xref: /third_party/mesa3d/src/freedreno/ir3/ir3_shader.c (revision bf215546)

/*
 * Copyright (C) 2014 Rob Clark <robclark@freedesktop.org>
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice (including the next
 * paragraph) shall be included in all copies or substantial portions of the
 * Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 *
 * Authors:
 *    Rob Clark <robclark@freedesktop.org>
 */

#include "util/format/u_format.h"
#include "util/u_atomic.h"
#include "util/u_math.h"
#include "util/u_memory.h"
#include "util/u_string.h"

#include "drm/freedreno_drmif.h"

#include "ir3_assembler.h"
#include "ir3_compiler.h"
#include "ir3_nir.h"
#include "ir3_parser.h"
#include "ir3_shader.h"

#include "isa/isa.h"

#include "disasm.h"

int
ir3_glsl_type_size(const struct glsl_type *type, bool bindless)
{
   return glsl_count_attribute_slots(type, false);
}

/* for vertex shader, the inputs are loaded into registers before the shader
 * is executed, so max_regs from the shader instructions might not properly
 * reflect the # of registers actually used, especially in case passthrough
 * varyings.
 *
 * Likewise, for fragment shader, we can have some regs which are passed
 * input values but never touched by the resulting shader (ie. as result
 * of dead code elimination or simply because we don't know how to turn
 * the reg off.
 */
static void
fixup_regfootprint(struct ir3_shader_variant *v)
{
   unsigned i;

   for (i = 0; i < v->inputs_count; i++) {
      /* skip frag inputs fetch via bary.f since their reg's are
       * not written by gpu before shader starts (and in fact the
       * regid's might not even be valid)
       */
      if (v->inputs[i].bary)
         continue;

      /* ignore high regs that are global to all threads in a warp
       * (they exist by default) (a5xx+)
       */
      if (v->inputs[i].regid >= regid(48, 0))
         continue;

      if (v->inputs[i].compmask) {
         unsigned n = util_last_bit(v->inputs[i].compmask) - 1;
         int32_t regid = v->inputs[i].regid + n;
         if (v->inputs[i].half) {
            if (!v->mergedregs) {
               v->info.max_half_reg = MAX2(v->info.max_half_reg, regid >> 2);
            } else {
               v->info.max_reg = MAX2(v->info.max_reg, regid >> 3);
            }
         } else {
            v->info.max_reg = MAX2(v->info.max_reg, regid >> 2);
         }
      }
   }

   for (i = 0; i < v->outputs_count; i++) {
      /* for ex, VS shaders with tess don't have normal varying outs: */
      if (!VALIDREG(v->outputs[i].regid))
         continue;
      int32_t regid = v->outputs[i].regid + 3;
      if (v->outputs[i].half) {
         if (!v->mergedregs) {
            v->info.max_half_reg = MAX2(v->info.max_half_reg, regid >> 2);
         } else {
            v->info.max_reg = MAX2(v->info.max_reg, regid >> 3);
         }
      } else {
         v->info.max_reg = MAX2(v->info.max_reg, regid >> 2);
      }
   }

   for (i = 0; i < v->num_sampler_prefetch; i++) {
      unsigned n = util_last_bit(v->sampler_prefetch[i].wrmask) - 1;
      int32_t regid = v->sampler_prefetch[i].dst + n;
      if (v->sampler_prefetch[i].half_precision) {
         if (!v->mergedregs) {
            v->info.max_half_reg = MAX2(v->info.max_half_reg, regid >> 2);
         } else {
            v->info.max_reg = MAX2(v->info.max_reg, regid >> 3);
         }
      } else {
         v->info.max_reg = MAX2(v->info.max_reg, regid >> 2);
      }
   }
}

/* wrapper for ir3_assemble() which does some info fixup based on
 * shader state.  Non-static since used by ir3_cmdline too.
 */
void *
ir3_shader_assemble(struct ir3_shader_variant *v)
{
   const struct ir3_compiler *compiler = v->compiler;
   struct ir3_info *info = &v->info;
   uint32_t *bin;

   ir3_collect_info(v);

   if (v->constant_data_size) {
      /* Make sure that where we're about to place the constant_data is safe
       * to indirectly upload from.
       */
      info->constant_data_offset =
         align(info->size, v->compiler->const_upload_unit * 16);
      info->size = info->constant_data_offset + v->constant_data_size;
   }

