Hi.
I have a proposal for possibly resolving this issue.
I believe that this situation occurs due to the way that the
Linux NFS client handles writes which modify partial pages.
The Linux NFS client handles partial page modifications by
allocating a page from the page cache, copying the data from
the user level into the page, and then keeping track of the
offset and length of the modified portions of the page. The
page is not marked as up to date because there are portions
of the page which do not contain valid file contents.
When a read call comes in for a portion of the page, the
contents of the page must be read in the from the server.
However, since the page may already contain some modified
data, that modified data must be written to the server
before the file contents can be read back in the from server.
And, since the writing and reading can not be done atomically,
the data must be written and committed to stable storage on
the server for safety purposes. This means either a
FILE_SYNC WRITE or a UNSTABLE WRITE followed by a COMMIT.
This has been discussed at length previously.
This algorithm could be described as modify-write-read. It
is most efficient when the application only updates pages
and does not read them.
My proposed solution is to add a heuristic to decide whether
to do this modify-write-read algorithm or switch to a read-
modify-write algorithm when initially allocating the page
in the write system call path. The heuristic uses the modes
that the file was opened with, the offset in the page to
read from, and the size of the region to read.
If the file was opened for reading in addition to writing
and the page would not be filled completely with data from
the user level, then read in the old contents of the page
and mark it as Uptodate before copying in the new data. If
the page would be completely filled with data from the user
level, then there would be no reason to read in the old
contents because they would just be copied over.
This would optimize for applications which randomly access
and update portions of files. The linkage editor for the
C compiler is an example of such a thing.
I tested the attached patch by using rpmbuild to build the
current Fedora rawhide kernel. The kernel without the
patch generated about 269,500 WRITE requests. The modified
kernel containing the patch generated about 261,000 WRITE
requests. Thus, about 8,500 fewer WRITE requests were
generated. I suspect that many of these additional
WRITE requests were probably FILE_SYNC requests to WRITE
a single page, but I didn't test this theory.
The difference between this patch and the previous one was
to remove the unneeded PageDirty() test. I then retested to
ensure that the resulting system continued to behave as
desired.
Thanx...
ps
Signed-off-by: Peter Staubach <staubach@redhat.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
* Remove smp_lock.h from files which don't need it (including some headers!)
* Add smp_lock.h to files which do need it
* Make smp_lock.h include conditional in hardirq.h
It's needed only for one kernel_locked() usage which is under CONFIG_PREEMPT
This will make hardirq.h inclusion cheaper for every PREEMPT=n config
(which includes allmodconfig/allyesconfig, BTW)
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Adds support for splice writes. It effectively calls
generic_file_splice_write() to do the writes.
We need not worry about O_APPEND case as the combination of splice()
writes and O_APPEND is disallowed. This patch propagates NFS write
errors back to the caller. The number of bytes written via splice are
being added to NFSIO_NORMALWRITTENBYTES as these are effectively
cached writes.
Signed-off-by: Suresh Jayaraman <sjayaraman@suse.de>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Follow up to Nick Piggin's patches to ensure that nfs_vm_page_mkwrite
returns with the page lock held, and sets the VM_FAULT_LOCKED flag.
See http://bugzilla.kernel.org/show_bug.cgi?id=12913
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit c2ec175c39 ("mm: page_mkwrite
change prototype to match fault") exposed a bug in the NFS
implementation of page_mkwrite. We should be returning 0 on success...
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
FS-Cache page management for NFS. This includes hooking the releasing and
invalidation of pages marked with PG_fscache (aka PG_private_2) and waiting for
completion of the write-to-cache flag (PG_fscache_write aka PG_owner_priv_2).
Signed-off-by: David Howells <dhowells@redhat.com>
Acked-by: Steve Dickson <steved@redhat.com>
Acked-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Acked-by: Al Viro <viro@zeniv.linux.org.uk>
Tested-by: Daire Byrne <Daire.Byrne@framestore.com>
Add comment banners to some NFS functions so that they can be modified by the
NFS fscache patches for further information.
Signed-off-by: David Howells <dhowells@redhat.com>
Acked-by: Steve Dickson <steved@redhat.com>
Acked-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Acked-by: Al Viro <viro@zeniv.linux.org.uk>
Tested-by: Daire Byrne <Daire.Byrne@framestore.com>
Change the page_mkwrite prototype to take a struct vm_fault, and return
VM_FAULT_xxx flags. There should be no functional change.
This makes it possible to return much more detailed error information to
the VM (and also can provide more information eg. virtual_address to the
driver, which might be important in some special cases).
