Commit Graph

1325 Commits

Author SHA1 Message Date
Suleiman Souhlal
679985d03a Allow EVFILT_VNODE events to work on every filesystem type, not just
UFS by:
- Making the pre and post hooks for the VOP functions work even when
DEBUG_VFS_LOCKS is not defined.
- Moving the KNOTE activations into the corresponding VOP hooks.
- Creating a MNTK_NOKNOTE flag for the mnt_kern_flag field of struct
mount that permits filesystems to disable the new behavior.
- Creating a default VOP_KQFILTER function: vfs_kqfilter()

My benchmarks have not revealed any performance degradation.

Reviewed by:	jeff, bde
Approved by:	rwatson, jmg (kqueue changes), grehan (mentor)
2005-06-09 20:20:31 +00:00
Ken Smith
6341095e0d This patch addresses a standards violation issue. The standards say a
file's access time should be updated when it gets executed.  A while
ago the mechanism used to exec was changed to use a more mmap based
mechanism and this behavior was broken as a side-effect of that.

A new vnode flag is added that gets set when the file gets executed,
and the VOP_SETATTR() vnode operation gets called.  The underlying
filesystem is expected to handle it based on its own semantics, some
filesystems don't support access time at all.  Those that do should
handle it in a way that does not block, does not generate I/O if possible,
etc.  In particular vn_start_write() has not been called.  The UFS code
handles it the same way as it would normally handle the access time if
a file was read - the IN_ACCESS flag gets set in the inode but no other
action happens at this point.  The actual time update will happen later
during a sync (which handles all the necessary locking).

Got me into this:	cperciva
Discussed with:		a lot with bde, a little with kan
Showed patches to:	phk, jeffr, standards@, arch@
Minor discussion on:	arch@
2005-05-31 19:39:52 +00:00
Jeff Roberson
204ec66d38 - Don't set our bio op to be a READ when we've just completed a write. There
are subtle differences in the read and write completion path.  Instead,
   grab an extra write ref so the write path can drop it when we recursively
   call bufdone().  I believe this may be the source of the wrong bufobj
   panics.

Reported by:	pho, kkenn
2005-05-30 07:04:15 +00:00
Kirk McKusick
6a52a06851 Allow removal of empty directories with high link counts. These can
occur on a filesystem running with soft updates after a crash and
before a background fsck has been run. To prevent discrepancies
from arising in a background fsck that may already be running,
the directory is removed but its inode is not freed and is left
with the residual reference count. When encountered by the
background fsck it will be reclaimed.
2005-05-18 22:18:21 +00:00
Jeff Roberson
6c71a2208d - Don't restrict the softdep stats to DEBUG kernels, they cost nothing to
export.  This was happening anyway since this file manually sets DEBUG.
 - Add a sysctl for the number of items on the worklist.
 - Use a more canonical loop restart in softdep_fsync_mountdev, it saves
   some code at the expense of a goto and makes me worry less about
   modifying a variable that should be private to the TAILQ_FOREACH_SAFE
   macro.
2005-05-03 11:03:29 +00:00
Jeff Roberson
2524c26de8 - Use bdone() directly instead of calling it indirectly through
ffs_rawreaddone().

Sponsored by:	Isilon Systems, Inc.
2005-04-30 11:28:19 +00:00
Pawel Jakub Dawidek
231b1be179 - Plug memory leak.
- Fix two style nits.

Found by:	Coverity Prevent analysis tool
Reviewed by:	rwatson
MFC after:	1 week
2005-04-16 10:57:49 +00:00
Jeff Roberson
4585e3ac5a - Change all filesystems and vfs_cache to relock the dvp once the child is
locked in the ISDOTDOT case.  Se vfs_lookup.c r1.79 for details.

Sponsored by:	Isilon Systems, Inc.
2005-04-13 10:59:09 +00:00
Jeff Roberson
ece9473efa - Consistently call 'vp' vp rather than ovp sometimes in ffs_truncate().
Do the same for oip.

