From 4bf0ebdd4fc8dbcab9333ff8805af35a91e6848b Mon Sep 17 00:00:00 2001
From: Marek Olšák <maraeo@gmail.com>
Date: Fri, 1 Mar 2013 16:31:49 +0100
Subject: r600g: use a single env var R600_DEBUG, disable bytecode dumping

Only the disassembler is used to dump shaders. Here's a few examples
how to use R600_DEBUG.

Log compute info:
  R600_DEBUG=compute

Dump all shaders:
  R600_DEBUG=fs,vs,gs,ps,cs

Dump pixel shaders only:
  R600_DEBUG=ps

Disable Hyper-Z:
  R600_DEBUG=nohyperz

Disable the LLVM backend:
  R600_DEBUG=nollvm

Or use any combination of the above, or print all options:
  R600_DEBUG=help

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
---
 src/gallium/drivers/r600/compute_memory_pool.c     | 30 +++++------
 src/gallium/drivers/r600/evergreen_compute.c       | 36 ++++++-------
 src/gallium/drivers/r600/evergreen_compute.h       | 20 +++----
 .../drivers/r600/evergreen_compute_internal.c      |  2 +-
 src/gallium/drivers/r600/r600_asm.c                | 11 +---
 src/gallium/drivers/r600/r600_isa.c                |  2 +-
 src/gallium/drivers/r600/r600_pipe.c               | 32 ++++++++---
 src/gallium/drivers/r600/r600_pipe.h               | 15 +++++-
 src/gallium/drivers/r600/r600_shader.c             | 62 +++++++++++++---------
 src/gallium/drivers/r600/r600_texture.c            |  7 ++-
 10 files changed, 122 insertions(+), 95 deletions(-)

(limited to 'src')

diff --git a/src/gallium/drivers/r600/compute_memory_pool.c b/src/gallium/drivers/r600/compute_memory_pool.c
index 8896909bf8a..ee1954e0818 100644
--- a/src/gallium/drivers/r600/compute_memory_pool.c
+++ b/src/gallium/drivers/r600/compute_memory_pool.c
@@ -52,7 +52,7 @@ struct compute_memory_pool* compute_memory_pool_new(
 	struct compute_memory_pool* pool = (struct compute_memory_pool*)
 				CALLOC(sizeof(struct compute_memory_pool), 1);
 
-	COMPUTE_DBG("* compute_memory_pool_new()\n");
+	COMPUTE_DBG(rscreen, "* compute_memory_pool_new()\n");
 
 	pool->screen = rscreen;
 	return pool;
@@ -62,7 +62,7 @@ static void compute_memory_pool_init(struct compute_memory_pool * pool,
 	unsigned initial_size_in_dw)
 {
 
-	COMPUTE_DBG("* compute_memory_pool_init() initial_size_in_dw = %ld\n",
+	COMPUTE_DBG(pool->screen, "* compute_memory_pool_init() initial_size_in_dw = %ld\n",
 		initial_size_in_dw);
 
 	pool->shadow = (uint32_t*)CALLOC(initial_size_in_dw, 4);
@@ -77,7 +77,7 @@ static void compute_memory_pool_init(struct compute_memory_pool * pool,
  */
 void compute_memory_pool_delete(struct compute_memory_pool* pool)
 {
-	COMPUTE_DBG("* compute_memory_pool_delete()\n");
+	COMPUTE_DBG(pool->screen, "* compute_memory_pool_delete()\n");
 	free(pool->shadow);
 	if (pool->bo) {
 		pool->screen->screen.resource_destroy((struct pipe_screen *)
@@ -100,7 +100,7 @@ int64_t compute_memory_prealloc_chunk(
 
 	assert(size_in_dw <= pool->size_in_dw);
 
-	COMPUTE_DBG("* compute_memory_prealloc_chunk() size_in_dw = %ld\n",
+	COMPUTE_DBG(pool->screen, "* compute_memory_prealloc_chunk() size_in_dw = %ld\n",
 		size_in_dw);
 
 	for (item = pool->item_list; item; item = item->next) {
@@ -130,7 +130,7 @@ struct compute_memory_item* compute_memory_postalloc_chunk(
 {
 	struct compute_memory_item* item;
 
-	COMPUTE_DBG("* compute_memory_postalloc_chunck() start_in_dw = %ld\n",
+	COMPUTE_DBG(pool->screen, "* compute_memory_postalloc_chunck() start_in_dw = %ld\n",
 		start_in_dw);
 
