gallium/radeon: remove old CS tracing

Cons:
- it was only integrated in r600g
- it doesn't work with GPUVM
- it records buffer contents at the end of IBs instead of at the beginning,
  so the replay isn't exact
- it lacks an IB parser and user-friendliness

A better solution is apitrace in combination with gallium/ddebug, which
has a complete IB parser and can pinpoint hanging CP packets.

Reviewed-by: Nicolai Hähnle <[email protected]>

20 files changed, 23 insertions, 476 deletions

diff --git a/src/gallium/drivers/r300/r300_context.c b/src/gallium/drivers/r300/r300_context.c
index 6fa892089ec..d100a9df55b 100644
--- a/src/gallium/drivers/r300/r300_context.c
+++ b/src/gallium/drivers/r300/r300_context.c
@@ -385,7 +385,7 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen,
     if (!r300->ctx)
         goto fail;
 
-    r300->cs = rws->cs_create(r300->ctx, RING_GFX, r300_flush_callback, r300, NULL);
+    r300->cs = rws->cs_create(r300->ctx, RING_GFX, r300_flush_callback, r300);
     if (r300->cs == NULL)
         goto fail;
 
diff --git a/src/gallium/drivers/r300/r300_flush.c b/src/gallium/drivers/r300/r300_flush.c
index 7a75b43a53e..63182cba2b2 100644
--- a/src/gallium/drivers/r300/r300_flush.c
+++ b/src/gallium/drivers/r300/r300_flush.c
@@ -53,7 +53,7 @@ static void r300_flush_and_cleanup(struct r300_context *r300, unsigned flags,
     }
 
     r300->flush_counter++;
-    r300->rws->cs_flush(r300->cs, flags, fence, 0);
+    r300->rws->cs_flush(r300->cs, flags, fence);
     r300->dirty_hw = 0;
 
     /* New kitchen sink, baby. */
@@ -88,11 +88,11 @@ void r300_flush(struct pipe_context *pipe,
              * and we cannot emit an empty CS. Let's write to some reg. */
             CS_LOCALS(r300);
             OUT_CS_REG(RB3D_COLOR_CHANNEL_MASK, 0);
-            r300->rws->cs_flush(r300->cs, flags, fence, 0);
+            r300->rws->cs_flush(r300->cs, flags, fence);
         } else {
             /* Even if hw is not dirty, we should at least reset the CS in case
              * the space checking failed for the first draw operation. */
-            r300->rws->cs_flush(r300->cs, flags, NULL, 0);
+            r300->rws->cs_flush(r300->cs, flags, NULL);
         }
     }
 
diff --git a/src/gallium/drivers/r600/r600_hw_context.c b/src/gallium/drivers/r600/r600_hw_context.c
index 4951297df42..7a6f957945b 100644
--- a/src/gallium/drivers/r600/r600_hw_context.c
+++ b/src/gallium/drivers/r600/r600_hw_context.c
@@ -57,18 +57,11 @@ void r600_need_cs_space(struct r600_context *ctx, unsigned num_dw,
 
 		/* The number of dwords all the dirty states would take. */
 		mask = ctx->dirty_atoms;
-		while (mask != 0) {
+		while (mask != 0)
 			num_dw += ctx->atoms[u_bit_scan64(&mask)]->num_dw;
-			if (ctx->screen->b.trace_bo) {
-				num_dw += R600_TRACE_CS_DWORDS;
-			}
-		}
 
 		/* The upper-bound of how much space a draw command would take. */
 		num_dw += R600_MAX_FLUSH_CS_DWORDS + R600_MAX_DRAW_CS_DWORDS;
-		if (ctx->screen->b.trace_bo) {
-			num_dw += R600_TRACE_CS_DWORDS;
-		}
 	}
 
 	/* Count in queries_suspend. */
@@ -273,7 +266,7 @@ void r600_context_gfx_flush(void *context, unsigned flags,
 	flags |= RADEON_FLUSH_KEEP_TILING_FLAGS;
 
 	/* Flush the CS. */
-	ctx->b.ws->cs_flush(cs, flags, fence, ctx->screen->b.cs_count++);
+	ctx->b.ws->cs_flush(cs, flags, fence);
 
 	r600_begin_new_cs(ctx);
 }
diff --git a/src/gallium/drivers/r600/r600_pipe.c b/src/gallium/drivers/r600/r600_pipe.c
index 7018088d204..88c500a162a 100644
--- a/src/gallium/drivers/r600/r600_pipe.c
+++ b/src/gallium/drivers/r600/r600_pipe.c
@@ -187,9 +187,7 @@ static struct pipe_context *r600_create_context(struct pipe_screen *screen,
 	}
 
 	rctx->b.gfx.cs = ws->cs_create(rctx->b.ctx, RING_GFX,
-				       r600_context_gfx_flush, rctx,
-				       rscreen->b.trace_bo ?
-					       rscreen->b.trace_bo->buf : NULL);
+				       r600_context_gfx_flush, rctx);
 	rctx->b.gfx.flush = r600_context_gfx_flush;
 
 	rctx->allocator_fetch_shader = u_suballocator_create(&rctx->b.b, 64 * 1024, 256,
diff --git a/src/gallium/drivers/r600/r600_pipe.h b/src/gallium/drivers/r600/r600_pipe.h
index f8a20398355..72aa64233a9 100644
--- a/src/gallium/drivers/r600/r600_pipe.h
+++ b/src/gallium/drivers/r600/r600_pipe.h
@@ -60,7 +60,6 @@
 /* the number of CS dwords for flushing and drawing */
 #define R600_MAX_FLUSH_CS_DWORDS	16
 #define R600_MAX_DRAW_CS_DWORDS		58
-#define R600_TRACE_CS_DWORDS		7
 
