freedreno/{a4xx,a5xx}: switch to CP_LOAD_STATE4

The layout of CP_LOAD_STATE packet is slightly different on a4xx+.
Switch to the a4xx+ specific CP_LOAD_STATE4 to get the correct encoding.

Signed-off-by: Rob Clark <[email protected]>

5 files changed, 124 insertions, 127 deletions

diff --git a/src/gallium/drivers/freedreno/a4xx/fd4_emit.c b/src/gallium/drivers/freedreno/a4xx/fd4_emit.c
index 2f3e0a6981b..e07694a6ed1 100644
--- a/src/gallium/drivers/freedreno/a4xx/fd4_emit.c
+++ b/src/gallium/drivers/freedreno/a4xx/fd4_emit.c
@@ -45,9 +45,9 @@
 #include "fd4_format.h"
 #include "fd4_zsa.h"
 
-static const enum adreno_state_block sb[] = {
-	[SHADER_VERTEX]   = SB_VERT_SHADER,
-	[SHADER_FRAGMENT] = SB_FRAG_SHADER,
+static const enum a4xx_state_block sb[] = {
+	[SHADER_VERTEX]   = SB4_VS_SHADER,
+	[SHADER_FRAGMENT] = SB4_FS_SHADER,
 };
 
 /* regid:          base const register
@@ -60,31 +60,31 @@ fd4_emit_const(struct fd_ringbuffer *ring, enum shader_t type,
 		const uint32_t *dwords, struct pipe_resource *prsc)
 {
 	uint32_t i, sz;
-	enum adreno_state_src src;
+	enum a4xx_state_src src;
 
 	debug_assert((regid % 4) == 0);
 	debug_assert((sizedwords % 4) == 0);
 
 	if (prsc) {
 		sz = 0;
-		src = 0x2;  // TODO ??
+		src = SS4_INDIRECT;
 	} else {
 		sz = sizedwords;
-		src = SS_DIRECT;
+		src = SS4_DIRECT;
 	}
 
-	OUT_PKT3(ring, CP_LOAD_STATE, 2 + sz);
-	OUT_RING(ring, CP_LOAD_STATE_0_DST_OFF(regid/4) |
-			CP_LOAD_STATE_0_STATE_SRC(src) |
-			CP_LOAD_STATE_0_STATE_BLOCK(sb[type]) |
-			CP_LOAD_STATE_0_NUM_UNIT(sizedwords/4));
+	OUT_PKT3(ring, CP_LOAD_STATE4, 2 + sz);
+	OUT_RING(ring, CP_LOAD_STATE4_0_DST_OFF(regid/4) |
+			CP_LOAD_STATE4_0_STATE_SRC(src) |
+			CP_LOAD_STATE4_0_STATE_BLOCK(sb[type]) |
+			CP_LOAD_STATE4_0_NUM_UNIT(sizedwords/4));
 	if (prsc) {
 		struct fd_bo *bo = fd_resource(prsc)->bo;
 		OUT_RELOC(ring, bo, offset,
-				CP_LOAD_STATE_1_STATE_TYPE(ST_CONSTANTS), 0);
+				CP_LOAD_STATE4_1_STATE_TYPE(ST4_CONSTANTS), 0);
 	} else {
-		OUT_RING(ring, CP_LOAD_STATE_1_EXT_SRC_ADDR(0) |
-				CP_LOAD_STATE_1_STATE_TYPE(ST_CONSTANTS));
+		OUT_RING(ring, CP_LOAD_STATE4_1_EXT_SRC_ADDR(0) |
+				CP_LOAD_STATE4_1_STATE_TYPE(ST4_CONSTANTS));
 		dwords = (uint32_t *)&((uint8_t *)dwords)[offset];
 	}
 	for (i = 0; i < sz; i++) {
@@ -101,13 +101,13 @@ fd4_emit_const_bo(struct fd_ringbuffer *ring, enum shader_t type, boolean write,
 
 	debug_assert((regid % 4) == 0);
 
