linux-2.6
16 years agofat: i_blocks warning fix
OGAWA Hirofumi [Thu, 6 Nov 2008 20:53:58 +0000 (12:53 -0800)] 
fat: i_blocks warning fix

blkcnt_t type depends on CONFIG_LSF. Use unsigned long long always for
printk().  But lazy to type it, so add "llu" and use it.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofat: ->i_pos race fix
OGAWA Hirofumi [Thu, 6 Nov 2008 20:53:57 +0000 (12:53 -0800)] 
fat: ->i_pos race fix

i_pos is 64bits value, hence it's not atomic to update.

Important place is fat_write_inode() only, other places without lock
are just for printk().

This adds lock for "BITS_PER_LONG == 32" kernel.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofat: mmu_private race fix
OGAWA Hirofumi [Thu, 6 Nov 2008 20:53:57 +0000 (12:53 -0800)] 
fat: mmu_private race fix

mmu_private is 64bits value, hence it's not atomic to update.

So, the access rule for mmu_private is we must hold ->i_mutex.  But,
fat_get_block() path doesn't follow the rule on non-allocation path.

This fixes by using i_size instead if non-allocation path.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofat: Add printf attribute to fat_fs_panic()
OGAWA Hirofumi [Thu, 6 Nov 2008 20:53:56 +0000 (12:53 -0800)] 
fat: Add printf attribute to fat_fs_panic()

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofat: Fix _fat_bmap() race
OGAWA Hirofumi [Thu, 6 Nov 2008 20:53:56 +0000 (12:53 -0800)] 
fat: Fix _fat_bmap() race

fat_get_cluster() assumes the requested blocknr isn't truncated during
read. _fat_bmap() doesn't follow this rule.

This protects it by ->i_mutex.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofat: Fix ATTR_RO for directory
OGAWA Hirofumi [Thu, 6 Nov 2008 20:53:55 +0000 (12:53 -0800)] 
fat: Fix ATTR_RO for directory

FAT has the ATTR_RO (read-only) attribute. But on Windows, the ATTR_RO
of the directory will be just ignored actually, and is used by only
applications as flag. E.g. it's setted for the customized folder by
Explorer.

http://msdn2.microsoft.com/en-us/library/aa969337.aspx

This adds "rodir" option. If user specified it, ATTR_RO is used as
read-only flag even if it's the directory. Otherwise, inode->i_mode
is not used to hold ATTR_RO (i.e. fat_mode_can_save_ro() returns 0).

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofat: Fix ATTR_RO in the case of (~umask & S_WUGO) == 0
OGAWA Hirofumi [Thu, 6 Nov 2008 20:53:54 +0000 (12:53 -0800)] 
fat: Fix ATTR_RO in the case of (~umask & S_WUGO) == 0

If inode->i_mode doesn't have S_WUGO, current code assumes it means
ATTR_RO.  However, if (~[ufd]mask & S_WUGO) == 0, inode->i_mode can't
hold S_WUGO. Therefore the updated directory entry will always have
ATTR_RO.

This adds fat_mode_can_hold_ro() to check it. And if inode->i_mode
can't hold, uses -i_attrs to hold ATTR_RO instead.

With this, we don't set ATTR_RO unless users change it via ioctl() if
(~[ufd]mask & S_WUGO) == 0.

And on FAT_IOCTL_GET_ATTRIBUTES path, this adds ->i_mutex to it for
not returning the partially updated attributes by FAT_IOCTL_SET_ATTRIBUTES
to userland.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofat: Cleanup FAT attribute stuff
OGAWA Hirofumi [Thu, 6 Nov 2008 20:53:54 +0000 (12:53 -0800)] 
fat: Cleanup FAT attribute stuff

This adds three helpers:

fat_make_attrs() - makes FAT attributes from inode.
fat_make_mode()  - makes mode_t from FAT attributes.
fat_save_attrs() - saves FAT attributes to inode.

