/*
 * Copyright © 2016 Red Hat.
 * Copyright © 2016 Bas Nieuwenhuizen
 *
 * based in part on anv driver which is:
 * Copyright © 2015 Intel Corporation
 *
 * 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.
 */

#include "nir/nir.h"
#include "nir/nir_builder.h"
#include "nir/nir_xfb_info.h"
#include "spirv/nir_spirv.h"
#include "util/disk_cache.h"
#include "util/mesa-sha1.h"
#include "util/u_atomic.h"
#include "radv_cs.h"
#include "radv_debug.h"
#include "radv_private.h"
#include "radv_shader.h"
#include "vk_util.h"

#include "util/debug.h"
#include "ac_binary.h"
#include "ac_exp_param.h"
#include "ac_nir.h"
#include "ac_shader_util.h"
#include "aco_interface.h"
#include "sid.h"
#include "vk_format.h"

struct radv_blend_state {
   uint32_t blend_enable_4bit;
   uint32_t need_src_alpha;

   uint32_t cb_color_control;
   uint32_t cb_target_mask;
   uint32_t cb_target_enabled_4bit;
   uint32_t sx_mrt_blend_opt[8];
   uint32_t cb_blend_control[8];

   uint32_t spi_shader_col_format;
   uint32_t col_format_is_int8;
   uint32_t col_format_is_int10;
   uint32_t cb_shader_mask;
   uint32_t db_alpha_to_mask;

   uint32_t commutative_4bit;

   bool single_cb_enable;
   bool mrt0_is_dual_src;
};

struct radv_dsa_order_invariance {
   /* Whether the final result in Z/S buffers is guaranteed to be
    * invariant under changes to the order in which fragments arrive.
    */
   bool zs;

   /* Whether the set of fragments that pass the combined Z/S test is
    * guaranteed to be invariant under changes to the order in which
    * fragments arrive.
    */
   bool pass_set;
};

static bool
radv_is_state_dynamic(const VkGraphicsPipelineCreateInfo *pCreateInfo, VkDynamicState state)
{
   if (pCreateInfo->pDynamicState) {
      uint32_t count = pCreateInfo->pDynamicState->dynamicStateCount;
      for (uint32_t i = 0; i < count; i++) {
         if (pCreateInfo->pDynamicState->pDynamicStates[i] == state)
            return true;
      }
   }

   return false;
}

static const VkPipelineMultisampleStateCreateInfo *
radv_pipeline_get_multisample_state(const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   if (!pCreateInfo->pRasterizationState->rasterizerDiscardEnable ||
       radv_is_state_dynamic(pCreateInfo, VK_DYNAMIC_STATE_RASTERIZER_DISCARD_ENABLE_EXT))
      return pCreateInfo->pMultisampleState;
   return NULL;
}

static const VkPipelineTessellationStateCreateInfo *
radv_pipeline_get_tessellation_state(const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   for (uint32_t i = 0; i < pCreateInfo->stageCount; i++) {
      if (pCreateInfo->pStages[i].stage == VK_SHADER_STAGE_TESSELLATION_CONTROL_BIT ||
          pCreateInfo->pStages[i].stage == VK_SHADER_STAGE_TESSELLATION_EVALUATION_BIT) {
         return pCreateInfo->pTessellationState;
      }
   }
   return NULL;
}

static const VkPipelineDepthStencilStateCreateInfo *
radv_pipeline_get_depth_stencil_state(const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   RADV_FROM_HANDLE(radv_render_pass, pass, pCreateInfo->renderPass);
   struct radv_subpass *subpass = pass->subpasses + pCreateInfo->subpass;

   if ((!pCreateInfo->pRasterizationState->rasterizerDiscardEnable &&
        subpass->depth_stencil_attachment) ||
       radv_is_state_dynamic(pCreateInfo, VK_DYNAMIC_STATE_RASTERIZER_DISCARD_ENABLE_EXT))
      return pCreateInfo->pDepthStencilState;
   return NULL;
}

static const VkPipelineColorBlendStateCreateInfo *
radv_pipeline_get_color_blend_state(const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   RADV_FROM_HANDLE(radv_render_pass, pass, pCreateInfo->renderPass);
   struct radv_subpass *subpass = pass->subpasses + pCreateInfo->subpass;

   if ((!pCreateInfo->pRasterizationState->rasterizerDiscardEnable && subpass->has_color_att) ||
       radv_is_state_dynamic(pCreateInfo, VK_DYNAMIC_STATE_RASTERIZER_DISCARD_ENABLE_EXT))
      return pCreateInfo->pColorBlendState;
   return NULL;
}

static bool
radv_pipeline_has_ngg(const struct radv_pipeline *pipeline)
{
   struct radv_shader_variant *variant = NULL;
   if (pipeline->shaders[MESA_SHADER_GEOMETRY])
      variant = pipeline->shaders[MESA_SHADER_GEOMETRY];
   else if (pipeline->shaders[MESA_SHADER_TESS_EVAL])
      variant = pipeline->shaders[MESA_SHADER_TESS_EVAL];
   else if (pipeline->shaders[MESA_SHADER_VERTEX])
      variant = pipeline->shaders[MESA_SHADER_VERTEX];
   else
      return false;
   return variant->info.is_ngg;
}

bool
radv_pipeline_has_ngg_passthrough(const struct radv_pipeline *pipeline)
{
   assert(radv_pipeline_has_ngg(pipeline));

   struct radv_shader_variant *variant = NULL;
   if (pipeline->shaders[MESA_SHADER_GEOMETRY])
      variant = pipeline->shaders[MESA_SHADER_GEOMETRY];
   else if (pipeline->shaders[MESA_SHADER_TESS_EVAL])
      variant = pipeline->shaders[MESA_SHADER_TESS_EVAL];
   else if (pipeline->shaders[MESA_SHADER_VERTEX])
      variant = pipeline->shaders[MESA_SHADER_VERTEX];
   else
      return false;
   return variant->info.is_ngg_passthrough;
}

bool
radv_pipeline_has_gs_copy_shader(const struct radv_pipeline *pipeline)
{
   return !!pipeline->gs_copy_shader;
}

static void
radv_pipeline_destroy(struct radv_device *device, struct radv_pipeline *pipeline,
                      const VkAllocationCallbacks *allocator)
{
   for (unsigned i = 0; i < MESA_SHADER_STAGES; ++i)
      if (pipeline->shaders[i])
         radv_shader_variant_destroy(device, pipeline->shaders[i]);

   if (pipeline->gs_copy_shader)
      radv_shader_variant_destroy(device, pipeline->gs_copy_shader);

   if (pipeline->cs.buf)
      free(pipeline->cs.buf);

   vk_object_base_finish(&pipeline->base);
   vk_free2(&device->vk.alloc, allocator, pipeline);
}

void
radv_DestroyPipeline(VkDevice _device, VkPipeline _pipeline,
                     const VkAllocationCallbacks *pAllocator)
{
   RADV_FROM_HANDLE(radv_device, device, _device);
   RADV_FROM_HANDLE(radv_pipeline, pipeline, _pipeline);

   if (!_pipeline)
      return;

   radv_pipeline_destroy(device, pipeline, pAllocator);
}

static uint32_t
get_hash_flags(const struct radv_device *device, bool stats)
{
   uint32_t hash_flags = 0;

   if (device->instance->debug_flags & RADV_DEBUG_NO_NGG)
      hash_flags |= RADV_HASH_SHADER_NO_NGG;
   if (device->physical_device->cs_wave_size == 32)
      hash_flags |= RADV_HASH_SHADER_CS_WAVE32;
   if (device->physical_device->ps_wave_size == 32)
      hash_flags |= RADV_HASH_SHADER_PS_WAVE32;
   if (device->physical_device->ge_wave_size == 32)
      hash_flags |= RADV_HASH_SHADER_GE_WAVE32;
   if (device->physical_device->use_llvm)
      hash_flags |= RADV_HASH_SHADER_LLVM;
   if (device->instance->debug_flags & RADV_DEBUG_DISCARD_TO_DEMOTE)
      hash_flags |= RADV_HASH_SHADER_DISCARD_TO_DEMOTE;
   if (device->instance->enable_mrt_output_nan_fixup)
      hash_flags |= RADV_HASH_SHADER_MRT_NAN_FIXUP;
   if (device->instance->debug_flags & RADV_DEBUG_INVARIANT_GEOM)
      hash_flags |= RADV_HASH_SHADER_INVARIANT_GEOM;
   if (stats)
      hash_flags |= RADV_HASH_SHADER_KEEP_STATISTICS;
   if (device->force_vrs != RADV_FORCE_VRS_2x2)
      hash_flags |= RADV_HASH_SHADER_FORCE_VRS_2x2;
   if (device->force_vrs != RADV_FORCE_VRS_2x1)
      hash_flags |= RADV_HASH_SHADER_FORCE_VRS_2x1;
   if (device->force_vrs != RADV_FORCE_VRS_1x2)
      hash_flags |= RADV_HASH_SHADER_FORCE_VRS_1x2;
   return hash_flags;
}

static void
radv_pipeline_init_scratch(const struct radv_device *device, struct radv_pipeline *pipeline)
{
   unsigned scratch_bytes_per_wave = 0;
   unsigned max_waves = 0;
   unsigned min_waves = 1;

   for (int i = 0; i < MESA_SHADER_STAGES; ++i) {
      if (pipeline->shaders[i] && pipeline->shaders[i]->config.scratch_bytes_per_wave) {
         unsigned max_stage_waves = device->scratch_waves;

         scratch_bytes_per_wave =
            MAX2(scratch_bytes_per_wave, pipeline->shaders[i]->config.scratch_bytes_per_wave);

         max_stage_waves =
            MIN2(max_stage_waves, 4 * device->physical_device->rad_info.num_good_compute_units *
                                     (256 / pipeline->shaders[i]->config.num_vgprs));
         max_waves = MAX2(max_waves, max_stage_waves);
      }
   }

   if (pipeline->shaders[MESA_SHADER_COMPUTE]) {
      unsigned group_size = pipeline->shaders[MESA_SHADER_COMPUTE]->info.cs.block_size[0] *
                            pipeline->shaders[MESA_SHADER_COMPUTE]->info.cs.block_size[1] *
                            pipeline->shaders[MESA_SHADER_COMPUTE]->info.cs.block_size[2];
      min_waves = MAX2(min_waves, round_up_u32(group_size, 64));
   }

   pipeline->scratch_bytes_per_wave = scratch_bytes_per_wave;
   pipeline->max_waves = max_waves;
}

static uint32_t
si_translate_blend_logic_op(VkLogicOp op)
{
   switch (op) {
   case VK_LOGIC_OP_CLEAR:
      return V_028808_ROP3_CLEAR;
   case VK_LOGIC_OP_AND:
      return V_028808_ROP3_AND;
   case VK_LOGIC_OP_AND_REVERSE:
      return V_028808_ROP3_AND_REVERSE;
   case VK_LOGIC_OP_COPY:
      return V_028808_ROP3_COPY;
   case VK_LOGIC_OP_AND_INVERTED:
      return V_028808_ROP3_AND_INVERTED;
   case VK_LOGIC_OP_NO_OP:
      return V_028808_ROP3_NO_OP;
   case VK_LOGIC_OP_XOR:
      return V_028808_ROP3_XOR;
   case VK_LOGIC_OP_OR:
      return V_028808_ROP3_OR;
   case VK_LOGIC_OP_NOR:
      return V_028808_ROP3_NOR;
   case VK_LOGIC_OP_EQUIVALENT:
      return V_028808_ROP3_EQUIVALENT;
   case VK_LOGIC_OP_INVERT:
      return V_028808_ROP3_INVERT;
   case VK_LOGIC_OP_OR_REVERSE:
      return V_028808_ROP3_OR_REVERSE;
   case VK_LOGIC_OP_COPY_INVERTED:
      return V_028808_ROP3_COPY_INVERTED;
   case VK_LOGIC_OP_OR_INVERTED:
      return V_028808_ROP3_OR_INVERTED;
   case VK_LOGIC_OP_NAND:
      return V_028808_ROP3_NAND;
   case VK_LOGIC_OP_SET:
      return V_028808_ROP3_SET;
   default:
      unreachable("Unhandled logic op");
   }
}

static uint32_t
si_translate_blend_function(VkBlendOp op)
{
   switch (op) {
   case VK_BLEND_OP_ADD:
      return V_028780_COMB_DST_PLUS_SRC;
   case VK_BLEND_OP_SUBTRACT:
      return V_028780_COMB_SRC_MINUS_DST;
   case VK_BLEND_OP_REVERSE_SUBTRACT:
      return V_028780_COMB_DST_MINUS_SRC;
   case VK_BLEND_OP_MIN:
      return V_028780_COMB_MIN_DST_SRC;
   case VK_BLEND_OP_MAX:
      return V_028780_COMB_MAX_DST_SRC;
   default:
      return 0;
   }
}

static uint32_t
si_translate_blend_factor(VkBlendFactor factor)
{
   switch (factor) {
   case VK_BLEND_FACTOR_ZERO:
      return V_028780_BLEND_ZERO;
   case VK_BLEND_FACTOR_ONE:
      return V_028780_BLEND_ONE;
   case VK_BLEND_FACTOR_SRC_COLOR:
      return V_028780_BLEND_SRC_COLOR;
   case VK_BLEND_FACTOR_ONE_MINUS_SRC_COLOR:
      return V_028780_BLEND_ONE_MINUS_SRC_COLOR;
   case VK_BLEND_FACTOR_DST_COLOR:
      return V_028780_BLEND_DST_COLOR;
   case VK_BLEND_FACTOR_ONE_MINUS_DST_COLOR:
      return V_028780_BLEND_ONE_MINUS_DST_COLOR;
   case VK_BLEND_FACTOR_SRC_ALPHA:
      return V_028780_BLEND_SRC_ALPHA;
   case VK_BLEND_FACTOR_ONE_MINUS_SRC_ALPHA:
      return V_028780_BLEND_ONE_MINUS_SRC_ALPHA;
   case VK_BLEND_FACTOR_DST_ALPHA:
      return V_028780_BLEND_DST_ALPHA;
   case VK_BLEND_FACTOR_ONE_MINUS_DST_ALPHA:
      return V_028780_BLEND_ONE_MINUS_DST_ALPHA;
   case VK_BLEND_FACTOR_CONSTANT_COLOR:
      return V_028780_BLEND_CONSTANT_COLOR;
   case VK_BLEND_FACTOR_ONE_MINUS_CONSTANT_COLOR:
      return V_028780_BLEND_ONE_MINUS_CONSTANT_COLOR;
   case VK_BLEND_FACTOR_CONSTANT_ALPHA:
      return V_028780_BLEND_CONSTANT_ALPHA;
   case VK_BLEND_FACTOR_ONE_MINUS_CONSTANT_ALPHA:
      return V_028780_BLEND_ONE_MINUS_CONSTANT_ALPHA;
   case VK_BLEND_FACTOR_SRC_ALPHA_SATURATE:
      return V_028780_BLEND_SRC_ALPHA_SATURATE;
   case VK_BLEND_FACTOR_SRC1_COLOR:
      return V_028780_BLEND_SRC1_COLOR;
   case VK_BLEND_FACTOR_ONE_MINUS_SRC1_COLOR:
      return V_028780_BLEND_INV_SRC1_COLOR;
   case VK_BLEND_FACTOR_SRC1_ALPHA:
      return V_028780_BLEND_SRC1_ALPHA;
   case VK_BLEND_FACTOR_ONE_MINUS_SRC1_ALPHA:
      return V_028780_BLEND_INV_SRC1_ALPHA;
   default:
      return 0;
   }
}

static uint32_t
si_translate_blend_opt_function(VkBlendOp op)
{
   switch (op) {
   case VK_BLEND_OP_ADD:
      return V_028760_OPT_COMB_ADD;
   case VK_BLEND_OP_SUBTRACT:
      return V_028760_OPT_COMB_SUBTRACT;
   case VK_BLEND_OP_REVERSE_SUBTRACT:
      return V_028760_OPT_COMB_REVSUBTRACT;
   case VK_BLEND_OP_MIN:
      return V_028760_OPT_COMB_MIN;
   case VK_BLEND_OP_MAX:
      return V_028760_OPT_COMB_MAX;
   default:
      return V_028760_OPT_COMB_BLEND_DISABLED;
   }
}

static uint32_t
si_translate_blend_opt_factor(VkBlendFactor factor, bool is_alpha)
{
   switch (factor) {
   case VK_BLEND_FACTOR_ZERO:
      return V_028760_BLEND_OPT_PRESERVE_NONE_IGNORE_ALL;
   case VK_BLEND_FACTOR_ONE:
      return V_028760_BLEND_OPT_PRESERVE_ALL_IGNORE_NONE;
   case VK_BLEND_FACTOR_SRC_COLOR:
      return is_alpha ? V_028760_BLEND_OPT_PRESERVE_A1_IGNORE_A0
                      : V_028760_BLEND_OPT_PRESERVE_C1_IGNORE_C0;
   case VK_BLEND_FACTOR_ONE_MINUS_SRC_COLOR:
      return is_alpha ? V_028760_BLEND_OPT_PRESERVE_A0_IGNORE_A1
                      : V_028760_BLEND_OPT_PRESERVE_C0_IGNORE_C1;
   case VK_BLEND_FACTOR_SRC_ALPHA:
      return V_028760_BLEND_OPT_PRESERVE_A1_IGNORE_A0;
   case VK_BLEND_FACTOR_ONE_MINUS_SRC_ALPHA:
      return V_028760_BLEND_OPT_PRESERVE_A0_IGNORE_A1;
   case VK_BLEND_FACTOR_SRC_ALPHA_SATURATE:
      return is_alpha ? V_028760_BLEND_OPT_PRESERVE_ALL_IGNORE_NONE
                      : V_028760_BLEND_OPT_PRESERVE_NONE_IGNORE_A0;
   default:
      return V_028760_BLEND_OPT_PRESERVE_NONE_IGNORE_NONE;
   }
}

/**
 * Get rid of DST in the blend factors by commuting the operands:
 *    func(src * DST, dst * 0) ---> func(src * 0, dst * SRC)
 */
static void
si_blend_remove_dst(VkBlendOp *func, VkBlendFactor *src_factor, VkBlendFactor *dst_factor,
                    VkBlendFactor expected_dst, VkBlendFactor replacement_src)
{
   if (*src_factor == expected_dst && *dst_factor == VK_BLEND_FACTOR_ZERO) {
      *src_factor = VK_BLEND_FACTOR_ZERO;
      *dst_factor = replacement_src;

      /* Commuting the operands requires reversing subtractions. */
      if (*func == VK_BLEND_OP_SUBTRACT)
         *func = VK_BLEND_OP_REVERSE_SUBTRACT;
      else if (*func == VK_BLEND_OP_REVERSE_SUBTRACT)
         *func = VK_BLEND_OP_SUBTRACT;
   }
}

static bool
si_blend_factor_uses_dst(VkBlendFactor factor)
{
   return factor == VK_BLEND_FACTOR_DST_COLOR || factor == VK_BLEND_FACTOR_DST_ALPHA ||
          factor == VK_BLEND_FACTOR_SRC_ALPHA_SATURATE ||
          factor == VK_BLEND_FACTOR_ONE_MINUS_DST_ALPHA ||
          factor == VK_BLEND_FACTOR_ONE_MINUS_DST_COLOR;
}

static bool
is_dual_src(VkBlendFactor factor)
{
   switch (factor) {
   case VK_BLEND_FACTOR_SRC1_COLOR:
   case VK_BLEND_FACTOR_ONE_MINUS_SRC1_COLOR:
   case VK_BLEND_FACTOR_SRC1_ALPHA:
   case VK_BLEND_FACTOR_ONE_MINUS_SRC1_ALPHA:
      return true;
   default:
      return false;
   }
}

static unsigned
radv_choose_spi_color_format(const struct radv_device *device, VkFormat vk_format,
                             bool blend_enable, bool blend_need_alpha)
{
   const struct util_format_description *desc = vk_format_description(vk_format);
   bool use_rbplus = device->physical_device->rad_info.rbplus_allowed;
   struct ac_spi_color_formats formats = {0};
   unsigned format, ntype, swap;

   format = radv_translate_colorformat(vk_format);
   ntype = radv_translate_color_numformat(vk_format, desc,
                                          vk_format_get_first_non_void_channel(vk_format));
   swap = radv_translate_colorswap(vk_format, false);

   ac_choose_spi_color_formats(format, swap, ntype, false, use_rbplus, &formats);

   if (blend_enable && blend_need_alpha)
      return formats.blend_alpha;
   else if (blend_need_alpha)
      return formats.alpha;
   else if (blend_enable)
      return formats.blend;
   else
      return formats.normal;
}

static bool
format_is_int8(VkFormat format)
{
   const struct util_format_description *desc = vk_format_description(format);
   int channel = vk_format_get_first_non_void_channel(format);

   return channel >= 0 && desc->channel[channel].pure_integer && desc->channel[channel].size == 8;
}

static bool
format_is_int10(VkFormat format)
{
   const struct util_format_description *desc = vk_format_description(format);

   if (desc->nr_channels != 4)
      return false;
   for (unsigned i = 0; i < 4; i++) {
      if (desc->channel[i].pure_integer && desc->channel[i].size == 10)
         return true;
   }
   return false;
}

static void
radv_pipeline_compute_spi_color_formats(const struct radv_pipeline *pipeline,
                                        const VkGraphicsPipelineCreateInfo *pCreateInfo,
                                        struct radv_blend_state *blend)
{
   RADV_FROM_HANDLE(radv_render_pass, pass, pCreateInfo->renderPass);
   struct radv_subpass *subpass = pass->subpasses + pCreateInfo->subpass;
   unsigned col_format = 0, is_int8 = 0, is_int10 = 0;
   unsigned num_targets;

   for (unsigned i = 0; i < (blend->single_cb_enable ? 1 : subpass->color_count); ++i) {
      unsigned cf;

      if (subpass->color_attachments[i].attachment == VK_ATTACHMENT_UNUSED ||
          !(blend->cb_target_mask & (0xfu << (i * 4)))) {
         cf = V_028714_SPI_SHADER_ZERO;
      } else {
         struct radv_render_pass_attachment *attachment =
            pass->attachments + subpass->color_attachments[i].attachment;
         bool blend_enable = blend->blend_enable_4bit & (0xfu << (i * 4));

         cf = radv_choose_spi_color_format(pipeline->device, attachment->format, blend_enable,
                                           blend->need_src_alpha & (1 << i));

         if (format_is_int8(attachment->format))
            is_int8 |= 1 << i;
         if (format_is_int10(attachment->format))
            is_int10 |= 1 << i;
      }

      col_format |= cf << (4 * i);
   }

   if (!(col_format & 0xf) && blend->need_src_alpha & (1 << 0)) {
      /* When a subpass doesn't have any color attachments, write the
       * alpha channel of MRT0 when alpha coverage is enabled because
       * the depth attachment needs it.
       */
      col_format |= V_028714_SPI_SHADER_32_AR;
   }

   /* If the i-th target format is set, all previous target formats must
    * be non-zero to avoid hangs.
    */
   num_targets = (util_last_bit(col_format) + 3) / 4;
   for (unsigned i = 0; i < num_targets; i++) {
      if (!(col_format & (0xfu << (i * 4)))) {
         col_format |= V_028714_SPI_SHADER_32_R << (i * 4);
      }
   }

   /* The output for dual source blending should have the same format as
    * the first output.
    */
   if (blend->mrt0_is_dual_src) {
      assert(!(col_format >> 4));
      col_format |= (col_format & 0xf) << 4;
   }

   blend->cb_shader_mask = ac_get_cb_shader_mask(col_format);
   blend->spi_shader_col_format = col_format;
   blend->col_format_is_int8 = is_int8;
   blend->col_format_is_int10 = is_int10;
}

/*
 * Ordered so that for each i,
 * radv_format_meta_fs_key(radv_fs_key_format_exemplars[i]) == i.
 */
const VkFormat radv_fs_key_format_exemplars[NUM_META_FS_KEYS] = {
   VK_FORMAT_R32_SFLOAT,
   VK_FORMAT_R32G32_SFLOAT,
   VK_FORMAT_R8G8B8A8_UNORM,
   VK_FORMAT_R16G16B16A16_UNORM,
   VK_FORMAT_R16G16B16A16_SNORM,
   VK_FORMAT_R16G16B16A16_UINT,
   VK_FORMAT_R16G16B16A16_SINT,
   VK_FORMAT_R32G32B32A32_SFLOAT,
   VK_FORMAT_R8G8B8A8_UINT,
   VK_FORMAT_R8G8B8A8_SINT,
   VK_FORMAT_A2R10G10B10_UINT_PACK32,
   VK_FORMAT_A2R10G10B10_SINT_PACK32,
};

unsigned
radv_format_meta_fs_key(struct radv_device *device, VkFormat format)
{
   unsigned col_format = radv_choose_spi_color_format(device, format, false, false);
   assert(col_format != V_028714_SPI_SHADER_32_AR);

   bool is_int8 = format_is_int8(format);
   bool is_int10 = format_is_int10(format);

   if (col_format == V_028714_SPI_SHADER_UINT16_ABGR && is_int8)
      return 8;
   else if (col_format == V_028714_SPI_SHADER_SINT16_ABGR && is_int8)
      return 9;
   else if (col_format == V_028714_SPI_SHADER_UINT16_ABGR && is_int10)
      return 10;
   else if (col_format == V_028714_SPI_SHADER_SINT16_ABGR && is_int10)
      return 11;
   else {
      if (col_format >= V_028714_SPI_SHADER_32_AR)
         --col_format; /* Skip V_028714_SPI_SHADER_32_AR  since there is no such VkFormat */

      --col_format; /* Skip V_028714_SPI_SHADER_ZERO */
      return col_format;
   }
}

static void
radv_blend_check_commutativity(struct radv_blend_state *blend, VkBlendOp op, VkBlendFactor src,
                               VkBlendFactor dst, unsigned chanmask)
{
   /* Src factor is allowed when it does not depend on Dst. */
   static const uint32_t src_allowed =
      (1u << VK_BLEND_FACTOR_ONE) | (1u << VK_BLEND_FACTOR_SRC_COLOR) |
      (1u << VK_BLEND_FACTOR_SRC_ALPHA) | (1u << VK_BLEND_FACTOR_SRC_ALPHA_SATURATE) |
      (1u << VK_BLEND_FACTOR_CONSTANT_COLOR) | (1u << VK_BLEND_FACTOR_CONSTANT_ALPHA) |
      (1u << VK_BLEND_FACTOR_SRC1_COLOR) | (1u << VK_BLEND_FACTOR_SRC1_ALPHA) |
      (1u << VK_BLEND_FACTOR_ZERO) | (1u << VK_BLEND_FACTOR_ONE_MINUS_SRC_COLOR) |
      (1u << VK_BLEND_FACTOR_ONE_MINUS_SRC_ALPHA) |
      (1u << VK_BLEND_FACTOR_ONE_MINUS_CONSTANT_COLOR) |
      (1u << VK_BLEND_FACTOR_ONE_MINUS_CONSTANT_ALPHA) |
      (1u << VK_BLEND_FACTOR_ONE_MINUS_SRC1_COLOR) | (1u << VK_BLEND_FACTOR_ONE_MINUS_SRC1_ALPHA);

   if (dst == VK_BLEND_FACTOR_ONE && (src_allowed & (1u << src))) {
      /* Addition is commutative, but floating point addition isn't
       * associative: subtle changes can be introduced via different
       * rounding. Be conservative, only enable for min and max.
       */
      if (op == VK_BLEND_OP_MAX || op == VK_BLEND_OP_MIN)
         blend->commutative_4bit |= chanmask;
   }
}

static struct radv_blend_state
radv_pipeline_init_blend_state(const struct radv_pipeline *pipeline,
                               const VkGraphicsPipelineCreateInfo *pCreateInfo,
                               const struct radv_graphics_pipeline_create_info *extra)
{
   const VkPipelineColorBlendStateCreateInfo *vkblend =
      radv_pipeline_get_color_blend_state(pCreateInfo);
   const VkPipelineMultisampleStateCreateInfo *vkms =
      radv_pipeline_get_multisample_state(pCreateInfo);
   struct radv_blend_state blend = {0};
   unsigned mode = V_028808_CB_NORMAL;
   int i;

   if (extra && extra->custom_blend_mode) {
      blend.single_cb_enable = true;
      mode = extra->custom_blend_mode;
   }

   blend.cb_color_control = 0;
   if (vkblend) {
      if (vkblend->logicOpEnable)
         blend.cb_color_control |= S_028808_ROP3(si_translate_blend_logic_op(vkblend->logicOp));
      else
         blend.cb_color_control |= S_028808_ROP3(V_028808_ROP3_COPY);
   }

   blend.db_alpha_to_mask = S_028B70_ALPHA_TO_MASK_OFFSET0(3) | S_028B70_ALPHA_TO_MASK_OFFSET1(1) |
                            S_028B70_ALPHA_TO_MASK_OFFSET2(0) | S_028B70_ALPHA_TO_MASK_OFFSET3(2) |
                            S_028B70_OFFSET_ROUND(1);

   if (vkms && vkms->alphaToCoverageEnable) {
      blend.db_alpha_to_mask |= S_028B70_ALPHA_TO_MASK_ENABLE(1);
      blend.need_src_alpha |= 0x1;
   }

   blend.cb_target_mask = 0;
   if (vkblend) {
      for (i = 0; i < vkblend->attachmentCount; i++) {
         const VkPipelineColorBlendAttachmentState *att = &vkblend->pAttachments[i];
         unsigned blend_cntl = 0;
         unsigned srcRGB_opt, dstRGB_opt, srcA_opt, dstA_opt;
         VkBlendOp eqRGB = att->colorBlendOp;
         VkBlendFactor srcRGB = att->srcColorBlendFactor;
         VkBlendFactor dstRGB = att->dstColorBlendFactor;
         VkBlendOp eqA = att->alphaBlendOp;
         VkBlendFactor srcA = att->srcAlphaBlendFactor;
         VkBlendFactor dstA = att->dstAlphaBlendFactor;

         blend.sx_mrt_blend_opt[i] = S_028760_COLOR_COMB_FCN(V_028760_OPT_COMB_BLEND_DISABLED) |
                                     S_028760_ALPHA_COMB_FCN(V_028760_OPT_COMB_BLEND_DISABLED);

         if (!att->colorWriteMask)
            continue;

         /* Ignore other blend targets if dual-source blending
          * is enabled to prevent wrong behaviour.
          */
         if (blend.mrt0_is_dual_src)
            continue;

         blend.cb_target_mask |= (unsigned)att->colorWriteMask << (4 * i);
         blend.cb_target_enabled_4bit |= 0xfu << (4 * i);
         if (!att->blendEnable) {
            blend.cb_blend_control[i] = blend_cntl;
            continue;
         }

         if (is_dual_src(srcRGB) || is_dual_src(dstRGB) || is_dual_src(srcA) || is_dual_src(dstA))
            if (i == 0)
               blend.mrt0_is_dual_src = true;

         if (eqRGB == VK_BLEND_OP_MIN || eqRGB == VK_BLEND_OP_MAX) {
            srcRGB = VK_BLEND_FACTOR_ONE;
            dstRGB = VK_BLEND_FACTOR_ONE;
         }
         if (eqA == VK_BLEND_OP_MIN || eqA == VK_BLEND_OP_MAX) {
            srcA = VK_BLEND_FACTOR_ONE;
            dstA = VK_BLEND_FACTOR_ONE;
         }

         radv_blend_check_commutativity(&blend, eqRGB, srcRGB, dstRGB, 0x7u << (4 * i));
         radv_blend_check_commutativity(&blend, eqA, srcA, dstA, 0x8u << (4 * i));

         /* Blending optimizations for RB+.
          * These transformations don't change the behavior.
          *
          * First, get rid of DST in the blend factors:
          *    func(src * DST, dst * 0) ---> func(src * 0, dst * SRC)
          */
         si_blend_remove_dst(&eqRGB, &srcRGB, &dstRGB, VK_BLEND_FACTOR_DST_COLOR,
                             VK_BLEND_FACTOR_SRC_COLOR);

         si_blend_remove_dst(&eqA, &srcA, &dstA, VK_BLEND_FACTOR_DST_COLOR,
                             VK_BLEND_FACTOR_SRC_COLOR);

         si_blend_remove_dst(&eqA, &srcA, &dstA, VK_BLEND_FACTOR_DST_ALPHA,
                             VK_BLEND_FACTOR_SRC_ALPHA);

         /* Look up the ideal settings from tables. */
         srcRGB_opt = si_translate_blend_opt_factor(srcRGB, false);
         dstRGB_opt = si_translate_blend_opt_factor(dstRGB, false);
         srcA_opt = si_translate_blend_opt_factor(srcA, true);
         dstA_opt = si_translate_blend_opt_factor(dstA, true);

         /* Handle interdependencies. */
         if (si_blend_factor_uses_dst(srcRGB))
            dstRGB_opt = V_028760_BLEND_OPT_PRESERVE_NONE_IGNORE_NONE;
         if (si_blend_factor_uses_dst(srcA))
            dstA_opt = V_028760_BLEND_OPT_PRESERVE_NONE_IGNORE_NONE;

         if (srcRGB == VK_BLEND_FACTOR_SRC_ALPHA_SATURATE &&
             (dstRGB == VK_BLEND_FACTOR_ZERO || dstRGB == VK_BLEND_FACTOR_SRC_ALPHA ||
              dstRGB == VK_BLEND_FACTOR_SRC_ALPHA_SATURATE))
            dstRGB_opt = V_028760_BLEND_OPT_PRESERVE_NONE_IGNORE_A0;

         /* Set the final value. */
         blend.sx_mrt_blend_opt[i] =
            S_028760_COLOR_SRC_OPT(srcRGB_opt) | S_028760_COLOR_DST_OPT(dstRGB_opt) |
            S_028760_COLOR_COMB_FCN(si_translate_blend_opt_function(eqRGB)) |
            S_028760_ALPHA_SRC_OPT(srcA_opt) | S_028760_ALPHA_DST_OPT(dstA_opt) |
            S_028760_ALPHA_COMB_FCN(si_translate_blend_opt_function(eqA));
         blend_cntl |= S_028780_ENABLE(1);

         blend_cntl |= S_028780_COLOR_COMB_FCN(si_translate_blend_function(eqRGB));
         blend_cntl |= S_028780_COLOR_SRCBLEND(si_translate_blend_factor(srcRGB));
         blend_cntl |= S_028780_COLOR_DESTBLEND(si_translate_blend_factor(dstRGB));
         if (srcA != srcRGB || dstA != dstRGB || eqA != eqRGB) {
            blend_cntl |= S_028780_SEPARATE_ALPHA_BLEND(1);
            blend_cntl |= S_028780_ALPHA_COMB_FCN(si_translate_blend_function(eqA));
            blend_cntl |= S_028780_ALPHA_SRCBLEND(si_translate_blend_factor(srcA));
            blend_cntl |= S_028780_ALPHA_DESTBLEND(si_translate_blend_factor(dstA));
         }
         blend.cb_blend_control[i] = blend_cntl;

         blend.blend_enable_4bit |= 0xfu << (i * 4);

         if (srcRGB == VK_BLEND_FACTOR_SRC_ALPHA || dstRGB == VK_BLEND_FACTOR_SRC_ALPHA ||
             srcRGB == VK_BLEND_FACTOR_SRC_ALPHA_SATURATE ||
             dstRGB == VK_BLEND_FACTOR_SRC_ALPHA_SATURATE ||
             srcRGB == VK_BLEND_FACTOR_ONE_MINUS_SRC_ALPHA ||
             dstRGB == VK_BLEND_FACTOR_ONE_MINUS_SRC_ALPHA)
            blend.need_src_alpha |= 1 << i;
      }
      for (i = vkblend->attachmentCount; i < 8; i++) {
         blend.cb_blend_control[i] = 0;
         blend.sx_mrt_blend_opt[i] = S_028760_COLOR_COMB_FCN(V_028760_OPT_COMB_BLEND_DISABLED) |
                                     S_028760_ALPHA_COMB_FCN(V_028760_OPT_COMB_BLEND_DISABLED);
      }
   }

   if (pipeline->device->physical_device->rad_info.has_rbplus) {
      /* Disable RB+ blend optimizations for dual source blending. */
      if (blend.mrt0_is_dual_src) {
         for (i = 0; i < 8; i++) {
            blend.sx_mrt_blend_opt[i] = S_028760_COLOR_COMB_FCN(V_028760_OPT_COMB_NONE) |
                                        S_028760_ALPHA_COMB_FCN(V_028760_OPT_COMB_NONE);
         }
      }

