[Mesa-dev] [PATCH] radeonsi: add cs tracing
Christian König
deathsimple at vodafone.de
Mon Mar 25 09:38:23 PDT 2013
Am 25.03.2013 17:01, schrieb j.glisse at gmail.com:
> From: Jerome Glisse <jglisse at redhat.com>
>
> Same as on r600, trace cs execution by writting cs offset after each
> states, this allow to pin point lockup inside command stream and
> narrow down the scope of lockup investigation.
>
> Signed-off-by: Jerome Glisse <jglisse at redhat.com>
Could your rewrite this to use an si_pm4_state instead of hand coding
it? It's cleaner and should reduce the needed code quite a bit.
Christian.
> ---
> src/gallium/drivers/radeonsi/r600_hw_context.c | 58 ++++++++++++++++++++++++++
> src/gallium/drivers/radeonsi/r600_texture.c | 2 +-
> src/gallium/drivers/radeonsi/radeonsi_pipe.c | 22 ++++++++++
> src/gallium/drivers/radeonsi/radeonsi_pipe.h | 12 ++++++
> src/gallium/drivers/radeonsi/radeonsi_pm4.c | 12 ++++++
> src/gallium/drivers/radeonsi/si_state_draw.c | 7 +++-
> 6 files changed, 111 insertions(+), 2 deletions(-)
>
> diff --git a/src/gallium/drivers/radeonsi/r600_hw_context.c b/src/gallium/drivers/radeonsi/r600_hw_context.c
> index bd348f9..3cd5d0e 100644
> --- a/src/gallium/drivers/radeonsi/r600_hw_context.c
> +++ b/src/gallium/drivers/radeonsi/r600_hw_context.c
> @@ -142,6 +142,12 @@ void si_need_cs_space(struct r600_context *ctx, unsigned num_dw,
> /* Save 16 dwords for the fence mechanism. */
> num_dw += 16;
>
> +#if R600_TRACE_CS
> + if (ctx->screen->trace_bo) {
> + num_dw += R600_TRACE_CS_DWORDS;
> + }
> +#endif
> +
> /* Flush if there's not enough space. */
> if (num_dw > RADEON_MAX_CMDBUF_DWORDS) {
> radeonsi_flush(&ctx->context, NULL, RADEON_FLUSH_ASYNC);
> @@ -206,9 +212,41 @@ void si_context_flush(struct r600_context *ctx, unsigned flags)
> /* force to keep tiling flags */
> flags |= RADEON_FLUSH_KEEP_TILING_FLAGS;
>
> +#if R600_TRACE_CS
> + if (ctx->screen->trace_bo) {
> + struct r600_screen *rscreen = ctx->screen;
> + unsigned i;
> +
> + for (i = 0; i < cs->cdw; i++) {
> + fprintf(stderr, "[%4d] [%5d] 0x%08x\n", rscreen->cs_count, i, cs->buf[i]);
> + }
> + rscreen->cs_count++;
> + }
> +#endif
> +
> /* Flush the CS. */
> ctx->ws->cs_flush(ctx->cs, flags);
>
> +#if R600_TRACE_CS
> + if (ctx->screen->trace_bo) {
> + struct r600_screen *rscreen = ctx->screen;
> + unsigned i;
> +
> + for (i = 0; i < 10; i++) {
> + usleep(5);
> + if (!ctx->ws->buffer_is_busy(rscreen->trace_bo->buf, RADEON_USAGE_READWRITE)) {
> + break;
> + }
> + }
> + if (i == 10) {
> + fprintf(stderr, "timeout on cs lockup likely happen at cs %d dw %d\n",
> + rscreen->trace_ptr[1], rscreen->trace_ptr[0]);
> + } else {
> + fprintf(stderr, "cs %d executed in %dms\n", rscreen->trace_ptr[1], i * 5);
> + }
> + }
> +#endif
> +
