| commit | 4acf80e7043469f9daca1dd43dc65188a2a8da3f | [log] [tgz] |
|---|---|---|
| author | Simon Pilgrim <llvm-dev@redking.me.uk> | Wed Feb 21 23:07:30 2018 +0000 |
| committer | Simon Pilgrim <llvm-dev@redking.me.uk> | Wed Feb 21 23:07:30 2018 +0000 |
| tree | 8e98620d0c55be0f1db2f89a30eb73ee7976897f | |
| parent | e4507fb8c94b605c8f89bb204ecdd7b51cc78758 [diff] |
[X86][MMX] Generlize MMX_MOVD64rr combines to accept v4i16/v8i8 build vectors as well as v2i32 Also handle both cases where the lower 32-bits of the MMX is undef or zero extended. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@325736 91177308-0d34-0410-b5e6-96231b3b80d8