filesystem supports additional operations using shared vnode locks.
Currently this is used to enable shared locks for open() and close() of
read-only file descriptors.
- When an ISOPEN namei() request is performed with LOCKSHARED, use a
shared vnode lock for the leaf vnode only if the mount point has the
extended shared flag set.
- Set LOCKSHARED in vn_open_cred() for requests that specify O_RDONLY but
not O_CREAT.
- Use a shared vnode lock around VOP_CLOSE() if the file was opened with
O_RDONLY and the mountpoint has the extended shared flag set.
- Adjust md(4) to upgrade the vnode lock on the vnode it gets back from
vn_open() since it now may only have a shared vnode lock.
- Don't enable shared vnode locks on FIFO vnodes in ZFS and UFS since
FIFO's require exclusive vnode locks for their open() and close()
routines. (My recent MPSAFE patches for UDF and cd9660 already included
this change.)
- Enable extended shared operations on UFS, cd9660, and UDF.
Submitted by: ups
Reviewed by: pjd (ZFS bits)
MFC after: 1 month
This bring huge amount of changes, I'll enumerate only user-visible changes:
- Delegated Administration
Allows regular users to perform ZFS operations, like file system
creation, snapshot creation, etc.
- L2ARC
Level 2 cache for ZFS - allows to use additional disks for cache.
Huge performance improvements mostly for random read of mostly
static content.
- slog
Allow to use additional disks for ZFS Intent Log to speed up
operations like fsync(2).
- vfs.zfs.super_owner
Allows regular users to perform privileged operations on files stored
on ZFS file systems owned by him. Very careful with this one.
- chflags(2)
Not all the flags are supported. This still needs work.
- ZFSBoot
Support to boot off of ZFS pool. Not finished, AFAIK.
Submitted by: dfr
- Snapshot properties
- New failure modes
Before if write requested failed, system paniced. Now one
can select from one of three failure modes:
- panic - panic on write error
- wait - wait for disk to reappear
- continue - serve read requests if possible, block write requests
- Refquota, refreservation properties
Just quota and reservation properties, but don't count space consumed
by children file systems, clones and snapshots.
- Sparse volumes
ZVOLs that don't reserve space in the pool.
- External attributes
Compatible with extattr(2).
- NFSv4-ACLs
Not sure about the status, might not be complete yet.
Submitted by: trasz
- Creation-time properties
- Regression tests for zpool(8) command.
Obtained from: OpenSolaris
- Implement real draining for vfs consumers by not relying on the
mnt_lock and using instead a refcount in order to keep track of lock
requesters.
- Due to the change above, remove the mnt_lock lockmgr because it is now
useless.
- Due to the change above, vfs_busy() is no more linked to a lockmgr.
Change so its KPI by removing the interlock argument and defining 2 new
flags for it: MBF_NOWAIT which basically replaces the LK_NOWAIT of the
old version (which was unlinked from the lockmgr alredy) and
MBF_MNTLSTLOCK which provides the ability to drop the mountlist_mtx
once the mnt interlock is held (ability still desired by most consumers).
- The stub used into vfs_mount_destroy(), that allows to override the
mnt_ref if running for more than 3 seconds, make it totally useless.
Remove it as it was thought to work into older versions.
If a problem of "refcount held never going away" should appear, we will
need to fix properly instead than trust on such hackish solution.
- Fix a bug where returning (with an error) from dounmount() was still
leaving the MNTK_MWAIT flag on even if it the waiters were actually
woken up. Just a place in vfs_mount_destroy() is left because it is
going to recycle the structure in any case, so it doesn't matter.
- Remove the markercnt refcount as it is useless.
This patch modifies VFS ABI and breaks KPI for vfs_busy() so manpages and
__FreeBSD_version will be modified accordingly.
Discussed with: kib
Tested by: pho
after each SYSINIT() macro invocation. This makes a number of
lightweight C parsers much happier with the FreeBSD kernel
source, including cflow's prcc and lxr.
MFC after: 1 month
Discussed with: imp, rink
always curthread.
As KPI gets broken by this patch, manpages and __FreeBSD_version will be
updated by further commits.