      /* RB+ doesn't work with dual source blending, logic op and
       * RESOLVE.
       */
      if (blend.mrt0_is_dual_src || (vkblend && vkblend->logicOpEnable) ||
          mode == V_028808_CB_RESOLVE)
         blend.cb_color_control |= S_028808_DISABLE_DUAL_QUAD(1);
   }

   if (blend.cb_target_mask)
      blend.cb_color_control |= S_028808_MODE(mode);
   else
      blend.cb_color_control |= S_028808_MODE(V_028808_CB_DISABLE);

   radv_pipeline_compute_spi_color_formats(pipeline, pCreateInfo, &blend);
   return blend;
}

static uint32_t
si_translate_fill(VkPolygonMode func)
{
   switch (func) {
   case VK_POLYGON_MODE_FILL:
      return V_028814_X_DRAW_TRIANGLES;
   case VK_POLYGON_MODE_LINE:
      return V_028814_X_DRAW_LINES;
   case VK_POLYGON_MODE_POINT:
      return V_028814_X_DRAW_POINTS;
   default:
      assert(0);
      return V_028814_X_DRAW_POINTS;
   }
}

static uint8_t
radv_pipeline_get_ps_iter_samples(const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   const VkPipelineMultisampleStateCreateInfo *vkms = pCreateInfo->pMultisampleState;
   RADV_FROM_HANDLE(radv_render_pass, pass, pCreateInfo->renderPass);
   struct radv_subpass *subpass = &pass->subpasses[pCreateInfo->subpass];
   uint32_t ps_iter_samples = 1;
   uint32_t num_samples;

   /* From the Vulkan 1.1.129 spec, 26.7. Sample Shading:
    *
    * "If the VK_AMD_mixed_attachment_samples extension is enabled and the
    *  subpass uses color attachments, totalSamples is the number of
    *  samples of the color attachments. Otherwise, totalSamples is the
    *  value of VkPipelineMultisampleStateCreateInfo::rasterizationSamples
    *  specified at pipeline creation time."
    */
   if (subpass->has_color_att) {
      num_samples = subpass->color_sample_count;
   } else {
      num_samples = vkms->rasterizationSamples;
   }

   if (vkms->sampleShadingEnable) {
      ps_iter_samples = ceilf(vkms->minSampleShading * num_samples);
      ps_iter_samples = util_next_power_of_two(ps_iter_samples);
   }
   return ps_iter_samples;
}

static bool
radv_is_depth_write_enabled(const VkPipelineDepthStencilStateCreateInfo *pCreateInfo)
{
   return pCreateInfo->depthTestEnable && pCreateInfo->depthWriteEnable &&
          pCreateInfo->depthCompareOp != VK_COMPARE_OP_NEVER;
}

static bool
radv_writes_stencil(const VkStencilOpState *state)
{
   return state->writeMask &&
          (state->failOp != VK_STENCIL_OP_KEEP || state->passOp != VK_STENCIL_OP_KEEP ||
           state->depthFailOp != VK_STENCIL_OP_KEEP);
}

static bool
radv_is_stencil_write_enabled(const VkPipelineDepthStencilStateCreateInfo *pCreateInfo)
{
   return pCreateInfo->stencilTestEnable &&
          (radv_writes_stencil(&pCreateInfo->front) || radv_writes_stencil(&pCreateInfo->back));
}

static bool
radv_is_ds_write_enabled(const VkPipelineDepthStencilStateCreateInfo *pCreateInfo)
{
   return radv_is_depth_write_enabled(pCreateInfo) || radv_is_stencil_write_enabled(pCreateInfo);
}

static bool
radv_order_invariant_stencil_op(VkStencilOp op)
{
   /* REPLACE is normally order invariant, except when the stencil
    * reference value is written by the fragment shader. Tracking this
    * interaction does not seem worth the effort, so be conservative.
    */
   return op != VK_STENCIL_OP_INCREMENT_AND_CLAMP && op != VK_STENCIL_OP_DECREMENT_AND_CLAMP &&
          op != VK_STENCIL_OP_REPLACE;
}

static bool
radv_order_invariant_stencil_state(const VkStencilOpState *state)
{
   /* Compute whether, assuming Z writes are disabled, this stencil state
    * is order invariant in the sense that the set of passing fragments as
    * well as the final stencil buffer result does not depend on the order
    * of fragments.
    */
   return !state->writeMask ||
          /* The following assumes that Z writes are disabled. */
          (state->compareOp == VK_COMPARE_OP_ALWAYS &&
           radv_order_invariant_stencil_op(state->passOp) &&
           radv_order_invariant_stencil_op(state->depthFailOp)) ||
          (state->compareOp == VK_COMPARE_OP_NEVER &&
           radv_order_invariant_stencil_op(state->failOp));
}

static bool
radv_pipeline_has_dynamic_ds_states(const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   VkDynamicState ds_states[] = {
      VK_DYNAMIC_STATE_DEPTH_TEST_ENABLE_EXT, VK_DYNAMIC_STATE_DEPTH_WRITE_ENABLE_EXT,
      VK_DYNAMIC_STATE_DEPTH_COMPARE_OP_EXT,  VK_DYNAMIC_STATE_STENCIL_TEST_ENABLE_EXT,
      VK_DYNAMIC_STATE_STENCIL_OP_EXT,
   };

   for (uint32_t i = 0; i < ARRAY_SIZE(ds_states); i++) {
      if (radv_is_state_dynamic(pCreateInfo, ds_states[i]))
         return true;
   }

   return false;
}

static bool
radv_pipeline_out_of_order_rast(struct radv_pipeline *pipeline,
                                const struct radv_blend_state *blend,
                                const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   RADV_FROM_HANDLE(radv_render_pass, pass, pCreateInfo->renderPass);
   struct radv_subpass *subpass = pass->subpasses + pCreateInfo->subpass;
   const VkPipelineDepthStencilStateCreateInfo *vkds =
      radv_pipeline_get_depth_stencil_state(pCreateInfo);
   const VkPipelineColorBlendStateCreateInfo *vkblend =
      radv_pipeline_get_color_blend_state(pCreateInfo);
   unsigned colormask = blend->cb_target_enabled_4bit;

   if (!pipeline->device->physical_device->out_of_order_rast_allowed)
      return false;

   /* Be conservative if a logic operation is enabled with color buffers. */
   if (colormask && vkblend && vkblend->logicOpEnable)
      return false;

   /* Be conservative if an extended dynamic depth/stencil state is
    * enabled because the driver can't update out-of-order rasterization
    * dynamically.
    */
   if (radv_pipeline_has_dynamic_ds_states(pCreateInfo))
      return false;

   /* Default depth/stencil invariance when no attachment is bound. */
   struct radv_dsa_order_invariance dsa_order_invariant = {.zs = true, .pass_set = true};

   if (vkds) {
      struct radv_render_pass_attachment *attachment =
         pass->attachments + subpass->depth_stencil_attachment->attachment;
      bool has_stencil = vk_format_has_stencil(attachment->format);
      struct radv_dsa_order_invariance order_invariance[2];
      struct radv_shader_variant *ps = pipeline->shaders[MESA_SHADER_FRAGMENT];

      /* Compute depth/stencil order invariance in order to know if
       * it's safe to enable out-of-order.
       */
      bool zfunc_is_ordered = vkds->depthCompareOp == VK_COMPARE_OP_NEVER ||
                              vkds->depthCompareOp == VK_COMPARE_OP_LESS ||
                              vkds->depthCompareOp == VK_COMPARE_OP_LESS_OR_EQUAL ||
                              vkds->depthCompareOp == VK_COMPARE_OP_GREATER ||
                              vkds->depthCompareOp == VK_COMPARE_OP_GREATER_OR_EQUAL;

      bool nozwrite_and_order_invariant_stencil =
         !radv_is_ds_write_enabled(vkds) ||
         (!radv_is_depth_write_enabled(vkds) && radv_order_invariant_stencil_state(&vkds->front) &&
          radv_order_invariant_stencil_state(&vkds->back));

      order_invariance[1].zs = nozwrite_and_order_invariant_stencil ||
                               (!radv_is_stencil_write_enabled(vkds) && zfunc_is_ordered);
      order_invariance[0].zs = !radv_is_depth_write_enabled(vkds) || zfunc_is_ordered;

      order_invariance[1].pass_set =
         nozwrite_and_order_invariant_stencil ||
         (!radv_is_stencil_write_enabled(vkds) && (vkds->depthCompareOp == VK_COMPARE_OP_ALWAYS ||
                                                   vkds->depthCompareOp == VK_COMPARE_OP_NEVER));
      order_invariance[0].pass_set =
         !radv_is_depth_write_enabled(vkds) || (vkds->depthCompareOp == VK_COMPARE_OP_ALWAYS ||
                                                vkds->depthCompareOp == VK_COMPARE_OP_NEVER);

      dsa_order_invariant = order_invariance[has_stencil];
      if (!dsa_order_invariant.zs)
         return false;

      /* The set of PS invocations is always order invariant,
       * except when early Z/S tests are requested.
       */
      if (ps && ps->info.ps.writes_memory && ps->info.ps.early_fragment_test &&
          !dsa_order_invariant.pass_set)
         return false;

      /* Determine if out-of-order rasterization should be disabled
       * when occlusion queries are used.
       */
      pipeline->graphics.disable_out_of_order_rast_for_occlusion = !dsa_order_invariant.pass_set;
   }

   /* No color buffers are enabled for writing. */
   if (!colormask)
      return true;

   unsigned blendmask = colormask & blend->blend_enable_4bit;

   if (blendmask) {
      /* Only commutative blending. */
      if (blendmask & ~blend->commutative_4bit)
         return false;

      if (!dsa_order_invariant.pass_set)
         return false;
   }

   if (colormask & ~blendmask)
      return false;

   return true;
}

static const VkConservativeRasterizationModeEXT
radv_get_conservative_raster_mode(const VkPipelineRasterizationStateCreateInfo *pCreateInfo)
{
   const VkPipelineRasterizationConservativeStateCreateInfoEXT *conservative_raster =
      vk_find_struct_const(pCreateInfo->pNext,
                           PIPELINE_RASTERIZATION_CONSERVATIVE_STATE_CREATE_INFO_EXT);

   if (!conservative_raster)
      return VK_CONSERVATIVE_RASTERIZATION_MODE_DISABLED_EXT;
   return conservative_raster->conservativeRasterizationMode;
}

static void
radv_pipeline_init_multisample_state(struct radv_pipeline *pipeline,
                                     const struct radv_blend_state *blend,
                                     const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   const VkPipelineMultisampleStateCreateInfo *vkms =
      radv_pipeline_get_multisample_state(pCreateInfo);
   struct radv_multisample_state *ms = &pipeline->graphics.ms;
   unsigned num_tile_pipes = pipeline->device->physical_device->rad_info.num_tile_pipes;
   const VkConservativeRasterizationModeEXT mode =
      radv_get_conservative_raster_mode(pCreateInfo->pRasterizationState);
   bool out_of_order_rast = false;
   int ps_iter_samples = 1;
   uint32_t mask = 0xffff;

   if (vkms) {
      ms->num_samples = vkms->rasterizationSamples;

      /* From the Vulkan 1.1.129 spec, 26.7. Sample Shading:
       *
       * "Sample shading is enabled for a graphics pipeline:
       *
       * - If the interface of the fragment shader entry point of the
       *   graphics pipeline includes an input variable decorated
       *   with SampleId or SamplePosition. In this case
       *   minSampleShadingFactor takes the value 1.0.
       * - Else if the sampleShadingEnable member of the
       *   VkPipelineMultisampleStateCreateInfo structure specified
       *   when creating the graphics pipeline is set to VK_TRUE. In
       *   this case minSampleShadingFactor takes the value of
       *   VkPipelineMultisampleStateCreateInfo::minSampleShading.
       *
       * Otherwise, sample shading is considered disabled."
       */
      if (pipeline->shaders[MESA_SHADER_FRAGMENT]->info.ps.uses_sample_shading) {
         ps_iter_samples = ms->num_samples;
      } else {
         ps_iter_samples = radv_pipeline_get_ps_iter_samples(pCreateInfo);
      }
   } else {
      ms->num_samples = 1;
   }

   const struct VkPipelineRasterizationStateRasterizationOrderAMD *raster_order =
      vk_find_struct_const(pCreateInfo->pRasterizationState->pNext,
                           PIPELINE_RASTERIZATION_STATE_RASTERIZATION_ORDER_AMD);
   if (raster_order && raster_order->rasterizationOrder == VK_RASTERIZATION_ORDER_RELAXED_AMD) {
      /* Out-of-order rasterization is explicitly enabled by the
       * application.
       */
      out_of_order_rast = true;
   } else {
      /* Determine if the driver can enable out-of-order
       * rasterization internally.
       */
      out_of_order_rast = radv_pipeline_out_of_order_rast(pipeline, blend, pCreateInfo);
   }

   ms->pa_sc_aa_config = 0;
   ms->db_eqaa = S_028804_HIGH_QUALITY_INTERSECTIONS(1) | S_028804_INCOHERENT_EQAA_READS(1) |
                 S_028804_INTERPOLATE_COMP_Z(1) | S_028804_STATIC_ANCHOR_ASSOCIATIONS(1);

   /* Adjust MSAA state if conservative rasterization is enabled. */
   if (mode != VK_CONSERVATIVE_RASTERIZATION_MODE_DISABLED_EXT) {
      ms->pa_sc_aa_config |= S_028BE0_AA_MASK_CENTROID_DTMN(1);

      ms->db_eqaa |=
         S_028804_ENABLE_POSTZ_OVERRASTERIZATION(1) | S_028804_OVERRASTERIZATION_AMOUNT(4);
   }

   ms->pa_sc_mode_cntl_1 =
      S_028A4C_WALK_FENCE_ENABLE(1) | // TODO linear dst fixes
      S_028A4C_WALK_FENCE_SIZE(num_tile_pipes == 2 ? 2 : 3) |
      S_028A4C_OUT_OF_ORDER_PRIMITIVE_ENABLE(out_of_order_rast) |
      S_028A4C_OUT_OF_ORDER_WATER_MARK(0x7) |
      /* always 1: */
      S_028A4C_WALK_ALIGN8_PRIM_FITS_ST(1) | S_028A4C_SUPERTILE_WALK_ORDER_ENABLE(1) |
      S_028A4C_TILE_WALK_ORDER_ENABLE(1) | S_028A4C_MULTI_SHADER_ENGINE_PRIM_DISCARD_ENABLE(1) |
      S_028A4C_FORCE_EOV_CNTDWN_ENABLE(1) | S_028A4C_FORCE_EOV_REZ_ENABLE(1);
   ms->pa_sc_mode_cntl_0 = S_028A48_ALTERNATE_RBS_PER_TILE(
                              pipeline->device->physical_device->rad_info.chip_class >= GFX9) |
                           S_028A48_VPORT_SCISSOR_ENABLE(1);

   const VkPipelineRasterizationLineStateCreateInfoEXT *rast_line = vk_find_struct_const(
      pCreateInfo->pRasterizationState->pNext, PIPELINE_RASTERIZATION_LINE_STATE_CREATE_INFO_EXT);
   if (rast_line) {
      ms->pa_sc_mode_cntl_0 |= S_028A48_LINE_STIPPLE_ENABLE(rast_line->stippledLineEnable);
      if (rast_line->lineRasterizationMode == VK_LINE_RASTERIZATION_MODE_BRESENHAM_EXT) {
         /* From the Vulkan spec 1.1.129:
          *
          * "When VK_LINE_RASTERIZATION_MODE_BRESENHAM_EXT lines
          *  are being rasterized, sample locations may all be
          *  treated as being at the pixel center (this may
          *  affect attribute and depth interpolation)."
          */
         ms->num_samples = 1;
      }
   }

   if (ms->num_samples > 1) {
      RADV_FROM_HANDLE(radv_render_pass, pass, pCreateInfo->renderPass);
      struct radv_subpass *subpass = &pass->subpasses[pCreateInfo->subpass];
      uint32_t z_samples =
         subpass->depth_stencil_attachment ? subpass->depth_sample_count : ms->num_samples;
      unsigned log_samples = util_logbase2(ms->num_samples);
      unsigned log_z_samples = util_logbase2(z_samples);
      unsigned log_ps_iter_samples = util_logbase2(ps_iter_samples);
      ms->pa_sc_mode_cntl_0 |= S_028A48_MSAA_ENABLE(1);
      ms->db_eqaa |= S_028804_MAX_ANCHOR_SAMPLES(log_z_samples) |
                     S_028804_PS_ITER_SAMPLES(log_ps_iter_samples) |
                     S_028804_MASK_EXPORT_NUM_SAMPLES(log_samples) |
                     S_028804_ALPHA_TO_MASK_NUM_SAMPLES(log_samples);
      ms->pa_sc_aa_config |=
         S_028BE0_MSAA_NUM_SAMPLES(log_samples) |
         S_028BE0_MAX_SAMPLE_DIST(radv_get_default_max_sample_dist(log_samples)) |
         S_028BE0_MSAA_EXPOSED_SAMPLES(log_samples) | /* CM_R_028BE0_PA_SC_AA_CONFIG */
         S_028BE0_COVERED_CENTROID_IS_CENTER(
            pipeline->device->physical_device->rad_info.chip_class >= GFX10_3);
      ms->pa_sc_mode_cntl_1 |= S_028A4C_PS_ITER_SAMPLE(ps_iter_samples > 1);
      if (ps_iter_samples > 1)
         pipeline->graphics.spi_baryc_cntl |= S_0286E0_POS_FLOAT_LOCATION(2);
   }

   if (vkms && vkms->pSampleMask) {
      mask = vkms->pSampleMask[0] & 0xffff;
   }

   ms->pa_sc_aa_mask[0] = mask | (mask << 16);
   ms->pa_sc_aa_mask[1] = mask | (mask << 16);
}

static void
gfx103_pipeline_init_vrs_state(struct radv_pipeline *pipeline,
                               const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   const VkPipelineMultisampleStateCreateInfo *vkms =
      radv_pipeline_get_multisample_state(pCreateInfo);
   struct radv_shader_variant *ps = pipeline->shaders[MESA_SHADER_FRAGMENT];
   struct radv_multisample_state *ms = &pipeline->graphics.ms;
   struct radv_vrs_state *vrs = &pipeline->graphics.vrs;

   if (vkms && (vkms->sampleShadingEnable || ps->info.ps.uses_sample_shading ||
                ps->info.ps.reads_sample_mask_in)) {
      /* Disable VRS and use the rates from PS_ITER_SAMPLES if:
       *
       * 1) sample shading is enabled or per-sample interpolation is
       *    used by the fragment shader
       * 2) the fragment shader reads gl_SampleMaskIn because the
       *    16-bit sample coverage mask isn't enough for MSAA8x and
       *    2x2 coarse shading isn't enough.
       */
      vrs->pa_cl_vrs_cntl = S_028848_SAMPLE_ITER_COMBINER_MODE(V_028848_VRS_COMB_MODE_OVERRIDE);

      /* Make sure sample shading is enabled even if only MSAA1x is
       * used because the SAMPLE_ITER combiner is in passthrough
       * mode if PS_ITER_SAMPLE is 0, and it uses the per-draw rate.
       * The default VRS rate when sample shading is enabled is 1x1.
       */
      if (!G_028A4C_PS_ITER_SAMPLE(ms->pa_sc_mode_cntl_1))
         ms->pa_sc_mode_cntl_1 |= S_028A4C_PS_ITER_SAMPLE(1);
   } else {
      vrs->pa_cl_vrs_cntl = S_028848_SAMPLE_ITER_COMBINER_MODE(V_028848_VRS_COMB_MODE_PASSTHRU);
   }

   /* The primitive combiner is always passthrough. */
   vrs->pa_cl_vrs_cntl |= S_028848_PRIMITIVE_RATE_COMBINER_MODE(V_028848_VRS_COMB_MODE_PASSTHRU);
}

static bool
radv_prim_can_use_guardband(enum VkPrimitiveTopology topology)
{
   switch (topology) {
   case VK_PRIMITIVE_TOPOLOGY_POINT_LIST:
   case VK_PRIMITIVE_TOPOLOGY_LINE_LIST:
   case VK_PRIMITIVE_TOPOLOGY_LINE_STRIP:
   case VK_PRIMITIVE_TOPOLOGY_LINE_LIST_WITH_ADJACENCY:
   case VK_PRIMITIVE_TOPOLOGY_LINE_STRIP_WITH_ADJACENCY:
      return false;
   case VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST:
   case VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP:
   case VK_PRIMITIVE_TOPOLOGY_TRIANGLE_FAN:
   case VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST_WITH_ADJACENCY:
   case VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP_WITH_ADJACENCY:
   case VK_PRIMITIVE_TOPOLOGY_PATCH_LIST:
      return true;
   default:
      unreachable("unhandled primitive type");
   }
}

static uint32_t
si_conv_gl_prim_to_gs_out(unsigned gl_prim)
{
   switch (gl_prim) {
   case 0: /* GL_POINTS */
      return V_028A6C_POINTLIST;
   case 1:      /* GL_LINES */
   case 3:      /* GL_LINE_STRIP */
   case 0xA:    /* GL_LINE_STRIP_ADJACENCY_ARB */
   case 0x8E7A: /* GL_ISOLINES */
      return V_028A6C_LINESTRIP;

   case 4:   /* GL_TRIANGLES */
   case 0xc: /* GL_TRIANGLES_ADJACENCY_ARB */
   case 5:   /* GL_TRIANGLE_STRIP */
   case 7:   /* GL_QUADS */
      return V_028A6C_TRISTRIP;
   default:
      assert(0);
      return 0;
   }
}

static uint32_t
si_conv_prim_to_gs_out(enum VkPrimitiveTopology topology)
{
   switch (topology) {
   case VK_PRIMITIVE_TOPOLOGY_POINT_LIST:
   case VK_PRIMITIVE_TOPOLOGY_PATCH_LIST:
      return V_028A6C_POINTLIST;
   case VK_PRIMITIVE_TOPOLOGY_LINE_LIST:
   case VK_PRIMITIVE_TOPOLOGY_LINE_STRIP:
   case VK_PRIMITIVE_TOPOLOGY_LINE_LIST_WITH_ADJACENCY:
   case VK_PRIMITIVE_TOPOLOGY_LINE_STRIP_WITH_ADJACENCY:
      return V_028A6C_LINESTRIP;
   case VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST:
   case VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP:
   case VK_PRIMITIVE_TOPOLOGY_TRIANGLE_FAN:
   case VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST_WITH_ADJACENCY:
   case VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP_WITH_ADJACENCY:
      return V_028A6C_TRISTRIP;
   default:
      assert(0);
      return 0;
   }
}

static uint64_t
radv_dynamic_state_mask(VkDynamicState state)
{
   switch (state) {
   case VK_DYNAMIC_STATE_VIEWPORT:
   case VK_DYNAMIC_STATE_VIEWPORT_WITH_COUNT_EXT:
      return RADV_DYNAMIC_VIEWPORT;
   case VK_DYNAMIC_STATE_SCISSOR:
   case VK_DYNAMIC_STATE_SCISSOR_WITH_COUNT_EXT:
      return RADV_DYNAMIC_SCISSOR;
   case VK_DYNAMIC_STATE_LINE_WIDTH:
      return RADV_DYNAMIC_LINE_WIDTH;
   case VK_DYNAMIC_STATE_DEPTH_BIAS:
      return RADV_DYNAMIC_DEPTH_BIAS;
   case VK_DYNAMIC_STATE_BLEND_CONSTANTS:
      return RADV_DYNAMIC_BLEND_CONSTANTS;
   case VK_DYNAMIC_STATE_DEPTH_BOUNDS:
      return RADV_DYNAMIC_DEPTH_BOUNDS;
   case VK_DYNAMIC_STATE_STENCIL_COMPARE_MASK:
      return RADV_DYNAMIC_STENCIL_COMPARE_MASK;
   case VK_DYNAMIC_STATE_STENCIL_WRITE_MASK:
      return RADV_DYNAMIC_STENCIL_WRITE_MASK;
   case VK_DYNAMIC_STATE_STENCIL_REFERENCE:
      return RADV_DYNAMIC_STENCIL_REFERENCE;
   case VK_DYNAMIC_STATE_DISCARD_RECTANGLE_EXT:
      return RADV_DYNAMIC_DISCARD_RECTANGLE;
   case VK_DYNAMIC_STATE_SAMPLE_LOCATIONS_EXT:
      return RADV_DYNAMIC_SAMPLE_LOCATIONS;
   case VK_DYNAMIC_STATE_LINE_STIPPLE_EXT:
      return RADV_DYNAMIC_LINE_STIPPLE;
   case VK_DYNAMIC_STATE_CULL_MODE_EXT:
      return RADV_DYNAMIC_CULL_MODE;
   case VK_DYNAMIC_STATE_FRONT_FACE_EXT:
      return RADV_DYNAMIC_FRONT_FACE;
   case VK_DYNAMIC_STATE_PRIMITIVE_TOPOLOGY_EXT:
      return RADV_DYNAMIC_PRIMITIVE_TOPOLOGY;
   case VK_DYNAMIC_STATE_DEPTH_TEST_ENABLE_EXT:
      return RADV_DYNAMIC_DEPTH_TEST_ENABLE;
   case VK_DYNAMIC_STATE_DEPTH_WRITE_ENABLE_EXT:
      return RADV_DYNAMIC_DEPTH_WRITE_ENABLE;
   case VK_DYNAMIC_STATE_DEPTH_COMPARE_OP_EXT:
      return RADV_DYNAMIC_DEPTH_COMPARE_OP;
   case VK_DYNAMIC_STATE_DEPTH_BOUNDS_TEST_ENABLE_EXT:
      return RADV_DYNAMIC_DEPTH_BOUNDS_TEST_ENABLE;
   case VK_DYNAMIC_STATE_STENCIL_TEST_ENABLE_EXT:
      return RADV_DYNAMIC_STENCIL_TEST_ENABLE;
   case VK_DYNAMIC_STATE_STENCIL_OP_EXT:
      return RADV_DYNAMIC_STENCIL_OP;
   case VK_DYNAMIC_STATE_VERTEX_INPUT_BINDING_STRIDE_EXT:
      return RADV_DYNAMIC_VERTEX_INPUT_BINDING_STRIDE;
   case VK_DYNAMIC_STATE_FRAGMENT_SHADING_RATE_KHR:
      return RADV_DYNAMIC_FRAGMENT_SHADING_RATE;
   case VK_DYNAMIC_STATE_PATCH_CONTROL_POINTS_EXT:
      return RADV_DYNAMIC_PATCH_CONTROL_POINTS;
   case VK_DYNAMIC_STATE_RASTERIZER_DISCARD_ENABLE_EXT:
      return RADV_DYNAMIC_RASTERIZER_DISCARD_ENABLE;
   case VK_DYNAMIC_STATE_DEPTH_BIAS_ENABLE_EXT:
      return RADV_DYNAMIC_DEPTH_BIAS_ENABLE;
   case VK_DYNAMIC_STATE_LOGIC_OP_EXT:
      return RADV_DYNAMIC_LOGIC_OP;
   case VK_DYNAMIC_STATE_PRIMITIVE_RESTART_ENABLE_EXT:
      return RADV_DYNAMIC_PRIMITIVE_RESTART_ENABLE;
   default:
      unreachable("Unhandled dynamic state");
   }
}

static uint64_t
radv_pipeline_needed_dynamic_state(const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   uint64_t states = RADV_DYNAMIC_ALL;

   /* If rasterization is disabled we do not care about any of the
    * dynamic states, since they are all rasterization related only,
    * except primitive topology, primitive restart enable and vertex
    * binding stride.
    */
   if (pCreateInfo->pRasterizationState->rasterizerDiscardEnable &&
       !radv_is_state_dynamic(pCreateInfo, VK_DYNAMIC_STATE_RASTERIZER_DISCARD_ENABLE_EXT))
      return RADV_DYNAMIC_PRIMITIVE_TOPOLOGY | RADV_DYNAMIC_VERTEX_INPUT_BINDING_STRIDE |
             RADV_DYNAMIC_PRIMITIVE_RESTART_ENABLE;

   if (!pCreateInfo->pRasterizationState->depthBiasEnable &&
       !radv_is_state_dynamic(pCreateInfo, VK_DYNAMIC_STATE_DEPTH_BIAS_ENABLE_EXT))
      states &= ~RADV_DYNAMIC_DEPTH_BIAS;

   if (!pCreateInfo->pDepthStencilState ||
       (!pCreateInfo->pDepthStencilState->depthBoundsTestEnable &&
        !radv_is_state_dynamic(pCreateInfo, VK_DYNAMIC_STATE_DEPTH_BOUNDS_TEST_ENABLE_EXT)))
      states &= ~RADV_DYNAMIC_DEPTH_BOUNDS;

   if (!pCreateInfo->pDepthStencilState ||
       (!pCreateInfo->pDepthStencilState->stencilTestEnable &&
        !radv_is_state_dynamic(pCreateInfo, VK_DYNAMIC_STATE_STENCIL_TEST_ENABLE_EXT)))
      states &= ~(RADV_DYNAMIC_STENCIL_COMPARE_MASK | RADV_DYNAMIC_STENCIL_WRITE_MASK |
                  RADV_DYNAMIC_STENCIL_REFERENCE);

   if (!vk_find_struct_const(pCreateInfo->pNext, PIPELINE_DISCARD_RECTANGLE_STATE_CREATE_INFO_EXT))
      states &= ~RADV_DYNAMIC_DISCARD_RECTANGLE;

   if (!pCreateInfo->pMultisampleState ||
       !vk_find_struct_const(pCreateInfo->pMultisampleState->pNext,
                             PIPELINE_SAMPLE_LOCATIONS_STATE_CREATE_INFO_EXT))
      states &= ~RADV_DYNAMIC_SAMPLE_LOCATIONS;

   if (!pCreateInfo->pRasterizationState ||
       !vk_find_struct_const(pCreateInfo->pRasterizationState->pNext,
                             PIPELINE_RASTERIZATION_LINE_STATE_CREATE_INFO_EXT))
      states &= ~RADV_DYNAMIC_LINE_STIPPLE;

   if (!vk_find_struct_const(pCreateInfo->pNext,
                             PIPELINE_FRAGMENT_SHADING_RATE_STATE_CREATE_INFO_KHR) &&
       !radv_is_state_dynamic(pCreateInfo, VK_DYNAMIC_STATE_FRAGMENT_SHADING_RATE_KHR))
      states &= ~RADV_DYNAMIC_FRAGMENT_SHADING_RATE;

   /* TODO: blend constants & line width. */

   return states;
}

static struct radv_ia_multi_vgt_param_helpers
radv_compute_ia_multi_vgt_param_helpers(struct radv_pipeline *pipeline)
{
   struct radv_ia_multi_vgt_param_helpers ia_multi_vgt_param = {0};
   const struct radv_device *device = pipeline->device;

   if (radv_pipeline_has_tess(pipeline))
      ia_multi_vgt_param.primgroup_size =
         pipeline->shaders[MESA_SHADER_TESS_CTRL]->info.num_tess_patches;
   else if (radv_pipeline_has_gs(pipeline))
      ia_multi_vgt_param.primgroup_size = 64;
   else
      ia_multi_vgt_param.primgroup_size = 128; /* recommended without a GS */

   /* GS requirement. */
   ia_multi_vgt_param.partial_es_wave = false;
   if (radv_pipeline_has_gs(pipeline) && device->physical_device->rad_info.chip_class <= GFX8)
      if (SI_GS_PER_ES / ia_multi_vgt_param.primgroup_size >= pipeline->device->gs_table_depth - 3)
         ia_multi_vgt_param.partial_es_wave = true;

   ia_multi_vgt_param.ia_switch_on_eoi = false;
   if (pipeline->shaders[MESA_SHADER_FRAGMENT]->info.ps.prim_id_input)
      ia_multi_vgt_param.ia_switch_on_eoi = true;
   if (radv_pipeline_has_gs(pipeline) && pipeline->shaders[MESA_SHADER_GEOMETRY]->info.uses_prim_id)
      ia_multi_vgt_param.ia_switch_on_eoi = true;
   if (radv_pipeline_has_tess(pipeline)) {
      /* SWITCH_ON_EOI must be set if PrimID is used. */
      if (pipeline->shaders[MESA_SHADER_TESS_CTRL]->info.uses_prim_id ||
          radv_get_shader(pipeline, MESA_SHADER_TESS_EVAL)->info.uses_prim_id)
         ia_multi_vgt_param.ia_switch_on_eoi = true;
   }

   ia_multi_vgt_param.partial_vs_wave = false;
   if (radv_pipeline_has_tess(pipeline)) {
      /* Bug with tessellation and GS on Bonaire and older 2 SE chips. */
      if ((device->physical_device->rad_info.family == CHIP_TAHITI ||
           device->physical_device->rad_info.family == CHIP_PITCAIRN ||
           device->physical_device->rad_info.family == CHIP_BONAIRE) &&
          radv_pipeline_has_gs(pipeline))
         ia_multi_vgt_param.partial_vs_wave = true;
      /* Needed for 028B6C_DISTRIBUTION_MODE != 0 */
      if (device->physical_device->rad_info.has_distributed_tess) {
         if (radv_pipeline_has_gs(pipeline)) {
            if (device->physical_device->rad_info.chip_class <= GFX8)
               ia_multi_vgt_param.partial_es_wave = true;
         } else {
            ia_multi_vgt_param.partial_vs_wave = true;
         }
      }
   }

   if (radv_pipeline_has_gs(pipeline)) {
      /* On these chips there is the possibility of a hang if the
       * pipeline uses a GS and partial_vs_wave is not set.
       *
       * This mostly does not hit 4-SE chips, as those typically set
       * ia_switch_on_eoi and then partial_vs_wave is set for pipelines
       * with GS due to another workaround.
       *
       * Reproducer: https://bugs.freedesktop.org/show_bug.cgi?id=109242
       */
      if (device->physical_device->rad_info.family == CHIP_TONGA ||
          device->physical_device->rad_info.family == CHIP_FIJI ||
          device->physical_device->rad_info.family == CHIP_POLARIS10 ||
          device->physical_device->rad_info.family == CHIP_POLARIS11 ||
          device->physical_device->rad_info.family == CHIP_POLARIS12 ||
          device->physical_device->rad_info.family == CHIP_VEGAM) {
         ia_multi_vgt_param.partial_vs_wave = true;
      }
   }

   ia_multi_vgt_param.base =
      S_028AA8_PRIMGROUP_SIZE(ia_multi_vgt_param.primgroup_size - 1) |
      /* The following field was moved to VGT_SHADER_STAGES_EN in GFX9. */
      S_028AA8_MAX_PRIMGRP_IN_WAVE(device->physical_device->rad_info.chip_class == GFX8 ? 2 : 0) |
      S_030960_EN_INST_OPT_BASIC(device->physical_device->rad_info.chip_class >= GFX9) |
      S_030960_EN_INST_OPT_ADV(device->physical_device->rad_info.chip_class >= GFX9);

   return ia_multi_vgt_param;
}

static void
radv_pipeline_init_input_assembly_state(struct radv_pipeline *pipeline,
                                        const VkGraphicsPipelineCreateInfo *pCreateInfo,
                                        const struct radv_graphics_pipeline_create_info *extra)
{
   const VkPipelineInputAssemblyStateCreateInfo *ia_state = pCreateInfo->pInputAssemblyState;
   struct radv_shader_variant *tes = pipeline->shaders[MESA_SHADER_TESS_EVAL];
   struct radv_shader_variant *gs = pipeline->shaders[MESA_SHADER_GEOMETRY];

   pipeline->graphics.can_use_guardband = radv_prim_can_use_guardband(ia_state->topology);

   if (radv_pipeline_has_gs(pipeline)) {
      if (si_conv_gl_prim_to_gs_out(gs->info.gs.output_prim) == V_028A6C_TRISTRIP)
         pipeline->graphics.can_use_guardband = true;
   } else if (radv_pipeline_has_tess(pipeline)) {
      if (!tes->info.tes.point_mode &&
          si_conv_gl_prim_to_gs_out(tes->info.tes.primitive_mode) == V_028A6C_TRISTRIP)
         pipeline->graphics.can_use_guardband = true;
   }

   if (extra && extra->use_rectlist) {
      pipeline->graphics.can_use_guardband = true;
   }

   pipeline->graphics.ia_multi_vgt_param = radv_compute_ia_multi_vgt_param_helpers(pipeline);
}

static void
radv_pipeline_init_dynamic_state(struct radv_pipeline *pipeline,
                                 const VkGraphicsPipelineCreateInfo *pCreateInfo,
                                 const struct radv_graphics_pipeline_create_info *extra)
{
   uint64_t needed_states = radv_pipeline_needed_dynamic_state(pCreateInfo);
   uint64_t states = needed_states;
   RADV_FROM_HANDLE(radv_render_pass, pass, pCreateInfo->renderPass);
   struct radv_subpass *subpass = &pass->subpasses[pCreateInfo->subpass];

   pipeline->dynamic_state = default_dynamic_state;
   pipeline->graphics.needed_dynamic_state = needed_states;

   if (pCreateInfo->pDynamicState) {
      /* Remove all of the states that are marked as dynamic */
      uint32_t count = pCreateInfo->pDynamicState->dynamicStateCount;
      for (uint32_t s = 0; s < count; s++)
         states &= ~radv_dynamic_state_mask(pCreateInfo->pDynamicState->pDynamicStates[s]);
   }

   struct radv_dynamic_state *dynamic = &pipeline->dynamic_state;

   if (needed_states & RADV_DYNAMIC_VIEWPORT) {
      assert(pCreateInfo->pViewportState);

      dynamic->viewport.count = pCreateInfo->pViewportState->viewportCount;
      if (states & RADV_DYNAMIC_VIEWPORT) {
         typed_memcpy(dynamic->viewport.viewports, pCreateInfo->pViewportState->pViewports,
                      pCreateInfo->pViewportState->viewportCount);
      }
   }

   if (needed_states & RADV_DYNAMIC_SCISSOR) {
      dynamic->scissor.count = pCreateInfo->pViewportState->scissorCount;
      if (states & RADV_DYNAMIC_SCISSOR) {
         typed_memcpy(dynamic->scissor.scissors, pCreateInfo->pViewportState->pScissors,
                      pCreateInfo->pViewportState->scissorCount);
      }
   }

   if (states & RADV_DYNAMIC_LINE_WIDTH) {
      assert(pCreateInfo->pRasterizationState);
      dynamic->line_width = pCreateInfo->pRasterizationState->lineWidth;
   }

   if (states & RADV_DYNAMIC_DEPTH_BIAS) {
      assert(pCreateInfo->pRasterizationState);
      dynamic->depth_bias.bias = pCreateInfo->pRasterizationState->depthBiasConstantFactor;
      dynamic->depth_bias.clamp = pCreateInfo->pRasterizationState->depthBiasClamp;
      dynamic->depth_bias.slope = pCreateInfo->pRasterizationState->depthBiasSlopeFactor;
   }