 	/* Check if we can insert it in the front of the list */
@@ -162,7 +162,7 @@ struct compute_memory_item* compute_memory_postalloc_chunk(
 void compute_memory_grow_pool(struct compute_memory_pool* pool,
 	struct pipe_context * pipe, int new_size_in_dw)
 {
-	COMPUTE_DBG("* compute_memory_grow_pool() new_size_in_dw = %d\n",
+	COMPUTE_DBG(pool->screen, "* compute_memory_grow_pool() new_size_in_dw = %d\n",
 		new_size_in_dw);
 
 	assert(new_size_in_dw >= pool->size_in_dw);
@@ -172,7 +172,7 @@ void compute_memory_grow_pool(struct compute_memory_pool* pool,
 	} else {
 		new_size_in_dw += 1024 - (new_size_in_dw % 1024);
 
-		COMPUTE_DBG("  Aligned size = %d\n", new_size_in_dw);
+		COMPUTE_DBG(pool->screen, "  Aligned size = %d\n", new_size_in_dw);
 
 		compute_memory_shadow(pool, pipe, 1);
 		pool->shadow = realloc(pool->shadow, new_size_in_dw*4);
@@ -195,7 +195,7 @@ void compute_memory_shadow(struct compute_memory_pool* pool,
 {
 	struct compute_memory_item chunk;
 
-	COMPUTE_DBG("* compute_memory_shadow() device_to_host = %d\n",
+	COMPUTE_DBG(pool->screen, "* compute_memory_shadow() device_to_host = %d\n",
 		device_to_host);
 
 	chunk.id = 0;
@@ -220,10 +220,10 @@ void compute_memory_finalize_pending(struct compute_memory_pool* pool,
 
 	int64_t start_in_dw = 0;
 
-	COMPUTE_DBG("* compute_memory_finalize_pending()\n");
+	COMPUTE_DBG(pool->screen, "* compute_memory_finalize_pending()\n");
 
 	for (item = pool->item_list; item; item = item->next) {
-		COMPUTE_DBG("  + list: offset = %i id = %i size = %i "
+		COMPUTE_DBG(pool->screen, "  + list: offset = %i id = %i size = %i "
 			"(%i bytes)\n",item->start_in_dw, item->id,
 			item->size_in_dw, item->size_in_dw * 4);
 	}
@@ -316,7 +316,7 @@ void compute_memory_finalize_pending(struct compute_memory_pool* pool,
 						pool->size_in_dw + need);
 			}
 		}
-		COMPUTE_DBG("  + Found space for Item %p id = %u "
+		COMPUTE_DBG(pool->screen, "  + Found space for Item %p id = %u "
 			"start_in_dw = %u (%u bytes) size_in_dw = %u (%u bytes)\n",
 			item, item->id, start_in_dw, start_in_dw * 4,
 			item->size_in_dw, item->size_in_dw * 4);
@@ -362,7 +362,7 @@ void compute_memory_free(struct compute_memory_pool* pool, int64_t id)
 {
 	struct compute_memory_item *item, *next;
 
-	COMPUTE_DBG("* compute_memory_free() id + %ld \n", id);
+	COMPUTE_DBG(pool->screen, "* compute_memory_free() id + %ld \n", id);
 
 	for (item = pool->item_list; item; item = next) {
 		next = item->next;
@@ -400,7 +400,7 @@ struct compute_memory_item* compute_memory_alloc(
 {
 	struct compute_memory_item *new_item = NULL, *last_item = NULL;
 
-	COMPUTE_DBG("* compute_memory_alloc() size_in_dw = %ld (%ld bytes)\n",
+	COMPUTE_DBG(pool->screen, "* compute_memory_alloc() size_in_dw = %ld (%ld bytes)\n",
 			size_in_dw, 4 * size_in_dw);
 
 	new_item = (struct compute_memory_item *)
@@ -421,7 +421,7 @@ struct compute_memory_item* compute_memory_alloc(
 		pool->item_list = new_item;
 	}
 
-	COMPUTE_DBG("  + Adding item %p id = %u size = %u (%u bytes)\n",
+	COMPUTE_DBG(pool->screen, "  + Adding item %p id = %u size = %u (%u bytes)\n",
 			new_item, new_item->id, new_item->size_in_dw,
 			new_item->size_in_dw * 4);
 	return new_item;
@@ -448,7 +448,7 @@ void compute_memory_transfer(
 
 	assert(gart);
 
