Pre-rfork code assumed inherent locking of a process's file descriptor
array. However, with the advent of rfork() the file descriptor table
could be shared between processes. This patch closes over a dozen
serious race conditions related to one thread manipulating the table
(e.g. closing or dup()ing a descriptor) while another is blocked in
an open(), close(), fcntl(), read(), write(), etc...
PR: kern/11629
Discussed with: Alexander Viro <viro@math.psu.edu>
mail:
The problem seems to originate with NFS's postop_attr
information that is returned with a read or write RPC.
Within a vm_fault context, the code cannot deal with
vnode_pager_setsize() shrinking a vnode.
The workaround in the patch below stops the nfsm_postop_attr()
macro from ever shrinking a vnode. If the new size in the
postop_attr information is smaller, then it just sets the
nfsnode n_attrstamp to 0 to stop the wrong size getting
used in the future. This change only affects postop_attr
attributes; the nfsm_loadattr() macro works as normal.
The change is implemented by adding a new argument to
nfs_loadattrcache() called 'dontshrink'. When this is
non-zero, nfs_loadattrcache() will never reduce the
vnode/nfsnode size; instead it zeros n_attrstamp.
There remain other was processes can get stuck in vmopar.
Submitted by: Ian Dowse <iedowse@maths.tcd.ie>
Reviewed by: dillon
Tested by: Vadim Belman <voland@lflat.org>
Add lockdestroy() and appropriate invocations, which corresponds to
lockinit() and must be called to clean up after a lockmgr lock is no
longer needed.
separately (nfs, cd9660 etc) or keept as a first element of structure
referenced by v_data pointer(ffs). Such organization leads to known problems
with stacked filesystems.
From this point vop_no*lock*() functions maintain only interlock lock.
vop_std*lock*() functions maintain built-in v_lock structure using lockmgr().
vop_sharedlock() is compatible with vop_stdunlock(), but maintains a shared
lock on vnode.
If filesystem wishes to export lockmgr compatible lock, it can put an address
of this lock to v_vnlock field. This indicates that the upper filesystem
can take advantage of it and use single lock structure for entire (or part)
of stack of vnodes. This field shouldn't be examined or modified by VFS code
except for initialization purposes.
Reviewed in general by: mckusick
include:
* Mutual exclusion is used instead of spl*(). See mutex(9). (Note: The
alpha port is still in transition and currently uses both.)
* Per-CPU idle processes.
* Interrupts are run in their own separate kernel threads and can be
preempted (i386 only).
Partially contributed by: BSDi (BSD/OS)
Submissions by (at least): cp, dfr, dillon, grog, jake, jhb, sheldonh
with the new snapshot code.
Update addaliasu to correctly implement the semantics of the old
checkalias function. When a device vnode first comes into existence,
check to see if an anonymous vnode for the same device was created
at boot time by bdevvp(). If so, adopt the bdevvp vnode rather than
creating a new vnode for the device. This corrects a problem which
caused the kernel to panic when taking a snapshot of the root
filesystem.
Change the calling convention of vn_write_suspend_wait() to be the
same as vn_start_write().
Split out softdep_flushworklist() from softdep_flushfiles() so that
it can be used to clear the work queue when suspending filesystem
operations.
Access to buffers becomes recursive so that snapshots can recursively
traverse their indirect blocks using ffs_copyonwrite() when checking
for the need for copy on write when flushing one of their own indirect
blocks. This eliminates a deadlock between the syncer daemon and a
process taking a snapshot.
Ensure that softdep_process_worklist() can never block because of a
snapshot being taken. This eliminates a problem with buffer starvation.
Cleanup change in ffs_sync() which did not synchronously wait when
MNT_WAIT was specified. The result was an unclean filesystem panic
when doing forcible unmount with heavy filesystem I/O in progress.
Return a zero'ed block when reading a block that was not in use at
the time that a snapshot was taken. Normally, these blocks should
never be read. However, the readahead code will occationally read
them which can cause unexpected behavior.
