diff options
| author | Dan Gohman <gohman@apple.com> | 2011-11-08 21:29:06 +0000 |
|---|---|---|
| committer | Dan Gohman <gohman@apple.com> | 2011-11-08 21:29:06 +0000 |
| commit | 9cae2d2225ba58a70ef8ff057feab6873f4af520 (patch) | |
| tree | a0c948c7431aeba1e389c126e0e2ce8f4b304b50 /test/CodeGen/X86/fold-pcmpeqd-0.ll | |
| parent | 3568a1051efb9a9edbd4914b04b44e9d7bc1b004 (diff) | |
| download | external_llvm-9cae2d2225ba58a70ef8ff057feab6873f4af520.zip external_llvm-9cae2d2225ba58a70ef8ff057feab6873f4af520.tar.gz external_llvm-9cae2d2225ba58a70ef8ff057feab6873f4af520.tar.bz2 | |
Add a hack to the scheduler to disable pseudo-two-address dependencies in
basic blocks containing calls. This works around a problem in which
these artificial dependencies can get tied up in calling seqeunce
scheduling in a way that makes the graph unschedulable with the current
approach of using artificial physical register dependencies for calling
sequences. This fixes PR11314.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@144124 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/CodeGen/X86/fold-pcmpeqd-0.ll')
| -rw-r--r-- | test/CodeGen/X86/fold-pcmpeqd-0.ll | 8 |
1 files changed, 1 insertions, 7 deletions
diff --git a/test/CodeGen/X86/fold-pcmpeqd-0.ll b/test/CodeGen/X86/fold-pcmpeqd-0.ll index 647bbdb..6095a9c 100644 --- a/test/CodeGen/X86/fold-pcmpeqd-0.ll +++ b/test/CodeGen/X86/fold-pcmpeqd-0.ll @@ -3,15 +3,9 @@ ; This testcase shouldn't need to spill the -1 value, ; so it should just use pcmpeqd to materialize an all-ones vector. -; For i386, cp load of -1 are folded. -; With -regalloc=greedy, the live range is split before spilling, so the first -; pcmpeq doesn't get folded as a constant pool load. - -; I386-NOT: pcmpeqd -; I386: orps LCPI0_2, %xmm +; I386: pcmpeqd ; I386-NOT: pcmpeqd -; I386: orps LCPI0_2, %xmm ; X86-64: pcmpeqd ; X86-64-NOT: pcmpeqd |