vsk abandoned this revision.
vsk added a comment.
I haven't measured the performance overhead of this patch. The failure case
it's supposed to protect against is exceptionally rare: we haven't seen it
happen again. In light of that, and considering that it clutters up the code, I
think it'd be
rsmith added a comment.
Have you done any performance testing on this code? In modules-heavy builds,
bitcode reading is *very* hot.
http://reviews.llvm.org/D17299
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin
benlangmuir added a comment.
This LGTM, but I would like Richard to take a look too.
http://reviews.llvm.org/D17299
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
vsk added a comment.
Gentle ping.
For context, we've had a few situations where corrupt pch files trigger
surprising assertion failures in our world builds.
http://reviews.llvm.org/D17299
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
htt
vsk created this revision.
vsk added reviewers: rsmith, benlangmuir.
vsk added a subscriber: cfe-commits.
If we try to read a corrupt pch, we can easily assert-fail or trigger invalid
memory accesses when manipulating ASTReader::RecordData objects. This problem
is easy enough to diagnose when asse
