On Wed, Jun 4, 2025 at 6:09 PM Amir Goldstein <amir73il@xxxxxxxxx> wrote: > > Jan, > > In v1 there was only patch 1 [1] to allow FAN_PRE_ACCESS events > on readdir (with FAN_ONDIR). > > Following your feedback on v1, v2 adds support for FAN_PATH_ACCESS > event so that a non-populated directory could be populted either on > first readdir or on first lookup. > > I am still tagging this as RFC for two semi-related reasons: > > 1) In my original draft of man-page for FAN_PATH_ACCESS [2], > I had introduced a new class FAN_CLASS_PRE_PATH, which FAN_PATH_ACCESS > requires and is defined as: > "Unlike FAN_CLASS_PRE_CONTENT, this class can be used along with > FAN_REPORT_DFID_NAME to report the names of the looked up files along > with O_PATH file descriptos in the new path lookup events." > > I am not sure if we really need FAN_CLASS_PRE_PATH, so wanted to ask > your opinion. > > The basic HSM (as implemented in my POC) does not need to get the lookup > name in the event - it populates dir on first readdir or lookup access. > So I think that support for (FAN_CLASS_PRE_CONTENT | FAN_REPORT_DFID_NAME) > could be added later per demand. > > 2) Current code does not generate FAN_PRE_ACCESS from vfs internal > lookup helpers such as lookup_one*() helpers from overalyfs and nfsd. > This is related to the API of reporting an O_PATH event->fd for > FAN_PATH_ACCESS event, which requires a mount. > > If we decide that we want to support FAN_PATH_ACCESS from all the > path-less lookup_one*() helpers, then we need to support reporting > FAN_PATH_ACCESS event with directory fid. > > If we allow FAN_PATH_ACCESS event from path-less vfs helpers, we still > have to allow setting FAN_PATH_ACCESS in a mount mark/ignore mask, because > we need to provide a way for HSM to opt-out of FAN_PATH_ACCESS events > on its "work" mount - the path via which directories are populated. > > There may be a middle ground: > - Pass optional path arg to __lookup_slow() (i.e. from walk_component()) > - Move fsnotify hook into __lookup_slow() > - fsnotify_lookup_perm() passes optional path data to fsnotify() > - fanotify_handle_event() returns -EPERM for FAN_PATH_ACCESS without > path data > > This way, if HSM is enabled on an sb and not ignored on specific dir > after it was populated, path lookup from syscall will trigger > FAN_PATH_ACCESS events and overalyfs/nfsd will fail to lookup inside > non-populated directories. > > Supporting populate events from overalyfs/nfsd could be implemented > later per demand by reporting directory fid instead of O_PATH fd. > > If you think that is worth checking, I can prepare a patch for the above > so we can expose it to performance regression bots. > > Better yet, if you have no issues with the implementation in this > patch set, maybe let it soak in for_next/for_testing as is to make > sure that it does not already introduce any performance regressions. > > Thoughts? > > Amir. > > Changes since v1: > - Jan's rewrite of patch 1 > - Add support for O_PATH event->fd > - Add FAN_PATH_ACCESS event > > [1] https://lore.kernel.org/all/20250402062707.1637811-1-amir73il@xxxxxxxxx/ > [2] https://github.com/amir73il/man-pages/commits/fan_pre_path Forgot to mention of course that I wrote an LTP test for both events on lookup and readdir: > [3] https://github.com/amir73il/ltp/commits/fan_hsm/ Thanks, Amir.
