freedreno/ir3: images can be arrays too

[mesa.git] / src / gallium / drivers / freedreno / a3xx / fd3_program.c

diff --git a/src/gallium/drivers/freedreno/a3xx/fd3_program.c b/src/gallium/drivers/freedreno/a3xx/fd3_program.c
index a0fa40d1c25be7c3e62f78cb4048071f5a4ea119..64eeb106e535ad8f95c97ade27a2dad7e170e7cd 100644 (file)
--- a/src/gallium/drivers/freedreno/a3xx/fd3_program.c
+++ b/src/gallium/drivers/freedreno/a3xx/fd3_program.c
@@ -28,6 +28,7 @@
 
 #include "pipe/p_state.h"
 #include "util/u_string.h"
+#include "util/u_math.h"
 #include "util/u_memory.h"
 #include "util/u_inlines.h"
 #include "util/u_format.h"
@@ -39,21 +40,13 @@
 #include "fd3_texture.h"
 #include "fd3_format.h"
 
-static void
-delete_shader_stateobj(struct fd3_shader_stateobj *so)
-{
-       ir3_shader_destroy(so->shader);
-       free(so);
-}
-
-static struct fd3_shader_stateobj *
+static struct ir3_shader *
 create_shader_stateobj(struct pipe_context *pctx, const struct pipe_shader_state *cso,
                enum shader_t type)
 {
-       struct fd3_shader_stateobj *so = CALLOC_STRUCT(fd3_shader_stateobj);
-       struct ir3_compiler *compiler = fd_context(pctx)->screen->compiler;
-       so->shader = ir3_shader_create(compiler, cso, type);
-       return so;
+       struct fd_context *ctx = fd_context(pctx);
+       struct ir3_compiler *compiler = ctx->screen->compiler;
+       return ir3_shader_create(compiler, cso, type, &ctx->debug);
 }
 
 static void *
@@ -66,8 +59,8 @@ fd3_fp_state_create(struct pipe_context *pctx,
 static void
 fd3_fp_state_delete(struct pipe_context *pctx, void *hwcso)
 {
-       struct fd3_shader_stateobj *so = hwcso;
-       delete_shader_stateobj(so);
+       struct ir3_shader *so = hwcso;
+       ir3_shader_destroy(so);
 }
 
 static void *
@@ -80,10 +73,24 @@ fd3_vp_state_create(struct pipe_context *pctx,
 static void
 fd3_vp_state_delete(struct pipe_context *pctx, void *hwcso)
 {
-       struct fd3_shader_stateobj *so = hwcso;
-       delete_shader_stateobj(so);
+       struct ir3_shader *so = hwcso;
+       ir3_shader_destroy(so);
 }
 
+bool
+fd3_needs_manual_clipping(const struct ir3_shader *shader,
+                                                 const struct pipe_rasterizer_state *rast)
+{
+       uint64_t outputs = ir3_shader_outputs(shader);
+
+       return (!rast->depth_clip ||
+                       util_bitcount(rast->clip_plane_enable) > 6 ||
+                       outputs & ((1ULL << VARYING_SLOT_CLIP_VERTEX) |
+                                          (1ULL << VARYING_SLOT_CLIP_DIST0) |
+                                          (1ULL << VARYING_SLOT_CLIP_DIST1)));
+}
+
+
 static void
 emit_shader(struct fd_ringbuffer *ring, const struct ir3_shader_variant *so)
 {
@@ -135,19 +142,12 @@ fd3_program_emit(struct fd_ringbuffer *ring, struct fd3_emit *emit,
        uint32_t fpbuffersz, vpbuffersz, fsoff;
        uint32_t pos_regid, posz_regid, psize_regid, color_regid[4] = {0};
        int constmode;
-       int i, j, k;
+       int i, j;
 
        debug_assert(nr <= ARRAY_SIZE(color_regid));
 
        vp = fd3_emit_get_vp(emit);
-
-       if (emit->key.binning_pass) {
-               /* use dummy stateobj to simplify binning vs non-binning: */
-               static const struct ir3_shader_variant binning_fp = {};
-               fp = &binning_fp;
-       } else {
-               fp = fd3_emit_get_fp(emit);
-       }
+       fp = fd3_emit_get_fp(emit);
 
        vsi = &vp->info;
        fsi = &fp->info;
@@ -266,45 +266,34 @@ fd3_program_emit(struct fd_ringbuffer *ring, struct fd3_emit *emit,
                        A3XX_SP_VS_PARAM_REG_PSIZEREGID(psize_regid) |
                        A3XX_SP_VS_PARAM_REG_TOTALVSOUTVAR(fp->varying_in));
 
