Merge remote-tracking branch 'origin/master' into pipe-video

Conflicts:
	configure.ac
	src/gallium/drivers/r600/r600_state_inlines.h
	src/gallium/tests/trivial/Makefile
	src/gallium/winsys/g3dvl/dri/XF86dri.c
	src/gallium/winsys/g3dvl/dri/driclient.c
	src/gallium/winsys/g3dvl/dri/driclient.h
	src/gallium/winsys/g3dvl/dri/xf86dri.h
	src/gallium/winsys/g3dvl/dri/xf86dristr.h
	src/gallium/winsys/r600/drm/r600_bo.c

18 files changed, 151 insertions, 120 deletions

diff --git a/src/gallium/auxiliary/draw/draw_llvm.c b/src/gallium/auxiliary/draw/draw_llvm.c
index 56c26f57cce..f33c9078c9c 100644
--- a/src/gallium/auxiliary/draw/draw_llvm.c
+++ b/src/gallium/auxiliary/draw/draw_llvm.c
@@ -1163,6 +1163,7 @@ draw_llvm_generate(struct draw_llvm *llvm, struct draw_llvm_variant *variant)
    struct lp_build_loop_state lp_loop;
    const int max_vertices = 4;
    LLVMValueRef outputs[PIPE_MAX_SHADER_OUTPUTS][NUM_CHANNELS];
+   LLVMValueRef fetch_max;
    void *code;
    struct lp_build_sampler_soa *sampler = 0;
    LLVMValueRef ret, ret_ptr;
@@ -1234,6 +1235,10 @@ draw_llvm_generate(struct draw_llvm *llvm, struct draw_llvm_variant *variant)
       draw_llvm_variant_key_samplers(&variant->key),
       context_ptr);
 
+   fetch_max = LLVMBuildSub(builder, count,
+                            lp_build_const_int32(gallivm, 1),
+                            "fetch_max");
+
 #if DEBUG_STORE
    lp_build_printf(builder, "start = %d, end = %d, step = %d\n",
                    start, end, step);
@@ -1257,6 +1262,12 @@ draw_llvm_generate(struct draw_llvm *llvm, struct draw_llvm_variant *variant)
             builder,
             lp_loop.counter,
             lp_build_const_int32(gallivm, i), "");
+
+         /* make sure we're not out of bounds which can happen
+          * if fetch_count % 4 != 0, because on the last iteration
+          * a few of the 4 vertex fetches will be out of bounds */
+         true_index = lp_build_min(&bld, true_index, fetch_max);
+
          for (j = 0; j < draw->pt.nr_vertex_elements; ++j) {
             struct pipe_vertex_element *velem = &draw->pt.vertex_element[j];
             LLVMValueRef vb_index = lp_build_const_int32(gallivm, velem->vertex_buffer_index);
diff --git a/src/gallium/auxiliary/draw/draw_pipe_aaline.c b/src/gallium/auxiliary/draw/draw_pipe_aaline.c
index 32af29ae144..458f85def2c 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_aaline.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_aaline.c
@@ -784,6 +784,14 @@ aaline_destroy(struct draw_stage *stage)
 
    draw_free_temp_verts( stage );
 
+   /* restore the old entry points */
+   pipe->create_fs_state = aaline->driver_create_fs_state;
+   pipe->bind_fs_state = aaline->driver_bind_fs_state;
+   pipe->delete_fs_state = aaline->driver_delete_fs_state;
+
+   pipe->bind_fragment_sampler_states = aaline->driver_bind_sampler_states;
+   pipe->set_fragment_sampler_views = aaline->driver_set_sampler_views;
+
    FREE( stage );
 }
 
diff --git a/src/gallium/auxiliary/draw/draw_pipe_aapoint.c b/src/gallium/auxiliary/draw/draw_pipe_aapoint.c
index 60f6380c503..9265c379de8 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_aapoint.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_aapoint.c
@@ -768,7 +768,16 @@ aapoint_reset_stipple_counter(struct draw_stage *stage)
 static void
 aapoint_destroy(struct draw_stage *stage)
 {
+   struct aapoint_stage* aapoint = aapoint_stage(stage);
+   struct pipe_context *pipe = stage->draw->pipe;
+
    draw_free_temp_verts( stage );
+
+   /* restore the old entry points */
+   pipe->create_fs_state = aapoint->driver_create_fs_state;
+   pipe->bind_fs_state = aapoint->driver_bind_fs_state;
+   pipe->delete_fs_state = aapoint->driver_delete_fs_state;
+
    FREE( stage );
 }
 