-	OUT_PKT3(ring, CP_LOAD_STATE, 2 + anum);
-	OUT_RING(ring, CP_LOAD_STATE_0_DST_OFF(regid/4) |
-			CP_LOAD_STATE_0_STATE_SRC(SS_DIRECT) |
-			CP_LOAD_STATE_0_STATE_BLOCK(sb[type]) |
-			CP_LOAD_STATE_0_NUM_UNIT(anum/4));
-	OUT_RING(ring, CP_LOAD_STATE_1_EXT_SRC_ADDR(0) |
-			CP_LOAD_STATE_1_STATE_TYPE(ST_CONSTANTS));
+	OUT_PKT3(ring, CP_LOAD_STATE4, 2 + anum);
+	OUT_RING(ring, CP_LOAD_STATE4_0_DST_OFF(regid/4) |
+			CP_LOAD_STATE4_0_STATE_SRC(SS4_DIRECT) |
+			CP_LOAD_STATE4_0_STATE_BLOCK(sb[type]) |
+			CP_LOAD_STATE4_0_NUM_UNIT(anum/4));
+	OUT_RING(ring, CP_LOAD_STATE4_1_EXT_SRC_ADDR(0) |
+			CP_LOAD_STATE4_1_STATE_TYPE(ST4_CONSTANTS));
 
 	for (i = 0; i < num; i++) {
 		if (prscs[i]) {
@@ -127,12 +127,12 @@ fd4_emit_const_bo(struct fd_ringbuffer *ring, enum shader_t type, boolean write,
 
 static void
 emit_textures(struct fd_context *ctx, struct fd_ringbuffer *ring,
-		enum adreno_state_block sb, struct fd_texture_stateobj *tex,
+		enum a4xx_state_block sb, struct fd_texture_stateobj *tex,
 		const struct ir3_shader_variant *v)
 {
 	static const uint32_t bcolor_reg[] = {
-			[SB_VERT_TEX] = REG_A4XX_TPL1_TP_VS_BORDER_COLOR_BASE_ADDR,
-			[SB_FRAG_TEX] = REG_A4XX_TPL1_TP_FS_BORDER_COLOR_BASE_ADDR,
+			[SB4_VS_TEX] = REG_A4XX_TPL1_TP_VS_BORDER_COLOR_BASE_ADDR,
+			[SB4_FS_TEX] = REG_A4XX_TPL1_TP_FS_BORDER_COLOR_BASE_ADDR,
 	};
 	struct fd4_context *fd4_ctx = fd4_context(ctx);
 	bool needs_border = false;
@@ -148,13 +148,13 @@ emit_textures(struct fd_context *ctx, struct fd_ringbuffer *ring,
 		num_samplers = align(tex->num_samplers, 2);
 
 		/* output sampler state: */
-		OUT_PKT3(ring, CP_LOAD_STATE, 2 + (2 * num_samplers));
-		OUT_RING(ring, CP_LOAD_STATE_0_DST_OFF(0) |
-				CP_LOAD_STATE_0_STATE_SRC(SS_DIRECT) |
-				CP_LOAD_STATE_0_STATE_BLOCK(sb) |
-				CP_LOAD_STATE_0_NUM_UNIT(num_samplers));
-		OUT_RING(ring, CP_LOAD_STATE_1_STATE_TYPE(ST_SHADER) |
-				CP_LOAD_STATE_1_EXT_SRC_ADDR(0));
+		OUT_PKT3(ring, CP_LOAD_STATE4, 2 + (2 * num_samplers));
+		OUT_RING(ring, CP_LOAD_STATE4_0_DST_OFF(0) |
+				CP_LOAD_STATE4_0_STATE_SRC(SS4_DIRECT) |
+				CP_LOAD_STATE4_0_STATE_BLOCK(sb) |
+				CP_LOAD_STATE4_0_NUM_UNIT(num_samplers));
+		OUT_RING(ring, CP_LOAD_STATE4_1_STATE_TYPE(ST4_SHADER) |
+				CP_LOAD_STATE4_1_EXT_SRC_ADDR(0));
 		for (i = 0; i < tex->num_samplers; i++) {
 			static const struct fd4_sampler_stateobj dummy_sampler = {};
 			const struct fd4_sampler_stateobj *sampler = tex->samplers[i] ?
@@ -176,13 +176,13 @@ emit_textures(struct fd_context *ctx, struct fd_ringbuffer *ring,
 		unsigned num_textures = tex->num_textures + v->astc_srgb.count;
 