   /* Pad out the size so that when turnip uploads the shaders in
    * sequence, the starting offset of the next one is properly aligned.
    */
   info->size = align(info->size, compiler->instr_align * sizeof(uint64_t));

   bin = isa_assemble(v);
   if (!bin)
      return NULL;

   /* Append the immediates after the end of the program.  This lets us emit
    * the immediates as an indirect load, while avoiding creating another BO.
    */
   if (v->constant_data_size)
      memcpy(&bin[info->constant_data_offset / 4], v->constant_data,
             v->constant_data_size);
   ralloc_free(v->constant_data);
   v->constant_data = NULL;

   /* NOTE: if relative addressing is used, we set constlen in
    * the compiler (to worst-case value) since we don't know in
    * the assembler what the max addr reg value can be:
    */
   v->constlen = MAX2(v->constlen, info->max_const + 1);

   if (v->constlen > ir3_const_state(v)->offsets.driver_param)
      v->need_driver_params = true;

   /* On a4xx and newer, constlen must be a multiple of 16 dwords even though
    * uploads are in units of 4 dwords. Round it up here to make calculations
    * regarding the shared constlen simpler.
    */
   if (compiler->gen >= 4)
      v->constlen = align(v->constlen, 4);

   /* Use the per-wave layout by default on a6xx for compute shaders. It
    * should result in better performance when loads/stores are to a uniform
    * index.
    */
   v->pvtmem_per_wave = compiler->gen >= 6 && !info->multi_dword_ldp_stp &&
                        ((v->type == MESA_SHADER_COMPUTE) ||
                         (v->type == MESA_SHADER_KERNEL));

   fixup_regfootprint(v);

   return bin;
}

static bool
try_override_shader_variant(struct ir3_shader_variant *v,
                            const char *identifier)
{
   assert(ir3_shader_override_path);

   char *name =
      ralloc_asprintf(NULL, "%s/%s.asm", ir3_shader_override_path, identifier);

   FILE *f = fopen(name, "r");

   if (!f) {
      ralloc_free(name);
      return false;
   }

   struct ir3_kernel_info info;
   info.numwg = INVALID_REG;
   v->ir = ir3_parse(v, &info, f);

   fclose(f);

   if (!v->ir) {
      fprintf(stderr, "Failed to parse %s\n", name);
      exit(1);
   }

   v->bin = ir3_shader_assemble(v);
   if (!v->bin) {
      fprintf(stderr, "Failed to assemble %s\n", name);
      exit(1);
   }

   ralloc_free(name);
   return true;
}

static void
assemble_variant(struct ir3_shader_variant *v)
{
   v->bin = ir3_shader_assemble(v);

   bool dbg_enabled = shader_debug_enabled(v->type);
   if (dbg_enabled || ir3_shader_override_path || v->disasm_info.write_disasm) {
      unsigned char sha1[21];
      char sha1buf[41];

      _mesa_sha1_compute(v->bin, v->info.size, sha1);
      _mesa_sha1_format(sha1buf, sha1);

      bool shader_overridden =
         ir3_shader_override_path && try_override_shader_variant(v, sha1buf);

      if (v->disasm_info.write_disasm) {
         char *stream_data = NULL;
         size_t stream_size = 0;
         FILE *stream = open_memstream(&stream_data, &stream_size);

         fprintf(stream,
                 "Native code%s for unnamed %s shader %s with sha1 %s:\n",
                 shader_overridden ? " (overridden)" : "", ir3_shader_stage(v),
                 v->name, sha1buf);
         ir3_shader_disasm(v, v->bin, stream);

         fclose(stream);

         v->disasm_info.disasm = ralloc_size(v, stream_size + 1);
         memcpy(v->disasm_info.disasm, stream_data, stream_size);
         v->disasm_info.disasm[stream_size] = 0;
         free(stream_data);
      }

      if (dbg_enabled || shader_overridden) {
         char *stream_data = NULL;
         size_t stream_size = 0;
         FILE *stream = open_memstream(&stream_data, &stream_size);

         fprintf(stream,
                 "Native code%s for unnamed %s shader %s with sha1 %s:\n",
                 shader_overridden ? " (overridden)" : "", ir3_shader_stage(v),
                 v->name, sha1buf);
         if (v->type == MESA_SHADER_FRAGMENT)
            fprintf(stream, "SIMD0\n");
         ir3_shader_disasm(v, v->bin, stream);
         fclose(stream);

         mesa_log_multiline(MESA_LOG_INFO, stream_data);
         free(stream_data);
      }
   }

