file are after snaplock, while other ffs device buffers are before
snaplock in global lock order. By itself, this could cause deadlock
when bdwrite() tries to flush dirty buffers on snapshotted ffs. If,
during the flush, COW activity for snapshot needs to allocate block
and ffs_alloccg() selects the cylinder group that is being written
by bdwrite(), then kernel would panic due to recursive buffer lock
acquision.
Avoid dealing with buffers in bdwrite() that are from other side of
snaplock divisor in the lock order then the buffer being written. Add
new BOP, bop_bdwrite(), to do dirty buffer flushing for same vnode in
the bdwrite(). Default implementation, bufbdflush(), refactors the code
from bdwrite(). For ffs device buffers, specialized implementation is
used.
Reviewed by: tegge, jeff, Russell Cattelan (cattelan xfs org, xfs changes)
Tested by: Peter Holm
X-MFC after: 3 weeks (if ever: it changes ABI)
This is easy to reproduce for EROFS. I am not sure if the attrs can be corrupt
for other NFS error responses. For now, disabling wcc pre-op attr checks and
post-op attr loads on NFS errors (sysctl'ed).
Reported by: Kris Kennaway
vnode into the vfs hash. Otherwise, another thread walking the hash can trip
on an nfsnode with an uninitialized or partially initialized fh.
Thanks to ups@ for spotting this race.
have to explicitly acquire Giant (although they need to be aware of this and
not hold any locks at that point). Remove the acquisitions of Giant in the
NFS client wrapping tprintf().
file. Leave the loser's lock(s) initialized, so the reclaim logic can
unconditionally destroy them when that race occurs (or if the vfs hash
insert happened to fail for some other reason). Thanks to ups@ for a
careful review of the code.
Reported by : Kris Kennaway
2) Reduce the acquisitions of the Giant lock in the nfs_socket.c paths significantly.
- We don't need to acquire Giant before tsleeping on lbolt anymore,
since jhb specialcased lbolt handling in msleep.
- nfs_up() needs to acquire Giant only if printing the "server up"
message.
- nfs_timer() held Giant for the duration of the NFS timer processing,
just because the printing of the message in nfs_down() needed it
(and we acquire other locks in nfs_timer()). The acquisition of
Giant is moved down into nfs_down() now, reducing the time Giant is
held in that path.
Reported by: Kris Kennaway
This bug results in data corruption with NFS/TCP. Writes are silently dropped
on EWOULDBLOCK (because socket send buffer is full and sockbuf timer fires).
Reviewed by: ups@
VA_MARK_ATIME feature to fix POSIX conformance fore execve() and mmap(),
we thought that it was optimized well enough for the one file system
that supports it (ffs) and harmless for other file systems (except
layered ones which already get the layering for VOP_SETATTR() wrong).
However, nfs_setattr() doesn't do much parameter checking, so when
it gets a combination of parameters that it doesn't understand, it
always does a Setattr RPC. This RPC can't do anything good, and for
VA_MARK_ATIME it is null except for wasting a lot of time.
This is the smallest and easiest to fix of several bugs that have
increased the number of RPCs for kernel builds on nfs by more than
100% since 2004-11-05. The real-time increase depends on network
latency and parallelization and can also be very large (approaching
the same percentage for unparallelized operations like "make depend"
on systems with fast CPUs and high-latency networks).
sync() and sync_fsync() without losing MNT_ASYNC. Add MNTK_ASYNC flag
which is set only when MNT_ASYNC is set and mnt_noasync is zero, and
check that flag instead of MNT_ASYNC before initiating async io.
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@
synchronized by the lock on the object containing the page.
Transition PG_WANTED and PG_SWAPINPROG to use the new field,
eliminating the need for holding the page queues lock when setting
or clearing these flags. Rename PG_WANTED and PG_SWAPINPROG to
VPO_WANTED and VPO_SWAPINPROG, respectively.