-       for (i = 0, j = -1; (i < 8) && (j < (int)fp->inputs_count); i++) {
+       struct ir3_shader_linkage l = {0};
+       ir3_link_shaders(&l, vp, fp);
+
+       for (i = 0, j = 0; (i < 16) && (j < l.cnt); i++) {
                uint32_t reg = 0;
 
                OUT_PKT0(ring, REG_A3XX_SP_VS_OUT_REG(i), 1);
 
-               j = ir3_next_varying(fp, j);
-               if (j < fp->inputs_count) {
-                       k = ir3_find_output(vp, fp->inputs[j].slot);
-                       reg |= A3XX_SP_VS_OUT_REG_A_REGID(vp->outputs[k].regid);
-                       reg |= A3XX_SP_VS_OUT_REG_A_COMPMASK(fp->inputs[j].compmask);
-               }
+               reg |= A3XX_SP_VS_OUT_REG_A_REGID(l.var[j].regid);
+               reg |= A3XX_SP_VS_OUT_REG_A_COMPMASK(l.var[j].compmask);
+               j++;
 
-               j = ir3_next_varying(fp, j);
-               if (j < fp->inputs_count) {
-                       k = ir3_find_output(vp, fp->inputs[j].slot);
-                       reg |= A3XX_SP_VS_OUT_REG_B_REGID(vp->outputs[k].regid);
-                       reg |= A3XX_SP_VS_OUT_REG_B_COMPMASK(fp->inputs[j].compmask);
-               }
+               reg |= A3XX_SP_VS_OUT_REG_B_REGID(l.var[j].regid);
+               reg |= A3XX_SP_VS_OUT_REG_B_COMPMASK(l.var[j].compmask);
+               j++;
 
                OUT_RING(ring, reg);
        }
 
-       for (i = 0, j = -1; (i < 4) && (j < (int)fp->inputs_count); i++) {
+       for (i = 0, j = 0; (i < 8) && (j < l.cnt); i++) {
                uint32_t reg = 0;
 
                OUT_PKT0(ring, REG_A3XX_SP_VS_VPC_DST_REG(i), 1);
 
-               j = ir3_next_varying(fp, j);
-               if (j < fp->inputs_count)
-                       reg |= A3XX_SP_VS_VPC_DST_REG_OUTLOC0(fp->inputs[j].inloc);
-               j = ir3_next_varying(fp, j);
-               if (j < fp->inputs_count)
-                       reg |= A3XX_SP_VS_VPC_DST_REG_OUTLOC1(fp->inputs[j].inloc);
-               j = ir3_next_varying(fp, j);
-               if (j < fp->inputs_count)
-                       reg |= A3XX_SP_VS_VPC_DST_REG_OUTLOC2(fp->inputs[j].inloc);
-               j = ir3_next_varying(fp, j);
-               if (j < fp->inputs_count)
-                       reg |= A3XX_SP_VS_VPC_DST_REG_OUTLOC3(fp->inputs[j].inloc);
+               reg |= A3XX_SP_VS_VPC_DST_REG_OUTLOC0(l.var[j++].loc + 8);
+               reg |= A3XX_SP_VS_VPC_DST_REG_OUTLOC1(l.var[j++].loc + 8);
+               reg |= A3XX_SP_VS_VPC_DST_REG_OUTLOC2(l.var[j++].loc + 8);
+               reg |= A3XX_SP_VS_VPC_DST_REG_OUTLOC3(l.var[j++].loc + 8);
 
                OUT_RING(ring, reg);
        }
@@ -393,12 +382,9 @@ fd3_program_emit(struct fd_ringbuffer *ring, struct fd3_emit *emit,
                         */
                        unsigned compmask = fp->inputs[j].compmask;
 
-                       /* TODO might be cleaner to just +8 in SP_VS_VPC_DST_REG
-                        * instead.. rather than -8 everywhere else..
-                        */
-                       uint32_t inloc = fp->inputs[j].inloc - 8;
+                       uint32_t inloc = fp->inputs[j].inloc;
 
-                       if ((fp->inputs[j].interpolate == INTERP_QUALIFIER_FLAT) ||
+                       if ((fp->inputs[j].interpolate == INTERP_MODE_FLAT) ||
                                        (fp->inputs[j].rasterflat && emit->rasterflat)) {
                                uint32_t loc = inloc;