 		/* emit texture state: */
-		OUT_PKT3(ring, CP_LOAD_STATE, 2 + (8 * num_textures));
-		OUT_RING(ring, CP_LOAD_STATE_0_DST_OFF(0) |
-				CP_LOAD_STATE_0_STATE_SRC(SS_DIRECT) |
-				CP_LOAD_STATE_0_STATE_BLOCK(sb) |
-				CP_LOAD_STATE_0_NUM_UNIT(num_textures));
-		OUT_RING(ring, CP_LOAD_STATE_1_STATE_TYPE(ST_CONSTANTS) |
-				CP_LOAD_STATE_1_EXT_SRC_ADDR(0));
+		OUT_PKT3(ring, CP_LOAD_STATE4, 2 + (8 * num_textures));
+		OUT_RING(ring, CP_LOAD_STATE4_0_DST_OFF(0) |
+				CP_LOAD_STATE4_0_STATE_SRC(SS4_DIRECT) |
+				CP_LOAD_STATE4_0_STATE_BLOCK(sb) |
+				CP_LOAD_STATE4_0_NUM_UNIT(num_textures));
+		OUT_RING(ring, CP_LOAD_STATE4_1_STATE_TYPE(ST4_CONSTANTS) |
+				CP_LOAD_STATE4_1_EXT_SRC_ADDR(0));
 		for (i = 0; i < tex->num_textures; i++) {
 			static const struct fd4_pipe_sampler_view dummy_view = {};
 			const struct fd4_pipe_sampler_view *view = tex->textures[i] ?
@@ -267,13 +267,13 @@ fd4_emit_gmem_restore_tex(struct fd_ringbuffer *ring, unsigned nr_bufs,
 	}
 
 	/* output sampler state: */
-	OUT_PKT3(ring, CP_LOAD_STATE, 2 + (2 * nr_bufs));
-	OUT_RING(ring, CP_LOAD_STATE_0_DST_OFF(0) |
-			CP_LOAD_STATE_0_STATE_SRC(SS_DIRECT) |
-			CP_LOAD_STATE_0_STATE_BLOCK(SB_FRAG_TEX) |
-			CP_LOAD_STATE_0_NUM_UNIT(nr_bufs));
-	OUT_RING(ring, CP_LOAD_STATE_1_STATE_TYPE(ST_SHADER) |
-			CP_LOAD_STATE_1_EXT_SRC_ADDR(0));
+	OUT_PKT3(ring, CP_LOAD_STATE4, 2 + (2 * nr_bufs));
+	OUT_RING(ring, CP_LOAD_STATE4_0_DST_OFF(0) |
+			CP_LOAD_STATE4_0_STATE_SRC(SS4_DIRECT) |
+			CP_LOAD_STATE4_0_STATE_BLOCK(SB4_FS_TEX) |
+			CP_LOAD_STATE4_0_NUM_UNIT(nr_bufs));
+	OUT_RING(ring, CP_LOAD_STATE4_1_STATE_TYPE(ST4_SHADER) |
+			CP_LOAD_STATE4_1_EXT_SRC_ADDR(0));
 	for (i = 0; i < nr_bufs; i++) {
 		OUT_RING(ring, A4XX_TEX_SAMP_0_XY_MAG(A4XX_TEX_NEAREST) |
 				A4XX_TEX_SAMP_0_XY_MIN(A4XX_TEX_NEAREST) |
@@ -284,13 +284,13 @@ fd4_emit_gmem_restore_tex(struct fd_ringbuffer *ring, unsigned nr_bufs,
 	}
 