Eliminate the assertion that the page queues lock is held in
vm_page_io_finish().
Eliminate the acquisition and release of the page queues lock
around calls to vm_page_io_finish() in kern_sendfile() and
vfs_unbusy_pages().
set the MTU prior to mounting root via NFS. This is required if the
server supports a higher than default MTU because the client will not
see the responses otherwise.
MFC after: 3 weeks
soreceive(), and sopoll(), which are wrappers for pru_sosend,
pru_soreceive, and pru_sopoll, and are now used univerally by socket
consumers rather than either directly invoking the old so*() functions
or directly invoking the protocol switch method (about an even split
prior to this commit).
This completes an architectural change that was begun in 1996 to permit
protocols to provide substitute implementations, as now used by UDP.
Consumers now uniformly invoke sosend(), soreceive(), and sopoll() to
perform these operations on sockets -- in particular, distributed file
systems and socket system calls.
Architectural head nod: sam, gnn, wollman
in nfs_strategy. Otherwise, for some buffers, signals would be ignored
at the intr mounts.
Reviewed by: mohan
MFC after: 1 month
Approved by: kan (mentor)
except in places dealing with ifaddr creation or destruction; and
in such special places incomplete ifaddrs should never be linked
to system-wide data structures. Therefore we can eliminate all the
superfluous checks for "ifa->ifa_addr != NULL" and get ready
to the system crashing honestly instead of masking possible bugs.
Suggested by: glebius, jhb, ru
If B_NOCACHE is set the pages of vm backed buffers will be invalidated.
However clean buffers can be backed by dirty VM pages so invalidating them
can lead to data loss.
Add support for flush dirty page in the data invalidation function
of some network file systems.
This fixes data losses during vnode recycling (and other code paths
using invalbuf(*,V_SAVE,*,*)) for data written using an mmaped file.
Collaborative effort by: jhb@,mohans@,peter@,ps@,ups@
Reviewed by: tegge@
MFC after: 7 days
client into the kernel by default, and many users won't use NFS,
don't start an extra 4 kernel threads that are unused. Once NFS
becomes active, it will start nfsiod's as it needs them.
We might consider mandating a minimum iod's equal to the number of
active NFS mounts (truncated to some value), which would force some
to remain available without having to create a new one if the file
system is mostly inactive.
PR: 70880
MFC after: 2 weeks
Prodded by: cel
Head nod: peter
Pointed out by: Joe <fbsd_user at a1poweruser dot com>
mimicing the NFS reference implementation.
NFS over TCP does not need fast retransmit timeouts, since network loss
and congestion are managed by the transport (TCP), unlike with NFS over
UDP. A long timeout prevents the unnecessary retransmission of non-
idempotent NFS requests.
Reviewed by: mohans, silby, rees?
Sponsored by: Network Appliance, Incorporated
the estimator to be more easily tuned and maintained.
There should be no functional change except there is now a lower limit
on the retransmit timeout to prevent the client from retransmitting
faster than the server's disks can fill requests, and an upper limit
to prevent the estimator from taking to long to retransmit during a
server outage.
Reviewed by: mohan, kris, silby
Sponsored by: Network Appliance, Incorporated
vnode after vflush() has succeeded. This would cause a dangling vnode
panic at unmount time otherwise. Other filesystems may have this problem
via their VFS_VGET() routines.
Found by: kris
Sponsored by: Isilon Systems, Inc.
The bug was that earlier, if a request was retransmitted,
we would do subsequent retransmits every 10 msecs.
This can cause data corruption under moderate loads by reordering
operations as seen by the client NFS attribute cache, and on the
server side when the retransmission occurs after the original request
has left the duplicate cache, since the operation will be committed
for a second time.
Further work on retransmission handling is needed (e.g. they are still
being done sent too often since they are scaled by HZ, and the size of
the dup cache is too small and easily overwhelmed on busy servers).
Submitted by: mohans
request, the FreeBSD NFS client will quickly back off to a excessively
long wait (days, then weeks) before retrying the request.