   /* Section 9.2 of the Vulkan 1.0.15 spec says:
    *
    *    pColorBlendState is [...] NULL if the pipeline has rasterization
    *    disabled or if the subpass of the render pass the pipeline is
    *    created against does not use any color attachments.
    */
   if (subpass->has_color_att && states & RADV_DYNAMIC_BLEND_CONSTANTS) {
      assert(pCreateInfo->pColorBlendState);
      typed_memcpy(dynamic->blend_constants, pCreateInfo->pColorBlendState->blendConstants, 4);
   }

   if (states & RADV_DYNAMIC_CULL_MODE) {
      dynamic->cull_mode = pCreateInfo->pRasterizationState->cullMode;
   }

   if (states & RADV_DYNAMIC_FRONT_FACE) {
      dynamic->front_face = pCreateInfo->pRasterizationState->frontFace;
   }

   if (states & RADV_DYNAMIC_PRIMITIVE_TOPOLOGY) {
      dynamic->primitive_topology = si_translate_prim(pCreateInfo->pInputAssemblyState->topology);
      if (extra && extra->use_rectlist) {
         dynamic->primitive_topology = V_008958_DI_PT_RECTLIST;
      }
   }

   /* If there is no depthstencil attachment, then don't read
    * pDepthStencilState. The Vulkan spec states that pDepthStencilState may
    * be NULL in this case. Even if pDepthStencilState is non-NULL, there is
    * no need to override the depthstencil defaults in
    * radv_pipeline::dynamic_state when there is no depthstencil attachment.
    *
    * Section 9.2 of the Vulkan 1.0.15 spec says:
    *
    *    pDepthStencilState is [...] NULL if the pipeline has rasterization
    *    disabled or if the subpass of the render pass the pipeline is created
    *    against does not use a depth/stencil attachment.
    */
   if (needed_states && subpass->depth_stencil_attachment) {
      if (states & RADV_DYNAMIC_DEPTH_BOUNDS) {
         dynamic->depth_bounds.min = pCreateInfo->pDepthStencilState->minDepthBounds;
         dynamic->depth_bounds.max = pCreateInfo->pDepthStencilState->maxDepthBounds;
      }

      if (states & RADV_DYNAMIC_STENCIL_COMPARE_MASK) {
         dynamic->stencil_compare_mask.front = pCreateInfo->pDepthStencilState->front.compareMask;
         dynamic->stencil_compare_mask.back = pCreateInfo->pDepthStencilState->back.compareMask;
      }

      if (states & RADV_DYNAMIC_STENCIL_WRITE_MASK) {
         dynamic->stencil_write_mask.front = pCreateInfo->pDepthStencilState->front.writeMask;
         dynamic->stencil_write_mask.back = pCreateInfo->pDepthStencilState->back.writeMask;
      }

      if (states & RADV_DYNAMIC_STENCIL_REFERENCE) {
         dynamic->stencil_reference.front = pCreateInfo->pDepthStencilState->front.reference;
         dynamic->stencil_reference.back = pCreateInfo->pDepthStencilState->back.reference;
      }

      if (states & RADV_DYNAMIC_DEPTH_TEST_ENABLE) {
         dynamic->depth_test_enable = pCreateInfo->pDepthStencilState->depthTestEnable;
      }

      if (states & RADV_DYNAMIC_DEPTH_WRITE_ENABLE) {
         dynamic->depth_write_enable = pCreateInfo->pDepthStencilState->depthWriteEnable;
      }

      if (states & RADV_DYNAMIC_DEPTH_COMPARE_OP) {
         dynamic->depth_compare_op = pCreateInfo->pDepthStencilState->depthCompareOp;
      }

      if (states & RADV_DYNAMIC_DEPTH_BOUNDS_TEST_ENABLE) {
         dynamic->depth_bounds_test_enable = pCreateInfo->pDepthStencilState->depthBoundsTestEnable;
      }

      if (states & RADV_DYNAMIC_STENCIL_TEST_ENABLE) {
         dynamic->stencil_test_enable = pCreateInfo->pDepthStencilState->stencilTestEnable;
      }

      if (states & RADV_DYNAMIC_STENCIL_OP) {
         dynamic->stencil_op.front.compare_op = pCreateInfo->pDepthStencilState->front.compareOp;
         dynamic->stencil_op.front.fail_op = pCreateInfo->pDepthStencilState->front.failOp;
         dynamic->stencil_op.front.pass_op = pCreateInfo->pDepthStencilState->front.passOp;
         dynamic->stencil_op.front.depth_fail_op =
            pCreateInfo->pDepthStencilState->front.depthFailOp;

         dynamic->stencil_op.back.compare_op = pCreateInfo->pDepthStencilState->back.compareOp;
         dynamic->stencil_op.back.fail_op = pCreateInfo->pDepthStencilState->back.failOp;
         dynamic->stencil_op.back.pass_op = pCreateInfo->pDepthStencilState->back.passOp;
         dynamic->stencil_op.back.depth_fail_op = pCreateInfo->pDepthStencilState->back.depthFailOp;
      }
   }

   const VkPipelineDiscardRectangleStateCreateInfoEXT *discard_rectangle_info =
      vk_find_struct_const(pCreateInfo->pNext, PIPELINE_DISCARD_RECTANGLE_STATE_CREATE_INFO_EXT);
   if (needed_states & RADV_DYNAMIC_DISCARD_RECTANGLE) {
      dynamic->discard_rectangle.count = discard_rectangle_info->discardRectangleCount;
      if (states & RADV_DYNAMIC_DISCARD_RECTANGLE) {
         typed_memcpy(dynamic->discard_rectangle.rectangles,
                      discard_rectangle_info->pDiscardRectangles,
                      discard_rectangle_info->discardRectangleCount);
      }
   }

   if (needed_states & RADV_DYNAMIC_SAMPLE_LOCATIONS) {
      const VkPipelineSampleLocationsStateCreateInfoEXT *sample_location_info =
         vk_find_struct_const(pCreateInfo->pMultisampleState->pNext,
                              PIPELINE_SAMPLE_LOCATIONS_STATE_CREATE_INFO_EXT);
      /* If sampleLocationsEnable is VK_FALSE, the default sample
       * locations are used and the values specified in
       * sampleLocationsInfo are ignored.
       */
      if (sample_location_info->sampleLocationsEnable) {
         const VkSampleLocationsInfoEXT *pSampleLocationsInfo =
            &sample_location_info->sampleLocationsInfo;

         assert(pSampleLocationsInfo->sampleLocationsCount <= MAX_SAMPLE_LOCATIONS);

         dynamic->sample_location.per_pixel = pSampleLocationsInfo->sampleLocationsPerPixel;
         dynamic->sample_location.grid_size = pSampleLocationsInfo->sampleLocationGridSize;
         dynamic->sample_location.count = pSampleLocationsInfo->sampleLocationsCount;
         typed_memcpy(&dynamic->sample_location.locations[0],
                      pSampleLocationsInfo->pSampleLocations,
                      pSampleLocationsInfo->sampleLocationsCount);
      }
   }

   const VkPipelineRasterizationLineStateCreateInfoEXT *rast_line_info = vk_find_struct_const(
      pCreateInfo->pRasterizationState->pNext, PIPELINE_RASTERIZATION_LINE_STATE_CREATE_INFO_EXT);
   if (needed_states & RADV_DYNAMIC_LINE_STIPPLE) {
      dynamic->line_stipple.factor = rast_line_info->lineStippleFactor;
      dynamic->line_stipple.pattern = rast_line_info->lineStipplePattern;
   }

   if (!(states & RADV_DYNAMIC_VERTEX_INPUT_BINDING_STRIDE))
      pipeline->graphics.uses_dynamic_stride = true;

   const VkPipelineFragmentShadingRateStateCreateInfoKHR *shading_rate = vk_find_struct_const(
      pCreateInfo->pNext, PIPELINE_FRAGMENT_SHADING_RATE_STATE_CREATE_INFO_KHR);
   if (states & RADV_DYNAMIC_FRAGMENT_SHADING_RATE) {
      dynamic->fragment_shading_rate.size = shading_rate->fragmentSize;
      for (int i = 0; i < 2; i++)
         dynamic->fragment_shading_rate.combiner_ops[i] = shading_rate->combinerOps[i];
   }

   if (states & RADV_DYNAMIC_DEPTH_BIAS_ENABLE) {
      dynamic->depth_bias_enable = pCreateInfo->pRasterizationState->depthBiasEnable;
   }

   if (states & RADV_DYNAMIC_PRIMITIVE_RESTART_ENABLE) {
      dynamic->primitive_restart_enable =
         !!pCreateInfo->pInputAssemblyState->primitiveRestartEnable;
   }

   if (states & RADV_DYNAMIC_RASTERIZER_DISCARD_ENABLE) {
      dynamic->rasterizer_discard_enable =
         pCreateInfo->pRasterizationState->rasterizerDiscardEnable;
   }

   pipeline->dynamic_state.mask = states;
}

static void
radv_pipeline_init_raster_state(struct radv_pipeline *pipeline,
                                const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   const VkPipelineRasterizationStateCreateInfo *raster_info = pCreateInfo->pRasterizationState;
   const VkPipelineRasterizationProvokingVertexStateCreateInfoEXT *provoking_vtx_info =
      vk_find_struct_const(raster_info->pNext,
                           PIPELINE_RASTERIZATION_PROVOKING_VERTEX_STATE_CREATE_INFO_EXT);
   bool provoking_vtx_last = false;

   if (provoking_vtx_info &&
       provoking_vtx_info->provokingVertexMode == VK_PROVOKING_VERTEX_MODE_LAST_VERTEX_EXT) {
      provoking_vtx_last = true;
   }

   pipeline->graphics.pa_su_sc_mode_cntl =
      S_028814_FACE(raster_info->frontFace) |
      S_028814_CULL_FRONT(!!(raster_info->cullMode & VK_CULL_MODE_FRONT_BIT)) |
      S_028814_CULL_BACK(!!(raster_info->cullMode & VK_CULL_MODE_BACK_BIT)) |
      S_028814_POLY_MODE(raster_info->polygonMode != VK_POLYGON_MODE_FILL) |
      S_028814_POLYMODE_FRONT_PTYPE(si_translate_fill(raster_info->polygonMode)) |
      S_028814_POLYMODE_BACK_PTYPE(si_translate_fill(raster_info->polygonMode)) |
      S_028814_POLY_OFFSET_FRONT_ENABLE(raster_info->depthBiasEnable ? 1 : 0) |
      S_028814_POLY_OFFSET_BACK_ENABLE(raster_info->depthBiasEnable ? 1 : 0) |
      S_028814_POLY_OFFSET_PARA_ENABLE(raster_info->depthBiasEnable ? 1 : 0) |
      S_028814_PROVOKING_VTX_LAST(provoking_vtx_last);

   if (pipeline->device->physical_device->rad_info.chip_class >= GFX10) {
      /* It should also be set if PERPENDICULAR_ENDCAP_ENA is set. */
      pipeline->graphics.pa_su_sc_mode_cntl |=
         S_028814_KEEP_TOGETHER_ENABLE(raster_info->polygonMode != VK_POLYGON_MODE_FILL);
   }

   bool depth_clip_disable = raster_info->depthClampEnable;
   const VkPipelineRasterizationDepthClipStateCreateInfoEXT *depth_clip_state =
      vk_find_struct_const(raster_info->pNext,
                           PIPELINE_RASTERIZATION_DEPTH_CLIP_STATE_CREATE_INFO_EXT);
   if (depth_clip_state) {
      depth_clip_disable = !depth_clip_state->depthClipEnable;
   }

   pipeline->graphics.pa_cl_clip_cntl =
      S_028810_DX_CLIP_SPACE_DEF(1) | // vulkan uses DX conventions.
      S_028810_ZCLIP_NEAR_DISABLE(depth_clip_disable ? 1 : 0) |
      S_028810_ZCLIP_FAR_DISABLE(depth_clip_disable ? 1 : 0) |
      S_028810_DX_RASTERIZATION_KILL(raster_info->rasterizerDiscardEnable ? 1 : 0) |
      S_028810_DX_LINEAR_ATTR_CLIP_ENA(1);
}

static void
radv_pipeline_init_depth_stencil_state(struct radv_pipeline *pipeline,
                                       const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   const VkPipelineDepthStencilStateCreateInfo *ds_info =
      radv_pipeline_get_depth_stencil_state(pCreateInfo);
   RADV_FROM_HANDLE(radv_render_pass, pass, pCreateInfo->renderPass);
   struct radv_subpass *subpass = pass->subpasses + pCreateInfo->subpass;
   struct radv_render_pass_attachment *attachment = NULL;
   uint32_t db_depth_control = 0;

   if (subpass->depth_stencil_attachment)
      attachment = pass->attachments + subpass->depth_stencil_attachment->attachment;

   bool has_depth_attachment = attachment && vk_format_has_depth(attachment->format);
   bool has_stencil_attachment = attachment && vk_format_has_stencil(attachment->format);

   if (ds_info) {
      if (has_depth_attachment) {
         db_depth_control = S_028800_Z_ENABLE(ds_info->depthTestEnable ? 1 : 0) |
                            S_028800_Z_WRITE_ENABLE(ds_info->depthWriteEnable ? 1 : 0) |
                            S_028800_ZFUNC(ds_info->depthCompareOp) |
                            S_028800_DEPTH_BOUNDS_ENABLE(ds_info->depthBoundsTestEnable ? 1 : 0);
      }

      if (has_stencil_attachment && ds_info->stencilTestEnable) {
         db_depth_control |= S_028800_STENCIL_ENABLE(1) | S_028800_BACKFACE_ENABLE(1);
         db_depth_control |= S_028800_STENCILFUNC(ds_info->front.compareOp);
         db_depth_control |= S_028800_STENCILFUNC_BF(ds_info->back.compareOp);
      }
   }

   pipeline->graphics.db_depth_control = db_depth_control;
}

static void
gfx9_get_gs_info(const struct radv_pipeline_key *key, const struct radv_pipeline *pipeline,
                 nir_shader **nir, struct radv_shader_info *infos, struct gfx9_gs_info *out)
{
   struct radv_shader_info *gs_info = &infos[MESA_SHADER_GEOMETRY];
   struct radv_es_output_info *es_info;
   if (pipeline->device->physical_device->rad_info.chip_class >= GFX9)
      es_info = nir[MESA_SHADER_TESS_CTRL] ? &gs_info->tes.es_info : &gs_info->vs.es_info;
   else
      es_info = nir[MESA_SHADER_TESS_CTRL] ? &infos[MESA_SHADER_TESS_EVAL].tes.es_info
                                           : &infos[MESA_SHADER_VERTEX].vs.es_info;

   unsigned gs_num_invocations = MAX2(gs_info->gs.invocations, 1);
   bool uses_adjacency;
   switch (key->topology) {
   case VK_PRIMITIVE_TOPOLOGY_LINE_LIST_WITH_ADJACENCY:
   case VK_PRIMITIVE_TOPOLOGY_LINE_STRIP_WITH_ADJACENCY:
   case VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST_WITH_ADJACENCY:
   case VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP_WITH_ADJACENCY:
      uses_adjacency = true;
      break;
   default:
      uses_adjacency = false;
      break;
   }

   /* All these are in dwords: */
   /* We can't allow using the whole LDS, because GS waves compete with
    * other shader stages for LDS space. */
   const unsigned max_lds_size = 8 * 1024;
   const unsigned esgs_itemsize = es_info->esgs_itemsize / 4;
   unsigned esgs_lds_size;

   /* All these are per subgroup: */
   const unsigned max_out_prims = 32 * 1024;
   const unsigned max_es_verts = 255;
   const unsigned ideal_gs_prims = 64;
   unsigned max_gs_prims, gs_prims;
   unsigned min_es_verts, es_verts, worst_case_es_verts;

   if (uses_adjacency || gs_num_invocations > 1)
      max_gs_prims = 127 / gs_num_invocations;
   else
      max_gs_prims = 255;

   /* MAX_PRIMS_PER_SUBGROUP = gs_prims * max_vert_out * gs_invocations.
    * Make sure we don't go over the maximum value.
    */
   if (gs_info->gs.vertices_out > 0) {
      max_gs_prims =
         MIN2(max_gs_prims, max_out_prims / (gs_info->gs.vertices_out * gs_num_invocations));
   }
   assert(max_gs_prims > 0);

   /* If the primitive has adjacency, halve the number of vertices
    * that will be reused in multiple primitives.
    */
   min_es_verts = gs_info->gs.vertices_in / (uses_adjacency ? 2 : 1);

   gs_prims = MIN2(ideal_gs_prims, max_gs_prims);
   worst_case_es_verts = MIN2(min_es_verts * gs_prims, max_es_verts);

   /* Compute ESGS LDS size based on the worst case number of ES vertices
    * needed to create the target number of GS prims per subgroup.
    */
   esgs_lds_size = esgs_itemsize * worst_case_es_verts;

   /* If total LDS usage is too big, refactor partitions based on ratio
    * of ESGS item sizes.
    */
   if (esgs_lds_size > max_lds_size) {
      /* Our target GS Prims Per Subgroup was too large. Calculate
       * the maximum number of GS Prims Per Subgroup that will fit
       * into LDS, capped by the maximum that the hardware can support.
       */
      gs_prims = MIN2((max_lds_size / (esgs_itemsize * min_es_verts)), max_gs_prims);
      assert(gs_prims > 0);
      worst_case_es_verts = MIN2(min_es_verts * gs_prims, max_es_verts);

      esgs_lds_size = esgs_itemsize * worst_case_es_verts;
      assert(esgs_lds_size <= max_lds_size);
   }

   /* Now calculate remaining ESGS information. */
   if (esgs_lds_size)
      es_verts = MIN2(esgs_lds_size / esgs_itemsize, max_es_verts);
   else
      es_verts = max_es_verts;

   /* Vertices for adjacency primitives are not always reused, so restore
    * it for ES_VERTS_PER_SUBGRP.
    */
   min_es_verts = gs_info->gs.vertices_in;

   /* For normal primitives, the VGT only checks if they are past the ES
    * verts per subgroup after allocating a full GS primitive and if they
    * are, kick off a new subgroup.  But if those additional ES verts are
    * unique (e.g. not reused) we need to make sure there is enough LDS
    * space to account for those ES verts beyond ES_VERTS_PER_SUBGRP.
    */
   es_verts -= min_es_verts - 1;

   uint32_t es_verts_per_subgroup = es_verts;
   uint32_t gs_prims_per_subgroup = gs_prims;
   uint32_t gs_inst_prims_in_subgroup = gs_prims * gs_num_invocations;
   uint32_t max_prims_per_subgroup = gs_inst_prims_in_subgroup * gs_info->gs.vertices_out;
   out->lds_size = align(esgs_lds_size, 128) / 128;
   out->vgt_gs_onchip_cntl = S_028A44_ES_VERTS_PER_SUBGRP(es_verts_per_subgroup) |
                             S_028A44_GS_PRIMS_PER_SUBGRP(gs_prims_per_subgroup) |
                             S_028A44_GS_INST_PRIMS_IN_SUBGRP(gs_inst_prims_in_subgroup);
   out->vgt_gs_max_prims_per_subgroup = S_028A94_MAX_PRIMS_PER_SUBGROUP(max_prims_per_subgroup);
   out->vgt_esgs_ring_itemsize = esgs_itemsize;
   assert(max_prims_per_subgroup <= max_out_prims);
}

static void
clamp_gsprims_to_esverts(unsigned *max_gsprims, unsigned max_esverts, unsigned min_verts_per_prim,
                         bool use_adjacency)
{
   unsigned max_reuse = max_esverts - min_verts_per_prim;
   if (use_adjacency)
      max_reuse /= 2;
   *max_gsprims = MIN2(*max_gsprims, 1 + max_reuse);
}

static unsigned
radv_get_num_input_vertices(nir_shader **nir)
{
   if (nir[MESA_SHADER_GEOMETRY]) {
      nir_shader *gs = nir[MESA_SHADER_GEOMETRY];

      return gs->info.gs.vertices_in;
   }

   if (nir[MESA_SHADER_TESS_CTRL]) {
      nir_shader *tes = nir[MESA_SHADER_TESS_EVAL];

      if (tes->info.tess.point_mode)
         return 1;
      if (tes->info.tess.primitive_mode == GL_ISOLINES)
         return 2;
      return 3;
   }

   return 3;
}

static void
gfx10_get_ngg_info(const struct radv_pipeline_key *key, struct radv_pipeline *pipeline,
                   nir_shader **nir, struct radv_shader_info *infos, struct gfx10_ngg_info *ngg)
{
   struct radv_shader_info *gs_info = &infos[MESA_SHADER_GEOMETRY];
   struct radv_es_output_info *es_info =
      nir[MESA_SHADER_TESS_CTRL] ? &gs_info->tes.es_info : &gs_info->vs.es_info;
   unsigned gs_type = nir[MESA_SHADER_GEOMETRY] ? MESA_SHADER_GEOMETRY : MESA_SHADER_VERTEX;
   unsigned max_verts_per_prim = radv_get_num_input_vertices(nir);
   unsigned min_verts_per_prim = gs_type == MESA_SHADER_GEOMETRY ? max_verts_per_prim : 1;
   unsigned gs_num_invocations = nir[MESA_SHADER_GEOMETRY] ? MAX2(gs_info->gs.invocations, 1) : 1;
   bool uses_adjacency;
   switch (key->topology) {
   case VK_PRIMITIVE_TOPOLOGY_LINE_LIST_WITH_ADJACENCY:
   case VK_PRIMITIVE_TOPOLOGY_LINE_STRIP_WITH_ADJACENCY:
   case VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST_WITH_ADJACENCY:
   case VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP_WITH_ADJACENCY:
      uses_adjacency = true;
      break;
   default:
      uses_adjacency = false;
      break;
   }

   /* All these are in dwords: */
   /* We can't allow using the whole LDS, because GS waves compete with
    * other shader stages for LDS space.
    *
    * TODO: We should really take the shader's internal LDS use into
    *       account. The linker will fail if the size is greater than
    *       8K dwords.
    */
   const unsigned max_lds_size = 8 * 1024 - 768;
   const unsigned target_lds_size = max_lds_size;
   unsigned esvert_lds_size = 0;
   unsigned gsprim_lds_size = 0;

   /* All these are per subgroup: */
   const unsigned min_esverts =
      pipeline->device->physical_device->rad_info.chip_class >= GFX10_3 ? 29 : 24;
   bool max_vert_out_per_gs_instance = false;
   unsigned max_esverts_base = 256;
   unsigned max_gsprims_base = 128; /* default prim group size clamp */

   /* Hardware has the following non-natural restrictions on the value
    * of GE_CNTL.VERT_GRP_SIZE based on based on the primitive type of
    * the draw:
    *  - at most 252 for any line input primitive type
    *  - at most 251 for any quad input primitive type
    *  - at most 251 for triangle strips with adjacency (this happens to
    *    be the natural limit for triangle *lists* with adjacency)
    */
   max_esverts_base = MIN2(max_esverts_base, 251 + max_verts_per_prim - 1);

   if (gs_type == MESA_SHADER_GEOMETRY) {
      unsigned max_out_verts_per_gsprim = gs_info->gs.vertices_out * gs_num_invocations;

      if (max_out_verts_per_gsprim <= 256) {
         if (max_out_verts_per_gsprim) {
            max_gsprims_base = MIN2(max_gsprims_base, 256 / max_out_verts_per_gsprim);
         }
      } else {
         /* Use special multi-cycling mode in which each GS
          * instance gets its own subgroup. Does not work with
          * tessellation. */
         max_vert_out_per_gs_instance = true;
         max_gsprims_base = 1;
         max_out_verts_per_gsprim = gs_info->gs.vertices_out;
      }

      esvert_lds_size = es_info->esgs_itemsize / 4;
      gsprim_lds_size = (gs_info->gs.gsvs_vertex_size / 4 + 1) * max_out_verts_per_gsprim;
   } else {
      /* VS and TES. */
      /* LDS size for passing data from GS to ES. */
      struct radv_streamout_info *so_info = nir[MESA_SHADER_TESS_CTRL]
                                               ? &infos[MESA_SHADER_TESS_EVAL].so
                                               : &infos[MESA_SHADER_VERTEX].so;

      if (so_info->num_outputs)
         esvert_lds_size = 4 * so_info->num_outputs + 1;

      /* GS stores Primitive IDs (one DWORD) into LDS at the address
       * corresponding to the ES thread of the provoking vertex. All
       * ES threads load and export PrimitiveID for their thread.
       */
      if (!nir[MESA_SHADER_TESS_CTRL] && infos[MESA_SHADER_VERTEX].vs.outinfo.export_prim_id)
         esvert_lds_size = MAX2(esvert_lds_size, 1);
   }

   unsigned max_gsprims = max_gsprims_base;
   unsigned max_esverts = max_esverts_base;

   if (esvert_lds_size)
      max_esverts = MIN2(max_esverts, target_lds_size / esvert_lds_size);
   if (gsprim_lds_size)
      max_gsprims = MIN2(max_gsprims, target_lds_size / gsprim_lds_size);

   max_esverts = MIN2(max_esverts, max_gsprims * max_verts_per_prim);
   clamp_gsprims_to_esverts(&max_gsprims, max_esverts, min_verts_per_prim, uses_adjacency);
   assert(max_esverts >= max_verts_per_prim && max_gsprims >= 1);

   if (esvert_lds_size || gsprim_lds_size) {
      /* Now that we have a rough proportionality between esverts
       * and gsprims based on the primitive type, scale both of them
       * down simultaneously based on required LDS space.
       *
       * We could be smarter about this if we knew how much vertex
       * reuse to expect.
       */
      unsigned lds_total = max_esverts * esvert_lds_size + max_gsprims * gsprim_lds_size;
      if (lds_total > target_lds_size) {
         max_esverts = max_esverts * target_lds_size / lds_total;
         max_gsprims = max_gsprims * target_lds_size / lds_total;

         max_esverts = MIN2(max_esverts, max_gsprims * max_verts_per_prim);
         clamp_gsprims_to_esverts(&max_gsprims, max_esverts, min_verts_per_prim, uses_adjacency);
         assert(max_esverts >= max_verts_per_prim && max_gsprims >= 1);
      }
   }

   /* Round up towards full wave sizes for better ALU utilization. */
   if (!max_vert_out_per_gs_instance) {
      unsigned orig_max_esverts;
      unsigned orig_max_gsprims;
      unsigned wavesize;

      if (gs_type == MESA_SHADER_GEOMETRY) {
         wavesize = gs_info->wave_size;
      } else {
         wavesize = nir[MESA_SHADER_TESS_CTRL] ? infos[MESA_SHADER_TESS_EVAL].wave_size
                                               : infos[MESA_SHADER_VERTEX].wave_size;
      }

      do {
         orig_max_esverts = max_esverts;
         orig_max_gsprims = max_gsprims;

         max_esverts = align(max_esverts, wavesize);
         max_esverts = MIN2(max_esverts, max_esverts_base);
         if (esvert_lds_size)
            max_esverts =
               MIN2(max_esverts, (max_lds_size - max_gsprims * gsprim_lds_size) / esvert_lds_size);
         max_esverts = MIN2(max_esverts, max_gsprims * max_verts_per_prim);

         /* Hardware restriction: minimum value of max_esverts */
         if (pipeline->device->physical_device->rad_info.chip_class == GFX10)
            max_esverts = MAX2(max_esverts, min_esverts - 1 + max_verts_per_prim);
         else
            max_esverts = MAX2(max_esverts, min_esverts);

         max_gsprims = align(max_gsprims, wavesize);
         max_gsprims = MIN2(max_gsprims, max_gsprims_base);
         if (gsprim_lds_size) {
            /* Don't count unusable vertices to the LDS
             * size. Those are vertices above the maximum
             * number of vertices that can occur in the
             * workgroup, which is e.g. max_gsprims * 3
             * for triangles.
             */
            unsigned usable_esverts = MIN2(max_esverts, max_gsprims * max_verts_per_prim);
            max_gsprims = MIN2(max_gsprims,
                               (max_lds_size - usable_esverts * esvert_lds_size) / gsprim_lds_size);
         }
         clamp_gsprims_to_esverts(&max_gsprims, max_esverts, min_verts_per_prim, uses_adjacency);
         assert(max_esverts >= max_verts_per_prim && max_gsprims >= 1);
      } while (orig_max_esverts != max_esverts || orig_max_gsprims != max_gsprims);

      /* Verify the restriction. */
      if (pipeline->device->physical_device->rad_info.chip_class == GFX10)
         assert(max_esverts >= min_esverts - 1 + max_verts_per_prim);
      else
         assert(max_esverts >= min_esverts);
   } else {
      /* Hardware restriction: minimum value of max_esverts */
      if (pipeline->device->physical_device->rad_info.chip_class == GFX10)
         max_esverts = MAX2(max_esverts, min_esverts - 1 + max_verts_per_prim);
      else
         max_esverts = MAX2(max_esverts, min_esverts);
   }

   unsigned max_out_vertices = max_vert_out_per_gs_instance ? gs_info->gs.vertices_out
                               : gs_type == MESA_SHADER_GEOMETRY
                                  ? max_gsprims * gs_num_invocations * gs_info->gs.vertices_out
                                  : max_esverts;
   assert(max_out_vertices <= 256);

   unsigned prim_amp_factor = 1;
   if (gs_type == MESA_SHADER_GEOMETRY) {
      /* Number of output primitives per GS input primitive after
       * GS instancing. */
      prim_amp_factor = gs_info->gs.vertices_out;
   }

   /* On Gfx10, the GE only checks against the maximum number of ES verts
    * after allocating a full GS primitive. So we need to ensure that
    * whenever this check passes, there is enough space for a full
    * primitive without vertex reuse.
    */
   if (pipeline->device->physical_device->rad_info.chip_class == GFX10)
      ngg->hw_max_esverts = max_esverts - max_verts_per_prim + 1;
   else
      ngg->hw_max_esverts = max_esverts;

   ngg->max_gsprims = max_gsprims;
   ngg->max_out_verts = max_out_vertices;
   ngg->prim_amp_factor = prim_amp_factor;
   ngg->max_vert_out_per_gs_instance = max_vert_out_per_gs_instance;
   ngg->ngg_emit_size = max_gsprims * gsprim_lds_size;

   /* Don't count unusable vertices. */
   ngg->esgs_ring_size = MIN2(max_esverts, max_gsprims * max_verts_per_prim) * esvert_lds_size * 4;

   if (gs_type == MESA_SHADER_GEOMETRY) {
      ngg->vgt_esgs_ring_itemsize = es_info->esgs_itemsize / 4;
   } else {
      ngg->vgt_esgs_ring_itemsize = 1;
   }

   assert(ngg->hw_max_esverts >= min_esverts); /* HW limitation */
}

static void
radv_pipeline_init_gs_ring_state(struct radv_pipeline *pipeline, const struct gfx9_gs_info *gs)
{
   struct radv_device *device = pipeline->device;
   unsigned num_se = device->physical_device->rad_info.max_se;
   unsigned wave_size = 64;
   unsigned max_gs_waves = 32 * num_se; /* max 32 per SE on GCN */
   /* On GFX6-GFX7, the value comes from VGT_GS_VERTEX_REUSE = 16.
    * On GFX8+, the value comes from VGT_VERTEX_REUSE_BLOCK_CNTL = 30 (+2).
    */
   unsigned gs_vertex_reuse =
      (device->physical_device->rad_info.chip_class >= GFX8 ? 32 : 16) * num_se;
   unsigned alignment = 256 * num_se;
   /* The maximum size is 63.999 MB per SE. */
   unsigned max_size = ((unsigned)(63.999 * 1024 * 1024) & ~255) * num_se;
   struct radv_shader_info *gs_info = &pipeline->shaders[MESA_SHADER_GEOMETRY]->info;

