OpenHarmony/third_party_mesa3d
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

intel/fs: fix size_read() for LOAD_PAYLOAD

Browse Source 
With Anv/Zink, the piglit test :

  arb_shader_storage_buffer_object-max-ssbo-size -auto -fbo fsexceed

is failing validation after copy propagation :

load_payload(8) vgrf15:F, vgrf1+0.12<0>:F, vgrf1+0.0<0>:F, vgrf1+0.4<0>:F, vgrf1+0.8<0>:F, vgrf1+0.12<0>:F
../src/intel/compiler/brw_fs_validate.cpp:191: A <= B failed
  A = inst->src[i].offset / REG_SIZE + regs_read(inst, i) = 2
  B = alloc.sizes[inst->src[i].nr] = 1

In most cases it works because src[0] would be at offset 0 and so
reading a full reg passes validation, but Anv/Zink started emitting
slightly different code adding an offset maybe the size read 2 GRFs.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/23126>
This commit is contained in:
Lionel Landwerlin
2023-05-19 11:10:01 +03:00
committed by Marge Bot
parent 1546a9de99
commit 21c7b55f6f
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
src/intel/compiler/brw_fs.cpp
View File

@@ -877,7 +877,7 @@ fs_inst::size_read(int arg) const
case SHADER_OPCODE_LOAD_PAYLOAD:
if (arg < this->header_size)
return REG_SIZE;
return retype(src[arg], BRW_REGISTER_TYPE_UD).component_size(8);
break;
case CS_OPCODE_CS_TERMINATE: