v3d,v3dv: stop trying to force 16-bit TMU output for shadow comparisons

In V3D we were doing this incorrectly by peeking into the sampler state
unconditionally, which is not correct if the TMU operations don't use
sampler state at all (like PBOs). This was causing us to fail the second
test in this sequence when both tests run back back to back in the same
process:

dEQP-GLES3.functional.texture.shadow.2d.linear.greater_or_equal_depth_component32f
dEQP-GLES3.functional.texture.specification.teximage2d_pbo.rg32f_cube

Here, the first test would setup sampler state for shadow comparisons and
the second test would setup a PBO upload, which would incorrectly pick
up the sampler state to decide about the TMU output size for the PBO
operation.

In V3DV we were doing this right looking through each texture/sampler
instruction and checking if they all involved shadow comparisons or had
relaxed precission, defaulting to 32-bit otherwise.

This special-casing for shadow comparisons also leaks from drivers
into the compiler where we are forced to emit some pieces of sampler
state for 32-bit outputs, so we had to special-case shadow instructions
there as well and we also had a fix for CS textures not having correct
sampler state representing shadow operations too. Finally,
we also had  at least a couple of bugs where forcing 32-bit TMU output
through V3D_DEBUG wasn't correctly forcing shadow comparisons to actually
be 32-bit in all the right places, leading to visual bugs with the
option enabled (Sponza being one example of this). This change eliminates
all of these issues.

Finally, the performance improvement observed from special casing shadow
comparison is negligible, and in specific scenarios it can even be
detrimental to performance due to increased register pressure (Sponza with
PCF filtering set to 4 is an example of this again).

Fixes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/8684
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22284>

src/broadcom/compiler/v3d40_tex.c

@@ -236,8 +236,7 @@ v3d40_vir_emit_tex(struct v3d_compile *c, nir_tex_instr *instr)
          * parameter if the output is 32 bit
          */
         bool output_type_32_bit =
-                c->key->sampler[sampler_idx].return_size == 32 &&
-                !instr->is_shadow;
+                c->key->sampler[sampler_idx].return_size == 32;
 
         struct V3D41_TMU_CONFIG_PARAMETER_0 p0_unpacked = {
         };

src/broadcom/compiler/vir.c

@@ -643,21 +643,6 @@ v3d_lower_nir(struct v3d_compile *c)
                 }
         }
 
-        /* CS textures may not have return_size reflecting the shadow state. */
-        nir_foreach_uniform_variable(var, c->s) {
-                const struct glsl_type *type = glsl_without_array(var->type);
-                unsigned array_len = MAX2(glsl_get_length(var->type), 1);
-
-                if (!glsl_type_is_sampler(type) ||
-                    !glsl_sampler_type_is_shadow(type))
-                        continue;
-
-                for (int i = 0; i < array_len; i++) {
-                        tex_options.lower_tex_packing[var->data.binding + i] =
-                                nir_lower_tex_packing_16;
-                }
-        }
-
         NIR_PASS(_, c->s, nir_lower_tex, &tex_options);
         NIR_PASS(_, c->s, nir_lower_system_values);
 

src/broadcom/vulkan/v3dv_pipeline.c

@@ -681,7 +681,7 @@ lower_tex_src(nir_builder *b,
    else  if (V3D_DBG(TMU_32BIT))
       return_size = 32;
    else
-      return_size = relaxed_precision || instr->is_shadow ? 16 : 32;
+      return_size = relaxed_precision ? 16 : 32;
 
    struct v3dv_descriptor_map *map =
       pipeline_get_descriptor_map(state->pipeline, binding_layout->type,

src/gallium/drivers/v3d/v3d_context.h

@@ -754,8 +754,7 @@ bool v3d_tex_format_supported(const struct v3d_device_info *devinfo,
 uint8_t v3d_get_rt_format(const struct v3d_device_info *devinfo, enum pipe_format f);
 uint8_t v3d_get_tex_format(const struct v3d_device_info *devinfo, enum pipe_format f);
 uint8_t v3d_get_tex_return_size(const struct v3d_device_info *devinfo,
-                                enum pipe_format f,
-                                enum pipe_tex_compare compare);
+                                enum pipe_format f);
 uint8_t v3d_get_tex_return_channels(const struct v3d_device_info *devinfo,
                                     enum pipe_format f);
 const uint8_t *v3d_get_format_swizzle(const struct v3d_device_info *devinfo,

