[PATCH] [x86] Add a feature flag for slow 32-byte unaligned memory accesses.

[Andrea Di Biagio]

Hi Sanjay,

The patch looks good to me (I left a minor comment on the test).

Thanks!
Andrea

================
Comment at: test/CodeGen/X86/unaligned-32-byte-memops.ll:1-3
@@ +1,4 @@
+; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=corei7-avx | FileCheck %s --check-prefix=SANDYB
+; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=btver2 | FileCheck %s --check-prefix=BTVER2
+; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=core-avx2 | FileCheck %s --check-prefix=HASWELL
+
----------------
Can you also add a RUN line for testing Ivy Bridge (core-avx-i) ?
Feature flag 'FeatureSlowUAMem32' is also added to Ivy Bridge so, I think you should test it (I guess you could reuse the same SANDYB checks).

http://reviews.llvm.org/D6355