 #define R600_MAX_USER_CONST_BUFFERS 13
 #define R600_MAX_DRIVER_CONST_BUFFERS 3
@@ -571,15 +570,10 @@ static inline void r600_mark_atom_dirty(struct r600_context *rctx,
 	r600_set_atom_dirty(rctx, atom, true);
 }
 
-void r600_trace_emit(struct r600_context *rctx);
-
 static inline void r600_emit_atom(struct r600_context *rctx, struct r600_atom *atom)
 {
 	atom->emit(&rctx->b, atom);
 	r600_set_atom_dirty(rctx, atom, false);
-	if (rctx->screen->b.trace_bo) {
-		r600_trace_emit(rctx);
-	}
 }
 
 static inline void r600_set_cso_state(struct r600_context *rctx,
diff --git a/src/gallium/drivers/r600/r600_state_common.c b/src/gallium/drivers/r600/r600_state_common.c
index 2211e07ceba..df41d3f028d 100644
--- a/src/gallium/drivers/r600/r600_state_common.c
+++ b/src/gallium/drivers/r600/r600_state_common.c
@@ -2029,10 +2029,6 @@ static void r600_draw_vbo(struct pipe_context *ctx, const struct pipe_draw_info
 		cs->buf[cs->cdw++] = EVENT_TYPE(EVENT_TYPE_SQ_NON_EVENT);
 	}
 
-	if (rctx->screen->b.trace_bo) {
-		r600_trace_emit(rctx);
-	}
-
 	/* Set the depth buffer as dirty. */
 	if (rctx->framebuffer.state.zsbuf) {
 		struct pipe_surface *surf = rctx->framebuffer.state.zsbuf;
@@ -2927,22 +2923,3 @@ void r600_init_common_state_functions(struct r600_context *rctx)
 	rctx->b.set_occlusion_query_state = r600_set_occlusion_query_state;
 	rctx->b.need_gfx_cs_space = r600_need_gfx_cs_space;
 }
-
-void r600_trace_emit(struct r600_context *rctx)
-{
-	struct r600_screen *rscreen = rctx->screen;
-	struct radeon_winsys_cs *cs = rctx->b.gfx.cs;
-	uint64_t va;
-	uint32_t reloc;
-
-	va = rscreen->b.trace_bo->gpu_address;
-	reloc = radeon_add_to_buffer_list(&rctx->b, &rctx->b.gfx, rscreen->b.trace_bo,
-				      RADEON_USAGE_READWRITE, RADEON_PRIO_TRACE);
-	radeon_emit(cs, PKT3(PKT3_MEM_WRITE, 3, 0));
-	radeon_emit(cs, va & 0xFFFFFFFFUL);
-	radeon_emit(cs, (va >> 32UL) & 0xFFUL);
-	radeon_emit(cs, cs->cdw);
-	radeon_emit(cs, rscreen->b.cs_count);
-	radeon_emit(cs, PKT3(PKT3_NOP, 0, 0));
-	radeon_emit(cs, reloc);
-}
diff --git a/src/gallium/drivers/radeon/r600_pipe_common.c b/src/gallium/drivers/radeon/r600_pipe_common.c
index ea028272ccd..eed9d83ee49 100644
--- a/src/gallium/drivers/radeon/r600_pipe_common.c
+++ b/src/gallium/drivers/radeon/r600_pipe_common.c
@@ -229,7 +229,7 @@ static void r600_flush_dma_ring(void *ctx, unsigned flags,
 	struct radeon_winsys_cs *cs = rctx->dma.cs;
 
 	if (cs->cdw)
-		rctx->ws->cs_flush(cs, flags, &rctx->last_sdma_fence, 0);
+		rctx->ws->cs_flush(cs, flags, &rctx->last_sdma_fence);
 	if (fence)
 		rctx->ws->fence_reference(fence, rctx->last_sdma_fence);
 }
@@ -318,7 +318,7 @@ bool r600_common_context_init(struct r600_common_context *rctx,
 	if (rscreen->info.has_sdma && !(rscreen->debug_flags & DBG_NO_ASYNC_DMA)) {
 		rctx->dma.cs = rctx->ws->cs_create(rctx->ctx, RING_DMA,
 						   r600_flush_dma_ring,
-						   rctx, NULL);
+						   rctx);
 		rctx->dma.flush = r600_flush_dma_ring;
 	}
 
@@ -379,7 +379,6 @@ static const struct debug_named_value common_debug_options[] = {
 	{ "tex", DBG_TEX, "Print texture info" },
 	{ "compute", DBG_COMPUTE, "Print compute info" },
 	{ "vm", DBG_VM, "Print virtual addresses when creating resources" },
-	{ "trace_cs", DBG_TRACE_CS, "Trace cs and write rlockup_<csid>.c file with faulty cs" },
 	{ "info", DBG_INFO, "Print driver information" },
 
