xref: /aosp_15_r20/external/mesa3d/src/gallium/drivers/llvmpipe/lp_state_fs.c (revision 6104692788411f58d303aa86923a9ff6ecaded22)

/**************************************************************************
 *
 * Copyright 2009 VMware, Inc.
 * Copyright 2007 VMware, Inc.
 * All Rights Reserved.
 *
 * 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, sub license, 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 NON-INFRINGEMENT.
 * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS 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.
 *
 **************************************************************************/

/**
 * @file
 * Code generate the whole fragment pipeline.
 *
 * The fragment pipeline consists of the following stages:
 * - early depth test
 * - fragment shader
 * - alpha test
 * - depth/stencil test
 * - blending
 *
 * This file has only the glue to assemble the fragment pipeline.  The actual
 * plumbing of converting Gallium state into LLVM IR is done elsewhere, in the
 * lp_bld_*.[ch] files, and in a complete generic and reusable way. Here we
 * muster the LLVM JIT execution engine to create a function that follows an
 * established binary interface and that can be called from C directly.
 *
 * A big source of complexity here is that we often want to run different
 * stages with different precisions and data types and precisions. For example,
 * the fragment shader needs typically to be done in floats, but the
 * depth/stencil test and blending is better done in the type that most closely
 * matches the depth/stencil and color buffer respectively.
 *
 * Since the width of a SIMD vector register stays the same regardless of the
 * element type, different types imply different number of elements, so we must
 * code generate more instances of the stages with larger types to be able to
 * feed/consume the stages with smaller types.
 *
 * @author Jose Fonseca <[email protected]>
 */

#include <limits.h>
#include "pipe/p_defines.h"
#include "util/u_inlines.h"
#include "util/u_memory.h"
#include "util/u_pointer.h"
#include "util/format/u_format.h"
#include "util/u_dump.h"
#include "util/u_string.h"
#include "util/u_dual_blend.h"
#include "util/u_upload_mgr.h"
#include "util/os_time.h"
#include "pipe/p_shader_tokens.h"
#include "draw/draw_context.h"
#include "nir/tgsi_to_nir.h"
#include "gallivm/lp_bld_type.h"
#include "gallivm/lp_bld_const.h"
#include "gallivm/lp_bld_conv.h"
#include "gallivm/lp_bld_init.h"
#include "gallivm/lp_bld_intr.h"
#include "gallivm/lp_bld_logic.h"
#include "gallivm/lp_bld_tgsi.h"
#include "gallivm/lp_bld_nir.h"
#include "gallivm/lp_bld_swizzle.h"
#include "gallivm/lp_bld_flow.h"
#include "gallivm/lp_bld_debug.h"
#include "gallivm/lp_bld_arit.h"
#include "gallivm/lp_bld_bitarit.h"
#include "gallivm/lp_bld_pack.h"
#include "gallivm/lp_bld_format.h"
#include "gallivm/lp_bld_quad.h"
#include "gallivm/lp_bld_gather.h"
#include "gallivm/lp_bld_jit_sample.h"

#include "lp_bld_alpha.h"
#include "lp_bld_blend.h"
#include "lp_bld_depth.h"
#include "lp_bld_interp.h"
#include "lp_context.h"
#include "lp_debug.h"
#include "lp_perf.h"
#include "lp_setup.h"
#include "lp_state.h"
#include "lp_tex_sample.h"
#include "lp_flush.h"
#include "lp_state_fs.h"
#include "lp_rast.h"
#include "nir/nir_to_tgsi_info.h"

#include "lp_screen.h"
#include "compiler/nir/nir_serialize.h"
#include "util/mesa-sha1.h"


/** Fragment shader number (for debugging) */
static unsigned fs_no = 0;


static void
load_unswizzled_block(struct gallivm_state *gallivm,
                      LLVMTypeRef base_type,
                      LLVMValueRef base_ptr,
                      LLVMValueRef stride,
                      unsigned block_width,
                      unsigned block_height,
                      LLVMValueRef* dst,
                      struct lp_type dst_type,
                      unsigned dst_count,
                      unsigned dst_alignment);
/**
 * Checks if a format description is an arithmetic format
 *
 * A format which has irregular channel sizes such as R3_G3_B2 or R5_G6_B5.
 */
static inline bool
is_arithmetic_format(const struct util_format_description *format_desc)
{
   bool arith = false;

   for (unsigned i = 0; i < format_desc->nr_channels; ++i) {
      arith |= format_desc->channel[i].size != format_desc->channel[0].size;
      arith |= (format_desc->channel[i].size % 8) != 0;
   }

   return arith;
}


/**
 * Checks if this format requires special handling due to required expansion
 * to floats for blending, and furthermore has "natural" packed AoS ->
 * unpacked SoA conversion.
 */
static inline bool
format_expands_to_float_soa(const struct util_format_description *format_desc)
{
   if (format_desc->format == PIPE_FORMAT_R11G11B10_FLOAT ||
       format_desc->colorspace == UTIL_FORMAT_COLORSPACE_SRGB) {
      return true;
   }
   return false;
}


/**
 * Retrieves the type representing the memory layout for a format
 *
 * e.g. RGBA16F = 4x half-float and R3G3B2 = 1x byte
 */
static inline void
lp_mem_type_from_format_desc(const struct util_format_description *format_desc,
                             struct lp_type* type)
{
   if (format_expands_to_float_soa(format_desc)) {
      /* just make this a uint with width of block */
      type->floating = false;
      type->fixed = false;
      type->sign = false;
      type->norm = false;
      type->width = format_desc->block.bits;
      type->length = 1;
      return;
   }

   int chan = util_format_get_first_non_void_channel(format_desc->format);

   memset(type, 0, sizeof(struct lp_type));
   type->floating = format_desc->channel[chan].type == UTIL_FORMAT_TYPE_FLOAT;
   type->fixed    = format_desc->channel[chan].type == UTIL_FORMAT_TYPE_FIXED;
   type->sign     = format_desc->channel[chan].type != UTIL_FORMAT_TYPE_UNSIGNED;
   type->norm     = format_desc->channel[chan].normalized;

   if (is_arithmetic_format(format_desc)) {
      type->width = 0;
      type->length = 1;

      for (unsigned i = 0; i < format_desc->nr_channels; ++i) {
         type->width += format_desc->channel[i].size;
      }
   } else {
      type->width = format_desc->channel[chan].size;
      type->length = format_desc->nr_channels;
   }
}


/**
 * Expand the relevant bits of mask_input to a n*4-dword mask for the
 * n*four pixels in n 2x2 quads.  This will set the n*four elements of the
 * quad mask vector to 0 or ~0.
 * Grouping is 01, 23 for 2 quad mode hence only 0 and 2 are valid
 * quad arguments with fs length 8.
 *
 * \param first_quad  which quad(s) of the quad group to test, in [0,3]
 * \param mask_input  bitwise mask for the whole 4x4 stamp
 */
static LLVMValueRef
generate_quad_mask(struct gallivm_state *gallivm,
                   struct lp_type fs_type,
                   unsigned first_quad,
                   unsigned sample,
                   LLVMValueRef mask_input) /* int64 */
{
   LLVMBuilderRef builder = gallivm->builder;
   LLVMTypeRef i32t = LLVMInt32TypeInContext(gallivm->context);
   LLVMValueRef bits[16];
   LLVMValueRef mask, bits_vec;

   /*
    * XXX: We'll need a different path for 16 x u8
    */
   assert(fs_type.width == 32);
   assert(fs_type.length <= ARRAY_SIZE(bits));
   struct lp_type mask_type = lp_int_type(fs_type);

   /*
    * mask_input >>= (quad * 4)
    */
   int shift;
   switch (first_quad) {
   case 0:
      shift = 0;
      break;
   case 1:
      assert(fs_type.length == 4);
      shift = 2;
      break;
   case 2:
      shift = 8;
      break;
   case 3:
      assert(fs_type.length == 4);
      shift = 10;
      break;
   default:
      assert(0);
      shift = 0;
   }

   mask_input = LLVMBuildLShr(builder, mask_input,
                              lp_build_const_int64(gallivm, 16 * sample), "");
   mask_input = LLVMBuildTrunc(builder, mask_input, i32t, "");
   mask_input = LLVMBuildAnd(builder, mask_input,
                             lp_build_const_int32(gallivm, 0xffff), "");
   mask_input = LLVMBuildLShr(builder, mask_input,
                              LLVMConstInt(i32t, shift, 0), "");

   /*
    * mask = { mask_input & (1 << i), for i in [0,3] }
    */
   mask = lp_build_broadcast(gallivm,
                             lp_build_vec_type(gallivm, mask_type),
                             mask_input);

   for (int i = 0; i < fs_type.length / 4; i++) {
      unsigned j = 2 * (i % 2) + (i / 2) * 8;
      bits[4*i + 0] = LLVMConstInt(i32t, 1ULL << (j + 0), 0);
      bits[4*i + 1] = LLVMConstInt(i32t, 1ULL << (j + 1), 0);
      bits[4*i + 2] = LLVMConstInt(i32t, 1ULL << (j + 4), 0);
      bits[4*i + 3] = LLVMConstInt(i32t, 1ULL << (j + 5), 0);
   }
   bits_vec = LLVMConstVector(bits, fs_type.length);
   mask = LLVMBuildAnd(builder, mask, bits_vec, "");

   /*
    * mask = mask == bits ? ~0 : 0
    */
   mask = lp_build_compare(gallivm,
                           mask_type, PIPE_FUNC_EQUAL,
                           mask, bits_vec);

   return mask;
}


#define EARLY_DEPTH_TEST  0x1
#define LATE_DEPTH_TEST   0x2
#define EARLY_DEPTH_WRITE 0x4
#define LATE_DEPTH_WRITE  0x8
#define EARLY_DEPTH_TEST_INFERRED  0x10 //only with EARLY_DEPTH_TEST

static unsigned
get_cbuf_location(nir_variable *var, unsigned slot)
{
   return (var->data.location - FRAG_RESULT_DATA0) + var->data.index + slot;
}

static int
find_output_by_frag_result(struct nir_shader *shader,
                           gl_frag_result frag_result)
{
   nir_foreach_shader_out_variable(var, shader) {
      int slots = nir_variable_count_slots(var, var->type);
      for (unsigned s = 0; s < slots; s++) {
         if (var->data.location + var->data.index + s == frag_result)
            return var->data.driver_location + s;
      }
   }

   return -1;
}

/**
 * Fetch the specified lp_jit_viewport structure for a given viewport_index.
 */
static LLVMValueRef
lp_llvm_viewport(LLVMTypeRef context_type,
                 LLVMValueRef context_ptr,
                 struct gallivm_state *gallivm,
                 LLVMValueRef viewport_index)
{
   LLVMBuilderRef builder = gallivm->builder;
   LLVMValueRef ptr;
   LLVMValueRef res;
   struct lp_type viewport_type =
      lp_type_float_vec(32, 32 * LP_JIT_VIEWPORT_NUM_FIELDS);
   LLVMTypeRef vtype = lp_build_vec_type(gallivm, viewport_type);

   ptr = lp_jit_context_viewports(gallivm, context_type, context_ptr);
   ptr = LLVMBuildPointerCast(builder, ptr,
            LLVMPointerType(vtype, 0), "");

   res = lp_build_pointer_get2(builder, vtype, ptr, viewport_index);

   return res;
}


static LLVMValueRef
lp_build_depth_clamp(struct gallivm_state *gallivm,
                     LLVMBuilderRef builder,
                     bool depth_clamp,
                     bool restrict_depth,
                     struct lp_type type,
                     LLVMTypeRef context_type,
                     LLVMValueRef context_ptr,
                     LLVMTypeRef thread_data_type,
                     LLVMValueRef thread_data_ptr,
                     LLVMValueRef z)
{
   LLVMValueRef viewport, min_depth, max_depth;
   LLVMValueRef viewport_index;
   struct lp_build_context f32_bld;

   assert(type.floating);
   lp_build_context_init(&f32_bld, gallivm, type);

   if (restrict_depth)
      z = lp_build_clamp(&f32_bld, z, f32_bld.zero, f32_bld.one);

   if (!depth_clamp)
      return z;

   /*
    * Assumes clamping of the viewport index will occur in setup/gs. Value
    * is passed through the rasterization stage via lp_rast_shader_inputs.
    *
    * See: draw_clamp_viewport_idx and lp_clamp_viewport_idx for clamping
    *      semantics.
    */
   viewport_index = lp_jit_thread_data_raster_state_viewport_index(gallivm,
                                                                   thread_data_type,
                                                                   thread_data_ptr);

   /*
    * Load the min and max depth from the lp_jit_context.viewports
    * array of lp_jit_viewport structures.
    */
   viewport = lp_llvm_viewport(context_type, context_ptr, gallivm, viewport_index);

   /* viewports[viewport_index].min_depth */
   min_depth = LLVMBuildExtractElement(builder, viewport,
                  lp_build_const_int32(gallivm, LP_JIT_VIEWPORT_MIN_DEPTH), "");
   min_depth = lp_build_broadcast_scalar(&f32_bld, min_depth);

   /* viewports[viewport_index].max_depth */
   max_depth = LLVMBuildExtractElement(builder, viewport,
                  lp_build_const_int32(gallivm, LP_JIT_VIEWPORT_MAX_DEPTH), "");
   max_depth = lp_build_broadcast_scalar(&f32_bld, max_depth);

   /*
    * Clamp to the min and max depth values for the given viewport.
    */
   return lp_build_clamp(&f32_bld, z, min_depth, max_depth);
}


static void
lp_build_sample_alpha_to_coverage(struct gallivm_state *gallivm,
                                  struct lp_type type,
                                  unsigned coverage_samples,
                                  LLVMValueRef num_loop,
                                  LLVMValueRef loop_counter,
                                  LLVMTypeRef coverage_mask_type,
                                  LLVMValueRef coverage_mask_store,
                                  LLVMValueRef alpha)
{
   struct lp_build_context bld;
   LLVMBuilderRef builder = gallivm->builder;
   float step = 1.0 / coverage_samples;

   lp_build_context_init(&bld, gallivm, type);
   for (unsigned s = 0; s < coverage_samples; s++) {
      LLVMValueRef alpha_ref_value = lp_build_const_vec(gallivm, type, step * s);
      LLVMValueRef test = lp_build_cmp(&bld, PIPE_FUNC_GREATER, alpha, alpha_ref_value);

      LLVMValueRef s_mask_idx = LLVMBuildMul(builder, lp_build_const_int32(gallivm, s), num_loop, "");
      s_mask_idx = LLVMBuildAdd(builder, s_mask_idx, loop_counter, "");
      LLVMValueRef s_mask_ptr = LLVMBuildGEP2(builder, coverage_mask_type,
                                              coverage_mask_store, &s_mask_idx, 1, "");
      LLVMValueRef s_mask = LLVMBuildLoad2(builder, coverage_mask_type, s_mask_ptr, "");
      s_mask = LLVMBuildAnd(builder, s_mask, test, "");
      LLVMBuildStore(builder, s_mask, s_mask_ptr);
   }
};


struct lp_build_fs_llvm_iface {
   struct lp_build_fs_iface base;
   struct lp_build_interp_soa_context *interp;
   struct lp_build_for_loop_state *loop_state;
   LLVMTypeRef mask_type;
   LLVMValueRef mask_store;
   LLVMValueRef sample_id;
   LLVMValueRef color_ptr_ptr;
   LLVMValueRef color_stride_ptr;
   LLVMValueRef color_sample_stride_ptr;
   LLVMValueRef zs_base_ptr;
   LLVMValueRef zs_stride;
   LLVMValueRef zs_sample_stride;
   const struct lp_fragment_shader_variant_key *key;
};


static LLVMValueRef
fs_interp(const struct lp_build_fs_iface *iface,
          struct lp_build_context *bld,
          unsigned attrib, unsigned chan,
          bool centroid, bool sample,
          LLVMValueRef attrib_indir,
          LLVMValueRef offsets[2])
{
   struct lp_build_fs_llvm_iface *fs_iface = (struct lp_build_fs_llvm_iface *)iface;
   struct lp_build_interp_soa_context *interp = fs_iface->interp;
   unsigned loc = TGSI_INTERPOLATE_LOC_CENTER;
   if (centroid)
      loc = TGSI_INTERPOLATE_LOC_CENTROID;
   if (sample)
      loc = TGSI_INTERPOLATE_LOC_SAMPLE;

   return lp_build_interp_soa(interp, bld->gallivm, fs_iface->loop_state->counter,
                              fs_iface->mask_type, fs_iface->mask_store,
                              attrib, chan, loc, attrib_indir, offsets);
}


/**
 * Convert depth-stencil format to a single component one, returning
 * PIPE_FORMAT_NONE if it doesn't contain the required component.
 */
static enum pipe_format
select_zs_component_format(enum pipe_format format,
                           bool fetch_stencil)
{
   const struct util_format_description* desc = util_format_description(format);
   if (fetch_stencil && !util_format_has_stencil(desc))
      return PIPE_FORMAT_NONE;
   if (!fetch_stencil && !util_format_has_depth(desc))
      return PIPE_FORMAT_NONE;

   switch (format) {
   case PIPE_FORMAT_Z24_UNORM_S8_UINT:
      return fetch_stencil ? PIPE_FORMAT_X24S8_UINT : PIPE_FORMAT_Z24X8_UNORM;
   case PIPE_FORMAT_S8_UINT_Z24_UNORM:
      return fetch_stencil ? PIPE_FORMAT_S8X24_UINT : PIPE_FORMAT_X8Z24_UNORM;
   case PIPE_FORMAT_Z32_FLOAT_S8X24_UINT:
      return fetch_stencil ? PIPE_FORMAT_X32_S8X24_UINT : format;
   default:
      return format;
   }
}

static void
fs_fb_fetch(const struct lp_build_fs_iface *iface,
            struct lp_build_context *bld,
            int location,
            LLVMValueRef result[4])
{
   struct lp_build_fs_llvm_iface *fs_iface = (struct lp_build_fs_llvm_iface *)iface;
   struct gallivm_state *gallivm = bld->gallivm;
   LLVMBuilderRef builder = gallivm->builder;
   LLVMTypeRef int32_type = LLVMInt32TypeInContext(gallivm->context);
   LLVMTypeRef int8_type = LLVMInt8TypeInContext(gallivm->context);
   LLVMTypeRef int8p_type = LLVMPointerType(int8_type, 0);
   const struct lp_fragment_shader_variant_key *key = fs_iface->key;

   LLVMValueRef buf_ptr;
   LLVMValueRef stride;
   enum pipe_format buf_format;

   const bool fetch_stencil = location == FRAG_RESULT_STENCIL;
   const bool fetch_zs = fetch_stencil || location == FRAG_RESULT_DEPTH;
   if (fetch_zs) {
      buf_ptr = fs_iface->zs_base_ptr;
      stride = fs_iface->zs_stride;
      buf_format = select_zs_component_format(key->zsbuf_format, fetch_stencil);
   } else {
      assert(location >= FRAG_RESULT_DATA0 && location <= FRAG_RESULT_DATA7);
      const int cbuf = location - FRAG_RESULT_DATA0;
      LLVMValueRef index = lp_build_const_int32(gallivm, cbuf);

      buf_ptr = LLVMBuildLoad2(builder, int8p_type,
                               LLVMBuildGEP2(builder, int8p_type,
                                             fs_iface->color_ptr_ptr, &index, 1, ""), "");
      stride = LLVMBuildLoad2(builder, int32_type,
                              LLVMBuildGEP2(builder, int32_type,
                                            fs_iface->color_stride_ptr, &index, 1, ""), "");
      buf_format = key->cbuf_format[cbuf];
   }

   const struct util_format_description* out_format_desc = util_format_description(buf_format);
   if (out_format_desc->format == PIPE_FORMAT_NONE) {
      result[0] = result[1] = result[2] = result[3] = bld->undef;
      return;
   }

   unsigned block_size = bld->type.length;
   unsigned block_height = key->resource_1d ? 1 : 2;
   unsigned block_width = block_size / block_height;

   if (key->multisample) {
      LLVMValueRef sample_stride;

      if (fetch_zs) {
         sample_stride = fs_iface->zs_sample_stride;
      } else {
         LLVMValueRef index = lp_build_const_int32(gallivm, location - FRAG_RESULT_DATA0);
         sample_stride = LLVMBuildLoad2(builder, int32_type,
                                       LLVMBuildGEP2(builder,
                                                     int32_type,
                                                     fs_iface->color_sample_stride_ptr,
                                                     &index, 1, ""), "");
      }

      LLVMValueRef sample_offset = LLVMBuildMul(builder, sample_stride, fs_iface->sample_id, "");
      buf_ptr = LLVMBuildGEP2(builder, int8_type,
                              buf_ptr, &sample_offset, 1, "");
   }

   /* fragment shader executes on 4x4 blocks. depending on vector width it can
    * execute 2 or 4 iterations.  only move to the next row once the top row
    * has completed 8 wide 1 iteration, 4 wide 2 iterations */
   LLVMValueRef x_offset = NULL, y_offset = NULL;
   if (!key->resource_1d) {
      LLVMValueRef counter = fs_iface->loop_state->counter;

      if (block_size == 4) {
         x_offset = LLVMBuildShl(builder,
                                 LLVMBuildAnd(builder, fs_iface->loop_state->counter, lp_build_const_int32(gallivm, 1), ""),
                                 lp_build_const_int32(gallivm, 1), "");
         counter = LLVMBuildLShr(builder, fs_iface->loop_state->counter, lp_build_const_int32(gallivm, 1), "");
      }
      y_offset = LLVMBuildMul(builder, counter, lp_build_const_int32(gallivm, 2), "");
   }

   LLVMValueRef offsets[4 * 4];
   for (unsigned i = 0; i < block_size; i++) {
      unsigned x = i % block_width;
      unsigned y = i / block_width;

      if (block_size == 8) {
         /* remap the raw slots into the fragment shader execution mode. */
         /* this math took me way too long to work out, I'm sure it's
          * overkill.
          */
         x = (i & 1) + ((i >> 2) << 1);
         if (!key->resource_1d)
            y = (i & 2) >> 1;
      }

      LLVMValueRef x_val;
      if (x_offset) {
         x_val = LLVMBuildAdd(builder, lp_build_const_int32(gallivm, x), x_offset, "");
         x_val = LLVMBuildMul(builder, x_val, lp_build_const_int32(gallivm, out_format_desc->block.bits / 8), "");
      } else {
         x_val = lp_build_const_int32(gallivm, x * (out_format_desc->block.bits / 8));
      }

      LLVMValueRef y_val = lp_build_const_int32(gallivm, y);
      if (y_offset)
         y_val = LLVMBuildAdd(builder, y_val, y_offset, "");
      y_val = LLVMBuildMul(builder, y_val, stride, "");

      offsets[i] = LLVMBuildAdd(builder, x_val, y_val, "");
   }
   LLVMValueRef offset = lp_build_gather_values(gallivm, offsets, block_size);

   struct lp_type texel_type = bld->type;
   if (out_format_desc->colorspace == UTIL_FORMAT_COLORSPACE_RGB &&
       out_format_desc->channel[0].pure_integer) {
      if (out_format_desc->channel[0].type == UTIL_FORMAT_TYPE_SIGNED) {
         texel_type = lp_type_int_vec(bld->type.width, bld->type.width * bld->type.length);
      } else if (out_format_desc->channel[0].type == UTIL_FORMAT_TYPE_UNSIGNED) {
         texel_type = lp_type_uint_vec(bld->type.width, bld->type.width * bld->type.length);
      }
   } else if (fetch_stencil) {
      texel_type = lp_type_uint_vec(bld->type.width, bld->type.width * bld->type.length);
   }

   lp_build_fetch_rgba_soa(gallivm, out_format_desc, texel_type,
                           true, buf_ptr, offset,
                           NULL, NULL, NULL, result);
}