   /* no need to keep the ir around beyond this point: */
   ir3_destroy(v->ir);
   v->ir = NULL;
}

static bool
compile_variant(struct ir3_shader *shader, struct ir3_shader_variant *v)
{
   int ret = ir3_compile_shader_nir(shader->compiler, shader, v);
   if (ret) {
      mesa_loge("compile failed! (%s:%s)", shader->nir->info.name,
                shader->nir->info.label);
      return false;
   }

   assemble_variant(v);
   if (!v->bin) {
      mesa_loge("assemble failed! (%s:%s)", shader->nir->info.name,
                shader->nir->info.label);
      return false;
   }

   return true;
}

/*
 * For creating normal shader variants, 'nonbinning' is NULL.  For
 * creating binning pass shader, it is link to corresponding normal
 * (non-binning) variant.
 */
static struct ir3_shader_variant *
alloc_variant(struct ir3_shader *shader, const struct ir3_shader_key *key,
              struct ir3_shader_variant *nonbinning, void *mem_ctx)
{
   /* hang the binning variant off it's non-binning counterpart instead
    * of the shader, to simplify the error cleanup paths
    */
   if (nonbinning)
      mem_ctx = nonbinning;
   struct ir3_shader_variant *v = rzalloc_size(mem_ctx, sizeof(*v));

   if (!v)
      return NULL;

   v->id = ++shader->variant_count;
   v->shader_id = shader->id;
   v->binning_pass = !!nonbinning;
   v->nonbinning = nonbinning;
   v->key = *key;
   v->type = shader->type;
   v->compiler = shader->compiler;
   v->mergedregs = shader->compiler->gen >= 6;
   v->stream_output = shader->stream_output;

   v->name = ralloc_strdup(v, shader->nir->info.name);

   struct shader_info *info = &shader->nir->info;
   switch (v->type) {
   case MESA_SHADER_TESS_CTRL:
   case MESA_SHADER_TESS_EVAL:
      v->tess.primitive_mode = info->tess._primitive_mode;
      v->tess.tcs_vertices_out = info->tess.tcs_vertices_out;
      v->tess.spacing = info->tess.spacing;
      v->tess.ccw = info->tess.ccw;
      v->tess.point_mode = info->tess.point_mode;
      break;

   case MESA_SHADER_GEOMETRY:
      v->gs.output_primitive = info->gs.output_primitive;
      v->gs.vertices_out = info->gs.vertices_out;
      v->gs.invocations = info->gs.invocations;
      v->gs.vertices_in = info->gs.vertices_in;
      break;

   case MESA_SHADER_FRAGMENT:
      v->fs.early_fragment_tests = info->fs.early_fragment_tests;
      v->fs.color_is_dual_source = info->fs.color_is_dual_source;
      break;

   case MESA_SHADER_COMPUTE:
   case MESA_SHADER_KERNEL:
      v->cs.req_input_mem = shader->cs.req_input_mem;
      v->cs.req_local_mem = shader->cs.req_local_mem;
      break;

   default:
      break;
   }

   v->num_ssbos = info->num_ssbos;
   v->num_ibos = info->num_ssbos + info->num_images;
   v->num_reserved_user_consts = shader->num_reserved_user_consts;
   v->api_wavesize = shader->api_wavesize;
   v->real_wavesize = shader->real_wavesize;

   if (!v->binning_pass) {
      v->const_state = rzalloc_size(v, sizeof(*v->const_state));
      v->const_state->shared_consts_enable = shader->shared_consts_enable;
   }

   return v;
}

static bool
needs_binning_variant(struct ir3_shader_variant *v)
{
   if ((v->type == MESA_SHADER_VERTEX) && ir3_has_binning_vs(&v->key))
      return true;
   return false;
}

static struct ir3_shader_variant *
create_variant(struct ir3_shader *shader, const struct ir3_shader_key *key,
               bool write_disasm, void *mem_ctx)
{
   struct ir3_shader_variant *v = alloc_variant(shader, key, NULL, mem_ctx);

   if (!v)
      goto fail;

   v->disasm_info.write_disasm = write_disasm;

   if (needs_binning_variant(v)) {
      v->binning = alloc_variant(shader, key, v, mem_ctx);
      if (!v->binning)
         goto fail;
      v->binning->disasm_info.write_disasm = write_disasm;
   }