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_debug.cpp b/src/gallium/auxiliary/gallivm/lp_bld_debug.cpp
index 0b724a34948..01e660ef7d9 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_debug.cpp
+++ b/src/gallium/auxiliary/gallivm/lp_bld_debug.cpp
@@ -207,7 +207,11 @@ lp_disassemble(const void* func)
    }
 
    raw_debug_ostream Out;
+#if HAVE_LLVM >= 0x0300
+   TargetMachine *TM = T->createTargetMachine(Triple, sys::getHostCPUName(), "");
+#else
    TargetMachine *TM = T->createTargetMachine(Triple, "");
+#endif
 
 #if HAVE_LLVM >= 0x0300
    unsigned int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
@@ -287,7 +291,11 @@ lp_disassemble(const void* func)
 
       pc += Size;
 
+#if HAVE_LLVM >= 0x0300
+      const MCInstrDesc &TID = TII->get(Inst.getOpcode());
+#else
       const TargetInstrDesc &TID = TII->get(Inst.getOpcode());
+#endif
 
       /*
        * Keep track of forward jumps to a nearby address.
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_misc.cpp b/src/gallium/auxiliary/gallivm/lp_bld_misc.cpp
index d2d7eccd92f..85fabc574b2 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_misc.cpp
+++ b/src/gallium/auxiliary/gallivm/lp_bld_misc.cpp
@@ -83,8 +83,12 @@ lp_set_target_options(void)
     * to only assume a 4 bytes alignment for backwards compatibility.
     */
 #if defined(PIPE_ARCH_X86)
+#if HAVE_LLVM >= 0x0300
+   llvm::StackAlignmentOverride = 4;
+#else
    llvm::StackAlignment = 4;
 #endif
+#endif
 
 #if defined(DEBUG) || defined(PROFILE)
    llvm::NoFramePointerElim = true;
diff --git a/src/gallium/auxiliary/indices/u_unfilled_gen.py b/src/gallium/auxiliary/indices/u_unfilled_gen.py
index 36896ce605d..085c47a114a 100644
--- a/src/gallium/auxiliary/indices/u_unfilled_gen.py
+++ b/src/gallium/auxiliary/indices/u_unfilled_gen.py
@@ -167,8 +167,8 @@ def trifan(intype, outtype):
 
 def polygon(intype, outtype):
     preamble(intype, outtype, prim='polygon')
-    print '  for (j = i = 0; j < nr; j+=6, i++) { '
-    do_tri( intype, outtype, 'out+j',  '0', 'i+1', 'i+2' );
+    print '  for (j = i = 0; j < nr; j+=2, i++) { '
+    line( intype, outtype, 'out+j', 'i', '(i+1)%(nr/2)' )
     print '   }'
     postamble()
 
diff --git a/src/gallium/auxiliary/indices/u_unfilled_indices.c b/src/gallium/auxiliary/indices/u_unfilled_indices.c
index 26c5d4d4c72..c353717d656 100644
--- a/src/gallium/auxiliary/indices/u_unfilled_indices.c
+++ b/src/gallium/auxiliary/indices/u_unfilled_indices.c
@@ -71,6 +71,11 @@ static void generate_linear_uint( unsigned nr,
 }
 
 
+/**
+ * Given a primitive type and number of vertices, return the number of vertices
+ * needed to draw the primitive with fill mode = PIPE_POLYGON_MODE_LINE using
+ * separate lines (PIPE_PRIM_LINES).
+ */
 static unsigned nr_lines( unsigned prim,
                           unsigned nr )
 {
@@ -86,7 +91,7 @@ static unsigned nr_lines( unsigned prim,
    case PIPE_PRIM_QUAD_STRIP:
       return (nr - 2) / 2 * 8;
    case PIPE_PRIM_POLYGON:
-      return (nr - 2) * 6;
+      return 2 * nr; /* a line (two verts) for each polygon edge */
    default:
       assert(0);
       return 0;
diff --git a/src/gallium/auxiliary/os/os_memory.h b/src/gallium/auxiliary/os/os_memory.h
index 556662d35e1..91a84a24bc8 100644
--- a/src/gallium/auxiliary/os/os_memory.h
+++ b/src/gallium/auxiliary/os/os_memory.h
@@ -39,7 +39,7 @@
 #include "pipe/p_compiler.h"
 