/**
 * Generate the fragment shader, depth/stencil test, and alpha tests.
 */
static void
generate_fs_loop(struct gallivm_state *gallivm,
                 struct lp_fragment_shader *shader,
                 const struct lp_fragment_shader_variant_key *key,
                 LLVMBuilderRef builder,
                 struct lp_type type,
                 LLVMTypeRef context_type,
                 LLVMValueRef context_ptr,
                 LLVMTypeRef resources_type,
                 LLVMValueRef resources_ptr,
                 LLVMTypeRef sample_pos_type,
                 LLVMValueRef sample_pos_array,
                 LLVMValueRef num_loop,
                 struct lp_build_interp_soa_context *interp,
                 const struct lp_build_sampler_soa *sampler,
                 const struct lp_build_image_soa *image,
                 LLVMTypeRef mask_type,
                 LLVMValueRef mask_store,
                 LLVMValueRef (*out_color)[4],
                 LLVMValueRef depth_base_ptr,
                 LLVMValueRef depth_stride,
                 LLVMValueRef depth_sample_stride,
                 LLVMValueRef color_ptr_ptr,
                 LLVMValueRef color_stride_ptr,
                 LLVMValueRef color_sample_stride_ptr,
                 LLVMValueRef facing,
                 LLVMTypeRef thread_data_type,
                 LLVMValueRef thread_data_ptr)
{
   struct lp_type int_type = lp_int_type(type);
   LLVMValueRef mask_ptr = NULL, mask_val = NULL;
   LLVMValueRef z;
   LLVMValueRef z_value, s_value;
   LLVMValueRef z_fb, s_fb;
   LLVMValueRef zs_samples = lp_build_const_int32(gallivm, key->zsbuf_nr_samples);
   LLVMValueRef z_out = NULL, s_out = NULL;
   struct lp_build_for_loop_state loop_state, sample_loop_state = {0};
   struct lp_build_mask_context mask;
   struct nir_shader *nir = shader->base.ir.nir;
   const bool dual_source_blend = key->blend.rt[0].blend_enable &&
                                  util_blend_state_is_dual(&key->blend, 0);
   const bool post_depth_coverage = nir->info.fs.post_depth_coverage;

   struct lp_bld_tgsi_system_values system_values;

   memset(&system_values, 0, sizeof(system_values));

   /* truncate then sign extend. */
   system_values.front_facing =
      LLVMBuildTrunc(gallivm->builder, facing,
                     LLVMInt1TypeInContext(gallivm->context), "");
   system_values.front_facing =
      LLVMBuildSExt(gallivm->builder, system_values.front_facing,
                    LLVMInt32TypeInContext(gallivm->context), "");
   system_values.view_index =
      lp_jit_thread_data_raster_state_view_index(gallivm,
                                                 thread_data_type,
                                                 thread_data_ptr);

   unsigned depth_mode;
   const struct util_format_description *zs_format_desc = NULL;
   if (key->depth.enabled ||
       key->stencil[0].enabled) {
      zs_format_desc = util_format_description(key->zsbuf_format);

      if (nir->info.fs.early_fragment_tests || nir->info.fs.post_depth_coverage) {
         depth_mode = EARLY_DEPTH_TEST | EARLY_DEPTH_WRITE;
      } else if (!(nir->info.outputs_written & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) &&
                 !(nir->info.outputs_written & BITFIELD64_BIT(FRAG_RESULT_STENCIL)) &&
                 !nir->info.fs.uses_fbfetch_output && !nir->info.writes_memory) {
         if (key->alpha.enabled ||
             key->blend.alpha_to_coverage ||
             nir->info.fs.uses_discard ||
             nir->info.outputs_written & BITFIELD64_BIT(FRAG_RESULT_SAMPLE_MASK)) {
            /* With alpha test and kill, can do the depth test early
             * and hopefully eliminate some quads.  But need to do a
             * special deferred depth write once the final mask value
             * is known. This only works though if there's either no
             * stencil test or the stencil value isn't written.
             */
            if (key->stencil[0].enabled && (key->stencil[0].writemask ||
                                            (key->stencil[1].enabled &&
                                             key->stencil[1].writemask)))
               depth_mode = LATE_DEPTH_TEST | LATE_DEPTH_WRITE;
            else
               depth_mode = EARLY_DEPTH_TEST | LATE_DEPTH_WRITE | EARLY_DEPTH_TEST_INFERRED;
         } else {
            depth_mode = EARLY_DEPTH_TEST | EARLY_DEPTH_WRITE | EARLY_DEPTH_TEST_INFERRED;
         }
      } else {
         depth_mode = LATE_DEPTH_TEST | LATE_DEPTH_WRITE;
      }

      if (!(key->depth.enabled && key->depth.writemask) &&
          !(key->stencil[0].enabled && (key->stencil[0].writemask ||
                                        (key->stencil[1].enabled &&
                                         key->stencil[1].writemask))))
         depth_mode &= ~(LATE_DEPTH_WRITE | EARLY_DEPTH_WRITE);
   } else {
      depth_mode = 0;
   }

   LLVMTypeRef vec_type = lp_build_vec_type(gallivm, type);
   LLVMTypeRef int_vec_type = lp_build_vec_type(gallivm, int_type);

   LLVMValueRef stencil_refs[2];
   stencil_refs[0] = lp_jit_context_stencil_ref_front_value(gallivm, context_type, context_ptr);
   stencil_refs[1] = lp_jit_context_stencil_ref_back_value(gallivm, context_type, context_ptr);
   /* convert scalar stencil refs into vectors */
   stencil_refs[0] = lp_build_broadcast(gallivm, int_vec_type, stencil_refs[0]);
   stencil_refs[1] = lp_build_broadcast(gallivm, int_vec_type, stencil_refs[1]);

   LLVMValueRef consts_ptr = lp_jit_resources_constants(gallivm, resources_type, resources_ptr);

   LLVMValueRef ssbo_ptr = lp_jit_resources_ssbos(gallivm, resources_type, resources_ptr);

   LLVMValueRef outputs[PIPE_MAX_SHADER_OUTPUTS][TGSI_NUM_CHANNELS];
   memset(outputs, 0, sizeof outputs);

   /* Allocate color storage for each fragment sample */
   LLVMValueRef color_store_size = num_loop;
   if (key->min_samples > 1)
      color_store_size = LLVMBuildMul(builder, num_loop, lp_build_const_int32(gallivm, key->min_samples), "");

   for (unsigned cbuf = 0; cbuf < key->nr_cbufs; cbuf++) {
      for (unsigned chan = 0; chan < TGSI_NUM_CHANNELS; ++chan) {
         out_color[cbuf][chan] = lp_build_array_alloca(gallivm,
                                                       lp_build_vec_type(gallivm,
                                                                         type),
                                                       color_store_size, "color");
      }
   }
   if (dual_source_blend) {
      assert(key->nr_cbufs <= 1);
      for (unsigned chan = 0; chan < TGSI_NUM_CHANNELS; ++chan) {
         out_color[1][chan] = lp_build_array_alloca(gallivm,
                                                    lp_build_vec_type(gallivm,
                                                                      type),
                                                    color_store_size, "color1");
      }
   }
   if (nir->info.outputs_written & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {
      z_out = lp_build_array_alloca(gallivm,
                                    lp_build_vec_type(gallivm, type),
                                    color_store_size, "depth");
   }

   if (nir->info.outputs_written & BITFIELD64_BIT(FRAG_RESULT_STENCIL)) {
      s_out = lp_build_array_alloca(gallivm,
                                    lp_build_vec_type(gallivm, type),
                                    color_store_size, "depth");
   }

   lp_build_for_loop_begin(&loop_state, gallivm,
                           lp_build_const_int32(gallivm, 0),
                           LLVMIntULT,
                           num_loop,
                           lp_build_const_int32(gallivm, 1));

   LLVMValueRef sample_mask_in;
   if (key->multisample) {
      sample_mask_in = lp_build_const_int_vec(gallivm, type, 0);
      /* create shader execution mask by combining all sample masks. */
      for (unsigned s = 0; s < key->coverage_samples; s++) {
         LLVMValueRef s_mask_idx = LLVMBuildMul(builder, num_loop, lp_build_const_int32(gallivm, s), "");
         s_mask_idx = LLVMBuildAdd(builder, s_mask_idx, loop_state.counter, "");
         LLVMValueRef s_mask = lp_build_pointer_get2(builder, mask_type, mask_store, s_mask_idx);
         if (s == 0)
            mask_val = s_mask;
         else
            mask_val = LLVMBuildOr(builder, s_mask, mask_val, "");

         LLVMValueRef mask_in = LLVMBuildAnd(builder, s_mask, lp_build_const_int_vec(gallivm, type, (1ll << s)), "");
         sample_mask_in = LLVMBuildOr(builder, sample_mask_in, mask_in, "");
      }
   } else {
      sample_mask_in = lp_build_const_int_vec(gallivm, type, 1);
      mask_ptr = LLVMBuildGEP2(builder, mask_type, mask_store,
                              &loop_state.counter, 1, "mask_ptr");
      mask_val = LLVMBuildLoad2(builder, mask_type, mask_ptr, "");

      LLVMValueRef mask_in = LLVMBuildAnd(builder, mask_val, lp_build_const_int_vec(gallivm, type, 1), "");
      sample_mask_in = LLVMBuildOr(builder, sample_mask_in, mask_in, "");
   }

   /* 'mask' will control execution based on quad's pixel alive/killed state */
   lp_build_mask_begin(&mask, gallivm, type, mask_val);

   if (!(depth_mode & EARLY_DEPTH_TEST))
      lp_build_mask_check(&mask);

   /* Create storage for recombining sample masks after early Z pass. */
   LLVMValueRef s_mask_or = lp_build_alloca(gallivm, int_vec_type, "cov_mask_early_depth");
   LLVMBuildStore(builder, LLVMConstNull(int_vec_type), s_mask_or);

   /* Create storage for post depth sample mask */
   LLVMValueRef post_depth_sample_mask_in = NULL;
   if (post_depth_coverage)
      post_depth_sample_mask_in = lp_build_alloca(gallivm, int_vec_type, "post_depth_sample_mask_in");

   LLVMValueRef s_mask = NULL, s_mask_ptr = NULL;
   LLVMValueRef z_sample_value_store = NULL, s_sample_value_store = NULL;
   LLVMValueRef z_fb_store = NULL, s_fb_store = NULL;
   LLVMTypeRef z_type = NULL, z_fb_type = NULL;

   /* Run early depth once per sample */
   if (key->multisample) {

      if (zs_format_desc) {
         struct lp_type zs_type = lp_depth_type(zs_format_desc, type.length);
         struct lp_type z_type = zs_type;
         struct lp_type s_type = zs_type;
         if (zs_format_desc->block.bits < type.width)
            z_type.width = type.width;
         if (zs_format_desc->block.bits == 8) {
            s_type.width = type.width;
         } else if (zs_format_desc->block.bits > 32) {
            z_type.width = z_type.width / 2;
            s_type.width = s_type.width / 2;
            s_type.floating = 0;
         }
         z_sample_value_store = lp_build_array_alloca(gallivm, lp_build_int_vec_type(gallivm, type),
                                                      zs_samples, "z_sample_store");
         s_sample_value_store = lp_build_array_alloca(gallivm, lp_build_int_vec_type(gallivm, type),
                                                      zs_samples, "s_sample_store");
         z_fb_store = lp_build_array_alloca(gallivm, lp_build_vec_type(gallivm, z_type),
                                            zs_samples, "z_fb_store");
         s_fb_store = lp_build_array_alloca(gallivm, lp_build_vec_type(gallivm, s_type),
                                            zs_samples, "s_fb_store");
      }
      lp_build_for_loop_begin(&sample_loop_state, gallivm,
                              lp_build_const_int32(gallivm, 0),
                              LLVMIntULT, lp_build_const_int32(gallivm, key->coverage_samples),
                              lp_build_const_int32(gallivm, 1));

      LLVMValueRef s_mask_idx = LLVMBuildMul(builder, sample_loop_state.counter, num_loop, "");
      s_mask_idx = LLVMBuildAdd(builder, s_mask_idx, loop_state.counter, "");
      s_mask_ptr = LLVMBuildGEP2(builder, mask_type, mask_store, &s_mask_idx, 1, "");

      s_mask = LLVMBuildLoad2(builder, mask_type, s_mask_ptr, "");
      s_mask = LLVMBuildAnd(builder, s_mask, mask_val, "");
   }


   /* for multisample Z needs to be interpolated at sample points for testing. */
   lp_build_interp_soa_update_pos_dyn(interp, gallivm, loop_state.counter,
                                      key->multisample
                                      ? sample_loop_state.counter : NULL);
   z = interp->pos[2];

   LLVMValueRef depth_ptr = depth_base_ptr;
   if (key->multisample) {
      LLVMValueRef sample_offset =
         LLVMBuildMul(builder, sample_loop_state.counter,
                      depth_sample_stride, "");
      depth_ptr = LLVMBuildGEP2(builder, LLVMInt8TypeInContext(gallivm->context),
                                depth_ptr, &sample_offset, 1, "");
   }

   if (depth_mode & EARLY_DEPTH_TEST) {
      z = lp_build_depth_clamp(gallivm, builder, key->depth_clamp,
                               key->restrict_depth_values, type,
                               context_type, context_ptr,
                               thread_data_type, thread_data_ptr, z);

      lp_build_depth_stencil_load_swizzled(gallivm, type,
                                           zs_format_desc, key->resource_1d,
                                           depth_ptr, depth_stride,
                                           &z_fb, &s_fb, loop_state.counter);
      lp_build_depth_stencil_test(gallivm,
                                  &key->depth,
                                  key->stencil,
                                  type,
                                  zs_format_desc,
                                  key->multisample ? NULL : &mask,
                                  &s_mask,
                                  stencil_refs,
                                  z, z_fb, s_fb,
                                  facing,
                                  &z_value, &s_value,
                                  !key->multisample,
                                  key->restrict_depth_values);

      if (depth_mode & EARLY_DEPTH_WRITE) {
         lp_build_depth_stencil_write_swizzled(gallivm, type,
                                               zs_format_desc, key->resource_1d,
                                               NULL, NULL, NULL, loop_state.counter,
                                               depth_ptr, depth_stride,
                                               z_value, s_value);
      }
      /*
       * Note mask check if stencil is enabled must be after ds write not
       * after stencil test otherwise new stencil values may not get written
       * if all fragments got killed by depth/stencil test.
       */
      if (key->stencil[0].enabled && !key->multisample)
         lp_build_mask_check(&mask);

      if (key->multisample) {
         z_fb_type = LLVMTypeOf(z_fb);
         z_type = LLVMTypeOf(z_value);
         lp_build_pointer_set(builder, z_sample_value_store, sample_loop_state.counter, LLVMBuildBitCast(builder, z_value, lp_build_int_vec_type(gallivm, type), ""));
         lp_build_pointer_set(builder, s_sample_value_store, sample_loop_state.counter, LLVMBuildBitCast(builder, s_value, lp_build_int_vec_type(gallivm, type), ""));
         lp_build_pointer_set(builder, z_fb_store, sample_loop_state.counter, z_fb);
         lp_build_pointer_set(builder, s_fb_store, sample_loop_state.counter, s_fb);
      }
      if (key->occlusion_count && !(depth_mode & EARLY_DEPTH_TEST_INFERRED)) {
         LLVMValueRef counter = lp_jit_thread_data_vis_counter(gallivm, thread_data_type, thread_data_ptr);
         lp_build_name(counter, "counter");
         lp_build_occlusion_count(gallivm, type,
                                 key->multisample ? s_mask : lp_build_mask_value(&mask), counter);
      }
   }

   if (key->multisample) {
      /*
       * Store the post-early Z coverage mask.
       * Recombine the resulting coverage masks post early Z into the fragment
       * shader execution mask.
       */
      LLVMValueRef tmp_s_mask_or = LLVMBuildLoad2(builder, int_vec_type, s_mask_or, "");
      tmp_s_mask_or = LLVMBuildOr(builder, tmp_s_mask_or, s_mask, "");
      LLVMBuildStore(builder, tmp_s_mask_or, s_mask_or);

      if (post_depth_coverage) {
         LLVMValueRef mask_bit_idx = LLVMBuildShl(builder, lp_build_const_int32(gallivm, 1), sample_loop_state.counter, "");
         LLVMValueRef post_depth_mask_in = LLVMBuildLoad2(builder, int_vec_type, post_depth_sample_mask_in, "");
         mask_bit_idx = LLVMBuildAnd(builder, s_mask, lp_build_broadcast(gallivm, int_vec_type, mask_bit_idx), "");
         post_depth_mask_in = LLVMBuildOr(builder, post_depth_mask_in, mask_bit_idx, "");
         LLVMBuildStore(builder, post_depth_mask_in, post_depth_sample_mask_in);
      }

      LLVMBuildStore(builder, s_mask, s_mask_ptr);

      lp_build_for_loop_end(&sample_loop_state);

      /* recombined all the coverage masks in the shader exec mask. */
      tmp_s_mask_or = LLVMBuildLoad2(builder, int_vec_type, s_mask_or, "");
      lp_build_mask_update(&mask, tmp_s_mask_or);

      if (key->min_samples == 1) {
         /* for multisample Z needs to be re interpolated at pixel center */
         lp_build_interp_soa_update_pos_dyn(interp, gallivm, loop_state.counter, NULL);
         z = interp->pos[2];
         lp_build_mask_update(&mask, tmp_s_mask_or);
      }
   } else {
      if (post_depth_coverage) {
         LLVMValueRef post_depth_mask_in = LLVMBuildAnd(builder, lp_build_mask_value(&mask), lp_build_const_int_vec(gallivm, type, 1), "");
         LLVMBuildStore(builder, post_depth_mask_in, post_depth_sample_mask_in);
      }
   }

   LLVMValueRef out_sample_mask_storage = NULL;
   if (nir->info.outputs_written & BITFIELD64_BIT(FRAG_RESULT_SAMPLE_MASK)) {
      out_sample_mask_storage = lp_build_alloca(gallivm, int_vec_type, "write_mask");
      if (key->min_samples > 1)
         LLVMBuildStore(builder, LLVMConstNull(int_vec_type), out_sample_mask_storage);
   }

   if (post_depth_coverage) {
      system_values.sample_mask_in = LLVMBuildLoad2(builder, int_vec_type, post_depth_sample_mask_in, "");
   } else {
      system_values.sample_mask_in = sample_mask_in;
   }
   if (key->multisample && key->min_samples > 1) {
      lp_build_for_loop_begin(&sample_loop_state, gallivm,
                              lp_build_const_int32(gallivm, 0),
                              LLVMIntULT,
                              lp_build_const_int32(gallivm, key->min_samples),
                              lp_build_const_int32(gallivm, 1));

      LLVMValueRef s_mask_idx = LLVMBuildMul(builder, sample_loop_state.counter, num_loop, "");
      s_mask_idx = LLVMBuildAdd(builder, s_mask_idx, loop_state.counter, "");
      s_mask_ptr = LLVMBuildGEP2(builder, mask_type, mask_store, &s_mask_idx, 1, "");
      s_mask = LLVMBuildLoad2(builder, mask_type, s_mask_ptr, "");
      lp_build_mask_force(&mask, s_mask);
      lp_build_interp_soa_update_pos_dyn(interp, gallivm, loop_state.counter, sample_loop_state.counter);
      system_values.sample_id = sample_loop_state.counter;
      system_values.sample_mask_in = LLVMBuildAnd(builder, system_values.sample_mask_in,
                                                  lp_build_broadcast(gallivm, int_vec_type,
                                                                     LLVMBuildShl(builder, lp_build_const_int32(gallivm, 1), sample_loop_state.counter, "")), "");
   } else {
      system_values.sample_id = lp_build_const_int32(gallivm, 0);

   }
   system_values.sample_pos = sample_pos_array;
   system_values.sample_pos_type = sample_pos_type;

   lp_build_interp_soa_update_inputs_dyn(interp, gallivm, loop_state.counter,
                                         mask_type, mask_store, sample_loop_state.counter);

   struct lp_build_fs_llvm_iface fs_iface = {
     .base.interp_fn = fs_interp,
     .base.fb_fetch = fs_fb_fetch,
     .interp = interp,
     .loop_state = &loop_state,
     .sample_id = system_values.sample_id,
     .mask_type = mask_type,
     .mask_store = mask_store,
     .color_ptr_ptr = color_ptr_ptr,
     .color_stride_ptr = color_stride_ptr,
     .color_sample_stride_ptr = color_sample_stride_ptr,
     .zs_base_ptr = depth_base_ptr,
     .zs_stride = depth_stride,
     .zs_sample_stride = depth_sample_stride,
     .key = key,
   };

   struct lp_build_tgsi_params params;
   memset(&params, 0, sizeof(params));

   params.type = type;
   params.mask = &mask;
   params.fs_iface = &fs_iface.base;
   params.consts_ptr = consts_ptr;
   params.system_values = &system_values;
   params.inputs = interp->inputs;
   params.num_inputs = interp->num_attribs - 1;
   params.context_type = context_type;
   params.context_ptr = context_ptr;
   params.resources_type = resources_type;
   params.resources_ptr = resources_ptr;
   params.thread_data_type = thread_data_type;
   params.thread_data_ptr = thread_data_ptr;
   params.sampler = sampler;
   params.info = &shader->info.base;
   params.ssbo_ptr = ssbo_ptr;
   params.image = image;
   params.aniso_filter_table = lp_jit_resources_aniso_filter_table(gallivm, resources_type, resources_ptr);

   /* Build the actual shader */
   lp_build_nir_soa(gallivm, nir, &params, outputs);

   /*
    * Must not count ps invocations if there's a null shader.
    * (It would be ok to count with null shader if there's d/s tests,
    * but only if there's d/s buffers too, which is different
    * to implicit rasterization disable which must not depend
    * on the d/s buffers.)
    * Could disable if there's no stats query, but maybe not worth it.
    */
   if (shader->info.base.num_instructions > 1) {
      LLVMValueRef invocs = lp_jit_thread_data_ps_invocations(gallivm, thread_data_type, thread_data_ptr);
      lp_build_occlusion_count(gallivm, type, lp_build_mask_value(&mask), invocs);
   }

   /* Alpha test */
   if (key->alpha.enabled) {
      int color0 = find_output_by_frag_result(nir, FRAG_RESULT_DATA0);

      if (color0 != -1 && outputs[color0][3]) {
         const struct util_format_description *cbuf_format_desc;
         LLVMValueRef alpha = LLVMBuildLoad2(builder, vec_type, outputs[color0][3], "alpha");
         LLVMValueRef alpha_ref_value;

         alpha_ref_value = lp_jit_context_alpha_ref_value(gallivm, context_type, context_ptr);
         alpha_ref_value = lp_build_broadcast(gallivm, vec_type, alpha_ref_value);

         cbuf_format_desc = util_format_description(key->cbuf_format[0]);

         lp_build_alpha_test(gallivm, key->alpha.func, type, cbuf_format_desc,
                             &mask, alpha, alpha_ref_value,
                             ((depth_mode & LATE_DEPTH_TEST) != 0) && !key->multisample);
      }
   }

