data;
+ }
+ else {
+ handle = ((struct cso_sampler *)cso_hash_iter_data(iter))->data;
+ }
+ }
+
+ ctx->vertex_samplers[idx] = handle;
+ return PIPE_OK;
+}
+
void cso_single_sampler_done( struct cso_context *ctx )
{
unsigned i;
@@ -398,7 +461,36 @@ void cso_single_sampler_done( struct cso_context *ctx )
memcpy(ctx->hw.samplers, ctx->samplers, ctx->nr_samplers * sizeof(void *));
ctx->hw.nr_samplers = ctx->nr_samplers;
- ctx->pipe->bind_sampler_states(ctx->pipe, ctx->nr_samplers, ctx->samplers);
+ ctx->pipe->bind_fragment_sampler_states(ctx->pipe, ctx->nr_samplers, ctx->samplers);
+ }
+}
+
+void
+cso_single_vertex_sampler_done(struct cso_context *ctx)
+{
+ unsigned i;
+
+ /* find highest non-null sampler */
+ for (i = PIPE_MAX_VERTEX_SAMPLERS; i > 0; i--) {
+ if (ctx->vertex_samplers[i - 1] != NULL)
+ break;
+ }
+
+ ctx->nr_vertex_samplers = i;
+
+ if (ctx->hw.nr_vertex_samplers != ctx->nr_vertex_samplers ||
+ memcmp(ctx->hw.vertex_samplers,
+ ctx->vertex_samplers,
+ ctx->nr_vertex_samplers * sizeof(void *)) != 0)
+ {
+ memcpy(ctx->hw.vertex_samplers,
+ ctx->vertex_samplers,
+ ctx->nr_vertex_samplers * sizeof(void *));
+ ctx->hw.nr_vertex_samplers = ctx->nr_vertex_samplers;
+
+ ctx->pipe->bind_vertex_sampler_states(ctx->pipe,
+ ctx->nr_vertex_samplers,
+ ctx->vertex_samplers);
}
}
@@ -447,6 +539,21 @@ void cso_restore_samplers(struct cso_context *ctx)
cso_single_sampler_done( ctx );
}
+void
+cso_save_vertex_samplers(struct cso_context *ctx)
+{
+ ctx->nr_vertex_samplers_saved = ctx->nr_vertex_samplers;
+ memcpy(ctx->vertex_samplers_saved, ctx->vertex_samplers, sizeof(ctx->vertex_samplers));
+}
+
+void
+cso_restore_vertex_samplers(struct cso_context *ctx)
+{
+ ctx->nr_vertex_samplers = ctx->nr_vertex_samplers_saved;
+ memcpy(ctx->vertex_samplers, ctx->vertex_samplers_saved, sizeof(ctx->vertex_samplers));
+ cso_single_vertex_sampler_done(ctx);
+}
+
enum pipe_error cso_set_sampler_textures( struct cso_context *ctx,
uint count,
@@ -461,7 +568,7 @@ enum pipe_error cso_set_sampler_textures( struct cso_context *ctx,
for ( ; i < PIPE_MAX_SAMPLERS; i++)
pipe_texture_reference(&ctx->textures[i], NULL);
- ctx->pipe->set_sampler_textures(ctx->pipe, count, textures);
+ ctx->pipe->set_fragment_sampler_textures(ctx->pipe, count, textures);
return PIPE_OK;
}
@@ -491,13 +598,71 @@ void cso_restore_sampler_textures( struct cso_context *ctx )
for ( ; i < PIPE_MAX_SAMPLERS; i++)
pipe_texture_reference(&ctx->textures[i], NULL);
- ctx->pipe->set_sampler_textures(ctx->pipe, ctx->nr_textures, ctx->textures);
+ ctx->pipe->set_fragment_sampler_textures(ctx->pipe, ctx->nr_textures, ctx->textures);
ctx->nr_textures_saved = 0;
}
+enum pipe_error
+cso_set_vertex_sampler_textures(struct cso_context *ctx,
+ uint count,
+ struct pipe_texture **textures)
+{
+ uint i;
+
+ ctx->nr_vertex_textures = count;
+
+ for (i = 0; i < count; i++) {
+ pipe_texture_reference(&ctx->vertex_textures[i], textures[i]);
+ }
+ for ( ; i < PIPE_MAX_VERTEX_SAMPLERS; i++) {
+ pipe_texture_reference(&ctx->vertex_textures[i], NULL);
+ }
+
+ ctx->pipe->set_vertex_sampler_textures(ctx->pipe, count, textures);
+
+ return PIPE_OK;
+}
+
+void
+cso_save_vertex_sampler_textures(struct cso_context *ctx)
+{
+ uint i;
+
+ ctx->nr_vertex_textures_saved = ctx->nr_vertex_textures;
+ for (i = 0; i < ctx->nr_vertex_textures; i++) {
+ assert(!ctx->vertex_textures_saved[i]);
+ pipe_texture_reference(&ctx->vertex_textures_saved[i], ctx->vertex_textures[i]);
+ }
+}
+
+void
+cso_restore_vertex_sampler_textures(struct cso_context *ctx)
+{
+ uint i;
+
+ ctx->nr_vertex_textures = ctx->nr_vertex_textures_saved;
+
+ for (i = 0; i < ctx->nr_vertex_textures; i++) {
+ pipe_texture_reference(&ctx->vertex_textures[i], NULL);
+ ctx->vertex_textures[i] = ctx->vertex_textures_saved[i];
+ ctx->vertex_textures_saved[i] = NULL;
+ }
+ for ( ; i < PIPE_MAX_VERTEX_SAMPLERS; i++) {
+ pipe_texture_reference(&ctx->vertex_textures[i], NULL);
+ }
+
+ ctx->pipe->set_vertex_sampler_textures(ctx->pipe,
+ ctx->nr_vertex_textures,
+ ctx->vertex_textures);
+
+ ctx->nr_vertex_textures_saved = 0;
+}
+
+
+
enum pipe_error cso_set_depth_stencil_alpha(struct cso_context *ctx,
const struct pipe_depth_stencil_alpha_state *templ)
{
@@ -863,3 +1028,38 @@ enum pipe_error cso_set_blend_color(struct cso_context *ctx,
}
return PIPE_OK;
}
+
+enum pipe_error cso_set_geometry_shader_handle(struct cso_context *ctx,
+ void *handle)
+{
+ if (ctx->geometry_shader != handle) {
+ ctx->geometry_shader = handle;
+ ctx->pipe->bind_gs_state(ctx->pipe, handle);
+ }
+ return PIPE_OK;
+}
+
+void cso_delete_geometry_shader(struct cso_context *ctx, void *handle)
+{
+ if (handle == ctx->geometry_shader) {
+ /* unbind before deleting */
+ ctx->pipe->bind_gs_state(ctx->pipe, NULL);
+ ctx->geometry_shader = NULL;
+ }
+ ctx->pipe->delete_gs_state(ctx->pipe, handle);
+}
+
+void cso_save_geometry_shader(struct cso_context *ctx)
+{
+ assert(!ctx->geometry_shader_saved);
+ ctx->geometry_shader_saved = ctx->geometry_shader;
+}
+
+void cso_restore_geometry_shader(struct cso_context *ctx)
+{
+ if (ctx->geometry_shader_saved != ctx->geometry_shader) {
+ ctx->pipe->bind_gs_state(ctx->pipe, ctx->geometry_shader_saved);
+ ctx->geometry_shader = ctx->geometry_shader_saved;
+ }
+ ctx->geometry_shader_saved = NULL;
+}
diff --git a/src/gallium/auxiliary/cso_cache/cso_context.h b/src/gallium/auxiliary/cso_cache/cso_context.h
index 69630e98bae..b9e313e32d6 100644
--- a/src/gallium/auxiliary/cso_cache/cso_context.h
+++ b/src/gallium/auxiliary/cso_cache/cso_context.h
@@ -84,6 +84,20 @@ enum pipe_error cso_single_sampler( struct cso_context *cso,
void cso_single_sampler_done( struct cso_context *cso );
+void
+cso_save_vertex_samplers(struct cso_context *cso);
+
+void
+cso_restore_vertex_samplers(struct cso_context *cso);
+
+enum pipe_error
+cso_single_vertex_sampler(struct cso_context *cso,
+ unsigned nr,
+ const struct pipe_sampler_state *states);
+
+void
+cso_single_vertex_sampler_done(struct cso_context *cso);
+
enum pipe_error cso_set_sampler_textures( struct cso_context *cso,
@@ -94,6 +108,17 @@ void cso_restore_sampler_textures( struct cso_context *cso );
+enum pipe_error
+cso_set_vertex_sampler_textures(struct cso_context *cso,
+ uint count,
+ struct pipe_texture **textures);
+void
+cso_save_vertex_sampler_textures(struct cso_context *cso);
+void
+cso_restore_vertex_sampler_textures(struct cso_context *cso);
+
+
+
/* These aren't really sensible -- most of the time the api provides
* object semantics for shaders anyway, and the cases where it doesn't
* (eg mesa's internall-generated texenv programs), it will be up to
@@ -121,6 +146,13 @@ void cso_save_vertex_shader(struct cso_context *cso);
void cso_restore_vertex_shader(struct cso_context *cso);
+enum pipe_error cso_set_geometry_shader_handle(struct cso_context *ctx,
+ void *handle);
+void cso_delete_geometry_shader(struct cso_context *ctx, void *handle);
+void cso_save_geometry_shader(struct cso_context *cso);
+void cso_restore_geometry_shader(struct cso_context *cso);
+
+
enum pipe_error cso_set_framebuffer(struct cso_context *cso,
const struct pipe_framebuffer_state *fb);
diff --git a/src/gallium/auxiliary/draw/Makefile b/src/gallium/auxiliary/draw/Makefile
deleted file mode 100644
index 5041dcc072b..00000000000
--- a/src/gallium/auxiliary/draw/Makefile
+++ /dev/null
@@ -1,46 +0,0 @@
-TOP = ../../../..
-include $(TOP)/configs/current
-
-LIBNAME = draw
-
-C_SOURCES = \
- draw_context.c \
- draw_pipe.c \
- draw_pipe_aaline.c \
- draw_pipe_aapoint.c \
- draw_pipe_clip.c \
- draw_pipe_cull.c \
- draw_pipe_flatshade.c \
- draw_pipe_offset.c \
- draw_pipe_pstipple.c \
- draw_pipe_stipple.c \
- draw_pipe_twoside.c \
- draw_pipe_unfilled.c \
- draw_pipe_util.c \
- draw_pipe_validate.c \
- draw_pipe_vbuf.c \
- draw_pipe_wide_line.c \
- draw_pipe_wide_point.c \
- draw_pt.c \
- draw_pt_elts.c \
- draw_pt_emit.c \
- draw_pt_fetch.c \
- draw_pt_fetch_emit.c \
- draw_pt_fetch_shade_emit.c \
- draw_pt_fetch_shade_pipeline.c \
- draw_pt_post_vs.c \
- draw_pt_util.c \
- draw_pt_varray.c \
- draw_pt_vcache.c \
- draw_vertex.c \
- draw_vs.c \
- draw_vs_varient.c \
- draw_vs_aos.c \
- draw_vs_aos_io.c \
- draw_vs_aos_machine.c \
- draw_vs_exec.c \
- draw_vs_llvm.c \
- draw_vs_ppc.c \
- draw_vs_sse.c
-
-include ../../Makefile.template
diff --git a/src/gallium/auxiliary/draw/SConscript b/src/gallium/auxiliary/draw/SConscript
deleted file mode 100644
index 5f05aa324a5..00000000000
--- a/src/gallium/auxiliary/draw/SConscript
+++ /dev/null
@@ -1,46 +0,0 @@
-Import('*')
-
-draw = env.ConvenienceLibrary(
- target = 'draw',
- source = [
- 'draw_context.c',
- 'draw_pipe.c',
- 'draw_pipe_aaline.c',
- 'draw_pipe_aapoint.c',
- 'draw_pipe_clip.c',
- 'draw_pipe_cull.c',
- 'draw_pipe_flatshade.c',
- 'draw_pipe_offset.c',
- 'draw_pipe_pstipple.c',
- 'draw_pipe_stipple.c',
- 'draw_pipe_twoside.c',
- 'draw_pipe_unfilled.c',
- 'draw_pipe_util.c',
- 'draw_pipe_validate.c',
- 'draw_pipe_vbuf.c',
- 'draw_pipe_wide_line.c',
- 'draw_pipe_wide_point.c',
- 'draw_pt.c',
- 'draw_pt_elts.c',
- 'draw_pt_emit.c',
- 'draw_pt_fetch.c',
- 'draw_pt_fetch_emit.c',
- 'draw_pt_fetch_shade_emit.c',
- 'draw_pt_fetch_shade_pipeline.c',
- 'draw_pt_post_vs.c',
- 'draw_pt_util.c',
- 'draw_pt_varray.c',
- 'draw_pt_vcache.c',
- 'draw_vertex.c',
- 'draw_vs.c',
- 'draw_vs_aos.c',
- 'draw_vs_aos_io.c',
- 'draw_vs_aos_machine.c',
- 'draw_vs_exec.c',
- 'draw_vs_llvm.c',
- 'draw_vs_ppc.c',
- 'draw_vs_sse.c',
- 'draw_vs_varient.c'
- ])
-
-auxiliaries.insert(0, draw)
diff --git a/src/gallium/auxiliary/draw/draw_context.c b/src/gallium/auxiliary/draw/draw_context.c
index a4f1fcddc1a..667aa46b208 100644
--- a/src/gallium/auxiliary/draw/draw_context.c
+++ b/src/gallium/auxiliary/draw/draw_context.c
@@ -36,6 +36,7 @@
#include "draw_context.h"
#include "draw_vbuf.h"
#include "draw_vs.h"
+#include "draw_gs.h"
#include "draw_pt.h"
#include "draw_pipe.h"
@@ -67,6 +68,9 @@ struct draw_context *draw_create( void )
if (!draw_vs_init( draw ))
goto fail;
+ if (!draw_gs_init( draw ))
+ goto fail;
+
return draw;
fail:
@@ -231,11 +235,19 @@ draw_set_mapped_vertex_buffer(struct draw_context *draw,
void
draw_set_mapped_constant_buffer(struct draw_context *draw,
- const void *buffer,
+ unsigned shader_type,
+ const void *buffer,
unsigned size )
{
- draw->pt.user.constants = buffer;
- draw_vs_set_constants( draw, (const float (*)[4])buffer, size );
+ debug_assert(shader_type == PIPE_SHADER_VERTEX ||
+ shader_type == PIPE_SHADER_GEOMETRY);
+ if (shader_type == PIPE_SHADER_VERTEX) {
+ draw->pt.user.vs_constants = buffer;
+ draw_vs_set_constants( draw, (const float (*)[4])buffer, size );
+ } else if (shader_type == PIPE_SHADER_GEOMETRY) {
+ draw->pt.user.gs_constants = buffer;
+ draw_gs_set_constants( draw, (const float (*)[4])buffer, size );
+ }
}
@@ -298,7 +310,7 @@ draw_set_force_passthrough( struct draw_context *draw, boolean enable )
* a post-transformed vertex.
*
* With this function, drivers that use the draw module should have no reason
- * to track the current vertex shader.
+ * to track the current vertex/geometry shader.
*
* Note that the draw module may sometimes generate vertices with extra
* attributes (such as texcoords for AA lines). The driver can call this
@@ -309,43 +321,59 @@ draw_set_force_passthrough( struct draw_context *draw, boolean enable )
* work for the drivers.
*/
int
-draw_find_vs_output(const struct draw_context *draw,
- uint semantic_name, uint semantic_index)
+draw_find_shader_output(const struct draw_context *draw,
+ uint semantic_name, uint semantic_index)
{
const struct draw_vertex_shader *vs = draw->vs.vertex_shader;
+ const struct draw_geometry_shader *gs = draw->gs.geometry_shader;
uint i;
- for (i = 0; i < vs->info.num_outputs; i++) {
- if (vs->info.output_semantic_name[i] == semantic_name &&
- vs->info.output_semantic_index[i] == semantic_index)
+ const struct tgsi_shader_info *info = &vs->info;
+
+ if (gs)
+ info = &gs->info;
+
+ for (i = 0; i < info->num_outputs; i++) {
+ if (info->output_semantic_name[i] == semantic_name &&
+ info->output_semantic_index[i] == semantic_index)
return i;
}
/* XXX there may be more than one extra vertex attrib.
* For example, simulated gl_FragCoord and gl_PointCoord.
*/
- if (draw->extra_vp_outputs.semantic_name == semantic_name &&
- draw->extra_vp_outputs.semantic_index == semantic_index) {
- return draw->extra_vp_outputs.slot;
+ if (draw->extra_shader_outputs.semantic_name == semantic_name &&
+ draw->extra_shader_outputs.semantic_index == semantic_index) {
+ return draw->extra_shader_outputs.slot;
}
+
return 0;
}
/**
- * Return number of vertex shader outputs.
+ * Return number of the shader outputs.
+ *
+ * If geometry shader is present, its output will be returned,
+ * if not vertex shader is used.
*/
uint
-draw_num_vs_outputs(const struct draw_context *draw)
+draw_num_shader_outputs(const struct draw_context *draw)
{
uint count = draw->vs.vertex_shader->info.num_outputs;
- if (draw->extra_vp_outputs.slot > 0)
+
+ /* if geometry shader is present, its outputs go to te
+ * driver, not the vertex shaders */
+ if (draw->gs.geometry_shader)
+ count = draw->gs.geometry_shader->info.num_outputs;
+
+ if (draw->extra_shader_outputs.slot > 0)
count++;
return count;
}
/**
- * Provide TGSI sampler objects for vertex shaders that use texture fetches.
+ * Provide TGSI sampler objects for vertex/geometry shaders that use texture fetches.
* This might only be used by software drivers for the time being.
*/
void
@@ -355,6 +383,8 @@ draw_texture_samplers(struct draw_context *draw,
{
draw->vs.num_samplers = num_samplers;
draw->vs.samplers = samplers;
+ draw->gs.num_samplers = num_samplers;
+ draw->gs.samplers = samplers;
}
@@ -366,13 +396,6 @@ void draw_set_render( struct draw_context *draw,
draw->render = render;
}
-void draw_set_edgeflags( struct draw_context *draw,
- const unsigned *edgeflag )
-{
- draw->pt.user.edgeflag = edgeflag;
-}
-
-
/**
@@ -428,3 +451,18 @@ void draw_do_flush( struct draw_context *draw, unsigned flags )
draw->flushing = FALSE;
}
}
+
+
+int draw_current_shader_outputs(struct draw_context *draw)
+{
+ if (draw->gs.geometry_shader)
+ return draw->gs.num_gs_outputs;
+ return draw->vs.num_vs_outputs;
+}
+
+int draw_current_shader_position_output(struct draw_context *draw)
+{
+ if (draw->gs.geometry_shader)
+ return draw->gs.position_output;
+ return draw->vs.position_output;
+}
diff --git a/src/gallium/auxiliary/draw/draw_context.h b/src/gallium/auxiliary/draw/draw_context.h
index d529e4e9a27..8a64c06efcd 100644
--- a/src/gallium/auxiliary/draw/draw_context.h
+++ b/src/gallium/auxiliary/draw/draw_context.h
@@ -45,6 +45,7 @@ struct pipe_context;
struct draw_context;
struct draw_stage;
struct draw_vertex_shader;
+struct draw_geometry_shader;
struct tgsi_sampler;
@@ -85,11 +86,11 @@ draw_install_pstipple_stage(struct draw_context *draw, struct pipe_context *pipe
int
-draw_find_vs_output(const struct draw_context *draw,
- uint semantic_name, uint semantic_index);
+draw_find_shader_output(const struct draw_context *draw,
+ uint semantic_name, uint semantic_index);
uint
-draw_num_vs_outputs(const struct draw_context *draw);
+draw_num_shader_outputs(const struct draw_context *draw);
void
@@ -112,6 +113,17 @@ void draw_delete_vertex_shader(struct draw_context *draw,
struct draw_vertex_shader *dvs);
+/*
+ * Geometry shader functions
+ */
+struct draw_geometry_shader *
+draw_create_geometry_shader(struct draw_context *draw,
+ const struct pipe_shader_state *shader);
+void draw_bind_geometry_shader(struct draw_context *draw,
+ struct draw_geometry_shader *dvs);
+void draw_delete_geometry_shader(struct draw_context *draw,
+ struct draw_geometry_shader *dvs);
+
/*
* Vertex data functions
@@ -140,12 +152,10 @@ void draw_set_mapped_vertex_buffer(struct draw_context *draw,
unsigned attr, const void *buffer);
void draw_set_mapped_constant_buffer(struct draw_context *draw,
+ unsigned shader_type,
const void *buffer,
unsigned size );
-void draw_set_edgeflags( struct draw_context *draw,
- const unsigned *edgeflag );
-
/***********************************************************************
* draw_prim.c
@@ -154,6 +164,14 @@ void draw_set_edgeflags( struct draw_context *draw,
void draw_arrays(struct draw_context *draw, unsigned prim,
unsigned start, unsigned count);
+void
+draw_arrays_instanced(struct draw_context *draw,
+ unsigned mode,
+ unsigned start,
+ unsigned count,
+ unsigned startInstance,
+ unsigned instanceCount);
+
void draw_flush(struct draw_context *draw);
diff --git a/src/gallium/auxiliary/draw/draw_gs.c b/src/gallium/auxiliary/draw/draw_gs.c
new file mode 100644
index 00000000000..5db2e755423
--- /dev/null
+++ b/src/gallium/auxiliary/draw/draw_gs.c
@@ -0,0 +1,338 @@
+/**************************************************************************
+ *
+ * Copyright 2009 VMWare Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+#include "draw_gs.h"
+
+#include "draw_private.h"
+#include "draw_context.h"
+
+#include "tgsi/tgsi_parse.h"
+#include "tgsi/tgsi_exec.h"
+
+#include "pipe/p_shader_tokens.h"
+
+#include "util/u_math.h"
+#include "util/u_memory.h"
+
+#define MAX_PRIM_VERTICES 6
+/* fixme: move it from here */
+#define MAX_PRIMITIVES 64
+
+boolean
+draw_gs_init( struct draw_context *draw )
+{
+ draw->gs.machine = tgsi_exec_machine_create();
+ if (!draw->gs.machine)
+ return FALSE;
+
+ draw->gs.machine->Primitives = align_malloc(
+ MAX_PRIMITIVES * sizeof(struct tgsi_exec_vector), 16);
+ if (!draw->gs.machine->Primitives)
+ return FALSE;
+ memset(draw->gs.machine->Primitives, 0,
+ MAX_PRIMITIVES * sizeof(struct tgsi_exec_vector));
+
+ return TRUE;
+}
+
+
+void draw_gs_set_constants( struct draw_context *draw,
+ const float (*constants)[4],
+ unsigned size )
+{
+}
+
+
+struct draw_geometry_shader *
+draw_create_geometry_shader(struct draw_context *draw,
+ const struct pipe_shader_state *state)
+{
+ struct draw_geometry_shader *gs;
+ int i;
+
+ gs = CALLOC_STRUCT(draw_geometry_shader);
+
+ if (!gs)
+ return NULL;
+
+ gs->state = *state;
+ gs->state.tokens = tgsi_dup_tokens(state->tokens);
+ if (!gs->state.tokens) {
+ FREE(gs);
+ return NULL;
+ }
+
+ tgsi_scan_shader(state->tokens, &gs->info);
+
+ /* setup the defaults */
+ gs->input_primitive = PIPE_PRIM_TRIANGLES;
+ gs->output_primitive = PIPE_PRIM_TRIANGLE_STRIP;
+ gs->max_output_vertices = 32;
+
+ for (i = 0; i < gs->info.num_properties; ++i) {
+ if (gs->info.properties[i].name ==
+ TGSI_PROPERTY_GS_INPUT_PRIM)
+ gs->input_primitive = gs->info.properties[i].data[0];
+ else if (gs->info.properties[i].name ==
+ TGSI_PROPERTY_GS_OUTPUT_PRIM)
+ gs->output_primitive = gs->info.properties[i].data[0];
+ else if (gs->info.properties[i].name ==
+ TGSI_PROPERTY_GS_MAX_VERTICES)
+ gs->max_output_vertices = gs->info.properties[i].data[0];
+ }
+
+ gs->machine = draw->gs.machine;
+
+ if (gs)
+ {
+ uint i;
+ for (i = 0; i < gs->info.num_outputs; i++) {
+ if (gs->info.output_semantic_name[i] == TGSI_SEMANTIC_POSITION &&
+ gs->info.output_semantic_index[i] == 0)
+ gs->position_output = i;
+ }
+ }
+
+ return gs;
+}
+
+void draw_bind_geometry_shader(struct draw_context *draw,
+ struct draw_geometry_shader *dgs)
+{
+ draw_do_flush(draw, DRAW_FLUSH_STATE_CHANGE);
+
+ if (dgs) {
+ draw->gs.geometry_shader = dgs;
+ draw->gs.num_gs_outputs = dgs->info.num_outputs;
+ draw->gs.position_output = dgs->position_output;
+ draw_geometry_shader_prepare(dgs, draw);
+ }
+ else {
+ draw->gs.geometry_shader = NULL;
+ draw->gs.num_gs_outputs = 0;
+ }
+}
+
+void draw_delete_geometry_shader(struct draw_context *draw,
+ struct draw_geometry_shader *dgs)
+{
+ FREE(dgs);
+}
+
+static INLINE int num_vertices_for_prim(int prim)
+{
+ switch(prim) {
+ case PIPE_PRIM_POINTS:
+ return 1;
+ case PIPE_PRIM_LINES:
+ return 2;
+ case PIPE_PRIM_LINE_LOOP:
+ return 2;
+ case PIPE_PRIM_LINE_STRIP:
+ return 2;
+ case PIPE_PRIM_TRIANGLES:
+ return 3;
+ case PIPE_PRIM_TRIANGLE_STRIP:
+ return 3;
+ case PIPE_PRIM_TRIANGLE_FAN:
+ return 3;
+ case PIPE_PRIM_LINES_ADJACENCY:
+ case PIPE_PRIM_LINE_STRIP_ADJACENCY:
+ return 4;
+ case PIPE_PRIM_TRIANGLES_ADJACENCY:
+ case PIPE_PRIM_TRIANGLE_STRIP_ADJACENCY:
+ return 6;
+ default:
+ assert(!"Bad geometry shader input");
+ return 0;
+ }
+}
+
+static void draw_fetch_geometry_input(struct draw_geometry_shader *shader,
+ int start_primitive,
+ int num_primitives,
+ const float (*input_ptr)[4],
+ unsigned input_vertex_stride,
+ unsigned inputs_from_vs)
+{
+ struct tgsi_exec_machine *machine = shader->machine;
+ unsigned slot, vs_slot, k, j;
+ unsigned num_vertices = num_vertices_for_prim(shader->input_primitive);
+ int idx = 0;
+
+ for (slot = 0, vs_slot = 0; slot < shader->info.num_inputs; slot++) {
+ /*debug_printf("Slot = %d (semantic = %d)\n", slot,
+ shader->info.input_semantic_name[slot]);*/
+ if (shader->info.input_semantic_name[slot] ==
+ TGSI_SEMANTIC_PRIMID) {
+ for (j = 0; j < num_primitives; ++j) {
+ machine->Inputs[idx].xyzw[0].f[j] = (float)start_primitive + j;
+ machine->Inputs[idx].xyzw[1].f[j] = (float)start_primitive + j;
+ machine->Inputs[idx].xyzw[2].f[j] = (float)start_primitive + j;
+ machine->Inputs[idx].xyzw[3].f[j] = (float)start_primitive + j;
+ }
+ ++idx;
+ } else {
+ for (j = 0; j < num_primitives; ++j) {
+ int vidx = idx;
+ const float (*prim_ptr)[4];
+ /*debug_printf(" %d) Prim (num_verts = %d)\n", start_primitive + j,
+ num_vertices);*/
+ prim_ptr = (const float (*)[4])(
+ (const char *)input_ptr +
+ (j * num_vertices * input_vertex_stride));
+
+ for (k = 0; k < num_vertices; ++k, ++vidx) {
+ const float (*input)[4];
+ input = (const float (*)[4])(
+ (const char *)prim_ptr + (k * input_vertex_stride));
+ vidx = k * TGSI_EXEC_MAX_INPUT_ATTRIBS + slot;
+ /*debug_printf("\t%d)(%d) Input vert:\n", vidx, k);*/
+#if 1
+ assert(!util_is_inf_or_nan(input[vs_slot][0]));
+ assert(!util_is_inf_or_nan(input[vs_slot][1]));
+ assert(!util_is_inf_or_nan(input[vs_slot][2]));
+ assert(!util_is_inf_or_nan(input[vs_slot][3]));
+#endif
+ machine->Inputs[vidx].xyzw[0].f[j] = input[vs_slot][0];
+ machine->Inputs[vidx].xyzw[1].f[j] = input[vs_slot][1];
+ machine->Inputs[vidx].xyzw[2].f[j] = input[vs_slot][2];
+ machine->Inputs[vidx].xyzw[3].f[j] = input[vs_slot][3];
+#if 0
+ debug_printf("\t\t%d %f %f %f %f\n", slot,
+ machine->Inputs[vidx].xyzw[0].f[j],
+ machine->Inputs[vidx].xyzw[1].f[j],
+ machine->Inputs[vidx].xyzw[2].f[j],
+ machine->Inputs[vidx].xyzw[3].f[j]);
+#endif
+ }
+ }
+ ++vs_slot;
+ idx += num_vertices;
+ }
+ }
+}
+
+static INLINE void
+draw_geometry_fetch_outputs(struct draw_geometry_shader *shader,
+ int num_primitives,
+ float (*output)[4],
+ unsigned vertex_size)
+{
+ struct tgsi_exec_machine *machine = shader->machine;
+ unsigned prim_idx, j, slot;
+
+ /* Unswizzle all output results.
+ */
+ /* FIXME: handle all the primitives produced by the gs, not just
+ * the first one
+ unsigned prim_count =
+ mach->Temps[TEMP_PRIMITIVE_I].xyzw[TEMP_PRIMITIVE_C].u[0];*/
+ for (prim_idx = 0; prim_idx < num_primitives; ++prim_idx) {
+ unsigned num_verts_per_prim = machine->Primitives[0];
+ for (j = 0; j < num_verts_per_prim; j++) {
+ int idx = (prim_idx * num_verts_per_prim + j) *
+ shader->info.num_outputs;
+#ifdef DEBUG_OUTPUTS
+ debug_printf("%d) Output vert:\n", idx);
+#endif
+ for (slot = 0; slot < shader->info.num_outputs; slot++) {
+ output[slot][0] = machine->Outputs[idx + slot].xyzw[0].f[prim_idx];
+ output[slot][1] = machine->Outputs[idx + slot].xyzw[1].f[prim_idx];
+ output[slot][2] = machine->Outputs[idx + slot].xyzw[2].f[prim_idx];
+ output[slot][3] = machine->Outputs[idx + slot].xyzw[3].f[prim_idx];
+#ifdef DEBUG_OUTPUTS
+ debug_printf("\t%d: %f %f %f %f\n", slot,
+ output[slot][0],
+ output[slot][1],
+ output[slot][2],
+ output[slot][3]);
+#endif
+ debug_assert(!util_is_inf_or_nan(output[slot][0]));
+ }
+ output = (float (*)[4])((char *)output + vertex_size);
+ }
+ }
+}
+
+void draw_geometry_shader_run(struct draw_geometry_shader *shader,
+ const float (*input)[4],
+ float (*output)[4],
+ const float (*constants)[4],
+ unsigned count,
+ unsigned input_stride,
+ unsigned vertex_size)
+{
+ struct tgsi_exec_machine *machine = shader->machine;
+ unsigned int i;
+ unsigned num_vertices = num_vertices_for_prim(shader->input_primitive);
+ unsigned num_primitives = count/num_vertices;
+ unsigned inputs_from_vs = 0;
+
+ machine->Consts = constants;
+
+ for (i = 0; i < shader->info.num_inputs; ++i) {
+ if (shader->info.input_semantic_name[i] != TGSI_SEMANTIC_PRIMID)
+ ++inputs_from_vs;
+ }
+
+ for (i = 0; i < num_primitives; ++i) {
+ unsigned int max_primitives = 1;
+
+ draw_fetch_geometry_input(shader, i, max_primitives, input,
+ input_stride, inputs_from_vs);
+
+ tgsi_set_exec_mask(machine,
+ 1,
+ max_primitives > 1,
+ max_primitives > 2,
+ max_primitives > 3);
+
+ /* run interpreter */
+ tgsi_exec_machine_run(machine);
+
+ draw_geometry_fetch_outputs(shader, max_primitives,
+ output, vertex_size);
+ }
+}
+
+void draw_geometry_shader_delete(struct draw_geometry_shader *shader)
+{
+ FREE((void*) shader->state.tokens);
+ FREE(shader);
+}
+
+void draw_geometry_shader_prepare(struct draw_geometry_shader *shader,
+ struct draw_context *draw)
+{
+ if (shader->machine->Tokens != shader->state.tokens) {
+ tgsi_exec_machine_bind_shader(shader->machine,
+ shader->state.tokens,
+ draw->gs.num_samplers,
+ draw->gs.samplers);
+ }
+}
diff --git a/src/gallium/auxiliary/draw/draw_gs.h b/src/gallium/auxiliary/draw/draw_gs.h
new file mode 100644
index 00000000000..d6a97d9c4ef
--- /dev/null
+++ b/src/gallium/auxiliary/draw/draw_gs.h
@@ -0,0 +1,76 @@
+/**************************************************************************
+ *
+ * Copyright 2009 VMWare Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+#ifndef DRAW_GS_H
+#define DRAW_GS_H
+
+#include "draw_context.h"
+#include "draw_private.h"
+
+
+#define MAX_TGSI_PRIMITIVES 4
+
+struct draw_context;
+
+/**
+ * Private version of the compiled geometry shader
+ */
+struct draw_geometry_shader {
+ struct draw_context *draw;
+
+ struct tgsi_exec_machine *machine;
+
+ /* This member will disappear shortly:*/
+ struct pipe_shader_state state;
+
+ struct tgsi_shader_info info;
+ unsigned position_output;
+
+ unsigned max_output_vertices;
+ unsigned input_primitive;
+ unsigned output_primitive;
+
+ /* Extracted from shader:
+ */
+ const float (*immediates)[4];
+};
+
+void draw_geometry_shader_run(struct draw_geometry_shader *shader,
+ const float (*input)[4],
+ float (*output)[4],
+ const float (*constants)[4],
+ unsigned count,
+ unsigned input_stride,
+ unsigned output_stride);
+
+void draw_geometry_shader_prepare(struct draw_geometry_shader *shader,
+ struct draw_context *draw);
+
+void draw_geometry_shader_delete(struct draw_geometry_shader *shader);
+
+
+#endif
diff --git a/src/gallium/auxiliary/draw/draw_pipe_aaline.c b/src/gallium/auxiliary/draw/draw_pipe_aaline.c
index 9f956715a22..4585dcdb48a 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_aaline.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_aaline.c
@@ -35,6 +35,8 @@
#include "pipe/p_context.h"
#include "pipe/p_defines.h"
#include "pipe/p_shader_tokens.h"
+
+#include "util/u_format.h"
#include "util/u_math.h"
#include "util/u_memory.h"
@@ -139,29 +141,29 @@ aa_transform_decl(struct tgsi_transform_context *ctx,
struct aa_transform_context *aactx = (struct aa_transform_context *) ctx;
if (decl->Declaration.File == TGSI_FILE_OUTPUT &&
- decl->Semantic.SemanticName == TGSI_SEMANTIC_COLOR &&
- decl->Semantic.SemanticIndex == 0) {
- aactx->colorOutput = decl->DeclarationRange.First;
+ decl->Semantic.Name == TGSI_SEMANTIC_COLOR &&
+ decl->Semantic.Index == 0) {
+ aactx->colorOutput = decl->Range.First;
}
else if (decl->Declaration.File == TGSI_FILE_SAMPLER) {
uint i;
- for (i = decl->DeclarationRange.First;
- i <= decl->DeclarationRange.Last; i++) {
+ for (i = decl->Range.First;
+ i <= decl->Range.Last; i++) {
aactx->samplersUsed |= 1 << i;
}
}
else if (decl->Declaration.File == TGSI_FILE_INPUT) {
- if ((int) decl->DeclarationRange.Last > aactx->maxInput)
- aactx->maxInput = decl->DeclarationRange.Last;
- if (decl->Semantic.SemanticName == TGSI_SEMANTIC_GENERIC &&
- (int) decl->Semantic.SemanticIndex > aactx->maxGeneric) {
- aactx->maxGeneric = decl->Semantic.SemanticIndex;
+ if ((int) decl->Range.Last > aactx->maxInput)
+ aactx->maxInput = decl->Range.Last;
+ if (decl->Semantic.Name == TGSI_SEMANTIC_GENERIC &&
+ (int) decl->Semantic.Index > aactx->maxGeneric) {
+ aactx->maxGeneric = decl->Semantic.Index;
}
}
else if (decl->Declaration.File == TGSI_FILE_TEMPORARY) {
uint i;
- for (i = decl->DeclarationRange.First;
- i <= decl->DeclarationRange.Last; i++) {
+ for (i = decl->Range.First;
+ i <= decl->Range.Last; i++) {
aactx->tempsUsed |= (1 << i);
}
}
@@ -228,30 +230,30 @@ aa_transform_inst(struct tgsi_transform_context *ctx,
/* XXX this could be linear... */
decl.Declaration.Interpolate = TGSI_INTERPOLATE_PERSPECTIVE;
decl.Declaration.Semantic = 1;
- decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC;
- decl.Semantic.SemanticIndex = aactx->maxGeneric + 1;
- decl.DeclarationRange.First =
- decl.DeclarationRange.Last = aactx->maxInput + 1;
+ decl.Semantic.Name = TGSI_SEMANTIC_GENERIC;
+ decl.Semantic.Index = aactx->maxGeneric + 1;
+ decl.Range.First =
+ decl.Range.Last = aactx->maxInput + 1;
ctx->emit_declaration(ctx, &decl);
/* declare new sampler */
decl = tgsi_default_full_declaration();
decl.Declaration.File = TGSI_FILE_SAMPLER;
- decl.DeclarationRange.First =
- decl.DeclarationRange.Last = aactx->freeSampler;
+ decl.Range.First =
+ decl.Range.Last = aactx->freeSampler;
ctx->emit_declaration(ctx, &decl);
/* declare new temp regs */
decl = tgsi_default_full_declaration();
decl.Declaration.File = TGSI_FILE_TEMPORARY;
- decl.DeclarationRange.First =
- decl.DeclarationRange.Last = aactx->texTemp;
+ decl.Range.First =
+ decl.Range.Last = aactx->texTemp;
ctx->emit_declaration(ctx, &decl);
decl = tgsi_default_full_declaration();
decl.Declaration.File = TGSI_FILE_TEMPORARY;
- decl.DeclarationRange.First =
- decl.DeclarationRange.Last = aactx->colorTemp;
+ decl.Range.First =
+ decl.Range.Last = aactx->colorTemp;
ctx->emit_declaration(ctx, &decl);
aactx->firstInstruction = FALSE;
@@ -265,14 +267,15 @@ aa_transform_inst(struct tgsi_transform_context *ctx,
newInst = tgsi_default_full_instruction();
newInst.Instruction.Opcode = TGSI_OPCODE_TEX;
newInst.Instruction.NumDstRegs = 1;
- newInst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullDstRegisters[0].DstRegister.Index = aactx->texTemp;
+ newInst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Dst[0].Register.Index = aactx->texTemp;
newInst.Instruction.NumSrcRegs = 2;
- newInst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT;
- newInst.FullSrcRegisters[0].SrcRegister.Index = aactx->maxInput + 1;
- newInst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER;
- newInst.FullSrcRegisters[1].SrcRegister.Index = aactx->freeSampler;
+ newInst.Instruction.Texture = TRUE;
+ newInst.Texture.Texture = TGSI_TEXTURE_2D;
+ newInst.Src[0].Register.File = TGSI_FILE_INPUT;
+ newInst.Src[0].Register.Index = aactx->maxInput + 1;
+ newInst.Src[1].Register.File = TGSI_FILE_SAMPLER;
+ newInst.Src[1].Register.Index = aactx->freeSampler;
ctx->emit_instruction(ctx, &newInst);
@@ -280,26 +283,26 @@ aa_transform_inst(struct tgsi_transform_context *ctx,
newInst = tgsi_default_full_instruction();
newInst.Instruction.Opcode = TGSI_OPCODE_MOV;
newInst.Instruction.NumDstRegs = 1;
- newInst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT;
- newInst.FullDstRegisters[0].DstRegister.Index = aactx->colorOutput;
- newInst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_XYZ;
+ newInst.Dst[0].Register.File = TGSI_FILE_OUTPUT;
+ newInst.Dst[0].Register.Index = aactx->colorOutput;
+ newInst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_XYZ;
newInst.Instruction.NumSrcRegs = 1;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[0].SrcRegister.Index = aactx->colorTemp;
+ newInst.Src[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[0].Register.Index = aactx->colorTemp;
ctx->emit_instruction(ctx, &newInst);
/* MUL alpha */
newInst = tgsi_default_full_instruction();
newInst.Instruction.Opcode = TGSI_OPCODE_MUL;
newInst.Instruction.NumDstRegs = 1;
- newInst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT;
- newInst.FullDstRegisters[0].DstRegister.Index = aactx->colorOutput;
- newInst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_W;
+ newInst.Dst[0].Register.File = TGSI_FILE_OUTPUT;
+ newInst.Dst[0].Register.Index = aactx->colorOutput;
+ newInst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_W;
newInst.Instruction.NumSrcRegs = 2;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[0].SrcRegister.Index = aactx->colorTemp;
- newInst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[1].SrcRegister.Index = aactx->texTemp;
+ newInst.Src[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[0].Register.Index = aactx->colorTemp;
+ newInst.Src[1].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[1].Register.Index = aactx->texTemp;
ctx->emit_instruction(ctx, &newInst);
/* END */
@@ -316,11 +319,11 @@ aa_transform_inst(struct tgsi_transform_context *ctx,
uint i;
for (i = 0; i < inst->Instruction.NumDstRegs; i++) {
- struct tgsi_full_dst_register *dst = &inst->FullDstRegisters[i];
- if (dst->DstRegister.File == TGSI_FILE_OUTPUT &&
- dst->DstRegister.Index == aactx->colorOutput) {
- dst->DstRegister.File = TGSI_FILE_TEMPORARY;
- dst->DstRegister.Index = aactx->colorTemp;
+ struct tgsi_full_dst_register *dst = &inst->Dst[i];
+ if (dst->Register.File == TGSI_FILE_OUTPUT &&
+ dst->Register.Index == aactx->colorOutput) {
+ dst->Register.File = TGSI_FILE_TEMPORARY;
+ dst->Register.Index = aactx->colorTemp;
}
}
@@ -398,10 +401,9 @@ aaline_create_texture(struct aaline_stage *aaline)
texTemp.target = PIPE_TEXTURE_2D;
texTemp.format = PIPE_FORMAT_A8_UNORM; /* XXX verify supported by driver! */
texTemp.last_level = MAX_TEXTURE_LEVEL;
- texTemp.width[0] = 1 << MAX_TEXTURE_LEVEL;
- texTemp.height[0] = 1 << MAX_TEXTURE_LEVEL;
- texTemp.depth[0] = 1;
- pf_get_block(texTemp.format, &texTemp.block);
+ texTemp.width0 = 1 << MAX_TEXTURE_LEVEL;
+ texTemp.height0 = 1 << MAX_TEXTURE_LEVEL;
+ texTemp.depth0 = 1;
aaline->texture = screen->texture_create(screen, &texTemp);
if (!aaline->texture)
@@ -413,11 +415,11 @@ aaline_create_texture(struct aaline_stage *aaline)
*/
for (level = 0; level <= MAX_TEXTURE_LEVEL; level++) {
struct pipe_transfer *transfer;
- const uint size = aaline->texture->width[level];
+ const uint size = u_minify(aaline->texture->width0, level);
ubyte *data;
uint i, j;
- assert(aaline->texture->width[level] == aaline->texture->height[level]);
+ assert(aaline->texture->width0 == aaline->texture->height0);
/* This texture is new, no need to flush.
*/
@@ -658,13 +660,13 @@ aaline_first_line(struct draw_stage *stage, struct prim_header *header)
}
/* update vertex attrib info */
- aaline->tex_slot = draw->vs.num_vs_outputs;
- aaline->pos_slot = draw->vs.position_output;
+ aaline->tex_slot = draw_current_shader_outputs(draw);
+ aaline->pos_slot = draw_current_shader_position_output(draw);;
/* advertise the extra post-transformed vertex attribute */
- draw->extra_vp_outputs.semantic_name = TGSI_SEMANTIC_GENERIC;
- draw->extra_vp_outputs.semantic_index = aaline->fs->generic_attrib;
- draw->extra_vp_outputs.slot = aaline->tex_slot;
+ draw->extra_shader_outputs.semantic_name = TGSI_SEMANTIC_GENERIC;
+ draw->extra_shader_outputs.semantic_index = aaline->fs->generic_attrib;
+ draw->extra_shader_outputs.slot = aaline->tex_slot;
/* how many samplers? */
/* we'll use sampler/texture[pstip->sampler_unit] for the stipple */
@@ -705,7 +707,7 @@ aaline_flush(struct draw_stage *stage, unsigned flags)
aaline->state.texture);
draw->suspend_flushing = FALSE;
- draw->extra_vp_outputs.slot = 0;
+ draw->extra_shader_outputs.slot = 0;
}
@@ -896,16 +898,16 @@ draw_install_aaline_stage(struct draw_context *draw, struct pipe_context *pipe)
aaline->driver_bind_fs_state = pipe->bind_fs_state;
aaline->driver_delete_fs_state = pipe->delete_fs_state;
- aaline->driver_bind_sampler_states = pipe->bind_sampler_states;
- aaline->driver_set_sampler_textures = pipe->set_sampler_textures;
+ aaline->driver_bind_sampler_states = pipe->bind_fragment_sampler_states;
+ aaline->driver_set_sampler_textures = pipe->set_fragment_sampler_textures;
/* override the driver's functions */
pipe->create_fs_state = aaline_create_fs_state;
pipe->bind_fs_state = aaline_bind_fs_state;
pipe->delete_fs_state = aaline_delete_fs_state;
- pipe->bind_sampler_states = aaline_bind_sampler_states;
- pipe->set_sampler_textures = aaline_set_sampler_textures;
+ pipe->bind_fragment_sampler_states = aaline_bind_sampler_states;
+ pipe->set_fragment_sampler_textures = aaline_set_sampler_textures;
/* Install once everything is known to be OK:
*/
diff --git a/src/gallium/auxiliary/draw/draw_pipe_aapoint.c b/src/gallium/auxiliary/draw/draw_pipe_aapoint.c
index ae1712fe122..d86717e5182 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_aapoint.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_aapoint.c
@@ -131,22 +131,22 @@ aa_transform_decl(struct tgsi_transform_context *ctx,
struct aa_transform_context *aactx = (struct aa_transform_context *) ctx;
if (decl->Declaration.File == TGSI_FILE_OUTPUT &&
- decl->Semantic.SemanticName == TGSI_SEMANTIC_COLOR &&
- decl->Semantic.SemanticIndex == 0) {
- aactx->colorOutput = decl->DeclarationRange.First;
+ decl->Semantic.Name == TGSI_SEMANTIC_COLOR &&
+ decl->Semantic.Index == 0) {
+ aactx->colorOutput = decl->Range.First;
}
else if (decl->Declaration.File == TGSI_FILE_INPUT) {
- if ((int) decl->DeclarationRange.Last > aactx->maxInput)
- aactx->maxInput = decl->DeclarationRange.Last;
- if (decl->Semantic.SemanticName == TGSI_SEMANTIC_GENERIC &&
- (int) decl->Semantic.SemanticIndex > aactx->maxGeneric) {
- aactx->maxGeneric = decl->Semantic.SemanticIndex;
+ if ((int) decl->Range.Last > aactx->maxInput)
+ aactx->maxInput = decl->Range.Last;
+ if (decl->Semantic.Name == TGSI_SEMANTIC_GENERIC &&
+ (int) decl->Semantic.Index > aactx->maxGeneric) {
+ aactx->maxGeneric = decl->Semantic.Index;
}
}
else if (decl->Declaration.File == TGSI_FILE_TEMPORARY) {
uint i;
- for (i = decl->DeclarationRange.First;
- i <= decl->DeclarationRange.Last; i++) {
+ for (i = decl->Range.First;
+ i <= decl->Range.Last; i++) {
aactx->tempsUsed |= (1 << i);
}
}
@@ -198,23 +198,23 @@ aa_transform_inst(struct tgsi_transform_context *ctx,
/* XXX this could be linear... */
decl.Declaration.Interpolate = TGSI_INTERPOLATE_PERSPECTIVE;
decl.Declaration.Semantic = 1;
- decl.Semantic.SemanticName = TGSI_SEMANTIC_GENERIC;
- decl.Semantic.SemanticIndex = aactx->maxGeneric + 1;
- decl.DeclarationRange.First =
- decl.DeclarationRange.Last = texInput;
+ decl.Semantic.Name = TGSI_SEMANTIC_GENERIC;
+ decl.Semantic.Index = aactx->maxGeneric + 1;
+ decl.Range.First =
+ decl.Range.Last = texInput;
ctx->emit_declaration(ctx, &decl);
/* declare new temp regs */
decl = tgsi_default_full_declaration();
decl.Declaration.File = TGSI_FILE_TEMPORARY;
- decl.DeclarationRange.First =
- decl.DeclarationRange.Last = tmp0;
+ decl.Range.First =
+ decl.Range.Last = tmp0;
ctx->emit_declaration(ctx, &decl);
decl = tgsi_default_full_declaration();
decl.Declaration.File = TGSI_FILE_TEMPORARY;
- decl.DeclarationRange.First =
- decl.DeclarationRange.Last = aactx->colorTemp;
+ decl.Range.First =
+ decl.Range.Last = aactx->colorTemp;
ctx->emit_declaration(ctx, &decl);
aactx->firstInstruction = FALSE;
@@ -234,30 +234,30 @@ aa_transform_inst(struct tgsi_transform_context *ctx,
newInst = tgsi_default_full_instruction();
newInst.Instruction.Opcode = TGSI_OPCODE_MUL;
newInst.Instruction.NumDstRegs = 1;
- newInst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullDstRegisters[0].DstRegister.Index = tmp0;
- newInst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_XY;
+ newInst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Dst[0].Register.Index = tmp0;
+ newInst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_XY;
newInst.Instruction.NumSrcRegs = 2;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT;
- newInst.FullSrcRegisters[0].SrcRegister.Index = texInput;
- newInst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_INPUT;
- newInst.FullSrcRegisters[1].SrcRegister.Index = texInput;
+ newInst.Src[0].Register.File = TGSI_FILE_INPUT;
+ newInst.Src[0].Register.Index = texInput;
+ newInst.Src[1].Register.File = TGSI_FILE_INPUT;
+ newInst.Src[1].Register.Index = texInput;
ctx->emit_instruction(ctx, &newInst);
/* ADD t0.x, t0.x, t0.y; # x^2 + y^2 */
newInst = tgsi_default_full_instruction();
newInst.Instruction.Opcode = TGSI_OPCODE_ADD;
newInst.Instruction.NumDstRegs = 1;
- newInst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullDstRegisters[0].DstRegister.Index = tmp0;
- newInst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X;
+ newInst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Dst[0].Register.Index = tmp0;
+ newInst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_X;
newInst.Instruction.NumSrcRegs = 2;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[0].SrcRegister.Index = tmp0;
- newInst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
- newInst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[1].SrcRegister.Index = tmp0;
- newInst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y;
+ newInst.Src[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[0].Register.Index = tmp0;
+ newInst.Src[0].Register.SwizzleX = TGSI_SWIZZLE_X;
+ newInst.Src[1].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[1].Register.Index = tmp0;
+ newInst.Src[1].Register.SwizzleX = TGSI_SWIZZLE_Y;
ctx->emit_instruction(ctx, &newInst);
#if NORMALIZE /* OPTIONAL normalization of length */
@@ -265,24 +265,24 @@ aa_transform_inst(struct tgsi_transform_context *ctx,
newInst = tgsi_default_full_instruction();
newInst.Instruction.Opcode = TGSI_OPCODE_RSQ;
newInst.Instruction.NumDstRegs = 1;
- newInst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullDstRegisters[0].DstRegister.Index = tmp0;
- newInst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X;
+ newInst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Dst[0].Register.Index = tmp0;
+ newInst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_X;
newInst.Instruction.NumSrcRegs = 1;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[0].SrcRegister.Index = tmp0;
+ newInst.Src[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[0].Register.Index = tmp0;
ctx->emit_instruction(ctx, &newInst);
/* RCP t0.x, t0.x; */
newInst = tgsi_default_full_instruction();
newInst.Instruction.Opcode = TGSI_OPCODE_RCP;
newInst.Instruction.NumDstRegs = 1;
- newInst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullDstRegisters[0].DstRegister.Index = tmp0;
- newInst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X;
+ newInst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Dst[0].Register.Index = tmp0;
+ newInst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_X;
newInst.Instruction.NumSrcRegs = 1;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[0].SrcRegister.Index = tmp0;
+ newInst.Src[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[0].Register.Index = tmp0;
ctx->emit_instruction(ctx, &newInst);
#endif
@@ -290,16 +290,16 @@ aa_transform_inst(struct tgsi_transform_context *ctx,
newInst = tgsi_default_full_instruction();
newInst.Instruction.Opcode = TGSI_OPCODE_SGT;
newInst.Instruction.NumDstRegs = 1;
- newInst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullDstRegisters[0].DstRegister.Index = tmp0;
- newInst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_Y;
+ newInst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Dst[0].Register.Index = tmp0;
+ newInst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_Y;
newInst.Instruction.NumSrcRegs = 2;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[0].SrcRegister.Index = tmp0;
- newInst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
- newInst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_INPUT;
- newInst.FullSrcRegisters[1].SrcRegister.Index = texInput;
- newInst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_W;
+ newInst.Src[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[0].Register.Index = tmp0;
+ newInst.Src[0].Register.SwizzleY = TGSI_SWIZZLE_X;
+ newInst.Src[1].Register.File = TGSI_FILE_INPUT;
+ newInst.Src[1].Register.Index = texInput;
+ newInst.Src[1].Register.SwizzleY = TGSI_SWIZZLE_W;
ctx->emit_instruction(ctx, &newInst);
/* KIL -tmp0.yyyy; # if -tmp0.y < 0, KILL */
@@ -307,13 +307,13 @@ aa_transform_inst(struct tgsi_transform_context *ctx,
newInst.Instruction.Opcode = TGSI_OPCODE_KIL;
newInst.Instruction.NumDstRegs = 0;
newInst.Instruction.NumSrcRegs = 1;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[0].SrcRegister.Index = tmp0;
- newInst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y;
- newInst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y;
- newInst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y;
- newInst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y;
- newInst.FullSrcRegisters[0].SrcRegister.Negate = 1;
+ newInst.Src[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[0].Register.Index = tmp0;
+ newInst.Src[0].Register.SwizzleX = TGSI_SWIZZLE_Y;
+ newInst.Src[0].Register.SwizzleY = TGSI_SWIZZLE_Y;
+ newInst.Src[0].Register.SwizzleZ = TGSI_SWIZZLE_Y;
+ newInst.Src[0].Register.SwizzleW = TGSI_SWIZZLE_Y;
+ newInst.Src[0].Register.Negate = 1;
ctx->emit_instruction(ctx, &newInst);
@@ -323,77 +323,77 @@ aa_transform_inst(struct tgsi_transform_context *ctx,
newInst = tgsi_default_full_instruction();
newInst.Instruction.Opcode = TGSI_OPCODE_SUB;
newInst.Instruction.NumDstRegs = 1;
- newInst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullDstRegisters[0].DstRegister.Index = tmp0;
- newInst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_Z;
+ newInst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Dst[0].Register.Index = tmp0;
+ newInst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_Z;
newInst.Instruction.NumSrcRegs = 2;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT;
- newInst.FullSrcRegisters[0].SrcRegister.Index = texInput;
- newInst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_W;
- newInst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_INPUT;
- newInst.FullSrcRegisters[1].SrcRegister.Index = texInput;
- newInst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Z;
+ newInst.Src[0].Register.File = TGSI_FILE_INPUT;
+ newInst.Src[0].Register.Index = texInput;
+ newInst.Src[0].Register.SwizzleZ = TGSI_SWIZZLE_W;
+ newInst.Src[1].Register.File = TGSI_FILE_INPUT;
+ newInst.Src[1].Register.Index = texInput;
+ newInst.Src[1].Register.SwizzleZ = TGSI_SWIZZLE_Z;
ctx->emit_instruction(ctx, &newInst);
/* RCP t0.z, t0.z; # t0.z = 1 / m */
newInst = tgsi_default_full_instruction();
newInst.Instruction.Opcode = TGSI_OPCODE_RCP;
newInst.Instruction.NumDstRegs = 1;
- newInst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullDstRegisters[0].DstRegister.Index = tmp0;
- newInst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_Z;
+ newInst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Dst[0].Register.Index = tmp0;
+ newInst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_Z;
newInst.Instruction.NumSrcRegs = 1;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[0].SrcRegister.Index = tmp0;
- newInst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Z;
+ newInst.Src[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[0].Register.Index = tmp0;
+ newInst.Src[0].Register.SwizzleX = TGSI_SWIZZLE_Z;
ctx->emit_instruction(ctx, &newInst);
/* SUB t0.y, 1, t0.x; # d = 1 - d */
newInst = tgsi_default_full_instruction();
newInst.Instruction.Opcode = TGSI_OPCODE_SUB;
newInst.Instruction.NumDstRegs = 1;
- newInst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullDstRegisters[0].DstRegister.Index = tmp0;
- newInst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_Y;
+ newInst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Dst[0].Register.Index = tmp0;
+ newInst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_Y;
newInst.Instruction.NumSrcRegs = 2;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT;
- newInst.FullSrcRegisters[0].SrcRegister.Index = texInput;
- newInst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_W;
- newInst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[1].SrcRegister.Index = tmp0;
- newInst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
+ newInst.Src[0].Register.File = TGSI_FILE_INPUT;
+ newInst.Src[0].Register.Index = texInput;
+ newInst.Src[0].Register.SwizzleY = TGSI_SWIZZLE_W;
+ newInst.Src[1].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[1].Register.Index = tmp0;
+ newInst.Src[1].Register.SwizzleY = TGSI_SWIZZLE_X;
ctx->emit_instruction(ctx, &newInst);
/* MUL t0.w, t0.y, t0.z; # coverage = d * m */
newInst = tgsi_default_full_instruction();
newInst.Instruction.Opcode = TGSI_OPCODE_MUL;
newInst.Instruction.NumDstRegs = 1;
- newInst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullDstRegisters[0].DstRegister.Index = tmp0;
- newInst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_W;
+ newInst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Dst[0].Register.Index = tmp0;
+ newInst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_W;
newInst.Instruction.NumSrcRegs = 2;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[0].SrcRegister.Index = tmp0;
- newInst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y;
- newInst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[1].SrcRegister.Index = tmp0;
- newInst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_Z;
+ newInst.Src[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[0].Register.Index = tmp0;
+ newInst.Src[0].Register.SwizzleW = TGSI_SWIZZLE_Y;
+ newInst.Src[1].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[1].Register.Index = tmp0;
+ newInst.Src[1].Register.SwizzleW = TGSI_SWIZZLE_Z;
ctx->emit_instruction(ctx, &newInst);
/* SLE t0.y, t0.x, tex.z; # bool b = distance <= k */
newInst = tgsi_default_full_instruction();
newInst.Instruction.Opcode = TGSI_OPCODE_SLE;
newInst.Instruction.NumDstRegs = 1;
- newInst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullDstRegisters[0].DstRegister.Index = tmp0;
- newInst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_Y;
+ newInst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Dst[0].Register.Index = tmp0;
+ newInst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_Y;
newInst.Instruction.NumSrcRegs = 2;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[0].SrcRegister.Index = tmp0;
- newInst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
- newInst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_INPUT;
- newInst.FullSrcRegisters[1].SrcRegister.Index = texInput;
- newInst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Z;
+ newInst.Src[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[0].Register.Index = tmp0;
+ newInst.Src[0].Register.SwizzleY = TGSI_SWIZZLE_X;
+ newInst.Src[1].Register.File = TGSI_FILE_INPUT;
+ newInst.Src[1].Register.Index = texInput;
+ newInst.Src[1].Register.SwizzleY = TGSI_SWIZZLE_Z;
ctx->emit_instruction(ctx, &newInst);
/* CMP t0.w, -t0.y, tex.w, t0.w;
@@ -405,29 +405,29 @@ aa_transform_inst(struct tgsi_transform_context *ctx,
newInst = tgsi_default_full_instruction();
newInst.Instruction.Opcode = TGSI_OPCODE_CMP;
newInst.Instruction.NumDstRegs = 1;
- newInst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullDstRegisters[0].DstRegister.Index = tmp0;
- newInst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_W;
+ newInst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Dst[0].Register.Index = tmp0;
+ newInst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_W;
newInst.Instruction.NumSrcRegs = 3;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[0].SrcRegister.Index = tmp0;
- newInst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y;
- newInst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y;
- newInst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y;
- newInst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y;
- newInst.FullSrcRegisters[0].SrcRegister.Negate = 1;
- newInst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_INPUT;
- newInst.FullSrcRegisters[1].SrcRegister.Index = texInput;
- newInst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_W;
- newInst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_W;
- newInst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_W;
- newInst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_W;
- newInst.FullSrcRegisters[2].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[2].SrcRegister.Index = tmp0;
- newInst.FullSrcRegisters[2].SrcRegister.SwizzleX = TGSI_SWIZZLE_W;
- newInst.FullSrcRegisters[2].SrcRegister.SwizzleY = TGSI_SWIZZLE_W;
- newInst.FullSrcRegisters[2].SrcRegister.SwizzleZ = TGSI_SWIZZLE_W;
- newInst.FullSrcRegisters[2].SrcRegister.SwizzleW = TGSI_SWIZZLE_W;
+ newInst.Src[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[0].Register.Index = tmp0;
+ newInst.Src[0].Register.SwizzleX = TGSI_SWIZZLE_Y;
+ newInst.Src[0].Register.SwizzleY = TGSI_SWIZZLE_Y;
+ newInst.Src[0].Register.SwizzleZ = TGSI_SWIZZLE_Y;
+ newInst.Src[0].Register.SwizzleW = TGSI_SWIZZLE_Y;
+ newInst.Src[0].Register.Negate = 1;
+ newInst.Src[1].Register.File = TGSI_FILE_INPUT;
+ newInst.Src[1].Register.Index = texInput;
+ newInst.Src[1].Register.SwizzleX = TGSI_SWIZZLE_W;
+ newInst.Src[1].Register.SwizzleY = TGSI_SWIZZLE_W;
+ newInst.Src[1].Register.SwizzleZ = TGSI_SWIZZLE_W;
+ newInst.Src[1].Register.SwizzleW = TGSI_SWIZZLE_W;
+ newInst.Src[2].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[2].Register.Index = tmp0;
+ newInst.Src[2].Register.SwizzleX = TGSI_SWIZZLE_W;
+ newInst.Src[2].Register.SwizzleY = TGSI_SWIZZLE_W;
+ newInst.Src[2].Register.SwizzleZ = TGSI_SWIZZLE_W;
+ newInst.Src[2].Register.SwizzleW = TGSI_SWIZZLE_W;
ctx->emit_instruction(ctx, &newInst);
}
@@ -439,26 +439,26 @@ aa_transform_inst(struct tgsi_transform_context *ctx,
newInst = tgsi_default_full_instruction();
newInst.Instruction.Opcode = TGSI_OPCODE_MOV;
newInst.Instruction.NumDstRegs = 1;
- newInst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT;
- newInst.FullDstRegisters[0].DstRegister.Index = aactx->colorOutput;
- newInst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_XYZ;
+ newInst.Dst[0].Register.File = TGSI_FILE_OUTPUT;
+ newInst.Dst[0].Register.Index = aactx->colorOutput;
+ newInst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_XYZ;
newInst.Instruction.NumSrcRegs = 1;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[0].SrcRegister.Index = aactx->colorTemp;
+ newInst.Src[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[0].Register.Index = aactx->colorTemp;
ctx->emit_instruction(ctx, &newInst);
/* MUL result.color.w, colorTemp, tmp0.w; */
newInst = tgsi_default_full_instruction();
newInst.Instruction.Opcode = TGSI_OPCODE_MUL;
newInst.Instruction.NumDstRegs = 1;
- newInst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT;
- newInst.FullDstRegisters[0].DstRegister.Index = aactx->colorOutput;
- newInst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_W;
+ newInst.Dst[0].Register.File = TGSI_FILE_OUTPUT;
+ newInst.Dst[0].Register.Index = aactx->colorOutput;
+ newInst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_W;
newInst.Instruction.NumSrcRegs = 2;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[0].SrcRegister.Index = aactx->colorTemp;
- newInst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[1].SrcRegister.Index = aactx->tmp0;
+ newInst.Src[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[0].Register.Index = aactx->colorTemp;
+ newInst.Src[1].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[1].Register.Index = aactx->tmp0;
ctx->emit_instruction(ctx, &newInst);
}
else {
@@ -468,11 +468,11 @@ aa_transform_inst(struct tgsi_transform_context *ctx,
uint i;
for (i = 0; i < inst->Instruction.NumDstRegs; i++) {
- struct tgsi_full_dst_register *dst = &inst->FullDstRegisters[i];
- if (dst->DstRegister.File == TGSI_FILE_OUTPUT &&
- dst->DstRegister.Index == aactx->colorOutput) {
- dst->DstRegister.File = TGSI_FILE_TEMPORARY;
- dst->DstRegister.Index = aactx->colorTemp;
+ struct tgsi_full_dst_register *dst = &inst->Dst[i];
+ if (dst->Register.File == TGSI_FILE_OUTPUT &&
+ dst->Register.Index == aactx->colorOutput) {
+ dst->Register.File = TGSI_FILE_TEMPORARY;
+ dst->Register.Index = aactx->colorTemp;
}
}
}
@@ -687,14 +687,14 @@ aapoint_first_point(struct draw_stage *stage, struct prim_header *header)
bind_aapoint_fragment_shader(aapoint);
/* update vertex attrib info */
- aapoint->tex_slot = draw->vs.num_vs_outputs;
+ aapoint->tex_slot = draw_current_shader_outputs(draw);
assert(aapoint->tex_slot > 0); /* output[0] is vertex pos */
- aapoint->pos_slot = draw->vs.position_output;
+ aapoint->pos_slot = draw_current_shader_position_output(draw);
- draw->extra_vp_outputs.semantic_name = TGSI_SEMANTIC_GENERIC;
- draw->extra_vp_outputs.semantic_index = aapoint->fs->generic_attrib;
- draw->extra_vp_outputs.slot = aapoint->tex_slot;
+ draw->extra_shader_outputs.semantic_name = TGSI_SEMANTIC_GENERIC;
+ draw->extra_shader_outputs.semantic_index = aapoint->fs->generic_attrib;
+ draw->extra_shader_outputs.slot = aapoint->tex_slot;
/* find psize slot in post-transform vertex */
aapoint->psize_slot = -1;
@@ -731,7 +731,7 @@ aapoint_flush(struct draw_stage *stage, unsigned flags)
aapoint->driver_bind_fs_state(pipe, aapoint->fs->driver_fs);
draw->suspend_flushing = FALSE;
- draw->extra_vp_outputs.slot = 0;
+ draw->extra_shader_outputs.slot = 0;
}
diff --git a/src/gallium/auxiliary/draw/draw_pipe_clip.c b/src/gallium/auxiliary/draw/draw_pipe_clip.c
index 0670268a196..51a6115ebf5 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_clip.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_clip.c
@@ -55,7 +55,7 @@
-struct clipper {
+struct clip_stage {
struct draw_stage stage; /**< base class */
/* Basically duplicate some of the flatshading logic here:
@@ -70,9 +70,9 @@ struct clipper {
/* This is a bit confusing:
*/
-static INLINE struct clipper *clipper_stage( struct draw_stage *stage )
+static INLINE struct clip_stage *clip_stage( struct draw_stage *stage )
{
- return (struct clipper *)stage;
+ return (struct clip_stage *)stage;
}
@@ -92,11 +92,12 @@ static void interp_attr( float *fdst,
fdst[3] = LINTERP( t, fout[3], fin[3] );
}
+
static void copy_colors( struct draw_stage *stage,
struct vertex_header *dst,
const struct vertex_header *src )
{
- const struct clipper *clipper = clipper_stage(stage);
+ const struct clip_stage *clipper = clip_stage(stage);
uint i;
for (i = 0; i < clipper->num_color_attribs; i++) {
const uint attr = clipper->color_attribs[i];
@@ -108,14 +109,14 @@ static void copy_colors( struct draw_stage *stage,
/* Interpolate between two vertices to produce a third.
*/
-static void interp( const struct clipper *clip,
+static void interp( const struct clip_stage *clip,
struct vertex_header *dst,
float t,
const struct vertex_header *out,
const struct vertex_header *in )
{
- const unsigned nr_attrs = clip->stage.draw->vs.num_vs_outputs;
- const unsigned pos_attr = clip->stage.draw->vs.position_output;
+ const unsigned nr_attrs = draw_current_shader_outputs(clip->stage.draw);
+ const unsigned pos_attr = draw_current_shader_position_output(clip->stage.draw);
unsigned j;
/* Vertex header.
@@ -179,7 +180,7 @@ static void emit_poly( struct draw_stage *stage,
header.v[2] = inlist[0]; /* keep in v[2] for flatshading */
if (i == n-1)
- header.flags |= edge_last;
+ header.flags |= edge_last;
if (0) {
const struct draw_vertex_shader *vs = stage->draw->vs.vertex_shader;
@@ -200,13 +201,14 @@ static void emit_poly( struct draw_stage *stage,
}
}
+
static INLINE float
dot4(const float *a, const float *b)
{
- return (a[0]*b[0] +
- a[1]*b[1] +
- a[2]*b[2] +
- a[3]*b[3]);
+ return (a[0] * b[0] +
+ a[1] * b[1] +
+ a[2] * b[2] +
+ a[3] * b[3]);
}
@@ -217,7 +219,7 @@ do_clip_tri( struct draw_stage *stage,
struct prim_header *header,
unsigned clipmask )
{
- struct clipper *clipper = clipper_stage( stage );
+ struct clip_stage *clipper = clip_stage( stage );
struct vertex_header *a[MAX_CLIPPED_VERTICES];
struct vertex_header *b[MAX_CLIPPED_VERTICES];
struct vertex_header **inlist = a;
@@ -280,6 +282,7 @@ do_clip_tri( struct draw_stage *stage,
dp_prev = dp;
}
+ /* swap in/out lists */
{
struct vertex_header **tmp = inlist;
inlist = outlist;
@@ -291,15 +294,11 @@ do_clip_tri( struct draw_stage *stage,
/* If flat-shading, copy color to new provoking vertex.
*/
if (clipper->flat && inlist[0] != header->v[2]) {
- if (1) {
- inlist[0] = dup_vert(stage, inlist[0], tmpnr++);
- }
+ inlist[0] = dup_vert(stage, inlist[0], tmpnr++);
copy_colors(stage, inlist[0], header->v[2]);
}
-
-
/* Emit the polygon as triangles to the setup stage:
*/
if (n >= 3)
@@ -314,7 +313,7 @@ do_clip_line( struct draw_stage *stage,
struct prim_header *header,
unsigned clipmask )
{
- const struct clipper *clipper = clipper_stage( stage );
+ const struct clip_stage *clipper = clip_stage( stage );
struct vertex_header *v0 = header->v[0];
struct vertex_header *v1 = header->v[1];
const float *pos0 = v0->clip;
@@ -416,13 +415,14 @@ clip_tri( struct draw_stage *stage,
}
}
+
/* Update state. Could further delay this until we hit the first
* primitive that really requires clipping.
*/
static void
clip_init_state( struct draw_stage *stage )
{
- struct clipper *clipper = clipper_stage( stage );
+ struct clip_stage *clipper = clip_stage( stage );
clipper->flat = stage->draw->rasterizer->flatshade ? TRUE : FALSE;
@@ -488,7 +488,7 @@ static void clip_destroy( struct draw_stage *stage )
*/
struct draw_stage *draw_clip_stage( struct draw_context *draw )
{
- struct clipper *clipper = CALLOC_STRUCT(clipper);
+ struct clip_stage *clipper = CALLOC_STRUCT(clip_stage);
if (clipper == NULL)
goto fail;
diff --git a/src/gallium/auxiliary/draw/draw_pipe_cull.c b/src/gallium/auxiliary/draw/draw_pipe_cull.c
index 0a70483858c..11b39db5990 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_cull.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_cull.c
@@ -55,7 +55,7 @@ static INLINE struct cull_stage *cull_stage( struct draw_stage *stage )
static void cull_tri( struct draw_stage *stage,
struct prim_header *header )
{
- const unsigned pos = stage->draw->vs.position_output;
+ const unsigned pos = draw_current_shader_position_output(stage->draw);
/* Window coords: */
const float *v0 = header->v[0]->data[pos];
diff --git a/src/gallium/auxiliary/draw/draw_pipe_offset.c b/src/gallium/auxiliary/draw/draw_pipe_offset.c
index 40798a5d6e7..e829492423e 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_offset.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_offset.c
@@ -63,7 +63,7 @@ static INLINE struct offset_stage *offset_stage( struct draw_stage *stage )
static void do_offset_tri( struct draw_stage *stage,
struct prim_header *header )
{
- const unsigned pos = stage->draw->vs.position_output;
+ const unsigned pos = draw_current_shader_position_output(stage->draw);
struct offset_stage *offset = offset_stage(stage);
float inv_det = 1.0f / header->det;
diff --git a/src/gallium/auxiliary/draw/draw_pipe_pstipple.c b/src/gallium/auxiliary/draw/draw_pipe_pstipple.c
index 283502cdf3e..0cc2b718641 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_pstipple.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_pstipple.c
@@ -38,6 +38,7 @@
#include "pipe/p_defines.h"
#include "pipe/p_shader_tokens.h"
+#include "util/u_format.h"
#include "util/u_math.h"
#include "util/u_memory.h"
@@ -133,20 +134,20 @@ pstip_transform_decl(struct tgsi_transform_context *ctx,
if (decl->Declaration.File == TGSI_FILE_SAMPLER) {
uint i;
- for (i = decl->DeclarationRange.First;
- i <= decl->DeclarationRange.Last; i++) {
+ for (i = decl->Range.First;
+ i <= decl->Range.Last; i++) {
pctx->samplersUsed |= 1 << i;
}
}
else if (decl->Declaration.File == TGSI_FILE_INPUT) {
- pctx->maxInput = MAX2(pctx->maxInput, (int) decl->DeclarationRange.Last);
- if (decl->Semantic.SemanticName == TGSI_SEMANTIC_POSITION)
- pctx->wincoordInput = (int) decl->DeclarationRange.First;
+ pctx->maxInput = MAX2(pctx->maxInput, (int) decl->Range.Last);
+ if (decl->Semantic.Name == TGSI_SEMANTIC_POSITION)
+ pctx->wincoordInput = (int) decl->Range.First;
}
else if (decl->Declaration.File == TGSI_FILE_TEMPORARY) {
uint i;
- for (i = decl->DeclarationRange.First;
- i <= decl->DeclarationRange.Last; i++) {
+ for (i = decl->Range.First;
+ i <= decl->Range.Last; i++) {
pctx->tempsUsed |= (1 << i);
}
}
@@ -226,25 +227,25 @@ pstip_transform_inst(struct tgsi_transform_context *ctx,
decl.Declaration.File = TGSI_FILE_INPUT;
decl.Declaration.Interpolate = TGSI_INTERPOLATE_LINEAR; /* XXX? */
decl.Declaration.Semantic = 1;
- decl.Semantic.SemanticName = TGSI_SEMANTIC_POSITION;
- decl.Semantic.SemanticIndex = 0;
- decl.DeclarationRange.First =
- decl.DeclarationRange.Last = wincoordInput;
+ decl.Semantic.Name = TGSI_SEMANTIC_POSITION;
+ decl.Semantic.Index = 0;
+ decl.Range.First =
+ decl.Range.Last = wincoordInput;
ctx->emit_declaration(ctx, &decl);
}
/* declare new sampler */
decl = tgsi_default_full_declaration();
decl.Declaration.File = TGSI_FILE_SAMPLER;
- decl.DeclarationRange.First =
- decl.DeclarationRange.Last = pctx->freeSampler;
+ decl.Range.First =
+ decl.Range.Last = pctx->freeSampler;
ctx->emit_declaration(ctx, &decl);
/* declare new temp regs */
decl = tgsi_default_full_declaration();
decl.Declaration.File = TGSI_FILE_TEMPORARY;
- decl.DeclarationRange.First =
- decl.DeclarationRange.Last = pctx->texTemp;
+ decl.Range.First =
+ decl.Range.Last = pctx->texTemp;
ctx->emit_declaration(ctx, &decl);
/* emit immediate = {1/32, 1/32, 1, 1}
@@ -280,27 +281,28 @@ pstip_transform_inst(struct tgsi_transform_context *ctx,
newInst = tgsi_default_full_instruction();
newInst.Instruction.Opcode = TGSI_OPCODE_MUL;
newInst.Instruction.NumDstRegs = 1;
- newInst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullDstRegisters[0].DstRegister.Index = pctx->texTemp;
+ newInst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Dst[0].Register.Index = pctx->texTemp;
newInst.Instruction.NumSrcRegs = 2;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT;
- newInst.FullSrcRegisters[0].SrcRegister.Index = wincoordInput;
- newInst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_IMMEDIATE;
- newInst.FullSrcRegisters[1].SrcRegister.Index = pctx->numImmed;
+ newInst.Src[0].Register.File = TGSI_FILE_INPUT;
+ newInst.Src[0].Register.Index = wincoordInput;
+ newInst.Src[1].Register.File = TGSI_FILE_IMMEDIATE;
+ newInst.Src[1].Register.Index = pctx->numImmed;
ctx->emit_instruction(ctx, &newInst);
/* TEX texTemp, texTemp, sampler; */
newInst = tgsi_default_full_instruction();
newInst.Instruction.Opcode = TGSI_OPCODE_TEX;
newInst.Instruction.NumDstRegs = 1;
- newInst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullDstRegisters[0].DstRegister.Index = pctx->texTemp;
+ newInst.Dst[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Dst[0].Register.Index = pctx->texTemp;
newInst.Instruction.NumSrcRegs = 2;
- newInst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[0].SrcRegister.Index = pctx->texTemp;
- newInst.FullSrcRegisters[1].SrcRegister.File = TGSI_FILE_SAMPLER;
- newInst.FullSrcRegisters[1].SrcRegister.Index = pctx->freeSampler;
+ newInst.Instruction.Texture = TRUE;
+ newInst.Texture.Texture = TGSI_TEXTURE_2D;
+ newInst.Src[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[0].Register.Index = pctx->texTemp;
+ newInst.Src[1].Register.File = TGSI_FILE_SAMPLER;
+ newInst.Src[1].Register.Index = pctx->freeSampler;
ctx->emit_instruction(ctx, &newInst);
/* KIL -texTemp; # if -texTemp < 0, KILL fragment */
@@ -308,9 +310,9 @@ pstip_transform_inst(struct tgsi_transform_context *ctx,
newInst.Instruction.Opcode = TGSI_OPCODE_KIL;
newInst.Instruction.NumDstRegs = 0;
newInst.Instruction.NumSrcRegs = 1;
- newInst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_TEMPORARY;
- newInst.FullSrcRegisters[0].SrcRegister.Index = pctx->texTemp;
- newInst.FullSrcRegisters[0].SrcRegister.Negate = 1;
+ newInst.Src[0].Register.File = TGSI_FILE_TEMPORARY;
+ newInst.Src[0].Register.Index = pctx->texTemp;
+ newInst.Src[0].Register.Negate = 1;
ctx->emit_instruction(ctx, &newInst);
}
@@ -427,10 +429,9 @@ pstip_create_texture(struct pstip_stage *pstip)
texTemp.target = PIPE_TEXTURE_2D;
texTemp.format = PIPE_FORMAT_A8_UNORM; /* XXX verify supported by driver! */
texTemp.last_level = 0;
- texTemp.width[0] = 32;
- texTemp.height[0] = 32;
- texTemp.depth[0] = 1;
- pf_get_block(texTemp.format, &texTemp.block);
+ texTemp.width0 = 32;
+ texTemp.height0 = 32;
+ texTemp.depth0 = 1;
pstip->texture = screen->texture_create(screen, &texTemp);
if (pstip->texture == NULL)
@@ -754,8 +755,8 @@ draw_install_pstipple_stage(struct draw_context *draw,
pstip->driver_bind_fs_state = pipe->bind_fs_state;
pstip->driver_delete_fs_state = pipe->delete_fs_state;
- pstip->driver_bind_sampler_states = pipe->bind_sampler_states;
- pstip->driver_set_sampler_textures = pipe->set_sampler_textures;
+ pstip->driver_bind_sampler_states = pipe->bind_fragment_sampler_states;
+ pstip->driver_set_sampler_textures = pipe->set_fragment_sampler_textures;
pstip->driver_set_polygon_stipple = pipe->set_polygon_stipple;
/* override the driver's functions */
@@ -763,8 +764,8 @@ draw_install_pstipple_stage(struct draw_context *draw,
pipe->bind_fs_state = pstip_bind_fs_state;
pipe->delete_fs_state = pstip_delete_fs_state;
- pipe->bind_sampler_states = pstip_bind_sampler_states;
- pipe->set_sampler_textures = pstip_set_sampler_textures;
+ pipe->bind_fragment_sampler_states = pstip_bind_sampler_states;
+ pipe->set_fragment_sampler_textures = pstip_set_sampler_textures;
pipe->set_polygon_stipple = pstip_set_polygon_stipple;
return TRUE;
diff --git a/src/gallium/auxiliary/draw/draw_pipe_stipple.c b/src/gallium/auxiliary/draw/draw_pipe_stipple.c
index 6e921bac278..70fbab9ea76 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_stipple.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_stipple.c
@@ -73,7 +73,8 @@ screen_interp( struct draw_context *draw,
const struct vertex_header *v1 )
{
uint attr;
- for (attr = 0; attr < draw->vs.num_vs_outputs; attr++) {
+ int num_outputs = draw_current_shader_outputs(draw);
+ for (attr = 0; attr < num_outputs; attr++) {
const float *val0 = v0->data[attr];
const float *val1 = v1->data[attr];
float *newv = dst->data[attr];
@@ -121,7 +122,7 @@ stipple_line(struct draw_stage *stage, struct prim_header *header)
struct stipple_stage *stipple = stipple_stage(stage);
struct vertex_header *v0 = header->v[0];
struct vertex_header *v1 = header->v[1];
- const unsigned pos = stage->draw->vs.position_output;
+ const unsigned pos = draw_current_shader_position_output(stage->draw);
const float *pos0 = v0->data[pos];
const float *pos1 = v1->data[pos];
float start = 0;
diff --git a/src/gallium/auxiliary/draw/draw_pipe_vbuf.c b/src/gallium/auxiliary/draw/draw_pipe_vbuf.c
index 1a5269c0de9..d40c0352401 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_vbuf.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_vbuf.c
@@ -138,7 +138,7 @@ emit_vertex( struct vbuf_stage *vbuf,
/* Note: we really do want data[0] here, not data[pos]:
*/
vbuf->translate->set_buffer(vbuf->translate, 0, vertex->data[0], 0);
- vbuf->translate->run(vbuf->translate, 0, 1, vbuf->vertex_ptr);
+ vbuf->translate->run(vbuf->translate, 0, 1, 0, vbuf->vertex_ptr);
if (0) draw_dump_emitted_vertex(vbuf->vinfo, (uint8_t *)vbuf->vertex_ptr);
@@ -271,10 +271,12 @@ vbuf_start_prim( struct vbuf_stage *vbuf, uint prim )
emit_sz = 0;
break;
}
-
+
+ hw_key.element[i].type = TRANSLATE_ELEMENT_NORMAL;
hw_key.element[i].input_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
hw_key.element[i].input_buffer = src_buffer;
hw_key.element[i].input_offset = src_offset;
+ hw_key.element[i].instance_divisor = 0;
hw_key.element[i].output_format = output_format;
hw_key.element[i].output_offset = dst_offset;
diff --git a/src/gallium/auxiliary/draw/draw_pipe_wide_line.c b/src/gallium/auxiliary/draw/draw_pipe_wide_line.c
index f32cbef983d..3073c870825 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_wide_line.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_wide_line.c
@@ -59,7 +59,7 @@ static void wideline_line( struct draw_stage *stage,
struct prim_header *header )
{
/*const struct wideline_stage *wide = wideline_stage(stage);*/
- const unsigned pos = stage->draw->vs.position_output;
+ const unsigned pos = draw_current_shader_position_output(stage->draw);
const float half_width = 0.5f * stage->draw->rasterizer->line_width;
struct prim_header tri;
diff --git a/src/gallium/auxiliary/draw/draw_pipe_wide_point.c b/src/gallium/auxiliary/draw/draw_pipe_wide_point.c
index 7d76a7dbf39..8dc50c0ab43 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_wide_point.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_wide_point.c
@@ -112,7 +112,7 @@ static void set_texcoords(const struct widepoint_stage *wide,
if (wide->point_coord_fs_input >= 0) {
/* put gl_PointCoord into the extra vertex slot */
- uint slot = wide->stage.draw->extra_vp_outputs.slot;
+ uint slot = wide->stage.draw->extra_shader_outputs.slot;
v->data[slot][0] = tc[0];
v->data[slot][1] = tc[1];
v->data[slot][2] = 0.0F;
@@ -130,7 +130,7 @@ static void widepoint_point( struct draw_stage *stage,
struct prim_header *header )
{
const struct widepoint_stage *wide = widepoint_stage(stage);
- const unsigned pos = stage->draw->vs.position_output;
+ const unsigned pos = draw_current_shader_position_output(stage->draw);
const boolean sprite = (boolean) stage->draw->rasterizer->point_sprite;
float half_size;
float left_adj, right_adj, bot_adj, top_adj;
@@ -257,13 +257,13 @@ static void widepoint_first_point( struct draw_stage *stage,
wide->point_coord_fs_input = find_pntc_input_attrib(draw);
/* setup extra vp output (point coord implemented as a texcoord) */
- draw->extra_vp_outputs.semantic_name = TGSI_SEMANTIC_GENERIC;
- draw->extra_vp_outputs.semantic_index = 0;
- draw->extra_vp_outputs.slot = draw->vs.num_vs_outputs;
+ draw->extra_shader_outputs.semantic_name = TGSI_SEMANTIC_GENERIC;
+ draw->extra_shader_outputs.semantic_index = 0;
+ draw->extra_shader_outputs.slot = draw_current_shader_outputs(draw);
}
else {
wide->point_coord_fs_input = -1;
- draw->extra_vp_outputs.slot = 0;
+ draw->extra_shader_outputs.slot = 0;
}
wide->psize_slot = -1;
@@ -287,7 +287,7 @@ static void widepoint_flush( struct draw_stage *stage, unsigned flags )
{
stage->point = widepoint_first_point;
stage->next->flush( stage->next, flags );
- stage->draw->extra_vp_outputs.slot = 0;
+ stage->draw->extra_shader_outputs.slot = 0;
}
diff --git a/src/gallium/auxiliary/draw/draw_private.h b/src/gallium/auxiliary/draw/draw_private.h
index 41fcb16a0a5..2a11b8c5459 100644
--- a/src/gallium/auxiliary/draw/draw_private.h
+++ b/src/gallium/auxiliary/draw/draw_private.h
@@ -142,8 +142,6 @@ struct draw_context
/* user-space vertex data, buffers */
struct {
- const unsigned *edgeflag;
-
/** vertex element/index buffer (ex: glDrawElements) */
const void *elts;
/** bytes per index (0, 1, 2 or 4) */
@@ -154,8 +152,9 @@ struct draw_context
/** vertex arrays */
const void *vbuffer[PIPE_MAX_ATTRIBS];
- /** constant buffer (for vertex shader) */
- const void *constants;
+ /** constant buffer (for vertex/geometry shader) */
+ const void *vs_constants;
+ const void *gs_constants;
} user;
boolean test_fse; /* enable FSE even though its not correct (eg for softpipe) */
@@ -184,6 +183,7 @@ struct draw_context
struct draw_vertex_shader *vertex_shader;
uint num_vs_outputs; /**< convenience, from vertex_shader */
uint position_output;
+ uint edgeflag_output;
/** TGSI program interpreter runtime state */
struct tgsi_exec_machine *machine;
@@ -212,6 +212,18 @@ struct draw_context
struct translate_cache *emit_cache;
} vs;
+ struct {
+ struct draw_geometry_shader *geometry_shader;
+ uint num_gs_outputs; /**< convenience, from geometry_shader */
+ uint position_output;
+
+ /** TGSI program interpreter runtime state */
+ struct tgsi_exec_machine *machine;
+
+ uint num_samplers;
+ struct tgsi_sampler **samplers;
+ } gs;
+
/* Clip derived state:
*/
float plane[12][4];
@@ -223,10 +235,12 @@ struct draw_context
uint semantic_name;
uint semantic_index;
int slot;
- } extra_vp_outputs;
+ } extra_shader_outputs;
unsigned reduced_prim;
+ unsigned instance_id;
+
void *driver_private;
};
@@ -246,6 +260,19 @@ void draw_vs_set_constants( struct draw_context *,
+/*******************************************************************************
+ * Geometry shading code:
+ */
+boolean draw_gs_init( struct draw_context *draw );
+void draw_gs_set_constants( struct draw_context *,
+ const float (*constants)[4],
+ unsigned size );
+
+/*******************************************************************************
+ * Common shading code:
+ */
+int draw_current_shader_outputs(struct draw_context *draw);
+int draw_current_shader_position_output(struct draw_context *draw);
/*******************************************************************************
* Vertex processing (was passthrough) code:
diff --git a/src/gallium/auxiliary/draw/draw_pt.c b/src/gallium/auxiliary/draw/draw_pt.c
index 4865a2d8542..a5ddec52863 100644
--- a/src/gallium/auxiliary/draw/draw_pt.c
+++ b/src/gallium/auxiliary/draw/draw_pt.c
@@ -280,20 +280,33 @@ void
draw_arrays(struct draw_context *draw, unsigned prim,
unsigned start, unsigned count)
{
- unsigned reduced_prim = u_reduced_prim(prim);
+ draw_arrays_instanced(draw, prim, start, count, 0, 1);
+}
+
+void
+draw_arrays_instanced(struct draw_context *draw,
+ unsigned mode,
+ unsigned start,
+ unsigned count,
+ unsigned startInstance,
+ unsigned instanceCount)
+{
+ unsigned reduced_prim = u_reduced_prim(mode);
+ unsigned instance;
+
if (reduced_prim != draw->reduced_prim) {
- draw_do_flush( draw, DRAW_FLUSH_STATE_CHANGE );
+ draw_do_flush(draw, DRAW_FLUSH_STATE_CHANGE);
draw->reduced_prim = reduced_prim;
}
if (0)
- draw_print_arrays(draw, prim, start, MIN2(count, 20));
+ draw_print_arrays(draw, mode, start, MIN2(count, 20));
#if 0
{
int i;
- debug_printf("draw_arrays(prim=%u start=%u count=%u):\n",
- prim, start, count);
+ debug_printf("draw_arrays(mode=%u start=%u count=%u):\n",
+ mode, start, count);
tgsi_dump(draw->vs.vertex_shader->state.tokens, 0);
debug_printf("Elements:\n");
for (i = 0; i < draw->pt.nr_vertex_elements; i++) {
@@ -311,15 +324,8 @@ draw_arrays(struct draw_context *draw, unsigned prim,
}
#endif
- /* drawing done here: */
- draw_pt_arrays(draw, prim, start, count);
-}
-
-boolean draw_pt_get_edgeflag( struct draw_context *draw,
- unsigned idx )
-{
- if (draw->pt.user.edgeflag)
- return (draw->pt.user.edgeflag[idx/32] & (1 << (idx%32))) != 0;
- else
- return 1;
+ for (instance = 0; instance < instanceCount; instance++) {
+ draw->instance_id = instance + startInstance;
+ draw_pt_arrays(draw, mode, start, count);
+ }
}
diff --git a/src/gallium/auxiliary/draw/draw_pt.h b/src/gallium/auxiliary/draw/draw_pt.h
index 7a17a9fb6b2..d5e0d92a605 100644
--- a/src/gallium/auxiliary/draw/draw_pt.h
+++ b/src/gallium/auxiliary/draw/draw_pt.h
@@ -149,11 +149,6 @@ struct draw_pt_middle_end *draw_pt_middle_fse( struct draw_context *draw );
struct draw_pt_middle_end *draw_pt_fetch_pipeline_or_emit(struct draw_context *draw);
-/* More helpers:
- */
-boolean draw_pt_get_edgeflag( struct draw_context *draw,
- unsigned idx );
-
/*******************************************************************************
* HW vertex emit:
@@ -188,7 +183,8 @@ struct pt_emit *draw_pt_emit_create( struct draw_context *draw );
struct pt_fetch;
void draw_pt_fetch_prepare( struct pt_fetch *fetch,
unsigned vertex_input_count,
- unsigned vertex_size );
+ unsigned vertex_size,
+ unsigned instance_id_index );
void draw_pt_fetch_run( struct pt_fetch *fetch,
const unsigned *elts,
@@ -217,7 +213,8 @@ boolean draw_pt_post_vs_run( struct pt_post_vs *pvs,
void draw_pt_post_vs_prepare( struct pt_post_vs *pvs,
boolean bypass_clipping,
boolean bypass_viewport,
- boolean opengl );
+ boolean opengl,
+ boolean need_edgeflags );
struct pt_post_vs *draw_pt_post_vs_create( struct draw_context *draw );
diff --git a/src/gallium/auxiliary/draw/draw_pt_emit.c b/src/gallium/auxiliary/draw/draw_pt_emit.c
index 064e16c295c..4fb53276bbe 100644
--- a/src/gallium/auxiliary/draw/draw_pt_emit.c
+++ b/src/gallium/auxiliary/draw/draw_pt_emit.c
@@ -121,10 +121,12 @@ void draw_pt_emit_prepare( struct pt_emit *emit,
emit_sz = 0;
break;
}
-
+
+ hw_key.element[i].type = TRANSLATE_ELEMENT_NORMAL;
hw_key.element[i].input_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
hw_key.element[i].input_buffer = src_buffer;
hw_key.element[i].input_offset = src_offset;
+ hw_key.element[i].instance_divisor = 0;
hw_key.element[i].output_format = output_format;
hw_key.element[i].output_offset = dst_offset;
@@ -204,6 +206,7 @@ void draw_pt_emit( struct pt_emit *emit,
translate->run( translate,
0,
vertex_count,
+ draw->instance_id,
hw_verts );
render->unmap_vertices( render,
@@ -263,6 +266,7 @@ void draw_pt_emit_linear(struct pt_emit *emit,
translate->run(translate,
0,
count,
+ draw->instance_id,
hw_verts);
if (0) {
diff --git a/src/gallium/auxiliary/draw/draw_pt_fetch.c b/src/gallium/auxiliary/draw/draw_pt_fetch.c
index 65c3a34c347..55e7a7b81ad 100644
--- a/src/gallium/auxiliary/draw/draw_pt_fetch.c
+++ b/src/gallium/auxiliary/draw/draw_pt_fetch.c
@@ -42,11 +42,11 @@ struct pt_fetch {
struct translate *translate;
unsigned vertex_size;
- boolean need_edgeflags;
struct translate_cache *cache;
};
+
/* Perform the fetch from API vertex elements & vertex buffers, to a
* contiguous set of float[4] attributes as required for the
* vertex_shader->run_linear() method.
