This revision was automatically updated to reflect the committed changes.
Closed by commit rL278937: [analyzer] Add LocationContext information to
SymbolMetadata. (authored by dergachev).
Changed prior to commit:
https://reviews.llvm.org/D21978?vs=62696&id=68368#toc
Repository:
rL LLVM
http
a.sidorin added a comment.
In http://reviews.llvm.org/D21978#476200, @NoQ wrote:
> I also never noticed that the block count gets reset on every stack frame,
> that's pretty unobvious.
Yes, that was surprising for me too.
> Uhm, and it's already there in SymbolConjured. Anyway, fixing the blo
NoQ added a comment.
I also never noticed that the block count gets reset on every stack frame,
that's pretty unobvious. Uhm, and it's already there in SymbolConjured. Anyway,
fixing the block count itself seems to be pretty complicated.
Also, maybe once the ScopeContext thing is ready, we'd be
a.sidorin created this revision.
a.sidorin added reviewers: zaks.anna, dcoughlin.
a.sidorin added a subscriber: cfe-commits.
Currently, SymbolMetadata are distinguished by statement and block count.
However, in case of inlining, block counts may be the same every time function
is called because