   /* Calculate the minimum size. */
   unsigned min_esgs_ring_size =
      align(gs->vgt_esgs_ring_itemsize * 4 * gs_vertex_reuse * wave_size, alignment);
   /* These are recommended sizes, not minimum sizes. */
   unsigned esgs_ring_size =
      max_gs_waves * 2 * wave_size * gs->vgt_esgs_ring_itemsize * 4 * gs_info->gs.vertices_in;
   unsigned gsvs_ring_size = max_gs_waves * 2 * wave_size * gs_info->gs.max_gsvs_emit_size;

   min_esgs_ring_size = align(min_esgs_ring_size, alignment);
   esgs_ring_size = align(esgs_ring_size, alignment);
   gsvs_ring_size = align(gsvs_ring_size, alignment);

   if (pipeline->device->physical_device->rad_info.chip_class <= GFX8)
      pipeline->graphics.esgs_ring_size = CLAMP(esgs_ring_size, min_esgs_ring_size, max_size);

   pipeline->graphics.gsvs_ring_size = MIN2(gsvs_ring_size, max_size);
}

struct radv_shader_variant *
radv_get_shader(const struct radv_pipeline *pipeline, gl_shader_stage stage)
{
   if (stage == MESA_SHADER_VERTEX) {
      if (pipeline->shaders[MESA_SHADER_VERTEX])
         return pipeline->shaders[MESA_SHADER_VERTEX];
      if (pipeline->shaders[MESA_SHADER_TESS_CTRL])
         return pipeline->shaders[MESA_SHADER_TESS_CTRL];
      if (pipeline->shaders[MESA_SHADER_GEOMETRY])
         return pipeline->shaders[MESA_SHADER_GEOMETRY];
   } else if (stage == MESA_SHADER_TESS_EVAL) {
      if (!radv_pipeline_has_tess(pipeline))
         return NULL;
      if (pipeline->shaders[MESA_SHADER_TESS_EVAL])
         return pipeline->shaders[MESA_SHADER_TESS_EVAL];
      if (pipeline->shaders[MESA_SHADER_GEOMETRY])
         return pipeline->shaders[MESA_SHADER_GEOMETRY];
   }
   return pipeline->shaders[stage];
}

static const struct radv_vs_output_info *
get_vs_output_info(const struct radv_pipeline *pipeline)
{
   if (radv_pipeline_has_gs(pipeline))
      if (radv_pipeline_has_ngg(pipeline))
         return &pipeline->shaders[MESA_SHADER_GEOMETRY]->info.vs.outinfo;
      else
         return &pipeline->gs_copy_shader->info.vs.outinfo;
   else if (radv_pipeline_has_tess(pipeline))
      return &pipeline->shaders[MESA_SHADER_TESS_EVAL]->info.tes.outinfo;
   else
      return &pipeline->shaders[MESA_SHADER_VERTEX]->info.vs.outinfo;
}

static void
radv_link_shaders(struct radv_pipeline *pipeline, nir_shader **shaders,
                  bool optimize_conservatively)
{
   nir_shader *ordered_shaders[MESA_SHADER_STAGES];
   int shader_count = 0;

   if (shaders[MESA_SHADER_FRAGMENT]) {
      ordered_shaders[shader_count++] = shaders[MESA_SHADER_FRAGMENT];
   }
   if (shaders[MESA_SHADER_GEOMETRY]) {
      ordered_shaders[shader_count++] = shaders[MESA_SHADER_GEOMETRY];
   }
   if (shaders[MESA_SHADER_TESS_EVAL]) {
      ordered_shaders[shader_count++] = shaders[MESA_SHADER_TESS_EVAL];
   }
   if (shaders[MESA_SHADER_TESS_CTRL]) {
      ordered_shaders[shader_count++] = shaders[MESA_SHADER_TESS_CTRL];
   }
   if (shaders[MESA_SHADER_VERTEX]) {
      ordered_shaders[shader_count++] = shaders[MESA_SHADER_VERTEX];
   }
   if (shaders[MESA_SHADER_COMPUTE]) {
      ordered_shaders[shader_count++] = shaders[MESA_SHADER_COMPUTE];
   }

   bool has_geom_tess = shaders[MESA_SHADER_GEOMETRY] || shaders[MESA_SHADER_TESS_CTRL];
   bool merged_gs = shaders[MESA_SHADER_GEOMETRY] &&
                    pipeline->device->physical_device->rad_info.chip_class >= GFX9;

   if (!optimize_conservatively && shader_count > 1) {
      unsigned first = ordered_shaders[shader_count - 1]->info.stage;
      unsigned last = ordered_shaders[0]->info.stage;

      if (ordered_shaders[0]->info.stage == MESA_SHADER_FRAGMENT &&
          ordered_shaders[1]->info.has_transform_feedback_varyings)
         nir_link_xfb_varyings(ordered_shaders[1], ordered_shaders[0]);

      for (int i = 1; i < shader_count; ++i) {
         nir_lower_io_arrays_to_elements(ordered_shaders[i], ordered_shaders[i - 1]);
      }

      for (int i = 0; i < shader_count; ++i) {
         nir_variable_mode mask = 0;

         if (ordered_shaders[i]->info.stage != first)
            mask = mask | nir_var_shader_in;

         if (ordered_shaders[i]->info.stage != last)
            mask = mask | nir_var_shader_out;

         if (nir_lower_io_to_scalar_early(ordered_shaders[i], mask)) {
            /* Optimize the new vector code and then remove dead vars */
            nir_copy_prop(ordered_shaders[i]);
            nir_opt_shrink_vectors(ordered_shaders[i],
                                   !pipeline->device->instance->disable_shrink_image_store);

            if (ordered_shaders[i]->info.stage != last) {
               /* Optimize swizzled movs of load_const for
                * nir_link_opt_varyings's constant propagation
                */
               nir_opt_constant_folding(ordered_shaders[i]);
               /* For nir_link_opt_varyings's duplicate input opt */
               nir_opt_cse(ordered_shaders[i]);
            }

            /* Run copy-propagation to help remove dead
             * output variables (some shaders have useless
             * copies to/from an output), so compaction
             * later will be more effective.
             *
             * This will have been done earlier but it might
             * not have worked because the outputs were vector.
             */
            if (ordered_shaders[i]->info.stage == MESA_SHADER_TESS_CTRL)
               nir_opt_copy_prop_vars(ordered_shaders[i]);

            nir_opt_dce(ordered_shaders[i]);
            nir_remove_dead_variables(
               ordered_shaders[i], nir_var_function_temp | nir_var_shader_in | nir_var_shader_out,
               NULL);
         }
      }
   }

   for (int i = 1; !optimize_conservatively && (i < shader_count); ++i) {
      if (nir_link_opt_varyings(ordered_shaders[i], ordered_shaders[i - 1])) {
         nir_opt_constant_folding(ordered_shaders[i - 1]);
         nir_opt_algebraic(ordered_shaders[i - 1]);
         nir_opt_dce(ordered_shaders[i - 1]);
      }

      nir_remove_dead_variables(ordered_shaders[i], nir_var_shader_out, NULL);
      nir_remove_dead_variables(ordered_shaders[i - 1], nir_var_shader_in, NULL);

      bool progress = nir_remove_unused_varyings(ordered_shaders[i], ordered_shaders[i - 1]);

      nir_compact_varyings(ordered_shaders[i], ordered_shaders[i - 1], true);

      if (ordered_shaders[i]->info.stage == MESA_SHADER_TESS_CTRL ||
          (ordered_shaders[i]->info.stage == MESA_SHADER_VERTEX && has_geom_tess) ||
          (ordered_shaders[i]->info.stage == MESA_SHADER_TESS_EVAL && merged_gs)) {
         nir_lower_io_to_vector(ordered_shaders[i], nir_var_shader_out);
         if (ordered_shaders[i]->info.stage == MESA_SHADER_TESS_CTRL)
            nir_vectorize_tess_levels(ordered_shaders[i]);
         nir_opt_combine_stores(ordered_shaders[i], nir_var_shader_out);
      }
      if (ordered_shaders[i - 1]->info.stage == MESA_SHADER_GEOMETRY ||
          ordered_shaders[i - 1]->info.stage == MESA_SHADER_TESS_CTRL ||
          ordered_shaders[i - 1]->info.stage == MESA_SHADER_TESS_EVAL) {
         nir_lower_io_to_vector(ordered_shaders[i - 1], nir_var_shader_in);
      }

      if (progress) {
         if (nir_lower_global_vars_to_local(ordered_shaders[i])) {
            ac_nir_lower_indirect_derefs(ordered_shaders[i],
                                         pipeline->device->physical_device->rad_info.chip_class);
            /* remove dead writes, which can remove input loads */
            nir_lower_vars_to_ssa(ordered_shaders[i]);
            nir_opt_dce(ordered_shaders[i]);
         }

         if (nir_lower_global_vars_to_local(ordered_shaders[i - 1])) {
            ac_nir_lower_indirect_derefs(ordered_shaders[i - 1],
                                         pipeline->device->physical_device->rad_info.chip_class);
         }
      }
   }
}

static void
radv_set_driver_locations(struct radv_pipeline *pipeline, nir_shader **shaders,
                          struct radv_shader_info infos[MESA_SHADER_STAGES])
{
   if (shaders[MESA_SHADER_FRAGMENT]) {
      nir_foreach_shader_out_variable(var, shaders[MESA_SHADER_FRAGMENT])
      {
         var->data.driver_location = var->data.location + var->data.index;
      }
   }

   if (!shaders[MESA_SHADER_VERTEX])
      return;

   bool has_tess = shaders[MESA_SHADER_TESS_CTRL];
   bool has_gs = shaders[MESA_SHADER_GEOMETRY];

   /* Merged stage for VS and TES */
   unsigned vs_info_idx = MESA_SHADER_VERTEX;
   unsigned tes_info_idx = MESA_SHADER_TESS_EVAL;

   /* Which stage is the last in the vertex, tess, geometry pipeline */
   unsigned last_vtg_stage = MESA_SHADER_VERTEX;

   if (pipeline->device->physical_device->rad_info.chip_class >= GFX9) {
      /* These are merged into the next stage */
      vs_info_idx = has_tess ? MESA_SHADER_TESS_CTRL : MESA_SHADER_GEOMETRY;
      tes_info_idx = has_gs ? MESA_SHADER_GEOMETRY : MESA_SHADER_TESS_EVAL;
   }

   nir_foreach_shader_in_variable (var, shaders[MESA_SHADER_VERTEX]) {
      var->data.driver_location = var->data.location;
   }

   if (has_tess) {
      nir_linked_io_var_info vs2tcs = nir_assign_linked_io_var_locations(
         shaders[MESA_SHADER_VERTEX], shaders[MESA_SHADER_TESS_CTRL]);
      nir_linked_io_var_info tcs2tes = nir_assign_linked_io_var_locations(
         shaders[MESA_SHADER_TESS_CTRL], shaders[MESA_SHADER_TESS_EVAL]);

      infos[MESA_SHADER_VERTEX].vs.num_linked_outputs = vs2tcs.num_linked_io_vars;
      infos[MESA_SHADER_TESS_CTRL].tcs.num_linked_inputs = vs2tcs.num_linked_io_vars;
      infos[MESA_SHADER_TESS_CTRL].tcs.num_linked_outputs = tcs2tes.num_linked_io_vars;
      infos[MESA_SHADER_TESS_CTRL].tcs.num_linked_patch_outputs = tcs2tes.num_linked_patch_io_vars;
      infos[MESA_SHADER_TESS_EVAL].tes.num_linked_inputs = tcs2tes.num_linked_io_vars;
      infos[MESA_SHADER_TESS_EVAL].tes.num_linked_patch_inputs = tcs2tes.num_linked_patch_io_vars;

      /* Copy data to merged stage */
      infos[vs_info_idx].vs.num_linked_outputs = vs2tcs.num_linked_io_vars;
      infos[tes_info_idx].tes.num_linked_inputs = tcs2tes.num_linked_io_vars;
      infos[tes_info_idx].tes.num_linked_patch_inputs = tcs2tes.num_linked_patch_io_vars;

      if (has_gs) {
         nir_linked_io_var_info tes2gs = nir_assign_linked_io_var_locations(
            shaders[MESA_SHADER_TESS_EVAL], shaders[MESA_SHADER_GEOMETRY]);

         infos[MESA_SHADER_TESS_EVAL].tes.num_linked_outputs = tes2gs.num_linked_io_vars;
         infos[MESA_SHADER_GEOMETRY].gs.num_linked_inputs = tes2gs.num_linked_io_vars;

         /* Copy data to merged stage */
         infos[tes_info_idx].tes.num_linked_outputs = tes2gs.num_linked_io_vars;

         last_vtg_stage = MESA_SHADER_GEOMETRY;

      } else {
         last_vtg_stage = MESA_SHADER_TESS_EVAL;
      }
   } else if (has_gs) {
      nir_linked_io_var_info vs2gs = nir_assign_linked_io_var_locations(
         shaders[MESA_SHADER_VERTEX], shaders[MESA_SHADER_GEOMETRY]);

      infos[MESA_SHADER_VERTEX].vs.num_linked_outputs = vs2gs.num_linked_io_vars;
      infos[MESA_SHADER_GEOMETRY].gs.num_linked_inputs = vs2gs.num_linked_io_vars;
      last_vtg_stage = MESA_SHADER_GEOMETRY;

      /* Copy data to merged stage */
      infos[vs_info_idx].vs.num_linked_outputs = vs2gs.num_linked_io_vars;
   }

   nir_foreach_shader_out_variable(var, shaders[last_vtg_stage])
   {
      var->data.driver_location = var->data.location;
   }
}

static uint32_t
radv_get_attrib_stride(const VkPipelineVertexInputStateCreateInfo *input_state,
                       uint32_t attrib_binding)
{
   for (uint32_t i = 0; i < input_state->vertexBindingDescriptionCount; i++) {
      const VkVertexInputBindingDescription *input_binding =
         &input_state->pVertexBindingDescriptions[i];

      if (input_binding->binding == attrib_binding)
         return input_binding->stride;
   }

   return 0;
}

static struct radv_pipeline_key
radv_generate_graphics_pipeline_key(const struct radv_pipeline *pipeline,
                                    const VkGraphicsPipelineCreateInfo *pCreateInfo,
                                    const struct radv_blend_state *blend)
{
   RADV_FROM_HANDLE(radv_render_pass, pass, pCreateInfo->renderPass);
   struct radv_subpass *subpass = pass->subpasses + pCreateInfo->subpass;
   const VkPipelineVertexInputStateCreateInfo *input_state = pCreateInfo->pVertexInputState;
   const VkPipelineVertexInputDivisorStateCreateInfoEXT *divisor_state =
      vk_find_struct_const(input_state->pNext, PIPELINE_VERTEX_INPUT_DIVISOR_STATE_CREATE_INFO_EXT);
   bool uses_dynamic_stride = false;

   struct radv_pipeline_key key;
   memset(&key, 0, sizeof(key));

   if (pCreateInfo->flags & VK_PIPELINE_CREATE_DISABLE_OPTIMIZATION_BIT)
      key.optimisations_disabled = 1;

   key.has_multiview_view_index = !!subpass->view_mask;

   uint32_t binding_input_rate = 0;
   uint32_t instance_rate_divisors[MAX_VERTEX_ATTRIBS];
   for (unsigned i = 0; i < input_state->vertexBindingDescriptionCount; ++i) {
      if (input_state->pVertexBindingDescriptions[i].inputRate) {
         unsigned binding = input_state->pVertexBindingDescriptions[i].binding;
         binding_input_rate |= 1u << binding;
         instance_rate_divisors[binding] = 1;
      }
   }
   if (divisor_state) {
      for (unsigned i = 0; i < divisor_state->vertexBindingDivisorCount; ++i) {
         instance_rate_divisors[divisor_state->pVertexBindingDivisors[i].binding] =
            divisor_state->pVertexBindingDivisors[i].divisor;
      }
   }

   if (pCreateInfo->pDynamicState) {
      uint32_t count = pCreateInfo->pDynamicState->dynamicStateCount;
      for (uint32_t i = 0; i < count; i++) {
         if (pCreateInfo->pDynamicState->pDynamicStates[i] ==
             VK_DYNAMIC_STATE_VERTEX_INPUT_BINDING_STRIDE_EXT) {
            uses_dynamic_stride = true;
            break;
         }
      }
   }

   for (unsigned i = 0; i < input_state->vertexAttributeDescriptionCount; ++i) {
      const VkVertexInputAttributeDescription *desc = &input_state->pVertexAttributeDescriptions[i];
      const struct util_format_description *format_desc;
      unsigned location = desc->location;
      unsigned binding = desc->binding;
      unsigned num_format, data_format;
      int first_non_void;

      if (binding_input_rate & (1u << binding)) {
         key.instance_rate_inputs |= 1u << location;
         key.instance_rate_divisors[location] = instance_rate_divisors[binding];
      }

      format_desc = vk_format_description(desc->format);
      first_non_void = vk_format_get_first_non_void_channel(desc->format);

      num_format = radv_translate_buffer_numformat(format_desc, first_non_void);
      data_format = radv_translate_buffer_dataformat(format_desc, first_non_void);

      key.vertex_attribute_formats[location] = data_format | (num_format << 4);
      key.vertex_attribute_bindings[location] = desc->binding;
      key.vertex_attribute_offsets[location] = desc->offset;

      if (!uses_dynamic_stride) {
         /* From the Vulkan spec 1.2.157:
          *
          * "If the bound pipeline state object was created
          *  with the
          *  VK_DYNAMIC_STATE_VERTEX_INPUT_BINDING_STRIDE_EXT
          *  dynamic state enabled then pStrides[i] specifies
          *  the distance in bytes between two consecutive
          *  elements within the corresponding buffer. In this
          *  case the VkVertexInputBindingDescription::stride
          *  state from the pipeline state object is ignored."
          *
          * Make sure the vertex attribute stride is zero to
          * avoid computing a wrong offset if it's initialized
          * to something else than zero.
          */
         key.vertex_attribute_strides[location] =
            radv_get_attrib_stride(input_state, desc->binding);
      }

      enum ac_fetch_format adjust = AC_FETCH_FORMAT_NONE;
      if (pipeline->device->physical_device->rad_info.chip_class <= GFX8 &&
          pipeline->device->physical_device->rad_info.family != CHIP_STONEY) {
         VkFormat format = input_state->pVertexAttributeDescriptions[i].format;
         switch (format) {
         case VK_FORMAT_A2R10G10B10_SNORM_PACK32:
         case VK_FORMAT_A2B10G10R10_SNORM_PACK32:
            adjust = AC_FETCH_FORMAT_SNORM;
            break;
         case VK_FORMAT_A2R10G10B10_SSCALED_PACK32:
         case VK_FORMAT_A2B10G10R10_SSCALED_PACK32:
            adjust = AC_FETCH_FORMAT_SSCALED;
            break;
         case VK_FORMAT_A2R10G10B10_SINT_PACK32:
         case VK_FORMAT_A2B10G10R10_SINT_PACK32:
            adjust = AC_FETCH_FORMAT_SINT;
            break;
         default:
            break;
         }
      }
      key.vertex_alpha_adjust[location] = adjust;

      switch (desc->format) {
      case VK_FORMAT_B8G8R8A8_UNORM:
      case VK_FORMAT_B8G8R8A8_SNORM:
      case VK_FORMAT_B8G8R8A8_USCALED:
      case VK_FORMAT_B8G8R8A8_SSCALED:
      case VK_FORMAT_B8G8R8A8_UINT:
      case VK_FORMAT_B8G8R8A8_SINT:
      case VK_FORMAT_B8G8R8A8_SRGB:
      case VK_FORMAT_A2R10G10B10_UNORM_PACK32:
      case VK_FORMAT_A2R10G10B10_SNORM_PACK32:
      case VK_FORMAT_A2R10G10B10_USCALED_PACK32:
      case VK_FORMAT_A2R10G10B10_SSCALED_PACK32:
      case VK_FORMAT_A2R10G10B10_UINT_PACK32:
      case VK_FORMAT_A2R10G10B10_SINT_PACK32:
         key.vertex_post_shuffle |= 1 << location;
         break;
      default:
         break;
      }
   }

   const VkPipelineTessellationStateCreateInfo *tess =
      radv_pipeline_get_tessellation_state(pCreateInfo);
   if (tess)
      key.tess_input_vertices = tess->patchControlPoints;

   const VkPipelineMultisampleStateCreateInfo *vkms =
      radv_pipeline_get_multisample_state(pCreateInfo);
   if (vkms && vkms->rasterizationSamples > 1) {
      uint32_t num_samples = vkms->rasterizationSamples;
      uint32_t ps_iter_samples = radv_pipeline_get_ps_iter_samples(pCreateInfo);
      key.num_samples = num_samples;
      key.log2_ps_iter_samples = util_logbase2(ps_iter_samples);
   }

   key.col_format = blend->spi_shader_col_format;
   if (pipeline->device->physical_device->rad_info.chip_class < GFX8) {
      key.is_int8 = blend->col_format_is_int8;
      key.is_int10 = blend->col_format_is_int10;
   }

   if (pipeline->device->physical_device->rad_info.chip_class >= GFX10) {
      key.topology = pCreateInfo->pInputAssemblyState->topology;

      const VkPipelineRasterizationStateCreateInfo *raster_info = pCreateInfo->pRasterizationState;
      const VkPipelineRasterizationProvokingVertexStateCreateInfoEXT *provoking_vtx_info =
         vk_find_struct_const(raster_info->pNext,
                              PIPELINE_RASTERIZATION_PROVOKING_VERTEX_STATE_CREATE_INFO_EXT);
      if (provoking_vtx_info &&
          provoking_vtx_info->provokingVertexMode == VK_PROVOKING_VERTEX_MODE_LAST_VERTEX_EXT) {
         key.provoking_vtx_last = true;
      }
   }
   return key;
}

static bool
radv_nir_stage_uses_xfb(const nir_shader *nir)
{
   nir_xfb_info *xfb = nir_gather_xfb_info(nir, NULL);
   bool uses_xfb = !!xfb;

   ralloc_free(xfb);
   return uses_xfb;
}

static void
radv_fill_shader_keys(struct radv_device *device, struct radv_shader_variant_key *keys,
                      const struct radv_pipeline_key *key, nir_shader **nir)
{
   keys[MESA_SHADER_VERTEX].vs.instance_rate_inputs = key->instance_rate_inputs;
   keys[MESA_SHADER_VERTEX].vs.post_shuffle = key->vertex_post_shuffle;
   for (unsigned i = 0; i < MAX_VERTEX_ATTRIBS; ++i) {
      keys[MESA_SHADER_VERTEX].vs.instance_rate_divisors[i] = key->instance_rate_divisors[i];
      keys[MESA_SHADER_VERTEX].vs.vertex_attribute_formats[i] = key->vertex_attribute_formats[i];
      keys[MESA_SHADER_VERTEX].vs.vertex_attribute_bindings[i] = key->vertex_attribute_bindings[i];
      keys[MESA_SHADER_VERTEX].vs.vertex_attribute_offsets[i] = key->vertex_attribute_offsets[i];
      keys[MESA_SHADER_VERTEX].vs.vertex_attribute_strides[i] = key->vertex_attribute_strides[i];
      keys[MESA_SHADER_VERTEX].vs.alpha_adjust[i] = key->vertex_alpha_adjust[i];
   }
   keys[MESA_SHADER_VERTEX].vs.outprim = si_conv_prim_to_gs_out(key->topology);
   keys[MESA_SHADER_VERTEX].vs.provoking_vtx_last = key->provoking_vtx_last;

   if (nir[MESA_SHADER_TESS_CTRL]) {
      keys[MESA_SHADER_VERTEX].vs_common_out.as_ls = true;
      keys[MESA_SHADER_TESS_CTRL].tcs.input_vertices = key->tess_input_vertices;
      keys[MESA_SHADER_TESS_CTRL].tcs.primitive_mode =
         nir[MESA_SHADER_TESS_EVAL]->info.tess.primitive_mode;
   }

   if (nir[MESA_SHADER_GEOMETRY]) {
      if (nir[MESA_SHADER_TESS_CTRL])
         keys[MESA_SHADER_TESS_EVAL].vs_common_out.as_es = true;
      else
         keys[MESA_SHADER_VERTEX].vs_common_out.as_es = true;
   }

   if (device->physical_device->use_ngg) {
      if (nir[MESA_SHADER_TESS_CTRL]) {
         keys[MESA_SHADER_TESS_EVAL].vs_common_out.as_ngg = true;
      } else {
         keys[MESA_SHADER_VERTEX].vs_common_out.as_ngg = true;
      }

      if (nir[MESA_SHADER_TESS_CTRL] && nir[MESA_SHADER_GEOMETRY] &&
          nir[MESA_SHADER_GEOMETRY]->info.gs.invocations *
                nir[MESA_SHADER_GEOMETRY]->info.gs.vertices_out >
             256) {
         /* Fallback to the legacy path if tessellation is
          * enabled with extreme geometry because
          * EN_MAX_VERT_OUT_PER_GS_INSTANCE doesn't work and it
          * might hang.
          */
         keys[MESA_SHADER_TESS_EVAL].vs_common_out.as_ngg = false;
      }

      gl_shader_stage last_xfb_stage = MESA_SHADER_VERTEX;

      for (int i = MESA_SHADER_VERTEX; i <= MESA_SHADER_GEOMETRY; i++) {
         if (nir[i])
            last_xfb_stage = i;
      }

      bool uses_xfb = nir[last_xfb_stage] && radv_nir_stage_uses_xfb(nir[last_xfb_stage]);

      if (!device->physical_device->use_ngg_streamout && uses_xfb) {
         if (nir[MESA_SHADER_TESS_CTRL])
            keys[MESA_SHADER_TESS_EVAL].vs_common_out.as_ngg = false;
         else
            keys[MESA_SHADER_VERTEX].vs_common_out.as_ngg = false;
      }

      /* Determine if the pipeline is eligible for the NGG passthrough
       * mode. It can't be enabled for geometry shaders, for NGG
       * streamout or for vertex shaders that export the primitive ID
       * (this is checked later because we don't have the info here.)
       */
      if (!nir[MESA_SHADER_GEOMETRY] && !uses_xfb) {
         if (nir[MESA_SHADER_TESS_CTRL] && keys[MESA_SHADER_TESS_EVAL].vs_common_out.as_ngg) {
            keys[MESA_SHADER_TESS_EVAL].vs_common_out.as_ngg_passthrough = true;
         } else if (nir[MESA_SHADER_VERTEX] && keys[MESA_SHADER_VERTEX].vs_common_out.as_ngg) {
            keys[MESA_SHADER_VERTEX].vs_common_out.as_ngg_passthrough = true;
         }
      }
   }

   for (int i = 0; i < MESA_SHADER_STAGES; ++i)
      keys[i].has_multiview_view_index = key->has_multiview_view_index;

   keys[MESA_SHADER_FRAGMENT].fs.col_format = key->col_format;
   keys[MESA_SHADER_FRAGMENT].fs.is_int8 = key->is_int8;
   keys[MESA_SHADER_FRAGMENT].fs.is_int10 = key->is_int10;
   keys[MESA_SHADER_FRAGMENT].fs.log2_ps_iter_samples = key->log2_ps_iter_samples;
   keys[MESA_SHADER_FRAGMENT].fs.num_samples = key->num_samples;

   if (nir[MESA_SHADER_COMPUTE]) {
      unsigned subgroup_size = key->compute_subgroup_size;
      unsigned req_subgroup_size = subgroup_size;
      bool require_full_subgroups = key->require_full_subgroups;

      if (!subgroup_size)
         subgroup_size = device->physical_device->cs_wave_size;

      unsigned local_size = nir[MESA_SHADER_COMPUTE]->info.cs.local_size[0] *
                            nir[MESA_SHADER_COMPUTE]->info.cs.local_size[1] *
                            nir[MESA_SHADER_COMPUTE]->info.cs.local_size[2];

      /* Games don't always request full subgroups when they should,
       * which can cause bugs if cswave32 is enabled.
       */
      if (device->physical_device->cs_wave_size == 32 &&
          nir[MESA_SHADER_COMPUTE]->info.cs.uses_wide_subgroup_intrinsics && !req_subgroup_size &&
          local_size % RADV_SUBGROUP_SIZE == 0)
         require_full_subgroups = true;

      if (require_full_subgroups && !req_subgroup_size) {
         /* don't use wave32 pretending to be wave64 */
         subgroup_size = RADV_SUBGROUP_SIZE;
      }

      keys[MESA_SHADER_COMPUTE].cs.subgroup_size = subgroup_size;
   }
}

static uint8_t
radv_get_wave_size(struct radv_device *device, const VkPipelineShaderStageCreateInfo *pStage,
                   gl_shader_stage stage, const struct radv_shader_variant_key *key,
                   const struct radv_shader_info *info)
{
   if (stage == MESA_SHADER_GEOMETRY && !info->is_ngg)
      return 64;
   else if (stage == MESA_SHADER_COMPUTE) {
      return key->cs.subgroup_size;
   } else if (stage == MESA_SHADER_FRAGMENT)
      return device->physical_device->ps_wave_size;
   else
      return device->physical_device->ge_wave_size;
}

static uint8_t
radv_get_ballot_bit_size(struct radv_device *device, const VkPipelineShaderStageCreateInfo *pStage,
                         gl_shader_stage stage, const struct radv_shader_variant_key *key)
{
   if (stage == MESA_SHADER_COMPUTE && key->cs.subgroup_size)
      return key->cs.subgroup_size;
   return 64;
}

static void
radv_fill_shader_info(struct radv_pipeline *pipeline,
                      const VkPipelineShaderStageCreateInfo **pStages,
                      struct radv_shader_variant_key *keys, struct radv_shader_info *infos,
                      nir_shader **nir)
{
   unsigned active_stages = 0;
   unsigned filled_stages = 0;

   for (int i = 0; i < MESA_SHADER_STAGES; i++) {
      if (nir[i])
         active_stages |= (1 << i);
   }

   if (nir[MESA_SHADER_FRAGMENT]) {
      radv_nir_shader_info_init(&infos[MESA_SHADER_FRAGMENT]);
      radv_nir_shader_info_pass(pipeline->device, nir[MESA_SHADER_FRAGMENT], pipeline->layout,
                                &keys[MESA_SHADER_FRAGMENT], &infos[MESA_SHADER_FRAGMENT]);

      /* TODO: These are no longer used as keys we should refactor this */
      keys[MESA_SHADER_VERTEX].vs_common_out.export_prim_id =
         infos[MESA_SHADER_FRAGMENT].ps.prim_id_input;
      keys[MESA_SHADER_VERTEX].vs_common_out.export_layer_id =
         infos[MESA_SHADER_FRAGMENT].ps.layer_input;
      keys[MESA_SHADER_VERTEX].vs_common_out.export_clip_dists =
         !!infos[MESA_SHADER_FRAGMENT].ps.num_input_clips_culls;
      keys[MESA_SHADER_VERTEX].vs_common_out.export_viewport_index =
         infos[MESA_SHADER_FRAGMENT].ps.viewport_index_input;
      keys[MESA_SHADER_TESS_EVAL].vs_common_out.export_prim_id =
         infos[MESA_SHADER_FRAGMENT].ps.prim_id_input;
      keys[MESA_SHADER_TESS_EVAL].vs_common_out.export_layer_id =
         infos[MESA_SHADER_FRAGMENT].ps.layer_input;
      keys[MESA_SHADER_TESS_EVAL].vs_common_out.export_clip_dists =
         !!infos[MESA_SHADER_FRAGMENT].ps.num_input_clips_culls;
      keys[MESA_SHADER_TESS_EVAL].vs_common_out.export_viewport_index =
         infos[MESA_SHADER_FRAGMENT].ps.viewport_index_input;

      /* NGG passthrough mode can't be enabled for vertex shaders
       * that export the primitive ID.
       *
       * TODO: I should really refactor the keys logic.
       */
      if (nir[MESA_SHADER_VERTEX] && keys[MESA_SHADER_VERTEX].vs_common_out.export_prim_id) {
         keys[MESA_SHADER_VERTEX].vs_common_out.as_ngg_passthrough = false;
      }

      filled_stages |= (1 << MESA_SHADER_FRAGMENT);
   }

   if (pipeline->device->physical_device->rad_info.chip_class >= GFX9 &&
       nir[MESA_SHADER_TESS_CTRL]) {
      struct nir_shader *combined_nir[] = {nir[MESA_SHADER_VERTEX], nir[MESA_SHADER_TESS_CTRL]};
      struct radv_shader_variant_key key = keys[MESA_SHADER_TESS_CTRL];
      key.tcs.vs_key = keys[MESA_SHADER_VERTEX].vs;

      radv_nir_shader_info_init(&infos[MESA_SHADER_TESS_CTRL]);

      for (int i = 0; i < 2; i++) {
         radv_nir_shader_info_pass(pipeline->device, combined_nir[i], pipeline->layout, &key,
                                   &infos[MESA_SHADER_TESS_CTRL]);
      }

      filled_stages |= (1 << MESA_SHADER_VERTEX);
      filled_stages |= (1 << MESA_SHADER_TESS_CTRL);
   }

   if (pipeline->device->physical_device->rad_info.chip_class >= GFX9 &&
       nir[MESA_SHADER_GEOMETRY]) {
      gl_shader_stage pre_stage =
         nir[MESA_SHADER_TESS_EVAL] ? MESA_SHADER_TESS_EVAL : MESA_SHADER_VERTEX;
      struct nir_shader *combined_nir[] = {nir[pre_stage], nir[MESA_SHADER_GEOMETRY]};

      radv_nir_shader_info_init(&infos[MESA_SHADER_GEOMETRY]);

      for (int i = 0; i < 2; i++) {
         radv_nir_shader_info_pass(pipeline->device, combined_nir[i], pipeline->layout,
                                   &keys[pre_stage], &infos[MESA_SHADER_GEOMETRY]);
      }

      filled_stages |= (1 << pre_stage);
      filled_stages |= (1 << MESA_SHADER_GEOMETRY);
   }

   active_stages ^= filled_stages;
   while (active_stages) {
      int i = u_bit_scan(&active_stages);
      radv_nir_shader_info_init(&infos[i]);
      radv_nir_shader_info_pass(pipeline->device, nir[i], pipeline->layout, &keys[i], &infos[i]);
   }

   for (int i = 0; i < MESA_SHADER_STAGES; i++) {
      if (nir[i]) {
         infos[i].wave_size = radv_get_wave_size(pipeline->device, pStages[i], i, &keys[i], &infos[i]);
         infos[i].ballot_bit_size =
            radv_get_ballot_bit_size(pipeline->device, pStages[i], i, &keys[i]);
      }
   }
}

static void
merge_tess_info(struct shader_info *tes_info, struct shader_info *tcs_info)
{
   /* The Vulkan 1.0.38 spec, section 21.1 Tessellator says:
    *
    *    "PointMode. Controls generation of points rather than triangles
    *     or lines. This functionality defaults to disabled, and is
    *     enabled if either shader stage includes the execution mode.
    *
    * and about Triangles, Quads, IsoLines, VertexOrderCw, VertexOrderCcw,
    * PointMode, SpacingEqual, SpacingFractionalEven, SpacingFractionalOdd,
    * and OutputVertices, it says:
    *
    *    "One mode must be set in at least one of the tessellation
    *     shader stages."
    *
    * So, the fields can be set in either the TCS or TES, but they must
    * agree if set in both.  Our backend looks at TES, so bitwise-or in
    * the values from the TCS.
    */
   assert(tcs_info->tess.tcs_vertices_out == 0 || tes_info->tess.tcs_vertices_out == 0 ||
          tcs_info->tess.tcs_vertices_out == tes_info->tess.tcs_vertices_out);
   tes_info->tess.tcs_vertices_out |= tcs_info->tess.tcs_vertices_out;

   assert(tcs_info->tess.spacing == TESS_SPACING_UNSPECIFIED ||
          tes_info->tess.spacing == TESS_SPACING_UNSPECIFIED ||
          tcs_info->tess.spacing == tes_info->tess.spacing);
   tes_info->tess.spacing |= tcs_info->tess.spacing;

   assert(tcs_info->tess.primitive_mode == 0 || tes_info->tess.primitive_mode == 0 ||
          tcs_info->tess.primitive_mode == tes_info->tess.primitive_mode);
   tes_info->tess.primitive_mode |= tcs_info->tess.primitive_mode;
   tes_info->tess.ccw |= tcs_info->tess.ccw;
   tes_info->tess.point_mode |= tcs_info->tess.point_mode;

   /* Copy the merged info back to the TCS */
   tcs_info->tess.tcs_vertices_out = tes_info->tess.tcs_vertices_out;
   tcs_info->tess.spacing = tes_info->tess.spacing;
   tcs_info->tess.primitive_mode = tes_info->tess.primitive_mode;
   tcs_info->tess.ccw = tes_info->tess.ccw;
   tcs_info->tess.point_mode = tes_info->tess.point_mode;
}

static void
gather_tess_info(struct radv_device *device, nir_shader **nir, struct radv_shader_info *infos,
                 const struct radv_pipeline_key *pipeline_key)
{
   merge_tess_info(&nir[MESA_SHADER_TESS_EVAL]->info, &nir[MESA_SHADER_TESS_CTRL]->info);