@@ -58,12 +58,14 @@ struct pt_fetch {
*/
void draw_pt_fetch_prepare( struct pt_fetch *fetch,
unsigned vs_input_count,
- unsigned vertex_size )
+ unsigned vertex_size,
+ unsigned instance_id_index )
{
struct draw_context *draw = fetch->draw;
unsigned nr_inputs;
- unsigned i, nr = 0;
+ unsigned i, nr = 0, ei = 0;
unsigned dst_offset = 0;
+ unsigned num_extra_inputs = 0;
struct translate_key key;
fetch->vertex_size = vertex_size;
@@ -78,9 +80,11 @@ void draw_pt_fetch_prepare( struct pt_fetch *fetch,
{
/* Need to set header->vertex_id = 0xffff somehow.
*/
+ key.element[nr].type = TRANSLATE_ELEMENT_NORMAL;
key.element[nr].input_format = PIPE_FORMAT_R32_FLOAT;
key.element[nr].input_buffer = draw->pt.nr_vertex_buffers;
key.element[nr].input_offset = 0;
+ key.element[nr].instance_divisor = 0;
key.element[nr].output_format = PIPE_FORMAT_R32_FLOAT;
key.element[nr].output_offset = dst_offset;
dst_offset += 1 * sizeof(float);
@@ -91,19 +95,36 @@ void draw_pt_fetch_prepare( struct pt_fetch *fetch,
*/
dst_offset += 4 * sizeof(float);
}
-
- assert( draw->pt.nr_vertex_elements >= vs_input_count );
- nr_inputs = MIN2( vs_input_count, draw->pt.nr_vertex_elements );
+ if (instance_id_index != ~0) {
+ num_extra_inputs++;
+ }
+
+ assert(draw->pt.nr_vertex_elements + num_extra_inputs >= vs_input_count);
+
+ nr_inputs = MIN2(vs_input_count, draw->pt.nr_vertex_elements + num_extra_inputs);
for (i = 0; i < nr_inputs; i++) {
- key.element[nr].input_format = draw->pt.vertex_element[i].src_format;
- key.element[nr].input_buffer = draw->pt.vertex_element[i].vertex_buffer_index;
- key.element[nr].input_offset = draw->pt.vertex_element[i].src_offset;
- key.element[nr].output_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
- key.element[nr].output_offset = dst_offset;
+ if (i == instance_id_index) {
+ key.element[nr].type = TRANSLATE_ELEMENT_INSTANCE_ID;
+ key.element[nr].input_format = PIPE_FORMAT_R32_USCALED;
+ key.element[nr].output_format = PIPE_FORMAT_R32_USCALED;
+ key.element[nr].output_offset = dst_offset;
+
+ dst_offset += sizeof(uint);
+ } else {
+ key.element[nr].type = TRANSLATE_ELEMENT_NORMAL;
+ key.element[nr].input_format = draw->pt.vertex_element[ei].src_format;
+ key.element[nr].input_buffer = draw->pt.vertex_element[ei].vertex_buffer_index;
+ key.element[nr].input_offset = draw->pt.vertex_element[ei].src_offset;
+ key.element[nr].instance_divisor = draw->pt.vertex_element[ei].instance_divisor;
+ key.element[nr].output_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
+ key.element[nr].output_offset = dst_offset;
+
+ ei++;
+ dst_offset += 4 * sizeof(float);
+ }
- dst_offset += 4 * sizeof(float);
nr++;
}
@@ -120,7 +141,12 @@ void draw_pt_fetch_prepare( struct pt_fetch *fetch,
fetch->translate = translate_cache_find(fetch->cache, &key);
{
- static struct vertex_header vh = { 0, 1, 0, UNDEFINED_VERTEX_ID, { .0f, .0f, .0f, .0f } };
+ static struct vertex_header vh = { 0,
+ 1,
+ 0,
+ UNDEFINED_VERTEX_ID,
+ { .0f, .0f, .0f, .0f } };
+
fetch->translate->set_buffer(fetch->translate,
draw->pt.nr_vertex_buffers,
&vh,
@@ -128,9 +154,6 @@ void draw_pt_fetch_prepare( struct pt_fetch *fetch,
}
}
- fetch->need_edgeflags = ((draw->rasterizer->fill_cw != PIPE_POLYGON_MODE_FILL ||
- draw->rasterizer->fill_ccw != PIPE_POLYGON_MODE_FILL) &&
- draw->pt.user.edgeflag);
}
@@ -156,19 +179,9 @@ void draw_pt_fetch_run( struct pt_fetch *fetch,
translate->run_elts( translate,
elts,
count,
+ draw->instance_id,
verts );
- /* Edgeflags are hard to fit into a translate program, populate
- * them separately if required. In the setup above they are
- * defaulted to one, so only need this if there is reason to change
- * that default:
- */
- if (fetch->need_edgeflags) {
- for (i = 0; i < count; i++) {
- struct vertex_header *vh = (struct vertex_header *)(verts + i * fetch->vertex_size);
- vh->edgeflag = draw_pt_get_edgeflag( draw, elts[i] );
- }
- }
}
@@ -192,19 +205,8 @@ void draw_pt_fetch_run_linear( struct pt_fetch *fetch,
translate->run( translate,
start,
count,
+ draw->instance_id,
verts );
-
- /* Edgeflags are hard to fit into a translate program, populate
- * them separately if required. In the setup above they are
- * defaulted to one, so only need this if there is reason to change
- * that default:
- */
- if (fetch->need_edgeflags) {
- for (i = 0; i < count; i++) {
- struct vertex_header *vh = (struct vertex_header *)(verts + i * fetch->vertex_size);
- vh->edgeflag = draw_pt_get_edgeflag( draw, start + i );
- }
- }
}
diff --git a/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c b/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c
index e7fe6b3b768..2a604470e9a 100644
--- a/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c
+++ b/src/gallium/auxiliary/draw/draw_pt_fetch_emit.c
@@ -166,9 +166,11 @@ static void fetch_emit_prepare( struct draw_pt_middle_end *middle,
continue;
}
+ key.element[i].type = TRANSLATE_ELEMENT_NORMAL;
key.element[i].input_format = input_format;
key.element[i].input_buffer = input_buffer;
key.element[i].input_offset = input_offset;
+ key.element[i].instance_divisor = src->instance_divisor;
key.element[i].output_format = output_format;
key.element[i].output_offset = dst_offset;
@@ -256,6 +258,7 @@ static void fetch_emit_run( struct draw_pt_middle_end *middle,
feme->translate->run_elts( feme->translate,
fetch_elts,
fetch_count,
+ draw->instance_id,
hw_verts );
if (0) {
@@ -314,6 +317,7 @@ static void fetch_emit_run_linear( struct draw_pt_middle_end *middle,
feme->translate->run( feme->translate,
start,
count,
+ draw->instance_id,
hw_verts );
if (0) {
@@ -374,6 +378,7 @@ static boolean fetch_emit_run_linear_elts( struct draw_pt_middle_end *middle,
feme->translate->run( feme->translate,
start,
count,
+ draw->instance_id,
hw_verts );
draw->render->unmap_vertices( draw->render, 0, (ushort)(count - 1) );
diff --git a/src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline.c b/src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline.c
index df6c265b7ec..279f4eec635 100644
--- a/src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline.c
+++ b/src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline.c
@@ -32,6 +32,7 @@
#include "draw/draw_vertex.h"
#include "draw/draw_pt.h"
#include "draw/draw_vs.h"
+#include "draw/draw_gs.h"
#include "translate/translate.h"
@@ -58,6 +59,8 @@ static void fetch_pipeline_prepare( struct draw_pt_middle_end *middle,
struct fetch_pipeline_middle_end *fpme = (struct fetch_pipeline_middle_end *)middle;
struct draw_context *draw = fpme->draw;
struct draw_vertex_shader *vs = draw->vs.vertex_shader;
+ unsigned i;
+ boolean instance_id_index = ~0;
/* Add one to num_outputs because the pipeline occasionally tags on
* an additional texcoord, eg for AA lines.
@@ -65,6 +68,15 @@ static void fetch_pipeline_prepare( struct draw_pt_middle_end *middle,
unsigned nr = MAX2( vs->info.num_inputs,
vs->info.num_outputs + 1 );
+ /* Scan for instanceID system value.
+ */
+ for (i = 0; i < vs->info.num_inputs; i++) {
+ if (vs->info.input_semantic_name[i] == TGSI_SEMANTIC_INSTANCEID) {
+ instance_id_index = i;
+ break;
+ }
+ }
+
fpme->prim = prim;
fpme->opt = opt;
@@ -78,16 +90,17 @@ static void fetch_pipeline_prepare( struct draw_pt_middle_end *middle,
draw_pt_fetch_prepare( fpme->fetch,
vs->info.num_inputs,
- fpme->vertex_size );
+ fpme->vertex_size,
+ instance_id_index );
/* XXX: it's not really gl rasterization rules we care about here,
* but gl vs dx9 clip spaces.
*/
draw_pt_post_vs_prepare( fpme->post_vs,
(boolean)draw->bypass_clipping,
(boolean)(draw->identity_viewport ||
- draw->rasterizer->bypass_vs_clip_and_viewport),
- (boolean)draw->rasterizer->gl_rasterization_rules );
-
+ draw->rasterizer->bypass_vs_clip_and_viewport),
+ (boolean)draw->rasterizer->gl_rasterization_rules,
+ (draw->vs.edgeflag_output ? true : false) );
if (!(opt & PT_PIPELINE)) {
draw_pt_emit_prepare( fpme->emit,
@@ -119,7 +132,8 @@ static void fetch_pipeline_run( struct draw_pt_middle_end *middle,
{
struct fetch_pipeline_middle_end *fpme = (struct fetch_pipeline_middle_end *)middle;
struct draw_context *draw = fpme->draw;
- struct draw_vertex_shader *shader = draw->vs.vertex_shader;
+ struct draw_vertex_shader *vshader = draw->vs.vertex_shader;
+ struct draw_geometry_shader *gshader = draw->gs.geometry_shader;
unsigned opt = fpme->opt;
unsigned alloc_count = align( fetch_count, 4 );
@@ -147,13 +161,21 @@ static void fetch_pipeline_run( struct draw_pt_middle_end *middle,
*/
if (opt & PT_SHADE)
{
- shader->run_linear(shader,
- (const float (*)[4])pipeline_verts->data,
- ( float (*)[4])pipeline_verts->data,
- (const float (*)[4])draw->pt.user.constants,
- fetch_count,
- fpme->vertex_size,
- fpme->vertex_size);
+ vshader->run_linear(vshader,
+ (const float (*)[4])pipeline_verts->data,
+ ( float (*)[4])pipeline_verts->data,
+ (const float (*)[4])draw->pt.user.vs_constants,
+ fetch_count,
+ fpme->vertex_size,
+ fpme->vertex_size);
+ if (gshader)
+ draw_geometry_shader_run(gshader,
+ (const float (*)[4])pipeline_verts->data,
+ ( float (*)[4])pipeline_verts->data,
+ (const float (*)[4])draw->pt.user.gs_constants,
+ fetch_count,
+ fpme->vertex_size,
+ fpme->vertex_size);
}
if (draw_pt_post_vs_run( fpme->post_vs,
@@ -196,6 +218,7 @@ static void fetch_pipeline_linear_run( struct draw_pt_middle_end *middle,
struct fetch_pipeline_middle_end *fpme = (struct fetch_pipeline_middle_end *)middle;
struct draw_context *draw = fpme->draw;
struct draw_vertex_shader *shader = draw->vs.vertex_shader;
+ struct draw_geometry_shader *geometry_shader = draw->gs.geometry_shader;
unsigned opt = fpme->opt;
unsigned alloc_count = align( count, 4 );
@@ -226,10 +249,19 @@ static void fetch_pipeline_linear_run( struct draw_pt_middle_end *middle,
shader->run_linear(shader,
(const float (*)[4])pipeline_verts->data,
( float (*)[4])pipeline_verts->data,
- (const float (*)[4])draw->pt.user.constants,
+ (const float (*)[4])draw->pt.user.vs_constants,
count,
fpme->vertex_size,
fpme->vertex_size);
+
+ if (geometry_shader)
+ draw_geometry_shader_run(geometry_shader,
+ (const float (*)[4])pipeline_verts->data,
+ ( float (*)[4])pipeline_verts->data,
+ (const float (*)[4])draw->pt.user.gs_constants,
+ count,
+ fpme->vertex_size,
+ fpme->vertex_size);
}
if (draw_pt_post_vs_run( fpme->post_vs,
@@ -270,6 +302,7 @@ static boolean fetch_pipeline_linear_run_elts( struct draw_pt_middle_end *middle
struct fetch_pipeline_middle_end *fpme = (struct fetch_pipeline_middle_end *)middle;
struct draw_context *draw = fpme->draw;
struct draw_vertex_shader *shader = draw->vs.vertex_shader;
+ struct draw_geometry_shader *geometry_shader = draw->gs.geometry_shader;
unsigned opt = fpme->opt;
unsigned alloc_count = align( count, 4 );
@@ -296,10 +329,19 @@ static boolean fetch_pipeline_linear_run_elts( struct draw_pt_middle_end *middle
shader->run_linear(shader,
(const float (*)[4])pipeline_verts->data,
( float (*)[4])pipeline_verts->data,
- (const float (*)[4])draw->pt.user.constants,
+ (const float (*)[4])draw->pt.user.vs_constants,
count,
fpme->vertex_size,
fpme->vertex_size);
+
+ if (geometry_shader)
+ draw_geometry_shader_run(geometry_shader,
+ (const float (*)[4])pipeline_verts->data,
+ ( float (*)[4])pipeline_verts->data,
+ (const float (*)[4])draw->pt.user.gs_constants,
+ count,
+ fpme->vertex_size,
+ fpme->vertex_size);
}
if (draw_pt_post_vs_run( fpme->post_vs,
diff --git a/src/gallium/auxiliary/draw/draw_pt_post_vs.c b/src/gallium/auxiliary/draw/draw_pt_post_vs.c
index 6c1cb48e8b8..55151823a14 100644
--- a/src/gallium/auxiliary/draw/draw_pt_post_vs.c
+++ b/src/gallium/auxiliary/draw/draw_pt_post_vs.c
@@ -100,7 +100,7 @@ static boolean post_vs_cliptest_viewport_gl( struct pt_post_vs *pvs,
struct vertex_header *out = vertices;
const float *scale = pvs->draw->viewport.scale;
const float *trans = pvs->draw->viewport.translate;
- const unsigned pos = pvs->draw->vs.position_output;
+ const unsigned pos = draw_current_shader_position_output(pvs->draw);
unsigned clipped = 0;
unsigned j;
@@ -147,6 +147,39 @@ static boolean post_vs_cliptest_viewport_gl( struct pt_post_vs *pvs,
+/* As above plus edgeflags
+ */
+static boolean
+post_vs_cliptest_viewport_gl_edgeflag(struct pt_post_vs *pvs,
+ struct vertex_header *vertices,
+ unsigned count,
+ unsigned stride )
+{
+ unsigned j;
+ boolean needpipe;
+
+ needpipe = post_vs_cliptest_viewport_gl( pvs, vertices, count, stride);
+
+ /* If present, copy edgeflag VS output into vertex header.
+ * Otherwise, leave header as is.
+ */
+ if (pvs->draw->vs.edgeflag_output) {
+ struct vertex_header *out = vertices;
+ int ef = pvs->draw->vs.edgeflag_output;
+
+ for (j = 0; j < count; j++) {
+ const float *edgeflag = out->data[ef];
+ out->edgeflag = !(edgeflag[0] != 1.0f);
+ needpipe |= !out->edgeflag;
+ out = (struct vertex_header *)( (char *)out + stride );
+ }
+ }
+ return needpipe;
+}
+
+
+
+
/* If bypass_clipping is set, skip cliptest and rhw divide.
*/
static boolean post_vs_viewport( struct pt_post_vs *pvs,
@@ -157,7 +190,7 @@ static boolean post_vs_viewport( struct pt_post_vs *pvs,
struct vertex_header *out = vertices;
const float *scale = pvs->draw->viewport.scale;
const float *trans = pvs->draw->viewport.translate;
- const unsigned pos = pvs->draw->vs.position_output;
+ const unsigned pos = draw_current_shader_position_output(pvs->draw);
unsigned j;
if (0) debug_printf("%s\n", __FUNCTION__);
@@ -201,17 +234,29 @@ boolean draw_pt_post_vs_run( struct pt_post_vs *pvs,
void draw_pt_post_vs_prepare( struct pt_post_vs *pvs,
boolean bypass_clipping,
boolean bypass_viewport,
- boolean opengl )
+ boolean opengl,
+ boolean need_edgeflags )
{
- if (bypass_clipping) {
- if (bypass_viewport)
- pvs->run = post_vs_none;
- else
- pvs->run = post_vs_viewport;
+ if (!need_edgeflags) {
+ if (bypass_clipping) {
+ if (bypass_viewport)
+ pvs->run = post_vs_none;
+ else
+ pvs->run = post_vs_viewport;
+ }
+ else {
+ /* if (opengl) */
+ pvs->run = post_vs_cliptest_viewport_gl;
+ }
}
else {
- /* if (opengl) */
- pvs->run = post_vs_cliptest_viewport_gl;
+ /* If we need to copy edgeflags to the vertex header, it should
+ * mean we're running the primitive pipeline. Hence the bypass
+ * flags should be false.
+ */
+ assert(!bypass_clipping);
+ assert(!bypass_viewport);
+ pvs->run = post_vs_cliptest_viewport_gl_edgeflag;
}
}
diff --git a/src/gallium/auxiliary/draw/draw_pt_util.c b/src/gallium/auxiliary/draw/draw_pt_util.c
index b61fa291436..17c3b8cec26 100644
--- a/src/gallium/auxiliary/draw/draw_pt_util.c
+++ b/src/gallium/auxiliary/draw/draw_pt_util.c
@@ -50,16 +50,32 @@ void draw_pt_split_prim(unsigned prim, unsigned *first, unsigned *incr)
*first = 2;
*incr = 1;
break;
+ case PIPE_PRIM_LINES_ADJACENCY:
+ *first = 4;
+ *incr = 2;
+ break;
+ case PIPE_PRIM_LINE_STRIP_ADJACENCY:
+ *first = 4;
+ *incr = 1;
+ break;
case PIPE_PRIM_TRIANGLES:
*first = 3;
*incr = 3;
break;
+ case PIPE_PRIM_TRIANGLES_ADJACENCY:
+ *first = 6;
+ *incr = 3;
+ break;
case PIPE_PRIM_TRIANGLE_STRIP:
case PIPE_PRIM_TRIANGLE_FAN:
case PIPE_PRIM_POLYGON:
*first = 3;
*incr = 1;
break;
+ case PIPE_PRIM_TRIANGLE_STRIP_ADJACENCY:
+ *first = 6;
+ *incr = 1;
+ break;
case PIPE_PRIM_QUADS:
*first = 4;
*incr = 4;
diff --git a/src/gallium/auxiliary/draw/draw_pt_varray_tmp_linear.h b/src/gallium/auxiliary/draw/draw_pt_varray_tmp_linear.h
index 010c7a18a7c..f0aec5febab 100644
--- a/src/gallium/auxiliary/draw/draw_pt_varray_tmp_linear.h
+++ b/src/gallium/auxiliary/draw/draw_pt_varray_tmp_linear.h
@@ -36,6 +36,10 @@ static void FUNC(struct draw_pt_front_end *frontend,
case PIPE_PRIM_TRIANGLE_STRIP:
case PIPE_PRIM_QUADS:
case PIPE_PRIM_QUAD_STRIP:
+ case PIPE_PRIM_LINES_ADJACENCY:
+ case PIPE_PRIM_LINE_STRIP_ADJACENCY:
+ case PIPE_PRIM_TRIANGLES_ADJACENCY:
+ case PIPE_PRIM_TRIANGLE_STRIP_ADJACENCY:
for (j = 0; j < count;) {
unsigned remaining = count - j;
unsigned nr = trim( MIN2(varray->driver_fetch_max, remaining), first, incr );
diff --git a/src/gallium/auxiliary/draw/draw_pt_vcache.c b/src/gallium/auxiliary/draw/draw_pt_vcache.c
index d3f179ced18..757c4874545 100644
--- a/src/gallium/auxiliary/draw/draw_pt_vcache.c
+++ b/src/gallium/auxiliary/draw/draw_pt_vcache.c
@@ -346,7 +346,8 @@ vcache_check_run( struct draw_pt_front_end *frontend,
vcache->fetch_max,
draw_count);
- if (max_index == 0xffffffff ||
+ if (max_index >= DRAW_PIPE_MAX_VERTICES ||
+ fetch_count >= UNDEFINED_VERTEX_ID ||
fetch_count > draw_count) {
if (0) debug_printf("fail\n");
goto fail;
diff --git a/src/gallium/auxiliary/draw/draw_vs.c b/src/gallium/auxiliary/draw/draw_vs.c
index 790e89ed820..35536895326 100644
--- a/src/gallium/auxiliary/draw/draw_vs.c
+++ b/src/gallium/auxiliary/draw/draw_vs.c
@@ -101,6 +101,9 @@ draw_create_vertex_shader(struct draw_context *draw,
if (vs->info.output_semantic_name[i] == TGSI_SEMANTIC_POSITION &&
vs->info.output_semantic_index[i] == 0)
vs->position_output = i;
+ else if (vs->info.output_semantic_name[i] == TGSI_SEMANTIC_EDGEFLAG &&
+ vs->info.output_semantic_index[i] == 0)
+ vs->edgeflag_output = i;
}
}
@@ -120,6 +123,7 @@ draw_bind_vertex_shader(struct draw_context *draw,
draw->vs.vertex_shader = dvs;
draw->vs.num_vs_outputs = dvs->info.num_outputs;
draw->vs.position_output = dvs->position_output;
+ draw->vs.edgeflag_output = dvs->edgeflag_output;
dvs->prepare( dvs, draw );
}
else {
diff --git a/src/gallium/auxiliary/draw/draw_vs.h b/src/gallium/auxiliary/draw/draw_vs.h
index 89ae158751a..00036cfe68b 100644
--- a/src/gallium/auxiliary/draw/draw_vs.h
+++ b/src/gallium/auxiliary/draw/draw_vs.h
@@ -43,6 +43,7 @@ struct draw_varient_input
enum pipe_format format;
unsigned buffer;
unsigned offset;
+ unsigned instance_divisor;
};
struct draw_varient_output
@@ -107,6 +108,7 @@ struct draw_vertex_shader {
struct tgsi_shader_info info;
unsigned position_output;
+ unsigned edgeflag_output;
/* Extracted from shader:
*/
diff --git a/src/gallium/auxiliary/draw/draw_vs_aos.c b/src/gallium/auxiliary/draw/draw_vs_aos.c
index 88bc790b621..1aaae4ab7a4 100644
--- a/src/gallium/auxiliary/draw/draw_vs_aos.c
+++ b/src/gallium/auxiliary/draw/draw_vs_aos.c
@@ -361,8 +361,8 @@ static struct x86_reg aos_get_shader_reg_ptr( struct aos_compilation *cp,
static struct x86_reg get_dst_ptr( struct aos_compilation *cp,
const struct tgsi_full_dst_register *dst )
{
- unsigned file = dst->DstRegister.File;
- unsigned idx = dst->DstRegister.Index;
+ unsigned file = dst->Register.File;
+ unsigned idx = dst->Register.Index;
unsigned i;
@@ -529,8 +529,8 @@ static struct x86_reg fetch_src( struct aos_compilation *cp,
const struct tgsi_full_src_register *src )
{
struct x86_reg arg0 = aos_get_shader_reg(cp,
- src->SrcRegister.File,
- src->SrcRegister.Index);
+ src->Register.File,
+ src->Register.Index);
unsigned i;
ubyte swz = 0;
unsigned negs = 0;
@@ -620,8 +620,8 @@ static void x87_fld_src( struct aos_compilation *cp,
unsigned channel )
{
struct x86_reg arg0 = aos_get_shader_reg_ptr(cp,
- src->SrcRegister.File,
- src->SrcRegister.Index);
+ src->Register.File,
+ src->Register.Index);
unsigned swizzle = tgsi_util_get_full_src_register_swizzle( src, channel );
unsigned neg = tgsi_util_get_full_src_register_sign_mode( src, channel );
@@ -669,15 +669,15 @@ static void store_dest( struct aos_compilation *cp,
{
struct x86_reg dst;
- switch (reg->DstRegister.WriteMask) {
+ switch (reg->Register.WriteMask) {
case 0:
return;
case TGSI_WRITEMASK_XYZW:
aos_adopt_xmm_reg(cp,
get_xmm_writable(cp, result),
- reg->DstRegister.File,
- reg->DstRegister.Index,
+ reg->Register.File,
+ reg->Register.Index,
TRUE);
return;
default:
@@ -685,10 +685,10 @@ static void store_dest( struct aos_compilation *cp,
}
dst = aos_get_shader_reg_xmm(cp,
- reg->DstRegister.File,
- reg->DstRegister.Index);
+ reg->Register.File,
+ reg->Register.Index);
- switch (reg->DstRegister.WriteMask) {
+ switch (reg->Register.WriteMask) {
case TGSI_WRITEMASK_X:
sse_movss(cp->func, dst, get_xmm(cp, result));
break;
@@ -710,14 +710,14 @@ static void store_dest( struct aos_compilation *cp,
break;
default:
- mask_write(cp, dst, result, reg->DstRegister.WriteMask);
+ mask_write(cp, dst, result, reg->Register.WriteMask);
break;
}
aos_adopt_xmm_reg(cp,
dst,
- reg->DstRegister.File,
- reg->DstRegister.Index,
+ reg->Register.File,
+ reg->Register.Index,
TRUE);
}
@@ -737,7 +737,7 @@ static void store_scalar_dest( struct aos_compilation *cp,
const struct tgsi_full_dst_register *reg,
struct x86_reg result )
{
- unsigned writemask = reg->DstRegister.WriteMask;
+ unsigned writemask = reg->Register.WriteMask;
struct x86_reg dst;
if (writemask != TGSI_WRITEMASK_X &&
@@ -754,12 +754,12 @@ static void store_scalar_dest( struct aos_compilation *cp,
result = get_xmm(cp, result);
dst = aos_get_shader_reg_xmm(cp,
- reg->DstRegister.File,
- reg->DstRegister.Index);
+ reg->Register.File,
+ reg->Register.Index);
- switch (reg->DstRegister.WriteMask) {
+ switch (reg->Register.WriteMask) {
case TGSI_WRITEMASK_X:
sse_movss(cp->func, dst, result);
break;
@@ -782,8 +782,8 @@ static void store_scalar_dest( struct aos_compilation *cp,
aos_adopt_xmm_reg(cp,
dst,
- reg->DstRegister.File,
- reg->DstRegister.Index,
+ reg->Register.File,
+ reg->Register.Index,
TRUE);
}
@@ -819,7 +819,7 @@ static void x87_fstp_dest4( struct aos_compilation *cp,
const struct tgsi_full_dst_register *dst )
{
struct x86_reg ptr = get_dst_ptr(cp, dst);
- unsigned writemask = dst->DstRegister.WriteMask;
+ unsigned writemask = dst->Register.WriteMask;
x87_fst_or_nop(cp->func, writemask, 0, ptr);
x87_fst_or_nop(cp->func, writemask, 1, ptr);
@@ -956,7 +956,7 @@ static void emit_print( struct aos_compilation *cp,
static boolean emit_ABS( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
+ struct x86_reg arg0 = fetch_src(cp, &op->Src[0]);
struct x86_reg neg = aos_get_internal(cp, IMM_NEGS);
struct x86_reg tmp = aos_get_xmm_reg(cp);
@@ -964,27 +964,27 @@ static boolean emit_ABS( struct aos_compilation *cp, const struct tgsi_full_inst
sse_mulps(cp->func, tmp, neg);
sse_maxps(cp->func, tmp, arg0);
- store_dest(cp, &op->FullDstRegisters[0], tmp);
+ store_dest(cp, &op->Dst[0], tmp);
return TRUE;
}
static boolean emit_ADD( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
- struct x86_reg arg1 = fetch_src(cp, &op->FullSrcRegisters[1]);
+ struct x86_reg arg0 = fetch_src(cp, &op->Src[0]);
+ struct x86_reg arg1 = fetch_src(cp, &op->Src[1]);
struct x86_reg dst = get_xmm_writable(cp, arg0);
sse_addps(cp->func, dst, arg1);
- store_dest(cp, &op->FullDstRegisters[0], dst);
+ store_dest(cp, &op->Dst[0], dst);
return TRUE;
}
static boolean emit_COS( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- x87_fld_src(cp, &op->FullSrcRegisters[0], 0);
+ x87_fld_src(cp, &op->Src[0], 0);
x87_fcos(cp->func);
- x87_fstp_dest4(cp, &op->FullDstRegisters[0]);
+ x87_fstp_dest4(cp, &op->Dst[0]);
return TRUE;
}
@@ -993,8 +993,8 @@ static boolean emit_COS( struct aos_compilation *cp, const struct tgsi_full_inst
*/
static boolean emit_DP3( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
- struct x86_reg arg1 = fetch_src(cp, &op->FullSrcRegisters[1]);
+ struct x86_reg arg0 = fetch_src(cp, &op->Src[0]);
+ struct x86_reg arg1 = fetch_src(cp, &op->Src[1]);
struct x86_reg tmp = aos_get_xmm_reg(cp);
struct x86_reg dst = get_xmm_writable(cp, arg0);
@@ -1007,14 +1007,14 @@ static boolean emit_DP3( struct aos_compilation *cp, const struct tgsi_full_inst
sse_addss(cp->func, dst, tmp);
aos_release_xmm_reg(cp, tmp.idx);
- store_scalar_dest(cp, &op->FullDstRegisters[0], dst);
+ store_scalar_dest(cp, &op->Dst[0], dst);
return TRUE;
}
static boolean emit_DP4( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
- struct x86_reg arg1 = fetch_src(cp, &op->FullSrcRegisters[1]);
+ struct x86_reg arg0 = fetch_src(cp, &op->Src[0]);
+ struct x86_reg arg1 = fetch_src(cp, &op->Src[1]);
struct x86_reg tmp = aos_get_xmm_reg(cp);
struct x86_reg dst = get_xmm_writable(cp, arg0);
@@ -1028,14 +1028,14 @@ static boolean emit_DP4( struct aos_compilation *cp, const struct tgsi_full_inst
sse_addss(cp->func, dst, tmp);
aos_release_xmm_reg(cp, tmp.idx);
- store_scalar_dest(cp, &op->FullDstRegisters[0], dst);
+ store_scalar_dest(cp, &op->Dst[0], dst);
return TRUE;
}
static boolean emit_DPH( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
- struct x86_reg arg1 = fetch_src(cp, &op->FullSrcRegisters[1]);
+ struct x86_reg arg0 = fetch_src(cp, &op->Src[0]);
+ struct x86_reg arg1 = fetch_src(cp, &op->Src[1]);
struct x86_reg tmp = aos_get_xmm_reg(cp);
struct x86_reg dst = get_xmm_writable(cp, arg0);
@@ -1051,14 +1051,14 @@ static boolean emit_DPH( struct aos_compilation *cp, const struct tgsi_full_inst
sse_addss(cp->func, dst, tmp);
aos_release_xmm_reg(cp, tmp.idx);
- store_scalar_dest(cp, &op->FullDstRegisters[0], dst);
+ store_scalar_dest(cp, &op->Dst[0], dst);
return TRUE;
}
static boolean emit_DST( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
- struct x86_reg arg1 = fetch_src(cp, &op->FullSrcRegisters[1]);
+ struct x86_reg arg0 = fetch_src(cp, &op->Src[0]);
+ struct x86_reg arg1 = fetch_src(cp, &op->Src[1]);
struct x86_reg dst = aos_get_xmm_reg(cp);
struct x86_reg tmp = aos_get_xmm_reg(cp);
struct x86_reg ones = aos_get_internal(cp, IMM_ONES);
@@ -1073,25 +1073,25 @@ static boolean emit_DST( struct aos_compilation *cp, const struct tgsi_full_inst
sse_mulps(cp->func, dst, tmp);
aos_release_xmm_reg(cp, tmp.idx);
- store_dest(cp, &op->FullDstRegisters[0], dst);
+ store_dest(cp, &op->Dst[0], dst);
return TRUE;
}
static boolean emit_LG2( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
x87_fld1(cp->func); /* 1 */
- x87_fld_src(cp, &op->FullSrcRegisters[0], 0); /* a0 1 */
+ x87_fld_src(cp, &op->Src[0], 0); /* a0 1 */
x87_fyl2x(cp->func); /* log2(a0) */
- x87_fstp_dest4(cp, &op->FullDstRegisters[0]);
+ x87_fstp_dest4(cp, &op->Dst[0]);
return TRUE;
}
#if 0
static boolean emit_EX2( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- x87_fld_src(cp, &op->FullSrcRegisters[0], 0);
+ x87_fld_src(cp, &op->Src[0], 0);
x87_emit_ex2(cp);
- x87_fstp_dest4(cp, &op->FullDstRegisters[0]);
+ x87_fstp_dest4(cp, &op->Dst[0]);
return TRUE;
}
#endif
@@ -1099,8 +1099,8 @@ static boolean emit_EX2( struct aos_compilation *cp, const struct tgsi_full_inst
static boolean emit_FLR( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg dst = get_dst_ptr(cp, &op->FullDstRegisters[0]);
- unsigned writemask = op->FullDstRegisters[0].DstRegister.WriteMask;
+ struct x86_reg dst = get_dst_ptr(cp, &op->Dst[0]);
+ unsigned writemask = op->Dst[0].Register.WriteMask;
int i;
set_fpu_round_neg_inf( cp );
@@ -1109,7 +1109,7 @@ static boolean emit_FLR( struct aos_compilation *cp, const struct tgsi_full_inst
*/
for (i = 3; i >= 0; i--) {
if (writemask & (1<<i)) {
- x87_fld_src(cp, &op->FullSrcRegisters[0], i);
+ x87_fld_src(cp, &op->Src[0], i);
}
}
@@ -1126,8 +1126,8 @@ static boolean emit_FLR( struct aos_compilation *cp, const struct tgsi_full_inst
static boolean emit_RND( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg dst = get_dst_ptr(cp, &op->FullDstRegisters[0]);
- unsigned writemask = op->FullDstRegisters[0].DstRegister.WriteMask;
+ struct x86_reg dst = get_dst_ptr(cp, &op->Dst[0]);
+ unsigned writemask = op->Dst[0].Register.WriteMask;
int i;
set_fpu_round_nearest( cp );
@@ -1136,7 +1136,7 @@ static boolean emit_RND( struct aos_compilation *cp, const struct tgsi_full_inst
*/
for (i = 3; i >= 0; i--) {
if (writemask & (1<<i)) {
- x87_fld_src(cp, &op->FullSrcRegisters[0], i);
+ x87_fld_src(cp, &op->Src[0], i);
}
}
@@ -1153,10 +1153,10 @@ static boolean emit_RND( struct aos_compilation *cp, const struct tgsi_full_inst
static boolean emit_FRC( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg dst = get_dst_ptr(cp, &op->FullDstRegisters[0]);
+ struct x86_reg dst = get_dst_ptr(cp, &op->Dst[0]);
struct x86_reg st0 = x86_make_reg(file_x87, 0);
struct x86_reg st1 = x86_make_reg(file_x87, 1);
- unsigned writemask = op->FullDstRegisters[0].DstRegister.WriteMask;
+ unsigned writemask = op->Dst[0].Register.WriteMask;
int i;
set_fpu_round_neg_inf( cp );
@@ -1166,7 +1166,7 @@ static boolean emit_FRC( struct aos_compilation *cp, const struct tgsi_full_inst
*/
for (i = 3; i >= 0; i--) {
if (writemask & (1<<i)) {
- x87_fld_src(cp, &op->FullSrcRegisters[0], i);
+ x87_fld_src(cp, &op->Src[0], i);
}
}
@@ -1190,7 +1190,7 @@ static boolean emit_FRC( struct aos_compilation *cp, const struct tgsi_full_inst
static boolean emit_LIT( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
struct x86_reg ecx = x86_make_reg( file_REG32, reg_CX );
- unsigned writemask = op->FullDstRegisters[0].DstRegister.WriteMask;
+ unsigned writemask = op->Dst[0].Register.WriteMask;
unsigned lit_count = cp->lit_count++;
struct x86_reg result, arg0;
unsigned i;
@@ -1209,10 +1209,10 @@ static boolean emit_LIT( struct aos_compilation *cp, const struct tgsi_full_inst
if (writemask != TGSI_WRITEMASK_XYZW)
result = x86_make_disp(cp->machine_EDX, Offset(struct aos_machine, tmp[0]));
else
- result = get_dst_ptr(cp, &op->FullDstRegisters[0]);
+ result = get_dst_ptr(cp, &op->Dst[0]);
- arg0 = fetch_src( cp, &op->FullSrcRegisters[0] );
+ arg0 = fetch_src( cp, &op->Src[0] );
if (arg0.file == file_XMM) {
struct x86_reg tmp = x86_make_disp(cp->machine_EDX,
Offset(struct aos_machine, tmp[1]));
@@ -1259,7 +1259,7 @@ static boolean emit_LIT( struct aos_compilation *cp, const struct tgsi_full_inst
if (writemask != TGSI_WRITEMASK_XYZW) {
store_dest( cp,
- &op->FullDstRegisters[0],
+ &op->Dst[0],
get_xmm_writable( cp, result ) );
}
@@ -1269,8 +1269,8 @@ static boolean emit_LIT( struct aos_compilation *cp, const struct tgsi_full_inst
#if 0
static boolean emit_inline_LIT( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg dst = get_dst_ptr(cp, &op->FullDstRegisters[0]);
- unsigned writemask = op->FullDstRegisters[0].DstRegister.WriteMask;
+ struct x86_reg dst = get_dst_ptr(cp, &op->Dst[0]);
+ unsigned writemask = op->Dst[0].Register.WriteMask;
if (writemask & TGSI_WRITEMASK_YZ) {
struct x86_reg st1 = x86_make_reg(file_x87, 1);
@@ -1286,13 +1286,13 @@ static boolean emit_inline_LIT( struct aos_compilation *cp, const struct tgsi_fu
*/
x87_fldz(cp->func); /* 1 0 */
#endif
- x87_fld_src(cp, &op->FullSrcRegisters[0], 1); /* a1 1 0 */
+ x87_fld_src(cp, &op->Src[0], 1); /* a1 1 0 */
x87_fcomi(cp->func, st2); /* a1 1 0 */
x87_fcmovb(cp->func, st1); /* a1' 1 0 */
x87_fstp(cp->func, st1); /* a1' 0 */
x87_fstp(cp->func, st1); /* a1' */
- x87_fld_src(cp, &op->FullSrcRegisters[0], 3); /* a3 a1' */
+ x87_fld_src(cp, &op->Src[0], 3); /* a3 a1' */
x87_fxch(cp->func, st1); /* a1' a3 */
@@ -1305,7 +1305,7 @@ static boolean emit_inline_LIT( struct aos_compilation *cp, const struct tgsi_fu
/* a0' = max2(a0, 0):
*/
x87_fldz(cp->func); /* 0 r2 */
- x87_fld_src(cp, &op->FullSrcRegisters[0], 0); /* a0 0 r2 */
+ x87_fld_src(cp, &op->Src[0], 0); /* a0 0 r2 */
x87_fcomi(cp->func, st1);
x87_fcmovb(cp->func, st1); /* a0' 0 r2 */
@@ -1333,58 +1333,58 @@ static boolean emit_inline_LIT( struct aos_compilation *cp, const struct tgsi_fu
static boolean emit_MAX( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
- struct x86_reg arg1 = fetch_src(cp, &op->FullSrcRegisters[1]);
+ struct x86_reg arg0 = fetch_src(cp, &op->Src[0]);
+ struct x86_reg arg1 = fetch_src(cp, &op->Src[1]);
struct x86_reg dst = get_xmm_writable(cp, arg0);
sse_maxps(cp->func, dst, arg1);
- store_dest(cp, &op->FullDstRegisters[0], dst);
+ store_dest(cp, &op->Dst[0], dst);
return TRUE;
}
static boolean emit_MIN( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
- struct x86_reg arg1 = fetch_src(cp, &op->FullSrcRegisters[1]);
+ struct x86_reg arg0 = fetch_src(cp, &op->Src[0]);
+ struct x86_reg arg1 = fetch_src(cp, &op->Src[1]);
struct x86_reg dst = get_xmm_writable(cp, arg0);
sse_minps(cp->func, dst, arg1);
- store_dest(cp, &op->FullDstRegisters[0], dst);
+ store_dest(cp, &op->Dst[0], dst);
return TRUE;
}
static boolean emit_MOV( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
+ struct x86_reg arg0 = fetch_src(cp, &op->Src[0]);
struct x86_reg dst = get_xmm_writable(cp, arg0);
/* potentially nothing to do */
- store_dest(cp, &op->FullDstRegisters[0], dst);
+ store_dest(cp, &op->Dst[0], dst);
return TRUE;
}
static boolean emit_MUL( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
- struct x86_reg arg1 = fetch_src(cp, &op->FullSrcRegisters[1]);
+ struct x86_reg arg0 = fetch_src(cp, &op->Src[0]);
+ struct x86_reg arg1 = fetch_src(cp, &op->Src[1]);
struct x86_reg dst = get_xmm_writable(cp, arg0);
sse_mulps(cp->func, dst, arg1);
- store_dest(cp, &op->FullDstRegisters[0], dst);
+ store_dest(cp, &op->Dst[0], dst);
return TRUE;
}
static boolean emit_MAD( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
- struct x86_reg arg1 = fetch_src(cp, &op->FullSrcRegisters[1]);
- struct x86_reg arg2 = fetch_src(cp, &op->FullSrcRegisters[2]);
+ struct x86_reg arg0 = fetch_src(cp, &op->Src[0]);
+ struct x86_reg arg1 = fetch_src(cp, &op->Src[1]);
+ struct x86_reg arg2 = fetch_src(cp, &op->Src[2]);
/* If we can't clobber old contents of arg0, get a temporary & copy
* it there, then clobber it...
@@ -1393,7 +1393,7 @@ static boolean emit_MAD( struct aos_compilation *cp, const struct tgsi_full_inst
sse_mulps(cp->func, arg0, arg1);
sse_addps(cp->func, arg0, arg2);
- store_dest(cp, &op->FullDstRegisters[0], arg0);
+ store_dest(cp, &op->Dst[0], arg0);
return TRUE;
}
@@ -1425,13 +1425,13 @@ static float PIPE_CDECL _exp2(float x)
static boolean emit_POW( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
#if 0
- x87_fld_src(cp, &op->FullSrcRegisters[1], 0); /* a1.x */
- x87_fld_src(cp, &op->FullSrcRegisters[0], 0); /* a0.x a1.x */
+ x87_fld_src(cp, &op->Src[1], 0); /* a1.x */
+ x87_fld_src(cp, &op->Src[0], 0); /* a0.x a1.x */
x87_fyl2x(cp->func); /* a1*log2(a0) */
x87_emit_ex2( cp ); /* 2^(a1*log2(a0)) */
- x87_fstp_dest4(cp, &op->FullDstRegisters[0]);
+ x87_fstp_dest4(cp, &op->Dst[0]);
#else
uint i;
@@ -1450,9 +1450,9 @@ static boolean emit_POW( struct aos_compilation *cp, const struct tgsi_full_inst
x86_lea( cp->func, cp->stack_ESP, x86_make_disp(cp->stack_ESP, -8) );
- x87_fld_src( cp, &op->FullSrcRegisters[1], 0 );
+ x87_fld_src( cp, &op->Src[1], 0 );
x87_fstp( cp->func, x86_make_disp( cp->stack_ESP, 4 ) );
- x87_fld_src( cp, &op->FullSrcRegisters[0], 0 );
+ x87_fld_src( cp, &op->Src[0], 0 );
x87_fstp( cp->func, x86_make_disp( cp->stack_ESP, 0 ) );
/* tmp_EAX has been pushed & will be restored below */
@@ -1467,7 +1467,7 @@ static boolean emit_POW( struct aos_compilation *cp, const struct tgsi_full_inst
*/
cp->func->x87_stack++;
- x87_fstp_dest4( cp, &op->FullDstRegisters[0] );
+ x87_fstp_dest4( cp, &op->Dst[0] );
#endif
return TRUE;
}
@@ -1493,7 +1493,7 @@ static boolean emit_EXPBASE2( struct aos_compilation *cp, const struct tgsi_full
x86_lea( cp->func, cp->stack_ESP, x86_make_disp(cp->stack_ESP, -4) );
- x87_fld_src( cp, &op->FullSrcRegisters[0], 0 );
+ x87_fld_src( cp, &op->Src[0], 0 );
x87_fstp( cp->func, x86_make_disp( cp->stack_ESP, 0 ) );
/* tmp_EAX has been pushed & will be restored below */
@@ -1508,7 +1508,7 @@ static boolean emit_EXPBASE2( struct aos_compilation *cp, const struct tgsi_full
*/
cp->func->x87_stack++;
- x87_fstp_dest4( cp, &op->FullDstRegisters[0] );
+ x87_fstp_dest4( cp, &op->Dst[0] );
return TRUE;
}
@@ -1517,7 +1517,7 @@ static boolean emit_EXPBASE2( struct aos_compilation *cp, const struct tgsi_full
static boolean emit_RCP( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
+ struct x86_reg arg0 = fetch_src(cp, &op->Src[0]);
struct x86_reg dst = aos_get_xmm_reg(cp);
if (cp->have_sse2) {
@@ -1531,7 +1531,7 @@ static boolean emit_RCP( struct aos_compilation *cp, const struct tgsi_full_inst
sse_divss(cp->func, dst, arg0);
}
- store_scalar_dest(cp, &op->FullDstRegisters[0], dst);
+ store_scalar_dest(cp, &op->Dst[0], dst);
return TRUE;
}
@@ -1551,14 +1551,14 @@ static boolean emit_RCP( struct aos_compilation *cp, const struct tgsi_full_inst
static boolean emit_RSQ( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
if (0) {
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
+ struct x86_reg arg0 = fetch_src(cp, &op->Src[0]);
struct x86_reg r = aos_get_xmm_reg(cp);
sse_rsqrtss(cp->func, r, arg0);
- store_scalar_dest(cp, &op->FullDstRegisters[0], r);
+ store_scalar_dest(cp, &op->Dst[0], r);
return TRUE;
}
else {
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
+ struct x86_reg arg0 = fetch_src(cp, &op->Src[0]);
struct x86_reg r = aos_get_xmm_reg(cp);
struct x86_reg neg_half = get_reg_ptr( cp, AOS_FILE_INTERNAL, IMM_RSQ );
@@ -1578,7 +1578,7 @@ static boolean emit_RSQ( struct aos_compilation *cp, const struct tgsi_full_inst
sse_addss( cp->func, tmp, one_point_five ); /* 1.5 - .5 * a * r * r */
sse_mulss( cp->func, r, tmp ); /* r * (1.5 - .5 * a * r * r) */
- store_scalar_dest(cp, &op->FullDstRegisters[0], r);
+ store_scalar_dest(cp, &op->Dst[0], r);
aos_release_xmm_reg(cp, tmp.idx);
@@ -1589,23 +1589,23 @@ static boolean emit_RSQ( struct aos_compilation *cp, const struct tgsi_full_inst
static boolean emit_SGE( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
- struct x86_reg arg1 = fetch_src(cp, &op->FullSrcRegisters[1]);
+ struct x86_reg arg0 = fetch_src(cp, &op->Src[0]);
+ struct x86_reg arg1 = fetch_src(cp, &op->Src[1]);
struct x86_reg ones = aos_get_internal(cp, IMM_ONES);
struct x86_reg dst = get_xmm_writable(cp, arg0);
sse_cmpps(cp->func, dst, arg1, cc_NotLessThan);
sse_andps(cp->func, dst, ones);
- store_dest(cp, &op->FullDstRegisters[0], dst);
+ store_dest(cp, &op->Dst[0], dst);
return TRUE;
}
static boolean emit_SIN( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- x87_fld_src(cp, &op->FullSrcRegisters[0], 0);
+ x87_fld_src(cp, &op->Src[0], 0);
x87_fsin(cp->func);
- x87_fstp_dest4(cp, &op->FullDstRegisters[0]);
+ x87_fstp_dest4(cp, &op->Dst[0]);
return TRUE;
}
@@ -1613,46 +1613,46 @@ static boolean emit_SIN( struct aos_compilation *cp, const struct tgsi_full_inst
static boolean emit_SLT( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
- struct x86_reg arg1 = fetch_src(cp, &op->FullSrcRegisters[1]);
+ struct x86_reg arg0 = fetch_src(cp, &op->Src[0]);
+ struct x86_reg arg1 = fetch_src(cp, &op->Src[1]);
struct x86_reg ones = aos_get_internal(cp, IMM_ONES);
struct x86_reg dst = get_xmm_writable(cp, arg0);
sse_cmpps(cp->func, dst, arg1, cc_LessThan);
sse_andps(cp->func, dst, ones);
- store_dest(cp, &op->FullDstRegisters[0], dst);
+ store_dest(cp, &op->Dst[0], dst);
return TRUE;
}
static boolean emit_SUB( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
- struct x86_reg arg1 = fetch_src(cp, &op->FullSrcRegisters[1]);
+ struct x86_reg arg0 = fetch_src(cp, &op->Src[0]);
+ struct x86_reg arg1 = fetch_src(cp, &op->Src[1]);
struct x86_reg dst = get_xmm_writable(cp, arg0);
sse_subps(cp->func, dst, arg1);
- store_dest(cp, &op->FullDstRegisters[0], dst);
+ store_dest(cp, &op->Dst[0], dst);
return TRUE;
}
static boolean emit_TRUNC( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
+ struct x86_reg arg0 = fetch_src(cp, &op->Src[0]);
struct x86_reg tmp0 = aos_get_xmm_reg(cp);
sse2_cvttps2dq(cp->func, tmp0, arg0);
sse2_cvtdq2ps(cp->func, tmp0, tmp0);
- store_dest(cp, &op->FullDstRegisters[0], tmp0);
+ store_dest(cp, &op->Dst[0], tmp0);
return TRUE;
}
static boolean emit_XPD( struct aos_compilation *cp, const struct tgsi_full_instruction *op )
{
- struct x86_reg arg0 = fetch_src(cp, &op->FullSrcRegisters[0]);
- struct x86_reg arg1 = fetch_src(cp, &op->FullSrcRegisters[1]);
+ struct x86_reg arg0 = fetch_src(cp, &op->Src[0]);
+ struct x86_reg arg1 = fetch_src(cp, &op->Src[1]);
struct x86_reg tmp0 = aos_get_xmm_reg(cp);
struct x86_reg tmp1 = aos_get_xmm_reg(cp);
@@ -1670,7 +1670,7 @@ static boolean emit_XPD( struct aos_compilation *cp, const struct tgsi_full_inst
aos_release_xmm_reg(cp, tmp0.idx);
- store_dest(cp, &op->FullDstRegisters[0], tmp1);
+ store_dest(cp, &op->Dst[0], tmp1);
return TRUE;
}
@@ -1897,10 +1897,10 @@ static void find_last_write_outputs( struct aos_compilation *cp )
continue;
for (i = 0; i < TGSI_FULL_MAX_DST_REGISTERS; i++) {
- if (parse.FullToken.FullInstruction.FullDstRegisters[i].DstRegister.File ==
+ if (parse.FullToken.FullInstruction.Dst[i].Register.File ==
TGSI_FILE_OUTPUT)
{
- unsigned idx = parse.FullToken.FullInstruction.FullDstRegisters[i].DstRegister.Index;
+ unsigned idx = parse.FullToken.FullInstruction.Dst[i].Register.Index;
cp->output_last_write[idx] = this_instruction;
}
}
diff --git a/src/gallium/auxiliary/draw/draw_vs_ppc.c b/src/gallium/auxiliary/draw/draw_vs_ppc.c
index ad184bd696d..da9f3e3d35c 100644
--- a/src/gallium/auxiliary/draw/draw_vs_ppc.c
+++ b/src/gallium/auxiliary/draw/draw_vs_ppc.c
@@ -98,9 +98,9 @@ vs_ppc_run_linear( struct draw_vertex_shader *base,
/* loop over verts */
for (i = 0; i < count; i += MAX_VERTICES) {
const uint max_vertices = MIN2(MAX_VERTICES, count - i);
- float inputs_soa[PIPE_MAX_SHADER_INPUTS][4][4] ALIGN16_ATTRIB;
- float outputs_soa[PIPE_MAX_SHADER_OUTPUTS][4][4] ALIGN16_ATTRIB;
- float temps_soa[TGSI_EXEC_NUM_TEMPS][4][4] ALIGN16_ATTRIB;
+ PIPE_ALIGN_VAR(16) float inputs_soa[PIPE_MAX_SHADER_INPUTS][4][4];
+ PIPE_ALIGN_VAR(16) float outputs_soa[PIPE_MAX_SHADER_OUTPUTS][4][4];
+ PIPE_ALIGN_VAR(16) float temps_soa[TGSI_EXEC_NUM_TEMPS][4][4];
uint attr;
/* convert (up to) four input verts to SoA format */
diff --git a/src/gallium/auxiliary/draw/draw_vs_varient.c b/src/gallium/auxiliary/draw/draw_vs_varient.c
index 7ee567d4789..9f40030f39f 100644
--- a/src/gallium/auxiliary/draw/draw_vs_varient.c
+++ b/src/gallium/auxiliary/draw/draw_vs_varient.c
@@ -142,16 +142,18 @@ static void PIPE_CDECL vsvg_run_elts( struct draw_vs_varient *varient,
vsvg->fetch->run_elts( vsvg->fetch,
elts,
count,
+ vsvg->draw->instance_id,
temp_buffer );
vsvg->base.vs->run_linear( vsvg->base.vs,
temp_buffer,
temp_buffer,
- (const float (*)[4])vsvg->base.vs->draw->pt.user.constants,
+ (const float (*)[4])vsvg->base.vs->draw->pt.user.vs_constants,
count,
temp_vertex_stride,
temp_vertex_stride);
+ /* FIXME: geometry shading? */
if (vsvg->base.key.clip) {
/* not really handling clipping, just do the rhw so we can
@@ -180,6 +182,7 @@ static void PIPE_CDECL vsvg_run_elts( struct draw_vs_varient *varient,
vsvg->emit->run( vsvg->emit,
0, count,
+ vsvg->draw->instance_id,
output_buffer );
FREE(temp_buffer);
@@ -202,12 +205,13 @@ static void PIPE_CDECL vsvg_run_linear( struct draw_vs_varient *varient,
vsvg->fetch->run( vsvg->fetch,
start,
count,
+ vsvg->draw->instance_id,
temp_buffer );
vsvg->base.vs->run_linear( vsvg->base.vs,
temp_buffer,
temp_buffer,
- (const float (*)[4])vsvg->base.vs->draw->pt.user.constants,
+ (const float (*)[4])vsvg->base.vs->draw->pt.user.vs_constants,
count,
temp_vertex_stride,
temp_vertex_stride);
@@ -238,6 +242,7 @@ static void PIPE_CDECL vsvg_run_linear( struct draw_vs_varient *varient,
vsvg->emit->run( vsvg->emit,
0, count,
+ vsvg->draw->instance_id,
output_buffer );
FREE(temp_buffer);
@@ -280,9 +285,11 @@ struct draw_vs_varient *draw_vs_varient_generic( struct draw_vertex_shader *vs,
fetch.nr_elements = key->nr_inputs;
fetch.output_stride = vsvg->temp_vertex_stride;
for (i = 0; i < key->nr_inputs; i++) {
+ fetch.element[i].type = TRANSLATE_ELEMENT_NORMAL;
fetch.element[i].input_format = key->element[i].in.format;
fetch.element[i].input_buffer = key->element[i].in.buffer;
fetch.element[i].input_offset = key->element[i].in.offset;
+ fetch.element[i].instance_divisor = 0;
fetch.element[i].output_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
fetch.element[i].output_offset = i * 4 * sizeof(float);
assert(fetch.element[i].output_offset < fetch.output_stride);
@@ -294,17 +301,21 @@ struct draw_vs_varient *draw_vs_varient_generic( struct draw_vertex_shader *vs,
for (i = 0; i < key->nr_outputs; i++) {
if (key->element[i].out.format != EMIT_1F_PSIZE)
{
+ emit.element[i].type = TRANSLATE_ELEMENT_NORMAL;
emit.element[i].input_format = PIPE_FORMAT_R32G32B32A32_FLOAT;
emit.element[i].input_buffer = 0;
emit.element[i].input_offset = key->element[i].out.vs_output * 4 * sizeof(float);
+ emit.element[i].instance_divisor = 0;
emit.element[i].output_format = draw_translate_vinfo_format(key->element[i].out.format);
emit.element[i].output_offset = key->element[i].out.offset;
assert(emit.element[i].input_offset <= fetch.output_stride);
}
else {
+ emit.element[i].type = TRANSLATE_ELEMENT_NORMAL;
emit.element[i].input_format = PIPE_FORMAT_R32_FLOAT;
emit.element[i].input_buffer = 1;
emit.element[i].input_offset = 0;
+ emit.element[i].instance_divisor = 0;
emit.element[i].output_format = PIPE_FORMAT_R32_FLOAT;
emit.element[i].output_offset = key->element[i].out.offset;
}
diff --git a/src/gallium/auxiliary/gallivm/Makefile b/src/gallium/auxiliary/gallivm/Makefile
deleted file mode 100644
index 5a96d94ec37..00000000000
--- a/src/gallium/auxiliary/gallivm/Makefile
+++ /dev/null
@@ -1,92 +0,0 @@
-# -*-makefile-*-
-TOP = ../../../..
-include $(TOP)/configs/current
-
-LIBNAME = gallivm
-
-
-GALLIVM_SOURCES = \
- gallivm.cpp \
- gallivm_cpu.cpp \
- instructions.cpp \
- loweringpass.cpp \
- tgsitollvm.cpp \
- storage.cpp \
- storagesoa.cpp \
- instructionssoa.cpp
-
-INC_SOURCES = gallivm_builtins.cpp gallivmsoabuiltins.cpp
-
-CPP_SOURCES = \
- $(GALLIVM_SOURCES)
-
-C_SOURCES =
-ASM_SOURCES =
-
-OBJECTS = $(C_SOURCES:.c=.o) \
- $(CPP_SOURCES:.cpp=.o) \
- $(ASM_SOURCES:.S=.o)
-
-### Include directories
-INCLUDES = \
- -I. \
- -I$(TOP)/src/gallium/drivers \
- -I$(TOP)/src/gallium/auxiliary \
- -I$(TOP)/src/gallium/include \
- -I$(TOP)/src/mesa \
- -I$(TOP)/include
-
-
-##### RULES #####
-
-.c.o:
- $(CC) -c $(INCLUDES) $(LLVM_CFLAGS) $(CFLAGS) $(DRIVER_DEFINES) $< -o $@
-
-.cpp.o:
- $(CXX) -c $(INCLUDES) $(LLVM_CXXFLAGS) $(CXXFLAGS) $(DRIVER_DEFINES) $< -o $@
-
-.S.o:
- $(CC) -c $(INCLUDES) $(CFLAGS) $(DRIVER_DEFINES) $< -o $@
-
-##### TARGETS #####
-
-default:: depend symlinks $(LIBNAME)
-
-
-$(LIBNAME): $(OBJECTS) Makefile
- $(TOP)/bin/mklib -o $@ -static $(OBJECTS)
-
-
-depend: $(C_SOURCES) $(CPP_SOURCES) $(ASM_SOURCES) $(INC_SOURCES)
- rm -f depend
- touch depend
- $(MKDEP) $(MKDEP_OPTIONS) $(DRIVER_DEFINES) $(INCLUDES) $(C_SOURCES) $(CPP_SOURCES) \
- $(ASM_SOURCES) $(INC_SOURCES) 2> /dev/null
-
-
-gallivm_builtins.cpp: llvm_builtins.c
- clang --emit-llvm < $< |llvm-as|opt -std-compile-opts > temp1.bin
- (echo "static const unsigned char llvm_builtins_data[] = {"; od -txC temp1.bin | sed -e "s/^[0-9]*//" -e s"/ \([0-9a-f][0-9a-f]\)/0x\1,/g" -e"\$$d" | sed -e"\$$s/,$$/,0x00};/") >$@
- rm temp1.bin
-
-gallivmsoabuiltins.cpp: soabuiltins.c
- clang --emit-llvm < $< |llvm-as|opt -std-compile-opts > temp2.bin
- (echo "static const unsigned char soabuiltins_data[] = {"; od -txC temp2.bin | sed -e "s/^[0-9]*//" -e s"/ \([0-9a-f][0-9a-f]\)/0x\1,/g" -e"\$$d" | sed -e"\$$s/,$$/,0x00};/") >$@
- rm temp2.bin
-
-# Emacs tags
-tags:
- etags `find . -name \*.[ch]` `find ../include`
-
-
-# Remove .o and backup files
-clean:
- -rm -f *.o */*.o *~ *.so *~ server/*.o
- -rm -f depend depend.bak
- -rm -f gallivm_builtins.cpp
- -rm -f gallivmsoabuiltins.cpp
-
-symlinks:
-
-
-include depend
diff --git a/src/gallium/auxiliary/gallivm/SConscript b/src/gallium/auxiliary/gallivm/SConscript
deleted file mode 100644
index c0aa51b90a9..00000000000
--- a/src/gallium/auxiliary/gallivm/SConscript
+++ /dev/null
@@ -1,16 +0,0 @@
-Import('*')
-
-gallivm = env.ConvenienceLibrary(
- target = 'gallivm',
- source = [
- 'gallivm.cpp',
- 'gallivm_cpu.cpp',
- 'instructions.cpp',
- 'loweringpass.cpp',
- 'tgsitollvm.cpp',
- 'storage.cpp',
- 'storagesoa.cpp',
- 'instructionssoa.cpp',
- ])
-
-auxiliaries.insert(0, gallivm)
diff --git a/src/gallium/auxiliary/gallivm/tgsitollvm.cpp b/src/gallium/auxiliary/gallivm/tgsitollvm.cpp
index bf84401e112..8f7d3b71004 100644
--- a/src/gallium/auxiliary/gallivm/tgsitollvm.cpp
+++ b/src/gallium/auxiliary/gallivm/tgsitollvm.cpp
@@ -94,8 +94,8 @@ translate_declaration(struct gallivm_ir *prog,
unsigned first, last, mask;
uint interp_method;
- first = decl->DeclarationRange.First;
- last = decl->DeclarationRange.Last;
+ first = decl->Range.First;
+ last = decl->Range.Last;
mask = decl->Declaration.UsageMask;
/* Do not touch WPOS.xy */
@@ -149,7 +149,7 @@ translate_declarationir(struct gallivm_ir *,
struct tgsi_full_declaration *)
{
if (decl->Declaration.File == TGSI_FILE_ADDRESS) {
- int idx = decl->DeclarationRange.First;
+ int idx = decl->Range.First;
storage->addAddress(idx);
}
}
@@ -234,26 +234,26 @@ translate_instruction(llvm::Module *module,
inputs[3] = 0;
for (int i = 0; i < inst->Instruction.NumSrcRegs; ++i) {
- struct tgsi_full_src_register *src = &inst->FullSrcRegisters[i];
+ struct tgsi_full_src_register *src = &inst->Src[i];
llvm::Value *val = 0;
llvm::Value *indIdx = 0;
- if (src->SrcRegister.Indirect) {
- indIdx = storage->addrElement(src->SrcRegisterInd.Index);
+ if (src->Register.Indirect) {
+ indIdx = storage->addrElement(src->Indirect.Index);
indIdx = storage->extractIndex(indIdx);
}
- if (src->SrcRegister.File == TGSI_FILE_CONSTANT) {
- val = storage->constElement(src->SrcRegister.Index, indIdx);
- } else if (src->SrcRegister.File == TGSI_FILE_INPUT) {
- val = storage->inputElement(src->SrcRegister.Index, indIdx);
- } else if (src->SrcRegister.File == TGSI_FILE_TEMPORARY) {
- val = storage->tempElement(src->SrcRegister.Index);
- } else if (src->SrcRegister.File == TGSI_FILE_OUTPUT) {
- val = storage->outputElement(src->SrcRegister.Index, indIdx);
- } else if (src->SrcRegister.File == TGSI_FILE_IMMEDIATE) {
- val = storage->immediateElement(src->SrcRegister.Index);
+ if (src->Register.File == TGSI_FILE_CONSTANT) {
+ val = storage->constElement(src->Register.Index, indIdx);
+ } else if (src->Register.File == TGSI_FILE_INPUT) {
+ val = storage->inputElement(src->Register.Index, indIdx);
+ } else if (src->Register.File == TGSI_FILE_TEMPORARY) {
+ val = storage->tempElement(src->Register.Index);
+ } else if (src->Register.File == TGSI_FILE_OUTPUT) {
+ val = storage->outputElement(src->Register.Index, indIdx);
+ } else if (src->Register.File == TGSI_FILE_IMMEDIATE) {
+ val = storage->immediateElement(src->Register.Index);
} else {
- fprintf(stderr, "ERROR: not supported llvm source %d\n", src->SrcRegister.File);
+ fprintf(stderr, "ERROR: not supported llvm source %d\n", src->Register.File);
return;
}
@@ -552,7 +552,7 @@ translate_instruction(llvm::Module *module,
break;
case TGSI_OPCODE_SHL:
break;
- case TGSI_OPCODE_SHR:
+ case TGSI_OPCODE_ISHR:
break;
case TGSI_OPCODE_AND:
break;
@@ -656,14 +656,14 @@ translate_instruction(llvm::Module *module,
/* store results */
for (int i = 0; i < inst->Instruction.NumDstRegs; ++i) {
- struct tgsi_full_dst_register *dst = &inst->FullDstRegisters[i];
-
- if (dst->DstRegister.File == TGSI_FILE_OUTPUT) {
- storage->setOutputElement(dst->DstRegister.Index, out, dst->DstRegister.WriteMask);
- } else if (dst->DstRegister.File == TGSI_FILE_TEMPORARY) {
- storage->setTempElement(dst->DstRegister.Index, out, dst->DstRegister.WriteMask);
- } else if (dst->DstRegister.File == TGSI_FILE_ADDRESS) {
- storage->setAddrElement(dst->DstRegister.Index, out, dst->DstRegister.WriteMask);
+ struct tgsi_full_dst_register *dst = &inst->Dst[i];
+
+ if (dst->Register.File == TGSI_FILE_OUTPUT) {
+ storage->setOutputElement(dst->Register.Index, out, dst->Register.WriteMask);
+ } else if (dst->Register.File == TGSI_FILE_TEMPORARY) {
+ storage->setTempElement(dst->Register.Index, out, dst->Register.WriteMask);
+ } else if (dst->Register.File == TGSI_FILE_ADDRESS) {
+ storage->setAddrElement(dst->Register.Index, out, dst->Register.WriteMask);
} else {
fprintf(stderr, "ERROR: unsupported LLVM destination!");
assert(!"wrong destination");
@@ -683,16 +683,16 @@ translate_instructionir(llvm::Module *module,
std::vector< std::vector<llvm::Value*> > inputs(inst->Instruction.NumSrcRegs);
for (int i = 0; i < inst->Instruction.NumSrcRegs; ++i) {
- struct tgsi_full_src_register *src = &inst->FullSrcRegisters[i];
+ struct tgsi_full_src_register *src = &inst->Src[i];
std::vector<llvm::Value*> val;
llvm::Value *indIdx = 0;
int swizzle = swizzleInt(src);
- if (src->SrcRegister.Indirect) {
- indIdx = storage->addrElement(src->SrcRegisterInd.Index);
+ if (src->Register.Indirect) {
+ indIdx = storage->addrElement(src->Indirect.Index);
}
- val = storage->load((enum tgsi_file_type)src->SrcRegister.File,
- src->SrcRegister.Index, swizzle, instr->getIRBuilder(), indIdx);
+ val = storage->load((enum tgsi_file_type)src->Register.File,
+ src->Register.Index, swizzle, instr->getIRBuilder(), indIdx);
inputs[i] = val;
}
@@ -919,7 +919,7 @@ translate_instructionir(llvm::Module *module,
break;
case TGSI_OPCODE_SHL:
break;
- case TGSI_OPCODE_SHR:
+ case TGSI_OPCODE_ISHR:
break;
case TGSI_OPCODE_AND:
break;
@@ -993,9 +993,9 @@ translate_instructionir(llvm::Module *module,
/* store results */
for (int i = 0; i < inst->Instruction.NumDstRegs; ++i) {
- struct tgsi_full_dst_register *dst = &inst->FullDstRegisters[i];
- storage->store((enum tgsi_file_type)dst->DstRegister.File,
- dst->DstRegister.Index, out, dst->DstRegister.WriteMask,
+ struct tgsi_full_dst_register *dst = &inst->Dst[i];
+ storage->store((enum tgsi_file_type)dst->Register.File,
+ dst->Register.Index, out, dst->Register.WriteMask,
instr->getIRBuilder() );
}
}
diff --git a/src/gallium/auxiliary/indices/Makefile b/src/gallium/auxiliary/indices/Makefile
deleted file mode 100644
index f2ebc3f410a..00000000000
--- a/src/gallium/auxiliary/indices/Makefile
+++ /dev/null
@@ -1,16 +0,0 @@
-TOP = ../../../..
-include $(TOP)/configs/current
-
-LIBNAME = indices
-
-C_SOURCES = \
- u_indices_gen.c \
- u_unfilled_gen.c
-
-include ../../Makefile.template
-
-u_indices_gen.c: u_indices_gen.py
- python $< > $@
-
-u_unfilled_gen.c: u_unfilled_gen.py
- python $< > $@
diff --git a/src/gallium/auxiliary/indices/SConscript b/src/gallium/auxiliary/indices/SConscript
deleted file mode 100644
index 712e215534f..00000000000
--- a/src/gallium/auxiliary/indices/SConscript
+++ /dev/null
@@ -1,28 +0,0 @@
-Import('*')
-
-from sys import executable as python_cmd
-
-env.CodeGenerate(
- target = 'u_indices_gen.c',
- script = 'u_indices_gen.py',
- source = [],
- command = python_cmd + ' $SCRIPT > $TARGET'
-)
-
-env.CodeGenerate(
- target = 'u_unfilled_gen.c',
- script = 'u_unfilled_gen.py',
- source = [],
- command = python_cmd + ' $SCRIPT > $TARGET'
-)
-
-indices = env.ConvenienceLibrary(
- target = 'indices',
- source = [
-# 'u_indices.c',
-# 'u_unfilled_indices.c',
- 'u_indices_gen.c',
- 'u_unfilled_gen.c',
- ])
-
-auxiliaries.insert(0, indices)
diff --git a/src/gallium/auxiliary/pipebuffer/Makefile b/src/gallium/auxiliary/pipebuffer/Makefile
deleted file mode 100644
index 1c00ba8d986..00000000000
--- a/src/gallium/auxiliary/pipebuffer/Makefile
+++ /dev/null
@@ -1,19 +0,0 @@
-TOP = ../../../..
-include $(TOP)/configs/current
-
-LIBNAME = pipebuffer
-
-C_SOURCES = \
- pb_buffer_fenced.c \
- pb_buffer_malloc.c \
- pb_bufmgr_alt.c \
- pb_bufmgr_cache.c \
- pb_bufmgr_debug.c \
- pb_bufmgr_fenced.c \
- pb_bufmgr_mm.c \
- pb_bufmgr_ondemand.c \
- pb_bufmgr_pool.c \
- pb_bufmgr_slab.c \
- pb_validate.c
-
-include ../../Makefile.template
diff --git a/src/gallium/auxiliary/pipebuffer/SConscript b/src/gallium/auxiliary/pipebuffer/SConscript
deleted file mode 100644
index 8e9f06abe45..00000000000
--- a/src/gallium/auxiliary/pipebuffer/SConscript
+++ /dev/null
@@ -1,19 +0,0 @@
-Import('*')
-
-pipebuffer = env.ConvenienceLibrary(
- target = 'pipebuffer',
- source = [
- 'pb_buffer_fenced.c',
- 'pb_buffer_malloc.c',
- 'pb_bufmgr_alt.c',
- 'pb_bufmgr_cache.c',
- 'pb_bufmgr_debug.c',
- 'pb_bufmgr_fenced.c',
- 'pb_bufmgr_mm.c',
- 'pb_bufmgr_ondemand.c',
- 'pb_bufmgr_pool.c',
- 'pb_bufmgr_slab.c',
- 'pb_validate.c',
- ])
-
-auxiliaries.insert(0, pipebuffer)
diff --git a/src/gallium/auxiliary/pipebuffer/pb_buffer.h b/src/gallium/auxiliary/pipebuffer/pb_buffer.h
index 4ef372233f0..eb7e84be848 100644
--- a/src/gallium/auxiliary/pipebuffer/pb_buffer.h
+++ b/src/gallium/auxiliary/pipebuffer/pb_buffer.h
@@ -237,8 +237,9 @@ pb_reference(struct pb_buffer **dst,
{
struct pb_buffer *old = *dst;
- if (pipe_reference((struct pipe_reference**)dst, &src->base.reference))
+ if (pipe_reference(&(*dst)->base.reference, &src->base.reference))
pb_destroy( old );
+ *dst = src;
}
diff --git a/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c b/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c
index 2ef4293d4d7..a4b78f14943 100644
--- a/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c
+++ b/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c
@@ -1,6 +1,6 @@
/**************************************************************************
*
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright 2007-2009 VMware, Inc.
* All Rights Reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a
@@ -18,7 +18,7 @@
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
* OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
* ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
@@ -221,6 +221,7 @@ _fenced_buffer_list_check_free(struct fenced_buffer_list *fenced_list,
struct pb_fence_ops *ops = fenced_list->ops;
struct list_head *curr, *next;
struct fenced_buffer *fenced_buf;
+ struct pb_buffer *pb_buf;
struct pipe_fence_handle *prev_fence = NULL;
curr = fenced_list->delayed.next;
diff --git a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_cache.c b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_cache.c
index 57d1ede45a4..7b34c8e3578 100644
--- a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_cache.c
+++ b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_cache.c
@@ -294,7 +294,7 @@ pb_cache_manager_create_buffer(struct pb_manager *_mgr,
LIST_DEL(&buf->head);
pipe_mutex_unlock(mgr->mutex);
/* Increase refcount */
- pb_reference((struct pb_buffer**)&buf, &buf->base);
+ pipe_reference(NULL, &buf->base.base.reference);
return &buf->base;
}
diff --git a/src/gallium/auxiliary/rbug/Makefile b/src/gallium/auxiliary/rbug/Makefile
deleted file mode 100644
index cd12e8468fc..00000000000
--- a/src/gallium/auxiliary/rbug/Makefile
+++ /dev/null
@@ -1,14 +0,0 @@
-TOP = ../../../..
-include $(TOP)/configs/current
-
-LIBNAME = rbug
-
-C_SOURCES = \
- rbug_connection.c \
- rbug_core.c \
- rbug_texture.c \
- rbug_context.c \
- rbug_shader.c \
- rbug_demarshal.c
-
-include ../../Makefile.template
diff --git a/src/gallium/auxiliary/rbug/SConscript b/src/gallium/auxiliary/rbug/SConscript
deleted file mode 100644
index 4a9afb45d3c..00000000000
--- a/src/gallium/auxiliary/rbug/SConscript
+++ /dev/null
@@ -1,14 +0,0 @@
-Import('*')
-
-rbug = env.ConvenienceLibrary(
- target = 'rbug',
- source = [
- 'rbug_core.c',
- 'rbug_shader.c',
- 'rbug_context.c',
- 'rbug_texture.c',
- 'rbug_demarshal.c',
- 'rbug_connection.c',
- ])
-
-auxiliaries.insert(0, rbug)
diff --git a/src/gallium/auxiliary/rtasm/Makefile b/src/gallium/auxiliary/rtasm/Makefile
deleted file mode 100644
index ab8ea464c6e..00000000000
--- a/src/gallium/auxiliary/rtasm/Makefile
+++ /dev/null
@@ -1,13 +0,0 @@
-TOP = ../../../..
-include $(TOP)/configs/current
-
-LIBNAME = rtasm
-
-C_SOURCES = \
- rtasm_cpu.c \
- rtasm_execmem.c \
- rtasm_x86sse.c \
- rtasm_ppc.c \
- rtasm_ppc_spe.c
-
-include ../../Makefile.template
diff --git a/src/gallium/auxiliary/rtasm/SConscript b/src/gallium/auxiliary/rtasm/SConscript
deleted file mode 100644
index eb48368accb..00000000000
--- a/src/gallium/auxiliary/rtasm/SConscript
+++ /dev/null
@@ -1,13 +0,0 @@
-Import('*')
-
-rtasm = env.ConvenienceLibrary(
- target = 'rtasm',
- source = [
- 'rtasm_cpu.c',
- 'rtasm_execmem.c',
- 'rtasm_x86sse.c',
- 'rtasm_ppc.c',
- 'rtasm_ppc_spe.c',
- ])
-
-auxiliaries.insert(0, rtasm)
diff --git a/src/gallium/auxiliary/rtasm/rtasm_x86sse.c b/src/gallium/auxiliary/rtasm/rtasm_x86sse.c
index 1acf3c373eb..f675427d987 100644
--- a/src/gallium/auxiliary/rtasm/rtasm_x86sse.c
+++ b/src/gallium/auxiliary/rtasm/rtasm_x86sse.c
@@ -673,6 +673,13 @@ void x86_and( struct x86_function *p,
emit_op_modrm( p, 0x23, 0x21, dst, src );
}
+void x86_div( struct x86_function *p,
+ struct x86_reg src )
+{
+ assert(src.file == file_REG32 && src.mod == mod_REG);
+ emit_op_modrm(p, 0xf7, 0, x86_make_reg(file_REG32, 6), src);
+}
+
/***********************************************************************
diff --git a/src/gallium/auxiliary/rtasm/rtasm_x86sse.h b/src/gallium/auxiliary/rtasm/rtasm_x86sse.h
index 731a6517968..f7612d416a0 100644
--- a/src/gallium/auxiliary/rtasm/rtasm_x86sse.h
+++ b/src/gallium/auxiliary/rtasm/rtasm_x86sse.h
@@ -244,6 +244,7 @@ void x86_sub( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
void x86_test( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
void x86_xor( struct x86_function *p, struct x86_reg dst, struct x86_reg src );
void x86_sahf( struct x86_function *p );
+void x86_div( struct x86_function *p, struct x86_reg src );
void x86_cdecl_caller_push_regs( struct x86_function *p );
diff --git a/src/gallium/auxiliary/sct/Makefile b/src/gallium/auxiliary/sct/Makefile
deleted file mode 100644
index a7d111b6891..00000000000
--- a/src/gallium/auxiliary/sct/Makefile
+++ /dev/null
@@ -1,9 +0,0 @@
-TOP = ../../../..