 	/* emit texture state: */
-	OUT_PKT3(ring, CP_LOAD_STATE, 2 + (8 * nr_bufs));
-	OUT_RING(ring, CP_LOAD_STATE_0_DST_OFF(0) |
-			CP_LOAD_STATE_0_STATE_SRC(SS_DIRECT) |
-			CP_LOAD_STATE_0_STATE_BLOCK(SB_FRAG_TEX) |
-			CP_LOAD_STATE_0_NUM_UNIT(nr_bufs));
-	OUT_RING(ring, CP_LOAD_STATE_1_STATE_TYPE(ST_CONSTANTS) |
-			CP_LOAD_STATE_1_EXT_SRC_ADDR(0));
+	OUT_PKT3(ring, CP_LOAD_STATE4, 2 + (8 * nr_bufs));
+	OUT_RING(ring, CP_LOAD_STATE4_0_DST_OFF(0) |
+			CP_LOAD_STATE4_0_STATE_SRC(SS4_DIRECT) |
+			CP_LOAD_STATE4_0_STATE_BLOCK(SB4_FS_TEX) |
+			CP_LOAD_STATE4_0_NUM_UNIT(nr_bufs));
+	OUT_RING(ring, CP_LOAD_STATE4_1_STATE_TYPE(ST4_CONSTANTS) |
+			CP_LOAD_STATE4_1_EXT_SRC_ADDR(0));
 	for (i = 0; i < nr_bufs; i++) {
 		if (bufs[i]) {
 			struct fd_resource *rsc = fd_resource(bufs[i]->texture);
@@ -747,14 +747,14 @@ fd4_emit_state(struct fd_context *ctx, struct fd_ringbuffer *ring,
 
 	if (dirty & FD_DIRTY_VERTTEX) {
 		if (vp->has_samp)
-			emit_textures(ctx, ring, SB_VERT_TEX, &ctx->verttex, vp);
+			emit_textures(ctx, ring, SB4_VS_TEX, &ctx->verttex, vp);
 		else
 			dirty &= ~FD_DIRTY_VERTTEX;
 	}
 
 	if (dirty & FD_DIRTY_FRAGTEX) {
 		if (fp->has_samp)
-			emit_textures(ctx, ring, SB_FRAG_TEX, &ctx->fragtex, fp);
+			emit_textures(ctx, ring, SB4_FS_TEX, &ctx->fragtex, fp);
 		else
 			dirty &= ~FD_DIRTY_FRAGTEX;
 	}
diff --git a/src/gallium/drivers/freedreno/a4xx/fd4_program.c b/src/gallium/drivers/freedreno/a4xx/fd4_program.c
index 3e751253379..d9f13af1d9b 100644
--- a/src/gallium/drivers/freedreno/a4xx/fd4_program.c
+++ b/src/gallium/drivers/freedreno/a4xx/fd4_program.c
@@ -94,32 +94,32 @@ emit_shader(struct fd_ringbuffer *ring, const struct ir3_shader_variant *so)
 	uint32_t i, sz, *bin;
 
 	if (so->type == SHADER_VERTEX) {
-		sb = SB_VERT_SHADER;
+		sb = SB4_VS_SHADER;
 	} else {
-		sb = SB_FRAG_SHADER;
+		sb = SB4_FS_SHADER;
 	}
 
 	if (fd_mesa_debug & FD_DBG_DIRECT) {
 		sz = si->sizedwords;
-		src = SS_DIRECT;
+		src = SS4_DIRECT;
 		bin = fd_bo_map(so->bo);
 	} else {
 		sz = 0;
-		src = 2;  // enums different on a4xx..
+		src = SS4_INDIRECT;
 		bin = NULL;
 	}
 