-	COMPUTE_DBG("* compute_memory_transfer() device_to_host = %d, "
+	COMPUTE_DBG(pool->screen, "* compute_memory_transfer() device_to_host = %d, "
 		"offset_in_chunk = %d, size = %d\n", device_to_host,
 		offset_in_chunk, size);
 
diff --git a/src/gallium/drivers/r600/evergreen_compute.c b/src/gallium/drivers/r600/evergreen_compute.c
index 8b3104e40a1..796fcf1951f 100644
--- a/src/gallium/drivers/r600/evergreen_compute.c
+++ b/src/gallium/drivers/r600/evergreen_compute.c
@@ -127,7 +127,7 @@ void *evergreen_create_compute_state(
 	const unsigned char * code;
 	unsigned i;
 
-	COMPUTE_DBG("*** evergreen_create_compute_state\n");
+	COMPUTE_DBG(ctx->screen, "*** evergreen_create_compute_state\n");
 
 	header = cso->prog;
 	code = cso->prog + sizeof(struct pipe_llvm_program_header);
@@ -166,7 +166,7 @@ static void evergreen_bind_compute_state(struct pipe_context *ctx_, void *state)
 {
 	struct r600_context *ctx = (struct r600_context *)ctx_;
 
-	COMPUTE_DBG("*** evergreen_bind_compute_state\n");
+	COMPUTE_DBG(ctx->screen, "*** evergreen_bind_compute_state\n");
 
 	ctx->cs_shader_state.shader = (struct r600_pipe_compute *)state;
 }
@@ -231,7 +231,7 @@ void evergreen_compute_upload_input(
 
 	for (i = 0; i < (kernel_parameters_offset_bytes / 4) +
 					(shader->input_size / 4); i++) {
-		COMPUTE_DBG("input %i : %i\n", i,
+		COMPUTE_DBG(ctx->screen, "input %i : %i\n", i,
 			((unsigned*)num_work_groups_start)[i]);
 	}
 
@@ -272,7 +272,7 @@ static void evergreen_emit_direct_dispatch(
 	num_waves = (block_layout[0] * block_layout[1] * block_layout[2] +
 			wave_divisor - 1) / wave_divisor;
 
-	COMPUTE_DBG("Using %u pipes, there are %u wavefronts per thread block\n",
+	COMPUTE_DBG(rctx->screen, "Using %u pipes, there are %u wavefronts per thread block\n",
 							num_pipes, num_waves);
 
 	/* XXX: Partition the LDS between PS/CS.  By default half (4096 dwords
@@ -378,7 +378,7 @@ static void compute_emit_cs(struct r600_context *ctx, const uint *block_layout,
 	for (i = 0; i < get_compute_resource_num(); i++) {
 		if (resources[i].enabled) {
 			int j;
-			COMPUTE_DBG("resnum: %i, cdw: %i\n", i, cs->cdw);
+			COMPUTE_DBG(ctx->screen, "resnum: %i, cdw: %i\n", i, cs->cdw);
 
 			for (j = 0; j < resources[i].cs_end; j++) {
 				if (resources[i].do_reloc[j]) {
@@ -417,9 +417,9 @@ static void compute_emit_cs(struct r600_context *ctx, const uint *block_layout,
 	r600_flush_emit(ctx);
 
 #if 0
-	COMPUTE_DBG("cdw: %i\n", cs->cdw);
+	COMPUTE_DBG(ctx->screen, "cdw: %i\n", cs->cdw);
 	for (i = 0; i < cs->cdw; i++) {
-		COMPUTE_DBG("%4i : 0x%08X\n", i, ctx->cs->buf[i]);
+		COMPUTE_DBG(ctx->screen, "%4i : 0x%08X\n", i, ctx->cs->buf[i]);
 	}
 #endif
 
@@ -433,11 +433,11 @@ static void compute_emit_cs(struct r600_context *ctx, const uint *block_layout,
 	ctx->pm4_dirty_cdwords = 0;
 	ctx->flags = 0;
 
-	COMPUTE_DBG("shader started\n");
+	COMPUTE_DBG(ctx->screen, "shader started\n");
 
 	ctx->ws->buffer_wait(onebo->buf, 0);
 
-	COMPUTE_DBG("...\n");
+	COMPUTE_DBG(ctx->screen, "...\n");
 }
 
 
@@ -479,7 +479,7 @@ static void evergreen_launch_grid(
 	struct r600_context *ctx = (struct r600_context *)ctx_;
 