Tested by: Andrea Barberio <insomniac at slackware dot it>
namespace in order to handle lockmgr fields in a controlled way instead
than spreading all around bogus stubs:
- VN_LOCK_AREC() allows lock recursion for a specified vnode
- VN_LOCK_ASHARE() allows lock sharing for a specified vnode
In FFS land:
- BUF_AREC() allows lock recursion for a specified buffer lock
- BUF_NOREC() disallows recursion for a specified buffer lock
Side note: union_subr.c::unionfs_node_update() is the only other function
directly handling lockmgr fields. As this is not simple to fix, it has
been left behind as "sole" exception.
conjuction with 'thread' argument passing which is always curthread.
Remove the unuseful extra-argument and pass explicitly curthread to lower
layer functions, when necessary.
KPI results broken by this change, which should affect several ports, so
version bumping and manpage update will be further committed.
Tested by: kris, pho, Diego Sardina <siarodx at gmail dot com>
Remove this argument and pass curthread directly to underlying
VOP_LOCK1() VFS method. This modify makes the code cleaner and in
particular remove an annoying dependence helping next lockmgr() cleanup.
KPI results, obviously, changed.
Manpage and FreeBSD_version will be updated through further commits.
As a side note, would be valuable to say that next commits will address
a similar cleanup about VFS methods, in particular vop_lock1 and
vop_unlock.
Tested by: Diego Sardina <siarodx at gmail dot com>,
Andrea Di Pasquale <whyx dot it at gmail dot com>
from Mac OS X Leopard--rationalize naming for entry points to
the following general forms:
mac_<object>_<method/action>
mac_<object>_check_<method/action>
The previous naming scheme was inconsistent and mostly
reversed from the new scheme. Also, make object types more
consistent and remove spaces from object types that contain
multiple parts ("posix_sem" -> "posixsem") to make mechanical
parsing easier. Introduce a new "netinet" object type for
certain IPv4/IPv6-related methods. Also simplify, slightly,
some entry point names.
All MAC policy modules will need to be recompiled, and modules
not updates as part of this commit will need to be modified to
conform to the new KPI.
Sponsored by: SPARTA (original patches against Mac OS X)
Obtained from: TrustedBSD Project, Apple Computer
and flags with an sxlock. This leads to a significant and measurable
performance improvement as a result of access to shared locking for
frequent lookup operations, reduced general overhead, and reduced overhead
in the event of contention. All of these are imported for threaded
applications where simultaneous access to a shared file descriptor array
occurs frequently. Kris has reported 2x-4x transaction rate improvements
on 8-core MySQL benchmarks; smaller improvements can be expected for many
workloads as a result of reduced overhead.
- Generally eliminate the distinction between "fast" and regular
acquisisition of the filedesc lock; the plan is that they will now all
be fast. Change all locking instances to either shared or exclusive
locks.
- Correct a bug (pointed out by kib) in fdfree() where previously msleep()
was called without the mutex held; sx_sleep() is now always called with
the sxlock held exclusively.
- Universally hold the struct file lock over changes to struct file,
rather than the filedesc lock or no lock. Always update the f_ops
field last. A further memory barrier is required here in the future
(discussed with jhb).
- Improve locking and reference management in linux_at(), which fails to
properly acquire vnode references before using vnode pointers. Annotate
improper use of vn_fullpath(), which will be replaced at a future date.
In fcntl(), we conservatively acquire an exclusive lock, even though in
some cases a shared lock may be sufficient, which should be revisited.
The dropping of the filedesc lock in fdgrowtable() is no longer required
as the sxlock can be held over the sleep operation; we should consider
removing that (pointed out by attilio).
Tested by: kris
Discussed with: jhb, kris, attilio, jeff
explicitly test and panic. This should not ever happen, but if it does,
this is a preferred failure mode to a NULL pointer dereference in kernel.
Coverity CID: 1716
Found with: Coverity Prevent(tm)
of the special handling for ".." and perform an ISDOTDOT VOP_LOOKUP()
for a filesystem root vnode. Handle this case inside lookup().