-	OUT_PKT3(ring, CP_LOAD_STATE, 2 + sz);
-	OUT_RING(ring, CP_LOAD_STATE_0_DST_OFF(0) |
-			CP_LOAD_STATE_0_STATE_SRC(src) |
-			CP_LOAD_STATE_0_STATE_BLOCK(sb) |
-			CP_LOAD_STATE_0_NUM_UNIT(so->instrlen));
+	OUT_PKT3(ring, CP_LOAD_STATE4, 2 + sz);
+	OUT_RING(ring, CP_LOAD_STATE4_0_DST_OFF(0) |
+			CP_LOAD_STATE4_0_STATE_SRC(src) |
+			CP_LOAD_STATE4_0_STATE_BLOCK(sb) |
+			CP_LOAD_STATE4_0_NUM_UNIT(so->instrlen));
 	if (bin) {
-		OUT_RING(ring, CP_LOAD_STATE_1_EXT_SRC_ADDR(0) |
-				CP_LOAD_STATE_1_STATE_TYPE(ST_SHADER));
+		OUT_RING(ring, CP_LOAD_STATE4_1_EXT_SRC_ADDR(0) |
+				CP_LOAD_STATE4_1_STATE_TYPE(ST4_SHADER));
 	} else {
 		OUT_RELOC(ring, so->bo, 0,
-				CP_LOAD_STATE_1_STATE_TYPE(ST_SHADER), 0);
+				CP_LOAD_STATE4_1_STATE_TYPE(ST4_SHADER), 0);
 	}
 
 	/* for how clever coverity is, it is sometimes rather dull, and
diff --git a/src/gallium/drivers/freedreno/a5xx/fd5_emit.c b/src/gallium/drivers/freedreno/a5xx/fd5_emit.c
index a12b143d8fa..cbb86d890c0 100644
--- a/src/gallium/drivers/freedreno/a5xx/fd5_emit.c
+++ b/src/gallium/drivers/freedreno/a5xx/fd5_emit.c
@@ -43,9 +43,9 @@
 #include "fd5_format.h"
 #include "fd5_zsa.h"
 
-static const enum adreno_state_block sb[] = {
-	[SHADER_VERTEX]   = SB_VERT_SHADER,
-	[SHADER_FRAGMENT] = SB_FRAG_SHADER,
+static const enum a4xx_state_block sb[] = {
+	[SHADER_VERTEX]   = SB4_VS_SHADER,
+	[SHADER_FRAGMENT] = SB4_FS_SHADER,
 };
 
 /* regid:          base const register
@@ -58,32 +58,32 @@ fd5_emit_const(struct fd_ringbuffer *ring, enum shader_t type,
 		const uint32_t *dwords, struct pipe_resource *prsc)
 {
 	uint32_t i, sz;
-	enum adreno_state_src src;
+	enum a4xx_state_src src;
 
 	debug_assert((regid % 4) == 0);
 	debug_assert((sizedwords % 4) == 0);
 
 	if (prsc) {
 		sz = 0;
-		src = 0x2;  // TODO ??
+		src = SS4_INDIRECT;
 	} else {
 		sz = sizedwords;
-		src = SS_DIRECT;
+		src = SS4_DIRECT;
 	}
 
-	OUT_PKT7(ring, CP_LOAD_STATE, 3 + sz);
-	OUT_RING(ring, CP_LOAD_STATE_0_DST_OFF(regid/4) |
-			CP_LOAD_STATE_0_STATE_SRC(src) |
-			CP_LOAD_STATE_0_STATE_BLOCK(sb[type]) |
-			CP_LOAD_STATE_0_NUM_UNIT(sizedwords/4));
+	OUT_PKT7(ring, CP_LOAD_STATE4, 3 + sz);
+	OUT_RING(ring, CP_LOAD_STATE4_0_DST_OFF(regid/4) |
+			CP_LOAD_STATE4_0_STATE_SRC(src) |
+			CP_LOAD_STATE4_0_STATE_BLOCK(sb[type]) |
+			CP_LOAD_STATE4_0_NUM_UNIT(sizedwords/4));
 	if (prsc) {
 		struct fd_bo *bo = fd_resource(prsc)->bo;
 		OUT_RELOC(ring, bo, offset,
-				CP_LOAD_STATE_1_STATE_TYPE(ST_CONSTANTS), 0);
+				CP_LOAD_STATE4_1_STATE_TYPE(ST4_CONSTANTS), 0);
 	} else {
-		OUT_RING(ring, CP_LOAD_STATE_1_EXT_SRC_ADDR(0) |
-				CP_LOAD_STATE_1_STATE_TYPE(ST_CONSTANTS));
-		OUT_RING(ring, CP_LOAD_STATE_2_EXT_SRC_ADDR_HI(0));
+		OUT_RING(ring, CP_LOAD_STATE4_1_EXT_SRC_ADDR(0) |
+				CP_LOAD_STATE4_1_STATE_TYPE(ST4_CONSTANTS));
+		OUT_RING(ring, CP_LOAD_STATE4_2_EXT_SRC_ADDR_HI(0));
 		dwords = (uint32_t *)&((uint8_t *)dwords)[offset];
 	}
 	for (i = 0; i < sz; i++) {
@@ -100,14 +100,14 @@ fd5_emit_const_bo(struct fd_ringbuffer *ring, enum shader_t type, boolean write,
 