 #ifdef HAVE_OPENCL 
-	COMPUTE_DBG("*** evergreen_launch_grid: pc = %u\n", pc);
+	COMPUTE_DBG(ctx->screen, "*** evergreen_launch_grid: pc = %u\n", pc);
 
 	struct r600_pipe_compute *shader = ctx->cs_shader_state.shader;
 	if (!shader->kernels[pc].code_bo) {
@@ -506,7 +506,7 @@ static void evergreen_set_compute_resources(struct pipe_context * ctx_,
 	struct r600_context *ctx = (struct r600_context *)ctx_;
 	struct r600_surface **resources = (struct r600_surface **)surfaces;
 
-	COMPUTE_DBG("*** evergreen_set_compute_resources: start = %u count = %u\n",
+	COMPUTE_DBG(ctx->screen, "*** evergreen_set_compute_resources: start = %u count = %u\n",
 			start, count);
 
 	for (int i = 0; i < count; i++)	{
@@ -579,7 +579,7 @@ static void evergreen_set_global_binding(
 	struct r600_resource_global **buffers =
 		(struct r600_resource_global **)resources;
 
-	COMPUTE_DBG("*** evergreen_set_global_binding first = %u n = %u\n",
+	COMPUTE_DBG(ctx->screen, "*** evergreen_set_global_binding first = %u n = %u\n",
 			first, n);
 
 	if (!resources) {
@@ -817,8 +817,8 @@ struct pipe_resource *r600_compute_global_buffer_create(
 	CALLOC(sizeof(struct r600_resource_global), 1);
 	rscreen = (struct r600_screen*)screen;
 
-	COMPUTE_DBG("*** r600_compute_global_buffer_create\n");
-	COMPUTE_DBG("width = %u array_size = %u\n", templ->width0,
+	COMPUTE_DBG(rscreen, "*** r600_compute_global_buffer_create\n");
+	COMPUTE_DBG(rscreen, "width = %u array_size = %u\n", templ->width0,
 			templ->array_size);
 
 	result->base.b.vtbl = &r600_global_buffer_vtbl;
@@ -877,7 +877,7 @@ void *r600_compute_global_transfer_map(
 
 	assert(resource->target == PIPE_BUFFER);
 
-	COMPUTE_DBG("* r600_compute_global_get_transfer()\n"
+	COMPUTE_DBG(rctx->screen, "* r600_compute_global_get_transfer()\n"
 			"level = %u, usage = %u, box(x = %u, y = %u, z = %u "
 			"width = %u, height = %u, depth = %u)\n", level, usage,
 			box->x, box->y, box->z, box->width, box->height,
@@ -898,7 +898,7 @@ void *r600_compute_global_transfer_map(
 
 	///TODO: do it better, mapping is not possible if the pool is too big
 
-	COMPUTE_DBG("* r600_compute_global_transfer_map()\n");
+	COMPUTE_DBG(rctx->screen, "* r600_compute_global_transfer_map()\n");
 
 	if (!(map = r600_buffer_mmap_sync_with_rings(rctx, buffer->chunk->pool->bo, transfer->usage))) {
 		util_slab_free(&rctx->pool_transfers, transfer);
@@ -907,7 +907,7 @@ void *r600_compute_global_transfer_map(
 
 	*ptransfer = transfer;
 
-	COMPUTE_DBG("Buffer: %p + %u (buffer offset in global memory) "
+	COMPUTE_DBG(rctx->screen, "Buffer: %p + %u (buffer offset in global memory) "
 		"+ %u (box.x)\n", map, buffer->chunk->start_in_dw, transfer->box.x);
 	return ((char*)(map + buffer->chunk->start_in_dw)) + transfer->box.x;
 }
@@ -925,7 +925,7 @@ void r600_compute_global_transfer_unmap(
 	ctx = (struct r600_context *)ctx_;
 	buffer = (struct r600_resource_global*)transfer->resource;
 
