[PATCH] D29994: Use pthreads for thread-local lsan allocator cache on darwin

[Kuba (Brecka) Mracek via Phabricator via llvm-commits]

kubamracek added a comment.

In https://reviews.llvm.org/D29994#683576, @fjricci wrote:

> I believe the issue is that we're using too small of a stack in that test, and perhaps we're just pushed over the limit by this new AllocatorCache. If I change `-DkStackSize=65536` to be a number `>=75392`, the test passes. Is this an acceptable solution? If so, what would be a good number to use?


Ok, that's even weirder.  Why should this change the local stack overhead?  Do we overflow the stack?  Can you actually post the backtrace from the debugger?  Why is there a nested ASan failure?


Repository:
  rL LLVM

https://reviews.llvm.org/D29994