   /* Number of tessellation patches per workgroup processed by the current pipeline. */
   unsigned num_patches = get_tcs_num_patches(
      pipeline_key->tess_input_vertices, nir[MESA_SHADER_TESS_CTRL]->info.tess.tcs_vertices_out,
      infos[MESA_SHADER_TESS_CTRL].tcs.num_linked_inputs,
      infos[MESA_SHADER_TESS_CTRL].tcs.num_linked_outputs,
      infos[MESA_SHADER_TESS_CTRL].tcs.num_linked_patch_outputs, device->tess_offchip_block_dw_size,
      device->physical_device->rad_info.chip_class, device->physical_device->rad_info.family);

   /* LDS size used by VS+TCS for storing TCS inputs and outputs. */
   unsigned tcs_lds_size = calculate_tess_lds_size(
      device->physical_device->rad_info.chip_class, pipeline_key->tess_input_vertices,
      nir[MESA_SHADER_TESS_CTRL]->info.tess.tcs_vertices_out,
      infos[MESA_SHADER_TESS_CTRL].tcs.num_linked_inputs, num_patches,
      infos[MESA_SHADER_TESS_CTRL].tcs.num_linked_outputs,
      infos[MESA_SHADER_TESS_CTRL].tcs.num_linked_patch_outputs);

   infos[MESA_SHADER_TESS_CTRL].num_tess_patches = num_patches;
   infos[MESA_SHADER_TESS_CTRL].tcs.num_lds_blocks = tcs_lds_size;
   infos[MESA_SHADER_TESS_CTRL].tcs.tes_reads_tess_factors =
      !!(nir[MESA_SHADER_TESS_EVAL]->info.inputs_read &
         (VARYING_BIT_TESS_LEVEL_INNER | VARYING_BIT_TESS_LEVEL_OUTER));
   infos[MESA_SHADER_TESS_CTRL].tcs.tes_inputs_read = nir[MESA_SHADER_TESS_EVAL]->info.inputs_read;
   infos[MESA_SHADER_TESS_CTRL].tcs.tes_patch_inputs_read =
      nir[MESA_SHADER_TESS_EVAL]->info.patch_inputs_read;

   infos[MESA_SHADER_TESS_EVAL].num_tess_patches = num_patches;
   infos[MESA_SHADER_GEOMETRY].num_tess_patches = num_patches;

   if (!radv_use_llvm_for_stage(device, MESA_SHADER_VERTEX)) {
      /* When the number of TCS input and output vertices are the same (typically 3):
       * - There is an equal amount of LS and HS invocations
       * - In case of merged LSHS shaders, the LS and HS halves of the shader
       *   always process the exact same vertex. We can use this knowledge to optimize them.
       *
       * We don't set tcs_in_out_eq if the float controls differ because that might
       * involve different float modes for the same block and our optimizer
       * doesn't handle a instruction dominating another with a different mode.
       */
      infos[MESA_SHADER_VERTEX].vs.tcs_in_out_eq =
         device->physical_device->rad_info.chip_class >= GFX9 &&
         pipeline_key->tess_input_vertices ==
            nir[MESA_SHADER_TESS_CTRL]->info.tess.tcs_vertices_out &&
         nir[MESA_SHADER_VERTEX]->info.float_controls_execution_mode ==
            nir[MESA_SHADER_TESS_CTRL]->info.float_controls_execution_mode;

      if (infos[MESA_SHADER_VERTEX].vs.tcs_in_out_eq)
         infos[MESA_SHADER_VERTEX].vs.tcs_temp_only_input_mask =
            nir[MESA_SHADER_TESS_CTRL]->info.inputs_read &
            nir[MESA_SHADER_VERTEX]->info.outputs_written &
            ~nir[MESA_SHADER_TESS_CTRL]->info.tess.tcs_cross_invocation_inputs_read &
            ~nir[MESA_SHADER_TESS_CTRL]->info.inputs_read_indirectly &
            ~nir[MESA_SHADER_VERTEX]->info.outputs_accessed_indirectly;

      /* Copy data to TCS so it can be accessed by the backend if they are merged. */
      infos[MESA_SHADER_TESS_CTRL].vs.tcs_in_out_eq = infos[MESA_SHADER_VERTEX].vs.tcs_in_out_eq;
      infos[MESA_SHADER_TESS_CTRL].vs.tcs_temp_only_input_mask =
         infos[MESA_SHADER_VERTEX].vs.tcs_temp_only_input_mask;
   }
}

static void
radv_init_feedback(const VkPipelineCreationFeedbackCreateInfoEXT *ext)
{
   if (!ext)
      return;

   if (ext->pPipelineCreationFeedback) {
      ext->pPipelineCreationFeedback->flags = 0;
      ext->pPipelineCreationFeedback->duration = 0;
   }

   for (unsigned i = 0; i < ext->pipelineStageCreationFeedbackCount; ++i) {
      ext->pPipelineStageCreationFeedbacks[i].flags = 0;
      ext->pPipelineStageCreationFeedbacks[i].duration = 0;
   }
}

static void
radv_start_feedback(VkPipelineCreationFeedbackEXT *feedback)
{
   if (!feedback)
      return;

   feedback->duration -= radv_get_current_time();
   feedback->flags = VK_PIPELINE_CREATION_FEEDBACK_VALID_BIT_EXT;
}

static void
radv_stop_feedback(VkPipelineCreationFeedbackEXT *feedback, bool cache_hit)
{
   if (!feedback)
      return;

   feedback->duration += radv_get_current_time();
   feedback->flags =
      VK_PIPELINE_CREATION_FEEDBACK_VALID_BIT_EXT |
      (cache_hit ? VK_PIPELINE_CREATION_FEEDBACK_APPLICATION_PIPELINE_CACHE_HIT_BIT_EXT : 0);
}

static bool
mem_vectorize_callback(unsigned align_mul, unsigned align_offset, unsigned bit_size,
                       unsigned num_components, nir_intrinsic_instr *low, nir_intrinsic_instr *high,
                       void *data)
{
   if (num_components > 4)
      return false;

   /* >128 bit loads are split except with SMEM */
   if (bit_size * num_components > 128)
      return false;

   uint32_t align;
   if (align_offset)
      align = 1 << (ffs(align_offset) - 1);
   else
      align = align_mul;

   switch (low->intrinsic) {
   case nir_intrinsic_load_global:
   case nir_intrinsic_store_global:
   case nir_intrinsic_store_ssbo:
   case nir_intrinsic_load_ssbo:
   case nir_intrinsic_load_ubo:
   case nir_intrinsic_load_push_constant: {
      unsigned max_components;
      if (align % 4 == 0)
         max_components = NIR_MAX_VEC_COMPONENTS;
      else if (align % 2 == 0)
         max_components = 16u / bit_size;
      else
         max_components = 8u / bit_size;
      return (align % (bit_size / 8u)) == 0 && num_components <= max_components;
   }
   case nir_intrinsic_load_deref:
   case nir_intrinsic_store_deref:
      assert(nir_deref_mode_is(nir_src_as_deref(low->src[0]), nir_var_mem_shared));
      FALLTHROUGH;
   case nir_intrinsic_load_shared:
   case nir_intrinsic_store_shared:
      if (bit_size * num_components ==
          96) /* 96 bit loads require 128 bit alignment and are split otherwise */
         return align % 16 == 0;
      else if (bit_size * num_components ==
               128) /* 128 bit loads require 64 bit alignment and are split otherwise */
         return align % 8 == 0;
      else
         return align % (bit_size == 8 ? 2 : 4) == 0;
   default:
      return false;
   }
   return false;
}

static unsigned
lower_bit_size_callback(const nir_instr *instr, void *_)
{
   struct radv_device *device = _;
   enum chip_class chip = device->physical_device->rad_info.chip_class;

   if (instr->type != nir_instr_type_alu)
      return 0;
   nir_alu_instr *alu = nir_instr_as_alu(instr);

   if (alu->dest.dest.ssa.bit_size & (8 | 16)) {
      unsigned bit_size = alu->dest.dest.ssa.bit_size;
      switch (alu->op) {
      case nir_op_iabs:
      case nir_op_bitfield_select:
      case nir_op_imul_high:
      case nir_op_umul_high:
      case nir_op_ineg:
      case nir_op_isign:
         return 32;
      case nir_op_imax:
      case nir_op_umax:
      case nir_op_imin:
      case nir_op_umin:
      case nir_op_ishr:
      case nir_op_ushr:
      case nir_op_ishl:
      case nir_op_uadd_sat:
         return (bit_size == 8 || !(chip >= GFX8 && nir_dest_is_divergent(alu->dest.dest))) ? 32
                                                                                            : 0;
      default:
         return 0;
      }
   }

   if (nir_src_bit_size(alu->src[0].src) & (8 | 16)) {
      unsigned bit_size = nir_src_bit_size(alu->src[0].src);
      switch (alu->op) {
      case nir_op_bit_count:
      case nir_op_find_lsb:
      case nir_op_ufind_msb:
      case nir_op_i2b1:
         return 32;
      case nir_op_ilt:
      case nir_op_ige:
      case nir_op_ieq:
      case nir_op_ine:
      case nir_op_ult:
      case nir_op_uge:
         return (bit_size == 8 || !(chip >= GFX8 && nir_dest_is_divergent(alu->dest.dest))) ? 32
                                                                                            : 0;
      default:
         return 0;
      }
   }

   return 0;
}

static bool
opt_vectorize_callback(const nir_instr *instr, void *_)
{
   assert(instr->type == nir_instr_type_alu);
   nir_alu_instr *alu = nir_instr_as_alu(instr);
   unsigned bit_size = alu->dest.dest.ssa.bit_size;
   if (bit_size != 16)
      return false;

   switch (alu->op) {
   case nir_op_fadd:
   case nir_op_fsub:
   case nir_op_fmul:
   case nir_op_fneg:
   case nir_op_fsat:
   case nir_op_fmin:
   case nir_op_fmax:
   case nir_op_iadd:
   case nir_op_isub:
   case nir_op_imul:
   case nir_op_imin:
   case nir_op_imax:
   case nir_op_umin:
   case nir_op_umax:
      return true;
   case nir_op_ishl: /* TODO: in NIR, these have 32bit shift operands */
   case nir_op_ishr: /* while Radeon needs 16bit operands when vectorized */
   case nir_op_ushr:
   default:
      return false;
   }
}

static nir_component_mask_t
non_uniform_access_callback(const nir_src *src, void *_)
{
   if (src->ssa->num_components == 1)
      return 0x1;
   return nir_chase_binding(*src).success ? 0x2 : 0x3;
}

VkResult
radv_create_shaders(struct radv_pipeline *pipeline, struct radv_device *device,
                    struct radv_pipeline_cache *cache, const struct radv_pipeline_key *pipeline_key,
                    const VkPipelineShaderStageCreateInfo **pStages,
                    const VkPipelineCreateFlags flags,
                    VkPipelineCreationFeedbackEXT *pipeline_feedback,
                    VkPipelineCreationFeedbackEXT **stage_feedbacks)
{
   struct vk_shader_module fs_m = {0};
   struct vk_shader_module *modules[MESA_SHADER_STAGES] = {
      0,
   };
   nir_shader *nir[MESA_SHADER_STAGES] = {0};
   struct radv_shader_binary *binaries[MESA_SHADER_STAGES] = {NULL};
   struct radv_shader_variant_key keys[MESA_SHADER_STAGES] = {{{{{0}}}}};
   struct radv_shader_info infos[MESA_SHADER_STAGES] = {0};
   unsigned char hash[20], gs_copy_hash[20];
   bool keep_executable_info =
      (flags & VK_PIPELINE_CREATE_CAPTURE_INTERNAL_REPRESENTATIONS_BIT_KHR) ||
      device->keep_shader_info;
   bool keep_statistic_info = (flags & VK_PIPELINE_CREATE_CAPTURE_STATISTICS_BIT_KHR) ||
                              (device->instance->debug_flags & RADV_DEBUG_DUMP_SHADER_STATS) ||
                              device->keep_shader_info;
   bool disable_optimizations = flags & VK_PIPELINE_CREATE_DISABLE_OPTIMIZATION_BIT;

   radv_start_feedback(pipeline_feedback);

   for (unsigned i = 0; i < MESA_SHADER_STAGES; ++i) {
      if (pStages[i]) {
         modules[i] = vk_shader_module_from_handle(pStages[i]->module);
         if (modules[i]->nir)
            _mesa_sha1_compute(modules[i]->nir->info.name, strlen(modules[i]->nir->info.name),
                               modules[i]->sha1);

         pipeline->active_stages |= mesa_to_vk_shader_stage(i);
      }
   }

   radv_hash_shaders(hash, pStages, pipeline->layout, pipeline_key,
                     get_hash_flags(device, keep_statistic_info));
   memcpy(gs_copy_hash, hash, 20);
   gs_copy_hash[0] ^= 1;

   pipeline->pipeline_hash = *(uint64_t *)hash;

   bool found_in_application_cache = true;
   if (modules[MESA_SHADER_GEOMETRY] && !keep_executable_info) {
      struct radv_shader_variant *variants[MESA_SHADER_STAGES] = {0};
      radv_create_shader_variants_from_pipeline_cache(device, cache, gs_copy_hash, variants,
                                                      &found_in_application_cache);
      pipeline->gs_copy_shader = variants[MESA_SHADER_GEOMETRY];
   }

   if (!keep_executable_info &&
       radv_create_shader_variants_from_pipeline_cache(device, cache, hash, pipeline->shaders,
                                                       &found_in_application_cache) &&
       (!modules[MESA_SHADER_GEOMETRY] || pipeline->gs_copy_shader)) {
      radv_stop_feedback(pipeline_feedback, found_in_application_cache);
      return VK_SUCCESS;
   }

   if (flags & VK_PIPELINE_CREATE_FAIL_ON_PIPELINE_COMPILE_REQUIRED_BIT_EXT) {
      radv_stop_feedback(pipeline_feedback, found_in_application_cache);
      return VK_PIPELINE_COMPILE_REQUIRED_EXT;
   }

   if (!modules[MESA_SHADER_FRAGMENT] && !modules[MESA_SHADER_COMPUTE]) {
      nir_builder fs_b = nir_builder_init_simple_shader(MESA_SHADER_FRAGMENT, NULL, "noop_fs");
      fs_m = vk_shader_module_from_nir(fs_b.shader);
      modules[MESA_SHADER_FRAGMENT] = &fs_m;
   }

   for (unsigned i = 0; i < MESA_SHADER_STAGES; ++i) {
      const VkPipelineShaderStageCreateInfo *stage = pStages[i];

      if (!modules[i])
         continue;

      radv_start_feedback(stage_feedbacks[i]);

      nir[i] = radv_shader_compile_to_nir(device, modules[i], stage ? stage->pName : "main", i,
                                          stage ? stage->pSpecializationInfo : NULL, flags,
                                          pipeline->layout, pipeline_key);

      /* We don't want to alter meta shaders IR directly so clone it
       * first.
       */
      if (nir[i]->info.name) {
         nir[i] = nir_shader_clone(NULL, nir[i]);
      }

      radv_stop_feedback(stage_feedbacks[i], false);
   }

   bool optimize_conservatively = flags & VK_PIPELINE_CREATE_DISABLE_OPTIMIZATION_BIT;

   radv_link_shaders(pipeline, nir, optimize_conservatively);
   radv_set_driver_locations(pipeline, nir, infos);

   for (int i = 0; i < MESA_SHADER_STAGES; ++i) {
      if (nir[i]) {
         radv_start_feedback(stage_feedbacks[i]);
         radv_optimize_nir(device, nir[i], optimize_conservatively, false);

         /* Gather info again, information such as outputs_read can be out-of-date. */
         nir_shader_gather_info(nir[i], nir_shader_get_entrypoint(nir[i]));
         radv_lower_io(device, nir[i]);

         radv_stop_feedback(stage_feedbacks[i], false);
      }
   }

   infos[MESA_SHADER_VERTEX].vs.as_ls = !!nir[MESA_SHADER_TESS_CTRL];
   infos[MESA_SHADER_VERTEX].vs.as_es = !!nir[MESA_SHADER_GEOMETRY] && !nir[MESA_SHADER_TESS_CTRL];
   infos[MESA_SHADER_TESS_EVAL].tes.as_es =
      !!nir[MESA_SHADER_GEOMETRY] && !!nir[MESA_SHADER_TESS_CTRL];

   if (nir[MESA_SHADER_TESS_CTRL]) {
      nir_lower_patch_vertices(nir[MESA_SHADER_TESS_EVAL],
                               nir[MESA_SHADER_TESS_CTRL]->info.tess.tcs_vertices_out, NULL);
      gather_tess_info(device, nir, infos, pipeline_key);
   }

   radv_fill_shader_keys(device, keys, pipeline_key, nir);
   radv_fill_shader_info(pipeline, pStages, keys, infos, nir);

   if ((nir[MESA_SHADER_VERTEX] && keys[MESA_SHADER_VERTEX].vs_common_out.as_ngg) ||
       (nir[MESA_SHADER_TESS_EVAL] && keys[MESA_SHADER_TESS_EVAL].vs_common_out.as_ngg)) {
      struct gfx10_ngg_info *ngg_info;

      if (nir[MESA_SHADER_GEOMETRY])
         ngg_info = &infos[MESA_SHADER_GEOMETRY].ngg_info;
      else if (nir[MESA_SHADER_TESS_CTRL])
         ngg_info = &infos[MESA_SHADER_TESS_EVAL].ngg_info;
      else
         ngg_info = &infos[MESA_SHADER_VERTEX].ngg_info;

      gfx10_get_ngg_info(pipeline_key, pipeline, nir, infos, ngg_info);
   } else if (nir[MESA_SHADER_GEOMETRY]) {
      struct gfx9_gs_info *gs_info = &infos[MESA_SHADER_GEOMETRY].gs_ring_info;

      gfx9_get_gs_info(pipeline_key, pipeline, nir, infos, gs_info);
   }

   for (int i = 0; i < MESA_SHADER_STAGES; ++i) {
      if (nir[i]) {
         radv_start_feedback(stage_feedbacks[i]);

         if (!radv_use_llvm_for_stage(device, i)) {
            nir_lower_non_uniform_access_options options = {
               .types = nir_lower_non_uniform_ubo_access | nir_lower_non_uniform_ssbo_access |
                        nir_lower_non_uniform_texture_access | nir_lower_non_uniform_image_access,
               .callback = &non_uniform_access_callback,
               .callback_data = NULL,
            };
            NIR_PASS_V(nir[i], nir_lower_non_uniform_access, &options);
         }
         NIR_PASS_V(nir[i], nir_lower_memory_model);

         bool lower_to_scalar = false;

         nir_load_store_vectorize_options vectorize_opts = {
            .modes = nir_var_mem_ssbo | nir_var_mem_ubo | nir_var_mem_push_const |
                     nir_var_mem_shared | nir_var_mem_global,
            .callback = mem_vectorize_callback,
            .robust_modes = 0,
         };

         if (device->robust_buffer_access2) {
            vectorize_opts.robust_modes =
               nir_var_mem_ubo | nir_var_mem_ssbo | nir_var_mem_global | nir_var_mem_push_const;
         }

         if (nir_opt_load_store_vectorize(nir[i], &vectorize_opts)) {
            NIR_PASS_V(nir[i], nir_copy_prop);
            lower_to_scalar = true;

            /* Gather info again, to update whether 8/16-bit are used. */
            nir_shader_gather_info(nir[i], nir_shader_get_entrypoint(nir[i]));
         }

         lower_to_scalar |=
            nir_opt_shrink_vectors(nir[i], !device->instance->disable_shrink_image_store);

         if (lower_to_scalar)
            nir_lower_alu_to_scalar(nir[i], NULL, NULL);

         /* lower ALU operations */
         /* TODO: Some 64-bit tests crash inside LLVM. */
         if (!radv_use_llvm_for_stage(device, i))
            nir_lower_int64(nir[i]);

         /* TODO: Implement nir_op_uadd_sat with LLVM. */
         if (!radv_use_llvm_for_stage(device, i))
            nir_opt_idiv_const(nir[i], 8);

         nir_lower_idiv(nir[i],
                        &(nir_lower_idiv_options){
                           .imprecise_32bit_lowering = false,
                           .allow_fp16 = device->physical_device->rad_info.chip_class >= GFX9,
                        });

         nir_opt_sink(nir[i], nir_move_load_input | nir_move_const_undef | nir_move_copies);
         nir_opt_move(nir[i], nir_move_load_input | nir_move_const_undef | nir_move_copies);

         /* Lower I/O intrinsics to memory instructions. */
         bool io_to_mem = radv_lower_io_to_mem(device, nir[i], &infos[i], pipeline_key);
         bool lowered_ngg = radv_lower_ngg(device, nir[i], !!nir[MESA_SHADER_GEOMETRY], &infos[i], pipeline_key, &keys[i]);

         /* optimize the lowered ALU operations */
         bool more_algebraic = true;
         while (more_algebraic) {
            more_algebraic = false;
            NIR_PASS_V(nir[i], nir_copy_prop);
            NIR_PASS_V(nir[i], nir_opt_dce);
            NIR_PASS_V(nir[i], nir_opt_constant_folding);
            NIR_PASS_V(nir[i], nir_opt_cse);
            NIR_PASS(more_algebraic, nir[i], nir_opt_algebraic);
         }

         if (io_to_mem || lowered_ngg || i == MESA_SHADER_COMPUTE)
            NIR_PASS_V(nir[i], nir_opt_offsets);

         /* Do late algebraic optimization to turn add(a,
          * neg(b)) back into subs, then the mandatory cleanup
          * after algebraic.  Note that it may produce fnegs,
          * and if so then we need to keep running to squash
          * fneg(fneg(a)).
          */
         bool more_late_algebraic = true;
         while (more_late_algebraic) {
            more_late_algebraic = false;
            NIR_PASS(more_late_algebraic, nir[i], nir_opt_algebraic_late);
            NIR_PASS_V(nir[i], nir_opt_constant_folding);
            NIR_PASS_V(nir[i], nir_copy_prop);
            NIR_PASS_V(nir[i], nir_opt_dce);
            NIR_PASS_V(nir[i], nir_opt_cse);
         }

         if (nir[i]->info.bit_sizes_int & (8 | 16)) {
            if (device->physical_device->rad_info.chip_class >= GFX8) {
               nir_convert_to_lcssa(nir[i], true, true);
               nir_divergence_analysis(nir[i]);
            }

            if (nir_lower_bit_size(nir[i], lower_bit_size_callback, device)) {
               NIR_PASS_V(nir[i], nir_opt_constant_folding);
               NIR_PASS_V(nir[i], nir_opt_dce);
            }

            if (device->physical_device->rad_info.chip_class >= GFX8)
               nir_opt_remove_phis(nir[i]); /* cleanup LCSSA phis */
         }
         if (((nir[i]->info.bit_sizes_int | nir[i]->info.bit_sizes_float) & 16) &&
             device->physical_device->rad_info.chip_class >= GFX9)
            NIR_PASS_V(nir[i], nir_opt_vectorize, opt_vectorize_callback, NULL);

         /* cleanup passes */
         nir_lower_load_const_to_scalar(nir[i]);
         nir_move_options move_opts = nir_move_const_undef | nir_move_load_ubo |
                                      nir_move_load_input | nir_move_comparisons | nir_move_copies;
         nir_opt_sink(nir[i], move_opts | nir_move_load_ssbo);
         nir_opt_move(nir[i], move_opts);

         radv_stop_feedback(stage_feedbacks[i], false);
      }
   }

   for (int i = 0; i < MESA_SHADER_STAGES; ++i) {
      if (radv_can_dump_shader(device, modules[i], false))
         nir_print_shader(nir[i], stderr);
   }

   if ((nir[MESA_SHADER_VERTEX] && keys[MESA_SHADER_VERTEX].vs_common_out.as_ngg) ||
       (nir[MESA_SHADER_TESS_EVAL] && keys[MESA_SHADER_TESS_EVAL].vs_common_out.as_ngg)) {
      struct gfx10_ngg_info *ngg_info;

      if (nir[MESA_SHADER_GEOMETRY])
         ngg_info = &infos[MESA_SHADER_GEOMETRY].ngg_info;
      else if (nir[MESA_SHADER_TESS_CTRL])
         ngg_info = &infos[MESA_SHADER_TESS_EVAL].ngg_info;
      else
         ngg_info = &infos[MESA_SHADER_VERTEX].ngg_info;

      gfx10_get_ngg_info(pipeline_key, pipeline, nir, infos, ngg_info);
   } else if (nir[MESA_SHADER_GEOMETRY]) {
      struct gfx9_gs_info *gs_info = &infos[MESA_SHADER_GEOMETRY].gs_ring_info;

      gfx9_get_gs_info(pipeline_key, pipeline, nir, infos, gs_info);
   }

   if (modules[MESA_SHADER_GEOMETRY]) {
      struct radv_shader_binary *gs_copy_binary = NULL;
      if (!radv_pipeline_has_ngg(pipeline)) {
         struct radv_shader_info info = {0};
         struct radv_shader_variant_key key = {0};

         key.has_multiview_view_index = keys[MESA_SHADER_GEOMETRY].has_multiview_view_index;

         radv_nir_shader_info_pass(device, nir[MESA_SHADER_GEOMETRY], pipeline->layout, &key,
                                   &info);
         info.wave_size = 64; /* Wave32 not supported. */
         info.ballot_bit_size = 64;

         pipeline->gs_copy_shader = radv_create_gs_copy_shader(
            device, nir[MESA_SHADER_GEOMETRY], &info, &gs_copy_binary, keep_executable_info,
            keep_statistic_info, keys[MESA_SHADER_GEOMETRY].has_multiview_view_index,
            disable_optimizations);
      }

      if (!keep_executable_info && pipeline->gs_copy_shader) {
         struct radv_shader_binary *gs_binaries[MESA_SHADER_STAGES] = {NULL};
         struct radv_shader_variant *gs_variants[MESA_SHADER_STAGES] = {0};

         gs_binaries[MESA_SHADER_GEOMETRY] = gs_copy_binary;
         gs_variants[MESA_SHADER_GEOMETRY] = pipeline->gs_copy_shader;

         radv_pipeline_cache_insert_shaders(device, cache, gs_copy_hash, gs_variants, gs_binaries);

         pipeline->gs_copy_shader = gs_variants[MESA_SHADER_GEOMETRY];
      }
      free(gs_copy_binary);
   }

   if (nir[MESA_SHADER_FRAGMENT]) {
      if (!pipeline->shaders[MESA_SHADER_FRAGMENT]) {
         radv_start_feedback(stage_feedbacks[MESA_SHADER_FRAGMENT]);

         pipeline->shaders[MESA_SHADER_FRAGMENT] = radv_shader_variant_compile(
            device, modules[MESA_SHADER_FRAGMENT], &nir[MESA_SHADER_FRAGMENT], 1, pipeline->layout,
            keys + MESA_SHADER_FRAGMENT, infos + MESA_SHADER_FRAGMENT, keep_executable_info,
            keep_statistic_info, disable_optimizations, &binaries[MESA_SHADER_FRAGMENT]);

         radv_stop_feedback(stage_feedbacks[MESA_SHADER_FRAGMENT], false);
      }
   }

   if (device->physical_device->rad_info.chip_class >= GFX9 && modules[MESA_SHADER_TESS_CTRL]) {
      if (!pipeline->shaders[MESA_SHADER_TESS_CTRL]) {
         struct nir_shader *combined_nir[] = {nir[MESA_SHADER_VERTEX], nir[MESA_SHADER_TESS_CTRL]};
         struct radv_shader_variant_key key = keys[MESA_SHADER_TESS_CTRL];
         key.tcs.vs_key = keys[MESA_SHADER_VERTEX].vs;

         radv_start_feedback(stage_feedbacks[MESA_SHADER_TESS_CTRL]);

         pipeline->shaders[MESA_SHADER_TESS_CTRL] = radv_shader_variant_compile(
            device, modules[MESA_SHADER_TESS_CTRL], combined_nir, 2, pipeline->layout, &key,
            &infos[MESA_SHADER_TESS_CTRL], keep_executable_info, keep_statistic_info,
            disable_optimizations, &binaries[MESA_SHADER_TESS_CTRL]);

         radv_stop_feedback(stage_feedbacks[MESA_SHADER_TESS_CTRL], false);
      }
      modules[MESA_SHADER_VERTEX] = NULL;
   }

   if (device->physical_device->rad_info.chip_class >= GFX9 && modules[MESA_SHADER_GEOMETRY]) {
      gl_shader_stage pre_stage =
         modules[MESA_SHADER_TESS_EVAL] ? MESA_SHADER_TESS_EVAL : MESA_SHADER_VERTEX;
      if (!pipeline->shaders[MESA_SHADER_GEOMETRY]) {
         struct nir_shader *combined_nir[] = {nir[pre_stage], nir[MESA_SHADER_GEOMETRY]};

         radv_start_feedback(stage_feedbacks[MESA_SHADER_GEOMETRY]);

         pipeline->shaders[MESA_SHADER_GEOMETRY] = radv_shader_variant_compile(
            device, modules[MESA_SHADER_GEOMETRY], combined_nir, 2, pipeline->layout,
            &keys[pre_stage], &infos[MESA_SHADER_GEOMETRY], keep_executable_info,
            keep_statistic_info, disable_optimizations, &binaries[MESA_SHADER_GEOMETRY]);

         radv_stop_feedback(stage_feedbacks[MESA_SHADER_GEOMETRY], false);
      }
      modules[pre_stage] = NULL;
   }

   for (int i = 0; i < MESA_SHADER_STAGES; ++i) {
      if (modules[i] && !pipeline->shaders[i]) {
         radv_start_feedback(stage_feedbacks[i]);

         pipeline->shaders[i] = radv_shader_variant_compile(
            device, modules[i], &nir[i], 1, pipeline->layout, keys + i, infos + i,
            keep_executable_info, keep_statistic_info, disable_optimizations, &binaries[i]);

         radv_stop_feedback(stage_feedbacks[i], false);
      }
   }

   if (!keep_executable_info) {
      radv_pipeline_cache_insert_shaders(device, cache, hash, pipeline->shaders, binaries);
   }

   for (int i = 0; i < MESA_SHADER_STAGES; ++i) {
      free(binaries[i]);
      if (nir[i]) {
         ralloc_free(nir[i]);

         if (radv_can_dump_shader_stats(device, modules[i])) {
            radv_dump_shader_stats(device, pipeline, i, stderr);
         }
      }
   }

   if (fs_m.nir)
      ralloc_free(fs_m.nir);

   radv_stop_feedback(pipeline_feedback, false);
   return VK_SUCCESS;
}

static uint32_t
radv_pipeline_stage_to_user_data_0(struct radv_pipeline *pipeline, gl_shader_stage stage,
                                   enum chip_class chip_class)
{
   bool has_gs = radv_pipeline_has_gs(pipeline);
   bool has_tess = radv_pipeline_has_tess(pipeline);
   bool has_ngg = radv_pipeline_has_ngg(pipeline);

   switch (stage) {
   case MESA_SHADER_FRAGMENT:
      return R_00B030_SPI_SHADER_USER_DATA_PS_0;
   case MESA_SHADER_VERTEX:
      if (has_tess) {
         if (chip_class >= GFX10) {
            return R_00B430_SPI_SHADER_USER_DATA_HS_0;
         } else if (chip_class == GFX9) {
            return R_00B430_SPI_SHADER_USER_DATA_LS_0;
         } else {
            return R_00B530_SPI_SHADER_USER_DATA_LS_0;
         }
      }

      if (has_gs) {
         if (chip_class >= GFX10) {
            return R_00B230_SPI_SHADER_USER_DATA_GS_0;
         } else {
            return R_00B330_SPI_SHADER_USER_DATA_ES_0;
         }
      }

      if (has_ngg)
         return R_00B230_SPI_SHADER_USER_DATA_GS_0;

      return R_00B130_SPI_SHADER_USER_DATA_VS_0;
   case MESA_SHADER_GEOMETRY:
      return chip_class == GFX9 ? R_00B330_SPI_SHADER_USER_DATA_ES_0
                                : R_00B230_SPI_SHADER_USER_DATA_GS_0;
   case MESA_SHADER_COMPUTE:
      return R_00B900_COMPUTE_USER_DATA_0;
   case MESA_SHADER_TESS_CTRL:
      return chip_class == GFX9 ? R_00B430_SPI_SHADER_USER_DATA_LS_0
                                : R_00B430_SPI_SHADER_USER_DATA_HS_0;
   case MESA_SHADER_TESS_EVAL:
      if (has_gs) {
         return chip_class >= GFX10 ? R_00B230_SPI_SHADER_USER_DATA_GS_0
                                    : R_00B330_SPI_SHADER_USER_DATA_ES_0;
      } else if (has_ngg) {
         return R_00B230_SPI_SHADER_USER_DATA_GS_0;
      } else {
         return R_00B130_SPI_SHADER_USER_DATA_VS_0;
      }
   default:
      unreachable("unknown shader");
   }
}

struct radv_bin_size_entry {
   unsigned bpp;
   VkExtent2D extent;
};

static VkExtent2D
radv_gfx9_compute_bin_size(const struct radv_pipeline *pipeline,
                           const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   static const struct radv_bin_size_entry color_size_table[][3][9] = {
      {
         /* One RB / SE */
         {
            /* One shader engine */
            {0, {128, 128}},
            {1, {64, 128}},
            {2, {32, 128}},
            {3, {16, 128}},
            {17, {0, 0}},
            {UINT_MAX, {0, 0}},
         },
         {
            /* Two shader engines */
            {0, {128, 128}},
            {2, {64, 128}},
            {3, {32, 128}},
            {5, {16, 128}},
            {17, {0, 0}},
            {UINT_MAX, {0, 0}},
         },
         {
            /* Four shader engines */
            {0, {128, 128}},
            {3, {64, 128}},
            {5, {16, 128}},
            {17, {0, 0}},
            {UINT_MAX, {0, 0}},
         },
      },
      {
         /* Two RB / SE */
         {
            /* One shader engine */
            {0, {128, 128}},
            {2, {64, 128}},
            {3, {32, 128}},
            {5, {16, 128}},
            {33, {0, 0}},
            {UINT_MAX, {0, 0}},
         },
         {
            /* Two shader engines */
            {0, {128, 128}},
            {3, {64, 128}},
            {5, {32, 128}},
            {9, {16, 128}},
            {33, {0, 0}},
            {UINT_MAX, {0, 0}},
         },
         {
            /* Four shader engines */
            {0, {256, 256}},
            {2, {128, 256}},
            {3, {128, 128}},
            {5, {64, 128}},
            {9, {16, 128}},
            {33, {0, 0}},
            {UINT_MAX, {0, 0}},
         },
      },
      {
         /* Four RB / SE */
         {
            /* One shader engine */
            {0, {128, 256}},
            {2, {128, 128}},
            {3, {64, 128}},
            {5, {32, 128}},
            {9, {16, 128}},
            {33, {0, 0}},
            {UINT_MAX, {0, 0}},
         },
         {
            /* Two shader engines */
            {0, {256, 256}},
            {2, {128, 256}},
            {3, {128, 128}},
            {5, {64, 128}},
            {9, {32, 128}},
            {17, {16, 128}},
            {33, {0, 0}},
            {UINT_MAX, {0, 0}},
         },
         {
            /* Four shader engines */
            {0, {256, 512}},
            {2, {256, 256}},
            {3, {128, 256}},
            {5, {128, 128}},
            {9, {64, 128}},
            {17, {16, 128}},
            {33, {0, 0}},
            {UINT_MAX, {0, 0}},
         },
      },
   };
   static const struct radv_bin_size_entry ds_size_table[][3][9] = {
      {
         // One RB / SE
         {
            // One shader engine
            {0, {128, 256}},
            {2, {128, 128}},
            {4, {64, 128}},
            {7, {32, 128}},
            {13, {16, 128}},
            {49, {0, 0}},
            {UINT_MAX, {0, 0}},
         },
         {
            // Two shader engines
            {0, {256, 256}},
            {2, {128, 256}},
            {4, {128, 128}},
            {7, {64, 128}},
            {13, {32, 128}},
            {25, {16, 128}},
            {49, {0, 0}},
            {UINT_MAX, {0, 0}},
         },
         {
            // Four shader engines
            {0, {256, 512}},
            {2, {256, 256}},
            {4, {128, 256}},
            {7, {128, 128}},
            {13, {64, 128}},
            {25, {16, 128}},
            {49, {0, 0}},
            {UINT_MAX, {0, 0}},
         },
      },
      {
         // Two RB / SE
         {
            // One shader engine
            {0, {256, 256}},
            {2, {128, 256}},
            {4, {128, 128}},
            {7, {64, 128}},
            {13, {32, 128}},
            {25, {16, 128}},
            {97, {0, 0}},
            {UINT_MAX, {0, 0}},
         },
         {
            // Two shader engines
            {0, {256, 512}},
            {2, {256, 256}},
            {4, {128, 256}},
            {7, {128, 128}},
            {13, {64, 128}},
            {25, {32, 128}},
            {49, {16, 128}},
            {97, {0, 0}},
            {UINT_MAX, {0, 0}},
         },
         {
            // Four shader engines
            {0, {512, 512}},
            {2, {256, 512}},
            {4, {256, 256}},
            {7, {128, 256}},
            {13, {128, 128}},
            {25, {64, 128}},
            {49, {16, 128}},
            {97, {0, 0}},
            {UINT_MAX, {0, 0}},
         },
      },
      {
         // Four RB / SE
         {
            // One shader engine
            {0, {256, 512}},
            {2, {256, 256}},
            {4, {128, 256}},
            {7, {128, 128}},
            {13, {64, 128}},
            {25, {32, 128}},
            {49, {16, 128}},
            {UINT_MAX, {0, 0}},
         },
         {
            // Two shader engines
            {0, {512, 512}},
            {2, {256, 512}},
            {4, {256, 256}},
            {7, {128, 256}},
            {13, {128, 128}},
            {25, {64, 128}},
            {49, {32, 128}},
            {97, {16, 128}},
            {UINT_MAX, {0, 0}},
         },
         {
            // Four shader engines
            {0, {512, 512}},
            {4, {256, 512}},
            {7, {256, 256}},
            {13, {128, 256}},
            {25, {128, 128}},
            {49, {64, 128}},
            {97, {16, 128}},
            {UINT_MAX, {0, 0}},
         },
      },
   };

