[PATCH] D14489: [AArch64] Applying load pair optimization for volatile load/store

[Chad Rosier via llvm-commits]

mcrosier added a comment.

I'm fine with deferring to your opinion, Tim, and thanks for the clarification.  If you're comfortable with the transformation then I'm on board..

However, if I understand you correctly the below is *not* a safe transformation (assuming one of the loads is volatile):

ldr x8, [x0, #8]
ldr x9, [x0]

into

ldp x9, x8, [x0]

correct?


http://reviews.llvm.org/D14489