> ctx->pm4_dirty_cdwords = 0;
> ctx->flags = 0;
>
> @@ -665,3 +703,23 @@ void r600_context_draw_opaque_count(struct r600_context *ctx, struct r600_so_tar
> cs->buf[cs->cdw++] = r600_context_bo_reloc(ctx, t->filled_size, RADEON_USAGE_READ);
>
> }
> +
> +#if R600_TRACE_CS
> +void r600_trace_emit(struct r600_context *rctx)
> +{
> + struct r600_screen *rscreen = rctx->screen;
> + struct radeon_winsys_cs *cs = rctx->cs;
> + uint64_t va;
> + uint32_t reloc;
> +
> + va = r600_resource_va(&rscreen->screen, (void*)rscreen->trace_bo);
> + reloc = r600_context_bo_reloc(rctx, rscreen->trace_bo, RADEON_USAGE_READWRITE);
> + cs->buf[cs->cdw++] = PKT3(PKT3_MEM_WRITE, 3, 0);
> + cs->buf[cs->cdw++] = va & 0xFFFFFFFFUL;
> + cs->buf[cs->cdw++] = (va >> 32UL) & 0xFFUL;
> + cs->buf[cs->cdw++] = cs->cdw;
> + cs->buf[cs->cdw++] = rscreen->cs_count;
> + cs->buf[cs->cdw++] = PKT3(PKT3_NOP, 0, 0);
> + cs->buf[cs->cdw++] = reloc;
> +}
> +#endif
> diff --git a/src/gallium/drivers/radeonsi/r600_texture.c b/src/gallium/drivers/radeonsi/r600_texture.c
> index 6cafc3d..3d074a3 100644
> --- a/src/gallium/drivers/radeonsi/r600_texture.c
> +++ b/src/gallium/drivers/radeonsi/r600_texture.c
> @@ -550,7 +550,7 @@ struct pipe_resource *si_texture_create(struct pipe_screen *screen,
>
> if (!(templ->flags & R600_RESOURCE_FLAG_TRANSFER) &&
> !(templ->bind & PIPE_BIND_SCANOUT)) {
> - array_mode = V_009910_ARRAY_2D_TILED_THIN1;
> + array_mode = V_009910_ARRAY_1D_TILED_THIN1;
> }
>
> r = r600_init_surface(rscreen, &surface, templ, array_mode,
> diff --git a/src/gallium/drivers/radeonsi/radeonsi_pipe.c b/src/gallium/drivers/radeonsi/radeonsi_pipe.c
> index c5dac29..a370d7e 100644
> --- a/src/gallium/drivers/radeonsi/radeonsi_pipe.c
> +++ b/src/gallium/drivers/radeonsi/radeonsi_pipe.c
> @@ -525,6 +525,14 @@ static void r600_destroy_screen(struct pipe_screen* pscreen)
> rscreen->ws->buffer_unmap(rscreen->fences.bo->cs_buf);
> si_resource_reference(&rscreen->fences.bo, NULL);
> }
> +
> +#if R600_TRACE_CS
> + if (rscreen->trace_bo) {
> + rscreen->ws->buffer_unmap(rscreen->trace_bo->cs_buf);
> + pipe_resource_reference((struct pipe_resource**)&rscreen->trace_bo, NULL);
> + }
> +#endif
> +
> pipe_mutex_destroy(rscreen->fences.mutex);
>
> rscreen->ws->destroy(rscreen->ws);
> @@ -727,5 +735,19 @@ struct pipe_screen *radeonsi_screen_create(struct radeon_winsys *ws)
> LIST_INITHEAD(&rscreen->fences.blocks);
> pipe_mutex_init(rscreen->fences.mutex);
>
> +#if R600_TRACE_CS
> + rscreen->cs_count = 0;
> + if (rscreen->info.drm_minor >= 28) {
> + rscreen->trace_bo = (struct si_resource*)pipe_buffer_create(&rscreen->screen,
> + PIPE_BIND_CUSTOM,
> + PIPE_USAGE_STAGING,
> + 4096);
> + if (rscreen->trace_bo) {
> + rscreen->trace_ptr = rscreen->ws->buffer_map(rscreen->trace_bo->cs_buf, NULL,