-include $(TOP)/configs/current
-
-LIBNAME = sct
-
-C_SOURCES = \
- sct.c
-
-include ../../Makefile.template
diff --git a/src/gallium/auxiliary/sct/SConscript b/src/gallium/auxiliary/sct/SConscript
deleted file mode 100644
index 76927d973f8..00000000000
--- a/src/gallium/auxiliary/sct/SConscript
+++ /dev/null
@@ -1,9 +0,0 @@
-Import('*')
-
-sct = env.ConvenienceLibrary(
- target = 'sct',
- source = [
- 'sct.c'
- ])
-
-auxiliaries.insert(0, sct)
diff --git a/src/gallium/auxiliary/sct/sct.c b/src/gallium/auxiliary/sct/sct.c
deleted file mode 100644
index 722d2b7e66e..00000000000
--- a/src/gallium/auxiliary/sct/sct.c
+++ /dev/null
@@ -1,453 +0,0 @@
-/**************************************************************************
- *
- * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- **************************************************************************/
-
-
-#include "util/u_memory.h"
-#include "pipe/p_state.h"
-#include "sct.h"
-
-
-struct texture_list
-{
- struct pipe_texture *texture;
- struct texture_list *next;
-};
-
-
-
-#define MAX_SURFACES ((PIPE_MAX_COLOR_BUFS) + 1)
-
-struct sct_context
-{
- const struct pipe_context *context;
-
- /** surfaces the context is drawing into */
- struct pipe_surface *surfaces[MAX_SURFACES];
-
- /** currently bound textures */
- struct pipe_texture *textures[PIPE_MAX_SAMPLERS];
-
- /** previously bound textures, used but not flushed */
- struct texture_list *textures_used;
-
- boolean needs_flush;
-
- struct sct_context *next;
-};
-
-
-
-struct sct_surface
-{
- const struct pipe_surface *surface;
-
- /** list of contexts drawing to this surface */
- struct sct_context_list *contexts;
-
- struct sct_surface *next;
-};
-
-
-
-/**
- * Find the surface_info for the given pipe_surface
- */
-static struct sct_surface *
-find_surface_info(struct surface_context_tracker *sct,
- const struct pipe_surface *surface)
-{
- struct sct_surface *si;
- for (si = sct->surfaces; si; si = si->next)
- if (si->surface == surface)
- return si;
- return NULL;
-}
-
-
-/**
- * As above, but create new surface_info if surface is new.
- */
-static struct sct_surface *
-find_create_surface_info(struct surface_context_tracker *sct,
- const struct pipe_surface *surface)
-{
- struct sct_surface *si = find_surface_info(sct, surface);
- if (si)
- return si;
-
- /* alloc new */
- si = CALLOC_STRUCT(sct_surface);
- if (si) {
- si->surface = surface;
-
- /* insert at head */
- si->next = sct->surfaces;
- sct->surfaces = si;
- }
-
- return si;
-}
-
-
-/**
- * Find a context_info for the given context.
- */
-static struct sct_context *
-find_context_info(struct surface_context_tracker *sct,
- const struct pipe_context *context)
-{
- struct sct_context *ci;
- for (ci = sct->contexts; ci; ci = ci->next)
- if (ci->context == context)
- return ci;
- return NULL;
-}
-
-
-/**
- * As above, but create new context_info if context is new.
- */
-static struct sct_context *
-find_create_context_info(struct surface_context_tracker *sct,
- const struct pipe_context *context)
-{
- struct sct_context *ci = find_context_info(sct, context);
- if (ci)
- return ci;
-
- /* alloc new */
- ci = CALLOC_STRUCT(sct_context);
- if (ci) {
- ci->context = context;
-
- /* insert at head */
- ci->next = sct->contexts;
- sct->contexts = ci;
- }
-
- return ci;
-}
-
-
-/**
- * Is the context already bound to the surface?
- */
-static boolean
-find_surface_context(const struct sct_surface *si,
- const struct pipe_context *context)
-{
- const struct sct_context_list *cl;
- for (cl = si->contexts; cl; cl = cl->next) {
- if (cl->context == context) {
- return TRUE;
- }
- }
- return FALSE;
-}
-
-
-/**
- * Add a context to the list of contexts associated with a surface.
- */
-static void
-add_context_to_surface(struct sct_surface *si,
- const struct pipe_context *context)
-{
- struct sct_context_list *cl = CALLOC_STRUCT(sct_context_list);
- if (cl) {
- cl->context = context;
- /* insert at head of list of contexts */
- cl->next = si->contexts;
- si->contexts = cl;
- }
-}
-
-
-/**
- * Remove a context from the list of contexts associated with a surface.
- */
-static void
-remove_context_from_surface(struct sct_surface *si,
- const struct pipe_context *context)
-{
- struct sct_context_list *prev = NULL, *curr, *next;
-
- for (curr = si->contexts; curr; curr = next) {
- if (curr->context == context) {
- /* remove */
- if (prev)
- prev->next = curr->next;
- else
- si->contexts = curr->next;
- next = curr->next;
- FREE(curr);
- }
- else {
- prev = curr;
- next = curr->next;
- }
- }
-}
-
-
-/**
- * Unbind context from surface.
- */
-static void
-unbind_context_surface(struct surface_context_tracker *sct,
- struct pipe_context *context,
- struct pipe_surface *surface)
-{
- struct sct_surface *si = find_surface_info(sct, surface);
- if (si) {
- remove_context_from_surface(si, context);
- }
-}
-
-
-/**
- * Bind context to a set of surfaces (color + Z).
- * Like MakeCurrent().
- */
-void
-sct_bind_surfaces(struct surface_context_tracker *sct,
- struct pipe_context *context,
- uint num_surf,
- struct pipe_surface **surfaces)
-{
- struct sct_context *ci = find_create_context_info(sct, context);
- uint i;
-
- if (!ci) {
- return; /* out of memory */
- }
-
- /* unbind currently bound surfaces */
- for (i = 0; i < MAX_SURFACES; i++) {
- if (ci->surfaces[i]) {
- unbind_context_surface(sct, context, ci->surfaces[i]);
- }
- }
-
- /* bind new surfaces */
- for (i = 0; i < num_surf; i++) {
- struct sct_surface *si = find_create_surface_info(sct, surfaces[i]);
- if (!find_surface_context(si, context)) {
- add_context_to_surface(si, context);
- }
- }
-}
-
-
-/**
- * Return list of contexts bound to a surface.
- */
-const struct sct_context_list *
-sct_get_surface_contexts(struct surface_context_tracker *sct,
- const struct pipe_surface *surface)
-{
- const struct sct_surface *si = find_surface_info(sct, surface);
- return si->contexts;
-}
-
-
-
-static boolean
-find_texture(const struct sct_context *ci,
- const struct pipe_texture *texture)
-{
- const struct texture_list *tl;
-
- for (tl = ci->textures_used; tl; tl = tl->next) {
- if (tl->texture == texture) {
- return TRUE;
- }
- }
- return FALSE;
-}
-
-
-/**
- * Add the given texture to the context's list of used textures.
- */
-static void
-add_texture_used(struct sct_context *ci,
- struct pipe_texture *texture)
-{
- if (!find_texture(ci, texture)) {
- /* add to list */
- struct texture_list *tl = CALLOC_STRUCT(texture_list);
- if (tl) {
- pipe_texture_reference(&tl->texture, texture);
- /* insert at head */
- tl->next = ci->textures_used;
- ci->textures_used = tl;
- }
- }
-}
-
-
-/**
- * Bind a texture to a rendering context.
- */
-void
-sct_bind_texture(struct surface_context_tracker *sct,
- struct pipe_context *context,
- uint unit,
- struct pipe_texture *tex)
-{
- struct sct_context *ci = find_context_info(sct, context);
-
- if (ci->textures[unit] != tex) {
- /* put texture on the 'used' list */
- add_texture_used(ci, tex);
- /* bind new */
- pipe_texture_reference(&ci->textures[unit], tex);
- }
-}
-
-
-/**
- * Check if the given texture has been used by the rendering context
- * since the last call to sct_flush_textures().
- */
-boolean
-sct_is_texture_used(struct surface_context_tracker *sct,
- const struct pipe_context *context,
- const struct pipe_texture *texture)
-{
- const struct sct_context *ci = find_context_info(sct, context);
- return find_texture(ci, texture);
-}
-
-
-/**
- * To be called when the image contents of a texture are changed, such
- * as for gl[Copy]TexSubImage().
- * XXX this may not be needed
- */
-void
-sct_update_texture(struct pipe_texture *tex)
-{
-
-}
-
-
-/**
- * When a scene is flushed/rendered we can release the list of
- * used textures.
- */
-void
-sct_flush_textures(struct surface_context_tracker *sct,
- struct pipe_context *context)
-{
- struct sct_context *ci = find_context_info(sct, context);
- struct texture_list *tl, *next;
- uint i;
-
- for (tl = ci->textures_used; tl; tl = next) {
- next = tl->next;
- pipe_texture_reference(&tl->texture, NULL);
- FREE(tl);
- }
- ci->textures_used = NULL;
-
- /* put the currently bound textures on the 'used' list */
- for (i = 0; i < PIPE_MAX_SAMPLERS; i++) {
- add_texture_used(ci, ci->textures[i]);
- }
-}
-
-
-
-void
-sct_destroy_context(struct surface_context_tracker *sct,
- struct pipe_context *context)
-{
- /* XXX should we require an unbinding first? */
- {
- struct sct_surface *si;
- for (si = sct->surfaces; si; si = si->next) {
- remove_context_from_surface(si, context);
- }
- }
-
- /* remove context from context_info list */
- {
- struct sct_context *ci, *next, *prev = NULL;
- for (ci = sct->contexts; ci; ci = next) {
- next = ci->next;
- if (ci->context == context) {
- if (prev)
- prev->next = ci->next;
- else
- sct->contexts = ci->next;
- FREE(ci);
- }
- else {
- prev = ci;
- }
- }
- }
-
-}
-
-
-void
-sct_destroy_surface(struct surface_context_tracker *sct,
- struct pipe_surface *surface)
-{
- if (1) {
- /* debug/sanity: no context should be bound to surface */
- struct sct_context *ci;
- uint i;
- for (ci = sct->contexts; ci; ci = ci->next) {
- for (i = 0; i < MAX_SURFACES; i++) {
- assert(ci->surfaces[i] != surface);
- }
- }
- }
-
- /* remove surface from sct_surface list */
- {
- struct sct_surface *si, *next, *prev = NULL;
- for (si = sct->surfaces; si; si = next) {
- next = si->next;
- if (si->surface == surface) {
- /* unlink */
- if (prev)
- prev->next = si->next;
- else
- sct->surfaces = si->next;
- FREE(si);
- }
- else {
- prev = si;
- }
- }
- }
-}
diff --git a/src/gallium/auxiliary/sct/sct.h b/src/gallium/auxiliary/sct/sct.h
deleted file mode 100644
index cf7c4d3bdfd..00000000000
--- a/src/gallium/auxiliary/sct/sct.h
+++ /dev/null
@@ -1,123 +0,0 @@
-/**************************************************************************
- *
- * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- **************************************************************************/
-
-/**
- * Surface/Context Tracking
- *
- * For some drivers, we need to monitor the binding between contexts and
- * surfaces/textures.
- * This code may evolve quite a bit...
- */
-
-
-#ifndef SCT_H
-#define SCT_H
-
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-
-struct pipe_context;
-struct pipe_surface;
-
-struct sct_context;
-struct sct_surface;
-
-
-/**
- * Per-device info, basically
- */
-struct surface_context_tracker
-{
- struct sct_context *contexts;
- struct sct_surface *surfaces;
-};
-
-
-
-/**
- * Simple linked list of contexts
- */
-struct sct_context_list
-{
- const struct pipe_context *context;
- struct sct_context_list *next;
-};
-
-
-
-extern void
-sct_bind_surfaces(struct surface_context_tracker *sct,
- struct pipe_context *context,
- uint num_surf,
- struct pipe_surface **surfaces);
-
-
-extern void
-sct_bind_texture(struct surface_context_tracker *sct,
- struct pipe_context *context,
- uint unit,
- struct pipe_texture *texture);
-
-
-extern void
-sct_update_texture(struct pipe_texture *tex);
-
-
-extern boolean
-sct_is_texture_used(struct surface_context_tracker *sct,
- const struct pipe_context *context,
- const struct pipe_texture *texture);
-
-extern void
-sct_flush_textures(struct surface_context_tracker *sct,
- struct pipe_context *context);
-
-
-extern const struct sct_context_list *
-sct_get_surface_contexts(struct surface_context_tracker *sct,
- const struct pipe_surface *surf);
-
-
-extern void
-sct_destroy_context(struct surface_context_tracker *sct,
- struct pipe_context *context);
-
-
-extern void
-sct_destroy_surface(struct surface_context_tracker *sct,
- struct pipe_surface *surface);
-
-
-
-#ifdef __cplusplus
-}
-#endif
-
-#endif /* SCT_H */
diff --git a/src/gallium/auxiliary/sct/usage.c b/src/gallium/auxiliary/sct/usage.c
deleted file mode 100644
index 6227f199628..00000000000
--- a/src/gallium/auxiliary/sct/usage.c
+++ /dev/null
@@ -1,61 +0,0 @@
-/* surface / context tracking */
-
-
-/*
-
-context A:
- render to texture T
-
-context B:
- texture from T
-
------------------------
-
-flush surface:
- which contexts are bound to the surface?
-
------------------------
-
-glTexSubImage():
- which contexts need to be flushed?
-
- */
-
-
-/*
-
-in MakeCurrent():
-
- call sct_bind_surfaces(context, list of surfaces) to update the
- dependencies between context and surfaces
-
-
-in SurfaceFlush(), or whatever it is in D3D:
-
- call sct_get_surface_contexts(surface) to get a list of contexts
- which are currently bound to the surface.
-
-
-
-in BindTexture():
-
- call sct_bind_texture(context, texture) to indicate that the texture
- is used in the scene.
-
-
-in glTexSubImage() or RenderToTexture():
-
- call sct_is_texture_used(context, texture) to determine if the texture
- has been used in the scene, but the scene's not flushed. If TRUE is
- returned it means the scene has to be rendered/flushed before the contents
- of the texture can be changed.
-
-
-in psb_scene_flush/terminate():
-
- call sct_flush_textures(context) to tell the SCT that the textures which
- were used in the scene can be released.
-
-
-
-*/
diff --git a/src/gallium/auxiliary/tgsi/Makefile b/src/gallium/auxiliary/tgsi/Makefile
deleted file mode 100644
index 5f0a580b096..00000000000
--- a/src/gallium/auxiliary/tgsi/Makefile
+++ /dev/null
@@ -1,22 +0,0 @@
-TOP = ../../../..
-include $(TOP)/configs/current
-
-LIBNAME = tgsi
-
-C_SOURCES = \
- tgsi_sanity.c \
- tgsi_build.c \
- tgsi_dump.c \
- tgsi_exec.c \
- tgsi_info.c \
- tgsi_iterate.c \
- tgsi_parse.c \
- tgsi_ppc.c \
- tgsi_scan.c \
- tgsi_sse2.c \
- tgsi_text.c \
- tgsi_transform.c \
- tgsi_ureg.c \
- tgsi_util.c
-
-include ../../Makefile.template
diff --git a/src/gallium/auxiliary/tgsi/SConscript b/src/gallium/auxiliary/tgsi/SConscript
deleted file mode 100644
index b6bc2924f06..00000000000
--- a/src/gallium/auxiliary/tgsi/SConscript
+++ /dev/null
@@ -1,23 +0,0 @@
-Import('*')
-
-tgsi = env.ConvenienceLibrary(
- target = 'tgsi',
- source = [
- 'tgsi_build.c',
- 'tgsi_dump.c',
- 'tgsi_dump_c.c',
- 'tgsi_exec.c',
- 'tgsi_info.c',
- 'tgsi_iterate.c',
- 'tgsi_parse.c',
- 'tgsi_sanity.c',
- 'tgsi_scan.c',
- 'tgsi_ppc.c',
- 'tgsi_sse2.c',
- 'tgsi_text.c',
- 'tgsi_transform.c',
- 'tgsi_ureg.c',
- 'tgsi_util.c',
- ])
-
-auxiliaries.insert(0, tgsi)
diff --git a/src/gallium/auxiliary/tgsi/tgsi-instruction-set.txt b/src/gallium/auxiliary/tgsi/tgsi-instruction-set.txt
index eb492076b7d..080fd4c7310 100644
--- a/src/gallium/auxiliary/tgsi/tgsi-instruction-set.txt
+++ b/src/gallium/auxiliary/tgsi/tgsi-instruction-set.txt
@@ -1129,3 +1129,35 @@ TGSI Instruction Specification
target Label of target instruction.
+
+3 Other tokens
+===============
+
+
+3.1 Declaration Semantic
+-------------------------
+
+
+ Follows Declaration token if Semantic bit is set.
+
+ Since its purpose is to link a shader with other stages of the pipeline,
+ it is valid to follow only those Declaration tokens that declare a register
+ either in INPUT or OUTPUT file.
+
+ SemanticName field contains the semantic name of the register being declared.
+ There is no default value.
+
+ SemanticIndex is an optional subscript that can be used to distinguish
+ different register declarations with the same semantic name. The default value
+ is 0.
+
+ The meanings of the individual semantic names are explained in the following
+ sections.
+
+
+3.1.1 FACE
+
+ Valid only in a fragment shader INPUT declaration.
+
+ FACE.x is negative when the primitive is back facing. FACE.x is positive
+ when the primitive is front facing.
diff --git a/src/gallium/auxiliary/tgsi/tgsi_build.c b/src/gallium/auxiliary/tgsi/tgsi_build.c
index 4fa10e2f7e3..de9cbc86305 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_build.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_build.c
@@ -30,21 +30,6 @@
#include "tgsi_build.h"
#include "tgsi_parse.h"
-/*
- * version
- */
-
-struct tgsi_version
-tgsi_build_version( void )
-{
- struct tgsi_version version;
-
- version.MajorVersion = 1;
- version.MinorVersion = 1;
- version.Padding = 0;
-
- return version;
-}
/*
* header
@@ -122,7 +107,6 @@ tgsi_default_declaration( void )
declaration.Centroid = 0;
declaration.Invariant = 0;
declaration.Padding = 0;
- declaration.Extended = 0;
return declaration;
}
@@ -173,7 +157,7 @@ tgsi_default_full_declaration( void )
struct tgsi_full_declaration full_declaration;
full_declaration.Declaration = tgsi_default_declaration();
- full_declaration.DeclarationRange = tgsi_default_declaration_range();
+ full_declaration.Range = tgsi_default_declaration_range();
full_declaration.Semantic = tgsi_default_declaration_semantic();
return full_declaration;
@@ -210,8 +194,8 @@ tgsi_build_full_declaration(
size++;
*dr = tgsi_build_declaration_range(
- full_decl->DeclarationRange.First,
- full_decl->DeclarationRange.Last,
+ full_decl->Range.First,
+ full_decl->Range.Last,
declaration,
header );
@@ -224,8 +208,8 @@ tgsi_build_full_declaration(
size++;
*ds = tgsi_build_declaration_semantic(
- full_decl->Semantic.SemanticName,
- full_decl->Semantic.SemanticIndex,
+ full_decl->Semantic.Name,
+ full_decl->Semantic.Index,
declaration,
header );
}
@@ -270,8 +254,8 @@ tgsi_default_declaration_semantic( void )
{
struct tgsi_declaration_semantic ds;
- ds.SemanticName = TGSI_SEMANTIC_POSITION;
- ds.SemanticIndex = 0;
+ ds.Name = TGSI_SEMANTIC_POSITION;
+ ds.Index = 0;
ds.Padding = 0;
return ds;
@@ -290,8 +274,8 @@ tgsi_build_declaration_semantic(
assert( semantic_index <= 0xFFFF );
ds = tgsi_default_declaration_semantic();
- ds.SemanticName = semantic_name;
- ds.SemanticIndex = semantic_index;
+ ds.Name = semantic_name;
+ ds.Index = semantic_index;
declaration_grow( declaration, header );
@@ -311,7 +295,6 @@ tgsi_default_immediate( void )
immediate.NrTokens = 1;
immediate.DataType = TGSI_IMM_FLOAT32;
immediate.Padding = 0;
- immediate.Extended = 0;
return immediate;
}
@@ -416,24 +399,26 @@ tgsi_default_instruction( void )
struct tgsi_instruction instruction;
instruction.Type = TGSI_TOKEN_TYPE_INSTRUCTION;
- instruction.NrTokens = 1;
+ instruction.NrTokens = 0;
instruction.Opcode = TGSI_OPCODE_MOV;
instruction.Saturate = TGSI_SAT_NONE;
+ instruction.Predicate = 0;
instruction.NumDstRegs = 1;
instruction.NumSrcRegs = 1;
+ instruction.Label = 0;
+ instruction.Texture = 0;
instruction.Padding = 0;
- instruction.Extended = 0;
return instruction;
}
struct tgsi_instruction
-tgsi_build_instruction(
- unsigned opcode,
- unsigned saturate,
- unsigned num_dst_regs,
- unsigned num_src_regs,
- struct tgsi_header *header )
+tgsi_build_instruction(unsigned opcode,
+ unsigned saturate,
+ unsigned predicate,
+ unsigned num_dst_regs,
+ unsigned num_src_regs,
+ struct tgsi_header *header)
{
struct tgsi_instruction instruction;
@@ -445,6 +430,7 @@ tgsi_build_instruction(
instruction = tgsi_default_instruction();
instruction.Opcode = opcode;
instruction.Saturate = saturate;
+ instruction.Predicate = predicate;
instruction.NumDstRegs = num_dst_regs;
instruction.NumSrcRegs = num_src_regs;
@@ -472,18 +458,16 @@ tgsi_default_full_instruction( void )
unsigned i;
full_instruction.Instruction = tgsi_default_instruction();
- full_instruction.InstructionExtLabel = tgsi_default_instruction_ext_label();
- full_instruction.InstructionExtTexture = tgsi_default_instruction_ext_texture();
- full_instruction.InstructionExtPredicate = tgsi_default_instruction_ext_predicate();
+ full_instruction.Predicate = tgsi_default_instruction_predicate();
+ full_instruction.Label = tgsi_default_instruction_label();
+ full_instruction.Texture = tgsi_default_instruction_texture();
for( i = 0; i < TGSI_FULL_MAX_DST_REGISTERS; i++ ) {
- full_instruction.FullDstRegisters[i] = tgsi_default_full_dst_register();
+ full_instruction.Dst[i] = tgsi_default_full_dst_register();
}
for( i = 0; i < TGSI_FULL_MAX_SRC_REGISTERS; i++ ) {
- full_instruction.FullSrcRegisters[i] = tgsi_default_full_src_register();
+ full_instruction.Src[i] = tgsi_default_full_src_register();
}
- full_instruction.Flags = 0x0;
-
return full_instruction;
}
@@ -504,77 +488,70 @@ tgsi_build_full_instruction(
instruction = (struct tgsi_instruction *) &tokens[size];
size++;
- *instruction = tgsi_build_instruction(
- full_inst->Instruction.Opcode,
- full_inst->Instruction.Saturate,
- full_inst->Instruction.NumDstRegs,
- full_inst->Instruction.NumSrcRegs,
- header );
+ *instruction = tgsi_build_instruction(full_inst->Instruction.Opcode,
+ full_inst->Instruction.Saturate,
+ full_inst->Instruction.Predicate,
+ full_inst->Instruction.NumDstRegs,
+ full_inst->Instruction.NumSrcRegs,
+ header);
prev_token = (struct tgsi_token *) instruction;
- if( tgsi_compare_instruction_ext_label(
- full_inst->InstructionExtLabel,
- tgsi_default_instruction_ext_label() ) ) {
- struct tgsi_instruction_ext_label *instruction_ext_label;
+ if (full_inst->Instruction.Predicate) {
+ struct tgsi_instruction_predicate *instruction_predicate;
- if( maxsize <= size )
+ if (maxsize <= size) {
return 0;
- instruction_ext_label =
- (struct tgsi_instruction_ext_label *) &tokens[size];
+ }
+ instruction_predicate = (struct tgsi_instruction_predicate *)&tokens[size];
size++;
- *instruction_ext_label = tgsi_build_instruction_ext_label(
- full_inst->InstructionExtLabel.Label,
- prev_token,
- instruction,
- header );
- prev_token = (struct tgsi_token *) instruction_ext_label;
+ *instruction_predicate =
+ tgsi_build_instruction_predicate(full_inst->Predicate.Index,
+ full_inst->Predicate.Negate,
+ full_inst->Predicate.SwizzleX,
+ full_inst->Predicate.SwizzleY,
+ full_inst->Predicate.SwizzleZ,
+ full_inst->Predicate.SwizzleW,
+ instruction,
+ header);
}
- if( tgsi_compare_instruction_ext_texture(
- full_inst->InstructionExtTexture,
- tgsi_default_instruction_ext_texture() ) ) {
- struct tgsi_instruction_ext_texture *instruction_ext_texture;
+ if (full_inst->Instruction.Label) {
+ struct tgsi_instruction_label *instruction_label;
if( maxsize <= size )
return 0;
- instruction_ext_texture =
- (struct tgsi_instruction_ext_texture *) &tokens[size];
+ instruction_label =
+ (struct tgsi_instruction_label *) &tokens[size];
size++;
- *instruction_ext_texture = tgsi_build_instruction_ext_texture(
- full_inst->InstructionExtTexture.Texture,
+ *instruction_label = tgsi_build_instruction_label(
+ full_inst->Label.Label,
prev_token,
instruction,
- header );
- prev_token = (struct tgsi_token *) instruction_ext_texture;
+ header );
+ prev_token = (struct tgsi_token *) instruction_label;
}
- if (tgsi_compare_instruction_ext_predicate(full_inst->InstructionExtPredicate,
- tgsi_default_instruction_ext_predicate())) {
- struct tgsi_instruction_ext_predicate *instruction_ext_predicate;
+ if (full_inst->Instruction.Texture) {
+ struct tgsi_instruction_texture *instruction_texture;
- if (maxsize <= size) {
+ if( maxsize <= size )
return 0;
- }
- instruction_ext_predicate = (struct tgsi_instruction_ext_predicate *)&tokens[size];
+ instruction_texture =
+ (struct tgsi_instruction_texture *) &tokens[size];
size++;
- *instruction_ext_predicate =
- tgsi_build_instruction_ext_predicate(full_inst->InstructionExtPredicate.SrcIndex,
- full_inst->InstructionExtPredicate.Negate,
- full_inst->InstructionExtPredicate.SwizzleX,
- full_inst->InstructionExtPredicate.SwizzleY,
- full_inst->InstructionExtPredicate.SwizzleZ,
- full_inst->InstructionExtPredicate.SwizzleW,
- prev_token,
- instruction,
- header);
- prev_token = (struct tgsi_token *)instruction_ext_predicate;
+ *instruction_texture = tgsi_build_instruction_texture(
+ full_inst->Texture.Texture,
+ prev_token,
+ instruction,
+ header );
+ prev_token = (struct tgsi_token *) instruction_texture;
}
for( i = 0; i < full_inst->Instruction.NumDstRegs; i++ ) {
- const struct tgsi_full_dst_register *reg = &full_inst->FullDstRegisters[i];
+ const struct tgsi_full_dst_register *reg = &full_inst->Dst[i];
struct tgsi_dst_register *dst_register;
struct tgsi_token *prev_token;
@@ -584,34 +561,15 @@ tgsi_build_full_instruction(
size++;
*dst_register = tgsi_build_dst_register(
- reg->DstRegister.File,
- reg->DstRegister.WriteMask,
- reg->DstRegister.Indirect,
- reg->DstRegister.Index,
+ reg->Register.File,
+ reg->Register.WriteMask,
+ reg->Register.Indirect,
+ reg->Register.Index,
instruction,
header );
prev_token = (struct tgsi_token *) dst_register;
- if( tgsi_compare_dst_register_ext_modulate(
- reg->DstRegisterExtModulate,
- tgsi_default_dst_register_ext_modulate() ) ) {
- struct tgsi_dst_register_ext_modulate *dst_register_ext_modulate;
-
- if( maxsize <= size )
- return 0;
- dst_register_ext_modulate =
- (struct tgsi_dst_register_ext_modulate *) &tokens[size];
- size++;
-
- *dst_register_ext_modulate = tgsi_build_dst_register_ext_modulate(
- reg->DstRegisterExtModulate.Modulate,
- prev_token,
- instruction,
- header );
- prev_token = (struct tgsi_token *) dst_register_ext_modulate;
- }
-
- if( reg->DstRegister.Indirect ) {
+ if( reg->Register.Indirect ) {
struct tgsi_src_register *ind;
if( maxsize <= size )
@@ -620,22 +578,23 @@ tgsi_build_full_instruction(
size++;
*ind = tgsi_build_src_register(
- reg->DstRegisterInd.File,
- reg->DstRegisterInd.SwizzleX,
- reg->DstRegisterInd.SwizzleY,
- reg->DstRegisterInd.SwizzleZ,
- reg->DstRegisterInd.SwizzleW,
- reg->DstRegisterInd.Negate,
- reg->DstRegisterInd.Indirect,
- reg->DstRegisterInd.Dimension,
- reg->DstRegisterInd.Index,
+ reg->Indirect.File,
+ reg->Indirect.SwizzleX,
+ reg->Indirect.SwizzleY,
+ reg->Indirect.SwizzleZ,
+ reg->Indirect.SwizzleW,
+ reg->Indirect.Negate,
+ reg->Indirect.Absolute,
+ reg->Indirect.Indirect,
+ reg->Indirect.Dimension,
+ reg->Indirect.Index,
instruction,
header );
}
}
for( i = 0; i < full_inst->Instruction.NumSrcRegs; i++ ) {
- const struct tgsi_full_src_register *reg = &full_inst->FullSrcRegisters[i];
+ const struct tgsi_full_src_register *reg = &full_inst->Src[i];
struct tgsi_src_register *src_register;
struct tgsi_token *prev_token;
@@ -645,43 +604,21 @@ tgsi_build_full_instruction(
size++;
*src_register = tgsi_build_src_register(
- reg->SrcRegister.File,
- reg->SrcRegister.SwizzleX,
- reg->SrcRegister.SwizzleY,
- reg->SrcRegister.SwizzleZ,
- reg->SrcRegister.SwizzleW,
- reg->SrcRegister.Negate,
- reg->SrcRegister.Indirect,
- reg->SrcRegister.Dimension,
- reg->SrcRegister.Index,
+ reg->Register.File,
+ reg->Register.SwizzleX,
+ reg->Register.SwizzleY,
+ reg->Register.SwizzleZ,
+ reg->Register.SwizzleW,
+ reg->Register.Negate,
+ reg->Register.Absolute,
+ reg->Register.Indirect,
+ reg->Register.Dimension,
+ reg->Register.Index,
instruction,
header );
prev_token = (struct tgsi_token *) src_register;
- if( tgsi_compare_src_register_ext_mod(
- reg->SrcRegisterExtMod,
- tgsi_default_src_register_ext_mod() ) ) {
- struct tgsi_src_register_ext_mod *src_register_ext_mod;
-
- if( maxsize <= size )
- return 0;
- src_register_ext_mod =
- (struct tgsi_src_register_ext_mod *) &tokens[size];
- size++;
-
- *src_register_ext_mod = tgsi_build_src_register_ext_mod(
- reg->SrcRegisterExtMod.Complement,
- reg->SrcRegisterExtMod.Bias,
- reg->SrcRegisterExtMod.Scale2X,
- reg->SrcRegisterExtMod.Absolute,
- reg->SrcRegisterExtMod.Negate,
- prev_token,
- instruction,
- header );
- prev_token = (struct tgsi_token *) src_register_ext_mod;
- }
-
- if( reg->SrcRegister.Indirect ) {
+ if( reg->Register.Indirect ) {
struct tgsi_src_register *ind;
if( maxsize <= size )
@@ -690,23 +627,24 @@ tgsi_build_full_instruction(
size++;
*ind = tgsi_build_src_register(
- reg->SrcRegisterInd.File,
- reg->SrcRegisterInd.SwizzleX,
- reg->SrcRegisterInd.SwizzleY,
- reg->SrcRegisterInd.SwizzleZ,
- reg->SrcRegisterInd.SwizzleW,
- reg->SrcRegisterInd.Negate,
- reg->SrcRegisterInd.Indirect,
- reg->SrcRegisterInd.Dimension,
- reg->SrcRegisterInd.Index,
+ reg->Indirect.File,
+ reg->Indirect.SwizzleX,
+ reg->Indirect.SwizzleY,
+ reg->Indirect.SwizzleZ,
+ reg->Indirect.SwizzleW,
+ reg->Indirect.Negate,
+ reg->Indirect.Absolute,
+ reg->Indirect.Indirect,
+ reg->Indirect.Dimension,
+ reg->Indirect.Index,
instruction,
header );
}
- if( reg->SrcRegister.Dimension ) {
+ if( reg->Register.Dimension ) {
struct tgsi_dimension *dim;
- assert( !reg->SrcRegisterDim.Dimension );
+ assert( !reg->Dimension.Dimension );
if( maxsize <= size )
return 0;
@@ -714,12 +652,12 @@ tgsi_build_full_instruction(
size++;
*dim = tgsi_build_dimension(
- reg->SrcRegisterDim.Indirect,
- reg->SrcRegisterDim.Index,
+ reg->Dimension.Indirect,
+ reg->Dimension.Index,
instruction,
header );
- if( reg->SrcRegisterDim.Indirect ) {
+ if( reg->Dimension.Indirect ) {
struct tgsi_src_register *ind;
if( maxsize <= size )
@@ -728,15 +666,16 @@ tgsi_build_full_instruction(
size++;
*ind = tgsi_build_src_register(
- reg->SrcRegisterDimInd.File,
- reg->SrcRegisterDimInd.SwizzleX,
- reg->SrcRegisterDimInd.SwizzleY,
- reg->SrcRegisterDimInd.SwizzleZ,
- reg->SrcRegisterDimInd.SwizzleW,
- reg->SrcRegisterDimInd.Negate,
- reg->SrcRegisterDimInd.Indirect,
- reg->SrcRegisterDimInd.Dimension,
- reg->SrcRegisterDimInd.Index,
+ reg->DimIndirect.File,
+ reg->DimIndirect.SwizzleX,
+ reg->DimIndirect.SwizzleY,
+ reg->DimIndirect.SwizzleZ,
+ reg->DimIndirect.SwizzleW,
+ reg->DimIndirect.Negate,
+ reg->DimIndirect.Absolute,
+ reg->DimIndirect.Indirect,
+ reg->DimIndirect.Dimension,
+ reg->DimIndirect.Index,
instruction,
header );
}
@@ -746,147 +685,103 @@ tgsi_build_full_instruction(
return size;
}
-/** test for inequality of 32-bit values pointed to by a and b */
-static INLINE boolean
-compare32(const void *a, const void *b)
+struct tgsi_instruction_predicate
+tgsi_default_instruction_predicate(void)
{
- return *((uint32_t *) a) != *((uint32_t *) b);
+ struct tgsi_instruction_predicate instruction_predicate;
+
+ instruction_predicate.SwizzleX = TGSI_SWIZZLE_X;
+ instruction_predicate.SwizzleY = TGSI_SWIZZLE_Y;
+ instruction_predicate.SwizzleZ = TGSI_SWIZZLE_Z;
+ instruction_predicate.SwizzleW = TGSI_SWIZZLE_W;
+ instruction_predicate.Negate = 0;
+ instruction_predicate.Index = 0;
+ instruction_predicate.Padding = 0;
+
+ return instruction_predicate;
}
-struct tgsi_instruction_ext_label
-tgsi_default_instruction_ext_label( void )
+struct tgsi_instruction_predicate
+tgsi_build_instruction_predicate(int index,
+ unsigned negate,
+ unsigned swizzleX,
+ unsigned swizzleY,
+ unsigned swizzleZ,
+ unsigned swizzleW,
+ struct tgsi_instruction *instruction,
+ struct tgsi_header *header)
{
- struct tgsi_instruction_ext_label instruction_ext_label;
+ struct tgsi_instruction_predicate instruction_predicate;
+
+ instruction_predicate = tgsi_default_instruction_predicate();
+ instruction_predicate.SwizzleX = swizzleX;
+ instruction_predicate.SwizzleY = swizzleY;
+ instruction_predicate.SwizzleZ = swizzleZ;
+ instruction_predicate.SwizzleW = swizzleW;
+ instruction_predicate.Negate = negate;
+ instruction_predicate.Index = index;
- instruction_ext_label.Type = TGSI_INSTRUCTION_EXT_TYPE_LABEL;
- instruction_ext_label.Label = 0;
- instruction_ext_label.Padding = 0;
- instruction_ext_label.Extended = 0;
+ instruction_grow(instruction, header);
- return instruction_ext_label;
+ return instruction_predicate;
}
-unsigned
-tgsi_compare_instruction_ext_label(
- struct tgsi_instruction_ext_label a,
- struct tgsi_instruction_ext_label b )
+struct tgsi_instruction_label
+tgsi_default_instruction_label( void )
{
- a.Padding = b.Padding = 0;
- a.Extended = b.Extended = 0;
- return compare32(&a, &b);
+ struct tgsi_instruction_label instruction_label;
+
+ instruction_label.Label = 0;
+ instruction_label.Padding = 0;
+
+ return instruction_label;
}
-struct tgsi_instruction_ext_label
-tgsi_build_instruction_ext_label(
+struct tgsi_instruction_label
+tgsi_build_instruction_label(
unsigned label,
struct tgsi_token *prev_token,
struct tgsi_instruction *instruction,
struct tgsi_header *header )
{
- struct tgsi_instruction_ext_label instruction_ext_label;
+ struct tgsi_instruction_label instruction_label;
- instruction_ext_label = tgsi_default_instruction_ext_label();
- instruction_ext_label.Label = label;
+ instruction_label = tgsi_default_instruction_label();
+ instruction_label.Label = label;
+ instruction->Label = 1;
- prev_token->Extended = 1;
instruction_grow( instruction, header );
- return instruction_ext_label;
+ return instruction_label;
}
-struct tgsi_instruction_ext_texture
-tgsi_default_instruction_ext_texture( void )
+struct tgsi_instruction_texture
+tgsi_default_instruction_texture( void )
{
- struct tgsi_instruction_ext_texture instruction_ext_texture;
+ struct tgsi_instruction_texture instruction_texture;
- instruction_ext_texture.Type = TGSI_INSTRUCTION_EXT_TYPE_TEXTURE;
- instruction_ext_texture.Texture = TGSI_TEXTURE_UNKNOWN;
- instruction_ext_texture.Padding = 0;
- instruction_ext_texture.Extended = 0;
-
- return instruction_ext_texture;
-}
+ instruction_texture.Texture = TGSI_TEXTURE_UNKNOWN;
+ instruction_texture.Padding = 0;
-unsigned
-tgsi_compare_instruction_ext_texture(
- struct tgsi_instruction_ext_texture a,
- struct tgsi_instruction_ext_texture b )
-{
- a.Padding = b.Padding = 0;
- a.Extended = b.Extended = 0;
- return compare32(&a, &b);
+ return instruction_texture;
}
-struct tgsi_instruction_ext_texture
-tgsi_build_instruction_ext_texture(
+struct tgsi_instruction_texture
+tgsi_build_instruction_texture(
unsigned texture,
struct tgsi_token *prev_token,
struct tgsi_instruction *instruction,
struct tgsi_header *header )
{
- struct tgsi_instruction_ext_texture instruction_ext_texture;
+ struct tgsi_instruction_texture instruction_texture;
- instruction_ext_texture = tgsi_default_instruction_ext_texture();
- instruction_ext_texture.Texture = texture;
+ instruction_texture = tgsi_default_instruction_texture();
+ instruction_texture.Texture = texture;
+ instruction->Texture = 1;
- prev_token->Extended = 1;
instruction_grow( instruction, header );
- return instruction_ext_texture;
-}
-
-struct tgsi_instruction_ext_predicate
-tgsi_default_instruction_ext_predicate(void)
-{
- struct tgsi_instruction_ext_predicate instruction_ext_predicate;
-
- instruction_ext_predicate.Type = TGSI_INSTRUCTION_EXT_TYPE_PREDICATE;
- instruction_ext_predicate.SwizzleX = TGSI_SWIZZLE_X;
- instruction_ext_predicate.SwizzleY = TGSI_SWIZZLE_Y;
- instruction_ext_predicate.SwizzleZ = TGSI_SWIZZLE_Z;
- instruction_ext_predicate.SwizzleW = TGSI_SWIZZLE_W;
- instruction_ext_predicate.Negate = 0;
- instruction_ext_predicate.SrcIndex = 0;
- instruction_ext_predicate.Padding = 0;
- instruction_ext_predicate.Extended = 0;
-
- return instruction_ext_predicate;
-}
-
-unsigned
-tgsi_compare_instruction_ext_predicate(struct tgsi_instruction_ext_predicate a,
- struct tgsi_instruction_ext_predicate b)
-{
- a.Padding = b.Padding = 0;
- a.Extended = b.Extended = 0;
- return compare32(&a, &b);
-}
-
-struct tgsi_instruction_ext_predicate
-tgsi_build_instruction_ext_predicate(unsigned index,
- unsigned negate,
- unsigned swizzleX,
- unsigned swizzleY,
- unsigned swizzleZ,
- unsigned swizzleW,
- struct tgsi_token *prev_token,
- struct tgsi_instruction *instruction,
- struct tgsi_header *header)
-{
- struct tgsi_instruction_ext_predicate instruction_ext_predicate;
-
- instruction_ext_predicate = tgsi_default_instruction_ext_predicate();
- instruction_ext_predicate.SwizzleX = swizzleX;
- instruction_ext_predicate.SwizzleY = swizzleY;
- instruction_ext_predicate.SwizzleZ = swizzleZ;
- instruction_ext_predicate.SwizzleW = swizzleW;
- instruction_ext_predicate.Negate = negate;
- instruction_ext_predicate.SrcIndex = index;
-
- prev_token->Extended = 1;
- instruction_grow(instruction, header);
-
- return instruction_ext_predicate;
+ return instruction_texture;
}
struct tgsi_src_register
@@ -900,10 +795,10 @@ tgsi_default_src_register( void )
src_register.SwizzleZ = TGSI_SWIZZLE_Z;
src_register.SwizzleW = TGSI_SWIZZLE_W;
src_register.Negate = 0;
+ src_register.Absolute = 0;
src_register.Indirect = 0;
src_register.Dimension = 0;
src_register.Index = 0;
- src_register.Extended = 0;
return src_register;
}
@@ -916,6 +811,7 @@ tgsi_build_src_register(
unsigned swizzle_z,
unsigned swizzle_w,
unsigned negate,
+ unsigned absolute,
unsigned indirect,
unsigned dimension,
int index,
@@ -939,6 +835,7 @@ tgsi_build_src_register(
src_register.SwizzleZ = swizzle_z;
src_register.SwizzleW = swizzle_w;
src_register.Negate = negate;
+ src_register.Absolute = absolute;
src_register.Indirect = indirect;
src_register.Dimension = dimension;
src_register.Index = index;
@@ -953,75 +850,15 @@ tgsi_default_full_src_register( void )
{
struct tgsi_full_src_register full_src_register;
- full_src_register.SrcRegister = tgsi_default_src_register();
- full_src_register.SrcRegisterExtMod = tgsi_default_src_register_ext_mod();
- full_src_register.SrcRegisterInd = tgsi_default_src_register();
- full_src_register.SrcRegisterDim = tgsi_default_dimension();
- full_src_register.SrcRegisterDimInd = tgsi_default_src_register();
+ full_src_register.Register = tgsi_default_src_register();
+ full_src_register.Indirect = tgsi_default_src_register();
+ full_src_register.Dimension = tgsi_default_dimension();
+ full_src_register.DimIndirect = tgsi_default_src_register();
return full_src_register;
}
-struct tgsi_src_register_ext_mod
-tgsi_default_src_register_ext_mod( void )
-{
- struct tgsi_src_register_ext_mod src_register_ext_mod;
-
- src_register_ext_mod.Type = TGSI_SRC_REGISTER_EXT_TYPE_MOD;
- src_register_ext_mod.Complement = 0;
- src_register_ext_mod.Bias = 0;
- src_register_ext_mod.Scale2X = 0;
- src_register_ext_mod.Absolute = 0;
- src_register_ext_mod.Negate = 0;
- src_register_ext_mod.Padding = 0;
- src_register_ext_mod.Extended = 0;
-
- return src_register_ext_mod;
-}
-
-unsigned
-tgsi_compare_src_register_ext_mod(
- struct tgsi_src_register_ext_mod a,
- struct tgsi_src_register_ext_mod b )
-{
- a.Padding = b.Padding = 0;
- a.Extended = b.Extended = 0;
- return compare32(&a, &b);
-}
-
-struct tgsi_src_register_ext_mod
-tgsi_build_src_register_ext_mod(
- unsigned complement,
- unsigned bias,
- unsigned scale_2x,
- unsigned absolute,
- unsigned negate,
- struct tgsi_token *prev_token,
- struct tgsi_instruction *instruction,
- struct tgsi_header *header )
-{
- struct tgsi_src_register_ext_mod src_register_ext_mod;
-
- assert( complement <= 1 );
- assert( bias <= 1 );
- assert( scale_2x <= 1 );
- assert( absolute <= 1 );
- assert( negate <= 1 );
-
- src_register_ext_mod = tgsi_default_src_register_ext_mod();
- src_register_ext_mod.Complement = complement;
- src_register_ext_mod.Bias = bias;
- src_register_ext_mod.Scale2X = scale_2x;
- src_register_ext_mod.Absolute = absolute;
- src_register_ext_mod.Negate = negate;
-
- prev_token->Extended = 1;
- instruction_grow( instruction, header );
-
- return src_register_ext_mod;
-}
-
struct tgsi_dimension
tgsi_default_dimension( void )
{
@@ -1031,7 +868,6 @@ tgsi_default_dimension( void )
dimension.Dimension = 0;
dimension.Padding = 0;
dimension.Index = 0;
- dimension.Extended = 0;
return dimension;
}
@@ -1065,7 +901,6 @@ tgsi_default_dst_register( void )
dst_register.Dimension = 0;
dst_register.Index = 0;
dst_register.Padding = 0;
- dst_register.Extended = 0;
return dst_register;
}
@@ -1101,53 +936,113 @@ tgsi_default_full_dst_register( void )
{
struct tgsi_full_dst_register full_dst_register;
- full_dst_register.DstRegister = tgsi_default_dst_register();
- full_dst_register.DstRegisterInd = tgsi_default_src_register();
- full_dst_register.DstRegisterExtModulate =
- tgsi_default_dst_register_ext_modulate();
+ full_dst_register.Register = tgsi_default_dst_register();
+ full_dst_register.Indirect = tgsi_default_src_register();
return full_dst_register;
}
-struct tgsi_dst_register_ext_modulate
-tgsi_default_dst_register_ext_modulate( void )
+struct tgsi_property
+tgsi_default_property( void )
{
- struct tgsi_dst_register_ext_modulate dst_register_ext_modulate;
+ struct tgsi_property property;
- dst_register_ext_modulate.Type = TGSI_DST_REGISTER_EXT_TYPE_MODULATE;
- dst_register_ext_modulate.Modulate = TGSI_MODULATE_1X;
- dst_register_ext_modulate.Padding = 0;
- dst_register_ext_modulate.Extended = 0;
+ property.Type = TGSI_TOKEN_TYPE_PROPERTY;
+ property.NrTokens = 1;
+ property.PropertyName = TGSI_PROPERTY_GS_INPUT_PRIM;
+ property.Padding = 0;
- return dst_register_ext_modulate;
+ return property;
}
-unsigned
-tgsi_compare_dst_register_ext_modulate(
- struct tgsi_dst_register_ext_modulate a,
- struct tgsi_dst_register_ext_modulate b )
+struct tgsi_property
+tgsi_build_property(unsigned property_name,
+ struct tgsi_header *header)
{
- a.Padding = b.Padding = 0;
- a.Extended = b.Extended = 0;
- return compare32(&a, &b);
+ struct tgsi_property property;
+
+ property = tgsi_default_property();
+ property.PropertyName = property_name;
+
+ header_bodysize_grow( header );
+
+ return property;
}
-struct tgsi_dst_register_ext_modulate
-tgsi_build_dst_register_ext_modulate(
- unsigned modulate,
- struct tgsi_token *prev_token,
- struct tgsi_instruction *instruction,
+
+struct tgsi_full_property
+tgsi_default_full_property( void )
+{
+ struct tgsi_full_property full_property;
+
+ full_property.Property = tgsi_default_property();
+ memset(full_property.u, 0,
+ sizeof(struct tgsi_property_data) * 8);
+
+ return full_property;
+}
+
+static void
+property_grow(
+ struct tgsi_property *property,
struct tgsi_header *header )
{
- struct tgsi_dst_register_ext_modulate dst_register_ext_modulate;
+ assert( property->NrTokens < 0xFF );
- assert( modulate <= TGSI_MODULATE_EIGHTH );
+ property->NrTokens++;
- dst_register_ext_modulate = tgsi_default_dst_register_ext_modulate();
- dst_register_ext_modulate.Modulate = modulate;
+ header_bodysize_grow( header );
+}
- prev_token->Extended = 1;
- instruction_grow( instruction, header );
+struct tgsi_property_data
+tgsi_build_property_data(
+ unsigned value,
+ struct tgsi_property *property,
+ struct tgsi_header *header )
+{
+ struct tgsi_property_data property_data;
+
+ property_data.Data = value;
+
+ property_grow( property, header );
+
+ return property_data;
+}
+
+unsigned
+tgsi_build_full_property(
+ const struct tgsi_full_property *full_prop,
+ struct tgsi_token *tokens,
+ struct tgsi_header *header,
+ unsigned maxsize )
+{
+ unsigned size = 0, i;
+ struct tgsi_property *property;
- return dst_register_ext_modulate;
+ if( maxsize <= size )
+ return 0;
+ property = (struct tgsi_property *) &tokens[size];
+ size++;
+
+ *property = tgsi_build_property(
+ full_prop->Property.PropertyName,
+ header );
+
+ assert( full_prop->Property.NrTokens <= 8 + 1 );
+
+ for( i = 0; i < full_prop->Property.NrTokens - 1; i++ ) {
+ struct tgsi_property_data *data;
+
+ if( maxsize <= size )
+ return 0;
+ data = (struct tgsi_property_data *) &tokens[size];
+ size++;
+
+ *data = tgsi_build_property_data(
+ full_prop->u[i].Data,
+ property,
+ header );
+ }
+
+ return size;
}
diff --git a/src/gallium/auxiliary/tgsi/tgsi_build.h b/src/gallium/auxiliary/tgsi/tgsi_build.h
index 669712eb8f9..9de2757fe40 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_build.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_build.h
@@ -36,12 +36,6 @@ struct tgsi_token;
extern "C" {
#endif
-/*
- * version
- */
-
-struct tgsi_version
-tgsi_build_version( void );
/*
* header
@@ -133,6 +127,34 @@ tgsi_build_full_immediate(
unsigned maxsize );
/*
+ * properties
+ */
+
+struct tgsi_property
+tgsi_default_property( void );
+
+struct tgsi_property
+tgsi_build_property(
+ unsigned property_name,
+ struct tgsi_header *header );
+
+struct tgsi_full_property
+tgsi_default_full_property( void );
+
+struct tgsi_property_data
+tgsi_build_property_data(
+ unsigned value,
+ struct tgsi_property *property,
+ struct tgsi_header *header );
+
+unsigned
+tgsi_build_full_property(
+ const struct tgsi_full_property *full_prop,
+ struct tgsi_token *tokens,
+ struct tgsi_header *header,
+ unsigned maxsize );
+
+/*
* instruction
*/
@@ -143,6 +165,7 @@ struct tgsi_instruction
tgsi_build_instruction(
unsigned opcode,
unsigned saturate,
+ unsigned predicate,
unsigned num_dst_regs,
unsigned num_src_regs,
struct tgsi_header *header );
@@ -157,54 +180,39 @@ tgsi_build_full_instruction(
struct tgsi_header *header,
unsigned maxsize );
-struct tgsi_instruction_ext_label
-tgsi_default_instruction_ext_label( void );
+struct tgsi_instruction_predicate
+tgsi_default_instruction_predicate(void);
-unsigned
-tgsi_compare_instruction_ext_label(
- struct tgsi_instruction_ext_label a,
- struct tgsi_instruction_ext_label b );
+struct tgsi_instruction_predicate
+tgsi_build_instruction_predicate(int index,
+ unsigned negate,
+ unsigned swizzleX,
+ unsigned swizzleY,
+ unsigned swizzleZ,
+ unsigned swizzleW,
+ struct tgsi_instruction *instruction,
+ struct tgsi_header *header);
+
+struct tgsi_instruction_label
+tgsi_default_instruction_label( void );
-struct tgsi_instruction_ext_label
-tgsi_build_instruction_ext_label(
+struct tgsi_instruction_label
+tgsi_build_instruction_label(
unsigned label,
struct tgsi_token *prev_token,
struct tgsi_instruction *instruction,
struct tgsi_header *header );
-struct tgsi_instruction_ext_texture
-tgsi_default_instruction_ext_texture( void );
+struct tgsi_instruction_texture
+tgsi_default_instruction_texture( void );
-unsigned
-tgsi_compare_instruction_ext_texture(
- struct tgsi_instruction_ext_texture a,
- struct tgsi_instruction_ext_texture b );
-
-struct tgsi_instruction_ext_texture
-tgsi_build_instruction_ext_texture(
+struct tgsi_instruction_texture
+tgsi_build_instruction_texture(
unsigned texture,
struct tgsi_token *prev_token,
struct tgsi_instruction *instruction,
struct tgsi_header *header );
-struct tgsi_instruction_ext_predicate
-tgsi_default_instruction_ext_predicate(void);
-
-unsigned
-tgsi_compare_instruction_ext_predicate(struct tgsi_instruction_ext_predicate a,
- struct tgsi_instruction_ext_predicate b);
-
-struct tgsi_instruction_ext_predicate
-tgsi_build_instruction_ext_predicate(unsigned index,
- unsigned negate,
- unsigned swizzleX,
- unsigned swizzleY,
- unsigned swizzleZ,
- unsigned swizzleW,
- struct tgsi_token *prev_token,
- struct tgsi_instruction *instruction,
- struct tgsi_header *header);
-
struct tgsi_src_register
tgsi_default_src_register( void );
@@ -216,6 +224,7 @@ tgsi_build_src_register(
unsigned swizzle_z,
unsigned swizzle_w,
unsigned negate,
+ unsigned absolute,
unsigned indirect,
unsigned dimension,
int index,
@@ -225,24 +234,6 @@ tgsi_build_src_register(
struct tgsi_full_src_register
tgsi_default_full_src_register( void );
-struct tgsi_src_register_ext_mod
-tgsi_default_src_register_ext_mod( void );
-
-unsigned
-tgsi_compare_src_register_ext_mod(
- struct tgsi_src_register_ext_mod a,
- struct tgsi_src_register_ext_mod b );
-
-struct tgsi_src_register_ext_mod
-tgsi_build_src_register_ext_mod(
- unsigned complement,
- unsigned bias,
- unsigned scale_2x,
- unsigned absolute,
- unsigned negate,
- struct tgsi_token *prev_token,
- struct tgsi_instruction *instruction,
- struct tgsi_header *header );
struct tgsi_dimension
tgsi_default_dimension( void );
@@ -269,20 +260,6 @@ tgsi_build_dst_register(
struct tgsi_full_dst_register
tgsi_default_full_dst_register( void );
-struct tgsi_dst_register_ext_modulate
-tgsi_default_dst_register_ext_modulate( void );
-
-unsigned
-tgsi_compare_dst_register_ext_modulate(
- struct tgsi_dst_register_ext_modulate a,
- struct tgsi_dst_register_ext_modulate b );
-
-struct tgsi_dst_register_ext_modulate
-tgsi_build_dst_register_ext_modulate(
- unsigned modulate,
- struct tgsi_token *prev_token,
- struct tgsi_instruction *instruction,
- struct tgsi_header *header );
#if defined __cplusplus
}
diff --git a/src/gallium/auxiliary/tgsi/tgsi_dump.c b/src/gallium/auxiliary/tgsi/tgsi_dump.c
index d16e64f9c59..d7ff262f30a 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_dump.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_dump.c
@@ -101,7 +101,8 @@ static const char *file_names[TGSI_FILE_COUNT] =
"ADDR",
"IMM",
"LOOP",
- "PRED"
+ "PRED",
+ "SV"
};
static const char *interpolate_names[] =
@@ -120,12 +121,17 @@ static const char *semantic_names[] =
"PSIZE",
"GENERIC",
"NORMAL",
- "FACE"
+ "FACE",
+ "EDGEFLAG",
+ "PRIM_ID",
+ "INSTANCEID"
};
static const char *immediate_type_names[] =
{
- "FLT32"
+ "FLT32",
+ "UINT32",
+ "INT32"
};
static const char *swizzle_names[] =
@@ -149,26 +155,42 @@ static const char *texture_names[] =
"SHADOWRECT"
};
+static const char *property_names[] =
+{
+ "GS_INPUT_PRIMITIVE",
+ "GS_OUTPUT_PRIMITIVE",
+ "GS_MAX_OUTPUT_VERTICES"
+};
-static const char *modulate_names[TGSI_MODULATE_COUNT] =
+static const char *primitive_names[] =
{
- "",
- "_2X",
- "_4X",
- "_8X",
- "_D2",
- "_D4",
- "_D8"
+ "POINTS",
+ "LINES",
+ "LINE_LOOP",
+ "LINE_STRIP",
+ "TRIANGLES",
+ "TRIANGLE_STRIP",
+ "TRIANGLE_FAN",
+ "QUADS",
+ "QUAD_STRIP",
+ "POLYGON"
};
+
static void
-_dump_register(
+_dump_register_decl(
struct dump_ctx *ctx,
uint file,
int first,
int last )
{
ENM( file, file_names );
+
+ /* all geometry shader inputs are two dimensional */
+ if (file == TGSI_FILE_INPUT &&
+ ctx->iter.processor.Processor == TGSI_PROCESSOR_GEOMETRY)
+ TXT("[]");
+
CHR( '[' );
SID( first );
if (first != last) {
@@ -179,6 +201,52 @@ _dump_register(
}
static void
+_dump_register_dst(
+ struct dump_ctx *ctx,
+ uint file,
+ int index)
+{
+ ENM( file, file_names );
+
+ CHR( '[' );
+ SID( index );
+ CHR( ']' );
+}
+
+
+static void
+_dump_register_src(
+ struct dump_ctx *ctx,
+ const struct tgsi_full_src_register *src )
+{
+ if (src->Register.Indirect) {
+ ENM( src->Register.File, file_names );
+ CHR( '[' );
+ ENM( src->Indirect.File, file_names );
+ CHR( '[' );
+ SID( src->Indirect.Index );
+ TXT( "]." );
+ ENM( src->Indirect.SwizzleX, swizzle_names );
+ if (src->Register.Index != 0) {
+ if (src->Register.Index > 0)
+ CHR( '+' );
+ SID( src->Register.Index );
+ }
+ CHR( ']' );
+ } else {
+ ENM( src->Register.File, file_names );
+ CHR( '[' );
+ SID( src->Register.Index );
+ CHR( ']' );
+ }
+ if (src->Register.Dimension) {
+ CHR( '[' );
+ SID( src->Dimension.Index );
+ CHR( ']' );
+ }
+}
+
+static void
_dump_register_ind(
struct dump_ctx *ctx,
uint file,
@@ -232,22 +300,22 @@ iter_declaration(
TXT( "DCL " );
- _dump_register(
+ _dump_register_decl(
ctx,
decl->Declaration.File,
- decl->DeclarationRange.First,
- decl->DeclarationRange.Last );
+ decl->Range.First,
+ decl->Range.Last );
_dump_writemask(
ctx,
decl->Declaration.UsageMask );
if (decl->Declaration.Semantic) {
TXT( ", " );
- ENM( decl->Semantic.SemanticName, semantic_names );
- if (decl->Semantic.SemanticIndex != 0 ||
- decl->Semantic.SemanticName == TGSI_SEMANTIC_GENERIC) {
+ ENM( decl->Semantic.Name, semantic_names );
+ if (decl->Semantic.Index != 0 ||
+ decl->Semantic.Name == TGSI_SEMANTIC_GENERIC) {
CHR( '[' );
- UID( decl->Semantic.SemanticIndex );
+ UID( decl->Semantic.Index );
CHR( ']' );
}
}
@@ -284,6 +352,50 @@ tgsi_dump_declaration(
}
static boolean
+iter_property(
+ struct tgsi_iterate_context *iter,
+ struct tgsi_full_property *prop )
+{
+ int i;
+ struct dump_ctx *ctx = (struct dump_ctx *)iter;
+
+ assert(Elements(property_names) == TGSI_PROPERTY_COUNT);
+
+ TXT( "PROPERTY " );
+ ENM(prop->Property.PropertyName, property_names);
+
+ if (prop->Property.NrTokens > 1)
+ TXT(" ");
+
+ for (i = 0; i < prop->Property.NrTokens - 1; ++i) {
+ switch (prop->Property.PropertyName) {
+ case TGSI_PROPERTY_GS_INPUT_PRIM:
+ case TGSI_PROPERTY_GS_OUTPUT_PRIM:
+ ENM(prop->u[i].Data, primitive_names);
+ break;
+ default:
+ SID( prop->u[i].Data );
+ break;
+ }
+ if (i < prop->Property.NrTokens - 2)
+ TXT( ", " );
+ }
+ EOL();
+
+ return TRUE;
+}
+
+void tgsi_dump_property(
+ const struct tgsi_full_property *prop )
+{
+ struct dump_ctx ctx;
+
+ ctx.printf = dump_ctx_printf;
+
+ iter_property( &ctx.iter, (struct tgsi_full_property *)prop );
+}
+
+static boolean
iter_immediate(
struct tgsi_iterate_context *iter,
struct tgsi_full_immediate *imm )
@@ -303,6 +415,12 @@ iter_immediate(
case TGSI_IMM_FLOAT32:
FLT( imm->u[i].Float );
break;
+ case TGSI_IMM_UINT32:
+ UID(imm->u[i].Uint);
+ break;
+ case TGSI_IMM_INT32:
+ SID(imm->u[i].Int);
+ break;
default:
assert( 0 );
}
@@ -363,99 +481,66 @@ iter_instruction(
}
for (i = 0; i < inst->Instruction.NumDstRegs; i++) {
- const struct tgsi_full_dst_register *dst = &inst->FullDstRegisters[i];
+ const struct tgsi_full_dst_register *dst = &inst->Dst[i];
if (!first_reg)
CHR( ',' );
CHR( ' ' );
- if (dst->DstRegister.Indirect) {
+ if (dst->Register.Indirect) {
_dump_register_ind(
ctx,
- dst->DstRegister.File,
- dst->DstRegister.Index,
- dst->DstRegisterInd.File,
- dst->DstRegisterInd.Index,
- dst->DstRegisterInd.SwizzleX );
+ dst->Register.File,
+ dst->Register.Index,
+ dst->Indirect.File,
+ dst->Indirect.Index,
+ dst->Indirect.SwizzleX );
}
else {
- _dump_register(
+ _dump_register_dst(
ctx,
- dst->DstRegister.File,
- dst->DstRegister.Index,
- dst->DstRegister.Index );
+ dst->Register.File,
+ dst->Register.Index );
}
- ENM( dst->DstRegisterExtModulate.Modulate, modulate_names );
- _dump_writemask( ctx, dst->DstRegister.WriteMask );
+ _dump_writemask( ctx, dst->Register.WriteMask );
first_reg = FALSE;
}
for (i = 0; i < inst->Instruction.NumSrcRegs; i++) {
- const struct tgsi_full_src_register *src = &inst->FullSrcRegisters[i];
+ const struct tgsi_full_src_register *src = &inst->Src[i];
if (!first_reg)
CHR( ',' );
CHR( ' ' );
- if (src->SrcRegisterExtMod.Negate)
- TXT( "-(" );
- if (src->SrcRegisterExtMod.Absolute)
- CHR( '|' );
- if (src->SrcRegisterExtMod.Scale2X)
- TXT( "2*(" );
- if (src->SrcRegisterExtMod.Bias)
- CHR( '(' );
- if (src->SrcRegisterExtMod.Complement)
- TXT( "1-(" );
- if (src->SrcRegister.Negate)
+ if (src->Register.Negate)
CHR( '-' );
+ if (src->Register.Absolute)
+ CHR( '|' );
- if (src->SrcRegister.Indirect) {
- _dump_register_ind(
- ctx,
- src->SrcRegister.File,
- src->SrcRegister.Index,
- src->SrcRegisterInd.File,
- src->SrcRegisterInd.Index,
- src->SrcRegisterInd.SwizzleX );
- }
- else {
- _dump_register(
- ctx,
- src->SrcRegister.File,
- src->SrcRegister.Index,
- src->SrcRegister.Index );
- }
+ _dump_register_src(ctx, src);
- if (src->SrcRegister.SwizzleX != TGSI_SWIZZLE_X ||
- src->SrcRegister.SwizzleY != TGSI_SWIZZLE_Y ||
- src->SrcRegister.SwizzleZ != TGSI_SWIZZLE_Z ||
- src->SrcRegister.SwizzleW != TGSI_SWIZZLE_W) {
+ if (src->Register.SwizzleX != TGSI_SWIZZLE_X ||
+ src->Register.SwizzleY != TGSI_SWIZZLE_Y ||
+ src->Register.SwizzleZ != TGSI_SWIZZLE_Z ||
+ src->Register.SwizzleW != TGSI_SWIZZLE_W) {
CHR( '.' );
- ENM( src->SrcRegister.SwizzleX, swizzle_names );
- ENM( src->SrcRegister.SwizzleY, swizzle_names );
- ENM( src->SrcRegister.SwizzleZ, swizzle_names );
- ENM( src->SrcRegister.SwizzleW, swizzle_names );
+ ENM( src->Register.SwizzleX, swizzle_names );
+ ENM( src->Register.SwizzleY, swizzle_names );
+ ENM( src->Register.SwizzleZ, swizzle_names );
+ ENM( src->Register.SwizzleW, swizzle_names );
}
- if (src->SrcRegisterExtMod.Complement)
- CHR( ')' );
- if (src->SrcRegisterExtMod.Bias)
- TXT( ")-.5" );
- if (src->SrcRegisterExtMod.Scale2X)
- CHR( ')' );
- if (src->SrcRegisterExtMod.Absolute)
+ if (src->Register.Absolute)
CHR( '|' );
- if (src->SrcRegisterExtMod.Negate)
- CHR( ')' );
first_reg = FALSE;
}
- if (inst->InstructionExtTexture.Texture != TGSI_TEXTURE_UNKNOWN) {
+ if (inst->Instruction.Texture) {
TXT( ", " );
- ENM( inst->InstructionExtTexture.Texture, texture_names );
+ ENM( inst->Texture.Texture, texture_names );
}
switch (inst->Instruction.Opcode) {
@@ -465,7 +550,7 @@ iter_instruction(
case TGSI_OPCODE_ENDLOOP:
case TGSI_OPCODE_CAL:
TXT( " :" );
- UID( inst->InstructionExtLabel.Label );
+ UID( inst->Label.Label );
break;
}
@@ -503,9 +588,6 @@ prolog(
{
struct dump_ctx *ctx = (struct dump_ctx *) iter;
ENM( iter->processor.Processor, processor_type_names );
- UID( iter->version.MajorVersion );
- CHR( '.' );
- UID( iter->version.MinorVersion );
EOL();
return TRUE;
}
@@ -521,6 +603,7 @@ tgsi_dump(
ctx.iter.iterate_instruction = iter_instruction;
ctx.iter.iterate_declaration = iter_declaration;
ctx.iter.iterate_immediate = iter_immediate;
+ ctx.iter.iterate_property = iter_property;
ctx.iter.epilog = NULL;
ctx.instno = 0;
@@ -575,6 +658,7 @@ tgsi_dump_str(
ctx.base.iter.iterate_instruction = iter_instruction;
ctx.base.iter.iterate_declaration = iter_declaration;
ctx.base.iter.iterate_immediate = iter_immediate;
+ ctx.base.iter.iterate_property = iter_property;
ctx.base.iter.epilog = NULL;
ctx.base.instno = 0;
diff --git a/src/gallium/auxiliary/tgsi/tgsi_dump.h b/src/gallium/auxiliary/tgsi/tgsi_dump.h
index ad1e647ec90..4cd27317b36 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_dump.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_dump.h
@@ -49,6 +49,7 @@ tgsi_dump(
struct tgsi_full_immediate;
struct tgsi_full_instruction;
struct tgsi_full_declaration;
+struct tgsi_full_property;
void
tgsi_dump_immediate(
@@ -63,6 +64,10 @@ void
tgsi_dump_declaration(
const struct tgsi_full_declaration *decl );
+void
+tgsi_dump_property(
+ const struct tgsi_full_property *prop );
+
#if defined __cplusplus
}
#endif
diff --git a/src/gallium/auxiliary/tgsi/tgsi_dump_c.c b/src/gallium/auxiliary/tgsi/tgsi_dump_c.c
index 4648051e29e..47fd1dd590e 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_dump_c.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_dump_c.c
@@ -113,13 +113,6 @@ static const char *TGSI_SATS[] =
"SAT_MINUS_PLUS_ONE"
};
-static const char *TGSI_INSTRUCTION_EXTS[] =
-{
- "",
- "INSTRUCTION_EXT_TYPE_LABEL",
- "INSTRUCTION_EXT_TYPE_TEXTURE"
-};
-
static const char *TGSI_SWIZZLES[] =
{
"SWIZZLE_X",
@@ -141,12 +134,6 @@ static const char *TGSI_TEXTURES[] =
"TEXTURE_SHADOWRECT"
};
-static const char *TGSI_SRC_REGISTER_EXTS[] =
-{
- "",
- "SRC_REGISTER_EXT_TYPE_MOD"
-};
-
static const char *TGSI_WRITEMASKS[] =
{
"0",
@@ -167,23 +154,6 @@ static const char *TGSI_WRITEMASKS[] =
"WRITEMASK_XYZW"
};
-static const char *TGSI_DST_REGISTER_EXTS[] =
-{
- "",
- "DST_REGISTER_EXT_TYPE_MODULATE"
-};
-
-static const char *TGSI_MODULATES[] =
-{
- "MODULATE_1X",
- "MODULATE_2X",
- "MODULATE_4X",
- "MODULATE_8X",
- "MODULATE_HALF",
- "MODULATE_QUARTER",
- "MODULATE_EIGHTH"
-};
-
static void
dump_declaration_verbose(
struct tgsi_full_declaration *decl,
@@ -216,6 +186,14 @@ dump_declaration_verbose(
TXT( "\nSemantic : " );
UID( decl->Declaration.Semantic );
}
+ if (deflt || fd->Declaration.Centroid != decl->Declaration.Centroid) {
+ TXT("\nCentroid : ");
+ UID(decl->Declaration.Centroid);
+ }
+ if (deflt || fd->Declaration.Invariant != decl->Declaration.Invariant) {
+ TXT("\nInvariant : ");
+ UID(decl->Declaration.Invariant);
+ }
if( ignored ) {
TXT( "\nPadding : " );
UIX( decl->Declaration.Padding );
@@ -223,16 +201,16 @@ dump_declaration_verbose(
EOL();
TXT( "\nFirst: " );
- UID( decl->DeclarationRange.First );
+ UID( decl->Range.First );
TXT( "\nLast : " );
- UID( decl->DeclarationRange.Last );
+ UID( decl->Range.Last );
if( decl->Declaration.Semantic ) {
EOL();
- TXT( "\nSemanticName : " );
- ENM( decl->Semantic.SemanticName, TGSI_SEMANTICS );
- TXT( "\nSemanticIndex: " );
- UID( decl->Semantic.SemanticIndex );
+ TXT( "\nName : " );
+ ENM( decl->Semantic.Name, TGSI_SEMANTICS );
+ TXT( "\nIndex: " );
+ UID( decl->Semantic.Index );
if( ignored ) {
TXT( "\nPadding : " );
UIX( decl->Semantic.Padding );
@@ -292,180 +270,122 @@ dump_instruction_verbose(
TXT( "\nNumSrcRegs : " );
UID( inst->Instruction.NumSrcRegs );
}
+ if (deflt || fi->Instruction.Predicate != inst->Instruction.Predicate) {
+ TXT("\nPredicate : ");
+ UID(inst->Instruction.Predicate);
+ }
+ if (deflt || fi->Instruction.Label != inst->Instruction.Label) {
+ TXT("\nLabel : ");
+ UID(inst->Instruction.Label);
+ }
+ if (deflt || fi->Instruction.Texture != inst->Instruction.Texture) {
+ TXT("\nTexture : ");
+ UID(inst->Instruction.Texture);
+ }
if( ignored ) {
TXT( "\nPadding : " );
UIX( inst->Instruction.Padding );
}
- if( deflt || tgsi_compare_instruction_ext_label( inst->InstructionExtLabel, fi->InstructionExtLabel ) ) {
+ if (deflt || inst->Instruction.Label) {
EOL();
- TXT( "\nType : " );
- ENM( inst->InstructionExtLabel.Type, TGSI_INSTRUCTION_EXTS );
- if( deflt || fi->InstructionExtLabel.Label != inst->InstructionExtLabel.Label ) {
+ if (deflt || fi->Label.Label != inst->Label.Label) {
TXT( "\nLabel : " );
- UID( inst->InstructionExtLabel.Label );
+ UID(inst->Label.Label);
}
if( ignored ) {
TXT( "\nPadding : " );
- UIX( inst->InstructionExtLabel.Padding );
- if( deflt || fi->InstructionExtLabel.Extended != inst->InstructionExtLabel.Extended ) {
- TXT( "\nExtended: " );
- UID( inst->InstructionExtLabel.Extended );
- }
+ UIX(inst->Label.Padding);
}
}
- if( deflt || tgsi_compare_instruction_ext_texture( inst->InstructionExtTexture, fi->InstructionExtTexture ) ) {
+ if (deflt || inst->Instruction.Texture) {
EOL();
- TXT( "\nType : " );
- ENM( inst->InstructionExtTexture.Type, TGSI_INSTRUCTION_EXTS );
- if( deflt || fi->InstructionExtTexture.Texture != inst->InstructionExtTexture.Texture ) {
+ if (deflt || fi->Texture.Texture != inst->Texture.Texture) {
TXT( "\nTexture : " );
- ENM( inst->InstructionExtTexture.Texture, TGSI_TEXTURES );
+ ENM(inst->Texture.Texture, TGSI_TEXTURES);
}
if( ignored ) {
TXT( "\nPadding : " );
- UIX( inst->InstructionExtTexture.Padding );
- if( deflt || fi->InstructionExtTexture.Extended != inst->InstructionExtTexture.Extended ) {
- TXT( "\nExtended: " );
- UID( inst->InstructionExtTexture.Extended );
- }
+ UIX(inst->Texture.Padding);
}
}
for( i = 0; i < inst->Instruction.NumDstRegs; i++ ) {
- struct tgsi_full_dst_register *dst = &inst->FullDstRegisters[i];
- struct tgsi_full_dst_register *fd = &fi->FullDstRegisters[i];
+ struct tgsi_full_dst_register *dst = &inst->Dst[i];
+ struct tgsi_full_dst_register *fd = &fi->Dst[i];
EOL();
TXT( "\nFile : " );
- ENM( dst->DstRegister.File, TGSI_FILES );
- if( deflt || fd->DstRegister.WriteMask != dst->DstRegister.WriteMask ) {
+ ENM( dst->Register.File, TGSI_FILES );
+ if( deflt || fd->Register.WriteMask != dst->Register.WriteMask ) {
TXT( "\nWriteMask: " );
- ENM( dst->DstRegister.WriteMask, TGSI_WRITEMASKS );
+ ENM( dst->Register.WriteMask, TGSI_WRITEMASKS );
}
if( ignored ) {
- if( deflt || fd->DstRegister.Indirect != dst->DstRegister.Indirect ) {
+ if( deflt || fd->Register.Indirect != dst->Register.Indirect ) {
TXT( "\nIndirect : " );
- UID( dst->DstRegister.Indirect );
+ UID( dst->Register.Indirect );
}
- if( deflt || fd->DstRegister.Dimension != dst->DstRegister.Dimension ) {
+ if( deflt || fd->Register.Dimension != dst->Register.Dimension ) {
TXT( "\nDimension: " );
- UID( dst->DstRegister.Dimension );
+ UID( dst->Register.Dimension );
}
}
- if( deflt || fd->DstRegister.Index != dst->DstRegister.Index ) {
+ if( deflt || fd->Register.Index != dst->Register.Index ) {
TXT( "\nIndex : " );
- SID( dst->DstRegister.Index );
+ SID( dst->Register.Index );
}
if( ignored ) {
TXT( "\nPadding : " );
- UIX( dst->DstRegister.Padding );
- if( deflt || fd->DstRegister.Extended != dst->DstRegister.Extended ) {
- TXT( "\nExtended : " );
- UID( dst->DstRegister.Extended );
- }
- }
-
- if( deflt || tgsi_compare_dst_register_ext_modulate( dst->DstRegisterExtModulate, fd->DstRegisterExtModulate ) ) {
- EOL();
- TXT( "\nType : " );
- ENM( dst->DstRegisterExtModulate.Type, TGSI_DST_REGISTER_EXTS );
- if( deflt || fd->DstRegisterExtModulate.Modulate != dst->DstRegisterExtModulate.Modulate ) {
- TXT( "\nModulate: " );
- ENM( dst->DstRegisterExtModulate.Modulate, TGSI_MODULATES );
- }
- if( ignored ) {
- TXT( "\nPadding : " );
- UIX( dst->DstRegisterExtModulate.Padding );
- if( deflt || fd->DstRegisterExtModulate.Extended != dst->DstRegisterExtModulate.Extended ) {
- TXT( "\nExtended: " );
- UID( dst->DstRegisterExtModulate.Extended );
- }
- }
+ UIX( dst->Register.Padding );
}
}
for( i = 0; i < inst->Instruction.NumSrcRegs; i++ ) {
- struct tgsi_full_src_register *src = &inst->FullSrcRegisters[i];
- struct tgsi_full_src_register *fs = &fi->FullSrcRegisters[i];
+ struct tgsi_full_src_register *src = &inst->Src[i];
+ struct tgsi_full_src_register *fs = &fi->Src[i];
EOL();
TXT( "\nFile : ");
- ENM( src->SrcRegister.File, TGSI_FILES );
- if( deflt || fs->SrcRegister.SwizzleX != src->SrcRegister.SwizzleX ) {
+ ENM( src->Register.File, TGSI_FILES );
+ if( deflt || fs->Register.SwizzleX != src->Register.SwizzleX ) {
TXT( "\nSwizzleX : " );
- ENM( src->SrcRegister.SwizzleX, TGSI_SWIZZLES );
+ ENM( src->Register.SwizzleX, TGSI_SWIZZLES );
}
- if( deflt || fs->SrcRegister.SwizzleY != src->SrcRegister.SwizzleY ) {
+ if( deflt || fs->Register.SwizzleY != src->Register.SwizzleY ) {
TXT( "\nSwizzleY : " );
- ENM( src->SrcRegister.SwizzleY, TGSI_SWIZZLES );
+ ENM( src->Register.SwizzleY, TGSI_SWIZZLES );
}
- if( deflt || fs->SrcRegister.SwizzleZ != src->SrcRegister.SwizzleZ ) {
+ if( deflt || fs->Register.SwizzleZ != src->Register.SwizzleZ ) {
TXT( "\nSwizzleZ : " );
- ENM( src->SrcRegister.SwizzleZ, TGSI_SWIZZLES );
+ ENM( src->Register.SwizzleZ, TGSI_SWIZZLES );
}
- if( deflt || fs->SrcRegister.SwizzleW != src->SrcRegister.SwizzleW ) {
+ if( deflt || fs->Register.SwizzleW != src->Register.SwizzleW ) {
TXT( "\nSwizzleW : " );
- ENM( src->SrcRegister.SwizzleW, TGSI_SWIZZLES );
+ ENM( src->Register.SwizzleW, TGSI_SWIZZLES );
+ }
+ if (deflt || fs->Register.Absolute != src->Register.Absolute) {
+ TXT("\nAbsolute : ");
+ UID(src->Register.Absolute);
}
- if( deflt || fs->SrcRegister.Negate != src->SrcRegister.Negate ) {
+ if( deflt || fs->Register.Negate != src->Register.Negate ) {
TXT( "\nNegate : " );
- UID( src->SrcRegister.Negate );
+ UID( src->Register.Negate );
}
if( ignored ) {
- if( deflt || fs->SrcRegister.Indirect != src->SrcRegister.Indirect ) {
+ if( deflt || fs->Register.Indirect != src->Register.Indirect ) {
TXT( "\nIndirect : " );
- UID( src->SrcRegister.Indirect );
+ UID( src->Register.Indirect );
}
- if( deflt || fs->SrcRegister.Dimension != src->SrcRegister.Dimension ) {
+ if( deflt || fs->Register.Dimension != src->Register.Dimension ) {
TXT( "\nDimension: " );
- UID( src->SrcRegister.Dimension );
+ UID( src->Register.Dimension );
}
}
- if( deflt || fs->SrcRegister.Index != src->SrcRegister.Index ) {
+ if( deflt || fs->Register.Index != src->Register.Index ) {
TXT( "\nIndex : " );
- SID( src->SrcRegister.Index );
- }
- if( ignored ) {
- if( deflt || fs->SrcRegister.Extended != src->SrcRegister.Extended ) {
- TXT( "\nExtended : " );
- UID( src->SrcRegister.Extended );
- }
- }
-
- if( deflt || tgsi_compare_src_register_ext_mod( src->SrcRegisterExtMod, fs->SrcRegisterExtMod ) ) {
- EOL();
- TXT( "\nType : " );
- ENM( src->SrcRegisterExtMod.Type, TGSI_SRC_REGISTER_EXTS );
- if( deflt || fs->SrcRegisterExtMod.Complement != src->SrcRegisterExtMod.Complement ) {
- TXT( "\nComplement: " );
- UID( src->SrcRegisterExtMod.Complement );
- }
- if( deflt || fs->SrcRegisterExtMod.Bias != src->SrcRegisterExtMod.Bias ) {
- TXT( "\nBias : " );
- UID( src->SrcRegisterExtMod.Bias );
- }
- if( deflt || fs->SrcRegisterExtMod.Scale2X != src->SrcRegisterExtMod.Scale2X ) {
- TXT( "\nScale2X : " );
- UID( src->SrcRegisterExtMod.Scale2X );
- }
- if( deflt || fs->SrcRegisterExtMod.Absolute != src->SrcRegisterExtMod.Absolute ) {
- TXT( "\nAbsolute : " );
- UID( src->SrcRegisterExtMod.Absolute );
- }
- if( deflt || fs->SrcRegisterExtMod.Negate != src->SrcRegisterExtMod.Negate ) {
- TXT( "\nNegate : " );
- UID( src->SrcRegisterExtMod.Negate );
- }
- if( ignored ) {
- TXT( "\nPadding : " );
- UIX( src->SrcRegisterExtMod.Padding );
- if( deflt || fs->SrcRegisterExtMod.Extended != src->SrcRegisterExtMod.Extended ) {
- TXT( "\nExtended : " );
- UID( src->SrcRegisterExtMod.Extended );
- }
- }
+ SID( src->Register.Index );
}
}
}
@@ -485,12 +405,6 @@ tgsi_dump_c(
TXT( "tgsi-dump begin -----------------" );
- TXT( "\nMajorVersion: " );
- UID( parse.FullVersion.Version.MajorVersion );
- TXT( "\nMinorVersion: " );
- UID( parse.FullVersion.Version.MinorVersion );
- EOL();
-
TXT( "\nHeaderSize: " );
UID( parse.FullHeader.Header.HeaderSize );
TXT( "\nBodySize : " );
@@ -510,10 +424,6 @@ tgsi_dump_c(
if( ignored ) {
TXT( "\nSize : " );
UID( parse.FullToken.Token.NrTokens );
- if( deflt || parse.FullToken.Token.Extended ) {
- TXT( "\nExtended : " );
- UID( parse.FullToken.Token.Extended );
- }
}
switch( parse.FullToken.Token.Type ) {
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.c b/src/gallium/auxiliary/tgsi/tgsi_exec.c
index 146000adeb4..118a638ab48 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_exec.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_exec.c
@@ -2,6 +2,7 @@
*
* Copyright 2007-2008 Tungsten Graphics, Inc., Cedar Park, Texas.
* All Rights Reserved.