   RADV_FROM_HANDLE(radv_render_pass, pass, pCreateInfo->renderPass);
   struct radv_subpass *subpass = pass->subpasses + pCreateInfo->subpass;
   VkExtent2D extent = {512, 512};

   unsigned log_num_rb_per_se =
      util_logbase2_ceil(pipeline->device->physical_device->rad_info.max_render_backends /
                         pipeline->device->physical_device->rad_info.max_se);
   unsigned log_num_se = util_logbase2_ceil(pipeline->device->physical_device->rad_info.max_se);

   unsigned total_samples = 1u << G_028BE0_MSAA_NUM_SAMPLES(pipeline->graphics.ms.pa_sc_aa_config);
   unsigned ps_iter_samples = 1u << G_028804_PS_ITER_SAMPLES(pipeline->graphics.ms.db_eqaa);
   unsigned effective_samples = total_samples;
   unsigned color_bytes_per_pixel = 0;

   const VkPipelineColorBlendStateCreateInfo *vkblend =
      radv_pipeline_get_color_blend_state(pCreateInfo);
   if (vkblend) {
      for (unsigned i = 0; i < subpass->color_count; i++) {
         if (!vkblend->pAttachments[i].colorWriteMask)
            continue;

         if (subpass->color_attachments[i].attachment == VK_ATTACHMENT_UNUSED)
            continue;

         VkFormat format = pass->attachments[subpass->color_attachments[i].attachment].format;
         color_bytes_per_pixel += vk_format_get_blocksize(format);
      }

      /* MSAA images typically don't use all samples all the time. */
      if (effective_samples >= 2 && ps_iter_samples <= 1)
         effective_samples = 2;
      color_bytes_per_pixel *= effective_samples;
   }

   const struct radv_bin_size_entry *color_entry = color_size_table[log_num_rb_per_se][log_num_se];
   while (color_entry[1].bpp <= color_bytes_per_pixel)
      ++color_entry;

   extent = color_entry->extent;

   if (subpass->depth_stencil_attachment) {
      struct radv_render_pass_attachment *attachment =
         pass->attachments + subpass->depth_stencil_attachment->attachment;

      /* Coefficients taken from AMDVLK */
      unsigned depth_coeff = vk_format_has_depth(attachment->format) ? 5 : 0;
      unsigned stencil_coeff = vk_format_has_stencil(attachment->format) ? 1 : 0;
      unsigned ds_bytes_per_pixel = 4 * (depth_coeff + stencil_coeff) * total_samples;

      const struct radv_bin_size_entry *ds_entry = ds_size_table[log_num_rb_per_se][log_num_se];
      while (ds_entry[1].bpp <= ds_bytes_per_pixel)
         ++ds_entry;

      if (ds_entry->extent.width * ds_entry->extent.height < extent.width * extent.height)
         extent = ds_entry->extent;
   }

   return extent;
}

static VkExtent2D
radv_gfx10_compute_bin_size(const struct radv_pipeline *pipeline,
                            const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   RADV_FROM_HANDLE(radv_render_pass, pass, pCreateInfo->renderPass);
   struct radv_subpass *subpass = pass->subpasses + pCreateInfo->subpass;
   VkExtent2D extent = {512, 512};

   const unsigned db_tag_size = 64;
   const unsigned db_tag_count = 312;
   const unsigned color_tag_size = 1024;
   const unsigned color_tag_count = 31;
   const unsigned fmask_tag_size = 256;
   const unsigned fmask_tag_count = 44;

   const unsigned rb_count = pipeline->device->physical_device->rad_info.max_render_backends;
   const unsigned pipe_count =
      MAX2(rb_count, pipeline->device->physical_device->rad_info.num_tcc_blocks);

   const unsigned db_tag_part = (db_tag_count * rb_count / pipe_count) * db_tag_size * pipe_count;
   const unsigned color_tag_part =
      (color_tag_count * rb_count / pipe_count) * color_tag_size * pipe_count;
   const unsigned fmask_tag_part =
      (fmask_tag_count * rb_count / pipe_count) * fmask_tag_size * pipe_count;

   const unsigned total_samples =
      1u << G_028BE0_MSAA_NUM_SAMPLES(pipeline->graphics.ms.pa_sc_aa_config);
   const unsigned samples_log = util_logbase2_ceil(total_samples);

   unsigned color_bytes_per_pixel = 0;
   unsigned fmask_bytes_per_pixel = 0;

   const VkPipelineColorBlendStateCreateInfo *vkblend =
      radv_pipeline_get_color_blend_state(pCreateInfo);
   if (vkblend) {
      for (unsigned i = 0; i < subpass->color_count; i++) {
         if (!vkblend->pAttachments[i].colorWriteMask)
            continue;

         if (subpass->color_attachments[i].attachment == VK_ATTACHMENT_UNUSED)
            continue;

         VkFormat format = pass->attachments[subpass->color_attachments[i].attachment].format;
         color_bytes_per_pixel += vk_format_get_blocksize(format);

         if (total_samples > 1) {
            assert(samples_log <= 3);
            const unsigned fmask_array[] = {0, 1, 1, 4};
            fmask_bytes_per_pixel += fmask_array[samples_log];
         }
      }

      color_bytes_per_pixel *= total_samples;
   }
   color_bytes_per_pixel = MAX2(color_bytes_per_pixel, 1);

   const unsigned color_pixel_count_log = util_logbase2(color_tag_part / color_bytes_per_pixel);
   extent.width = 1ull << ((color_pixel_count_log + 1) / 2);
   extent.height = 1ull << (color_pixel_count_log / 2);

   if (fmask_bytes_per_pixel) {
      const unsigned fmask_pixel_count_log = util_logbase2(fmask_tag_part / fmask_bytes_per_pixel);

      const VkExtent2D fmask_extent =
         (VkExtent2D){.width = 1ull << ((fmask_pixel_count_log + 1) / 2),
                      .height = 1ull << (color_pixel_count_log / 2)};

      if (fmask_extent.width * fmask_extent.height < extent.width * extent.height)
         extent = fmask_extent;
   }

   if (subpass->depth_stencil_attachment) {
      struct radv_render_pass_attachment *attachment =
         pass->attachments + subpass->depth_stencil_attachment->attachment;

      /* Coefficients taken from AMDVLK */
      unsigned depth_coeff = vk_format_has_depth(attachment->format) ? 5 : 0;
      unsigned stencil_coeff = vk_format_has_stencil(attachment->format) ? 1 : 0;
      unsigned db_bytes_per_pixel = (depth_coeff + stencil_coeff) * total_samples;

      const unsigned db_pixel_count_log = util_logbase2(db_tag_part / db_bytes_per_pixel);

      const VkExtent2D db_extent = (VkExtent2D){.width = 1ull << ((db_pixel_count_log + 1) / 2),
                                                .height = 1ull << (color_pixel_count_log / 2)};

      if (db_extent.width * db_extent.height < extent.width * extent.height)
         extent = db_extent;
   }

   extent.width = MAX2(extent.width, 128);
   extent.height = MAX2(extent.width, 64);

   return extent;
}

static void
radv_pipeline_init_disabled_binning_state(struct radv_pipeline *pipeline,
                                          const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   uint32_t pa_sc_binner_cntl_0 = S_028C44_BINNING_MODE(V_028C44_DISABLE_BINNING_USE_LEGACY_SC) |
                                  S_028C44_DISABLE_START_OF_PRIM(1);
   uint32_t db_dfsm_control = S_028060_PUNCHOUT_MODE(V_028060_FORCE_OFF);

   if (pipeline->device->physical_device->rad_info.chip_class >= GFX10) {
      RADV_FROM_HANDLE(radv_render_pass, pass, pCreateInfo->renderPass);
      struct radv_subpass *subpass = pass->subpasses + pCreateInfo->subpass;
      const VkPipelineColorBlendStateCreateInfo *vkblend =
         radv_pipeline_get_color_blend_state(pCreateInfo);
      unsigned min_bytes_per_pixel = 0;

      if (vkblend) {
         for (unsigned i = 0; i < subpass->color_count; i++) {
            if (!vkblend->pAttachments[i].colorWriteMask)
               continue;

            if (subpass->color_attachments[i].attachment == VK_ATTACHMENT_UNUSED)
               continue;

            VkFormat format = pass->attachments[subpass->color_attachments[i].attachment].format;
            unsigned bytes = vk_format_get_blocksize(format);
            if (!min_bytes_per_pixel || bytes < min_bytes_per_pixel)
               min_bytes_per_pixel = bytes;
         }
      }

      pa_sc_binner_cntl_0 =
         S_028C44_BINNING_MODE(V_028C44_DISABLE_BINNING_USE_NEW_SC) | S_028C44_BIN_SIZE_X(0) |
         S_028C44_BIN_SIZE_Y(0) | S_028C44_BIN_SIZE_X_EXTEND(2) |       /* 128 */
         S_028C44_BIN_SIZE_Y_EXTEND(min_bytes_per_pixel <= 4 ? 2 : 1) | /* 128 or 64 */
         S_028C44_DISABLE_START_OF_PRIM(1);
   }

   pipeline->graphics.binning.pa_sc_binner_cntl_0 = pa_sc_binner_cntl_0;
   pipeline->graphics.binning.db_dfsm_control = db_dfsm_control;
}

struct radv_binning_settings
radv_get_binning_settings(const struct radv_physical_device *pdev)
{
   struct radv_binning_settings settings;
   if (pdev->rad_info.has_dedicated_vram) {
      if (pdev->rad_info.max_render_backends > 4) {
         settings.context_states_per_bin = 1;
         settings.persistent_states_per_bin = 1;
      } else {
         settings.context_states_per_bin = 3;
         settings.persistent_states_per_bin = 8;
      }
      settings.fpovs_per_batch = 63;
   } else {
      /* The context states are affected by the scissor bug. */
      settings.context_states_per_bin = 6;
      /* 32 causes hangs for RAVEN. */
      settings.persistent_states_per_bin = 16;
      settings.fpovs_per_batch = 63;
   }

   if (pdev->rad_info.has_gfx9_scissor_bug)
      settings.context_states_per_bin = 1;

   return settings;
}

static void
radv_pipeline_init_binning_state(struct radv_pipeline *pipeline,
                                 const VkGraphicsPipelineCreateInfo *pCreateInfo,
                                 const struct radv_blend_state *blend)
{
   if (pipeline->device->physical_device->rad_info.chip_class < GFX9)
      return;

   VkExtent2D bin_size;
   if (pipeline->device->physical_device->rad_info.chip_class >= GFX10) {
      bin_size = radv_gfx10_compute_bin_size(pipeline, pCreateInfo);
   } else if (pipeline->device->physical_device->rad_info.chip_class == GFX9) {
      bin_size = radv_gfx9_compute_bin_size(pipeline, pCreateInfo);
   } else
      unreachable("Unhandled generation for binning bin size calculation");

   if (pipeline->device->pbb_allowed && bin_size.width && bin_size.height) {
      struct radv_binning_settings settings =
         radv_get_binning_settings(pipeline->device->physical_device);

      bool disable_start_of_prim = true;
      uint32_t db_dfsm_control = S_028060_PUNCHOUT_MODE(V_028060_FORCE_OFF);

      const struct radv_shader_variant *ps = pipeline->shaders[MESA_SHADER_FRAGMENT];

      if (pipeline->device->dfsm_allowed && ps && !ps->info.ps.can_discard &&
          !ps->info.ps.writes_memory && blend->cb_target_enabled_4bit) {
         db_dfsm_control = S_028060_PUNCHOUT_MODE(V_028060_AUTO);
         disable_start_of_prim = (blend->blend_enable_4bit & blend->cb_target_enabled_4bit) != 0;
      }

      const uint32_t pa_sc_binner_cntl_0 =
         S_028C44_BINNING_MODE(V_028C44_BINNING_ALLOWED) |
         S_028C44_BIN_SIZE_X(bin_size.width == 16) | S_028C44_BIN_SIZE_Y(bin_size.height == 16) |
         S_028C44_BIN_SIZE_X_EXTEND(util_logbase2(MAX2(bin_size.width, 32)) - 5) |
         S_028C44_BIN_SIZE_Y_EXTEND(util_logbase2(MAX2(bin_size.height, 32)) - 5) |
         S_028C44_CONTEXT_STATES_PER_BIN(settings.context_states_per_bin - 1) |
         S_028C44_PERSISTENT_STATES_PER_BIN(settings.persistent_states_per_bin - 1) |
         S_028C44_DISABLE_START_OF_PRIM(disable_start_of_prim) |
         S_028C44_FPOVS_PER_BATCH(settings.fpovs_per_batch) | S_028C44_OPTIMAL_BIN_SELECTION(1);

      pipeline->graphics.binning.pa_sc_binner_cntl_0 = pa_sc_binner_cntl_0;
      pipeline->graphics.binning.db_dfsm_control = db_dfsm_control;
   } else
      radv_pipeline_init_disabled_binning_state(pipeline, pCreateInfo);
}

static void
radv_pipeline_generate_depth_stencil_state(struct radeon_cmdbuf *ctx_cs,
                                           const struct radv_pipeline *pipeline,
                                           const VkGraphicsPipelineCreateInfo *pCreateInfo,
                                           const struct radv_graphics_pipeline_create_info *extra)
{
   const VkPipelineDepthStencilStateCreateInfo *vkds =
      radv_pipeline_get_depth_stencil_state(pCreateInfo);
   RADV_FROM_HANDLE(radv_render_pass, pass, pCreateInfo->renderPass);
   struct radv_subpass *subpass = pass->subpasses + pCreateInfo->subpass;
   struct radv_shader_variant *ps = pipeline->shaders[MESA_SHADER_FRAGMENT];
   struct radv_render_pass_attachment *attachment = NULL;
   uint32_t db_render_control = 0, db_render_override2 = 0;
   uint32_t db_render_override = 0;

   if (subpass->depth_stencil_attachment)
      attachment = pass->attachments + subpass->depth_stencil_attachment->attachment;

   bool has_depth_attachment = attachment && vk_format_has_depth(attachment->format);

   if (vkds && has_depth_attachment) {
      /* from amdvlk: For 4xAA and 8xAA need to decompress on flush for better performance */
      db_render_override2 |= S_028010_DECOMPRESS_Z_ON_FLUSH(attachment->samples > 2);

      if (pipeline->device->physical_device->rad_info.chip_class >= GFX10_3)
         db_render_override2 |= S_028010_CENTROID_COMPUTATION_MODE(1);
   }

   if (attachment && extra) {
      db_render_control |= S_028000_DEPTH_CLEAR_ENABLE(extra->db_depth_clear);
      db_render_control |= S_028000_STENCIL_CLEAR_ENABLE(extra->db_stencil_clear);

      db_render_control |= S_028000_RESUMMARIZE_ENABLE(extra->resummarize_enable);
      db_render_control |= S_028000_DEPTH_COMPRESS_DISABLE(extra->depth_compress_disable);
      db_render_control |= S_028000_STENCIL_COMPRESS_DISABLE(extra->stencil_compress_disable);
      db_render_override2 |=
         S_028010_DISABLE_ZMASK_EXPCLEAR_OPTIMIZATION(extra->db_depth_disable_expclear);
      db_render_override2 |=
         S_028010_DISABLE_SMEM_EXPCLEAR_OPTIMIZATION(extra->db_stencil_disable_expclear);
   }

   db_render_override |= S_02800C_FORCE_HIS_ENABLE0(V_02800C_FORCE_DISABLE) |
                         S_02800C_FORCE_HIS_ENABLE1(V_02800C_FORCE_DISABLE);

   if (!pCreateInfo->pRasterizationState->depthClampEnable && ps->info.ps.writes_z) {
      /* From VK_EXT_depth_range_unrestricted spec:
       *
       * "The behavior described in Primitive Clipping still applies.
       *  If depth clamping is disabled the depth values are still
       *  clipped to 0 ≤ zc ≤ wc before the viewport transform. If
       *  depth clamping is enabled the above equation is ignored and
       *  the depth values are instead clamped to the VkViewport
       *  minDepth and maxDepth values, which in the case of this
       *  extension can be outside of the 0.0 to 1.0 range."
       */
      db_render_override |= S_02800C_DISABLE_VIEWPORT_CLAMP(1);
   }

   radeon_set_context_reg(ctx_cs, R_028000_DB_RENDER_CONTROL, db_render_control);
   radeon_set_context_reg(ctx_cs, R_02800C_DB_RENDER_OVERRIDE, db_render_override);
   radeon_set_context_reg(ctx_cs, R_028010_DB_RENDER_OVERRIDE2, db_render_override2);
}

static void
radv_pipeline_generate_blend_state(struct radeon_cmdbuf *ctx_cs,
                                   const struct radv_pipeline *pipeline,
                                   const struct radv_blend_state *blend)
{
   radeon_set_context_reg_seq(ctx_cs, R_028780_CB_BLEND0_CONTROL, 8);
   radeon_emit_array(ctx_cs, blend->cb_blend_control, 8);
   radeon_set_context_reg(ctx_cs, R_028808_CB_COLOR_CONTROL, blend->cb_color_control);
   radeon_set_context_reg(ctx_cs, R_028B70_DB_ALPHA_TO_MASK, blend->db_alpha_to_mask);

   if (pipeline->device->physical_device->rad_info.has_rbplus) {

      radeon_set_context_reg_seq(ctx_cs, R_028760_SX_MRT0_BLEND_OPT, 8);
      radeon_emit_array(ctx_cs, blend->sx_mrt_blend_opt, 8);
   }

   radeon_set_context_reg(ctx_cs, R_028714_SPI_SHADER_COL_FORMAT, blend->spi_shader_col_format);

   radeon_set_context_reg(ctx_cs, R_028238_CB_TARGET_MASK, blend->cb_target_mask);
   radeon_set_context_reg(ctx_cs, R_02823C_CB_SHADER_MASK, blend->cb_shader_mask);
}

static void
radv_pipeline_generate_raster_state(struct radeon_cmdbuf *ctx_cs,
                                    const struct radv_pipeline *pipeline,
                                    const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   const VkPipelineRasterizationStateCreateInfo *vkraster = pCreateInfo->pRasterizationState;
   const VkConservativeRasterizationModeEXT mode = radv_get_conservative_raster_mode(vkraster);
   uint32_t pa_sc_conservative_rast = S_028C4C_NULL_SQUAD_AA_MASK_ENABLE(1);

   radeon_set_context_reg(ctx_cs, R_028BDC_PA_SC_LINE_CNTL, S_028BDC_DX10_DIAMOND_TEST_ENA(1));

   /* Conservative rasterization. */
   if (mode != VK_CONSERVATIVE_RASTERIZATION_MODE_DISABLED_EXT) {
      pa_sc_conservative_rast = S_028C4C_PREZ_AA_MASK_ENABLE(1) | S_028C4C_POSTZ_AA_MASK_ENABLE(1) |
                                S_028C4C_CENTROID_SAMPLE_OVERRIDE(1);

      if (mode == VK_CONSERVATIVE_RASTERIZATION_MODE_OVERESTIMATE_EXT) {
         pa_sc_conservative_rast |=
            S_028C4C_OVER_RAST_ENABLE(1) | S_028C4C_OVER_RAST_SAMPLE_SELECT(0) |
            S_028C4C_UNDER_RAST_ENABLE(0) | S_028C4C_UNDER_RAST_SAMPLE_SELECT(1) |
            S_028C4C_PBB_UNCERTAINTY_REGION_ENABLE(1);
      } else {
         assert(mode == VK_CONSERVATIVE_RASTERIZATION_MODE_UNDERESTIMATE_EXT);
         pa_sc_conservative_rast |=
            S_028C4C_OVER_RAST_ENABLE(0) | S_028C4C_OVER_RAST_SAMPLE_SELECT(1) |
            S_028C4C_UNDER_RAST_ENABLE(1) | S_028C4C_UNDER_RAST_SAMPLE_SELECT(0) |
            S_028C4C_PBB_UNCERTAINTY_REGION_ENABLE(0);
      }
   }

   radeon_set_context_reg(ctx_cs, R_028C4C_PA_SC_CONSERVATIVE_RASTERIZATION_CNTL,
                          pa_sc_conservative_rast);
}

static void
radv_pipeline_generate_multisample_state(struct radeon_cmdbuf *ctx_cs,
                                         const struct radv_pipeline *pipeline)
{
   const struct radv_multisample_state *ms = &pipeline->graphics.ms;

   radeon_set_context_reg_seq(ctx_cs, R_028C38_PA_SC_AA_MASK_X0Y0_X1Y0, 2);
   radeon_emit(ctx_cs, ms->pa_sc_aa_mask[0]);
   radeon_emit(ctx_cs, ms->pa_sc_aa_mask[1]);

   radeon_set_context_reg(ctx_cs, R_028804_DB_EQAA, ms->db_eqaa);
   radeon_set_context_reg(ctx_cs, R_028A48_PA_SC_MODE_CNTL_0, ms->pa_sc_mode_cntl_0);
   radeon_set_context_reg(ctx_cs, R_028A4C_PA_SC_MODE_CNTL_1, ms->pa_sc_mode_cntl_1);
   radeon_set_context_reg(ctx_cs, R_028BE0_PA_SC_AA_CONFIG, ms->pa_sc_aa_config);

   /* The exclusion bits can be set to improve rasterization efficiency
    * if no sample lies on the pixel boundary (-8 sample offset). It's
    * currently always TRUE because the driver doesn't support 16 samples.
    */
   bool exclusion = pipeline->device->physical_device->rad_info.chip_class >= GFX7;
   radeon_set_context_reg(
      ctx_cs, R_02882C_PA_SU_PRIM_FILTER_CNTL,
      S_02882C_XMAX_RIGHT_EXCLUSION(exclusion) | S_02882C_YMAX_BOTTOM_EXCLUSION(exclusion));

   /* GFX9: Flush DFSM when the AA mode changes. */
   if (pipeline->device->dfsm_allowed) {
      radeon_emit(ctx_cs, PKT3(PKT3_EVENT_WRITE, 0, 0));
      radeon_emit(ctx_cs, EVENT_TYPE(V_028A90_FLUSH_DFSM) | EVENT_INDEX(0));
   }
}

static void
radv_pipeline_generate_vgt_gs_mode(struct radeon_cmdbuf *ctx_cs,
                                   const struct radv_pipeline *pipeline)
{
   const struct radv_vs_output_info *outinfo = get_vs_output_info(pipeline);
   const struct radv_shader_variant *vs = pipeline->shaders[MESA_SHADER_TESS_EVAL]
                                             ? pipeline->shaders[MESA_SHADER_TESS_EVAL]
                                             : pipeline->shaders[MESA_SHADER_VERTEX];
   unsigned vgt_primitiveid_en = 0;
   uint32_t vgt_gs_mode = 0;

   if (radv_pipeline_has_ngg(pipeline))
      return;

   if (radv_pipeline_has_gs(pipeline)) {
      const struct radv_shader_variant *gs = pipeline->shaders[MESA_SHADER_GEOMETRY];

      vgt_gs_mode = ac_vgt_gs_mode(gs->info.gs.vertices_out,
                                   pipeline->device->physical_device->rad_info.chip_class);
   } else if (outinfo->export_prim_id || vs->info.uses_prim_id) {
      vgt_gs_mode = S_028A40_MODE(V_028A40_GS_SCENARIO_A);
      vgt_primitiveid_en |= S_028A84_PRIMITIVEID_EN(1);
   }

   radeon_set_context_reg(ctx_cs, R_028A84_VGT_PRIMITIVEID_EN, vgt_primitiveid_en);
   radeon_set_context_reg(ctx_cs, R_028A40_VGT_GS_MODE, vgt_gs_mode);
}

static void
radv_pipeline_generate_hw_vs(struct radeon_cmdbuf *ctx_cs, struct radeon_cmdbuf *cs,
                             const struct radv_pipeline *pipeline,
                             const struct radv_shader_variant *shader)
{
   uint64_t va = radv_buffer_get_va(shader->bo) + shader->bo_offset;

   radeon_set_sh_reg_seq(cs, R_00B120_SPI_SHADER_PGM_LO_VS, 4);
   radeon_emit(cs, va >> 8);
   radeon_emit(cs, S_00B124_MEM_BASE(va >> 40));
   radeon_emit(cs, shader->config.rsrc1);
   radeon_emit(cs, shader->config.rsrc2);

   const struct radv_vs_output_info *outinfo = get_vs_output_info(pipeline);
   unsigned clip_dist_mask, cull_dist_mask, total_mask;
   clip_dist_mask = outinfo->clip_dist_mask;
   cull_dist_mask = outinfo->cull_dist_mask;
   total_mask = clip_dist_mask | cull_dist_mask;

   bool writes_primitive_shading_rate =
      outinfo->writes_primitive_shading_rate || pipeline->device->force_vrs != RADV_FORCE_VRS_NONE;
   bool misc_vec_ena = outinfo->writes_pointsize || outinfo->writes_layer ||
                       outinfo->writes_viewport_index || writes_primitive_shading_rate;
   unsigned spi_vs_out_config, nparams;

   /* VS is required to export at least one param. */
   nparams = MAX2(outinfo->param_exports, 1);
   spi_vs_out_config = S_0286C4_VS_EXPORT_COUNT(nparams - 1);

   if (pipeline->device->physical_device->rad_info.chip_class >= GFX10) {
      spi_vs_out_config |= S_0286C4_NO_PC_EXPORT(outinfo->param_exports == 0);
   }

   radeon_set_context_reg(ctx_cs, R_0286C4_SPI_VS_OUT_CONFIG, spi_vs_out_config);

   radeon_set_context_reg(
      ctx_cs, R_02870C_SPI_SHADER_POS_FORMAT,
      S_02870C_POS0_EXPORT_FORMAT(V_02870C_SPI_SHADER_4COMP) |
         S_02870C_POS1_EXPORT_FORMAT(outinfo->pos_exports > 1 ? V_02870C_SPI_SHADER_4COMP
                                                              : V_02870C_SPI_SHADER_NONE) |
         S_02870C_POS2_EXPORT_FORMAT(outinfo->pos_exports > 2 ? V_02870C_SPI_SHADER_4COMP
                                                              : V_02870C_SPI_SHADER_NONE) |
         S_02870C_POS3_EXPORT_FORMAT(outinfo->pos_exports > 3 ? V_02870C_SPI_SHADER_4COMP
                                                              : V_02870C_SPI_SHADER_NONE));

   radeon_set_context_reg(ctx_cs, R_02881C_PA_CL_VS_OUT_CNTL,
                          S_02881C_USE_VTX_POINT_SIZE(outinfo->writes_pointsize) |
                             S_02881C_USE_VTX_RENDER_TARGET_INDX(outinfo->writes_layer) |
                             S_02881C_USE_VTX_VIEWPORT_INDX(outinfo->writes_viewport_index) |
                             S_02881C_USE_VTX_VRS_RATE(writes_primitive_shading_rate) |
                             S_02881C_VS_OUT_MISC_VEC_ENA(misc_vec_ena) |
                             S_02881C_VS_OUT_MISC_SIDE_BUS_ENA(misc_vec_ena) |
                             S_02881C_VS_OUT_CCDIST0_VEC_ENA((total_mask & 0x0f) != 0) |
                             S_02881C_VS_OUT_CCDIST1_VEC_ENA((total_mask & 0xf0) != 0) |
                             cull_dist_mask << 8 | clip_dist_mask);

   if (pipeline->device->physical_device->rad_info.chip_class <= GFX8)
      radeon_set_context_reg(ctx_cs, R_028AB4_VGT_REUSE_OFF, outinfo->writes_viewport_index);
}

static void
radv_pipeline_generate_hw_es(struct radeon_cmdbuf *cs, const struct radv_pipeline *pipeline,
                             const struct radv_shader_variant *shader)
{
   uint64_t va = radv_buffer_get_va(shader->bo) + shader->bo_offset;

   radeon_set_sh_reg_seq(cs, R_00B320_SPI_SHADER_PGM_LO_ES, 4);
   radeon_emit(cs, va >> 8);
   radeon_emit(cs, S_00B324_MEM_BASE(va >> 40));
   radeon_emit(cs, shader->config.rsrc1);
   radeon_emit(cs, shader->config.rsrc2);
}

static void
radv_pipeline_generate_hw_ls(struct radeon_cmdbuf *cs, const struct radv_pipeline *pipeline,
                             const struct radv_shader_variant *shader)
{
   unsigned num_lds_blocks = pipeline->shaders[MESA_SHADER_TESS_CTRL]->info.tcs.num_lds_blocks;
   uint64_t va = radv_buffer_get_va(shader->bo) + shader->bo_offset;
   uint32_t rsrc2 = shader->config.rsrc2;

   radeon_set_sh_reg_seq(cs, R_00B520_SPI_SHADER_PGM_LO_LS, 2);
   radeon_emit(cs, va >> 8);
   radeon_emit(cs, S_00B524_MEM_BASE(va >> 40));

   rsrc2 |= S_00B52C_LDS_SIZE(num_lds_blocks);
   if (pipeline->device->physical_device->rad_info.chip_class == GFX7 &&
       pipeline->device->physical_device->rad_info.family != CHIP_HAWAII)
      radeon_set_sh_reg(cs, R_00B52C_SPI_SHADER_PGM_RSRC2_LS, rsrc2);

   radeon_set_sh_reg_seq(cs, R_00B528_SPI_SHADER_PGM_RSRC1_LS, 2);
   radeon_emit(cs, shader->config.rsrc1);
   radeon_emit(cs, rsrc2);
}

static void
radv_pipeline_generate_hw_ngg(struct radeon_cmdbuf *ctx_cs, struct radeon_cmdbuf *cs,
                              const struct radv_pipeline *pipeline,
                              const struct radv_shader_variant *shader)
{
   uint64_t va = radv_buffer_get_va(shader->bo) + shader->bo_offset;
   gl_shader_stage es_type =
      radv_pipeline_has_tess(pipeline) ? MESA_SHADER_TESS_EVAL : MESA_SHADER_VERTEX;
   struct radv_shader_variant *es = es_type == MESA_SHADER_TESS_EVAL
                                       ? pipeline->shaders[MESA_SHADER_TESS_EVAL]
                                       : pipeline->shaders[MESA_SHADER_VERTEX];
   const struct gfx10_ngg_info *ngg_state = &shader->info.ngg_info;

   radeon_set_sh_reg_seq(cs, R_00B320_SPI_SHADER_PGM_LO_ES, 2);
   radeon_emit(cs, va >> 8);
   radeon_emit(cs, S_00B324_MEM_BASE(va >> 40));
   radeon_set_sh_reg_seq(cs, R_00B228_SPI_SHADER_PGM_RSRC1_GS, 2);
   radeon_emit(cs, shader->config.rsrc1);
   radeon_emit(cs, shader->config.rsrc2);

   const struct radv_vs_output_info *outinfo = get_vs_output_info(pipeline);
   unsigned clip_dist_mask, cull_dist_mask, total_mask;
   clip_dist_mask = outinfo->clip_dist_mask;
   cull_dist_mask = outinfo->cull_dist_mask;
   total_mask = clip_dist_mask | cull_dist_mask;

   bool writes_primitive_shading_rate =
      outinfo->writes_primitive_shading_rate || pipeline->device->force_vrs != RADV_FORCE_VRS_NONE;
   bool misc_vec_ena = outinfo->writes_pointsize || outinfo->writes_layer ||
                       outinfo->writes_viewport_index || writes_primitive_shading_rate;
   bool es_enable_prim_id = outinfo->export_prim_id || (es && es->info.uses_prim_id);
   bool break_wave_at_eoi = false;
   unsigned ge_cntl;
   unsigned nparams;

   if (es_type == MESA_SHADER_TESS_EVAL) {
      struct radv_shader_variant *gs = pipeline->shaders[MESA_SHADER_GEOMETRY];

      if (es_enable_prim_id || (gs && gs->info.uses_prim_id))
         break_wave_at_eoi = true;
   }

   nparams = MAX2(outinfo->param_exports, 1);
   radeon_set_context_reg(
      ctx_cs, R_0286C4_SPI_VS_OUT_CONFIG,
      S_0286C4_VS_EXPORT_COUNT(nparams - 1) | S_0286C4_NO_PC_EXPORT(outinfo->param_exports == 0));

   radeon_set_context_reg(ctx_cs, R_028708_SPI_SHADER_IDX_FORMAT,
                          S_028708_IDX0_EXPORT_FORMAT(V_028708_SPI_SHADER_1COMP));
   radeon_set_context_reg(
      ctx_cs, R_02870C_SPI_SHADER_POS_FORMAT,
      S_02870C_POS0_EXPORT_FORMAT(V_02870C_SPI_SHADER_4COMP) |
         S_02870C_POS1_EXPORT_FORMAT(outinfo->pos_exports > 1 ? V_02870C_SPI_SHADER_4COMP
                                                              : V_02870C_SPI_SHADER_NONE) |
         S_02870C_POS2_EXPORT_FORMAT(outinfo->pos_exports > 2 ? V_02870C_SPI_SHADER_4COMP
                                                              : V_02870C_SPI_SHADER_NONE) |
         S_02870C_POS3_EXPORT_FORMAT(outinfo->pos_exports > 3 ? V_02870C_SPI_SHADER_4COMP
                                                              : V_02870C_SPI_SHADER_NONE));

   radeon_set_context_reg(ctx_cs, R_02881C_PA_CL_VS_OUT_CNTL,
                          S_02881C_USE_VTX_POINT_SIZE(outinfo->writes_pointsize) |
                             S_02881C_USE_VTX_RENDER_TARGET_INDX(outinfo->writes_layer) |
                             S_02881C_USE_VTX_VIEWPORT_INDX(outinfo->writes_viewport_index) |
                             S_02881C_USE_VTX_VRS_RATE(writes_primitive_shading_rate) |
                             S_02881C_VS_OUT_MISC_VEC_ENA(misc_vec_ena) |
                             S_02881C_VS_OUT_MISC_SIDE_BUS_ENA(misc_vec_ena) |
                             S_02881C_VS_OUT_CCDIST0_VEC_ENA((total_mask & 0x0f) != 0) |
                             S_02881C_VS_OUT_CCDIST1_VEC_ENA((total_mask & 0xf0) != 0) |
                             cull_dist_mask << 8 | clip_dist_mask);

   radeon_set_context_reg(ctx_cs, R_028A84_VGT_PRIMITIVEID_EN,
                          S_028A84_PRIMITIVEID_EN(es_enable_prim_id) |
                             S_028A84_NGG_DISABLE_PROVOK_REUSE(outinfo->export_prim_id));

   radeon_set_context_reg(ctx_cs, R_028AAC_VGT_ESGS_RING_ITEMSIZE,
                          ngg_state->vgt_esgs_ring_itemsize);

