itessier added a comment.
Thanks for the detailed info! I will try to get something out when I get some
free cycles.
Repository:
rC Clang
https://reviews.llvm.org/D52440
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.o
rjmccall added a comment.
In https://reviews.llvm.org/D52440#1298816, @itessier wrote:
> > This is problematic because we're not necessarily in a scope that usefully
> > limits the duration of cleanups — we don't push full-expression scopes when
> > emitting an arbitrary statement. Probably we
itessier added a comment.
> This is problematic because we're not necessarily in a scope that usefully
> limits the duration of cleanups — we don't push full-expression scopes when
> emitting an arbitrary statement. Probably we should, but we don't.
> If you'd like to take a look at solving t
rjmccall added inline comments.
Comment at: lib/CodeGen/CGExpr.cpp:176
+ pushFullExprCleanup(NormalEHLifetimeMarker,
+ Slot.getAddress(), Size);
+}
This is problematic because we're not necessarily in a scope tha
itessier created this revision.
itessier added a reviewer: rjmccall.
Herald added a subscriber: cfe-commits.
Clang is not emitting lifetime markers for temporary function parameters, so
more stack space is potentially being allocated than necessary.
A new parameter is added to the CreateAggTemp
