[PATCH] D131009: [analyzer] Fixing a bug raising false positives of stack block object leaking under ARC

[Ziqing Luo via Phabricator via cfe-commits]

ziqingluo-90 added inline comments.


================
Comment at: clang/lib/StaticAnalyzer/Checkers/StackAddrEscapeChecker.cpp:313-315
+      // Under ARC, blocks are retained and released automatically:
+      if (isArcManagedBlock(Referred, Ctx))
+        return false;
----------------
NoQ wrote:
> Aha ok, it sounds like we can no longer be sure that the block is on the stack at this point, did I get it right?
> 
> In this case I think it's more productive to have the block's memory space be `UnknownSpaceRegion` from the start, so that it fell through the memory space check, both here and at other call sites of `isArcManagedBlock()` (so it can be removed), and in any other code that relies on memory spaces (so this mistake is never made again).
//" ..., did I get it right?"//  Yes.

This suggestion makes sense to me.  To my understanding, I need to modify the symbolic execution engine to address it.  So shall I do it in a new patch?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D131009/new/

https://reviews.llvm.org/D131009