-	COMPUTE_DBG("* r600_compute_global_transfer_unmap()\n");
+	COMPUTE_DBG(ctx->screen, "* r600_compute_global_transfer_unmap()\n");
 
 	ctx->ws->buffer_unmap(buffer->chunk->pool->bo->cs_buf);
 	util_slab_free(&ctx->pool_transfers, transfer);
diff --git a/src/gallium/drivers/r600/evergreen_compute.h b/src/gallium/drivers/r600/evergreen_compute.h
index db57520a74c..69e41cc82aa 100644
--- a/src/gallium/drivers/r600/evergreen_compute.h
+++ b/src/gallium/drivers/r600/evergreen_compute.h
@@ -54,20 +54,16 @@ void r600_compute_global_transfer_inline_write( struct pipe_context *, struct pi
                                                 unsigned usage, const struct pipe_box *, const void *data, unsigned stride, unsigned layer_stride);
 
 
-static inline void COMPUTE_DBG(const char *fmt, ...)
+static inline void COMPUTE_DBG(struct r600_screen *rscreen, const char *fmt, ...)
 {
-   static bool check_debug = false, debug = false;
+	if (!(rscreen->debug_flags & DBG_COMPUTE)) {
+		return;
+	}
 
-   if (!check_debug) {
-		debug = debug_get_bool_option("R600_COMPUTE_DEBUG", FALSE);
-   }
-
-   if (debug) {
-      va_list ap;
-      va_start(ap, fmt);
-      _debug_vprintf(fmt, ap);
-      va_end(ap);
-   }
+	va_list ap;
+	va_start(ap, fmt);
+	_debug_vprintf(fmt, ap);
+	va_end(ap);
 }
 
 #endif
diff --git a/src/gallium/drivers/r600/evergreen_compute_internal.c b/src/gallium/drivers/r600/evergreen_compute_internal.c
index 60bb3bfb045..c671478a268 100644
--- a/src/gallium/drivers/r600/evergreen_compute_internal.c
+++ b/src/gallium/drivers/r600/evergreen_compute_internal.c
@@ -281,7 +281,7 @@ void evergreen_set_rat(
 
 	rctx = pipe->ctx;
 
-	COMPUTE_DBG("bind rat: %i \n", id);
+	COMPUTE_DBG(rctx->screen, "bind rat: %i \n", id);
 
 	/* Create the RAT surface */
 	memset(&rat_templ, 0, sizeof(rat_templ));
diff --git a/src/gallium/drivers/r600/r600_asm.c b/src/gallium/drivers/r600/r600_asm.c
index cb3e7a63025..c661af9c205 100644
--- a/src/gallium/drivers/r600/r600_asm.c
+++ b/src/gallium/drivers/r600/r600_asm.c
@@ -2465,7 +2465,6 @@ void *r600_create_vertex_fetch_shader(struct pipe_context *ctx,
 				      const struct pipe_vertex_element *elements)
 {
 	struct r600_context *rctx = (struct r600_context *)ctx;
-	static int dump_shaders = -1;
 	struct r600_bytecode bc;
 	struct r600_bytecode_vtx vtx;
 	const struct util_format_description *desc;
@@ -2571,15 +2570,7 @@ void *r600_create_vertex_fetch_shader(struct pipe_context *ctx,
 		return NULL;
 	}
 
-	if (dump_shaders == -1)
-		dump_shaders = debug_get_num_option("R600_DUMP_SHADERS", 0);
-
-	if (dump_shaders & 1) {
-		fprintf(stderr, "--------------------------------------------------------------\n");
-		r600_bytecode_dump(&bc);
-		fprintf(stderr, "______________________________________________________________\n");
-	}
-	if (dump_shaders & 2) {
+	if (rctx->screen->debug_flags & DBG_FS) {
 		fprintf(stderr, "--------------------------------------------------------------\n");
 		r600_bytecode_disasm(&bc);
 		fprintf(stderr, "______________________________________________________________\n");
diff --git a/src/gallium/drivers/r600/r600_isa.c b/src/gallium/drivers/r600/r600_isa.c
index 20b17594f49..54cdfff8bbd 100644
--- a/src/gallium/drivers/r600/r600_isa.c
+++ b/src/gallium/drivers/r600/r600_isa.c
@@ -37,7 +37,7 @@ int r600_isa_init(struct r600_context *ctx, struct r600_isa *isa) {
 #if defined R600_USE_LLVM || defined HAVE_OPENCL
 	unsigned i, use_llvm;
 