 	debug_assert((regid % 4) == 0);
 
-	OUT_PKT7(ring, CP_LOAD_STATE, 3 + (2 * anum));
-	OUT_RING(ring, CP_LOAD_STATE_0_DST_OFF(regid/4) |
-			CP_LOAD_STATE_0_STATE_SRC(SS_DIRECT) |
-			CP_LOAD_STATE_0_STATE_BLOCK(sb[type]) |
-			CP_LOAD_STATE_0_NUM_UNIT(anum/2));
-	OUT_RING(ring, CP_LOAD_STATE_1_EXT_SRC_ADDR(0) |
-			CP_LOAD_STATE_1_STATE_TYPE(ST_CONSTANTS));
-	OUT_RING(ring, CP_LOAD_STATE_2_EXT_SRC_ADDR_HI(0));
+	OUT_PKT7(ring, CP_LOAD_STATE4, 3 + (2 * anum));
+	OUT_RING(ring, CP_LOAD_STATE4_0_DST_OFF(regid/4) |
+			CP_LOAD_STATE4_0_STATE_SRC(SS4_DIRECT) |
+			CP_LOAD_STATE4_0_STATE_BLOCK(sb[type]) |
+			CP_LOAD_STATE4_0_NUM_UNIT(anum/2));
+	OUT_RING(ring, CP_LOAD_STATE4_1_EXT_SRC_ADDR(0) |
+			CP_LOAD_STATE4_1_STATE_TYPE(ST4_CONSTANTS));
+	OUT_RING(ring, CP_LOAD_STATE4_2_EXT_SRC_ADDR_HI(0));
 
 	for (i = 0; i < num; i++) {
 		if (prscs[i]) {
@@ -276,22 +276,22 @@ emit_border_color(struct fd_context *ctx, struct fd_ringbuffer *ring)
 
 static bool
 emit_textures(struct fd_context *ctx, struct fd_ringbuffer *ring,
-		enum adreno_state_block sb, struct fd_texture_stateobj *tex)
+		enum a4xx_state_block sb, struct fd_texture_stateobj *tex)
 {
 	bool needs_border = false;
-	unsigned bcolor_offset = (sb == SB_FRAG_TEX) ? ctx->verttex.num_samplers : 0;
+	unsigned bcolor_offset = (sb == SB4_FS_TEX) ? ctx->verttex.num_samplers : 0;
 	unsigned i;
 
 	if (tex->num_samplers > 0) {
 		/* output sampler state: */
-		OUT_PKT7(ring, CP_LOAD_STATE, 3 + (4 * tex->num_samplers));
-		OUT_RING(ring, CP_LOAD_STATE_0_DST_OFF(0) |
-				CP_LOAD_STATE_0_STATE_SRC(SS_DIRECT) |
-				CP_LOAD_STATE_0_STATE_BLOCK(sb) |
-				CP_LOAD_STATE_0_NUM_UNIT(tex->num_samplers));
-		OUT_RING(ring, CP_LOAD_STATE_1_STATE_TYPE(ST_SHADER) |
-				CP_LOAD_STATE_1_EXT_SRC_ADDR(0));
-		OUT_RING(ring, CP_LOAD_STATE_2_EXT_SRC_ADDR_HI(0));
+		OUT_PKT7(ring, CP_LOAD_STATE4, 3 + (4 * tex->num_samplers));
+		OUT_RING(ring, CP_LOAD_STATE4_0_DST_OFF(0) |
+				CP_LOAD_STATE4_0_STATE_SRC(SS4_DIRECT) |
+				CP_LOAD_STATE4_0_STATE_BLOCK(sb) |
+				CP_LOAD_STATE4_0_NUM_UNIT(tex->num_samplers));
+		OUT_RING(ring, CP_LOAD_STATE4_1_STATE_TYPE(ST4_SHADER) |
+				CP_LOAD_STATE4_1_EXT_SRC_ADDR(0));
+		OUT_RING(ring, CP_LOAD_STATE4_2_EXT_SRC_ADDR_HI(0));
 		for (i = 0; i < tex->num_samplers; i++) {
 			static const struct fd5_sampler_stateobj dummy_sampler = {};
 			const struct fd5_sampler_stateobj *sampler = tex->samplers[i] ?
@@ -311,14 +311,14 @@ emit_textures(struct fd_context *ctx, struct fd_ringbuffer *ring,
 		unsigned num_textures = tex->num_textures;
 