   /* Emulate Alpha to Coverage with Alpha test */
   if (key->blend.alpha_to_coverage) {
      int color0 = find_output_by_frag_result(nir, FRAG_RESULT_DATA0);

      if (color0 != -1 && outputs[color0][3]) {
         LLVMValueRef alpha = LLVMBuildLoad2(builder, vec_type, outputs[color0][3], "alpha");

         if (!key->multisample) {
            lp_build_alpha_to_coverage(gallivm, type,
                                       &mask, alpha,
                                       (depth_mode & LATE_DEPTH_TEST) != 0);
         } else {
            lp_build_sample_alpha_to_coverage(gallivm, type, key->coverage_samples, num_loop,
                                              loop_state.counter,
                                              mask_type, mask_store, alpha);
         }
      }
   }

   if (key->blend.alpha_to_one) {
      nir_foreach_shader_out_variable(var, nir) {
         if (var->data.location < FRAG_RESULT_DATA0)
            continue;
         int slots = nir_variable_count_slots(var, var->type);
         for (unsigned s = 0; s < slots; s++) {
            unsigned cbuf = get_cbuf_location(var, s);
            if ((cbuf < key->nr_cbufs) || (cbuf == 1 && dual_source_blend))
               if (outputs[cbuf][3]) {
                  LLVMBuildStore(builder, lp_build_const_vec(gallivm, type, 1.0),
                                 outputs[cbuf][3]);
               }
         }
      }
   }

   if (nir->info.outputs_written & BITFIELD64_BIT(FRAG_RESULT_SAMPLE_MASK)) {
      LLVMValueRef output_smask = NULL;
      int smaski = find_output_by_frag_result(nir, FRAG_RESULT_SAMPLE_MASK);

      struct lp_build_context smask_bld;
      lp_build_context_init(&smask_bld, gallivm, int_type);

      assert(smaski >= 0);
      output_smask = LLVMBuildLoad2(builder, vec_type, outputs[smaski][0], "smask");
      output_smask = LLVMBuildBitCast(builder, output_smask, smask_bld.vec_type, "");
      if (!key->multisample && key->no_ms_sample_mask_out) {
         output_smask = lp_build_and(&smask_bld, output_smask, smask_bld.one);
         output_smask = lp_build_cmp(&smask_bld, PIPE_FUNC_NOTEQUAL, output_smask, smask_bld.zero);
         lp_build_mask_update(&mask, output_smask);
      }

      if (key->min_samples > 1) {
         /* only the bit corresponding to this sample is to be used. */
         LLVMValueRef tmp_mask = LLVMBuildLoad2(builder, int_vec_type, out_sample_mask_storage, "tmp_mask");
         LLVMValueRef out_smask_idx = LLVMBuildShl(builder, lp_build_const_int32(gallivm, 1), sample_loop_state.counter, "");
         LLVMValueRef smask_bit = LLVMBuildAnd(builder, output_smask, lp_build_broadcast(gallivm, int_vec_type, out_smask_idx), "");
         output_smask = LLVMBuildOr(builder, tmp_mask, smask_bit, "");
      }

      LLVMBuildStore(builder, output_smask, out_sample_mask_storage);
   }

   if (nir->info.outputs_written & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {
      int pos0 = find_output_by_frag_result(nir, FRAG_RESULT_DEPTH);

      LLVMValueRef out = LLVMBuildLoad2(builder, vec_type, outputs[pos0][2], "");
      LLVMValueRef idx = loop_state.counter;
      if (key->min_samples > 1)
         idx = LLVMBuildAdd(builder, idx,
                            LLVMBuildMul(builder, sample_loop_state.counter, num_loop, ""), "");
      LLVMValueRef ptr = LLVMBuildGEP2(builder, vec_type, z_out, &idx, 1, "");
      LLVMBuildStore(builder, out, ptr);
   }

   if (nir->info.outputs_written & BITFIELD64_BIT(FRAG_RESULT_STENCIL)) {
      int sten_out = find_output_by_frag_result(nir, FRAG_RESULT_STENCIL);

      LLVMValueRef out = LLVMBuildLoad2(builder, vec_type,
                                        outputs[sten_out][1], "output.s");
      LLVMValueRef idx = loop_state.counter;
      if (key->min_samples > 1)
         idx = LLVMBuildAdd(builder, idx,
                            LLVMBuildMul(builder, sample_loop_state.counter, num_loop, ""), "");
      LLVMValueRef ptr = LLVMBuildGEP2(builder, vec_type, s_out, &idx, 1, "");
      LLVMBuildStore(builder, out, ptr);
   }

   bool has_cbuf0_write = false;
   /* Color write - per fragment sample */
   nir_foreach_shader_out_variable(var, nir) {
      if (var->data.location < FRAG_RESULT_DATA0)
         continue;
      int slots = nir_variable_count_slots(var, var->type);

      for (unsigned s = 0; s < slots; s++) {
         unsigned cbuf = get_cbuf_location(var, s);
         unsigned attrib = var->data.driver_location + s;
         if ((cbuf < key->nr_cbufs) || (cbuf == 1 && dual_source_blend)) {
            if (cbuf == 0) {
               /* XXX: there is an edge case with FB fetch where gl_FragColor and
                * gl_LastFragData[0] are used together. This creates both
                * FRAG_RESULT_COLOR and FRAG_RESULT_DATA* output variables. This
                * loop then writes to cbuf 0 twice, owerwriting the correct value
                * from gl_FragColor with some garbage. This case is excercised in
                * one of deqp tests.  A similar bug can happen if
                * gl_SecondaryFragColorEXT and gl_LastFragData[1] are mixed in
                * the same fashion...  This workaround will break if
                * gl_LastFragData[0] goes in outputs list before
                * gl_FragColor. This doesn't seem to happen though.
                */
               if (has_cbuf0_write)
                  continue;
               has_cbuf0_write = true;
            }

            for (unsigned chan = 0; chan < TGSI_NUM_CHANNELS; ++chan) {
               if (outputs[attrib][chan]) {
                  /* XXX: just initialize outputs to point at colors[] and
                   * skip this.
                   */
                  LLVMValueRef out = LLVMBuildLoad2(builder, vec_type, outputs[attrib][chan], "");
                  LLVMValueRef color_ptr;
                  LLVMValueRef color_idx = loop_state.counter;
                  if (key->min_samples > 1)
                     color_idx = LLVMBuildAdd(builder, color_idx,
                                              LLVMBuildMul(builder, sample_loop_state.counter, num_loop, ""), "");
                  color_ptr = LLVMBuildGEP2(builder, vec_type, out_color[cbuf][chan],
                                            &color_idx, 1, "");
                  lp_build_name(out, "color%u.%c", attrib, "rgba"[chan]);
                  LLVMBuildStore(builder, out, color_ptr);
               }
            }
         }
      }
   }

   if (key->multisample && key->min_samples > 1) {
      LLVMBuildStore(builder, lp_build_mask_value(&mask), s_mask_ptr);
      lp_build_for_loop_end(&sample_loop_state);
   }

   if (key->multisample) {
      /* execute depth test for each sample */
      lp_build_for_loop_begin(&sample_loop_state, gallivm,
                              lp_build_const_int32(gallivm, 0),
                              LLVMIntULT, lp_build_const_int32(gallivm, key->coverage_samples),
                              lp_build_const_int32(gallivm, 1));

      /* load the per-sample coverage mask */
      LLVMValueRef s_mask_idx = LLVMBuildMul(builder, sample_loop_state.counter, num_loop, "");
      s_mask_idx = LLVMBuildAdd(builder, s_mask_idx, loop_state.counter, "");
      s_mask_ptr = LLVMBuildGEP2(builder, mask_type, mask_store, &s_mask_idx, 1, "");

      /* combine the execution mask post fragment shader with the coverage mask. */
      s_mask = LLVMBuildLoad2(builder, mask_type, s_mask_ptr, "");
      if (key->min_samples == 1)
         s_mask = LLVMBuildAnd(builder, s_mask, lp_build_mask_value(&mask), "");

      /* if the shader writes sample mask use that,
       * but only if this isn't genuine early-depth to avoid breaking occlusion query */
      if (nir->info.outputs_written & BITFIELD64_BIT(FRAG_RESULT_SAMPLE_MASK) &&
          (!(depth_mode & EARLY_DEPTH_TEST) || (depth_mode & (EARLY_DEPTH_TEST_INFERRED)))) {
         LLVMValueRef out_smask_idx = LLVMBuildShl(builder, lp_build_const_int32(gallivm, 1), sample_loop_state.counter, "");
         out_smask_idx = lp_build_broadcast(gallivm, int_vec_type, out_smask_idx);
         LLVMValueRef output_smask = LLVMBuildLoad2(builder, int_vec_type, out_sample_mask_storage, "");
         LLVMValueRef smask_bit = LLVMBuildAnd(builder, output_smask, out_smask_idx, "");
         LLVMValueRef cmp = LLVMBuildICmp(builder, LLVMIntNE, smask_bit, lp_build_const_int_vec(gallivm, int_type, 0), "");
         smask_bit = LLVMBuildSExt(builder, cmp, int_vec_type, "");

         s_mask = LLVMBuildAnd(builder, s_mask, smask_bit, "");
      }
   }

   depth_ptr = depth_base_ptr;
   if (key->multisample) {
      LLVMValueRef sample_offset = LLVMBuildMul(builder, sample_loop_state.counter, depth_sample_stride, "");
      depth_ptr = LLVMBuildGEP2(builder, LLVMInt8TypeInContext(gallivm->context),
                                depth_ptr, &sample_offset, 1, "");
   }

   /* Late Z test */
   if (depth_mode & LATE_DEPTH_TEST) {
      if (nir->info.outputs_written & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {
         LLVMValueRef idx = loop_state.counter;
         if (key->min_samples > 1)
            idx = LLVMBuildAdd(builder, idx,
                               LLVMBuildMul(builder, sample_loop_state.counter, num_loop, ""), "");
         LLVMValueRef ptr = LLVMBuildGEP2(builder, vec_type, z_out, &idx, 1, "");
         z = LLVMBuildLoad2(builder, vec_type, ptr, "output.z");
      } else {
         if (key->multisample) {
            lp_build_interp_soa_update_pos_dyn(interp, gallivm, loop_state.counter, key->multisample ? sample_loop_state.counter : NULL);
            z = interp->pos[2];
         }
      }

      /*
       * Clamp according to ARB_depth_clamp semantics.
       */
      z = lp_build_depth_clamp(gallivm, builder, key->depth_clamp,
                               key->restrict_depth_values, type,
                               context_type, context_ptr,
                               thread_data_type, thread_data_ptr, z);

      if (nir->info.outputs_written & BITFIELD64_BIT(FRAG_RESULT_STENCIL)) {
         LLVMValueRef idx = loop_state.counter;
         if (key->min_samples > 1)
            idx = LLVMBuildAdd(builder, idx,
                               LLVMBuildMul(builder, sample_loop_state.counter, num_loop, ""), "");
         LLVMValueRef ptr = LLVMBuildGEP2(builder, vec_type, s_out, &idx, 1, "");
         stencil_refs[0] = LLVMBuildLoad2(builder, vec_type, ptr, "output.s");
         /* there's only one value, and spec says to discard additional bits */
         LLVMValueRef s_max_mask = lp_build_const_int_vec(gallivm, int_type, 255);
         stencil_refs[0] = LLVMBuildBitCast(builder, stencil_refs[0], int_vec_type, "");
         stencil_refs[0] = LLVMBuildAnd(builder, stencil_refs[0], s_max_mask, "");
         stencil_refs[1] = stencil_refs[0];
      }

      lp_build_depth_stencil_load_swizzled(gallivm, type,
                                           zs_format_desc, key->resource_1d,
                                           depth_ptr, depth_stride,
                                           &z_fb, &s_fb, loop_state.counter);

      lp_build_depth_stencil_test(gallivm,
                                  &key->depth,
                                  key->stencil,
                                  type,
                                  zs_format_desc,
                                  key->multisample ? NULL : &mask,
                                  &s_mask,
                                  stencil_refs,
                                  z, z_fb, s_fb,
                                  facing,
                                  &z_value, &s_value,
                                  false,
                                  key->restrict_depth_values);
      /* Late Z write */
      if (depth_mode & LATE_DEPTH_WRITE) {
         lp_build_depth_stencil_write_swizzled(gallivm, type,
                                               zs_format_desc, key->resource_1d,
                                               NULL, NULL, NULL, loop_state.counter,
                                               depth_ptr, depth_stride,
                                               z_value, s_value);
      }
   } else if ((depth_mode & EARLY_DEPTH_TEST) &&
              (depth_mode & LATE_DEPTH_WRITE)) {
      /* Need to apply a reduced mask to the depth write.  Reload the
       * depth value, update from zs_value with the new mask value and
       * write that out.
       */
      if (key->multisample) {
         z_value = LLVMBuildBitCast(builder, lp_build_pointer_get2(builder, int_vec_type, z_sample_value_store, sample_loop_state.counter), z_type, "");
         s_value = lp_build_pointer_get2(builder, int_vec_type, s_sample_value_store, sample_loop_state.counter);
         z_fb = LLVMBuildBitCast(builder, lp_build_pointer_get2(builder, int_vec_type, z_fb_store, sample_loop_state.counter), z_fb_type, "");
         s_fb = lp_build_pointer_get2(builder, int_vec_type, s_fb_store, sample_loop_state.counter);
      }
      lp_build_depth_stencil_write_swizzled(gallivm, type,
                                            zs_format_desc, key->resource_1d,
                                            key->multisample ? s_mask : lp_build_mask_value(&mask), z_fb, s_fb, loop_state.counter,
                                            depth_ptr, depth_stride,
                                            z_value, s_value);
   }

   if (key->occlusion_count && (!(depth_mode & EARLY_DEPTH_TEST) || (depth_mode & EARLY_DEPTH_TEST_INFERRED))) {
      LLVMValueRef counter = lp_jit_thread_data_vis_counter(gallivm, thread_data_type, thread_data_ptr);
      lp_build_name(counter, "counter");

      lp_build_occlusion_count(gallivm, type,
                               key->multisample ? s_mask : lp_build_mask_value(&mask), counter);
   }

   /* if this is genuine early-depth in the shader, write samplemask now
    * after occlusion count has been updated
    */
   if (key->multisample &&
       nir->info.outputs_written & BITFIELD64_BIT(FRAG_RESULT_SAMPLE_MASK) &&
       (depth_mode & (EARLY_DEPTH_TEST_INFERRED | EARLY_DEPTH_TEST)) == EARLY_DEPTH_TEST) {
      /* if the shader writes sample mask use that */
         LLVMValueRef out_smask_idx = LLVMBuildShl(builder, lp_build_const_int32(gallivm, 1), sample_loop_state.counter, "");
         out_smask_idx = lp_build_broadcast(gallivm, int_vec_type, out_smask_idx);
         LLVMValueRef output_smask = LLVMBuildLoad2(builder, int_vec_type, out_sample_mask_storage, "");
         LLVMValueRef smask_bit = LLVMBuildAnd(builder, output_smask, out_smask_idx, "");
         LLVMValueRef cmp = LLVMBuildICmp(builder, LLVMIntNE, smask_bit, lp_build_const_int_vec(gallivm, int_type, 0), "");
         smask_bit = LLVMBuildSExt(builder, cmp, int_vec_type, "");

         s_mask = LLVMBuildAnd(builder, s_mask, smask_bit, "");
   }


   if (key->multisample) {
      /* store the sample mask for this loop */
      LLVMBuildStore(builder, s_mask, s_mask_ptr);
      lp_build_for_loop_end(&sample_loop_state);
   }

   mask_val = lp_build_mask_end(&mask);
   if (!key->multisample)
      LLVMBuildStore(builder, mask_val, mask_ptr);
   lp_build_for_loop_end(&loop_state);
}


/**
 * This function will reorder pixels from the fragment shader SoA to memory
 * layout AoS
 *
 * Fragment Shader outputs pixels in small 2x2 blocks
 *  e.g. (0, 0), (1, 0), (0, 1), (1, 1) ; (2, 0) ...
 *
 * However in memory pixels are stored in rows
 *  e.g. (0, 0), (1, 0), (2, 0), (3, 0) ; (0, 1) ...
 *
 * @param type            fragment shader type (4x or 8x float)
 * @param num_fs          number of fs_src
 * @param is_1d           whether we're outputting to a 1d resource
 * @param dst_channels    number of output channels
 * @param fs_src          output from fragment shader
 * @param dst             pointer to store result
 * @param pad_inline      is channel padding inline or at end of row
 * @return                the number of dsts
 */
static int
generate_fs_twiddle(struct gallivm_state *gallivm,
                    struct lp_type type,
                    unsigned num_fs,
                    unsigned dst_channels,
                    LLVMValueRef fs_src[][4],
                    LLVMValueRef* dst,
                    bool pad_inline)
{
   LLVMValueRef src[16];
   unsigned pixels = type.length / 4;
   unsigned src_channels = dst_channels < 3 ? dst_channels : 4;
   unsigned src_count = num_fs * src_channels;

   assert(pixels == 2 || pixels == 1);
   assert(num_fs * src_channels <= ARRAY_SIZE(src));

   /*
    * Transpose from SoA -> AoS
    */
   for (unsigned i = 0; i < num_fs; ++i) {
      lp_build_transpose_aos_n(gallivm, type, &fs_src[i][0], src_channels,
                               &src[i * src_channels]);
   }

   /*
    * Pick transformation options
    */
   bool swizzle_pad = false;
   bool twiddle = false;
   bool split = false;
   unsigned reorder_group = 0;

   if (dst_channels == 1) {
      twiddle = true;
      if (pixels == 2) {
         split = true;
      }
   } else if (dst_channels == 2) {
      if (pixels == 1) {
         reorder_group = 1;
      }
   } else if (dst_channels > 2) {
      if (pixels == 1) {
         reorder_group = 2;
      } else {
         twiddle = true;
      }

      if (!pad_inline && dst_channels == 3 && pixels > 1) {
         swizzle_pad = true;
      }
   }

   /*
    * Split the src in half
    */
   if (split) {
      for (unsigned i = num_fs; i > 0; --i) {
         src[(i - 1)*2 + 1] = lp_build_extract_range(gallivm, src[i - 1], 4, 4);
         src[(i - 1)*2 + 0] = lp_build_extract_range(gallivm, src[i - 1], 0, 4);
      }

      src_count *= 2;
      type.length = 4;
   }

   /*
    * Ensure pixels are in memory order
    */
   if (reorder_group) {
      /* Twiddle pixels by reordering the array, e.g.:
       *
       * src_count =  8 -> 0 2 1 3 4 6 5 7
       * src_count = 16 -> 0 1 4 5 2 3 6 7 8 9 12 13 10 11 14 15
       */
      const unsigned reorder_sw[] = { 0, 2, 1, 3 };

      for (unsigned i = 0; i < src_count; ++i) {
         unsigned group = i / reorder_group;
         unsigned block = (group / 4) * 4 * reorder_group;
         unsigned j = block + (reorder_sw[group % 4] * reorder_group) + (i % reorder_group);
         dst[i] = src[j];
      }
   } else if (twiddle) {
      /* Twiddle pixels across elements of array */
      /*
       * XXX: we should avoid this in some cases, but would need to tell
       * lp_build_conv to reorder (or deal with it ourselves).
       */
      lp_bld_quad_twiddle(gallivm, type, src, src_count, dst);
   } else {
      /* Do nothing */
      memcpy(dst, src, sizeof(LLVMValueRef) * src_count);
   }

   /*
    * Moves any padding between pixels to the end
    * e.g. RGBXRGBX -> RGBRGBXX
    */
   if (swizzle_pad) {
      unsigned char swizzles[16];
      unsigned elems = pixels * dst_channels;

      for (unsigned i = 0; i < type.length; ++i) {
         if (i < elems)
            swizzles[i] = i % dst_channels + (i / dst_channels) * 4;
         else
            swizzles[i] = LP_BLD_SWIZZLE_DONTCARE;
      }

      for (unsigned i = 0; i < src_count; ++i) {
         dst[i] = lp_build_swizzle_aos_n(gallivm, dst[i], swizzles,
                                         type.length, type.length);
      }
   }

   return src_count;
}


/*
 * Untwiddle and transpose, much like the above.
 * However, this is after conversion, so we get packed vectors.
 * At this time only handle 4x16i8 rgba / 2x16i8 rg / 1x16i8 r data,
 * the vectors will look like:
 * r0r1r4r5r2r3r6r7r8r9r12... (albeit color channels may
 * be swizzled here). Extending to 16bit should be trivial.
 * Should also be extended to handle twice wide vectors with AVX2...
 */
static void
fs_twiddle_transpose(struct gallivm_state *gallivm,
                     struct lp_type type,
                     LLVMValueRef *src,
                     unsigned src_count,
                     LLVMValueRef *dst)
{
   struct lp_type type64, type16, type32;
   LLVMTypeRef type64_t, type8_t, type16_t, type32_t;
   LLVMBuilderRef builder = gallivm->builder;
   LLVMValueRef tmp[4], shuf[8];
   for (unsigned j = 0; j < 2; j++) {
      shuf[j*4 + 0] = lp_build_const_int32(gallivm, j*4 + 0);
      shuf[j*4 + 1] = lp_build_const_int32(gallivm, j*4 + 2);
      shuf[j*4 + 2] = lp_build_const_int32(gallivm, j*4 + 1);
      shuf[j*4 + 3] = lp_build_const_int32(gallivm, j*4 + 3);
   }

   assert(src_count == 4 || src_count == 2 || src_count == 1);
   assert(type.width == 8);
   assert(type.length == 16);

   type8_t = lp_build_vec_type(gallivm, type);

   type64 = type;
   type64.length /= 8;
   type64.width *= 8;
   type64_t = lp_build_vec_type(gallivm, type64);

   type16 = type;
   type16.length /= 2;
   type16.width *= 2;
   type16_t = lp_build_vec_type(gallivm, type16);

   type32 = type;
   type32.length /= 4;
   type32.width *= 4;
   type32_t = lp_build_vec_type(gallivm, type32);

   lp_build_transpose_aos_n(gallivm, type, src, src_count, tmp);

   if (src_count == 1) {
      /* transpose was no-op, just untwiddle */
      LLVMValueRef shuf_vec;
      shuf_vec = LLVMConstVector(shuf, 8);
      tmp[0] = LLVMBuildBitCast(builder, src[0], type16_t, "");
      tmp[0] = LLVMBuildShuffleVector(builder, tmp[0], tmp[0], shuf_vec, "");
      dst[0] = LLVMBuildBitCast(builder, tmp[0], type8_t, "");
   } else if (src_count == 2) {
      LLVMValueRef shuf_vec;
      shuf_vec = LLVMConstVector(shuf, 4);

      for (unsigned i = 0; i < 2; i++) {
         tmp[i] = LLVMBuildBitCast(builder, tmp[i], type32_t, "");
         tmp[i] = LLVMBuildShuffleVector(builder, tmp[i], tmp[i], shuf_vec, "");
         dst[i] = LLVMBuildBitCast(builder, tmp[i], type8_t, "");
      }
   } else {
      for (unsigned j = 0; j < 2; j++) {
         LLVMValueRef lo, hi, lo2, hi2;
          /*
          * Note that if we only really have 3 valid channels (rgb)
          * and we don't need alpha we could substitute a undef here
          * for the respective channel (causing llvm to drop conversion
          * for alpha).
          */
         /* we now have rgba0rgba1rgba4rgba5 etc, untwiddle */
         lo2 = LLVMBuildBitCast(builder, tmp[j*2], type64_t, "");
         hi2 = LLVMBuildBitCast(builder, tmp[j*2 + 1], type64_t, "");
         lo = lp_build_interleave2(gallivm, type64, lo2, hi2, 0);
         hi = lp_build_interleave2(gallivm, type64, lo2, hi2, 1);
         dst[j*2] = LLVMBuildBitCast(builder, lo, type8_t, "");
         dst[j*2 + 1] = LLVMBuildBitCast(builder, hi, type8_t, "");
      }
   }
}


/**
 * Load an unswizzled block of pixels from memory
 */
static void
load_unswizzled_block(struct gallivm_state *gallivm,
                      LLVMTypeRef base_type,
                      LLVMValueRef base_ptr,
                      LLVMValueRef stride,
                      unsigned block_width,
                      unsigned block_height,
                      LLVMValueRef* dst,
                      struct lp_type dst_type,
                      unsigned dst_count,
                      unsigned dst_alignment)
{
   LLVMBuilderRef builder = gallivm->builder;
   const unsigned row_size = dst_count / block_height;

