linux-toradex.git/fs, branch v3.2.1

xfs: fix acl count validation in xfs_acl_from_disk()

2012-01-12T19:29:46+00:00

commit 093019cf1b18dd31b2c3b77acce4e000e2cbc9ce upstream.

Commit fa8b18ed didn't prevent the integer overflow and possible
memory corruption.  "count" can go negative and bypass the check.

Signed-off-by: Xi Wang 
Reviewed-by: Christoph Hellwig 
Signed-off-by: Ben Myers 
Signed-off-by: Greg Kroah-Hartman

udf: Fix deadlock when converting file from in-ICB one to normal one

2012-01-12T19:29:37+00:00

commit d2eb8c359309ec45d6bf5b147303ab8e13be86ea upstream.

During BKL removal in 2.6.38, conversion of files from in-ICB format to normal
format got broken. We call ->writepage with i_data_sem held but udf_get_block()
also acquires i_data_sem thus creating A-A deadlock.

We fix the problem by dropping i_data_sem before calling ->writepage() which is
safe since i_mutex still protects us against any changes in the file. Also fix
pagelock - i_data_sem lock inversion in udf_expand_file_adinicb() by dropping
i_data_sem before calling find_or_create_page().

Reported-by: Matthias Matiak 
Tested-by: Matthias Matiak 
Reviewed-by: Namjae Jeon 
Signed-off-by: Jan Kara 
Signed-off-by: Greg Kroah-Hartman

ext3: Don't warn from writepage when readonly inode is spotted after error

2012-01-12T19:29:32+00:00

commit 33c104d415e92a51aaf638dc3d93920cfa601e5c upstream.

WARN_ON_ONCE(IS_RDONLY(inode)) tends to trip when filesystem hits error and is
remounted read-only. This unnecessarily scares users (well, they should be
scared because of filesystem error, but the stack trace distracts them from the
right source of their fear ;-). We could as well just remove the WARN_ON but
it's not hard to fix it to not trip on filesystem with errors and not use more
cycles in the common case so that's what we do.

Signed-off-by: Jan Kara 
Signed-off-by: Greg Kroah-Hartman

reiserfs: Force inode evictions before umount to avoid crash

2012-01-12T19:29:30+00:00

commit a9e36da655e54545c3289b2a0700b5c443de0edd upstream.

This patch fixes a crash in reiserfs_delete_xattrs during umount.

When shrink_dcache_for_umount clears the dcache from
generic_shutdown_super, delayed evictions are forced to disk. If an
evicted inode has extended attributes associated with it, it will
need to walk the xattr tree to locate and remove them.

But since shrink_dcache_for_umount will BUG if it encounters active
dentries, the xattr tree must be released before it's called or it will
crash during every umount.

This patch forces the evictions to occur before generic_shutdown_super
by calling shrink_dcache_sb first. The additional evictions caused
by the removal of each associated xattr file and dir will be automatically
handled as they're added to the LRU list.

CC: reiserfs-devel@vger.kernel.org
Signed-off-by: Jeff Mahoney 
Signed-off-by: Jan Kara 
Signed-off-by: Greg Kroah-Hartman

reiserfs: Fix quota mount option parsing

2012-01-12T19:29:30+00:00

commit a06d789b424190e9f59da391681f908486db2554 upstream.

When jqfmt mount option is not specified on remount, we mistakenly clear
s_jquota_fmt value stored in superblock. Fix the problem.

CC: reiserfs-devel@vger.kernel.org
Signed-off-by: Jan Kara 
Signed-off-by: Greg Kroah-Hartman

ore: FIX breakage when MISC_FILESYSTEMS is not set

2012-01-12T19:29:29+00:00

commit 831c2dc5f47c1dc79c32229d75065ada1dcc66e1 upstream.

As Reported by Randy Dunlap

When MISC_FILESYSTEMS is not enabled and NFS4.1 is:

fs/built-in.o: In function `objio_alloc_io_state':
objio_osd.c:(.text+0xcb525): undefined reference to `ore_get_rw_state'
fs/built-in.o: In function `_write_done':
objio_osd.c:(.text+0xcb58d): undefined reference to `ore_check_io'
fs/built-in.o: In function `_read_done':
...

When MISC_FILESYSTEMS, which is more of a GUI thing then anything else,
is not selected. exofs/Kconfig is never examined during Kconfig,
and it can not do it's magic stuff to automatically select everything
needed.

We must split exofs/Kconfig in two. The ore one is always included.
And the exofs one is left in it's old place in the menu.

Reported-by: Randy Dunlap 
Signed-off-by: Boaz Harrosh 
Signed-off-by: Greg Kroah-Hartman

ore: Must support none-PAGE-aligned IO

2012-01-12T19:29:29+00:00

commit 724577ca355795b0a25c93ccbeee927871ca1a77 upstream.

NFS might send us offsets that are not PAGE aligned. So
we must read in the reminder of the first/last pages, in cases
we need it for Parity calculations.

We only add an sg segments to read the partial page. But
we don't mark it as read=true because it is a lock-for-write
page.

TODO: In some cases (IO spans a single unit) we can just
adjust the raid_unit offset/length, but this is left for
later Kernels.

Signed-off-by: Boaz Harrosh 
Signed-off-by: Greg Kroah-Hartman

ore: fix BUG_ON, too few sgs when reading

2012-01-12T19:29:28+00:00

commit 361aba569f55dd159b850489a3538253afbb3973 upstream.

When reading RAID5 files, in rare cases, we calculated too
few sg segments. There should be two extra for the beginning
and end partial units.

Also "too few sg segments" should not be a BUG_ON there is
all the mechanics in place to handle it, as a short read.
So just return -ENOMEM and the rest of the code will gracefully
split the IO.

Signed-off-by: Boaz Harrosh 
Signed-off-by: Greg Kroah-Hartman

ore: Fix crash in case of an IO error.

2012-01-12T19:29:27+00:00

commit ffefb8eaa367e8a5c14f779233d9da1fbc23d164 upstream.

The users of ore_check_io() expect the reported device
(In case of error) to be indexed relative to the passed-in
ore_components table, and not the logical dev index.

This causes a crash inside objlayoutdriver in case of
an IO error.

Signed-off-by: Boaz Harrosh 
Signed-off-by: Greg Kroah-Hartman

minixfs: misplaced checks lead to dentry leak

2012-01-04T23:03:06+00:00

bitmap size sanity checks should be done *before* allocating ->s_root;
there their cleanup on failure would be correct.  As it is, we do iput()
on root inode, but leak the root dentry...

Signed-off-by: Al Viro 
Acked-by: Josh Boyer 
Signed-off-by: Linus Torvalds