linux/fs/autofs4
Ian Kent 092a53452b autofs: take more care to not update last_used on path walk
GUI environments seem to be becoming more agressive at scanning
filesystems, to the point where autofs cannot expire mounts at all.

This is one key reason the update of the autofs dentry info last_used
field is done in the expire system when the dentry is seen to be in use.

But somewhere along the way instances of the update has crept back into
the autofs path walk functions which, with the more aggressive file
access patterns, is preventing expiration.

Changing the update in the path walk functions allows autofs to at least
make progress in spite of frequent immediate re-mounts from file
accesses.

Link: http://lkml.kernel.org/r/148577167169.9801.1377050092212016834.stgit@pluto.themaw.net
Signed-off-by: Ian Kent <raven@themaw.net>
Cc: Tomohiro Kusumi <tkusumi@tuxera.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-02-27 18:43:45 -08:00
..
autofs_i.h autofs: change autofs4_wait() to take struct path 2016-12-03 20:51:48 -05:00
dev-ioctl.c autofs: remove duplicated AUTOFS_DEV_IOCTL_SIZE definition 2017-02-27 18:43:45 -08:00
expire.c autofs - dont hold spin lock over direct mount expire 2016-12-03 20:51:50 -05:00
init.c autofs4: coding style fixes 2016-03-15 16:55:16 -07:00
inode.c don't open-code file_inode() 2016-12-04 18:29:28 -05:00
Kconfig fs/Kconfig: move autofs, autofs4 out 2009-01-22 13:15:54 +03:00
Makefile
root.c autofs: take more care to not update last_used on path walk 2017-02-27 18:43:45 -08:00
symlink.c vfs: remove ".readlink = generic_readlink" assignments 2016-12-09 16:45:04 +01:00
waitq.c fs: Better permission checking for submounts 2017-02-02 04:36:12 +13:00