gallium: allow setting of the internal stream output offset

D3D10 allows setting of the internal offset of a buffer, which is
in general only incremented via actual stream output writes. By
allowing setting of the internal offset draw_auto is capable
of rendering from buffers which have not been actually streamed
out to. Our interface didn't allow. This change functionally
shouldn't make any difference to OpenGL where instead of an
append_bitmask you just get a real array where -1 means append
(like in D3D) and 0 means do not append.

Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

11 files changed, 34 insertions, 25 deletions

diff --git a/src/gallium/drivers/galahad/glhd_context.c b/src/gallium/drivers/galahad/glhd_context.c
index fa743bd..2e61e59 100644
--- a/src/gallium/drivers/galahad/glhd_context.c
+++ b/src/gallium/drivers/galahad/glhd_context.c
@@ -613,12 +613,12 @@ static INLINE void
 galahad_context_set_stream_output_targets(struct pipe_context *_pipe,
                                           unsigned num_targets,
                                           struct pipe_stream_output_target **tgs,
-                                          unsigned append_bitmask)
+                                          const unsigned *offsets)
 {
    struct galahad_context *glhd_pipe = galahad_context(_pipe);
    struct pipe_context *pipe = glhd_pipe->pipe;
 
-   pipe->set_stream_output_targets(pipe, num_targets, tgs, append_bitmask);
+   pipe->set_stream_output_targets(pipe, num_targets, tgs, offsets);
 }
 
 static void
diff --git a/src/gallium/drivers/ilo/ilo_state.c b/src/gallium/drivers/ilo/ilo_state.c
index 43dc347..1b97eaa 100644
--- a/src/gallium/drivers/ilo/ilo_state.c
+++ b/src/gallium/drivers/ilo/ilo_state.c
@@ -856,10 +856,11 @@ static void
 ilo_set_stream_output_targets(struct pipe_context *pipe,
                               unsigned num_targets,
                               struct pipe_stream_output_target **targets,
-                              unsigned append_bitmask)
+                              const unsigned *offset)
 {
    struct ilo_context *ilo = ilo_context(pipe);
    unsigned i;
+   unsigned append_bitmask = 0;
 
    if (!targets)
       num_targets = 0;
@@ -868,8 +869,11 @@ ilo_set_stream_output_targets(struct pipe_context *pipe,
    if (!ilo->so.count && !num_targets)
       return;
 
-   for (i = 0; i < num_targets; i++)
+   for (i = 0; i < num_targets; i++) {
       pipe_so_target_reference(&ilo->so.states[i], targets[i]);
+      if (offset[i] == (unsigned)-1)
+         append_bitmask |= 1 << i;
+   }
 
    for (; i < ilo->so.count; i++)
       pipe_so_target_reference(&ilo->so.states[i], NULL);
diff --git a/src/gallium/drivers/llvmpipe/lp_state_so.c b/src/gallium/drivers/llvmpipe/lp_state_so.c
index fa58f79..2af04cd 100644
--- a/src/gallium/drivers/llvmpipe/lp_state_so.c
+++ b/src/gallium/drivers/llvmpipe/lp_state_so.c
@@ -64,17 +64,17 @@ static void
 llvmpipe_set_so_targets(struct pipe_context *pipe,
                         unsigned num_targets,
                         struct pipe_stream_output_target **targets,
-                        unsigned append_bitmask)
+                        const unsigned *offsets)
 {
    struct llvmpipe_context *llvmpipe = llvmpipe_context(pipe);
    int i;
    for (i = 0; i < num_targets; i++) {
+      const boolean append = (offsets[i] == (unsigned)-1);
       pipe_so_target_reference((struct pipe_stream_output_target **)&llvmpipe->so_targets[i], targets[i]);
-      /* if we're not appending then lets reset the internal
-         data of our so target */
-      if (!(append_bitmask & (1 << i)) && llvmpipe->so_targets[i]) {
-         llvmpipe->so_targets[i]->internal_offset = 0;
-         llvmpipe->so_targets[i]->emitted_vertices = 0;
+      /* If we're not appending then lets set the internal
+         offset to what was requested */
+      if (!append && llvmpipe->so_targets[i]) {
+         llvmpipe->so_targets[i]->internal_offset = offsets[i];
       }
    }
 
diff --git a/src/gallium/drivers/noop/noop_state.c b/src/gallium/drivers/noop/noop_state.c
index 9c62c27..5cb37b6 100644
--- a/src/gallium/drivers/noop/noop_state.c
+++ b/src/gallium/drivers/noop/noop_state.c
@@ -274,7 +274,7 @@ static void noop_stream_output_target_destroy(struct pipe_context *ctx,
 static void noop_set_stream_output_targets(struct pipe_context *ctx,
                            unsigned num_targets,
                            struct pipe_stream_output_target **targets,
-                           unsigned append_bitmask)
+                           const unsigned *offsets)
 {
 }
 