> + PIPE_TRANSFER_UNSYNCHRONIZED);
> + }
> + }
> +#endif
> +
> return &rscreen->screen;
> }
> diff --git a/src/gallium/drivers/radeonsi/radeonsi_pipe.h b/src/gallium/drivers/radeonsi/radeonsi_pipe.h
> index d0f04f4..f5c03ce 100644
> --- a/src/gallium/drivers/radeonsi/radeonsi_pipe.h
> +++ b/src/gallium/drivers/radeonsi/radeonsi_pipe.h
> @@ -47,6 +47,9 @@
> #define R600_BIG_ENDIAN 0
> #endif
>
> +#define R600_TRACE_CS 0
> +#define R600_TRACE_CS_DWORDS 7
> +
> struct r600_pipe_fences {
> struct si_resource *bo;
> unsigned *data;
> @@ -67,6 +70,11 @@ struct r600_screen {
> struct r600_tiling_info tiling_info;
> struct util_slab_mempool pool_buffers;
> struct r600_pipe_fences fences;
> +#if R600_TRACE_CS
> + struct si_resource *trace_bo;
> + uint32_t *trace_ptr;
> + unsigned cs_count;
> +#endif
> };
>
> struct si_pipe_sampler_view {
> @@ -226,6 +234,10 @@ void r600_translate_index_buffer(struct r600_context *r600,
> struct pipe_index_buffer *ib,
> unsigned count);
>
> +#if R600_TRACE_CS
> +void r600_trace_emit(struct r600_context *rctx);
> +#endif
> +
> /*
> * common helpers
> */
> diff --git a/src/gallium/drivers/radeonsi/radeonsi_pm4.c b/src/gallium/drivers/radeonsi/radeonsi_pm4.c
> index 79a2521..8e01738 100644
> --- a/src/gallium/drivers/radeonsi/radeonsi_pm4.c
> +++ b/src/gallium/drivers/radeonsi/radeonsi_pm4.c
> @@ -199,6 +199,12 @@ unsigned si_pm4_dirty_dw(struct r600_context *rctx)
> continue;
>
> count += state->ndw;
> +#if R600_TRACE_CS
> + /* for tracing each states */
> + if (rctx->screen->trace_bo) {
> + count += R600_TRACE_CS_DWORDS;
> + }
> +#endif
> }
>
> return count;
> @@ -219,6 +225,12 @@ void si_pm4_emit(struct r600_context *rctx, struct si_pm4_state *state)
> }
>
> cs->cdw += state->ndw;
> +
> +#if R600_TRACE_CS
> + if (rctx->screen->trace_bo) {
> + r600_trace_emit(rctx);
> + }
> +#endif
> }
>
> void si_pm4_emit_dirty(struct r600_context *rctx)
> diff --git a/src/gallium/drivers/radeonsi/si_state_draw.c b/src/gallium/drivers/radeonsi/si_state_draw.c
> index a78751b..1e1d1cc 100644
> --- a/src/gallium/drivers/radeonsi/si_state_draw.c
> +++ b/src/gallium/drivers/radeonsi/si_state_draw.c
> @@ -579,6 +579,12 @@ void si_draw_vbo(struct pipe_context *ctx, const struct pipe_draw_info *info)
> si_pm4_emit_dirty(rctx);
> rctx->pm4_dirty_cdwords = 0;
>
> +#if R600_TRACE_CS
> + if (rctx->screen->trace_bo) {
> + r600_trace_emit(rctx);
> + }
> +#endif
> +
> #if 0
> /* Enable stream out if needed. */
> if (rctx->streamout_start) {
> @@ -587,7 +593,6 @@ void si_draw_vbo(struct pipe_context *ctx, const struct pipe_draw_info *info)
> }
> #endif
>
> -
> rctx->flags |= R600_CONTEXT_DST_CACHES_DIRTY;
>
> /* Set the depth buffer as dirty. */
More information about the mesa-dev
mailing list