Clean up the debugging code that ensures that no blocks be written
on a filesystem while it is suspended. Snapshots must explicitly
label the blocks that they are writing during the suspension so that
they do not cause a `write on suspended filesystem' panic.
Reorganize ffs_copyonwrite() to eliminate a deadlock and also to
prevent a race condition that would permit the same block to be
copied twice. This change eliminates an unexpected soft updates
inconsistency in fsck caused by the double allocation.
Use bqrelse rather than brelse for buffers that will be needed
soon again by the snapshot code. This improves snapshot performance.
the gating of system calls that cause modifications to the underlying
filesystem. The gating can be enabled by any filesystem that needs
to consistently suspend operations by adding the vop_stdgetwritemount
to their set of vnops. Once gating is enabled, the function
vfs_write_suspend stops all new write operations to a filesystem,
allows any filesystem modifying system calls already in progress
to complete, then sync's the filesystem to disk and returns. The
function vfs_write_resume allows the suspended write operations to
begin again. Gating is not added by default for all filesystems as
for SMP systems it adds two extra locks to such critical kernel
paths as the write system call. Thus, gating should only be added
as needed.
Details on the use and current status of snapshots in FFS can be
found in /sys/ufs/ffs/README.snapshot so for brevity and timelyness
is not included here. Unless and until you create a snapshot file,
these changes should have no effect on your system (famous last words).
<sys/bio.h>.
<sys/bio.h> is now a prerequisite for <sys/buf.h> but it shall
not be made a nested include according to bdes teachings on the
subject of nested includes.
Diskdrivers and similar stuff below specfs::strategy() should no
longer need to include <sys/buf.> unless they need caching of data.
Still a few bogus uses of struct buf to track down.
Repocopy by: peter
Exceptions:
Vinum untouched. This means that it cannot be compiled.
Greg Lehey is on the case.
CCD not converted yet, casts to struct buf (still safe)
atapi-cd casts to struct buf to examine B_PHYS
(Much of this done by script)
Move B_ORDERED flag to b_ioflags and call it BIO_ORDERED.
Move b_pblkno and b_iodone_chain to struct bio while we transition, they
will be obsoleted once bio structs chain/stack.
Add bio_queue field for struct bio aware disksort.
Address a lot of stylistic issues brought up by bde.
reserve, in maximal NFS packets. Originally only 2 packets worth of
space was reserved. The default is now 4, which appears to greatly
improve performance for slow to mid-speed machines on gigabit networks.
Add documentation and correct some prior documentation.
Problem Researched by: Andrew Gallatin <gallatin@cs.duke.edu>
Approved by: jkh
substitute BUF_WRITE(foo) for VOP_BWRITE(foo->b_vp, foo)
substitute BUF_STRATEGY(foo) for VOP_STRATEGY(foo->b_vp, foo)
This patch is machine generated except for the ccd.c and buf.h parts.
field in struct buf: b_iocmd. The b_iocmd is enforced to have
exactly one bit set.
B_WRITE was bogusly defined as zero giving rise to obvious coding
mistakes.
Also eliminate the redundant struct buf flag B_CALL, it can just
as efficiently be done by comparing b_iodone to NULL.
Should you get a panic or drop into the debugger, complaining about
"b_iocmd", don't continue. It is likely to write on your disk
where it should have been reading.
This change is a step in the direction towards a stackable BIO capability.
A lot of this patch were machine generated (Thanks to style(9) compliance!)
Vinum users: Greg has not had time to test this yet, be careful.
struct contains a major union for which lph_slp was being initialized
only for TCP connections, but accessed for all types of connections
leading to a crash. Also, a conditional controlling an nfs_slplock()
call contained an improper paren grouping, causing a second crash in
the UDP case.
The nqhost structure has been reorganized and lph_slp has been made a
normal structural field rather then a union field, and properly
initialized for all connection types.
Approved by: jkh
there is nothing we can do about it. In fact, after further review
there simply are not very many instances of the two structures NFS
checks for 'bloat' so I've decided to simply rip the checks out entirely.