Change the behavior of the FreeBSD NFS client to match the behavior of
the reference NFS client implementation (Solaris). This provides a fixed
delay of 10 seconds between each retry by default. A sysctl, called
nfs3_jukebox_delay, is now available to tune the delay. Unlike Solaris,
the sysctl value on FreeBSD is in seconds, rather than in HZ.
Sponsored by: Network Appliance, Incorporated
Reviewed by: rick
Approved by: silby
MFC after: 3 days
The client's READDIRPLUS logic skips the attributes and
filehandle of the ".." entry. If the server doesn't send
attributes but does send a filehandle for "..", the
client's logic doesn't account for the extra "value
follows" field that indicates whether the filehandle is
present, causing the remaining entries in the reply
to be ignored.
Sponsored by: Network Appliance, Inc.
Reviewed by: rick, mohans
Approved by: silby
MFC after: 2 weeks
last few days. I tracked it down to the fact that nfs_reclaim()
is setting vp->v_data to NULL _before_ calling vnode_destroy_object().
After silence from the mailing list I checked further and discovered
that ufs_reclaim() is unique among FreeBSD filesystems for calling
vnode_destroy_object() early, long before tossing v_data or much
of anything else, for that matter. The rest, including NFS, appear
to be identical, as if they were just clones of one original routine.
The enclosed patch fixes all file systems in essentially the same
way, by moving the call to vnode_destroy_object() to early in the
routine (before the call to vfs_hash_remove(), if any). I have
only tested NFS, but I've now run for over eighteen hours with the
patch where I wouldn't get past four or five without it.
Submitted by: Frank Mayhar
Requested by: Mohan Srinivasan
MFC After: 1 week
event of an error, does the right thing, in terms of setting
the error flags in the buf header. That fixes a crash from
bstrategy().
- Treat ETIMEDOUT as a "recoverable" error, causing the buffer
to be re-dirtied. ETIMEDOUT can occur on soft mounts, when
the number of retries are exceeded, and we don't want data loss
in that case.
Submitted by: Mohan Srinivasan
buffers *and* there are no buffers queued up for writing. The bug
was that NMODIFIED was being cleared even while there were buffers
scheduled to be written out, which leads to all sorts of interesting
bugs - one where the file could shrink (because of a post-op getattr
load, say) causing data in buffer(s) queued for write to be tossed,
resulting in data corruption.
Submitted by: Mohan Srinivasan
- Prefer '_' to ' ', as it results in more easily parsed results in
memory monitoring tools such as vmstat.
- Remove punctuation that is incompatible with using memory type names
as file names, such as '/' characters.
- Disambiguate some collisions by adding subsystem prefixes to some
memory types.
- Generally prefer lower case to upper case.
- If the same type is defined in multiple architecture directories,
attempt to use the same name in additional cases.
Not all instances were caught in this change, so more work is required to
finish this conversion. Similar changes are required for UMA zone names.
It allows to specify options for NFS root file system.
Currently supported options are: soft, intr, conn, lockd.
I'm adding this functionality mostly for 'lockd' option, which is only
honored when performing the initial mount and will be silently ignored
if used while updating the mount options.
This will allow to use flock(2) without the need of using varmfs or
rpc.lockd and friends.
Example of use:
boot.nfsroot.options="intr,lockd"
MFC after: 2 weeks
as they both interact with the tty code (!MPSAFE) and may sleep if the
tty buffer is full (per comment).
Modify all consumers of uprintf() and tprintf() to hold Giant around
calls into these functions. In most cases, this means adding an
acquisition of Giant immediately around the function. In some cases
(nfs_timer()), it means acquiring Giant higher up in the callout.
With these changes, UFS no longer panics on SMP when either blocks are
exhausted or inodes are exhausted under load due to races in the tty
code when running without Giant.
NB: Some reduction in calls to uprintf() in the svr4 code is probably
desirable.