 	/* shaders */
@@ -893,19 +892,6 @@ bool r600_common_screen_init(struct r600_common_screen *rscreen,
 	pipe_mutex_init(rscreen->aux_context_lock);
 	pipe_mutex_init(rscreen->gpu_load_mutex);
 
-	if (((rscreen->info.drm_major == 2 && rscreen->info.drm_minor >= 28) ||
-	     rscreen->info.drm_major == 3) &&
-	    (rscreen->debug_flags & DBG_TRACE_CS)) {
-		rscreen->trace_bo = (struct r600_resource*)pipe_buffer_create(&rscreen->b,
-										PIPE_BIND_CUSTOM,
-										PIPE_USAGE_STAGING,
-										4096);
-		if (rscreen->trace_bo) {
-			rscreen->trace_ptr = rscreen->ws->buffer_map(rscreen->trace_bo->buf, NULL,
-									PIPE_TRANSFER_UNSYNCHRONIZED);
-		}
-	}
-
 	if (rscreen->debug_flags & DBG_INFO) {
 		printf("pci_id = 0x%x\n", rscreen->info.pci_id);
 		printf("family = %i (%s)\n", rscreen->info.family,
@@ -951,9 +937,6 @@ void r600_destroy_common_screen(struct r600_common_screen *rscreen)
 	pipe_mutex_destroy(rscreen->aux_context_lock);
 	rscreen->aux_context->destroy(rscreen->aux_context);
 
-	if (rscreen->trace_bo)
-		pipe_resource_reference((struct pipe_resource**)&rscreen->trace_bo, NULL);
-
 	rscreen->ws->destroy(rscreen->ws);
 	FREE(rscreen);
 }
diff --git a/src/gallium/drivers/radeon/r600_pipe_common.h b/src/gallium/drivers/radeon/r600_pipe_common.h
index cf8dcf7ea88..a9de71a8734 100644
--- a/src/gallium/drivers/radeon/r600_pipe_common.h
+++ b/src/gallium/drivers/radeon/r600_pipe_common.h
@@ -61,7 +61,7 @@
 /* gap - reuse */
 #define DBG_COMPUTE		(1 << 2)
 #define DBG_VM			(1 << 3)
-#define DBG_TRACE_CS		(1 << 4)
+/* gap - reuse */
 /* shader logging */
 #define DBG_FS			(1 << 5)
 #define DBG_VS			(1 << 6)
@@ -303,10 +303,6 @@ struct r600_common_screen {
 	struct pipe_context		*aux_context;
 	pipe_mutex			aux_context_lock;
 
-	struct r600_resource		*trace_bo;
-	uint32_t			*trace_ptr;
-	unsigned			cs_count;
-
 	/* This must be in the screen, because UE4 uses one context for
 	 * compilation and another one for rendering.
 	 */
diff --git a/src/gallium/drivers/radeon/radeon_uvd.c b/src/gallium/drivers/radeon/radeon_uvd.c
index b8efc58eaab..233f46091a4 100644
--- a/src/gallium/drivers/radeon/radeon_uvd.c
+++ b/src/gallium/drivers/radeon/radeon_uvd.c
@@ -92,7 +92,7 @@ struct ruvd_decoder {
 /* flush IB to the hardware */
 static void flush(struct ruvd_decoder *dec)
 {
-	dec->ws->cs_flush(dec->cs, RADEON_FLUSH_ASYNC, NULL, 0);
+	dec->ws->cs_flush(dec->cs, RADEON_FLUSH_ASYNC, NULL);
 }
 
 /* add a new set register command to the IB */
@@ -1142,7 +1142,7 @@ struct pipe_video_codec *ruvd_create_decoder(struct pipe_context *context,
 	dec->stream_handle = rvid_alloc_stream_handle();
 	dec->screen = context->screen;
 	dec->ws = ws;
-	dec->cs = ws->cs_create(rctx->ctx, RING_UVD, NULL, NULL, NULL);
+	dec->cs = ws->cs_create(rctx->ctx, RING_UVD, NULL, NULL);
 	if (!dec->cs) {
 		RVID_ERR("Can't get command submission context.\n");
 		goto error;
diff --git a/src/gallium/drivers/radeon/radeon_vce.c b/src/gallium/drivers/radeon/radeon_vce.c
index 087d9422c04..2ab74e9eb6c 100644
--- a/src/gallium/drivers/radeon/radeon_vce.c
+++ b/src/gallium/drivers/radeon/radeon_vce.c
@@ -56,7 +56,7 @@
  */
 static void flush(struct rvce_encoder *enc)
 {
-	enc->ws->cs_flush(enc->cs, RADEON_FLUSH_ASYNC, NULL, 0);
+	enc->ws->cs_flush(enc->cs, RADEON_FLUSH_ASYNC, NULL);
 	enc->task_info_idx = 0;
 	enc->bs_idx = 0;
 }
@@ -429,7 +429,7 @@ struct pipe_video_codec *rvce_create_encoder(struct pipe_context *context,
 
