diff options
| author | Bill Schmidt <wschmidt@linux.vnet.ibm.com> | 2014-10-19 21:29:21 +0000 |
|---|---|---|
| committer | Bill Schmidt <wschmidt@linux.vnet.ibm.com> | 2014-10-19 21:29:21 +0000 |
| commit | 941a3244ffe0fbed51f977ace7358fbf5c924479 (patch) | |
| tree | 8d89b02538508a320e03e9c4559145560fad74c2 /llvm/test/CodeGen/PowerPC/unsafe-math.ll | |
| parent | 87982a1e9b42b18e803a2dda3a79ff665b7b2243 (diff) | |
| download | bcm5719-llvm-941a3244ffe0fbed51f977ace7358fbf5c924479.tar.gz bcm5719-llvm-941a3244ffe0fbed51f977ace7358fbf5c924479.zip | |
[PowerPC] Clean up -mattr=+vsx tests to always specify -mcpu
We recently discovered an issue that reinforces what a good idea it is
to always specify -mcpu in our code generation tests, particularly for
-mattr=+vsx. This patch ensures that all tests that specify
-mattr=+vsx also specify -mcpu=pwr7 or -mcpu=pwr8, as appropriate.
Some of the uses of -mattr=+vsx added recently don't make much sense
(when specified for -mtriple=powerpc-apple-darwin8 or -march=ppc32,
for example). For cases like this I've just removed the extra VSX
test commands; there's enough coverage without them.
llvm-svn: 220173
Diffstat (limited to 'llvm/test/CodeGen/PowerPC/unsafe-math.ll')
| -rw-r--r-- | llvm/test/CodeGen/PowerPC/unsafe-math.ll | 3 |
1 files changed, 0 insertions, 3 deletions
diff --git a/llvm/test/CodeGen/PowerPC/unsafe-math.ll b/llvm/test/CodeGen/PowerPC/unsafe-math.ll index 1a269da1ac3..f6430270eac 100644 --- a/llvm/test/CodeGen/PowerPC/unsafe-math.ll +++ b/llvm/test/CodeGen/PowerPC/unsafe-math.ll @@ -1,9 +1,6 @@ ; RUN: llc < %s -mattr=-vsx -march=ppc32 | grep fmul | count 2 ; RUN: llc < %s -mattr=-vsx -march=ppc32 -enable-unsafe-fp-math | \ ; RUN: grep fmul | count 1 -; RUN: llc < %s -mattr=+vsx -march=ppc32 | grep xsmuldp | count 2 -; RUN: llc < %s -mattr=+vsx -march=ppc32 -enable-unsafe-fp-math | \ -; RUN: grep xsmuldp | count 1 define double @foo(double %X) nounwind { %tmp1 = fmul double %X, 1.23 |