Submitted by: Andrew Gallatin <gallatin@cs.duke.edu>
into vnode dirtyblkhd we append it to the list instead of prepend it to
the list in order to maintain a 'forward' locality of reference, which
is arguably better then 'reverse'. The original algorithm did things this
way to but at a huge time cost.
Enhance the append interlock for NFS writes to handle intr/soft mounts
better.
Fix the hysteresis for NFS async daemon I/O requests to reduce the
number of unnecessary context switches.
Modify handling of NFS mount options. Any given user option that is
too high now defaults to the kernel maximum for that option rather then
the kernel default for that option.
Reviewed by: Alfred Perlstein <bright@wintelcom.net>
was calling nfs_flush() and then clearing the NMODIFIED bit. This is
not legal since there might still be dirty buffers after the nfs_flush
(for example, pending commits). The clearing of this bit in turn prevented
a necessary vinvalbuf() from occuring leaving left over dirty buffers
even after truncating the file in a new operation. The fix is to
simply not clear NMODIFIED.
Also added a sysctl vfs.nfs.nfsv3_commit_on_close which, if set to 1,
will cause close() to do a stage 1 write AND a stage 2 commit
synchronously. By default only the stage 1 write is done synchronously.
Reviewed by: Alfred Perlstein <bright@wintelcom.net>
is an application space macro and the applications are supposed to be free
to use it as they please (but cannot). This is consistant with the other
BSD's who made this change quite some time ago. More commits to come.
NFSSERVER defined, useful for userland fileservers that want to
use a filehandle type interface to the filesystem.
Submitted by: Assar Westerlund assar@stacken.kth.se
PR: kern/15452
generate the NFSv3 Version id. boottime itself may change, sometimes
once every tick if you are running xntpd, which really throws off
clients. Clients will tend to throw away what they believe to be
stale data too often, and can get into long loops rewriting the same
data over and over again because they believe the server has rebooted
over and over again due to the changing version id.
Approved by: jkh
occur due to np->n_size potentially changing if nfs_getcacheblk()
blocks in nfs_write().
Second, under -current we must supply the proper bufsize when obtaining
buffers that straddle the EOF, but due to the fact that np->n_size can
change out from under us it is possible that we may specify the wrong
buffer size and wind up truncating dirty data written by another
process.
Both problems are solved by implementing nfs_rslock(), which allows us
to lock around sensitive buffer cache operations such as those that
occur when appending to a file.
It is believed that this race is responsible for causing dirtyoff/dirtyend
and (in stable) validoff/validend to exceed the buffer size. Therefore
we have now added a warning printf for the dirtyoff/end case in current.
However, we have introduced a new problem which we need to fix at some
point, and that is that soft or intr NFS mounts may become
uninterruptable from the point of view of process A which is stuck waiting
on rslock while process B is stuck doing the rpc. To unstick process A,
process B would have to be interrupted first.
Reviewed by: Alfred Perlstein <bright@wintelcom.net>
cannot unilaterally pass data to a client it can reduce the physical
disk transaction overhead by reading larger blocks. This results in
better pipelining of requests/responses over the network and an almost
100% increase in cpu efficiency on the server. On a 100BaseTX network
NFS read performance increases from 8.5 MBytes/sec to 10 MB/sec (maxed
out), and cpu efficiency increases from 72% idle to 80% idle on the server.
Reviewed by: Alfred Perlstein <bright@wintelcom.net>
NFS packets, mainly initializing structure pointers to NULL which
are conditionally freed prior to return.
PR: kern/15249
Submitted by: Ian Dowse <iedowse@maths.tcd.ie>
blocks of zeros could wind up in a file written to over NFS by a client.
The problem only occurs a few times per several gigabytes of data. This
problem turned out to be bug #3 below.
bug #1:
B_CLUSTEROK must be cleared when an NFS buffer is reverted from
stage 2 (ready for commit rpc) to stage 1 (ready for write).
Reversions can occur when a dirty NFS buffer is redirtied with new
data.