 	enc->screen = context->screen;
 	enc->ws = ws;
-	enc->cs = ws->cs_create(rctx->ctx, RING_VCE, rvce_cs_flush, enc, NULL);
+	enc->cs = ws->cs_create(rctx->ctx, RING_VCE, rvce_cs_flush, enc);
 	if (!enc->cs) {
 		RVID_ERR("Can't get command submission context.\n");
 		goto error;
diff --git a/src/gallium/drivers/radeon/radeon_winsys.h b/src/gallium/drivers/radeon/radeon_winsys.h
index daa15db2812..d35e963133e 100644
--- a/src/gallium/drivers/radeon/radeon_winsys.h
+++ b/src/gallium/drivers/radeon/radeon_winsys.h
@@ -593,14 +593,12 @@ struct radeon_winsys {
      * \param ring_type The ring type (GFX, DMA, UVD)
      * \param flush     Flush callback function associated with the command stream.
      * \param user      User pointer that will be passed to the flush callback.
-     * \param trace_buf Trace buffer when tracing is enabled
      */
     struct radeon_winsys_cs *(*cs_create)(struct radeon_winsys_ctx *ctx,
                                           enum ring_type ring_type,
                                           void (*flush)(void *ctx, unsigned flags,
 							struct pipe_fence_handle **fence),
-                                          void *flush_ctx,
-                                          struct pb_buffer *trace_buf);
+                                          void *flush_ctx);
 
     /**
      * Destroy a command stream.
@@ -673,12 +671,10 @@ struct radeon_winsys {
      * \param flags,      RADEON_FLUSH_ASYNC or 0.
      * \param fence       Pointer to a fence. If non-NULL, a fence is inserted
      *                    after the CS and is returned through this parameter.
-     * \param cs_trace_id A unique identifier of the cs, used for tracing.
      */
     void (*cs_flush)(struct radeon_winsys_cs *cs,
                      unsigned flags,
-                     struct pipe_fence_handle **fence,
-                     uint32_t cs_trace_id);
+                     struct pipe_fence_handle **fence);
 
     /**
      * Return TRUE if a buffer is referenced by a command stream.
diff --git a/src/gallium/drivers/radeonsi/si_hw_context.c b/src/gallium/drivers/radeonsi/si_hw_context.c
index b5a4034cc12..8c900a4ecb6 100644
--- a/src/gallium/drivers/radeonsi/si_hw_context.c
+++ b/src/gallium/drivers/radeonsi/si_hw_context.c
@@ -118,8 +118,7 @@ void si_context_gfx_flush(void *context, unsigned flags,
 	}
 
 	/* Flush the CS. */
-	ws->cs_flush(cs, flags, &ctx->last_gfx_fence,
-		     ctx->screen->b.cs_count++);
+	ws->cs_flush(cs, flags, &ctx->last_gfx_fence);
 
 	if (fence)
 		ws->fence_reference(fence, ctx->last_gfx_fence);
diff --git a/src/gallium/drivers/radeonsi/si_pipe.c b/src/gallium/drivers/radeonsi/si_pipe.c
index 8b50a49cba0..042cfc764fd 100644
--- a/src/gallium/drivers/radeonsi/si_pipe.c
+++ b/src/gallium/drivers/radeonsi/si_pipe.c
@@ -140,9 +140,8 @@ static struct pipe_context *si_create_context(struct pipe_screen *screen,
 		sctx->b.b.create_video_buffer = vl_video_buffer_create;
 	}
 
-	sctx->b.gfx.cs = ws->cs_create(sctx->b.ctx, RING_GFX, si_context_gfx_flush,
-				       sctx, sscreen->b.trace_bo ?
-					       sscreen->b.trace_bo->buf : NULL);
+	sctx->b.gfx.cs = ws->cs_create(sctx->b.ctx, RING_GFX,
+				       si_context_gfx_flush, sctx);
 	sctx->b.gfx.flush = si_context_gfx_flush;
 
 	/* Border colors. */
diff --git a/src/gallium/winsys/amdgpu/drm/amdgpu_cs.c b/src/gallium/winsys/amdgpu/drm/amdgpu_cs.c
index 83da740f649..a9fc55f4a5a 100644
--- a/src/gallium/winsys/amdgpu/drm/amdgpu_cs.c
+++ b/src/gallium/winsys/amdgpu/drm/amdgpu_cs.c
@@ -335,8 +335,7 @@ amdgpu_cs_create(struct radeon_winsys_ctx *rwctx,
                  enum ring_type ring_type,
                  void (*flush)(void *ctx, unsigned flags,
                                struct pipe_fence_handle **fence),
-                 void *flush_ctx,
-                 struct pb_buffer *trace_buf)
+                 void *flush_ctx)
 {
    struct amdgpu_ctx *ctx = (struct amdgpu_ctx*)rwctx;
    struct amdgpu_cs *cs;
@@ -609,8 +608,7 @@ DEBUG_GET_ONCE_BOOL_OPTION(all_bos, "RADEON_ALL_BOS", FALSE)
 