Pointed out by:	glebius
2005-04-05 08:49:41 +00:00
Jeff Roberson
bcc8f66c8b - Use M_ZERO rather than explicitly calling bzero().
- Don't intermingle direct calls to lockmgr and indirect calls through
   VOPs.  This will be important in the future.
 - Dont lock the devvp's interlock just to release it on the next line by
   passing LK_INTERLOCK to lockmgr.
 - Restructure ffs_snapshot_unmount so we don't call free() with the
   devvp's interlock locked.
2005-04-03 12:03:44 +00:00
Jeff Roberson
41d4783d49 - In ffs_sync we need to pass LK_SLEEPFAIL in when we lock the vnode
because it may change identities while we're sleeping on the lock.
   Otherwise we may bail out of ffs_sync() early due to an error from
   deadfs.
 - Collapse a VOP_UNLOCK, vrele into a single vput().
2005-04-03 10:38:18 +00:00
Jeff Roberson
153910e0f5 - Move the contents of softdep_disk_prewrite into ffs_geom_strategy to fix
two bugs.
 - ffs_disk_prewrite was pulling the vp from the buf and checking for
   COPYONWRITE, when really it wanted the vp from the bufobj that we're
   writing to, which is the devvp.  This lead to us skipping the copy on
   write to all file data, which significantly broke snapshots for the
   last few months.
 - When the SOFTUPDATES option was not included in the kernel config we
   would also skip the copy on write check, which would effectively disable
   snapshots.
 - Remove an invalid mp_fixme().

Debugging tips from:	mckusick
Reported by:		iedowse, others
Discussed with:		phk
2005-04-03 10:29:55 +00:00
Jeff Roberson
278c5a6efa - Fix botched LK_NOWAIT removal. I mistakenly thought this compiled as
part of GENERIC.
2005-03-31 05:58:14 +00:00
Jeff Roberson
aa7ba42796 - FFS supports shared locks, clear LK_NOSHARE from our vnode locks.
Sponsored by:	Isilon Systems, Inc.
2005-03-31 05:23:20 +00:00
Jeff Roberson
ec3db02a3e - Set LK_NOSHARE for snapshot locks. snapshots require exclusive only
access.
 - Remove the hack from ffs_lock() to implement LK_NOSHARE in a ffs
   specific way.

Sponsored by:	Isilon Systems, Inc.
2005-03-31 05:21:17 +00:00
Jeff Roberson
f247a5240d - LK_NOPAUSE is a nop now.
Sponsored by:   Isilon Systems, Inc.
2005-03-31 04:37:09 +00:00
Jeff Roberson
52f6886551 - Remove wantparent, it is no longer necessary. An assert in vfs_lookup.c
prevents any callers from doing a modifying op without
   LOCKPARENT or WANTPARENT.  It wasn't even properly used in the CREATE
   or DELETE cases.
2005-03-29 13:16:38 +00:00
Jeff Roberson
d6919865fa - Upgrade a shared lock request to exclusive in ffs_vget() if we have
to create the vnode.

Sponsored by:	Isilon Systems, Inc.
2005-03-29 10:10:51 +00:00
Jeff Roberson
a69c43548d - Honor the cn_lkflags passed from namei() when locking the leaf.
Sponsored by:	Isilon Systems, Inc.
2005-03-29 10:10:01 +00:00
Jeff Roberson
e19881ff08 - UFS no longer uses PDIRUNLOCK to track the parent state. Instead, we now
rely on ufs to always leave the parent locked except in the ISDOTDOT
   case.  Adjust asserts to deal with these changes.

Sponsored by:	Isilon Systems, Inc.
2005-03-28 09:35:58 +00:00
Jeff Roberson
eddcb03d02 - We no longer have to bother with PDIRUNLOCK, lookup() handles it for us.
Sponsored by:   Isilon Systems, Inc.
2005-03-28 09:34:36 +00:00
David Schultz
188f6433f6 When the softupdates worklist gets too long, threads that attempt to
add more work are forced to process two worklist items first.
However, processing an item may generate additional work, causing the
unlucky thread to recursively process the worklist.  Add a per-thread
flag to detect this situation and avoid the recursion.  This should
fix the stack overflows that could occur while removing large
directory trees.

