lockf: Do not block in vfs_busy()
ClosedPublic
Actions

Authored by markj on Thu, Sep 24, 1:51 PM.

Tags

None

Referenced Files

	F173415877: D59982.id187616.diff
	Fri, Sep 25, 7:45 PM

	F173378581: D59982.id187598.diff
	Fri, Sep 25, 1:47 PM

	F173367653: D59982.diff
	Fri, Sep 25, 12:04 PM

	F173328450: D59982.diff
	Fri, Sep 25, 6:35 AM

	F173301460: D59982.id187598.diff
	Fri, Sep 25, 3:06 AM

	F173300004: D59982.id187616.diff
	Fri, Sep 25, 2:54 AM

	F173295091: D59982.id187616.diff
	Fri, Sep 25, 1:57 AM

	F173295090: D59982.id187598.diff
	Fri, Sep 25, 1:57 AM

View All 12 Files

Subscribers

imp

olce

Details

Reviewers

kib

Commits

rGdd16a5f53b0b: lockf: Do not block in vfs_busy()

Summary

I believe a race is possible otherwise: vfs_busy() may return after an
unmounted filesystem has been removed from the global mount list.
That is, vfs_busy() will block until vfs_mount_destroy() sets
MNTK_REFEXPIRE, and at that point the mountpoint has been removed from
the mountlist, so TAILQ_FOREACH can return an invalid value.

Simply do not block if the mountpoint is being unmounted.

Fixes: eca39864f702 ("Add sysctl KERN_LOCKF")