Submitted by: tegge
PR: 92785
MFC after: 1 week
--------------------------
[Deadlock] is caused by a lock order reversal in vfs_lookup(), where
[some] process is trying to lock a directory vnode, that is the parent
directory of covered vnode) while holding an exclusive vnode lock on
covering vnode.
A simplified scenario:
root fs var fs
/ A / (/var) D
/var B /log (/var/log) E
vfs lock C vfs lock F
Within each file system, the lock order is clear: C->A->B and F->D->E
When traversing across mounts, the system can choose between two lock orders,
but everything must then follow that lock order:
L1: C->A->B
|
+->F->D->E
L2: F->D->E
|
+->C->A->B
The lookup() process for namei("/var") mixes those two lock orders:
VOP_LOOKUP() obtains B while A is held
vfs_busy() obtains a shared lock on F while A and B are held (follows L1,
violates L2)
vput() releases lock on B
VOP_UNLOCK() releases lock on A
VFS_ROOT() obtains lock on D while shared lock on F is held
vfs_unbusy() releases shared lock on F
vn_lock() obtains lock on A while D is held (violates L1, follows L2)
dounmount() follows L1 (B is locked while F is drained).
Without unmount activity, vfs_busy() will always succeed without blocking
and the deadlock isn't triggered (the system behaves as if L2 is followed).
With unmount, you can get 4 processes in a deadlock:
p1: holds D, want A (in lookup())
p2: holds shared lock on F, want D (in VFS_ROOT())
p3: holds B, want drain lock on F (in dounmount())
p4: holds A, want B (in VOP_LOOKUP())
You can have more than one instance of p2.
The reversal was introduced in revision 1.81 of src/sys/kern/vfs_lookup.c and
MFCed to revision 1.80.2.1, probably to avoid a cascade of vnode locks when nfs
servers are dead (VFS_ROOT() just hangs) spreading to the root fs root vnode.
- Tor Egge
To fix the LOR, ups@ noted that when crossing the mount point, ni_dvp
is actually not used by the callers of namei. Thus, placeholder deadfs
vnode vp_crossmp is introduced that is filled into ni_dvp.
Idea by: ups
Reviewed by: tegge, ups, jeff, rwatson (mac interaction)
Tested by: Peter Holm
MFC after: 2 weeks
begun with a repo-copy of mac.h to mac_framework.h. sys/mac.h now
contains the userspace and user<->kernel API and definitions, with all
in-kernel interfaces moved to mac_framework.h, which is now included
across most of the kernel instead.
This change is the first step in a larger cleanup and sweep of MAC
Framework interfaces in the kernel, and will not be MFC'd.
Obtained from: TrustedBSD Project
Sponsored by: SPARTA
adds a FS type specific flag indicating that the FS supports shared
vnode lock lookups, adds some logic in vfs_lookup.c to test this flag
and set lock flags appropriately.
- amd on 6.x is a non-starter (without this change). Using amd under
heavy load results in a deadlock (with cascading vnode locks all the
way to the root) very quickly.
- This change should also fix the more general problem of cascading
vnode deadlocks when an NFS server goes down.
Ideally, we wouldn't need these changes, as enabling shared vnode lock
lookups globally would work. Unfortunately, UFS, for example isn't
ready for shared vnode lock lookups, crashing pretty quickly.
This change is the result of discussions with Stephan Uphoff (ups@).
Reviewed by: ups@
than trying to optimize it into a single lock. This adds more calls to
lock giant with non smpsafe filesystems but is the only way to reliably
hold the correct lock.
- Remove an invalid assert in the mountedhere case in lookup and fix the
code to properly deal with the scenario. We can actually have a lookup
that returns dp == dvp with mountedhere set with certain unmount races.
Tested by: kris
Reported by: kris/mohans
- Move the vn_lock of the dvp until after we've unbusied the filesystem
to avoid a LOR with the mount point lock.
- In the v_mountedhere while loop we acquire a new instance of giant each
time through without releasing the first. This would cause us to leak
Giant.