   /* Ensure block exactly fits into dst */
   assert((block_width * block_height) % dst_count == 0);

   for (unsigned i = 0; i < dst_count; ++i) {
      unsigned x = i % row_size;
      unsigned y = i / row_size;

      LLVMValueRef bx = lp_build_const_int32(gallivm, x * (dst_type.width / 8) * dst_type.length);
      LLVMValueRef by = LLVMBuildMul(builder, lp_build_const_int32(gallivm, y), stride, "");

      LLVMValueRef gep[2];
      LLVMValueRef dst_ptr;

      gep[0] = lp_build_const_int32(gallivm, 0);
      gep[1] = LLVMBuildAdd(builder, bx, by, "");

      dst_ptr = LLVMBuildGEP2(builder, base_type, base_ptr, gep, 2, "");
      dst_ptr = LLVMBuildBitCast(builder, dst_ptr,
                                 LLVMPointerType(lp_build_vec_type(gallivm, dst_type), 0), "");

      dst[i] = LLVMBuildLoad2(builder,
                              lp_build_vec_type(gallivm, dst_type),
                              dst_ptr, "");

      LLVMSetAlignment(dst[i], dst_alignment);
   }
}


/**
 * Store an unswizzled block of pixels to memory
 */
static void
store_unswizzled_block(struct gallivm_state *gallivm,
                       LLVMTypeRef base_type,
                       LLVMValueRef base_ptr,
                       LLVMValueRef stride,
                       unsigned block_width,
                       unsigned block_height,
                       LLVMValueRef src[],   // [src_count]
                       struct lp_type src_type,
                       unsigned src_count,
                       unsigned src_alignment)
{
   LLVMBuilderRef builder = gallivm->builder;
   const unsigned row_size = src_count / block_height;

   /* Ensure src exactly fits into block */
   assert((block_width * block_height) % src_count == 0);

   for (unsigned i = 0; i < src_count; ++i) {
      unsigned x = i % row_size;
      unsigned y = i / row_size;

      LLVMValueRef bx = lp_build_const_int32(gallivm, x * (src_type.width / 8) * src_type.length);
      LLVMValueRef by = LLVMBuildMul(builder, lp_build_const_int32(gallivm, y), stride, "");

      LLVMValueRef gep[2];
      LLVMValueRef src_ptr;

      gep[0] = lp_build_const_int32(gallivm, 0);
      gep[1] = LLVMBuildAdd(builder, bx, by, "");

      src_ptr = LLVMBuildGEP2(builder, base_type, base_ptr, gep, 2, "");
      src_ptr = LLVMBuildBitCast(builder, src_ptr,
                                 LLVMPointerType(lp_build_vec_type(gallivm, src_type), 0), "");

      src_ptr = LLVMBuildStore(builder, src[i], src_ptr);

      LLVMSetAlignment(src_ptr, src_alignment);
   }
}



/**
 * Retrieves the type for a format which is usable in the blending code.
 *
 * e.g. RGBA16F = 4x float, R3G3B2 = 3x byte
 */
static inline void
lp_blend_type_from_format_desc(const struct util_format_description *format_desc,
                               struct lp_type* type)
{
   if (format_expands_to_float_soa(format_desc)) {
      /* always use ordinary floats for blending */
      type->floating = true;
      type->fixed = false;
      type->sign = true;
      type->norm = false;
      type->width = 32;
      type->length = 4;
      return;
   }

   const int chan = util_format_get_first_non_void_channel(format_desc->format);

   memset(type, 0, sizeof(struct lp_type));
   type->floating = format_desc->channel[chan].type == UTIL_FORMAT_TYPE_FLOAT;
   type->fixed    = format_desc->channel[chan].type == UTIL_FORMAT_TYPE_FIXED;
   type->sign     = format_desc->channel[chan].type != UTIL_FORMAT_TYPE_UNSIGNED;
   type->norm     = format_desc->channel[chan].normalized;
   type->width    = format_desc->channel[chan].size;
   type->length   = format_desc->nr_channels;

   for (unsigned i = 1; i < format_desc->nr_channels; ++i) {
      if (format_desc->channel[i].size > type->width)
         type->width = format_desc->channel[i].size;
   }

   if (type->floating) {
      type->width = 32;
   } else {
      if (type->width <= 8) {
         type->width = 8;
      } else if (type->width <= 16) {
         type->width = 16;
      } else {
         type->width = 32;
      }
   }

   if (is_arithmetic_format(format_desc) && type->length == 3) {
      type->length = 4;
   }
}


/**
 * Scale a normalized value from src_bits to dst_bits.
 *
 * The exact calculation is
 *
 *    dst = iround(src * dst_mask / src_mask)
 *
 *  or with integer rounding
 *
 *    dst = src * (2*dst_mask + sign(src)*src_mask) / (2*src_mask)
 *
 *  where
 *
 *    src_mask = (1 << src_bits) - 1
 *    dst_mask = (1 << dst_bits) - 1
 *
 * but we try to avoid division and multiplication through shifts.
 */
static inline LLVMValueRef
scale_bits(struct gallivm_state *gallivm,
           int src_bits,
           int dst_bits,
           LLVMValueRef src,
           struct lp_type src_type)
{
   LLVMBuilderRef builder = gallivm->builder;
   LLVMValueRef result = src;

   if (dst_bits < src_bits) {
      int delta_bits = src_bits - dst_bits;

      if (delta_bits <= dst_bits) {

         if (dst_bits == 4) {
            struct lp_type flt_type =
               lp_type_float_vec(32, src_type.length * 32);

            result = lp_build_unsigned_norm_to_float(gallivm, src_bits,
                                                     flt_type, src);
            result = lp_build_clamped_float_to_unsigned_norm(gallivm, flt_type,
                                                             dst_bits, result);
            result = LLVMBuildTrunc(gallivm->builder, result,
                                    lp_build_int_vec_type(gallivm, src_type),
                                    "");
            return result;
         }

         /*
          * Approximate the rescaling with a single shift.
          *
          * This gives the wrong rounding.
          */

         result = LLVMBuildLShr(builder, src,
                                lp_build_const_int_vec(gallivm, src_type,
                                                       delta_bits),
                                "");
      } else {
         /*
          * Try more accurate rescaling.
          */

         /*
          * Drop the least significant bits to make space for the
          * multiplication.
          *
          * XXX: A better approach would be to use a wider integer type as
          * intermediate.  But this is enough to convert alpha from 16bits ->
          * 2 when rendering to PIPE_FORMAT_R10G10B10A2_UNORM.
          */
         result = LLVMBuildLShr(builder, src,
                                lp_build_const_int_vec(gallivm, src_type,
                                                       dst_bits),
                                "");


         result = LLVMBuildMul(builder, result,
                               lp_build_const_int_vec(gallivm, src_type,
                                                      (1LL << dst_bits) - 1),
                               "");

         /*
          * Add a rounding term before the division.
          *
          * TODO: Handle signed integers too.
          */
         if (!src_type.sign) {
            result = LLVMBuildAdd(builder, result,
                                  lp_build_const_int_vec(gallivm, src_type,
                                                    (1LL << (delta_bits - 1))),
                                  "");
         }

         /*
          * Approximate the division by src_mask with a src_bits shift.
          *
          * Given the src has already been shifted by dst_bits, all we need
          * to do is to shift by the difference.
          */

         result = LLVMBuildLShr(builder,
                                result,
                                lp_build_const_int_vec(gallivm, src_type, delta_bits),
                                "");
      }

   } else if (dst_bits > src_bits) {
      /* Scale up bits */
      int db = dst_bits - src_bits;

      /* Shift left by difference in bits */
      result = LLVMBuildShl(builder,
                            src,
                            lp_build_const_int_vec(gallivm, src_type, db),
                            "");

      if (db <= src_bits) {
         /* Enough bits in src to fill the remainder */
         LLVMValueRef lower = LLVMBuildLShr(builder,
                                            src,
                                            lp_build_const_int_vec(gallivm, src_type, src_bits - db),
                                            "");

         result = LLVMBuildOr(builder, result, lower, "");
      } else if (db > src_bits) {
         /* Need to repeatedly copy src bits to fill remainder in dst */
         unsigned n;

         for (n = src_bits; n < dst_bits; n *= 2) {
            LLVMValueRef shuv = lp_build_const_int_vec(gallivm, src_type, n);

            result = LLVMBuildOr(builder,
                                 result,
                                 LLVMBuildLShr(builder, result, shuv, ""),
                                 "");
         }
      }
   }

   return result;
}

/**
 * If RT is a smallfloat (needing denorms) format
 */
static inline int
have_smallfloat_format(struct lp_type dst_type,
                       enum pipe_format format)
{
   return ((dst_type.floating && dst_type.width != 32) ||
    /* due to format handling hacks this format doesn't have floating set
     * here (and actually has width set to 32 too) so special case this.
     */
    (format == PIPE_FORMAT_R11G11B10_FLOAT));
}


/**
 * Convert from memory format to blending format
 *
 * e.g. GL_R3G3B2 is 1 byte in memory but 3 bytes for blending
 */
static void
convert_to_blend_type(struct gallivm_state *gallivm,
                      unsigned block_size,
                      const struct util_format_description *src_fmt,
                      struct lp_type src_type,
                      struct lp_type dst_type,
                      LLVMValueRef* src, // and dst
                      unsigned num_srcs)
{
   LLVMValueRef *dst = src;
   LLVMBuilderRef builder = gallivm->builder;
   struct lp_type blend_type;
   struct lp_type mem_type;
   unsigned i, j;
   unsigned pixels = block_size / num_srcs;
   bool is_arith;

   /*
    * full custom path for packed floats and srgb formats - none of the later
    * functions would do anything useful, and given the lp_type representation
    * they can't be fixed. Should really have some SoA blend path for these
    * kind of formats rather than hacking them in here.
    */
   if (format_expands_to_float_soa(src_fmt)) {
      LLVMValueRef tmpsrc[4];
      /*
       * This is pretty suboptimal for this case blending in SoA would be much
       * better, since conversion gets us SoA values so need to convert back.
       */
      assert(src_type.width == 32 || src_type.width == 16);
      assert(dst_type.floating);
      assert(dst_type.width == 32);
      assert(dst_type.length % 4 == 0);
      assert(num_srcs % 4 == 0);

      if (src_type.width == 16) {
         /* expand 4x16bit values to 4x32bit */
         struct lp_type type32x4 = src_type;
         LLVMTypeRef ltype32x4;
         unsigned num_fetch = dst_type.length == 8 ? num_srcs / 2 : num_srcs / 4;
         type32x4.width = 32;
         ltype32x4 = lp_build_vec_type(gallivm, type32x4);
         for (i = 0; i < num_fetch; i++) {
            src[i] = LLVMBuildZExt(builder, src[i], ltype32x4, "");
         }
         src_type.width = 32;
      }
      for (i = 0; i < 4; i++) {
         tmpsrc[i] = src[i];
      }
      for (i = 0; i < num_srcs / 4; i++) {
         LLVMValueRef tmpsoa[4];
         LLVMValueRef tmps = tmpsrc[i];
         if (dst_type.length == 8) {
            LLVMValueRef shuffles[8];
            unsigned j;
            /* fetch was 4 values but need 8-wide output values */
            tmps = lp_build_concat(gallivm, &tmpsrc[i * 2], src_type, 2);
            /*
             * for 8-wide aos transpose would give us wrong order not matching
             * incoming converted fs values and mask. ARGH.
             */
            for (j = 0; j < 4; j++) {
               shuffles[j] = lp_build_const_int32(gallivm, j * 2);
               shuffles[j + 4] = lp_build_const_int32(gallivm, j * 2 + 1);
            }
            tmps = LLVMBuildShuffleVector(builder, tmps, tmps,
                                          LLVMConstVector(shuffles, 8), "");
         }
         if (src_fmt->format == PIPE_FORMAT_R11G11B10_FLOAT) {
            lp_build_r11g11b10_to_float(gallivm, tmps, tmpsoa);
         } else {
            lp_build_unpack_rgba_soa(gallivm, src_fmt, dst_type, tmps, tmpsoa);
         }
         lp_build_transpose_aos(gallivm, dst_type, tmpsoa, &src[i * 4]);
      }
      return;
   }

   lp_mem_type_from_format_desc(src_fmt, &mem_type);
   lp_blend_type_from_format_desc(src_fmt, &blend_type);

   /* Is the format arithmetic */
   is_arith = blend_type.length * blend_type.width != mem_type.width * mem_type.length;
   is_arith &= !(mem_type.width == 16 && mem_type.floating);

   /* Pad if necessary */
   if (!is_arith && src_type.length < dst_type.length) {
      for (i = 0; i < num_srcs; ++i) {
         dst[i] = lp_build_pad_vector(gallivm, src[i], dst_type.length);
      }

      src_type.length = dst_type.length;
   }

   /* Special case for half-floats */
   if (mem_type.width == 16 && mem_type.floating) {
      assert(blend_type.width == 32 && blend_type.floating);
      lp_build_conv_auto(gallivm, src_type, &dst_type, dst, num_srcs, dst);
      is_arith = false;
   }

   if (!is_arith) {
      return;
   }

   src_type.width = blend_type.width * blend_type.length;
   blend_type.length *= pixels;
   src_type.length *= pixels / (src_type.length / mem_type.length);

   for (i = 0; i < num_srcs; ++i) {
      LLVMValueRef chans;
      LLVMValueRef res = NULL;

      dst[i] = LLVMBuildZExt(builder, src[i], lp_build_vec_type(gallivm, src_type), "");

      for (j = 0; j < src_fmt->nr_channels; ++j) {
         unsigned mask = 0;
         unsigned sa = src_fmt->channel[j].shift;
#if UTIL_ARCH_LITTLE_ENDIAN
         unsigned from_lsb = j;
#else
         unsigned from_lsb = (blend_type.length / pixels) - j - 1;
#endif

         mask = (1 << src_fmt->channel[j].size) - 1;

         /* Extract bits from source */
         chans = LLVMBuildLShr(builder,
                               dst[i],
                               lp_build_const_int_vec(gallivm, src_type, sa),
                               "");

         chans = LLVMBuildAnd(builder,
                              chans,
                              lp_build_const_int_vec(gallivm, src_type, mask),
                              "");

         /* Scale bits */
         if (src_type.norm) {
            chans = scale_bits(gallivm, src_fmt->channel[j].size,
                               blend_type.width, chans, src_type);
         }

         /* Insert bits into correct position */
         chans = LLVMBuildShl(builder,
                              chans,
                              lp_build_const_int_vec(gallivm, src_type, from_lsb * blend_type.width),
                              "");

         if (j == 0) {
            res = chans;
         } else {
            res = LLVMBuildOr(builder, res, chans, "");
         }
      }

      dst[i] = LLVMBuildBitCast(builder, res, lp_build_vec_type(gallivm, blend_type), "");
   }
}


/**
 * Convert from blending format to memory format
 *
 * e.g. GL_R3G3B2 is 3 bytes for blending but 1 byte in memory
 */
static void
convert_from_blend_type(struct gallivm_state *gallivm,
                        unsigned block_size,
                        const struct util_format_description *src_fmt,
                        struct lp_type src_type,
                        struct lp_type dst_type,
                        LLVMValueRef* src, // and dst
                        unsigned num_srcs)
{
   LLVMValueRef* dst = src;
   unsigned i, j, k;
   struct lp_type mem_type;
   struct lp_type blend_type;
   LLVMBuilderRef builder = gallivm->builder;
   unsigned pixels = block_size / num_srcs;
   bool is_arith;

   /*
    * full custom path for packed floats and srgb formats - none of the later
    * functions would do anything useful, and given the lp_type representation
    * they can't be fixed. Should really have some SoA blend path for these
    * kind of formats rather than hacking them in here.
    */
   if (format_expands_to_float_soa(src_fmt)) {
      /*
       * This is pretty suboptimal for this case blending in SoA would be much
       * better - we need to transpose the AoS values back to SoA values for
       * conversion/packing.
       */
      assert(src_type.floating);
      assert(src_type.width == 32);
      assert(src_type.length % 4 == 0);
      assert(dst_type.width == 32 || dst_type.width == 16);

      for (i = 0; i < num_srcs / 4; i++) {
         LLVMValueRef tmpsoa[4], tmpdst;
         lp_build_transpose_aos(gallivm, src_type, &src[i * 4], tmpsoa);
         /* really really need SoA here */

         if (src_fmt->format == PIPE_FORMAT_R11G11B10_FLOAT) {
            tmpdst = lp_build_float_to_r11g11b10(gallivm, tmpsoa);
         } else {
            tmpdst = lp_build_float_to_srgb_packed(gallivm, src_fmt,
                                                   src_type, tmpsoa);
         }

         if (src_type.length == 8) {
            LLVMValueRef tmpaos, shuffles[8];
            unsigned j;
            /*
             * for 8-wide aos transpose has given us wrong order not matching
             * output order. HMPF. Also need to split the output values
             * manually.
             */
            for (j = 0; j < 4; j++) {
               shuffles[j * 2] = lp_build_const_int32(gallivm, j);
               shuffles[j * 2 + 1] = lp_build_const_int32(gallivm, j + 4);
            }
            tmpaos = LLVMBuildShuffleVector(builder, tmpdst, tmpdst,
                                            LLVMConstVector(shuffles, 8), "");
            src[i * 2] = lp_build_extract_range(gallivm, tmpaos, 0, 4);
            src[i * 2 + 1] = lp_build_extract_range(gallivm, tmpaos, 4, 4);
         } else {
            src[i] = tmpdst;
         }
      }
      if (dst_type.width == 16) {
         struct lp_type type16x8 = dst_type;
         struct lp_type type32x4 = dst_type;
         LLVMTypeRef ltype16x4, ltypei64, ltypei128;
         unsigned num_fetch = src_type.length == 8 ? num_srcs / 2 : num_srcs / 4;
         type16x8.length = 8;
         type32x4.width = 32;
         ltypei128 = LLVMIntTypeInContext(gallivm->context, 128);
         ltypei64 = LLVMIntTypeInContext(gallivm->context, 64);
         ltype16x4 = lp_build_vec_type(gallivm, dst_type);
         /* We could do vector truncation but it doesn't generate very good code */
         for (i = 0; i < num_fetch; i++) {
            src[i] = lp_build_pack2(gallivm, type32x4, type16x8,
                                    src[i], lp_build_zero(gallivm, type32x4));
            src[i] = LLVMBuildBitCast(builder, src[i], ltypei128, "");
            src[i] = LLVMBuildTrunc(builder, src[i], ltypei64, "");
            src[i] = LLVMBuildBitCast(builder, src[i], ltype16x4, "");
         }
      }
      return;
   }

   lp_mem_type_from_format_desc(src_fmt, &mem_type);
   lp_blend_type_from_format_desc(src_fmt, &blend_type);

   is_arith = (blend_type.length * blend_type.width != mem_type.width * mem_type.length);

   /* Special case for half-floats */
   if (mem_type.width == 16 && mem_type.floating) {
      int length = dst_type.length;
      assert(blend_type.width == 32 && blend_type.floating);

      dst_type.length = src_type.length;

      lp_build_conv_auto(gallivm, src_type, &dst_type, dst, num_srcs, dst);

      dst_type.length = length;
      is_arith = false;
   }

   /* Remove any padding */
   if (!is_arith && (src_type.length % mem_type.length)) {
      src_type.length -= (src_type.length % mem_type.length);

      for (i = 0; i < num_srcs; ++i) {
         dst[i] = lp_build_extract_range(gallivm, dst[i], 0, src_type.length);
      }
   }

   /* No bit arithmetic to do */
   if (!is_arith) {
      return;
   }

   src_type.length = pixels;
   src_type.width = blend_type.length * blend_type.width;
   dst_type.length = pixels;

   for (i = 0; i < num_srcs; ++i) {
      LLVMValueRef chans;
      LLVMValueRef res = NULL;

      dst[i] = LLVMBuildBitCast(builder, src[i], lp_build_vec_type(gallivm, src_type), "");

      for (j = 0; j < src_fmt->nr_channels; ++j) {
         unsigned mask = 0;
         unsigned sa = src_fmt->channel[j].shift;
         unsigned sz_a = src_fmt->channel[j].size;
#if UTIL_ARCH_LITTLE_ENDIAN
         unsigned from_lsb = j;
#else
         unsigned from_lsb = blend_type.length - j - 1;
#endif

         assert(blend_type.width > src_fmt->channel[j].size);

         for (k = 0; k < blend_type.width; ++k) {
            mask |= 1 << k;
         }

         /* Extract bits */
         chans = LLVMBuildLShr(builder,
                               dst[i],
                               lp_build_const_int_vec(gallivm, src_type,
                                                      from_lsb * blend_type.width),
                               "");

         chans = LLVMBuildAnd(builder,
                              chans,
                              lp_build_const_int_vec(gallivm, src_type, mask),
                              "");

         /* Scale down bits */
         if (src_type.norm) {
            chans = scale_bits(gallivm, blend_type.width,
                               src_fmt->channel[j].size, chans, src_type);
         } else if (!src_type.floating && sz_a < blend_type.width) {
            LLVMValueRef mask_val = lp_build_const_int_vec(gallivm, src_type, (1UL << sz_a) - 1);
            LLVMValueRef mask = LLVMBuildICmp(builder, LLVMIntUGT, chans, mask_val, "");
            chans = LLVMBuildSelect(builder, mask, mask_val, chans, "");
         }

         /* Insert bits */
         chans = LLVMBuildShl(builder,
                              chans,
                              lp_build_const_int_vec(gallivm, src_type, sa),
                              "");

         sa += src_fmt->channel[j].size;

         if (j == 0) {
            res = chans;
         } else {
            res = LLVMBuildOr(builder, res, chans, "");
         }
      }

      assert (dst_type.width != 24);

      dst[i] = LLVMBuildTrunc(builder, res, lp_build_vec_type(gallivm, dst_type), "");
   }
}


/**
 * Convert alpha to same blend type as src
 */
static void
convert_alpha(struct gallivm_state *gallivm,
              struct lp_type row_type,
              struct lp_type alpha_type,
              const unsigned block_size,
              const unsigned block_height,
              const unsigned src_count,
              const unsigned dst_channels,
              const bool pad_inline,
              LLVMValueRef* src_alpha)
{
   LLVMBuilderRef builder = gallivm->builder;
   const unsigned length = row_type.length;
   row_type.length = alpha_type.length;

   /* Twiddle the alpha to match pixels */
   lp_bld_quad_twiddle(gallivm, alpha_type, src_alpha, block_height, src_alpha);

   /*
    * TODO this should use single lp_build_conv call for
    * src_count == 1 && dst_channels == 1 case (dropping the concat below)
    */
   for (unsigned i = 0; i < block_height; ++i) {
      lp_build_conv(gallivm, alpha_type, row_type, &src_alpha[i], 1,
                    &src_alpha[i], 1);
   }

   alpha_type = row_type;
   row_type.length = length;

   /* If only one channel we can only need the single alpha value per pixel */
   if (src_count == 1 && dst_channels == 1) {
      lp_build_concat_n(gallivm, alpha_type, src_alpha, block_height,
                        src_alpha, src_count);
   } else {
      /* If there are more srcs than rows then we need to split alpha up */
      if (src_count > block_height) {
         for (unsigned i = src_count; i > 0; --i) {
            unsigned pixels = block_size / src_count;
            unsigned idx = i - 1;

            src_alpha[idx] =
               lp_build_extract_range(gallivm, src_alpha[(idx * pixels) / 4],
                                      (idx * pixels) % 4, pixels);
         }
      }