   if (ir3_disk_cache_retrieve(shader, v))
      return v;

   if (!shader->nir_finalized) {
      ir3_nir_post_finalize(shader);

      if (ir3_shader_debug & IR3_DBG_DISASM) {
         mesa_logi("dump nir%d: type=%d", shader->id, shader->type);
         nir_log_shaderi(shader->nir);
      }

      if (v->disasm_info.write_disasm) {
         v->disasm_info.nir = nir_shader_as_str(shader->nir, v);
      }

      shader->nir_finalized = true;
   }

   if (!compile_variant(shader, v))
      goto fail;

   if (needs_binning_variant(v) && !compile_variant(shader, v->binning))
      goto fail;

   ir3_disk_cache_store(shader, v);

   return v;

fail:
   ralloc_free(v);
   return NULL;
}

struct ir3_shader_variant *
ir3_shader_create_variant(struct ir3_shader *shader,
                          const struct ir3_shader_key *key,
                          bool keep_ir)
{
   return create_variant(shader, key, keep_ir, NULL);
}

static inline struct ir3_shader_variant *
shader_variant(struct ir3_shader *shader, const struct ir3_shader_key *key)
{
   struct ir3_shader_variant *v;

   for (v = shader->variants; v; v = v->next)
      if (ir3_shader_key_equal(key, &v->key))
         return v;

   return NULL;
}

struct ir3_shader_variant *
ir3_shader_get_variant(struct ir3_shader *shader,
                       const struct ir3_shader_key *key, bool binning_pass,
                       bool write_disasm, bool *created)
{
   mtx_lock(&shader->variants_lock);
   struct ir3_shader_variant *v = shader_variant(shader, key);

   if (!v) {
      /* compile new variant if it doesn't exist already: */
      v = create_variant(shader, key, write_disasm, shader);
      if (v) {
         v->next = shader->variants;
         shader->variants = v;
         *created = true;
      }
   }

   if (v && binning_pass) {
      v = v->binning;
      assert(v);
   }

   mtx_unlock(&shader->variants_lock);

   return v;
}

void
ir3_shader_destroy(struct ir3_shader *shader)
{
   ralloc_free(shader->nir);
   mtx_destroy(&shader->variants_lock);
   ralloc_free(shader);
}

/**
 * Creates a bitmask of the used bits of the shader key by this particular
 * shader.  Used by the gallium driver to skip state-dependent recompiles when
 * possible.
 */
static void
ir3_setup_used_key(struct ir3_shader *shader)
{
   nir_shader *nir = shader->nir;
   struct shader_info *info = &nir->info;
   struct ir3_shader_key *key = &shader->key_mask;

   /* This key flag is just used to make for a cheaper ir3_shader_key_equal
    * check in the common case.
    */
   key->has_per_samp = true;

   key->safe_constlen = true;

   /* When clip/cull distances are natively supported, we only use
    * ucp_enables to determine whether to lower legacy clip planes to
    * gl_ClipDistance.
    */
   if (info->stage != MESA_SHADER_COMPUTE && (info->stage != MESA_SHADER_FRAGMENT || !shader->compiler->has_clip_cull))
      key->ucp_enables = 0xff;

   if (info->stage == MESA_SHADER_FRAGMENT) {
      key->fastc_srgb = ~0;
      key->fsamples = ~0;
      memset(key->fsampler_swizzles, 0xff, sizeof(key->fsampler_swizzles));

      if (info->inputs_read & VARYING_BITS_COLOR) {
         key->rasterflat = true;
      }

      if (info->inputs_read & VARYING_BIT_LAYER) {
         key->layer_zero = true;
      }

      if (info->inputs_read & VARYING_BIT_VIEWPORT) {
         key->view_zero = true;
      }