diff --git a/src/gallium/drivers/nouveau/nv50/nv50_state.c b/src/gallium/drivers/nouveau/nv50/nv50_state.c
index c03d729..862636b 100644
--- a/src/gallium/drivers/nouveau/nv50/nv50_state.c
+++ b/src/gallium/drivers/nouveau/nv50/nv50_state.c
@@ -1055,7 +1055,7 @@ static void
 nv50_set_stream_output_targets(struct pipe_context *pipe,
                                unsigned num_targets,
                                struct pipe_stream_output_target **targets,
-                               unsigned append_mask)
+                               const unsigned *offsets)
 {
    struct nv50_context *nv50 = nv50_context(pipe);
    unsigned i;
@@ -1066,7 +1066,8 @@ nv50_set_stream_output_targets(struct pipe_context *pipe,
 
    for (i = 0; i < num_targets; ++i) {
       const boolean changed = nv50->so_target[i] != targets[i];
-      if (!changed && (append_mask & (1 << i)))
+      const boolean append = (offsets[i] == (unsigned)-1);
+      if (!changed && append)
          continue;
       nv50->so_targets_dirty |= 1 << i;
 
@@ -1075,7 +1076,7 @@ nv50_set_stream_output_targets(struct pipe_context *pipe,
          serialize = FALSE;
       }
 
-      if (targets[i] && !(append_mask & (1 << i)))
+      if (targets[i] && !append)
          nv50_so_target(targets[i])->clean = TRUE;
 
       pipe_so_target_reference(&nv50->so_target[i], targets[i]);
diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_state.c b/src/gallium/drivers/nouveau/nvc0/nvc0_state.c
index dec5355..88dbaa1 100644
--- a/src/gallium/drivers/nouveau/nvc0/nvc0_state.c
+++ b/src/gallium/drivers/nouveau/nvc0/nvc0_state.c
@@ -1031,7 +1031,7 @@ static void
 nvc0_set_transform_feedback_targets(struct pipe_context *pipe,
                                     unsigned num_targets,
                                     struct pipe_stream_output_target **targets,
-                                    unsigned append_mask)
+                                    unsigned *offsets)
 {
    struct nvc0_context *nvc0 = nvc0_context(pipe);
    unsigned i;
@@ -1040,14 +1040,15 @@ nvc0_set_transform_feedback_targets(struct pipe_context *pipe,
    assert(num_targets <= 4);
 
    for (i = 0; i < num_targets; ++i) {
-      if (nvc0->tfbbuf[i] == targets[i] && (append_mask & (1 << i)))
+      boolean append = (offsets[i] == ((unsigned)-1));
+      if (nvc0->tfbbuf[i] == targets[i] && append)
          continue;
       nvc0->tfbbuf_dirty |= 1 << i;
 
       if (nvc0->tfbbuf[i] && nvc0->tfbbuf[i] != targets[i])
          nvc0_so_target_save_offset(pipe, nvc0->tfbbuf[i], i, &serialize);
 
-      if (targets[i] && !(append_mask & (1 << i)))
+      if (targets[i] && !append)
          nvc0_so_target(targets[i])->clean = TRUE;
 
       pipe_so_target_reference(&nvc0->tfbbuf[i], targets[i]);
diff --git a/src/gallium/drivers/radeon/r600_pipe_common.h b/src/gallium/drivers/radeon/r600_pipe_common.h
index cbd3f0e..a178e9c 100644
--- a/src/gallium/drivers/radeon/r600_pipe_common.h
+++ b/src/gallium/drivers/radeon/r600_pipe_common.h
@@ -413,7 +413,7 @@ void r600_streamout_buffers_dirty(struct r600_common_context *rctx);
 void r600_set_streamout_targets(struct pipe_context *ctx,
 				unsigned num_targets,
 				struct pipe_stream_output_target **targets,
-				unsigned append_bitmask);
+				unsigned *offset);
 void r600_emit_streamout_end(struct r600_common_context *rctx);
 void r600_streamout_init(struct r600_common_context *rctx);
 
diff --git a/src/gallium/drivers/radeon/r600_streamout.c b/src/gallium/drivers/radeon/r600_streamout.c
index adc11e0..b6cf858 100644
--- a/src/gallium/drivers/radeon/r600_streamout.c
+++ b/src/gallium/drivers/radeon/r600_streamout.c
@@ -108,10 +108,11 @@ void r600_streamout_buffers_dirty(struct r600_common_context *rctx)
 void r600_set_streamout_targets(struct pipe_context *ctx,
 				unsigned num_targets,
 				struct pipe_stream_output_target **targets,
-				unsigned append_bitmask)
+				const unsigned *offsets)
 {
 	struct r600_common_context *rctx = (struct r600_common_context *)ctx;
 	unsigned i;
+        unsigned append_bitmask = 0;
 
