[PATCH] D56949: FileOutputBuffer: handle mmap(2) failure

[Rui Ueyama via Phabricator via llvm-commits]

ruiu added a comment.

> I wonder if there are other places where this fallback is already performed at a higher level that we could presumably now remove some code?  It seems like other tools such as clang were already handling the lack of mmap since the bug reporter didn't hit an issue until link time.

I don't think testing whether this fallback will be triggered or not can be done at a higher level prior to actually creating a file and call mmap(2) on the file, as the condition of the failure depends on the filesystem. Even on the same system and possibly with the same pathname, it may or may not fail, depending on the filesystem of the destination directory.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D56949/new/

https://reviews.llvm.org/D56949