 static void amdgpu_cs_flush(struct radeon_winsys_cs *rcs,
                             unsigned flags,
-                            struct pipe_fence_handle **fence,
-                            uint32_t cs_trace_id)
+                            struct pipe_fence_handle **fence)
 {
    struct amdgpu_cs *cs = amdgpu_cs(rcs);
    struct amdgpu_winsys *ws = cs->ctx->ws;
diff --git a/src/gallium/winsys/radeon/drm/Makefile.am b/src/gallium/winsys/radeon/drm/Makefile.am
index 0320aca01f9..b413b0b93a0 100644
--- a/src/gallium/winsys/radeon/drm/Makefile.am
+++ b/src/gallium/winsys/radeon/drm/Makefile.am
@@ -8,5 +8,3 @@ AM_CFLAGS = \
 noinst_LTLIBRARIES = libradeonwinsys.la
 
 libradeonwinsys_la_SOURCES = $(C_SOURCES)
-
-EXTRA_DIST = $(TOOLS_HDR)
diff --git a/src/gallium/winsys/radeon/drm/Makefile.sources b/src/gallium/winsys/radeon/drm/Makefile.sources
index a00c84d35b3..2762c91e216 100644
--- a/src/gallium/winsys/radeon/drm/Makefile.sources
+++ b/src/gallium/winsys/radeon/drm/Makefile.sources
@@ -2,12 +2,8 @@ C_SOURCES := \
 	radeon_drm_bo.c \
 	radeon_drm_bo.h \
 	radeon_drm_cs.c \
-	radeon_drm_cs_dump.c \
 	radeon_drm_cs.h \
 	radeon_drm_public.h \
 	radeon_drm_surface.c \
 	radeon_drm_winsys.c \
 	radeon_drm_winsys.h
-
-TOOLS_HDR := \
-	radeon_ctx.h
diff --git a/src/gallium/winsys/radeon/drm/radeon_ctx.h b/src/gallium/winsys/radeon/drm/radeon_ctx.h
deleted file mode 100644
index 5618b3a8d00..00000000000
--- a/src/gallium/winsys/radeon/drm/radeon_ctx.h
+++ /dev/null
@@ -1,205 +0,0 @@
-/*
- * Copyright 2011 Jerome Glisse <[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
- * on 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
- * THE AUTHOR(S) AND/OR THEIR 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.
- *
- * Authors:
- *      Jérôme Glisse
- */
-#ifndef RADEON_CTX_H
-#define RADEON_CTX_H
-
-#define _FILE_OFFSET_BITS 64
-#include <sys/mman.h>
-
-#include <errno.h>
-#include <stdio.h>
-#include <stdlib.h>
-#include <stdint.h>
-#include <string.h>
-#include "xf86drm.h"
-#include "radeon_drm.h"
-
-struct ctx {
-    int                         fd;
-};
-
-struct bo {
-    uint32_t                    handle;
-    uint32_t                    alignment;
-    uint64_t                    size;
-    uint64_t                    va;
-    void                        *ptr;
-};
-
-static void ctx_init(struct ctx *ctx)
-{
-    ctx->fd = drmOpen("radeon", NULL);
-    if (ctx->fd < 0) {
-        fprintf(stderr, "failed to open radeon drm device file\n");
-        exit(-1);
-    }
-}
-
-static void bo_wait(struct ctx *ctx, struct bo *bo)
-{
-    struct drm_radeon_gem_wait_idle args;
-    void *ptr;
-    int r;
-
-    /* Zero out args to make valgrind happy */
-    memset(&args, 0, sizeof(args));
-    args.handle = bo->handle;
-    do {
-        r = drmCommandWrite(ctx->fd, DRM_RADEON_GEM_WAIT_IDLE, &args, sizeof(args));
-    } while (r == -EBUSY);
-}
-
-
-static void ctx_cs(struct ctx *ctx, uint32_t *cs, uint32_t cs_flags[2], unsigned ndw,
-                   struct bo **bo, uint32_t *bo_relocs, unsigned nbo)
-{
-    struct drm_radeon_cs args;
-    struct drm_radeon_cs_chunk chunks[3];
-    uint64_t chunk_array[3];
-    unsigned i;
-    int r;
-
-    /* update handle */
-    for (i = 0; i < nbo; i++) {
-        bo_relocs[i*4+0] = bo[i]->handle;
-    }
-
-    args.num_chunks = 2;
-    if (cs_flags[0] || cs_flags[1]) {
-        /* enable RADEON_CHUNK_ID_FLAGS */
-        args.num_chunks = 3;
-    }
-    args.chunks = (uint64_t)(uintptr_t)chunk_array;
-    chunks[0].chunk_id = RADEON_CHUNK_ID_IB;
-    chunks[0].length_dw = ndw;
-    chunks[0].chunk_data = (uintptr_t)cs;
-    chunks[1].chunk_id = RADEON_CHUNK_ID_RELOCS;
-    chunks[1].length_dw = nbo * 4;
-    chunks[1].chunk_data = (uintptr_t)bo_relocs;
-    chunks[2].chunk_id = RADEON_CHUNK_ID_FLAGS;
-    chunks[2].length_dw = 2;
-    chunks[2].chunk_data = (uintptr_t)cs_flags;
-    chunk_array[0] = (uintptr_t)&chunks[0];
-    chunk_array[1] = (uintptr_t)&chunks[1];
-    chunk_array[2] = (uintptr_t)&chunks[2];
-
