* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mfasheh/ocfs2: (138 commits)
  ocfs2: Access the right buffer_head in ocfs2_merge_rec_left.
  ocfs2: use min_t in ocfs2_quota_read()
  ocfs2: remove unneeded lvb casts
  ocfs2: Add xattr support checking in init_security
  ocfs2: alloc xattr bucket in ocfs2_xattr_set_handle
  ocfs2: calculate and reserve credits for xattr value in mknod
  ocfs2/xattr: fix credits calculation during index create
  ocfs2/xattr: Always updating ctime during xattr set.
  ocfs2/xattr: Remove extend_trans call and add its credits from the beginning
  ocfs2/dlm: Fix race during lockres mastery
  ocfs2/dlm: Fix race in adding/removing lockres' to/from the tracking list
  ocfs2/dlm: Hold off sending lockres drop ref message while lockres is migrating
  ocfs2/dlm: Clean up errors in dlm_proxy_ast_handler()
  ocfs2/dlm: Fix a race between migrate request and exit domain
  ocfs2: One more hamming code optimization.
  ocfs2: Another hamming code optimization.
  ocfs2: Don't hand-code xor in ocfs2_hamming_encode().
  ocfs2: Enable metadata checksums.
  ocfs2: Validate superblock with checksum and ecc.
  ocfs2: Checksum and ECC for directory blocks.
  ...

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6:
  inotify: fix type errors in interfaces
  fix breakage in reiserfs_new_inode()
  fix the treatment of jfs special inodes
  vfs: remove duplicate code in get_fs_type()
  add a vfs_fsync helper
  sys_execve and sys_uselib do not call into fsnotify
  zero i_uid/i_gid on inode allocation
  inode->i_op is never NULL
  ntfs: don't NULL i_op
  isofs check for NULL ->i_op in root directory is dead code
  affs: do not zero ->i_op
  kill suid bit only for regular files
  vfs: lseek(fd, 0, SEEK_CUR) race condition

An XFS workload showed up a bug in the lockless pagecache patch. Basically it
would go into an "infinite" loop, although it would sometimes be able to break
out of the loop! The reason is a missing compiler barrier in the "increment
reference count unless it was zero" case of the lockless pagecache protocol in
the gang lookup functions.

This would cause the compiler to use a cached value of struct page pointer to
retry the operation with, rather than reload it. So the page might have been
removed from pagecache and freed (refcount==0) but the lookup would not correctly
notice the page is no longer in pagecache, and keep attempting to increment the
refcount and failing, until the page gets reallocated for something else. This
isn't a data corruption because the condition will be detected if the page has
been reallocated. However it can result in a lockup.

Linus points out that ACCESS_ONCE is also required in that pointer load, even
if it's absence is not causing a bug on our particular build. The most general
way to solve this is just to put an rcu_dereference in radix_tree_deref_slot.

Assembly of find_get_pages,
before:
.L220:
        movq    (%rbx), %rax    #* ivtmp.1162, tmp82
        movq    (%rax), %rdi    #, prephitmp.1149
.L218:
        testb   $1, %dil        #, prephitmp.1149
        jne     .L217   #,
        testq   %rdi, %rdi      # prephitmp.1149
        je      .L203   #,
        cmpq    $-1, %rdi       #, prephitmp.1149
        je      .L217   #,
        movl    8(%rdi), %esi   # <variable>._count.counter, c
        testl   %esi, %esi      # c
        je      .L218   #,

after:
.L212:
        movq    (%rbx), %rax    #* ivtmp.1109, tmp81
        movq    (%rax), %rdi    #, ret
        testb   $1, %dil        #, ret
        jne     .L211   #,
        testq   %rdi, %rdi      # ret
        je      .L197   #,
        cmpq    $-1, %rdi       #, ret
        je      .L211   #,
        movl    8(%rdi), %esi   # <variable>._count.counter, c
        testl   %esi, %esi      # c
        je      .L212   #,

(notice the obvious infinite loop in the first example, if page->count remains 0)
Signed-off-by: NNick Piggin <npiggin@suse.de>
Reviewed-by: NPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Signed-off-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Signed-off-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

... if you revert a commit, revert the fixups elsewhere that had been
triggered by it.  Such as 8c56250f
(lockdep, UML: fix compilation when CONFIG_TRACE_IRQFLAGS_SUPPORT is not set).
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

We need to make asm-offsets.h contents visible for objects built
with userland headers.  Instead of creating a symlink, just have the
file with equivalent include (relative to location of header) created
once.  That kills the last symlink used in arch/um builds.