      /* Only used for deciding on behavior of
       * nir_intrinsic_load_barycentric_sample, or the centroid demotion
       * on older HW.
       */
      key->msaa = info->fs.uses_sample_qualifier ||
                  (shader->compiler->gen < 6 &&
                   (BITSET_TEST(info->system_values_read,
                                SYSTEM_VALUE_BARYCENTRIC_PERSP_CENTROID) ||
                    BITSET_TEST(info->system_values_read,
                                SYSTEM_VALUE_BARYCENTRIC_LINEAR_CENTROID)));
   } else if (info->stage == MESA_SHADER_COMPUTE) {
      key->fastc_srgb = ~0;
      key->fsamples = ~0;
      memset(key->fsampler_swizzles, 0xff, sizeof(key->fsampler_swizzles));
   } else {
      key->tessellation = ~0;
      key->has_gs = true;

      if (info->stage == MESA_SHADER_VERTEX) {
         key->vastc_srgb = ~0;
         key->vsamples = ~0;
         memset(key->vsampler_swizzles, 0xff, sizeof(key->vsampler_swizzles));
      }

      if (info->stage == MESA_SHADER_TESS_CTRL)
         key->tcs_store_primid = true;
   }
}

/* Given an array of constlen's, decrease some of them so that the sum stays
 * within "combined_limit" while trying to fairly share the reduction. Returns
 * a bitfield of which stages should be trimmed.
 */
static uint32_t
trim_constlens(unsigned *constlens, unsigned first_stage, unsigned last_stage,
               unsigned combined_limit, unsigned safe_limit)
{
   unsigned cur_total = 0;
   for (unsigned i = first_stage; i <= last_stage; i++) {
      cur_total += constlens[i];
   }

   unsigned max_stage = 0;
   unsigned max_const = 0;
   uint32_t trimmed = 0;

   while (cur_total > combined_limit) {
      for (unsigned i = first_stage; i <= last_stage; i++) {
         if (constlens[i] >= max_const) {
            max_stage = i;
            max_const = constlens[i];
         }
      }

      assert(max_const > safe_limit);
      trimmed |= 1 << max_stage;
      cur_total = cur_total - max_const + safe_limit;
      constlens[max_stage] = safe_limit;
   }

   return trimmed;
}

/* Figures out which stages in the pipeline to use the "safe" constlen for, in
 * order to satisfy all shared constlen limits.
 */
uint32_t
ir3_trim_constlen(struct ir3_shader_variant **variants,
                  const struct ir3_compiler *compiler)
{
   unsigned constlens[MESA_SHADER_STAGES] = {};

   for (unsigned i = 0; i < MESA_SHADER_STAGES; i++) {
      if (variants[i])
         constlens[i] = variants[i]->constlen;
   }

   uint32_t trimmed = 0;
   STATIC_ASSERT(MESA_SHADER_STAGES <= 8 * sizeof(trimmed));

   bool shared_consts_enable =
      ir3_const_state(variants[MESA_SHADER_VERTEX])->shared_consts_enable;

   /* Use a hw quirk for geometry shared consts, not matched with actual
    * shared consts size (on a6xx).
    */
   uint32_t shared_consts_size_geom = shared_consts_enable ?
      compiler->geom_shared_consts_size_quirk : 0;

   uint32_t shared_consts_size = shared_consts_enable ?
      compiler->shared_consts_size : 0;

   uint32_t safe_shared_consts_size = shared_consts_enable  ?
      ALIGN_POT(MAX2(DIV_ROUND_UP(shared_consts_size_geom, 4),
                     DIV_ROUND_UP(shared_consts_size, 5)), 4) : 0;

   /* There are two shared limits to take into account, the geometry limit on
    * a6xx and the total limit. The frag limit on a6xx only matters for a
    * single stage, so it's always satisfied with the first variant.
    */
   if (compiler->gen >= 6) {
      trimmed |=
         trim_constlens(constlens, MESA_SHADER_VERTEX, MESA_SHADER_GEOMETRY,
                        compiler->max_const_geom - shared_consts_size_geom,
                        compiler->max_const_safe - safe_shared_consts_size);
   }
   trimmed |=
      trim_constlens(constlens, MESA_SHADER_VERTEX, MESA_SHADER_FRAGMENT,
                     compiler->max_const_pipeline - shared_consts_size,
                     compiler->max_const_safe - safe_shared_consts_size);

   return trimmed;
}

struct ir3_shader *
ir3_shader_from_nir(struct ir3_compiler *compiler, nir_shader *nir,
                    const struct ir3_shader_options *options,
                    struct ir3_stream_output_info *stream_output)
{
   struct ir3_shader *shader = rzalloc_size(NULL, sizeof(*shader));