Otherwise the VFS/BIO system may end up thinking that a stage 1
NFS buffer is clusterable. Stage 1 NFS buffers are not clusterable.
bug #2:
B_CLUSTEROK was inappropriately set for a 'short' NFS buffer (short
buffers only occur near the EOF of the file). Change to only set
when the buffer is a full biosize (usually 8K). This bug has no
effect but should be fixed in -current anyway. It need not be
backported.
bug #3:
B_NEEDCOMMIT was inappropriately set in nfs_flush() (which is
typically only called by the update daemon). nfs_flush()
does a multi-pass loop but due to the lack of vnode locking it
is possible for new buffers to be added to the dirtyblkhd list
while a flush operation is going on. This may result in nfs_flush()
setting B_NEEDCOMMIT on a buffer which has *NOT* yet gone through its
stage 1 write, causing only the commit rpc to be made and thus
causing the contents of the buffer to be thrown away (never sent to
the server).
The patch also contains some cleanup, which only applies to the commit
into -current.
Reviewed by: dg, julian
Originally Reported by: Dan Nelson <dnelson@emsphone.com>
* lockstatus() and VOP_ISLOCKED() gets a new process argument and a new
return value: LK_EXCLOTHER, when the lock is held exclusively by another
process.
* The ASSERT_VOP_(UN)LOCKED family is extended to use what this gives them
* Extend the vnode_if.src format to allow more exact specification than
locked/unlocked.
This commit should not do any semantic changes unless you are using
DEBUG_VFS_LOCKS.
Discussed with: grog, mch, peter, phk
Reviewed by: peter
a 0 error code. The problem occured with NFSv2 mounts and also with
any NFSv3 mount returning an EEXIST error (which is translated to 0
prior to return). The reply to the rpc only contains the file handle
for the no-error case under NFSv3. The error case under NFSv3 and
all cases under NFSv2 do *not* return the file handle. The fix is
to do a secondary lookup to obtain the file handle and thus be able
to generate a return vnode for the situations where the rpc reply
does not contain the required information.
The bug was originally introduced when VOP_SYMLINK semantics were
changed for -CURRENT. The NFS symlink implementation was not properly
modified to go along with the change despite the fact that three
people reviewed the code. It took four attempts to get the current
fix correct with five people. Is NFS obfuscated? Ha!
Reviewed by: Alfred Perlstein <bright@wintelcom.net>
Testing and Discussion: "Viren R.Shah" <viren@rstcorp.com>, Eivind Eklund <eivind@FreeBSD.ORG>, Ian Dowse <iedowse@maths.tcd.ie>
of element [4] in both, which goes beyond the end of the array, leaving
[0], [1], [2], and [3]. This bug did not cause any problems since
the overrun fields are initialized after the bogus array init but
needs to be fixed anyway.
Submitted by: Ian Dowse <iedowse@maths.tcd.ie>
Note: Previous commit to these files (except coda_vnops and devfs_vnops)
that claimed to remove WILLRELE from VOP_RENAME actually removed it from
VOP_MKNOD.
when returning an error. Bug fix was extracted from the PR. The PR
is not yet entirely resolved by this commit.
PR: kern/13049
Reviewed by: Matt Dillon <dillon@freebsd.org>
Submitted by: Ian Dowse <iedowse@maths.tcd.ie>
Merge the contents (less some trivial bordering the silly comments)
of <vm/vm_prot.h> and <vm/vm_inherit.h> into <vm/vm.h>. This puts
the #defines for the vm_inherit_t and vm_prot_t types next to their
typedefs.
This paves the road for the commit to follow shortly: change
useracc() to use VM_PROT_{READ|WRITE} rather than B_{READ|WRITE}
as argument.
error for a directory. I have made this change after a great deal of
review although I cannot be absolutely sure that this meets the spec.
