[PATCH] D25088: [x86] Fixing PR28755 by precomputing the address used in CMPXCHG8B

[Alexander Ivchenko via llvm-commits]

aivchenk marked 3 inline comments as done.
aivchenk added inline comments.


> qcolombet wrote in X86ISelLowering.cpp:24615
> Split the NFC changes in a different patch.

Uploaded as: https://reviews.llvm.org/D25192

> qcolombet wrote in X86ISelLowering.cpp:24754
> Shouldn't we check for is32Bit or something?
> Indeed, LEA32r would not be valid for 64-bit arch and the helping regalloc shouldn't be needed for 64-bit arch, right?

Right. I checked for "TRI->getBaseRegister() == X86::ESI", which is only true for i686, but I agree that it would be better to check directly for 32 bits

https://reviews.llvm.org/D25088