Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

My understanding of the issue was that the reason to make this change was that some file systems fail (CIFS, at the time) instantly when fsync() is called on directories. As this is an inherent property of the file system, retrying wouldn't change anything. Returning the error to the caller is what we want to avoid as the software should work with these file systems, even if this means being able to make fewer guarantees.

That said, this was added as an application-specific work-around for a 2018 (no, even older) bug in CIFS it may not be needed anymore, nor necessary to upstream.

Returning the error to the caller is what we want to avoid as the software should work with these file systems, even if this means being able to make fewer guarantees.

Understandable. The problem is that this solution makes fewer guarantees on every file system, not just CIFS. fsync for CIFS becoming a no-op in the kernel really is the best solution.