src/glsl/nir/nir_lower_vec_to_movs.c

   1 /*
   2  * Copyright © 2014 Intel Corporation
   3  *
   4  * Permission is hereby granted, free of charge, to any person obtaining a
   5  * copy of this software and associated documentation files (the "Software"),
   6  * to deal in the Software without restriction, including without limitation
   7  * the rights to use, copy, modify, merge, publish, distribute, sublicense,
   8  * and/or sell copies of the Software, and to permit persons to whom the
   9  * Software is furnished to do so, subject to the following conditions:
  10  *
  11  * The above copyright notice and this permission notice (including the next
  12  * paragraph) shall be included in all copies or substantial portions of the
  13  * Software.
  14  *
  15  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  16  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  17  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
  18  * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  19  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
  20  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
  21  * IN THE SOFTWARE.
  22  *
  23  * Authors:
  24  *    Jason Ekstrand (jason@jlekstrand.net)
  25  *
  26  */
  27
  28 #include "nir.h"
  29
  30 /*
  31  * Implements a simple pass that lowers vecN instructions to a series of
  32  * moves with partial writes.
  33  */
  34
  35 static bool
  36 lower_vec_to_movs_block(nir_block *block, void *mem_ctx)
  37 {
  38    nir_foreach_instr_safe(block, instr) {
  39       if (instr->type != nir_instr_type_alu)
  40          continue;
  41
  42       nir_alu_instr *vec = (nir_alu_instr *)instr;
  43
  44       switch (vec->op) {
  45       case nir_op_vec2:
  46       case nir_op_vec3:
  47       case nir_op_vec4:
  48          break;
  49       default:
  50          continue; /* The loop */
  51       }
  52
  53       for (unsigned i = 0, src_idx = 0; i < 4; i++) {
  54          if (!(vec->dest.write_mask & (1 << i)))
  55             continue;
  56
  57          assert(src_idx < nir_op_infos[vec->op].num_inputs);
  58
  59          nir_alu_instr *mov = nir_alu_instr_create(mem_ctx, nir_op_imov);
  60          mov->src[0].src = nir_src_copy(vec->src[src_idx].src, mem_ctx);
  61          mov->src[0].negate = vec->src[src_idx].negate;
  62          mov->src[0].abs = vec->src[src_idx].abs;
  63
  64          /* We only care about the one swizzle */
  65          mov->src[0].swizzle[i] = vec->src[src_idx].swizzle[0];
  66
  67          mov->dest.dest = nir_dest_copy(vec->dest.dest, mem_ctx);
  68          mov->dest.saturate = vec->dest.saturate;
  69          mov->dest.write_mask = (1u << i);
  70
  71          nir_instr_insert_before(&vec->instr, &mov->instr);
  72
  73          src_idx++;
  74       }
  75
  76       nir_instr_remove(&vec->instr);
  77       ralloc_free(vec);
  78    }
  79
  80    return true;
  81 }
  82
  83 static void
  84 nir_lower_vec_to_movs_impl(nir_function_impl *impl)
  85 {
  86    nir_foreach_block(impl, lower_vec_to_movs_block, ralloc_parent(impl));
  87 }
  88
  89 void
  90 nir_lower_vec_to_movs(nir_shader *shader)
  91 {
  92    nir_foreach_overload(shader, overload) {
  93       if (overload->impl)
  94          nir_lower_vec_to_movs_impl(overload->impl);
  95    }
  96 }