 
-#if defined(PIPE_OS_EMBEDDED)
+#if defined(PIPE_SUBSYSTEM_EMBEDDED)
 
 #ifdef __cplusplus
 extern "C" {
diff --git a/src/gallium/auxiliary/os/os_misc.h b/src/gallium/auxiliary/os/os_misc.h
index d59f9819fec..48522dac4d7 100644
--- a/src/gallium/auxiliary/os/os_misc.h
+++ b/src/gallium/auxiliary/os/os_misc.h
@@ -58,8 +58,6 @@ extern "C" {
 #  define os_break()  __debugbreak()
 #elif defined(PIPE_OS_UNIX)
 #  define os_break() kill(getpid(), SIGTRAP)
-#elif defined(PIPE_OS_EMBEDDED)
-void os_break(void);
 #else
 #  define os_break() abort()
 #endif
@@ -70,8 +68,6 @@ void os_break(void);
  */
 #if defined(DEBUG) || defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY) || defined(PIPE_SUBSYSTEM_WINDOWS_MINIPORT)
 #  define os_abort() os_break()
-#elif defined(PIPE_OS_EMBEDDED)
-void os_abort(void);
 #else
 #  define os_abort() abort()
 #endif
diff --git a/src/gallium/auxiliary/os/os_thread.h b/src/gallium/auxiliary/os/os_thread.h
index 6b4281ad661..8f1245bff55 100644
--- a/src/gallium/auxiliary/os/os_thread.h
+++ b/src/gallium/auxiliary/os/os_thread.h
@@ -40,7 +40,7 @@
 #include "util/u_debug.h" /* for assert */
 
 
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_EMBEDDED) || defined(PIPE_OS_CYGWIN)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_CYGWIN)
 
 #include <pthread.h> /* POSIX threads headers */
 #include <stdio.h> /* for perror() */
@@ -314,7 +314,7 @@ typedef int64_t pipe_condvar;
  * pipe_barrier
  */
 
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_EMBEDDED)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_HAIKU)
 
 typedef pthread_barrier_t pipe_barrier;
 
@@ -442,7 +442,7 @@ pipe_semaphore_wait(pipe_semaphore *sema)
  */
 
 typedef struct {
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_EMBEDDED) || defined(PIPE_OS_CYGWIN)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_CYGWIN)
    pthread_key_t key;
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_USER)
    DWORD key;
@@ -457,7 +457,7 @@ typedef struct {
 static INLINE void
 pipe_tsd_init(pipe_tsd *tsd)
 {
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_EMBEDDED) || defined(PIPE_OS_CYGWIN)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_CYGWIN)
    if (pthread_key_create(&tsd->key, NULL/*free*/) != 0) {
       perror("pthread_key_create(): failed to allocate key for thread specific data");
       exit(-1);
@@ -474,7 +474,7 @@ pipe_tsd_get(pipe_tsd *tsd)
    if (tsd->initMagic != (int) PIPE_TSD_INIT_MAGIC) {
       pipe_tsd_init(tsd);
    }
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_EMBEDDED) || defined(PIPE_OS_CYGWIN)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_CYGWIN)
    return pthread_getspecific(tsd->key);
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_USER)
    assert(0);
@@ -491,7 +491,7 @@ pipe_tsd_set(pipe_tsd *tsd, void *value)
    if (tsd->initMagic != (int) PIPE_TSD_INIT_MAGIC) {
       pipe_tsd_init(tsd);
    }
-#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_EMBEDDED) || defined(PIPE_OS_CYGWIN)
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) || defined(PIPE_OS_SOLARIS) || defined(PIPE_OS_APPLE) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_CYGWIN)
    if (pthread_setspecific(tsd->key, value) != 0) {
       perror("pthread_set_specific() failed");
       exit(-1);
diff --git a/src/gallium/auxiliary/os/os_time.c b/src/gallium/auxiliary/os/os_time.c
index 325f316784c..73d86296d91 100644
--- a/src/gallium/auxiliary/os/os_time.c
+++ b/src/gallium/auxiliary/os/os_time.c
@@ -35,8 +35,6 @@
 