This is required for a subsequent fix. And will also make it easier to
merge page_mkwrite() with fault() in future.
Signed-off-by: Nick Piggin <npiggin@suse.de>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: Trond Myklebust <trond.myklebust@fys.uio.no>
Cc: Miklos Szeredi <miklos@szeredi.hu>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <joel.becker@oracle.com>
Cc: Artem Bityutskiy <dedekind@infradead.org>
Cc: Felix Blyakher <felixb@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Close-to-open cache consistency rules really only require us to flush out
writes on calls to close(), and require us to revalidate attributes on the
very last close of the file.
Currently we appear to be doing a lot of extra attribute revalidation
and cache flushes.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Bryan Wu reports that when compiling NFS on nommu machines he gets a
"defined but not used" error on nfs_file_mmap().
The easiest fix is simply to get rid of the special casing in NFS, and
just always call generic_file_mmap() to set up the file.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
The following patch is a combination of a patch by myself and Peter
Staubach.
Trond: If we allow other processes to dirty pages while a process is doing
a consistency sync to disk, we can end up never making progress.
Peter: Attached is a patch which addresses a continuing problem with
the NFS client generating out of order WRITE requests. While
this is compliant with all of the current protocol
specifications, there are servers in the market which can not
handle out of order WRITE requests very well. Also, this may
lead to sub-optimal block allocations in the underlying file
system on the server. This may cause the read throughputs to
be reduced when reading the file from the server.
Peter: There has been a lot of work recently done to address out of
order issues on a systemic level. However, the NFS client is
still susceptible to the problem. Out of order WRITE
requests can occur when pdflush is in the middle of writing
out pages while the process dirtying the pages calls
generic_file_buffered_write which calls
generic_perform_write which calls
balance_dirty_pages_rate_limited which ends up calling
writeback_inodes which ends up calling back into the NFS
client to writes out dirty pages for the same file that
pdflush happens to be working with.
Signed-off-by: Peter Staubach <staubach@redhat.com>
[modification by Trond to merge the two similar patches]
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
With the write_begin/write_end aops, page_symlink was broken because it
could no longer pass a GFP_NOFS type mask into the point where the
allocations happened. They are done in write_begin, which would always
assume that the filesystem can be entered from reclaim. This bug could
cause filesystem deadlocks.
The funny thing with having a gfp_t mask there is that it doesn't really
allow the caller to arbitrarily tinker with the context in which it can be
called. It couldn't ever be GFP_ATOMIC, for example, because it needs to
take the page lock. The only thing any callers care about is __GFP_FS
anyway, so turn that into a single flag.
Add a new flag for write_begin, AOP_FLAG_NOFS. Filesystems can now act on
this flag in their write_begin function. Change __grab_cache_page to
accept a nofs argument as well, to honour that flag (while we're there,
change the name to grab_cache_page_write_begin which is more instructive
and does away with random leading underscores).
This is really a more flexible way to go in the end anyway -- if a
filesystem happens to want any extra allocations aside from the pagecache
ones in ints write_begin function, it may now use GFP_KERNEL (rather than
GFP_NOFS) for common case allocations (eg. ocfs2_alloc_write_ctxt, for a
random example).
[kosaki.motohiro@jp.fujitsu.com: fix ubifs]
[kosaki.motohiro@jp.fujitsu.com: fix fuse]
Signed-off-by: Nick Piggin <npiggin@suse.de>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: <stable@kernel.org> [2.6.28.x]
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
[ Cleaned up the calling convention: just pass in the AOP flags
untouched to the grab_cache_page_write_begin() function. That
just simplifies everybody, and may even allow future expansion of the
logic. - Linus ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We *do* now allow bsd flocks over nfs.
Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
After the BKL removal patches were applied to the rest of the NFS code, the
BKL protection in nfs_file_llseek() is no longer sufficient to ensure that
inode->i_size is read safely in generic_file_llseek_unlocked().
In order to fix the situation, we either have to replace the naked read of
inode->i_size in generic_file_llseek_unlocked() with i_size_read(), or the
whole thing needs to be executed under the inode->i_lock;
In order to avoid disrupting other filesystems, avoid touching
generic_file_llseek_unlocked() for now...