+ * Copyright 2009-2010 VMware, Inc. All rights Reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the
@@ -60,21 +61,337 @@
#include "util/u_memory.h"
#include "util/u_math.h"
-#define FAST_MATH 1
-/** for tgsi_full_instruction::Flags */
-#define SOA_DEPENDENCY_FLAG 0x1
+#define FAST_MATH 1
#define TILE_TOP_LEFT 0
#define TILE_TOP_RIGHT 1
#define TILE_BOTTOM_LEFT 2
#define TILE_BOTTOM_RIGHT 3
+static void
+micro_abs(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = fabsf(src->f[0]);
+ dst->f[1] = fabsf(src->f[1]);
+ dst->f[2] = fabsf(src->f[2]);
+ dst->f[3] = fabsf(src->f[3]);
+}
+
+static void
+micro_arl(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->i[0] = (int)floorf(src->f[0]);
+ dst->i[1] = (int)floorf(src->f[1]);
+ dst->i[2] = (int)floorf(src->f[2]);
+ dst->i[3] = (int)floorf(src->f[3]);
+}
+
+static void
+micro_arr(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->i[0] = (int)floorf(src->f[0] + 0.5f);
+ dst->i[1] = (int)floorf(src->f[1] + 0.5f);
+ dst->i[2] = (int)floorf(src->f[2] + 0.5f);
+ dst->i[3] = (int)floorf(src->f[3] + 0.5f);
+}
+
+static void
+micro_ceil(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = ceilf(src->f[0]);
+ dst->f[1] = ceilf(src->f[1]);
+ dst->f[2] = ceilf(src->f[2]);
+ dst->f[3] = ceilf(src->f[3]);
+}
+
+static void
+micro_cos(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = cosf(src->f[0]);
+ dst->f[1] = cosf(src->f[1]);
+ dst->f[2] = cosf(src->f[2]);
+ dst->f[3] = cosf(src->f[3]);
+}
+
+static void
+micro_ddx(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] =
+ dst->f[1] =
+ dst->f[2] =
+ dst->f[3] = src->f[TILE_BOTTOM_RIGHT] - src->f[TILE_BOTTOM_LEFT];
+}
+
+static void
+micro_ddy(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] =
+ dst->f[1] =
+ dst->f[2] =
+ dst->f[3] = src->f[TILE_BOTTOM_LEFT] - src->f[TILE_TOP_LEFT];
+}
+
+static void
+micro_exp2(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+#if FAST_MATH
+ dst->f[0] = util_fast_exp2(src->f[0]);
+ dst->f[1] = util_fast_exp2(src->f[1]);
+ dst->f[2] = util_fast_exp2(src->f[2]);
+ dst->f[3] = util_fast_exp2(src->f[3]);
+#else
+#if DEBUG
+ /* Inf is okay for this instruction, so clamp it to silence assertions. */
+ uint i;
+ union tgsi_exec_channel clamped;
+
+ for (i = 0; i < 4; i++) {
+ if (src->f[i] > 127.99999f) {
+ clamped.f[i] = 127.99999f;
+ } else if (src->f[i] < -126.99999f) {
+ clamped.f[i] = -126.99999f;
+ } else {
+ clamped.f[i] = src->f[i];
+ }
+ }
+ src = &clamped;
+#endif /* DEBUG */
+
+ dst->f[0] = powf(2.0f, src->f[0]);
+ dst->f[1] = powf(2.0f, src->f[1]);
+ dst->f[2] = powf(2.0f, src->f[2]);
+ dst->f[3] = powf(2.0f, src->f[3]);
+#endif /* FAST_MATH */
+}
+
+static void
+micro_flr(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = floorf(src->f[0]);
+ dst->f[1] = floorf(src->f[1]);
+ dst->f[2] = floorf(src->f[2]);
+ dst->f[3] = floorf(src->f[3]);
+}
+
+static void
+micro_frc(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = src->f[0] - floorf(src->f[0]);
+ dst->f[1] = src->f[1] - floorf(src->f[1]);
+ dst->f[2] = src->f[2] - floorf(src->f[2]);
+ dst->f[3] = src->f[3] - floorf(src->f[3]);
+}
+
+static void
+micro_iabs(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->i[0] = src->i[0] >= 0 ? src->i[0] : -src->i[0];
+ dst->i[1] = src->i[1] >= 0 ? src->i[1] : -src->i[1];
+ dst->i[2] = src->i[2] >= 0 ? src->i[2] : -src->i[2];
+ dst->i[3] = src->i[3] >= 0 ? src->i[3] : -src->i[3];
+}
+
+static void
+micro_ineg(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->i[0] = -src->i[0];
+ dst->i[1] = -src->i[1];
+ dst->i[2] = -src->i[2];
+ dst->i[3] = -src->i[3];
+}
+
+static void
+micro_lg2(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+#if FAST_MATH
+ dst->f[0] = util_fast_log2(src->f[0]);
+ dst->f[1] = util_fast_log2(src->f[1]);
+ dst->f[2] = util_fast_log2(src->f[2]);
+ dst->f[3] = util_fast_log2(src->f[3]);
+#else
+ dst->f[0] = logf(src->f[0]) * 1.442695f;
+ dst->f[1] = logf(src->f[1]) * 1.442695f;
+ dst->f[2] = logf(src->f[2]) * 1.442695f;
+ dst->f[3] = logf(src->f[3]) * 1.442695f;
+#endif
+}
+
+static void
+micro_lrp(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = src[0].f[0] * (src[1].f[0] - src[2].f[0]) + src[2].f[0];
+ dst->f[1] = src[0].f[1] * (src[1].f[1] - src[2].f[1]) + src[2].f[1];
+ dst->f[2] = src[0].f[2] * (src[1].f[2] - src[2].f[2]) + src[2].f[2];
+ dst->f[3] = src[0].f[3] * (src[1].f[3] - src[2].f[3]) + src[2].f[3];
+}
+
+static void
+micro_mad(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = src[0].f[0] * src[1].f[0] + src[2].f[0];
+ dst->f[1] = src[0].f[1] * src[1].f[1] + src[2].f[1];
+ dst->f[2] = src[0].f[2] * src[1].f[2] + src[2].f[2];
+ dst->f[3] = src[0].f[3] * src[1].f[3] + src[2].f[3];
+}
+
+static void
+micro_mov(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->u[0] = src->u[0];
+ dst->u[1] = src->u[1];
+ dst->u[2] = src->u[2];
+ dst->u[3] = src->u[3];
+}
+
+static void
+micro_rcp(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = 1.0f / src->f[0];
+ dst->f[1] = 1.0f / src->f[1];
+ dst->f[2] = 1.0f / src->f[2];
+ dst->f[3] = 1.0f / src->f[3];
+}
+
+static void
+micro_rnd(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = floorf(src->f[0] + 0.5f);
+ dst->f[1] = floorf(src->f[1] + 0.5f);
+ dst->f[2] = floorf(src->f[2] + 0.5f);
+ dst->f[3] = floorf(src->f[3] + 0.5f);
+}
+
+static void
+micro_rsq(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = 1.0f / sqrtf(fabsf(src->f[0]));
+ dst->f[1] = 1.0f / sqrtf(fabsf(src->f[1]));
+ dst->f[2] = 1.0f / sqrtf(fabsf(src->f[2]));
+ dst->f[3] = 1.0f / sqrtf(fabsf(src->f[3]));
+}
+
+static void
+micro_seq(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = src[0].f[0] == src[1].f[0] ? 1.0f : 0.0f;
+ dst->f[1] = src[0].f[1] == src[1].f[1] ? 1.0f : 0.0f;
+ dst->f[2] = src[0].f[2] == src[1].f[2] ? 1.0f : 0.0f;
+ dst->f[3] = src[0].f[3] == src[1].f[3] ? 1.0f : 0.0f;
+}
+
+static void
+micro_sge(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = src[0].f[0] >= src[1].f[0] ? 1.0f : 0.0f;
+ dst->f[1] = src[0].f[1] >= src[1].f[1] ? 1.0f : 0.0f;
+ dst->f[2] = src[0].f[2] >= src[1].f[2] ? 1.0f : 0.0f;
+ dst->f[3] = src[0].f[3] >= src[1].f[3] ? 1.0f : 0.0f;
+}
+
+static void
+micro_sgn(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = src->f[0] < 0.0f ? -1.0f : src->f[0] > 0.0f ? 1.0f : 0.0f;
+ dst->f[1] = src->f[1] < 0.0f ? -1.0f : src->f[1] > 0.0f ? 1.0f : 0.0f;
+ dst->f[2] = src->f[2] < 0.0f ? -1.0f : src->f[2] > 0.0f ? 1.0f : 0.0f;
+ dst->f[3] = src->f[3] < 0.0f ? -1.0f : src->f[3] > 0.0f ? 1.0f : 0.0f;
+}
+
+static void
+micro_sgt(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = src[0].f[0] > src[1].f[0] ? 1.0f : 0.0f;
+ dst->f[1] = src[0].f[1] > src[1].f[1] ? 1.0f : 0.0f;
+ dst->f[2] = src[0].f[2] > src[1].f[2] ? 1.0f : 0.0f;
+ dst->f[3] = src[0].f[3] > src[1].f[3] ? 1.0f : 0.0f;
+}
+
+static void
+micro_sin(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = sinf(src->f[0]);
+ dst->f[1] = sinf(src->f[1]);
+ dst->f[2] = sinf(src->f[2]);
+ dst->f[3] = sinf(src->f[3]);
+}
+
+static void
+micro_sle(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = src[0].f[0] <= src[1].f[0] ? 1.0f : 0.0f;
+ dst->f[1] = src[0].f[1] <= src[1].f[1] ? 1.0f : 0.0f;
+ dst->f[2] = src[0].f[2] <= src[1].f[2] ? 1.0f : 0.0f;
+ dst->f[3] = src[0].f[3] <= src[1].f[3] ? 1.0f : 0.0f;
+}
+
+static void
+micro_slt(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = src[0].f[0] < src[1].f[0] ? 1.0f : 0.0f;
+ dst->f[1] = src[0].f[1] < src[1].f[1] ? 1.0f : 0.0f;
+ dst->f[2] = src[0].f[2] < src[1].f[2] ? 1.0f : 0.0f;
+ dst->f[3] = src[0].f[3] < src[1].f[3] ? 1.0f : 0.0f;
+}
+
+static void
+micro_sne(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = src[0].f[0] != src[1].f[0] ? 1.0f : 0.0f;
+ dst->f[1] = src[0].f[1] != src[1].f[1] ? 1.0f : 0.0f;
+ dst->f[2] = src[0].f[2] != src[1].f[2] ? 1.0f : 0.0f;
+ dst->f[3] = src[0].f[3] != src[1].f[3] ? 1.0f : 0.0f;
+}
+
+static void
+micro_trunc(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = (float)(int)src->f[0];
+ dst->f[1] = (float)(int)src->f[1];
+ dst->f[2] = (float)(int)src->f[2];
+ dst->f[3] = (float)(int)src->f[3];
+}
+
+
#define CHAN_X 0
#define CHAN_Y 1
#define CHAN_Z 2
#define CHAN_W 3
+enum tgsi_exec_datatype {
+ TGSI_EXEC_DATA_FLOAT,
+ TGSI_EXEC_DATA_INT,
+ TGSI_EXEC_DATA_UINT
+};
+
/*
* Shorthand locations of various utility registers (_I = Index, _C = Channel)
*/
@@ -110,10 +427,10 @@
#define TEMP_P0 TGSI_EXEC_TEMP_P0
#define IS_CHANNEL_ENABLED(INST, CHAN)\
- ((INST).FullDstRegisters[0].DstRegister.WriteMask & (1 << (CHAN)))
+ ((INST).Dst[0].Register.WriteMask & (1 << (CHAN)))
#define IS_CHANNEL_ENABLED2(INST, CHAN)\
- ((INST).FullDstRegisters[1].DstRegister.WriteMask & (1 << (CHAN)))
+ ((INST).Dst[1].Register.WriteMask & (1 << (CHAN)))
#define FOR_EACH_ENABLED_CHANNEL(INST, CHAN)\
for (CHAN = 0; CHAN < NUM_CHANNELS; CHAN++)\
@@ -126,23 +443,19 @@
/** The execution mask depends on the conditional mask and the loop mask */
#define UPDATE_EXEC_MASK(MACH) \
- MACH->ExecMask = MACH->CondMask & MACH->LoopMask & MACH->ContMask & MACH->FuncMask
+ MACH->ExecMask = MACH->CondMask & MACH->LoopMask & MACH->ContMask & MACH->Switch.mask & MACH->FuncMask
static const union tgsi_exec_channel ZeroVec =
{ { 0.0, 0.0, 0.0, 0.0 } };
-#ifdef DEBUG
-static void
-check_inf_or_nan(const union tgsi_exec_channel *chan)
-{
- assert(!util_is_inf_or_nan(chan->f[0]));
- assert(!util_is_inf_or_nan(chan->f[1]));
- assert(!util_is_inf_or_nan(chan->f[2]));
- assert(!util_is_inf_or_nan(chan->f[3]));
-}
-#endif
+#define CHECK_INF_OR_NAN(chan) do {\
+ assert(!util_is_inf_or_nan((chan)->f[0]));\
+ assert(!util_is_inf_or_nan((chan)->f[1]));\
+ assert(!util_is_inf_or_nan((chan)->f[2]));\
+ assert(!util_is_inf_or_nan((chan)->f[3]));\
+ } while (0)
#ifdef DEBUG
@@ -191,7 +504,7 @@ tgsi_check_soa_dependencies(const struct tgsi_full_instruction *inst)
{
uint i, chan;
- uint writemask = inst->FullDstRegisters[0].DstRegister.WriteMask;
+ uint writemask = inst->Dst[0].Register.WriteMask;
if (writemask == TGSI_WRITEMASK_X ||
writemask == TGSI_WRITEMASK_Y ||
writemask == TGSI_WRITEMASK_Z ||
@@ -203,15 +516,15 @@ tgsi_check_soa_dependencies(const struct tgsi_full_instruction *inst)
/* loop over src regs */
for (i = 0; i < inst->Instruction.NumSrcRegs; i++) {
- if ((inst->FullSrcRegisters[i].SrcRegister.File ==
- inst->FullDstRegisters[0].DstRegister.File) &&
- (inst->FullSrcRegisters[i].SrcRegister.Index ==
- inst->FullDstRegisters[0].DstRegister.Index)) {
+ if ((inst->Src[i].Register.File ==
+ inst->Dst[0].Register.File) &&
+ (inst->Src[i].Register.Index ==
+ inst->Dst[0].Register.Index)) {
/* loop over dest channels */
uint channelsWritten = 0x0;
FOR_EACH_ENABLED_CHANNEL(*inst, chan) {
/* check if we're reading a channel that's been written */
- uint swizzle = tgsi_util_get_full_src_register_swizzle(&inst->FullSrcRegisters[i], chan);
+ uint swizzle = tgsi_util_get_full_src_register_swizzle(&inst->Src[i], chan);
if (channelsWritten & (1 << swizzle)) {
return TRUE;
}
@@ -295,6 +608,14 @@ tgsi_exec_machine_bind_shader(
* sizeof(struct tgsi_full_declaration));
maxDeclarations += 10;
}
+ if (parse.FullToken.FullDeclaration.Declaration.File == TGSI_FILE_OUTPUT) {
+ unsigned reg;
+ for (reg = parse.FullToken.FullDeclaration.Range.First;
+ reg <= parse.FullToken.FullDeclaration.Range.Last;
+ ++reg) {
+ ++mach->NumOutputs;
+ }
+ }
memcpy(declarations + numDeclarations,
&parse.FullToken.FullDeclaration,
sizeof(declarations[0]));
@@ -332,20 +653,6 @@ tgsi_exec_machine_bind_shader(
maxInstructions += 10;
}
- if (tgsi_check_soa_dependencies(&parse.FullToken.FullInstruction)) {
- uint opcode = parse.FullToken.FullInstruction.Instruction.Opcode;
- parse.FullToken.FullInstruction.Flags = SOA_DEPENDENCY_FLAG;
- /* XXX we only handle SOA dependencies properly for MOV/SWZ
- * at this time!
- */
- if (opcode != TGSI_OPCODE_MOV) {
- debug_printf("Warning: SOA dependency in instruction"
- " is not handled:\n");
- tgsi_dump_instruction(&parse.FullToken.FullInstruction,
- numInstructions);
- }
- }
-
memcpy(instructions + numInstructions,
&parse.FullToken.FullInstruction,
sizeof(instructions[0]));
@@ -353,6 +660,9 @@ tgsi_exec_machine_bind_shader(
numInstructions++;
break;
+ case TGSI_TOKEN_TYPE_PROPERTY:
+ break;
+
default:
assert( 0 );
}
@@ -386,6 +696,8 @@ tgsi_exec_machine_create( void )
memset(mach, 0, sizeof(*mach));
mach->Addrs = &mach->Temps[TGSI_EXEC_TEMP_ADDR];
+ mach->MaxGeometryShaderOutputs = TGSI_MAX_TOTAL_VERTICES;
+ mach->Predicates = &mach->Temps[TGSI_EXEC_TEMP_P0];
/* Setup constants. */
for( i = 0; i < 4; i++ ) {
@@ -426,18 +738,6 @@ tgsi_exec_machine_destroy(struct tgsi_exec_machine *mach)
align_free(mach);
}
-
-static void
-micro_abs(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src )
-{
- dst->f[0] = fabsf( src->f[0] );
- dst->f[1] = fabsf( src->f[1] );
- dst->f[2] = fabsf( src->f[2] );
- dst->f[3] = fabsf( src->f[3] );
-}
-
static void
micro_add(
union tgsi_exec_channel *dst,
@@ -450,76 +750,6 @@ micro_add(
dst->f[3] = src0->f[3] + src1->f[3];
}
-#if 0
-static void
-micro_iadd(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0,
- const union tgsi_exec_channel *src1 )
-{
- dst->i[0] = src0->i[0] + src1->i[0];
- dst->i[1] = src0->i[1] + src1->i[1];
- dst->i[2] = src0->i[2] + src1->i[2];
- dst->i[3] = src0->i[3] + src1->i[3];
-}
-#endif
-
-static void
-micro_and(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0,
- const union tgsi_exec_channel *src1 )
-{
- dst->u[0] = src0->u[0] & src1->u[0];
- dst->u[1] = src0->u[1] & src1->u[1];
- dst->u[2] = src0->u[2] & src1->u[2];
- dst->u[3] = src0->u[3] & src1->u[3];
-}
-
-static void
-micro_ceil(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src )
-{
- dst->f[0] = ceilf( src->f[0] );
- dst->f[1] = ceilf( src->f[1] );
- dst->f[2] = ceilf( src->f[2] );
- dst->f[3] = ceilf( src->f[3] );
-}
-
-static void
-micro_cos(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src )
-{
- dst->f[0] = cosf( src->f[0] );
- dst->f[1] = cosf( src->f[1] );
- dst->f[2] = cosf( src->f[2] );
- dst->f[3] = cosf( src->f[3] );
-}
-
-static void
-micro_ddx(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src )
-{
- dst->f[0] =
- dst->f[1] =
- dst->f[2] =
- dst->f[3] = src->f[TILE_BOTTOM_RIGHT] - src->f[TILE_BOTTOM_LEFT];
-}
-
-static void
-micro_ddy(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src )
-{
- dst->f[0] =
- dst->f[1] =
- dst->f[2] =
- dst->f[3] = src->f[TILE_TOP_LEFT] - src->f[TILE_BOTTOM_LEFT];
-}
-
static void
micro_div(
union tgsi_exec_channel *dst,
@@ -540,81 +770,6 @@ micro_div(
}
}
-#if 0
-static void
-micro_udiv(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0,
- const union tgsi_exec_channel *src1 )
-{
- dst->u[0] = src0->u[0] / src1->u[0];
- dst->u[1] = src0->u[1] / src1->u[1];
- dst->u[2] = src0->u[2] / src1->u[2];
- dst->u[3] = src0->u[3] / src1->u[3];
-}
-#endif
-
-static void
-micro_eq(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0,
- const union tgsi_exec_channel *src1,
- const union tgsi_exec_channel *src2,
- const union tgsi_exec_channel *src3 )
-{
- dst->f[0] = src0->f[0] == src1->f[0] ? src2->f[0] : src3->f[0];
- dst->f[1] = src0->f[1] == src1->f[1] ? src2->f[1] : src3->f[1];
- dst->f[2] = src0->f[2] == src1->f[2] ? src2->f[2] : src3->f[2];
- dst->f[3] = src0->f[3] == src1->f[3] ? src2->f[3] : src3->f[3];
-}
-
-#if 0
-static void
-micro_ieq(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0,
- const union tgsi_exec_channel *src1,
- const union tgsi_exec_channel *src2,
- const union tgsi_exec_channel *src3 )
-{
- dst->i[0] = src0->i[0] == src1->i[0] ? src2->i[0] : src3->i[0];
- dst->i[1] = src0->i[1] == src1->i[1] ? src2->i[1] : src3->i[1];
- dst->i[2] = src0->i[2] == src1->i[2] ? src2->i[2] : src3->i[2];
- dst->i[3] = src0->i[3] == src1->i[3] ? src2->i[3] : src3->i[3];
-}
-#endif
-
-static void
-micro_exp2(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src)
-{
-#if FAST_MATH
- dst->f[0] = util_fast_exp2( src->f[0] );
- dst->f[1] = util_fast_exp2( src->f[1] );
- dst->f[2] = util_fast_exp2( src->f[2] );
- dst->f[3] = util_fast_exp2( src->f[3] );
-#else
- dst->f[0] = powf( 2.0f, src->f[0] );
- dst->f[1] = powf( 2.0f, src->f[1] );
- dst->f[2] = powf( 2.0f, src->f[2] );
- dst->f[3] = powf( 2.0f, src->f[3] );
-#endif
-}
-
-#if 0
-static void
-micro_f2ut(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src )
-{
- dst->u[0] = (uint) src->f[0];
- dst->u[1] = (uint) src->f[1];
- dst->u[2] = (uint) src->f[2];
- dst->u[3] = (uint) src->f[3];
-}
-#endif
-
static void
micro_float_clamp(union tgsi_exec_channel *dst,
const union tgsi_exec_channel *src)
@@ -642,71 +797,6 @@ micro_float_clamp(union tgsi_exec_channel *dst,
}
static void
-micro_flr(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src )
-{
- dst->f[0] = floorf( src->f[0] );
- dst->f[1] = floorf( src->f[1] );
- dst->f[2] = floorf( src->f[2] );
- dst->f[3] = floorf( src->f[3] );
-}
-
-static void
-micro_frc(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src )
-{
- dst->f[0] = src->f[0] - floorf( src->f[0] );
- dst->f[1] = src->f[1] - floorf( src->f[1] );
- dst->f[2] = src->f[2] - floorf( src->f[2] );
- dst->f[3] = src->f[3] - floorf( src->f[3] );
-}
-
-static void
-micro_i2f(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src )
-{
- dst->f[0] = (float) src->i[0];
- dst->f[1] = (float) src->i[1];
- dst->f[2] = (float) src->i[2];
- dst->f[3] = (float) src->i[3];
-}
-
-static void
-micro_lg2(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src )
-{
-#if FAST_MATH
- dst->f[0] = util_fast_log2( src->f[0] );
- dst->f[1] = util_fast_log2( src->f[1] );
- dst->f[2] = util_fast_log2( src->f[2] );
- dst->f[3] = util_fast_log2( src->f[3] );
-#else
- dst->f[0] = logf( src->f[0] ) * 1.442695f;
- dst->f[1] = logf( src->f[1] ) * 1.442695f;
- dst->f[2] = logf( src->f[2] ) * 1.442695f;
- dst->f[3] = logf( src->f[3] ) * 1.442695f;
-#endif
-}
-
-static void
-micro_le(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0,
- const union tgsi_exec_channel *src1,
- const union tgsi_exec_channel *src2,
- const union tgsi_exec_channel *src3 )
-{
- dst->f[0] = src0->f[0] <= src1->f[0] ? src2->f[0] : src3->f[0];
- dst->f[1] = src0->f[1] <= src1->f[1] ? src2->f[1] : src3->f[1];
- dst->f[2] = src0->f[2] <= src1->f[2] ? src2->f[2] : src3->f[2];
- dst->f[3] = src0->f[3] <= src1->f[3] ? src2->f[3] : src3->f[3];
-}
-
-static void
micro_lt(
union tgsi_exec_channel *dst,
const union tgsi_exec_channel *src0,
@@ -720,38 +810,6 @@ micro_lt(
dst->f[3] = src0->f[3] < src1->f[3] ? src2->f[3] : src3->f[3];
}
-#if 0
-static void
-micro_ilt(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0,
- const union tgsi_exec_channel *src1,
- const union tgsi_exec_channel *src2,
- const union tgsi_exec_channel *src3 )
-{
- dst->i[0] = src0->i[0] < src1->i[0] ? src2->i[0] : src3->i[0];
- dst->i[1] = src0->i[1] < src1->i[1] ? src2->i[1] : src3->i[1];
- dst->i[2] = src0->i[2] < src1->i[2] ? src2->i[2] : src3->i[2];
- dst->i[3] = src0->i[3] < src1->i[3] ? src2->i[3] : src3->i[3];
-}
-#endif
-
-#if 0
-static void
-micro_ult(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0,
- const union tgsi_exec_channel *src1,
- const union tgsi_exec_channel *src2,
- const union tgsi_exec_channel *src3 )
-{
- dst->u[0] = src0->u[0] < src1->u[0] ? src2->u[0] : src3->u[0];
- dst->u[1] = src0->u[1] < src1->u[1] ? src2->u[1] : src3->u[1];
- dst->u[2] = src0->u[2] < src1->u[2] ? src2->u[2] : src3->u[2];
- dst->u[3] = src0->u[3] < src1->u[3] ? src2->u[3] : src3->u[3];
-}
-#endif
-
static void
micro_max(
union tgsi_exec_channel *dst,
@@ -764,34 +822,6 @@ micro_max(
dst->f[3] = src0->f[3] > src1->f[3] ? src0->f[3] : src1->f[3];
}
-#if 0
-static void
-micro_imax(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0,
- const union tgsi_exec_channel *src1 )
-{
- dst->i[0] = src0->i[0] > src1->i[0] ? src0->i[0] : src1->i[0];
- dst->i[1] = src0->i[1] > src1->i[1] ? src0->i[1] : src1->i[1];
- dst->i[2] = src0->i[2] > src1->i[2] ? src0->i[2] : src1->i[2];
- dst->i[3] = src0->i[3] > src1->i[3] ? src0->i[3] : src1->i[3];
-}
-#endif
-
-#if 0
-static void
-micro_umax(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0,
- const union tgsi_exec_channel *src1 )
-{
- dst->u[0] = src0->u[0] > src1->u[0] ? src0->u[0] : src1->u[0];
- dst->u[1] = src0->u[1] > src1->u[1] ? src0->u[1] : src1->u[1];
- dst->u[2] = src0->u[2] > src1->u[2] ? src0->u[2] : src1->u[2];
- dst->u[3] = src0->u[3] > src1->u[3] ? src0->u[3] : src1->u[3];
-}
-#endif
-
static void
micro_min(
union tgsi_exec_channel *dst,
@@ -804,48 +834,6 @@ micro_min(
dst->f[3] = src0->f[3] < src1->f[3] ? src0->f[3] : src1->f[3];
}
-#if 0
-static void
-micro_imin(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0,
- const union tgsi_exec_channel *src1 )
-{
- dst->i[0] = src0->i[0] < src1->i[0] ? src0->i[0] : src1->i[0];
- dst->i[1] = src0->i[1] < src1->i[1] ? src0->i[1] : src1->i[1];
- dst->i[2] = src0->i[2] < src1->i[2] ? src0->i[2] : src1->i[2];
- dst->i[3] = src0->i[3] < src1->i[3] ? src0->i[3] : src1->i[3];
-}
-#endif
-
-#if 0
-static void
-micro_umin(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0,
- const union tgsi_exec_channel *src1 )
-{
- dst->u[0] = src0->u[0] < src1->u[0] ? src0->u[0] : src1->u[0];
- dst->u[1] = src0->u[1] < src1->u[1] ? src0->u[1] : src1->u[1];
- dst->u[2] = src0->u[2] < src1->u[2] ? src0->u[2] : src1->u[2];
- dst->u[3] = src0->u[3] < src1->u[3] ? src0->u[3] : src1->u[3];
-}
-#endif
-
-#if 0
-static void
-micro_umod(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0,
- const union tgsi_exec_channel *src1 )
-{
- dst->u[0] = src0->u[0] % src1->u[0];
- dst->u[1] = src0->u[1] % src1->u[1];
- dst->u[2] = src0->u[2] % src1->u[2];
- dst->u[3] = src0->u[3] % src1->u[3];
-}
-#endif
-
static void
micro_mul(
union tgsi_exec_channel *dst,
@@ -860,20 +848,6 @@ micro_mul(
#if 0
static void
-micro_imul(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0,
- const union tgsi_exec_channel *src1 )
-{
- dst->i[0] = src0->i[0] * src1->i[0];
- dst->i[1] = src0->i[1] * src1->i[1];
- dst->i[2] = src0->i[2] * src1->i[2];
- dst->i[3] = src0->i[3] * src1->i[3];
-}
-#endif
-
-#if 0
-static void
micro_imul64(
union tgsi_exec_channel *dst0,
union tgsi_exec_channel *dst1,
@@ -937,42 +911,6 @@ micro_neg(
dst->f[3] = -src->f[3];
}
-#if 0
-static void
-micro_ineg(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src )
-{
- dst->i[0] = -src->i[0];
- dst->i[1] = -src->i[1];
- dst->i[2] = -src->i[2];
- dst->i[3] = -src->i[3];
-}
-#endif
-
-static void
-micro_not(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src )
-{
- dst->u[0] = ~src->u[0];
- dst->u[1] = ~src->u[1];
- dst->u[2] = ~src->u[2];
- dst->u[3] = ~src->u[3];
-}
-
-static void
-micro_or(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0,
- const union tgsi_exec_channel *src1 )
-{
- dst->u[0] = src0->u[0] | src1->u[0];
- dst->u[1] = src0->u[1] | src1->u[1];
- dst->u[2] = src0->u[2] | src1->u[2];
- dst->u[3] = src0->u[3] | src1->u[3];
-}
-
static void
micro_pow(
union tgsi_exec_channel *dst,
@@ -993,88 +931,6 @@ micro_pow(
}
static void
-micro_rnd(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src )
-{
- dst->f[0] = floorf( src->f[0] + 0.5f );
- dst->f[1] = floorf( src->f[1] + 0.5f );
- dst->f[2] = floorf( src->f[2] + 0.5f );
- dst->f[3] = floorf( src->f[3] + 0.5f );
-}
-
-static void
-micro_sgn(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src )
-{
- dst->f[0] = src->f[0] < 0.0f ? -1.0f : src->f[0] > 0.0f ? 1.0f : 0.0f;
- dst->f[1] = src->f[1] < 0.0f ? -1.0f : src->f[1] > 0.0f ? 1.0f : 0.0f;
- dst->f[2] = src->f[2] < 0.0f ? -1.0f : src->f[2] > 0.0f ? 1.0f : 0.0f;
- dst->f[3] = src->f[3] < 0.0f ? -1.0f : src->f[3] > 0.0f ? 1.0f : 0.0f;
-}
-
-static void
-micro_shl(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0,
- const union tgsi_exec_channel *src1 )
-{
- dst->i[0] = src0->i[0] << src1->i[0];
- dst->i[1] = src0->i[1] << src1->i[1];
- dst->i[2] = src0->i[2] << src1->i[2];
- dst->i[3] = src0->i[3] << src1->i[3];
-}
-
-static void
-micro_ishr(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0,
- const union tgsi_exec_channel *src1 )
-{
- dst->i[0] = src0->i[0] >> src1->i[0];
- dst->i[1] = src0->i[1] >> src1->i[1];
- dst->i[2] = src0->i[2] >> src1->i[2];
- dst->i[3] = src0->i[3] >> src1->i[3];
-}
-
-static void
-micro_trunc(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0 )
-{
- dst->f[0] = (float) (int) src0->f[0];
- dst->f[1] = (float) (int) src0->f[1];
- dst->f[2] = (float) (int) src0->f[2];
- dst->f[3] = (float) (int) src0->f[3];
-}
-
-#if 0
-static void
-micro_ushr(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0,
- const union tgsi_exec_channel *src1 )
-{
- dst->u[0] = src0->u[0] >> src1->u[0];
- dst->u[1] = src0->u[1] >> src1->u[1];
- dst->u[2] = src0->u[2] >> src1->u[2];
- dst->u[3] = src0->u[3] >> src1->u[3];
-}
-#endif
-
-static void
-micro_sin(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src )
-{
- dst->f[0] = sinf( src->f[0] );
- dst->f[1] = sinf( src->f[1] );
- dst->f[2] = sinf( src->f[2] );
- dst->f[3] = sinf( src->f[3] );
-}
-
-static void
micro_sqrt( union tgsi_exec_channel *dst,
const union tgsi_exec_channel *src )
{
@@ -1096,31 +952,6 @@ micro_sub(
dst->f[3] = src0->f[3] - src1->f[3];
}
-#if 0
-static void
-micro_u2f(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src )
-{
- dst->f[0] = (float) src->u[0];
- dst->f[1] = (float) src->u[1];
- dst->f[2] = (float) src->u[2];
- dst->f[3] = (float) src->u[3];
-}
-#endif
-
-static void
-micro_xor(
- union tgsi_exec_channel *dst,
- const union tgsi_exec_channel *src0,
- const union tgsi_exec_channel *src1 )
-{
- dst->u[0] = src0->u[0] ^ src1->u[0];
- dst->u[1] = src0->u[1] ^ src1->u[1];
- dst->u[2] = src0->u[2] ^ src1->u[2];
- dst->u[3] = src0->u[3] ^ src1->u[3];
-}
-
static void
fetch_src_file_channel(
const struct tgsi_exec_machine *mach,
@@ -1156,6 +987,7 @@ fetch_src_file_channel(
break;
case TGSI_FILE_INPUT:
+ case TGSI_FILE_SYSTEM_VALUE:
chan->u[0] = mach->Inputs[index->i[0]].xyzw[swizzle].u[0];
chan->u[1] = mach->Inputs[index->i[1]].xyzw[swizzle].u[1];
chan->u[2] = mach->Inputs[index->i[2]].xyzw[swizzle].u[2];
@@ -1193,10 +1025,10 @@ fetch_src_file_channel(
assert(index->i[1] < TGSI_EXEC_NUM_PREDS);
assert(index->i[2] < TGSI_EXEC_NUM_PREDS);
assert(index->i[3] < TGSI_EXEC_NUM_PREDS);
- chan->u[0] = mach->Addrs[0].xyzw[swizzle].u[0];
- chan->u[1] = mach->Addrs[0].xyzw[swizzle].u[1];
- chan->u[2] = mach->Addrs[0].xyzw[swizzle].u[2];
- chan->u[3] = mach->Addrs[0].xyzw[swizzle].u[3];
+ chan->u[0] = mach->Predicates[0].xyzw[swizzle].u[0];
+ chan->u[1] = mach->Predicates[0].xyzw[swizzle].u[1];
+ chan->u[2] = mach->Predicates[0].xyzw[swizzle].u[2];
+ chan->u[3] = mach->Predicates[0].xyzw[swizzle].u[3];
break;
case TGSI_FILE_OUTPUT:
@@ -1226,11 +1058,11 @@ fetch_src_file_channel(
}
static void
-fetch_source(
- const struct tgsi_exec_machine *mach,
- union tgsi_exec_channel *chan,
- const struct tgsi_full_src_register *reg,
- const uint chan_index )
+fetch_source(const struct tgsi_exec_machine *mach,
+ union tgsi_exec_channel *chan,
+ const struct tgsi_full_src_register *reg,
+ const uint chan_index,
+ enum tgsi_exec_datatype src_datatype)
{
union tgsi_exec_channel index;
uint swizzle;
@@ -1239,13 +1071,13 @@ fetch_source(
*
* file[1],
* where:
- * file = SrcRegister.File
- * [1] = SrcRegister.Index
+ * file = Register.File
+ * [1] = Register.Index
*/
index.i[0] =
index.i[1] =
index.i[2] =
- index.i[3] = reg->SrcRegister.Index;
+ index.i[3] = reg->Register.Index;
/* There is an extra source register that indirectly subscripts
* a register file. The direct index now becomes an offset
@@ -1253,11 +1085,11 @@ fetch_source(
*
* file[ind[2].x+1],
* where:
- * ind = SrcRegisterInd.File
- * [2] = SrcRegisterInd.Index
- * .x = SrcRegisterInd.SwizzleX
+ * ind = Indirect.File
+ * [2] = Indirect.Index
+ * .x = Indirect.SwizzleX
*/
- if (reg->SrcRegister.Indirect) {
+ if (reg->Register.Indirect) {
union tgsi_exec_channel index2;
union tgsi_exec_channel indir_index;
const uint execmask = mach->ExecMask;
@@ -1267,22 +1099,22 @@ fetch_source(
index2.i[0] =
index2.i[1] =
index2.i[2] =
- index2.i[3] = reg->SrcRegisterInd.Index;
+ index2.i[3] = reg->Indirect.Index;
/* get current value of address register[swizzle] */
- swizzle = tgsi_util_get_src_register_swizzle( &reg->SrcRegisterInd, CHAN_X );
+ swizzle = tgsi_util_get_src_register_swizzle( &reg->Indirect, CHAN_X );
fetch_src_file_channel(
mach,
- reg->SrcRegisterInd.File,
+ reg->Indirect.File,
swizzle,
&index2,
&indir_index );
/* add value of address register to the offset */
- index.i[0] += (int) indir_index.f[0];
- index.i[1] += (int) indir_index.f[1];
- index.i[2] += (int) indir_index.f[2];
- index.i[3] += (int) indir_index.f[3];
+ index.i[0] += indir_index.i[0];
+ index.i[1] += indir_index.i[1];
+ index.i[2] += indir_index.i[2];
+ index.i[3] += indir_index.i[3];
/* for disabled execution channels, zero-out the index to
* avoid using a potential garbage value.
@@ -1299,15 +1131,16 @@ fetch_source(
*
* file[1][3] == file[1*sizeof(file[1])+3],
* where:
- * [3] = SrcRegisterDim.Index
+ * [3] = Dimension.Index
*/
- if (reg->SrcRegister.Dimension) {
+ if (reg->Register.Dimension) {
/* The size of the first-order array depends on the register file type.
* We need to multiply the index to the first array to get an effective,
* "flat" index that points to the beginning of the second-order array.
*/
- switch (reg->SrcRegister.File) {
+ switch (reg->Register.File) {
case TGSI_FILE_INPUT:
+ case TGSI_FILE_SYSTEM_VALUE:
index.i[0] *= TGSI_EXEC_MAX_INPUT_ATTRIBS;
index.i[1] *= TGSI_EXEC_MAX_INPUT_ATTRIBS;
index.i[2] *= TGSI_EXEC_MAX_INPUT_ATTRIBS;
@@ -1323,10 +1156,10 @@ fetch_source(
assert( 0 );
}
- index.i[0] += reg->SrcRegisterDim.Index;
- index.i[1] += reg->SrcRegisterDim.Index;
- index.i[2] += reg->SrcRegisterDim.Index;
- index.i[3] += reg->SrcRegisterDim.Index;
+ index.i[0] += reg->Dimension.Index;
+ index.i[1] += reg->Dimension.Index;
+ index.i[2] += reg->Dimension.Index;
+ index.i[3] += reg->Dimension.Index;
/* Again, the second subscript index can be addressed indirectly
* identically to the first one.
@@ -1335,11 +1168,11 @@ fetch_source(
*
* file[1][ind[4].y+3],
* where:
- * ind = SrcRegisterDimInd.File
- * [4] = SrcRegisterDimInd.Index
- * .y = SrcRegisterDimInd.SwizzleX
+ * ind = DimIndirect.File
+ * [4] = DimIndirect.Index
+ * .y = DimIndirect.SwizzleX
*/
- if (reg->SrcRegisterDim.Indirect) {
+ if (reg->Dimension.Indirect) {
union tgsi_exec_channel index2;
union tgsi_exec_channel indir_index;
const uint execmask = mach->ExecMask;
@@ -1348,20 +1181,20 @@ fetch_source(
index2.i[0] =
index2.i[1] =
index2.i[2] =
- index2.i[3] = reg->SrcRegisterDimInd.Index;
+ index2.i[3] = reg->DimIndirect.Index;
- swizzle = tgsi_util_get_src_register_swizzle( &reg->SrcRegisterDimInd, CHAN_X );
+ swizzle = tgsi_util_get_src_register_swizzle( &reg->DimIndirect, CHAN_X );
fetch_src_file_channel(
mach,
- reg->SrcRegisterDimInd.File,
+ reg->DimIndirect.File,
swizzle,
&index2,
&indir_index );
- index.i[0] += (int) indir_index.f[0];
- index.i[1] += (int) indir_index.f[1];
- index.i[2] += (int) indir_index.f[2];
- index.i[3] += (int) indir_index.f[3];
+ index.i[0] += indir_index.i[0];
+ index.i[1] += indir_index.i[1];
+ index.i[2] += indir_index.i[2];
+ index.i[3] += indir_index.i[3];
/* for disabled execution channels, zero-out the index to
* avoid using a potential garbage value.
@@ -1373,7 +1206,7 @@ fetch_source(
}
/* If by any chance there was a need for a 3D array of register
- * files, we would have to check whether SrcRegisterDim is followed
+ * files, we would have to check whether Dimension is followed
* by a dimension register and continue the saga.
*/
}
@@ -1381,41 +1214,35 @@ fetch_source(
swizzle = tgsi_util_get_full_src_register_swizzle( reg, chan_index );
fetch_src_file_channel(
mach,
- reg->SrcRegister.File,
+ reg->Register.File,
swizzle,
&index,
chan );
- switch (tgsi_util_get_full_src_register_sign_mode( reg, chan_index )) {
- case TGSI_UTIL_SIGN_CLEAR:
- micro_abs( chan, chan );
- break;
-
- case TGSI_UTIL_SIGN_SET:
- micro_abs( chan, chan );
- micro_neg( chan, chan );
- break;
-
- case TGSI_UTIL_SIGN_TOGGLE:
- micro_neg( chan, chan );
- break;
-
- case TGSI_UTIL_SIGN_KEEP:
- break;
+ if (reg->Register.Absolute) {
+ if (src_datatype == TGSI_EXEC_DATA_FLOAT) {
+ micro_abs(chan, chan);
+ } else {
+ micro_iabs(chan, chan);
+ }
}
- if (reg->SrcRegisterExtMod.Complement) {
- micro_sub( chan, &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], chan );
+ if (reg->Register.Negate) {
+ if (src_datatype == TGSI_EXEC_DATA_FLOAT) {
+ micro_neg(chan, chan);
+ } else {
+ micro_ineg(chan, chan);
+ }
}
}
static void
-store_dest(
- struct tgsi_exec_machine *mach,
- const union tgsi_exec_channel *chan,
- const struct tgsi_full_dst_register *reg,
- const struct tgsi_full_instruction *inst,
- uint chan_index )
+store_dest(struct tgsi_exec_machine *mach,
+ const union tgsi_exec_channel *chan,
+ const struct tgsi_full_dst_register *reg,
+ const struct tgsi_full_instruction *inst,
+ uint chan_index,
+ enum tgsi_exec_datatype dst_datatype)
{
uint i;
union tgsi_exec_channel null;
@@ -1424,9 +1251,9 @@ store_dest(
int offset = 0; /* indirection offset */
int index;
-#ifdef DEBUG
- check_inf_or_nan(chan);
-#endif
+ if (dst_datatype == TGSI_EXEC_DATA_FLOAT) {
+ CHECK_INF_OR_NAN(chan);
+ }
/* There is an extra source register that indirectly subscripts
* a register file. The direct index now becomes an offset
@@ -1434,11 +1261,11 @@ store_dest(
*
* file[ind[2].x+1],
* where:
- * ind = DstRegisterInd.File
- * [2] = DstRegisterInd.Index
- * .x = DstRegisterInd.SwizzleX
+ * ind = Indirect.File
+ * [2] = Indirect.Index
+ * .x = Indirect.SwizzleX
*/
- if (reg->DstRegister.Indirect) {
+ if (reg->Register.Indirect) {
union tgsi_exec_channel index;
union tgsi_exec_channel indir_index;
uint swizzle;
@@ -1447,49 +1274,65 @@ store_dest(
index.i[0] =
index.i[1] =
index.i[2] =
- index.i[3] = reg->DstRegisterInd.Index;
+ index.i[3] = reg->Indirect.Index;
/* get current value of address register[swizzle] */
- swizzle = tgsi_util_get_src_register_swizzle( &reg->DstRegisterInd, CHAN_X );
+ swizzle = tgsi_util_get_src_register_swizzle( &reg->Indirect, CHAN_X );
/* fetch values from the address/indirection register */
fetch_src_file_channel(
mach,
- reg->DstRegisterInd.File,
+ reg->Indirect.File,
swizzle,
&index,
&indir_index );
/* save indirection offset */
- offset = (int) indir_index.f[0];
+ offset = indir_index.i[0];
}
- switch (reg->DstRegister.File) {
+ switch (reg->Register.File) {
case TGSI_FILE_NULL:
dst = &null;
break;
case TGSI_FILE_OUTPUT:
index = mach->Temps[TEMP_OUTPUT_I].xyzw[TEMP_OUTPUT_C].u[0]
- + reg->DstRegister.Index;
+ + reg->Register.Index;
dst = &mach->Outputs[offset + index].xyzw[chan_index];
+#if 0
+ if (TGSI_PROCESSOR_GEOMETRY == mach->Processor) {
+ fprintf(stderr, "STORING OUT[%d] mask(%d), = (", offset + index, execmask);
+ for (i = 0; i < QUAD_SIZE; i++)
+ if (execmask & (1 << i))
+ fprintf(stderr, "%f, ", chan->f[i]);
+ fprintf(stderr, ")\n");
+ }
+#endif
break;
case TGSI_FILE_TEMPORARY:
- index = reg->DstRegister.Index;
+ index = reg->Register.Index;
assert( index < TGSI_EXEC_NUM_TEMPS );
dst = &mach->Temps[offset + index].xyzw[chan_index];
break;
case TGSI_FILE_ADDRESS:
- index = reg->DstRegister.Index;
+ index = reg->Register.Index;
dst = &mach->Addrs[index].xyzw[chan_index];
break;
+ case TGSI_FILE_LOOP:
+ assert(reg->Register.Index == 0);
+ assert(mach->LoopCounterStackTop > 0);
+ assert(chan_index == CHAN_X);
+ dst = &mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[chan_index];
+ break;
+
case TGSI_FILE_PREDICATE:
- index = reg->DstRegister.Index;
+ index = reg->Register.Index;
assert(index < TGSI_EXEC_NUM_PREDS);
- dst = &mach->Addrs[index].xyzw[chan_index];
+ dst = &mach->Predicates[index].xyzw[chan_index];
break;
default:
@@ -1497,6 +1340,47 @@ store_dest(
return;
}
+ if (inst->Instruction.Predicate) {
+ uint swizzle;
+ union tgsi_exec_channel *pred;
+
+ switch (chan_index) {
+ case CHAN_X:
+ swizzle = inst->Predicate.SwizzleX;
+ break;
+ case CHAN_Y:
+ swizzle = inst->Predicate.SwizzleY;
+ break;
+ case CHAN_Z:
+ swizzle = inst->Predicate.SwizzleZ;
+ break;
+ case CHAN_W:
+ swizzle = inst->Predicate.SwizzleW;
+ break;
+ default:
+ assert(0);
+ return;
+ }
+
+ assert(inst->Predicate.Index == 0);
+
+ pred = &mach->Predicates[inst->Predicate.Index].xyzw[swizzle];
+
+ if (inst->Predicate.Negate) {
+ for (i = 0; i < QUAD_SIZE; i++) {
+ if (pred->u[i]) {
+ execmask &= ~(1 << i);
+ }
+ }
+ } else {
+ for (i = 0; i < QUAD_SIZE; i++) {
+ if (!pred->u[i]) {
+ execmask &= ~(1 << i);
+ }
+ }
+ }
+ }
+
switch (inst->Instruction.Saturate) {
case TGSI_SAT_NONE:
for (i = 0; i < QUAD_SIZE; i++)
@@ -1534,10 +1418,10 @@ store_dest(
}
#define FETCH(VAL,INDEX,CHAN)\
- fetch_source (mach, VAL, &inst->FullSrcRegisters[INDEX], CHAN)
+ fetch_source(mach, VAL, &inst->Src[INDEX], CHAN, TGSI_EXEC_DATA_FLOAT)
#define STORE(VAL,INDEX,CHAN)\
- store_dest (mach, VAL, &inst->FullDstRegisters[INDEX], inst, CHAN )
+ store_dest(mach, VAL, &inst->Dst[INDEX], inst, CHAN, TGSI_EXEC_DATA_FLOAT)
/**
@@ -1563,7 +1447,7 @@ exec_kil(struct tgsi_exec_machine *mach,
/* unswizzle channel */
swizzle = tgsi_util_get_full_src_register_swizzle (
- &inst->FullSrcRegisters[0],
+ &inst->Src[0],
chan_index);
/* check if the component has not been already tested */
@@ -1595,6 +1479,35 @@ exec_kilp(struct tgsi_exec_machine *mach,
mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0] |= kilmask;
}
+static void
+emit_vertex(struct tgsi_exec_machine *mach)
+{
+ /* FIXME: check for exec mask correctly
+ unsigned i;
+ for (i = 0; i < QUAD_SIZE; ++i) {
+ if ((mach->ExecMask & (1 << i)))
+ */
+ if (mach->ExecMask) {
+ mach->Temps[TEMP_OUTPUT_I].xyzw[TEMP_OUTPUT_C].u[0] += mach->NumOutputs;
+ mach->Primitives[mach->Temps[TEMP_PRIMITIVE_I].xyzw[TEMP_PRIMITIVE_C].u[0]]++;
+ }
+}
+
+static void
+emit_primitive(struct tgsi_exec_machine *mach)
+{
+ unsigned *prim_count = &mach->Temps[TEMP_PRIMITIVE_I].xyzw[TEMP_PRIMITIVE_C].u[0];
+ /* FIXME: check for exec mask correctly
+ unsigned i;
+ for (i = 0; i < QUAD_SIZE; ++i) {
+ if ((mach->ExecMask & (1 << i)))
+ */
+ if (mach->ExecMask) {
+ ++(*prim_count);
+ debug_assert((*prim_count * mach->NumOutputs) < mach->MaxGeometryShaderOutputs);
+ mach->Primitives[*prim_count] = 0;
+ }
+}
/*
* Fetch a four texture samples using STR texture coordinates.
@@ -1604,7 +1517,8 @@ fetch_texel( struct tgsi_sampler *sampler,
const union tgsi_exec_channel *s,
const union tgsi_exec_channel *t,
const union tgsi_exec_channel *p,
- float lodbias, /* XXX should be float[4] */
+ const union tgsi_exec_channel *c0,
+ enum tgsi_sampler_control control,
union tgsi_exec_channel *r,
union tgsi_exec_channel *g,
union tgsi_exec_channel *b,
@@ -1613,7 +1527,7 @@ fetch_texel( struct tgsi_sampler *sampler,
uint j;
float rgba[NUM_CHANNELS][QUAD_SIZE];
- sampler->get_samples(sampler, s->f, t->f, p->f, lodbias, rgba);
+ sampler->get_samples(sampler, s->f, t->f, p->f, c0->f, control, rgba);
for (j = 0; j < 4; j++) {
r->f[j] = rgba[0][j];
@@ -1624,102 +1538,156 @@ fetch_texel( struct tgsi_sampler *sampler,
}
+#define TEX_MODIFIER_NONE 0
+#define TEX_MODIFIER_PROJECTED 1
+#define TEX_MODIFIER_LOD_BIAS 2
+#define TEX_MODIFIER_EXPLICIT_LOD 3
+
+
static void
exec_tex(struct tgsi_exec_machine *mach,
const struct tgsi_full_instruction *inst,
- boolean biasLod,
- boolean projected)
+ uint modifier)
{
- const uint unit = inst->FullSrcRegisters[1].SrcRegister.Index;
+ const uint unit = inst->Src[1].Register.Index;
union tgsi_exec_channel r[4];
+ const union tgsi_exec_channel *lod = &ZeroVec;
+ enum tgsi_sampler_control control;
uint chan_index;
- float lodBias;
- /* debug_printf("Sampler %u unit %u\n", sampler, unit); */
+ if (modifier != TEX_MODIFIER_NONE) {
+ FETCH(&r[3], 0, CHAN_W);
+ if (modifier != TEX_MODIFIER_PROJECTED) {
+ lod = &r[3];
+ }
+ }
+
+ if (modifier == TEX_MODIFIER_EXPLICIT_LOD) {
+ control = tgsi_sampler_lod_explicit;
+ } else {
+ control = tgsi_sampler_lod_bias;
+ }
- switch (inst->InstructionExtTexture.Texture) {
+ switch (inst->Texture.Texture) {
case TGSI_TEXTURE_1D:
case TGSI_TEXTURE_SHADOW1D:
-
FETCH(&r[0], 0, CHAN_X);
- if (projected) {
- FETCH(&r[1], 0, CHAN_W);
- micro_div( &r[0], &r[0], &r[1] );
- }
-
- if (biasLod) {
- FETCH(&r[1], 0, CHAN_W);
- lodBias = r[2].f[0];
+ if (modifier == TEX_MODIFIER_PROJECTED) {
+ micro_div(&r[0], &r[0], &r[3]);
}
- else
- lodBias = 0.0;
fetch_texel(mach->Samplers[unit],
- &r[0], &ZeroVec, &ZeroVec, lodBias, /* S, T, P, BIAS */
- &r[0], &r[1], &r[2], &r[3]); /* R, G, B, A */
+ &r[0], &ZeroVec, &ZeroVec, lod, /* S, T, P, LOD */
+ control,
+ &r[0], &r[1], &r[2], &r[3]); /* R, G, B, A */
break;
case TGSI_TEXTURE_2D:
case TGSI_TEXTURE_RECT:
case TGSI_TEXTURE_SHADOW2D:
case TGSI_TEXTURE_SHADOWRECT:
-
FETCH(&r[0], 0, CHAN_X);
FETCH(&r[1], 0, CHAN_Y);
FETCH(&r[2], 0, CHAN_Z);
- if (projected) {
- FETCH(&r[3], 0, CHAN_W);
- micro_div( &r[0], &r[0], &r[3] );
- micro_div( &r[1], &r[1], &r[3] );
- micro_div( &r[2], &r[2], &r[3] );
- }
-
- if (biasLod) {
- FETCH(&r[3], 0, CHAN_W);
- lodBias = r[3].f[0];
+ if (modifier == TEX_MODIFIER_PROJECTED) {
+ micro_div(&r[0], &r[0], &r[3]);
+ micro_div(&r[1], &r[1], &r[3]);
+ micro_div(&r[2], &r[2], &r[3]);
}
- else
- lodBias = 0.0;
fetch_texel(mach->Samplers[unit],
- &r[0], &r[1], &r[2], lodBias, /* inputs */
+ &r[0], &r[1], &r[2], lod, /* S, T, P, LOD */
+ control,
&r[0], &r[1], &r[2], &r[3]); /* outputs */
break;
case TGSI_TEXTURE_3D:
case TGSI_TEXTURE_CUBE:
-
FETCH(&r[0], 0, CHAN_X);
FETCH(&r[1], 0, CHAN_Y);
FETCH(&r[2], 0, CHAN_Z);
- if (projected) {
- FETCH(&r[3], 0, CHAN_W);
- micro_div( &r[0], &r[0], &r[3] );
- micro_div( &r[1], &r[1], &r[3] );
- micro_div( &r[2], &r[2], &r[3] );
+ if (modifier == TEX_MODIFIER_PROJECTED) {
+ micro_div(&r[0], &r[0], &r[3]);
+ micro_div(&r[1], &r[1], &r[3]);
+ micro_div(&r[2], &r[2], &r[3]);
}
- if (biasLod) {
- FETCH(&r[3], 0, CHAN_W);
- lodBias = r[3].f[0];
- }
- else
- lodBias = 0.0;
+ fetch_texel(mach->Samplers[unit],
+ &r[0], &r[1], &r[2], lod,
+ control,
+ &r[0], &r[1], &r[2], &r[3]);
+ break;
+
+ default:
+ assert(0);
+ }
+
+ FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+ STORE(&r[chan_index], 0, chan_index);
+ }
+}
+
+static void
+exec_txd(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ const uint unit = inst->Src[3].Register.Index;
+ union tgsi_exec_channel r[4];
+ uint chan_index;
+
+ /*
+ * XXX: This is fake TXD -- the derivatives are not taken into account, yet.
+ */
+
+ switch (inst->Texture.Texture) {
+ case TGSI_TEXTURE_1D:
+ case TGSI_TEXTURE_SHADOW1D:
+
+ FETCH(&r[0], 0, CHAN_X);
+
+ fetch_texel(mach->Samplers[unit],
+ &r[0], &ZeroVec, &ZeroVec, &ZeroVec, /* S, T, P, BIAS */
+ tgsi_sampler_lod_bias,
+ &r[0], &r[1], &r[2], &r[3]); /* R, G, B, A */
+ break;
+
+ case TGSI_TEXTURE_2D:
+ case TGSI_TEXTURE_RECT:
+ case TGSI_TEXTURE_SHADOW2D:
+ case TGSI_TEXTURE_SHADOWRECT:
+
+ FETCH(&r[0], 0, CHAN_X);
+ FETCH(&r[1], 0, CHAN_Y);
+ FETCH(&r[2], 0, CHAN_Z);
+
+ fetch_texel(mach->Samplers[unit],
+ &r[0], &r[1], &r[2], &ZeroVec, /* inputs */
+ tgsi_sampler_lod_bias,
+ &r[0], &r[1], &r[2], &r[3]); /* outputs */
+ break;
+
+ case TGSI_TEXTURE_3D:
+ case TGSI_TEXTURE_CUBE:
+
+ FETCH(&r[0], 0, CHAN_X);
+ FETCH(&r[1], 0, CHAN_Y);
+ FETCH(&r[2], 0, CHAN_Z);
fetch_texel(mach->Samplers[unit],
- &r[0], &r[1], &r[2], lodBias,
+ &r[0], &r[1], &r[2], &ZeroVec,
+ tgsi_sampler_lod_bias,
&r[0], &r[1], &r[2], &r[3]);
break;
default:
- assert (0);
+ assert(0);
}
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- STORE( &r[chan_index], 0, chan_index );
+ FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+ STORE(&r[chan_index], 0, chan_index);
}
}
@@ -1792,53 +1760,59 @@ typedef void (* eval_coef_func)(
unsigned chan );
static void
-exec_declaration(
- struct tgsi_exec_machine *mach,
- const struct tgsi_full_declaration *decl )
+exec_declaration(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_declaration *decl)
{
- if( mach->Processor == TGSI_PROCESSOR_FRAGMENT ) {
- if( decl->Declaration.File == TGSI_FILE_INPUT ) {
- unsigned first, last, mask;
- eval_coef_func eval;
+ if (mach->Processor == TGSI_PROCESSOR_FRAGMENT) {
+ if (decl->Declaration.File == TGSI_FILE_INPUT ||
+ decl->Declaration.File == TGSI_FILE_SYSTEM_VALUE) {
+ uint first, last, mask;
- first = decl->DeclarationRange.First;
- last = decl->DeclarationRange.Last;
+ first = decl->Range.First;
+ last = decl->Range.Last;
mask = decl->Declaration.UsageMask;
- switch( decl->Declaration.Interpolate ) {
- case TGSI_INTERPOLATE_CONSTANT:
- eval = eval_constant_coef;
- break;
-
- case TGSI_INTERPOLATE_LINEAR:
- eval = eval_linear_coef;
- break;
-
- case TGSI_INTERPOLATE_PERSPECTIVE:
- eval = eval_perspective_coef;
- break;
+ if (decl->Semantic.Name == TGSI_SEMANTIC_POSITION) {
+ assert(decl->Semantic.Index == 0);
+ assert(first == last);
+ assert(mask == TGSI_WRITEMASK_XYZW);
- default:
- assert( 0 );
- return;
- }
+ mach->Inputs[first] = mach->QuadPos;
+ } else if (decl->Semantic.Name == TGSI_SEMANTIC_FACE) {
+ uint i;
- if( mask == TGSI_WRITEMASK_XYZW ) {
- unsigned i, j;
+ assert(decl->Semantic.Index == 0);
+ assert(first == last);
- for( i = first; i <= last; i++ ) {
- for( j = 0; j < NUM_CHANNELS; j++ ) {
- eval( mach, i, j );
- }
+ for (i = 0; i < QUAD_SIZE; i++) {
+ mach->Inputs[first].xyzw[0].f[i] = mach->Face;
+ }
+ } else {
+ eval_coef_func eval;
+ uint i, j;
+
+ switch (decl->Declaration.Interpolate) {
+ case TGSI_INTERPOLATE_CONSTANT:
+ eval = eval_constant_coef;
+ break;
+
+ case TGSI_INTERPOLATE_LINEAR:
+ eval = eval_linear_coef;
+ break;
+
+ case TGSI_INTERPOLATE_PERSPECTIVE:
+ eval = eval_perspective_coef;
+ break;
+
+ default:
+ assert(0);
+ return;
}
- }
- else {
- unsigned i, j;
- for( j = 0; j < NUM_CHANNELS; j++ ) {
- if( mask & (1 << j) ) {
- for( i = first; i <= last; i++ ) {
- eval( mach, i, j );
+ for (j = 0; j < NUM_CHANNELS; j++) {
+ if (mask & (1 << j)) {
+ for (i = first; i <= last; i++) {
+ eval(mach, i, j);
}
}
}
@@ -1847,6 +1821,585 @@ exec_declaration(
}
}
+typedef void (* micro_op)(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src);
+
+static void
+exec_scalar_unary(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst,
+ micro_op op,
+ enum tgsi_exec_datatype dst_datatype,
+ enum tgsi_exec_datatype src_datatype)
+{
+ unsigned int chan;
+ union tgsi_exec_channel src;
+ union tgsi_exec_channel dst;
+
+ fetch_source(mach, &src, &inst->Src[0], CHAN_X, src_datatype);
+ op(&dst, &src);
+ for (chan = 0; chan < NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ store_dest(mach, &dst, &inst->Dst[0], inst, chan, dst_datatype);
+ }
+ }
+}
+
+static void
+exec_vector_unary(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst,
+ micro_op op,
+ enum tgsi_exec_datatype dst_datatype,
+ enum tgsi_exec_datatype src_datatype)
+{
+ unsigned int chan;
+ struct tgsi_exec_vector dst;
+
+ for (chan = 0; chan < NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ union tgsi_exec_channel src;
+
+ fetch_source(mach, &src, &inst->Src[0], chan, src_datatype);
+ op(&dst.xyzw[chan], &src);
+ }
+ }
+ for (chan = 0; chan < NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ store_dest(mach, &dst.xyzw[chan], &inst->Dst[0], inst, chan, dst_datatype);
+ }
+ }
+}
+
+static void
+exec_vector_binary(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst,
+ micro_op op,
+ enum tgsi_exec_datatype dst_datatype,
+ enum tgsi_exec_datatype src_datatype)
+{
+ unsigned int chan;
+ struct tgsi_exec_vector dst;
+
+ for (chan = 0; chan < NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ union tgsi_exec_channel src[2];
+
+ fetch_source(mach, &src[0], &inst->Src[0], chan, src_datatype);
+ fetch_source(mach, &src[1], &inst->Src[1], chan, src_datatype);
+ op(&dst.xyzw[chan], src);
+ }
+ }
+ for (chan = 0; chan < NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ store_dest(mach, &dst.xyzw[chan], &inst->Dst[0], inst, chan, dst_datatype);
+ }
+ }
+}
+
+static void
+exec_vector_trinary(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst,
+ micro_op op,
+ enum tgsi_exec_datatype dst_datatype,
+ enum tgsi_exec_datatype src_datatype)
+{
+ unsigned int chan;
+ struct tgsi_exec_vector dst;
+
+ for (chan = 0; chan < NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ union tgsi_exec_channel src[3];
+
+ fetch_source(mach, &src[0], &inst->Src[0], chan, src_datatype);
+ fetch_source(mach, &src[1], &inst->Src[1], chan, src_datatype);
+ fetch_source(mach, &src[2], &inst->Src[2], chan, src_datatype);
+ op(&dst.xyzw[chan], src);
+ }
+ }
+ for (chan = 0; chan < NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ store_dest(mach, &dst.xyzw[chan], &inst->Dst[0], inst, chan, dst_datatype);
+ }
+ }
+}
+
+static void
+exec_dp3(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ unsigned int chan;
+ union tgsi_exec_channel arg[3];
+
+ fetch_source(mach, &arg[0], &inst->Src[0], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ micro_mul(&arg[2], &arg[0], &arg[1]);
+
+ for (chan = CHAN_Y; chan <= CHAN_Z; chan++) {
+ fetch_source(mach, &arg[0], &inst->Src[0], chan, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], chan, TGSI_EXEC_DATA_FLOAT);
+ micro_mad(&arg[2], arg);
+ }
+
+ for (chan = 0; chan < NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ store_dest(mach, &arg[2], &inst->Dst[0], inst, chan, TGSI_EXEC_DATA_FLOAT);
+ }
+ }
+}
+
+static void
+exec_dp4(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ unsigned int chan;
+ union tgsi_exec_channel arg[3];
+
+ fetch_source(mach, &arg[0], &inst->Src[0], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ micro_mul(&arg[2], &arg[0], &arg[1]);
+
+ for (chan = CHAN_Y; chan <= CHAN_W; chan++) {
+ fetch_source(mach, &arg[0], &inst->Src[0], chan, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], chan, TGSI_EXEC_DATA_FLOAT);
+ micro_mad(&arg[2], arg);
+ }
+
+ for (chan = 0; chan < NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ store_dest(mach, &arg[2], &inst->Dst[0], inst, chan, TGSI_EXEC_DATA_FLOAT);
+ }
+ }
+}
+
+static void
+exec_dp2a(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ unsigned int chan;
+ union tgsi_exec_channel arg[3];
+
+ fetch_source(mach, &arg[0], &inst->Src[0], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ micro_mul(&arg[2], &arg[0], &arg[1]);
+
+ fetch_source(mach, &arg[0], &inst->Src[0], CHAN_Y, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_Y, TGSI_EXEC_DATA_FLOAT);
+ micro_mad(&arg[0], arg);
+
+ fetch_source(mach, &arg[1], &inst->Src[2], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ micro_add(&arg[0], &arg[0], &arg[1]);
+
+ for (chan = 0; chan < NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ store_dest(mach, &arg[0], &inst->Dst[0], inst, chan, TGSI_EXEC_DATA_FLOAT);
+ }
+ }
+}
+
+static void
+exec_dph(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ unsigned int chan;
+ union tgsi_exec_channel arg[3];
+
+ fetch_source(mach, &arg[0], &inst->Src[0], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ micro_mul(&arg[2], &arg[0], &arg[1]);
+
+ fetch_source(mach, &arg[0], &inst->Src[0], CHAN_Y, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_Y, TGSI_EXEC_DATA_FLOAT);
+ micro_mad(&arg[2], arg);
+
+ fetch_source(mach, &arg[0], &inst->Src[0], CHAN_Z, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_Z, TGSI_EXEC_DATA_FLOAT);
+ micro_mad(&arg[0], arg);
+
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_W, TGSI_EXEC_DATA_FLOAT);
+ micro_add(&arg[0], &arg[0], &arg[1]);
+
+ for (chan = 0; chan < NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ store_dest(mach, &arg[0], &inst->Dst[0], inst, chan, TGSI_EXEC_DATA_FLOAT);
+ }
+ }
+}
+
+static void
+exec_dp2(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ unsigned int chan;
+ union tgsi_exec_channel arg[3];
+
+ fetch_source(mach, &arg[0], &inst->Src[0], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_X, TGSI_EXEC_DATA_FLOAT);
+ micro_mul(&arg[2], &arg[0], &arg[1]);
+
+ fetch_source(mach, &arg[0], &inst->Src[0], CHAN_Y, TGSI_EXEC_DATA_FLOAT);
+ fetch_source(mach, &arg[1], &inst->Src[1], CHAN_Y, TGSI_EXEC_DATA_FLOAT);
+ micro_mad(&arg[2], arg);
+
+ for (chan = 0; chan < NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ store_dest(mach, &arg[2], &inst->Dst[0], inst, chan, TGSI_EXEC_DATA_FLOAT);
+ }
+ }
+}
+
+static void
+exec_break(struct tgsi_exec_machine *mach)
+{
+ if (mach->BreakType == TGSI_EXEC_BREAK_INSIDE_LOOP) {
+ /* turn off loop channels for each enabled exec channel */
+ mach->LoopMask &= ~mach->ExecMask;
+ /* Todo: if mach->LoopMask == 0, jump to end of loop */
+ UPDATE_EXEC_MASK(mach);
+ } else {
+ assert(mach->BreakType == TGSI_EXEC_BREAK_INSIDE_SWITCH);
+
+ mach->Switch.mask = 0x0;
+
+ UPDATE_EXEC_MASK(mach);
+ }
+}
+
+static void
+exec_switch(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ assert(mach->SwitchStackTop < TGSI_EXEC_MAX_SWITCH_NESTING);
+ assert(mach->BreakStackTop < TGSI_EXEC_MAX_BREAK_STACK);
+
+ mach->SwitchStack[mach->SwitchStackTop++] = mach->Switch;
+ fetch_source(mach, &mach->Switch.selector, &inst->Src[0], CHAN_X, TGSI_EXEC_DATA_UINT);
+ mach->Switch.mask = 0x0;
+ mach->Switch.defaultMask = 0x0;
+
+ mach->BreakStack[mach->BreakStackTop++] = mach->BreakType;
+ mach->BreakType = TGSI_EXEC_BREAK_INSIDE_SWITCH;
+
+ UPDATE_EXEC_MASK(mach);
+}
+
+static void
+exec_case(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ uint prevMask = mach->SwitchStack[mach->SwitchStackTop - 1].mask;
+ union tgsi_exec_channel src;
+ uint mask = 0;
+
+ fetch_source(mach, &src, &inst->Src[0], CHAN_X, TGSI_EXEC_DATA_UINT);
+
+ if (mach->Switch.selector.u[0] == src.u[0]) {
+ mask |= 0x1;
+ }
+ if (mach->Switch.selector.u[1] == src.u[1]) {
+ mask |= 0x2;
+ }
+ if (mach->Switch.selector.u[2] == src.u[2]) {
+ mask |= 0x4;
+ }
+ if (mach->Switch.selector.u[3] == src.u[3]) {
+ mask |= 0x8;
+ }
+
+ mach->Switch.defaultMask |= mask;
+
+ mach->Switch.mask |= mask & prevMask;
+
+ UPDATE_EXEC_MASK(mach);
+}
+
+static void
+exec_default(struct tgsi_exec_machine *mach)
+{
+ uint prevMask = mach->SwitchStack[mach->SwitchStackTop - 1].mask;
+
+ mach->Switch.mask |= ~mach->Switch.defaultMask & prevMask;
+
+ UPDATE_EXEC_MASK(mach);
+}
+
+static void
+exec_endswitch(struct tgsi_exec_machine *mach)
+{
+ mach->Switch = mach->SwitchStack[--mach->SwitchStackTop];
+ mach->BreakType = mach->BreakStack[--mach->BreakStackTop];
+
+ UPDATE_EXEC_MASK(mach);
+}
+
+static void
+micro_i2f(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = (float)src->i[0];
+ dst->f[1] = (float)src->i[1];
+ dst->f[2] = (float)src->i[2];
+ dst->f[3] = (float)src->i[3];
+}
+
+static void
+micro_not(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->u[0] = ~src->u[0];
+ dst->u[1] = ~src->u[1];
+ dst->u[2] = ~src->u[2];
+ dst->u[3] = ~src->u[3];
+}
+
+static void
+micro_shl(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->u[0] = src[0].u[0] << src[1].u[0];
+ dst->u[1] = src[0].u[1] << src[1].u[1];
+ dst->u[2] = src[0].u[2] << src[1].u[2];
+ dst->u[3] = src[0].u[3] << src[1].u[3];
+}
+
+static void
+micro_and(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->u[0] = src[0].u[0] & src[1].u[0];
+ dst->u[1] = src[0].u[1] & src[1].u[1];
+ dst->u[2] = src[0].u[2] & src[1].u[2];
+ dst->u[3] = src[0].u[3] & src[1].u[3];
+}
+
+static void
+micro_or(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->u[0] = src[0].u[0] | src[1].u[0];
+ dst->u[1] = src[0].u[1] | src[1].u[1];
+ dst->u[2] = src[0].u[2] | src[1].u[2];
+ dst->u[3] = src[0].u[3] | src[1].u[3];
+}
+
+static void
+micro_xor(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->u[0] = src[0].u[0] ^ src[1].u[0];
+ dst->u[1] = src[0].u[1] ^ src[1].u[1];
+ dst->u[2] = src[0].u[2] ^ src[1].u[2];
+ dst->u[3] = src[0].u[3] ^ src[1].u[3];
+}
+
+static void
+micro_f2i(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->i[0] = (int)src->f[0];
+ dst->i[1] = (int)src->f[1];
+ dst->i[2] = (int)src->f[2];
+ dst->i[3] = (int)src->f[3];
+}
+
+static void
+micro_idiv(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->i[0] = src[0].i[0] / src[1].i[0];
+ dst->i[1] = src[0].i[1] / src[1].i[1];
+ dst->i[2] = src[0].i[2] / src[1].i[2];
+ dst->i[3] = src[0].i[3] / src[1].i[3];
+}
+
+static void
+micro_imax(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->i[0] = src[0].i[0] > src[1].i[0] ? src[0].i[0] : src[1].i[0];
+ dst->i[1] = src[0].i[1] > src[1].i[1] ? src[0].i[1] : src[1].i[1];
+ dst->i[2] = src[0].i[2] > src[1].i[2] ? src[0].i[2] : src[1].i[2];
+ dst->i[3] = src[0].i[3] > src[1].i[3] ? src[0].i[3] : src[1].i[3];
+}
+
+static void
+micro_imin(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->i[0] = src[0].i[0] < src[1].i[0] ? src[0].i[0] : src[1].i[0];
+ dst->i[1] = src[0].i[1] < src[1].i[1] ? src[0].i[1] : src[1].i[1];
+ dst->i[2] = src[0].i[2] < src[1].i[2] ? src[0].i[2] : src[1].i[2];
+ dst->i[3] = src[0].i[3] < src[1].i[3] ? src[0].i[3] : src[1].i[3];
+}
+
+static void
+micro_isge(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->i[0] = src[0].i[0] >= src[1].i[0] ? -1 : 0;
+ dst->i[1] = src[0].i[1] >= src[1].i[1] ? -1 : 0;
+ dst->i[2] = src[0].i[2] >= src[1].i[2] ? -1 : 0;
+ dst->i[3] = src[0].i[3] >= src[1].i[3] ? -1 : 0;
+}
+
+static void
+micro_ishr(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->i[0] = src[0].i[0] >> src[1].i[0];
+ dst->i[1] = src[0].i[1] >> src[1].i[1];
+ dst->i[2] = src[0].i[2] >> src[1].i[2];
+ dst->i[3] = src[0].i[3] >> src[1].i[3];
+}
+
+static void
+micro_islt(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->i[0] = src[0].i[0] < src[1].i[0] ? -1 : 0;
+ dst->i[1] = src[0].i[1] < src[1].i[1] ? -1 : 0;
+ dst->i[2] = src[0].i[2] < src[1].i[2] ? -1 : 0;
+ dst->i[3] = src[0].i[3] < src[1].i[3] ? -1 : 0;
+}
+
+static void
+micro_f2u(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->u[0] = (uint)src->f[0];
+ dst->u[1] = (uint)src->f[1];
+ dst->u[2] = (uint)src->f[2];
+ dst->u[3] = (uint)src->f[3];
+}
+
+static void
+micro_u2f(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->f[0] = (float)src->u[0];
+ dst->f[1] = (float)src->u[1];
+ dst->f[2] = (float)src->u[2];
+ dst->f[3] = (float)src->u[3];
+}
+
+static void
+micro_uadd(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->u[0] = src[0].u[0] + src[1].u[0];
+ dst->u[1] = src[0].u[1] + src[1].u[1];
+ dst->u[2] = src[0].u[2] + src[1].u[2];
+ dst->u[3] = src[0].u[3] + src[1].u[3];
+}
+
+static void
+micro_udiv(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->u[0] = src[0].u[0] / src[1].u[0];
+ dst->u[1] = src[0].u[1] / src[1].u[1];
+ dst->u[2] = src[0].u[2] / src[1].u[2];
+ dst->u[3] = src[0].u[3] / src[1].u[3];
+}
+
+static void
+micro_umad(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->u[0] = src[0].u[0] * src[1].u[0] + src[2].u[0];
+ dst->u[1] = src[0].u[1] * src[1].u[1] + src[2].u[1];
+ dst->u[2] = src[0].u[2] * src[1].u[2] + src[2].u[2];
+ dst->u[3] = src[0].u[3] * src[1].u[3] + src[2].u[3];
+}
+
+static void
+micro_umax(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->u[0] = src[0].u[0] > src[1].u[0] ? src[0].u[0] : src[1].u[0];
+ dst->u[1] = src[0].u[1] > src[1].u[1] ? src[0].u[1] : src[1].u[1];
+ dst->u[2] = src[0].u[2] > src[1].u[2] ? src[0].u[2] : src[1].u[2];
+ dst->u[3] = src[0].u[3] > src[1].u[3] ? src[0].u[3] : src[1].u[3];
+}
+
+static void
+micro_umin(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->u[0] = src[0].u[0] < src[1].u[0] ? src[0].u[0] : src[1].u[0];
+ dst->u[1] = src[0].u[1] < src[1].u[1] ? src[0].u[1] : src[1].u[1];
+ dst->u[2] = src[0].u[2] < src[1].u[2] ? src[0].u[2] : src[1].u[2];
+ dst->u[3] = src[0].u[3] < src[1].u[3] ? src[0].u[3] : src[1].u[3];
+}
+
+static void
+micro_umod(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->u[0] = src[0].u[0] % src[1].u[0];
+ dst->u[1] = src[0].u[1] % src[1].u[1];
+ dst->u[2] = src[0].u[2] % src[1].u[2];
+ dst->u[3] = src[0].u[3] % src[1].u[3];
+}
+
+static void
+micro_umul(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->u[0] = src[0].u[0] * src[1].u[0];
+ dst->u[1] = src[0].u[1] * src[1].u[1];
+ dst->u[2] = src[0].u[2] * src[1].u[2];
+ dst->u[3] = src[0].u[3] * src[1].u[3];
+}
+
+static void
+micro_useq(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->u[0] = src[0].u[0] == src[1].u[0] ? ~0 : 0;
+ dst->u[1] = src[0].u[1] == src[1].u[1] ? ~0 : 0;
+ dst->u[2] = src[0].u[2] == src[1].u[2] ? ~0 : 0;
+ dst->u[3] = src[0].u[3] == src[1].u[3] ? ~0 : 0;
+}
+
+static void
+micro_usge(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->u[0] = src[0].u[0] >= src[1].u[0] ? ~0 : 0;
+ dst->u[1] = src[0].u[1] >= src[1].u[1] ? ~0 : 0;
+ dst->u[2] = src[0].u[2] >= src[1].u[2] ? ~0 : 0;
+ dst->u[3] = src[0].u[3] >= src[1].u[3] ? ~0 : 0;
+}
+
+static void
+micro_ushr(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->u[0] = src[0].u[0] >> src[1].u[0];
+ dst->u[1] = src[0].u[1] >> src[1].u[1];
+ dst->u[2] = src[0].u[2] >> src[1].u[2];
+ dst->u[3] = src[0].u[3] >> src[1].u[3];
+}
+
+static void
+micro_uslt(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->u[0] = src[0].u[0] < src[1].u[0] ? ~0 : 0;
+ dst->u[1] = src[0].u[1] < src[1].u[1] ? ~0 : 0;
+ dst->u[2] = src[0].u[2] < src[1].u[2] ? ~0 : 0;
+ dst->u[3] = src[0].u[3] < src[1].u[3] ? ~0 : 0;
+}
+
+static void
+micro_usne(union tgsi_exec_channel *dst,
+ const union tgsi_exec_channel *src)
+{
+ dst->u[0] = src[0].u[0] != src[1].u[0] ? ~0 : 0;
+ dst->u[1] = src[0].u[1] != src[1].u[1] ? ~0 : 0;
+ dst->u[2] = src[0].u[2] != src[1].u[2] ? ~0 : 0;
+ dst->u[3] = src[0].u[3] != src[1].u[3] ? ~0 : 0;
+}
+
static void
exec_instruction(
struct tgsi_exec_machine *mach,
@@ -1855,50 +2408,24 @@ exec_instruction(
{
uint chan_index;
union tgsi_exec_channel r[10];
+ union tgsi_exec_channel d[8];
(*pc)++;
switch (inst->Instruction.Opcode) {
case TGSI_OPCODE_ARL:
- case TGSI_OPCODE_FLR:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- micro_flr( &r[0], &r[0] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_unary(mach, inst, micro_arl, TGSI_EXEC_DATA_INT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_MOV:
- if (inst->Flags & SOA_DEPENDENCY_FLAG) {
- /* Do all fetches into temp regs, then do all stores to avoid
- * intermediate/accidental clobbering. This could be done all the
- * time for MOV but for other instructions we'll need more temps...