Additionally, both generated headers can become dependencies of
archprepare now, killing the misuse of prepare.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The problems lie in the types used for some inotify interfaces, both at the kernel level and at the glibc level. This mail addresses the kernel problem. I will follow up with some suggestions for glibc changes.

For the sys_inotify_rm_watch() interface, the type of the 'wd' argument is
currently 'u32', it should be '__s32' .  That is Robert's suggestion, and
is consistent with the other declarations of watch descriptors in the
kernel source, in particular, the inotify_event structure in
include/linux/inotify.h:

struct inotify_event {
        __s32           wd;             /* watch descriptor */
        __u32           mask;           /* watch mask */
        __u32           cookie;         /* cookie to synchronize two events */
        __u32           len;            /* length (including nulls) of name */
        char            name[0];        /* stub for possible name */
};

The patch makes the changes needed for inotify_rm_watch().
Signed-off-by: NMichael Kerrisk <mtk.manpages@googlemail.com>
Cc: Robert Love <rlove@google.com>
Cc: Vegard Nossum <vegard.nossum@gmail.com>
Cc: Ulrich Drepper <drepper@redhat.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

now that we use ih.key earlier, we need to do all its setup early enough
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

We used to put them on a single list, without any locking.  Racy.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

save 14 bytes:

   text    data     bss     dec     hex filename
   1354      32       4    1390     56e fs/filesystems.o.before
   text    data     bss     dec     hex filename
   1340      32       4    1376     560 fs/filesystems.o
Signed-off-by: NLi Zefan <lizf@cn.fujitsu.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Fsync currently has a fdatawrite/fdatawait pair around the method call,
and a mutex_lock/unlock of the inode mutex.  All callers of fsync have
to duplicate this, but we have a few and most of them don't quite get
it right.  This patch adds a new vfs_fsync that takes care of this.
It's a little more complicated as usual as ->fsync might get a NULL file
pointer and just a dentry from nfsd, but otherwise gets afile and we
want to take the mapping and file operations from it when it is there.

Notes on the fsync callers:

 - ecryptfs wasn't calling filemap_fdatawrite / filemap_fdatawait on the
   	lower file
 - coda wasn't calling filemap_fdatawrite / filemap_fdatawait on the host
	file, and returning 0 when ->fsync was missing
 - shm wasn't calling either filemap_fdatawrite / filemap_fdatawait nor
   taking i_mutex.  Now given that shared memory doesn't have disk
   backing not doing anything in fsync seems fine and I left it out of
   the vfs_fsync conversion for now, but in that case we might just
   not pass it through to the lower file at all but just call the no-op
   simple_sync_file directly.

[and now actually export vfs_fsync]
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

sys_execve and sys_uselib do not call into fsnotify so inotify does not get
open events for these types of syscalls.  This patch simply makes the
requisite fsnotify calls.
Signed-off-by: NEric Paris <eparis@redhat.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

... and don't bother in callers.  Don't bother with zeroing i_blocks,
while we are at it - it's already been zeroed.

i_mode is not worth the effort; it has no common default value.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

We used to have rather schizophrenic set of checks for NULL ->i_op even
though it had been eliminated years ago.  You'd need to go out of your
way to set it to NULL explicitly _and_ a bunch of code would die on
such inodes anyway.  After killing two remaining places that still
did that bogosity, all that crap can go away.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

it's already set to empty table (and no, ntfs doesn't have any explicit
checks for NULL ->i_op or NULL ->i_fop)
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

for one thing it never happens, for another we check that inode
is a directory right after that place anyway (and we'd already
checked that reading it from disk has not failed).
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

it is already set to empty table and should never be NULL
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

We don't have to do it because it is useless for non regular files.
In fact block device may trigger this path without dentry->d_inode->i_mutex.