Sponsored by: Isilon Systems, Inc.
and vnode attribute information for looked up vnodes during the lookup
operation. This will allow consumers of namei() to specify that this
information be added to the in-process audit record.
Submitted by: wsalamon
Obtained from: TrustedBSD Project
without Giant held. Do this by tracking the vfslocked state for
the directory seperate from the child. This is only important
in the case where we cross a mountpoint.
Sponsored by: Isilon Systems, Inc.
MFC After: 3 days
lookup() instead of EPERM when a DELETE or RENAME operation is
attempted on "..".
In kern_unlink(), remap EINVAL errors returned from namei() to EPERM
to match existing (and POSIX required) behaviour.
Discussed with: bde
MFC after: 3 days
the last component of the path name is "..". This keeps VOP_LOOKUP()
from locking vnodes in reverse order.
Tested by: Denis Shaposhnikov <dsh AT vlink DOT ru>
MFC after: 3 days
links and the execution of ELF binaries. Two problems were found:
1) The link path wasn't tagged as being MP safe and thus was not properly
protected.
2) The ELF interpreter vnode wasnt being locked in namei(9) and thus was
insufficiently protected.
This commit makes the following changes:
-Sets the MPSAFE flag in NDINIT for symbolic link paths
-Sets the MPSAFE flag in NDINIT and introduce a vfslocked variable which
will be used to instruct VFS_UNLOCK_GIANT to unlock Giant if it has been
picked up.
-Drop in an assertion into vfs_lookup which ensures that if the MPSAFE
flag is NOT set, that we have picked up giant. If not panic (if WITNESS
compiled into the kernel). This should help us find conditions where vnode
operations are in-sufficiently protected.
This is a RELENG_6 candidate.
Discussed with: jeff
MFC after: 4 days
points in lookup(). The lock can be dropped safely around VFS_ROOT because
LOCKPARENT semantics with child and perent vnodes coming from different FSes
does not really have any meaningful use. On the other hard, this prevents
easily triggered deadlock on systems using automounter daemon.
case. There are bugs in some which didn't unlock in the ISDOTDOT case
to begin with that need to be addressed seperately. This simplifies
things anyway.
- Fix relookup() to prevent it from vrele()'ing the dvp while the vp
is locked. Catch up to other lookup changes.
Sponsored by: Isilon Systems, Inc.
Reported by: Peter Wemm
when vrele() acquires the directory lock in the wrong order. Fix this
via the following changes:
- Keep the directory locked after VOP_LOOKUP() until we've determined
what we're going to do with the child. This allows us to remove the
complicated post LOOKUP code which determins whether we should lock or
unlock the parent. This means we may have to vput() in the appropriate
cases later, rather than doing an unsafe vrele.
- in NDFREE() keep two flags to indicate whether we need to unlock vp or
dvp. This allows us to vput rather than vrele in the appropriate
cases without rechecking the flags. Move the code to handle dvp after
we handle vp.
- Remove some dead code from namei() that was the result of changes to
VFS_LOCK_GIANT().
Sponsored by: Isilon Systems, Inc.
acquire shared locks on intermediate directories.
- For the LASTCN, we may have to LK_UPGRADE the parent directory before
we lookup the last component.
- Acquire VFS_ROOT and dp locks based on the cn_lkflag.
Sponsored by: Isilon Systems, Inc.
- Assert that REMOVE, CREATE, and RENAME callers have WANTPARENT
or LOCKPARENT set. You can't complete any of these operations without
at least a reference to the parent. Many filesystems check for this case
even though it isn't possible in the current system.
calling VOP_LOOKUP(). Rather than having each filesystem check the
LOCKPARENT flag, we simply check it once here and unlock as required.
The only unusual case is ISDOTDOT, where we require an unlocked vnode
on return. Relocking this vnode with the child locked is allowed since
the child is actually its parent.
- Add a few asserts for some unusual conditions that I do not believe can
happen. These will later go away and turn into implementations for these
conditions.
Sponsored by: Isilon Systems, Inc.
necessary since we disable the shared locks in vfs_cache, but it is
prefered that the option not leak out into filesystems when it is
disabled.
Sponsored by: Isilon Systems, Inc.
