• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*
2  * Copyright (C) 2020 Icecream95 <ixn@disroot.org>
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 FROM,
20  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21  * SOFTWARE.
22  */
23 
24 #include "pan_ir.h"
25 #include "compiler/nir/nir_builder.h"
26 
27 /* OpenCL uses 64-bit types for some intrinsic functions, including
28  * global_invocation_id(). This could be worked around during conversion to
29  * MIR, except that global_invocation_id is a vec3, and the 128-bit registers
30  * on Midgard can only hold a 64-bit vec2.
31  * Rather than attempting to add hacky 64-bit vec3 support, convert these
32  * intrinsics to 32-bit and add a cast back to 64-bit, and rely on NIR not
33  * vectorizing back to vec3.
34  */
35 
36 static bool
nir_lower_64bit_intrin_instr(nir_builder * b,nir_instr * instr,void * data)37 nir_lower_64bit_intrin_instr(nir_builder *b, nir_instr *instr, void *data)
38 {
39         if (instr->type != nir_instr_type_intrinsic)
40                 return false;
41 
42         nir_intrinsic_instr *intr = nir_instr_as_intrinsic(instr);
43 
44         switch (intr->intrinsic) {
45         case nir_intrinsic_load_global_invocation_id:
46         case nir_intrinsic_load_global_invocation_id_zero_base:
47         case nir_intrinsic_load_workgroup_id:
48         case nir_intrinsic_load_num_workgroups:
49                 break;
50 
51         default:
52                 return false;
53         }
54 
55         if (nir_dest_bit_size(intr->dest) != 64)
56                 return false;
57 
58         b->cursor = nir_after_instr(instr);
59 
60         assert(intr->dest.is_ssa);
61         intr->dest.ssa.bit_size = 32;
62 
63         nir_ssa_def *conv = nir_u2u64(b, &intr->dest.ssa);
64 
65         nir_ssa_def_rewrite_uses_after(&intr->dest.ssa, conv,
66                                        conv->parent_instr);
67 
68         return true;
69 }
70 
71 bool
pan_nir_lower_64bit_intrin(nir_shader * shader)72 pan_nir_lower_64bit_intrin(nir_shader *shader)
73 {
74         return nir_shader_instructions_pass(shader,
75                                             nir_lower_64bit_intrin_instr,
76                                             nir_metadata_block_index | nir_metadata_dominance,
77                                             NULL);
78 }
79