- */
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[chan_index], 0, chan_index );
- }
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- STORE( &r[chan_index], 0, chan_index );
- }
- }
- else {
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- STORE( &r[0], 0, chan_index );
- }
- }
+ exec_vector_unary(mach, inst, micro_mov, TGSI_EXEC_DATA_UINT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_LIT:
- if (IS_CHANNEL_ENABLED( *inst, CHAN_X )) {
- STORE( &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], 0, CHAN_X );
- }
-
if (IS_CHANNEL_ENABLED( *inst, CHAN_Y ) || IS_CHANNEL_ENABLED( *inst, CHAN_Z )) {
FETCH( &r[0], 0, CHAN_X );
if (IS_CHANNEL_ENABLED( *inst, CHAN_Y )) {
- micro_max( &r[0], &r[0], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] );
- STORE( &r[0], 0, CHAN_Y );
+ micro_max(&d[CHAN_Y], &r[0], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C]);
}
if (IS_CHANNEL_ENABLED( *inst, CHAN_Z )) {
@@ -1909,34 +2436,30 @@ exec_instruction(
micro_min( &r[2], &r[2], &mach->Temps[TEMP_128_I].xyzw[TEMP_128_C] );
micro_max( &r[2], &r[2], &mach->Temps[TEMP_M128_I].xyzw[TEMP_M128_C] );
micro_pow( &r[1], &r[1], &r[2] );
- micro_lt( &r[0], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] );
- STORE( &r[0], 0, CHAN_Z );
+ micro_lt(&d[CHAN_Z], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C]);
}
- }
+ if (IS_CHANNEL_ENABLED(*inst, CHAN_Y)) {
+ STORE(&d[CHAN_Y], 0, CHAN_Y);
+ }
+ if (IS_CHANNEL_ENABLED(*inst, CHAN_Z)) {
+ STORE(&d[CHAN_Z], 0, CHAN_Z);
+ }
+ }
+ if (IS_CHANNEL_ENABLED( *inst, CHAN_X )) {
+ STORE( &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], 0, CHAN_X );
+ }
if (IS_CHANNEL_ENABLED( *inst, CHAN_W )) {
STORE( &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], 0, CHAN_W );
}
break;
case TGSI_OPCODE_RCP:
- /* TGSI_OPCODE_RECIP */
- FETCH( &r[0], 0, CHAN_X );
- micro_div( &r[0], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &r[0] );
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- STORE( &r[0], 0, chan_index );
- }
+ exec_scalar_unary(mach, inst, micro_rcp, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_RSQ:
- /* TGSI_OPCODE_RECIPSQRT */
- FETCH( &r[0], 0, CHAN_X );
- micro_abs( &r[0], &r[0] );
- micro_sqrt( &r[0], &r[0] );
- micro_div( &r[0], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &r[0] );
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- STORE( &r[0], 0, chan_index );
- }
+ exec_scalar_unary(mach, inst, micro_rsq, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_EXP:
@@ -1981,14 +2504,13 @@ exec_instruction(
break;
case TGSI_OPCODE_MUL:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index )
- {
+ FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
FETCH(&r[0], 0, chan_index);
FETCH(&r[1], 1, chan_index);
-
- micro_mul( &r[0], &r[0], &r[1] );
-
- STORE(&r[0], 0, chan_index);
+ micro_mul(&d[chan_index], &r[0], &r[1]);
+ }
+ FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+ STORE(&d[chan_index], 0, chan_index);
}
break;
@@ -1996,82 +2518,45 @@ exec_instruction(
FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
FETCH( &r[0], 0, chan_index );
FETCH( &r[1], 1, chan_index );
- micro_add( &r[0], &r[0], &r[1] );
- STORE( &r[0], 0, chan_index );
+ micro_add(&d[chan_index], &r[0], &r[1]);
+ }
+ FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+ STORE(&d[chan_index], 0, chan_index);
}
break;
case TGSI_OPCODE_DP3:
- /* TGSI_OPCODE_DOT3 */
- FETCH( &r[0], 0, CHAN_X );
- FETCH( &r[1], 1, CHAN_X );
- micro_mul( &r[0], &r[0], &r[1] );
-
- FETCH( &r[1], 0, CHAN_Y );
- FETCH( &r[2], 1, CHAN_Y );
- micro_mul( &r[1], &r[1], &r[2] );
- micro_add( &r[0], &r[0], &r[1] );
-
- FETCH( &r[1], 0, CHAN_Z );
- FETCH( &r[2], 1, CHAN_Z );
- micro_mul( &r[1], &r[1], &r[2] );
- micro_add( &r[0], &r[0], &r[1] );
-
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- STORE( &r[0], 0, chan_index );
- }
+ exec_dp3(mach, inst);
break;
- case TGSI_OPCODE_DP4:
- /* TGSI_OPCODE_DOT4 */
- FETCH(&r[0], 0, CHAN_X);
- FETCH(&r[1], 1, CHAN_X);
-
- micro_mul( &r[0], &r[0], &r[1] );
-
- FETCH(&r[1], 0, CHAN_Y);
- FETCH(&r[2], 1, CHAN_Y);
-
- micro_mul( &r[1], &r[1], &r[2] );
- micro_add( &r[0], &r[0], &r[1] );
-
- FETCH(&r[1], 0, CHAN_Z);
- FETCH(&r[2], 1, CHAN_Z);
-
- micro_mul( &r[1], &r[1], &r[2] );
- micro_add( &r[0], &r[0], &r[1] );
-
- FETCH(&r[1], 0, CHAN_W);
- FETCH(&r[2], 1, CHAN_W);
-
- micro_mul( &r[1], &r[1], &r[2] );
- micro_add( &r[0], &r[0], &r[1] );
-
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- STORE( &r[0], 0, chan_index );
- }
+ case TGSI_OPCODE_DP4:
+ exec_dp4(mach, inst);
break;
case TGSI_OPCODE_DST:
- if (IS_CHANNEL_ENABLED( *inst, CHAN_X )) {
- STORE( &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], 0, CHAN_X );
- }
-
if (IS_CHANNEL_ENABLED( *inst, CHAN_Y )) {
FETCH( &r[0], 0, CHAN_Y );
FETCH( &r[1], 1, CHAN_Y);
- micro_mul( &r[0], &r[0], &r[1] );
- STORE( &r[0], 0, CHAN_Y );
+ micro_mul(&d[CHAN_Y], &r[0], &r[1]);
}
-
if (IS_CHANNEL_ENABLED( *inst, CHAN_Z )) {
- FETCH( &r[0], 0, CHAN_Z );
- STORE( &r[0], 0, CHAN_Z );
+ FETCH(&d[CHAN_Z], 0, CHAN_Z);
}
-
if (IS_CHANNEL_ENABLED( *inst, CHAN_W )) {
- FETCH( &r[0], 1, CHAN_W );
- STORE( &r[0], 0, CHAN_W );
+ FETCH(&d[CHAN_W], 1, CHAN_W);
+ }
+
+ if (IS_CHANNEL_ENABLED(*inst, CHAN_X)) {
+ STORE(&mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], 0, CHAN_X);
+ }
+ if (IS_CHANNEL_ENABLED(*inst, CHAN_Y)) {
+ STORE(&d[CHAN_Y], 0, CHAN_Y);
+ }
+ if (IS_CHANNEL_ENABLED(*inst, CHAN_Z)) {
+ STORE(&d[CHAN_Z], 0, CHAN_Z);
+ }
+ if (IS_CHANNEL_ENABLED(*inst, CHAN_W)) {
+ STORE(&d[CHAN_W], 0, CHAN_W);
}
break;
@@ -2081,9 +2566,10 @@ exec_instruction(
FETCH(&r[1], 1, chan_index);
/* XXX use micro_min()?? */
- micro_lt( &r[0], &r[0], &r[1], &r[0], &r[1] );
-
- STORE(&r[0], 0, chan_index);
+ micro_lt(&d[chan_index], &r[0], &r[1], &r[0], &r[1]);
+ }
+ FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+ STORE(&d[chan_index], 0, chan_index);
}
break;
@@ -2093,67 +2579,38 @@ exec_instruction(
FETCH(&r[1], 1, chan_index);
/* XXX use micro_max()?? */
- micro_lt( &r[0], &r[0], &r[1], &r[1], &r[0] );
-
- STORE(&r[0], 0, chan_index );
+ micro_lt(&d[chan_index], &r[0], &r[1], &r[1], &r[0] );
+ }
+ FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+ STORE(&d[chan_index], 0, chan_index);
}
break;
case TGSI_OPCODE_SLT:
- /* TGSI_OPCODE_SETLT */
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- FETCH( &r[1], 1, chan_index );
- micro_lt( &r[0], &r[0], &r[1], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_binary(mach, inst, micro_slt, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_SGE:
- /* TGSI_OPCODE_SETGE */
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- FETCH( &r[1], 1, chan_index );
- micro_le( &r[0], &r[1], &r[0], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_binary(mach, inst, micro_sge, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_MAD:
- /* TGSI_OPCODE_MADD */
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- FETCH( &r[1], 1, chan_index );
- micro_mul( &r[0], &r[0], &r[1] );
- FETCH( &r[1], 2, chan_index );
- micro_add( &r[0], &r[0], &r[1] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_trinary(mach, inst, micro_mad, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_SUB:
FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
FETCH(&r[0], 0, chan_index);
FETCH(&r[1], 1, chan_index);
-
- micro_sub( &r[0], &r[0], &r[1] );
-
- STORE(&r[0], 0, chan_index);
+ micro_sub(&d[chan_index], &r[0], &r[1]);
+ }
+ FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+ STORE(&d[chan_index], 0, chan_index);
}
break;
case TGSI_OPCODE_LRP:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH(&r[0], 0, chan_index);
- FETCH(&r[1], 1, chan_index);
- FETCH(&r[2], 2, chan_index);
-
- micro_sub( &r[1], &r[1], &r[2] );
- micro_mul( &r[0], &r[0], &r[1] );
- micro_add( &r[0], &r[0], &r[2] );
-
- STORE(&r[0], 0, chan_index);
- }
+ exec_vector_trinary(mach, inst, micro_lrp, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_CND:
@@ -2161,35 +2618,19 @@ exec_instruction(
FETCH(&r[0], 0, chan_index);
FETCH(&r[1], 1, chan_index);
FETCH(&r[2], 2, chan_index);
- micro_lt(&r[0], &mach->Temps[TEMP_HALF_I].xyzw[TEMP_HALF_C], &r[2], &r[0], &r[1]);
- STORE(&r[0], 0, chan_index);
+ micro_lt(&d[chan_index], &mach->Temps[TEMP_HALF_I].xyzw[TEMP_HALF_C], &r[2], &r[0], &r[1]);
+ }
+ FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+ STORE(&d[chan_index], 0, chan_index);
}
break;
case TGSI_OPCODE_DP2A:
- FETCH( &r[0], 0, CHAN_X );
- FETCH( &r[1], 1, CHAN_X );
- micro_mul( &r[0], &r[0], &r[1] );
-
- FETCH( &r[1], 0, CHAN_Y );
- FETCH( &r[2], 1, CHAN_Y );
- micro_mul( &r[1], &r[1], &r[2] );
- micro_add( &r[0], &r[0], &r[1] );
-
- FETCH( &r[2], 2, CHAN_X );
- micro_add( &r[0], &r[0], &r[2] );
-
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- STORE( &r[0], 0, chan_index );
- }
+ exec_dp2a(mach, inst);
break;
case TGSI_OPCODE_FRC:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- micro_frc( &r[0], &r[0] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_unary(mach, inst, micro_frc, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_CLAMP:
@@ -2198,40 +2639,27 @@ exec_instruction(
FETCH(&r[1], 1, chan_index);
micro_max(&r[0], &r[0], &r[1]);
FETCH(&r[1], 2, chan_index);
- micro_min(&r[0], &r[0], &r[1]);
- STORE(&r[0], 0, chan_index);
+ micro_min(&d[chan_index], &r[0], &r[1]);
}
+ FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+ STORE(&d[chan_index], 0, chan_index);
+ }
+ break;
+
+ case TGSI_OPCODE_FLR:
+ exec_vector_unary(mach, inst, micro_flr, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_ROUND:
- case TGSI_OPCODE_ARR:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- micro_rnd( &r[0], &r[0] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_unary(mach, inst, micro_rnd, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_EX2:
- FETCH(&r[0], 0, CHAN_X);
-
-#if FAST_MATH
- micro_exp2( &r[0], &r[0] );
-#else
- micro_pow( &r[0], &mach->Temps[TEMP_2_I].xyzw[TEMP_2_C], &r[0] );
-#endif
-
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- STORE( &r[0], 0, chan_index );
- }
+ exec_scalar_unary(mach, inst, micro_exp2, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_LG2:
- FETCH( &r[0], 0, CHAN_X );
- micro_lg2( &r[0], &r[0] );
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- STORE( &r[0], 0, chan_index );
- }
+ exec_scalar_unary(mach, inst, micro_lg2, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_POW:
@@ -2255,11 +2683,7 @@ exec_instruction(
FETCH(&r[4], 1, CHAN_Y);
micro_mul( &r[5], &r[3], &r[4] );
- micro_sub( &r[2], &r[2], &r[5] );
-
- if (IS_CHANNEL_ENABLED( *inst, CHAN_X )) {
- STORE( &r[2], 0, CHAN_X );
- }
+ micro_sub(&d[CHAN_X], &r[2], &r[5]);
FETCH(&r[2], 1, CHAN_X);
@@ -2268,34 +2692,29 @@ exec_instruction(
FETCH(&r[5], 0, CHAN_X);
micro_mul( &r[1], &r[1], &r[5] );
- micro_sub( &r[3], &r[3], &r[1] );
-
- if (IS_CHANNEL_ENABLED( *inst, CHAN_Y )) {
- STORE( &r[3], 0, CHAN_Y );
- }
+ micro_sub(&d[CHAN_Y], &r[3], &r[1]);
micro_mul( &r[5], &r[5], &r[4] );
micro_mul( &r[0], &r[0], &r[2] );
- micro_sub( &r[5], &r[5], &r[0] );
+ micro_sub(&d[CHAN_Z], &r[5], &r[0]);
- if (IS_CHANNEL_ENABLED( *inst, CHAN_Z )) {
- STORE( &r[5], 0, CHAN_Z );
+ if (IS_CHANNEL_ENABLED(*inst, CHAN_X)) {
+ STORE(&d[CHAN_X], 0, CHAN_X);
+ }
+ if (IS_CHANNEL_ENABLED(*inst, CHAN_Y)) {
+ STORE(&d[CHAN_Y], 0, CHAN_Y);
+ }
+ if (IS_CHANNEL_ENABLED(*inst, CHAN_Z)) {
+ STORE(&d[CHAN_Z], 0, CHAN_Z);
}
-
if (IS_CHANNEL_ENABLED( *inst, CHAN_W )) {
STORE( &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], 0, CHAN_W );
}
break;
- case TGSI_OPCODE_ABS:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH(&r[0], 0, chan_index);
-
- micro_abs( &r[0], &r[0] );
-
- STORE(&r[0], 0, chan_index);
- }
- break;
+ case TGSI_OPCODE_ABS:
+ exec_vector_unary(mach, inst, micro_abs, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
+ break;
case TGSI_OPCODE_RCC:
FETCH(&r[0], 0, CHAN_X);
@@ -2307,56 +2726,19 @@ exec_instruction(
break;
case TGSI_OPCODE_DPH:
- FETCH(&r[0], 0, CHAN_X);
- FETCH(&r[1], 1, CHAN_X);
-
- micro_mul( &r[0], &r[0], &r[1] );
-
- FETCH(&r[1], 0, CHAN_Y);
- FETCH(&r[2], 1, CHAN_Y);
-
- micro_mul( &r[1], &r[1], &r[2] );
- micro_add( &r[0], &r[0], &r[1] );
-
- FETCH(&r[1], 0, CHAN_Z);
- FETCH(&r[2], 1, CHAN_Z);
-
- micro_mul( &r[1], &r[1], &r[2] );
- micro_add( &r[0], &r[0], &r[1] );
-
- FETCH(&r[1], 1, CHAN_W);
-
- micro_add( &r[0], &r[0], &r[1] );
-
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- STORE( &r[0], 0, chan_index );
- }
+ exec_dph(mach, inst);
break;
case TGSI_OPCODE_COS:
- FETCH(&r[0], 0, CHAN_X);
-
- micro_cos( &r[0], &r[0] );
-
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- STORE( &r[0], 0, chan_index );
- }
+ exec_scalar_unary(mach, inst, micro_cos, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_DDX:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- micro_ddx( &r[0], &r[0] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_unary(mach, inst, micro_ddx, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_DDY:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- micro_ddy( &r[0], &r[0] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_unary(mach, inst, micro_ddy, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_KILP:
@@ -2433,14 +2815,7 @@ exec_instruction(
break;
case TGSI_OPCODE_SEQ:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- FETCH( &r[1], 1, chan_index );
- micro_eq( &r[0], &r[0], &r[1],
- &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C],
- &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_binary(mach, inst, micro_seq, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_SFL:
@@ -2450,38 +2825,19 @@ exec_instruction(
break;
case TGSI_OPCODE_SGT:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- FETCH( &r[1], 1, chan_index );
- micro_le( &r[0], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_binary(mach, inst, micro_sgt, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_SIN:
- FETCH( &r[0], 0, CHAN_X );
- micro_sin( &r[0], &r[0] );
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- STORE( &r[0], 0, chan_index );
- }
+ exec_scalar_unary(mach, inst, micro_sin, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_SLE:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- FETCH( &r[1], 1, chan_index );
- micro_le( &r[0], &r[0], &r[1], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_binary(mach, inst, micro_sle, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_SNE:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- FETCH( &r[1], 1, chan_index );
- micro_eq( &r[0], &r[0], &r[1], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_binary(mach, inst, micro_sne, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_STR:
@@ -2494,14 +2850,14 @@ exec_instruction(
/* simple texture lookup */
/* src[0] = texcoord */
/* src[1] = sampler unit */
- exec_tex(mach, inst, FALSE, FALSE);
+ exec_tex(mach, inst, TEX_MODIFIER_NONE);
break;
case TGSI_OPCODE_TXB:
/* Texture lookup with lod bias */
/* src[0] = texcoord (src[0].w = LOD bias) */
/* src[1] = sampler unit */
- exec_tex(mach, inst, TRUE, FALSE);
+ exec_tex(mach, inst, TEX_MODIFIER_LOD_BIAS);
break;
case TGSI_OPCODE_TXD:
@@ -2510,21 +2866,21 @@ exec_instruction(
/* src[1] = d[strq]/dx */
/* src[2] = d[strq]/dy */
/* src[3] = sampler unit */
- assert (0);
+ exec_txd(mach, inst);
break;
case TGSI_OPCODE_TXL:
/* Texture lookup with explit LOD */
/* src[0] = texcoord (src[0].w = LOD) */
/* src[1] = sampler unit */
- exec_tex(mach, inst, TRUE, FALSE);
+ exec_tex(mach, inst, TEX_MODIFIER_EXPLICIT_LOD);
break;
case TGSI_OPCODE_TXP:
/* Texture lookup with projection */
/* src[0] = texcoord (src[0].w = projection) */
/* src[1] = sampler unit */
- exec_tex(mach, inst, FALSE, TRUE);
+ exec_tex(mach, inst, TEX_MODIFIER_PROJECTED);
break;
case TGSI_OPCODE_UP2H:
@@ -2554,13 +2910,8 @@ exec_instruction(
micro_mul(&r[3], &r[3], &r[1]);
micro_add(&r[2], &r[2], &r[3]);
FETCH(&r[3], 0, CHAN_X);
- micro_add(&r[2], &r[2], &r[3]);
- if (IS_CHANNEL_ENABLED(*inst, CHAN_X)) {
- STORE(&r[2], 0, CHAN_X);
- }
- if (IS_CHANNEL_ENABLED(*inst, CHAN_Z)) {
- STORE(&r[2], 0, CHAN_Z);
- }
+ micro_add(&d[CHAN_X], &r[2], &r[3]);
+
}
if (IS_CHANNEL_ENABLED(*inst, CHAN_Y) ||
IS_CHANNEL_ENABLED(*inst, CHAN_W)) {
@@ -2570,13 +2921,20 @@ exec_instruction(
micro_mul(&r[3], &r[3], &r[1]);
micro_add(&r[2], &r[2], &r[3]);
FETCH(&r[3], 0, CHAN_Y);
- micro_add(&r[2], &r[2], &r[3]);
- if (IS_CHANNEL_ENABLED(*inst, CHAN_Y)) {
- STORE(&r[2], 0, CHAN_Y);
- }
- if (IS_CHANNEL_ENABLED(*inst, CHAN_W)) {
- STORE(&r[2], 0, CHAN_W);
- }
+ micro_add(&d[CHAN_Y], &r[2], &r[3]);
+
+ }
+ if (IS_CHANNEL_ENABLED(*inst, CHAN_X)) {
+ STORE(&d[CHAN_X], 0, CHAN_X);
+ }
+ if (IS_CHANNEL_ENABLED(*inst, CHAN_Y)) {
+ STORE(&d[CHAN_Y], 0, CHAN_Y);
+ }
+ if (IS_CHANNEL_ENABLED(*inst, CHAN_Z)) {
+ STORE(&d[CHAN_X], 0, CHAN_Z);
+ }
+ if (IS_CHANNEL_ENABLED(*inst, CHAN_W)) {
+ STORE(&d[CHAN_Y], 0, CHAN_W);
}
break;
@@ -2584,6 +2942,10 @@ exec_instruction(
assert (0);
break;
+ case TGSI_OPCODE_ARR:
+ exec_vector_unary(mach, inst, micro_arr, TGSI_EXEC_DATA_INT, TGSI_EXEC_DATA_FLOAT);
+ break;
+
case TGSI_OPCODE_BRA:
assert (0);
break;
@@ -2603,6 +2965,8 @@ exec_instruction(
mach->CallStack[mach->CallStackTop].CondStackTop = mach->CondStackTop;
mach->CallStack[mach->CallStackTop].LoopStackTop = mach->LoopStackTop;
mach->CallStack[mach->CallStackTop].ContStackTop = mach->ContStackTop;
+ mach->CallStack[mach->CallStackTop].SwitchStackTop = mach->SwitchStackTop;
+ mach->CallStack[mach->CallStackTop].BreakStackTop = mach->BreakStackTop;
/* note that PC was already incremented above */
mach->CallStack[mach->CallStackTop].ReturnAddr = *pc;
@@ -2610,16 +2974,21 @@ exec_instruction(
/* Second, push the Cond, Loop, Cont, Func stacks */
assert(mach->CondStackTop < TGSI_EXEC_MAX_COND_NESTING);
- mach->CondStack[mach->CondStackTop++] = mach->CondMask;
assert(mach->LoopStackTop < TGSI_EXEC_MAX_LOOP_NESTING);
- mach->LoopStack[mach->LoopStackTop++] = mach->LoopMask;
assert(mach->ContStackTop < TGSI_EXEC_MAX_LOOP_NESTING);
- mach->ContStack[mach->ContStackTop++] = mach->ContMask;
+ assert(mach->SwitchStackTop < TGSI_EXEC_MAX_SWITCH_NESTING);
+ assert(mach->BreakStackTop < TGSI_EXEC_MAX_BREAK_STACK);
assert(mach->FuncStackTop < TGSI_EXEC_MAX_CALL_NESTING);
+
+ mach->CondStack[mach->CondStackTop++] = mach->CondMask;
+ mach->LoopStack[mach->LoopStackTop++] = mach->LoopMask;
+ mach->ContStack[mach->ContStackTop++] = mach->ContMask;
+ mach->SwitchStack[mach->SwitchStackTop++] = mach->Switch;
+ mach->BreakStack[mach->BreakStackTop++] = mach->BreakType;
mach->FuncStack[mach->FuncStackTop++] = mach->FuncMask;
/* Finally, jump to the subroutine */
- *pc = inst->InstructionExtLabel.Label;
+ *pc = inst->Label.Label;
}
break;
@@ -2648,6 +3017,12 @@ exec_instruction(
mach->ContStackTop = mach->CallStack[mach->CallStackTop].ContStackTop;
mach->ContMask = mach->ContStack[mach->ContStackTop];
+ mach->SwitchStackTop = mach->CallStack[mach->CallStackTop].SwitchStackTop;
+ mach->Switch = mach->SwitchStack[mach->SwitchStackTop];
+
+ mach->BreakStackTop = mach->CallStack[mach->CallStackTop].BreakStackTop;
+ mach->BreakType = mach->BreakStack[mach->BreakStackTop];
+
assert(mach->FuncStackTop > 0);
mach->FuncMask = mach->FuncStack[--mach->FuncStackTop];
@@ -2658,12 +3033,7 @@ exec_instruction(
break;
case TGSI_OPCODE_SSG:
- /* TGSI_OPCODE_SGN */
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- micro_sgn( &r[0], &r[0] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_unary(mach, inst, micro_sgn, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_CMP:
@@ -2671,10 +3041,10 @@ exec_instruction(
FETCH(&r[0], 0, chan_index);
FETCH(&r[1], 1, chan_index);
FETCH(&r[2], 2, chan_index);
-
- micro_lt( &r[0], &r[0], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &r[1], &r[2] );
-
- STORE(&r[0], 0, chan_index);
+ micro_lt(&d[chan_index], &r[0], &mach->Temps[TEMP_0_I].xyzw[TEMP_0_C], &r[1], &r[2]);
+ }
+ FOR_EACH_ENABLED_CHANNEL(*inst, chan_index) {
+ STORE(&d[chan_index], 0, chan_index);
}
break;
@@ -2770,18 +3140,7 @@ exec_instruction(
break;
case TGSI_OPCODE_DP2:
- FETCH( &r[0], 0, CHAN_X );
- FETCH( &r[1], 1, CHAN_X );
- micro_mul( &r[0], &r[0], &r[1] );
-
- FETCH( &r[1], 0, CHAN_Y );
- FETCH( &r[2], 1, CHAN_Y );
- micro_mul( &r[1], &r[1], &r[2] );
- micro_add( &r[0], &r[0], &r[1] );
-
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- STORE( &r[0], 0, chan_index );
- }
+ exec_dp2(mach, inst);
break;
case TGSI_OPCODE_IF:
@@ -2847,71 +3206,31 @@ exec_instruction(
break;
case TGSI_OPCODE_CEIL:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- micro_ceil( &r[0], &r[0] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_unary(mach, inst, micro_ceil, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_I2F:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- micro_i2f( &r[0], &r[0] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_unary(mach, inst, micro_i2f, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_INT);
break;
case TGSI_OPCODE_NOT:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- micro_not( &r[0], &r[0] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_unary(mach, inst, micro_not, TGSI_EXEC_DATA_UINT, TGSI_EXEC_DATA_UINT);
break;
case TGSI_OPCODE_TRUNC:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- micro_trunc( &r[0], &r[0] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_unary(mach, inst, micro_trunc, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_FLOAT);
break;
case TGSI_OPCODE_SHL:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- FETCH( &r[1], 1, chan_index );
- micro_shl( &r[0], &r[0], &r[1] );
- STORE( &r[0], 0, chan_index );
- }
- break;
-
- case TGSI_OPCODE_SHR:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- FETCH( &r[1], 1, chan_index );
- micro_ishr( &r[0], &r[0], &r[1] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_binary(mach, inst, micro_shl, TGSI_EXEC_DATA_UINT, TGSI_EXEC_DATA_UINT);
break;
case TGSI_OPCODE_AND:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- FETCH( &r[1], 1, chan_index );
- micro_and( &r[0], &r[0], &r[1] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_binary(mach, inst, micro_and, TGSI_EXEC_DATA_UINT, TGSI_EXEC_DATA_UINT);
break;
case TGSI_OPCODE_OR:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- FETCH( &r[1], 1, chan_index );
- micro_or( &r[0], &r[0], &r[1] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_binary(mach, inst, micro_or, TGSI_EXEC_DATA_UINT, TGSI_EXEC_DATA_UINT);
break;
case TGSI_OPCODE_MOD:
@@ -2919,12 +3238,7 @@ exec_instruction(
break;
case TGSI_OPCODE_XOR:
- FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
- FETCH( &r[0], 0, chan_index );
- FETCH( &r[1], 1, chan_index );
- micro_xor( &r[0], &r[0], &r[1] );
- STORE( &r[0], 0, chan_index );
- }
+ exec_vector_binary(mach, inst, micro_xor, TGSI_EXEC_DATA_UINT, TGSI_EXEC_DATA_UINT);
break;
case TGSI_OPCODE_SAD:
@@ -2940,13 +3254,11 @@ exec_instruction(
break;
case TGSI_OPCODE_EMIT:
- mach->Temps[TEMP_OUTPUT_I].xyzw[TEMP_OUTPUT_C].u[0] += 16;
- mach->Primitives[mach->Temps[TEMP_PRIMITIVE_I].xyzw[TEMP_PRIMITIVE_C].u[0]]++;
+ emit_vertex(mach);
break;
case TGSI_OPCODE_ENDPRIM:
- mach->Temps[TEMP_PRIMITIVE_I].xyzw[TEMP_PRIMITIVE_C].u[0]++;
- mach->Primitives[mach->Temps[TEMP_PRIMITIVE_I].xyzw[TEMP_PRIMITIVE_C].u[0]] = 0;
+ emit_primitive(mach);
break;
case TGSI_OPCODE_BGNFOR:
@@ -2954,43 +3266,62 @@ exec_instruction(
for (chan_index = 0; chan_index < 3; chan_index++) {
FETCH( &mach->LoopCounterStack[mach->LoopCounterStackTop].xyzw[chan_index], 0, chan_index );
}
- STORE( &mach->LoopCounterStack[mach->LoopCounterStackTop].xyzw[CHAN_Y], 0, CHAN_X );
++mach->LoopCounterStackTop;
+ STORE(&mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_X], 0, CHAN_X);
+ /* update LoopMask */
+ if (mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y].f[0] <= 0.0f) {
+ mach->LoopMask &= ~0x1;
+ }
+ if (mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y].f[1] <= 0.0f) {
+ mach->LoopMask &= ~0x2;
+ }
+ if (mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y].f[2] <= 0.0f) {
+ mach->LoopMask &= ~0x4;
+ }
+ if (mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y].f[3] <= 0.0f) {
+ mach->LoopMask &= ~0x8;
+ }
+ /* TODO: if mach->LoopMask == 0, jump to end of loop */
+ UPDATE_EXEC_MASK(mach);
/* fall-through (for now) */
case TGSI_OPCODE_BGNLOOP:
/* push LoopMask and ContMasks */
assert(mach->LoopStackTop < TGSI_EXEC_MAX_LOOP_NESTING);
- mach->LoopStack[mach->LoopStackTop++] = mach->LoopMask;
assert(mach->ContStackTop < TGSI_EXEC_MAX_LOOP_NESTING);
- mach->ContStack[mach->ContStackTop++] = mach->ContMask;
assert(mach->LoopLabelStackTop < TGSI_EXEC_MAX_LOOP_NESTING);
+ assert(mach->BreakStackTop < TGSI_EXEC_MAX_BREAK_STACK);
+
+ mach->LoopStack[mach->LoopStackTop++] = mach->LoopMask;
+ mach->ContStack[mach->ContStackTop++] = mach->ContMask;
mach->LoopLabelStack[mach->LoopLabelStackTop++] = *pc - 1;
+ mach->BreakStack[mach->BreakStackTop++] = mach->BreakType;
+ mach->BreakType = TGSI_EXEC_BREAK_INSIDE_LOOP;
break;
case TGSI_OPCODE_ENDFOR:
assert(mach->LoopCounterStackTop > 0);
- micro_sub( &mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_X],
- &mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_X],
- &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C] );
+ micro_sub(&mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y],
+ &mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y],
+ &mach->Temps[TEMP_1_I].xyzw[TEMP_1_C]);
/* update LoopMask */
- if( mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_X].f[0] <= 0) {
+ if (mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y].f[0] <= 0.0f) {
mach->LoopMask &= ~0x1;
}
- if( mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_X].f[1] <= 0 ) {
+ if (mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y].f[1] <= 0.0f) {
mach->LoopMask &= ~0x2;
}
- if( mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_X].f[2] <= 0 ) {
+ if (mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y].f[2] <= 0.0f) {
mach->LoopMask &= ~0x4;
}
- if( mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_X].f[3] <= 0 ) {
+ if (mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y].f[3] <= 0.0f) {
mach->LoopMask &= ~0x8;
}
- micro_add( &mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y],
- &mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Y],
- &mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Z]);
+ micro_add(&mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_X],
+ &mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_X],
+ &mach->LoopCounterStack[mach->LoopCounterStackTop - 1].xyzw[CHAN_Z]);
assert(mach->LoopLabelStackTop > 0);
inst = mach->Instructions + mach->LoopLabelStack[mach->LoopLabelStackTop - 1];
- STORE( &mach->LoopCounterStack[mach->LoopCounterStackTop].xyzw[CHAN_Y], 0, CHAN_X );
+ STORE(&mach->LoopCounterStack[mach->LoopCounterStackTop].xyzw[CHAN_X], 0, CHAN_X);
/* Restore ContMask, but don't pop */
assert(mach->ContStackTop > 0);
mach->ContMask = mach->ContStack[mach->ContStackTop - 1];
@@ -3011,6 +3342,8 @@ exec_instruction(
--mach->LoopLabelStackTop;
assert(mach->LoopCounterStackTop > 0);
--mach->LoopCounterStackTop;
+
+ mach->BreakType = mach->BreakStack[--mach->BreakStackTop];
}
UPDATE_EXEC_MASK(mach);
break;
@@ -3034,15 +3367,14 @@ exec_instruction(
mach->ContMask = mach->ContStack[--mach->ContStackTop];
assert(mach->LoopLabelStackTop > 0);
--mach->LoopLabelStackTop;
+
+ mach->BreakType = mach->BreakStack[--mach->BreakStackTop];
}
UPDATE_EXEC_MASK(mach);
break;
case TGSI_OPCODE_BRK:
- /* turn off loop channels for each enabled exec channel */
- mach->LoopMask &= ~mach->ExecMask;
- /* Todo: if mach->LoopMask == 0, jump to end of loop */
- UPDATE_EXEC_MASK(mach);
+ exec_break(mach);
break;
case TGSI_OPCODE_CONT:
@@ -3057,18 +3389,171 @@ exec_instruction(
break;
case TGSI_OPCODE_ENDSUB:
- /* no-op */
+ /*
+ * XXX: This really should be a no-op. We should never reach this opcode.
+ */
+
+ assert(mach->CallStackTop > 0);
+ mach->CallStackTop--;
+
+ mach->CondStackTop = mach->CallStack[mach->CallStackTop].CondStackTop;
+ mach->CondMask = mach->CondStack[mach->CondStackTop];
+
+ mach->LoopStackTop = mach->CallStack[mach->CallStackTop].LoopStackTop;
+ mach->LoopMask = mach->LoopStack[mach->LoopStackTop];
+
+ mach->ContStackTop = mach->CallStack[mach->CallStackTop].ContStackTop;
+ mach->ContMask = mach->ContStack[mach->ContStackTop];
+
+ mach->SwitchStackTop = mach->CallStack[mach->CallStackTop].SwitchStackTop;
+ mach->Switch = mach->SwitchStack[mach->SwitchStackTop];
+
+ mach->BreakStackTop = mach->CallStack[mach->CallStackTop].BreakStackTop;
+ mach->BreakType = mach->BreakStack[mach->BreakStackTop];
+
+ assert(mach->FuncStackTop > 0);
+ mach->FuncMask = mach->FuncStack[--mach->FuncStackTop];
+
+ *pc = mach->CallStack[mach->CallStackTop].ReturnAddr;
+
+ UPDATE_EXEC_MASK(mach);
break;
case TGSI_OPCODE_NOP:
break;
+ case TGSI_OPCODE_BREAKC:
+ FETCH(&r[0], 0, CHAN_X);
+ /* update CondMask */
+ if (r[0].u[0] && (mach->ExecMask & 0x1)) {
+ mach->LoopMask &= ~0x1;
+ }
+ if (r[0].u[1] && (mach->ExecMask & 0x2)) {
+ mach->LoopMask &= ~0x2;
+ }
+ if (r[0].u[2] && (mach->ExecMask & 0x4)) {
+ mach->LoopMask &= ~0x4;
+ }
+ if (r[0].u[3] && (mach->ExecMask & 0x8)) {
+ mach->LoopMask &= ~0x8;
+ }
+ /* Todo: if mach->LoopMask == 0, jump to end of loop */
+ UPDATE_EXEC_MASK(mach);
+ break;
+
+ case TGSI_OPCODE_F2I:
+ exec_vector_unary(mach, inst, micro_f2i, TGSI_EXEC_DATA_INT, TGSI_EXEC_DATA_FLOAT);
+ break;
+
+ case TGSI_OPCODE_IDIV:
+ exec_vector_binary(mach, inst, micro_idiv, TGSI_EXEC_DATA_INT, TGSI_EXEC_DATA_INT);
+ break;
+
+ case TGSI_OPCODE_IMAX:
+ exec_vector_binary(mach, inst, micro_imax, TGSI_EXEC_DATA_INT, TGSI_EXEC_DATA_INT);
+ break;
+
+ case TGSI_OPCODE_IMIN:
+ exec_vector_binary(mach, inst, micro_imin, TGSI_EXEC_DATA_INT, TGSI_EXEC_DATA_INT);
+ break;
+
+ case TGSI_OPCODE_INEG:
+ exec_vector_unary(mach, inst, micro_ineg, TGSI_EXEC_DATA_INT, TGSI_EXEC_DATA_INT);
+ break;
+
+ case TGSI_OPCODE_ISGE:
+ exec_vector_binary(mach, inst, micro_isge, TGSI_EXEC_DATA_INT, TGSI_EXEC_DATA_INT);
+ break;
+
+ case TGSI_OPCODE_ISHR:
+ exec_vector_binary(mach, inst, micro_ishr, TGSI_EXEC_DATA_INT, TGSI_EXEC_DATA_INT);
+ break;
+
+ case TGSI_OPCODE_ISLT:
+ exec_vector_binary(mach, inst, micro_islt, TGSI_EXEC_DATA_INT, TGSI_EXEC_DATA_INT);
+ break;
+
+ case TGSI_OPCODE_F2U:
+ exec_vector_unary(mach, inst, micro_f2u, TGSI_EXEC_DATA_UINT, TGSI_EXEC_DATA_FLOAT);
+ break;
+
+ case TGSI_OPCODE_U2F:
+ exec_vector_unary(mach, inst, micro_u2f, TGSI_EXEC_DATA_FLOAT, TGSI_EXEC_DATA_UINT);
+ break;
+
+ case TGSI_OPCODE_UADD:
+ exec_vector_binary(mach, inst, micro_uadd, TGSI_EXEC_DATA_UINT, TGSI_EXEC_DATA_UINT);
+ break;
+
+ case TGSI_OPCODE_UDIV:
+ exec_vector_binary(mach, inst, micro_udiv, TGSI_EXEC_DATA_UINT, TGSI_EXEC_DATA_UINT);
+ break;
+
+ case TGSI_OPCODE_UMAD:
+ exec_vector_trinary(mach, inst, micro_umad, TGSI_EXEC_DATA_UINT, TGSI_EXEC_DATA_UINT);
+ break;
+
+ case TGSI_OPCODE_UMAX:
+ exec_vector_binary(mach, inst, micro_umax, TGSI_EXEC_DATA_UINT, TGSI_EXEC_DATA_UINT);
+ break;
+
+ case TGSI_OPCODE_UMIN:
+ exec_vector_binary(mach, inst, micro_umin, TGSI_EXEC_DATA_UINT, TGSI_EXEC_DATA_UINT);
+ break;
+
+ case TGSI_OPCODE_UMOD:
+ exec_vector_binary(mach, inst, micro_umod, TGSI_EXEC_DATA_UINT, TGSI_EXEC_DATA_UINT);
+ break;
+
+ case TGSI_OPCODE_UMUL:
+ exec_vector_binary(mach, inst, micro_umul, TGSI_EXEC_DATA_UINT, TGSI_EXEC_DATA_UINT);
+ break;
+
+ case TGSI_OPCODE_USEQ:
+ exec_vector_binary(mach, inst, micro_useq, TGSI_EXEC_DATA_UINT, TGSI_EXEC_DATA_UINT);
+ break;
+
+ case TGSI_OPCODE_USGE:
+ exec_vector_binary(mach, inst, micro_usge, TGSI_EXEC_DATA_UINT, TGSI_EXEC_DATA_UINT);
+ break;
+
+ case TGSI_OPCODE_USHR:
+ exec_vector_binary(mach, inst, micro_ushr, TGSI_EXEC_DATA_UINT, TGSI_EXEC_DATA_UINT);
+ break;
+
+ case TGSI_OPCODE_USLT:
+ exec_vector_binary(mach, inst, micro_uslt, TGSI_EXEC_DATA_UINT, TGSI_EXEC_DATA_UINT);
+ break;
+
+ case TGSI_OPCODE_USNE:
+ exec_vector_binary(mach, inst, micro_usne, TGSI_EXEC_DATA_UINT, TGSI_EXEC_DATA_UINT);
+ break;
+
+ case TGSI_OPCODE_SWITCH:
+ exec_switch(mach, inst);
+ break;
+
+ case TGSI_OPCODE_CASE:
+ exec_case(mach, inst);
+ break;
+
+ case TGSI_OPCODE_DEFAULT:
+ exec_default(mach);
+ break;
+
+ case TGSI_OPCODE_ENDSWITCH:
+ exec_endswitch(mach);
+ break;
+
default:
assert( 0 );
}
}
+#define DEBUG_EXECUTION 0
+
+
/**
* Run TGSI interpreter.
* \return bitmask of "alive" quad components
@@ -3085,9 +3570,13 @@ tgsi_exec_machine_run( struct tgsi_exec_machine *mach )
mach->FuncMask = 0xf;
mach->ExecMask = 0xf;
+ mach->Switch.mask = 0xf;
+
assert(mach->CondStackTop == 0);
assert(mach->LoopStackTop == 0);
assert(mach->ContStackTop == 0);
+ assert(mach->SwitchStackTop == 0);
+ assert(mach->BreakStackTop == 0);
assert(mach->CallStackTop == 0);
mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0] = 0;
@@ -3111,10 +3600,67 @@ tgsi_exec_machine_run( struct tgsi_exec_machine *mach )
exec_declaration( mach, mach->Declarations+i );
}
- /* execute instructions, until pc is set to -1 */
- while (pc != -1) {
- assert(pc < (int) mach->NumInstructions);
- exec_instruction( mach, mach->Instructions + pc, &pc );
+ {
+#if DEBUG_EXECUTION
+ struct tgsi_exec_vector temps[TGSI_EXEC_NUM_TEMPS + TGSI_EXEC_NUM_TEMP_EXTRAS];
+ struct tgsi_exec_vector outputs[PIPE_MAX_ATTRIBS];
+ uint inst = 1;
+
+ memcpy(temps, mach->Temps, sizeof(temps));
+ memcpy(outputs, mach->Outputs, sizeof(outputs));
+#endif
+
+ /* execute instructions, until pc is set to -1 */
+ while (pc != -1) {
+
+#if DEBUG_EXECUTION
+ uint i;
+
+ tgsi_dump_instruction(&mach->Instructions[pc], inst++);
+#endif
+
+ assert(pc < (int) mach->NumInstructions);
+ exec_instruction(mach, mach->Instructions + pc, &pc);
+
+#if DEBUG_EXECUTION
+ for (i = 0; i < TGSI_EXEC_NUM_TEMPS + TGSI_EXEC_NUM_TEMP_EXTRAS; i++) {
+ if (memcmp(&temps[i], &mach->Temps[i], sizeof(temps[i]))) {
+ uint j;
+
+ memcpy(&temps[i], &mach->Temps[i], sizeof(temps[i]));
+ debug_printf("TEMP[%2u] = ", i);
+ for (j = 0; j < 4; j++) {
+ if (j > 0) {
+ debug_printf(" ");
+ }
+ debug_printf("(%6f %u, %6f %u, %6f %u, %6f %u)\n",
+ temps[i].xyzw[0].f[j], temps[i].xyzw[0].u[j],
+ temps[i].xyzw[1].f[j], temps[i].xyzw[1].u[j],
+ temps[i].xyzw[2].f[j], temps[i].xyzw[2].u[j],
+ temps[i].xyzw[3].f[j], temps[i].xyzw[3].u[j]);
+ }
+ }
+ }
+ for (i = 0; i < PIPE_MAX_ATTRIBS; i++) {
+ if (memcmp(&outputs[i], &mach->Outputs[i], sizeof(outputs[i]))) {
+ uint j;
+
+ memcpy(&outputs[i], &mach->Outputs[i], sizeof(outputs[i]));
+ debug_printf("OUT[%2u] = ", i);
+ for (j = 0; j < 4; j++) {
+ if (j > 0) {
+ debug_printf(" ");
+ }
+ debug_printf("(%6f %u, %6f %u, %6f %u, %6f %u)\n",
+ outputs[i].xyzw[0].f[j], outputs[i].xyzw[0].u[j],
+ outputs[i].xyzw[1].f[j], outputs[i].xyzw[1].u[j],
+ outputs[i].xyzw[2].f[j], outputs[i].xyzw[2].u[j],
+ outputs[i].xyzw[3].f[j], outputs[i].xyzw[3].u[j]);
+ }
+ }
+ }
+#endif
+ }
}
#if 0
@@ -3128,5 +3674,12 @@ tgsi_exec_machine_run( struct tgsi_exec_machine *mach )
}
#endif
+ assert(mach->CondStackTop == 0);
+ assert(mach->LoopStackTop == 0);
+ assert(mach->ContStackTop == 0);
+ assert(mach->SwitchStackTop == 0);
+ assert(mach->BreakStackTop == 0);
+ assert(mach->CallStackTop == 0);
+
return ~mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0];
}
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.h b/src/gallium/auxiliary/tgsi/tgsi_exec.h
index 471f591dd6e..59e3b445cc3 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_exec.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_exec.h
@@ -2,6 +2,7 @@
*
* Copyright 2007-2008 Tungsten Graphics, Inc., Cedar Park, Texas.
* All Rights Reserved.
+ * Copyright 2009-2010 VMware, Inc. All rights Reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the
@@ -35,11 +36,13 @@
extern "C" {
#endif
+
#define MAX_LABELS (4 * 1024) /**< basically, max instructions */
#define NUM_CHANNELS 4 /* R,G,B,A */
#define QUAD_SIZE 4 /* 4 pixel/quad */
+
/**
* Registers may be treated as float, signed int or unsigned int.
*/
@@ -69,6 +72,11 @@ struct tgsi_interp_coef
float dady[NUM_CHANNELS];
};
+enum tgsi_sampler_control {
+ tgsi_sampler_lod_bias,
+ tgsi_sampler_lod_explicit
+};
+
/**
* Information for sampling textures, which must be implemented
* by code outside the TGSI executor.
@@ -80,7 +88,8 @@ struct tgsi_sampler
const float s[QUAD_SIZE],
const float t[QUAD_SIZE],
const float p[QUAD_SIZE],
- float lodbias,
+ const float c0[QUAD_SIZE],
+ enum tgsi_sampler_control control,
float rgba[NUM_CHANNELS][QUAD_SIZE]);
};
@@ -179,6 +188,7 @@ struct tgsi_exec_labels
#define TGSI_EXEC_MAX_COND_NESTING 32
#define TGSI_EXEC_MAX_LOOP_NESTING 32
+#define TGSI_EXEC_MAX_SWITCH_NESTING 32
#define TGSI_EXEC_MAX_CALL_NESTING 32
/* The maximum number of input attributes per vertex. For 2D
@@ -191,6 +201,14 @@ struct tgsi_exec_labels
*/
#define TGSI_EXEC_MAX_CONST_BUFFER 4096
+/* The maximum number of vertices per primitive */
+#define TGSI_MAX_PRIM_VERTICES 6
+
+/* The maximum number of primitives to be generated */
+#define TGSI_MAX_PRIMITIVES 64
+
+/* The maximum total number of vertices */
+#define TGSI_MAX_TOTAL_VERTICES (TGSI_MAX_PRIM_VERTICES * TGSI_MAX_PRIMITIVES * PIPE_MAX_ATTRIBS)
/** function call/activation record */
struct tgsi_call_record
@@ -198,10 +216,29 @@ struct tgsi_call_record
uint CondStackTop;
uint LoopStackTop;
uint ContStackTop;
+ int SwitchStackTop;
+ int BreakStackTop;
uint ReturnAddr;
};
+/* Switch-case block state. */
+struct tgsi_switch_record {
+ uint mask; /**< execution mask */
+ union tgsi_exec_channel selector; /**< a value case statements are compared to */
+ uint defaultMask; /**< non-execute mask for default case */
+};
+
+
+enum tgsi_break_type {
+ TGSI_EXEC_BREAK_INSIDE_LOOP,
+ TGSI_EXEC_BREAK_INSIDE_SWITCH
+};
+
+
+#define TGSI_EXEC_MAX_BREAK_STACK (TGSI_EXEC_MAX_LOOP_NESTING + TGSI_EXEC_MAX_SWITCH_NESTING)
+
+
/**
* Run-time virtual machine state for executing TGSI shader.
*/
@@ -214,10 +251,11 @@ struct tgsi_exec_machine
float Imms[TGSI_EXEC_NUM_IMMEDIATES][4];
- struct tgsi_exec_vector Inputs[PIPE_MAX_ATTRIBS];
- struct tgsi_exec_vector Outputs[PIPE_MAX_ATTRIBS];
+ struct tgsi_exec_vector Inputs[TGSI_MAX_PRIM_VERTICES * PIPE_MAX_ATTRIBS];
+ struct tgsi_exec_vector Outputs[TGSI_MAX_TOTAL_VERTICES];
struct tgsi_exec_vector *Addrs;
+ struct tgsi_exec_vector *Predicates;
struct tgsi_sampler **Samplers;
@@ -228,10 +266,13 @@ struct tgsi_exec_machine
/* GEOMETRY processor only. */
unsigned *Primitives;
+ unsigned NumOutputs;
+ unsigned MaxGeometryShaderOutputs;
/* FRAGMENT processor only. */
const struct tgsi_interp_coef *InterpCoefs;
struct tgsi_exec_vector QuadPos;
+ float Face; /**< +1 if front facing, -1 if back facing */
/* Conditional execution masks */
uint CondMask; /**< For IF/ELSE/ENDIF */
@@ -240,6 +281,12 @@ struct tgsi_exec_machine
uint FuncMask; /**< For function calls */
uint ExecMask; /**< = CondMask & LoopMask */
+ /* Current switch-case state. */
+ struct tgsi_switch_record Switch;
+
+ /* Current break type. */
+ enum tgsi_break_type BreakType;
+
/** Condition mask stack (for nested conditionals) */
uint CondStack[TGSI_EXEC_MAX_COND_NESTING];
int CondStackTop;
@@ -252,7 +299,7 @@ struct tgsi_exec_machine
uint LoopLabelStack[TGSI_EXEC_MAX_LOOP_NESTING];
int LoopLabelStackTop;
- /** Loop counter stack (x = count, y = current, z = step) */
+ /** Loop counter stack (x = index, y = counter, z = step) */
struct tgsi_exec_vector LoopCounterStack[TGSI_EXEC_MAX_LOOP_NESTING];
int LoopCounterStackTop;
@@ -260,6 +307,13 @@ struct tgsi_exec_machine
uint ContStack[TGSI_EXEC_MAX_LOOP_NESTING];
int ContStackTop;
+ /** Switch case stack */
+ struct tgsi_switch_record SwitchStack[TGSI_EXEC_MAX_SWITCH_NESTING];
+ int SwitchStackTop;
+
+ enum tgsi_break_type BreakStack[TGSI_EXEC_MAX_BREAK_STACK];
+ int BreakStackTop;
+
/** Function execution mask stack (for executing subroutine code) */
uint FuncStack[TGSI_EXEC_MAX_CALL_NESTING];
int FuncStackTop;
diff --git a/src/gallium/auxiliary/tgsi/tgsi_info.c b/src/gallium/auxiliary/tgsi/tgsi_info.c
index be375cabb8b..de0e09cdbae 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_info.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_info.c
@@ -119,7 +119,7 @@ static const struct tgsi_opcode_info opcode_info[TGSI_OPCODE_LAST] =
{ 1, 1, 0, 0, 0, 0, "NOT", TGSI_OPCODE_NOT },
{ 1, 1, 0, 0, 0, 0, "TRUNC", TGSI_OPCODE_TRUNC },
{ 1, 2, 0, 0, 0, 0, "SHL", TGSI_OPCODE_SHL },
- { 1, 2, 0, 0, 0, 0, "SHR", TGSI_OPCODE_SHR },
+ { 0, 0, 0, 0, 0, 0, "", 88 }, /* removed */
{ 1, 2, 0, 0, 0, 0, "AND", TGSI_OPCODE_AND },
{ 1, 2, 0, 0, 0, 0, "OR", TGSI_OPCODE_OR },
{ 1, 2, 0, 0, 0, 0, "MOD", TGSI_OPCODE_MOD },
@@ -149,7 +149,33 @@ static const struct tgsi_opcode_info opcode_info[TGSI_OPCODE_LAST] =
{ 0, 1, 0, 0, 0, 0, "BREAKC", TGSI_OPCODE_BREAKC },
{ 0, 1, 0, 0, 0, 0, "KIL", TGSI_OPCODE_KIL },
{ 0, 0, 0, 0, 0, 0, "END", TGSI_OPCODE_END },
- { 0, 0, 0, 0, 0, 0, "", 118 } /* removed */
+ { 0, 0, 0, 0, 0, 0, "", 118 }, /* removed */
+ { 1, 1, 0, 0, 0, 0, "F2I", TGSI_OPCODE_F2I },
+ { 1, 2, 0, 0, 0, 0, "IDIV", TGSI_OPCODE_IDIV },
+ { 1, 2, 0, 0, 0, 0, "IMAX", TGSI_OPCODE_IMAX },
+ { 1, 2, 0, 0, 0, 0, "IMIN", TGSI_OPCODE_IMIN },
+ { 1, 1, 0, 0, 0, 0, "INEG", TGSI_OPCODE_INEG },
+ { 1, 2, 0, 0, 0, 0, "ISGE", TGSI_OPCODE_ISGE },
+ { 1, 2, 0, 0, 0, 0, "ISHR", TGSI_OPCODE_ISHR },
+ { 1, 2, 0, 0, 0, 0, "ISLT", TGSI_OPCODE_ISLT },
+ { 1, 1, 0, 0, 0, 0, "F2U", TGSI_OPCODE_F2U },
+ { 1, 1, 0, 0, 0, 0, "U2F", TGSI_OPCODE_U2F },
+ { 1, 2, 0, 0, 0, 0, "UADD", TGSI_OPCODE_UADD },
+ { 1, 2, 0, 0, 0, 0, "UDIV", TGSI_OPCODE_UDIV },
+ { 1, 3, 0, 0, 0, 0, "UMAD", TGSI_OPCODE_UMAD },
+ { 1, 2, 0, 0, 0, 0, "UMAX", TGSI_OPCODE_UMAX },
+ { 1, 2, 0, 0, 0, 0, "UMIN", TGSI_OPCODE_UMIN },
+ { 1, 2, 0, 0, 0, 0, "UMOD", TGSI_OPCODE_UMOD },
+ { 1, 2, 0, 0, 0, 0, "UMUL", TGSI_OPCODE_UMUL },
+ { 1, 2, 0, 0, 0, 0, "USEQ", TGSI_OPCODE_USEQ },
+ { 1, 2, 0, 0, 0, 0, "USGE", TGSI_OPCODE_USGE },
+ { 1, 2, 0, 0, 0, 0, "USHR", TGSI_OPCODE_USHR },
+ { 1, 2, 0, 0, 0, 0, "USLT", TGSI_OPCODE_USLT },
+ { 1, 2, 0, 0, 0, 0, "USNE", TGSI_OPCODE_USNE },
+ { 0, 1, 0, 0, 0, 0, "SWITCH", TGSI_OPCODE_SWITCH },
+ { 0, 1, 0, 0, 0, 0, "CASE", TGSI_OPCODE_CASE },
+ { 0, 0, 0, 0, 0, 0, "DEFAULT", TGSI_OPCODE_DEFAULT },
+ { 0, 0, 0, 0, 0, 0, "ENDSWITCH", TGSI_OPCODE_ENDSWITCH }
};
const struct tgsi_opcode_info *
diff --git a/src/gallium/auxiliary/tgsi/tgsi_iterate.c b/src/gallium/auxiliary/tgsi/tgsi_iterate.c
index d88c2558d81..0ba5fe48419 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_iterate.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_iterate.c
@@ -39,7 +39,6 @@ tgsi_iterate_shader(
return FALSE;
ctx->processor = parse.FullHeader.Processor;
- ctx->version = parse.FullVersion.Version;
if (ctx->prolog)
if (!ctx->prolog( ctx ))
@@ -67,6 +66,12 @@ tgsi_iterate_shader(
goto fail;
break;
+ case TGSI_TOKEN_TYPE_PROPERTY:
+ if (ctx->iterate_property)
+ if (!ctx->iterate_property( ctx, &parse.FullToken.FullProperty ))
+ goto fail;
+ break;
+
default:
assert( 0 );
}
diff --git a/src/gallium/auxiliary/tgsi/tgsi_iterate.h b/src/gallium/auxiliary/tgsi/tgsi_iterate.h
index ec7b85bf63d..8d67f22c429 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_iterate.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_iterate.h
@@ -57,11 +57,15 @@ struct tgsi_iterate_context
struct tgsi_full_immediate *imm );
boolean
+ (* iterate_property)(
+ struct tgsi_iterate_context *ctx,
+ struct tgsi_full_property *prop );
+
+ boolean
(* epilog)(
struct tgsi_iterate_context *ctx );
struct tgsi_processor processor;
- struct tgsi_version version;
};
boolean
diff --git a/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h b/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h
index b34263da489..e4af15c156f 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_opcode_tmp.h
@@ -124,7 +124,6 @@ OP11(I2F)
OP11(NOT)
OP11(TRUNC)
OP12(SHL)
-OP12(SHR)
OP12(AND)
OP12(OR)
OP12(MOD)
@@ -146,6 +145,28 @@ OP01(IFC)
OP01(BREAKC)
OP01(KIL)
OP00(END)
+OP11(F2I)
+OP12(IDIV)
+OP12(IMAX)
+OP12(IMIN)
+OP11(INEG)
+OP12(ISGE)
+OP12(ISHR)
+OP12(ISLT)
+OP11(F2U)
+OP11(U2F)
+OP12(UADD)
+OP12(UDIV)
+OP13(UMAD)
+OP12(UMAX)
+OP12(UMIN)
+OP12(UMOD)
+OP12(UMUL)
+OP12(USEQ)
+OP12(USGE)
+OP12(USHR)
+OP12(USLT)
+OP12(USNE)
#undef OP00
diff --git a/src/gallium/auxiliary/tgsi/tgsi_parse.c b/src/gallium/auxiliary/tgsi/tgsi_parse.c
index 83f9df1183e..8c7062d850c 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_parse.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_parse.c
@@ -28,44 +28,23 @@
#include "util/u_debug.h"
#include "pipe/p_shader_tokens.h"
#include "tgsi_parse.h"
-#include "tgsi_build.h"
#include "util/u_memory.h"
-void
-tgsi_full_token_init(
- union tgsi_full_token *full_token )
-{
- full_token->Token.Type = TGSI_TOKEN_TYPE_DECLARATION;
-}
-
-void
-tgsi_full_token_free(
- union tgsi_full_token *full_token )
-{
-}
-
unsigned
tgsi_parse_init(
struct tgsi_parse_context *ctx,
const struct tgsi_token *tokens )
{
- ctx->FullVersion.Version = *(struct tgsi_version *) &tokens[0];
- if( ctx->FullVersion.Version.MajorVersion > 1 ) {
- return TGSI_PARSE_ERROR;
- }
-
- ctx->FullHeader.Header = *(struct tgsi_header *) &tokens[1];
+ ctx->FullHeader.Header = *(struct tgsi_header *) &tokens[0];
if( ctx->FullHeader.Header.HeaderSize >= 2 ) {
- ctx->FullHeader.Processor = *(struct tgsi_processor *) &tokens[2];
+ ctx->FullHeader.Processor = *(struct tgsi_processor *) &tokens[1];
}
else {
- ctx->FullHeader.Processor = tgsi_default_processor();
+ return TGSI_PARSE_ERROR;
}
ctx->Tokens = tokens;
- ctx->Position = 1 + ctx->FullHeader.Header.HeaderSize;
-
- tgsi_full_token_init( &ctx->FullToken );
+ ctx->Position = ctx->FullHeader.Header.HeaderSize;
return TGSI_PARSE_OK;
}
@@ -74,7 +53,6 @@ void
tgsi_parse_free(
struct tgsi_parse_context *ctx )
{
- tgsi_full_token_free( &ctx->FullToken );
}
boolean
@@ -82,7 +60,7 @@ tgsi_parse_end_of_tokens(
struct tgsi_parse_context *ctx )
{
return ctx->Position >=
- 1 + ctx->FullHeader.Header.HeaderSize + ctx->FullHeader.Header.BodySize;
+ ctx->FullHeader.Header.HeaderSize + ctx->FullHeader.Header.BodySize;
}
@@ -119,9 +97,6 @@ tgsi_parse_token(
struct tgsi_token token;
unsigned i;
- tgsi_full_token_free( &ctx->FullToken );
- tgsi_full_token_init( &ctx->FullToken );
-
next_token( ctx, &token );
switch( token.Type ) {
@@ -129,10 +104,10 @@ tgsi_parse_token(
{
struct tgsi_full_declaration *decl = &ctx->FullToken.FullDeclaration;
- *decl = tgsi_default_full_declaration();
+ memset(decl, 0, sizeof *decl);
copy_token(&decl->Declaration, &token);
- next_token( ctx, &decl->DeclarationRange );
+ next_token( ctx, &decl->Range );
if( decl->Declaration.Semantic ) {
next_token( ctx, &decl->Semantic );
@@ -144,18 +119,29 @@ tgsi_parse_token(
case TGSI_TOKEN_TYPE_IMMEDIATE:
{
struct tgsi_full_immediate *imm = &ctx->FullToken.FullImmediate;
+ uint imm_count;
- *imm = tgsi_default_full_immediate();
+ memset(imm, 0, sizeof *imm);
copy_token(&imm->Immediate, &token);
- assert( !imm->Immediate.Extended );
+
+ imm_count = imm->Immediate.NrTokens - 1;
switch (imm->Immediate.DataType) {
case TGSI_IMM_FLOAT32:
- {
- uint imm_count = imm->Immediate.NrTokens - 1;
- for (i = 0; i < imm_count; i++) {
- next_token(ctx, &imm->u[i]);
- }
+ for (i = 0; i < imm_count; i++) {
+ next_token(ctx, &imm->u[i].Float);
+ }
+ break;
+
+ case TGSI_IMM_UINT32:
+ for (i = 0; i < imm_count; i++) {
+ next_token(ctx, &imm->u[i].Uint);
+ }
+ break;
+
+ case TGSI_IMM_INT32:
+ for (i = 0; i < imm_count; i++) {
+ next_token(ctx, &imm->u[i].Int);
}
break;
@@ -169,137 +155,76 @@ tgsi_parse_token(
case TGSI_TOKEN_TYPE_INSTRUCTION:
{
struct tgsi_full_instruction *inst = &ctx->FullToken.FullInstruction;
- unsigned extended;
- *inst = tgsi_default_full_instruction();
+ memset(inst, 0, sizeof *inst);
copy_token(&inst->Instruction, &token);
- extended = inst->Instruction.Extended;
-
- while( extended ) {
- struct tgsi_src_register_ext token;
-
- next_token( ctx, &token );
- switch( token.Type ) {
- case TGSI_INSTRUCTION_EXT_TYPE_LABEL:
- copy_token(&inst->InstructionExtLabel, &token);
- break;
-
- case TGSI_INSTRUCTION_EXT_TYPE_TEXTURE:
- copy_token(&inst->InstructionExtTexture, &token);
- break;
-
- case TGSI_INSTRUCTION_EXT_TYPE_PREDICATE:
- copy_token(&inst->InstructionExtPredicate, &token);
- break;
+ if (inst->Instruction.Predicate) {
+ next_token(ctx, &inst->Predicate);
+ }
- default:
- assert( 0 );
- }
+ if (inst->Instruction.Label) {
+ next_token( ctx, &inst->Label);
+ }
- extended = token.Extended;
+ if (inst->Instruction.Texture) {
+ next_token( ctx, &inst->Texture);
}
assert( inst->Instruction.NumDstRegs <= TGSI_FULL_MAX_DST_REGISTERS );
for( i = 0; i < inst->Instruction.NumDstRegs; i++ ) {
- unsigned extended;
- next_token( ctx, &inst->FullDstRegisters[i].DstRegister );
+ next_token( ctx, &inst->Dst[i].Register );
/*
* No support for indirect or multi-dimensional addressing.
*/
- assert( !inst->FullDstRegisters[i].DstRegister.Dimension );
-
- extended = inst->FullDstRegisters[i].DstRegister.Extended;
-
- while( extended ) {
- struct tgsi_src_register_ext token;
+ assert( !inst->Dst[i].Register.Dimension );
- next_token( ctx, &token );
-
- switch( token.Type ) {
- case TGSI_DST_REGISTER_EXT_TYPE_MODULATE:
- copy_token(&inst->FullDstRegisters[i].DstRegisterExtModulate,
- &token);
- break;
-
- default:
- assert( 0 );
- }
-
- extended = token.Extended;
- }
-
- if( inst->FullDstRegisters[i].DstRegister.Indirect ) {
- next_token( ctx, &inst->FullDstRegisters[i].DstRegisterInd );
+ if( inst->Dst[i].Register.Indirect ) {
+ next_token( ctx, &inst->Dst[i].Indirect );
/*
* No support for indirect or multi-dimensional addressing.
*/
- assert( !inst->FullDstRegisters[i].DstRegisterInd.Indirect );
- assert( !inst->FullDstRegisters[i].DstRegisterInd.Dimension );
- assert( !inst->FullDstRegisters[i].DstRegisterInd.Extended );
+ assert( !inst->Dst[i].Indirect.Dimension );
+ assert( !inst->Dst[i].Indirect.Indirect );
}
}
assert( inst->Instruction.NumSrcRegs <= TGSI_FULL_MAX_SRC_REGISTERS );
for( i = 0; i < inst->Instruction.NumSrcRegs; i++ ) {
- unsigned extended;
-
- next_token( ctx, &inst->FullSrcRegisters[i].SrcRegister );
-
- extended = inst->FullSrcRegisters[i].SrcRegister.Extended;
-
- while( extended ) {
- struct tgsi_src_register_ext token;
- next_token( ctx, &token );
+ next_token( ctx, &inst->Src[i].Register );
- switch( token.Type ) {
- case TGSI_SRC_REGISTER_EXT_TYPE_MOD:
- copy_token(&inst->FullSrcRegisters[i].SrcRegisterExtMod,
- &token);
- break;
-
- default:
- assert( 0 );
- }
-
- extended = token.Extended;
- }
-
- if( inst->FullSrcRegisters[i].SrcRegister.Indirect ) {
- next_token( ctx, &inst->FullSrcRegisters[i].SrcRegisterInd );
+ if( inst->Src[i].Register.Indirect ) {
+ next_token( ctx, &inst->Src[i].Indirect );
/*
* No support for indirect or multi-dimensional addressing.
*/
- assert( !inst->FullSrcRegisters[i].SrcRegisterInd.Indirect );
- assert( !inst->FullSrcRegisters[i].SrcRegisterInd.Dimension );
- assert( !inst->FullSrcRegisters[i].SrcRegisterInd.Extended );
+ assert( !inst->Src[i].Indirect.Indirect );
+ assert( !inst->Src[i].Indirect.Dimension );
}
- if( inst->FullSrcRegisters[i].SrcRegister.Dimension ) {
- next_token( ctx, &inst->FullSrcRegisters[i].SrcRegisterDim );
+ if( inst->Src[i].Register.Dimension ) {
+ next_token( ctx, &inst->Src[i].Dimension );
/*
* No support for multi-dimensional addressing.
*/
- assert( !inst->FullSrcRegisters[i].SrcRegisterDim.Dimension );
- assert( !inst->FullSrcRegisters[i].SrcRegisterDim.Extended );
+ assert( !inst->Src[i].Dimension.Dimension );
- if( inst->FullSrcRegisters[i].SrcRegisterDim.Indirect ) {
- next_token( ctx, &inst->FullSrcRegisters[i].SrcRegisterDimInd );
+ if( inst->Src[i].Dimension.Indirect ) {
+ next_token( ctx, &inst->Src[i].DimIndirect );
/*
* No support for indirect or multi-dimensional addressing.
*/
- assert( !inst->FullSrcRegisters[i].SrcRegisterInd.Indirect );
- assert( !inst->FullSrcRegisters[i].SrcRegisterInd.Dimension );
- assert( !inst->FullSrcRegisters[i].SrcRegisterInd.Extended );
+ assert( !inst->Src[i].Indirect.Indirect );
+ assert( !inst->Src[i].Indirect.Dimension );
}
}
}
@@ -307,6 +232,22 @@ tgsi_parse_token(
break;
}
+ case TGSI_TOKEN_TYPE_PROPERTY:
+ {
+ struct tgsi_full_property *prop = &ctx->FullToken.FullProperty;
+ uint prop_count;
+
+ memset(prop, 0, sizeof *prop);
+ copy_token(&prop->Property, &token);
+
+ prop_count = prop->Property.NrTokens - 1;
+ for (i = 0; i < prop_count; i++) {
+ next_token(ctx, &prop->u[i]);
+ }
+
+ break;
+ }
+
default:
assert( 0 );
}
@@ -319,8 +260,7 @@ tgsi_num_tokens(const struct tgsi_token *tokens)
struct tgsi_parse_context ctx;
if (tgsi_parse_init(&ctx, tokens) == TGSI_PARSE_OK) {
unsigned len = (ctx.FullHeader.Header.HeaderSize +
- ctx.FullHeader.Header.BodySize +
- 1);
+ ctx.FullHeader.Header.BodySize);
return len;
}
return 0;
diff --git a/src/gallium/auxiliary/tgsi/tgsi_parse.h b/src/gallium/auxiliary/tgsi/tgsi_parse.h
index 76f1676d85d..439a57269b7 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_parse.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_parse.h
@@ -34,11 +34,6 @@
extern "C" {
#endif
-struct tgsi_full_version
-{
- struct tgsi_version Version;
-};
-
struct tgsi_full_header
{
struct tgsi_header Header;
@@ -47,24 +42,22 @@ struct tgsi_full_header
struct tgsi_full_dst_register
{
- struct tgsi_dst_register DstRegister;
- struct tgsi_src_register DstRegisterInd;
- struct tgsi_dst_register_ext_modulate DstRegisterExtModulate;
+ struct tgsi_dst_register Register;
+ struct tgsi_src_register Indirect;
};
struct tgsi_full_src_register
{
- struct tgsi_src_register SrcRegister;
- struct tgsi_src_register_ext_mod SrcRegisterExtMod;
- struct tgsi_src_register SrcRegisterInd;
- struct tgsi_dimension SrcRegisterDim;
- struct tgsi_src_register SrcRegisterDimInd;
+ struct tgsi_src_register Register;
+ struct tgsi_src_register Indirect;
+ struct tgsi_dimension Dimension;
+ struct tgsi_src_register DimIndirect;
};
struct tgsi_full_declaration
{
struct tgsi_declaration Declaration;
- struct tgsi_declaration_range DeclarationRange;
+ struct tgsi_declaration_range Range;
struct tgsi_declaration_semantic Semantic;
};
@@ -74,18 +67,23 @@ struct tgsi_full_immediate
union tgsi_immediate_data u[4];
};
+struct tgsi_full_property
+{
+ struct tgsi_property Property;
+ struct tgsi_property_data u[8];
+};
+
#define TGSI_FULL_MAX_DST_REGISTERS 2
#define TGSI_FULL_MAX_SRC_REGISTERS 4 /* TXD has 4 */
struct tgsi_full_instruction
{
struct tgsi_instruction Instruction;
- struct tgsi_instruction_ext_label InstructionExtLabel;
- struct tgsi_instruction_ext_texture InstructionExtTexture;
- struct tgsi_instruction_ext_predicate InstructionExtPredicate;
- struct tgsi_full_dst_register FullDstRegisters[TGSI_FULL_MAX_DST_REGISTERS];
- struct tgsi_full_src_register FullSrcRegisters[TGSI_FULL_MAX_SRC_REGISTERS];
- uint Flags; /**< user-defined usage */
+ struct tgsi_instruction_predicate Predicate;
+ struct tgsi_instruction_label Label;
+ struct tgsi_instruction_texture Texture;
+ struct tgsi_full_dst_register Dst[TGSI_FULL_MAX_DST_REGISTERS];
+ struct tgsi_full_src_register Src[TGSI_FULL_MAX_SRC_REGISTERS];
};
union tgsi_full_token
@@ -94,21 +92,13 @@ union tgsi_full_token
struct tgsi_full_declaration FullDeclaration;
struct tgsi_full_immediate FullImmediate;
struct tgsi_full_instruction FullInstruction;
+ struct tgsi_full_property FullProperty;
};
-void
-tgsi_full_token_init(
- union tgsi_full_token *full_token );
-
-void
-tgsi_full_token_free(
- union tgsi_full_token *full_token );
-
struct tgsi_parse_context
{
const struct tgsi_token *Tokens;
unsigned Position;
- struct tgsi_full_version FullVersion;
struct tgsi_full_header FullHeader;
union tgsi_full_token FullToken;
};
diff --git a/src/gallium/auxiliary/tgsi/tgsi_ppc.c b/src/gallium/auxiliary/tgsi/tgsi_ppc.c
index 617fd7f6be1..ad553c71a57 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_ppc.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_ppc.c
@@ -51,7 +51,8 @@
* Since it's pretty much impossible to form PPC vector immediates, load
* them from memory here:
*/
-const float ppc_builtin_constants[] ALIGN16_ATTRIB = {
+PIPE_ALIGN_VAR(16) const float
+ppc_builtin_constants[] = {
1.0f, -128.0f, 128.0, 0.0
};
@@ -60,7 +61,7 @@ const float ppc_builtin_constants[] ALIGN16_ATTRIB = {
for (CHAN = 0; CHAN < NUM_CHANNELS; CHAN++)
#define IS_DST0_CHANNEL_ENABLED( INST, CHAN )\
- ((INST).FullDstRegisters[0].DstRegister.WriteMask & (1 << (CHAN)))
+ ((INST).Dst[0].Register.WriteMask & (1 << (CHAN)))
#define IF_IS_DST0_CHANNEL_ENABLED( INST, CHAN )\
if (IS_DST0_CHANNEL_ENABLED( INST, CHAN ))
@@ -156,8 +157,8 @@ init_gen_context(struct gen_context *gen, struct ppc_function *func)
static boolean
is_ppc_vec_temporary(const struct tgsi_full_src_register *reg)
{
- return (reg->SrcRegister.File == TGSI_FILE_TEMPORARY &&
- reg->SrcRegister.Index < MAX_PPC_TEMPS);
+ return (reg->Register.File == TGSI_FILE_TEMPORARY &&
+ reg->Register.Index < MAX_PPC_TEMPS);
}
@@ -167,8 +168,8 @@ is_ppc_vec_temporary(const struct tgsi_full_src_register *reg)
static boolean
is_ppc_vec_temporary_dst(const struct tgsi_full_dst_register *reg)
{
- return (reg->DstRegister.File == TGSI_FILE_TEMPORARY &&
- reg->DstRegister.Index < MAX_PPC_TEMPS);
+ return (reg->Register.File == TGSI_FILE_TEMPORARY &&
+ reg->Register.Index < MAX_PPC_TEMPS);
}
@@ -291,10 +292,11 @@ emit_fetch(struct gen_context *gen,
case TGSI_SWIZZLE_Y:
case TGSI_SWIZZLE_Z:
case TGSI_SWIZZLE_W:
- switch (reg->SrcRegister.File) {
+ switch (reg->Register.File) {
case TGSI_FILE_INPUT:
+ case TGSI_FILE_SYSTEM_VALUE:
{
- int offset = (reg->SrcRegister.Index * 4 + swizzle) * 16;
+ int offset = (reg->Register.Index * 4 + swizzle) * 16;
int offset_reg = emit_li_offset(gen, offset);
dst_vec = ppc_allocate_vec_register(gen->f);
ppc_lvx(gen->f, dst_vec, gen->inputs_reg, offset_reg);
@@ -303,11 +305,11 @@ emit_fetch(struct gen_context *gen,
case TGSI_FILE_TEMPORARY:
if (is_ppc_vec_temporary(reg)) {
/* use PPC vec register */
- dst_vec = gen->temps_map[reg->SrcRegister.Index][swizzle];
+ dst_vec = gen->temps_map[reg->Register.Index][swizzle];
}
else {
/* use memory-based temp register "file" */
- int offset = (reg->SrcRegister.Index * 4 + swizzle) * 16;
+ int offset = (reg->Register.Index * 4 + swizzle) * 16;
int offset_reg = emit_li_offset(gen, offset);
dst_vec = ppc_allocate_vec_register(gen->f);
ppc_lvx(gen->f, dst_vec, gen->temps_reg, offset_reg);
@@ -315,7 +317,7 @@ emit_fetch(struct gen_context *gen,
break;
case TGSI_FILE_IMMEDIATE:
{
- int offset = (reg->SrcRegister.Index * 4 + swizzle) * 4;
+ int offset = (reg->Register.Index * 4 + swizzle) * 4;
int offset_reg = emit_li_offset(gen, offset);
dst_vec = ppc_allocate_vec_register(gen->f);
/* Load 4-byte word into vector register.
@@ -331,7 +333,7 @@ emit_fetch(struct gen_context *gen,
break;
case TGSI_FILE_CONSTANT:
{
- int offset = (reg->SrcRegister.Index * 4 + swizzle) * 4;
+ int offset = (reg->Register.Index * 4 + swizzle) * 4;
int offset_reg = emit_li_offset(gen, offset);
dst_vec = ppc_allocate_vec_register(gen->f);
/* Load 4-byte word into vector register.
@@ -404,9 +406,9 @@ equal_src_locs(const struct tgsi_full_src_register *a, uint chan_a,
{
int swz_a, swz_b;
int sign_a, sign_b;
- if (a->SrcRegister.File != b->SrcRegister.File)
+ if (a->Register.File != b->Register.File)
return FALSE;
- if (a->SrcRegister.Index != b->SrcRegister.Index)
+ if (a->Register.Index != b->Register.Index)
return FALSE;
swz_a = tgsi_util_get_full_src_register_swizzle(a, chan_a);
swz_b = tgsi_util_get_full_src_register_swizzle(b, chan_b);
@@ -431,7 +433,7 @@ get_src_vec(struct gen_context *gen,
struct tgsi_full_instruction *inst, int src_reg, uint chan)
{
const const struct tgsi_full_src_register *src =
- &inst->FullSrcRegisters[src_reg];
+ &inst->Src[src_reg];
int vec;
uint i;
@@ -482,10 +484,10 @@ get_dst_vec(struct gen_context *gen,
const struct tgsi_full_instruction *inst,
unsigned chan_index)
{
- const struct tgsi_full_dst_register *reg = &inst->FullDstRegisters[0];
+ const struct tgsi_full_dst_register *reg = &inst->Dst[0];
if (is_ppc_vec_temporary_dst(reg)) {
- int vec = gen->temps_map[reg->DstRegister.Index][chan_index];
+ int vec = gen->temps_map[reg->Register.Index][chan_index];
return vec;
}
else {
@@ -505,12 +507,12 @@ emit_store(struct gen_context *gen,
unsigned chan_index,
boolean free_vec)
{
- const struct tgsi_full_dst_register *reg = &inst->FullDstRegisters[0];
+ const struct tgsi_full_dst_register *reg = &inst->Dst[0];
- switch (reg->DstRegister.File) {
+ switch (reg->Register.File) {
case TGSI_FILE_OUTPUT:
{
- int offset = (reg->DstRegister.Index * 4 + chan_index) * 16;
+ int offset = (reg->Register.Index * 4 + chan_index) * 16;
int offset_reg = emit_li_offset(gen, offset);
ppc_stvx(gen->f, src_vec, gen->outputs_reg, offset_reg);
}
@@ -518,14 +520,14 @@ emit_store(struct gen_context *gen,
case TGSI_FILE_TEMPORARY:
if (is_ppc_vec_temporary_dst(reg)) {
if (!free_vec) {
- int dst_vec = gen->temps_map[reg->DstRegister.Index][chan_index];
+ int dst_vec = gen->temps_map[reg->Register.Index][chan_index];
if (dst_vec != src_vec)
ppc_vmove(gen->f, dst_vec, src_vec);
}
free_vec = FALSE;
}
else {
- int offset = (reg->DstRegister.Index * 4 + chan_index) * 16;
+ int offset = (reg->Register.Index * 4 + chan_index) * 16;
int offset_reg = emit_li_offset(gen, offset);
ppc_stvx(gen->f, src_vec, gen->temps_reg, offset_reg);
}
@@ -535,7 +537,7 @@ emit_store(struct gen_context *gen,
emit_addrs(
func,
xmm,
- reg->DstRegister.Index,
+ reg->Register.Index,
chan_index );
break;
#endif
@@ -1173,13 +1175,14 @@ emit_declaration(
struct ppc_function *func,
struct tgsi_full_declaration *decl )
{
- if( decl->Declaration.File == TGSI_FILE_INPUT ) {
+ if( decl->Declaration.File == TGSI_FILE_INPUT ||
+ decl->Declaration.File == TGSI_FILE_SYSTEM_VALUE ) {
#if 0
unsigned first, last, mask;
unsigned i, j;
- first = decl->DeclarationRange.First;
- last = decl->DeclarationRange.Last;
+ first = decl->Range.First;
+ last = decl->Range.Last;
mask = decl->Declaration.UsageMask;
for( i = first; i <= last; i++ ) {
@@ -1339,6 +1342,9 @@ tgsi_emit_ppc(const struct tgsi_token *tokens,
}
break;
+ case TGSI_TOKEN_TYPE_PROPERTY:
+ break;
+
default:
ok = 0;
assert( 0 );
diff --git a/src/gallium/auxiliary/tgsi/tgsi_sanity.c b/src/gallium/auxiliary/tgsi/tgsi_sanity.c
index 36e27ea52f4..7f1c8e5dd68 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_sanity.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_sanity.c
@@ -26,32 +26,112 @@
**************************************************************************/
#include "util/u_debug.h"
+#include "util/u_memory.h"
+#include "util/u_prim.h"
+#include "cso_cache/cso_hash.h"
#include "tgsi_sanity.h"
#include "tgsi_info.h"
#include "tgsi_iterate.h"
-typedef uint reg_flag;
-
-#define BITS_IN_REG_FLAG (sizeof( reg_flag ) * 8)
-
-#define MAX_REGISTERS 256
-#define MAX_REG_FLAGS ((MAX_REGISTERS + BITS_IN_REG_FLAG - 1) / BITS_IN_REG_FLAG)
+typedef struct {
+ uint file : 28;
+ /* max 2 dimensions */
+ uint dimensions : 4;
+ uint indices[2];
+} scan_register;
struct sanity_check_ctx
{
struct tgsi_iterate_context iter;
+ struct cso_hash *regs_decl;
+ struct cso_hash *regs_used;
+ struct cso_hash *regs_ind_used;
- reg_flag regs_decl[TGSI_FILE_COUNT][MAX_REG_FLAGS];
- reg_flag regs_used[TGSI_FILE_COUNT][MAX_REG_FLAGS];
- boolean regs_ind_used[TGSI_FILE_COUNT];
uint num_imms;
uint num_instructions;
uint index_of_END;
uint errors;
uint warnings;
+ uint implied_array_size;
};
+static INLINE unsigned
+scan_register_key(const scan_register *reg)
+{
+ unsigned key = reg->file;
+ key |= (reg->indices[0] << 4);
+ key |= (reg->indices[1] << 18);
+
+ return key;
+}
+
+static void
+fill_scan_register1d(scan_register *reg,
+ uint file, uint index)
+{
+ reg->file = file;
+ reg->dimensions = 1;
+ reg->indices[0] = index;
+ reg->indices[1] = 0;
+}
+
+static void
+fill_scan_register2d(scan_register *reg,
+ uint file, uint index1, uint index2)
+{
+ reg->file = file;
+ reg->dimensions = 2;
+ reg->indices[0] = index1;
+ reg->indices[1] = index2;
+}
+
+static void
+scan_register_dst(scan_register *reg,
+ struct tgsi_full_dst_register *dst)
+{
+ fill_scan_register1d(reg,
+ dst->Register.File,
+ dst->Register.Index);
+}
+
+static void
+scan_register_src(scan_register *reg,
+ struct tgsi_full_src_register *src)
+{
+ if (src->Register.Dimension) {
+ /*FIXME: right now we don't support indirect
+ * multidimensional addressing */
+ debug_assert(!src->Dimension.Indirect);
+ fill_scan_register2d(reg,
+ src->Register.File,
+ src->Register.Index,
+ src->Dimension.Index);
+ } else {
+ fill_scan_register1d(reg,
+ src->Register.File,
+ src->Register.Index);
+ }
+}
+
+static scan_register *
+create_scan_register_src(struct tgsi_full_src_register *src)
+{
+ scan_register *reg = MALLOC(sizeof(scan_register));
+ scan_register_src(reg, src);
+
+ return reg;
+}
+
+static scan_register *
+create_scan_register_dst(struct tgsi_full_dst_register *dst)
+{
+ scan_register *reg = MALLOC(sizeof(scan_register));
+ scan_register_dst(reg, dst);
+
+ return reg;
+}
+
static void
report_error(
struct sanity_check_ctx *ctx,
@@ -99,12 +179,12 @@ check_file_name(
static boolean
is_register_declared(
struct sanity_check_ctx *ctx,
- uint file,
- int index )
+ const scan_register *reg)
{
- assert( index >= 0 && index < MAX_REGISTERS );
-
- return (ctx->regs_decl[file][index / BITS_IN_REG_FLAG] & (1 << (index % BITS_IN_REG_FLAG))) ? TRUE : FALSE;
+ void *data = cso_hash_find_data_from_template(
+ ctx->regs_decl, scan_register_key(reg),
+ (void*)reg, sizeof(scan_register));
+ return data ? TRUE : FALSE;
}
static boolean
@@ -112,23 +192,37 @@ is_any_register_declared(
struct sanity_check_ctx *ctx,
uint file )
{
- uint i;
+ struct cso_hash_iter iter =
+ cso_hash_first_node(ctx->regs_decl);
- for (i = 0; i < MAX_REG_FLAGS; i++)
- if (ctx->regs_decl[file][i])
+ while (!cso_hash_iter_is_null(iter)) {
+ scan_register *reg = (scan_register *)cso_hash_iter_data(iter);
+ if (reg->file == file)
return TRUE;
+ iter = cso_hash_iter_next(iter);
+ }
+
return FALSE;
}
static boolean
is_register_used(
struct sanity_check_ctx *ctx,
- uint file,
- int index )
+ scan_register *reg)
{
- assert( index < MAX_REGISTERS );
+ void *data = cso_hash_find_data_from_template(
+ ctx->regs_used, scan_register_key(reg),
+ reg, sizeof(scan_register));
+ return data ? TRUE : FALSE;
+}
- return (ctx->regs_used[file][index / BITS_IN_REG_FLAG] & (1 << (index % BITS_IN_REG_FLAG))) ? TRUE : FALSE;
+
+static boolean
+is_ind_register_used(
+ struct sanity_check_ctx *ctx,
+ scan_register *reg)
+{
+ return cso_hash_contains(ctx->regs_ind_used, reg->file);
}
static const char *file_names[TGSI_FILE_COUNT] =
@@ -148,31 +242,42 @@ static const char *file_names[TGSI_FILE_COUNT] =
static boolean
check_register_usage(
struct sanity_check_ctx *ctx,
- uint file,
- int index,
+ scan_register *reg,
const char *name,
boolean indirect_access )
{
- if (!check_file_name( ctx, file ))
+ if (!check_file_name( ctx, reg->file )) {
+ FREE(reg);
return FALSE;
+ }
if (indirect_access) {
/* Note that 'index' is an offset relative to the value of the
- * address register. No range checking done here.
- */
- if (!is_any_register_declared( ctx, file ))
- report_error( ctx, "%s: Undeclared %s register", file_names[file], name );
- ctx->regs_ind_used[file] = TRUE;
+ * address register. No range checking done here.*/
+ reg->indices[0] = 0;
+ reg->indices[1] = 0;
+ if (!is_any_register_declared( ctx, reg->file ))
+ report_error( ctx, "%s: Undeclared %s register", file_names[reg->file], name );
+ if (!is_ind_register_used(ctx, reg))
+ cso_hash_insert(ctx->regs_ind_used, reg->file, reg);
+ else
+ FREE(reg);
}
else {
- if (index < 0 || index >= MAX_REGISTERS) {
- report_error( ctx, "%s[%d]: Invalid %s index", file_names[file], index, name );
- return FALSE;
+ if (!is_register_declared( ctx, reg )) {
+ if (reg->dimensions == 2) {
+ report_error( ctx, "%s[%d][%d]: Undeclared %s register", file_names[reg->file],
+ reg->indices[0], reg->indices[1], name );
+ }
+ else {
+ report_error( ctx, "%s[%d]: Undeclared %s register", file_names[reg->file],
+ reg->indices[0], name );
+ }
}
-
- if (!is_register_declared( ctx, file, index ))
- report_error( ctx, "%s[%d]: Undeclared %s register", file_names[file], index, name );
- ctx->regs_used[file][index / BITS_IN_REG_FLAG] |= (1 << (index % BITS_IN_REG_FLAG));
+ if (!is_register_used( ctx, reg ))
+ cso_hash_insert(ctx->regs_used, scan_register_key(reg), reg);
+ else
+ FREE(reg);
}
return TRUE;
}
@@ -210,43 +315,43 @@ iter_instruction(
* Mark the registers as used.
*/
for (i = 0; i < inst->Instruction.NumDstRegs; i++) {
+ scan_register *reg = create_scan_register_dst(&inst->Dst[i]);
check_register_usage(
ctx,
- inst->FullDstRegisters[i].DstRegister.File,
- inst->FullDstRegisters[i].DstRegister.Index,
+ reg,
"destination",
FALSE );
}
for (i = 0; i < inst->Instruction.NumSrcRegs; i++) {
+ scan_register *reg = create_scan_register_src(&inst->Src[i]);
check_register_usage(
ctx,
- inst->FullSrcRegisters[i].SrcRegister.File,
- inst->FullSrcRegisters[i].SrcRegister.Index,
+ reg,
"source",
- (boolean)inst->FullSrcRegisters[i].SrcRegister.Indirect );
- if (inst->FullSrcRegisters[i].SrcRegister.Indirect) {
- uint file;
- int index;
-
- file = inst->FullSrcRegisters[i].SrcRegisterInd.File;
- index = inst->FullSrcRegisters[i].SrcRegisterInd.Index;
+ (boolean)inst->Src[i].Register.Indirect );
+ if (inst->Src[i].Register.Indirect) {
+ scan_register *ind_reg = MALLOC(sizeof(scan_register));
+
+ fill_scan_register1d(ind_reg,
+ inst->Src[i].Indirect.File,
+ inst->Src[i].Indirect.Index);
+ if (!(reg->file == TGSI_FILE_ADDRESS || reg->file == TGSI_FILE_LOOP) ||
+ reg->indices[0] != 0) {
+ report_warning(ctx, "Indirect register neither ADDR[0] nor LOOP[0]");
+ }
check_register_usage(
ctx,
- file,
- index,
+ reg,
"indirect",
FALSE );
- if (!(file == TGSI_FILE_ADDRESS || file == TGSI_FILE_LOOP) || index != 0) {
- report_warning(ctx, "Indirect register neither ADDR[0] nor LOOP[0]");
- }
}
}
switch (inst->Instruction.Opcode) {
case TGSI_OPCODE_BGNFOR:
case TGSI_OPCODE_ENDFOR:
- if (inst->FullDstRegisters[0].DstRegister.File != TGSI_FILE_LOOP ||
- inst->FullDstRegisters[0].DstRegister.Index != 0) {
+ if (inst->Dst[0].Register.File != TGSI_FILE_LOOP ||
+ inst->Dst[0].Register.Index != 0) {
report_error(ctx, "Destination register must be LOOP[0]");
}
break;
@@ -254,8 +359,8 @@ iter_instruction(
switch (inst->Instruction.Opcode) {
case TGSI_OPCODE_BGNFOR:
- if (inst->FullSrcRegisters[0].SrcRegister.File != TGSI_FILE_CONSTANT &&
- inst->FullSrcRegisters[0].SrcRegister.File != TGSI_FILE_IMMEDIATE) {
+ if (inst->Src[0].Register.File != TGSI_FILE_CONSTANT &&
+ inst->Src[0].Register.File != TGSI_FILE_IMMEDIATE) {
report_error(ctx, "Source register file must be either CONST or IMM");
}
break;
@@ -266,6 +371,19 @@ iter_instruction(
return TRUE;
}
+static void
+check_and_declare(struct sanity_check_ctx *ctx,
+ scan_register *reg)
+{
+ if (is_register_declared( ctx, reg))
+ report_error( ctx, "%s[%u]: The same register declared more than once",
+ file_names[reg->file], reg->indices[0] );
+ cso_hash_insert(ctx->regs_decl,
+ scan_register_key(reg),
+ reg);
+}
+
+
static boolean
iter_declaration(
struct tgsi_iterate_context *iter,
@@ -286,10 +404,22 @@ iter_declaration(
file = decl->Declaration.File;
if (!check_file_name( ctx, file ))
return TRUE;
- for (i = decl->DeclarationRange.First; i <= decl->DeclarationRange.Last; i++) {
- if (is_register_declared( ctx, file, i ))
- report_error( ctx, "%s[%u]: The same register declared more than once", file_names[file], i );
- ctx->regs_decl[file][i / BITS_IN_REG_FLAG] |= (1 << (i % BITS_IN_REG_FLAG));
+ for (i = decl->Range.First; i <= decl->Range.Last; i++) {
+ /* declared TGSI_FILE_INPUT's for geometry processor
+ * have an implied second dimension */
+ if (file == TGSI_FILE_INPUT &&
+ ctx->iter.processor.Processor == TGSI_PROCESSOR_GEOMETRY) {
+ uint vert;
+ for (vert = 0; vert < ctx->implied_array_size; ++vert) {
+ scan_register *reg = MALLOC(sizeof(scan_register));
+ fill_scan_register2d(reg, file, vert, i);
+ check_and_declare(ctx, reg);
+ }
+ } else {
+ scan_register *reg = MALLOC(sizeof(scan_register));
+ fill_scan_register1d(reg, file, i);
+ check_and_declare(ctx, reg);
+ }
}
return TRUE;
@@ -301,8 +431,7 @@ iter_immediate(
struct tgsi_full_immediate *imm )
{
struct sanity_check_ctx *ctx = (struct sanity_check_ctx *) iter;
-
- assert( ctx->num_imms < MAX_REGISTERS );
+ scan_register *reg;
/* No immediates allowed after the first instruction.
*/
@@ -311,12 +440,16 @@ iter_immediate(
/* Mark the register as declared.
*/
- ctx->regs_decl[TGSI_FILE_IMMEDIATE][ctx->num_imms / BITS_IN_REG_FLAG] |= (1 << (ctx->num_imms % BITS_IN_REG_FLAG));
+ reg = MALLOC(sizeof(scan_register));
+ fill_scan_register1d(reg, TGSI_FILE_IMMEDIATE, ctx->num_imms);
+ cso_hash_insert(ctx->regs_decl, scan_register_key(reg), reg);
ctx->num_imms++;
/* Check data type validity.
*/
- if (imm->Immediate.DataType != TGSI_IMM_FLOAT32) {
+ if (imm->Immediate.DataType != TGSI_IMM_FLOAT32 &&
+ imm->Immediate.DataType != TGSI_IMM_UINT32 &&
+ imm->Immediate.DataType != TGSI_IMM_INT32) {
report_error( ctx, "(%u): Invalid immediate data type", imm->Immediate.DataType );
return TRUE;
}
@@ -324,12 +457,26 @@ iter_immediate(
return TRUE;
}
+
+static boolean
+iter_property(
+ struct tgsi_iterate_context *iter,
+ struct tgsi_full_property *prop )
+{
+ struct sanity_check_ctx *ctx = (struct sanity_check_ctx *) iter;
+
+ if (iter->processor.Processor == TGSI_PROCESSOR_GEOMETRY &&
+ prop->Property.PropertyName == TGSI_PROPERTY_GS_INPUT_PRIM) {
+ ctx->implied_array_size = u_vertices_per_prim(prop->u[0].Data);
+ }
+ return TRUE;
+}
+
static boolean
epilog(
struct tgsi_iterate_context *iter )
{
struct sanity_check_ctx *ctx = (struct sanity_check_ctx *) iter;
- uint file;
/* There must be an END instruction somewhere.
*/
@@ -339,13 +486,17 @@ epilog(
/* Check if all declared registers were used.