 #include "pipe/p_config.h"
 
-#if !defined(PIPE_OS_EMBEDDED)
-
 #if defined(PIPE_OS_UNIX)
 #  include <sys/time.h> /* timeval */
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_DISPLAY)
@@ -123,6 +121,3 @@ os_time_sleep(int64_t usecs)
 }
 
 #endif
-
-
-#endif /* !PIPE_OS_EMBEDDED */
diff --git a/src/gallium/auxiliary/target-helpers/inline_debug_helper.h b/src/gallium/auxiliary/target-helpers/inline_debug_helper.h
index 0433da6141d..c6630c40f5c 100644
--- a/src/gallium/auxiliary/target-helpers/inline_debug_helper.h
+++ b/src/gallium/auxiliary/target-helpers/inline_debug_helper.h
@@ -22,6 +22,10 @@
 #include "galahad/glhd_public.h"
 #endif
 
+#ifdef GALLIUM_NOOP
+#include "noop/noop_public.h"
+#endif
+
 static INLINE struct pipe_screen *
 debug_screen_wrap(struct pipe_screen *screen)
 {
@@ -38,6 +42,10 @@ debug_screen_wrap(struct pipe_screen *screen)
    screen = galahad_screen_create(screen);
 #endif
 
+#if defined(GALLIUM_NOOP)
+   screen = noop_screen_create(screen);
+#endif
+
    return screen;
 }
 
diff --git a/src/gallium/auxiliary/target-helpers/inline_noop_helper.h b/src/gallium/auxiliary/target-helpers/inline_noop_helper.h
deleted file mode 100644
index 77c7cfd0c20..00000000000
--- a/src/gallium/auxiliary/target-helpers/inline_noop_helper.h
+++ /dev/null
@@ -1,51 +0,0 @@
-
-#ifndef INLINE_DEBUG_HELPER_H
-#define INLINE_DEBUG_HELPER_H
-
-#include "pipe/p_compiler.h"
-#include "util/u_debug.h"
-
-
-/* Helper function to wrap a screen with
- * one or more debug driver: rbug, trace.
- */
-
-#ifdef GALLIUM_TRACE
-#include "trace/tr_public.h"
-#endif
-
-#ifdef GALLIUM_RBUG
-#include "rbug/rbug_public.h"
-#endif
-
-#ifdef GALLIUM_GALAHAD
-#include "galahad/glhd_public.h"
-#endif
-
-#ifdef GALLIUM_NOOP
-#include "noop/noop_public.h"
-#endif
-
-static INLINE struct pipe_screen *
-debug_screen_wrap(struct pipe_screen *screen)
-{
-#if defined(GALLIUM_RBUG)
-   screen = rbug_screen_create(screen);
-#endif
-
-#if defined(GALLIUM_TRACE)
-   screen = trace_screen_create(screen);
-#endif
-
-#if defined(GALLIUM_GALAHAD)
-   screen = galahad_screen_create(screen);
-#endif
-
-#if defined(GALLIUM_NOOP)
-   screen = noop_screen_create(screen);
-#endif
-
-   return screen;
-}
-
-#endif
diff --git a/src/gallium/auxiliary/util/u_debug.c b/src/gallium/auxiliary/util/u_debug.c
index 36ce4b57713..004df439ff5 100644
--- a/src/gallium/auxiliary/util/u_debug.c
+++ b/src/gallium/auxiliary/util/u_debug.c
@@ -48,7 +48,7 @@
 
 void _debug_vprintf(const char *format, va_list ap)
 {
-#if defined(PIPE_OS_WINDOWS) || defined(PIPE_OS_EMBEDDED)
+#if defined(PIPE_OS_WINDOWS) || defined(PIPE_SUBSYSTEM_EMBEDDED)
    /* We buffer until we find a newline. */
    static char buf[4096] = {'\0'};
    size_t len = strlen(buf);
diff --git a/src/gallium/auxiliary/util/u_upload_mgr.c b/src/gallium/auxiliary/util/u_upload_mgr.c
index 9562acb8210..e50db6d67fe 100644
--- a/src/gallium/auxiliary/util/u_upload_mgr.c
+++ b/src/gallium/auxiliary/util/u_upload_mgr.c
@@ -72,6 +72,22 @@ struct u_upload_mgr *u_upload_create( struct pipe_context *pipe,
    return upload;
 }
 