-	use_llvm = debug_get_bool_option("R600_LLVM", TRUE);
+	use_llvm = !(ctx->screen->debug_flags & DBG_NO_LLVM);
 
 	if (!use_llvm)
 		return 0;
diff --git a/src/gallium/drivers/r600/r600_pipe.c b/src/gallium/drivers/r600/r600_pipe.c
index f5bf0cec892..ee297793fe6 100644
--- a/src/gallium/drivers/r600/r600_pipe.c
+++ b/src/gallium/drivers/r600/r600_pipe.c
@@ -29,6 +29,7 @@
 #include <errno.h>
 #include "pipe/p_shader_tokens.h"
 #include "util/u_blitter.h"
+#include "util/u_debug.h"
 #include "util/u_format_s3tc.h"
 #include "util/u_memory.h"
 #include "util/u_simple_shaders.h"
@@ -38,6 +39,27 @@
 #include "vl/vl_video_buffer.h"
 #include "os/os_time.h"
 
+static const struct debug_named_value debug_options[] = {
+	/* logging */
+	{ "texdepth", DBG_TEX_DEPTH, "Print texture depth info" },
+	{ "compute", DBG_COMPUTE, "Print compute info" },
+
+	/* shaders */
+	{ "fs", DBG_FS, "Print fetch shaders" },
+	{ "vs", DBG_VS, "Print vertex shaders" },
+	{ "gs", DBG_GS, "Print geometry shaders" },
+	{ "ps", DBG_PS, "Print pixel shaders" },
+	{ "cs", DBG_CS, "Print compute shaders" },
+
+	/* features */
+	{ "nohyperz", DBG_NO_HYPERZ, "Disable Hyper-Z" },
+#if defined(R600_USE_LLVM)
+	{ "nollvm", DBG_NO_LLVM, "Disable the LLVM shader compiler" },
+#endif
+
+	DEBUG_NAMED_VALUE_END /* must be last */
+};
+
 /*
  * pipe_context
  */
@@ -1071,8 +1093,10 @@ struct pipe_screen *r600_screen_create(struct radeon_winsys *ws)
 	rscreen->ws = ws;
 	ws->query_info(ws, &rscreen->info);
 
+	rscreen->debug_flags = debug_get_flags_option("R600_DEBUG", debug_options, 0);
 	rscreen->family = rscreen->info.family;
 	rscreen->chip_class = rscreen->info.chip_class;
+
 	if (rscreen->family == CHIP_UNKNOWN) {
 		fprintf(stderr, "r600: Unknown chipset 0x%04X\n", rscreen->info.pci_id);
 		FREE(rscreen);
@@ -1164,14 +1188,6 @@ struct pipe_screen *r600_screen_create(struct radeon_winsys *ws)
 	LIST_INITHEAD(&rscreen->fences.blocks);
 	pipe_mutex_init(rscreen->fences.mutex);
 
-	/* Hyperz is very lockup prone any code that touch related part should be
-	 * carefully tested especialy on r6xx/r7xx Development show that some piglit
-	 * case were triggering lockup quickly such as :
-	 * piglit/bin/depthstencil-render-miplevels 1024 d=s=z24_s8
-	 */
-	rscreen->use_hyperz = debug_get_bool_option("R600_HYPERZ", TRUE);
-	rscreen->use_hyperz = rscreen->info.drm_minor >= 26 ? rscreen->use_hyperz : FALSE;
-
 	rscreen->global_pool = compute_memory_pool_new(rscreen);
 
 #if R600_TRACE_CS
diff --git a/src/gallium/drivers/r600/r600_pipe.h b/src/gallium/drivers/r600/r600_pipe.h
index c178285e8e5..3eb2968119a 100644
--- a/src/gallium/drivers/r600/r600_pipe.h
+++ b/src/gallium/drivers/r600/r600_pipe.h
@@ -223,9 +223,23 @@ typedef boolean (*r600g_dma_blit_t)(struct pipe_context *ctx,
 				unsigned src_level,
 				const struct pipe_box *src_box);
 
+/* logging */
+#define DBG_TEX_DEPTH		(1 << 0)
+#define DBG_COMPUTE		(1 << 1)
+/* shaders */
+#define DBG_FS			(1 << 8)
+#define DBG_VS			(1 << 9)
+#define DBG_GS			(1 << 10)
+#define DBG_PS			(1 << 11)
+#define DBG_CS			(1 << 12)
+/* features */
+#define DBG_NO_HYPERZ		(1 << 16)
+#define DBG_NO_LLVM		(1 << 17)
+
 struct r600_screen {
 	struct pipe_screen		screen;
 	struct radeon_winsys		*ws;
+	unsigned			debug_flags;
 	unsigned			family;
 	enum chip_class			chip_class;
 	struct radeon_info		info;
@@ -233,7 +247,6 @@ struct r600_screen {
 	bool				has_msaa;
 	bool				has_cp_dma;
 	enum r600_msaa_texture_mode	msaa_texture_support;
-	bool				use_hyperz;
 	struct r600_tiling_info		tiling_info;
 	struct r600_pipe_fences		fences;
 
