StephenTozer added inline comments.
================
Comment at: clang/lib/CodeGen/CGCall.cpp:3492-3499
+ // Ignore fake uses and the instructions that load their
+ // operands.
+ if (Intrinsic->getIntrinsicID() == llvm::Intrinsic::fake_use) {
+ const llvm::Value *FakeUseArg = Intrinsic->getArgOperand(0);
+ if (++II == IE || &*II != FakeUseArg)
+ break;
+ continue;
----------------
jmorse wrote:
> It's a bit ugly to switch away from having a range-based loop, am I right in
> thinking this is because you need to skip the load-to-fake-use? Annoying but
> necessary I guess. Could I request an explicit "II = std::next(II);" though
> to really ram home to the reader that we're skipping an extra instruction,
> using the pre-increment is liable to be missed, and we're not trying to be
> super-concise here IMO.
>
> The addition of the "I" variable looks like it might be un-necessary, that's
> only _used_ by GetStoreIfValid which has a domination assignment above it no?
> Annoying but necessary I guess.
Probably not necessary actually - the addition of an extra variable could cover
this case without removing the for-range.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D157613/new/
https://reviews.llvm.org/D157613
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
