[PATCH] D103745: [dfsan] Add full fast8 support

[Andrew via Phabricator via cfe-commits]

browneee added a comment.

Sorry, I was thinking/describing fast-16 to fast-8.

The same reasons apply for legacy mode (2^16 labels) to fast-8.

- simplify code (smaller codesize overhead)
- share more code and shadow/origin memory layout with MSan (which also has a 1:1 shadow)
- reduce memory overhead

In this case it also reduces runtime overhead.

Our experience was legacy mode did not scale to large programs. The choice we saw was do multiple runs with fast-8, or be unable to run with legacy.
Using fast-8 we can successfully run dfsan with large programs.

Having said this, we also fixed many other issues (e.g. memory leaks) to make fast-8 scale. So our experience comparing legacy and fast-8 mode is not quite fair.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D103745/new/

https://reviews.llvm.org/D103745