freedreno/a6xx: smaller hammer for fb barrier

We just need to do a sequence of commands to flush the cache.

Signed-off-by: Rob Clark <[email protected]>
Reviewed-by: Kristian H. Kristensen <[email protected]>

3 files changed, 48 insertions, 0 deletions

diff --git a/src/gallium/drivers/freedreno/a6xx/fd6_emit.c b/src/gallium/drivers/freedreno/a6xx/fd6_emit.c
index c17bd49be4d..fca00204e54 100644
--- a/src/gallium/drivers/freedreno/a6xx/fd6_emit.c
+++ b/src/gallium/drivers/freedreno/a6xx/fd6_emit.c
@@ -1297,6 +1297,41 @@ fd6_mem_to_mem(struct fd_ringbuffer *ring, struct pipe_resource *dst,
 	}
 }
 
+/* this is *almost* the same as fd6_cache_flush().. which I guess
+ * could be re-worked to be something a bit more generic w/ param
+ * indicating what needs to be flushed..  although that would mean
+ * figuring out which events trigger what state to flush..
+ */
+static void
+fd6_framebuffer_barrier(struct fd_context *ctx)
+{
+	struct fd6_context *fd6_ctx = fd6_context(ctx);
+	struct fd_batch *batch = ctx->batch;
+	struct fd_ringbuffer *ring = batch->draw;
+	unsigned seqno;
+
+	seqno = fd6_event_write(batch, ring, CACHE_FLUSH_AND_INV_EVENT, true);
+
+	OUT_PKT7(ring, CP_WAIT_REG_MEM, 6);
+	OUT_RING(ring, 0x00000013);
+	OUT_RELOC(ring, fd6_ctx->blit_mem, 0, 0, 0);
+	OUT_RING(ring, seqno);
+	OUT_RING(ring, 0xffffffff);
+	OUT_RING(ring, 0x00000010);
+
+	fd6_event_write(batch, ring, UNK_1D, true);
+	fd6_event_write(batch, ring, UNK_1C, true);
+
+	seqno = fd6_event_write(batch, ring, CACHE_FLUSH_TS, true);
+
+	fd6_event_write(batch, ring, 0x31, false);
+
+	OUT_PKT7(ring, CP_UNK_A6XX_14, 4);
+	OUT_RING(ring, 0x00000000);
+	OUT_RELOC(ring, fd6_ctx->blit_mem, 0, 0, 0);
+	OUT_RING(ring, seqno);
+}
+
 void
 fd6_emit_init(struct pipe_context *pctx)
 {
@@ -1305,4 +1340,5 @@ fd6_emit_init(struct pipe_context *pctx)
 	ctx->emit_const_bo = fd6_emit_const_bo;
 	ctx->emit_ib = fd6_emit_ib;
 	ctx->mem_to_mem = fd6_mem_to_mem;
+	ctx->framebuffer_barrier = fd6_framebuffer_barrier;
 }
diff --git a/src/gallium/drivers/freedreno/freedreno_context.c b/src/gallium/drivers/freedreno/freedreno_context.c
index bf26d00bc1e..c30bfe44dd5 100644
--- a/src/gallium/drivers/freedreno/freedreno_context.c
+++ b/src/gallium/drivers/freedreno/freedreno_context.c
@@ -88,6 +88,15 @@ out:
 static void
 fd_texture_barrier(struct pipe_context *pctx, unsigned flags)
 {
+	if (flags == PIPE_TEXTURE_BARRIER_FRAMEBUFFER) {
+		struct fd_context *ctx = fd_context(pctx);
+
+		if (ctx->framebuffer_barrier) {
+			ctx->framebuffer_barrier(ctx);
+			return;
+		}
+	}
+
 	/* On devices that could sample from GMEM we could possibly do better.
 	 * Or if we knew that we were doing GMEM bypass we could just emit a
 	 * cache flush, perhaps?  But we don't know if future draws would cause
diff --git a/src/gallium/drivers/freedreno/freedreno_context.h b/src/gallium/drivers/freedreno/freedreno_context.h
index 4d22f8c470e..1fcebded21b 100644
--- a/src/gallium/drivers/freedreno/freedreno_context.h
+++ b/src/gallium/drivers/freedreno/freedreno_context.h
@@ -352,6 +352,9 @@ struct fd_context {
 			unsigned dst_off, struct pipe_resource *src, unsigned src_off,
 			unsigned sizedwords);
 
+	/* handling for barriers: */
+	void (*framebuffer_barrier)(struct fd_context *ctx);
+
 	/*
 	 * Common pre-cooked VBO state (used for a3xx and later):
 	 */