src/gallium/drivers/v3d/v3d_formats.c

@@ -95,7 +95,7 @@ v3d_get_tex_format(const struct v3d_device_info *devinfo, enum pipe_format f)
 
 uint8_t
 v3d_get_tex_return_size(const struct v3d_device_info *devinfo,
-                        enum pipe_format f, enum pipe_tex_compare compare)
+                        enum pipe_format f)
 {
         const struct v3d_format *vf = get_format(devinfo, f);
 
@@ -108,9 +108,6 @@ v3d_get_tex_return_size(const struct v3d_device_info *devinfo,
         if (V3D_DBG(TMU_32BIT))
                 return 32;
 
-        if (compare == PIPE_TEX_COMPARE_R_TO_TEXTURE)
-                return 16;
-
         return vf->return_size;
 }
 

src/gallium/drivers/v3d/v3d_program.c

@@ -467,22 +467,12 @@ v3d_setup_shared_key(struct v3d_context *v3d, struct v3d_key *key,
         for (int i = 0; i < texstate->num_textures; i++) {
                 struct pipe_sampler_view *sampler = texstate->textures[i];
                 struct v3d_sampler_view *v3d_sampler = v3d_sampler_view(sampler);
-                struct pipe_sampler_state *sampler_state =
-                        texstate->samplers[i];
 
                 if (!sampler)
                         continue;
 
-                assert(sampler->target == PIPE_BUFFER || sampler_state);
-
-                unsigned compare_mode = sampler_state ?
-                        sampler_state->compare_mode :
-                        PIPE_TEX_COMPARE_NONE;
-
                 key->sampler[i].return_size =
-                        v3d_get_tex_return_size(devinfo,
-                                                sampler->format,
-                                                compare_mode);
+                        v3d_get_tex_return_size(devinfo, sampler->format);
 
                 /* For 16-bit, we set up the sampler to always return 2
                  * channels (meaning no recompiles for most statechanges),

src/gallium/drivers/v3d/v3dx_emit.c

@@ -97,8 +97,7 @@ swizzled_border_color(const struct v3d_device_info *devinfo,
          * For swizzling in the shader, we don't do any pre-swizzling of the
          * border color.
          */
-        if (v3d_get_tex_return_size(devinfo, sview->base.format,
-                                    sampler->compare_mode) != 32)
+        if (v3d_get_tex_return_size(devinfo, sview->base.format) != 32)
                 swiz = desc->swizzle[swiz];
 
         switch (swiz) {
@@ -131,8 +130,7 @@ emit_one_texture(struct v3d_context *v3d, struct v3d_texture_stateobj *stage_tex
         v3d_bo_set_reference(&stage_tex->texture_state[i].bo,
                              job->indirect.bo);
 
-        uint32_t return_size = v3d_get_tex_return_size(devinfo, psview->format,
-                                                       psampler->compare_mode);
+        uint32_t return_size = v3d_get_tex_return_size(devinfo, psview->format);
 
         struct V3D33_TEXTURE_SHADER_STATE unpacked = {
                 /* XXX */

src/gallium/drivers/v3d/v3dx_state.c

@@ -1114,8 +1114,7 @@ v3d_create_sampler_view(struct pipe_context *pctx, struct pipe_resource *prsc,
                         }
                 }
         } else {
-                if (v3d_get_tex_return_size(&screen->devinfo, sample_format,
-                                           PIPE_TEX_COMPARE_NONE) == 32) {
+                if (v3d_get_tex_return_size(&screen->devinfo, sample_format) == 32) {
                         if (util_format_is_alpha(sample_format))
                                 so->sampler_variant = V3D_SAMPLER_STATE_32_A;
                         else