The issue devolves into whether changes in an underlying (UFS) directory
can cause NFS directory blocks to be renumbered. My read of the code
indicates that NFS directory blocks will not be renumbered, which means
that the cookies should still remain valid after a change is made to
the underlying directory. This being the case, a cookie error should
not be returned when a change is made to the underlying directory and,
instead, the NFS client should rely on mtime detection to invalidate and
reload the directory.
The use of mtime is problematic in of itself, due to insufficient
resolution, which is why I believe the original conservative error
handling was done. Still, there have been dozens of bug reports by
people needing solaris<->FreeBSD interoperability and these have to
be accomodated.
-----------------------------
The core of the signalling code has been rewritten to operate
on the new sigset_t. No methodological changes have been made.
Most references to a sigset_t object are through macros (see
signalvar.h) to create a level of abstraction and to provide
a basis for further improvements.
The NSIG constant has not been changed to reflect the maximum
number of signals possible. The reason is that it breaks
programs (especially shells) which assume that all signals
have a non-null name in sys_signame. See src/bin/sh/trap.c
for an example. Instead _SIG_MAXSIG has been introduced to
hold the maximum signal possible with the new sigset_t.
struct sigprop has been moved from signalvar.h to kern_sig.c
because a) it is only used there, and b) access must be done
though function sigprop(). The latter because the table doesn't
holds properties for all signals, but only for the first NSIG
signals.
signal.h has been reorganized to make reading easier and to
add the new and/or modified structures. The "old" structures
are moved to signalvar.h to prevent namespace polution.
Especially the coda filesystem suffers from the change, because
it contained lines like (p->p_sigmask == SIGIO), which is easy
to do for integral types, but not for compound types.
NOTE: kdump (and port linux_kdump) must be recompiled.
Thanks to Garrett Wollman and Daniel Eischen for pressing the
importance of changing sigreturn as well.
previously issued synchronously even if async daemons (nfsiod's) were
available. The commit has been moved from the strategy code to the doio
code in order to asynchronize it.
Removed use of lastr in preparation for removal of vnode->v_lastr. It
has been replaced with seqcount, which is already supported by the system
and, in fact, gives us a better heuristic for sequential detection then
lastr ever did.
Made major performance improvements to the server side commit. The
server previously fsync'd the entire file for each commit rpc. The
server now bawrite()s only those buffers related to the offset/size
specified in the commit rpc.
Note that we do not commit the meta-data yet. This works still needs
to be done.
Note that a further optimization can be done (and has not yet been done)
on the client: we can merge multiple potential commit rpc's into a
single rpc with a greater file offset/size range and greatly reduce
rpc traffic.
Reviewed by: Alan Cox <alc@cs.rice.edu>, David Greenman <dg@root.com>
reasonable defaults.
This avoids confusing and ugly casting to eopnotsupp or making dummy functions.
Bogus casting of filesystem sysctls to eopnotsupp() have been removed.
This should make *_vfsops.c more readable and reduce bloat.
Reviewed by: msmith, eivind
Approved by: phk
Tested by: Jeroen Ruigrok/Asmodai <asmodai@wxs.nl>
Make the alias list a SLIST.
Drop the "fast recycling" optimization of vnodes (including
the returning of a prexisting but stale vnode from checkalias).
It doesn't buy us anything now that we don't hardlimit
vnodes anymore.
Rename checkalias2() and checkalias() to addalias() and
addaliasu() - which takes dev_t and udev_t arg respectively.
Make the revoke syscalls use vcount() instead of VALIASED.
Remove VALIASED flag, we don't need it now and it is faster
to traverse the much shorter lists than to maintain the
flag.
vfs_mountedon() can check the dev_t directly, all the vnodes
point to the same one.
Print the devicename in specfs/vprint().
Remove a couple of stale LFS vnode flags.
Remove unimplemented/unused LK_DRAINED;
works correctly in if/else etc. egcs had probably picked up most of the
problems here before with "ambiguous braces" etc, but this should
increase the robustness a bit. Based on an idea from Eivind Eklund.
correct the pointers afterwards.