Tested by:	kris
Reviewed by:	mckusick
2005-03-25 17:30:31 +00:00
Jeff Roberson
080c061ad0 - Call VFS_ROOT() with LK_EXCLUSIVE.
Sponsored by:	Isilon Systems, Inc.
2005-03-24 07:33:45 +00:00
Jeff Roberson
469ec10c1e - Update the ufs_root() prototype.
- Pass the ufs_root() flags argument to VFS_VGET() to allow callers to
   specify shared locks.

Sponsored by:	Isilon Systems, Inc.
2005-03-24 07:32:50 +00:00
Jeff Roberson
23d15e852d - Lock the clearing of v_data in ufs_reclaim() to prevent a pagefault
in ffs_lock() when it acesses v_data without the vnlock.

Sponsored by:	Isilon Systems, Inc.
2005-03-17 11:58:43 +00:00
Poul-Henning Kamp
51f5ce0c8c Add two arguments to the vfs_hash() KPI so that filesystems which do
not have unique hashes (NFS) can also use it.
2005-03-16 11:20:51 +00:00
Poul-Henning Kamp
de68347b1b Don't hold a reference on the disk vnode for each inode. 2005-03-15 20:50:58 +00:00
Poul-Henning Kamp
45c26fa2b6 Improve the vfs_hash() API: vput() the unneeded vnode centrally to
avoid replicating the vput in all the filesystems.
2005-03-15 20:00:03 +00:00
Poul-Henning Kamp
e82ef95c11 Simplify the vfs_hash calling convention. 2005-03-15 08:07:07 +00:00
Jeff Roberson
4483fe9227 - Destroy the vnode object earlier in VOP_RECLAIM as we need more of
the vnode valid before the vm flushes pages.
 - Get rid of some extraneous uses of the vnode interlock.

Sponsored by:	Isilon Systems, Inc.
2005-03-15 01:42:58 +00:00
Poul-Henning Kamp
14bc0685ac Use vfs_hash instead of home-rolled. 2005-03-14 10:21:16 +00:00
Jeff Roberson
9cbe5da9d5 - It is not legal to access v_data without the vnode lock or interlock
held.  Grab the vnode interlock if LK_INTERLOCK has not been passed in
   so that we can inspect v_data in ffs_lock().

Sponsored by:	Isilon Systems, Inc.
2005-03-13 12:04:12 +00:00
Jeff Roberson
fe68abe291 - The VI_DOOMED flag now signals the end of a vnode's relationship with
the filesystem.  Check that rather than VI_XLOCK.
 - Shorten ffs_reload by one step.  The old check for an inactive vnode
   was slightly racey, and the code which deals with still active vnodes
   is not much more expensive.

Sponsored by:	Isilon Systems, Inc.
2005-03-13 12:03:14 +00:00
Jeff Roberson
fdcc82276e - The VI_DOOMED flag now signals the end of a vnode's relationship with
the filesystem.  Check that rather than VI_XLOCK.

Sponsored by:	Isilon Systems, Inc.
2005-03-13 12:01:50 +00:00
Jeff Roberson
b5411d4fcb - Fix an assert now that the XLOCK no longer exists.
Sponsored by:	Isilon Systems, Inc.
2005-03-13 12:00:41 +00:00
Jeff Roberson
b6ee8476d3 - In ufs_mknod(), hold the lock across the call to vgone() as that is now
required.
 - In ufs_close(), don't do the EAGAIN vrele hack, the top layer now calls
   vn_start_write before the lock is acquired as it should.

Sponsored by:	Isilon Systems, Inc.
2005-03-13 11:59:14 +00:00
Jeff Roberson
38d504db44 - Don't drop the lock in ufs_inactive().
- Also in ufs_inactive, don't acquire the vnode interlock where it isn't
   strictly needed.  Also owning the vnode interlock while calling vprint()
   will cause locking assertions to trip.

Sponsored by:	Isilon Systems, Inc.
2005-03-13 11:57:39 +00:00
Jeff Roberson
41766826eb - Fix anoter dyslexic moment; an atomic_set_int should've become ACTIVESET,
not ACTIVECLEAR.