   /* NGG specific registers. */
   struct radv_shader_variant *gs = pipeline->shaders[MESA_SHADER_GEOMETRY];
   uint32_t gs_num_invocations = gs ? gs->info.gs.invocations : 1;

   radeon_set_context_reg(
      ctx_cs, R_028A44_VGT_GS_ONCHIP_CNTL,
      S_028A44_ES_VERTS_PER_SUBGRP(ngg_state->hw_max_esverts) |
         S_028A44_GS_PRIMS_PER_SUBGRP(ngg_state->max_gsprims) |
         S_028A44_GS_INST_PRIMS_IN_SUBGRP(ngg_state->max_gsprims * gs_num_invocations));
   radeon_set_context_reg(ctx_cs, R_0287FC_GE_MAX_OUTPUT_PER_SUBGROUP,
                          S_0287FC_MAX_VERTS_PER_SUBGROUP(ngg_state->max_out_verts));
   radeon_set_context_reg(ctx_cs, R_028B4C_GE_NGG_SUBGRP_CNTL,
                          S_028B4C_PRIM_AMP_FACTOR(ngg_state->prim_amp_factor) |
                             S_028B4C_THDS_PER_SUBGRP(0)); /* for fast launch */
   radeon_set_context_reg(
      ctx_cs, R_028B90_VGT_GS_INSTANCE_CNT,
      S_028B90_CNT(gs_num_invocations) | S_028B90_ENABLE(gs_num_invocations > 1) |
         S_028B90_EN_MAX_VERT_OUT_PER_GS_INSTANCE(ngg_state->max_vert_out_per_gs_instance));

   /* User edge flags are set by the pos exports. If user edge flags are
    * not used, we must use hw-generated edge flags and pass them via
    * the prim export to prevent drawing lines on internal edges of
    * decomposed primitives (such as quads) with polygon mode = lines.
    *
    * TODO: We should combine hw-generated edge flags with user edge
    *       flags in the shader.
    */
   radeon_set_context_reg(
      ctx_cs, R_028838_PA_CL_NGG_CNTL,
      S_028838_INDEX_BUF_EDGE_FLAG_ENA(!radv_pipeline_has_tess(pipeline) &&
                                       !radv_pipeline_has_gs(pipeline)) |
         /* Reuse for NGG. */
         S_028838_VERTEX_REUSE_DEPTH(
            pipeline->device->physical_device->rad_info.chip_class >= GFX10_3 ? 30 : 0));

   ge_cntl = S_03096C_PRIM_GRP_SIZE(ngg_state->max_gsprims) |
             S_03096C_VERT_GRP_SIZE(256) | /* 256 = disable vertex grouping */
             S_03096C_BREAK_WAVE_AT_EOI(break_wave_at_eoi);

   /* Bug workaround for a possible hang with non-tessellation cases.
    * Tessellation always sets GE_CNTL.VERT_GRP_SIZE = 0
    *
    * Requirement: GE_CNTL.VERT_GRP_SIZE = VGT_GS_ONCHIP_CNTL.ES_VERTS_PER_SUBGRP - 5
    */
   if (pipeline->device->physical_device->rad_info.chip_class == GFX10 &&
       !radv_pipeline_has_tess(pipeline) && ngg_state->hw_max_esverts != 256) {
      ge_cntl &= C_03096C_VERT_GRP_SIZE;

      if (ngg_state->hw_max_esverts > 5) {
         ge_cntl |= S_03096C_VERT_GRP_SIZE(ngg_state->hw_max_esverts - 5);
      }
   }

   radeon_set_uconfig_reg(ctx_cs, R_03096C_GE_CNTL, ge_cntl);
}

static void
radv_pipeline_generate_hw_hs(struct radeon_cmdbuf *cs, const struct radv_pipeline *pipeline,
                             const struct radv_shader_variant *shader)
{
   uint64_t va = radv_buffer_get_va(shader->bo) + shader->bo_offset;

   if (pipeline->device->physical_device->rad_info.chip_class >= GFX9) {
      if (pipeline->device->physical_device->rad_info.chip_class >= GFX10) {
         radeon_set_sh_reg_seq(cs, R_00B520_SPI_SHADER_PGM_LO_LS, 2);
         radeon_emit(cs, va >> 8);
         radeon_emit(cs, S_00B524_MEM_BASE(va >> 40));
      } else {
         radeon_set_sh_reg_seq(cs, R_00B410_SPI_SHADER_PGM_LO_LS, 2);
         radeon_emit(cs, va >> 8);
         radeon_emit(cs, S_00B414_MEM_BASE(va >> 40));
      }

      radeon_set_sh_reg_seq(cs, R_00B428_SPI_SHADER_PGM_RSRC1_HS, 2);
      radeon_emit(cs, shader->config.rsrc1);
      radeon_emit(cs, shader->config.rsrc2);
   } else {
      radeon_set_sh_reg_seq(cs, R_00B420_SPI_SHADER_PGM_LO_HS, 4);
      radeon_emit(cs, va >> 8);
      radeon_emit(cs, S_00B424_MEM_BASE(va >> 40));
      radeon_emit(cs, shader->config.rsrc1);
      radeon_emit(cs, shader->config.rsrc2);
   }
}

static void
radv_pipeline_generate_vertex_shader(struct radeon_cmdbuf *ctx_cs, struct radeon_cmdbuf *cs,
                                     const struct radv_pipeline *pipeline)
{
   struct radv_shader_variant *vs;

   /* Skip shaders merged into HS/GS */
   vs = pipeline->shaders[MESA_SHADER_VERTEX];
   if (!vs)
      return;

   if (vs->info.vs.as_ls)
      radv_pipeline_generate_hw_ls(cs, pipeline, vs);
   else if (vs->info.vs.as_es)
      radv_pipeline_generate_hw_es(cs, pipeline, vs);
   else if (vs->info.is_ngg)
      radv_pipeline_generate_hw_ngg(ctx_cs, cs, pipeline, vs);
   else
      radv_pipeline_generate_hw_vs(ctx_cs, cs, pipeline, vs);
}

static void
radv_pipeline_generate_tess_shaders(struct radeon_cmdbuf *ctx_cs, struct radeon_cmdbuf *cs,
                                    const struct radv_pipeline *pipeline)
{
   struct radv_shader_variant *tes, *tcs;

   tcs = pipeline->shaders[MESA_SHADER_TESS_CTRL];
   tes = pipeline->shaders[MESA_SHADER_TESS_EVAL];

   if (tes) {
      if (tes->info.is_ngg) {
         radv_pipeline_generate_hw_ngg(ctx_cs, cs, pipeline, tes);
      } else if (tes->info.tes.as_es)
         radv_pipeline_generate_hw_es(cs, pipeline, tes);
      else
         radv_pipeline_generate_hw_vs(ctx_cs, cs, pipeline, tes);
   }

   radv_pipeline_generate_hw_hs(cs, pipeline, tcs);

   if (pipeline->device->physical_device->rad_info.chip_class >= GFX10 &&
       !radv_pipeline_has_gs(pipeline) && !radv_pipeline_has_ngg(pipeline)) {
      radeon_set_context_reg(ctx_cs, R_028A44_VGT_GS_ONCHIP_CNTL,
                             S_028A44_ES_VERTS_PER_SUBGRP(250) | S_028A44_GS_PRIMS_PER_SUBGRP(126) |
                                S_028A44_GS_INST_PRIMS_IN_SUBGRP(126));
   }
}

static void
radv_pipeline_generate_tess_state(struct radeon_cmdbuf *ctx_cs,
                                  const struct radv_pipeline *pipeline,
                                  const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   struct radv_shader_variant *tes = radv_get_shader(pipeline, MESA_SHADER_TESS_EVAL);
   unsigned type = 0, partitioning = 0, topology = 0, distribution_mode = 0;
   unsigned num_tcs_input_cp, num_tcs_output_cp, num_patches;
   unsigned ls_hs_config;

   num_tcs_input_cp = pCreateInfo->pTessellationState->patchControlPoints;
   num_tcs_output_cp =
      pipeline->shaders[MESA_SHADER_TESS_CTRL]->info.tcs.tcs_vertices_out; // TCS VERTICES OUT
   num_patches = pipeline->shaders[MESA_SHADER_TESS_CTRL]->info.num_tess_patches;

   ls_hs_config = S_028B58_NUM_PATCHES(num_patches) | S_028B58_HS_NUM_INPUT_CP(num_tcs_input_cp) |
                  S_028B58_HS_NUM_OUTPUT_CP(num_tcs_output_cp);

   if (pipeline->device->physical_device->rad_info.chip_class >= GFX7) {
      radeon_set_context_reg_idx(ctx_cs, R_028B58_VGT_LS_HS_CONFIG, 2, ls_hs_config);
   } else {
      radeon_set_context_reg(ctx_cs, R_028B58_VGT_LS_HS_CONFIG, ls_hs_config);
   }

   switch (tes->info.tes.primitive_mode) {
   case GL_TRIANGLES:
      type = V_028B6C_TESS_TRIANGLE;
      break;
   case GL_QUADS:
      type = V_028B6C_TESS_QUAD;
      break;
   case GL_ISOLINES:
      type = V_028B6C_TESS_ISOLINE;
      break;
   }

   switch (tes->info.tes.spacing) {
   case TESS_SPACING_EQUAL:
      partitioning = V_028B6C_PART_INTEGER;
      break;
   case TESS_SPACING_FRACTIONAL_ODD:
      partitioning = V_028B6C_PART_FRAC_ODD;
      break;
   case TESS_SPACING_FRACTIONAL_EVEN:
      partitioning = V_028B6C_PART_FRAC_EVEN;
      break;
   default:
      break;
   }

   bool ccw = tes->info.tes.ccw;
   const VkPipelineTessellationDomainOriginStateCreateInfo *domain_origin_state =
      vk_find_struct_const(pCreateInfo->pTessellationState,
                           PIPELINE_TESSELLATION_DOMAIN_ORIGIN_STATE_CREATE_INFO);

   if (domain_origin_state &&
       domain_origin_state->domainOrigin != VK_TESSELLATION_DOMAIN_ORIGIN_UPPER_LEFT)
      ccw = !ccw;

   if (tes->info.tes.point_mode)
      topology = V_028B6C_OUTPUT_POINT;
   else if (tes->info.tes.primitive_mode == GL_ISOLINES)
      topology = V_028B6C_OUTPUT_LINE;
   else if (ccw)
      topology = V_028B6C_OUTPUT_TRIANGLE_CCW;
   else
      topology = V_028B6C_OUTPUT_TRIANGLE_CW;

   if (pipeline->device->physical_device->rad_info.has_distributed_tess) {
      if (pipeline->device->physical_device->rad_info.family == CHIP_FIJI ||
          pipeline->device->physical_device->rad_info.family >= CHIP_POLARIS10)
         distribution_mode = V_028B6C_TRAPEZOIDS;
      else
         distribution_mode = V_028B6C_DONUTS;
   } else
      distribution_mode = V_028B6C_NO_DIST;

   radeon_set_context_reg(ctx_cs, R_028B6C_VGT_TF_PARAM,
                          S_028B6C_TYPE(type) | S_028B6C_PARTITIONING(partitioning) |
                             S_028B6C_TOPOLOGY(topology) |
                             S_028B6C_DISTRIBUTION_MODE(distribution_mode));
}

static void
radv_pipeline_generate_hw_gs(struct radeon_cmdbuf *ctx_cs, struct radeon_cmdbuf *cs,
                             const struct radv_pipeline *pipeline,
                             const struct radv_shader_variant *gs)
{
   const struct gfx9_gs_info *gs_state = &gs->info.gs_ring_info;
   unsigned gs_max_out_vertices;
   const uint8_t *num_components;
   uint8_t max_stream;
   unsigned offset;
   uint64_t va;

   gs_max_out_vertices = gs->info.gs.vertices_out;
   max_stream = gs->info.gs.max_stream;
   num_components = gs->info.gs.num_stream_output_components;

   offset = num_components[0] * gs_max_out_vertices;

   radeon_set_context_reg_seq(ctx_cs, R_028A60_VGT_GSVS_RING_OFFSET_1, 3);
   radeon_emit(ctx_cs, offset);
   if (max_stream >= 1)
      offset += num_components[1] * gs_max_out_vertices;
   radeon_emit(ctx_cs, offset);
   if (max_stream >= 2)
      offset += num_components[2] * gs_max_out_vertices;
   radeon_emit(ctx_cs, offset);
   if (max_stream >= 3)
      offset += num_components[3] * gs_max_out_vertices;
   radeon_set_context_reg(ctx_cs, R_028AB0_VGT_GSVS_RING_ITEMSIZE, offset);

   radeon_set_context_reg_seq(ctx_cs, R_028B5C_VGT_GS_VERT_ITEMSIZE, 4);
   radeon_emit(ctx_cs, num_components[0]);
   radeon_emit(ctx_cs, (max_stream >= 1) ? num_components[1] : 0);
   radeon_emit(ctx_cs, (max_stream >= 2) ? num_components[2] : 0);
   radeon_emit(ctx_cs, (max_stream >= 3) ? num_components[3] : 0);

   uint32_t gs_num_invocations = gs->info.gs.invocations;
   radeon_set_context_reg(
      ctx_cs, R_028B90_VGT_GS_INSTANCE_CNT,
      S_028B90_CNT(MIN2(gs_num_invocations, 127)) | S_028B90_ENABLE(gs_num_invocations > 0));

   radeon_set_context_reg(ctx_cs, R_028AAC_VGT_ESGS_RING_ITEMSIZE,
                          gs_state->vgt_esgs_ring_itemsize);

   va = radv_buffer_get_va(gs->bo) + gs->bo_offset;

   if (pipeline->device->physical_device->rad_info.chip_class >= GFX9) {
      if (pipeline->device->physical_device->rad_info.chip_class >= GFX10) {
         radeon_set_sh_reg_seq(cs, R_00B320_SPI_SHADER_PGM_LO_ES, 2);
         radeon_emit(cs, va >> 8);
         radeon_emit(cs, S_00B324_MEM_BASE(va >> 40));
      } else {
         radeon_set_sh_reg_seq(cs, R_00B210_SPI_SHADER_PGM_LO_ES, 2);
         radeon_emit(cs, va >> 8);
         radeon_emit(cs, S_00B214_MEM_BASE(va >> 40));
      }

      radeon_set_sh_reg_seq(cs, R_00B228_SPI_SHADER_PGM_RSRC1_GS, 2);
      radeon_emit(cs, gs->config.rsrc1);
      radeon_emit(cs, gs->config.rsrc2 | S_00B22C_LDS_SIZE(gs_state->lds_size));

      radeon_set_context_reg(ctx_cs, R_028A44_VGT_GS_ONCHIP_CNTL, gs_state->vgt_gs_onchip_cntl);
      radeon_set_context_reg(ctx_cs, R_028A94_VGT_GS_MAX_PRIMS_PER_SUBGROUP,
                             gs_state->vgt_gs_max_prims_per_subgroup);
   } else {
      radeon_set_sh_reg_seq(cs, R_00B220_SPI_SHADER_PGM_LO_GS, 4);
      radeon_emit(cs, va >> 8);
      radeon_emit(cs, S_00B224_MEM_BASE(va >> 40));
      radeon_emit(cs, gs->config.rsrc1);
      radeon_emit(cs, gs->config.rsrc2);
   }

   radv_pipeline_generate_hw_vs(ctx_cs, cs, pipeline, pipeline->gs_copy_shader);
}

static void
radv_pipeline_generate_geometry_shader(struct radeon_cmdbuf *ctx_cs, struct radeon_cmdbuf *cs,
                                       const struct radv_pipeline *pipeline)
{
   struct radv_shader_variant *gs;

   gs = pipeline->shaders[MESA_SHADER_GEOMETRY];
   if (!gs)
      return;

   if (gs->info.is_ngg)
      radv_pipeline_generate_hw_ngg(ctx_cs, cs, pipeline, gs);
   else
      radv_pipeline_generate_hw_gs(ctx_cs, cs, pipeline, gs);

   radeon_set_context_reg(ctx_cs, R_028B38_VGT_GS_MAX_VERT_OUT, gs->info.gs.vertices_out);
}

static uint32_t
offset_to_ps_input(uint32_t offset, bool flat_shade, bool explicit, bool float16)
{
   uint32_t ps_input_cntl;
   if (offset <= AC_EXP_PARAM_OFFSET_31) {
      ps_input_cntl = S_028644_OFFSET(offset);
      if (flat_shade || explicit)
         ps_input_cntl |= S_028644_FLAT_SHADE(1);
      if (explicit) {
         /* Force parameter cache to be read in passthrough
          * mode.
          */
         ps_input_cntl |= S_028644_OFFSET(1 << 5);
      }
      if (float16) {
         ps_input_cntl |= S_028644_FP16_INTERP_MODE(1) | S_028644_ATTR0_VALID(1);
      }
   } else {
      /* The input is a DEFAULT_VAL constant. */
      assert(offset >= AC_EXP_PARAM_DEFAULT_VAL_0000 && offset <= AC_EXP_PARAM_DEFAULT_VAL_1111);
      offset -= AC_EXP_PARAM_DEFAULT_VAL_0000;
      ps_input_cntl = S_028644_OFFSET(0x20) | S_028644_DEFAULT_VAL(offset);
   }
   return ps_input_cntl;
}

static void
radv_pipeline_generate_ps_inputs(struct radeon_cmdbuf *ctx_cs, const struct radv_pipeline *pipeline)
{
   struct radv_shader_variant *ps = pipeline->shaders[MESA_SHADER_FRAGMENT];
   const struct radv_vs_output_info *outinfo = get_vs_output_info(pipeline);
   uint32_t ps_input_cntl[32];

   unsigned ps_offset = 0;

   if (ps->info.ps.prim_id_input) {
      unsigned vs_offset = outinfo->vs_output_param_offset[VARYING_SLOT_PRIMITIVE_ID];
      if (vs_offset != AC_EXP_PARAM_UNDEFINED) {
         ps_input_cntl[ps_offset] = offset_to_ps_input(vs_offset, true, false, false);
         ++ps_offset;
      }
   }

   if (ps->info.ps.layer_input || ps->info.needs_multiview_view_index) {
      unsigned vs_offset = outinfo->vs_output_param_offset[VARYING_SLOT_LAYER];
      if (vs_offset != AC_EXP_PARAM_UNDEFINED)
         ps_input_cntl[ps_offset] = offset_to_ps_input(vs_offset, true, false, false);
      else
         ps_input_cntl[ps_offset] =
            offset_to_ps_input(AC_EXP_PARAM_DEFAULT_VAL_0000, true, false, false);
      ++ps_offset;
   }

   if (ps->info.ps.viewport_index_input) {
      unsigned vs_offset = outinfo->vs_output_param_offset[VARYING_SLOT_VIEWPORT];
      if (vs_offset != AC_EXP_PARAM_UNDEFINED)
         ps_input_cntl[ps_offset] = offset_to_ps_input(vs_offset, true, false, false);
      else
         ps_input_cntl[ps_offset] =
            offset_to_ps_input(AC_EXP_PARAM_DEFAULT_VAL_0000, true, false, false);
      ++ps_offset;
   }

   if (ps->info.ps.has_pcoord) {
      unsigned val;
      val = S_028644_PT_SPRITE_TEX(1) | S_028644_OFFSET(0x20);
      ps_input_cntl[ps_offset] = val;
      ps_offset++;
   }

   if (ps->info.ps.num_input_clips_culls) {
      unsigned vs_offset;

      vs_offset = outinfo->vs_output_param_offset[VARYING_SLOT_CLIP_DIST0];
      if (vs_offset != AC_EXP_PARAM_UNDEFINED) {
         ps_input_cntl[ps_offset] = offset_to_ps_input(vs_offset, false, false, false);
         ++ps_offset;
      }

      vs_offset = outinfo->vs_output_param_offset[VARYING_SLOT_CLIP_DIST1];
      if (vs_offset != AC_EXP_PARAM_UNDEFINED && ps->info.ps.num_input_clips_culls > 4) {
         ps_input_cntl[ps_offset] = offset_to_ps_input(vs_offset, false, false, false);
         ++ps_offset;
      }
   }

   for (unsigned i = 0; i < 32 && (1u << i) <= ps->info.ps.input_mask; ++i) {
      unsigned vs_offset;
      bool flat_shade;
      bool explicit;
      bool float16;
      if (!(ps->info.ps.input_mask & (1u << i)))
         continue;

      vs_offset = outinfo->vs_output_param_offset[VARYING_SLOT_VAR0 + i];
      if (vs_offset == AC_EXP_PARAM_UNDEFINED) {
         ps_input_cntl[ps_offset] = S_028644_OFFSET(0x20);
         ++ps_offset;
         continue;
      }

      flat_shade = !!(ps->info.ps.flat_shaded_mask & (1u << ps_offset));
      explicit = !!(ps->info.ps.explicit_shaded_mask & (1u << ps_offset));
      float16 = !!(ps->info.ps.float16_shaded_mask & (1u << ps_offset));

      ps_input_cntl[ps_offset] = offset_to_ps_input(vs_offset, flat_shade, explicit, float16);
      ++ps_offset;
   }

   if (ps_offset) {
      radeon_set_context_reg_seq(ctx_cs, R_028644_SPI_PS_INPUT_CNTL_0, ps_offset);
      for (unsigned i = 0; i < ps_offset; i++) {
         radeon_emit(ctx_cs, ps_input_cntl[i]);
      }
   }
}

static uint32_t
radv_compute_db_shader_control(const struct radv_device *device,
                               const struct radv_pipeline *pipeline,
                               const struct radv_shader_variant *ps)
{
   unsigned conservative_z_export = V_02880C_EXPORT_ANY_Z;
   unsigned z_order;
   if (ps->info.ps.early_fragment_test || !ps->info.ps.writes_memory)
      z_order = V_02880C_EARLY_Z_THEN_LATE_Z;
   else
      z_order = V_02880C_LATE_Z;

   if (ps->info.ps.depth_layout == FRAG_DEPTH_LAYOUT_GREATER)
      conservative_z_export = V_02880C_EXPORT_GREATER_THAN_Z;
   else if (ps->info.ps.depth_layout == FRAG_DEPTH_LAYOUT_LESS)
      conservative_z_export = V_02880C_EXPORT_LESS_THAN_Z;

   bool disable_rbplus = device->physical_device->rad_info.has_rbplus &&
                         !device->physical_device->rad_info.rbplus_allowed;

   /* It shouldn't be needed to export gl_SampleMask when MSAA is disabled
    * but this appears to break Project Cars (DXVK). See
    * https://bugs.freedesktop.org/show_bug.cgi?id=109401
    */
   bool mask_export_enable = ps->info.ps.writes_sample_mask;

   return S_02880C_Z_EXPORT_ENABLE(ps->info.ps.writes_z) |
          S_02880C_STENCIL_TEST_VAL_EXPORT_ENABLE(ps->info.ps.writes_stencil) |
          S_02880C_KILL_ENABLE(!!ps->info.ps.can_discard) |
          S_02880C_MASK_EXPORT_ENABLE(mask_export_enable) |
          S_02880C_CONSERVATIVE_Z_EXPORT(conservative_z_export) | S_02880C_Z_ORDER(z_order) |
          S_02880C_DEPTH_BEFORE_SHADER(ps->info.ps.early_fragment_test) |
          S_02880C_PRE_SHADER_DEPTH_COVERAGE_ENABLE(ps->info.ps.post_depth_coverage) |
          S_02880C_EXEC_ON_HIER_FAIL(ps->info.ps.writes_memory) |
          S_02880C_EXEC_ON_NOOP(ps->info.ps.writes_memory) |
          S_02880C_DUAL_QUAD_DISABLE(disable_rbplus);
}

static void
radv_pipeline_generate_fragment_shader(struct radeon_cmdbuf *ctx_cs, struct radeon_cmdbuf *cs,
                                       struct radv_pipeline *pipeline)
{
   struct radv_shader_variant *ps;
   uint64_t va;
   assert(pipeline->shaders[MESA_SHADER_FRAGMENT]);

   ps = pipeline->shaders[MESA_SHADER_FRAGMENT];
   va = radv_buffer_get_va(ps->bo) + ps->bo_offset;

   radeon_set_sh_reg_seq(cs, R_00B020_SPI_SHADER_PGM_LO_PS, 4);
   radeon_emit(cs, va >> 8);
   radeon_emit(cs, S_00B024_MEM_BASE(va >> 40));
   radeon_emit(cs, ps->config.rsrc1);
   radeon_emit(cs, ps->config.rsrc2);

   radeon_set_context_reg(ctx_cs, R_02880C_DB_SHADER_CONTROL,
                          radv_compute_db_shader_control(pipeline->device, pipeline, ps));

   radeon_set_context_reg(ctx_cs, R_0286CC_SPI_PS_INPUT_ENA, ps->config.spi_ps_input_ena);

   radeon_set_context_reg(ctx_cs, R_0286D0_SPI_PS_INPUT_ADDR, ps->config.spi_ps_input_addr);

   radeon_set_context_reg(
      ctx_cs, R_0286D8_SPI_PS_IN_CONTROL,
      S_0286D8_NUM_INTERP(ps->info.ps.num_interp) | S_0286D8_PS_W32_EN(ps->info.wave_size == 32));

   radeon_set_context_reg(ctx_cs, R_0286E0_SPI_BARYC_CNTL, pipeline->graphics.spi_baryc_cntl);

   radeon_set_context_reg(
      ctx_cs, R_028710_SPI_SHADER_Z_FORMAT,
      ac_get_spi_shader_z_format(ps->info.ps.writes_z, ps->info.ps.writes_stencil,
                                 ps->info.ps.writes_sample_mask));

   if (pipeline->device->dfsm_allowed) {
      /* optimise this? */
      radeon_emit(cs, PKT3(PKT3_EVENT_WRITE, 0, 0));
      radeon_emit(cs, EVENT_TYPE(V_028A90_FLUSH_DFSM) | EVENT_INDEX(0));
   }
}

static void
radv_pipeline_generate_vgt_vertex_reuse(struct radeon_cmdbuf *ctx_cs,
                                        const struct radv_pipeline *pipeline)
{
   if (pipeline->device->physical_device->rad_info.family < CHIP_POLARIS10 ||
       pipeline->device->physical_device->rad_info.chip_class >= GFX10)
      return;

   unsigned vtx_reuse_depth = 30;
   if (radv_pipeline_has_tess(pipeline) &&
       radv_get_shader(pipeline, MESA_SHADER_TESS_EVAL)->info.tes.spacing ==
          TESS_SPACING_FRACTIONAL_ODD) {
      vtx_reuse_depth = 14;
   }
   radeon_set_context_reg(ctx_cs, R_028C58_VGT_VERTEX_REUSE_BLOCK_CNTL,
                          S_028C58_VTX_REUSE_DEPTH(vtx_reuse_depth));
}

static void
radv_pipeline_generate_vgt_shader_config(struct radeon_cmdbuf *ctx_cs,
                                         const struct radv_pipeline *pipeline)
{
   uint32_t stages = 0;
   if (radv_pipeline_has_tess(pipeline)) {
      stages |= S_028B54_LS_EN(V_028B54_LS_STAGE_ON) | S_028B54_HS_EN(1) | S_028B54_DYNAMIC_HS(1);

      if (radv_pipeline_has_gs(pipeline))
         stages |= S_028B54_ES_EN(V_028B54_ES_STAGE_DS) | S_028B54_GS_EN(1);
      else if (radv_pipeline_has_ngg(pipeline))
         stages |= S_028B54_ES_EN(V_028B54_ES_STAGE_DS);
      else
         stages |= S_028B54_VS_EN(V_028B54_VS_STAGE_DS);
   } else if (radv_pipeline_has_gs(pipeline)) {
      stages |= S_028B54_ES_EN(V_028B54_ES_STAGE_REAL) | S_028B54_GS_EN(1);
   } else if (radv_pipeline_has_ngg(pipeline)) {
      stages |= S_028B54_ES_EN(V_028B54_ES_STAGE_REAL);
   }

   if (radv_pipeline_has_ngg(pipeline)) {
      stages |= S_028B54_PRIMGEN_EN(1);
      if (pipeline->streamout_shader)
         stages |= S_028B54_NGG_WAVE_ID_EN(1);
      if (radv_pipeline_has_ngg_passthrough(pipeline))
         stages |= S_028B54_PRIMGEN_PASSTHRU_EN(1);
   } else if (radv_pipeline_has_gs(pipeline)) {
      stages |= S_028B54_VS_EN(V_028B54_VS_STAGE_COPY_SHADER);
   }

   if (pipeline->device->physical_device->rad_info.chip_class >= GFX9)
      stages |= S_028B54_MAX_PRIMGRP_IN_WAVE(2);

   if (pipeline->device->physical_device->rad_info.chip_class >= GFX10) {
      uint8_t hs_size = 64, gs_size = 64, vs_size = 64;

      if (radv_pipeline_has_tess(pipeline))
         hs_size = pipeline->shaders[MESA_SHADER_TESS_CTRL]->info.wave_size;

      if (pipeline->shaders[MESA_SHADER_GEOMETRY]) {
         vs_size = gs_size = pipeline->shaders[MESA_SHADER_GEOMETRY]->info.wave_size;
         if (radv_pipeline_has_gs_copy_shader(pipeline))
            vs_size = pipeline->gs_copy_shader->info.wave_size;
      } else if (pipeline->shaders[MESA_SHADER_TESS_EVAL])
         vs_size = pipeline->shaders[MESA_SHADER_TESS_EVAL]->info.wave_size;
      else if (pipeline->shaders[MESA_SHADER_VERTEX])
         vs_size = pipeline->shaders[MESA_SHADER_VERTEX]->info.wave_size;

      if (radv_pipeline_has_ngg(pipeline))
         gs_size = vs_size;

      /* legacy GS only supports Wave64 */
      stages |= S_028B54_HS_W32_EN(hs_size == 32 ? 1 : 0) |
                S_028B54_GS_W32_EN(gs_size == 32 ? 1 : 0) |
                S_028B54_VS_W32_EN(vs_size == 32 ? 1 : 0);
   }

   radeon_set_context_reg(ctx_cs, R_028B54_VGT_SHADER_STAGES_EN, stages);
}

static void
radv_pipeline_generate_cliprect_rule(struct radeon_cmdbuf *ctx_cs,
                                     const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   const VkPipelineDiscardRectangleStateCreateInfoEXT *discard_rectangle_info =
      vk_find_struct_const(pCreateInfo->pNext, PIPELINE_DISCARD_RECTANGLE_STATE_CREATE_INFO_EXT);
   uint32_t cliprect_rule = 0;

   if (!discard_rectangle_info) {
      cliprect_rule = 0xffff;
   } else {
      for (unsigned i = 0; i < (1u << MAX_DISCARD_RECTANGLES); ++i) {
         /* Interpret i as a bitmask, and then set the bit in
          * the mask if that combination of rectangles in which
          * the pixel is contained should pass the cliprect
          * test.
          */
         unsigned relevant_subset = i & ((1u << discard_rectangle_info->discardRectangleCount) - 1);

         if (discard_rectangle_info->discardRectangleMode ==
                VK_DISCARD_RECTANGLE_MODE_INCLUSIVE_EXT &&
             !relevant_subset)
            continue;

         if (discard_rectangle_info->discardRectangleMode ==
                VK_DISCARD_RECTANGLE_MODE_EXCLUSIVE_EXT &&
             relevant_subset)
            continue;

         cliprect_rule |= 1u << i;
      }
   }

   radeon_set_context_reg(ctx_cs, R_02820C_PA_SC_CLIPRECT_RULE, cliprect_rule);
}

static void
gfx10_pipeline_generate_ge_cntl(struct radeon_cmdbuf *ctx_cs, struct radv_pipeline *pipeline)
{
   bool break_wave_at_eoi = false;
   unsigned primgroup_size;
   unsigned vertgroup_size = 256; /* 256 = disable vertex grouping */

   if (radv_pipeline_has_tess(pipeline)) {
      primgroup_size = pipeline->shaders[MESA_SHADER_TESS_CTRL]->info.num_tess_patches;
   } else if (radv_pipeline_has_gs(pipeline)) {
      const struct gfx9_gs_info *gs_state =
         &pipeline->shaders[MESA_SHADER_GEOMETRY]->info.gs_ring_info;
      unsigned vgt_gs_onchip_cntl = gs_state->vgt_gs_onchip_cntl;
      primgroup_size = G_028A44_GS_PRIMS_PER_SUBGRP(vgt_gs_onchip_cntl);
   } else {
      primgroup_size = 128; /* recommended without a GS and tess */
   }

   if (radv_pipeline_has_tess(pipeline)) {
      if (pipeline->shaders[MESA_SHADER_TESS_CTRL]->info.uses_prim_id ||
          radv_get_shader(pipeline, MESA_SHADER_TESS_EVAL)->info.uses_prim_id)
         break_wave_at_eoi = true;
   }

   radeon_set_uconfig_reg(ctx_cs, R_03096C_GE_CNTL,
                          S_03096C_PRIM_GRP_SIZE(primgroup_size) |
                             S_03096C_VERT_GRP_SIZE(vertgroup_size) |
                             S_03096C_PACKET_TO_ONE_PA(0) /* line stipple */ |
                             S_03096C_BREAK_WAVE_AT_EOI(break_wave_at_eoi));
}

static void
radv_pipeline_generate_vgt_gs_out(struct radeon_cmdbuf *ctx_cs,
                                  const struct radv_pipeline *pipeline,
                                  const VkGraphicsPipelineCreateInfo *pCreateInfo,
                                  const struct radv_graphics_pipeline_create_info *extra)
{
   uint32_t gs_out;

   if (radv_pipeline_has_gs(pipeline)) {
      gs_out =
         si_conv_gl_prim_to_gs_out(pipeline->shaders[MESA_SHADER_GEOMETRY]->info.gs.output_prim);
   } else if (radv_pipeline_has_tess(pipeline)) {
      if (pipeline->shaders[MESA_SHADER_TESS_EVAL]->info.tes.point_mode) {
         gs_out = V_028A6C_POINTLIST;
      } else {
         gs_out = si_conv_gl_prim_to_gs_out(
            pipeline->shaders[MESA_SHADER_TESS_EVAL]->info.tes.primitive_mode);
      }
   } else {
      gs_out = si_conv_prim_to_gs_out(pCreateInfo->pInputAssemblyState->topology);
   }

   if (extra && extra->use_rectlist) {
      gs_out = V_028A6C_TRISTRIP;
      if (radv_pipeline_has_ngg(pipeline))
         gs_out = V_028A6C_RECTLIST;
   }

   radeon_set_context_reg(ctx_cs, R_028A6C_VGT_GS_OUT_PRIM_TYPE, gs_out);
}

static bool
gfx103_pipeline_vrs_coarse_shading(const struct radv_pipeline *pipeline)
{
   struct radv_shader_variant *ps = pipeline->shaders[MESA_SHADER_FRAGMENT];
   struct radv_device *device = pipeline->device;

   if (device->instance->debug_flags & RADV_DEBUG_NO_VRS_FLAT_SHADING)
      return false;

   if (!ps->info.ps.allow_flat_shading)
      return false;

   return true;
}

static void
gfx103_pipeline_generate_vrs_state(struct radeon_cmdbuf *ctx_cs,
                                   const struct radv_pipeline *pipeline,
                                   const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   uint32_t mode = V_028064_VRS_COMB_MODE_PASSTHRU;
   uint8_t rate_x = 0, rate_y = 0;
   bool enable_vrs = false;

   if (vk_find_struct_const(pCreateInfo->pNext,
                            PIPELINE_FRAGMENT_SHADING_RATE_STATE_CREATE_INFO_KHR) ||
       radv_is_state_dynamic(pCreateInfo, VK_DYNAMIC_STATE_FRAGMENT_SHADING_RATE_KHR)) {
      /* Enable draw call VRS because it's explicitly requested.  */
      enable_vrs = true;
   } else if (gfx103_pipeline_vrs_coarse_shading(pipeline)) {
      /* Enable VRS coarse shading 2x2 if the driver determined that
       * it's safe to enable.
       */
      mode = V_028064_VRS_COMB_MODE_OVERRIDE;
      rate_x = rate_y = 1;
   } else if (pipeline->device->force_vrs != RADV_FORCE_VRS_NONE) {
      /* Force enable vertex VRS if requested by the user. */
      radeon_set_context_reg(
         ctx_cs, R_028848_PA_CL_VRS_CNTL,
         S_028848_SAMPLE_ITER_COMBINER_MODE(V_028848_VRS_COMB_MODE_OVERRIDE) |
            S_028848_VERTEX_RATE_COMBINER_MODE(V_028848_VRS_COMB_MODE_OVERRIDE));