Then this replaces: MSDOS_MKMODE() by fat_make_mode(), fat_attr() by
fat_make_attrs(), ->i_attrs = attr & ATTR_UNUSED by fat_save_attrs().
And for root inode, those is used with ATTR_DIR instead of bogus
ATTR_NONE.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofat: Cleanup msdos_lookup()
OGAWA Hirofumi [Thu, 6 Nov 2008 20:53:53 +0000 (12:53 -0800)] 
fat: Cleanup msdos_lookup()

Use same style with vfat_lookup().

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofat: Kill d_invalidate() in vfat_lookup()
OGAWA Hirofumi [Thu, 6 Nov 2008 20:53:52 +0000 (12:53 -0800)] 
fat: Kill d_invalidate() in vfat_lookup()

d_invalidate() for positive dentry doesn't work in some cases
(vfsmount, nfsd, and maybe others). shrink_dcache_parent() by
d_invalidate() is pointless for vfat usage at all.

So, this kills it, and intead of it uses d_move().

To save old behavior, this returns alias simply for directory (don't
change pwd, etc..). the directory lookup shouldn't be important for
performance.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofat: Fix/Cleanup dcache handling for vfat
OGAWA Hirofumi [Thu, 6 Nov 2008 20:53:51 +0000 (12:53 -0800)] 
fat: Fix/Cleanup dcache handling for vfat

- Add comments for handling dcache of vfat.

- Separate case-sensitive case and case-insensitive to
  vfat_revalidate() and vfat_ci_revalidate().

  vfat_revalidate() doesn't need to drop case-insensitive negative
  dentry on creation path.

- Current code is missing to set ->d_revalidate to the negative dentry
  created by unlink/etc..

  This sets ->d_revalidate always, and returns 1 for positive
  dentry. Now, we don't need to change ->d_op dynamically anymore,
  so this just uses sb->s_root->d_op to set ->d_op.

- d_find_alias() may return DCACHE_DISCONNECTED dentry. It's not
  the interesting dentry there. This checks it.

- Add missing LOOKUP_PARENT check. We don't need to drop the valid
  negative dentry for (LOOKUP_CREATE | LOOKUP_PARENT) lookup.

- For consistent filename on creation path, this drops negative dentry
  if we can't see intent.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agovfat: Fix vfat_find() error path in vfat_lookup()
OGAWA Hirofumi [Thu, 6 Nov 2008 20:53:51 +0000 (12:53 -0800)] 
vfat: Fix vfat_find() error path in vfat_lookup()

Current vfat_lookup() creates negetive dentry blindly if vfat_find()
returned a error. It's wrong. If the error isn't -ENOENT, just return
error.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofat: use fat_detach() in fat_clear_inode()
OGAWA Hirofumi [Thu, 6 Nov 2008 20:53:50 +0000 (12:53 -0800)] 
fat: use fat_detach() in fat_clear_inode()

Use fat_detach() instead of opencoding it.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofat: Fix fat_ent_update_ptr() for FAT12
OGAWA Hirofumi [Thu, 6 Nov 2008 20:53:49 +0000 (12:53 -0800)] 
fat: Fix fat_ent_update_ptr() for FAT12

This fixes the missing update for bhs/nr_bhs in case the caller
accessed from block boundary to first block of boundary.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofat: improve fat_hash()
OGAWA Hirofumi [Thu, 6 Nov 2008 20:53:49 +0000 (12:53 -0800)] 
fat: improve fat_hash()

fat_hash() is using the algorithm known as bad. Instead of it, this
uses hash_32(). The following is the summary of test.

old hash:
hash func (1000 times): 33489 cycles
total inodes in hash table: 70926
largest bucket contains: 696
smallest bucket contains: 54

new hash:
hash func (1000 times): 33129 cycles
total inodes in hash table: 70926
largest bucket contains: 315
smallest bucket contains: 236

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofat: cleanup fat_parse_long() error handling
Darren Jenkins [Thu, 6 Nov 2008 20:53:48 +0000 (12:53 -0800)] 
fat: cleanup fat_parse_long() error handling

Coverity CID 2332 & 2333 RESOURCE_LEAK

In fat_search_long() if fat_parse_long() returns a -ve value we return
without first freeing unicode.  This patch free's them on this error path.

The above was false positive on current tree, but this change is more
clean, so apply as cleanup.