+void u_upload_unmap( struct u_upload_mgr *upload )
+{
+   if (upload->transfer) {
+      struct pipe_box *box = &upload->transfer->box;
+      if (upload->offset > box->x) {
+
+         pipe_buffer_flush_mapped_range(upload->pipe, upload->transfer,
+                                        box->x, upload->offset - box->x);
+      }
+      pipe_transfer_unmap(upload->pipe, upload->transfer);
+      pipe_transfer_destroy(upload->pipe, upload->transfer);
+      upload->transfer = NULL;
+      upload->map = NULL;
+   }
+}
+
 /* Release old buffer.
  * 
  * This must usually be called prior to firing the command stream
@@ -84,15 +100,7 @@ struct u_upload_mgr *u_upload_create( struct pipe_context *pipe,
 void u_upload_flush( struct u_upload_mgr *upload )
 {
    /* Unmap and unreference the upload buffer. */
-   if (upload->transfer) {
-      if (upload->offset) {
-         pipe_buffer_flush_mapped_range(upload->pipe, upload->transfer,
-                                        0, upload->offset);
-      }
-      pipe_transfer_unmap(upload->pipe, upload->transfer);
-      pipe_transfer_destroy(upload->pipe, upload->transfer);
-      upload->transfer = NULL;
-   }
+   u_upload_unmap(upload);
    pipe_resource_reference( &upload->buffer, NULL );
    upload->size = 0;
 }
@@ -172,6 +180,15 @@ enum pipe_error u_upload_alloc( struct u_upload_mgr *upload,
 
    offset = MAX2(upload->offset, alloc_offset);
 
+   if (!upload->map) {
+      upload->map = pipe_buffer_map_range(upload->pipe, upload->buffer,
+					  offset, upload->size - offset,
+					  PIPE_TRANSFER_WRITE |
+					  PIPE_TRANSFER_FLUSH_EXPLICIT |
+					  PIPE_TRANSFER_UNSYNCHRONIZED,
+					  &upload->transfer);
+   }
+
    assert(offset < upload->buffer->width0);
    assert(offset + size <= upload->buffer->width0);
    assert(size);
@@ -223,10 +240,11 @@ enum pipe_error u_upload_buffer( struct u_upload_mgr *upload,
    struct pipe_transfer *transfer = NULL;
    const char *map = NULL;
 
-   map = (const char *)pipe_buffer_map(upload->pipe,
-				       inbuf,
-				       PIPE_TRANSFER_READ,
-				       &transfer);
+   map = (const char *)pipe_buffer_map_range(upload->pipe,
+                                             inbuf,
+                                             offset, size,
+                                             PIPE_TRANSFER_READ,
+                                             &transfer);
 
    if (map == NULL) {
       ret = PIPE_ERROR_OUT_OF_MEMORY;
@@ -239,7 +257,7 @@ enum pipe_error u_upload_buffer( struct u_upload_mgr *upload,
    ret = u_upload_data( upload,
                         min_out_offset,
                         size,
-                        map + offset,
+                        map,
                         out_offset,
                         outbuf, flushed );
 
diff --git a/src/gallium/auxiliary/util/u_upload_mgr.h b/src/gallium/auxiliary/util/u_upload_mgr.h
index c9a2ffeb572..98915139801 100644
--- a/src/gallium/auxiliary/util/u_upload_mgr.h
+++ b/src/gallium/auxiliary/util/u_upload_mgr.h
@@ -56,15 +56,27 @@ struct u_upload_mgr *u_upload_create( struct pipe_context *pipe,
  */
 void u_upload_destroy( struct u_upload_mgr *upload );
 
-/* Unmap and release old buffer.
+/* Unmap and release old upload buffer.
  * 
+ * This is like u_upload_unmap() except the upload buffer is released for
+ * recycling. This should be called on real hardware flushes on systems
+ * that don't support the PIPE_TRANSFER_UNSYNCHRONIZED flag, as otherwise
+ * the next u_upload_buffer will cause a sync on the buffer.
+ */
+
+void u_upload_flush( struct u_upload_mgr *upload );
+
+/**
+ * Unmap upload buffer
+ *
+ * \param upload           Upload manager
+ *
  * This must usually be called prior to firing the command stream
  * which references the upload buffer, as many memory managers either
  * don't like firing a mapped buffer or cause subsequent maps of a
- * fired buffer to wait.  For now, it's easiest just to grab a new
- * buffer.
+ * fired buffer to wait.
  */
-void u_upload_flush( struct u_upload_mgr *upload );
+void u_upload_unmap( struct u_upload_mgr *upload );
 