(akpm: concerns were expressed (by me) about S_ISDIR inodes)
Signed-off-by: NDmitri Monakhov <dmonakhov@openvz.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

This patch fixes a race condition in lseek. While it is expected that
unpredictable behaviour may result while repositioning the offset of a
file descriptor concurrently with reading/writing to the same file
descriptor, this should not happen when merely *reading* the file
descriptor's offset.

Unfortunately, the only portable way in Unix to read a file
descriptor's offset is lseek(fd, 0, SEEK_CUR); however executing this
concurrently with read/write may mess up the position.

[with fixes from akpm]
Signed-off-by: NAlain Knaff <alain@knaff.lu>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

In commit "ocfs2: Use metadata-specific ocfs2_journal_access_*()
functions", the wrong buffer_head is accessed. So change it
to the right buffer_head.
Signed-off-by: NTao Ma <tao.ma@oracle.com>
Acked-by: NJoel Becker <joel.becker@oracle.com>
Signed-off-by: NMark Fasheh <mfasheh@suse.com>

This is preferred to min().
Signed-off-by: NMark Fasheh <mfasheh@suse.com>

dlmglue.c has lots of code which casts the return value of ocfs2_dlm_lvb().
This is pointless however, as ocfs2_dlm_lvb() returns void *.
Signed-off-by: NMark Fasheh <mfasheh@suse.com>

We must check whether ocfs2 volume support xattr in init_security,
if not support xattr and security is enable, would cause failure of mknod.
Signed-off-by: NTiger Yang <tiger.yang@oracle.com>
Signed-off-by: NMark Fasheh <mfasheh@suse.com>

In extreme situation, may need xattr bucket for setting
security entry and acl entries during mknod. This only
happens when block size is too small.
Signed-off-by: NTiger Yang <tiger.yang@oracle.com>
Signed-off-by: NMark Fasheh <mfasheh@suse.com>

We extend the credits for xattr's large value in set_value_outside
before, this can give rise to a credits issue when we set one security
entry and two acl entries duing mknod. As we remove extend_trans form
set_value_outside, we must calculate and reserve the credits for
xattr's large value in mknod.
Signed-off-by: NTiger Yang <tiger.yang@oracle.com>
Signed-off-by: NMark Fasheh <mfasheh@suse.com>

When creating a xattr index block, the old calculation forget
to add credits for the meta change of the alloc file. So add
more credits and more comments to explain it.
Signed-off-by: NTao Ma <tao.ma@oracle.com>
Signed-off-by: NMark Fasheh <mfasheh@suse.com>

In xattr set, we should always update ctime if the operation goes
sucessfully. The old one mistakenly put it in ocfs2_xattr_set_entry
which is only called when we set xattr in inode or xattr block. The
side benefit is that it resolve the bug 1052 since in that scenario,
ocfs2_calc_xattr_set_need only calc out the xattr set credits while
ocfs2_xattr_set_entry update the inode also which isn't concerned with
the process of xattr set.
Signed-off-by: NTao Ma <tao.ma@oracle.com>
Signed-off-by: NMark Fasheh <mfasheh@suse.com>

Actually, when setting a new xattr value, we know it from the very
beginning, and it isn't like the extension of bucket in which case
we can't figure it out. So remove ocfs2_extend_trans in that function
and calculate it before the transaction. It also relieve acl operation
from the worry about the side effect of ocfs2_extend_trans.
Signed-off-by: NTao Ma <tao.ma@oracle.com>
Signed-off-by: NMark Fasheh <mfasheh@suse.com>

dlm_get_lock_resource() is supposed to return a lock resource with a proper
master. If multiple concurrent threads attempt to lookup the lockres for the
same lockid while the lock mastery in underway, one or more threads are likely
to return a lockres without a proper master.