      /* If the shader is using discard, turn off coarse shading
       * because discard at 2x2 pixel granularity degrades quality
       * too much. MIN allows sample shading but not coarse shading.
       */
      struct radv_shader_variant *ps = pipeline->shaders[MESA_SHADER_FRAGMENT];

      mode = ps->info.ps.can_discard ? V_028064_VRS_COMB_MODE_MIN : V_028064_VRS_COMB_MODE_PASSTHRU;
   }

   radeon_set_context_reg(ctx_cs, R_028A98_VGT_DRAW_PAYLOAD_CNTL, S_028A98_EN_VRS_RATE(enable_vrs));

   radeon_set_context_reg(ctx_cs, R_028064_DB_VRS_OVERRIDE_CNTL,
                          S_028064_VRS_OVERRIDE_RATE_COMBINER_MODE(mode) |
                             S_028064_VRS_OVERRIDE_RATE_X(rate_x) |
                             S_028064_VRS_OVERRIDE_RATE_Y(rate_y));
}

static void
radv_pipeline_generate_pm4(struct radv_pipeline *pipeline,
                           const VkGraphicsPipelineCreateInfo *pCreateInfo,
                           const struct radv_graphics_pipeline_create_info *extra,
                           const struct radv_blend_state *blend)
{
   struct radeon_cmdbuf *ctx_cs = &pipeline->ctx_cs;
   struct radeon_cmdbuf *cs = &pipeline->cs;

   cs->max_dw = 64;
   ctx_cs->max_dw = 256;
   cs->buf = malloc(4 * (cs->max_dw + ctx_cs->max_dw));
   ctx_cs->buf = cs->buf + cs->max_dw;

   radv_pipeline_generate_depth_stencil_state(ctx_cs, pipeline, pCreateInfo, extra);
   radv_pipeline_generate_blend_state(ctx_cs, pipeline, blend);
   radv_pipeline_generate_raster_state(ctx_cs, pipeline, pCreateInfo);
   radv_pipeline_generate_multisample_state(ctx_cs, pipeline);
   radv_pipeline_generate_vgt_gs_mode(ctx_cs, pipeline);
   radv_pipeline_generate_vertex_shader(ctx_cs, cs, pipeline);

   if (radv_pipeline_has_tess(pipeline)) {
      radv_pipeline_generate_tess_shaders(ctx_cs, cs, pipeline);
      radv_pipeline_generate_tess_state(ctx_cs, pipeline, pCreateInfo);
   }

   radv_pipeline_generate_geometry_shader(ctx_cs, cs, pipeline);
   radv_pipeline_generate_fragment_shader(ctx_cs, cs, pipeline);
   radv_pipeline_generate_ps_inputs(ctx_cs, pipeline);
   radv_pipeline_generate_vgt_vertex_reuse(ctx_cs, pipeline);
   radv_pipeline_generate_vgt_shader_config(ctx_cs, pipeline);
   radv_pipeline_generate_cliprect_rule(ctx_cs, pCreateInfo);
   radv_pipeline_generate_vgt_gs_out(ctx_cs, pipeline, pCreateInfo, extra);

   if (pipeline->device->physical_device->rad_info.chip_class >= GFX10 &&
       !radv_pipeline_has_ngg(pipeline))
      gfx10_pipeline_generate_ge_cntl(ctx_cs, pipeline);

   if (pipeline->device->physical_device->rad_info.chip_class >= GFX10_3)
      gfx103_pipeline_generate_vrs_state(ctx_cs, pipeline, pCreateInfo);

   pipeline->ctx_cs_hash = _mesa_hash_data(ctx_cs->buf, ctx_cs->cdw * 4);

   assert(ctx_cs->cdw <= ctx_cs->max_dw);
   assert(cs->cdw <= cs->max_dw);
}

static void
radv_pipeline_init_vertex_input_state(struct radv_pipeline *pipeline,
                                      const VkGraphicsPipelineCreateInfo *pCreateInfo)
{
   const struct radv_shader_info *info = &radv_get_shader(pipeline, MESA_SHADER_VERTEX)->info;
   const VkPipelineVertexInputStateCreateInfo *vi_info = pCreateInfo->pVertexInputState;

   for (uint32_t i = 0; i < vi_info->vertexBindingDescriptionCount; i++) {
      const VkVertexInputBindingDescription *desc = &vi_info->pVertexBindingDescriptions[i];

      pipeline->binding_stride[desc->binding] = desc->stride;
   }

   for (uint32_t i = 0; i < vi_info->vertexAttributeDescriptionCount; i++) {
      const VkVertexInputAttributeDescription *desc = &vi_info->pVertexAttributeDescriptions[i];

      uint32_t end = desc->offset + vk_format_get_blocksize(desc->format);
      pipeline->attrib_ends[desc->location] = end;
      pipeline->attrib_bindings[desc->location] = desc->binding;
   }

   pipeline->use_per_attribute_vb_descs = info->vs.use_per_attribute_vb_descs;
   pipeline->vb_desc_usage_mask = info->vs.vb_desc_usage_mask;
}

static struct radv_shader_variant *
radv_pipeline_get_streamout_shader(struct radv_pipeline *pipeline)
{
   int i;

   for (i = MESA_SHADER_GEOMETRY; i >= MESA_SHADER_VERTEX; i--) {
      struct radv_shader_variant *shader = radv_get_shader(pipeline, i);

      if (shader && shader->info.so.num_outputs > 0)
         return shader;
   }

   return NULL;
}

static void
radv_pipeline_init_shader_stages_state(struct radv_pipeline *pipeline)
{
   struct radv_device *device = pipeline->device;

   for (unsigned i = 0; i < MESA_SHADER_STAGES; i++) {
      pipeline->user_data_0[i] = radv_pipeline_stage_to_user_data_0(
         pipeline, i, device->physical_device->rad_info.chip_class);

      if (pipeline->shaders[i]) {
         pipeline->need_indirect_descriptor_sets |=
            pipeline->shaders[i]->info.need_indirect_descriptor_sets;
      }
   }

   struct radv_userdata_info *loc =
      radv_lookup_user_sgpr(pipeline, MESA_SHADER_VERTEX, AC_UD_VS_BASE_VERTEX_START_INSTANCE);
   if (loc->sgpr_idx != -1) {
      pipeline->graphics.vtx_base_sgpr = pipeline->user_data_0[MESA_SHADER_VERTEX];
      pipeline->graphics.vtx_base_sgpr += loc->sgpr_idx * 4;
      pipeline->graphics.vtx_emit_num = loc->num_sgprs;
      pipeline->graphics.uses_drawid =
         radv_get_shader(pipeline, MESA_SHADER_VERTEX)->info.vs.needs_draw_id;
      pipeline->graphics.uses_baseinstance =
         radv_get_shader(pipeline, MESA_SHADER_VERTEX)->info.vs.needs_base_instance;
   }
}

static VkResult
radv_pipeline_init(struct radv_pipeline *pipeline, struct radv_device *device,
                   struct radv_pipeline_cache *cache,
                   const VkGraphicsPipelineCreateInfo *pCreateInfo,
                   const struct radv_graphics_pipeline_create_info *extra)
{
   VkResult result;

   pipeline->device = device;
   pipeline->layout = radv_pipeline_layout_from_handle(pCreateInfo->layout);
   assert(pipeline->layout);

   struct radv_blend_state blend = radv_pipeline_init_blend_state(pipeline, pCreateInfo, extra);

   const VkPipelineCreationFeedbackCreateInfoEXT *creation_feedback =
      vk_find_struct_const(pCreateInfo->pNext, PIPELINE_CREATION_FEEDBACK_CREATE_INFO_EXT);
   radv_init_feedback(creation_feedback);

   VkPipelineCreationFeedbackEXT *pipeline_feedback =
      creation_feedback ? creation_feedback->pPipelineCreationFeedback : NULL;

   const VkPipelineShaderStageCreateInfo *pStages[MESA_SHADER_STAGES] = {
      0,
   };
   VkPipelineCreationFeedbackEXT *stage_feedbacks[MESA_SHADER_STAGES] = {0};
   for (uint32_t i = 0; i < pCreateInfo->stageCount; i++) {
      gl_shader_stage stage = ffs(pCreateInfo->pStages[i].stage) - 1;
      pStages[stage] = &pCreateInfo->pStages[i];
      if (creation_feedback)
         stage_feedbacks[stage] = &creation_feedback->pPipelineStageCreationFeedbacks[i];
   }

   struct radv_pipeline_key key =
      radv_generate_graphics_pipeline_key(pipeline, pCreateInfo, &blend);

   result = radv_create_shaders(pipeline, device, cache, &key, pStages, pCreateInfo->flags,
                                pipeline_feedback, stage_feedbacks);
   if (result != VK_SUCCESS)
      return result;

   pipeline->graphics.spi_baryc_cntl = S_0286E0_FRONT_FACE_ALL_BITS(1);
   radv_pipeline_init_multisample_state(pipeline, &blend, pCreateInfo);
   radv_pipeline_init_input_assembly_state(pipeline, pCreateInfo, extra);
   radv_pipeline_init_dynamic_state(pipeline, pCreateInfo, extra);
   radv_pipeline_init_raster_state(pipeline, pCreateInfo);
   radv_pipeline_init_depth_stencil_state(pipeline, pCreateInfo);

   if (pipeline->device->physical_device->rad_info.chip_class >= GFX10_3)
      gfx103_pipeline_init_vrs_state(pipeline, pCreateInfo);

   /* Ensure that some export memory is always allocated, for two reasons:
    *
    * 1) Correctness: The hardware ignores the EXEC mask if no export
    *    memory is allocated, so KILL and alpha test do not work correctly
    *    without this.
    * 2) Performance: Every shader needs at least a NULL export, even when
    *    it writes no color/depth output. The NULL export instruction
    *    stalls without this setting.
    *
    * Don't add this to CB_SHADER_MASK.
    *
    * GFX10 supports pixel shaders without exports by setting both the
    * color and Z formats to SPI_SHADER_ZERO. The hw will skip export
    * instructions if any are present.
    */
   struct radv_shader_variant *ps = pipeline->shaders[MESA_SHADER_FRAGMENT];
   if ((pipeline->device->physical_device->rad_info.chip_class <= GFX9 ||
        ps->info.ps.can_discard) &&
       !blend.spi_shader_col_format) {
      if (!ps->info.ps.writes_z && !ps->info.ps.writes_stencil && !ps->info.ps.writes_sample_mask)
         blend.spi_shader_col_format = V_028714_SPI_SHADER_32_R;
   }

   if (extra && (extra->custom_blend_mode == V_028808_CB_ELIMINATE_FAST_CLEAR ||
                 extra->custom_blend_mode == V_028808_CB_FMASK_DECOMPRESS ||
                 extra->custom_blend_mode == V_028808_CB_DCC_DECOMPRESS ||
                 extra->custom_blend_mode == V_028808_CB_RESOLVE)) {
      /* According to the CB spec states, CB_SHADER_MASK should be
       * set to enable writes to all four channels of MRT0.
       */
      blend.cb_shader_mask = 0xf;
   }

   pipeline->graphics.col_format = blend.spi_shader_col_format;
   pipeline->graphics.cb_target_mask = blend.cb_target_mask;

   if (radv_pipeline_has_gs(pipeline) && !radv_pipeline_has_ngg(pipeline)) {
      struct radv_shader_variant *gs = pipeline->shaders[MESA_SHADER_GEOMETRY];

      radv_pipeline_init_gs_ring_state(pipeline, &gs->info.gs_ring_info);
   }

   if (radv_pipeline_has_tess(pipeline)) {
      pipeline->graphics.tess_patch_control_points =
         pCreateInfo->pTessellationState->patchControlPoints;
   }

   radv_pipeline_init_vertex_input_state(pipeline, pCreateInfo);
   radv_pipeline_init_binning_state(pipeline, pCreateInfo, &blend);
   radv_pipeline_init_shader_stages_state(pipeline);
   radv_pipeline_init_scratch(device, pipeline);

   /* Find the last vertex shader stage that eventually uses streamout. */
   pipeline->streamout_shader = radv_pipeline_get_streamout_shader(pipeline);

   pipeline->graphics.is_ngg = radv_pipeline_has_ngg(pipeline);

   radv_pipeline_generate_pm4(pipeline, pCreateInfo, extra, &blend);

   return result;
}

VkResult
radv_graphics_pipeline_create(VkDevice _device, VkPipelineCache _cache,
                              const VkGraphicsPipelineCreateInfo *pCreateInfo,
                              const struct radv_graphics_pipeline_create_info *extra,
                              const VkAllocationCallbacks *pAllocator, VkPipeline *pPipeline)
{
   RADV_FROM_HANDLE(radv_device, device, _device);
   RADV_FROM_HANDLE(radv_pipeline_cache, cache, _cache);
   struct radv_pipeline *pipeline;
   VkResult result;

   pipeline = vk_zalloc2(&device->vk.alloc, pAllocator, sizeof(*pipeline), 8,
                         VK_SYSTEM_ALLOCATION_SCOPE_OBJECT);
   if (pipeline == NULL)
      return vk_error(device->instance, VK_ERROR_OUT_OF_HOST_MEMORY);

   vk_object_base_init(&device->vk, &pipeline->base, VK_OBJECT_TYPE_PIPELINE);

   result = radv_pipeline_init(pipeline, device, cache, pCreateInfo, extra);
   if (result != VK_SUCCESS) {
      radv_pipeline_destroy(device, pipeline, pAllocator);
      return result;
   }

   *pPipeline = radv_pipeline_to_handle(pipeline);

   return VK_SUCCESS;
}

VkResult
radv_CreateGraphicsPipelines(VkDevice _device, VkPipelineCache pipelineCache, uint32_t count,
                             const VkGraphicsPipelineCreateInfo *pCreateInfos,
                             const VkAllocationCallbacks *pAllocator, VkPipeline *pPipelines)
{
   VkResult result = VK_SUCCESS;
   unsigned i = 0;

   for (; i < count; i++) {
      VkResult r;
      r = radv_graphics_pipeline_create(_device, pipelineCache, &pCreateInfos[i], NULL, pAllocator,
                                        &pPipelines[i]);
      if (r != VK_SUCCESS) {
         result = r;
         pPipelines[i] = VK_NULL_HANDLE;

         if (pCreateInfos[i].flags & VK_PIPELINE_CREATE_EARLY_RETURN_ON_FAILURE_BIT_EXT)
            break;
      }
   }

   for (; i < count; ++i)
      pPipelines[i] = VK_NULL_HANDLE;

   return result;
}

static void
radv_pipeline_generate_hw_cs(struct radeon_cmdbuf *cs, const struct radv_pipeline *pipeline)
{
   struct radv_shader_variant *shader = pipeline->shaders[MESA_SHADER_COMPUTE];
   uint64_t va = radv_buffer_get_va(shader->bo) + shader->bo_offset;
   struct radv_device *device = pipeline->device;

   radeon_set_sh_reg_seq(cs, R_00B830_COMPUTE_PGM_LO, 2);
   radeon_emit(cs, va >> 8);
   radeon_emit(cs, S_00B834_DATA(va >> 40));

   radeon_set_sh_reg_seq(cs, R_00B848_COMPUTE_PGM_RSRC1, 2);
   radeon_emit(cs, shader->config.rsrc1);
   radeon_emit(cs, shader->config.rsrc2);
   if (device->physical_device->rad_info.chip_class >= GFX10) {
      radeon_set_sh_reg(cs, R_00B8A0_COMPUTE_PGM_RSRC3, shader->config.rsrc3);
   }
}

static void
radv_pipeline_generate_compute_state(struct radeon_cmdbuf *cs, const struct radv_pipeline *pipeline)
{
   struct radv_shader_variant *shader = pipeline->shaders[MESA_SHADER_COMPUTE];
   struct radv_device *device = pipeline->device;
   unsigned threads_per_threadgroup;
   unsigned threadgroups_per_cu = 1;
   unsigned waves_per_threadgroup;
   unsigned max_waves_per_sh = 0;

   /* Calculate best compute resource limits. */
   threads_per_threadgroup =
      shader->info.cs.block_size[0] * shader->info.cs.block_size[1] * shader->info.cs.block_size[2];
   waves_per_threadgroup = DIV_ROUND_UP(threads_per_threadgroup, shader->info.wave_size);

   if (device->physical_device->rad_info.chip_class >= GFX10 && waves_per_threadgroup == 1)
      threadgroups_per_cu = 2;

   radeon_set_sh_reg(
      cs, R_00B854_COMPUTE_RESOURCE_LIMITS,
      ac_get_compute_resource_limits(&device->physical_device->rad_info, waves_per_threadgroup,
                                     max_waves_per_sh, threadgroups_per_cu));

   radeon_set_sh_reg_seq(cs, R_00B81C_COMPUTE_NUM_THREAD_X, 3);
   radeon_emit(cs, S_00B81C_NUM_THREAD_FULL(shader->info.cs.block_size[0]));
   radeon_emit(cs, S_00B81C_NUM_THREAD_FULL(shader->info.cs.block_size[1]));
   radeon_emit(cs, S_00B81C_NUM_THREAD_FULL(shader->info.cs.block_size[2]));
}

static void
radv_compute_generate_pm4(struct radv_pipeline *pipeline)
{
   struct radv_device *device = pipeline->device;
   struct radeon_cmdbuf *cs = &pipeline->cs;

   cs->max_dw = device->physical_device->rad_info.chip_class >= GFX10 ? 19 : 16;
   cs->buf = malloc(cs->max_dw * 4);

   radv_pipeline_generate_hw_cs(cs, pipeline);
   radv_pipeline_generate_compute_state(cs, pipeline);

   assert(pipeline->cs.cdw <= pipeline->cs.max_dw);
}

static struct radv_pipeline_key
radv_generate_compute_pipeline_key(struct radv_pipeline *pipeline,
                                   const VkComputePipelineCreateInfo *pCreateInfo)
{
   const VkPipelineShaderStageCreateInfo *stage = &pCreateInfo->stage;
   struct radv_pipeline_key key;
   memset(&key, 0, sizeof(key));

   if (pCreateInfo->flags & VK_PIPELINE_CREATE_DISABLE_OPTIMIZATION_BIT)
      key.optimisations_disabled = 1;

   const VkPipelineShaderStageRequiredSubgroupSizeCreateInfoEXT *subgroup_size =
      vk_find_struct_const(stage->pNext,
                           PIPELINE_SHADER_STAGE_REQUIRED_SUBGROUP_SIZE_CREATE_INFO_EXT);

   if (subgroup_size) {
      assert(subgroup_size->requiredSubgroupSize == 32 ||
             subgroup_size->requiredSubgroupSize == 64);
      key.compute_subgroup_size = subgroup_size->requiredSubgroupSize;
   } else if (stage->flags & VK_PIPELINE_SHADER_STAGE_CREATE_REQUIRE_FULL_SUBGROUPS_BIT_EXT) {
      key.require_full_subgroups = true;
   }

   return key;
}

static VkResult
radv_compute_pipeline_create(VkDevice _device, VkPipelineCache _cache,
                             const VkComputePipelineCreateInfo *pCreateInfo,
                             const VkAllocationCallbacks *pAllocator, VkPipeline *pPipeline)
{
   RADV_FROM_HANDLE(radv_device, device, _device);
   RADV_FROM_HANDLE(radv_pipeline_cache, cache, _cache);
   const VkPipelineShaderStageCreateInfo *pStages[MESA_SHADER_STAGES] = {
      0,
   };
   VkPipelineCreationFeedbackEXT *stage_feedbacks[MESA_SHADER_STAGES] = {0};
   struct radv_pipeline *pipeline;
   VkResult result;

   pipeline = vk_zalloc2(&device->vk.alloc, pAllocator, sizeof(*pipeline), 8,
                         VK_SYSTEM_ALLOCATION_SCOPE_OBJECT);
   if (pipeline == NULL)
      return vk_error(device->instance, VK_ERROR_OUT_OF_HOST_MEMORY);

   vk_object_base_init(&device->vk, &pipeline->base, VK_OBJECT_TYPE_PIPELINE);

   pipeline->device = device;
   pipeline->layout = radv_pipeline_layout_from_handle(pCreateInfo->layout);
   assert(pipeline->layout);

   const VkPipelineCreationFeedbackCreateInfoEXT *creation_feedback =
      vk_find_struct_const(pCreateInfo->pNext, PIPELINE_CREATION_FEEDBACK_CREATE_INFO_EXT);
   radv_init_feedback(creation_feedback);

   VkPipelineCreationFeedbackEXT *pipeline_feedback =
      creation_feedback ? creation_feedback->pPipelineCreationFeedback : NULL;
   if (creation_feedback)
      stage_feedbacks[MESA_SHADER_COMPUTE] = &creation_feedback->pPipelineStageCreationFeedbacks[0];

   pStages[MESA_SHADER_COMPUTE] = &pCreateInfo->stage;

   struct radv_pipeline_key key = radv_generate_compute_pipeline_key(pipeline, pCreateInfo);

   result = radv_create_shaders(pipeline, device, cache, &key, pStages, pCreateInfo->flags,
                                pipeline_feedback, stage_feedbacks);
   if (result != VK_SUCCESS) {
      radv_pipeline_destroy(device, pipeline, pAllocator);
      return result;
   }

   pipeline->user_data_0[MESA_SHADER_COMPUTE] = radv_pipeline_stage_to_user_data_0(
      pipeline, MESA_SHADER_COMPUTE, device->physical_device->rad_info.chip_class);
   pipeline->need_indirect_descriptor_sets |=
      pipeline->shaders[MESA_SHADER_COMPUTE]->info.need_indirect_descriptor_sets;
   radv_pipeline_init_scratch(device, pipeline);

   radv_compute_generate_pm4(pipeline);

   *pPipeline = radv_pipeline_to_handle(pipeline);

   return VK_SUCCESS;
}

VkResult
radv_CreateComputePipelines(VkDevice _device, VkPipelineCache pipelineCache, uint32_t count,
                            const VkComputePipelineCreateInfo *pCreateInfos,
                            const VkAllocationCallbacks *pAllocator, VkPipeline *pPipelines)
{
   VkResult result = VK_SUCCESS;

   unsigned i = 0;
   for (; i < count; i++) {
      VkResult r;
      r = radv_compute_pipeline_create(_device, pipelineCache, &pCreateInfos[i], pAllocator,
                                       &pPipelines[i]);
      if (r != VK_SUCCESS) {
         result = r;
         pPipelines[i] = VK_NULL_HANDLE;

         if (pCreateInfos[i].flags & VK_PIPELINE_CREATE_EARLY_RETURN_ON_FAILURE_BIT_EXT)
            break;
      }
   }

   for (; i < count; ++i)
      pPipelines[i] = VK_NULL_HANDLE;

   return result;
}

static uint32_t
radv_get_executable_count(const struct radv_pipeline *pipeline)
{
   uint32_t ret = 0;
   for (int i = 0; i < MESA_SHADER_STAGES; ++i) {
      if (!pipeline->shaders[i])
         continue;

      if (i == MESA_SHADER_GEOMETRY && !radv_pipeline_has_ngg(pipeline)) {
         ret += 2u;
      } else {
         ret += 1u;
      }
   }
   return ret;
}

static struct radv_shader_variant *
radv_get_shader_from_executable_index(const struct radv_pipeline *pipeline, int index,
                                      gl_shader_stage *stage)
{
   for (int i = 0; i < MESA_SHADER_STAGES; ++i) {
      if (!pipeline->shaders[i])
         continue;
      if (!index) {
         *stage = i;
         return pipeline->shaders[i];
      }

      --index;

      if (i == MESA_SHADER_GEOMETRY && !radv_pipeline_has_ngg(pipeline)) {
         if (!index) {
            *stage = i;
            return pipeline->gs_copy_shader;
         }
         --index;
      }
   }

   *stage = -1;
   return NULL;
}

/* Basically strlcpy (which does not exist on linux) specialized for
 * descriptions. */
static void
desc_copy(char *desc, const char *src)
{
   int len = strlen(src);
   assert(len < VK_MAX_DESCRIPTION_SIZE);
   memcpy(desc, src, len);
   memset(desc + len, 0, VK_MAX_DESCRIPTION_SIZE - len);
}

VkResult
radv_GetPipelineExecutablePropertiesKHR(VkDevice _device, const VkPipelineInfoKHR *pPipelineInfo,
                                        uint32_t *pExecutableCount,
                                        VkPipelineExecutablePropertiesKHR *pProperties)
{
   RADV_FROM_HANDLE(radv_pipeline, pipeline, pPipelineInfo->pipeline);
   const uint32_t total_count = radv_get_executable_count(pipeline);

   if (!pProperties) {
      *pExecutableCount = total_count;
      return VK_SUCCESS;
   }

   const uint32_t count = MIN2(total_count, *pExecutableCount);
   for (unsigned i = 0, executable_idx = 0; i < MESA_SHADER_STAGES && executable_idx < count; ++i) {
      if (!pipeline->shaders[i])
         continue;
      pProperties[executable_idx].stages = mesa_to_vk_shader_stage(i);
      const char *name = NULL;
      const char *description = NULL;
      switch (i) {
      case MESA_SHADER_VERTEX:
         name = "Vertex Shader";
         description = "Vulkan Vertex Shader";
         break;
      case MESA_SHADER_TESS_CTRL:
         if (!pipeline->shaders[MESA_SHADER_VERTEX]) {
            pProperties[executable_idx].stages |= VK_SHADER_STAGE_VERTEX_BIT;
            name = "Vertex + Tessellation Control Shaders";
            description = "Combined Vulkan Vertex and Tessellation Control Shaders";
         } else {
            name = "Tessellation Control Shader";
            description = "Vulkan Tessellation Control Shader";
         }
         break;
      case MESA_SHADER_TESS_EVAL:
         name = "Tessellation Evaluation Shader";
         description = "Vulkan Tessellation Evaluation Shader";
         break;
      case MESA_SHADER_GEOMETRY:
         if (radv_pipeline_has_tess(pipeline) && !pipeline->shaders[MESA_SHADER_TESS_EVAL]) {
            pProperties[executable_idx].stages |= VK_SHADER_STAGE_TESSELLATION_EVALUATION_BIT;
            name = "Tessellation Evaluation + Geometry Shaders";
            description = "Combined Vulkan Tessellation Evaluation and Geometry Shaders";
         } else if (!radv_pipeline_has_tess(pipeline) && !pipeline->shaders[MESA_SHADER_VERTEX]) {
            pProperties[executable_idx].stages |= VK_SHADER_STAGE_VERTEX_BIT;
            name = "Vertex + Geometry Shader";
            description = "Combined Vulkan Vertex and Geometry Shaders";
         } else {
            name = "Geometry Shader";
            description = "Vulkan Geometry Shader";
         }
         break;
      case MESA_SHADER_FRAGMENT:
         name = "Fragment Shader";
         description = "Vulkan Fragment Shader";
         break;
      case MESA_SHADER_COMPUTE:
         name = "Compute Shader";
         description = "Vulkan Compute Shader";
         break;
      }

      pProperties[executable_idx].subgroupSize = pipeline->shaders[i]->info.wave_size;
      desc_copy(pProperties[executable_idx].name, name);
      desc_copy(pProperties[executable_idx].description, description);

      ++executable_idx;
      if (i == MESA_SHADER_GEOMETRY && !radv_pipeline_has_ngg(pipeline)) {
         assert(pipeline->gs_copy_shader);
         if (executable_idx >= count)
            break;

         pProperties[executable_idx].stages = VK_SHADER_STAGE_GEOMETRY_BIT;
         pProperties[executable_idx].subgroupSize = 64;
         desc_copy(pProperties[executable_idx].name, "GS Copy Shader");
         desc_copy(pProperties[executable_idx].description,
                   "Extra shader stage that loads the GS output ringbuffer into the rasterizer");

         ++executable_idx;
      }
   }

   VkResult result = *pExecutableCount < total_count ? VK_INCOMPLETE : VK_SUCCESS;
   *pExecutableCount = count;
   return result;
}

VkResult
radv_GetPipelineExecutableStatisticsKHR(VkDevice _device,
                                        const VkPipelineExecutableInfoKHR *pExecutableInfo,
                                        uint32_t *pStatisticCount,
                                        VkPipelineExecutableStatisticKHR *pStatistics)
{
   RADV_FROM_HANDLE(radv_device, device, _device);
   RADV_FROM_HANDLE(radv_pipeline, pipeline, pExecutableInfo->pipeline);
   gl_shader_stage stage;
   struct radv_shader_variant *shader =
      radv_get_shader_from_executable_index(pipeline, pExecutableInfo->executableIndex, &stage);

   enum chip_class chip_class = device->physical_device->rad_info.chip_class;
   unsigned lds_increment = chip_class >= GFX7 ? 512 : 256;
   unsigned max_waves = radv_get_max_waves(device, shader, stage);

   VkPipelineExecutableStatisticKHR *s = pStatistics;
   VkPipelineExecutableStatisticKHR *end = s + (pStatistics ? *pStatisticCount : 0);
   VkResult result = VK_SUCCESS;

   if (s < end) {
      desc_copy(s->name, "SGPRs");
      desc_copy(s->description, "Number of SGPR registers allocated per subgroup");
      s->format = VK_PIPELINE_EXECUTABLE_STATISTIC_FORMAT_UINT64_KHR;
      s->value.u64 = shader->config.num_sgprs;
   }
   ++s;

   if (s < end) {
      desc_copy(s->name, "VGPRs");
      desc_copy(s->description, "Number of VGPR registers allocated per subgroup");
      s->format = VK_PIPELINE_EXECUTABLE_STATISTIC_FORMAT_UINT64_KHR;
      s->value.u64 = shader->config.num_vgprs;
   }
   ++s;

   if (s < end) {
      desc_copy(s->name, "Spilled SGPRs");
      desc_copy(s->description, "Number of SGPR registers spilled per subgroup");
      s->format = VK_PIPELINE_EXECUTABLE_STATISTIC_FORMAT_UINT64_KHR;
      s->value.u64 = shader->config.spilled_sgprs;
   }
   ++s;

   if (s < end) {
      desc_copy(s->name, "Spilled VGPRs");
      desc_copy(s->description, "Number of VGPR registers spilled per subgroup");
      s->format = VK_PIPELINE_EXECUTABLE_STATISTIC_FORMAT_UINT64_KHR;
      s->value.u64 = shader->config.spilled_vgprs;
   }
   ++s;

   if (s < end) {
      desc_copy(s->name, "PrivMem VGPRs");
      desc_copy(s->description, "Number of VGPRs stored in private memory per subgroup");
      s->format = VK_PIPELINE_EXECUTABLE_STATISTIC_FORMAT_UINT64_KHR;
      s->value.u64 = shader->info.private_mem_vgprs;
   }
   ++s;

   if (s < end) {
      desc_copy(s->name, "Code size");
      desc_copy(s->description, "Code size in bytes");
      s->format = VK_PIPELINE_EXECUTABLE_STATISTIC_FORMAT_UINT64_KHR;
      s->value.u64 = shader->exec_size;
   }
   ++s;

   if (s < end) {
      desc_copy(s->name, "LDS size");
      desc_copy(s->description, "LDS size in bytes per workgroup");
      s->format = VK_PIPELINE_EXECUTABLE_STATISTIC_FORMAT_UINT64_KHR;
      s->value.u64 = shader->config.lds_size * lds_increment;
   }
   ++s;

   if (s < end) {
      desc_copy(s->name, "Scratch size");
      desc_copy(s->description, "Private memory in bytes per subgroup");
      s->format = VK_PIPELINE_EXECUTABLE_STATISTIC_FORMAT_UINT64_KHR;
      s->value.u64 = shader->config.scratch_bytes_per_wave;
   }
   ++s;

   if (s < end) {
      desc_copy(s->name, "Subgroups per SIMD");
      desc_copy(s->description, "The maximum number of subgroups in flight on a SIMD unit");
      s->format = VK_PIPELINE_EXECUTABLE_STATISTIC_FORMAT_UINT64_KHR;
      s->value.u64 = max_waves;
   }
   ++s;

   if (shader->statistics) {
      for (unsigned i = 0; i < aco_num_statistics; i++) {
         const struct aco_compiler_statistic_info *info = &aco_statistic_infos[i];
         if (s < end) {
            desc_copy(s->name, info->name);
            desc_copy(s->description, info->desc);
            s->format = VK_PIPELINE_EXECUTABLE_STATISTIC_FORMAT_UINT64_KHR;
            s->value.u64 = shader->statistics[i];
         }
         ++s;
      }
   }

   if (!pStatistics)
      *pStatisticCount = s - pStatistics;
   else if (s > end) {
      *pStatisticCount = end - pStatistics;
      result = VK_INCOMPLETE;
   } else {
      *pStatisticCount = s - pStatistics;
   }

   return result;
}

static VkResult
radv_copy_representation(void *data, size_t *data_size, const char *src)
{
   size_t total_size = strlen(src) + 1;

   if (!data) {
      *data_size = total_size;
      return VK_SUCCESS;
   }

   size_t size = MIN2(total_size, *data_size);

   memcpy(data, src, size);
   if (size)
      *((char *)data + size - 1) = 0;
   return size < total_size ? VK_INCOMPLETE : VK_SUCCESS;
}

VkResult
radv_GetPipelineExecutableInternalRepresentationsKHR(
   VkDevice device, const VkPipelineExecutableInfoKHR *pExecutableInfo,
   uint32_t *pInternalRepresentationCount,
   VkPipelineExecutableInternalRepresentationKHR *pInternalRepresentations)
{
   RADV_FROM_HANDLE(radv_pipeline, pipeline, pExecutableInfo->pipeline);
   gl_shader_stage stage;
   struct radv_shader_variant *shader =
      radv_get_shader_from_executable_index(pipeline, pExecutableInfo->executableIndex, &stage);

   VkPipelineExecutableInternalRepresentationKHR *p = pInternalRepresentations;
   VkPipelineExecutableInternalRepresentationKHR *end =
      p + (pInternalRepresentations ? *pInternalRepresentationCount : 0);
   VkResult result = VK_SUCCESS;
   /* optimized NIR */
   if (p < end) {
      p->isText = true;
      desc_copy(p->name, "NIR Shader(s)");
      desc_copy(p->description, "The optimized NIR shader(s)");
      if (radv_copy_representation(p->pData, &p->dataSize, shader->nir_string) != VK_SUCCESS)
         result = VK_INCOMPLETE;
   }
   ++p;

   /* backend IR */
   if (p < end) {
      p->isText = true;
      if (radv_use_llvm_for_stage(pipeline->device, stage)) {
         desc_copy(p->name, "LLVM IR");
         desc_copy(p->description, "The LLVM IR after some optimizations");
      } else {
         desc_copy(p->name, "ACO IR");
         desc_copy(p->description, "The ACO IR after some optimizations");
      }
      if (radv_copy_representation(p->pData, &p->dataSize, shader->ir_string) != VK_SUCCESS)
         result = VK_INCOMPLETE;
   }
   ++p;

   /* Disassembler */
   if (p < end) {
      p->isText = true;
      desc_copy(p->name, "Assembly");
      desc_copy(p->description, "Final Assembly");
      if (radv_copy_representation(p->pData, &p->dataSize, shader->disasm_string) != VK_SUCCESS)
         result = VK_INCOMPLETE;
   }
   ++p;

   if (!pInternalRepresentations)
      *pInternalRepresentationCount = p - pInternalRepresentations;
   else if (p > end) {
      result = VK_INCOMPLETE;
      *pInternalRepresentationCount = end - pInternalRepresentations;
   } else {
      *pInternalRepresentationCount = p - pInternalRepresentations;
   }

   return result;
}