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
All the NFSv4 stateful operations are already protected by other locks (in
particular by the rpc_sequence locks.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
All instances are set to nfs_open(), so we should just remove the redundant
indirection. Ditto for the file_release op
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Clean up: some fops use NFSDBG_FILE, some use NFSDBG_VFS. Let's use
NFSDBG_FILE for all fops, and consistently report file names instead
of inode numbers.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Recent work in fs/nfs/file.c neglected to add appropriate trace debugging
for the NFS client's address space operations.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Clean up: Report the same debugging info and count function calls the
same for files and directories in nfs_opendir() and nfs_file_open().
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Clean up: Report the same debugging info in nfs_llseek_dir() and
nfs_llseek_file().
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Clean up: Report the same debugging info, count function calls the same,
and use similar function naming in nfs_fsync_dir() and nfs_fsync().
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
If a file is being extended, and we're creating a hole, we might as well
declare the entire page to be up to date.
This patch significantly improves the write performance for sparse files
in the case where lseek(SEEK_END) is used to append several non-contiguous
writes at intervals of < PAGE_SIZE.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
NFSv2 file locking currently fails the Connectathon tests, because the
calls to the VFS locking code do not return an EINVAL error if the
struct file_lock overflows the 32-bit boundaries.
The problem is due to the fact that we occasionally call helpers from
fs/locks.c in order to avoid RPC calls to the server when we know that a
local process holds the lock. These helpers are, of course, always
64-bit enabled, so EINVAL is not returned in cases when it would if
the call had gone to the NLM code.
For consistency, we therefore add support for a bounds-checking helper.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
- Replace remote_llseek with generic_file_llseek_unlocked (to force compilation
failures in all users)
- Change all users to either use generic_file_llseek_unlocked directly or
take the BKL around. I changed the file systems who don't use the BKL
for anything (CIFS, GFS) to call it directly. NCPFS and SMBFS and NFS
take the BKL, but explicitely in their own source now.
I moved them all over in a single patch to avoid unbisectable sections.
Open problem: 32bit kernels can corrupt fpos because its modification
is not atomic, but they can do that anyways because there's other paths who
modify it without BKL.
Do we need a special lock for the pos/f_version = 0 checks?
Trond says the NFS BKL is likely not needed, but keep it for now
until his full audit.
v2: Use generic_file_llseek_unlocked instead of remote_llseek_unlocked
and factor duplicated code (suggested by hch)
Cc: Trond.Myklebust@netapp.com
Cc: swhiteho@redhat.com
Cc: sfrench@samba.org
Cc: vandrove@vc.cvut.cz
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
Both NLM and NFSv4 should be able to clean up adequately in the case where
the user interrupts the RPC call...
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Since O_DIRECT is a standard feature that is enabled in most distros,
eliminate the CONFIG_NFS_DIRECTIO build option, and change the
fs/nfs/Makefile to always build in the NFS direct I/O engine.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Clean up: commit 4899f9c8 added nfs_write_end(), which introduces a
conditional expression that returns an unsigned integer in one arm and
a signed integer in the other.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
The current model locks the page twice for no good reason. Optimise by
inlining the parts of nfs_write_begin()/nfs_write_end() that we care about.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Otherwise, we do end up breaking close-to-open semantics. We also end up
breaking some of the silly-rename tests in Connectathon on some setups.
Please refer to the bug-report at
http://bugzilla.linux-nfs.org/show_bug.cgi?id=150
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
The __mandatory_lock(inode) macro makes the same check, but makes the code
more readable.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Cc: Trond Myklebust <trond.myklebust@fys.uio.no>
Cc: "J. Bruce Fields" <bfields@fieldses.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
This helps prevent huge queues of background writes from building up
whenever the server runs out of disk or quota space, or if someone changes
the file access modes behind our backs.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
The addition of nfs_page_mkwrite means that We should no longer need to
create requests inside nfs_writepage()
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Ryusuke Konishi says:
The recent truncate_complete_page() clears the dirty flag from a page
before calling a_ops->invalidatepage(),
^^^^^^
static void
truncate_complete_page(struct address_space *mapping, struct page *page)
{
...
cancel_dirty_page(page, PAGE_CACHE_SIZE); <--- Inserted here at
kernel 2.6.20
if (PagePrivate(page))
do_invalidatepage(page, 0); ---> will call
a_ops->invalidatepage()
...
}
and this is disturbing nfs_wb_page_priority() from calling
nfs_writepage_locked() that is expected to handle the pending
request (=nfs_page) associated with the page.
int nfs_wb_page_priority(struct inode *inode, struct page *page, int how)
{
...
if (clear_page_dirty_for_io(page)) {
ret = nfs_writepage_locked(page, &wbc);
if (ret < 0)
goto out;
}
...