-    fprintf(stderr, "emiting cs %ddw with %d bo\n", ndw, nbo);
-    r = drmCommandWriteRead(ctx->fd, DRM_RADEON_CS, &args, sizeof(args));
-    if (r) {
-        fprintf(stderr, "cs submission failed with %d\n", r);
-        return;
-    }
-}
-
-static void bo_map(struct ctx *ctx, struct bo *bo)
-{
-    struct drm_radeon_gem_mmap args;
-    void *ptr;
-    int r;
-
-    /* Zero out args to make valgrind happy */
-    memset(&args, 0, sizeof(args));
-    args.handle = bo->handle;
-    args.offset = 0;
-    args.size = (uint64_t)bo->size;
-    r = drmCommandWriteRead(ctx->fd, DRM_RADEON_GEM_MMAP, &args, sizeof(args));
-    if (r) {
-        fprintf(stderr, "error mapping %p 0x%08X (error = %d)\n", bo, bo->handle, r);
-        exit(-1);
-    }
-    ptr = mmap(0, args.size, PROT_READ|PROT_WRITE, MAP_SHARED, ctx->fd, args.addr_ptr);
-    if (ptr == MAP_FAILED) {
-        fprintf(stderr, "%s failed to map bo\n", __func__);
-        exit(-1);
-    }
-    bo->ptr = ptr;
-}
-
-static void bo_va(struct ctx *ctx, struct bo *bo)
-{
-    struct drm_radeon_gem_va args;
-    int r;
-
-    args.handle = bo->handle;
-    args.vm_id = 0;
-    args.operation = RADEON_VA_MAP;
-    args.flags = RADEON_VM_PAGE_READABLE | RADEON_VM_PAGE_WRITEABLE | RADEON_VM_PAGE_SNOOPED;
-    args.offset = bo->va;
-    r = drmCommandWriteRead(ctx->fd, DRM_RADEON_GEM_VA, &args, sizeof(args));
-    if (r && args.operation == RADEON_VA_RESULT_ERROR) {
-        fprintf(stderr, "radeon: Failed to allocate virtual address for buffer:\n");
-        fprintf(stderr, "radeon:    size      : %d bytes\n", bo->size);
-        fprintf(stderr, "radeon:    alignment : %d bytes\n", bo->alignment);
-        fprintf(stderr, "radeon:    va        : 0x%016llx\n", (unsigned long long)bo->va);
-        exit(-1);
-    }
-}
-
-static struct bo *bo_new(struct ctx *ctx, unsigned ndw, uint32_t *data, uint64_t va, uint32_t alignment)
-{
-    struct drm_radeon_gem_create args;
-    struct bo *bo;
-    int r;
-
-    bo = calloc(1, sizeof(*bo));
-    if (bo == NULL) {
-        fprintf(stderr, "failed to malloc bo struct\n");
-        exit(-1);
-    }
-    bo->size = ndw * 4ULL;
-    bo->va = va;
-    bo->alignment = alignment;
-
-    args.size = bo->size;
-    args.alignment = bo->alignment;
-    args.initial_domain = RADEON_GEM_DOMAIN_GTT;
-    args.flags = 0;
-    args.handle = 0;
-
-    r = drmCommandWriteRead(ctx->fd, DRM_RADEON_GEM_CREATE, &args, sizeof(args));
-    bo->handle = args.handle;
-    if (r) {
-        fprintf(stderr, "Failed to allocate :\n");
-        fprintf(stderr, "   size      : %d bytes\n", bo->size);
-        fprintf(stderr, "   alignment : %d bytes\n", bo->alignment);
-        free(bo);
-        exit(-1);
-    }
-
-    if (data) {
-        bo_map(ctx, bo);
-        memcpy(bo->ptr, data, bo->size);
-    }
-
-    if (va) {
-        bo_va(ctx, bo);
-    }
-
-    return bo;
-}
-
-
-#endif
diff --git a/src/gallium/winsys/radeon/drm/radeon_drm_cs.c b/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
index 155a13008a4..b50e19c0381 100644
--- a/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
+++ b/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
@@ -168,8 +168,7 @@ radeon_drm_cs_create(struct radeon_winsys_ctx *ctx,
                      enum ring_type ring_type,
                      void (*flush)(void *ctx, unsigned flags,
                                    struct pipe_fence_handle **fence),
-                     void *flush_ctx,
-                     struct pb_buffer *trace_buf)
+                     void *flush_ctx)
 {
     struct radeon_drm_winsys *ws = (struct radeon_drm_winsys*)ctx;
     struct radeon_drm_cs *cs;
@@ -183,7 +182,6 @@ radeon_drm_cs_create(struct radeon_winsys_ctx *ctx,
     cs->ws = ws;
     cs->flush_cs = flush;
     cs->flush_data = flush_ctx;
-    cs->trace_buf = (struct radeon_bo*)trace_buf;
 
     if (!radeon_init_cs_context(&cs->csc1, cs->ws)) {
         FREE(cs);
@@ -439,10 +437,6 @@ void radeon_drm_cs_emit_ioctl_oneshot(struct radeon_drm_cs *cs, struct radeon_cs
         }
     }
 