NB: In the case of nfs_timer(), calling uprintf() while holding a mutex,
or even in a callout at all, is a bad idea, and will generate warnings
and potential upset. This needs to be fixed, but was a problem before
this change.
NB: uprintf()/tprintf() sleeping is generally a bad ideas, as is having
non-MPSAFE tty code.
MFC after: 1 week
writers that want to extend the file. It was also used to serialize
readers that might want to read the last block of the file (with a
writer extending the file). Now that we support vnode locking for
NFS, the rslock is unnecessary. Writers grab the exclusive vnode
lock before writing and readers grab the shared (or in some cases
the exclusive) lock.
Submitted by: Mohan Srinivasan
- Fix nfsm_disct() so that after pulling up data, the remaining data
is aligned if necessary.
- Fix nfs_clnt_tcp_soupcall() to bcopy() the rpc length out of the
mbuf (instead of casting m_data to a uint32).
Submitted by: Pyun YongHyeon
Reviewed by: Mohan Srinivasan
pending discussion of how implementation would proceed. Applications
like -lc_r expect select(3) to match the EAGAIN-status of IO
functions.
Approved by: re
atomic write request, it can fill the buffer cache with the entirety
of that write in order to handle retries. However, it never drops
the vnode lock, or else it wouldn't be atomic, so it ends up waiting
indefinitely for more buf memory that cannot be gotten as it has it
all, and it waits in an uncancellable state.
To fix this, hibufspace is exported and scaled to a reasonable
fraction. This is used as the limit of how much of an atomic write
request by the NFS client will be handled asynchronously. If the
request is larger than this, it will be turned into a synchronous
request which won't deadlock the system. It's possible this value is
far off from what is required by some, so it shall be tunable as soon
as mount_nfs(8) learns of the new field.
The slowdown between an asynchronous and a synchronous write on NFS
appears to be on the order of 2x-4x.
General nod by: gad
MFC after: 2 weeks
More testing: wes
PR: kern/79208
re-sent instead of timing out.
don't log an error message on reconnection, which is not an error.
remove unused nfs_mrep_before_tsleep.
Reviewed by: Mohan Srinivasan
Approved by: alfred
as they have no connection with the expected MNT_* flags. This bug
was exposed 18 months ago when the assignments to f_flags in
vfs_syscalls.c were moved to before the VFS_STATFS() call. It was
fixed in the CSRG source 10 years ago, but we never picked up that
change.
PR: kern/80390
MFC after: 1 week
the MNT_RDONLY flag if the "ro" option was passed in from userland, and
clears it otherwise. In the diskless case, the MNT_RDONLY flag is already
set when this code is reached, but there are no mount options, so it was
incorrectly cleared. Change the logic so the MNT_RDONLY flag is set if the
"ro" option was specified, and left alone otherwise.
Note that the NFS code will still happily let you mount a filesystem RW
even if the server exports it RO. I'm not sure how to fix that.
- Network filesystems are written with a special idiom that checks the
cache first, and may even unlock dvp before discovering that a network
round-trip is required to resolve the name. I believe dvp is prevented
from being recycled even in the forced unmount case by the shared lock
on the mount point. If not, this code should grow checks for VI_DOOMED
after it relocks dvp or it will access NULL v_data fields.
Sponsored by: Isilon Systems, Inc.
these filesystems will support shared locks until they are explicitly
modified to do so. Careful review must be done to ensure that this
is safe for each individual filesystem.
Sponsored by: Isilon Systems, Inc.
non-maskable).
- The NFS client needs to guard against spurious wakeups
while waiting for the response. ltrace causes the process
under question to wakeup (possibly from ptrace()), which
causes NFS to wakeup from tsleep without the response being
delivered.
Submitted by: Mohan Srinivasan
that NFS ever started using it. Long time ago I added the necessary
vhold()/vdrop() calls to replace it, but forgot to remove the v_id code.
Do it now.
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.