*/
- for (file = TGSI_FILE_NULL; file < TGSI_FILE_COUNT; file++) {
- uint i;
-
- for (i = 0; i < MAX_REGISTERS; i++) {
- if (is_register_declared( ctx, file, i ) && !is_register_used( ctx, file, i ) && !ctx->regs_ind_used[file]) {
- report_warning( ctx, "%s[%u]: Register never used", file_names[file], i );
+ {
+ struct cso_hash_iter iter =
+ cso_hash_first_node(ctx->regs_decl);
+
+ while (!cso_hash_iter_is_null(iter)) {
+ scan_register *reg = (scan_register *)cso_hash_iter_data(iter);
+ if (!is_register_used(ctx, reg) && !is_ind_register_used(ctx, reg)) {
+ report_warning( ctx, "%s[%u]: Register never used",
+ file_names[reg->file], reg->indices[0] );
}
+ iter = cso_hash_iter_next(iter);
}
}
@@ -357,6 +508,19 @@ epilog(
return TRUE;
}
+static void
+regs_hash_destroy(struct cso_hash *hash)
+{
+ struct cso_hash_iter iter = cso_hash_first_node(hash);
+ while (!cso_hash_iter_is_null(iter)) {
+ scan_register *reg = (scan_register *)cso_hash_iter_data(iter);
+ iter = cso_hash_erase(hash, iter);
+ assert(reg->file < TGSI_FILE_COUNT);
+ FREE(reg);
+ }
+ cso_hash_delete(hash);
+}
+
boolean
tgsi_sanity_check(
const struct tgsi_token *tokens )
@@ -367,20 +531,26 @@ tgsi_sanity_check(
ctx.iter.iterate_instruction = iter_instruction;
ctx.iter.iterate_declaration = iter_declaration;
ctx.iter.iterate_immediate = iter_immediate;
+ ctx.iter.iterate_property = iter_property;
ctx.iter.epilog = epilog;
- memset( ctx.regs_decl, 0, sizeof( ctx.regs_decl ) );
- memset( ctx.regs_used, 0, sizeof( ctx.regs_used ) );
- memset( ctx.regs_ind_used, 0, sizeof( ctx.regs_ind_used ) );
+ ctx.regs_decl = cso_hash_create();
+ ctx.regs_used = cso_hash_create();
+ ctx.regs_ind_used = cso_hash_create();
+
ctx.num_imms = 0;
ctx.num_instructions = 0;
ctx.index_of_END = ~0;
ctx.errors = 0;
ctx.warnings = 0;
+ ctx.implied_array_size = 0;
if (!tgsi_iterate_shader( tokens, &ctx.iter ))
return FALSE;
+ regs_hash_destroy(ctx.regs_decl);
+ regs_hash_destroy(ctx.regs_used);
+ regs_hash_destroy(ctx.regs_ind_used);
return ctx.errors == 0;
}
diff --git a/src/gallium/auxiliary/tgsi/tgsi_scan.c b/src/gallium/auxiliary/tgsi/tgsi_scan.c
index f9c16f1b6cb..a6cc773003a 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_scan.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_scan.c
@@ -35,7 +35,6 @@
#include "util/u_math.h"
-#include "tgsi/tgsi_build.h"
#include "tgsi/tgsi_parse.h"
#include "tgsi/tgsi_scan.h"
@@ -97,14 +96,15 @@ tgsi_scan_shader(const struct tgsi_token *tokens,
uint i;
for (i = 0; i < fullinst->Instruction.NumSrcRegs; i++) {
const struct tgsi_full_src_register *src =
- &fullinst->FullSrcRegisters[i];
- if (src->SrcRegister.File == TGSI_FILE_INPUT) {
- const int ind = src->SrcRegister.Index;
+ &fullinst->Src[i];
+ if (src->Register.File == TGSI_FILE_INPUT ||
+ src->Register.File == TGSI_FILE_SYSTEM_VALUE) {
+ const int ind = src->Register.Index;
if (info->input_semantic_name[ind] == TGSI_SEMANTIC_FOG) {
- if (src->SrcRegister.SwizzleX == TGSI_SWIZZLE_X) {
+ if (src->Register.SwizzleX == TGSI_SWIZZLE_X) {
info->uses_fogcoord = TRUE;
}
- else if (src->SrcRegister.SwizzleX == TGSI_SWIZZLE_Y) {
+ else if (src->Register.SwizzleX == TGSI_SWIZZLE_Y) {
info->uses_frontfacing = TRUE;
}
}
@@ -120,8 +120,8 @@ tgsi_scan_shader(const struct tgsi_token *tokens,
= &parse.FullToken.FullDeclaration;
const uint file = fulldecl->Declaration.File;
uint reg;
- for (reg = fulldecl->DeclarationRange.First;
- reg <= fulldecl->DeclarationRange.Last;
+ for (reg = fulldecl->Range.First;
+ reg <= fulldecl->Range.Last;
reg++) {
/* only first 32 regs will appear in this bitfield */
@@ -129,25 +129,29 @@ tgsi_scan_shader(const struct tgsi_token *tokens,
info->file_count[file]++;
info->file_max[file] = MAX2(info->file_max[file], (int)reg);
- if (file == TGSI_FILE_INPUT) {
- info->input_semantic_name[reg] = (ubyte)fulldecl->Semantic.SemanticName;
- info->input_semantic_index[reg] = (ubyte)fulldecl->Semantic.SemanticIndex;
+ if (file == TGSI_FILE_INPUT || file == TGSI_FILE_SYSTEM_VALUE) {
+ info->input_semantic_name[reg] = (ubyte)fulldecl->Semantic.Name;
+ info->input_semantic_index[reg] = (ubyte)fulldecl->Semantic.Index;
info->input_interpolate[reg] = (ubyte)fulldecl->Declaration.Interpolate;
info->num_inputs++;
}
else if (file == TGSI_FILE_OUTPUT) {
- info->output_semantic_name[reg] = (ubyte)fulldecl->Semantic.SemanticName;
- info->output_semantic_index[reg] = (ubyte)fulldecl->Semantic.SemanticIndex;
+ info->output_semantic_name[reg] = (ubyte)fulldecl->Semantic.Name;
+ info->output_semantic_index[reg] = (ubyte)fulldecl->Semantic.Index;
info->num_outputs++;
- }
- /* special case */
- if (procType == TGSI_PROCESSOR_FRAGMENT &&
- file == TGSI_FILE_OUTPUT &&
- fulldecl->Semantic.SemanticName == TGSI_SEMANTIC_POSITION) {
- info->writes_z = TRUE;
+ /* extra info for special outputs */
+ if (procType == TGSI_PROCESSOR_FRAGMENT &&
+ fulldecl->Semantic.Name == TGSI_SEMANTIC_POSITION) {
+ info->writes_z = TRUE;
+ }
+ if (procType == TGSI_PROCESSOR_VERTEX &&
+ fulldecl->Semantic.Name == TGSI_SEMANTIC_EDGEFLAG) {
+ info->writes_edgeflag = TRUE;
+ }
}
- }
+
+ }
}
break;
@@ -161,6 +165,19 @@ tgsi_scan_shader(const struct tgsi_token *tokens,
info->file_max[file] = MAX2(info->file_max[file], (int)reg);
}
break;
+ case TGSI_TOKEN_TYPE_PROPERTY:
+ {
+ const struct tgsi_full_property *fullprop
+ = &parse.FullToken.FullProperty;
+
+ info->properties[info->num_properties].name =
+ fullprop->Property.PropertyName;
+ memcpy(info->properties[info->num_properties].data,
+ fullprop->u, 8 * sizeof(unsigned));;
+
+ ++info->num_properties;
+ }
+ break;
default:
assert( 0 );
@@ -206,29 +223,26 @@ tgsi_is_passthrough_shader(const struct tgsi_token *tokens)
struct tgsi_full_instruction *fullinst =
&parse.FullToken.FullInstruction;
const struct tgsi_full_src_register *src =
- &fullinst->FullSrcRegisters[0];
+ &fullinst->Src[0];
const struct tgsi_full_dst_register *dst =
- &fullinst->FullDstRegisters[0];
+ &fullinst->Dst[0];
/* Do a whole bunch of checks for a simple move */
if (fullinst->Instruction.Opcode != TGSI_OPCODE_MOV ||
- src->SrcRegister.File != TGSI_FILE_INPUT ||
- dst->DstRegister.File != TGSI_FILE_OUTPUT ||
- src->SrcRegister.Index != dst->DstRegister.Index ||
-
- src->SrcRegister.Negate ||
- src->SrcRegisterExtMod.Negate ||
- src->SrcRegisterExtMod.Absolute ||
- src->SrcRegisterExtMod.Scale2X ||
- src->SrcRegisterExtMod.Bias ||
- src->SrcRegisterExtMod.Complement ||
-
- src->SrcRegister.SwizzleX != TGSI_SWIZZLE_X ||
- src->SrcRegister.SwizzleY != TGSI_SWIZZLE_Y ||
- src->SrcRegister.SwizzleZ != TGSI_SWIZZLE_Z ||
- src->SrcRegister.SwizzleW != TGSI_SWIZZLE_W ||
-
- dst->DstRegister.WriteMask != TGSI_WRITEMASK_XYZW)
+ (src->Register.File != TGSI_FILE_INPUT &&
+ src->Register.File != TGSI_FILE_SYSTEM_VALUE) ||
+ dst->Register.File != TGSI_FILE_OUTPUT ||
+ src->Register.Index != dst->Register.Index ||
+
+ src->Register.Negate ||
+ src->Register.Absolute ||
+
+ src->Register.SwizzleX != TGSI_SWIZZLE_X ||
+ src->Register.SwizzleY != TGSI_SWIZZLE_Y ||
+ src->Register.SwizzleZ != TGSI_SWIZZLE_Z ||
+ src->Register.SwizzleW != TGSI_SWIZZLE_W ||
+
+ dst->Register.WriteMask != TGSI_WRITEMASK_XYZW)
{
tgsi_parse_free(&parse);
return FALSE;
@@ -240,6 +254,8 @@ tgsi_is_passthrough_shader(const struct tgsi_token *tokens)
/* fall-through */
case TGSI_TOKEN_TYPE_IMMEDIATE:
/* fall-through */
+ case TGSI_TOKEN_TYPE_PROPERTY:
+ /* fall-through */
default:
; /* no-op */
}
diff --git a/src/gallium/auxiliary/tgsi/tgsi_scan.h b/src/gallium/auxiliary/tgsi/tgsi_scan.h
index 8a7ee0c7e4f..dae5376c24a 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_scan.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_scan.h
@@ -33,7 +33,6 @@
#include "pipe/p_state.h"
#include "pipe/p_shader_tokens.h"
-
/**
* Shader summary info
*/
@@ -58,11 +57,17 @@ struct tgsi_shader_info
uint opcode_count[TGSI_OPCODE_LAST]; /**< opcode histogram */
boolean writes_z; /**< does fragment shader write Z value? */
+ boolean writes_edgeflag; /**< vertex shader outputs edgeflag */
boolean uses_kill; /**< KIL or KILP instruction used? */
boolean uses_fogcoord; /**< fragment shader uses fog coord? */
boolean uses_frontfacing; /**< fragment shader uses front/back-face flag? */
-};
+ struct {
+ unsigned name;
+ unsigned data[8];
+ } properties[TGSI_PROPERTY_COUNT];
+ uint num_properties;
+};
extern void
tgsi_scan_shader(const struct tgsi_token *tokens,
diff --git a/src/gallium/auxiliary/tgsi/tgsi_sse2.c b/src/gallium/auxiliary/tgsi/tgsi_sse2.c
index 2c39defde50..a85cc4659e0 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_sse2.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_sse2.c
@@ -2,6 +2,7 @@
*
* Copyright 2007-2008 Tungsten Graphics, Inc., Cedar Park, Texas.
* All Rights Reserved.
+ * Copyright 2009-2010 VMware, Inc. All rights Reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the
@@ -58,7 +59,7 @@
for (CHAN = 0; CHAN < NUM_CHANNELS; CHAN++)
#define IS_DST0_CHANNEL_ENABLED( INST, CHAN )\
- ((INST).FullDstRegisters[0].DstRegister.WriteMask & (1 << (CHAN)))
+ ((INST).Dst[0].Register.WriteMask & (1 << (CHAN)))
#define IF_IS_DST0_CHANNEL_ENABLED( INST, CHAN )\
if (IS_DST0_CHANNEL_ENABLED( INST, CHAN ))
@@ -1267,31 +1268,32 @@ emit_fetch(
case TGSI_SWIZZLE_Y:
case TGSI_SWIZZLE_Z:
case TGSI_SWIZZLE_W:
- switch (reg->SrcRegister.File) {
+ switch (reg->Register.File) {
case TGSI_FILE_CONSTANT:
emit_const(
func,
xmm,
- reg->SrcRegister.Index,
+ reg->Register.Index,
swizzle,
- reg->SrcRegister.Indirect,
- reg->SrcRegisterInd.File,
- reg->SrcRegisterInd.Index );
+ reg->Register.Indirect,
+ reg->Indirect.File,
+ reg->Indirect.Index );
break;
case TGSI_FILE_IMMEDIATE:
emit_immediate(
func,
xmm,
- reg->SrcRegister.Index,
+ reg->Register.Index,
swizzle );
break;
case TGSI_FILE_INPUT:
+ case TGSI_FILE_SYSTEM_VALUE:
emit_inputf(
func,
xmm,
- reg->SrcRegister.Index,
+ reg->Register.Index,
swizzle );
break;
@@ -1299,7 +1301,7 @@ emit_fetch(
emit_tempf(
func,
xmm,
- reg->SrcRegister.Index,
+ reg->Register.Index,
swizzle );
break;
@@ -1331,7 +1333,7 @@ emit_fetch(
}
#define FETCH( FUNC, INST, XMM, INDEX, CHAN )\
- emit_fetch( FUNC, XMM, &(INST).FullSrcRegisters[INDEX], CHAN )
+ emit_fetch( FUNC, XMM, &(INST).Src[INDEX], CHAN )
/**
* Register store.
@@ -1371,12 +1373,12 @@ emit_store(
}
- switch( reg->DstRegister.File ) {
+ switch( reg->Register.File ) {
case TGSI_FILE_OUTPUT:
emit_output(
func,
xmm,
- reg->DstRegister.Index,
+ reg->Register.Index,
chan_index );
break;
@@ -1384,7 +1386,7 @@ emit_store(
emit_temps(
func,
xmm,
- reg->DstRegister.Index,
+ reg->Register.Index,
chan_index );
break;
@@ -1392,7 +1394,7 @@ emit_store(
emit_addrs(
func,
xmm,
- reg->DstRegister.Index,
+ reg->Register.Index,
chan_index );
break;
@@ -1402,7 +1404,7 @@ emit_store(
}
#define STORE( FUNC, INST, XMM, INDEX, CHAN )\
- emit_store( FUNC, XMM, &(INST).FullDstRegisters[INDEX], &(INST), CHAN )
+ emit_store( FUNC, XMM, &(INST).Dst[INDEX], &(INST), CHAN )
static void PIPE_CDECL
@@ -1417,13 +1419,13 @@ fetch_texel( struct tgsi_sampler **sampler,
sampler, *sampler,
store );
- debug_printf("lodbias %f\n", store[12]);
-
for (j = 0; j < 4; j++)
- debug_printf("sample %d texcoord %f %f\n",
+ debug_printf("sample %d texcoord %f %f %f lodbias %f\n",
j,
store[0+j],
- store[4+j]);
+ store[4+j],
+ store[8 + j],
+ store[12 + j]);
#endif
{
@@ -1432,7 +1434,8 @@ fetch_texel( struct tgsi_sampler **sampler,
&store[0], /* s */
&store[4], /* t */
&store[8], /* r */
- store[12], /* lodbias */
+ &store[12], /* lodbias */
+ tgsi_sampler_lod_bias,
rgba); /* results */
memcpy( store, rgba, 16 * sizeof(float));
@@ -1459,12 +1462,13 @@ emit_tex( struct x86_function *func,
boolean lodbias,
boolean projected)
{
- const uint unit = inst->FullSrcRegisters[1].SrcRegister.Index;
+ const uint unit = inst->Src[1].Register.Index;
struct x86_reg args[2];
unsigned count;
unsigned i;
- switch (inst->InstructionExtTexture.Texture) {
+ assert(inst->Instruction.Texture);
+ switch (inst->Texture.Texture) {
case TGSI_TEXTURE_1D:
count = 1;
break;
@@ -1719,15 +1723,15 @@ indirect_temp_reference(const struct tgsi_full_instruction *inst)
{
uint i;
for (i = 0; i < inst->Instruction.NumSrcRegs; i++) {
- const struct tgsi_full_src_register *reg = &inst->FullSrcRegisters[i];
- if (reg->SrcRegister.File == TGSI_FILE_TEMPORARY &&
- reg->SrcRegister.Indirect)
+ const struct tgsi_full_src_register *reg = &inst->Src[i];
+ if (reg->Register.File == TGSI_FILE_TEMPORARY &&
+ reg->Register.Indirect)
return TRUE;
}
for (i = 0; i < inst->Instruction.NumDstRegs; i++) {
- const struct tgsi_full_dst_register *reg = &inst->FullDstRegisters[i];
- if (reg->DstRegister.File == TGSI_FILE_TEMPORARY &&
- reg->DstRegister.Indirect)
+ const struct tgsi_full_dst_register *reg = &inst->Dst[i];
+ if (reg->Register.File == TGSI_FILE_TEMPORARY &&
+ reg->Register.Indirect)
return TRUE;
}
return FALSE;
@@ -2253,7 +2257,7 @@ emit_instruction(
case TGSI_OPCODE_KIL:
/* conditional kill */
- emit_kil( func, &inst->FullSrcRegisters[0] );
+ emit_kil( func, &inst->Src[0] );
break;
case TGSI_OPCODE_PK2H:
@@ -2514,7 +2518,7 @@ emit_instruction(
break;
case TGSI_OPCODE_TXL:
- emit_tex( func, inst, TRUE, FALSE );
+ return 0;
break;
case TGSI_OPCODE_TXP:
@@ -2586,7 +2590,7 @@ emit_instruction(
return 0;
break;
- case TGSI_OPCODE_SHR:
+ case TGSI_OPCODE_ISHR:
return 0;
break;
@@ -2642,12 +2646,13 @@ emit_declaration(
struct x86_function *func,
struct tgsi_full_declaration *decl )
{
- if( decl->Declaration.File == TGSI_FILE_INPUT ) {
+ if( decl->Declaration.File == TGSI_FILE_INPUT ||
+ decl->Declaration.File == TGSI_FILE_SYSTEM_VALUE ) {
unsigned first, last, mask;
unsigned i, j;
- first = decl->DeclarationRange.First;
- last = decl->DeclarationRange.Last;
+ first = decl->Range.First;
+ last = decl->Range.Last;
mask = decl->Declaration.UsageMask;
for( i = first; i <= last; i++ ) {
@@ -2961,6 +2966,9 @@ tgsi_emit_sse2(
num_immediates++;
}
break;
+ case TGSI_TOKEN_TYPE_PROPERTY:
+ /* we just ignore them for now */
+ break;
default:
ok = 0;
diff --git a/src/gallium/auxiliary/tgsi/tgsi_text.c b/src/gallium/auxiliary/tgsi/tgsi_text.c
index d2b03ffb2fc..9fcffeda368 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_text.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_text.c
@@ -27,6 +27,9 @@
#include "util/u_debug.h"
#include "util/u_memory.h"
+#include "util/u_prim.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_inlines.h"
#include "tgsi_text.h"
#include "tgsi_build.h"
#include "tgsi_info.h"
@@ -59,6 +62,23 @@ static boolean uprcase( char c )
return c;
}
+/*
+ * Ignore case of str1 and assume str1 is already uppercase.
+ * Return TRUE iff str1 and str2 are equal.
+ */
+static int
+streq_nocase_uprcase(const char *str1,
+ const char *str2)
+{
+ while (*str1 && *str2) {
+ if (*str1 != uprcase(*str2))
+ return FALSE;
+ str1++;
+ str2++;
+ }
+ return TRUE;
+}
+
static boolean str_match_no_case( const char **pcur, const char *str )
{
const char *cur = *pcur;
@@ -110,6 +130,20 @@ static boolean parse_uint( const char **pcur, uint *val )
return FALSE;
}
+static boolean parse_identifier( const char **pcur, char *ret )
+{
+ const char *cur = *pcur;
+ int i = 0;
+ if (is_alpha_underscore( cur )) {
+ ret[i++] = *cur++;
+ while (is_alpha_underscore( cur ))
+ ret[i++] = *cur++;
+ *pcur = cur;
+ return TRUE;
+ }
+ return FALSE;
+}
+
/* Parse floating point.
*/
static boolean parse_float( const char **pcur, float *val )
@@ -163,28 +197,43 @@ struct translate_ctx
struct tgsi_token *tokens_cur;
struct tgsi_token *tokens_end;
struct tgsi_header *header;
+ unsigned processor : 4;
+ int implied_array_size : 5;
};
static void report_error( struct translate_ctx *ctx, const char *msg )
{
- debug_printf( "\nError: %s", msg );
+ int line = 1;
+ int column = 1;
+ const char *itr = ctx->text;
+
+ while (itr != ctx->cur) {
+ if (*itr == '\n') {
+ column = 1;
+ ++line;
+ }
+ ++column;
+ ++itr;
+ }
+
+ debug_printf( "\nTGSI asm error: %s [%d : %d] \n", msg, line, column );
}
/* Parse shader header.
* Return TRUE for one of the following headers.
- * FRAG1.1
- * GEOM1.1
- * VERT1.1
+ * FRAG
+ * GEOM
+ * VERT
*/
static boolean parse_header( struct translate_ctx *ctx )
{
uint processor;
- if (str_match_no_case( &ctx->cur, "FRAG1.1" ))
+ if (str_match_no_case( &ctx->cur, "FRAG" ))
processor = TGSI_PROCESSOR_FRAGMENT;
- else if (str_match_no_case( &ctx->cur, "VERT1.1" ))
+ else if (str_match_no_case( &ctx->cur, "VERT" ))
processor = TGSI_PROCESSOR_VERTEX;
- else if (str_match_no_case( &ctx->cur, "GEOM1.1" ))
+ else if (str_match_no_case( &ctx->cur, "GEOM" ))
processor = TGSI_PROCESSOR_GEOMETRY;
else {
report_error( ctx, "Unknown header" );
@@ -193,16 +242,13 @@ static boolean parse_header( struct translate_ctx *ctx )
if (ctx->tokens_cur >= ctx->tokens_end)
return FALSE;
- *(struct tgsi_version *) ctx->tokens_cur++ = tgsi_build_version();
-
- if (ctx->tokens_cur >= ctx->tokens_end)
- return FALSE;
ctx->header = (struct tgsi_header *) ctx->tokens_cur++;
*ctx->header = tgsi_build_header();
if (ctx->tokens_cur >= ctx->tokens_end)
return FALSE;
*(struct tgsi_processor *) ctx->tokens_cur++ = tgsi_build_processor( processor, ctx->header );
+ ctx->processor = processor;
return TRUE;
}
@@ -233,7 +279,8 @@ static const char *file_names[TGSI_FILE_COUNT] =
"ADDR",
"IMM",
"LOOP",
- "PRED"
+ "PRED",
+ "SV"
};
static boolean
@@ -298,92 +345,36 @@ parse_opt_writemask(
return TRUE;
}
-/* <register_file_bracket> ::= <file> `['
- */
static boolean
-parse_register_file_bracket(
- struct translate_ctx *ctx,
- uint *file )
-{
- if (!parse_file( &ctx->cur, file )) {
- report_error( ctx, "Unknown register file" );
- return FALSE;
- }
- eat_opt_white( &ctx->cur );
- if (*ctx->cur != '[') {
- report_error( ctx, "Expected `['" );
- return FALSE;
- }
- ctx->cur++;
- return TRUE;
-}
+parse_register_dst( struct translate_ctx *ctx,
+ uint *file,
+ int *index );
-/* <register_file_bracket_index> ::= <register_file_bracket> <uint>
- */
-static boolean
-parse_register_file_bracket_index(
- struct translate_ctx *ctx,
- uint *file,
- int *index )
-{
- uint uindex;
+struct parsed_src_bracket {
+ int index;
- if (!parse_register_file_bracket( ctx, file ))
- return FALSE;
- eat_opt_white( &ctx->cur );
- if (!parse_uint( &ctx->cur, &uindex )) {
- report_error( ctx, "Expected literal unsigned integer" );
- return FALSE;
- }
- *index = (int) uindex;
- return TRUE;
-}
+ uint ind_file;
+ int ind_index;
+ uint ind_comp;
+};
-/* Parse destination register operand.
- * <register_dst> ::= <register_file_bracket_index> `]'
- */
-static boolean
-parse_register_dst(
- struct translate_ctx *ctx,
- uint *file,
- int *index )
-{
- if (!parse_register_file_bracket_index( ctx, file, index ))
- return FALSE;
- eat_opt_white( &ctx->cur );
- if (*ctx->cur != ']') {
- report_error( ctx, "Expected `]'" );
- return FALSE;
- }
- ctx->cur++;
- return TRUE;
-}
-/* Parse source register operand.
- * <register_src> ::= <register_file_bracket_index> `]' |
- * <register_file_bracket> <register_dst> [`.' (`x' | `y' | `z' | `w')] `]' |
- * <register_file_bracket> <register_dst> [`.' (`x' | `y' | `z' | `w')] `+' <uint> `]' |
- * <register_file_bracket> <register_dst> [`.' (`x' | `y' | `z' | `w')] `-' <uint> `]'
- */
static boolean
-parse_register_src(
+parse_register_src_bracket(
struct translate_ctx *ctx,
- uint *file,
- int *index,
- uint *ind_file,
- int *ind_index,
- uint *ind_comp)
+ struct parsed_src_bracket *brackets)
{
const char *cur;
uint uindex;
- *ind_comp = TGSI_SWIZZLE_X;
- if (!parse_register_file_bracket( ctx, file ))
- return FALSE;
+ memset(brackets, 0, sizeof(struct parsed_src_bracket));
+
eat_opt_white( &ctx->cur );
+
cur = ctx->cur;
- if (parse_file( &cur, ind_file )) {
- if (!parse_register_dst( ctx, ind_file, ind_index ))
+ if (parse_file( &cur, &brackets->ind_file )) {
+ if (!parse_register_dst( ctx, &brackets->ind_file,
+ &brackets->ind_index ))
return FALSE;
eat_opt_white( &ctx->cur );
@@ -393,16 +384,16 @@ parse_register_src(
switch (uprcase(*ctx->cur)) {
case 'X':
- *ind_comp = TGSI_SWIZZLE_X;
+ brackets->ind_comp = TGSI_SWIZZLE_X;
break;
case 'Y':
- *ind_comp = TGSI_SWIZZLE_Y;
+ brackets->ind_comp = TGSI_SWIZZLE_Y;
break;
case 'Z':
- *ind_comp = TGSI_SWIZZLE_Z;
+ brackets->ind_comp = TGSI_SWIZZLE_Z;
break;
case 'W':
- *ind_comp = TGSI_SWIZZLE_W;
+ brackets->ind_comp = TGSI_SWIZZLE_W;
break;
default:
report_error(ctx, "Expected indirect register swizzle component `x', `y', `z' or `w'");
@@ -423,12 +414,12 @@ parse_register_src(
return FALSE;
}
if (negate)
- *index = -(int) uindex;
+ brackets->index = -(int) uindex;
else
- *index = (int) uindex;
+ brackets->index = (int) uindex;
}
else {
- *index = 0;
+ brackets->index = 0;
}
}
else {
@@ -436,9 +427,9 @@ parse_register_src(
report_error( ctx, "Expected literal unsigned integer" );
return FALSE;
}
- *index = (int) uindex;
- *ind_file = TGSI_FILE_NULL;
- *ind_index = 0;
+ brackets->index = (int) uindex;
+ brackets->ind_file = TGSI_FILE_NULL;
+ brackets->ind_index = 0;
}
eat_opt_white( &ctx->cur );
if (*ctx->cur != ']') {
@@ -449,20 +440,123 @@ parse_register_src(
return TRUE;
}
-/* Parse register declaration.
- * <register_dcl> ::= <register_file_bracket_index> `]' |
- * <register_file_bracket_index> `..' <index> `]'
+static boolean
+parse_opt_register_src_bracket(
+ struct translate_ctx *ctx,
+ struct parsed_src_bracket *brackets,
+ int *parsed_brackets)
+{
+ const char *cur = ctx->cur;
+
+ *parsed_brackets = 0;
+
+ eat_opt_white( &cur );
+ if (cur[0] == '[') {
+ ++cur;
+ ctx->cur = cur;
+
+ if (!parse_register_src_bracket(ctx, brackets))
+ return FALSE;
+
+ *parsed_brackets = 1;
+ }
+
+ return TRUE;
+}
+
+/* <register_file_bracket> ::= <file> `['
*/
static boolean
-parse_register_dcl(
+parse_register_file_bracket(
+ struct translate_ctx *ctx,
+ uint *file )
+{
+ if (!parse_file( &ctx->cur, file )) {
+ report_error( ctx, "Unknown register file" );
+ return FALSE;
+ }
+ eat_opt_white( &ctx->cur );
+ if (*ctx->cur != '[') {
+ report_error( ctx, "Expected `['" );
+ return FALSE;
+ }
+ ctx->cur++;
+ return TRUE;
+}
+
+/* <register_file_bracket_index> ::= <register_file_bracket> <uint>
+ */
+static boolean
+parse_register_file_bracket_index(
+ struct translate_ctx *ctx,
+ uint *file,
+ int *index )
+{
+ uint uindex;
+
+ if (!parse_register_file_bracket( ctx, file ))
+ return FALSE;
+ eat_opt_white( &ctx->cur );
+ if (!parse_uint( &ctx->cur, &uindex )) {
+ report_error( ctx, "Expected literal unsigned integer" );
+ return FALSE;
+ }
+ *index = (int) uindex;
+ return TRUE;
+}
+
+/* Parse source register operand.
+ * <register_src> ::= <register_file_bracket_index> `]' |
+ * <register_file_bracket> <register_dst> [`.' (`x' | `y' | `z' | `w')] `]' |
+ * <register_file_bracket> <register_dst> [`.' (`x' | `y' | `z' | `w')] `+' <uint> `]' |
+ * <register_file_bracket> <register_dst> [`.' (`x' | `y' | `z' | `w')] `-' <uint> `]'
+ */
+static boolean
+parse_register_src(
struct translate_ctx *ctx,
uint *file,
- int *first,
- int *last )
+ struct parsed_src_bracket *brackets)
{
- if (!parse_register_file_bracket_index( ctx, file, first ))
+
+ brackets->ind_comp = TGSI_SWIZZLE_X;
+ if (!parse_register_file_bracket( ctx, file ))
return FALSE;
+ if (!parse_register_src_bracket( ctx, brackets ))
+ return FALSE;
+
+ return TRUE;
+}
+
+struct parsed_dcl_bracket {
+ uint first;
+ uint last;
+};
+
+static boolean
+parse_register_dcl_bracket(
+ struct translate_ctx *ctx,
+ struct parsed_dcl_bracket *bracket)
+{
+ uint uindex;
+ memset(bracket, 0, sizeof(struct parsed_dcl_bracket));
+
eat_opt_white( &ctx->cur );
+
+ if (!parse_uint( &ctx->cur, &uindex )) {
+ /* it can be an empty bracket [] which means its range
+ * is from 0 to some implied size */
+ if (ctx->cur[0] == ']' && ctx->implied_array_size != 0) {
+ bracket->first = 0;
+ bracket->last = ctx->implied_array_size - 1;
+ goto cleanup;
+ }
+ report_error( ctx, "Expected literal unsigned integer" );
+ return FALSE;
+ }
+ bracket->first = (int) uindex;
+
+ eat_opt_white( &ctx->cur );
+
if (ctx->cur[0] == '.' && ctx->cur[1] == '.') {
uint uindex;
@@ -472,12 +566,14 @@ parse_register_dcl(
report_error( ctx, "Expected literal integer" );
return FALSE;
}
- *last = (int) uindex;
+ bracket->last = (int) uindex;
eat_opt_white( &ctx->cur );
}
else {
- *last = *first;
+ bracket->last = bracket->first;
}
+
+cleanup:
if (*ctx->cur != ']') {
report_error( ctx, "Expected `]' or `..'" );
return FALSE;
@@ -486,16 +582,70 @@ parse_register_dcl(
return TRUE;
}
-static const char *modulate_names[TGSI_MODULATE_COUNT] =
+/* Parse register declaration.
+ * <register_dcl> ::= <register_file_bracket_index> `]' |
+ * <register_file_bracket_index> `..' <index> `]'
+ */
+static boolean
+parse_register_dcl(
+ struct translate_ctx *ctx,
+ uint *file,
+ struct parsed_dcl_bracket *brackets,
+ int *num_brackets)
{
- "_1X",
- "_2X",
- "_4X",
- "_8X",
- "_D2",
- "_D4",
- "_D8"
-};
+ const char *cur;
+
+ *num_brackets = 0;
+
+ if (!parse_register_file_bracket( ctx, file ))
+ return FALSE;
+ if (!parse_register_dcl_bracket( ctx, &brackets[0] ))
+ return FALSE;
+
+ *num_brackets = 1;
+
+ cur = ctx->cur;
+ eat_opt_white( &cur );
+
+ if (cur[0] == '[') {
+ ++cur;
+ ctx->cur = cur;
+ if (!parse_register_dcl_bracket( ctx, &brackets[1] ))
+ return FALSE;
+ /* for geometry shader we don't really care about
+ * the first brackets it's always the size of the
+ * input primitive. so we want to declare just
+ * the index relevant to the semantics which is in
+ * the second bracket */
+ if (ctx->processor == TGSI_PROCESSOR_GEOMETRY) {
+ brackets[0] = brackets[1];
+ }
+ *num_brackets = 2;
+ }
+
+ return TRUE;
+}
+
+
+/* Parse destination register operand.
+ * <register_dst> ::= <register_file_bracket_index> `]'
+ */
+static boolean
+parse_register_dst(
+ struct translate_ctx *ctx,
+ uint *file,
+ int *index )
+{
+ if (!parse_register_file_bracket_index( ctx, file, index ))
+ return FALSE;
+ eat_opt_white( &ctx->cur );
+ if (*ctx->cur != ']') {
+ report_error( ctx, "Expected `]'" );
+ return FALSE;
+ }
+ ctx->cur++;
+ return TRUE;
+}
static boolean
parse_dst_operand(
@@ -512,26 +662,13 @@ parse_dst_operand(
cur = ctx->cur;
eat_opt_white( &cur );
- if (*cur == '_') {
- uint i;
-
- for (i = 0; i < TGSI_MODULATE_COUNT; i++) {
- if (str_match_no_case( &cur, modulate_names[i] )) {
- if (!is_digit_alpha_underscore( cur )) {
- dst->DstRegisterExtModulate.Modulate = i;
- ctx->cur = cur;
- break;
- }
- }
- }
- }
if (!parse_opt_writemask( ctx, &writemask ))
return FALSE;
- dst->DstRegister.File = file;
- dst->DstRegister.Index = index;
- dst->DstRegister.WriteMask = writemask;
+ dst->Register.File = file;
+ dst->Register.Index = index;
+ dst->Register.WriteMask = writemask;
return TRUE;
}
@@ -577,162 +714,59 @@ parse_src_operand(
struct translate_ctx *ctx,
struct tgsi_full_src_register *src )
{
- const char *cur;
- float value;
uint file;
- int index;
- uint ind_file;
- int ind_index;
- uint ind_comp;
uint swizzle[4];
- boolean parsed_ext_negate_paren = FALSE;
boolean parsed_swizzle;
+ struct parsed_src_bracket bracket[2];
+ int parsed_opt_brackets;
if (*ctx->cur == '-') {
- cur = ctx->cur;
- cur++;
- eat_opt_white( &cur );
- if (*cur == '(') {
- cur++;
- src->SrcRegisterExtMod.Negate = 1;
- eat_opt_white( &cur );
- ctx->cur = cur;
- parsed_ext_negate_paren = TRUE;
- }
- else if (*cur == '|') {
- cur++;
- src->SrcRegisterExtMod.Negate = 1;
- src->SrcRegisterExtMod.Absolute = 1;
- eat_opt_white(&cur);
- ctx->cur = cur;
- }
- }
- else if (*ctx->cur == '|') {
ctx->cur++;
eat_opt_white( &ctx->cur );
- src->SrcRegisterExtMod.Absolute = 1;
+ src->Register.Negate = 1;
}
- if (*ctx->cur == '-') {
+ if (*ctx->cur == '|') {
ctx->cur++;
eat_opt_white( &ctx->cur );
- src->SrcRegister.Negate = 1;
+ src->Register.Absolute = 1;
}
- cur = ctx->cur;
- if (parse_float( &cur, &value )) {
- if (value == 2.0f) {
- eat_opt_white( &cur );
- if (*cur != '*') {
- report_error( ctx, "Expected `*'" );
- return FALSE;
- }
- cur++;
- if (*cur != '(') {
- report_error( ctx, "Expected `('" );
- return FALSE;
- }
- cur++;
- src->SrcRegisterExtMod.Scale2X = 1;
- eat_opt_white( &cur );
- ctx->cur = cur;
- }
- }
+ if (!parse_register_src(ctx, &file, &bracket[0]))
+ return FALSE;
+ if (!parse_opt_register_src_bracket(ctx, &bracket[1], &parsed_opt_brackets))
+ return FALSE;
- if (*ctx->cur == '(') {
- ctx->cur++;
- eat_opt_white( &ctx->cur );
- src->SrcRegisterExtMod.Bias = 1;
+ src->Register.File = file;
+ src->Register.Index = bracket[0].index;
+ if (bracket[0].ind_file != TGSI_FILE_NULL) {
+ src->Register.Indirect = 1;
+ src->Indirect.File = bracket[0].ind_file;
+ src->Indirect.Index = bracket[0].ind_index;
+ src->Indirect.SwizzleX = bracket[0].ind_comp;
+ src->Indirect.SwizzleY = bracket[0].ind_comp;
+ src->Indirect.SwizzleZ = bracket[0].ind_comp;
+ src->Indirect.SwizzleW = bracket[0].ind_comp;
}
-
- cur = ctx->cur;
- if (parse_float( &cur, &value )) {
- if (value == 1.0f) {
- eat_opt_white( &cur );
- if (*cur != '-') {
- report_error( ctx, "Expected `-'" );
- return FALSE;
- }
- cur++;
- if (*cur != '(') {
- report_error( ctx, "Expected `('" );
- return FALSE;
- }
- cur++;
- src->SrcRegisterExtMod.Complement = 1;
- eat_opt_white( &cur );
- ctx->cur = cur;
- }
- }
-
- if (!parse_register_src(ctx, &file, &index, &ind_file, &ind_index, &ind_comp))
- return FALSE;
- src->SrcRegister.File = file;
- src->SrcRegister.Index = index;
- if (ind_file != TGSI_FILE_NULL) {
- src->SrcRegister.Indirect = 1;
- src->SrcRegisterInd.File = ind_file;
- src->SrcRegisterInd.Index = ind_index;
- src->SrcRegisterInd.SwizzleX = ind_comp;
- src->SrcRegisterInd.SwizzleY = ind_comp;
- src->SrcRegisterInd.SwizzleZ = ind_comp;
- src->SrcRegisterInd.SwizzleW = ind_comp;
+ if (parsed_opt_brackets) {
+ src->Register.Dimension = 1;
+ src->Dimension.Indirect = 0;
+ src->Dimension.Dimension = 0;
+ src->Dimension.Index = bracket[1].index;
}
/* Parse optional swizzle.
*/
if (parse_optional_swizzle( ctx, swizzle, &parsed_swizzle )) {
if (parsed_swizzle) {
- src->SrcRegister.SwizzleX = swizzle[0];
- src->SrcRegister.SwizzleY = swizzle[1];
- src->SrcRegister.SwizzleZ = swizzle[2];
- src->SrcRegister.SwizzleW = swizzle[3];
- }
- }
-
- if (src->SrcRegisterExtMod.Complement) {
- eat_opt_white( &ctx->cur );
- if (*ctx->cur != ')') {
- report_error( ctx, "Expected `)'" );
- return FALSE;
- }
- ctx->cur++;
- }
-
- if (src->SrcRegisterExtMod.Bias) {
- eat_opt_white( &ctx->cur );
- if (*ctx->cur != ')') {
- report_error( ctx, "Expected `)'" );
- return FALSE;
- }
- ctx->cur++;
- eat_opt_white( &ctx->cur );
- if (*ctx->cur != '-') {
- report_error( ctx, "Expected `-'" );
- return FALSE;
- }
- ctx->cur++;
- eat_opt_white( &ctx->cur );
- if (!parse_float( &ctx->cur, &value )) {
- report_error( ctx, "Expected literal floating point" );
- return FALSE;
- }
- if (value != 0.5f) {
- report_error( ctx, "Expected 0.5" );
- return FALSE;
+ src->Register.SwizzleX = swizzle[0];
+ src->Register.SwizzleY = swizzle[1];
+ src->Register.SwizzleZ = swizzle[2];
+ src->Register.SwizzleW = swizzle[3];
}
}
- if (src->SrcRegisterExtMod.Scale2X) {
- eat_opt_white( &ctx->cur );
- if (*ctx->cur != ')') {
- report_error( ctx, "Expected `)'" );
- return FALSE;
- }
- ctx->cur++;
- }
-
- if (src->SrcRegisterExtMod.Absolute) {
+ if (src->Register.Absolute) {
eat_opt_white( &ctx->cur );
if (*ctx->cur != '|') {
report_error( ctx, "Expected `|'" );
@@ -741,14 +775,6 @@ parse_src_operand(
ctx->cur++;
}
- if (parsed_ext_negate_paren) {
- eat_opt_white( &ctx->cur );
- if (*ctx->cur != ')') {
- report_error( ctx, "Expected `)'" );
- return FALSE;
- }
- ctx->cur++;
- }
return TRUE;
}
@@ -840,11 +866,11 @@ parse_instruction(
}
if (i < info->num_dst) {
- if (!parse_dst_operand( ctx, &inst.FullDstRegisters[i] ))
+ if (!parse_dst_operand( ctx, &inst.Dst[i] ))
return FALSE;
}
else if (i < info->num_dst + info->num_src) {
- if (!parse_src_operand( ctx, &inst.FullSrcRegisters[i - info->num_dst] ))
+ if (!parse_src_operand( ctx, &inst.Src[i - info->num_dst] ))
return FALSE;
}
else {
@@ -853,7 +879,8 @@ parse_instruction(
for (j = 0; j < TGSI_TEXTURE_COUNT; j++) {
if (str_match_no_case( &ctx->cur, texture_names[j] )) {
if (!is_digit_alpha_underscore( ctx->cur )) {
- inst.InstructionExtTexture.Texture = j;
+ inst.Instruction.Texture = 1;
+ inst.Texture.Texture = j;
break;
}
}
@@ -879,7 +906,8 @@ parse_instruction(
report_error( ctx, "Expected a label" );
return FALSE;
}
- inst.InstructionExtLabel.Label = target;
+ inst.Instruction.Label = 1;
+ inst.Label.Label = target;
}
advance = tgsi_build_full_instruction(
@@ -903,7 +931,9 @@ static const char *semantic_names[TGSI_SEMANTIC_COUNT] =
"PSIZE",
"GENERIC",
"NORMAL",
- "FACE"
+ "FACE",
+ "EDGEFLAG",
+ "PRIM_ID"
};
static const char *interpolate_names[TGSI_INTERPOLATE_COUNT] =
@@ -917,8 +947,8 @@ static boolean parse_declaration( struct translate_ctx *ctx )
{
struct tgsi_full_declaration decl;
uint file;
- int first;
- int last;
+ struct parsed_dcl_bracket brackets[2];
+ int num_brackets;
uint writemask;
const char *cur;
uint advance;
@@ -930,7 +960,7 @@ static boolean parse_declaration( struct translate_ctx *ctx )
report_error( ctx, "Syntax error" );
return FALSE;
}
- if (!parse_register_dcl( ctx, &file, &first, &last ))
+ if (!parse_register_dcl( ctx, &file, brackets, &num_brackets))
return FALSE;
if (!parse_opt_writemask( ctx, &writemask ))
return FALSE;
@@ -938,8 +968,8 @@ static boolean parse_declaration( struct translate_ctx *ctx )
decl = tgsi_default_full_declaration();
decl.Declaration.File = file;
decl.Declaration.UsageMask = writemask;
- decl.DeclarationRange.First = first;
- decl.DeclarationRange.Last = last;
+ decl.Range.First = brackets[0].first;
+ decl.Range.Last = brackets[0].last;
cur = ctx->cur;
eat_opt_white( &cur );
@@ -970,13 +1000,13 @@ static boolean parse_declaration( struct translate_ctx *ctx )
}
cur2++;
- decl.Semantic.SemanticIndex = index;
+ decl.Semantic.Index = index;
cur = cur2;
}
decl.Declaration.Semantic = 1;
- decl.Semantic.SemanticName = i;
+ decl.Semantic.Name = i;
ctx->cur = cur;
break;
@@ -1082,6 +1112,110 @@ static boolean parse_immediate( struct translate_ctx *ctx )
return TRUE;
}
+static const char *property_names[] =
+{
+ "GS_INPUT_PRIMITIVE",
+ "GS_OUTPUT_PRIMITIVE",
+ "GS_MAX_OUTPUT_VERTICES"
+};
+
+static const char *primitive_names[] =
+{
+ "POINTS",
+ "LINES",
+ "LINE_LOOP",
+ "LINE_STRIP",
+ "TRIANGLES",
+ "TRIANGLE_STRIP",
+ "TRIANGLE_FAN",
+ "QUADS",
+ "QUAD_STRIP",
+ "POLYGON"
+};
+
+static boolean
+parse_primitive( const char **pcur, uint *primitive )
+{
+ uint i;
+
+ for (i = 0; i < PIPE_PRIM_MAX; i++) {
+ const char *cur = *pcur;
+
+ if (str_match_no_case( &cur, primitive_names[i])) {
+ *primitive = i;
+ *pcur = cur;
+ return TRUE;
+ }
+ }
+ return FALSE;
+}
+
+
+static boolean parse_property( struct translate_ctx *ctx )
+{
+ struct tgsi_full_property prop;
+ uint property_name;
+ uint values[8];
+ uint advance;
+ char id[64];
+
+ if (!eat_white( &ctx->cur )) {
+ report_error( ctx, "Syntax error" );
+ return FALSE;
+ }
+ if (!parse_identifier( &ctx->cur, id )) {
+ report_error( ctx, "Syntax error" );
+ return FALSE;
+ }
+ for (property_name = 0; property_name < TGSI_PROPERTY_COUNT;
+ ++property_name) {
+ if (streq_nocase_uprcase(property_names[property_name], id)) {
+ break;
+ }
+ }
+ if (property_name >= TGSI_PROPERTY_COUNT) {
+ debug_printf( "\nError: Unknown property : '%s'", id );
+ return FALSE;
+ }
+
+ eat_opt_white( &ctx->cur );
+ switch(property_name) {
+ case TGSI_PROPERTY_GS_INPUT_PRIM:
+ case TGSI_PROPERTY_GS_OUTPUT_PRIM:
+ if (!parse_primitive(&ctx->cur, &values[0] )) {
+ report_error( ctx, "Unknown primitive name as property!" );
+ return FALSE;
+ }
+ if (property_name == TGSI_PROPERTY_GS_INPUT_PRIM &&
+ ctx->processor == TGSI_PROCESSOR_GEOMETRY) {
+ ctx->implied_array_size = u_vertices_per_prim(values[0]);
+ }
+ break;
+ default:
+ if (!parse_uint(&ctx->cur, &values[0] )) {
+ report_error( ctx, "Expected unsigned integer as property!" );
+ return FALSE;
+ }
+ }
+
+ prop = tgsi_default_full_property();
+ prop.Property.PropertyName = property_name;
+ prop.Property.NrTokens += 1;
+ prop.u[0].Data = values[0];
+
+ advance = tgsi_build_full_property(
+ &prop,
+ ctx->tokens_cur,
+ ctx->header,
+ (uint) (ctx->tokens_end - ctx->tokens_cur) );
+ if (advance == 0)
+ return FALSE;
+ ctx->tokens_cur += advance;
+
+ return TRUE;
+}
+
+
static boolean translate( struct translate_ctx *ctx )
{
eat_opt_white( &ctx->cur );
@@ -1090,7 +1224,6 @@ static boolean translate( struct translate_ctx *ctx )
while (*ctx->cur != '\0') {
uint label_val = 0;
-
if (!eat_white( &ctx->cur )) {
report_error( ctx, "Syntax error" );
return FALSE;
@@ -1098,7 +1231,6 @@ static boolean translate( struct translate_ctx *ctx )
if (*ctx->cur == '\0')
break;
-
if (parse_label( ctx, &label_val )) {
if (!parse_instruction( ctx, TRUE ))
return FALSE;
@@ -1111,6 +1243,10 @@ static boolean translate( struct translate_ctx *ctx )
if (!parse_immediate( ctx ))
return FALSE;
}
+ else if (str_match_no_case( &ctx->cur, "PROPERTY" )) {
+ if (!parse_property( ctx ))
+ return FALSE;
+ }
else if (!parse_instruction( ctx, FALSE )) {
return FALSE;
}
diff --git a/src/gallium/auxiliary/tgsi/tgsi_transform.c b/src/gallium/auxiliary/tgsi/tgsi_transform.c
index bc9c18fd4a7..ae875f29abf 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_transform.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_transform.c
@@ -79,6 +79,19 @@ emit_immediate(struct tgsi_transform_context *ctx,
}
+static void
+emit_property(struct tgsi_transform_context *ctx,
+ const struct tgsi_full_property *prop)
+{
+ uint ti = ctx->ti;
+
+ ti += tgsi_build_full_property(prop,
+ ctx->tokens_out + ti,
+ ctx->header,
+ ctx->max_tokens_out - ti);
+ ctx->ti = ti;
+}
+
/**
* Apply user-defined transformations to the input shader to produce
@@ -110,6 +123,7 @@ tgsi_transform_shader(const struct tgsi_token *tokens_in,
ctx->emit_instruction = emit_instruction;
ctx->emit_declaration = emit_declaration;
ctx->emit_immediate = emit_immediate;
+ ctx->emit_property = emit_property;
ctx->tokens_out = tokens_out;
ctx->max_tokens_out = max_tokens_out;
@@ -130,15 +144,13 @@ tgsi_transform_shader(const struct tgsi_token *tokens_in,
/**
** Setup output shader
**/
- *(struct tgsi_version *) &tokens_out[0] = tgsi_build_version();
-
- ctx->header = (struct tgsi_header *) (tokens_out + 1);
+ ctx->header = (struct tgsi_header *)tokens_out;
*ctx->header = tgsi_build_header();
- processor = (struct tgsi_processor *) (tokens_out + 2);
+ processor = (struct tgsi_processor *) (tokens_out + 1);
*processor = tgsi_build_processor( procType, ctx->header );
- ctx->ti = 3;
+ ctx->ti = 2;
/**
@@ -184,6 +196,17 @@ tgsi_transform_shader(const struct tgsi_token *tokens_in,
ctx->emit_immediate(ctx, fullimm);
}
break;
+ case TGSI_TOKEN_TYPE_PROPERTY:
+ {
+ struct tgsi_full_property *fullprop
+ = &parse.FullToken.FullProperty;
+
+ if (ctx->transform_property)
+ ctx->transform_property(ctx, fullprop);
+ else
+ ctx->emit_property(ctx, fullprop);
+ }
+ break;
default:
assert( 0 );
@@ -215,7 +238,7 @@ tgsi_transform_foo( struct tgsi_token *tokens_out,
uint max_tokens_out )
{
const char *text =
- "FRAG1.1\n"
+ "FRAG\n"
"DCL IN[0], COLOR, CONSTANT\n"
"DCL OUT[0], COLOR\n"
" 0: MOV OUT[0], IN[0]\n"
diff --git a/src/gallium/auxiliary/tgsi/tgsi_transform.h b/src/gallium/auxiliary/tgsi/tgsi_transform.h
index a121adbaef4..818478e277a 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_transform.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_transform.h
@@ -53,6 +53,8 @@ struct tgsi_transform_context
void (*transform_immediate)(struct tgsi_transform_context *ctx,
struct tgsi_full_immediate *imm);
+ void (*transform_property)(struct tgsi_transform_context *ctx,
+ struct tgsi_full_property *prop);
/**
* Called at end of input program to allow caller to append extra
@@ -73,6 +75,8 @@ struct tgsi_transform_context
const struct tgsi_full_declaration *decl);
void (*emit_immediate)(struct tgsi_transform_context *ctx,
const struct tgsi_full_immediate *imm);
+ void (*emit_property)(struct tgsi_transform_context *ctx,
+ const struct tgsi_full_property *prop);
struct tgsi_header *header;
uint max_tokens_out;
diff --git a/src/gallium/auxiliary/tgsi/tgsi_ureg.c b/src/gallium/auxiliary/tgsi/tgsi_ureg.c
index bf39cf54094..2b51672b8ea 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_ureg.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_ureg.c
@@ -37,7 +37,6 @@
#include "util/u_math.h"
union tgsi_any_token {
- struct tgsi_version version;
struct tgsi_header header;
struct tgsi_processor processor;
struct tgsi_token token;
@@ -47,14 +46,12 @@ union tgsi_any_token {
struct tgsi_immediate imm;
union tgsi_immediate_data imm_data;
struct tgsi_instruction insn;
- struct tgsi_instruction_ext_label insn_ext_label;
- struct tgsi_instruction_ext_texture insn_ext_texture;
- struct tgsi_instruction_ext_predicate insn_ext_predicate;
+ struct tgsi_instruction_predicate insn_predicate;
+ struct tgsi_instruction_label insn_label;
+ struct tgsi_instruction_texture insn_texture;
struct tgsi_src_register src;
- struct tgsi_src_register_ext_mod src_ext_mod;
struct tgsi_dimension dim;
struct tgsi_dst_register dst;
- struct tgsi_dst_register_ext_modulate dst_ext_mod;
unsigned value;
};
@@ -67,11 +64,13 @@ struct ureg_tokens {
};
#define UREG_MAX_INPUT PIPE_MAX_ATTRIBS
+#define UREG_MAX_SYSTEM_VALUE PIPE_MAX_ATTRIBS
#define UREG_MAX_OUTPUT PIPE_MAX_ATTRIBS
#define UREG_MAX_CONSTANT_RANGE 32
#define UREG_MAX_IMMEDIATE 32
#define UREG_MAX_TEMP 256
#define UREG_MAX_ADDR 2
+#define UREG_MAX_LOOP 1
#define UREG_MAX_PRED 1
#define DOMAIN_DECL 0
@@ -92,14 +91,31 @@ struct ureg_program
unsigned vs_inputs[UREG_MAX_INPUT/32];
struct {
+ unsigned index;
+ } gs_input[UREG_MAX_INPUT];
+ unsigned nr_gs_inputs;
+
+ struct {
+ unsigned index;
+ unsigned semantic_name;
+ unsigned semantic_index;
+ } system_value[UREG_MAX_SYSTEM_VALUE];
+ unsigned nr_system_values;
+
+ struct {
unsigned semantic_name;
unsigned semantic_index;
} output[UREG_MAX_OUTPUT];
unsigned nr_outputs;
struct {
- float v[4];
+ union {
+ float f[4];
+ unsigned u[4];
+ int i[4];
+ } value;
unsigned nr;
+ unsigned type;
} immediate[UREG_MAX_IMMEDIATE];
unsigned nr_immediates;
@@ -117,6 +133,7 @@ struct ureg_program
unsigned nr_addrs;
unsigned nr_preds;
+ unsigned nr_loops;
unsigned nr_instructions;
struct ureg_tokens domain[2];
@@ -279,6 +296,41 @@ ureg_DECL_vs_input( struct ureg_program *ureg,
}
+struct ureg_src
+ureg_DECL_gs_input(struct ureg_program *ureg,
+ unsigned index)
+{
+ if (ureg->nr_gs_inputs < UREG_MAX_INPUT) {
+ ureg->gs_input[ureg->nr_gs_inputs].index = index;
+ ureg->nr_gs_inputs++;
+ } else {
+ set_bad(ureg);
+ }
+
+ /* XXX: Add suport for true 2D input registers. */
+ return ureg_src_register(TGSI_FILE_INPUT, index);
+}
+
+
+struct ureg_src
+ureg_DECL_system_value(struct ureg_program *ureg,
+ unsigned index,
+ unsigned semantic_name,
+ unsigned semantic_index)
+{
+ if (ureg->nr_system_values < UREG_MAX_SYSTEM_VALUE) {
+ ureg->system_value[ureg->nr_system_values].index = index;
+ ureg->system_value[ureg->nr_system_values].semantic_name = semantic_name;
+ ureg->system_value[ureg->nr_system_values].semantic_index = semantic_index;
+ ureg->nr_system_values++;
+ } else {
+ set_bad(ureg);
+ }
+
+ return ureg_src_register(TGSI_FILE_SYSTEM_VALUE, index);
+}
+
+
struct ureg_dst
ureg_DECL_output( struct ureg_program *ureg,
unsigned name,
@@ -350,6 +402,7 @@ struct ureg_src ureg_DECL_constant(struct ureg_program *ureg,
i = ureg->nr_constant_ranges++;
ureg->constant_range[i].first = index;
ureg->constant_range[i].last = index;
+ goto out;
}
/* Collapse all ranges down to one:
@@ -417,6 +470,19 @@ struct ureg_dst ureg_DECL_address( struct ureg_program *ureg )
return ureg_dst_register( TGSI_FILE_ADDRESS, 0 );
}
+/* Allocate a new loop register.
+ */
+struct ureg_dst
+ureg_DECL_loop(struct ureg_program *ureg)
+{
+ if (ureg->nr_loops < UREG_MAX_LOOP) {
+ return ureg_dst_register(TGSI_FILE_LOOP, ureg->nr_loops++);
+ }
+
+ assert(0);
+ return ureg_dst_register(TGSI_FILE_LOOP, 0);
+}
+
/* Allocate a new predicate register.
*/
struct ureg_dst
@@ -452,22 +518,22 @@ struct ureg_src ureg_DECL_sampler( struct ureg_program *ureg,
}
-
-
-static int match_or_expand_immediate( const float *v,
- unsigned nr,
- float *v2,
- unsigned *nr2,
- unsigned *swizzle )
+static int
+match_or_expand_immediate( const unsigned *v,
+ unsigned nr,
+ unsigned *v2,
+ unsigned *pnr2,
+ unsigned *swizzle )
{
+ unsigned nr2 = *pnr2;
unsigned i, j;
-
+
*swizzle = 0;
for (i = 0; i < nr; i++) {
boolean found = FALSE;
- for (j = 0; j < *nr2 && !found; j++) {
+ for (j = 0; j < nr2 && !found; j++) {
if (v[i] == v2[j]) {
*swizzle |= j << (i * 2);
found = TRUE;
@@ -475,24 +541,28 @@ static int match_or_expand_immediate( const float *v,
}
if (!found) {
- if (*nr2 >= 4)
+ if (nr2 >= 4) {
return FALSE;
+ }
- v2[*nr2] = v[i];
- *swizzle |= *nr2 << (i * 2);
- (*nr2)++;
+ v2[nr2] = v[i];
+ *swizzle |= nr2 << (i * 2);
+ nr2++;
}
}
+ /* Actually expand immediate only when fully succeeded.
+ */
+ *pnr2 = nr2;
return TRUE;
}
-
-
-struct ureg_src ureg_DECL_immediate( struct ureg_program *ureg,
- const float *v,
- unsigned nr )
+static struct ureg_src
+decl_immediate( struct ureg_program *ureg,
+ const unsigned *v,
+ unsigned nr,
+ unsigned type )
{
unsigned i, j;
unsigned swizzle;
@@ -502,38 +572,82 @@ struct ureg_src ureg_DECL_immediate( struct ureg_program *ureg,
*/
for (i = 0; i < ureg->nr_immediates; i++) {
- if (match_or_expand_immediate( v,
- nr,
- ureg->immediate[i].v,
- &ureg->immediate[i].nr,
- &swizzle ))
+ if (ureg->immediate[i].type != type) {
+ continue;
+ }
+ if (match_or_expand_immediate(v,
+ nr,
+ ureg->immediate[i].value.u,
+ &ureg->immediate[i].nr,
+ &swizzle)) {
goto out;
+ }
}
if (ureg->nr_immediates < UREG_MAX_IMMEDIATE) {
i = ureg->nr_immediates++;
- if (match_or_expand_immediate( v,
- nr,
- ureg->immediate[i].v,
- &ureg->immediate[i].nr,
- &swizzle ))
+ ureg->immediate[i].type = type;
+ if (match_or_expand_immediate(v,
+ nr,
+ ureg->immediate[i].value.u,
+ &ureg->immediate[i].nr,
+ &swizzle)) {
goto out;
+ }
}
- set_bad( ureg );
+ set_bad(ureg);
out:
/* Make sure that all referenced elements are from this immediate.
* Has the effect of making size-one immediates into scalars.
*/
- for (j = nr; j < 4; j++)
+ for (j = nr; j < 4; j++) {
swizzle |= (swizzle & 0x3) << (j * 2);
+ }
- return ureg_swizzle( ureg_src_register( TGSI_FILE_IMMEDIATE, i ),
- (swizzle >> 0) & 0x3,
- (swizzle >> 2) & 0x3,
- (swizzle >> 4) & 0x3,
- (swizzle >> 6) & 0x3);
+ return ureg_swizzle(ureg_src_register(TGSI_FILE_IMMEDIATE, i),
+ (swizzle >> 0) & 0x3,
+ (swizzle >> 2) & 0x3,
+ (swizzle >> 4) & 0x3,
+ (swizzle >> 6) & 0x3);
+}
+
+
+struct ureg_src
+ureg_DECL_immediate( struct ureg_program *ureg,
+ const float *v,
+ unsigned nr )
+{
+ union {
+ float f[4];
+ unsigned u[4];
+ } fu;
+ unsigned int i;
+
+ for (i = 0; i < nr; i++) {
+ fu.f[i] = v[i];
+ }
+
+ return decl_immediate(ureg, fu.u, nr, TGSI_IMM_FLOAT32);
+}
+
+
+struct ureg_src
+ureg_DECL_immediate_uint( struct ureg_program *ureg,
+ const unsigned *v,
+ unsigned nr )
+{
+ return decl_immediate(ureg, v, nr, TGSI_IMM_UINT32);
+}
+
+
+struct ureg_src
+ureg_DECL_immediate_int( struct ureg_program *ureg,
+ const int *v,
+ unsigned nr )
+{
+ return decl_immediate(ureg, (const unsigned *)v, nr, TGSI_IMM_INT32);
}
@@ -541,9 +655,7 @@ void
ureg_emit_src( struct ureg_program *ureg,
struct ureg_src src )
{
- unsigned size = (1 +
- (src.Absolute ? 1 : 0) +
- (src.Indirect ? 1 : 0));
+ unsigned size = 1 + (src.Indirect ? 1 : 0);
union tgsi_any_token *out = get_tokens( ureg, DOMAIN_INSN, size );
unsigned n = 0;
@@ -560,17 +672,8 @@ ureg_emit_src( struct ureg_program *ureg,
out[n].src.SwizzleW = src.SwizzleW;
out[n].src.Index = src.Index;
out[n].src.Negate = src.Negate;
+ out[0].src.Absolute = src.Absolute;
n++;
-
- if (src.Absolute) {
- out[0].src.Extended = 1;
- out[0].src.Negate = 0;
- out[n].value = 0;
- out[n].src_ext_mod.Type = TGSI_SRC_REGISTER_EXT_TYPE_MOD;
- out[n].src_ext_mod.Absolute = 1;
- out[n].src_ext_mod.Negate = src.Negate;
- n++;
- }
if (src.Indirect) {
out[0].src.Indirect = 1;
@@ -661,35 +764,27 @@ ureg_emit_insn(struct ureg_program *ureg,
validate( opcode, num_dst, num_src );
out = get_tokens( ureg, DOMAIN_INSN, count );
- out[0].value = 0;
- out[0].insn.Type = TGSI_TOKEN_TYPE_INSTRUCTION;
- out[0].insn.NrTokens = 0;
+ out[0].insn = tgsi_default_instruction();
out[0].insn.Opcode = opcode;
out[0].insn.Saturate = saturate;
out[0].insn.NumDstRegs = num_dst;
out[0].insn.NumSrcRegs = num_src;
- out[0].insn.Padding = 0;
result.insn_token = ureg->domain[DOMAIN_INSN].count - count;
+ result.extended_token = result.insn_token;
if (predicate) {
- out[0].insn.Extended = 1;
- out[1].insn_ext_predicate = tgsi_default_instruction_ext_predicate();
- out[1].insn_ext_predicate.Negate = pred_negate;
- out[1].insn_ext_predicate.SwizzleX = pred_swizzle_x;
- out[1].insn_ext_predicate.SwizzleY = pred_swizzle_y;
- out[1].insn_ext_predicate.SwizzleZ = pred_swizzle_z;
- out[1].insn_ext_predicate.SwizzleW = pred_swizzle_w;
-
- result.extended_token = result.insn_token + 1;
- } else {
- out[0].insn.Extended = 0;
-
- result.extended_token = result.insn_token;
+ out[0].insn.Predicate = 1;
+ out[1].insn_predicate = tgsi_default_instruction_predicate();
+ out[1].insn_predicate.Negate = pred_negate;
+ out[1].insn_predicate.SwizzleX = pred_swizzle_x;
+ out[1].insn_predicate.SwizzleY = pred_swizzle_y;
+ out[1].insn_predicate.SwizzleZ = pred_swizzle_z;
+ out[1].insn_predicate.SwizzleW = pred_swizzle_w;
}
ureg->nr_instructions++;
-
+
return result;
}
@@ -705,13 +800,11 @@ ureg_emit_label(struct ureg_program *ureg,
return;
out = get_tokens( ureg, DOMAIN_INSN, 1 );
- insn = retrieve_token( ureg, DOMAIN_INSN, extended_token );
+ out[0].value = 0;
- insn->token.Extended = 1;
+ insn = retrieve_token( ureg, DOMAIN_INSN, extended_token );
+ insn->insn.Label = 1;
- out[0].value = 0;
- out[0].insn_ext_label.Type = TGSI_INSTRUCTION_EXT_TYPE_LABEL;
-
*label_token = ureg->domain[DOMAIN_INSN].count - 1;
}
@@ -734,8 +827,7 @@ ureg_fixup_label(struct ureg_program *ureg,
{
union tgsi_any_token *out = retrieve_token( ureg, DOMAIN_INSN, label_token );
- assert(out->insn_ext_label.Type == TGSI_INSTRUCTION_EXT_TYPE_LABEL);
- out->insn_ext_label.Label = instruction_number;
+ out->insn_label.Label = instruction_number;
}
@@ -749,11 +841,10 @@ ureg_emit_texture(struct ureg_program *ureg,
out = get_tokens( ureg, DOMAIN_INSN, 1 );
insn = retrieve_token( ureg, DOMAIN_INSN, extended_token );
- insn->token.Extended = 1;
+ insn->insn.Texture = 1;
out[0].value = 0;
- out[0].insn_ext_texture.Type = TGSI_INSTRUCTION_EXT_TYPE_TEXTURE;
- out[0].insn_ext_texture.Texture = target;
+ out[0].insn_texture.Texture = target;
}
@@ -918,8 +1009,8 @@ static void emit_decl( struct ureg_program *ureg,
out[1].decl_range.Last = index;
out[2].value = 0;
- out[2].decl_semantic.SemanticName = semantic_name;
- out[2].decl_semantic.SemanticIndex = semantic_index;
+ out[2].decl_semantic.Name = semantic_name;
+ out[2].decl_semantic.Index = semantic_index;
}
@@ -944,22 +1035,23 @@ static void emit_decl_range( struct ureg_program *ureg,
out[1].decl_range.Last = first + count - 1;
}
-static void emit_immediate( struct ureg_program *ureg,
- const float *v )
+static void
+emit_immediate( struct ureg_program *ureg,
+ const unsigned *v,
+ unsigned type )
{
union tgsi_any_token *out = get_tokens( ureg, DOMAIN_DECL, 5 );
out[0].value = 0;
out[0].imm.Type = TGSI_TOKEN_TYPE_IMMEDIATE;
out[0].imm.NrTokens = 5;
- out[0].imm.DataType = TGSI_IMM_FLOAT32;
+ out[0].imm.DataType = type;
out[0].imm.Padding = 0;
- out[0].imm.Extended = 0;
- out[1].imm_data.Float = v[0];
- out[2].imm_data.Float = v[1];
- out[3].imm_data.Float = v[2];
- out[4].imm_data.Float = v[3];
+ out[1].imm_data.Uint = v[0];
+ out[2].imm_data.Uint = v[1];
+ out[3].imm_data.Uint = v[2];
+ out[4].imm_data.Uint = v[3];
}
@@ -975,8 +1067,7 @@ static void emit_decls( struct ureg_program *ureg )
emit_decl_range( ureg, TGSI_FILE_INPUT, i, 1 );
}
}
- }
- else {
+ } else if (ureg->processor == TGSI_PROCESSOR_FRAGMENT) {
for (i = 0; i < ureg->nr_fs_inputs; i++) {
emit_decl( ureg,
TGSI_FILE_INPUT,
@@ -985,6 +1076,22 @@ static void emit_decls( struct ureg_program *ureg )
ureg->fs_input[i].semantic_index,
ureg->fs_input[i].interp );
}
+ } else {
+ for (i = 0; i < ureg->nr_gs_inputs; i++) {
+ emit_decl_range(ureg,
+ TGSI_FILE_INPUT,
+ ureg->gs_input[i].index,
+ 1);
+ }
+ }
+
+ for (i = 0; i < ureg->nr_system_values; i++) {
+ emit_decl(ureg,
+ TGSI_FILE_SYSTEM_VALUE,
+ ureg->system_value[i].index,
+ ureg->system_value[i].semantic_name,
+ ureg->system_value[i].semantic_index,
+ TGSI_INTERPOLATE_CONSTANT);
}
for (i = 0; i < ureg->nr_outputs; i++) {
@@ -1023,6 +1130,13 @@ static void emit_decls( struct ureg_program *ureg )
0, ureg->nr_addrs );
}
+ if (ureg->nr_loops) {
+ emit_decl_range(ureg,
+ TGSI_FILE_LOOP,
+ 0,
+ ureg->nr_loops);
+ }
+
if (ureg->nr_preds) {
emit_decl_range(ureg,
TGSI_FILE_PREDICATE,
@@ -1032,7 +1146,8 @@ static void emit_decls( struct ureg_program *ureg )
for (i = 0; i < ureg->nr_immediates; i++) {
emit_immediate( ureg,
- ureg->immediate[i].v );
+ ureg->immediate[i].value.u,
+ ureg->immediate[i].type );
}
}
@@ -1055,26 +1170,22 @@ static void copy_instructions( struct ureg_program *ureg )
static void
fixup_header_size(struct ureg_program *ureg)
{
- union tgsi_any_token *out = retrieve_token( ureg, DOMAIN_DECL, 1 );
+ union tgsi_any_token *out = retrieve_token( ureg, DOMAIN_DECL, 0 );
- out->header.BodySize = ureg->domain[DOMAIN_DECL].count - 3;
+ out->header.BodySize = ureg->domain[DOMAIN_DECL].count - 2;
}
static void
emit_header( struct ureg_program *ureg )
{
- union tgsi_any_token *out = get_tokens( ureg, DOMAIN_DECL, 3 );
-
- out[0].version.MajorVersion = 1;
- out[0].version.MinorVersion = 1;
- out[0].version.Padding = 0;
+ union tgsi_any_token *out = get_tokens( ureg, DOMAIN_DECL, 2 );
- out[1].header.HeaderSize = 2;
- out[1].header.BodySize = 0;
+ out[0].header.HeaderSize = 2;
+ out[0].header.BodySize = 0;
- out[2].processor.Processor = ureg->processor;
- out[2].processor.Padding = 0;
+ out[1].processor.Processor = ureg->processor;
+ out[1].processor.Padding = 0;
}
diff --git a/src/gallium/auxiliary/tgsi/tgsi_ureg.h b/src/gallium/auxiliary/tgsi/tgsi_ureg.h
index dae42911947..38e2fd8d0a9 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_ureg.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_ureg.h
@@ -133,6 +133,16 @@ struct ureg_src
ureg_DECL_vs_input( struct ureg_program *,
unsigned index );
+struct ureg_src
+ureg_DECL_gs_input(struct ureg_program *,
+ unsigned index);
+
+struct ureg_src
+ureg_DECL_system_value(struct ureg_program *,
+ unsigned index,
+ unsigned semantic_name,
+ unsigned semantic_index);
+
struct ureg_dst
ureg_DECL_output( struct ureg_program *,
unsigned semantic_name,
@@ -144,6 +154,16 @@ ureg_DECL_immediate( struct ureg_program *,
unsigned nr );
struct ureg_src
+ureg_DECL_immediate_uint( struct ureg_program *,
+ const unsigned *v,
+ unsigned nr );
+
+struct ureg_src
+ureg_DECL_immediate_int( struct ureg_program *,
+ const int *v,
+ unsigned nr );
+
+struct ureg_src
ureg_DECL_constant( struct ureg_program *,
unsigned index );
@@ -158,6 +178,9 @@ struct ureg_dst
ureg_DECL_address( struct ureg_program * );
struct ureg_dst
+ureg_DECL_loop( struct ureg_program * );
+
+struct ureg_dst
ureg_DECL_predicate(struct ureg_program *);
/* Supply an index to the sampler declaration as this is the hook to
@@ -214,6 +237,90 @@ ureg_imm1f( struct ureg_program *ureg,
return ureg_DECL_immediate( ureg, v, 1 );
}
+static INLINE struct ureg_src
+ureg_imm4u( struct ureg_program *ureg,
+ unsigned a, unsigned b,
+ unsigned c, unsigned d)
+{
+ unsigned v[4];
+ v[0] = a;
+ v[1] = b;
+ v[2] = c;
+ v[3] = d;
+ return ureg_DECL_immediate_uint( ureg, v, 4 );
+}
+
+static INLINE struct ureg_src
+ureg_imm3u( struct ureg_program *ureg,
+ unsigned a, unsigned b,
+ unsigned c)
+{
+ unsigned v[3];
+ v[0] = a;
+ v[1] = b;
+ v[2] = c;
+ return ureg_DECL_immediate_uint( ureg, v, 3 );
+}
+
+static INLINE struct ureg_src
+ureg_imm2u( struct ureg_program *ureg,
+ unsigned a, unsigned b)
+{
+ unsigned v[2];
+ v[0] = a;
+ v[1] = b;
+ return ureg_DECL_immediate_uint( ureg, v, 2 );
+}
+
+static INLINE struct ureg_src
+ureg_imm1u( struct ureg_program *ureg,
+ unsigned a)
+{
+ return ureg_DECL_immediate_uint( ureg, &a, 1 );
+}
+
+static INLINE struct ureg_src
+ureg_imm4i( struct ureg_program *ureg,
+ int a, int b,
+ int c, int d)
+{
+ int v[4];
+ v[0] = a;
+ v[1] = b;
+ v[2] = c;
+ v[3] = d;
+ return ureg_DECL_immediate_int( ureg, v, 4 );
+}
+
+static INLINE struct ureg_src
+ureg_imm3i( struct ureg_program *ureg,
+ int a, int b,
+ int c)
+{
+ int v[3];
+ v[0] = a;
+ v[1] = b;
+ v[2] = c;
+ return ureg_DECL_immediate_int( ureg, v, 3 );
+}
+
+static INLINE struct ureg_src
+ureg_imm2i( struct ureg_program *ureg,
+ int a, int b)
+{
+ int v[2];
+ v[0] = a;
+ v[1] = b;
+ return ureg_DECL_immediate_int( ureg, v, 2 );
+}
+
+static INLINE struct ureg_src
+ureg_imm1i( struct ureg_program *ureg,
+ int a)
+{
+ return ureg_DECL_immediate_int( ureg, &a, 1 );
+}
+
/***********************************************************************
* Functions for patching up labels
*/
diff --git a/src/gallium/auxiliary/tgsi/tgsi_util.c b/src/gallium/auxiliary/tgsi/tgsi_util.c
index 4dee1be9e8c..f4ca9e21ed9 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_util.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_util.c
@@ -76,7 +76,7 @@ tgsi_util_get_full_src_register_swizzle(
unsigned component )
{
return tgsi_util_get_src_register_swizzle(
- &reg->SrcRegister,
+ &reg->Register,
component );
}
@@ -111,10 +111,10 @@ tgsi_util_get_full_src_register_sign_mode(
{
unsigned sign_mode;
- if( reg->SrcRegisterExtMod.Absolute ) {
+ if( reg->Register.Absolute ) {
/* Consider only the post-abs negation. */
- if( reg->SrcRegisterExtMod.Negate ) {
+ if( reg->Register.Negate ) {
sign_mode = TGSI_UTIL_SIGN_SET;
}
else {
@@ -122,17 +122,7 @@ tgsi_util_get_full_src_register_sign_mode(
}
}
else {
- /* Accumulate the three negations. */
-
- unsigned negate;
-
- negate = reg->SrcRegister.Negate;
-
- if( reg->SrcRegisterExtMod.Negate ) {
- negate = !negate;
- }
-
- if( negate ) {
+ if( reg->Register.Negate ) {
sign_mode = TGSI_UTIL_SIGN_TOGGLE;
}
else {
@@ -151,27 +141,23 @@ tgsi_util_set_full_src_register_sign_mode(
switch (sign_mode)
{
case TGSI_UTIL_SIGN_CLEAR:
- reg->SrcRegister.Negate = 0;
- reg->SrcRegisterExtMod.Absolute = 1;
- reg->SrcRegisterExtMod.Negate = 0;
+ reg->Register.Negate = 0;
+ reg->Register.Absolute = 1;
break;
case TGSI_UTIL_SIGN_SET:
- reg->SrcRegister.Negate = 0;
- reg->SrcRegisterExtMod.Absolute = 1;
- reg->SrcRegisterExtMod.Negate = 1;
+ reg->Register.Absolute = 1;
+ reg->Register.Negate = 1;
break;
case TGSI_UTIL_SIGN_TOGGLE:
- reg->SrcRegister.Negate = 1;
- reg->SrcRegisterExtMod.Absolute = 0;
- reg->SrcRegisterExtMod.Negate = 0;
+ reg->Register.Negate = 1;
+ reg->Register.Absolute = 0;
break;
case TGSI_UTIL_SIGN_KEEP:
- reg->SrcRegister.Negate = 0;
- reg->SrcRegisterExtMod.Absolute = 0;
- reg->SrcRegisterExtMod.Negate = 0;
+ reg->Register.Negate = 0;
+ reg->Register.Absolute = 0;
break;
default:
diff --git a/src/gallium/auxiliary/translate/Makefile b/src/gallium/auxiliary/translate/Makefile
deleted file mode 100644
index 3c82f8ae037..00000000000
--- a/src/gallium/auxiliary/translate/Makefile
+++ /dev/null
@@ -1,12 +0,0 @@
-TOP = ../../../..
-include $(TOP)/configs/current
-
-LIBNAME = translate
-
-C_SOURCES = \
- translate_generic.c \
- translate_sse.c \
- translate.c \
- translate_cache.c
-
-include ../../Makefile.template
diff --git a/src/gallium/auxiliary/translate/SConscript b/src/gallium/auxiliary/translate/SConscript
deleted file mode 100644
index 9553a675372..00000000000
--- a/src/gallium/auxiliary/translate/SConscript
+++ /dev/null
@@ -1,12 +0,0 @@
-Import('*')
-
-translate = env.ConvenienceLibrary(
- target = 'translate',
- source = [
- 'translate_generic.c',
- 'translate_sse.c',
- 'translate.c',
- 'translate_cache.c',
- ])
-
-auxiliaries.insert(0, translate)
diff --git a/src/gallium/auxiliary/translate/translate.h b/src/gallium/auxiliary/translate/translate.h
index 34526eb0617..54ed2c1a4be 100644
--- a/src/gallium/auxiliary/translate/translate.h
+++ b/src/gallium/auxiliary/translate/translate.h
@@ -44,12 +44,19 @@
#include "pipe/p_format.h"
#include "pipe/p_state.h"
+enum translate_element_type {
+ TRANSLATE_ELEMENT_NORMAL,
+ TRANSLATE_ELEMENT_INSTANCE_ID
+};
+
struct translate_element
{
+ enum translate_element_type type;
enum pipe_format input_format;
enum pipe_format output_format;
unsigned input_buffer:8;
unsigned input_offset:24;
+ unsigned instance_divisor;
unsigned output_offset;
};
@@ -74,11 +81,13 @@ struct translate {
void (PIPE_CDECL *run_elts)( struct translate *,
const unsigned *elts,
unsigned count,
+ unsigned instance_id,
void *output_buffer);
void (PIPE_CDECL *run)( struct translate *,
unsigned start,
unsigned count,
+ unsigned instance_id,
void *output_buffer);
};
@@ -103,8 +112,13 @@ static INLINE int translate_keysize( const struct translate_key *key )
static INLINE int translate_key_compare( const struct translate_key *a,
const struct translate_key *b )
{
- int keysize = translate_keysize(a);
- return memcmp(a, b, keysize);
+ int keysize_a = translate_keysize(a);
+ int keysize_b = translate_keysize(b);
+
+ if (keysize_a != keysize_b) {
+ return keysize_a - keysize_b;
+ }
+ return memcmp(a, b, keysize_a);
}
diff --git a/src/gallium/auxiliary/translate/translate_generic.c b/src/gallium/auxiliary/translate/translate_generic.c
index 266e7ee81e6..24727d49888 100644
--- a/src/gallium/auxiliary/translate/translate_generic.c
+++ b/src/gallium/auxiliary/translate/translate_generic.c
@@ -46,9 +46,12 @@ struct translate_generic {
struct translate translate;
struct {
+ enum translate_element_type type;
+
fetch_func fetch;
unsigned buffer;
unsigned input_offset;
+ unsigned instance_divisor;
emit_func emit;
unsigned output_offset;
@@ -568,6 +571,7 @@ static emit_func get_emit_func( enum pipe_format format )
static void PIPE_CDECL generic_run_elts( struct translate *translate,
const unsigned *elts,
unsigned count,
+ unsigned instance_id,
void *output_buffer )
{
struct translate_generic *tg = translate_generic(translate);
@@ -583,13 +587,20 @@ static void PIPE_CDECL generic_run_elts( struct translate *translate,
for (attr = 0; attr < nr_attrs; attr++) {
float data[4];
-
- const char *src = (tg->attrib[attr].input_ptr +
- tg->attrib[attr].input_stride * elt);
+ const char *src;
char *dst = (vert +
tg->attrib[attr].output_offset);
+ if (tg->attrib[attr].instance_divisor) {
+ src = tg->attrib[attr].input_ptr +
+ tg->attrib[attr].input_stride *
+ (instance_id / tg->attrib[attr].instance_divisor);
+ } else {
+ src = tg->attrib[attr].input_ptr +
+ tg->attrib[attr].input_stride * elt;
+ }
+
tg->attrib[attr].fetch( src, data );
if (0) debug_printf("vert %d/%d attr %d: %f %f %f %f\n",
@@ -607,6 +618,7 @@ static void PIPE_CDECL generic_run_elts( struct translate *translate,
static void PIPE_CDECL generic_run( struct translate *translate,
unsigned start,
unsigned count,
+ unsigned instance_id,
void *output_buffer )
{
struct translate_generic *tg = translate_generic(translate);
@@ -623,13 +635,25 @@ static void PIPE_CDECL generic_run( struct translate *translate,
for (attr = 0; attr < nr_attrs; attr++) {
float data[4];
- const char *src = (tg->attrib[attr].input_ptr +
- tg->attrib[attr].input_stride * elt);
-
char *dst = (vert +
tg->attrib[attr].output_offset);
- tg->attrib[attr].fetch( src, data );
+ if (tg->attrib[attr].type == TRANSLATE_ELEMENT_NORMAL) {
+ const char *src;
+
+ if (tg->attrib[attr].instance_divisor) {
+ src = tg->attrib[attr].input_ptr +
+ tg->attrib[attr].input_stride *
+ (instance_id / tg->attrib[attr].instance_divisor);
+ } else {
+ src = tg->attrib[attr].input_ptr +
+ tg->attrib[attr].input_stride * elt;
+ }
+
+ tg->attrib[attr].fetch( src, data );
+ } else {
+ data[0] = (float)instance_id;
+ }
if (0) debug_printf("vert %d attr %d: %f %f %f %f\n",
i, attr, data[0], data[1], data[2], data[3]);
@@ -683,10 +707,12 @@ struct translate *translate_generic_create( const struct translate_key *key )
tg->translate.run = generic_run;
for (i = 0; i < key->nr_elements; i++) {
+ tg->attrib[i].type = key->element[i].type;
tg->attrib[i].fetch = get_fetch_func(key->element[i].input_format);
tg->attrib[i].buffer = key->element[i].input_buffer;
tg->attrib[i].input_offset = key->element[i].input_offset;
+ tg->attrib[i].instance_divisor = key->element[i].instance_divisor;
tg->attrib[i].emit = get_emit_func(key->element[i].output_format);
tg->attrib[i].output_offset = key->element[i].output_offset;
diff --git a/src/gallium/auxiliary/translate/translate_sse.c b/src/gallium/auxiliary/translate/translate_sse.c
index b62db8d8f33..c13e7427387 100644
--- a/src/gallium/auxiliary/translate/translate_sse.c
+++ b/src/gallium/auxiliary/translate/translate_sse.c
@@ -49,19 +49,29 @@
typedef void (PIPE_CDECL *run_func)( struct translate *translate,
unsigned start,
unsigned count,
- void *output_buffer );
+ unsigned instance_id,
+ void *output_buffer);
typedef void (PIPE_CDECL *run_elts_func)( struct translate *translate,
const unsigned *elts,
unsigned count,
- void *output_buffer );
+ unsigned instance_id,
+ void *output_buffer);
struct translate_buffer {
const void *base_ptr;
unsigned stride;
- void *ptr; /* updated per vertex */
};
+struct translate_buffer_varient {
+ unsigned buffer_index;
+ unsigned instance_divisor;
+ void *ptr; /* updated either per vertex or per instance */
+};
+
+
+#define ELEMENT_BUFFER_INSTANCE_ID 1001
+
struct translate_sse {
struct translate translate;
@@ -81,6 +91,16 @@ struct translate_sse {
struct translate_buffer buffer[PIPE_MAX_ATTRIBS];
unsigned nr_buffers;
+ /* Multiple buffer varients can map to a single buffer. */
+ struct translate_buffer_varient buffer_varient[PIPE_MAX_ATTRIBS];
+ unsigned nr_buffer_varients;
+
+ /* Multiple elements can map to a single buffer varient. */
+ unsigned element_to_buffer_varient[PIPE_MAX_ATTRIBS];
+
+ boolean use_instancing;
+ unsigned instance_id;
+
run_func gen_run;
run_elts_func gen_run_elts;
@@ -359,32 +379,61 @@ static boolean init_inputs( struct translate_sse *p,
boolean linear )
{
unsigned i;
- if (linear) {
- for (i = 0; i < p->nr_buffers; i++) {
+ struct x86_reg instance_id = x86_make_disp(p->machine_EDX,
+ get_offset(p, &p->instance_id));
+
+ for (i = 0; i < p->nr_buffer_varients; i++) {
+ struct translate_buffer_varient *varient = &p->buffer_varient[i];
+ struct translate_buffer *buffer = &p->buffer[varient->buffer_index];
+
+ if (linear || varient->instance_divisor) {
struct x86_reg buf_stride = x86_make_disp(p->machine_EDX,
- get_offset(p, &p->buffer[i].stride));
+ get_offset(p, &buffer->stride));
struct x86_reg buf_ptr = x86_make_disp(p->machine_EDX,
- get_offset(p, &p->buffer[i].ptr));
+ get_offset(p, &varient->ptr));
struct x86_reg buf_base_ptr = x86_make_disp(p->machine_EDX,
- get_offset(p, &p->buffer[i].base_ptr));
+ get_offset(p, &buffer->base_ptr));
struct x86_reg elt = p->idx_EBX;
- struct x86_reg tmp = p->tmp_EAX;
-
+ struct x86_reg tmp_EAX = p->tmp_EAX;
/* Calculate pointer to first attrib:
+ * base_ptr + stride * index, where index depends on instance divisor
*/
- x86_mov(p->func, tmp, buf_stride);
- x86_imul(p->func, tmp, elt);
- x86_add(p->func, tmp, buf_base_ptr);
+ if (varient->instance_divisor) {
+ /* Our index is instance ID divided by instance divisor.
+ */
+ x86_mov(p->func, tmp_EAX, instance_id);
+
+ if (varient->instance_divisor != 1) {
+ struct x86_reg tmp_EDX = p->machine_EDX;
+ struct x86_reg tmp_ECX = p->outbuf_ECX;
+
+ /* TODO: Add x86_shr() to rtasm and use it whenever
+ * instance divisor is power of two.