 		/* emit texture state: */
-		OUT_PKT7(ring, CP_LOAD_STATE, 3 + (12 * num_textures));
-		OUT_RING(ring, CP_LOAD_STATE_0_DST_OFF(0) |
-				CP_LOAD_STATE_0_STATE_SRC(SS_DIRECT) |
-				CP_LOAD_STATE_0_STATE_BLOCK(sb) |
-				CP_LOAD_STATE_0_NUM_UNIT(num_textures));
-		OUT_RING(ring, CP_LOAD_STATE_1_STATE_TYPE(ST_CONSTANTS) |
-				CP_LOAD_STATE_1_EXT_SRC_ADDR(0));
-		OUT_RING(ring, CP_LOAD_STATE_2_EXT_SRC_ADDR_HI(0));
+		OUT_PKT7(ring, CP_LOAD_STATE4, 3 + (12 * num_textures));
+		OUT_RING(ring, CP_LOAD_STATE4_0_DST_OFF(0) |
+				CP_LOAD_STATE4_0_STATE_SRC(SS4_DIRECT) |
+				CP_LOAD_STATE4_0_STATE_BLOCK(sb) |
+				CP_LOAD_STATE4_0_NUM_UNIT(num_textures));
+		OUT_RING(ring, CP_LOAD_STATE4_1_STATE_TYPE(ST4_CONSTANTS) |
+				CP_LOAD_STATE4_1_EXT_SRC_ADDR(0));
+		OUT_RING(ring, CP_LOAD_STATE4_2_EXT_SRC_ADDR_HI(0));
 		for (i = 0; i < tex->num_textures; i++) {
 			static const struct fd5_pipe_sampler_view dummy_view = {};
 			const struct fd5_pipe_sampler_view *view = tex->textures[i] ?
@@ -653,7 +653,7 @@ fd5_emit_state(struct fd_context *ctx, struct fd_ringbuffer *ring,
 
 	if (dirty & FD_DIRTY_VERTTEX) {
 		if (vp->has_samp) {
-			needs_border |= emit_textures(ctx, ring, SB_VERT_TEX, &ctx->verttex);
+			needs_border |= emit_textures(ctx, ring, SB4_VS_TEX, &ctx->verttex);
 			OUT_PKT4(ring, REG_A5XX_TPL1_VS_TEX_COUNT, 1);
 			OUT_RING(ring, ctx->verttex.num_textures);
 		} else {
@@ -663,7 +663,7 @@ fd5_emit_state(struct fd_context *ctx, struct fd_ringbuffer *ring,
 
 	if (dirty & FD_DIRTY_FRAGTEX) {
 		if (fp->has_samp) {
-			needs_border |= emit_textures(ctx, ring, SB_FRAG_TEX, &ctx->fragtex);
+			needs_border |= emit_textures(ctx, ring, SB4_FS_TEX, &ctx->fragtex);
 			OUT_PKT4(ring, REG_A5XX_TPL1_FS_TEX_COUNT, 1);
 			OUT_RING(ring, ctx->fragtex.num_textures);
 		} else {
diff --git a/src/gallium/drivers/freedreno/a5xx/fd5_format.h b/src/gallium/drivers/freedreno/a5xx/fd5_format.h
index ca410eba451..b052aa52960 100644
--- a/src/gallium/drivers/freedreno/a5xx/fd5_format.h
+++ b/src/gallium/drivers/freedreno/a5xx/fd5_format.h
@@ -31,9 +31,6 @@
 
