]> granicus.if.org Git - llvm/commit
AMDGPU: Disable AMDGPUPromoteAlloca pass for shader calling conventions.
authorNicolai Haehnle <nhaehnle@gmail.com>
Mon, 18 Jul 2016 09:02:47 +0000 (09:02 +0000)
committerNicolai Haehnle <nhaehnle@gmail.com>
Mon, 18 Jul 2016 09:02:47 +0000 (09:02 +0000)
commit0c05ce4746dbc79b7a9efecace559b2366b3f04b
treec09e8eb78f42cfc360e0cb3ecf237a4534cba81c
parent7e13fe031b877b14c315d4f585f7bdc8f2e8395e
AMDGPU: Disable AMDGPUPromoteAlloca pass for shader calling conventions.

Summary:
The work item intrinsics are not available for the shader
calling conventions. And even if we did hook them up most
shader stages haves some extra restrictions on the amount
of available LDS.

Reviewers: tstellarAMD, arsenm

Subscribers: nhaehnle, arsenm, llvm-commits, kzhuravl

Differential Revision: https://reviews.llvm.org/D20728

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@275779 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/AMDGPU/AMDGPUPromoteAlloca.cpp
test/CodeGen/AMDGPU/promote-alloca-shaders.ll [new file with mode: 0644]