r600g: make all scissor states use single atom

As suggested by Marek Olšák, we can use single atom to track all scissor
states. This will allow to simplify dirty atom handling later.

Signed-off-by: Marek Olšák <[email protected]>

6 files changed, 62 insertions, 40 deletions

diff --git a/src/gallium/drivers/r600/evergreen_state.c b/src/gallium/drivers/r600/evergreen_state.c
index 0d4b5987246..3a7f583ecfa 100644
--- a/src/gallium/drivers/r600/evergreen_state.c
+++ b/src/gallium/drivers/r600/evergreen_state.c
@@ -892,27 +892,38 @@ static void evergreen_set_scissor_states(struct pipe_context *ctx,
 					const struct pipe_scissor_state *state)
 {
 	struct r600_context *rctx = (struct r600_context *)ctx;
+	struct r600_scissor_state *rstate = &rctx->scissor;
 	int i;
 
-	for (i = start_slot; i < start_slot + num_scissors; i++) {
-		rctx->scissor[i].scissor = state[i - start_slot];
-		r600_mark_atom_dirty(rctx, &rctx->scissor[i].atom);
-	}
+	for (i = start_slot; i < start_slot + num_scissors; i++)
+		rstate->scissor[i] = state[i - start_slot];
+	rstate->dirty_mask |= ((1 << num_scissors) - 1) << start_slot;
+	rstate->atom.num_dw = util_bitcount(rstate->dirty_mask) * 4;
+	r600_mark_atom_dirty(rctx, &rstate->atom);
 }
 
 static void evergreen_emit_scissor_state(struct r600_context *rctx, struct r600_atom *atom)
 {
 	struct radeon_winsys_cs *cs = rctx->b.rings.gfx.cs;
-	struct r600_scissor_state *rstate = (struct r600_scissor_state *)atom;
-	struct pipe_scissor_state *state = &rstate->scissor;
-	unsigned offset = rstate->idx * 4 * 2;
+	struct r600_scissor_state *rstate = &rctx->scissor;
+	struct pipe_scissor_state *state;
+	uint32_t dirty_mask;
+	unsigned i, offset;
 	uint32_t tl, br;
 
-	evergreen_get_scissor_rect(rctx, state->minx, state->miny, state->maxx, state->maxy, &tl, &br);
+	dirty_mask = rstate->dirty_mask;
+	while (dirty_mask != 0) {
+		i = u_bit_scan(&dirty_mask);
+		state = &rstate->scissor[i];
+		evergreen_get_scissor_rect(rctx, state->minx, state->miny, state->maxx, state->maxy, &tl, &br);
 
-	radeon_set_context_reg_seq(cs, R_028250_PA_SC_VPORT_SCISSOR_0_TL + offset, 2);
-	radeon_emit(cs, tl);
-	radeon_emit(cs, br);
+		offset = i * 4 * 2;
+		radeon_set_context_reg_seq(cs, R_028250_PA_SC_VPORT_SCISSOR_0_TL + offset, 2);
+		radeon_emit(cs, tl);
+		radeon_emit(cs, br);
+	}
+	rstate->dirty_mask = 0;
+	rstate->atom.num_dw = 0;
 }
 
 /**
@@ -3491,11 +3502,10 @@ void evergreen_init_state_functions(struct r600_context *rctx)
 	r600_init_atom(rctx, &rctx->dsa_state.atom, id++, r600_emit_cso_state, 0);
 	r600_init_atom(rctx, &rctx->poly_offset_state.atom, id++, evergreen_emit_polygon_offset, 6);
 	r600_init_atom(rctx, &rctx->rasterizer_state.atom, id++, r600_emit_cso_state, 0);
+	r600_init_atom(rctx, &rctx->scissor.atom, id++, evergreen_emit_scissor_state, 0);
 	for (i = 0; i < R600_MAX_VIEWPORTS; i++) {
 		r600_init_atom(rctx, &rctx->viewport[i].atom, id++, r600_emit_viewport_state, 8);
-		r600_init_atom(rctx, &rctx->scissor[i].atom, id++, evergreen_emit_scissor_state, 4);
 		rctx->viewport[i].idx = i;
-		rctx->scissor[i].idx = i;
 	}
 	r600_init_atom(rctx, &rctx->stencil_ref.atom, id++, r600_emit_stencil_ref, 4);
 	r600_init_atom(rctx, &rctx->vertex_fetch_shader.atom, id++, evergreen_emit_vertex_fetch_shader, 5);
diff --git a/src/gallium/drivers/r600/r600_blit.c b/src/gallium/drivers/r600/r600_blit.c
index 08b2f644cad..8774cb5d655 100644
--- a/src/gallium/drivers/r600/r600_blit.c
+++ b/src/gallium/drivers/r600/r600_blit.c
@@ -66,7 +66,7 @@ static void r600_blitter_begin(struct pipe_context *ctx, enum r600_blitter_op op
 
