[PATCH] D68193: In openFileForRead don't cache erroneous entries if the error relates to them being directories. Add tests.

[Michael Spencer via Phabricator via cfe-commits]

Bigcheese added a comment.

In D68193#1701426 <https://reviews.llvm.org/D68193#1701426>, @kousikk wrote:

> @arphaman I don't mind changing this if there are race conditions as you say, but isn't the assumption of the tool that the filesystem remains unchanged for a single run of the tool? If so, should we actually throw error conditions instead of crashing in those cases?


This is true for a normal build, but it is not true with modules. All file system access goes through this, even for modules. I'm currently fixing an issue where we cache the non-existence of the module cache directory, causing the subsequent module load to fail. There is also an issue due to minimization where if we return the underlying stat result, the file size after minimization is different.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D68193/new/

https://reviews.llvm.org/D68193