   mtx_init(&shader->variants_lock, mtx_plain);
   shader->compiler = compiler;
   shader->id = p_atomic_inc_return(&shader->compiler->shader_count);
   shader->type = nir->info.stage;
   if (stream_output)
      memcpy(&shader->stream_output, stream_output,
             sizeof(shader->stream_output));
   shader->num_reserved_user_consts = options->reserved_user_consts;
   shader->api_wavesize = options->api_wavesize;
   shader->real_wavesize = options->real_wavesize;
   shader->shared_consts_enable = options->shared_consts_enable;
   shader->nir = nir;

   ir3_disk_cache_init_shader_key(compiler, shader);

   ir3_setup_used_key(shader);

   return shader;
}

static void
dump_reg(FILE *out, const char *name, uint32_t r)
{
   if (r != regid(63, 0)) {
      const char *reg_type = (r & HALF_REG_ID) ? "hr" : "r";
      fprintf(out, "; %s: %s%d.%c\n", name, reg_type, (r & ~HALF_REG_ID) >> 2,
              "xyzw"[r & 0x3]);
   }
}

static void
dump_output(FILE *out, struct ir3_shader_variant *so, unsigned slot,
            const char *name)
{
   uint32_t regid;
   regid = ir3_find_output_regid(so, slot);
   dump_reg(out, name, regid);
}

static const char *
input_name(struct ir3_shader_variant *so, int i)
{
   if (so->inputs[i].sysval) {
      return gl_system_value_name(so->inputs[i].slot);
   } else if (so->type == MESA_SHADER_VERTEX) {
      return gl_vert_attrib_name(so->inputs[i].slot);
   } else {
      return gl_varying_slot_name_for_stage(so->inputs[i].slot, so->type);
   }
}

static const char *
output_name(struct ir3_shader_variant *so, int i)
{
   if (so->type == MESA_SHADER_FRAGMENT) {
      return gl_frag_result_name(so->outputs[i].slot);
   } else {
      switch (so->outputs[i].slot) {
      case VARYING_SLOT_GS_HEADER_IR3:
         return "GS_HEADER";
      case VARYING_SLOT_GS_VERTEX_FLAGS_IR3:
         return "GS_VERTEX_FLAGS";
      case VARYING_SLOT_TCS_HEADER_IR3:
         return "TCS_HEADER";
      default:
         return gl_varying_slot_name_for_stage(so->outputs[i].slot, so->type);
      }
   }
}

static void
dump_const_state(struct ir3_shader_variant *so, FILE *out)
{
   const struct ir3_const_state *cs = ir3_const_state(so);
   const struct ir3_ubo_analysis_state *us = &cs->ubo_state;

   fprintf(out, "; num_ubos:           %u\n", cs->num_ubos);
   fprintf(out, "; num_driver_params:  %u\n", cs->num_driver_params);
   fprintf(out, "; offsets:\n");
   if (cs->offsets.ubo != ~0)
      fprintf(out, ";   ubo:              c%u.x\n", cs->offsets.ubo);
   if (cs->offsets.image_dims != ~0)
      fprintf(out, ";   image_dims:       c%u.x\n", cs->offsets.image_dims);
   if (cs->offsets.kernel_params != ~0)
      fprintf(out, ";   kernel_params:    c%u.x\n", cs->offsets.kernel_params);
   if (cs->offsets.driver_param != ~0)
      fprintf(out, ";   driver_param:     c%u.x\n", cs->offsets.driver_param);
   if (cs->offsets.tfbo != ~0)
      fprintf(out, ";   tfbo:             c%u.x\n", cs->offsets.tfbo);
   if (cs->offsets.primitive_param != ~0)
      fprintf(out, ";   primitive_params: c%u.x\n", cs->offsets.primitive_param);
   if (cs->offsets.primitive_map != ~0)
      fprintf(out, ";   primitive_map:    c%u.x\n", cs->offsets.primitive_map);
   fprintf(out, "; ubo_state:\n");
   fprintf(out, ";   num_enabled:      %u\n", us->num_enabled);
   for (unsigned i = 0; i < us->num_enabled; i++) {
      const struct ir3_ubo_range *r = &us->range[i];

      assert((r->offset % 16) == 0);

      fprintf(out, ";   range[%u]:\n", i);
      fprintf(out, ";     block:          %u\n", r->ubo.block);
      if (r->ubo.bindless)
         fprintf(out, ";     bindless_base:  %u\n", r->ubo.bindless_base);
      fprintf(out, ";     offset:         c%u.x\n", r->offset/16);

      unsigned size = r->end - r->start;
      assert((size % 16) == 0);

