[PATCH] D38466: [ TargetLowering, AMDGPU] Use the return value of UpdateNodeOperands();
Simon Pilgrim via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Mon Oct 2 10:45:04 PDT 2017
RKSimon added inline comments.
================
Comment at: lib/Target/AMDGPU/SIISelLowering.cpp:6489
+ Node = DAG.UpdateNodeOperands(Node, Ops);
return Node;
}
----------------
```
return DAG.UpdateNodeOperands(Node, Ops);
```
================
Comment at: test/CodeGen/AMDGPU/simplifydemandedbits-recursion.ll:1
+; RUN: (ulimit -t 5; llc -march=amdgcn < %s | FileCheck -check-prefix=GCN %s)
+
----------------
arsenm wrote:
> Does this actually work? I've never seen another test use ulimit or a sub shell. 5 seconds seems high also.
Does this work on windows builds?
================
Comment at: test/CodeGen/AMDGPU/simplifydemandedbits-recursion.ll:16
+
+; GCN: s_endpgm
+define amdgpu_kernel void @foo(float addrspace(1)* noalias nocapture readonly, float addrspace(1)* noalias nocapture readonly, float addrspace(1)* noalias nocapture, float) local_unnamed_addr !reqd_work_group_size !{i32 8, i32 16, i32 1} {
----------------
You're not testing that foo() is empty - just that it includes s_endpgm
https://reviews.llvm.org/D38466
More information about the llvm-commits
mailing list