 	if (op & R600_SAVE_FRAGMENT_STATE) {
 		util_blitter_save_viewport(rctx->blitter, &rctx->viewport[0].state);
-		util_blitter_save_scissor(rctx->blitter, &rctx->scissor[0].scissor);
+		util_blitter_save_scissor(rctx->blitter, &rctx->scissor.scissor[0]);
 		util_blitter_save_fragment_shader(rctx->blitter, rctx->ps_shader);
 		util_blitter_save_blend(rctx->blitter, rctx->blend_state.cso);
 		util_blitter_save_depth_stencil_alpha(rctx->blitter, rctx->dsa_state.cso);
diff --git a/src/gallium/drivers/r600/r600_hw_context.c b/src/gallium/drivers/r600/r600_hw_context.c
index 2fe29e91c4f..c5403252d36 100644
--- a/src/gallium/drivers/r600/r600_hw_context.c
+++ b/src/gallium/drivers/r600/r600_hw_context.c
@@ -308,8 +308,10 @@ void r600_begin_new_cs(struct r600_context *ctx)
 	r600_mark_atom_dirty(ctx, &ctx->poly_offset_state.atom);
 	r600_mark_atom_dirty(ctx, &ctx->vgt_state.atom);
 	r600_mark_atom_dirty(ctx, &ctx->sample_mask.atom);
+	ctx->scissor.dirty_mask = (1 << R600_MAX_VIEWPORTS) - 1;
+	ctx->scissor.atom.num_dw = R600_MAX_VIEWPORTS * 4;
+	r600_mark_atom_dirty(ctx, &ctx->scissor.atom);
 	for (i = 0; i < R600_MAX_VIEWPORTS; i++) {
-		r600_mark_atom_dirty(ctx, &ctx->scissor[i].atom);
 		r600_mark_atom_dirty(ctx, &ctx->viewport[i].atom);
 	}
 	if (ctx->b.chip_class < EVERGREEN) {
diff --git a/src/gallium/drivers/r600/r600_pipe.h b/src/gallium/drivers/r600/r600_pipe.h
index 8d5fd99e65a..e09bee15d46 100644
--- a/src/gallium/drivers/r600/r600_pipe.h
+++ b/src/gallium/drivers/r600/r600_pipe.h
@@ -38,7 +38,7 @@
 
 #include "tgsi/tgsi_scan.h"
 
-#define R600_NUM_ATOMS 75
+#define R600_NUM_ATOMS 60
 
 #define R600_MAX_VIEWPORTS 16
 
@@ -393,9 +393,9 @@ struct r600_cso_state
 struct r600_scissor_state
 {
 	struct r600_atom		atom;
-	struct pipe_scissor_state	scissor;
+	struct pipe_scissor_state	scissor[R600_MAX_VIEWPORTS];
+	uint32_t			dirty_mask;
 	bool				enable; /* r6xx only */
-	int idx;
 };
 
 struct r600_fetch_shader {
@@ -458,7 +458,7 @@ struct r600_context {
 	struct r600_poly_offset_state	poly_offset_state;
 	struct r600_cso_state		rasterizer_state;
 	struct r600_sample_mask		sample_mask;
-	struct r600_scissor_state	scissor[R600_MAX_VIEWPORTS];
+	struct r600_scissor_state	scissor;
 	struct r600_seamless_cube_map	seamless_cube_map;
 	struct r600_config_state	config_state;
 	struct r600_stencil_ref_state	stencil_ref;
diff --git a/src/gallium/drivers/r600/r600_state.c b/src/gallium/drivers/r600/r600_state.c
index f2d24a3e37a..64a22e6d305 100644
--- a/src/gallium/drivers/r600/r600_state.c
+++ b/src/gallium/drivers/r600/r600_state.c
@@ -769,21 +769,32 @@ static void r600_set_polygon_stipple(struct pipe_context *ctx,
 static void r600_emit_scissor_state(struct r600_context *rctx, struct r600_atom *atom)
 {
 	struct radeon_winsys_cs *cs = rctx->b.rings.gfx.cs;
-	struct r600_scissor_state *rstate = (struct r600_scissor_state *)atom;
-	struct pipe_scissor_state *state = &rstate->scissor;
-	unsigned offset = rstate->idx * 4 * 2;
+	struct r600_scissor_state *rstate = &rctx->scissor;
+	struct pipe_scissor_state *state;
+	uint32_t dirty_mask;
+	unsigned i, offset;
 