      /* If there is a src for each pixel broadcast the alpha across whole
       * row
       */
      if (src_count == block_size) {
         for (unsigned i = 0; i < src_count; ++i) {
            src_alpha[i] = lp_build_broadcast(gallivm,
                              lp_build_vec_type(gallivm, row_type), src_alpha[i]);
         }
      } else {
         unsigned pixels = block_size / src_count;
         unsigned channels = pad_inline ? TGSI_NUM_CHANNELS : dst_channels;
         unsigned alpha_span = 1;
         LLVMValueRef shuffles[LP_MAX_VECTOR_LENGTH];

         /* Check if we need 2 src_alphas for our shuffles */
         if (pixels > alpha_type.length) {
            alpha_span = 2;
         }

         /* Broadcast alpha across all channels, e.g. a1a2 to a1a1a1a1a2a2a2a2 */
         for (unsigned j = 0; j < row_type.length; ++j) {
            if (j < pixels * channels) {
               shuffles[j] = lp_build_const_int32(gallivm, j / channels);
            } else {
               shuffles[j] = LLVMGetUndef(LLVMInt32TypeInContext(gallivm->context));
            }
         }

         for (unsigned i = 0; i < src_count; ++i) {
            unsigned idx1 = i, idx2 = i;

            if (alpha_span > 1){
               idx1 *= alpha_span;
               idx2 = idx1 + 1;
            }

            src_alpha[i] = LLVMBuildShuffleVector(builder,
                                                  src_alpha[idx1],
                                                  src_alpha[idx2],
                                                  LLVMConstVector(shuffles, row_type.length),
                                                  "");
         }
      }
   }
}


/**
 * Generates the blend function for unswizzled colour buffers
 * Also generates the read & write from colour buffer
 */
static void
generate_unswizzled_blend(struct gallivm_state *gallivm,
                          unsigned rt,
                          struct lp_fragment_shader_variant *variant,
                          enum pipe_format out_format,
                          unsigned int num_fs,
                          struct lp_type fs_type,
                          LLVMValueRef* fs_mask,
                          LLVMValueRef fs_out_color[PIPE_MAX_COLOR_BUFS][TGSI_NUM_CHANNELS][4],
                          LLVMTypeRef context_type,
                          LLVMValueRef context_ptr,
                          LLVMTypeRef color_type,
                          LLVMValueRef color_ptr,
                          LLVMValueRef stride,
                          unsigned partial_mask,
                          bool do_branch)
{
   const unsigned alpha_channel = 3;
   const unsigned block_width = LP_RASTER_BLOCK_SIZE;
   const unsigned block_height = LP_RASTER_BLOCK_SIZE;
   const unsigned block_size = block_width * block_height;
   const unsigned lp_integer_vector_width = 128;

   LLVMBuilderRef builder = gallivm->builder;
   LLVMValueRef fs_src[4][TGSI_NUM_CHANNELS];
   LLVMValueRef fs_src1[4][TGSI_NUM_CHANNELS];
   LLVMValueRef src_alpha[4 * 4];
   LLVMValueRef src1_alpha[4 * 4] = { NULL };
   LLVMValueRef src_mask[4 * 4];
   LLVMValueRef src[4 * 4];
   LLVMValueRef src1[4 * 4];
   LLVMValueRef dst[4 * 4];

   struct lp_build_mask_context mask_ctx;

   unsigned char swizzle[TGSI_NUM_CHANNELS];
   unsigned src_channels = TGSI_NUM_CHANNELS;

   const struct util_format_description *out_format_desc =
      util_format_description(out_format);

   bool pad_inline = is_arithmetic_format(out_format_desc);
   const bool dual_source_blend =
      variant->key.blend.rt[0].blend_enable &&
      util_blend_state_is_dual(&variant->key.blend, 0);

   const bool is_1d = variant->key.resource_1d;
   const unsigned num_fullblock_fs = is_1d ? 2 * num_fs : num_fs;
   LLVMValueRef fpstate = NULL;

   LLVMTypeRef fs_vec_type = lp_build_vec_type(gallivm, fs_type);

   /* Get type from output format */
   struct lp_type row_type, dst_type;
   lp_blend_type_from_format_desc(out_format_desc, &row_type);
   lp_mem_type_from_format_desc(out_format_desc, &dst_type);

   /*
    * Technically this code should go into lp_build_smallfloat_to_float
    * and lp_build_float_to_smallfloat but due to the
    * http://llvm.org/bugs/show_bug.cgi?id=6393
    * llvm reorders the mxcsr intrinsics in a way that breaks the code.
    * So the ordering is important here and there shouldn't be any
    * llvm ir instrunctions in this function before
    * this, otherwise half-float format conversions won't work
    * (again due to llvm bug #6393).
    */
   if (have_smallfloat_format(dst_type, out_format)) {
      /* We need to make sure that denorms are ok for half float
         conversions */
      fpstate = lp_build_fpstate_get(gallivm);
      lp_build_fpstate_set_denorms_zero(gallivm, false);
   }

   struct lp_type mask_type = lp_int32_vec4_type();
   mask_type.length = fs_type.length;

   for (unsigned i = num_fs; i < num_fullblock_fs; i++) {
      fs_mask[i] = lp_build_zero(gallivm, mask_type);
   }

   /* Do not bother executing code when mask is empty.. */
   if (do_branch) {
      LLVMValueRef check_mask =
         LLVMConstNull(lp_build_int_vec_type(gallivm, mask_type));

      for (unsigned i = 0; i < num_fullblock_fs; ++i) {
         check_mask = LLVMBuildOr(builder, check_mask, fs_mask[i], "");
      }

      lp_build_mask_begin(&mask_ctx, gallivm, mask_type, check_mask);
      lp_build_mask_check(&mask_ctx);
   }

   partial_mask |= !variant->opaque;
   LLVMValueRef i32_zero = lp_build_const_int32(gallivm, 0);

   LLVMValueRef undef_src_val = lp_build_undef(gallivm, fs_type);

   row_type.length = fs_type.length;
   unsigned vector_width =
      dst_type.floating ? lp_native_vector_width : lp_integer_vector_width;

   /* Compute correct swizzle and count channels */
   memset(swizzle, LP_BLD_SWIZZLE_DONTCARE, TGSI_NUM_CHANNELS);
   unsigned dst_channels = 0;

   bool has_alpha = false;
   for (unsigned i = 0; i < TGSI_NUM_CHANNELS; ++i) {
      /* Ensure channel is used */
      if (out_format_desc->swizzle[i] >= TGSI_NUM_CHANNELS) {
         continue;
      }

      /* Ensure not already written to (happens in case with GL_ALPHA) */
      if (swizzle[out_format_desc->swizzle[i]] < TGSI_NUM_CHANNELS) {
         continue;
      }

      /* Ensure we haven't already found all channels */
      if (dst_channels >= out_format_desc->nr_channels) {
         continue;
      }

      swizzle[out_format_desc->swizzle[i]] = i;
      ++dst_channels;

      if (i == alpha_channel) {
         has_alpha = true;
      }
   }

   if (format_expands_to_float_soa(out_format_desc)) {
      /*
       * the code above can't work for layout_other
       * for srgb it would sort of work but we short-circuit swizzles, etc.
       * as that is done as part of unpack / pack.
       */
      dst_channels = 4; /* HACK: this is fake 4 really but need it due to transpose stuff later */
      has_alpha = true;
      swizzle[0] = 0;
      swizzle[1] = 1;
      swizzle[2] = 2;
      swizzle[3] = 3;
      pad_inline = true; /* HACK: prevent rgbxrgbx->rgbrgbxx conversion later */
   }

   /* If 3 channels then pad to include alpha for 4 element transpose */
   if (dst_channels == 3) {
      assert (!has_alpha);
      for (unsigned i = 0; i < TGSI_NUM_CHANNELS; i++) {
         if (swizzle[i] > TGSI_NUM_CHANNELS)
            swizzle[i] = 3;
      }
      if (out_format_desc->nr_channels == 4) {
         dst_channels = 4;
         /*
          * We use alpha from the color conversion, not separate one.
          * We had to include it for transpose, hence it will get converted
          * too (albeit when doing transpose after conversion, that would
          * no longer be the case necessarily).
          * (It works only with 4 channel dsts, e.g. rgbx formats, because
          * otherwise we really have padding, not alpha, included.)
          */
         has_alpha = true;
      }
   }

   /*
    * Load shader output
    */
   for (unsigned i = 0; i < num_fullblock_fs; ++i) {
      /* Always load alpha for use in blending */
      LLVMValueRef alpha;
      if (i < num_fs) {
         alpha = LLVMBuildLoad2(builder, fs_vec_type,
                                fs_out_color[rt][alpha_channel][i], "");
      } else {
         alpha = undef_src_val;
      }

      /* Load each channel */
      for (unsigned j = 0; j < dst_channels; ++j) {
         assert(swizzle[j] < 4);
         if (i < num_fs) {
            fs_src[i][j] = LLVMBuildLoad2(builder, fs_vec_type,
                                          fs_out_color[rt][swizzle[j]][i], "");
         } else {
            fs_src[i][j] = undef_src_val;
         }
      }

      /* If 3 channels then pad to include alpha for 4 element transpose */
      /*
       * XXX If we include that here maybe could actually use it instead of
       * separate alpha for blending?
       * (Difficult though we actually convert pad channels, not alpha.)
       */
      if (dst_channels == 3 && !has_alpha) {
         fs_src[i][3] = alpha;
      }

      /* We split the row_mask and row_alpha as we want 128bit interleave */
      if (fs_type.length == 8) {
         src_mask[i*2 + 0]  = lp_build_extract_range(gallivm, fs_mask[i],
                                                     0, src_channels);
         src_mask[i*2 + 1]  = lp_build_extract_range(gallivm, fs_mask[i],
                                                     src_channels,
                                                     src_channels);

         src_alpha[i*2 + 0] = lp_build_extract_range(gallivm, alpha,
                                                     0, src_channels);
         src_alpha[i*2 + 1] = lp_build_extract_range(gallivm, alpha,
                                                     src_channels,
                                                     src_channels);
      } else {
         src_mask[i] = fs_mask[i];
         src_alpha[i] = alpha;
      }
   }
   if (dual_source_blend) {
      /* same as above except different src/dst, skip masks and comments... */
      for (unsigned i = 0; i < num_fullblock_fs; ++i) {
         LLVMValueRef alpha;
         if (i < num_fs) {
            alpha = LLVMBuildLoad2(builder, fs_vec_type,
                                   fs_out_color[1][alpha_channel][i], "");
         } else {
            alpha = undef_src_val;
         }

         for (unsigned j = 0; j < dst_channels; ++j) {
            assert(swizzle[j] < 4);
            if (i < num_fs) {
               fs_src1[i][j] = LLVMBuildLoad2(builder, fs_vec_type,
                                              fs_out_color[1][swizzle[j]][i], "");
            } else {
               fs_src1[i][j] = undef_src_val;
            }
         }
         if (dst_channels == 3 && !has_alpha) {
            fs_src1[i][3] = alpha;
         }
         if (fs_type.length == 8) {
            src1_alpha[i*2 + 0] = lp_build_extract_range(gallivm, alpha, 0, src_channels);
            src1_alpha[i*2 + 1] = lp_build_extract_range(gallivm, alpha,
                                                         src_channels, src_channels);
         } else {
            src1_alpha[i] = alpha;
         }
      }
   }

   if (util_format_is_pure_integer(out_format)) {
      /*
       * In this case fs_type was really ints or uints disguised as floats,
       * fix that up now.
       */
      fs_type.floating = 0;
      fs_type.sign = dst_type.sign;
      fs_vec_type = lp_build_vec_type(gallivm, fs_type);
      for (unsigned i = 0; i < num_fullblock_fs; ++i) {
         for (unsigned j = 0; j < dst_channels; ++j) {
            fs_src[i][j] = LLVMBuildBitCast(builder, fs_src[i][j],
                                            fs_vec_type, "");
         }
         if (dst_channels == 3 && !has_alpha) {
            fs_src[i][3] = LLVMBuildBitCast(builder, fs_src[i][3],
                                            fs_vec_type, "");
         }
      }
   }

   /*
    * We actually should generally do conversion first (for non-1d cases)
    * when the blend format is 8 or 16 bits. The reason is obvious,
    * there's 2 or 4 times less vectors to deal with for the interleave...
    * Albeit for the AVX (not AVX2) case there's no benefit with 16 bit
    * vectors (as it can do 32bit unpack with 256bit vectors, but 8/16bit
    * unpack only with 128bit vectors).
    * Note: for 16bit sizes really need matching pack conversion code
    */
   bool twiddle_after_convert = false;
   if (!is_1d && dst_channels != 3 && dst_type.width == 8) {
      twiddle_after_convert = true;
   }

   /*
    * Pixel twiddle from fragment shader order to memory order
    */
   unsigned src_count;
   if (!twiddle_after_convert) {
      src_count = generate_fs_twiddle(gallivm, fs_type, num_fullblock_fs,
                                      dst_channels, fs_src, src, pad_inline);
      if (dual_source_blend) {
         generate_fs_twiddle(gallivm, fs_type, num_fullblock_fs, dst_channels,
                             fs_src1, src1, pad_inline);
      }
   } else {
      src_count = num_fullblock_fs * dst_channels;
      /*
       * We reorder things a bit here, so the cases for 4-wide and 8-wide
       * (AVX) turn out the same later when untwiddling/transpose (albeit
       * for true AVX2 path untwiddle needs to be different).
       * For now just order by colors first (so we can use unpack later).
       */
      for (unsigned j = 0; j < num_fullblock_fs; j++) {
         for (unsigned i = 0; i < dst_channels; i++) {
            src[i*num_fullblock_fs + j] = fs_src[j][i];
            if (dual_source_blend) {
               src1[i*num_fullblock_fs + j] = fs_src1[j][i];
            }
         }
      }
   }

   src_channels = dst_channels < 3 ? dst_channels : 4;
   if (src_count != num_fullblock_fs * src_channels) {
      unsigned ds = src_count / (num_fullblock_fs * src_channels);
      row_type.length /= ds;
      fs_type.length = row_type.length;
      fs_vec_type = lp_build_vec_type(gallivm, fs_type);
   }

   struct lp_type blend_type = row_type;
   mask_type.length = 4;

   /* Convert src to row_type */
   if (dual_source_blend) {
      struct lp_type old_row_type = row_type;
      lp_build_conv_auto(gallivm, fs_type, &row_type, src, src_count, src);
      src_count = lp_build_conv_auto(gallivm, fs_type, &old_row_type,
                                     src1, src_count, src1);
   } else {
      src_count = lp_build_conv_auto(gallivm, fs_type, &row_type,
                                     src, src_count, src);
   }

   /* If the rows are not an SSE vector, combine them to become SSE size! */
   if ((row_type.width * row_type.length) % 128) {
      unsigned bits = row_type.width * row_type.length;
      unsigned combined;

      assert(src_count >= (vector_width / bits));

      const unsigned dst_count = src_count / (vector_width / bits);

      combined = lp_build_concat_n(gallivm, row_type, src, src_count,
                                   src, dst_count);
      if (dual_source_blend) {
         lp_build_concat_n(gallivm, row_type, src1, src_count, src1, dst_count);
      }

      row_type.length *= combined;
      src_count /= combined;

      bits = row_type.width * row_type.length;
      assert(bits == 128 || bits == 256);
   }

   if (twiddle_after_convert) {
      fs_twiddle_transpose(gallivm, row_type, src, src_count, src);
      if (dual_source_blend) {
         fs_twiddle_transpose(gallivm, row_type, src1, src_count, src1);
      }
   }

   /*
    * Blend Colour conversion
    */
   LLVMValueRef blend_color =
      lp_jit_context_f_blend_color(gallivm, context_type, context_ptr);
   blend_color = LLVMBuildPointerCast(builder, blend_color,
                                      LLVMPointerType(fs_vec_type, 0),
                                      "");
   blend_color = LLVMBuildLoad2(builder, fs_vec_type,
                                LLVMBuildGEP2(builder, fs_vec_type,
                                              blend_color,
                                              &i32_zero, 1, ""), "");

   /* Convert */
   lp_build_conv(gallivm, fs_type, blend_type, &blend_color, 1,
                 &blend_color, 1);

   if (out_format_desc->colorspace == UTIL_FORMAT_COLORSPACE_SRGB) {
      /*
       * since blending is done with floats, there was no conversion.
       * However, the rules according to fixed point renderbuffers still
       * apply, that is we must clamp inputs to 0.0/1.0.
       * (This would apply to separate alpha conversion too but we currently
       * force has_alpha to be true.)
       * TODO: should skip this with "fake" blend, since post-blend conversion
       * will clamp anyway.
       * TODO: could also skip this if fragment color clamping is enabled.
       * We don't support it natively so it gets baked into the shader
       * however, so can't really tell here.
       */
      struct lp_build_context f32_bld;
      assert(row_type.floating);
      lp_build_context_init(&f32_bld, gallivm, row_type);
      for (unsigned i = 0; i < src_count; i++) {
         src[i] = lp_build_clamp_zero_one_nanzero(&f32_bld, src[i]);
      }
      if (dual_source_blend) {
         for (unsigned i = 0; i < src_count; i++) {
            src1[i] = lp_build_clamp_zero_one_nanzero(&f32_bld, src1[i]);
         }
      }
      /* probably can't be different than row_type but better safe than sorry... */
      lp_build_context_init(&f32_bld, gallivm, blend_type);
      blend_color = lp_build_clamp(&f32_bld, blend_color,
                                   f32_bld.zero, f32_bld.one);
   }

   /* Extract alpha */
   LLVMValueRef blend_alpha =
      lp_build_extract_broadcast(gallivm, blend_type, row_type,
                                 blend_color,
                                 lp_build_const_int32(gallivm, 3));

   /* Swizzle to appropriate channels, e.g. from RGBA to BGRA BGRA */
   pad_inline &= (dst_channels * (block_size / src_count) * row_type.width)
      != vector_width;
   if (pad_inline) {
      /* Use all 4 channels e.g. from RGBA RGBA to RGxx RGxx */
      blend_color = lp_build_swizzle_aos_n(gallivm, blend_color, swizzle,
                                           TGSI_NUM_CHANNELS, row_type.length);
   } else {
      /* Only use dst_channels e.g. RGBA RGBA to RG RG xxxx */
      blend_color = lp_build_swizzle_aos_n(gallivm, blend_color, swizzle,
                                           dst_channels, row_type.length);
   }

   /*
    * Mask conversion
    */
   lp_bld_quad_twiddle(gallivm, mask_type, &src_mask[0],
                       block_height, &src_mask[0]);

   if (src_count < block_height) {
      lp_build_concat_n(gallivm, mask_type, src_mask, 4, src_mask, src_count);
   } else if (src_count > block_height) {
      for (unsigned i = src_count; i > 0; --i) {
         unsigned pixels = block_size / src_count;
         unsigned idx = i - 1;

         src_mask[idx] = lp_build_extract_range(gallivm,
                                                src_mask[(idx * pixels) / 4],
                                                (idx * pixels) % 4, pixels);
      }
   }

   assert(mask_type.width == 32);

   for (unsigned i = 0; i < src_count; ++i) {
      unsigned pixels = block_size / src_count;
      unsigned pixel_width = row_type.width * dst_channels;

      if (pixel_width == 24) {
         mask_type.width = 8;
         mask_type.length = vector_width / mask_type.width;
      } else {
         mask_type.length = pixels;
         mask_type.width = row_type.width * dst_channels;

         /*
          * If mask_type width is smaller than 32bit, this doesn't quite
          * generate the most efficient code (could use some pack).
          */
         src_mask[i] = LLVMBuildIntCast(builder, src_mask[i],
                                        lp_build_int_vec_type(gallivm,
                                                              mask_type), "");

         mask_type.length *= dst_channels;
         mask_type.width /= dst_channels;
      }

      src_mask[i] = LLVMBuildBitCast(builder, src_mask[i],
                                     lp_build_int_vec_type(gallivm, mask_type),
                                     "");
      src_mask[i] = lp_build_pad_vector(gallivm, src_mask[i], row_type.length);
   }

   /*
    * Alpha conversion
    */
   if (!has_alpha) {
      struct lp_type alpha_type = fs_type;
      alpha_type.length = 4;
      convert_alpha(gallivm, row_type, alpha_type,
                    block_size, block_height,
                    src_count, dst_channels,
                    pad_inline, src_alpha);
      if (dual_source_blend) {
         convert_alpha(gallivm, row_type, alpha_type,
                       block_size, block_height,
                       src_count, dst_channels,
                       pad_inline, src1_alpha);
      }
   }


   /*
    * Load dst from memory
    */
   unsigned dst_count;
   if (src_count < block_height) {
      dst_count = block_height;
   } else {
      dst_count = src_count;
   }

   dst_type.length *= block_size / dst_count;

   if (format_expands_to_float_soa(out_format_desc)) {
      /*
       * we need multiple values at once for the conversion, so can as well
       * load them vectorized here too instead of concatenating later.
       * (Still need concatenation later for 8-wide vectors).
       */
      dst_count = block_height;
      dst_type.length = block_width;
   }

   /*
    * Compute the alignment of the destination pointer in bytes
    * We fetch 1-4 pixels, if the format has pot alignment then those fetches
    * are always aligned by MIN2(16, fetch_width) except for buffers (not
    * 1d tex but can't distinguish here) so need to stick with per-pixel
    * alignment in this case.
    */
   unsigned dst_alignment;
   if (is_1d) {
      dst_alignment = (out_format_desc->block.bits + 7)/(out_format_desc->block.width * 8);
   } else {
      dst_alignment = dst_type.length * dst_type.width / 8;
   }
   /* Force power-of-two alignment by extracting only the least-significant-bit */
   dst_alignment = 1 << (ffs(dst_alignment) - 1);
   /*
    * Resource base and stride pointers are aligned to 16 bytes, so that's
    * the maximum alignment we can guarantee
    */
   dst_alignment = MIN2(16, dst_alignment);

   struct lp_type ls_type = dst_type;

   if (dst_count > src_count) {
      if ((dst_type.width == 8 || dst_type.width == 16) &&
          util_is_power_of_two_or_zero(dst_type.length) &&
          dst_type.length * dst_type.width < 128) {
         /*
          * Never try to load values as 4xi8 which we will then
          * concatenate to larger vectors. This gives llvm a real
          * headache (the problem is the type legalizer (?) will
          * try to load that as 4xi8 zext to 4xi32 to fill the vector,
          * then the shuffles to concatenate are more or less impossible
          * - llvm is easily capable of generating a sequence of 32
          * pextrb/pinsrb instructions for that. Albeit it appears to
          * be fixed in llvm 4.0. So, load and concatenate with 32bit
          * width to avoid the trouble (16bit seems not as bad, llvm
          * probably recognizes the load+shuffle as only one shuffle
          * is necessary, but we can do just the same anyway).
          */
         ls_type.length = dst_type.length * dst_type.width / 32;
         ls_type.width = 32;
      }
   }

   if (is_1d) {
      load_unswizzled_block(gallivm, color_type, color_ptr, stride, block_width, 1,
                            dst, ls_type, dst_count / 4, dst_alignment);
      for (unsigned i = dst_count / 4; i < dst_count; i++) {
         dst[i] = lp_build_undef(gallivm, ls_type);
      }
   } else {
      load_unswizzled_block(gallivm, color_type, color_ptr, stride, block_width,
                            block_height, dst, ls_type, dst_count,
                            dst_alignment);
   }