 	/* Stop streamout. */
 	if (rctx->streamout.num_targets && rctx->streamout.begin_emitted) {
@@ -122,6 +123,8 @@ void r600_set_streamout_targets(struct pipe_context *ctx,
 	for (i = 0; i < num_targets; i++) {
 		pipe_so_target_reference((struct pipe_stream_output_target**)&rctx->streamout.targets[i], targets[i]);
 		r600_context_add_resource_size(ctx, targets[i]->buffer);
+		if (offsets[i] == ((unsigned)-1))
+			append_bitmask |=  1 << i;
 	}
 	for (; i < rctx->streamout.num_targets; i++) {
 		pipe_so_target_reference((struct pipe_stream_output_target**)&rctx->streamout.targets[i], NULL);
diff --git a/src/gallium/drivers/radeonsi/si_descriptors.c b/src/gallium/drivers/radeonsi/si_descriptors.c
index 06b29d4..6b0ff91 100644
--- a/src/gallium/drivers/radeonsi/si_descriptors.c
+++ b/src/gallium/drivers/radeonsi/si_descriptors.c
@@ -605,7 +605,7 @@ void si_set_ring_buffer(struct pipe_context *ctx, uint shader, uint slot,
 static void si_set_streamout_targets(struct pipe_context *ctx,
 				     unsigned num_targets,
 				     struct pipe_stream_output_target **targets,
-				     unsigned append_bitmask)
+				     const unsigned *offsets)
 {
 	struct si_context *sctx = (struct si_context *)ctx;
 	struct si_buffer_resources *buffers = &sctx->rw_buffers[PIPE_SHADER_VERTEX];
@@ -618,7 +618,7 @@ static void si_set_streamout_targets(struct pipe_context *ctx,
 	 */
 
 	/* Set the VGT regs. */
-	r600_set_streamout_targets(ctx, num_targets, targets, append_bitmask);
+	r600_set_streamout_targets(ctx, num_targets, targets, offsets);
 
 	/* Set the shader resources.*/
 	for (i = 0; i < num_targets; i++) {
diff --git a/src/gallium/drivers/softpipe/sp_state_so.c b/src/gallium/drivers/softpipe/sp_state_so.c
index f456237..3878e4a 100644
--- a/src/gallium/drivers/softpipe/sp_state_so.c
+++ b/src/gallium/drivers/softpipe/sp_state_so.c
@@ -63,7 +63,7 @@ static void
 softpipe_set_so_targets(struct pipe_context *pipe,
                         unsigned num_targets,
                         struct pipe_stream_output_target **targets,
-                        unsigned append_bitmask)
+                        const unsigned *offsets)
 {
    struct softpipe_context *softpipe = softpipe_context(pipe);
    unsigned i;
diff --git a/src/gallium/drivers/trace/tr_context.c b/src/gallium/drivers/trace/tr_context.c
index c10e010..3e99598 100644
--- a/src/gallium/drivers/trace/tr_context.c
+++ b/src/gallium/drivers/trace/tr_context.c
@@ -1080,7 +1080,7 @@ static INLINE void
 trace_context_set_stream_output_targets(struct pipe_context *_pipe,
                                         unsigned num_targets,
                                         struct pipe_stream_output_target **tgs,
-                                        unsigned append_bitmask)
+                                        const unsigned *offsets)
 {
    struct trace_context *tr_ctx = trace_context(_pipe);
    struct pipe_context *pipe = tr_ctx->pipe;
@@ -1090,9 +1090,9 @@ trace_context_set_stream_output_targets(struct pipe_context *_pipe,
    trace_dump_arg(ptr, pipe);
    trace_dump_arg(uint, num_targets);
    trace_dump_arg_array(ptr, tgs, num_targets);
-   trace_dump_arg(uint, append_bitmask);
+   trace_dump_arg_array(uint, offsets, num_targets);
 
-   pipe->set_stream_output_targets(pipe, num_targets, tgs, append_bitmask);
+   pipe->set_stream_output_targets(pipe, num_targets, tgs, offsets);
 
    trace_dump_call_end();
 }