 /**
  * Sub-allocate new memory from the upload buffer.
diff --git a/src/gallium/auxiliary/util/u_vbuf_mgr.c b/src/gallium/auxiliary/util/u_vbuf_mgr.c
index 04149525ea7..374fc336b83 100644
--- a/src/gallium/auxiliary/util/u_vbuf_mgr.c
+++ b/src/gallium/auxiliary/util/u_vbuf_mgr.c
@@ -152,9 +152,9 @@ void u_vbuf_mgr_destroy(struct u_vbuf_mgr *mgrb)
 }
 
 
-static void u_vbuf_translate_begin(struct u_vbuf_mgr_priv *mgr,
-                                   int min_index, int max_index,
-                                   boolean *upload_flushed)
+static enum u_vbuf_return_flags
+u_vbuf_translate_begin(struct u_vbuf_mgr_priv *mgr,
+                       int min_index, int max_index)
 {
    struct translate_key key;
    struct translate_element *te;
@@ -166,6 +166,7 @@ static void u_vbuf_translate_begin(struct u_vbuf_mgr_priv *mgr,
    struct pipe_resource *out_buffer = NULL;
    unsigned i, num_verts, out_offset;
    struct pipe_vertex_element new_velems[PIPE_MAX_ATTRIBS];
+   boolean upload_flushed = FALSE;
 
    memset(&key, 0, sizeof(key));
    memset(tr_elem_index, 0xff, sizeof(tr_elem_index));
@@ -248,7 +249,7 @@ static void u_vbuf_translate_begin(struct u_vbuf_mgr_priv *mgr,
    u_upload_alloc(mgr->b.uploader,
                   key.output_stride * min_index,
                   key.output_stride * num_verts,
-                  &out_offset, &out_buffer, upload_flushed,
+                  &out_offset, &out_buffer, &upload_flushed,
                   (void**)&out_map);
 
    out_offset -= key.output_stride * min_index;
@@ -308,6 +309,8 @@ static void u_vbuf_translate_begin(struct u_vbuf_mgr_priv *mgr,
    }
 
    pipe_resource_reference(&out_buffer, NULL);
+
+   return upload_flushed ? U_VBUF_UPLOAD_FLUSHED : 0;
 }
 
 static void u_vbuf_translate_end(struct u_vbuf_mgr_priv *mgr)
@@ -510,14 +513,15 @@ void u_vbuf_mgr_set_vertex_buffers(struct u_vbuf_mgr *mgrb,
    mgr->b.nr_real_vertex_buffers = count;
 }
 
-static void u_vbuf_upload_buffers(struct u_vbuf_mgr_priv *mgr,
-                                  int min_index, int max_index,
-                                  unsigned instance_count,
-                                  boolean *upload_flushed)
+static enum u_vbuf_return_flags
+u_vbuf_upload_buffers(struct u_vbuf_mgr_priv *mgr,
+                      int min_index, int max_index,
+                      unsigned instance_count)
 {
    unsigned i, nr = mgr->ve->count;
    unsigned count = max_index + 1 - min_index;
    boolean uploaded[PIPE_MAX_ATTRIBS] = {0};
+   enum u_vbuf_return_flags retval = 0;
 
    for (i = 0; i < nr; i++) {
       unsigned index = mgr->ve->ve[i].vertex_buffer_index;
@@ -537,6 +541,11 @@ static void u_vbuf_upload_buffers(struct u_vbuf_mgr_priv *mgr,
          } else if (vb->stride) {
             first = vb->stride * min_index;
             size = vb->stride * count;
+
+            /* Unusual case when stride is smaller than the format size.
+             * XXX This won't work with interleaved arrays. */
+            if (mgr->ve->native_format_size[i] > vb->stride)
+               size += mgr->ve->native_format_size[i] - vb->stride;
          } else {
             first = 0;
             size = mgr->ve->native_format_size[i];
@@ -551,11 +560,14 @@ static void u_vbuf_upload_buffers(struct u_vbuf_mgr_priv *mgr,
          vb->buffer_offset -= first;
 