   /*
    * Convert from dst/output format to src/blending format.
    *
    * This is necessary as we can only read 1 row from memory at a time,
    * so the minimum dst_count will ever be at this point is 4.
    *
    * With, for example, R8 format you can have all 16 pixels in a 128 bit
    * vector, this will take the 4 dsts and combine them into 1 src so we can
    * perform blending on all 16 pixels in that single vector at once.
    */
   if (dst_count > src_count) {
      if (ls_type.length != dst_type.length && ls_type.length == 1) {
         LLVMTypeRef elem_type = lp_build_elem_type(gallivm, ls_type);
         LLVMTypeRef ls_vec_type = LLVMVectorType(elem_type, 1);
         for (unsigned i = 0; i < dst_count; i++) {
            dst[i] = LLVMBuildBitCast(builder, dst[i], ls_vec_type, "");
         }
      }

      lp_build_concat_n(gallivm, ls_type, dst, 4, dst, src_count);

      if (ls_type.length != dst_type.length) {
         struct lp_type tmp_type = dst_type;
         tmp_type.length = dst_type.length * 4 / src_count;
         for (unsigned i = 0; i < src_count; i++) {
            dst[i] = LLVMBuildBitCast(builder, dst[i],
                                      lp_build_vec_type(gallivm, tmp_type), "");
         }
      }
   }

   /*
    * Blending
    */
   /* XXX this is broken for RGB8 formats -
    * they get expanded from 12 to 16 elements (to include alpha)
    * by convert_to_blend_type then reduced to 15 instead of 12
    * by convert_from_blend_type (a simple fix though breaks A8...).
    * R16G16B16 also crashes differently however something going wrong
    * inside llvm handling npot vector sizes seemingly.
    * It seems some cleanup could be done here (like skipping conversion/blend
    * when not needed).
    */
   convert_to_blend_type(gallivm, block_size, out_format_desc, dst_type,
                         row_type, dst, src_count);

   /*
    * FIXME: Really should get logic ops / masks out of generic blend / row
    * format. Logic ops will definitely not work on the blend float format
    * used for SRGB here and I think OpenGL expects this to work as expected
    * (that is incoming values converted to srgb then logic op applied).
    */
   for (unsigned i = 0; i < src_count; ++i) {
      dst[i] = lp_build_blend_aos(gallivm,
                                  &variant->key.blend,
                                  out_format,
                                  row_type,
                                  rt,
                                  src[i],
                                  has_alpha ? NULL : src_alpha[i],
                                  src1[i],
                                  has_alpha ? NULL : src1_alpha[i],
                                  dst[i],
                                  partial_mask ? src_mask[i] : NULL,
                                  blend_color,
                                  has_alpha ? NULL : blend_alpha,
                                  swizzle,
                                  pad_inline ? 4 : dst_channels);
   }

   convert_from_blend_type(gallivm, block_size, out_format_desc,
                           row_type, dst_type, dst, src_count);

   /* Split the blend rows back to memory rows */
   if (dst_count > src_count) {
      row_type.length = dst_type.length * (dst_count / src_count);

      if (src_count == 1) {
         dst[1] = lp_build_extract_range(gallivm, dst[0],
                                         row_type.length / 2,
                                         row_type.length / 2);
         dst[0] = lp_build_extract_range(gallivm, dst[0],
                                         0, row_type.length / 2);

         row_type.length /= 2;
         src_count *= 2;
      }

      dst[3] = lp_build_extract_range(gallivm, dst[1], row_type.length / 2,
                                      row_type.length / 2);
      dst[2] = lp_build_extract_range(gallivm, dst[1], 0, row_type.length / 2);
      dst[1] = lp_build_extract_range(gallivm, dst[0], row_type.length / 2,
                                      row_type.length / 2);
      dst[0] = lp_build_extract_range(gallivm, dst[0], 0, row_type.length / 2);

      row_type.length /= 2;
      src_count *= 2;
   }

   /*
    * Store blend result to memory
    */
   if (is_1d) {
      store_unswizzled_block(gallivm, color_type, color_ptr, stride, block_width, 1,
                             dst, dst_type, dst_count / 4, dst_alignment);
   } else {
      store_unswizzled_block(gallivm, color_type, color_ptr, stride, block_width,
                             block_height,
                             dst, dst_type, dst_count, dst_alignment);
   }

   if (do_branch) {
      lp_build_mask_end(&mask_ctx);
   }

   if (fpstate) {
      lp_build_fpstate_set(gallivm, fpstate);
   }
}


/**
 * Generate the runtime callable function for the whole fragment pipeline.
 * Note that the function which we generate operates on a block of 16
 * pixels at at time.  The block contains 2x2 quads.  Each quad contains
 * 2x2 pixels.
 */
static void
generate_fragment(struct llvmpipe_context *lp,
                  struct lp_fragment_shader *shader,
                  struct lp_fragment_shader_variant *variant,
                  unsigned partial_mask)
{
   assert(partial_mask == RAST_WHOLE ||
          partial_mask == RAST_EDGE_TEST);

   struct nir_shader *nir = shader->base.ir.nir;
   struct gallivm_state *gallivm = variant->gallivm;
   struct lp_fragment_shader_variant_key *key = &variant->key;
   struct lp_shader_input inputs[PIPE_MAX_SHADER_INPUTS];
   LLVMTypeRef fs_elem_type;
   LLVMTypeRef blend_vec_type;
   LLVMTypeRef arg_types[16];
   LLVMTypeRef func_type;
   LLVMTypeRef int32_type = LLVMInt32TypeInContext(gallivm->context);
   LLVMTypeRef int32p_type = LLVMPointerType(int32_type, 0);
   LLVMTypeRef int8_type = LLVMInt8TypeInContext(gallivm->context);
   LLVMTypeRef int8p_type = LLVMPointerType(int8_type, 0);
   LLVMValueRef context_ptr;
   LLVMValueRef resources_ptr;
   LLVMValueRef x;
   LLVMValueRef y;
   LLVMValueRef a0_ptr;
   LLVMValueRef dadx_ptr;
   LLVMValueRef dady_ptr;
   LLVMValueRef color_ptr_ptr;
   LLVMValueRef stride_ptr;
   LLVMValueRef color_sample_stride_ptr;
   LLVMValueRef depth_ptr;
   LLVMValueRef depth_stride;
   LLVMValueRef depth_sample_stride;
   LLVMValueRef mask_input;
   LLVMValueRef thread_data_ptr;
   LLVMBasicBlockRef block;
   LLVMBuilderRef builder;
   struct lp_build_interp_soa_context interp;
   LLVMValueRef fs_mask[(16 / 4) * LP_MAX_SAMPLES];
   LLVMValueRef fs_out_color[LP_MAX_SAMPLES][PIPE_MAX_COLOR_BUFS][TGSI_NUM_CHANNELS][16 / 4];
   LLVMValueRef function;
   LLVMValueRef facing;
   const bool dual_source_blend = key->blend.rt[0].blend_enable &&
                                  util_blend_state_is_dual(&key->blend, 0);

   assert(lp_native_vector_width / 32 >= 4);

   /* Adjust color input interpolation according to flatshade state:
    */
   nir_foreach_shader_in_variable(var, nir) {
      unsigned idx = var->data.driver_location;
      unsigned slots = nir_variable_count_slots(var, var->type);
      memcpy(&inputs[idx], &shader->inputs[idx], (sizeof inputs[0] * slots));
      for (unsigned s = 0; s < slots; s++) {
         if (inputs[idx + s].interp == LP_INTERP_COLOR)
            inputs[idx + s].interp = key->flatshade ? LP_INTERP_CONSTANT : LP_INTERP_PERSPECTIVE;
      }
   }

   /* TODO: actually pick these based on the fs and color buffer
    * characteristics. */

   struct lp_type fs_type;
   memset(&fs_type, 0, sizeof fs_type);
   fs_type.floating = true;      /* floating point values */
   fs_type.sign = true;          /* values are signed */
   fs_type.norm = false;         /* values are not limited to [0,1] or [-1,1] */
   fs_type.width = 32;           /* 32-bit float */
   fs_type.length = MIN2(lp_native_vector_width / 32, 16); /* n*4 elements per vector */

   struct lp_type blend_type;
   memset(&blend_type, 0, sizeof blend_type);
   blend_type.floating = false; /* values are integers */
   blend_type.sign = false;     /* values are unsigned */
   blend_type.norm = true;      /* values are in [0,1] or [-1,1] */
   blend_type.width = 8;        /* 8-bit ubyte values */
   blend_type.length = 16;      /* 16 elements per vector */

   /*
    * Generate the function prototype. Any change here must be reflected in
    * lp_jit.h's lp_jit_frag_func function pointer type, and vice-versa.
    */

   fs_elem_type = lp_build_elem_type(gallivm, fs_type);

   blend_vec_type = lp_build_vec_type(gallivm, blend_type);

   char func_name[64];
   snprintf(func_name, sizeof(func_name), "fs_variant_%s",
            partial_mask ? "partial" : "whole");

   arg_types[0] = variant->jit_context_ptr_type;       /* context */
   arg_types[1] = variant->jit_resources_ptr_type;       /* context */
   arg_types[2] = int32_type;                          /* x */
   arg_types[3] = int32_type;                          /* y */
   arg_types[4] = int32_type;                          /* facing */
   arg_types[5] = LLVMPointerType(fs_elem_type, 0);    /* a0 */
   arg_types[6] = LLVMPointerType(fs_elem_type, 0);    /* dadx */
   arg_types[7] = LLVMPointerType(fs_elem_type, 0);    /* dady */
   arg_types[8] = LLVMPointerType(int8p_type, 0);  /* color */
   arg_types[9] = int8p_type;       /* depth */
   arg_types[10] = LLVMInt64TypeInContext(gallivm->context);  /* mask_input */
   arg_types[11] = variant->jit_thread_data_ptr_type;  /* per thread data */
   arg_types[12] = int32p_type;     /* stride */
   arg_types[13] = int32_type;                         /* depth_stride */
   arg_types[14] = int32p_type;     /* color sample strides */
   arg_types[15] = int32_type;                         /* depth sample stride */

   func_type = LLVMFunctionType(LLVMVoidTypeInContext(gallivm->context),
                                arg_types, ARRAY_SIZE(arg_types), 0);

   function = LLVMAddFunction(gallivm->module, func_name, func_type);
   LLVMSetFunctionCallConv(function, LLVMCCallConv);

   variant->function[partial_mask] = function;
   variant->function_name[partial_mask] = MALLOC(strlen(func_name)+1);
   strcpy(variant->function_name[partial_mask], func_name);

   /* XXX: need to propagate noalias down into color param now we are
    * passing a pointer-to-pointer?
    */
   for (unsigned i = 0; i < ARRAY_SIZE(arg_types); ++i)
      if (LLVMGetTypeKind(arg_types[i]) == LLVMPointerTypeKind)
         lp_add_function_attr(function, i + 1, LP_FUNC_ATTR_NOALIAS);

   if (variant->gallivm->cache->data_size) {
      gallivm_stub_func(gallivm, function);
      return;
   }

   context_ptr  = LLVMGetParam(function, 0);
   resources_ptr  = LLVMGetParam(function, 1);
   x            = LLVMGetParam(function, 2);
   y            = LLVMGetParam(function, 3);
   facing       = LLVMGetParam(function, 4);
   a0_ptr       = LLVMGetParam(function, 5);
   dadx_ptr     = LLVMGetParam(function, 6);
   dady_ptr     = LLVMGetParam(function, 7);
   color_ptr_ptr = LLVMGetParam(function, 8);
   depth_ptr    = LLVMGetParam(function, 9);
   mask_input   = LLVMGetParam(function, 10);
   thread_data_ptr  = LLVMGetParam(function, 11);
   stride_ptr   = LLVMGetParam(function, 12);
   depth_stride = LLVMGetParam(function, 13);
   color_sample_stride_ptr = LLVMGetParam(function, 14);
   depth_sample_stride = LLVMGetParam(function, 15);

   lp_build_name(context_ptr, "context");
   lp_build_name(resources_ptr, "resources");
   lp_build_name(x, "x");
   lp_build_name(y, "y");
   lp_build_name(a0_ptr, "a0");
   lp_build_name(dadx_ptr, "dadx");
   lp_build_name(dady_ptr, "dady");
   lp_build_name(color_ptr_ptr, "color_ptr_ptr");
   lp_build_name(depth_ptr, "depth");
   lp_build_name(mask_input, "mask_input");
   lp_build_name(thread_data_ptr, "thread_data");
   lp_build_name(stride_ptr, "stride_ptr");
   lp_build_name(depth_stride, "depth_stride");
   lp_build_name(color_sample_stride_ptr, "color_sample_stride_ptr");
   lp_build_name(depth_sample_stride, "depth_sample_stride");

   /*
    * Function body
    */

   block = LLVMAppendBasicBlockInContext(gallivm->context, function, "entry");
   builder = gallivm->builder;
   assert(builder);
   LLVMPositionBuilderAtEnd(builder, block);

   /* code generated texture sampling */
   struct lp_build_sampler_soa *sampler =
      lp_llvm_sampler_soa_create(lp_fs_variant_key_samplers(key),
                                 MAX2(key->nr_samplers,
                                      key->nr_sampler_views));
   struct lp_build_image_soa *image =
      lp_bld_llvm_image_soa_create(lp_fs_variant_key_images(key), key->nr_images);

   unsigned num_fs = 16 / fs_type.length; /* number of loops per 4x4 stamp */
   /* for 1d resources only run "upper half" of stamp */
   if (key->resource_1d)
      num_fs /= 2;

   {
      LLVMValueRef num_loop = lp_build_const_int32(gallivm, num_fs);
      LLVMTypeRef mask_type = lp_build_int_vec_type(gallivm, fs_type);
      LLVMValueRef num_loop_samp =
         lp_build_const_int32(gallivm, num_fs * key->coverage_samples);
      LLVMValueRef mask_store =
         lp_build_array_alloca(gallivm, mask_type,
                               num_loop_samp, "mask_store");
      LLVMTypeRef flt_type = LLVMFloatTypeInContext(gallivm->context);
      LLVMValueRef glob_sample_pos =
         LLVMAddGlobal(gallivm->module,
                       LLVMArrayType(flt_type, key->coverage_samples * 2), "");
      LLVMSetLinkage(glob_sample_pos, LLVMInternalLinkage);
      LLVMValueRef sample_pos_array;

      if (key->multisample && key->coverage_samples == 4) {
         LLVMValueRef sample_pos_arr[8];
         for (unsigned i = 0; i < 4; i++) {
            sample_pos_arr[i * 2] = LLVMConstReal(flt_type,
                                                  lp_sample_pos_4x[i][0]);
            sample_pos_arr[i * 2 + 1] = LLVMConstReal(flt_type,
                                                      lp_sample_pos_4x[i][1]);
         }
         sample_pos_array =
            LLVMConstArray(LLVMFloatTypeInContext(gallivm->context),
                           sample_pos_arr, 8);
      } else {
         LLVMValueRef sample_pos_arr[2];
         sample_pos_arr[0] = LLVMConstReal(flt_type, 0.5);
         sample_pos_arr[1] = LLVMConstReal(flt_type, 0.5);
         sample_pos_array =
            LLVMConstArray(LLVMFloatTypeInContext(gallivm->context),
                           sample_pos_arr, 2);
      }
      LLVMSetInitializer(glob_sample_pos, sample_pos_array);

      LLVMValueRef color_store[PIPE_MAX_COLOR_BUFS][TGSI_NUM_CHANNELS];
      bool pixel_center_integer = nir->info.fs.pixel_center_integer;

      /*
       * The shader input interpolation info is not explicitely baked in the
       * shader key, but everything it derives from (TGSI, and flatshade) is
       * already included in the shader key.
       */
      lp_build_interp_soa_init(&interp,
                               gallivm,
                               nir->num_inputs,
                               inputs,
                               pixel_center_integer,
                               key->coverage_samples,
                               LLVMTypeOf(sample_pos_array),
                               glob_sample_pos,
                               num_loop,
                               builder, fs_type,
                               a0_ptr, dadx_ptr, dady_ptr,
                               x, y);

      for (unsigned i = 0; i < num_fs; i++) {
         if (key->multisample) {
            LLVMValueRef smask_val =
               LLVMBuildLoad2(builder, int32_type,
                              lp_jit_context_sample_mask(gallivm, variant->jit_context_type, context_ptr),
                              "");

            /*
             * For multisampling, extract the per-sample mask from the
             * incoming 64-bit mask, store to the per sample mask storage. Or
             * all of them together to generate the fragment shader
             * mask. (sample shading TODO).  Take the incoming state coverage
             * mask into account.
             */
            for (unsigned s = 0; s < key->coverage_samples; s++) {
               LLVMValueRef sindexi =
                  lp_build_const_int32(gallivm, i + (s * num_fs));
               LLVMValueRef sample_mask_ptr =
                  LLVMBuildGEP2(builder, mask_type, mask_store, &sindexi, 1,
                                "sample_mask_ptr");
               LLVMValueRef s_mask =
                  generate_quad_mask(gallivm, fs_type,
                                     i * fs_type.length / 4, s, mask_input);
               LLVMValueRef smask_bit =
                  LLVMBuildAnd(builder, smask_val,
                               lp_build_const_int32(gallivm, (1 << s)), "");
               LLVMValueRef cmp =
                  LLVMBuildICmp(builder, LLVMIntNE, smask_bit,
                                lp_build_const_int32(gallivm, 0), "");
               smask_bit = LLVMBuildSExt(builder, cmp, int32_type, "");
               smask_bit = lp_build_broadcast(gallivm, mask_type, smask_bit);

               s_mask = LLVMBuildAnd(builder, s_mask, smask_bit, "");
               LLVMBuildStore(builder, s_mask, sample_mask_ptr);
            }
         } else {
            LLVMValueRef mask;
            LLVMValueRef indexi = lp_build_const_int32(gallivm, i);
            LLVMValueRef mask_ptr = LLVMBuildGEP2(builder, mask_type, mask_store,
                                                  &indexi, 1, "mask_ptr");

            if (partial_mask) {
               mask = generate_quad_mask(gallivm, fs_type,
                                         i * fs_type.length / 4, 0, mask_input);
            } else {
               mask = lp_build_const_int_vec(gallivm, fs_type, ~0);
            }
            LLVMBuildStore(builder, mask, mask_ptr);
         }
      }

      generate_fs_loop(gallivm,
                       shader, key,
                       builder,
                       fs_type,
                       variant->jit_context_type,
                       context_ptr,
                       variant->jit_resources_type,
                       resources_ptr,
                       LLVMTypeOf(sample_pos_array),
                       glob_sample_pos,
                       num_loop,
                       &interp,
                       sampler,
                       image,
                       mask_type,
                       mask_store, /* output */
                       color_store,
                       depth_ptr,
                       depth_stride,
                       depth_sample_stride,
                       color_ptr_ptr,
                       stride_ptr,
                       color_sample_stride_ptr,
                       facing,
                       variant->jit_thread_data_type,
                       thread_data_ptr);

      LLVMTypeRef fs_vec_type = lp_build_vec_type(gallivm, fs_type);
      for (unsigned i = 0; i < num_fs; i++) {
         LLVMValueRef ptr;
         for (unsigned s = 0; s < key->coverage_samples; s++) {
            int idx = (i + (s * num_fs));
            LLVMValueRef sindexi = lp_build_const_int32(gallivm, idx);
            ptr = LLVMBuildGEP2(builder, mask_type, mask_store, &sindexi, 1, "");

            fs_mask[idx] = LLVMBuildLoad2(builder, mask_type, ptr, "smask");
         }

         for (unsigned s = 0; s < key->min_samples; s++) {
            /* This is fucked up need to reorganize things */
            int idx = s * num_fs + i;
            LLVMValueRef sindexi = lp_build_const_int32(gallivm, idx);
            for (unsigned cbuf = 0; cbuf < key->nr_cbufs; cbuf++) {
               for (unsigned chan = 0; chan < TGSI_NUM_CHANNELS; ++chan) {
                  ptr = LLVMBuildGEP2(builder, fs_vec_type,
                                      color_store[cbuf][chan],
                                      &sindexi, 1, "");
                  fs_out_color[s][cbuf][chan][i] = ptr;
               }
            }
            if (dual_source_blend) {
               /* only support one dual source blend target hence always use
                * output 1
                */
               for (unsigned chan = 0; chan < TGSI_NUM_CHANNELS; ++chan) {
                  ptr = LLVMBuildGEP2(builder, fs_vec_type,
                                      color_store[1][chan],
                                      &sindexi, 1, "");
                  fs_out_color[s][1][chan][i] = ptr;
               }
            }
         }
      }
   }

   lp_bld_llvm_sampler_soa_destroy(sampler);
   lp_bld_llvm_image_soa_destroy(image);

   /* Loop over color outputs / color buffers to do blending */
   for (unsigned cbuf = 0; cbuf < key->nr_cbufs; cbuf++) {
      if (key->cbuf_format[cbuf] != PIPE_FORMAT_NONE &&
          (key->blend.rt[cbuf].blend_enable || key->blend.logicop_enable ||
           find_output_by_frag_result(nir, FRAG_RESULT_DATA0 + cbuf) != -1)) {
         LLVMValueRef color_ptr;
         LLVMValueRef stride;
         LLVMValueRef sample_stride = NULL;
         LLVMValueRef index = lp_build_const_int32(gallivm, cbuf);

         bool do_branch = ((key->depth.enabled
                            || key->stencil[0].enabled
                            || key->alpha.enabled)
                           && !nir->info.fs.uses_discard);

         color_ptr = LLVMBuildLoad2(builder, int8p_type,
                                    LLVMBuildGEP2(builder, int8p_type, color_ptr_ptr,
                                                 &index, 1, ""),
                                    "");

         stride = LLVMBuildLoad2(builder, int32_type,
                                 LLVMBuildGEP2(builder, int32_type, stride_ptr,
                                             &index, 1, ""),
                                 "");

         if (key->cbuf_nr_samples[cbuf] > 1)
            sample_stride = LLVMBuildLoad2(builder, int32_type,
                                           LLVMBuildGEP2(builder,
                                                         int32_type,
                                                         color_sample_stride_ptr,
                                                         &index, 1, ""), "");

         for (unsigned s = 0; s < key->cbuf_nr_samples[cbuf]; s++) {
            unsigned mask_idx = num_fs * (key->multisample ? s : 0);
            unsigned out_idx = key->min_samples == 1 ? 0 : s;
            LLVMValueRef out_ptr = color_ptr;

            if (sample_stride) {
               LLVMValueRef sample_offset =
                  LLVMBuildMul(builder, sample_stride,
                               lp_build_const_int32(gallivm, s), "");
               out_ptr = LLVMBuildGEP2(builder, int8_type, out_ptr, &sample_offset, 1, "");
            }
            out_ptr = LLVMBuildBitCast(builder, out_ptr,
                                       LLVMPointerType(blend_vec_type, 0), "");

            lp_build_name(out_ptr, "color_ptr%d", cbuf);

            generate_unswizzled_blend(gallivm, cbuf, variant,
                                      key->cbuf_format[cbuf],
                                      num_fs, fs_type, &fs_mask[mask_idx],
                                      fs_out_color[out_idx],
                                      variant->jit_context_type,
                                      context_ptr, blend_vec_type, out_ptr, stride,
                                      partial_mask, do_branch);
         }
      }
   }