      fprintf(out, ";     size:           %u vec4 (%ub -> %ub)\n", (size/16), r->start, r->end);
   }
}

void
ir3_shader_disasm(struct ir3_shader_variant *so, uint32_t *bin, FILE *out)
{
   struct ir3 *ir = so->ir;
   struct ir3_register *reg;
   const char *type = ir3_shader_stage(so);
   uint8_t regid;
   unsigned i;

   dump_const_state(so, out);

   foreach_input_n (instr, i, ir) {
      reg = instr->dsts[0];
      regid = reg->num;
      fprintf(out, "@in(%sr%d.%c)\tin%d",
              (reg->flags & IR3_REG_HALF) ? "h" : "", (regid >> 2),
              "xyzw"[regid & 0x3], i);

      if (reg->wrmask > 0x1)
         fprintf(out, " (wrmask=0x%x)", reg->wrmask);
      fprintf(out, "\n");
   }

   /* print pre-dispatch texture fetches: */
   for (i = 0; i < so->num_sampler_prefetch; i++) {
      const struct ir3_sampler_prefetch *fetch = &so->sampler_prefetch[i];
      fprintf(out,
              "@tex(%sr%d.%c)\tsrc=%u, samp=%u, tex=%u, wrmask=0x%x, cmd=%u\n",
              fetch->half_precision ? "h" : "", fetch->dst >> 2,
              "xyzw"[fetch->dst & 0x3], fetch -> src, fetch -> samp_id,
              fetch -> tex_id, fetch -> wrmask, fetch -> cmd);
   }

   const struct ir3_const_state *const_state = ir3_const_state(so);
   for (i = 0; i < DIV_ROUND_UP(const_state->immediates_count, 4); i++) {
      fprintf(out, "@const(c%d.x)\t", const_state->offsets.immediate + i);
      fprintf(out, "0x%08x, 0x%08x, 0x%08x, 0x%08x\n",
              const_state->immediates[i * 4 + 0],
              const_state->immediates[i * 4 + 1],
              const_state->immediates[i * 4 + 2],
              const_state->immediates[i * 4 + 3]);
   }

   isa_decode(bin, so->info.sizedwords * 4, out,
              &(struct isa_decode_options){
                 .gpu_id = fd_dev_gpu_id(ir->compiler->dev_id),
                 .show_errors = true,
                 .branch_labels = true,
              });

   fprintf(out, "; %s: outputs:", type);
   for (i = 0; i < so->outputs_count; i++) {
      uint8_t regid = so->outputs[i].regid;
      const char *reg_type = so->outputs[i].half ? "hr" : "r";
      fprintf(out, " %s%d.%c (%s)", reg_type, (regid >> 2), "xyzw"[regid & 0x3],
              output_name(so, i));
   }
   fprintf(out, "\n");

   fprintf(out, "; %s: inputs:", type);
   for (i = 0; i < so->inputs_count; i++) {
      uint8_t regid = so->inputs[i].regid;
      fprintf(out, " r%d.%c (%s slot=%d cm=%x,il=%u,b=%u)", (regid >> 2),
              "xyzw"[regid & 0x3], input_name(so, i), so -> inputs[i].slot,
              so->inputs[i].compmask, so->inputs[i].inloc, so->inputs[i].bary);
   }
   fprintf(out, "\n");

   /* print generic shader info: */
   fprintf(
      out,
      "; %s prog %d/%d: %u instr, %u nops, %u non-nops, %u mov, %u cov, %u dwords\n",
      type, so->shader_id, so->id, so->info.instrs_count, so->info.nops_count,
      so->info.instrs_count - so->info.nops_count, so->info.mov_count,
      so->info.cov_count, so->info.sizedwords);

   fprintf(out,
           "; %s prog %d/%d: %u last-baryf, %d half, %d full, %u constlen\n",
           type, so->shader_id, so->id, so->info.last_baryf,
           so->info.max_half_reg + 1, so->info.max_reg + 1, so->constlen);

   fprintf(
      out,
      "; %s prog %d/%d: %u cat0, %u cat1, %u cat2, %u cat3, %u cat4, %u cat5, %u cat6, %u cat7, \n",
      type, so->shader_id, so->id, so->info.instrs_per_cat[0],
      so->info.instrs_per_cat[1], so->info.instrs_per_cat[2],
      so->info.instrs_per_cat[3], so->info.instrs_per_cat[4],
      so->info.instrs_per_cat[5], so->info.instrs_per_cat[6],
      so->info.instrs_per_cat[7]);

   fprintf(
      out,
      "; %s prog %d/%d: %u sstall, %u (ss), %u systall, %u (sy), %d loops\n",
      type, so->shader_id, so->id, so->info.sstall, so->info.ss,
      so->info.systall, so->info.sy, so->loops);