This patch makes the threads wait in dlm_get_lock_resource() while the mastery
is underway, ensuring all threads return the lockres with a proper master.

This issue is known to be limited to users using the flock() syscall. For all
other fs operations, the ocfs2 dlmglue layer serializes the dlm op for each
lockid.

Users encountering this bug will see flock() return EINVAL and dmesg have the
following error:
ERROR: Dlm error "DLM_BADARGS" while calling dlmlock on resource <LOCKID>: bad api args
Reported-by: NColy Li <coyli@suse.de>
Signed-off-by: NSunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: NMark Fasheh <mfasheh@suse.com>

This patch adds a new lock, dlm->tracking_lock, to protect adding/removing
lockres' to/from the dlm->tracking_list. We were previously using dlm->spinlock
for the same, but that proved inadequate as we could be freeing a lockres from
a context that did not hold that lock. As the new lock only protects this list,
we can explicitly take it when removing the lockres from the tracking list.

This bug was exposed when testing multiple processes concurrently flock() the
same file.
Signed-off-by: NSunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: NMark Fasheh <mfasheh@suse.com>

During lockres purge, o2dlm sends a drop reference message to the lockres
master. This patch delays the message if the lockres is being migrated.

Fixes oss bugzilla#1012
http://oss.oracle.com/bugzilla/show_bug.cgi?id=1012Signed-off-by: NSunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: NMark Fasheh <mfasheh@suse.com>

Patch cleans printed errors in dlm_proxy_ast_handler(). The errors now includes
the node number that sent the (b)ast. Also it reduces the number of endian swaps
of the cookie.
Signed-off-by: NSunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: NMark Fasheh <mfasheh@suse.com>

Patch address a racing migrate request message and an exit domain message.
Instead of blocking exit domains for the duration of the migrate, we ignore
failure to deliver that message. This is because an exiting domain should
not have any active locks and thus has no role to play in the migration.
Signed-off-by: NSunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: NMark Fasheh <mfasheh@suse.com>

The previous optimization used a fast find-highest-bit-set operation to
give us a good starting point in calc_code_bit().  This version lets the
caller cache the previous code buffer bit offset.  Thus, the next call
always starts where the last one left off.

This reduces the calculation another 39%, for a total 80% reduction from
the original, naive implementation.  At least, on my machine.  This also
brings the parity calculation to within an order of magnitude of the
crc32 calculation.
Signed-off-by: NJoel Becker <joel.becker@oracle.com>
Signed-off-by: NMark Fasheh <mfasheh@suse.com>

In the calc_code_bit() function, we must find all powers of two beneath
the code bit number, *after* it's shifted by those powers of two.  This
requires a loop to see where it ends up.

We can optimize it by starting at its most significant bit.  This shaves
32% off the time, for a total of 67.6% shaved off of the original, naive
implementation.
Signed-off-by: NJoel Becker <joel.becker@oracle.com>
Signed-off-by: NMark Fasheh <mfasheh@suse.com>

When I wrote ocfs2_hamming_encode(), I was following documentation of
the algorithm and didn't have quite the (possibly still imperfect) grasp
of it I do now.  As part of this, I literally hand-coded xor.  I would
test a bit, and then add that bit via xor to the parity word.

I can, of course, just do a single xor of the parity word and the source
word (the code buffer bit offset).  This cuts CPU usage by 53% on a
mostly populated buffer (an inode containing utmp.h inline).

Joel
Signed-off-by: NJoel Becker <joel.becker@oracle.com>
Signed-off-by: NMark Fasheh <mfasheh@suse.com>

Add OCFS2_FEATURE_INCOMPAT_META_ECC to the list of supported features.
Signed-off-by: NJoel Becker <joel.becker@oracle.com>
Signed-off-by: NMark Fasheh <mfasheh@suse.com>

The superblock is read via a raw call.  Validate it after we find it
from its signature.
Signed-off-by: NJoel Becker <joel.becker@oracle.com>
Signed-off-by: NMark Fasheh <mfasheh@suse.com>