On Thu, May 8, 2025 at 10:46 PM Amir Goldstein <amir73il@xxxxxxxxx> wrote: > > On Sat, Apr 19, 2025 at 1:48 PM Amir Goldstein <amir73il@xxxxxxxxx> wrote: > > > > On Sat, Apr 19, 2025 at 12:07 PM Amir Goldstein <amir73il@xxxxxxxxx> wrote: > > > > > > Jan, > > > > > > This v2 is following a two years leap from the RFC path [1]. > > > the code is based on the mntns fix patches I posted and is available > > > on my github [2]. > > > > > > Since then, Christian added support for open_by_handle_at(2) > > > to admin inside userns, which makes watching FS_USERNS_MOUNT > > > sb more useful. > > > > > > And this should also be useful for Miklos' mntns mount tree watch > > > inside userns. > > > > > > Tested sb/mount watches inside userns manually with fsnotifywatch -S > > > and -M with some changes to inotify-tools [3]. > > > > > > Ran mount-notify test manually inside userns and saw that it works > > > after this change. > > > > > > I was going to write a variant of mount-notify selftest that clones > > > also a userns, but did not get to it. > > > > > > Christian, Miklos, > > > > > > If you guys have interest and time in this work, it would be nice if > > > you can help with this test variant or give me some pointers. > > > > > > I can work on the test and address review comments when I get back from > > > vacation around rc5 time, but wanted to get this out soon for review. > > > > > > > FWIW, this is my failed attempt to copy what statmount_test_ns does > > to mount-notify_test_ns: > > > > https://github.com/amir73il/linux/commits/fanotify_selftests/ > > > > Hi Jan, > > This selftests branch is now updated. > The test is working as expected and verifies the changes in this patch set. > > Would you consider queuing the fanotify patches for v6.6? > > We need to collaborate the merge of the selftests with Christian > because my selftests branch has some cleanups with a minor > conflict with Christian's vfs/vfs-6.16.mount branch. > > Maybe you will carry only the fanotify patches to v6.6 and > Christian will carry the tests in a separate branch? > because the fanotify patches and the tests do not actually depend > on each other to build, only for the test to pass. FYI, the selftests to verify these fanotify patches have been queued on Christian's vfs-6.16.selftests branch. Thanks, Amir.