diff --git a/src/gallium/drivers/r600/r600_shader.c b/src/gallium/drivers/r600/r600_shader.c
index 86424634d0f..949191a952a 100644
--- a/src/gallium/drivers/r600/r600_shader.c
+++ b/src/gallium/drivers/r600/r600_shader.c
@@ -108,6 +108,33 @@ static int r600_shader_from_tgsi(struct r600_screen *rscreen,
 				 struct r600_pipe_shader *pipeshader,
 				 struct r600_shader_key key);
 
+static unsigned tgsi_get_processor_type(const struct tgsi_token *tokens)
+{
+	struct tgsi_parse_context parse;
+
+	if (tgsi_parse_init( &parse, tokens ) != TGSI_PARSE_OK) {
+		debug_printf("tgsi_parse_init() failed in %s:%i!\n", __func__, __LINE__);
+		return ~0;
+	}
+	return parse.FullHeader.Processor.Processor;
+}
+
+static bool r600_can_dump_shader(struct r600_screen *rscreen, unsigned processor_type)
+{
+	switch (processor_type) {
+	case TGSI_PROCESSOR_VERTEX:
+		return (rscreen->debug_flags & DBG_VS) != 0;
+	case TGSI_PROCESSOR_GEOMETRY:
+		return (rscreen->debug_flags & DBG_GS) != 0;
+	case TGSI_PROCESSOR_FRAGMENT:
+		return (rscreen->debug_flags & DBG_PS) != 0;
+	case TGSI_PROCESSOR_COMPUTE:
+		return (rscreen->debug_flags & DBG_CS) != 0;
+	default:
+		return false;
+	}
+}
+
 static void r600_dump_streamout(struct pipe_stream_output_info *so)
 {
 	unsigned i;
@@ -132,19 +159,14 @@ int r600_pipe_shader_create(struct pipe_context *ctx,
 			    struct r600_pipe_shader *shader,
 			    struct r600_shader_key key)
 {
-	static int dump_shaders = -1;
 	struct r600_context *rctx = (struct r600_context *)ctx;
 	struct r600_pipe_shader_selector *sel = shader->selector;
 	int r;
+	bool dump = r600_can_dump_shader(rctx->screen, tgsi_get_processor_type(sel->tokens));
 
 	shader->shader.bc.isa = rctx->isa;
 
-	/* Would like some magic "get_bool_option_once" routine.
-	*/
-	if (dump_shaders == -1)
-		dump_shaders = debug_get_num_option("R600_DUMP_SHADERS", 0);
-
-	if (dump_shaders) {
+	if (dump) {
 		fprintf(stderr, "--------------------------------------------------------------\n");
 		tgsi_dump(sel->tokens, 0);
 
@@ -162,12 +184,7 @@ int r600_pipe_shader_create(struct pipe_context *ctx,
 		R600_ERR("building bytecode failed !\n");
 		return r;
 	}
-	if (dump_shaders & 1) {
-		fprintf(stderr, "--------------------------------------------------------------\n");
-		r600_bytecode_dump(&shader->shader.bc);
-		fprintf(stderr, "______________________________________________________________\n");
-	}
-	if (dump_shaders & 2) {
+	if (dump) {
 		fprintf(stderr, "--------------------------------------------------------------\n");
 		r600_bytecode_disasm(&shader->shader.bc);
 		fprintf(stderr, "______________________________________________________________\n");
@@ -260,11 +277,7 @@ int r600_compute_shader_create(struct pipe_context * ctx,
 	unsigned char * bytes;
 	unsigned byte_count;
 	struct r600_shader_ctx shader_ctx;
-	unsigned dump = 0;
-
-	if (debug_get_bool_option("R600_DUMP_SHADERS", FALSE)) {
-		dump = 1;
-	}
+	bool dump = (r600_ctx->screen->debug_flags & DBG_CS) != 0;
 
 	r600_llvm_compile(mod, &bytes, &byte_count, r600_ctx->family , dump);
 	shader_ctx.bc = bytecode;
@@ -278,7 +291,7 @@ int r600_compute_shader_create(struct pipe_context * ctx,
 	}
 	r600_bytecode_build(shader_ctx.bc);
 	if (dump) {
-		r600_bytecode_dump(shader_ctx.bc);
+		r600_bytecode_disasm(shader_ctx.bc);
 	}
 	free(bytes);
 	return 1;
@@ -1240,7 +1253,7 @@ static int r600_shader_from_tgsi(struct r600_screen *rscreen,
 	unsigned inst_byte_count = 0;
 