It's kinda bogus that we generate a 24 (?) byte filehandle (2 x int32
fsid and 16 byte VFS fhandle) and pad it out to 64 bytes for NFSv3 with
garbage. The whole point of NFSv3's variable filehandle length was
to allow for shorter handles, both in memory and over the wire. I plan
on taking a shot at fixing this shortly.
I did some tcpdumping the other day and noticed that GETATTR calls
were frequently followed by an ACCESS call to the same file. The
attached patch changes nfs_getattr to fill the access cache as a side
effect. This is accomplished by calling ACCESS rather than
GETATTR. This implies a modest overhead of 4 bytes in the request and
8 bytes in the response compared to doing a vanilla GETATTR.
...
[The patch comprises two parts] The first
is the "real" patch, the second counts misses and hits rather than
fills and hits. The difference is subtle but important because both
nfs_getattr and nfs_access now fill the cache. It also changes the
default value of nfsaccess_cache_timeout to better match the attribute
cache. IMHO, file timestamps change much more frequently than
protection bits.
Submitted by: Bjoern Groenvall <bg@sics.se>
Reviewed by: dillon (partially)
vnodes are locked and never unlocked, which leads to processes starting
to wedge up after doing a mount -o nfsv3,tcp,rdirplus foo:/fs /fs; ls /fs.
The second is that sometimes cnp is accessed without having been
properly initialized: cnp->cn_nameptr points to an earlier name while
"len" contains the length of a current name of different size. This
leads to an attempt to dereference *(cn->cn_nameptr + len) which will
sometimes cause a page fault and a panic.
With these two fixes, client side readdirplus works correctly with
FreeBSD, IRIX 6.5.4 and Solaris 2.5.1 and 2.6 servers.
Submitted by: Matthew Dillon <dillon@backplane.com>
being incremented by 4 bytes too few each time through the loop, which
allows more data into the mbuf chain that we really want. In the worst
case, when we're using 32K read/write sizes with a TCP client, this causes
readdirplus replies to sometimes exceed NFS_MAXPACKET which leads to a
panic. This problem cropped up for me using an IRIX 6.5.4 NFSv3 TCP client
with 32K read/write sizes, however supposedly it can be triggered by
WinNT NFS servers too. In theory, it can probably be triggered by any
NFS v3 implementation using TCP as long as it's using the maxiumum block
size.
Reviewed by: Matthew Dillon <dillon@backplane.com>
Reviewed by: Matthew Dillon <dillon@apollo.backplane.com>
The following ugly hack to the exit path of nfs_readlinkrpc() circumvents
an Auspex bug: for symlinks longer than 112 (0x70) they return a 1024 byte
xdr string - the correct data with many nulls appended. Without this fix
namei returns ENAMETOOLONG, at least it does on our source base and on
FreeBSD 3.0. Note we do not (and should not) rely upon their null padding.
lockmgr locks. This commit should be functionally equivalent to the old
semantics. That is, all buffer locking is done with LK_EXCLUSIVE
requests. Changes to take advantage of LK_SHARED and LK_RECURSIVE will
be done in future commits.
mount point for seeing whether or not the new nfsnode is already in the
hash queue. We're pretty much guaranteed that the old nfsnode is already
in the hash queue. Wank! Infinite Loop! Looks like just a minor typo....
(ah the influence of fortran ... np && np2... why not nfsnode_the_first &&
nfsnode_the_second???)...
make sure we've freed any allocated resources (to avoid a memory leak)
and and do the right thing with respect to the nfs node hash lock we'd
acquired.
txdr_hyper and fxdr_hyper tweaks to avoid excessive CPU order knowledge.
nfs_serv.c: don't call nfsm_adj() with negative values, windows clients
could crash servers when doing a readdir of a large directory.
nfs_socket.c: Use IP_PORTRANGE to get a priviliged port without a spin
loop trying to bind(). Don't clobber a mbuf pointer or we get panics
on a NFS3ERR_JUKEBOX error from a server when reusing a freed mbuf.
nfs_subs.c: Don't loose st_blocks on NFSv2 mounts when > 2GB.
Obtained from: OpenBSD
