radeonsi: Reinitialize all descriptors in CE preamble.

This fixes a problem with the CE preamble and restoring only stuff in the
preamble when needed.

To illustrate suppose we have two graphics IB's 1 and 2, which  are submitted in
that order. Furthermore suppose IB 1 does not use CE ram, but IB 2 does, and we
have a context switch at the start of IB 1, but not between IB 1 and IB 2.

The old code put the CE RAM loads in the preamble of IB 2. As the preamble of
IB 1 does not have the loads and the preamble of IB 2 does not get executed, the
old values are not load into CE RAM.

Fix this by always restoring the entire CE RAM.

v2: - Just load all descriptor set buffers instead of load and store the entire
      CE RAM.
    - Leave the ce_ram_dirty tracking in place for the non-preamble case.

Signed-off-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

Note: This commit differs from the one in master - 54f755fa0fd
("radeonsi: Reinitialize all descriptors in CE preamble.")

3 files changed, 17 insertions, 2 deletions

diff --git a/src/gallium/drivers/radeonsi/si_descriptors.c b/src/gallium/drivers/radeonsi/si_descriptors.c
index 69fdb086d28..57d2ae60a6d 100644
--- a/src/gallium/drivers/radeonsi/si_descriptors.c
+++ b/src/gallium/drivers/radeonsi/si_descriptors.c
@@ -159,7 +159,7 @@ static bool si_ce_upload(struct si_context *sctx, unsigned ce_offset, unsigned s
 	return true;
 }
 
-static void si_reinitialize_ce_ram(struct si_context *sctx,
+static void si_ce_reinitialize_descriptors(struct si_context *sctx,
                             struct si_descriptors *desc)
 {
 	if (desc->buffer) {
@@ -185,6 +185,17 @@ static void si_reinitialize_ce_ram(struct si_context *sctx,
 	desc->ce_ram_dirty = false;
 }
 
+void si_ce_reinitialize_all_descriptors(struct si_context *sctx)
+{
+	for (int i = 0; i < SI_NUM_SHADERS; i++) {
+		 si_ce_reinitialize_descriptors(sctx, &sctx->const_buffers[i].desc);
+		 si_ce_reinitialize_descriptors(sctx, &sctx->shader_buffers[i].desc);
+		 si_ce_reinitialize_descriptors(sctx, &sctx->samplers[i].views.desc);
+		 si_ce_reinitialize_descriptors(sctx, &sctx->images[i].desc);
+	}
+	 si_ce_reinitialize_descriptors(sctx, &sctx->rw_buffers.desc);
+}
+
 void si_ce_enable_loads(struct radeon_winsys_cs *ib)
 {
 	radeon_emit(ib, PKT3(PKT3_CONTEXT_CONTROL, 1, 0));
@@ -206,7 +217,7 @@ static bool si_upload_descriptors(struct si_context *sctx,
 		uint32_t const* list = (uint32_t const*)desc->list;
 
 		if (desc->ce_ram_dirty)
-			si_reinitialize_ce_ram(sctx, desc);
+			si_ce_reinitialize_descriptors(sctx, desc);
 
 		while(desc->dirty_mask) {
 			int begin, count;
diff --git a/src/gallium/drivers/radeonsi/si_hw_context.c b/src/gallium/drivers/radeonsi/si_hw_context.c
index dcf206df216..1a887c1a4e4 100644
--- a/src/gallium/drivers/radeonsi/si_hw_context.c
+++ b/src/gallium/drivers/radeonsi/si_hw_context.c
@@ -207,6 +207,9 @@ void si_begin_new_cs(struct si_context *ctx)
 	else if (ctx->ce_ib)
 		si_ce_enable_loads(ctx->ce_ib);
 
+	if (ctx->ce_preamble_ib)
+		si_ce_reinitialize_all_descriptors(ctx);
+
 	ctx->framebuffer.dirty_cbufs = (1 << 8) - 1;
 	ctx->framebuffer.dirty_zsbuf = true;
 	si_mark_atom_dirty(ctx, &ctx->framebuffer.atom);
diff --git a/src/gallium/drivers/radeonsi/si_state.h b/src/gallium/drivers/radeonsi/si_state.h
index a3589d4611d..aea98ae464f 100644
--- a/src/gallium/drivers/radeonsi/si_state.h
+++ b/src/gallium/drivers/radeonsi/si_state.h
@@ -250,6 +250,7 @@ struct si_buffer_resources {
 	} while(0)
 
 /* si_descriptors.c */
+void si_ce_reinitialize_all_descriptors(struct si_context *sctx);
 void si_ce_enable_loads(struct radeon_winsys_cs *ib);
 void si_set_ring_buffer(struct pipe_context *ctx, uint slot,
 			struct pipe_resource *buffer,