 #ifdef R600_USE_LLVM
-	use_llvm = debug_get_bool_option("R600_LLVM", TRUE);
+	use_llvm = !(ctx->screen->debug_flags & DBG_NO_LLVM);
 #endif
 	ctx.bc = &shader->bc;
 	ctx.shader = shader;
@@ -1416,7 +1429,8 @@ static int r600_shader_from_tgsi(struct r600_screen *rscreen,
 	if (use_llvm) {
 		struct radeon_llvm_context radeon_llvm_ctx;
 		LLVMModuleRef mod;
-		unsigned dump = 0;
+		bool dump = r600_can_dump_shader(rscreen, ctx.type);
+
 		memset(&radeon_llvm_ctx, 0, sizeof(radeon_llvm_ctx));
 		radeon_llvm_ctx.type = ctx.type;
 		radeon_llvm_ctx.two_side = shader->two_side;
@@ -1430,11 +1444,9 @@ static int r600_shader_from_tgsi(struct r600_screen *rscreen,
 		radeon_llvm_ctx.clip_vertex = ctx.cv_output;
 		radeon_llvm_ctx.alpha_to_one = key.alpha_to_one;
 		mod = r600_tgsi_llvm(&radeon_llvm_ctx, tokens);
-		if (debug_get_bool_option("R600_DUMP_SHADERS", FALSE)) {
-			dump = 1;
-		}
+
 		if (r600_llvm_compile(mod, &inst_bytes, &inst_byte_count,
-							rscreen->family, dump)) {
+				      rscreen->family, dump)) {
 			FREE(inst_bytes);
 			radeon_llvm_dispose(&radeon_llvm_ctx);
 			use_llvm = 0;
diff --git a/src/gallium/drivers/r600/r600_texture.c b/src/gallium/drivers/r600/r600_texture.c
index 1c071cb2b74..484045ed4f0 100644
--- a/src/gallium/drivers/r600/r600_texture.c
+++ b/src/gallium/drivers/r600/r600_texture.c
@@ -400,8 +400,6 @@ static void r600_texture_allocate_cmask(struct r600_screen *rscreen,
 #endif
 }
 
-DEBUG_GET_ONCE_BOOL_OPTION(print_texdepth, "R600_PRINT_TEXDEPTH", FALSE);
-
 static struct r600_texture *
 r600_texture_create_object(struct pipe_screen *screen,
 			   const struct pipe_resource *base,
@@ -457,7 +455,8 @@ r600_texture_create_object(struct pipe_screen *screen,
 	rtex->htile = NULL;
 	if (!(base->flags & (R600_RESOURCE_FLAG_TRANSFER | R600_RESOURCE_FLAG_FLUSHED_DEPTH)) &&
 	    util_format_is_depth_or_stencil(base->format) &&
-	    rscreen->use_hyperz &&
+	    rscreen->info.drm_minor >= 26 &&
+	    !(rscreen->debug_flags & DBG_NO_HYPERZ) &&
 	    base->target == PIPE_TEXTURE_2D &&
 	    rtex->surface.level[0].nblk_x >= 32 &&
 	    rtex->surface.level[0].nblk_y >= 32) {
@@ -511,7 +510,7 @@ r600_texture_create_object(struct pipe_screen *screen,
 		rscreen->ws->buffer_unmap(resource->cs_buf);
 	}
 
-	if (debug_get_option_print_texdepth() && rtex->is_depth && rtex->non_disp_tiling) {
+	if (rscreen->debug_flags & DBG_TEX_DEPTH && rtex->is_depth && rtex->non_disp_tiling) {
 		printf("Texture: npix_x=%u, npix_y=%u, npix_z=%u, blk_w=%u, "
 		       "blk_h=%u, blk_d=%u, array_size=%u, last_level=%u, "
 		       "bpe=%u, nsamples=%u, flags=%u\n",
-- 
cgit v1.2.3