projects / platform / upstream / llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: b71203a) | patch
AMDGPU/GlobalISel: Switch target in argument test
authorMatt Arsenault <Matthew.Arsenault@amd.com>
Tue, 25 Feb 2020 20:00:37 +0000 (15:00 -0500)
committerMatt Arsenault <arsenm2@gmail.com>
Wed, 4 Mar 2020 21:40:06 +0000 (16:40 -0500)
commitf70e7dc17d78a83265a89158ee549f1147d7d500
tree15546e698976355616dbddf4d1464297864b40b2tree | snapshot
parentb71203a7512a7a7a85fe943d5a9de68948e7d3d0commit | diff
AMDGPU/GlobalISel: Switch target in argument test

Since this is still largely relying on the DAG argument type lowering
code, this has inherited the problem where i16 vectors have a
different ABI on targets with and without legal i16. Switch to using a
target with legal i16, so the i16 vector argument tests are more
useful.
llvm/test/CodeGen/AMDGPU/GlobalISel/irtranslator-function-args.ll diff | blob | history
Domain: System / Toolchain;