[LTP] [PATCH v4] syscalls/sync_file_range: add partial file sync test-cases

Amir Goldstein amir73il@gmail.com
Wed Jun 19 12:28:26 CEST 2019


On Wed, Jun 19, 2019 at 12:58 PM Cyril Hrubis <chrubis@suse.cz> wrote:
>
> Hi!
> > > With these modification the test was stable and worked fine for more
> > > than 100 iterations for me. It still fails for FUSE based filesystems
> > > but that is to be expected since FUSE does not implement sync_file_range
> > > yet.
> > >
> >
> > What is the proposed way to resolve this failure?
> > If FUSE does not implement sync_file_range, shouldn't test detect
> > it and return TCONF?
>
> Well the call returns success but does not sync anything on FUSE based
> filesystems so the choices here are:
>
> 1) Disable the test of FUSE
> 2) Keep the test failing and ignore the failures
>
> Which one do you prefer?
>

I prefer the first option.

My preference may seem inconsistent with my opinion that
F_SETLEASE tests should NOT be disabled for overlayfs.

The difference is that sync_file_range() API does not promise
any real guaranties vs. F_SETLEASE that has a very clear
definition of how it should behave.
A subtle difference, but I think it matters.

BTW, I did already fix the overlayfs F_SETLEASE bug.
It is queued for 5.3 (by file locks maintainer).
I will post fixes to LTP F_SETLEASE tests once the kernel
fix is merged.

Thanks,
Amir.


More information about the ltp mailing list