[libcxx-commits] [PATCH] D96842: [dfsan] Do not specialize vector<bool> for DFSan
stephan.yichao.zhao via Phabricator via libcxx-commits
libcxx-commits at lists.llvm.org
Wed Feb 24 09:59:13 PST 2021
stephan.yichao.zhao updated this revision to Diff 326131.
stephan.yichao.zhao added a comment.
Before doing a complete test, this shows how one potential change would look like.
Please have a look and see if this is on the right way.
What the change does are
1. extended all operations of __bit_reference/_bit_iterator with dfsan label propagation The two classes take an additional iterator of dfsan label storage
2. vector<bool> uses <__bit_reference>, and allocates dfsan label storage We use vector<int> as the storage, so we can get all operations like copy/fill for free.
3. bitset uses <__bit_reference>. So it got changed, and this increases dfsan accuracy too. Its change is more than vector<bool> because its logic is not only in __bit_reference.
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 39208 bytes
Desc: not available
More information about the libcxx-commits