   /* print shader type specific info: */
   switch (so->type) {
   case MESA_SHADER_VERTEX:
      dump_output(out, so, VARYING_SLOT_POS, "pos");
      dump_output(out, so, VARYING_SLOT_PSIZ, "psize");
      break;
   case MESA_SHADER_FRAGMENT:
      dump_reg(out, "pos (ij_pixel)",
               ir3_find_sysval_regid(so, SYSTEM_VALUE_BARYCENTRIC_PERSP_PIXEL));
      dump_reg(
         out, "pos (ij_centroid)",
         ir3_find_sysval_regid(so, SYSTEM_VALUE_BARYCENTRIC_PERSP_CENTROID));
      dump_reg(out, "pos (center_rhw)",
               ir3_find_sysval_regid(so, SYSTEM_VALUE_BARYCENTRIC_PERSP_CENTER_RHW));
      dump_output(out, so, FRAG_RESULT_DEPTH, "posz");
      if (so->color0_mrt) {
         dump_output(out, so, FRAG_RESULT_COLOR, "color");
      } else {
         dump_output(out, so, FRAG_RESULT_DATA0, "data0");
         dump_output(out, so, FRAG_RESULT_DATA1, "data1");
         dump_output(out, so, FRAG_RESULT_DATA2, "data2");
         dump_output(out, so, FRAG_RESULT_DATA3, "data3");
         dump_output(out, so, FRAG_RESULT_DATA4, "data4");
         dump_output(out, so, FRAG_RESULT_DATA5, "data5");
         dump_output(out, so, FRAG_RESULT_DATA6, "data6");
         dump_output(out, so, FRAG_RESULT_DATA7, "data7");
      }
      dump_reg(out, "fragcoord",
               ir3_find_sysval_regid(so, SYSTEM_VALUE_FRAG_COORD));
      dump_reg(out, "fragface",
               ir3_find_sysval_regid(so, SYSTEM_VALUE_FRONT_FACE));
      break;
   default:
      /* TODO */
      break;
   }

   fprintf(out, "\n");
}

uint64_t
ir3_shader_outputs(const struct ir3_shader *so)
{
   return so->nir->info.outputs_written;
}

/* Add any missing varyings needed for stream-out.  Otherwise varyings not
 * used by fragment shader will be stripped out.
 */
void
ir3_link_stream_out(struct ir3_shader_linkage *l,
                    const struct ir3_shader_variant *v)
{
   const struct ir3_stream_output_info *strmout = &v->stream_output;

   /*
    * First, any stream-out varyings not already in linkage map (ie. also
    * consumed by frag shader) need to be added:
    */
   for (unsigned i = 0; i < strmout->num_outputs; i++) {
      const struct ir3_stream_output *out = &strmout->output[i];
      unsigned k = out->register_index;
      unsigned compmask =
         (1 << (out->num_components + out->start_component)) - 1;
      unsigned idx, nextloc = 0;

      /* psize/pos need to be the last entries in linkage map, and will
       * get added link_stream_out, so skip over them:
       */
      if ((v->outputs[k].slot == VARYING_SLOT_PSIZ) ||
          (v->outputs[k].slot == VARYING_SLOT_POS))
         continue;

      for (idx = 0; idx < l->cnt; idx++) {
         if (l->var[idx].slot == v->outputs[k].slot)
            break;
         nextloc = MAX2(nextloc, l->var[idx].loc + 4);
      }

      /* add if not already in linkage map: */
      if (idx == l->cnt) {
         ir3_link_add(l, v->outputs[k].slot, v->outputs[k].regid,
                      compmask, nextloc);
      }

      /* expand component-mask if needed, ie streaming out all components
       * but frag shader doesn't consume all components:
       */
      if (compmask & ~l->var[idx].compmask) {
         l->var[idx].compmask |= compmask;
         l->max_loc = MAX2(
            l->max_loc, l->var[idx].loc + util_last_bit(l->var[idx].compmask));
      }
   }
}