diff options
| author | Chandler Carruth <chandlerc@gmail.com> | 2014-07-26 01:52:13 +0000 |
|---|---|---|
| committer | Chandler Carruth <chandlerc@gmail.com> | 2014-07-26 01:52:13 +0000 |
| commit | 1bf4d19172879e42eeb595e6337e8b358f6469c9 (patch) | |
| tree | 1a7819a3c23231c930dc8ab8e8ef79b0a24bce2e /llvm/test/CodeGen/X86/vector-idiv.ll | |
| parent | e87e5ce5d687a7f1223faa652add3b3b937ae30f (diff) | |
| download | bcm5719-llvm-1bf4d19172879e42eeb595e6337e8b358f6469c9.tar.gz bcm5719-llvm-1bf4d19172879e42eeb595e6337e8b358f6469c9.zip | |
[x86] Fix PR20355 (and dups) by not using unsigned multiplication when
signed multiplication is requested. While there is not a difference in
the *low* half of the result, the *high* half (used specifically to
implement the signed division by these constants) certainly is used. The
test case I've nuked was actively asserting wrong code.
There is a delightful solution to doing signed multiplication even when
we don't have it that Richard Smith has crafted, but I'll add the
machinery back and implement that in a follow-up patch. This at least
restores correctness.
llvm-svn: 214007
Diffstat (limited to 'llvm/test/CodeGen/X86/vector-idiv.ll')
| -rw-r--r-- | llvm/test/CodeGen/X86/vector-idiv.ll | 16 |
1 files changed, 5 insertions, 11 deletions
diff --git a/llvm/test/CodeGen/X86/vector-idiv.ll b/llvm/test/CodeGen/X86/vector-idiv.ll index ec1ce3da5e1..0214a47f44c 100644 --- a/llvm/test/CodeGen/X86/vector-idiv.ll +++ b/llvm/test/CodeGen/X86/vector-idiv.ll @@ -131,18 +131,12 @@ define <4 x i32> @test8(<4 x i32> %a) { ; SSE41: psrad $2 ; SSE41: padd +; FIXME: scalarized -- there is no signed multiply in SSE. ; SSE-LABEL: test8: -; SSE: pmuludq -; SSE: pshufd $49 -; SSE: pshufd $49 -; SSE: pmuludq -; SSE: shufps $-35 -; SSE: pshufd $-40 -; SSE: psubd -; SSE: padd -; SSE: psrld $31 -; SSE: psrad $2 -; SSE: padd +; SSE: imulq +; SSE: imulq +; SSE: imulq +; SSE: imulq ; AVX-LABEL: test8: ; AVX: vpmuldq |
