diff options
| author | Johannes Doerfert <doerfert@cs.uni-saarland.de> | 2016-04-09 14:30:11 +0000 |
|---|---|---|
| committer | Johannes Doerfert <doerfert@cs.uni-saarland.de> | 2016-04-09 14:30:11 +0000 |
| commit | b3410db2b7dfd9ccec9d782b74477a1d0a2febad (patch) | |
| tree | 378b99faccb0e38b0f154462b1d3b1896019d283 /polly/test/Isl/CodeGen/OpenMP | |
| parent | cc05ee52424fdab29984da46c026768311509c9d (diff) | |
| download | bcm5719-llvm-b3410db2b7dfd9ccec9d782b74477a1d0a2febad.tar.gz bcm5719-llvm-b3410db2b7dfd9ccec9d782b74477a1d0a2febad.zip | |
[FIX] Do not recompute SCEVs but pass them to subfunctions
This reverts commit 2879c53e80e05497f408f21ce470d122e9f90f94.
Additionally, it adds SDiv and SRem instructions to the set of values
discovered by the findValues function even if we add the operands to
be able to recompute the SCEVs. In subfunctions we do not want to
recompute SDiv and SRem instructions but pass them instead as they
might have been created through the IslExprBuilder and are more
complicated than simple SDiv/SRem instructions in the code.
llvm-svn: 265873
Diffstat (limited to 'polly/test/Isl/CodeGen/OpenMP')
| -rw-r--r-- | polly/test/Isl/CodeGen/OpenMP/recomputed-srem.ll | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/polly/test/Isl/CodeGen/OpenMP/recomputed-srem.ll b/polly/test/Isl/CodeGen/OpenMP/recomputed-srem.ll index bfbfce176f8..c05830c2c23 100644 --- a/polly/test/Isl/CodeGen/OpenMP/recomputed-srem.ll +++ b/polly/test/Isl/CodeGen/OpenMP/recomputed-srem.ll @@ -1,10 +1,10 @@ ; RUN: opt %loadPolly -polly-codegen -polly-parallel \ ; RUN: -polly-parallel-force -S < %s | FileCheck %s ; -; Test to verify that we recompute %rem96 in the parallel subfunction. +; Test to verify that we pass %rem96 to the parallel subfunction. ; -; CHECK: %rem96polly = srem i32 %polly.subfunc.arg.n2, 16 -; CHECK-NEXT: br label %polly.par.checkNext +; CHECK: %[[R:[0-9]*]] = getelementptr inbounds { i32, i32, i64, float*, float*, i32 }, { i32, i32, i64, float*, float*, i32 }* %polly.par.userContext1, i32 0, i32 5 +; CHECK-NEXT: %polly.subfunc.arg.rem96 = load i32, i32* %[[R]] ; target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" |