   LLVMBuildRetVoid(builder);

   gallivm_verify_function(gallivm, function);
}


static void
dump_fs_variant_key(struct lp_fragment_shader_variant_key *key)
{
   debug_printf("fs variant %p:\n", (void *) key);

   if (key->flatshade) {
      debug_printf("flatshade = 1\n");
   }
   if (key->depth_clamp)
      debug_printf("depth_clamp = 1\n");

   if (key->restrict_depth_values)
      debug_printf("restrict_depth_values = 1\n");

   if (key->multisample) {
      debug_printf("multisample = 1\n");
      debug_printf("coverage samples = %d\n", key->coverage_samples);
      debug_printf("min samples = %d\n", key->min_samples);
   }
   for (unsigned i = 0; i < key->nr_cbufs; ++i) {
      debug_printf("cbuf_format[%u] = %s\n", i, util_format_name(key->cbuf_format[i]));
      debug_printf("cbuf nr_samples[%u] = %d\n", i, key->cbuf_nr_samples[i]);
   }
   if (key->depth.enabled || key->stencil[0].enabled) {
      debug_printf("depth.format = %s\n", util_format_name(key->zsbuf_format));
      debug_printf("depth nr_samples = %d\n", key->zsbuf_nr_samples);
   }
   if (key->depth.enabled) {
      debug_printf("depth.func = %s\n", util_str_func(key->depth.func, true));
      debug_printf("depth.writemask = %u\n", key->depth.writemask);
   }

   for (unsigned i = 0; i < 2; ++i) {
      if (key->stencil[i].enabled) {
         debug_printf("stencil[%u].func = %s\n", i, util_str_func(key->stencil[i].func, true));
         debug_printf("stencil[%u].fail_op = %s\n", i, util_str_stencil_op(key->stencil[i].fail_op, true));
         debug_printf("stencil[%u].zpass_op = %s\n", i, util_str_stencil_op(key->stencil[i].zpass_op, true));
         debug_printf("stencil[%u].zfail_op = %s\n", i, util_str_stencil_op(key->stencil[i].zfail_op, true));
         debug_printf("stencil[%u].valuemask = 0x%x\n", i, key->stencil[i].valuemask);
         debug_printf("stencil[%u].writemask = 0x%x\n", i, key->stencil[i].writemask);
      }
   }

   if (key->alpha.enabled) {
      debug_printf("alpha.func = %s\n", util_str_func(key->alpha.func, true));
   }

   if (key->occlusion_count) {
      debug_printf("occlusion_count = 1\n");
   }

   if (key->blend.logicop_enable) {
      debug_printf("blend.logicop_func = %s\n", util_str_logicop(key->blend.logicop_func, true));
   } else if (key->blend.rt[0].blend_enable) {
      debug_printf("blend.rgb_func = %s\n",   util_str_blend_func  (key->blend.rt[0].rgb_func, true));
      debug_printf("blend.rgb_src_factor = %s\n",   util_str_blend_factor(key->blend.rt[0].rgb_src_factor, true));
      debug_printf("blend.rgb_dst_factor = %s\n",   util_str_blend_factor(key->blend.rt[0].rgb_dst_factor, true));
      debug_printf("blend.alpha_func = %s\n",       util_str_blend_func  (key->blend.rt[0].alpha_func, true));
      debug_printf("blend.alpha_src_factor = %s\n", util_str_blend_factor(key->blend.rt[0].alpha_src_factor, true));
      debug_printf("blend.alpha_dst_factor = %s\n", util_str_blend_factor(key->blend.rt[0].alpha_dst_factor, true));
   }
   debug_printf("blend.colormask = 0x%x\n", key->blend.rt[0].colormask);
   if (key->blend.alpha_to_coverage) {
      debug_printf("blend.alpha_to_coverage is enabled\n");
   }
   for (unsigned i = 0; i < key->nr_samplers; ++i) {
      const struct lp_sampler_static_state *samplers = lp_fs_variant_key_samplers(key);
      const struct lp_static_sampler_state *sampler = &samplers[i].sampler_state;
      debug_printf("sampler[%u] = \n", i);
      debug_printf("  .wrap = %s %s %s\n",
                   util_str_tex_wrap(sampler->wrap_s, true),
                   util_str_tex_wrap(sampler->wrap_t, true),
                   util_str_tex_wrap(sampler->wrap_r, true));
      debug_printf("  .min_img_filter = %s\n",
                   util_str_tex_filter(sampler->min_img_filter, true));
      debug_printf("  .min_mip_filter = %s\n",
                   util_str_tex_mipfilter(sampler->min_mip_filter, true));
      debug_printf("  .mag_img_filter = %s\n",
                   util_str_tex_filter(sampler->mag_img_filter, true));
      if (sampler->compare_mode != PIPE_TEX_COMPARE_NONE)
         debug_printf("  .compare_func = %s\n", util_str_func(sampler->compare_func, true));
      debug_printf("  .normalized_coords = %u\n", sampler->normalized_coords);
      debug_printf("  .min_max_lod_equal = %u\n", sampler->min_max_lod_equal);
      debug_printf("  .lod_bias_non_zero = %u\n", sampler->lod_bias_non_zero);
      debug_printf("  .apply_min_lod = %u\n", sampler->apply_min_lod);
      debug_printf("  .apply_max_lod = %u\n", sampler->apply_max_lod);
      debug_printf("  .reduction_mode = %u\n", sampler->reduction_mode);
      debug_printf("  .aniso = %u\n", sampler->aniso);
   }
   for (unsigned i = 0; i < key->nr_sampler_views; ++i) {
      const struct lp_sampler_static_state *samplers = lp_fs_variant_key_samplers(key);
      const struct lp_static_texture_state *texture = &samplers[i].texture_state;
      debug_printf("texture[%u] = \n", i);
      debug_printf("  .format = %s\n",
                   util_format_name(texture->format));
      debug_printf("  .target = %s\n",
                   util_str_tex_target(texture->target, true));
      debug_printf("  .level_zero_only = %u\n",
                   texture->level_zero_only);
      debug_printf("  .pot = %u %u %u\n",
                   texture->pot_width,
                   texture->pot_height,
                   texture->pot_depth);
   }
   struct lp_image_static_state *images = lp_fs_variant_key_images(key);
   for (unsigned i = 0; i < key->nr_images; ++i) {
      const struct lp_static_texture_state *image = &images[i].image_state;
      debug_printf("image[%u] = \n", i);
      debug_printf("  .format = %s\n",
                   util_format_name(image->format));
      debug_printf("  .target = %s\n",
                   util_str_tex_target(image->target, true));
      debug_printf("  .level_zero_only = %u\n",
                   image->level_zero_only);
      debug_printf("  .pot = %u %u %u\n",
                   image->pot_width,
                   image->pot_height,
                   image->pot_depth);
   }
}


const char *
lp_debug_fs_kind(enum lp_fs_kind kind)
{
   switch (kind) {
   case LP_FS_KIND_GENERAL:
      return "GENERAL";
   case LP_FS_KIND_BLIT_RGBA:
      return "BLIT_RGBA";
   case LP_FS_KIND_BLIT_RGB1:
      return "BLIT_RGB1";
   case LP_FS_KIND_AERO_MINIFICATION:
      return "AERO_MINIFICATION";
   case LP_FS_KIND_LLVM_LINEAR:
      return "LLVM_LINEAR";
   default:
      return "unknown";
   }
}


void
lp_debug_fs_variant(struct lp_fragment_shader_variant *variant)
{
   debug_printf("llvmpipe: Fragment shader #%u variant #%u:\n",
                variant->shader->no, variant->no);
   nir_print_shader(variant->shader->base.ir.nir, stderr);
   dump_fs_variant_key(&variant->key);
   debug_printf("variant->opaque = %u\n", variant->opaque);
   debug_printf("variant->potentially_opaque = %u\n", variant->potentially_opaque);
   debug_printf("variant->blit = %u\n", variant->blit);
   debug_printf("shader->kind = %s\n", lp_debug_fs_kind(variant->shader->kind));
   debug_printf("\n");
}


static void
lp_fs_get_ir_cache_key(struct lp_fragment_shader_variant *variant,
                       unsigned char ir_sha1_cache_key[20])
{
   struct blob blob = { 0 };
   unsigned ir_size;
   void *ir_binary;

   blob_init(&blob);
   nir_serialize(&blob, variant->shader->base.ir.nir, true);
   ir_binary = blob.data;
   ir_size = blob.size;

   struct mesa_sha1 ctx;
   _mesa_sha1_init(&ctx);
   _mesa_sha1_update(&ctx, &variant->key, variant->shader->variant_key_size);
   _mesa_sha1_update(&ctx, ir_binary, ir_size);
   _mesa_sha1_final(&ctx, ir_sha1_cache_key);

   blob_finish(&blob);
}


/**
 * Generate a new fragment shader variant from the shader code and
 * other state indicated by the key.
 */
static struct lp_fragment_shader_variant *
generate_variant(struct llvmpipe_context *lp,
                 struct lp_fragment_shader *shader,
                 const struct lp_fragment_shader_variant_key *key)
{
   struct nir_shader *nir = shader->base.ir.nir;
   struct lp_fragment_shader_variant *variant =
      MALLOC(sizeof *variant + shader->variant_key_size - sizeof variant->key);
   if (!variant)
      return NULL;

   memset(variant, 0, sizeof(*variant));

   pipe_reference_init(&variant->reference, 1);
   lp_fs_reference(lp, &variant->shader, shader);

   memcpy(&variant->key, key, shader->variant_key_size);

   struct llvmpipe_screen *screen = llvmpipe_screen(lp->pipe.screen);
   struct lp_cached_code cached = { 0 };
   unsigned char ir_sha1_cache_key[20];
   bool needs_caching = false;
   if (shader->base.ir.nir) {
      lp_fs_get_ir_cache_key(variant, ir_sha1_cache_key);

      lp_disk_cache_find_shader(screen, &cached, ir_sha1_cache_key);
      if (!cached.data_size)
         needs_caching = true;
   }

   char module_name[64];
   snprintf(module_name, sizeof(module_name), "fs%u_variant%u",
            shader->no, shader->variants_created);
   variant->gallivm = gallivm_create(module_name, &lp->context, &cached);
   if (!variant->gallivm) {
      FREE(variant);
      return NULL;
   }

   variant->list_item_global.base = variant;
   variant->list_item_local.base = variant;
   variant->no = shader->variants_created++;

   /*
    * Determine whether we are touching all channels in the color buffer.
    */
   const struct util_format_description *cbuf0_format_desc = NULL;
   bool fullcolormask = false;
   if (key->nr_cbufs == 1) {
      cbuf0_format_desc = util_format_description(key->cbuf_format[0]);
      fullcolormask = util_format_colormask_full(cbuf0_format_desc,
                                                 key->blend.rt[0].colormask);
   }

   /* The scissor is ignored here as only tiles inside the scissoring
    * rectangle will refer to this.
    */
   const bool no_kill =
         fullcolormask &&
         !key->stencil[0].enabled &&
         !key->alpha.enabled &&
         !key->multisample &&
         !key->blend.alpha_to_coverage &&
         !key->depth.enabled &&
         !nir->info.fs.uses_discard &&
         !(nir->info.outputs_written & BITFIELD64_BIT(FRAG_RESULT_SAMPLE_MASK)) &&
         !nir->info.fs.uses_fbfetch_output;

   variant->opaque =
         no_kill &&
         !key->blend.logicop_enable &&
         !key->blend.rt[0].blend_enable
         ? true : false;

   variant->potentially_opaque =
         no_kill &&
         !key->blend.logicop_enable &&
         key->blend.rt[0].blend_enable &&
         key->blend.rt[0].rgb_func == PIPE_BLEND_ADD &&
         key->blend.rt[0].rgb_dst_factor == PIPE_BLENDFACTOR_INV_SRC_ALPHA &&
         key->blend.rt[0].alpha_func == key->blend.rt[0].rgb_func &&
         key->blend.rt[0].alpha_dst_factor == key->blend.rt[0].rgb_dst_factor &&
         shader->base.type == PIPE_SHADER_IR_TGSI &&
         /*
          * FIXME: for NIR, all of the fields of info.xxx (except info.base)
          * are zeros, hence shader analysis (here and elsewhere) using these
          * bits cannot work and will silently fail (cbuf is the only pointer
          * field, hence causing a crash).
          */
         shader->info.cbuf[0][3].file != TGSI_FILE_NULL
         ? true : false;

   /* We only care about opaque blits for now */
   if (variant->opaque &&
       (shader->kind == LP_FS_KIND_BLIT_RGBA ||
        shader->kind == LP_FS_KIND_BLIT_RGB1)) {
      const struct lp_sampler_static_state *samp0 =
         lp_fs_variant_key_sampler_idx(key, 0);
      assert(samp0);

      const enum pipe_format texture_format = samp0->texture_state.format;
      const enum pipe_texture_target target = samp0->texture_state.target;
      const unsigned min_img_filter = samp0->sampler_state.min_img_filter;
      const unsigned mag_img_filter = samp0->sampler_state.mag_img_filter;

      unsigned min_mip_filter;
      if (samp0->texture_state.level_zero_only) {
         min_mip_filter = PIPE_TEX_MIPFILTER_NONE;
      } else {
         min_mip_filter = samp0->sampler_state.min_mip_filter;
      }

      if (target == PIPE_TEXTURE_2D &&
          min_img_filter == PIPE_TEX_FILTER_NEAREST &&
          mag_img_filter == PIPE_TEX_FILTER_NEAREST &&
          min_mip_filter == PIPE_TEX_MIPFILTER_NONE &&
          ((texture_format &&
            util_is_format_compatible(util_format_description(texture_format),
                                      cbuf0_format_desc)) ||
           (shader->kind == LP_FS_KIND_BLIT_RGB1 &&
            (texture_format == PIPE_FORMAT_B8G8R8A8_UNORM ||
             texture_format == PIPE_FORMAT_B8G8R8X8_UNORM) &&
            (key->cbuf_format[0] == PIPE_FORMAT_B8G8R8A8_UNORM ||
             key->cbuf_format[0] == PIPE_FORMAT_B8G8R8X8_UNORM)))) {
         variant->blit = 1;
      }
   }

   /* Determine whether this shader + pipeline state is a candidate for
    * the linear path.
    */
   const bool linear_pipeline =
         !key->stencil[0].enabled &&
         !key->depth.enabled &&
         !nir->info.fs.uses_discard &&
         !key->blend.logicop_enable &&
         (key->cbuf_format[0] == PIPE_FORMAT_B8G8R8A8_UNORM ||
          key->cbuf_format[0] == PIPE_FORMAT_B8G8R8X8_UNORM ||
          key->cbuf_format[0] == PIPE_FORMAT_R8G8B8A8_UNORM ||
          key->cbuf_format[0] == PIPE_FORMAT_R8G8B8X8_UNORM);

   memcpy(&variant->key, key, sizeof *key);

   if ((LP_DEBUG & DEBUG_FS) || (gallivm_debug & GALLIVM_DEBUG_IR)) {
      lp_debug_fs_variant(variant);
   }

   llvmpipe_fs_variant_fastpath(variant);

   lp_jit_init_types(variant);

   if (variant->jit_function[RAST_EDGE_TEST] == NULL)
      generate_fragment(lp, shader, variant, RAST_EDGE_TEST);

   if (variant->jit_function[RAST_WHOLE] == NULL) {
      if (variant->opaque) {
         /* Specialized shader, which doesn't need to read the color buffer. */
         generate_fragment(lp, shader, variant, RAST_WHOLE);
      }
   }

   if (linear_pipeline) {
      /* Currently keeping both the old fastpaths and new linear path
       * active.  The older code is still somewhat faster for the cases
       * it covers.
       *
       * XXX: consider restricting this to aero-mode only.
       */
      if (fullcolormask &&
          !key->alpha.enabled &&
          !key->blend.alpha_to_coverage) {
         llvmpipe_fs_variant_linear_fastpath(variant);
      }

      /* If the original fastpath doesn't cover this variant, try the new
       * code:
       */
      if (variant->jit_linear == NULL) {
         if (shader->kind == LP_FS_KIND_BLIT_RGBA ||
             shader->kind == LP_FS_KIND_BLIT_RGB1 ||
             shader->kind == LP_FS_KIND_LLVM_LINEAR) {
            llvmpipe_fs_variant_linear_llvm(lp, shader, variant);
         }
      }
   } else {
      if (LP_DEBUG & DEBUG_LINEAR) {
         lp_debug_fs_variant(variant);
         debug_printf("    ----> no linear path for this variant\n");
      }
   }

   /*
    * Compile everything
    */

#if GALLIVM_USE_ORCJIT
/* module has been moved into ORCJIT after gallivm_compile_module */
   variant->nr_instrs += lp_build_count_ir_module(variant->gallivm->module);

   gallivm_compile_module(variant->gallivm);
#else
   gallivm_compile_module(variant->gallivm);

   variant->nr_instrs += lp_build_count_ir_module(variant->gallivm->module);
#endif

   if (variant->function[RAST_EDGE_TEST]) {
      variant->jit_function[RAST_EDGE_TEST] = (lp_jit_frag_func)
            gallivm_jit_function(variant->gallivm,
                                 variant->function[RAST_EDGE_TEST],
                                 variant->function_name[RAST_EDGE_TEST]);
   }

   if (variant->function[RAST_WHOLE]) {
      variant->jit_function[RAST_WHOLE] = (lp_jit_frag_func)
         gallivm_jit_function(variant->gallivm,
                              variant->function[RAST_WHOLE],
                              variant->function_name[RAST_WHOLE]);
   } else if (!variant->jit_function[RAST_WHOLE]) {
      variant->jit_function[RAST_WHOLE] = (lp_jit_frag_func)
         variant->jit_function[RAST_EDGE_TEST];
   }

   if (linear_pipeline) {
      if (variant->linear_function) {
         variant->jit_linear_llvm = (lp_jit_linear_llvm_func)
            gallivm_jit_function(variant->gallivm, variant->linear_function,
                                 variant->linear_function_name);
      }

      /*
       * This must be done after LLVM compilation, as it will call the JIT'ed
       * code to determine active inputs.
       */
      lp_linear_check_variant(variant);
   }

   if (needs_caching) {
      lp_disk_cache_insert_shader(screen, &cached, ir_sha1_cache_key);
   }

   gallivm_free_ir(variant->gallivm);

   return variant;
}


static void *
llvmpipe_create_fs_state(struct pipe_context *pipe,
                         const struct pipe_shader_state *templ)
{
   struct llvmpipe_context *llvmpipe = llvmpipe_context(pipe);

   struct lp_fragment_shader *shader = CALLOC_STRUCT(lp_fragment_shader);
   if (!shader)
      return NULL;

   pipe_reference_init(&shader->reference, 1);
   shader->no = fs_no++;
   list_inithead(&shader->variants.list);

   shader->base.type = PIPE_SHADER_IR_NIR;

   if (templ->type == PIPE_SHADER_IR_TGSI) {
      shader->base.ir.nir = tgsi_to_nir(templ->tokens, pipe->screen, false);
   } else {
      shader->base.ir.nir = templ->ir.nir;
   }

   /* lower FRAG_RESULT_COLOR -> DATA[0-7] to correctly handle unused attachments */
   nir_shader *nir = shader->base.ir.nir;
   NIR_PASS_V(nir, nir_lower_fragcolor, nir->info.fs.color_is_dual_source ? 1 : 8);

   nir_shader_gather_info(nir, nir_shader_get_entrypoint(nir));
   nir_tgsi_scan_shader(nir, &shader->info.base, true);
   shader->info.num_texs = shader->info.base.opcode_count[TGSI_OPCODE_TEX];

   llvmpipe_register_shader(pipe, &shader->base);

   shader->draw_data = draw_create_fragment_shader(llvmpipe->draw, templ);
   if (shader->draw_data == NULL) {
      FREE(shader);
      return NULL;
   }

   const int nr_samplers = BITSET_LAST_BIT(nir->info.samplers_used);
   const int nr_sampler_views = BITSET_LAST_BIT(nir->info.textures_used);
   const int nr_images = BITSET_LAST_BIT(nir->info.images_used);

   shader->variant_key_size = lp_fs_variant_key_size(MAX2(nr_samplers,
                                                          nr_sampler_views),
                                                     nr_images);

   nir_foreach_shader_in_variable(var, nir) {
      unsigned idx = var->data.driver_location;
      unsigned slots = nir_variable_count_slots(var, var->type);

      if (var->data.centroid)
         shader->inputs[idx].location = TGSI_INTERPOLATE_LOC_CENTROID;
      if (var->data.sample)
         shader->inputs[idx].location = TGSI_INTERPOLATE_LOC_SAMPLE;

      enum glsl_base_type base_type =
         glsl_get_base_type(glsl_without_array(var->type));
      switch (var->data.interpolation) {
      case INTERP_MODE_NONE:
         if (glsl_base_type_is_integer(base_type) || var->data.per_primitive) {
            shader->inputs[idx].interp = LP_INTERP_CONSTANT;
            break;
         }
         if (var->data.location == VARYING_SLOT_COL0 ||
             var->data.location == VARYING_SLOT_COL1) {
            shader->inputs[idx].interp = LP_INTERP_COLOR;
            break;
         }
         FALLTHROUGH;
      case INTERP_MODE_SMOOTH:
         shader->inputs[idx].interp = LP_INTERP_PERSPECTIVE;
         break;
      case INTERP_MODE_NOPERSPECTIVE:
         shader->inputs[idx].interp = LP_INTERP_LINEAR;
         break;
      case INTERP_MODE_FLAT:
         shader->inputs[idx].interp = LP_INTERP_CONSTANT;
         break;
      }

      /* XXX this is a completely pointless index map... */
      shader->inputs[idx].src_index = idx + 1;
      if (var->data.location == VARYING_SLOT_FACE)
         shader->inputs[idx].interp = LP_INTERP_FACING;
      else if (var->data.location == VARYING_SLOT_POS) {
         shader->inputs[idx].src_index = 0;
         shader->inputs[idx].interp = LP_INTERP_POSITION;
      }

      shader->inputs[idx].usage_mask = shader->info.base.input_usage_mask[idx];
      for (unsigned s = 1; s < slots; s++) {
         shader->inputs[idx + s] = shader->inputs[idx];
         shader->inputs[idx + s].src_index = idx + s + 1;
         shader->inputs[idx + s].usage_mask = shader->info.base.input_usage_mask[idx + s];
      }
   }

   llvmpipe_fs_analyse_nir(shader);

   return shader;
}


static void
llvmpipe_bind_fs_state(struct pipe_context *pipe, void *fs)
{
   struct llvmpipe_context *llvmpipe = llvmpipe_context(pipe);
   struct lp_fragment_shader *lp_fs = (struct lp_fragment_shader *)fs;
   if (llvmpipe->fs == lp_fs)
      return;

   draw_bind_fragment_shader(llvmpipe->draw,
                             (lp_fs ? lp_fs->draw_data : NULL));

   lp_fs_reference(llvmpipe, &llvmpipe->fs, lp_fs);

   /* invalidate the setup link, NEW_FS will make it update */
   lp_setup_set_fs_variant(llvmpipe->setup, NULL);
   llvmpipe->dirty |= LP_NEW_FS;
}


/**
 * Remove shader variant from two lists: the shader's variant list
 * and the context's variant list.
 */
static void
llvmpipe_remove_shader_variant(struct llvmpipe_context *lp,
                               struct lp_fragment_shader_variant *variant)
{
   if ((LP_DEBUG & DEBUG_FS) || (gallivm_debug & GALLIVM_DEBUG_IR)) {
      debug_printf("llvmpipe: del fs #%u var %u v created %u v cached %u "
                   "v total cached %u inst %u total inst %u\n",
                   variant->shader->no, variant->no,
                   variant->shader->variants_created,
                   variant->shader->variants_cached,
                   lp->nr_fs_variants, variant->nr_instrs, lp->nr_fs_instrs);
   }

   /* remove from shader's list */
   list_del(&variant->list_item_local.list);
   variant->shader->variants_cached--;

