calixte marked an inline comment as done.
calixte added inline comments.
================
Comment at: llvm/lib/Transforms/Instrumentation/GCOVProfiling.cpp:635
void GCOVProfiler::AddFlushBeforeForkAndExec() {
- SmallVector<Instruction *, 2> ForkAndExecs;
+ SmallVector<std::pair<bool, CallInst *>, 2> ForkAndExecs;
for (auto &F : M->functions()) {
----------------
marco-c wrote:
> calixte wrote:
> > marco-c wrote:
> > > Since we are now mostly doing different things on forks and execs, we
> > > could remove this vector and just do the operations directly instead of
> > > adding to the vec.
> > If we make the insertions of new code in the second for loop, we'll
> > invalidate the iterator used in this loop.
> M->getOrInsertFunction is what would invalidate the iterator?
>
> You could clean things up a bit by having two vectors then, one for forks and
> one for execs.
Builder.Create** will invalidate the iterator
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D74953/new/
https://reviews.llvm.org/D74953
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