-    if (cs->trace_buf) {
-        radeon_dump_cs_on_lockup(cs, csc);
-    }
-
     for (i = 0; i < csc->crelocs; i++)
         p_atomic_dec(&csc->relocs_bo[i].bo->num_active_ioctls);
 
@@ -467,8 +461,7 @@ DEBUG_GET_ONCE_BOOL_OPTION(noop, "RADEON_NOOP", FALSE)
 
 static void radeon_drm_cs_flush(struct radeon_winsys_cs *rcs,
                                 unsigned flags,
-                                struct pipe_fence_handle **fence,
-                                uint32_t cs_trace_id)
+                                struct pipe_fence_handle **fence)
 {
     struct radeon_drm_cs *cs = radeon_drm_cs(rcs);
     struct radeon_cs_context *tmp;
@@ -520,8 +513,6 @@ static void radeon_drm_cs_flush(struct radeon_winsys_cs *rcs,
     cs->csc = cs->cst;
     cs->cst = tmp;
 
-    cs->cst->cs_trace_id = cs_trace_id;
-
     /* If the CS is not empty or overflowed, emit it in a separate thread. */
     if (cs->base.cdw && cs->base.cdw <= cs->base.max_dw && !debug_get_option_noop()) {
         unsigned i, crelocs;
diff --git a/src/gallium/winsys/radeon/drm/radeon_drm_cs.h b/src/gallium/winsys/radeon/drm/radeon_drm_cs.h
index 81f66f56d99..4ffa91ae8b2 100644
--- a/src/gallium/winsys/radeon/drm/radeon_drm_cs.h
+++ b/src/gallium/winsys/radeon/drm/radeon_drm_cs.h
@@ -43,8 +43,6 @@ struct radeon_cs_context {
     uint64_t                    chunk_array[3];
     uint32_t                    flags[2];
 
-    uint32_t                    cs_trace_id;
-
     /* Buffers. */
     unsigned                    nrelocs;
     unsigned                    crelocs;
@@ -80,7 +78,6 @@ struct radeon_drm_cs {
     void *flush_data;
 
     pipe_semaphore flush_completed;
-    struct radeon_bo                    *trace_buf;
 };
 
 int radeon_lookup_buffer(struct radeon_cs_context *csc, struct radeon_bo *bo);
@@ -126,6 +123,4 @@ void radeon_drm_cs_sync_flush(struct radeon_winsys_cs *rcs);
 void radeon_drm_cs_init_functions(struct radeon_drm_winsys *ws);
 void radeon_drm_cs_emit_ioctl_oneshot(struct radeon_drm_cs *cs, struct radeon_cs_context *csc);
 