 #include "a5xx.xml.h"
 
-// XXX temp hack
-#define CP_LOAD_STATE_2_EXT_SRC_ADDR_HI CP_LOAD_STATE4_2_EXT_SRC_ADDR_HI
-
 enum a5xx_vtx_fmt fd5_pipe2vtx(enum pipe_format format);
 enum a5xx_tex_fmt fd5_pipe2tex(enum pipe_format format);
 enum a5xx_color_fmt fd5_pipe2color(enum pipe_format format);
diff --git a/src/gallium/drivers/freedreno/a5xx/fd5_program.c b/src/gallium/drivers/freedreno/a5xx/fd5_program.c
index 62f2d0e82f4..b8722a5d038 100644
--- a/src/gallium/drivers/freedreno/a5xx/fd5_program.c
+++ b/src/gallium/drivers/freedreno/a5xx/fd5_program.c
@@ -88,38 +88,38 @@ static void
 emit_shader(struct fd_ringbuffer *ring, const struct ir3_shader_variant *so)
 {
 	const struct ir3_info *si = &so->info;
-	enum adreno_state_block sb;
-	enum adreno_state_src src;
+	enum a4xx_state_block sb;
+	enum a4xx_state_src src;
 	uint32_t i, sz, *bin;
 
 	if (so->type == SHADER_VERTEX) {
-		sb = SB_VERT_SHADER;
+		sb = SB4_VS_SHADER;
 	} else {
-		sb = SB_FRAG_SHADER;
+		sb = SB4_FS_SHADER;
 	}
 
 	if (fd_mesa_debug & FD_DBG_DIRECT) {
 		sz = si->sizedwords;
-		src = SS_DIRECT;
+		src = SS4_DIRECT;
 		bin = fd_bo_map(so->bo);
 	} else {
 		sz = 0;
-		src = 2;  // enums different on a5xx..
+		src = SS4_INDIRECT;
 		bin = NULL;
 	}
 
-	OUT_PKT7(ring, CP_LOAD_STATE, 3 + sz);
-	OUT_RING(ring, CP_LOAD_STATE_0_DST_OFF(0) |
-			CP_LOAD_STATE_0_STATE_SRC(src) |
-			CP_LOAD_STATE_0_STATE_BLOCK(sb) |
-			CP_LOAD_STATE_0_NUM_UNIT(so->instrlen));
+	OUT_PKT7(ring, CP_LOAD_STATE4, 3 + sz);
+	OUT_RING(ring, CP_LOAD_STATE4_0_DST_OFF(0) |
+			CP_LOAD_STATE4_0_STATE_SRC(src) |
+			CP_LOAD_STATE4_0_STATE_BLOCK(sb) |
+			CP_LOAD_STATE4_0_NUM_UNIT(so->instrlen));
 	if (bin) {
-		OUT_RING(ring, CP_LOAD_STATE_1_EXT_SRC_ADDR(0) |
-				CP_LOAD_STATE_1_STATE_TYPE(ST_SHADER));
-		OUT_RING(ring, CP_LOAD_STATE_2_EXT_SRC_ADDR_HI(0));
+		OUT_RING(ring, CP_LOAD_STATE4_1_EXT_SRC_ADDR(0) |
+				CP_LOAD_STATE4_1_STATE_TYPE(ST4_SHADER));
+		OUT_RING(ring, CP_LOAD_STATE4_2_EXT_SRC_ADDR_HI(0));
 	} else {
 		OUT_RELOC(ring, so->bo, 0,
-				CP_LOAD_STATE_1_STATE_TYPE(ST_SHADER), 0);
+				CP_LOAD_STATE4_1_STATE_TYPE(ST4_SHADER), 0);
 	}
 
 	/* for how clever coverity is, it is sometimes rather dull, and