}
Since truncate_complete_page() will get rid of the page after
a_ops->invalidatepage() returns, the request (=nfs_page) associated
with the page becomes a garbage in nfs_inode->nfs_page_tree.
------------------------
Fix this by ensuring that nfs_wb_page_priority() recognises that it may
also need to clear out non-dirty pages that have an nfs_page associated
with them.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Since posix_test_lock(), like fcntl() and ->lock(), indicates absence or
presence of a conflict lock by setting fl_type to, respectively, F_UNLCK
or something other than F_UNLCK, the return value is no longer needed.
Signed-off-by: "J. Bruce Fields" <bfields@citi.umich.edu>
As Peter Staubach says elsewhere
(http://marc.info/?l=linux-kernel&m=118113649526444&w=2):
> The problem is that some file system such as NFSv2 and NFSv3 do
> not have sufficient support to be able to support leases correctly.
> In particular for these two file systems, there is no over the wire
> protocol support.
>
> Currently, these two file systems fail the fcntl(F_SETLEASE) call
> accidentally, due to a reference counting difference. These file
> systems should fail more consciously, with a proper error to
> indicate that the call is invalid for them.
Define an nfs setlease method that just returns -EINVAL.
If someone can demonstrate a real need, perhaps we could reenable
them in the presence of the "nolock" mount option.
Signed-off-by: "J. Bruce Fields" <bfields@citi.umich.edu>
Cc: Peter Staubach <staubach@redhat.com>
Cc: Trond Myklebust <Trond.Myklebust@netapp.com>
First thing mm.h does is including sched.h solely for can_do_mlock() inline
function which has "current" dereference inside. By dealing with can_do_mlock()
mm.h can be detached from sched.h which is good. See below, why.
This patch
a) removes unconditional inclusion of sched.h from mm.h
b) makes can_do_mlock() normal function in mm/mlock.c
c) exports can_do_mlock() to not break compilation
d) adds sched.h inclusions back to files that were getting it indirectly.
e) adds less bloated headers to some files (asm/signal.h, jiffies.h) that were
getting them indirectly
Net result is:
a) mm.h users would get less code to open, read, preprocess, parse, ... if
they don't need sched.h
b) sched.h stops being dependency for significant number of files:
on x86_64 allmodconfig touching sched.h results in recompile of 4083 files,
after patch it's only 3744 (-8.3%).
Cross-compile tested on
all arm defconfigs, all mips defconfigs, all powerpc defconfigs,
alpha alpha-up
arm
i386 i386-up i386-defconfig i386-allnoconfig
ia64 ia64-up
m68k
mips
parisc parisc-up
powerpc powerpc-up
s390 s390-up
sparc sparc-up
sparc64 sparc64-up
um-x86_64
x86_64 x86_64-up x86_64-defconfig x86_64-allnoconfig
as well as my two usual configs.
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
posix_test_lock() and ->lock() do the same job but have gratuitously
different interfaces. Modify posix_test_lock() so the two agree,
simplifying some code in the process.
Signed-off-by: Marc Eshel <eshel@almaden.ibm.com>
Signed-off-by: "J. Bruce Fields" <bfields@citi.umich.edu>
Many struct inode_operations in the kernel can be "const". Marking them const
moves these to the .rodata section, which avoids false sharing with potential
dirty data. In addition it'll catch accidental writes at compile time to
these shared resources.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
But keep it as a dprintk
The message can be generated in a quite normal situation:
If a 'lock' request is interrupted, then the lock client needs to
record that the server has the lock, incase it does.
When we come the unlock, the server might say it doesn't, even
though we think it does (or might) and this generates the message.
Signed-off-by: Neil Brown <neilb@suse.de>
Acked-by: Trond Myklebust <trond.myklebust@fys.uio.no>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
NFS: Fix race in nfs_release_page()
invalidate_inode_pages2() may find the dirty bit has been set on a page
owing to the fact that the page may still be mapped after it was locked.
Only after the call to unmap_mapping_range() are we sure that the page
can no longer be dirtied.
In order to fix this, NFS has hooked the releasepage() method and tries
to write the page out between the call to unmap_mapping_range() and the
call to remove_mapping(). This, however leads to deadlocks in the page
reclaim code, where the page may be locked without holding a reference
to the inode or dentry.
Fix is to add a new address_space_operation, launder_page(), which will
attempt to write out a dirty page without releasing the page lock.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Also, the bare SetPageDirty() can skew all sort of accounting leading to
other nasties.
[akpm@osdl.org: cleanup]
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Peter Staubach