+ */
+
+ x86_push(p->func, tmp_EDX);
+ x86_push(p->func, tmp_ECX);
+ x86_xor(p->func, tmp_EDX, tmp_EDX);
+ x86_mov_reg_imm(p->func, tmp_ECX, varient->instance_divisor);
+ x86_div(p->func, tmp_ECX); /* EAX = EDX:EAX / ECX */
+ x86_pop(p->func, tmp_ECX);
+ x86_pop(p->func, tmp_EDX);
+ }
+ } else {
+ x86_mov(p->func, tmp_EAX, elt);
+ }
+ x86_imul(p->func, tmp_EAX, buf_stride);
+ x86_add(p->func, tmp_EAX, buf_base_ptr);
/* In the linear case, keep the buffer pointer instead of the
* index number.
*/
- if (p->nr_buffers == 1)
- x86_mov( p->func, elt, tmp );
+ if (linear && p->nr_buffer_varients == 1)
+ x86_mov(p->func, elt, tmp_EAX);
else
- x86_mov( p->func, buf_ptr, tmp );
+ x86_mov(p->func, buf_ptr, tmp_EAX);
}
}
@@ -394,31 +443,36 @@ static boolean init_inputs( struct translate_sse *p,
static struct x86_reg get_buffer_ptr( struct translate_sse *p,
boolean linear,
- unsigned buf_idx,
+ unsigned var_idx,
struct x86_reg elt )
{
- if (linear && p->nr_buffers == 1) {
+ if (var_idx == ELEMENT_BUFFER_INSTANCE_ID) {
+ return x86_make_disp(p->machine_EDX,
+ get_offset(p, &p->instance_id));
+ }
+ if (linear && p->nr_buffer_varients == 1) {
return p->idx_EBX;
}
- else if (linear) {
+ else if (linear || p->buffer_varient[var_idx].instance_divisor) {
struct x86_reg ptr = p->tmp_EAX;
struct x86_reg buf_ptr =
x86_make_disp(p->machine_EDX,
- get_offset(p, &p->buffer[buf_idx].ptr));
+ get_offset(p, &p->buffer_varient[var_idx].ptr));
x86_mov(p->func, ptr, buf_ptr);
return ptr;
}
else {
struct x86_reg ptr = p->tmp_EAX;
+ const struct translate_buffer_varient *varient = &p->buffer_varient[var_idx];
struct x86_reg buf_stride =
x86_make_disp(p->machine_EDX,
- get_offset(p, &p->buffer[buf_idx].stride));
+ get_offset(p, &p->buffer[varient->buffer_index].stride));
struct x86_reg buf_base_ptr =
x86_make_disp(p->machine_EDX,
- get_offset(p, &p->buffer[buf_idx].base_ptr));
+ get_offset(p, &p->buffer[varient->buffer_index].base_ptr));
@@ -436,28 +490,33 @@ static struct x86_reg get_buffer_ptr( struct translate_sse *p,
static boolean incr_inputs( struct translate_sse *p,
boolean linear )
{
- if (linear && p->nr_buffers == 1) {
+ if (linear && p->nr_buffer_varients == 1) {
struct x86_reg stride = x86_make_disp(p->machine_EDX,
get_offset(p, &p->buffer[0].stride));
- x86_add(p->func, p->idx_EBX, stride);
- sse_prefetchnta(p->func, x86_make_disp(p->idx_EBX, 192));
+ if (p->buffer_varient[0].instance_divisor == 0) {
+ x86_add(p->func, p->idx_EBX, stride);
+ sse_prefetchnta(p->func, x86_make_disp(p->idx_EBX, 192));
+ }
}
else if (linear) {
unsigned i;
/* Is this worthwhile??
*/
- for (i = 0; i < p->nr_buffers; i++) {
+ for (i = 0; i < p->nr_buffer_varients; i++) {
+ struct translate_buffer_varient *varient = &p->buffer_varient[i];
struct x86_reg buf_ptr = x86_make_disp(p->machine_EDX,
- get_offset(p, &p->buffer[i].ptr));
+ get_offset(p, &varient->ptr));
struct x86_reg buf_stride = x86_make_disp(p->machine_EDX,
- get_offset(p, &p->buffer[i].stride));
+ get_offset(p, &p->buffer[varient->buffer_index].stride));
- x86_mov(p->func, p->tmp_EAX, buf_ptr);
- x86_add(p->func, p->tmp_EAX, buf_stride);
- if (i == 0) sse_prefetchnta(p->func, x86_make_disp(p->tmp_EAX, 192));
- x86_mov(p->func, buf_ptr, p->tmp_EAX);
+ if (varient->instance_divisor == 0) {
+ x86_mov(p->func, p->tmp_EAX, buf_ptr);
+ x86_add(p->func, p->tmp_EAX, buf_stride);
+ if (i == 0) sse_prefetchnta(p->func, x86_make_disp(p->tmp_EAX, 192));
+ x86_mov(p->func, buf_ptr, p->tmp_EAX);
+ }
}
}
else {
@@ -514,7 +573,18 @@ static boolean build_vertex_emit( struct translate_sse *p,
x86_mov(p->func, p->machine_EDX, x86_fn_arg(p->func, 1));
x86_mov(p->func, p->idx_EBX, x86_fn_arg(p->func, 2));
x86_mov(p->func, p->count_ESI, x86_fn_arg(p->func, 3));
- x86_mov(p->func, p->outbuf_ECX, x86_fn_arg(p->func, 4));
+ x86_mov(p->func, p->outbuf_ECX, x86_fn_arg(p->func, 5));
+
+ /* Load instance ID.
+ */
+ if (p->use_instancing) {
+ x86_mov(p->func,
+ p->tmp_EAX,
+ x86_fn_arg(p->func, 4));
+ x86_mov(p->func,
+ x86_make_disp(p->machine_EDX, get_offset(p, &p->instance_id)),
+ p->tmp_EAX);
+ }
/* Get vertex count, compare to zero
*/
@@ -531,17 +601,18 @@ static boolean build_vertex_emit( struct translate_sse *p,
label = x86_get_label(p->func);
{
struct x86_reg elt = linear ? p->idx_EBX : x86_deref(p->idx_EBX);
- int last_vb = -1;
+ int last_varient = -1;
struct x86_reg vb;
for (j = 0; j < p->translate.key.nr_elements; j++) {
const struct translate_element *a = &p->translate.key.element[j];
+ unsigned varient = p->element_to_buffer_varient[j];
/* Figure out source pointer address:
*/
- if (a->input_buffer != last_vb) {
- last_vb = a->input_buffer;
- vb = get_buffer_ptr(p, linear, a->input_buffer, elt);
+ if (varient != last_varient) {
+ last_varient = varient;
+ vb = get_buffer_ptr(p, linear, varient, elt);
}
if (!translate_attr( p, a,
@@ -624,6 +695,7 @@ static void translate_sse_release( struct translate *translate )
static void PIPE_CDECL translate_sse_run_elts( struct translate *translate,
const unsigned *elts,
unsigned count,
+ unsigned instance_id,
void *output_buffer )
{
struct translate_sse *p = (struct translate_sse *)translate;
@@ -631,12 +703,14 @@ static void PIPE_CDECL translate_sse_run_elts( struct translate *translate,
p->gen_run_elts( translate,
elts,
count,
- output_buffer );
+ instance_id,
+ output_buffer);
}
static void PIPE_CDECL translate_sse_run( struct translate *translate,
unsigned start,
unsigned count,
+ unsigned instance_id,
void *output_buffer )
{
struct translate_sse *p = (struct translate_sse *)translate;
@@ -644,7 +718,8 @@ static void PIPE_CDECL translate_sse_run( struct translate *translate,
p->gen_run( translate,
start,
count,
- output_buffer );
+ instance_id,
+ output_buffer);
}
@@ -666,8 +741,37 @@ struct translate *translate_sse2_create( const struct translate_key *key )
p->translate.run_elts = translate_sse_run_elts;
p->translate.run = translate_sse_run;
- for (i = 0; i < key->nr_elements; i++)
- p->nr_buffers = MAX2( p->nr_buffers, key->element[i].input_buffer + 1 );
+ for (i = 0; i < key->nr_elements; i++) {
+ if (key->element[i].type == TRANSLATE_ELEMENT_NORMAL) {
+ unsigned j;
+
+ p->nr_buffers = MAX2(p->nr_buffers, key->element[i].input_buffer + 1);
+
+ if (key->element[i].instance_divisor) {
+ p->use_instancing = TRUE;
+ }
+
+ /*
+ * Map vertex element to vertex buffer varient.
+ */
+ for (j = 0; j < p->nr_buffer_varients; j++) {
+ if (p->buffer_varient[j].buffer_index == key->element[i].input_buffer &&
+ p->buffer_varient[j].instance_divisor == key->element[i].instance_divisor) {
+ break;
+ }
+ }
+ if (j == p->nr_buffer_varients) {
+ p->buffer_varient[j].buffer_index = key->element[i].input_buffer;
+ p->buffer_varient[j].instance_divisor = key->element[i].instance_divisor;
+ p->nr_buffer_varients++;
+ }
+ p->element_to_buffer_varient[i] = j;
+ } else {
+ assert(key->element[i].type == TRANSLATE_ELEMENT_INSTANCE_ID);
+
+ p->element_to_buffer_varient[i] = ELEMENT_BUFFER_INSTANCE_ID;
+ }
+ }
if (0) debug_printf("nr_buffers: %d\n", p->nr_buffers);
diff --git a/src/gallium/auxiliary/util/Makefile b/src/gallium/auxiliary/util/Makefile
deleted file mode 100644
index 7e47ec6d3ba..00000000000
--- a/src/gallium/auxiliary/util/Makefile
+++ /dev/null
@@ -1,46 +0,0 @@
-TOP = ../../../..
-include $(TOP)/configs/current
-
-LIBNAME = util
-
-C_SOURCES = \
- u_debug.c \
- u_debug_dump.c \
- u_debug_symbol.c \
- u_debug_stack.c \
- u_blit.c \
- u_bitmask.c \
- u_cache.c \
- u_cpu_detect.c \
- u_draw_quad.c \
- u_format.c \
- u_format_access.c \
- u_format_table.c \
- u_gen_mipmap.c \
- u_handle_table.c \
- u_hash_table.c \
- u_hash.c \
- u_keymap.c \
- u_linear.c \
- u_network.c \
- u_math.c \
- u_mm.c \
- u_rect.c \
- u_simple_shaders.c \
- u_snprintf.c \
- u_stream_stdc.c \
- u_stream_wd.c \
- u_surface.c \
- u_tile.c \
- u_time.c \
- u_timed_winsys.c \
- u_upload_mgr.c \
- u_simple_screen.c
-
-include ../../Makefile.template
-
-u_format_table.c: u_format_table.py u_format_parse.py u_format.csv
- python u_format_table.py u_format.csv > $@
-
-u_format_access.c: u_format_access.py u_format_parse.py u_format.csv
- python u_format_access.py u_format.csv > $@
diff --git a/src/gallium/auxiliary/util/SConscript b/src/gallium/auxiliary/util/SConscript
deleted file mode 100644
index 8d99106d0b8..00000000000
--- a/src/gallium/auxiliary/util/SConscript
+++ /dev/null
@@ -1,58 +0,0 @@
-Import('*')
-
-env.Clone()
-
-env.Append(CPPPATH = ['.'])
-
-env.CodeGenerate(
- target = 'u_format_table.c',
- script = 'u_format_table.py',
- source = ['u_format.csv'],
- command = 'python $SCRIPT $SOURCE > $TARGET'
-)
-
-env.CodeGenerate(
- target = 'u_format_access.c',
- script = 'u_format_access.py',
- source = ['u_format.csv'],
- command = 'python $SCRIPT $SOURCE > $TARGET'
-)
-
-util = env.ConvenienceLibrary(
- target = 'util',
- source = [
- 'u_bitmask.c',
- 'u_blit.c',
- 'u_cache.c',
- 'u_cpu_detect.c',
- 'u_debug.c',
- 'u_debug_dump.c',
- 'u_debug_memory.c',
- 'u_debug_stack.c',
- 'u_debug_symbol.c',
- 'u_draw_quad.c',
- 'u_format.c',
- 'u_format_access.c',
- 'u_format_table.c',
- 'u_gen_mipmap.c',
- 'u_handle_table.c',
- 'u_hash.c',
- 'u_hash_table.c',
- 'u_keymap.c',
- 'u_network.c',
- 'u_math.c',
- 'u_mm.c',
- 'u_rect.c',
- 'u_simple_shaders.c',
- 'u_snprintf.c',
- 'u_stream_stdc.c',
- 'u_stream_wd.c',
- 'u_surface.c',
- 'u_tile.c',
- 'u_time.c',
- 'u_timed_winsys.c',
- 'u_upload_mgr.c',
- 'u_simple_screen.c',
- ])
-
-auxiliaries.insert(0, util)
diff --git a/src/gallium/auxiliary/util/u_blit.c b/src/gallium/auxiliary/util/u_blit.c
index 50386425995..3f74e2aa8b8 100644
--- a/src/gallium/auxiliary/util/u_blit.c
+++ b/src/gallium/auxiliary/util/u_blit.c
@@ -42,6 +42,7 @@
#include "util/u_blit.h"
#include "util/u_draw_quad.h"
+#include "util/u_format.h"
#include "util/u_math.h"
#include "util/u_memory.h"
#include "util/u_simple_shaders.h"
@@ -126,7 +127,8 @@ util_create_blit(struct pipe_context *pipe, struct cso_context *cso)
}
/* fragment shader */
- ctx->fs[TGSI_WRITEMASK_XYZW] = util_make_fragment_tex_shader(pipe);
+ ctx->fs[TGSI_WRITEMASK_XYZW] =
+ util_make_fragment_tex_shader(pipe, TGSI_TEXTURE_2D);
ctx->vbuf = NULL;
/* init vertex data that doesn't change */
@@ -354,10 +356,9 @@ util_blit_pixels_writemask(struct blit_state *ctx,
texTemp.target = PIPE_TEXTURE_2D;
texTemp.format = src->format;
texTemp.last_level = 0;
- texTemp.width[0] = srcW;
- texTemp.height[0] = srcH;
- texTemp.depth[0] = 1;
- pf_get_block(src->format, &texTemp.block);
+ texTemp.width0 = srcW;
+ texTemp.height0 = srcH;
+ texTemp.depth0 = 1;
tex = screen->texture_create(screen, &texTemp);
if (!tex)
@@ -389,10 +390,10 @@ util_blit_pixels_writemask(struct blit_state *ctx,
}
else {
pipe_texture_reference(&tex, src->texture);
- s0 = srcX0 / (float)tex->width[0];
- s1 = srcX1 / (float)tex->width[0];
- t0 = srcY0 / (float)tex->height[0];
- t1 = srcY1 / (float)tex->height[0];
+ s0 = srcX0 / (float)tex->width0;
+ s1 = srcX1 / (float)tex->width0;
+ t0 = srcY0 / (float)tex->height0;
+ t1 = srcY1 / (float)tex->height0;
}
@@ -421,7 +422,9 @@ util_blit_pixels_writemask(struct blit_state *ctx,
cso_set_sampler_textures(ctx->cso, 1, &tex);
if (ctx->fs[writemask] == NULL)
- ctx->fs[writemask] = util_make_fragment_tex_shader_writemask(pipe, writemask);
+ ctx->fs[writemask] =
+ util_make_fragment_tex_shader_writemask(pipe, TGSI_TEXTURE_2D,
+ writemask);
/* shaders */
cso_set_fragment_shader_handle(ctx->cso, ctx->fs[writemask]);
@@ -518,13 +521,13 @@ util_blit_pixels_tex(struct blit_state *ctx,
assert(filter == PIPE_TEX_MIPFILTER_NEAREST ||
filter == PIPE_TEX_MIPFILTER_LINEAR);
- assert(tex->width[0] != 0);
- assert(tex->height[0] != 0);
+ assert(tex->width0 != 0);
+ assert(tex->height0 != 0);
- s0 = srcX0 / (float)tex->width[0];
- s1 = srcX1 / (float)tex->width[0];
- t0 = srcY0 / (float)tex->height[0];
- t1 = srcY1 / (float)tex->height[0];
+ s0 = srcX0 / (float)tex->width0;
+ s1 = srcX1 / (float)tex->width0;
+ t0 = srcY0 / (float)tex->height0;
+ t1 = srcY1 / (float)tex->height0;
assert(ctx->pipe->screen->is_format_supported(ctx->pipe->screen, dst->format,
PIPE_TEXTURE_2D,
diff --git a/src/gallium/auxiliary/util/u_blitter.c b/src/gallium/auxiliary/util/u_blitter.c
new file mode 100644
index 00000000000..249a0375fc5
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_blitter.c
@@ -0,0 +1,820 @@
+/**************************************************************************
+ *
+ * Copyright 2009 Marek Olšák <[email protected]>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+/**
+ * @file
+ * Blitter utility to facilitate acceleration of the clear, surface_copy,
+ * and surface_fill functions.
+ *
+ * @author Marek Olšák
+ */
+
+#include "pipe/p_context.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_inlines.h"
+#include "pipe/p_shader_tokens.h"
+#include "pipe/p_state.h"
+
+#include "util/u_format.h"
+#include "util/u_memory.h"
+#include "util/u_math.h"
+#include "util/u_blitter.h"
+#include "util/u_draw_quad.h"
+#include "util/u_pack_color.h"
+#include "util/u_rect.h"
+#include "util/u_simple_shaders.h"
+#include "util/u_texture.h"
+
+#define INVALID_PTR ((void*)~0)
+
+struct blitter_context_priv
+{
+ struct blitter_context blitter;
+
+ struct pipe_context *pipe; /**< pipe context */
+ struct pipe_buffer *vbuf; /**< quad */
+
+ float vertices[4][2][4]; /**< {pos, color} or {pos, texcoord} */
+
+ /* Templates for various state objects. */
+ struct pipe_depth_stencil_alpha_state template_dsa;
+ struct pipe_sampler_state template_sampler_state;
+
+ /* Constant state objects. */
+ /* Vertex shaders. */
+ void *vs_col; /**< Vertex shader which passes {pos, color} to the output */
+ void *vs_tex; /**<Vertex shader which passes {pos, texcoord} to the output.*/
+
+ /* Fragment shaders. */
+ /* FS which outputs a color to multiple color buffers. */
+ void *fs_col[PIPE_MAX_COLOR_BUFS];
+
+ /* FS which outputs a color from a texture,
+ where the index is PIPE_TEXTURE_* to be sampled. */
+ void *fs_texfetch_col[PIPE_MAX_TEXTURE_TYPES];
+
+ /* FS which outputs a depth from a texture,
+ where the index is PIPE_TEXTURE_* to be sampled. */
+ void *fs_texfetch_depth[PIPE_MAX_TEXTURE_TYPES];
+
+ /* Blend state. */
+ void *blend_write_color; /**< blend state with writemask of RGBA */
+ void *blend_keep_color; /**< blend state with writemask of 0 */
+
+ /* Depth stencil alpha state. */
+ void *dsa_write_depth_stencil[0xff]; /**< indices are stencil clear values */
+ void *dsa_write_depth_keep_stencil;
+ void *dsa_keep_depth_stencil;
+
+ /* Sampler state for clamping to a miplevel. */
+ void *sampler_state[PIPE_MAX_TEXTURE_LEVELS];
+
+ /* Rasterizer state. */
+ void *rs_state;
+};
+
+struct blitter_context *util_blitter_create(struct pipe_context *pipe)
+{
+ struct blitter_context_priv *ctx;
+ struct pipe_blend_state blend;
+ struct pipe_depth_stencil_alpha_state *dsa;
+ struct pipe_rasterizer_state rs_state;
+ struct pipe_sampler_state *sampler_state;
+ unsigned i;
+
+ ctx = CALLOC_STRUCT(blitter_context_priv);
+ if (!ctx)
+ return NULL;
+
+ ctx->pipe = pipe;
+
+ /* init state objects for them to be considered invalid */
+ ctx->blitter.saved_blend_state = INVALID_PTR;
+ ctx->blitter.saved_dsa_state = INVALID_PTR;
+ ctx->blitter.saved_rs_state = INVALID_PTR;
+ ctx->blitter.saved_fs = INVALID_PTR;
+ ctx->blitter.saved_vs = INVALID_PTR;
+ ctx->blitter.saved_fb_state.nr_cbufs = ~0;
+ ctx->blitter.saved_num_textures = ~0;
+ ctx->blitter.saved_num_sampler_states = ~0;
+
+ /* blend state objects */
+ memset(&blend, 0, sizeof(blend));
+ ctx->blend_keep_color = pipe->create_blend_state(pipe, &blend);
+
+ blend.colormask = PIPE_MASK_RGBA;
+ ctx->blend_write_color = pipe->create_blend_state(pipe, &blend);
+
+ /* depth stencil alpha state objects */
+ dsa = &ctx->template_dsa;
+ ctx->dsa_keep_depth_stencil =
+ pipe->create_depth_stencil_alpha_state(pipe, dsa);
+
+ dsa->depth.enabled = 1;
+ dsa->depth.writemask = 1;
+ dsa->depth.func = PIPE_FUNC_ALWAYS;
+ ctx->dsa_write_depth_keep_stencil =
+ pipe->create_depth_stencil_alpha_state(pipe, dsa);
+
+ dsa->stencil[0].enabled = 1;
+ dsa->stencil[0].func = PIPE_FUNC_ALWAYS;
+ dsa->stencil[0].fail_op = PIPE_STENCIL_OP_REPLACE;
+ dsa->stencil[0].zpass_op = PIPE_STENCIL_OP_REPLACE;
+ dsa->stencil[0].zfail_op = PIPE_STENCIL_OP_REPLACE;
+ dsa->stencil[0].valuemask = 0xff;
+ dsa->stencil[0].writemask = 0xff;
+ /* The DSA state objects which write depth and stencil are created
+ * on-demand. */
+
+ /* sampler state */
+ sampler_state = &ctx->template_sampler_state;
+ sampler_state->wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+ sampler_state->wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+ sampler_state->wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE;
+ /* The sampler state objects which sample from a specified mipmap level
+ * are created on-demand. */
+
+ /* rasterizer state */
+ memset(&rs_state, 0, sizeof(rs_state));
+ rs_state.front_winding = PIPE_WINDING_CW;
+ rs_state.cull_mode = PIPE_WINDING_NONE;
+ rs_state.bypass_vs_clip_and_viewport = 1;
+ rs_state.gl_rasterization_rules = 1;
+ rs_state.flatshade = 1;
+ ctx->rs_state = pipe->create_rasterizer_state(pipe, &rs_state);
+
+ /* fragment shaders are created on-demand */
+
+ /* vertex shaders */
+ {
+ const uint semantic_names[] = { TGSI_SEMANTIC_POSITION,
+ TGSI_SEMANTIC_COLOR };
+ const uint semantic_indices[] = { 0, 0 };
+ ctx->vs_col =
+ util_make_vertex_passthrough_shader(pipe, 2, semantic_names,
+ semantic_indices);
+ }
+ {
+ const uint semantic_names[] = { TGSI_SEMANTIC_POSITION,
+ TGSI_SEMANTIC_GENERIC };
+ const uint semantic_indices[] = { 0, 0 };
+ ctx->vs_tex =
+ util_make_vertex_passthrough_shader(pipe, 2, semantic_names,
+ semantic_indices);
+ }
+
+ /* set invariant vertex coordinates */
+ for (i = 0; i < 4; i++)
+ ctx->vertices[i][0][3] = 1; /*v.w*/
+
+ /* create the vertex buffer */
+ ctx->vbuf = pipe_buffer_create(ctx->pipe->screen,
+ 32,
+ PIPE_BUFFER_USAGE_VERTEX,
+ sizeof(ctx->vertices));
+
+ return &ctx->blitter;
+}
+
+void util_blitter_destroy(struct blitter_context *blitter)
+{
+ struct blitter_context_priv *ctx = (struct blitter_context_priv*)blitter;
+ struct pipe_context *pipe = ctx->pipe;
+ int i;
+
+ pipe->delete_blend_state(pipe, ctx->blend_write_color);
+ pipe->delete_blend_state(pipe, ctx->blend_keep_color);
+ pipe->delete_depth_stencil_alpha_state(pipe, ctx->dsa_keep_depth_stencil);
+ pipe->delete_depth_stencil_alpha_state(pipe,
+ ctx->dsa_write_depth_keep_stencil);
+
+ for (i = 0; i < 0xff; i++)
+ if (ctx->dsa_write_depth_stencil[i])
+ pipe->delete_depth_stencil_alpha_state(pipe,
+ ctx->dsa_write_depth_stencil[i]);
+
+ pipe->delete_rasterizer_state(pipe, ctx->rs_state);
+ pipe->delete_vs_state(pipe, ctx->vs_col);
+ pipe->delete_vs_state(pipe, ctx->vs_tex);
+
+ for (i = 0; i < PIPE_MAX_TEXTURE_TYPES; i++) {
+ if (ctx->fs_texfetch_col[i])
+ pipe->delete_fs_state(pipe, ctx->fs_texfetch_col[i]);
+ if (ctx->fs_texfetch_depth[i])
+ pipe->delete_fs_state(pipe, ctx->fs_texfetch_depth[i]);
+ }
+
+ for (i = 0; i < PIPE_MAX_COLOR_BUFS && ctx->fs_col[i]; i++)
+ if (ctx->fs_col[i])
+ pipe->delete_fs_state(pipe, ctx->fs_col[i]);
+
+ for (i = 0; i < PIPE_MAX_TEXTURE_LEVELS; i++)
+ if (ctx->sampler_state[i])
+ pipe->delete_sampler_state(pipe, ctx->sampler_state[i]);
+
+ pipe_buffer_reference(&ctx->vbuf, NULL);
+ FREE(ctx);
+}
+
+static void blitter_check_saved_CSOs(struct blitter_context_priv *ctx)
+{
+ /* make sure these CSOs have been saved */
+ assert(ctx->blitter.saved_blend_state != INVALID_PTR &&
+ ctx->blitter.saved_dsa_state != INVALID_PTR &&
+ ctx->blitter.saved_rs_state != INVALID_PTR &&
+ ctx->blitter.saved_fs != INVALID_PTR &&
+ ctx->blitter.saved_vs != INVALID_PTR);
+}
+
+static void blitter_restore_CSOs(struct blitter_context_priv *ctx)
+{
+ struct pipe_context *pipe = ctx->pipe;
+
+ /* restore the state objects which are always required to be saved */
+ pipe->bind_blend_state(pipe, ctx->blitter.saved_blend_state);
+ pipe->bind_depth_stencil_alpha_state(pipe, ctx->blitter.saved_dsa_state);
+ pipe->bind_rasterizer_state(pipe, ctx->blitter.saved_rs_state);
+ pipe->bind_fs_state(pipe, ctx->blitter.saved_fs);
+ pipe->bind_vs_state(pipe, ctx->blitter.saved_vs);
+
+ ctx->blitter.saved_blend_state = INVALID_PTR;
+ ctx->blitter.saved_dsa_state = INVALID_PTR;
+ ctx->blitter.saved_rs_state = INVALID_PTR;
+ ctx->blitter.saved_fs = INVALID_PTR;
+ ctx->blitter.saved_vs = INVALID_PTR;
+
+ /* restore the state objects which are required to be saved before copy/fill
+ */
+ if (ctx->blitter.saved_fb_state.nr_cbufs != ~0) {
+ pipe->set_framebuffer_state(pipe, &ctx->blitter.saved_fb_state);
+ ctx->blitter.saved_fb_state.nr_cbufs = ~0;
+ }
+
+ if (ctx->blitter.saved_num_sampler_states != ~0) {
+ pipe->bind_fragment_sampler_states(pipe,
+ ctx->blitter.saved_num_sampler_states,
+ ctx->blitter.saved_sampler_states);
+ ctx->blitter.saved_num_sampler_states = ~0;
+ }
+
+ if (ctx->blitter.saved_num_textures != ~0) {
+ pipe->set_fragment_sampler_textures(pipe,
+ ctx->blitter.saved_num_textures,
+ ctx->blitter.saved_textures);
+ ctx->blitter.saved_num_textures = ~0;
+ }
+}
+
+static void blitter_set_rectangle(struct blitter_context_priv *ctx,
+ unsigned x1, unsigned y1,
+ unsigned x2, unsigned y2,
+ float depth)
+{
+ int i;
+
+ /* set vertex positions */
+ ctx->vertices[0][0][0] = x1; /*v0.x*/
+ ctx->vertices[0][0][1] = y1; /*v0.y*/
+
+ ctx->vertices[1][0][0] = x2; /*v1.x*/
+ ctx->vertices[1][0][1] = y1; /*v1.y*/
+
+ ctx->vertices[2][0][0] = x2; /*v2.x*/
+ ctx->vertices[2][0][1] = y2; /*v2.y*/
+
+ ctx->vertices[3][0][0] = x1; /*v3.x*/
+ ctx->vertices[3][0][1] = y2; /*v3.y*/
+
+ for (i = 0; i < 4; i++)
+ ctx->vertices[i][0][2] = depth; /*z*/
+}
+
+static void blitter_set_clear_color(struct blitter_context_priv *ctx,
+ const float *rgba)
+{
+ int i;
+
+ for (i = 0; i < 4; i++) {
+ ctx->vertices[i][1][0] = rgba[0];
+ ctx->vertices[i][1][1] = rgba[1];
+ ctx->vertices[i][1][2] = rgba[2];
+ ctx->vertices[i][1][3] = rgba[3];
+ }
+}
+
+static void blitter_set_texcoords_2d(struct blitter_context_priv *ctx,
+ struct pipe_surface *surf,
+ unsigned x1, unsigned y1,
+ unsigned x2, unsigned y2)
+{
+ int i;
+ float s1 = x1 / (float)surf->width;
+ float t1 = y1 / (float)surf->height;
+ float s2 = x2 / (float)surf->width;
+ float t2 = y2 / (float)surf->height;
+
+ ctx->vertices[0][1][0] = s1; /*t0.s*/
+ ctx->vertices[0][1][1] = t1; /*t0.t*/
+
+ ctx->vertices[1][1][0] = s2; /*t1.s*/
+ ctx->vertices[1][1][1] = t1; /*t1.t*/
+
+ ctx->vertices[2][1][0] = s2; /*t2.s*/
+ ctx->vertices[2][1][1] = t2; /*t2.t*/
+
+ ctx->vertices[3][1][0] = s1; /*t3.s*/
+ ctx->vertices[3][1][1] = t2; /*t3.t*/
+
+ for (i = 0; i < 4; i++) {
+ ctx->vertices[i][1][2] = 0; /*r*/
+ ctx->vertices[i][1][3] = 1; /*q*/
+ }
+}
+
+static void blitter_set_texcoords_3d(struct blitter_context_priv *ctx,
+ struct pipe_surface *surf,
+ unsigned x1, unsigned y1,
+ unsigned x2, unsigned y2)
+{
+ int i;
+ float depth = u_minify(surf->texture->depth0, surf->level);
+ float r = surf->zslice / depth;
+
+ blitter_set_texcoords_2d(ctx, surf, x1, y1, x2, y2);
+
+ for (i = 0; i < 4; i++)
+ ctx->vertices[i][1][2] = r; /*r*/
+}
+
+static void blitter_set_texcoords_cube(struct blitter_context_priv *ctx,
+ struct pipe_surface *surf,
+ unsigned x1, unsigned y1,
+ unsigned x2, unsigned y2)
+{
+ int i;
+ float s1 = x1 / (float)surf->width;
+ float t1 = y1 / (float)surf->height;
+ float s2 = x2 / (float)surf->width;
+ float t2 = y2 / (float)surf->height;
+ float st[4][2];
+
+ st[0][0] = s1;
+ st[0][1] = t1;
+ st[1][0] = s2;
+ st[1][1] = t1;
+ st[2][0] = s2;
+ st[2][1] = t2;
+ st[3][0] = s1;
+ st[3][1] = t2;
+
+ util_map_texcoords2d_onto_cubemap(surf->face,
+ /* pointer, stride in floats */
+ &st[0][0], 2,
+ &ctx->vertices[0][1][0], 8);
+
+ for (i = 0; i < 4; i++)
+ ctx->vertices[i][1][3] = 1; /*q*/
+}
+
+static void blitter_draw_quad(struct blitter_context_priv *ctx)
+{
+ struct pipe_context *pipe = ctx->pipe;
+
+ /* write vertices and draw them */
+ pipe_buffer_write(pipe->screen, ctx->vbuf,
+ 0, sizeof(ctx->vertices), ctx->vertices);
+
+ util_draw_vertex_buffer(pipe, ctx->vbuf, 0, PIPE_PRIM_TRIANGLE_FAN,
+ 4, /* verts */
+ 2); /* attribs/vert */
+}
+
+static INLINE
+void *blitter_get_state_write_depth_stencil(
+ struct blitter_context_priv *ctx,
+ unsigned stencil)
+{
+ struct pipe_context *pipe = ctx->pipe;
+
+ stencil &= 0xff;
+
+ /* Create the DSA state on-demand. */
+ if (!ctx->dsa_write_depth_stencil[stencil]) {
+ ctx->template_dsa.stencil[0].ref_value = stencil;
+
+ ctx->dsa_write_depth_stencil[stencil] =
+ pipe->create_depth_stencil_alpha_state(pipe, &ctx->template_dsa);
+ }
+
+ return ctx->dsa_write_depth_stencil[stencil];
+}
+
+static INLINE
+void **blitter_get_sampler_state(struct blitter_context_priv *ctx,
+ int miplevel)
+{
+ struct pipe_context *pipe = ctx->pipe;
+ struct pipe_sampler_state *sampler_state = &ctx->template_sampler_state;
+
+ assert(miplevel < PIPE_MAX_TEXTURE_LEVELS);
+
+ /* Create the sampler state on-demand. */
+ if (!ctx->sampler_state[miplevel]) {
+ sampler_state->lod_bias = miplevel;
+ sampler_state->min_lod = miplevel;
+ sampler_state->max_lod = miplevel;
+
+ ctx->sampler_state[miplevel] = pipe->create_sampler_state(pipe,
+ sampler_state);
+ }
+
+ /* Return void** so that it can be passed to bind_fragment_sampler_states
+ * directly. */
+ return &ctx->sampler_state[miplevel];
+}
+
+static INLINE
+void *blitter_get_fs_col(struct blitter_context_priv *ctx, unsigned num_cbufs)
+{
+ struct pipe_context *pipe = ctx->pipe;
+ unsigned index = num_cbufs ? num_cbufs - 1 : 0;
+
+ assert(num_cbufs <= PIPE_MAX_COLOR_BUFS);
+
+ if (!ctx->fs_col[index])
+ ctx->fs_col[index] =
+ util_make_fragment_clonecolor_shader(pipe, num_cbufs);
+
+ return ctx->fs_col[index];
+}
+
+static INLINE
+void *blitter_get_fs_texfetch_col(struct blitter_context_priv *ctx,
+ unsigned tex_target)
+{
+ struct pipe_context *pipe = ctx->pipe;
+
+ assert(tex_target < PIPE_MAX_TEXTURE_TYPES);
+
+ /* Create the fragment shader on-demand. */
+ if (!ctx->fs_texfetch_col[tex_target]) {
+ switch (tex_target) {
+ case PIPE_TEXTURE_1D:
+ ctx->fs_texfetch_col[PIPE_TEXTURE_1D] =
+ util_make_fragment_tex_shader(pipe, TGSI_TEXTURE_1D);
+ break;
+ case PIPE_TEXTURE_2D:
+ ctx->fs_texfetch_col[PIPE_TEXTURE_2D] =
+ util_make_fragment_tex_shader(pipe, TGSI_TEXTURE_2D);
+ break;
+ case PIPE_TEXTURE_3D:
+ ctx->fs_texfetch_col[PIPE_TEXTURE_3D] =
+ util_make_fragment_tex_shader(pipe, TGSI_TEXTURE_3D);
+ break;
+ case PIPE_TEXTURE_CUBE:
+ ctx->fs_texfetch_col[PIPE_TEXTURE_CUBE] =
+ util_make_fragment_tex_shader(pipe, TGSI_TEXTURE_CUBE);
+ break;
+ default:;
+ }
+ }
+
+ return ctx->fs_texfetch_col[tex_target];
+}
+
+static INLINE
+void *blitter_get_fs_texfetch_depth(struct blitter_context_priv *ctx,
+ unsigned tex_target)
+{
+ struct pipe_context *pipe = ctx->pipe;
+
+ assert(tex_target < PIPE_MAX_TEXTURE_TYPES);
+
+ /* Create the fragment shader on-demand. */
+ if (!ctx->fs_texfetch_depth[tex_target]) {
+ switch (tex_target) {
+ case PIPE_TEXTURE_1D:
+ ctx->fs_texfetch_depth[PIPE_TEXTURE_1D] =
+ util_make_fragment_tex_shader_writedepth(pipe, TGSI_TEXTURE_1D);
+ break;
+ case PIPE_TEXTURE_2D:
+ ctx->fs_texfetch_depth[PIPE_TEXTURE_2D] =
+ util_make_fragment_tex_shader_writedepth(pipe, TGSI_TEXTURE_2D);
+ break;
+ case PIPE_TEXTURE_3D:
+ ctx->fs_texfetch_depth[PIPE_TEXTURE_3D] =
+ util_make_fragment_tex_shader_writedepth(pipe, TGSI_TEXTURE_3D);
+ break;
+ case PIPE_TEXTURE_CUBE:
+ ctx->fs_texfetch_depth[PIPE_TEXTURE_CUBE] =
+ util_make_fragment_tex_shader_writedepth(pipe,TGSI_TEXTURE_CUBE);
+ break;
+ default:;
+ }
+ }
+
+ return ctx->fs_texfetch_depth[tex_target];
+}
+
+void util_blitter_clear(struct blitter_context *blitter,
+ unsigned width, unsigned height,
+ unsigned num_cbufs,
+ unsigned clear_buffers,
+ const float *rgba,
+ double depth, unsigned stencil)
+{
+ struct blitter_context_priv *ctx = (struct blitter_context_priv*)blitter;
+ struct pipe_context *pipe = ctx->pipe;
+
+ assert(num_cbufs <= PIPE_MAX_COLOR_BUFS);
+
+ blitter_check_saved_CSOs(ctx);
+
+ /* bind CSOs */
+ if (clear_buffers & PIPE_CLEAR_COLOR)
+ pipe->bind_blend_state(pipe, ctx->blend_write_color);
+ else
+ pipe->bind_blend_state(pipe, ctx->blend_keep_color);
+
+ if (clear_buffers & PIPE_CLEAR_DEPTHSTENCIL)
+ pipe->bind_depth_stencil_alpha_state(pipe,
+ blitter_get_state_write_depth_stencil(ctx, stencil));
+ else
+ pipe->bind_depth_stencil_alpha_state(pipe, ctx->dsa_keep_depth_stencil);
+
+ pipe->bind_rasterizer_state(pipe, ctx->rs_state);
+ pipe->bind_fs_state(pipe, blitter_get_fs_col(ctx, num_cbufs));
+ pipe->bind_vs_state(pipe, ctx->vs_col);
+
+ blitter_set_clear_color(ctx, rgba);
+ blitter_set_rectangle(ctx, 0, 0, width, height, depth);
+ blitter_draw_quad(ctx);
+ blitter_restore_CSOs(ctx);
+}
+
+static boolean
+is_overlap(unsigned sx1, unsigned sx2, unsigned sy1, unsigned sy2,
+ unsigned dx1, unsigned dx2, unsigned dy1, unsigned dy2)
+{
+ if (sx1 >= dx2 || sx2 <= dx1 || sy1 >= dy2 || sy2 <= dy1) {
+ return FALSE;
+ } else {
+ return TRUE;
+ }
+}
+
+static void util_blitter_do_copy(struct blitter_context *blitter,
+ struct pipe_surface *dst,
+ unsigned dstx, unsigned dsty,
+ struct pipe_surface *src,
+ unsigned srcx, unsigned srcy,
+ unsigned width, unsigned height,
+ boolean is_depth)
+{
+ struct blitter_context_priv *ctx = (struct blitter_context_priv*)blitter;
+ struct pipe_context *pipe = ctx->pipe;
+ struct pipe_framebuffer_state fb_state;
+
+ assert(blitter->saved_fb_state.nr_cbufs != ~0);
+ assert(blitter->saved_num_textures != ~0);
+ assert(blitter->saved_num_sampler_states != ~0);
+ assert(src->texture->target < PIPE_MAX_TEXTURE_TYPES);
+
+ /* bind CSOs */
+ fb_state.width = dst->width;
+ fb_state.height = dst->height;
+
+ if (is_depth) {
+ pipe->bind_blend_state(pipe, ctx->blend_keep_color);
+ pipe->bind_depth_stencil_alpha_state(pipe,
+ ctx->dsa_write_depth_keep_stencil);
+ pipe->bind_fs_state(pipe,
+ blitter_get_fs_texfetch_depth(ctx, src->texture->target));
+
+ fb_state.nr_cbufs = 0;
+ fb_state.zsbuf = dst;
+ } else {
+ pipe->bind_blend_state(pipe, ctx->blend_write_color);
+ pipe->bind_depth_stencil_alpha_state(pipe, ctx->dsa_keep_depth_stencil);
+ pipe->bind_fs_state(pipe,
+ blitter_get_fs_texfetch_col(ctx, src->texture->target));
+
+ fb_state.nr_cbufs = 1;
+ fb_state.cbufs[0] = dst;
+ fb_state.zsbuf = 0;
+ }
+
+ pipe->bind_rasterizer_state(pipe, ctx->rs_state);
+ pipe->bind_vs_state(pipe, ctx->vs_tex);
+ pipe->bind_fragment_sampler_states(pipe, 1,
+ blitter_get_sampler_state(ctx, src->level));
+ pipe->set_fragment_sampler_textures(pipe, 1, &src->texture);
+ pipe->set_framebuffer_state(pipe, &fb_state);
+
+ /* set texture coordinates */
+ switch (src->texture->target) {
+ case PIPE_TEXTURE_1D:
+ case PIPE_TEXTURE_2D:
+ blitter_set_texcoords_2d(ctx, src, srcx, srcy,
+ srcx+width, srcy+height);
+ break;
+ case PIPE_TEXTURE_3D:
+ blitter_set_texcoords_3d(ctx, src, srcx, srcy,
+ srcx+width, srcy+height);
+ break;
+ case PIPE_TEXTURE_CUBE:
+ blitter_set_texcoords_cube(ctx, src, srcx, srcy,
+ srcx+width, srcy+height);
+ break;
+ default:
+ assert(0);
+ }
+
+ blitter_set_rectangle(ctx, dstx, dsty, dstx+width, dsty+height, 0);
+ blitter_draw_quad(ctx);
+
+}
+
+static void util_blitter_overlap_copy(struct blitter_context *blitter,
+ struct pipe_surface *dst,
+ unsigned dstx, unsigned dsty,
+ struct pipe_surface *src,
+ unsigned srcx, unsigned srcy,
+ unsigned width, unsigned height)
+{
+ struct blitter_context_priv *ctx = (struct blitter_context_priv*)blitter;
+ struct pipe_context *pipe = ctx->pipe;
+ struct pipe_screen *screen = pipe->screen;
+
+ struct pipe_texture texTemp;
+ struct pipe_texture *texture;
+ struct pipe_surface *tex_surf;
+
+ /* check whether the states are properly saved */
+ blitter_check_saved_CSOs(ctx);
+
+ memset(&texTemp, 0, sizeof(texTemp));
+ texTemp.target = PIPE_TEXTURE_2D;
+ texTemp.format = dst->texture->format; /* XXX verify supported by driver! */
+ texTemp.last_level = 0;
+ texTemp.width0 = width;
+ texTemp.height0 = height;
+ texTemp.depth0 = 1;
+
+ texture = screen->texture_create(screen, &texTemp);
+ if (!texture)
+ return;
+
+ tex_surf = screen->get_tex_surface(screen, texture, 0, 0, 0,
+ PIPE_BUFFER_USAGE_GPU_READ |
+ PIPE_BUFFER_USAGE_GPU_WRITE);
+
+ /* blit from the src to the temp */
+ util_blitter_do_copy(blitter, tex_surf, 0, 0,
+ src, srcx, srcy,
+ width, height,
+ FALSE);
+ util_blitter_do_copy(blitter, dst, dstx, dsty,
+ tex_surf, 0, 0,
+ width, height,
+ FALSE);
+ pipe_surface_reference(&tex_surf, NULL);
+ pipe_texture_reference(&texture, NULL);
+ blitter_restore_CSOs(ctx);
+}
+
+void util_blitter_copy(struct blitter_context *blitter,
+ struct pipe_surface *dst,
+ unsigned dstx, unsigned dsty,
+ struct pipe_surface *src,
+ unsigned srcx, unsigned srcy,
+ unsigned width, unsigned height,
+ boolean ignore_stencil)
+{
+ struct blitter_context_priv *ctx = (struct blitter_context_priv*)blitter;
+ struct pipe_context *pipe = ctx->pipe;
+ struct pipe_screen *screen = pipe->screen;
+ boolean is_stencil, is_depth;
+ unsigned dst_tex_usage;
+
+ /* give up if textures are not set */
+ assert(dst->texture && src->texture);
+ if (!dst->texture || !src->texture)
+ return;
+
+ if (dst->texture == src->texture) {
+ if (is_overlap(srcx, srcx + width, srcy, srcy + height,
+ dstx, dstx + width, dsty, dsty + height)) {
+ util_blitter_overlap_copy(blitter, dst, dstx, dsty, src, srcx, srcy,
+ width, height);
+ return;
+ }
+ }
+
+ is_depth = util_format_get_component_bits(src->format, UTIL_FORMAT_COLORSPACE_ZS, 0) != 0;
+ is_stencil = util_format_get_component_bits(src->format, UTIL_FORMAT_COLORSPACE_ZS, 1) != 0;
+ dst_tex_usage = is_depth || is_stencil ? PIPE_TEXTURE_USAGE_DEPTH_STENCIL :
+ PIPE_TEXTURE_USAGE_RENDER_TARGET;
+
+ /* check if we can sample from and render to the surfaces */
+ /* (assuming copying a stencil buffer is not possible) */
+ if ((!ignore_stencil && is_stencil) ||
+ !screen->is_format_supported(screen, dst->format, dst->texture->target,
+ dst_tex_usage, 0) ||
+ !screen->is_format_supported(screen, src->format, src->texture->target,
+ PIPE_TEXTURE_USAGE_SAMPLER, 0)) {
+ util_surface_copy(pipe, FALSE, dst, dstx, dsty, src, srcx, srcy,
+ width, height);
+ return;
+ }
+
+ /* check whether the states are properly saved */
+ blitter_check_saved_CSOs(ctx);
+ util_blitter_do_copy(blitter,
+ dst, dstx, dsty,
+ src, srcx, srcy,
+ width, height, is_depth);
+ blitter_restore_CSOs(ctx);
+}
+
+void util_blitter_fill(struct blitter_context *blitter,
+ struct pipe_surface *dst,
+ unsigned dstx, unsigned dsty,
+ unsigned width, unsigned height,
+ unsigned value)
+{
+ struct blitter_context_priv *ctx = (struct blitter_context_priv*)blitter;
+ struct pipe_context *pipe = ctx->pipe;
+ struct pipe_screen *screen = pipe->screen;
+ struct pipe_framebuffer_state fb_state;
+ float rgba[4];
+ ubyte ub_rgba[4] = {0};
+ union util_color color;
+ int i;
+
+ assert(dst->texture);
+ if (!dst->texture)
+ return;
+
+ /* check if we can render to the surface */
+ if (util_format_is_depth_or_stencil(dst->format) || /* unlikely, but you never know */
+ !screen->is_format_supported(screen, dst->format, dst->texture->target,
+ PIPE_TEXTURE_USAGE_RENDER_TARGET, 0)) {
+ util_surface_fill(pipe, dst, dstx, dsty, width, height, value);
+ return;
+ }
+
+ /* unpack the color */
+ color.ui = value;
+ util_unpack_color_ub(dst->format, &color,
+ ub_rgba, ub_rgba+1, ub_rgba+2, ub_rgba+3);
+ for (i = 0; i < 4; i++)
+ rgba[i] = ubyte_to_float(ub_rgba[i]);
+
+ /* check the saved state */
+ blitter_check_saved_CSOs(ctx);
+ assert(blitter->saved_fb_state.nr_cbufs != ~0);
+
+ /* bind CSOs */
+ pipe->bind_blend_state(pipe, ctx->blend_write_color);
+ pipe->bind_depth_stencil_alpha_state(pipe, ctx->dsa_keep_depth_stencil);
+ pipe->bind_rasterizer_state(pipe, ctx->rs_state);
+ pipe->bind_fs_state(pipe, blitter_get_fs_col(ctx, 1));
+ pipe->bind_vs_state(pipe, ctx->vs_col);
+
+ /* set a framebuffer state */
+ fb_state.width = dst->width;
+ fb_state.height = dst->height;
+ fb_state.nr_cbufs = 1;
+ fb_state.cbufs[0] = dst;
+ fb_state.zsbuf = 0;
+ pipe->set_framebuffer_state(pipe, &fb_state);
+
+ blitter_set_clear_color(ctx, rgba);
+ blitter_set_rectangle(ctx, 0, 0, width, height, 0);
+ blitter_draw_quad(ctx);
+ blitter_restore_CSOs(ctx);
+}
diff --git a/src/gallium/auxiliary/util/u_blitter.h b/src/gallium/auxiliary/util/u_blitter.h
new file mode 100644
index 00000000000..3da5a6ca525
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_blitter.h
@@ -0,0 +1,230 @@
+/**************************************************************************
+ *
+ * Copyright 2009 Marek Olšák <[email protected]>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+#ifndef U_BLITTER_H
+#define U_BLITTER_H
+
+#include "util/u_memory.h"
+
+#include "pipe/p_state.h"
+
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+struct pipe_context;
+
+struct blitter_context
+{
+ /* Private members, really. */
+ void *saved_blend_state; /**< blend state */
+ void *saved_dsa_state; /**< depth stencil alpha state */
+ void *saved_rs_state; /**< rasterizer state */
+ void *saved_fs, *saved_vs; /**< fragment shader, vertex shader */
+
+ struct pipe_framebuffer_state saved_fb_state; /**< framebuffer state */
+
+ int saved_num_sampler_states;
+ void *saved_sampler_states[32];
+
+ int saved_num_textures;
+ struct pipe_texture *saved_textures[32]; /* is 32 enough? */
+};
+
+/**
+ * Create a blitter context.
+ */
+struct blitter_context *util_blitter_create(struct pipe_context *pipe);
+
+/**
+ * Destroy a blitter context.
+ */
+void util_blitter_destroy(struct blitter_context *blitter);
+
+/*
+ * These CSOs must be saved before any of the following functions is called:
+ * - blend state
+ * - depth stencil alpha state
+ * - rasterizer state
+ * - vertex shader
+ * - fragment shader
+ */
+
+/**
+ * Clear a specified set of currently bound buffers to specified values.
+ */
+void util_blitter_clear(struct blitter_context *blitter,
+ unsigned width, unsigned height,
+ unsigned num_cbufs,
+ unsigned clear_buffers,
+ const float *rgba,
+ double depth, unsigned stencil);
+
+/**
+ * Copy a block of pixels from one surface to another.
+ *
+ * You can copy from any color format to any other color format provided
+ * the former can be sampled and the latter can be rendered to. Otherwise,
+ * a software fallback path is taken and both surfaces must be of the same
+ * format.
+ *
+ * The same holds for depth-stencil formats with the exception that stencil
+ * cannot be copied unless you set ignore_stencil to FALSE. In that case,
+ * a software fallback path is taken and both surfaces must be of the same
+ * format.
+ *
+ * Use pipe_screen->is_format_supported to know your options.
+ *
+ * These states must be saved in the blitter in addition to the state objects
+ * already required to be saved:
+ * - framebuffer state
+ * - fragment sampler states
+ * - fragment sampler textures
+ */
+void util_blitter_copy(struct blitter_context *blitter,
+ struct pipe_surface *dst,
+ unsigned dstx, unsigned dsty,
+ struct pipe_surface *src,
+ unsigned srcx, unsigned srcy,
+ unsigned width, unsigned height,
+ boolean ignore_stencil);
+
+/**
+ * Fill a region of a surface with a constant value.
+ *
+ * If the surface cannot be rendered to or it's a depth-stencil format,
+ * a software fallback path is taken.
+ *
+ * These states must be saved in the blitter in addition to the state objects
+ * already required to be saved:
+ * - framebuffer state
+ */
+void util_blitter_fill(struct blitter_context *blitter,
+ struct pipe_surface *dst,
+ unsigned dstx, unsigned dsty,
+ unsigned width, unsigned height,
+ unsigned value);
+
+/**
+ * Copy all pixels from one surface to another.
+ *
+ * The rules are the same as in util_blitter_copy with the addition that
+ * surfaces must have the same size.
+ */
+static INLINE
+void util_blitter_copy_surface(struct blitter_context *blitter,
+ struct pipe_surface *dst,
+ struct pipe_surface *src,
+ boolean ignore_stencil)
+{
+ assert(dst->width == src->width && dst->height == src->height);
+
+ util_blitter_copy(blitter, dst, 0, 0, src, 0, 0, src->width, src->height,
+ ignore_stencil);
+}
+
+
+/* The functions below should be used to save currently bound constant state
+ * objects inside a driver. The objects are automatically restored at the end
+ * of the util_blitter_{clear, fill, copy, copy_surface} functions and then
+ * forgotten.
+ *
+ * CSOs not listed here are not affected by util_blitter. */
+
+static INLINE
+void util_blitter_save_blend(struct blitter_context *blitter,
+ void *state)
+{
+ blitter->saved_blend_state = state;
+}
+
+static INLINE
+void util_blitter_save_depth_stencil_alpha(struct blitter_context *blitter,
+ void *state)
+{
+ blitter->saved_dsa_state = state;
+}
+
+static INLINE
+void util_blitter_save_rasterizer(struct blitter_context *blitter,
+ void *state)
+{
+ blitter->saved_rs_state = state;
+}
+
+static INLINE
+void util_blitter_save_fragment_shader(struct blitter_context *blitter,
+ void *fs)
+{
+ blitter->saved_fs = fs;
+}
+
+static INLINE
+void util_blitter_save_vertex_shader(struct blitter_context *blitter,
+ void *vs)
+{
+ blitter->saved_vs = vs;
+}
+
+static INLINE
+void util_blitter_save_framebuffer(struct blitter_context *blitter,
+ struct pipe_framebuffer_state *state)
+{
+ blitter->saved_fb_state = *state;
+}
+
+static INLINE
+void util_blitter_save_fragment_sampler_states(
+ struct blitter_context *blitter,
+ int num_sampler_states,
+ void **sampler_states)
+{
+ assert(num_sampler_states <= Elements(blitter->saved_sampler_states));
+
+ blitter->saved_num_sampler_states = num_sampler_states;
+ memcpy(blitter->saved_sampler_states, sampler_states,
+ num_sampler_states * sizeof(void *));
+}
+
+static INLINE
+void util_blitter_save_fragment_sampler_textures(
+ struct blitter_context *blitter,
+ int num_textures,
+ struct pipe_texture **textures)
+{
+ assert(num_textures <= Elements(blitter->saved_textures));
+
+ blitter->saved_num_textures = num_textures;
+ memcpy(blitter->saved_textures, textures,
+ num_textures * sizeof(struct pipe_texture *));
+}
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif
diff --git a/src/gallium/auxiliary/util/u_clear.h b/src/gallium/auxiliary/util/u_clear.h
index 6be5ca2297d..2c32db61756 100644
--- a/src/gallium/auxiliary/util/u_clear.h
+++ b/src/gallium/auxiliary/util/u_clear.h
@@ -46,13 +46,13 @@ util_clear(struct pipe_context *pipe,
{
if (buffers & PIPE_CLEAR_COLOR) {
struct pipe_surface *ps = framebuffer->cbufs[0];
- unsigned color = 0;
+ union util_color uc;
- util_pack_color(rgba, ps->format, &color);
+ util_pack_color(rgba, ps->format, &uc);
if (pipe->surface_fill) {
- pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, color);
+ pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, uc.ui);
} else {
- util_surface_fill(pipe, ps, 0, 0, ps->width, ps->height, color);
+ util_surface_fill(pipe, ps, 0, 0, ps->width, ps->height, uc.ui);
}
}
diff --git a/src/gallium/auxiliary/util/u_debug.c b/src/gallium/auxiliary/util/u_debug.c
index be5eb87e474..9b4e6ca2a73 100644
--- a/src/gallium/auxiliary/util/u_debug.c
+++ b/src/gallium/auxiliary/util/u_debug.c
@@ -64,11 +64,13 @@
#include "pipe/p_format.h"
#include "pipe/p_state.h"
#include "pipe/p_inlines.h"
+#include "util/u_format.h"
#include "util/u_memory.h"
#include "util/u_string.h"
#include "util/u_stream.h"
#include "util/u_math.h"
#include "util/u_tile.h"
+#include "util/u_prim.h"
#ifdef PIPE_SUBSYSTEM_WINDOWS_DISPLAY
@@ -602,6 +604,32 @@ const char *pf_name( enum pipe_format format )
}
+
+static const struct debug_named_value pipe_prim_names[] = {
+#ifdef DEBUG
+ DEBUG_NAMED_VALUE(PIPE_PRIM_POINTS),
+ DEBUG_NAMED_VALUE(PIPE_PRIM_LINES),
+ DEBUG_NAMED_VALUE(PIPE_PRIM_LINE_LOOP),
+ DEBUG_NAMED_VALUE(PIPE_PRIM_LINE_STRIP),
+ DEBUG_NAMED_VALUE(PIPE_PRIM_TRIANGLES),
+ DEBUG_NAMED_VALUE(PIPE_PRIM_TRIANGLE_STRIP),
+ DEBUG_NAMED_VALUE(PIPE_PRIM_TRIANGLE_FAN),
+ DEBUG_NAMED_VALUE(PIPE_PRIM_QUADS),
+ DEBUG_NAMED_VALUE(PIPE_PRIM_QUAD_STRIP),
+ DEBUG_NAMED_VALUE(PIPE_PRIM_POLYGON),
+#endif
+ DEBUG_NAMED_VALUE_END
+};
+
+
+const char *u_prim_name( unsigned prim )
+{
+ return debug_dump_enum(pipe_prim_names, prim);
+}
+
+
+
+
#ifdef DEBUG
void debug_dump_image(const char *prefix,
unsigned format, unsigned cpp,
@@ -671,10 +699,10 @@ void debug_dump_surface(const char *prefix,
goto error;
debug_dump_image(prefix,
- transfer->format,
- transfer->block.size,
- transfer->nblocksx,
- transfer->nblocksy,
+ texture->format,
+ util_format_get_blocksize(texture->format),
+ util_format_get_nblocksx(texture->format, transfer->width),
+ util_format_get_nblocksy(texture->format, transfer->height),
transfer->stride,
data);
diff --git a/src/gallium/auxiliary/util/u_debug_dump.c b/src/gallium/auxiliary/util/u_debug_dump.c
index 09866880aea..61624d05c0a 100644
--- a/src/gallium/auxiliary/util/u_debug_dump.c
+++ b/src/gallium/auxiliary/util/u_debug_dump.c
@@ -255,15 +255,13 @@ DEFINE_DEBUG_DUMP_CONTINUOUS(tex_mipfilter)
static const char *
debug_dump_tex_filter_names[] = {
"PIPE_TEX_FILTER_NEAREST",
- "PIPE_TEX_FILTER_LINEAR",
- "PIPE_TEX_FILTER_ANISO"
+ "PIPE_TEX_FILTER_LINEAR"
};
static const char *
debug_dump_tex_filter_short_names[] = {
"nearest",
- "linear",
- "aniso"
+ "linear"
};
DEFINE_DEBUG_DUMP_CONTINUOUS(tex_filter)
diff --git a/src/gallium/auxiliary/util/u_dl.c b/src/gallium/auxiliary/util/u_dl.c
new file mode 100644
index 00000000000..b42b429d4d7
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_dl.c
@@ -0,0 +1,79 @@
+/**************************************************************************
+ *
+ * Copyright 2009 VMware, Inc.
+ * Copyright 1999-2008 Brian Paul
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ **************************************************************************/
+
+
+#include "pipe/p_config.h"
+
+#if defined(PIPE_OS_UNIX)
+#include <dlfcn.h>
+#endif
+#if defined(PIPE_OS_WINDOWS)
+#include <windows.h>
+#endif
+
+#include "u_dl.h"
+
+
+struct util_dl_library *
+util_dl_open(const char *filename)
+{
+#if defined(PIPE_OS_UNIX)
+ return (struct util_dl_library *)dlopen(filename, RTLD_LAZY | RTLD_GLOBAL);
+#elif defined(PIPE_OS_WINDOWS)
+ return (struct util_dl_library *)LoadLibraryA(filename);
+#else
+ return NULL;
+#endif
+}
+
+
+util_dl_proc
+util_dl_get_proc_address(struct util_dl_library *library,
+ const char *procname)
+{
+#if defined(PIPE_OS_UNIX)
+ return (util_dl_proc)dlsym((void *)library, procname);
+#elif defined(PIPE_OS_WINDOWS)
+ return (util_dl_proc)GetProcAddress((HMODULE)library, procname);
+#else
+ return (util_dl_proc)NULL;
+#endif
+}
+
+
+void
+util_dl_close(struct util_dl_library *library)
+{
+#if defined(PIPE_OS_UNIX)
+ dlclose((void *)library);
+#elif defined(PIPE_OS_WINDOWS)
+ FreeLibrary((HMODULE)library);
+#else
+ (void)library;
+#endif
+}
diff --git a/src/gallium/auxiliary/util/u_dl.h b/src/gallium/auxiliary/util/u_dl.h
new file mode 100644
index 00000000000..85296c58af6
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_dl.h
@@ -0,0 +1,73 @@
+/**************************************************************************
+ *
+ * Copyright 2009 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ **************************************************************************/
+
+
+#ifndef U_DL_H_
+#define U_DL_H_
+
+
+#include "pipe/p_config.h"
+
+
+#if defined(PIPE_OS_WINDOWS)
+# define UTIL_DL_EXT ".dll"
+#elif defined(PIPE_OS_APPLE)
+# define UTIL_DL_EXT ".dylib"
+#else
+# define UTIL_DL_EXT ".so"
+#endif
+
+
+struct util_dl_library;
+
+
+typedef void (*util_dl_proc)(void);
+
+
+/**
+ * Open a library dynamically.
+ */
+struct util_dl_library *
+util_dl_open(const char *filename);
+
+
+/**
+ * Lookup a function in a library.
+ */
+util_dl_proc
+util_dl_get_proc_address(struct util_dl_library *library,
+ const char *procname);
+
+
+/**
+ * Close a library.
+ */
+void
+util_dl_close(struct util_dl_library *library);
+
+
+#endif /* U_DL_H_ */
diff --git a/src/gallium/auxiliary/util/u_format.c b/src/gallium/auxiliary/util/u_format.c
index 98ea13b60b5..e0724a1a8be 100644
--- a/src/gallium/auxiliary/util/u_format.c
+++ b/src/gallium/auxiliary/util/u_format.c
@@ -32,15 +32,14 @@
const struct util_format_description *
util_format_description(enum pipe_format format)
{
- const struct util_format_description *desc = util_format_description_table;
+ const struct util_format_description *desc;
- while(TRUE) {
- if(desc->format == format)
- return desc;
+ if (format >= PIPE_FORMAT_COUNT) {
+ return NULL;
+ }
- if(desc->format == PIPE_FORMAT_NONE)
- return NULL;
+ desc = &util_format_description_table[format];
+ assert(desc->format == format);
- ++desc;
- };
+ return desc;
}
diff --git a/src/gallium/auxiliary/util/u_format.csv b/src/gallium/auxiliary/util/u_format.csv
index f1bf94f17dd..9f16b42944e 100644
--- a/src/gallium/auxiliary/util/u_format.csv
+++ b/src/gallium/auxiliary/util/u_format.csv
@@ -11,6 +11,8 @@ PIPE_FORMAT_A8_UNORM , arith , 1, 1, un8 , , , , 000x,
PIPE_FORMAT_I8_UNORM , arith , 1, 1, un8 , , , , xxxx, rgb
PIPE_FORMAT_A8L8_UNORM , arith , 1, 1, un8 , un8 , , , xxxy, rgb
PIPE_FORMAT_L16_UNORM , arith , 1, 1, un16, , , , xxx1, rgb
+PIPE_FORMAT_YCBCR , yuv , 2, 1, x32 , , , , xyz1, yuv
+PIPE_FORMAT_YCBCR_REV , yuv , 2, 1, x32 , , , , xyz1, yuv
PIPE_FORMAT_Z16_UNORM , array , 1, 1, un16, , , , x___, zs
PIPE_FORMAT_Z32_UNORM , array , 1, 1, un32, , , , x___, zs
PIPE_FORMAT_Z32_FLOAT , array , 1, 1, f32 , , , , x___, zs
@@ -74,9 +76,9 @@ PIPE_FORMAT_R8G8_SNORM , array , 1, 1, sn8 , sn8 , , , xy01,
PIPE_FORMAT_R8G8B8_SNORM , array , 1, 1, sn8 , sn8 , sn8 , , xyz1, rgb
PIPE_FORMAT_R8G8B8A8_SNORM , array , 1, 1, sn8 , sn8 , sn8 , sn8 , xyzw, rgb
PIPE_FORMAT_R8G8B8X8_SNORM , array , 1, 1, sn8 , sn8 , sn8 , sn8 , xyz1, rgb
-PIPE_FORMAT_B6G5R5_SNORM , arith , 1, 1, sn5 , sn5 , sn6 , , zyx1, rgb
-PIPE_FORMAT_A8B8G8R8_SNORM , arith , 1, 1, sn8 , sn8 , sn8 , sn8 , zyxw, rgb
-PIPE_FORMAT_X8B8G8R8_SNORM , arith , 1, 1, sn8 , sn8 , sn8 , sn8 , zyx1, rgb
+PIPE_FORMAT_B6G5R5_SNORM , arith , 1, 1, sn5 , sn5 , sn6 , , xyz1, rgb
+PIPE_FORMAT_A8B8G8R8_SNORM , array , 1, 1, sn8 , sn8 , sn8 , sn8 , wzyx, rgb
+PIPE_FORMAT_X8B8G8R8_SNORM , array , 1, 1, sn8 , sn8 , sn8 , sn8 , wzy1, rgb
PIPE_FORMAT_R8_SSCALED , array , 1, 1, s8 , , , , x001, rgb
PIPE_FORMAT_R8G8_SSCALED , array , 1, 1, s8 , s8 , , , xy01, rgb
PIPE_FORMAT_R8G8B8_SSCALED , array , 1, 1, s8 , s8 , s8 , , xyz1, rgb
@@ -88,12 +90,20 @@ PIPE_FORMAT_R32G32B32_FIXED , array , 1, 1, h32 , h32 , h32 , , xyz1,
PIPE_FORMAT_R32G32B32A32_FIXED , array , 1, 1, h32 , h32 , h32 , h32 , xyzw, rgb
PIPE_FORMAT_L8_SRGB , arith , 1, 1, u8 , , , , xxx1, srgb
PIPE_FORMAT_A8L8_SRGB , arith , 1, 1, u8 , u8 , , , xxxy, srgb
-PIPE_FORMAT_R8G8B8_SRGB , arith , 1, 1, u8 , u8 , u8 , , xyz1, srgb
-PIPE_FORMAT_R8G8B8A8_SRGB , arith , 1, 1, u8 , u8 , u8 , u8 , xyzw, srgb
-PIPE_FORMAT_R8G8B8X8_SRGB , arith , 1, 1, u8 , u8 , u8 , u8 , xyz1, srgb
-PIPE_FORMAT_A8R8G8B8_SRGB , arith , 1, 1, u8 , u8 , u8 , u8 , wxyz, srgb
-PIPE_FORMAT_X8R8G8B8_SRGB , arith , 1, 1, u8 , u8 , u8 , u8 , 1xyz, srgb
-PIPE_FORMAT_B8G8R8A8_SRGB , arith , 1, 1, u8 , u8 , u8 , u8 , zyxw, srgb
-PIPE_FORMAT_B8G8R8X8_SRGB , arith , 1, 1, u8 , u8 , u8 , u8 , zyx1, srgb
-PIPE_FORMAT_X8UB8UG8SR8S_NORM , arith , 1, 1, sn8 , sn8 , un8 , x8 , 1zyx, rgb
+PIPE_FORMAT_R8G8B8_SRGB , array , 1, 1, u8 , u8 , u8 , , xyz1, srgb
+PIPE_FORMAT_R8G8B8A8_SRGB , array , 1, 1, u8 , u8 , u8 , u8 , xyzw, srgb
+PIPE_FORMAT_R8G8B8X8_SRGB , array , 1, 1, u8 , u8 , u8 , u8 , xyz1, srgb
+PIPE_FORMAT_A8R8G8B8_SRGB , array , 1, 1, u8 , u8 , u8 , u8 , yzwx, srgb
+PIPE_FORMAT_X8R8G8B8_SRGB , array , 1, 1, u8 , u8 , u8 , u8 , yzw1, srgb
+PIPE_FORMAT_B8G8R8A8_SRGB , array , 1, 1, u8 , u8 , u8 , u8 , zyxw, srgb
+PIPE_FORMAT_B8G8R8X8_SRGB , array , 1, 1, u8 , u8 , u8 , u8 , zyx1, srgb
+PIPE_FORMAT_X8UB8UG8SR8S_NORM , array , 1, 1, sn8 , sn8 , un8 , x8 , wzy1, rgb
PIPE_FORMAT_B6UG5SR5S_NORM , arith , 1, 1, sn5 , sn5 , un6 , , xyz1, rgb
+PIPE_FORMAT_DXT1_RGB , dxt , 4, 4, x64 , , , , xyz1, rgb
+PIPE_FORMAT_DXT1_RGBA , dxt , 4, 4, x64 , , , , xyzw, rgb
+PIPE_FORMAT_DXT3_RGBA , dxt , 4, 4, x128, , , , xyzw, rgb
+PIPE_FORMAT_DXT5_RGBA , dxt , 4, 4, x128, , , , xyzw, rgb
+PIPE_FORMAT_DXT1_SRGB , dxt , 4, 4, x64 , , , , xyz1, srgb
+PIPE_FORMAT_DXT1_SRGBA , dxt , 4, 4, x64 , , , , xyzw, srgb
+PIPE_FORMAT_DXT3_SRGBA , dxt , 4, 4, x128, , , , xyzw, srgb
+PIPE_FORMAT_DXT5_SRGBA , dxt , 4, 4, x128, , , , xyzw, srgb
diff --git a/src/gallium/auxiliary/util/u_format.h b/src/gallium/auxiliary/util/u_format.h
index bd27f346924..a558923b2ed 100644
--- a/src/gallium/auxiliary/util/u_format.h
+++ b/src/gallium/auxiliary/util/u_format.h
@@ -32,11 +32,51 @@
#include "pipe/p_format.h"
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/**
+ * Describe how to best pack/unpack pixels into/from the prescribed format.
+ *
+ * These are used for automatic code generation of pixel packing and unpacking
+ * routines (in compile time, e.g., u_format_access.py, or in runtime, like
+ * llvmpipe does).
+ *
+ * Thumb rule is: if you're not code generating pixel packing/unpacking then
+ * these are irrelevant for you.
+ *
+ * Note that this can be deduced from other values in util_format_description
+ * structure. This is by design, to make code generation of pixel
+ * packing/unpacking/sampling routines simple and efficient.
+ *
+ * XXX: This should be renamed to something like util_format_pack.
+ */
enum util_format_layout {
+ /**
+ * Single scalar component.
+ */
UTIL_FORMAT_LAYOUT_SCALAR = 0,
+
+ /**
+ * One or more components of mixed integer formats, arithmetically encoded
+ * in a word up to 32bits.
+ */
UTIL_FORMAT_LAYOUT_ARITH = 1,
+
+ /**
+ * One or more components, no mixed formats, each with equal power of two
+ * number of bytes.
+ */
UTIL_FORMAT_LAYOUT_ARRAY = 2,
+
+ /**
+ * XXX: Not used yet. These might go away and be replaced by a single entry,
+ * for formats where multiple pixels have to be
+ * read in order to determine a single pixel value (i.e., block.width > 1
+ * || block.height > 1)
+ */
UTIL_FORMAT_LAYOUT_YUV = 3,
UTIL_FORMAT_LAYOUT_DXT = 4
};
@@ -50,7 +90,7 @@ struct util_format_block
/** Block height in pixels */
unsigned height;
- /** Block size in bytes */
+ /** Block size in bits */
unsigned bits;
};
@@ -111,6 +151,250 @@ const struct util_format_description *
util_format_description(enum pipe_format format);
+/*
+ * Format query functions.
+ */
+
+static INLINE boolean
+util_format_is_compressed(enum pipe_format format)
+{
+ const struct util_format_description *desc = util_format_description(format);
+
+ assert(format);
+ if (!format) {
+ return FALSE;
+ }
+
+ return desc->layout == UTIL_FORMAT_LAYOUT_DXT ? TRUE : FALSE;
+}
+
+static INLINE boolean
+util_format_is_depth_or_stencil(enum pipe_format format)
+{
+ const struct util_format_description *desc = util_format_description(format);
+
+ assert(format);
+ if (!format) {
+ return FALSE;
+ }
+
+ return desc->colorspace == UTIL_FORMAT_COLORSPACE_ZS ? TRUE : FALSE;
+}
+
+static INLINE boolean
+util_format_is_depth_and_stencil(enum pipe_format format)
+{
+ const struct util_format_description *desc = util_format_description(format);
+
+ assert(format);
+ if (!format) {
+ return FALSE;
+ }
+
+ if (desc->colorspace != UTIL_FORMAT_COLORSPACE_ZS) {
+ return FALSE;
+ }
+
+ return (desc->swizzle[0] != UTIL_FORMAT_SWIZZLE_NONE &&
+ desc->swizzle[1] != UTIL_FORMAT_SWIZZLE_NONE) ? TRUE : FALSE;
+}
+
+
+/**
+ * Return total bits needed for the pixel format per block.
+ */
+static INLINE uint
+util_format_get_blocksizebits(enum pipe_format format)
+{
+ const struct util_format_description *desc = util_format_description(format);
+
+ assert(format);
+ if (!format) {
+ return 0;
+ }
+
+ return desc->block.bits;
+}
+
+/**
+ * Return bytes per block (not pixel) for the given format.
+ */
+static INLINE uint
+util_format_get_blocksize(enum pipe_format format)
+{
+ uint bits = util_format_get_blocksizebits(format);
+
+ assert(bits % 8 == 0);
+
+ return bits / 8;
+}
+
+static INLINE uint
+util_format_get_blockwidth(enum pipe_format format)
+{
+ const struct util_format_description *desc = util_format_description(format);
+
+ assert(format);
+ if (!format) {
+ return 1;
+ }
+
+ switch (desc->layout) {
+ case UTIL_FORMAT_LAYOUT_YUV:
+ return 2;
+ case UTIL_FORMAT_LAYOUT_DXT:
+ return 4;
+ default:
+ return 1;
+ }
+}
+
+static INLINE uint
+util_format_get_blockheight(enum pipe_format format)
+{
+ const struct util_format_description *desc = util_format_description(format);
+
+ assert(format);
+ if (!format) {
+ return 1;
+ }
+
+ switch (desc->layout) {
+ case UTIL_FORMAT_LAYOUT_DXT:
+ return 4;
+ default:
+ return 1;
+ }
+}
+
+static INLINE unsigned
+util_format_get_nblocksx(enum pipe_format format,
+ unsigned x)
+{
+ unsigned blockwidth = util_format_get_blockwidth(format);
+ return (x + blockwidth - 1) / blockwidth;
+}
+
+static INLINE unsigned
+util_format_get_nblocksy(enum pipe_format format,
+ unsigned y)
+{
+ unsigned blockheight = util_format_get_blockheight(format);
+ return (y + blockheight - 1) / blockheight;
+}
+
+static INLINE unsigned
+util_format_get_nblocks(enum pipe_format format,
+ unsigned width,
+ unsigned height)
+{
+ return util_format_get_nblocksx(format, width) * util_format_get_nblocksy(format, height);
+}
+
+static INLINE size_t
+util_format_get_stride(enum pipe_format format,
+ unsigned width)
+{
+ return util_format_get_nblocksx(format, width) * util_format_get_blocksize(format);
+}
+
+static INLINE size_t
+util_format_get_2d_size(enum pipe_format format,
+ size_t stride,
+ unsigned height)
+{
+ return util_format_get_nblocksy(format, height) * stride;
+}
+
+static INLINE uint
+util_format_get_component_bits(enum pipe_format format,
+ enum util_format_colorspace colorspace,
+ uint component)
+{
+ const struct util_format_description *desc = util_format_description(format);
+ enum util_format_colorspace desc_colorspace;
+
+ assert(format);
+ if (!format) {
+ return 0;
+ }
+
+ assert(component < 4);
+
+ /* Treat RGB and SRGB as equivalent. */
+ if (colorspace == UTIL_FORMAT_COLORSPACE_SRGB) {
+ colorspace = UTIL_FORMAT_COLORSPACE_RGB;
+ }
+ if (desc->colorspace == UTIL_FORMAT_COLORSPACE_SRGB) {
+ desc_colorspace = UTIL_FORMAT_COLORSPACE_RGB;
+ } else {
+ desc_colorspace = desc->colorspace;
+ }
+
+ if (desc_colorspace != colorspace) {
+ return 0;
+ }
+
+ switch (desc->swizzle[component]) {
+ case UTIL_FORMAT_SWIZZLE_X:
+ return desc->channel[0].size;
+ case UTIL_FORMAT_SWIZZLE_Y:
+ return desc->channel[1].size;
+ case UTIL_FORMAT_SWIZZLE_Z:
+ return desc->channel[2].size;
+ case UTIL_FORMAT_SWIZZLE_W:
+ return desc->channel[3].size;
+ default:
+ return 0;
+ }
+}
+
+static INLINE boolean
+util_format_has_alpha(enum pipe_format format)
+{
+ const struct util_format_description *desc = util_format_description(format);
+
+ assert(format);
+ if (!format) {
+ return FALSE;
+ }
+
+ switch (desc->layout) {
+ case UTIL_FORMAT_LAYOUT_SCALAR:
+ case UTIL_FORMAT_LAYOUT_ARITH:
+ case UTIL_FORMAT_LAYOUT_ARRAY:
+ /* FIXME: pf_get_component_bits( PIPE_FORMAT_A8L8_UNORM, PIPE_FORMAT_COMP_A ) should not return 0 right? */
+ if (format == PIPE_FORMAT_A8_UNORM ||
+ format == PIPE_FORMAT_A8L8_UNORM ||
+ format == PIPE_FORMAT_A8L8_SRGB) {
+ return TRUE;
+ }
+ return util_format_get_component_bits(format, UTIL_FORMAT_COLORSPACE_RGB, 3) != 0;
+ case UTIL_FORMAT_LAYOUT_YUV:
+ return FALSE;
+ case UTIL_FORMAT_LAYOUT_DXT:
+ switch (format) {
+ case PIPE_FORMAT_DXT1_RGBA:
+ case PIPE_FORMAT_DXT3_RGBA:
+ case PIPE_FORMAT_DXT5_RGBA:
+ case PIPE_FORMAT_DXT1_SRGBA:
+ case PIPE_FORMAT_DXT3_SRGBA:
+ case PIPE_FORMAT_DXT5_SRGBA:
+ return TRUE;
+ default:
+ return FALSE;
+ }
+ default:
+ assert(0);
+ return FALSE;
+ }
+}
+
+
+/*
+ * Format access functions.
+ */
+
void
util_format_read_4f(enum pipe_format format,
float *dst, unsigned dst_stride,
@@ -135,4 +419,8 @@ util_format_write_4ub(enum pipe_format format,
void *dst, unsigned dst_stride,
unsigned x, unsigned y, unsigned w, unsigned h);
+#ifdef __cplusplus
+} // extern "C" {
+#endif
+
#endif /* ! U_FORMAT_H */
diff --git a/src/gallium/auxiliary/util/u_format_access.py b/src/gallium/auxiliary/util/u_format_access.py
index eeb1a9657fd..0b05ddb9312 100644
--- a/src/gallium/auxiliary/util/u_format_access.py
+++ b/src/gallium/auxiliary/util/u_format_access.py
@@ -325,14 +325,14 @@ def generate_format_read(format, dst_type, dst_native_type, dst_suffix):
elif swizzle == SWIZZLE_0:
value = '0'
elif swizzle == SWIZZLE_1:
- value = '1'
+ value = get_one(dst_type)
else:
assert False
elif format.colorspace == 'zs':
if i < 3:
value = 'z'
else:
- value = '1'
+ value = get_one(dst_type)
else:
assert False
print ' *dst_pixel++ = %s; /* %s */' % (value, 'rgba'[i])
diff --git a/src/gallium/auxiliary/util/u_format_table.py b/src/gallium/auxiliary/util/u_format_table.py
index 8834568e8ee..571cab55dc8 100755
--- a/src/gallium/auxiliary/util/u_format_table.py
+++ b/src/gallium/auxiliary/util/u_format_table.py
@@ -44,11 +44,10 @@ def colorspace_map(colorspace):
colorspace_channels_map = {
- 'rgb': 'rgba',
- 'rgba': 'rgba',
- 'zs': 'zs',
- 'yuv': ['y1', 'y2', 'u', 'v'],
- 'dxt': []
+ 'rgb': ['r', 'g', 'b', 'a'],
+ 'srgb': ['sr', 'sg', 'sb', 'a'],
+ 'zs': ['z', 's'],
+ 'yuv': ['y', 'u', 'v'],
}
@@ -90,11 +89,20 @@ def write_format_table(formats):
print 'const struct util_format_description'
print 'util_format_description_table[] = '
print "{"
+ print " {"
+ print " PIPE_FORMAT_NONE,"
+ print " \"PIPE_FORMAT_NONE\","
+ print " {0, 0, 0},"
+ print " 0,"
+ print " {{0, 0, 0}, {0, 0, 0}, {0, 0, 0}, {0, 0, 0}},"
+ print " {0, 0, 0, 0},"
+ print " 0"
+ print " },"
for format in formats:
print " {"
print " %s," % (format.name,)
print " \"%s\"," % (format.name,)
- print " {%u, %u, %u}, /* block */" % (format.block_width, format.block_height, format.block_size())
+ print " {%u, %u, %u},\t/* block */" % (format.block_width, format.block_height, format.block_size())
print " %s," % (layout_map(format.layout),)
print " {"
for i in range(4):
@@ -103,7 +111,7 @@ def write_format_table(formats):
sep = ","
else:
sep = ""
- print " {%s, %s, %u}%s /* %s */" % (kind_map[type.kind], bool_map(type.norm), type.size, sep, "xyzw"[i])
+ print " {%s, %s, %u}%s\t/* %s */" % (kind_map[type.kind], bool_map(type.norm), type.size, sep, "xyzw"[i])
print " },"
print " {"
for i in range(4):
@@ -113,22 +121,13 @@ def write_format_table(formats):
else:
sep = ""
try:
- comment = layout_channels_map[format.layout][i]
- except:
+ comment = colorspace_channels_map[format.colorspace][i]
+ except (KeyError, IndexError):
comment = 'ignored'
- print " %s%s /* %s */" % (swizzle_map[swizzle], sep, comment)
+ print " %s%s\t/* %s */" % (swizzle_map[swizzle], sep, comment)
print " },"
print " %s," % (colorspace_map(format.colorspace),)
print " },"
- print " {"
- print " PIPE_FORMAT_NONE,"
- print " \"PIPE_FORMAT_NONE\","
- print " {0, 0, 0},"
- print " 0,"
- print " {{0, 0, 0}, {0, 0, 0}, {0, 0, 0}, {0, 0, 0}},"
- print " {0, 0, 0, 0},"
- print " 0"
- print " },"
print "};"
diff --git a/src/gallium/auxiliary/util/u_gen_mipmap.c b/src/gallium/auxiliary/util/u_gen_mipmap.c
index aa823aa218b..76023794dcd 100644
--- a/src/gallium/auxiliary/util/u_gen_mipmap.c
+++ b/src/gallium/auxiliary/util/u_gen_mipmap.c
@@ -41,10 +41,13 @@
#include "pipe/p_shader_tokens.h"
#include "pipe/p_state.h"
+#include "util/u_format.h"
#include "util/u_memory.h"
#include "util/u_draw_quad.h"
#include "util/u_gen_mipmap.h"
#include "util/u_simple_shaders.h"
+#include "util/u_math.h"
+#include "util/u_texture.h"
#include "cso_cache/cso_context.h"
@@ -60,7 +63,7 @@ struct gen_mipmap_state
struct pipe_sampler_state sampler;
void *vs;
- void *fs;
+ void *fs2d, *fsCube;
struct pipe_buffer *vbuf; /**< quad vertices */
unsigned vbuf_slot;
@@ -995,7 +998,7 @@ reduce_2d(enum pipe_format pformat,
{
enum dtype datatype;
uint comps;
- const int bpt = pf_get_size(pformat);
+ const int bpt = util_format_get_blocksize(pformat);
const ubyte *srcA, *srcB;
ubyte *dst;
int row;
@@ -1034,7 +1037,7 @@ reduce_3d(enum pipe_format pformat,
int dstWidth, int dstHeight, int dstDepth,
int dstRowStride, ubyte *dstPtr)
{
- const int bpt = pf_get_size(pformat);
+ const int bpt = util_format_get_blocksize(pformat);
const int border = 0;
int img, row;
int bytesPerSrcImage, bytesPerDstImage;
@@ -1125,12 +1128,12 @@ make_1d_mipmap(struct gen_mipmap_state *ctx,
srcTrans = screen->get_tex_transfer(screen, pt, face, srcLevel, zslice,
PIPE_TRANSFER_READ, 0, 0,
- pt->width[srcLevel],
- pt->height[srcLevel]);
+ u_minify(pt->width0, srcLevel),
+ u_minify(pt->height0, srcLevel));
dstTrans = screen->get_tex_transfer(screen, pt, face, dstLevel, zslice,
PIPE_TRANSFER_WRITE, 0, 0,
- pt->width[dstLevel],
- pt->height[dstLevel]);
+ u_minify(pt->width0, dstLevel),
+ u_minify(pt->height0, dstLevel));
srcMap = (ubyte *) screen->transfer_map(screen, srcTrans);
dstMap = (ubyte *) screen->transfer_map(screen, dstTrans);
@@ -1158,8 +1161,8 @@ make_2d_mipmap(struct gen_mipmap_state *ctx,
const uint zslice = 0;
uint dstLevel;
- assert(pt->block.width == 1);
- assert(pt->block.height == 1);
+ assert(util_format_get_blockwidth(pt->format) == 1);
+ assert(util_format_get_blockheight(pt->format) == 1);
for (dstLevel = baseLevel + 1; dstLevel <= lastLevel; dstLevel++) {
const uint srcLevel = dstLevel - 1;
@@ -1168,12 +1171,12 @@ make_2d_mipmap(struct gen_mipmap_state *ctx,
srcTrans = screen->get_tex_transfer(screen, pt, face, srcLevel, zslice,
PIPE_TRANSFER_READ, 0, 0,
- pt->width[srcLevel],
- pt->height[srcLevel]);
+ u_minify(pt->width0, srcLevel),
+ u_minify(pt->height0, srcLevel));
dstTrans = screen->get_tex_transfer(screen, pt, face, dstLevel, zslice,
PIPE_TRANSFER_WRITE, 0, 0,
- pt->width[dstLevel],
- pt->height[dstLevel]);
+ u_minify(pt->width0, dstLevel),
+ u_minify(pt->height0, dstLevel));
srcMap = (ubyte *) screen->transfer_map(screen, srcTrans);
dstMap = (ubyte *) screen->transfer_map(screen, dstTrans);
@@ -1203,8 +1206,8 @@ make_3d_mipmap(struct gen_mipmap_state *ctx,
struct pipe_screen *screen = pipe->screen;
uint dstLevel, zslice = 0;
- assert(pt->block.width == 1);
- assert(pt->block.height == 1);
+ assert(util_format_get_blockwidth(pt->format) == 1);
+ assert(util_format_get_blockheight(pt->format) == 1);
for (dstLevel = baseLevel + 1; dstLevel <= lastLevel; dstLevel++) {
const uint srcLevel = dstLevel - 1;
@@ -1213,12 +1216,12 @@ make_3d_mipmap(struct gen_mipmap_state *ctx,
srcTrans = screen->get_tex_transfer(screen, pt, face, srcLevel, zslice,
PIPE_TRANSFER_READ, 0, 0,
- pt->width[srcLevel],
- pt->height[srcLevel]);
+ u_minify(pt->width0, srcLevel),
+ u_minify(pt->height0, srcLevel));
dstTrans = screen->get_tex_transfer(screen, pt, face, dstLevel, zslice,
PIPE_TRANSFER_WRITE, 0, 0,
- pt->width[dstLevel],
- pt->height[dstLevel]);
+ u_minify(pt->width0, dstLevel),
+ u_minify(pt->height0, dstLevel));
srcMap = (ubyte *) screen->transfer_map(screen, srcTrans);
dstMap = (ubyte *) screen->transfer_map(screen, dstTrans);
@@ -1316,7 +1319,8 @@ util_create_gen_mipmap(struct pipe_context *pipe,
}
/* fragment shader */
- ctx->fs = util_make_fragment_tex_shader(pipe);
+ ctx->fs2d = util_make_fragment_tex_shader(pipe, TGSI_TEXTURE_2D);
+ ctx->fsCube = util_make_fragment_tex_shader(pipe, TGSI_TEXTURE_CUBE);
/* vertex data that doesn't change */
for (i = 0; i < 4; i++) {
@@ -1382,59 +1386,9 @@ set_vertex_data(struct gen_mipmap_state *ctx,
static const float st[4][2] = {
{0.0f, 0.0f}, {1.0f, 0.0f}, {1.0f, 1.0f}, {0.0f, 1.0f}
};
- float rx, ry, rz;
- uint i;
-
- /* loop over quad verts */
- for (i = 0; i < 4; i++) {
- /* Compute sc = +/-scale and tc = +/-scale.