   /* remove from context's list */
   list_del(&variant->list_item_global.list);
   lp->nr_fs_variants--;
   lp->nr_fs_instrs -= variant->nr_instrs;
}


void
llvmpipe_destroy_shader_variant(struct llvmpipe_context *lp,
                                struct lp_fragment_shader_variant *variant)
{
   gallivm_destroy(variant->gallivm);
   lp_fs_reference(lp, &variant->shader, NULL);
   if (variant->function_name[RAST_EDGE_TEST])
      FREE(variant->function_name[RAST_EDGE_TEST]);
   if (variant->function_name[RAST_WHOLE])
      FREE(variant->function_name[RAST_WHOLE]);
   if (variant->linear_function_name)
      FREE(variant->linear_function_name);
   FREE(variant);
}


void
llvmpipe_destroy_fs(struct llvmpipe_context *llvmpipe,
                    struct lp_fragment_shader *shader)
{
   /* Delete draw module's data */
   draw_delete_fragment_shader(llvmpipe->draw, shader->draw_data);

   ralloc_free(shader->base.ir.nir);
   assert(shader->variants_cached == 0);
   FREE(shader);
}


static void
llvmpipe_delete_fs_state(struct pipe_context *pipe, void *fs)
{
   struct llvmpipe_context *llvmpipe = llvmpipe_context(pipe);
   struct lp_fragment_shader *shader = fs;
   struct lp_fs_variant_list_item *li, *next;

   /* Delete all the variants */
   LIST_FOR_EACH_ENTRY_SAFE(li, next, &shader->variants.list, list) {
      struct lp_fragment_shader_variant *variant;
      variant = li->base;
      llvmpipe_remove_shader_variant(llvmpipe, li->base);
      lp_fs_variant_reference(llvmpipe, &variant, NULL);
   }

   lp_fs_reference(llvmpipe, &shader, NULL);
}


static void
llvmpipe_set_constant_buffer(struct pipe_context *pipe,
                             enum pipe_shader_type shader, uint index,
                             bool take_ownership,
                             const struct pipe_constant_buffer *cb)
{
   struct llvmpipe_context *llvmpipe = llvmpipe_context(pipe);
   struct pipe_constant_buffer *constants = &llvmpipe->constants[shader][index];

   assert(shader < PIPE_SHADER_MESH_TYPES);
   assert(index < ARRAY_SIZE(llvmpipe->constants[shader]));

   /* note: reference counting */
   util_copy_constant_buffer(&llvmpipe->constants[shader][index], cb,
                             take_ownership);

   /* user_buffer is only valid until the next set_constant_buffer (at most,
    * possibly until shader deletion), so we need to upload it now to make
    * sure it doesn't get updated/freed out from under us.
    */
   if (constants->user_buffer) {
      u_upload_data(llvmpipe->pipe.const_uploader, 0, constants->buffer_size,
                    16, constants->user_buffer, &constants->buffer_offset,
                    &constants->buffer);
   }
   if (constants->buffer) {
       if (!(constants->buffer->bind & PIPE_BIND_CONSTANT_BUFFER)) {
         debug_printf("Illegal set constant without bind flag\n");
         constants->buffer->bind |= PIPE_BIND_CONSTANT_BUFFER;
      }
      llvmpipe_flush_resource(pipe, constants->buffer, 0, true, true, false, "set_constant_buffer");
   }

   switch (shader) {
   case PIPE_SHADER_VERTEX:
   case PIPE_SHADER_GEOMETRY:
   case PIPE_SHADER_TESS_CTRL:
   case PIPE_SHADER_TESS_EVAL: {
      const unsigned size = cb ? cb->buffer_size : 0;

      const uint8_t *data = NULL;
      if (constants->buffer) {
         data = (uint8_t *) llvmpipe_resource_data(constants->buffer)
            + constants->buffer_offset;
      }

      draw_set_mapped_constant_buffer(llvmpipe->draw, shader,
                                      index, data, size);
      break;
   }
   case PIPE_SHADER_COMPUTE:
      llvmpipe->cs_dirty |= LP_CSNEW_CONSTANTS;
      break;
   case PIPE_SHADER_FRAGMENT:
      llvmpipe->dirty |= LP_NEW_FS_CONSTANTS;
      break;
   case PIPE_SHADER_TASK:
      llvmpipe->dirty |= LP_NEW_TASK_CONSTANTS;
      break;
   case PIPE_SHADER_MESH:
      llvmpipe->dirty |= LP_NEW_MESH_CONSTANTS;
      break;
   default:
      unreachable("Illegal shader type");
      break;
   }
}


static void
llvmpipe_set_shader_buffers(struct pipe_context *pipe,
                            enum pipe_shader_type shader, unsigned start_slot,
                            unsigned count,
                            const struct pipe_shader_buffer *buffers,
                            unsigned writable_bitmask)
{
   struct llvmpipe_context *llvmpipe = llvmpipe_context(pipe);

   unsigned i, idx;
   for (i = start_slot, idx = 0; i < start_slot + count; i++, idx++) {
      const struct pipe_shader_buffer *buffer = buffers ? &buffers[idx] : NULL;

      util_copy_shader_buffer(&llvmpipe->ssbos[shader][i], buffer);

      if (buffer && buffer->buffer) {
         bool read_only = !(writable_bitmask & (1 << idx));
         llvmpipe_flush_resource(pipe, buffer->buffer, 0, read_only, false,
                                 false, "buffer");
      }

      switch (shader) {
      case PIPE_SHADER_VERTEX:
      case PIPE_SHADER_GEOMETRY:
      case PIPE_SHADER_TESS_CTRL:
      case PIPE_SHADER_TESS_EVAL: {
         const unsigned size = buffer ? buffer->buffer_size : 0;
         const uint8_t *data = NULL;
         if (buffer && buffer->buffer)
            data = (uint8_t *) llvmpipe_resource_data(buffer->buffer);
         if (data)
            data += buffer->buffer_offset;
         draw_set_mapped_shader_buffer(llvmpipe->draw, shader,
                                       i, data, size);
         break;
      }
      case PIPE_SHADER_COMPUTE:
         llvmpipe->cs_dirty |= LP_CSNEW_SSBOS;
         break;
      case PIPE_SHADER_TASK:
         llvmpipe->dirty |= LP_NEW_TASK_SSBOS;
         break;
      case PIPE_SHADER_MESH:
         llvmpipe->dirty |= LP_NEW_MESH_SSBOS;
         break;
      case PIPE_SHADER_FRAGMENT:
         llvmpipe->fs_ssbo_write_mask &= ~(((1 << count) - 1) << start_slot);
         llvmpipe->fs_ssbo_write_mask |= writable_bitmask << start_slot;
         llvmpipe->dirty |= LP_NEW_FS_SSBOS;
         break;
      default:
         unreachable("Illegal shader type");
         break;
      }
   }
}


static void
llvmpipe_set_shader_images(struct pipe_context *pipe,
                           enum pipe_shader_type shader, unsigned start_slot,
                           unsigned count, unsigned unbind_num_trailing_slots,
                           const struct pipe_image_view *images)
{
   struct llvmpipe_context *llvmpipe = llvmpipe_context(pipe);
   unsigned i, idx;

   draw_flush(llvmpipe->draw);
   for (i = start_slot, idx = 0; i < start_slot + count; i++, idx++) {
      const struct pipe_image_view *image = images ? &images[idx] : NULL;

      util_copy_image_view(&llvmpipe->images[shader][i], image);

      if (image && image->resource) {
         bool read_only = !(image->access & PIPE_IMAGE_ACCESS_WRITE);
         llvmpipe_flush_resource(pipe, image->resource, 0, read_only, false,
                                 false, "image");
      }
   }

   llvmpipe->num_images[shader] = start_slot + count;
   switch (shader) {
   case PIPE_SHADER_VERTEX:
   case PIPE_SHADER_GEOMETRY:
   case PIPE_SHADER_TESS_CTRL:
   case PIPE_SHADER_TESS_EVAL:
      draw_set_images(llvmpipe->draw, shader, llvmpipe->images[shader],
                      start_slot + count);
      break;
   case PIPE_SHADER_COMPUTE:
      llvmpipe->cs_dirty |= LP_CSNEW_IMAGES;
      break;
   case PIPE_SHADER_FRAGMENT:
      llvmpipe->dirty |= LP_NEW_FS_IMAGES;
      break;
   case PIPE_SHADER_TASK:
      llvmpipe->dirty |= LP_NEW_TASK_IMAGES;
      break;
   case PIPE_SHADER_MESH:
      llvmpipe->dirty |= LP_NEW_MESH_IMAGES;
      break;
   default:
      unreachable("Illegal shader type");
      break;
   }

   if (unbind_num_trailing_slots) {
      llvmpipe_set_shader_images(pipe, shader, start_slot + count,
                                 unbind_num_trailing_slots, 0, NULL);
   }
}


/**
 * Return the blend factor equivalent to a destination alpha of one.
 */
static inline enum pipe_blendfactor
force_dst_alpha_one(enum pipe_blendfactor factor, bool clamped_zero)
{
   switch (factor) {
   case PIPE_BLENDFACTOR_DST_ALPHA:
      return PIPE_BLENDFACTOR_ONE;
   case PIPE_BLENDFACTOR_INV_DST_ALPHA:
      return PIPE_BLENDFACTOR_ZERO;
   case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE:
      if (clamped_zero)
         return PIPE_BLENDFACTOR_ZERO;
      else
         return PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE;
   default:
      return factor;
   }
}


/**
 * We need to generate several variants of the fragment pipeline to match
 * all the combinations of the contributing state atoms.
 *
 * TODO: there is actually no reason to tie this to context state -- the
 * generated code could be cached globally in the screen.
 */
static struct lp_fragment_shader_variant_key *
make_variant_key(struct llvmpipe_context *lp,
                 struct lp_fragment_shader *shader,
                 char *store)
{
   struct lp_fragment_shader_variant_key *key =
      (struct lp_fragment_shader_variant_key *)store;
   struct nir_shader *nir = shader->base.ir.nir;

   memset(key, 0, sizeof(*key));

   if (lp->framebuffer.zsbuf) {
      const enum pipe_format zsbuf_format = lp->framebuffer.zsbuf->format;
      const struct util_format_description *zsbuf_desc =
         util_format_description(zsbuf_format);

      if (lp->depth_stencil->depth_enabled &&
          util_format_has_depth(zsbuf_desc)) {
         key->zsbuf_format = zsbuf_format;
         key->depth.enabled = lp->depth_stencil->depth_enabled;
         key->depth.writemask = lp->depth_stencil->depth_writemask;
         key->depth.func = lp->depth_stencil->depth_func;
      }
      if (lp->depth_stencil->stencil[0].enabled &&
          util_format_has_stencil(zsbuf_desc)) {
         key->zsbuf_format = zsbuf_format;
         memcpy(&key->stencil, &lp->depth_stencil->stencil,
                sizeof key->stencil);
      }
      if (llvmpipe_resource_is_1d(lp->framebuffer.zsbuf->texture)) {
         key->resource_1d = true;
      }
      key->zsbuf_nr_samples =
         util_res_sample_count(lp->framebuffer.zsbuf->texture);

      /*
       * Restrict depth values if the API is clamped (GL, VK with ext)
       * for non float Z buffer
       */
      key->restrict_depth_values =
         !(lp->rasterizer->unclamped_fragment_depth_values &&
           util_format_get_depth_only(zsbuf_format) == PIPE_FORMAT_Z32_FLOAT);
   }

   /*
    * Propagate the depth clamp setting from the rasterizer state.
    */
   key->depth_clamp = lp->rasterizer->depth_clamp;

   /* alpha test only applies if render buffer 0 is non-integer
    * (or does not exist)
    */
   if (!lp->framebuffer.nr_cbufs ||
       !lp->framebuffer.cbufs[0] ||
       !util_format_is_pure_integer(lp->framebuffer.cbufs[0]->format)) {
      key->alpha.enabled = lp->depth_stencil->alpha_enabled;
   }
   if (key->alpha.enabled) {
      key->alpha.func = lp->depth_stencil->alpha_func;
      /* alpha.ref_value is passed in jit_context */
   }

   key->flatshade = lp->rasterizer->flatshade;
   key->multisample = lp->rasterizer->multisample;
   key->no_ms_sample_mask_out = lp->rasterizer->no_ms_sample_mask_out;
   if (lp->active_occlusion_queries && !lp->queries_disabled) {
      key->occlusion_count = true;
   }

   memcpy(&key->blend, lp->blend, sizeof key->blend);

   key->coverage_samples = 1;
   key->min_samples = 1;
   if (key->multisample) {
      key->coverage_samples =
         util_framebuffer_get_num_samples(&lp->framebuffer);
      /* Per EXT_shader_framebuffer_fetch spec:
       *
       *   "1. How is framebuffer data treated during multisample rendering?
       *
       *    RESOLVED: Reading the value of gl_LastFragData produces a
       *    different result for each sample. This implies that all or part
       *    of the shader be run once for each sample, but has no additional
       *    implications on fragment shader input variables which may still
       *    be interpolated per pixel by the implementation."
       *
       * ARM_shader_framebuffer_fetch_depth_stencil spec further says:
       *
       *   "(1) When multisampling is enabled, does the shader run per sample?
       *
       *    RESOLVED.
       *
       *    This behavior is inherited from either
       *    EXT_shader_framebuffer_fetch or ARM_shader_framebuffer_fetch as
       *    described in the interactions section.  If neither extension is
       *    supported, the shader runs once per fragment."
       *
       * Therefore we should always enable per-sample shading when FB fetch is
       * used.
       */
      if (lp->min_samples > 1 || nir->info.fs.uses_fbfetch_output)
         key->min_samples = key->coverage_samples;
   }
   key->nr_cbufs = lp->framebuffer.nr_cbufs;

   if (!key->blend.independent_blend_enable) {
      // we always need independent blend otherwise the fixups below won't work
      for (unsigned i = 1; i < key->nr_cbufs; i++) {
         memcpy(&key->blend.rt[i], &key->blend.rt[0],
                sizeof(key->blend.rt[0]));
      }
      key->blend.independent_blend_enable = 1;
   }

   for (unsigned i = 0; i < lp->framebuffer.nr_cbufs; i++) {
      struct pipe_rt_blend_state *blend_rt = &key->blend.rt[i];

      if (lp->framebuffer.cbufs[i]) {
         const enum pipe_format format = lp->framebuffer.cbufs[i]->format;

         key->cbuf_format[i] = format;
         key->cbuf_nr_samples[i] =
            util_res_sample_count(lp->framebuffer.cbufs[i]->texture);

         /*
          * Figure out if this is a 1d resource. Note that OpenGL allows crazy
          * mixing of 2d textures with height 1 and 1d textures, so make sure
          * we pick 1d if any cbuf or zsbuf is 1d.
          */
         if (llvmpipe_resource_is_1d(lp->framebuffer.cbufs[i]->texture)) {
            key->resource_1d = true;
         }

         const struct util_format_description *format_desc =
            util_format_description(format);
         assert(format_desc->colorspace == UTIL_FORMAT_COLORSPACE_RGB ||
                format_desc->colorspace == UTIL_FORMAT_COLORSPACE_SRGB);

         /*
          * Mask out color channels not present in the color buffer.
          */
         blend_rt->colormask &= util_format_colormask(format_desc);

         /*
          * Disable blend for integer formats.
          */
         if (util_format_is_pure_integer(format)) {
            blend_rt->blend_enable = 0;
         }

         /*
          * Our swizzled render tiles always have an alpha channel, but the
          * linear render target format often does not, so force here the dst
          * alpha to be one.
          *
          * This is not a mere optimization. Wrong results will be produced if
          * the dst alpha is used, the dst format does not have alpha, and the
          * previous rendering was not flushed from the swizzled to linear
          * buffer. For example, NonPowTwo DCT.
          *
          * TODO: This should be generalized to all channels for better
          * performance, but only alpha causes correctness issues.
          *
          * Also, force rgb/alpha func/factors match, to make AoS blending
          * easier.
          */
         if (format_desc->swizzle[3] > PIPE_SWIZZLE_W ||
             format_desc->swizzle[3] == format_desc->swizzle[0]) {
            // Doesn't cover mixed snorm/unorm but can't render to them anyway
            bool clamped_zero = !util_format_is_float(format) &&
                                !util_format_is_snorm(format);
            blend_rt->rgb_src_factor =
               force_dst_alpha_one(blend_rt->rgb_src_factor, clamped_zero);
            blend_rt->rgb_dst_factor =
               force_dst_alpha_one(blend_rt->rgb_dst_factor, clamped_zero);
            blend_rt->alpha_func       = blend_rt->rgb_func;
            blend_rt->alpha_src_factor = blend_rt->rgb_src_factor;
            blend_rt->alpha_dst_factor = blend_rt->rgb_dst_factor;
         }
      } else {
         /* no color buffer for this fragment output */
         key->cbuf_format[i] = PIPE_FORMAT_NONE;
         key->cbuf_nr_samples[i] = 0;
         blend_rt->colormask = 0x0;
         blend_rt->blend_enable = 0;
      }
   }

   /* This value will be the same for all the variants of a given shader:
    */
   key->nr_samplers = BITSET_LAST_BIT(nir->info.samplers_used);
   key->nr_sampler_views = BITSET_LAST_BIT(nir->info.textures_used);

   struct lp_sampler_static_state *fs_sampler =
      lp_fs_variant_key_samplers(key);

   memset(fs_sampler, 0,
          MAX2(key->nr_samplers, key->nr_sampler_views) * sizeof *fs_sampler);

   for (unsigned i = 0; i < key->nr_samplers; ++i) {
      if (BITSET_TEST(nir->info.samplers_used, i)) {
         lp_sampler_static_sampler_state(&fs_sampler[i].sampler_state,
                                         lp->samplers[PIPE_SHADER_FRAGMENT][i]);
      }
   }

   /*
    * XXX If TGSI_FILE_SAMPLER_VIEW exists assume all texture opcodes
    * are dx10-style? Can't really have mixed opcodes, at least not
    * if we want to skip the holes here (without rescanning tgsi).
    */
   if (key->nr_sampler_views) {
      for (unsigned i = 0; i < key->nr_sampler_views; ++i) {
         /*
          * Note sview may exceed what's representable by file_mask.
          * This will still work, the only downside is that not actually
          * used views may be included in the shader key.
          */
         if (BITSET_TEST(nir->info.textures_used, i)) {
            lp_sampler_static_texture_state(&fs_sampler[i].texture_state,
                                  lp->sampler_views[PIPE_SHADER_FRAGMENT][i]);
         }
      }
   } else {
      key->nr_sampler_views = key->nr_samplers;
      for (unsigned i = 0; i < key->nr_sampler_views; ++i) {
         if (BITSET_TEST(nir->info.samplers_used, i)) {
            lp_sampler_static_texture_state(&fs_sampler[i].texture_state,
                                 lp->sampler_views[PIPE_SHADER_FRAGMENT][i]);
         }
      }
   }

   struct lp_image_static_state *lp_image = lp_fs_variant_key_images(key);
   key->nr_images = BITSET_LAST_BIT(nir->info.images_used);
   if (key->nr_images)
      memset(lp_image, 0,
             key->nr_images * sizeof *lp_image);
   for (unsigned i = 0; i < key->nr_images; ++i) {
      if (BITSET_TEST(nir->info.images_used, i)) {
         lp_sampler_static_texture_state_image(&lp_image[i].image_state,
                                      &lp->images[PIPE_SHADER_FRAGMENT][i]);
      }
   }

   if (shader->kind == LP_FS_KIND_AERO_MINIFICATION) {
      struct lp_sampler_static_state *samp0 =
         lp_fs_variant_key_sampler_idx(key, 0);
      assert(samp0);
      samp0->sampler_state.min_img_filter = PIPE_TEX_FILTER_NEAREST;
      samp0->sampler_state.mag_img_filter = PIPE_TEX_FILTER_NEAREST;
   }

   return key;
}


/**
 * Update fragment shader state.  This is called just prior to drawing
 * something when some fragment-related state has changed.
 */
void
llvmpipe_update_fs(struct llvmpipe_context *lp)
{
   struct lp_fragment_shader *shader = lp->fs;

   char store[LP_FS_MAX_VARIANT_KEY_SIZE];
   const struct lp_fragment_shader_variant_key *key =
      make_variant_key(lp, shader, store);

   struct lp_fragment_shader_variant *variant = NULL;
   struct lp_fs_variant_list_item *li;
   /* Search the variants for one which matches the key */
   LIST_FOR_EACH_ENTRY(li, &shader->variants.list, list) {
      if (memcmp(&li->base->key, key, shader->variant_key_size) == 0) {
         variant = li->base;
         break;
      }
   }

   if (variant) {
      /* Move this variant to the head of the list to implement LRU
       * deletion of shader's when we have too many.
       */
      list_move_to(&variant->list_item_global.list, &lp->fs_variants_list.list);
   } else {
      /* variant not found, create it now */

      if (LP_DEBUG & DEBUG_FS) {
         debug_printf("%u variants,\t%u instrs,\t%u instrs/variant\n",
                      lp->nr_fs_variants,
                      lp->nr_fs_instrs,
                      lp->nr_fs_variants ? lp->nr_fs_instrs / lp->nr_fs_variants : 0);
      }

      /* First, check if we've exceeded the max number of shader variants.
       * If so, free 6.25% of them (the least recently used ones).
       */
      const unsigned variants_to_cull =
         lp->nr_fs_variants >= LP_MAX_SHADER_VARIANTS
         ? LP_MAX_SHADER_VARIANTS / 16 : 0;

      if (variants_to_cull ||
          lp->nr_fs_instrs >= LP_MAX_SHADER_INSTRUCTIONS) {
         if (gallivm_debug & GALLIVM_DEBUG_PERF) {
            debug_printf("Evicting FS: %u fs variants,\t%u total variants,"
                         "\t%u instrs,\t%u instrs/variant\n",
                         shader->variants_cached,
                         lp->nr_fs_variants, lp->nr_fs_instrs,
                         lp->nr_fs_instrs / lp->nr_fs_variants);
         }

         /*
          * We need to re-check lp->nr_fs_variants because an arbitrarily
          * large number of shader variants (potentially all of them) could
          * be pending for destruction on flush.
          */

         for (unsigned i = 0;
              i < variants_to_cull ||
                 lp->nr_fs_instrs >= LP_MAX_SHADER_INSTRUCTIONS;
              i++) {
            struct lp_fs_variant_list_item *item;
            if (list_is_empty(&lp->fs_variants_list.list)) {
               break;
            }
            item = list_last_entry(&lp->fs_variants_list.list,
                                   struct lp_fs_variant_list_item, list);
            assert(item);
            assert(item->base);
            llvmpipe_remove_shader_variant(lp, item->base);
            struct lp_fragment_shader_variant *variant = item->base;
            lp_fs_variant_reference(lp, &variant, NULL);
         }
      }

      /*
       * Generate the new variant.
       */
      int64_t t0 = os_time_get();
      variant = generate_variant(lp, shader, key);
      int64_t t1 = os_time_get();
      int64_t dt = t1 - t0;
      LP_COUNT_ADD(llvm_compile_time, dt);
      LP_COUNT_ADD(nr_llvm_compiles, 2);  /* emit vs. omit in/out test */

      /* Put the new variant into the list */
      if (variant) {
         list_add(&variant->list_item_local.list, &shader->variants.list);
         list_add(&variant->list_item_global.list, &lp->fs_variants_list.list);
         lp->nr_fs_variants++;
         lp->nr_fs_instrs += variant->nr_instrs;
         shader->variants_cached++;
      }
   }

   /* Bind this variant */
   lp_setup_set_fs_variant(lp->setup, variant);
}


void
llvmpipe_init_fs_funcs(struct llvmpipe_context *llvmpipe)
{
   llvmpipe->pipe.create_fs_state = llvmpipe_create_fs_state;
   llvmpipe->pipe.bind_fs_state   = llvmpipe_bind_fs_state;
   llvmpipe->pipe.delete_fs_state = llvmpipe_delete_fs_state;
   llvmpipe->pipe.set_constant_buffer = llvmpipe_set_constant_buffer;
   llvmpipe->pipe.set_shader_buffers = llvmpipe_set_shader_buffers;
   llvmpipe->pipe.set_shader_images = llvmpipe_set_shader_images;
}