          uploaded[index] = TRUE;
-         *upload_flushed = *upload_flushed || flushed;
+         if (flushed)
+            retval |= U_VBUF_UPLOAD_FLUSHED;
       } else {
          assert(mgr->b.real_vertex_buffer[index]);
       }
    }
+
+   return retval;
 }
 
 static void u_vbuf_mgr_compute_max_index(struct u_vbuf_mgr_priv *mgr)
@@ -597,14 +609,13 @@ static void u_vbuf_mgr_compute_max_index(struct u_vbuf_mgr_priv *mgr)
    }
 }
 
-void u_vbuf_mgr_draw_begin(struct u_vbuf_mgr *mgrb,
-                           const struct pipe_draw_info *info,
-                           boolean *buffers_updated,
-                           boolean *uploader_flushed)
+enum u_vbuf_return_flags
+u_vbuf_mgr_draw_begin(struct u_vbuf_mgr *mgrb,
+                      const struct pipe_draw_info *info)
 {
    struct u_vbuf_mgr_priv *mgr = (struct u_vbuf_mgr_priv*)mgrb;
-   boolean bufs_updated = FALSE, upload_flushed = FALSE;
    int min_index, max_index;
+   enum u_vbuf_return_flags retval = 0;
 
    u_vbuf_mgr_compute_max_index(mgr);
 
@@ -617,27 +628,20 @@ void u_vbuf_mgr_draw_begin(struct u_vbuf_mgr *mgrb,
 
    /* Translate vertices with non-native layouts or formats. */
    if (mgr->incompatible_vb_layout || mgr->ve->incompatible_layout) {
-      u_vbuf_translate_begin(mgr, min_index, max_index, &upload_flushed);
+      retval |= u_vbuf_translate_begin(mgr, min_index, max_index);
 
       if (mgr->fallback_ve) {
-         bufs_updated = TRUE;
+         retval |= U_VBUF_BUFFERS_UPDATED;
       }
    }
 
    /* Upload user buffers. */
    if (mgr->any_user_vbs) {
-      u_vbuf_upload_buffers(mgr, min_index, max_index, info->instance_count,
-                            &upload_flushed);
-      bufs_updated = TRUE;
-   }
-
-   /* Set the return values. */
-   if (buffers_updated) {
-      *buffers_updated = bufs_updated;
-   }
-   if (uploader_flushed) {
-      *uploader_flushed = upload_flushed;
+      retval |= u_vbuf_upload_buffers(mgr, min_index, max_index,
+                                      info->instance_count);
+      retval |= U_VBUF_BUFFERS_UPDATED;
    }
+   return retval;
 }
 
 void u_vbuf_mgr_draw_end(struct u_vbuf_mgr *mgrb)
diff --git a/src/gallium/auxiliary/util/u_vbuf_mgr.h b/src/gallium/auxiliary/util/u_vbuf_mgr.h
index 9380dce4f72..4e6372435d8 100644
--- a/src/gallium/auxiliary/util/u_vbuf_mgr.h
+++ b/src/gallium/auxiliary/util/u_vbuf_mgr.h
@@ -78,6 +78,11 @@ enum u_fetch_alignment {
    U_VERTEX_FETCH_DWORD_ALIGNED
 };
 
+enum u_vbuf_return_flags {
+   U_VBUF_BUFFERS_UPDATED = 1,
+   U_VBUF_UPLOAD_FLUSHED = 2
+};
+
 
 struct u_vbuf_mgr *
 u_vbuf_mgr_create(struct pipe_context *pipe,
@@ -105,10 +110,9 @@ void u_vbuf_mgr_set_vertex_buffers(struct u_vbuf_mgr *mgr,
                                    unsigned count,
                                    const struct pipe_vertex_buffer *bufs);
 
-void u_vbuf_mgr_draw_begin(struct u_vbuf_mgr *mgr,
-                           const struct pipe_draw_info *info,
-                           boolean *buffers_updated,
-                           boolean *uploader_flushed);
+enum u_vbuf_return_flags
+u_vbuf_mgr_draw_begin(struct u_vbuf_mgr *mgr,
+                      const struct pipe_draw_info *info);
 
 void u_vbuf_mgr_draw_end(struct u_vbuf_mgr *mgr);