- * Not +/-1 to avoid cube face selection ambiguity near the edges,
- * though that can still sometimes happen with this scale factor...
- */
- const float scale = 0.9999f;
- const float sc = (2.0f * st[i][0] - 1.0f) * scale;
- const float tc = (2.0f * st[i][1] - 1.0f) * scale;
-
- switch (face) {
- case PIPE_TEX_FACE_POS_X:
- rx = 1.0f;
- ry = -tc;
- rz = -sc;
- break;
- case PIPE_TEX_FACE_NEG_X:
- rx = -1.0f;
- ry = -tc;
- rz = sc;
- break;
- case PIPE_TEX_FACE_POS_Y:
- rx = sc;
- ry = 1.0f;
- rz = tc;
- break;
- case PIPE_TEX_FACE_NEG_Y:
- rx = sc;
- ry = -1.0f;
- rz = -tc;
- break;
- case PIPE_TEX_FACE_POS_Z:
- rx = sc;
- ry = -tc;
- rz = 1.0f;
- break;
- case PIPE_TEX_FACE_NEG_Z:
- rx = -sc;
- ry = -tc;
- rz = -1.0f;
- break;
- default:
- rx = ry = rz = 0.0f;
- assert(0);
- }
- ctx->vertices[i][1][0] = rx; /*s*/
- ctx->vertices[i][1][1] = ry; /*t*/
- ctx->vertices[i][1][2] = rz; /*r*/
- }
+ util_map_texcoords2d_onto_cubemap(face, &st[0][0], 2,
+ &ctx->vertices[0][1][0], 8);
}
else {
/* 1D/2D */
@@ -1474,7 +1428,8 @@ util_destroy_gen_mipmap(struct gen_mipmap_state *ctx)
struct pipe_context *pipe = ctx->pipe;
pipe->delete_vs_state(pipe, ctx->vs);
- pipe->delete_fs_state(pipe, ctx->fs);
+ pipe->delete_fs_state(pipe, ctx->fs2d);
+ pipe->delete_fs_state(pipe, ctx->fsCube);
pipe_buffer_reference(&ctx->vbuf, NULL);
@@ -1512,6 +1467,7 @@ util_gen_mipmap(struct gen_mipmap_state *ctx,
struct pipe_context *pipe = ctx->pipe;
struct pipe_screen *screen = pipe->screen;
struct pipe_framebuffer_state fb;
+ void *fs = (pt->target == PIPE_TEXTURE_CUBE) ? ctx->fsCube : ctx->fs2d;
uint dstLevel;
uint zslice = 0;
uint offset;
@@ -1549,7 +1505,7 @@ util_gen_mipmap(struct gen_mipmap_state *ctx,
cso_set_depth_stencil_alpha(ctx->cso, &ctx->depthstencil);
cso_set_rasterizer(ctx->cso, &ctx->rasterizer);
- cso_set_fragment_shader_handle(ctx->cso, ctx->fs);
+ cso_set_fragment_shader_handle(ctx->cso, fs);
cso_set_vertex_shader_handle(ctx->cso, ctx->vs);
/* init framebuffer state */
@@ -1575,8 +1531,8 @@ util_gen_mipmap(struct gen_mipmap_state *ctx,
* Setup framebuffer / dest surface
*/
fb.cbufs[0] = surf;
- fb.width = pt->width[dstLevel];
- fb.height = pt->height[dstLevel];
+ fb.width = u_minify(pt->width0, dstLevel);
+ fb.height = u_minify(pt->height0, dstLevel);
cso_set_framebuffer(ctx->cso, &fb);
/*
@@ -1597,8 +1553,8 @@ util_gen_mipmap(struct gen_mipmap_state *ctx,
offset = set_vertex_data(ctx,
pt->target,
face,
- (float) pt->width[dstLevel],
- (float) pt->height[dstLevel]);
+ (float) u_minify(pt->width0, dstLevel),
+ (float) u_minify(pt->height0, dstLevel));
util_draw_vertex_buffer(ctx->pipe,
ctx->vbuf,
diff --git a/src/gallium/auxiliary/util/u_linear.c b/src/gallium/auxiliary/util/u_linear.c
index a1dce3f5cf4..f1aef216771 100644
--- a/src/gallium/auxiliary/util/u_linear.c
+++ b/src/gallium/auxiliary/util/u_linear.c
@@ -82,7 +82,7 @@ void pipe_linear_from_tile(struct pipe_tile_info *t, const void *src_ptr,
void
pipe_linear_fill_info(struct pipe_tile_info *t,
- const struct pipe_format_block *block,
+ const struct u_linear_format_block *block,
unsigned tile_width, unsigned tile_height,
unsigned tiles_x, unsigned tiles_y)
{
diff --git a/src/gallium/auxiliary/util/u_linear.h b/src/gallium/auxiliary/util/u_linear.h
index b74308ffa3d..42c40b2aa75 100644
--- a/src/gallium/auxiliary/util/u_linear.h
+++ b/src/gallium/auxiliary/util/u_linear.h
@@ -35,6 +35,19 @@
#include "pipe/p_format.h"
+struct u_linear_format_block
+{
+ /** Block size in bytes */
+ unsigned size;
+
+ /** Block width in pixels */
+ unsigned width;
+
+ /** Block height in pixels */
+ unsigned height;
+};
+
+
struct pipe_tile_info
{
unsigned size;
@@ -49,10 +62,10 @@ struct pipe_tile_info
unsigned rows;
/* Describe the tile in pixels */
- struct pipe_format_block tile;
+ struct u_linear_format_block tile;
/* Describe each block within the tile */
- struct pipe_format_block block;
+ struct u_linear_format_block block;
};
void pipe_linear_to_tile(size_t src_stride, const void *src_ptr,
@@ -71,7 +84,7 @@ void pipe_linear_from_tile(struct pipe_tile_info *t, const void *src_ptr,
* @tiles_y number of tiles in y axis
*/
void pipe_linear_fill_info(struct pipe_tile_info *t,
- const struct pipe_format_block *block,
+ const struct u_linear_format_block *block,
unsigned tile_width, unsigned tile_height,
unsigned tiles_x, unsigned tiles_y);
diff --git a/src/gallium/auxiliary/util/u_math.h b/src/gallium/auxiliary/util/u_math.h
index 75b075f160d..b2969a210a7 100644
--- a/src/gallium/auxiliary/util/u_math.h
+++ b/src/gallium/auxiliary/util/u_math.h
@@ -491,6 +491,47 @@ util_next_power_of_two(unsigned x)
/**
+ * Return number of bits set in n.
+ */
+static INLINE unsigned
+util_bitcount(unsigned n)
+{
+#if defined(PIPE_CC_GCC)
+ return __builtin_popcount(n);
+#else
+ /* K&R classic bitcount.
+ *
+ * For each iteration, clear the LSB from the bitfield.
+ * Requires only one iteration per set bit, instead of
+ * one iteration per bit less than highest set bit.
+ */
+ unsigned bits = 0;
+ for (bits; n; bits++) {
+ n &= n - 1;
+ }
+ return bits;
+#endif
+}
+
+
+/**
+ * Reverse byte order of a 32 bit word.
+ */
+static INLINE uint32_t
+util_bswap32(uint32_t n)
+{
+#if defined(PIPE_CC_GCC) && (PIPE_CC_GCC_VERSION >= 403)
+ return __builtin_bswap32(n);
+#else
+ return (n >> 24) |
+ ((n >> 8) & 0x0000ff00) |
+ ((n << 8) & 0x00ff0000) |
+ (n << 24);
+#endif
+}
+
+
+/**
* Clamp X to [MIN, MAX].
* This is a macro to allow float, int, uint, etc. types.
*/
@@ -499,6 +540,9 @@ util_next_power_of_two(unsigned x)
#define MIN2( A, B ) ( (A)<(B) ? (A) : (B) )
#define MAX2( A, B ) ( (A)>(B) ? (A) : (B) )
+#define MIN3( A, B, C ) MIN2( MIN2( A, B ), C )
+#define MAX3( A, B, C ) MAX2( MAX2( A, B ), C )
+
static INLINE int
align(int value, int alignment)
@@ -507,9 +551,9 @@ align(int value, int alignment)
}
static INLINE unsigned
-minify(unsigned value)
+u_minify(unsigned value, unsigned levels)
{
- return MAX2(1, value >> 1);
+ return MAX2(1, value >> levels);
}
#ifndef COPY_4V
@@ -539,6 +583,18 @@ do { \
#endif
+static INLINE uint32_t util_unsigned_fixed(float value, unsigned frac_bits)
+{
+ return value < 0 ? 0 : (uint32_t)(value * (1<<frac_bits));
+}
+
+static INLINE int32_t util_signed_fixed(float value, unsigned frac_bits)
+{
+ return (int32_t)(value * (1<<frac_bits));
+}
+
+
+
#ifdef __cplusplus
}
#endif
diff --git a/src/gallium/auxiliary/util/u_pack_color.h b/src/gallium/auxiliary/util/u_pack_color.h
index 9dacc6d83db..43eb0153ee7 100644
--- a/src/gallium/auxiliary/util/u_pack_color.h
+++ b/src/gallium/auxiliary/util/u_pack_color.h
@@ -37,106 +37,110 @@
#include "pipe/p_compiler.h"
#include "pipe/p_format.h"
+#include "util/u_format.h"
#include "util/u_math.h"
+
+union util_color {
+ ubyte ub;
+ ushort us;
+ uint ui;
+ float f[4];
+};
+
/**
* Pack ubyte R,G,B,A into dest pixel.
*/
static INLINE void
util_pack_color_ub(ubyte r, ubyte g, ubyte b, ubyte a,
- enum pipe_format format, void *dest)
+ enum pipe_format format, union util_color *uc)
{
switch (format) {
case PIPE_FORMAT_R8G8B8A8_UNORM:
{
- uint *d = (uint *) dest;
- *d = (r << 24) | (g << 16) | (b << 8) | a;
+ uc->ui = (r << 24) | (g << 16) | (b << 8) | a;
}
return;
case PIPE_FORMAT_R8G8B8X8_UNORM:
{
- uint *d = (uint *) dest;
- *d = (r << 24) | (g << 16) | (b << 8) | 0xff;
+ uc->ui = (r << 24) | (g << 16) | (b << 8) | 0xff;
}
return;
case PIPE_FORMAT_A8R8G8B8_UNORM:
{
- uint *d = (uint *) dest;
- *d = (a << 24) | (r << 16) | (g << 8) | b;
+ uc->ui = (a << 24) | (r << 16) | (g << 8) | b;
}
return;
case PIPE_FORMAT_X8R8G8B8_UNORM:
{
- uint *d = (uint *) dest;
- *d = (0xff << 24) | (r << 16) | (g << 8) | b;
+ uc->ui = (0xff << 24) | (r << 16) | (g << 8) | b;
}
return;
case PIPE_FORMAT_B8G8R8A8_UNORM:
{
- uint *d = (uint *) dest;
- *d = (b << 24) | (g << 16) | (r << 8) | a;
+ uc->ui = (b << 24) | (g << 16) | (r << 8) | a;
}
return;
case PIPE_FORMAT_B8G8R8X8_UNORM:
{
- uint *d = (uint *) dest;
- *d = (b << 24) | (g << 16) | (r << 8) | 0xff;
+ uc->ui = (b << 24) | (g << 16) | (r << 8) | 0xff;
}
return;
case PIPE_FORMAT_R5G6B5_UNORM:
{
- ushort *d = (ushort *) dest;
- *d = ((r & 0xf8) << 8) | ((g & 0xfc) << 3) | (b >> 3);
+ uc->us = ((r & 0xf8) << 8) | ((g & 0xfc) << 3) | (b >> 3);
}
return;
case PIPE_FORMAT_A1R5G5B5_UNORM:
{
- ushort *d = (ushort *) dest;
- *d = ((a & 0x80) << 8) | ((r & 0xf8) << 7) | ((g & 0xf8) << 2) | (b >> 3);
+ uc->us = ((a & 0x80) << 8) | ((r & 0xf8) << 7) | ((g & 0xf8) << 2) | (b >> 3);
}
return;
case PIPE_FORMAT_A4R4G4B4_UNORM:
{
- ushort *d = (ushort *) dest;
- *d = ((a & 0xf0) << 8) | ((r & 0xf0) << 4) | ((g & 0xf0) << 0) | (b >> 4);
+ uc->us = ((a & 0xf0) << 8) | ((r & 0xf0) << 4) | ((g & 0xf0) << 0) | (b >> 4);
}
return;
case PIPE_FORMAT_A8_UNORM:
{
- ubyte *d = (ubyte *) dest;
- *d = a;
+ uc->ub = a;
}
return;
case PIPE_FORMAT_L8_UNORM:
case PIPE_FORMAT_I8_UNORM:
{
- ubyte *d = (ubyte *) dest;
- *d = r;
+ uc->ub = a;
}
return;
case PIPE_FORMAT_R32G32B32A32_FLOAT:
{
- float *d = (float *) dest;
- d[0] = (float)r / 255.0f;
- d[1] = (float)g / 255.0f;
- d[2] = (float)b / 255.0f;
- d[3] = (float)a / 255.0f;
+ uc->f[0] = (float)r / 255.0f;
+ uc->f[1] = (float)g / 255.0f;
+ uc->f[2] = (float)b / 255.0f;
+ uc->f[3] = (float)a / 255.0f;
}
return;
case PIPE_FORMAT_R32G32B32_FLOAT:
{
- float *d = (float *) dest;
- d[0] = (float)r / 255.0f;
- d[1] = (float)g / 255.0f;
- d[2] = (float)b / 255.0f;
+ uc->f[0] = (float)r / 255.0f;
+ uc->f[1] = (float)g / 255.0f;
+ uc->f[2] = (float)b / 255.0f;
}
return;
- /* XXX lots more cases to add */
+ /* Handle other cases with a generic function.
+ */
default:
- debug_print_format("gallium: unhandled format in util_pack_color_ub()", format);
- assert(0);
+ {
+ ubyte src[4];
+
+ src[0] = r;
+ src[1] = g;
+ src[2] = b;
+ src[3] = a;
+ util_format_write_4ub(format, src, 0, uc, 0, 0, 0, 1, 1);
+ }
}
}
@@ -145,13 +149,13 @@ util_pack_color_ub(ubyte r, ubyte g, ubyte b, ubyte a,
* Unpack RGBA from a packed pixel, returning values as ubytes in [0,255].
*/
static INLINE void
-util_unpack_color_ub(enum pipe_format format, const void *src,
+util_unpack_color_ub(enum pipe_format format, union util_color *uc,
ubyte *r, ubyte *g, ubyte *b, ubyte *a)
{
switch (format) {
case PIPE_FORMAT_R8G8B8A8_UNORM:
{
- uint p = ((const uint *) src)[0];
+ uint p = uc->ui;
*r = (ubyte) ((p >> 24) & 0xff);
*g = (ubyte) ((p >> 16) & 0xff);
*b = (ubyte) ((p >> 8) & 0xff);
@@ -160,7 +164,7 @@ util_unpack_color_ub(enum pipe_format format, const void *src,
return;
case PIPE_FORMAT_R8G8B8X8_UNORM:
{
- uint p = ((const uint *) src)[0];
+ uint p = uc->ui;
*r = (ubyte) ((p >> 24) & 0xff);
*g = (ubyte) ((p >> 16) & 0xff);
*b = (ubyte) ((p >> 8) & 0xff);
@@ -169,7 +173,7 @@ util_unpack_color_ub(enum pipe_format format, const void *src,
return;
case PIPE_FORMAT_A8R8G8B8_UNORM:
{
- uint p = ((const uint *) src)[0];
+ uint p = uc->ui;
*r = (ubyte) ((p >> 16) & 0xff);
*g = (ubyte) ((p >> 8) & 0xff);
*b = (ubyte) ((p >> 0) & 0xff);
@@ -178,7 +182,7 @@ util_unpack_color_ub(enum pipe_format format, const void *src,
return;
case PIPE_FORMAT_X8R8G8B8_UNORM:
{
- uint p = ((const uint *) src)[0];
+ uint p = uc->ui;
*r = (ubyte) ((p >> 16) & 0xff);
*g = (ubyte) ((p >> 8) & 0xff);
*b = (ubyte) ((p >> 0) & 0xff);
@@ -187,7 +191,7 @@ util_unpack_color_ub(enum pipe_format format, const void *src,
return;
case PIPE_FORMAT_B8G8R8A8_UNORM:
{
- uint p = ((const uint *) src)[0];
+ uint p = uc->ui;
*r = (ubyte) ((p >> 8) & 0xff);
*g = (ubyte) ((p >> 16) & 0xff);
*b = (ubyte) ((p >> 24) & 0xff);
@@ -196,7 +200,7 @@ util_unpack_color_ub(enum pipe_format format, const void *src,
return;
case PIPE_FORMAT_B8G8R8X8_UNORM:
{
- uint p = ((const uint *) src)[0];
+ uint p = uc->ui;
*r = (ubyte) ((p >> 8) & 0xff);
*g = (ubyte) ((p >> 16) & 0xff);
*b = (ubyte) ((p >> 24) & 0xff);
@@ -205,7 +209,7 @@ util_unpack_color_ub(enum pipe_format format, const void *src,
return;
case PIPE_FORMAT_R5G6B5_UNORM:
{
- ushort p = ((const ushort *) src)[0];
+ ushort p = uc->us;
*r = (ubyte) (((p >> 8) & 0xf8) | ((p >> 13) & 0x7));
*g = (ubyte) (((p >> 3) & 0xfc) | ((p >> 9) & 0x3));
*b = (ubyte) (((p << 3) & 0xf8) | ((p >> 2) & 0x7));
@@ -214,7 +218,7 @@ util_unpack_color_ub(enum pipe_format format, const void *src,
return;
case PIPE_FORMAT_A1R5G5B5_UNORM:
{
- ushort p = ((const ushort *) src)[0];
+ ushort p = uc->us;
*r = (ubyte) (((p >> 7) & 0xf8) | ((p >> 12) & 0x7));
*g = (ubyte) (((p >> 2) & 0xf8) | ((p >> 7) & 0x7));
*b = (ubyte) (((p << 3) & 0xf8) | ((p >> 2) & 0x7));
@@ -223,7 +227,7 @@ util_unpack_color_ub(enum pipe_format format, const void *src,
return;
case PIPE_FORMAT_A4R4G4B4_UNORM:
{
- ushort p = ((const ushort *) src)[0];
+ ushort p = uc->us;
*r = (ubyte) (((p >> 4) & 0xf0) | ((p >> 8) & 0xf));
*g = (ubyte) (((p >> 0) & 0xf0) | ((p >> 4) & 0xf));
*b = (ubyte) (((p << 4) & 0xf0) | ((p >> 0) & 0xf));
@@ -232,27 +236,27 @@ util_unpack_color_ub(enum pipe_format format, const void *src,
return;
case PIPE_FORMAT_A8_UNORM:
{
- ubyte p = ((const ubyte *) src)[0];
+ ubyte p = uc->ub;
*r = *g = *b = (ubyte) 0xff;
*a = p;
}
return;
case PIPE_FORMAT_L8_UNORM:
{
- ubyte p = ((const ubyte *) src)[0];
+ ubyte p = uc->ub;
*r = *g = *b = p;
*a = (ubyte) 0xff;
}
return;
case PIPE_FORMAT_I8_UNORM:
{
- ubyte p = ((const ubyte *) src)[0];
+ ubyte p = uc->ub;
*r = *g = *b = *a = p;
}
return;
case PIPE_FORMAT_R32G32B32A32_FLOAT:
{
- const float *p = (const float *) src;
+ const float *p = &uc->f[0];
*r = float_to_ubyte(p[0]);
*g = float_to_ubyte(p[1]);
*b = float_to_ubyte(p[2]);
@@ -261,7 +265,7 @@ util_unpack_color_ub(enum pipe_format format, const void *src,
return;
case PIPE_FORMAT_R32G32B32_FLOAT:
{
- const float *p = (const float *) src;
+ const float *p = &uc->f[0];
*r = float_to_ubyte(p[0]);
*g = float_to_ubyte(p[1]);
*b = float_to_ubyte(p[2]);
@@ -271,7 +275,7 @@ util_unpack_color_ub(enum pipe_format format, const void *src,
case PIPE_FORMAT_R32G32_FLOAT:
{
- const float *p = (const float *) src;
+ const float *p = &uc->f[0];
*r = float_to_ubyte(p[0]);
*g = float_to_ubyte(p[1]);
*b = *a = (ubyte) 0xff;
@@ -280,34 +284,40 @@ util_unpack_color_ub(enum pipe_format format, const void *src,
case PIPE_FORMAT_R32_FLOAT:
{
- const float *p = (const float *) src;
+ const float *p = &uc->f[0];
*r = float_to_ubyte(p[0]);
*g = *b = *a = (ubyte) 0xff;
}
return;
- /* XXX lots more cases to add */
+ /* Handle other cases with a generic function.
+ */
default:
- debug_print_format("gallium: unhandled format in util_unpack_color_ub()",
- format);
- assert(0);
+ {
+ ubyte dst[4];
+
+ util_format_read_4ub(format, dst, 0, uc, 0, 0, 0, 1, 1);
+ *r = dst[0];
+ *g = dst[1];
+ *b = dst[2];
+ *a = dst[3];
+ }
}
}
-
/**
* Note rgba outside [0,1] will be clamped for int pixel formats.
*/
static INLINE void
-util_pack_color(const float rgba[4], enum pipe_format format, void *dest)
+util_pack_color(const float rgba[4], enum pipe_format format, union util_color *uc)
{
ubyte r = 0;
ubyte g = 0;
ubyte b = 0;
ubyte a = 0;
- if (pf_size_x(format) <= 8) {
+ if (util_format_get_component_bits(format, UTIL_FORMAT_COLORSPACE_RGB, 0) <= 8) {
/* format uses 8-bit components or less */
r = float_to_ubyte(rgba[0]);
g = float_to_ubyte(rgba[1]);
@@ -318,92 +328,80 @@ util_pack_color(const float rgba[4], enum pipe_format format, void *dest)
switch (format) {
case PIPE_FORMAT_R8G8B8A8_UNORM:
{
- uint *d = (uint *) dest;
- *d = (r << 24) | (g << 16) | (b << 8) | a;
+ uc->ui = (r << 24) | (g << 16) | (b << 8) | a;
}
return;
case PIPE_FORMAT_R8G8B8X8_UNORM:
{
- uint *d = (uint *) dest;
- *d = (r << 24) | (g << 16) | (b << 8) | 0xff;
+ uc->ui = (r << 24) | (g << 16) | (b << 8) | 0xff;
}
return;
case PIPE_FORMAT_A8R8G8B8_UNORM:
{
- uint *d = (uint *) dest;
- *d = (a << 24) | (r << 16) | (g << 8) | b;
+ uc->ui = (a << 24) | (r << 16) | (g << 8) | b;
}
return;
case PIPE_FORMAT_X8R8G8B8_UNORM:
{
- uint *d = (uint *) dest;
- *d = (0xff << 24) | (r << 16) | (g << 8) | b;
+ uc->ui = (0xff << 24) | (r << 16) | (g << 8) | b;
}
return;
case PIPE_FORMAT_B8G8R8A8_UNORM:
{
- uint *d = (uint *) dest;
- *d = (b << 24) | (g << 16) | (r << 8) | a;
+ uc->ui = (b << 24) | (g << 16) | (r << 8) | a;
}
return;
case PIPE_FORMAT_B8G8R8X8_UNORM:
{
- uint *d = (uint *) dest;
- *d = (b << 24) | (g << 16) | (r << 8) | 0xff;
+ uc->ui = (b << 24) | (g << 16) | (r << 8) | 0xff;
}
return;
case PIPE_FORMAT_R5G6B5_UNORM:
{
- ushort *d = (ushort *) dest;
- *d = ((r & 0xf8) << 8) | ((g & 0xfc) << 3) | (b >> 3);
+ uc->us = ((r & 0xf8) << 8) | ((g & 0xfc) << 3) | (b >> 3);
}
return;
case PIPE_FORMAT_A1R5G5B5_UNORM:
{
- ushort *d = (ushort *) dest;
- *d = ((a & 0x80) << 8) | ((r & 0xf8) << 7) | ((g & 0xf8) << 2) | (b >> 3);
+ uc->us = ((a & 0x80) << 8) | ((r & 0xf8) << 7) | ((g & 0xf8) << 2) | (b >> 3);
}
return;
case PIPE_FORMAT_A4R4G4B4_UNORM:
{
- ushort *d = (ushort *) dest;
- *d = ((a & 0xf0) << 8) | ((r & 0xf0) << 4) | ((g & 0xf0) << 0) | (b >> 4);
+ uc->ub = ((a & 0xf0) << 8) | ((r & 0xf0) << 4) | ((g & 0xf0) << 0) | (b >> 4);
}
return;
case PIPE_FORMAT_A8_UNORM:
{
- ubyte *d = (ubyte *) dest;
- *d = a;
+ uc->ub = a;
}
return;
case PIPE_FORMAT_L8_UNORM:
case PIPE_FORMAT_I8_UNORM:
{
- ubyte *d = (ubyte *) dest;
- *d = r;
+ uc->ub = r;
}
return;
case PIPE_FORMAT_R32G32B32A32_FLOAT:
{
- float *d = (float *) dest;
- d[0] = rgba[0];
- d[1] = rgba[1];
- d[2] = rgba[2];
- d[3] = rgba[3];
+ uc->f[0] = rgba[0];
+ uc->f[1] = rgba[1];
+ uc->f[2] = rgba[2];
+ uc->f[3] = rgba[3];
}
return;
case PIPE_FORMAT_R32G32B32_FLOAT:
{
- float *d = (float *) dest;
- d[0] = rgba[0];
- d[1] = rgba[1];
- d[2] = rgba[2];
+ uc->f[0] = rgba[0];
+ uc->f[1] = rgba[1];
+ uc->f[2] = rgba[2];
}
return;
- /* XXX lots more cases to add */
+
+ /* Handle other cases with a generic function.
+ */
default:
- debug_print_format("gallium: unhandled format in util_pack_color()", format);
- assert(0);
+ util_format_write_4f(format, rgba, 0, uc, 0, 0, 0, 1, 1);
}
}
diff --git a/src/gallium/auxiliary/util/u_prim.h b/src/gallium/auxiliary/util/u_prim.h
index a9b533eea70..10a874f3416 100644
--- a/src/gallium/auxiliary/util/u_prim.h
+++ b/src/gallium/auxiliary/util/u_prim.h
@@ -135,4 +135,39 @@ static INLINE unsigned u_reduced_prim( unsigned pipe_prim )
}
}
+static INLINE unsigned
+u_vertices_per_prim(int primitive)
+{
+ switch(primitive) {
+ case PIPE_PRIM_POINTS:
+ return 1;
+ case PIPE_PRIM_LINES:
+ case PIPE_PRIM_LINE_LOOP:
+ case PIPE_PRIM_LINE_STRIP:
+ return 2;
+ case PIPE_PRIM_TRIANGLES:
+ case PIPE_PRIM_TRIANGLE_STRIP:
+ case PIPE_PRIM_TRIANGLE_FAN:
+ return 3;
+ case PIPE_PRIM_LINES_ADJACENCY:
+ case PIPE_PRIM_LINE_STRIP_ADJACENCY:
+ return 4;
+ case PIPE_PRIM_TRIANGLES_ADJACENCY:
+ case PIPE_PRIM_TRIANGLE_STRIP_ADJACENCY:
+ return 6;
+
+ /* following primitives should never be used
+ * with geometry shaders abd their size is
+ * undefined */
+ case PIPE_PRIM_POLYGON:
+ case PIPE_PRIM_QUADS:
+ case PIPE_PRIM_QUAD_STRIP:
+ default:
+ debug_printf("Unrecognized geometry shader primitive");
+ return 3;
+ }
+}
+
+const char *u_prim_name( unsigned pipe_prim );
+
#endif
diff --git a/src/gallium/auxiliary/util/u_rect.c b/src/gallium/auxiliary/util/u_rect.c
index 9866b6fc8a0..8479161c744 100644
--- a/src/gallium/auxiliary/util/u_rect.c
+++ b/src/gallium/auxiliary/util/u_rect.c
@@ -34,17 +34,18 @@
#include "pipe/p_format.h"
#include "pipe/p_context.h"
#include "pipe/p_screen.h"
+#include "util/u_format.h"
#include "util/u_rect.h"
/**
* Copy 2D rect from one place to another.
* Position and sizes are in pixels.
- * src_pitch may be negative to do vertical flip of pixels from source.
+ * src_stride may be negative to do vertical flip of pixels from source.
*/
void
util_copy_rect(ubyte * dst,
- const struct pipe_format_block *block,
+ enum pipe_format format,
unsigned dst_stride,
unsigned dst_x,
unsigned dst_y,
@@ -53,31 +54,30 @@ util_copy_rect(ubyte * dst,
const ubyte * src,
int src_stride,
unsigned src_x,
- int src_y)
+ unsigned src_y)
{
unsigned i;
int src_stride_pos = src_stride < 0 ? -src_stride : src_stride;
-
- assert(block->size > 0);
- assert(block->width > 0);
- assert(block->height > 0);
- assert(src_x >= 0);
- assert(src_y >= 0);
- assert(dst_x >= 0);
- assert(dst_y >= 0);
-
- dst_x /= block->width;
- dst_y /= block->height;
- width = (width + block->width - 1)/block->width;
- height = (height + block->height - 1)/block->height;
- src_x /= block->width;
- src_y /= block->height;
+ int blocksize = util_format_get_blocksize(format);
+ int blockwidth = util_format_get_blockwidth(format);
+ int blockheight = util_format_get_blockheight(format);
+
+ assert(blocksize > 0);
+ assert(blockwidth > 0);
+ assert(blockheight > 0);
+
+ dst_x /= blockwidth;
+ dst_y /= blockheight;
+ width = (width + blockwidth - 1)/blockwidth;
+ height = (height + blockheight - 1)/blockheight;
+ src_x /= blockwidth;
+ src_y /= blockheight;
- dst += dst_x * block->size;
- src += src_x * block->size;
+ dst += dst_x * blocksize;
+ src += src_x * blocksize;
dst += dst_y * dst_stride;
src += src_y * src_stride_pos;
- width *= block->size;
+ width *= blocksize;
if (width == dst_stride && width == src_stride)
memcpy(dst, src, height * width);
@@ -92,7 +92,7 @@ util_copy_rect(ubyte * dst,
void
util_fill_rect(ubyte * dst,
- const struct pipe_format_block *block,
+ enum pipe_format format,
unsigned dst_stride,
unsigned dst_x,
unsigned dst_y,
@@ -102,23 +102,24 @@ util_fill_rect(ubyte * dst,
{
unsigned i, j;
unsigned width_size;
-
- assert(block->size > 0);
- assert(block->width > 0);
- assert(block->height > 0);
- assert(dst_x >= 0);
- assert(dst_y >= 0);
-
- dst_x /= block->width;
- dst_y /= block->height;
- width = (width + block->width - 1)/block->width;
- height = (height + block->height - 1)/block->height;
+ int blocksize = util_format_get_blocksize(format);
+ int blockwidth = util_format_get_blockwidth(format);
+ int blockheight = util_format_get_blockheight(format);
+
+ assert(blocksize > 0);
+ assert(blockwidth > 0);
+ assert(blockheight > 0);
+
+ dst_x /= blockwidth;
+ dst_y /= blockheight;
+ width = (width + blockwidth - 1)/blockwidth;
+ height = (height + blockheight - 1)/blockheight;
- dst += dst_x * block->size;
+ dst += dst_x * blocksize;
dst += dst_y * dst_stride;
- width_size = width * block->size;
+ width_size = width * blocksize;
- switch (block->size) {
+ switch (blocksize) {
case 1:
if(dst_stride == width_size)
memset(dst, (ubyte) value, height * width_size);
@@ -172,10 +173,15 @@ util_surface_copy(struct pipe_context *pipe,
struct pipe_transfer *src_trans, *dst_trans;
void *dst_map;
const void *src_map;
+ enum pipe_format src_format, dst_format;
assert(src->texture && dst->texture);
if (!src->texture || !dst->texture)
return;
+
+ src_format = src->texture->format;
+ dst_format = dst->texture->format;
+
src_trans = screen->get_tex_transfer(screen,
src->texture,
src->face,
@@ -192,9 +198,9 @@ util_surface_copy(struct pipe_context *pipe,
PIPE_TRANSFER_WRITE,
dst_x, dst_y, w, h);
- assert(dst_trans->block.size == src_trans->block.size);
- assert(dst_trans->block.width == src_trans->block.width);
- assert(dst_trans->block.height == src_trans->block.height);
+ assert(util_format_get_blocksize(dst_format) == util_format_get_blocksize(src_format));
+ assert(util_format_get_blockwidth(dst_format) == util_format_get_blockwidth(src_format));
+ assert(util_format_get_blockheight(dst_format) == util_format_get_blockheight(src_format));
src_map = pipe->screen->transfer_map(screen, src_trans);
dst_map = pipe->screen->transfer_map(screen, dst_trans);
@@ -205,7 +211,7 @@ util_surface_copy(struct pipe_context *pipe,
if (src_map && dst_map) {
/* If do_flip, invert src_y position and pass negative src stride */
util_copy_rect(dst_map,
- &dst_trans->block,
+ dst_format,
dst_trans->stride,
0, 0,
w, h,
@@ -259,11 +265,11 @@ util_surface_fill(struct pipe_context *pipe,
if (dst_map) {
assert(dst_trans->stride > 0);
- switch (dst_trans->block.size) {
+ switch (util_format_get_blocksize(dst_trans->texture->format)) {
case 1:
case 2:
case 4:
- util_fill_rect(dst_map, &dst_trans->block, dst_trans->stride,
+ util_fill_rect(dst_map, dst_trans->texture->format, dst_trans->stride,
0, 0, width, height, value);
break;
case 8:
diff --git a/src/gallium/auxiliary/util/u_rect.h b/src/gallium/auxiliary/util/u_rect.h
index daa50834d36..b44d821904b 100644
--- a/src/gallium/auxiliary/util/u_rect.h
+++ b/src/gallium/auxiliary/util/u_rect.h
@@ -42,13 +42,13 @@ struct pipe_surface;
extern void
-util_copy_rect(ubyte * dst, const struct pipe_format_block *block,
+util_copy_rect(ubyte * dst, enum pipe_format format,
unsigned dst_stride, unsigned dst_x, unsigned dst_y,
unsigned width, unsigned height, const ubyte * src,
- int src_stride, unsigned src_x, int src_y);
+ int src_stride, unsigned src_x, unsigned src_y);
extern void
-util_fill_rect(ubyte * dst, const struct pipe_format_block *block,
+util_fill_rect(ubyte * dst, enum pipe_format format,
unsigned dst_stride, unsigned dst_x, unsigned dst_y,
unsigned width, unsigned height, uint32_t value);
diff --git a/src/gallium/auxiliary/util/u_simple_shaders.c b/src/gallium/auxiliary/util/u_simple_shaders.c
index 1c8b157d91f..b751e29ab62 100644
--- a/src/gallium/auxiliary/util/u_simple_shaders.c
+++ b/src/gallium/auxiliary/util/u_simple_shaders.c
@@ -2,6 +2,7 @@
*
* Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
* All Rights Reserved.
+ * Copyright 2009 Marek Olšák <[email protected]>
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the
@@ -30,6 +31,7 @@
* Simple vertex/fragment shader generators.
*
* @author Brian Paul
+ Marek Olšák
*/
@@ -42,13 +44,15 @@
/**
* Make simple vertex pass-through shader.
+ * \param num_attribs number of attributes to pass through
+ * \param semantic_names array of semantic names for each attribute
+ * \param semantic_indexes array of semantic indexes for each attribute
*/
void *
util_make_vertex_passthrough_shader(struct pipe_context *pipe,
uint num_attribs,
const uint *semantic_names,
const uint *semantic_indexes)
-
{
struct ureg_program *ureg;
uint i;
@@ -76,8 +80,6 @@ util_make_vertex_passthrough_shader(struct pipe_context *pipe,
}
-
-
/**
* Make simple fragment texture shader:
* IMM {0,0,0,1} // (if writemask != 0xf)
@@ -87,6 +89,7 @@ util_make_vertex_passthrough_shader(struct pipe_context *pipe,
*/
void *
util_make_fragment_tex_shader_writemask(struct pipe_context *pipe,
+ unsigned tex_target,
unsigned writemask )
{
struct ureg_program *ureg;
@@ -116,20 +119,71 @@ util_make_fragment_tex_shader_writemask(struct pipe_context *pipe,
ureg_TEX( ureg,
ureg_writemask(out, writemask),
- TGSI_TEXTURE_2D, tex, sampler );
+ tex_target, tex, sampler );
ureg_END( ureg );
return ureg_create_shader_and_destroy( ureg, pipe );
}
+
+/**
+ * Make a simple fragment shader that sets the output color to a color
+ * taken from a texture.
+ * \param tex_target one of PIPE_TEXTURE_x
+ */
void *
-util_make_fragment_tex_shader(struct pipe_context *pipe )
+util_make_fragment_tex_shader(struct pipe_context *pipe, unsigned tex_target )
{
return util_make_fragment_tex_shader_writemask( pipe,
+ tex_target,
TGSI_WRITEMASK_XYZW );
}
+/**
+ * Make a simple fragment texture shader which reads an X component from
+ * a texture and writes it as depth.
+ */
+void *
+util_make_fragment_tex_shader_writedepth(struct pipe_context *pipe,
+ unsigned tex_target)
+{
+ struct ureg_program *ureg;
+ struct ureg_src sampler;
+ struct ureg_src tex;
+ struct ureg_dst out, depth;
+ struct ureg_src imm;
+
+ ureg = ureg_create( TGSI_PROCESSOR_FRAGMENT );
+ if (ureg == NULL)
+ return NULL;
+
+ sampler = ureg_DECL_sampler( ureg, 0 );
+
+ tex = ureg_DECL_fs_input( ureg,
+ TGSI_SEMANTIC_GENERIC, 0,
+ TGSI_INTERPOLATE_PERSPECTIVE );
+
+ out = ureg_DECL_output( ureg,
+ TGSI_SEMANTIC_COLOR,
+ 0 );
+
+ depth = ureg_DECL_output( ureg,
+ TGSI_SEMANTIC_POSITION,
+ 0 );
+
+ imm = ureg_imm4f( ureg, 0, 0, 0, 1 );
+
+ ureg_MOV( ureg, out, imm );
+
+ ureg_TEX( ureg,
+ ureg_writemask(depth, TGSI_WRITEMASK_Z),
+ tex_target, tex, sampler );
+ ureg_END( ureg );
+
+ return ureg_create_shader_and_destroy( ureg, pipe );
+}
+
/**
* Make simple fragment color pass-through shader.
@@ -137,9 +191,22 @@ util_make_fragment_tex_shader(struct pipe_context *pipe )
void *
util_make_fragment_passthrough_shader(struct pipe_context *pipe)
{
+ return util_make_fragment_clonecolor_shader(pipe, 1);
+}
+
+
+/**
+ * Make a fragment shader that copies the input color to N output colors.
+ */
+void *
+util_make_fragment_clonecolor_shader(struct pipe_context *pipe, int num_cbufs)
+{
struct ureg_program *ureg;
struct ureg_src src;
- struct ureg_dst dst;
+ struct ureg_dst dst[PIPE_MAX_COLOR_BUFS];
+ int i;
+
+ assert(num_cbufs <= PIPE_MAX_COLOR_BUFS);
ureg = ureg_create( TGSI_PROCESSOR_FRAGMENT );
if (ureg == NULL)
@@ -148,12 +215,13 @@ util_make_fragment_passthrough_shader(struct pipe_context *pipe)
src = ureg_DECL_fs_input( ureg, TGSI_SEMANTIC_COLOR, 0,
TGSI_INTERPOLATE_PERSPECTIVE );
- dst = ureg_DECL_output( ureg, TGSI_SEMANTIC_COLOR, 0 );
+ for (i = 0; i < num_cbufs; i++)
+ dst[i] = ureg_DECL_output( ureg, TGSI_SEMANTIC_COLOR, i );
+
+ for (i = 0; i < num_cbufs; i++)
+ ureg_MOV( ureg, dst[i], src );
- ureg_MOV( ureg, dst, src );
ureg_END( ureg );
return ureg_create_shader_and_destroy( ureg, pipe );
}
-
-
diff --git a/src/gallium/auxiliary/util/u_simple_shaders.h b/src/gallium/auxiliary/util/u_simple_shaders.h
index d2e80d6eb4d..6e760942e25 100644
--- a/src/gallium/auxiliary/util/u_simple_shaders.h
+++ b/src/gallium/auxiliary/util/u_simple_shaders.h
@@ -51,16 +51,25 @@ util_make_vertex_passthrough_shader(struct pipe_context *pipe,
extern void *
util_make_fragment_tex_shader_writemask(struct pipe_context *pipe,
- unsigned writemask );
+ unsigned tex_target,
+ unsigned writemask);
extern void *
-util_make_fragment_tex_shader(struct pipe_context *pipe);
+util_make_fragment_tex_shader(struct pipe_context *pipe, unsigned tex_target);
+
+
+extern void *
+util_make_fragment_tex_shader_writedepth(struct pipe_context *pipe,
+ unsigned tex_target);
extern void *
util_make_fragment_passthrough_shader(struct pipe_context *pipe);
+extern void *
+util_make_fragment_clonecolor_shader(struct pipe_context *pipe, int num_cbufs);
+
#ifdef __cplusplus
}
#endif
diff --git a/src/gallium/auxiliary/util/u_surface.c b/src/gallium/auxiliary/util/u_surface.c
index 85e443204e3..35c49782043 100644
--- a/src/gallium/auxiliary/util/u_surface.c
+++ b/src/gallium/auxiliary/util/u_surface.c
@@ -36,6 +36,7 @@
#include "pipe/p_state.h"
#include "pipe/p_defines.h"
+#include "util/u_format.h"
#include "util/u_surface.h"
@@ -79,10 +80,9 @@ util_create_rgba_surface(struct pipe_screen *screen,
templ.target = target;
templ.format = format;
templ.last_level = 0;
- templ.width[0] = width;
- templ.height[0] = height;
- templ.depth[0] = 1;
- pf_get_block(format, &templ.block);
+ templ.width0 = width;
+ templ.height0 = height;
+ templ.depth0 = 1;
templ.tex_usage = usage;
*textureOut = screen->texture_create(screen, &templ);
diff --git a/src/gallium/auxiliary/util/u_texture.c b/src/gallium/auxiliary/util/u_texture.c
new file mode 100644
index 00000000000..cd477ab640f
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_texture.c
@@ -0,0 +1,102 @@
+/**************************************************************************
+ *
+ * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * Copyright 2008 VMware, Inc. All rights reserved.
+ * Copyright 2009 Marek Olšák <[email protected]>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+/**
+ * @file
+ * Texture mapping utility functions.
+ *
+ * @author Brian Paul
+ * Marek Olšák
+ */
+
+#include "pipe/p_defines.h"
+
+#include "util/u_texture.h"
+
+void util_map_texcoords2d_onto_cubemap(unsigned face,
+ const float *in_st, unsigned in_stride,
+ float *out_str, unsigned out_stride)
+{
+ int i;
+ float rx, ry, rz;
+
+ /* loop over quad verts */
+ for (i = 0; i < 4; i++) {
+ /* Compute sc = +/-scale and tc = +/-scale.
+ * Not +/-1 to avoid cube face selection ambiguity near the edges,
+ * though that can still sometimes happen with this scale factor...
+ */
+ const float scale = 0.9999f;
+ const float sc = (2 * in_st[0] - 1) * scale;
+ const float tc = (2 * in_st[1] - 1) * scale;
+
+ switch (face) {
+ case PIPE_TEX_FACE_POS_X:
+ rx = 1;
+ ry = -tc;
+ rz = -sc;
+ break;
+ case PIPE_TEX_FACE_NEG_X:
+ rx = -1;
+ ry = -tc;
+ rz = sc;
+ break;
+ case PIPE_TEX_FACE_POS_Y:
+ rx = sc;
+ ry = 1;
+ rz = tc;
+ break;
+ case PIPE_TEX_FACE_NEG_Y:
+ rx = sc;
+ ry = -1;
+ rz = -tc;
+ break;
+ case PIPE_TEX_FACE_POS_Z:
+ rx = sc;
+ ry = -tc;
+ rz = 1;
+ break;
+ case PIPE_TEX_FACE_NEG_Z:
+ rx = -sc;
+ ry = -tc;
+ rz = -1;
+ break;
+ default:
+ rx = ry = rz = 0;
+ assert(0);
+ }
+
+ out_str[0] = rx; /*s*/
+ out_str[1] = ry; /*t*/
+ out_str[2] = rz; /*r*/
+
+ in_st += in_stride;
+ out_str += out_stride;
+ }
+}
diff --git a/src/gallium/auxiliary/util/u_texture.h b/src/gallium/auxiliary/util/u_texture.h
new file mode 100644
index 00000000000..93b2f1e4c97
--- /dev/null
+++ b/src/gallium/auxiliary/util/u_texture.h
@@ -0,0 +1,54 @@
+/**************************************************************************
+ *
+ * Copyright 2009 Marek Olšák <[email protected]>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+#ifndef U_TEXTURE_H
+#define U_TEXTURE_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/**
+ * Convert 2D texture coordinates of 4 vertices into cubemap coordinates
+ * in the given face.
+ * Coordinates must be in the range [0,1].
+ *
+ * \param face Cubemap face.
+ * \param in_st 4 pairs of 2D texture coordinates to convert.
+ * \param in_stride Stride of in_st in floats.
+ * \param out_str STR cubemap texture coordinates to compute.
+ * \param out_stride Stride of out_str in floats.
+ */
+void util_map_texcoords2d_onto_cubemap(unsigned face,
+ const float *in_st, unsigned in_stride,
+ float *out_str, unsigned out_stride);
+
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif
diff --git a/src/gallium/auxiliary/util/u_tile.c b/src/gallium/auxiliary/util/u_tile.c
index 8a22f584bee..1ba82bb21f0 100644
--- a/src/gallium/auxiliary/util/u_tile.c
+++ b/src/gallium/auxiliary/util/u_tile.c
@@ -34,6 +34,7 @@
#include "pipe/p_defines.h"
#include "pipe/p_inlines.h"
+#include "util/u_format.h"
#include "util/u_math.h"
#include "util/u_memory.h"
#include "util/u_rect.h"
@@ -52,7 +53,7 @@ pipe_get_tile_raw(struct pipe_transfer *pt,
const void *src;
if (dst_stride == 0)
- dst_stride = pf_get_nblocksx(&pt->block, w) * pt->block.size;
+ dst_stride = util_format_get_stride(pt->texture->format, w);
if (pipe_clip_tile(x, y, &w, &h, pt))
return;
@@ -62,7 +63,7 @@ pipe_get_tile_raw(struct pipe_transfer *pt,
if(!src)
return;
- util_copy_rect(dst, &pt->block, dst_stride, 0, 0, w, h, src, pt->stride, x, y);
+ util_copy_rect(dst, pt->texture->format, dst_stride, 0, 0, w, h, src, pt->stride, x, y);
screen->transfer_unmap(screen, pt);
}
@@ -78,9 +79,10 @@ pipe_put_tile_raw(struct pipe_transfer *pt,
{
struct pipe_screen *screen = pt->texture->screen;
void *dst;
+ enum pipe_format format = pt->texture->format;
if (src_stride == 0)
- src_stride = pf_get_nblocksx(&pt->block, w) * pt->block.size;
+ src_stride = util_format_get_stride(format, w);
if (pipe_clip_tile(x, y, &w, &h, pt))
return;
@@ -90,7 +92,7 @@ pipe_put_tile_raw(struct pipe_transfer *pt,
if(!dst)
return;
- util_copy_rect(dst, &pt->block, pt->stride, x, y, w, h, src, src_stride, 0, 0);
+ util_copy_rect(dst, format, pt->stride, x, y, w, h, src, src_stride, 0, 0);
screen->transfer_unmap(screen, pt);
}
@@ -246,6 +248,53 @@ b8g8r8a8_put_tile_rgba(unsigned *dst,
}
+/*** PIPE_FORMAT_R8G8B8A8_UNORM ***/
+
+static void
+r8g8b8a8_get_tile_rgba(const unsigned *src,
+ unsigned w, unsigned h,
+ float *p,
+ unsigned dst_stride)
+{
+ unsigned i, j;
+
+ for (i = 0; i < h; i++) {
+ float *pRow = p;
+ for (j = 0; j < w; j++, pRow += 4) {
+ const unsigned pixel = *src++;
+ pRow[0] = ubyte_to_float((pixel >> 24) & 0xff);
+ pRow[1] = ubyte_to_float((pixel >> 16) & 0xff);
+ pRow[2] = ubyte_to_float((pixel >> 8) & 0xff);
+ pRow[3] = ubyte_to_float((pixel >> 0) & 0xff);
+ }
+ p += dst_stride;
+ }
+}
+
+
+static void
+r8g8b8a8_put_tile_rgba(unsigned *dst,
+ unsigned w, unsigned h,
+ const float *p,
+ unsigned src_stride)
+{
+ unsigned i, j;
+
+ for (i = 0; i < h; i++) {
+ const float *pRow = p;
+ for (j = 0; j < w; j++, pRow += 4) {
+ unsigned r, g, b, a;
+ r = float_to_ubyte(pRow[0]);
+ g = float_to_ubyte(pRow[1]);
+ b = float_to_ubyte(pRow[2]);
+ a = float_to_ubyte(pRow[3]);
+ *dst++ = (r << 24) | (g << 16) | (b << 8) | a;
+ }
+ p += src_stride;
+ }
+}
+
+
/*** PIPE_FORMAT_A1R5G5B5_UNORM ***/
static void
@@ -1106,27 +1155,6 @@ ycbcr_get_tile_rgba(const ushort *src,
}
-static void
-fake_get_tile_rgba(const ushort *src,
- unsigned w, unsigned h,
- float *p,
- unsigned dst_stride)
-{
- unsigned i, j;
-
- for (i = 0; i < h; i++) {
- float *pRow = p;
- for (j = 0; j < w; j++, pRow += 4) {
- pRow[0] =
- pRow[1] =
- pRow[2] =
- pRow[3] = (i ^ j) & 1 ? 1.0f : 0.0f;
- }
- p += dst_stride;
- }
-}
-
-
void
pipe_tile_raw_to_rgba(enum pipe_format format,
void *src,
@@ -1143,6 +1171,9 @@ pipe_tile_raw_to_rgba(enum pipe_format format,
case PIPE_FORMAT_B8G8R8A8_UNORM:
b8g8r8a8_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
break;
+ case PIPE_FORMAT_R8G8B8A8_UNORM:
+ r8g8b8a8_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
+ break;
case PIPE_FORMAT_A1R5G5B5_UNORM:
a1r5g5b5_get_tile_rgba((ushort *) src, w, h, dst, dst_stride);
break;
@@ -1206,8 +1237,10 @@ pipe_tile_raw_to_rgba(enum pipe_format format,
ycbcr_get_tile_rgba((ushort *) src, w, h, dst, dst_stride, TRUE);
break;
default:
- debug_printf("%s: unsupported format %s\n", __FUNCTION__, pf_name(format));
- fake_get_tile_rgba(src, w, h, dst, dst_stride);
+ util_format_read_4f(format,
+ dst, dst_stride * sizeof(float),
+ src, util_format_get_stride(format, w),
+ 0, 0, w, h);
}
}
@@ -1219,21 +1252,22 @@ pipe_get_tile_rgba(struct pipe_transfer *pt,
{
unsigned dst_stride = w * 4;
void *packed;
+ enum pipe_format format = pt->texture->format;
if (pipe_clip_tile(x, y, &w, &h, pt))
return;
- packed = MALLOC(pf_get_nblocks(&pt->block, w, h) * pt->block.size);
+ packed = MALLOC(util_format_get_nblocks(format, w, h) * util_format_get_blocksize(format));
if (!packed)
return;
- if(pt->format == PIPE_FORMAT_YCBCR || pt->format == PIPE_FORMAT_YCBCR_REV)
+ if(format == PIPE_FORMAT_YCBCR || format == PIPE_FORMAT_YCBCR_REV)
assert((x & 1) == 0);
pipe_get_tile_raw(pt, x, y, w, h, packed, 0);
- pipe_tile_raw_to_rgba(pt->format, packed, w, h, p, dst_stride);
+ pipe_tile_raw_to_rgba(format, packed, w, h, p, dst_stride);
FREE(packed);
}
@@ -1246,16 +1280,17 @@ pipe_put_tile_rgba(struct pipe_transfer *pt,
{
unsigned src_stride = w * 4;
void *packed;
+ enum pipe_format format = pt->texture->format;
if (pipe_clip_tile(x, y, &w, &h, pt))
return;
- packed = MALLOC(pf_get_nblocks(&pt->block, w, h) * pt->block.size);
+ packed = MALLOC(util_format_get_nblocks(format, w, h) * util_format_get_blocksize(format));
if (!packed)
return;
- switch (pt->format) {
+ switch (format) {
case PIPE_FORMAT_A8R8G8B8_UNORM:
a8r8g8b8_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);
break;
@@ -1265,6 +1300,9 @@ pipe_put_tile_rgba(struct pipe_transfer *pt,
case PIPE_FORMAT_B8G8R8A8_UNORM:
b8g8r8a8_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);
break;
+ case PIPE_FORMAT_R8G8B8A8_UNORM:
+ r8g8b8a8_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);
+ break;
case PIPE_FORMAT_A1R5G5B5_UNORM:
a1r5g5b5_put_tile_rgba((ushort *) packed, w, h, p, src_stride);
break;
@@ -1274,9 +1312,6 @@ pipe_put_tile_rgba(struct pipe_transfer *pt,
case PIPE_FORMAT_R8G8B8_UNORM:
r8g8b8_put_tile_rgba((ubyte *) packed, w, h, p, src_stride);
break;
- case PIPE_FORMAT_R8G8B8A8_UNORM:
- assert(0);
- break;
case PIPE_FORMAT_A4R4G4B4_UNORM:
a4r4g4b4_put_tile_rgba((ushort *) packed, w, h, p, src_stride);
break;
@@ -1322,7 +1357,7 @@ pipe_put_tile_rgba(struct pipe_transfer *pt,
/*z24s8_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);*/
break;
default:
- debug_printf("%s: unsupported format %s\n", __FUNCTION__, pf_name(pt->format));
+ debug_printf("%s: unsupported format %s\n", __FUNCTION__, pf_name(format));
}
pipe_put_tile_raw(pt, x, y, w, h, packed, 0);
@@ -1344,6 +1379,7 @@ pipe_get_tile_z(struct pipe_transfer *pt,
ubyte *map;
uint *pDest = z;
uint i, j;
+ enum pipe_format format = pt->texture->format;
if (pipe_clip_tile(x, y, &w, &h, pt))
return;
@@ -1354,7 +1390,7 @@ pipe_get_tile_z(struct pipe_transfer *pt,
return;
}
- switch (pt->format) {
+ switch (format) {
case PIPE_FORMAT_Z32_UNORM:
{
const uint *ptrc
@@ -1428,6 +1464,7 @@ pipe_put_tile_z(struct pipe_transfer *pt,
const uint *ptrc = zSrc;
ubyte *map;
uint i, j;
+ enum pipe_format format = pt->texture->format;
if (pipe_clip_tile(x, y, &w, &h, pt))
return;
@@ -1438,7 +1475,7 @@ pipe_put_tile_z(struct pipe_transfer *pt,
return;
}
- switch (pt->format) {
+ switch (format) {
case PIPE_FORMAT_Z32_UNORM:
{
uint *pDest = (uint *) (map + y * pt->stride + x*4);
diff --git a/src/gallium/auxiliary/util/u_upload_mgr.h b/src/gallium/auxiliary/util/u_upload_mgr.h
index 745b5834af6..e158bed9d04 100644
--- a/src/gallium/auxiliary/util/u_upload_mgr.h
+++ b/src/gallium/auxiliary/util/u_upload_mgr.h
@@ -32,6 +32,8 @@
#ifndef U_UPLOAD_MGR_H
#define U_UPLOAD_MGR_H
+#include "pipe/p_defines.h"
+
struct pipe_screen;
struct pipe_buffer;
struct u_upload_mgr;
diff --git a/src/gallium/auxiliary/vl/Makefile b/src/gallium/auxiliary/vl/Makefile
deleted file mode 100644
index 4314c1e8d69..00000000000
--- a/src/gallium/auxiliary/vl/Makefile
+++ /dev/null
@@ -1,13 +0,0 @@
-TOP = ../../../..
-include $(TOP)/configs/current
-
-LIBNAME = vl
-
-C_SOURCES = \
- vl_bitstream_parser.c \
- vl_mpeg12_mc_renderer.c \
- vl_compositor.c \
- vl_csc.c \
- vl_shader_build.c
-
-include ../../Makefile.template
diff --git a/src/gallium/auxiliary/vl/SConscript b/src/gallium/auxiliary/vl/SConscript
deleted file mode 100644
index aed69f5efed..00000000000
--- a/src/gallium/auxiliary/vl/SConscript
+++ /dev/null
@@ -1,13 +0,0 @@
-Import('*')
-
-vl = env.ConvenienceLibrary(
- target = 'vl',
- source = [
- 'vl_bitstream_parser.c',
- 'vl_mpeg12_mc_renderer.c',
- 'vl_compositor.c',
- 'vl_csc.c',
- 'vl_shader_build.c',
- ])
-
-auxiliaries.insert(0, vl)
diff --git a/src/gallium/auxiliary/vl/vl_compositor.c b/src/gallium/auxiliary/vl/vl_compositor.c
index cda6dc134a0..fc2a1c59a6b 100644
--- a/src/gallium/auxiliary/vl/vl_compositor.c
+++ b/src/gallium/auxiliary/vl/vl_compositor.c
@@ -95,12 +95,11 @@ create_vert_shader(struct vl_compositor *c)
assert(c);
tokens = (struct tgsi_token*)MALLOC(max_tokens * sizeof(struct tgsi_token));
- *(struct tgsi_version*)&tokens[0] = tgsi_build_version();
- header = (struct tgsi_header*)&tokens[1];
+ header = (struct tgsi_header*)&tokens[0];
*header = tgsi_build_header();
- *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
+ *(struct tgsi_processor*)&tokens[1] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
- ti = 3;
+ ti = 2;
/*
* decl i0 ; Vertex pos
@@ -172,12 +171,11 @@ create_frag_shader(struct vl_compositor *c)
assert(c);
tokens = (struct tgsi_token*)MALLOC(max_tokens * sizeof(struct tgsi_token));
- *(struct tgsi_version*)&tokens[0] = tgsi_build_version();
- header = (struct tgsi_header*)&tokens[1];
+ header = (struct tgsi_header*)&tokens[0];
*header = tgsi_build_header();
- *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
+ *(struct tgsi_processor*)&tokens[1] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
- ti = 3;
+ ti = 2;
/* decl i0 ; Texcoords for s0 */
decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, 1, 0, 0, TGSI_INTERPOLATE_LINEAR);
@@ -213,7 +211,7 @@ create_frag_shader(struct vl_compositor *c)
*/
for (i = 0; i < 4; ++i) {
inst = vl_inst3(TGSI_OPCODE_DP4, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i);
- inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
+ inst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_X << i;
ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
}
@@ -455,8 +453,8 @@ void vl_compositor_render(struct vl_compositor *compositor,
assert(dst_area);
assert(picture_type == PIPE_MPEG12_PICTURE_TYPE_FRAME);
- compositor->fb_state.width = dst_surface->width[0];
- compositor->fb_state.height = dst_surface->height[0];
+ compositor->fb_state.width = dst_surface->width0;
+ compositor->fb_state.height = dst_surface->height0;
compositor->fb_state.cbufs[0] = compositor->pipe->screen->get_tex_surface
(
compositor->pipe->screen,
@@ -479,8 +477,8 @@ void vl_compositor_render(struct vl_compositor *compositor,
compositor->pipe->set_framebuffer_state(compositor->pipe, &compositor->fb_state);
compositor->pipe->set_viewport_state(compositor->pipe, &compositor->viewport);
compositor->pipe->set_scissor_state(compositor->pipe, &compositor->scissor);
- compositor->pipe->bind_sampler_states(compositor->pipe, 1, &compositor->sampler);
- compositor->pipe->set_sampler_textures(compositor->pipe, 1, &src_surface);
+ compositor->pipe->bind_fragment_sampler_states(compositor->pipe, 1, &compositor->sampler);
+ compositor->pipe->set_fragment_sampler_textures(compositor->pipe, 1, &src_surface);
compositor->pipe->bind_vs_state(compositor->pipe, compositor->vertex_shader);
compositor->pipe->bind_fs_state(compositor->pipe, compositor->fragment_shader);
compositor->pipe->set_vertex_buffers(compositor->pipe, 2, compositor->vertex_bufs);
@@ -504,12 +502,12 @@ void vl_compositor_render(struct vl_compositor *compositor,
vs_consts->dst_trans.z = 0;
vs_consts->dst_trans.w = 0;
- vs_consts->src_scale.x = src_area->w / (float)src_surface->width[0];
- vs_consts->src_scale.y = src_area->h / (float)src_surface->height[0];
+ vs_consts->src_scale.x = src_area->w / (float)src_surface->width0;
+ vs_consts->src_scale.y = src_area->h / (float)src_surface->height0;
vs_consts->src_scale.z = 1;
vs_consts->src_scale.w = 1;
- vs_consts->src_trans.x = src_area->x / (float)src_surface->width[0];
- vs_consts->src_trans.y = src_area->y / (float)src_surface->height[0];
+ vs_consts->src_trans.x = src_area->x / (float)src_surface->width0;
+ vs_consts->src_trans.y = src_area->y / (float)src_surface->height0;
vs_consts->src_trans.z = 0;
vs_consts->src_trans.w = 0;
diff --git a/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c b/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c
index bbe0d5fa4c9..caf581aca60 100644
--- a/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c
+++ b/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c
@@ -29,6 +29,7 @@
#include <assert.h>
#include <pipe/p_context.h>
#include <pipe/p_inlines.h>
+#include <util/u_format.h>
#include <util/u_math.h>
#include <util/u_memory.h>
#include <tgsi/tgsi_parse.h>
@@ -115,12 +116,11 @@ create_intra_vert_shader(struct vl_mpeg12_mc_renderer *r)
assert(r);
tokens = (struct tgsi_token *) malloc(max_tokens * sizeof(struct tgsi_token));
- *(struct tgsi_version *) &tokens[0] = tgsi_build_version();
- header = (struct tgsi_header *) &tokens[1];
+ header = (struct tgsi_header *) &tokens[0];
*header = tgsi_build_header();
- *(struct tgsi_processor *) &tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
+ *(struct tgsi_processor *) &tokens[1] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
- ti = 3;
+ ti = 2;
/*
* decl i0 ; Vertex pos
@@ -185,12 +185,11 @@ create_intra_frag_shader(struct vl_mpeg12_mc_renderer *r)
assert(r);
tokens = (struct tgsi_token *) malloc(max_tokens * sizeof(struct tgsi_token));
- *(struct tgsi_version *) &tokens[0] = tgsi_build_version();
- header = (struct tgsi_header *) &tokens[1];
+ header = (struct tgsi_header *) &tokens[0];
*header = tgsi_build_header();
- *(struct tgsi_processor *) &tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
+ *(struct tgsi_processor *) &tokens[1] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
- ti = 3;
+ ti = 2;
/*
* decl i0 ; Luma texcoords
@@ -237,10 +236,10 @@ create_intra_frag_shader(struct vl_mpeg12_mc_renderer *r)
ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
- inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
- inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
- inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
- inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
+ inst.Src[0].Register.SwizzleX = TGSI_SWIZZLE_X;
+ inst.Src[0].Register.SwizzleY = TGSI_SWIZZLE_X;
+ inst.Src[0].Register.SwizzleZ = TGSI_SWIZZLE_X;
+ inst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_X << i;
ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
}
@@ -278,12 +277,11 @@ create_frame_pred_vert_shader(struct vl_mpeg12_mc_renderer *r)
assert(r);
tokens = (struct tgsi_token *) malloc(max_tokens * sizeof(struct tgsi_token));
- *(struct tgsi_version *) &tokens[0] = tgsi_build_version();
- header = (struct tgsi_header *) &tokens[1];
+ header = (struct tgsi_header *) &tokens[0];
*header = tgsi_build_header();
- *(struct tgsi_processor *) &tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
+ *(struct tgsi_processor *) &tokens[1] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
- ti = 3;
+ ti = 2;
/*
* decl i0 ; Vertex pos
@@ -363,12 +361,11 @@ create_frame_pred_frag_shader(struct vl_mpeg12_mc_renderer *r)
assert(r);
tokens = (struct tgsi_token *) malloc(max_tokens * sizeof(struct tgsi_token));
- *(struct tgsi_version *) &tokens[0] = tgsi_build_version();
- header = (struct tgsi_header *) &tokens[1];
+ header = (struct tgsi_header *) &tokens[0];
*header = tgsi_build_header();
- *(struct tgsi_processor *) &tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
+ *(struct tgsi_processor *) &tokens[1] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
- ti = 3;
+ ti = 2;
/*
* decl i0 ; Luma texcoords
@@ -417,10 +414,10 @@ create_frame_pred_frag_shader(struct vl_mpeg12_mc_renderer *r)
ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
- inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
- inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
- inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
- inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
+ inst.Src[0].Register.SwizzleX = TGSI_SWIZZLE_X;
+ inst.Src[0].Register.SwizzleY = TGSI_SWIZZLE_X;
+ inst.Src[0].Register.SwizzleZ = TGSI_SWIZZLE_X;
+ inst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_X << i;
ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
}
@@ -474,12 +471,11 @@ create_frame_bi_pred_vert_shader(struct vl_mpeg12_mc_renderer *r)
assert(r);
tokens = (struct tgsi_token *) malloc(max_tokens * sizeof(struct tgsi_token));
- *(struct tgsi_version *) &tokens[0] = tgsi_build_version();
- header = (struct tgsi_header *) &tokens[1];
+ header = (struct tgsi_header *) &tokens[0];
*header = tgsi_build_header();
- *(struct tgsi_processor *) &tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
+ *(struct tgsi_processor *) &tokens[1] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header);
- ti = 3;
+ ti = 2;
/*
* decl i0 ; Vertex pos
@@ -567,12 +563,11 @@ create_frame_bi_pred_frag_shader(struct vl_mpeg12_mc_renderer *r)
assert(r);
tokens = (struct tgsi_token *) malloc(max_tokens * sizeof(struct tgsi_token));
- *(struct tgsi_version *) &tokens[0] = tgsi_build_version();
- header = (struct tgsi_header *) &tokens[1];
+ header = (struct tgsi_header *) &tokens[0];
*header = tgsi_build_header();
- *(struct tgsi_processor *) &tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
+ *(struct tgsi_processor *) &tokens[1] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header);
- ti = 3;
+ ti = 2;
/*
* decl i0 ; Luma texcoords
@@ -626,10 +621,10 @@ create_frame_bi_pred_frag_shader(struct vl_mpeg12_mc_renderer *r)
ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1);
- inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
- inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
- inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
- inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i;
+ inst.Src[0].Register.SwizzleX = TGSI_SWIZZLE_X;
+ inst.Src[0].Register.SwizzleY = TGSI_SWIZZLE_X;
+ inst.Src[0].Register.SwizzleZ = TGSI_SWIZZLE_X;
+ inst.Dst[0].Register.WriteMask = TGSI_WRITEMASK_X << i;
ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
}
@@ -648,10 +643,10 @@ create_frame_bi_pred_frag_shader(struct vl_mpeg12_mc_renderer *r)
/* lerp t1, c1.x, t1, t2 ; Blend past and future texels */
inst = vl_inst4(TGSI_OPCODE_LRP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2);
- inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X;
- inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X;
- inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X;
- inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X;
+ inst.Src[0].Register.SwizzleX = TGSI_SWIZZLE_X;
+ inst.Src[0].Register.SwizzleY = TGSI_SWIZZLE_X;
+ inst.Src[0].Register.SwizzleZ = TGSI_SWIZZLE_X;
+ inst.Src[0].Register.SwizzleW = TGSI_SWIZZLE_X;
ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti);
/* add o0, t0, t1 ; Add past/future ref and differential to form final output */
@@ -689,7 +684,7 @@ xfer_buffers_map(struct vl_mpeg12_mc_renderer *r)
(
r->pipe->screen, r->textures.all[i],
0, 0, 0, PIPE_TRANSFER_WRITE, 0, 0,
- r->textures.all[i]->width[0], r->textures.all[i]->height[0]
+ r->textures.all[i]->width0, r->textures.all[i]->height0
);
r->texels[i] = r->pipe->screen->transfer_map(r->pipe->screen, r->tex_transfer[i]);
@@ -843,26 +838,25 @@ init_buffers(struct vl_mpeg12_mc_renderer *r)
/* TODO: Accomodate HW that can't do this and also for cases when this isn't precise enough */
template.format = PIPE_FORMAT_R16_SNORM;
template.last_level = 0;
- template.width[0] = r->pot_buffers ?
+ template.width0 = r->pot_buffers ?
util_next_power_of_two(r->picture_width) : r->picture_width;
- template.height[0] = r->pot_buffers ?
+ template.height0 = r->pot_buffers ?
util_next_power_of_two(r->picture_height) : r->picture_height;
- template.depth[0] = 1;
- pf_get_block(template.format, &template.block);
+ template.depth0 = 1;
template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER | PIPE_TEXTURE_USAGE_DYNAMIC;
r->textures.individual.y = r->pipe->screen->texture_create(r->pipe->screen, &template);
if (r->chroma_format == PIPE_VIDEO_CHROMA_FORMAT_420) {
- template.width[0] = r->pot_buffers ?
+ template.width0 = r->pot_buffers ?
util_next_power_of_two(r->picture_width / 2) :
r->picture_width / 2;
- template.height[0] = r->pot_buffers ?
+ template.height0 = r->pot_buffers ?
util_next_power_of_two(r->picture_height / 2) :
r->picture_height / 2;
}
else if (r->chroma_format == PIPE_VIDEO_CHROMA_FORMAT_422)
- template.height[0] = r->pot_buffers ?
+ template.height0 = r->pot_buffers ?
util_next_power_of_two(r->picture_height / 2) :
r->picture_height / 2;
@@ -1294,8 +1288,8 @@ flush(struct vl_mpeg12_mc_renderer *r)
PIPE_BUFFER_USAGE_CPU_WRITE | PIPE_BUFFER_USAGE_DISCARD
);
- vs_consts->denorm.x = r->surface->width[0];
- vs_consts->denorm.y = r->surface->height[0];
+ vs_consts->denorm.x = r->surface->width0;
+ vs_consts->denorm.y = r->surface->height0;
pipe_buffer_unmap(r->pipe->screen, r->vs_const_buf.buffer);
@@ -1307,8 +1301,8 @@ flush(struct vl_mpeg12_mc_renderer *r)
if (num_macroblocks[MACROBLOCK_TYPE_INTRA] > 0) {
r->pipe->set_vertex_buffers(r->pipe, 1, r->vertex_bufs.all);
r->pipe->set_vertex_elements(r->pipe, 4, r->vertex_elems);
- r->pipe->set_sampler_textures(r->pipe, 3, r->textures.all);
- r->pipe->bind_sampler_states(r->pipe, 3, r->samplers.all);
+ r->pipe->set_fragment_sampler_textures(r->pipe, 3, r->textures.all);
+ r->pipe->bind_fragment_sampler_states(r->pipe, 3, r->samplers.all);
r->pipe->bind_vs_state(r->pipe, r->i_vs);
r->pipe->bind_fs_state(r->pipe, r->i_fs);
@@ -1321,8 +1315,8 @@ flush(struct vl_mpeg12_mc_renderer *r)
r->pipe->set_vertex_buffers(r->pipe, 2, r->vertex_bufs.all);
r->pipe->set_vertex_elements(r->pipe, 6, r->vertex_elems);
r->textures.individual.ref[0] = r->past;
- r->pipe->set_sampler_textures(r->pipe, 4, r->textures.all);
- r->pipe->bind_sampler_states(r->pipe, 4, r->samplers.all);
+ r->pipe->set_fragment_sampler_textures(r->pipe, 4, r->textures.all);
+ r->pipe->bind_fragment_sampler_states(r->pipe, 4, r->samplers.all);
r->pipe->bind_vs_state(r->pipe, r->p_vs[0]);
r->pipe->bind_fs_state(r->pipe, r->p_fs[0]);
@@ -1335,8 +1329,8 @@ flush(struct vl_mpeg12_mc_renderer *r)
r->pipe->set_vertex_buffers(r->pipe, 2, r->vertex_bufs.all);
r->pipe->set_vertex_elements(r->pipe, 6, r->vertex_elems);
r->textures.individual.ref[0] = r->past;
- r->pipe->set_sampler_textures(r->pipe, 4, r->textures.all);
- r->pipe->bind_sampler_states(r->pipe, 4, r->samplers.all);
+ r->pipe->set_fragment_sampler_textures(r->pipe, 4, r->textures.all);
+ r->pipe->bind_fragment_sampler_states(r->pipe, 4, r->samplers.all);
r->pipe->bind_vs_state(r->pipe, r->p_vs[1]);
r->pipe->bind_fs_state(r->pipe, r->p_fs[1]);
@@ -1349,8 +1343,8 @@ flush(struct vl_mpeg12_mc_renderer *r)
r->pipe->set_vertex_buffers(r->pipe, 2, r->vertex_bufs.all);
r->pipe->set_vertex_elements(r->pipe, 6, r->vertex_elems);
r->textures.individual.ref[0] = r->future;
- r->pipe->set_sampler_textures(r->pipe, 4, r->textures.all);
- r->pipe->bind_sampler_states(r->pipe, 4, r->samplers.all);
+ r->pipe->set_fragment_sampler_textures(r->pipe, 4, r->textures.all);
+ r->pipe->bind_fragment_sampler_states(r->pipe, 4, r->samplers.all);
r->pipe->bind_vs_state(r->pipe, r->p_vs[0]);
r->pipe->bind_fs_state(r->pipe, r->p_fs[0]);
@@ -1363,8 +1357,8 @@ flush(struct vl_mpeg12_mc_renderer *r)
r->pipe->set_vertex_buffers(r->pipe, 2, r->vertex_bufs.all);
r->pipe->set_vertex_elements(r->pipe, 6, r->vertex_elems);
r->textures.individual.ref[0] = r->future;
- r->pipe->set_sampler_textures(r->pipe, 4, r->textures.all);
- r->pipe->bind_sampler_states(r->pipe, 4, r->samplers.all);
+ r->pipe->set_fragment_sampler_textures(r->pipe, 4, r->textures.all);
+ r->pipe->bind_fragment_sampler_states(r->pipe, 4, r->samplers.all);
r->pipe->bind_vs_state(r->pipe, r->p_vs[1]);
r->pipe->bind_fs_state(r->pipe, r->p_fs[1]);
@@ -1378,8 +1372,8 @@ flush(struct vl_mpeg12_mc_renderer *r)
r->pipe->set_vertex_elements(r->pipe, 8, r->vertex_elems);
r->textures.individual.ref[0] = r->past;
r->textures.individual.ref[1] = r->future;
- r->pipe->set_sampler_textures(r->pipe, 5, r->textures.all);
- r->pipe->bind_sampler_states(r->pipe, 5, r->samplers.all);
+ r->pipe->set_fragment_sampler_textures(r->pipe, 5, r->textures.all);
+ r->pipe->bind_fragment_sampler_states(r->pipe, 5, r->samplers.all);
r->pipe->bind_vs_state(r->pipe, r->b_vs[0]);
r->pipe->bind_fs_state(r->pipe, r->b_fs[0]);
@@ -1393,8 +1387,8 @@ flush(struct vl_mpeg12_mc_renderer *r)
r->pipe->set_vertex_elements(r->pipe, 8, r->vertex_elems);
r->textures.individual.ref[0] = r->past;
r->textures.individual.ref[1] = r->future;
- r->pipe->set_sampler_textures(r->pipe, 5, r->textures.all);
- r->pipe->bind_sampler_states(r->pipe, 5, r->samplers.all);
+ r->pipe->set_fragment_sampler_textures(r->pipe, 5, r->textures.all);
+ r->pipe->bind_fragment_sampler_states(r->pipe, 5, r->samplers.all);
r->pipe->bind_vs_state(r->pipe, r->b_vs[1]);
r->pipe->bind_fs_state(r->pipe, r->b_fs[1]);
@@ -1461,7 +1455,7 @@ grab_blocks(struct vl_mpeg12_mc_renderer *r, unsigned mbx, unsigned mby,
assert(r);
assert(blocks);
- tex_pitch = r->tex_transfer[0]->stride / r->tex_transfer[0]->block.size;
+ tex_pitch = r->tex_transfer[0]->stride / util_format_get_blocksize(r->tex_transfer[0]->texture->format);
texels = r->texels[0] + mbpy * tex_pitch + mbpx;
for (y = 0; y < 2; ++y) {
@@ -1500,7 +1494,7 @@ grab_blocks(struct vl_mpeg12_mc_renderer *r, unsigned mbx, unsigned mby,
mbpy /= 2;
for (tb = 0; tb < 2; ++tb) {
- tex_pitch = r->tex_transfer[tb + 1]->stride / r->tex_transfer[tb + 1]->block.size;
+ tex_pitch = r->tex_transfer[tb + 1]->stride / util_format_get_blocksize(r->tex_transfer[tb + 1]->texture->format);
texels = r->texels[tb + 1] + mbpy * tex_pitch + mbpx;
if ((cbp >> (1 - tb)) & 1) {
@@ -1644,8 +1638,8 @@ vl_mpeg12_mc_renderer_render_macroblocks(struct vl_mpeg12_mc_renderer
renderer->past = past;
renderer->future = future;
renderer->fence = fence;
- renderer->surface_tex_inv_size.x = 1.0f / surface->width[0];
- renderer->surface_tex_inv_size.y = 1.0f / surface->height[0];
+ renderer->surface_tex_inv_size.x = 1.0f / surface->width0;
+ renderer->surface_tex_inv_size.y = 1.0f / surface->height0;
}
while (num_macroblocks) {
diff --git a/src/gallium/auxiliary/vl/vl_shader_build.c b/src/gallium/auxiliary/vl/vl_shader_build.c
index faa20a903cd..d011ef97bd6 100644
--- a/src/gallium/auxiliary/vl/vl_shader_build.c
+++ b/src/gallium/auxiliary/vl/vl_shader_build.c
@@ -36,10 +36,10 @@ struct tgsi_full_declaration vl_decl_input(unsigned int name, unsigned int index
decl.Declaration.File = TGSI_FILE_INPUT;
decl.Declaration.Semantic = 1;
- decl.Semantic.SemanticName = name;
- decl.Semantic.SemanticIndex = index;
- decl.DeclarationRange.First = first;
- decl.DeclarationRange.Last = last;
+ decl.Semantic.Name = name;
+ decl.Semantic.Index = index;
+ decl.Range.First = first;
+ decl.Range.Last = last;
return decl;
}
@@ -64,11 +64,11 @@ struct tgsi_full_declaration vl_decl_interpolated_input
decl.Declaration.File = TGSI_FILE_INPUT;
decl.Declaration.Semantic = 1;
- decl.Semantic.SemanticName = name;
- decl.Semantic.SemanticIndex = index;
+ decl.Semantic.Name = name;
+ decl.Semantic.Index = index;
decl.Declaration.Interpolate = interpolation;;
- decl.DeclarationRange.First = first;
- decl.DeclarationRange.Last = last;
+ decl.Range.First = first;
+ decl.Range.Last = last;
return decl;
}
@@ -79,10 +79,10 @@ struct tgsi_full_declaration vl_decl_constants(unsigned int name, unsigned int i
decl.Declaration.File = TGSI_FILE_CONSTANT;
decl.Declaration.Semantic = 1;
- decl.Semantic.SemanticName = name;
- decl.Semantic.SemanticIndex = index;
- decl.DeclarationRange.First = first;
- decl.DeclarationRange.Last = last;
+ decl.Semantic.Name = name;
+ decl.Semantic.Index = index;
+ decl.Range.First = first;
+ decl.Range.Last = last;
return decl;
}
@@ -93,10 +93,10 @@ struct tgsi_full_declaration vl_decl_output(unsigned int name, unsigned int inde
decl.Declaration.File = TGSI_FILE_OUTPUT;
decl.Declaration.Semantic = 1;
- decl.Semantic.SemanticName = name;
- decl.Semantic.SemanticIndex = index;
- decl.DeclarationRange.First = first;
- decl.DeclarationRange.Last = last;
+ decl.Semantic.Name = name;
+ decl.Semantic.Index = index;
+ decl.Range.First = first;
+ decl.Range.Last = last;
return decl;
}
@@ -107,8 +107,8 @@ struct tgsi_full_declaration vl_decl_temps(unsigned int first, unsigned int last
decl = tgsi_default_full_declaration();
decl.Declaration.File = TGSI_FILE_TEMPORARY;
- decl.DeclarationRange.First = first;
- decl.DeclarationRange.Last = last;
+ decl.Range.First = first;
+ decl.Range.Last = last;
return decl;
}
@@ -119,8 +119,8 @@ struct tgsi_full_declaration vl_decl_samplers(unsigned int first, unsigned int l
decl = tgsi_default_full_declaration();
decl.Declaration.File = TGSI_FILE_SAMPLER;
- decl.DeclarationRange.First = first;
- decl.DeclarationRange.Last = last;
+ decl.Range.First = first;
+ decl.Range.Last = last;
return decl;
}
@@ -138,11 +138,11 @@ struct tgsi_full_instruction vl_inst2
inst.Instruction.Opcode = opcode;
inst.Instruction.NumDstRegs = 1;
- inst.FullDstRegisters[0].DstRegister.File = dst_file;
- inst.FullDstRegisters[0].DstRegister.Index = dst_index;
+ inst.Dst[0].Register.File = dst_file;
+ inst.Dst[0].Register.Index = dst_index;
inst.Instruction.NumSrcRegs = 1;
- inst.FullSrcRegisters[0].SrcRegister.File = src_file;
- inst.FullSrcRegisters[0].SrcRegister.Index = src_index;
+ inst.Src[0].Register.File = src_file;
+ inst.Src[0].Register.Index = src_index;
return inst;
}
@@ -162,13 +162,13 @@ struct tgsi_full_instruction vl_inst3
inst.Instruction.Opcode = opcode;
inst.Instruction.NumDstRegs = 1;
- inst.FullDstRegisters[0].DstRegister.File = dst_file;
- inst.FullDstRegisters[0].DstRegister.Index = dst_index;
+ inst.Dst[0].Register.File = dst_file;
+ inst.Dst[0].Register.Index = dst_index;
inst.Instruction.NumSrcRegs = 2;
- inst.FullSrcRegisters[0].SrcRegister.File = src1_file;
- inst.FullSrcRegisters[0].SrcRegister.Index = src1_index;
- inst.FullSrcRegisters[1].SrcRegister.File = src2_file;
- inst.FullSrcRegisters[1].SrcRegister.Index = src2_index;
+ inst.Src[0].Register.File = src1_file;
+ inst.Src[0].Register.Index = src1_index;
+ inst.Src[1].Register.File = src2_file;
+ inst.Src[1].Register.Index = src2_index;
return inst;
}
@@ -188,14 +188,15 @@ struct tgsi_full_instruction vl_tex
inst.Instruction.Opcode = TGSI_OPCODE_TEX;
inst.Instruction.NumDstRegs = 1;
- inst.FullDstRegisters[0].DstRegister.File = dst_file;
- inst.FullDstRegisters[0].DstRegister.Index = dst_index;
+ inst.Dst[0].Register.File = dst_file;
+ inst.Dst[0].Register.Index = dst_index;
inst.Instruction.NumSrcRegs = 2;
- inst.InstructionExtTexture.Texture = tex;
- inst.FullSrcRegisters[0].SrcRegister.File = src1_file;
- inst.FullSrcRegisters[0].SrcRegister.Index = src1_index;
- inst.FullSrcRegisters[1].SrcRegister.File = src2_file;
- inst.FullSrcRegisters[1].SrcRegister.Index = src2_index;
+ inst.Instruction.Texture = 1;
+ inst.Texture.Texture = tex;
+ inst.Src[0].Register.File = src1_file;
+ inst.Src[0].Register.Index = src1_index;
+ inst.Src[1].Register.File = src2_file;
+ inst.Src[1].Register.Index = src2_index;
return inst;
}
@@ -217,15 +218,15 @@ struct tgsi_full_instruction vl_inst4
inst.Instruction.Opcode = opcode;
inst.Instruction.NumDstRegs = 1;
- inst.FullDstRegisters[0].DstRegister.File = dst_file;
- inst.FullDstRegisters[0].DstRegister.Index = dst_index;
+ inst.Dst[0].Register.File = dst_file;
+ inst.Dst[0].Register.Index = dst_index;
inst.Instruction.NumSrcRegs = 3;
- inst.FullSrcRegisters[0].SrcRegister.File = src1_file;
- inst.FullSrcRegisters[0].SrcRegister.Index = src1_index;
- inst.FullSrcRegisters[1].SrcRegister.File = src2_file;
- inst.FullSrcRegisters[1].SrcRegister.Index = src2_index;
- inst.FullSrcRegisters[2].SrcRegister.File = src3_file;
- inst.FullSrcRegisters[2].SrcRegister.Index = src3_index;
+ inst.Src[0].Register.File = src1_file;
+ inst.Src[0].Register.Index = src1_index;
+ inst.Src[1].Register.File = src2_file;
+ inst.Src[1].Register.Index = src2_index;
+ inst.Src[2].Register.File = src3_file;
+ inst.Src[2].Register.Index = src3_index;
return inst;
}