-	if (rctx->b.chip_class != R600 || rctx->scissor[0].enable) {
-		radeon_set_context_reg_seq(cs, R_028250_PA_SC_VPORT_SCISSOR_0_TL + offset, 2);
-		radeon_emit(cs, S_028240_TL_X(state->minx) | S_028240_TL_Y(state->miny) |
-				     S_028240_WINDOW_OFFSET_DISABLE(1));
-		radeon_emit(cs, S_028244_BR_X(state->maxx) | S_028244_BR_Y(state->maxy));
-	} else {
+	if (rctx->b.chip_class == R600 && !rctx->scissor.enable) {
 		radeon_set_context_reg_seq(cs, R_028250_PA_SC_VPORT_SCISSOR_0_TL, 2);
 		radeon_emit(cs, S_028240_TL_X(0) | S_028240_TL_Y(0) |
 				     S_028240_WINDOW_OFFSET_DISABLE(1));
 		radeon_emit(cs, S_028244_BR_X(8192) | S_028244_BR_Y(8192));
+		return;
+	}
+
+	dirty_mask = rstate->dirty_mask;
+	while (dirty_mask != 0)
+	{
+		i = u_bit_scan(&dirty_mask);
+		offset = i * 4 * 2;
+		state = &rstate->scissor[i];
+		radeon_set_context_reg_seq(cs, R_028250_PA_SC_VPORT_SCISSOR_0_TL + offset, 2);
+		radeon_emit(cs, S_028240_TL_X(state->minx) | S_028240_TL_Y(state->miny) |
+				     S_028240_WINDOW_OFFSET_DISABLE(1));
+		radeon_emit(cs, S_028244_BR_X(state->maxx) | S_028244_BR_Y(state->maxy));
 	}
+	rstate->dirty_mask = 0;
+	rstate->atom.num_dw = 0;
 }
 
 static void r600_set_scissor_states(struct pipe_context *ctx,
@@ -792,18 +803,18 @@ static void r600_set_scissor_states(struct pipe_context *ctx,
                                     const struct pipe_scissor_state *state)
 {
 	struct r600_context *rctx = (struct r600_context *)ctx;
+	struct r600_scissor_state *rstate = &rctx->scissor;
 	int i;
 
-	for (i = start_slot ; i < start_slot + num_scissors; i++) {
-		rctx->scissor[i].scissor = state[i - start_slot];
-	}
+	for (i = start_slot ; i < start_slot + num_scissors; i++)
+		rstate->scissor[i] = state[i - start_slot];
+	rstate->dirty_mask |= ((1 << num_scissors) - 1) << start_slot;
+	rstate->atom.num_dw = util_bitcount(rstate->dirty_mask) * 4;
 
-	if (rctx->b.chip_class == R600 && !rctx->scissor[0].enable)
+	if (rctx->b.chip_class == R600 && !rstate->enable)
 		return;
 
-	for (i = start_slot ; i < start_slot + num_scissors; i++) {
-		r600_mark_atom_dirty(rctx, &rctx->scissor[i].atom);
-	}
+	r600_mark_atom_dirty(rctx, &rstate->atom);
 }
 
 static struct r600_resource *r600_buffer_create_helper(struct r600_screen *rscreen,
@@ -3065,10 +3076,9 @@ void r600_init_state_functions(struct r600_context *rctx)
 	r600_init_atom(rctx, &rctx->dsa_state.atom, id++, r600_emit_cso_state, 0);
 	r600_init_atom(rctx, &rctx->poly_offset_state.atom, id++, r600_emit_polygon_offset, 6);
 	r600_init_atom(rctx, &rctx->rasterizer_state.atom, id++, r600_emit_cso_state, 0);
+	r600_init_atom(rctx, &rctx->scissor.atom, id++, r600_emit_scissor_state, 0);
 	for (i = 0;i < R600_MAX_VIEWPORTS; i++) {
-		r600_init_atom(rctx, &rctx->scissor[i].atom, id++, r600_emit_scissor_state, 4);
 		r600_init_atom(rctx, &rctx->viewport[i].atom, id++, r600_emit_viewport_state, 8);
-		rctx->scissor[i].idx = i;
 		rctx->viewport[i].idx = i;
 	}
 	r600_init_atom(rctx, &rctx->config_state.atom, id++, r600_emit_config_state, 3);
diff --git a/src/gallium/drivers/r600/r600_state_common.c b/src/gallium/drivers/r600/r600_state_common.c
index 24ed74b40d6..d2b6ebec6a4 100644
--- a/src/gallium/drivers/r600/r600_state_common.c
+++ b/src/gallium/drivers/r600/r600_state_common.c
@@ -372,9 +372,9 @@ static void r600_bind_rs_state(struct pipe_context *ctx, void *state)
 
 	/* Workaround for a missing scissor enable on r600. */
 	if (rctx->b.chip_class == R600 &&
-	    rs->scissor_enable != rctx->scissor[0].enable) {
-		rctx->scissor[0].enable = rs->scissor_enable;
-		r600_mark_atom_dirty(rctx, &rctx->scissor[0].atom);
+	    rs->scissor_enable != rctx->scissor.enable) {
+		rctx->scissor.enable = rs->scissor_enable;
+		r600_mark_atom_dirty(rctx, &rctx->scissor.atom);
 	}
 
 	/* Re-emit PA_SC_LINE_STIPPLE. */