Submitted by:	iedowse
2005-03-01 07:38:45 +00:00
Poul-Henning Kamp
7ce296cf04 Remove debug printout of major/minor numbers, print name instead. 2005-02-27 21:16:26 +00:00
Sam Leffler
d5bbad8372 use uiomove return value instead of always returning 0 when doing a
readlink of a fast link

Noticed by:	Coverity Prevent analysis tool
Reviewed by:	phk
2005-02-27 18:58:31 +00:00
Jeff Roberson
1a4a9672f1 - Add VOP locking asserts in several functions that have been implicated in
recent deadlocks.
2005-02-22 23:56:42 +00:00
Xin LI
a16baf37b9 The recomputation of file system summary at mount time can be a
very slow process, especially for large file systems that is just
recovered from a crash.

Since the summary is already re-sync'ed every 30 second, we will
not lag behind too much after a crash.  With this consideration
in mind, it is more reasonable to transfer the responsibility to
background fsck, to reduce the delay after a crash.

Add a new sysctl variable, vfs.ffs.compute_summary_at_mount, to
control this behavior.  When set to nonzero, we will get the
"old" behavior, that the summary is computed immediately at mount
time.

Add five new sysctl variables to adjust ndir, nbfree, nifree,
nffree and numclusters respectively.  Teach fsck_ffs about these
API, however, intentionally not to check the existence, since
kernels without these sysctls must have recomputed the summary
and hence no adjustments are necessary.

This change has eliminated the usual tens of minutes of delay of
mounting large dirty volumes.

Reviewed by:	mckusick
MFC After:	1 week
2005-02-20 08:02:15 +00:00
Poul-Henning Kamp
dfd4be14bd Try to unbreak the vnode locking around vop_reclaim() (based mostly on
patch from kan@).

Pull bufobj_invalbuf() out of vinvalbuf() and make g_vfs call it on
close.  This is not yet a generally safe function, but for this very
specific use it is safe.  This solves the problem with buffers not
being flushed by unmount or after failed mount attempts.
2005-02-19 11:44:57 +00:00
Xin LI
d5128ab2af When clearing a fragment, it's possible that the length is zero.
Reviewed by:	mckusick
MFC After:	1 week
2005-02-19 07:31:33 +00:00
Jeff Roberson
a8127ebb5d - Remove the unused and unsafe ufs_ihashlookup. This function returned a
vnode pointer that could not be used since no locks were held.

Sponsored by:	Isilon Systems, Inc.
2005-02-14 20:51:39 +00:00
Poul-Henning Kamp
1121c39497 Make non-SOFTUPDATES kernels compile again.
Integrate the stubfile into the main file now that license issues have been
long resolved.
2005-02-11 08:13:31 +00:00
Poul-Henning Kamp
adf4157738 Make a some SYSCTL_NODEs and some of FFS's VFS_ methods static. 2005-02-10 12:20:08 +00:00
Jeff Roberson
a3caf16e99 - In the softupdates case for ffs_truncate() we use vinvalbuf() to
invalidate pending io and dependencies.  However, vinvalbuf() rightfully
   does not call vnode_pager_setsize() for us.  We must do this here.  This
   could potentially have caused numerous kinds of bugs, but it was
   specifically causing msync() deadlocks because msync() was writing
   flushing pages that should not have been valid.

Sponsored by:	Isilon Systems, Inc.
Reported by:	kkenn
2005-02-09 23:05:20 +00:00
Poul-Henning Kamp
365b18aa89 style polishing. 2005-02-09 12:22:16 +00:00
Colin Percival
79653046d8 Add a new sysctl, "security.jail.chflags_allowed", which controls the
behaviour of chflags within a jail.  If set to 0 (the default), then a
jailed root user is treated as an unprivileged user; if set to 1, then
a jailed root user is treated the same as an unjailed root user.

This is necessary to allow "make installworld" to work inside a jail,
since it attempts to manipulate the system immutable flag on certain
files.

Discussed with:	csjp, rwatson
MFC after:	2 weeks
2005-02-08 21:31:11 +00:00