-void radeon_dump_cs_on_lockup(struct radeon_drm_cs *cs, struct radeon_cs_context *csc);
-
 #endif
diff --git a/src/gallium/winsys/radeon/drm/radeon_drm_cs_dump.c b/src/gallium/winsys/radeon/drm/radeon_drm_cs_dump.c
deleted file mode 100644
index 99585956a49..00000000000
--- a/src/gallium/winsys/radeon/drm/radeon_drm_cs_dump.c
+++ /dev/null
@@ -1,161 +0,0 @@
-/*
- * Copyright © 2013 Jérôme Glisse
- * 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.
- */
-/*
- * Authors:
- *      Jérôme Glisse <[email protected]>
- */
-#include <stdio.h>
-#include <stdlib.h>
-#include <stdint.h>
-#include <inttypes.h>
-#include <xf86drm.h>
-#include "radeon_drm_cs.h"
-#include "radeon_drm_bo.h"
-
-#define RADEON_CS_DUMP_AFTER_MS_TIMEOUT         500
-
-void radeon_dump_cs_on_lockup(struct radeon_drm_cs *cs, struct radeon_cs_context *csc)
-{
-    struct drm_radeon_gem_busy args;
-    FILE *dump;
-    unsigned i, lockup;
-    uint32_t *ptr;
-    char fname[32];
-
-    /* only dump the first cs to cause a lockup */
-    if (!csc->crelocs) {
-        /* can not determine if there was a lockup if no bo were use by
-         * the cs and most likely in such case no lockup occurs
-         */
-        return;
-    }
-
-    memset(&args, 0, sizeof(args));
-    args.handle = csc->relocs_bo[0].bo->handle;
-    for (i = 0; i < RADEON_CS_DUMP_AFTER_MS_TIMEOUT; i++) {
-        usleep(1);
-        lockup = drmCommandWriteRead(csc->fd, DRM_RADEON_GEM_BUSY, &args, sizeof(args));
-        if (!lockup) {
-            break;
-        }
-    }
-    if (!lockup || i < RADEON_CS_DUMP_AFTER_MS_TIMEOUT) {
-        return;
-    }
-
-    ptr = radeon_bo_do_map(cs->trace_buf);
-    fprintf(stderr, "timeout on cs lockup likely happen at cs 0x%08x dw 0x%08x\n", ptr[1], ptr[0]);
-
-    if (csc->cs_trace_id != ptr[1]) {
-        return;
-    }
-
-    /* ok we are most likely facing a lockup write the standalone replay file */
-    snprintf(fname, sizeof(fname), "rlockup_0x%08x.c", csc->cs_trace_id);
-    dump = fopen(fname, "w");
-    if (dump == NULL) {
-        return;
-    }
-    fprintf(dump, "/* To build this file you will need to copy radeon_ctx.h\n");
-    fprintf(dump, " * in same directory. You can find radeon_ctx.h in mesa tree :\n");
-    fprintf(dump, " * mesa/src/gallium/winsys/radeon/drm/radeon_ctx.h\n");
-    fprintf(dump, " * Build with :\n");
-    fprintf(dump, " * gcc -O0 -g `pkg-config --cflags --libs libdrm` %s -o rlockup_0x%08x \n", fname, csc->cs_trace_id);
-    fprintf(dump, " */\n");
-    fprintf(dump, " /* timeout on cs lockup likely happen at cs 0x%08x dw 0x%08x*/\n", ptr[1], ptr[0]);
-    fprintf(dump, "#include <stdio.h>\n");
-    fprintf(dump, "#include <stdint.h>\n");
-    fprintf(dump, "#include \"radeon_ctx.h\"\n");
-    fprintf(dump, "\n");
-    fprintf(dump, "#define ARRAY_SIZE(x)  (sizeof(x)/sizeof(x[0]))\n");
-    fprintf(dump, "\n");
-
-    for (i = 0; i < csc->crelocs; i++) {
-        unsigned j, ndw = (csc->relocs_bo[i].bo->base.size + 3) >> 2;
-
-        ptr = radeon_bo_do_map(csc->relocs_bo[i].bo);
-        if (ptr) {
-            fprintf(dump, "static uint32_t bo_%04d_data[%d] = {\n   ", i, ndw);
-            for (j = 0; j < ndw; j++) {
-                if (j && !(j % 8)) {
-                    uint32_t offset = (j - 8) << 2;
-                    fprintf(dump, "  /* [0x%08x] va[0x%016"PRIx64"] */\n   ", offset, offset + csc->relocs_bo[i].bo->va);
-                }
-                fprintf(dump, " 0x%08x,", ptr[j]);
-            }
-            fprintf(dump, "};\n\n");
-        }
-    }
-
-    fprintf(dump, "static uint32_t bo_relocs[%d] = {\n", csc->crelocs * 4);
-    for (i = 0; i < csc->crelocs; i++) {
-        fprintf(dump, "    0x%08x, 0x%08x, 0x%08x, 0x%08x,\n",
-                0, csc->relocs[i].read_domains, csc->relocs[i].write_domain, csc->relocs[i].flags);
-    }
-    fprintf(dump, "};\n\n");
-
-    fprintf(dump, "/* cs %d dw */\n", csc->chunks[0].length_dw);
-    fprintf(dump, "static uint32_t cs[] = {\n");
-    ptr = csc->buf;
-    for (i = 0; i < csc->chunks[0].length_dw; i++) {
-        fprintf(dump, "    0x%08x,\n", ptr[i]);
-    }
-    fprintf(dump, "};\n\n");
-
-    fprintf(dump, "static uint32_t cs_flags[2] = {\n");
-    fprintf(dump, "    0x%08x,\n", csc->flags[0]);
-    fprintf(dump, "    0x%08x,\n", csc->flags[1]);
-    fprintf(dump, "};\n\n");
-
-    fprintf(dump, "int main(int argc, char *argv[])\n");
-    fprintf(dump, "{\n");
-    fprintf(dump, "    struct bo *bo[%d];\n", csc->crelocs);
-    fprintf(dump, "    struct ctx ctx;\n");
-    fprintf(dump, "\n");
-    fprintf(dump, "    ctx_init(&ctx);\n");
-    fprintf(dump, "\n");
-
-    for (i = 0; i < csc->crelocs; i++) {
-        unsigned ndw = (csc->relocs_bo[i].bo->base.size + 3) >> 2;
-        uint32_t *ptr;
-
-        ptr = radeon_bo_do_map(csc->relocs_bo[i].bo);
-        if (ptr) {
-            fprintf(dump, "    bo[%d] = bo_new(&ctx, %d, bo_%04d_data, 0x%016"PRIx64", 0x%08x);\n",
-                    i, ndw, i, csc->relocs_bo[i].bo->va, csc->relocs_bo[i].bo->base.alignment);
-        } else {
-            fprintf(dump, "    bo[%d] = bo_new(&ctx, %d, NULL, 0x%016"PRIx64", 0x%08x);\n",
-                    i, ndw, csc->relocs_bo[i].bo->va, csc->relocs_bo[i].bo->base.alignment);
-        }
-    }
-    fprintf(dump, "\n");
-    fprintf(dump, "    ctx_cs(&ctx, cs, cs_flags, ARRAY_SIZE(cs), bo, bo_relocs, %d);\n", csc->crelocs);
-    fprintf(dump, "\n");
-    fprintf(dump, "    fprintf(stderr, \"waiting for cs execution to end ....\\n\");\n");
-    fprintf(dump, "    bo_wait(&ctx, bo[0]);\n");
-    fprintf(dump, "}\n");
-    fclose(dump);
-}