[hirofumi@mail.parknet.co.jp: fix coding style]
Signed-off-by: Darren Jenkins <darrenrjenkins@gmail.com>
Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofat: use generic_file_llseek() for directory
OGAWA Hirofumi [Thu, 6 Nov 2008 20:53:47 +0000 (12:53 -0800)] 
fat: use generic_file_llseek() for directory

Since fat_dir_ioctl() was already fixed (i.e. called under ->i_mutex),
and __fat_readdir() doesn't take BKL anymore. So, BKL for ->llseek()
is pointless, and we have to use generic_file_llseek().

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofat: Fix and cleanup timestamp conversion
OGAWA Hirofumi [Thu, 6 Nov 2008 20:53:47 +0000 (12:53 -0800)] 
fat: Fix and cleanup timestamp conversion

This cleans date_dos2unix()/fat_date_unix2dos() up. New code should be
much more readable.

And this fixes those old functions. Those doesn't handle 2100
correctly. 2100 isn't leap year, but old one handles it as leap year.
Also, with this, centi sec is handled and is fixed.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofat: split include/msdos_fs.h
OGAWA Hirofumi [Thu, 6 Nov 2008 20:53:46 +0000 (12:53 -0800)] 
fat: split include/msdos_fs.h

This splits __KERNEL__ stuff in include/msdos_fs.h into fs/fat/fat.h.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofat: move fs/vfat/* and fs/msdos/* to fs/fat
OGAWA Hirofumi [Thu, 6 Nov 2008 20:53:45 +0000 (12:53 -0800)] 
fat: move fs/vfat/* and fs/msdos/* to fs/fat

This just moves those files, but change link order from MSDOS, VFAT to
VFAT, MSDOS.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofat: document additional vfat mount options
Bart Trojanowski [Thu, 6 Nov 2008 20:53:44 +0000 (12:53 -0800)] 
fat: document additional vfat mount options

While debugging a sync mount regression on vfat I noticed that there were
mount options parsed by the driver that were not documented.

[hirofumi@mail.parknet.co.jp: fix some parts]
Signed-off-by: Bart Trojanowski <bart@jukie.net>
Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoSAM9 watchdog: update for moved headers
Andrew Victor [Thu, 6 Nov 2008 20:53:42 +0000 (12:53 -0800)] 
SAM9 watchdog: update for moved headers

The architecture header files were recently moved from
include/asm-arm/mach-at91/ to arch/arm/mach-at91/include/mach/.  The SAM9
watchdog driver still includes a header from the old location.

Signed-off-by: Andrew Victor <linux@maxim.org.za>
Cc: Wim Van Sebroeck <wim@iguana.be>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc-cmos: fix boot log message
Frans Pop [Thu, 6 Nov 2008 20:53:41 +0000 (12:53 -0800)] 
rtc-cmos: fix boot log message

-rtc0: alarms up to one month, y3k, 114 bytes nvram, , hpet irqs irqs
+rtc0: alarms up to one month, y3k, 114 bytes nvram, hpet irqs

Signed-off-by: Frans Pop <elendil@planet.nl>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoatmel_serial: keep clock off when it's not needed
David Brownell [Thu, 6 Nov 2008 20:53:40 +0000 (12:53 -0800)] 
atmel_serial: keep clock off when it's not needed

The atmel_serial driver is mismanaging its clock by leaving it on at all
times ...  the whole point of clock management is to leave it off unless
it's actively needed, which conserves power!!

Although the kernel doesn't actually hang without my fix, it does
discard quite a lot of early console output.

The result still looks correct:

          usart        users= 1 on   35000000 Hz, for atmel_usart.0
          usart        users= 0 off  35000000 Hz, for atmel_usart.2

when using ttyS0 as serial console.

[haavard.skinnemoen@atmel.com: Make sure clock is enabled early for console]
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Haavard Skinnemoen <haavard.skinnemoen@atmel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoDocumentation/kernel-parameters.txt: update 'isolcpus' kernel option
Li Zefan [Thu, 6 Nov 2008 20:53:39 +0000 (12:53 -0800)] 
Documentation/kernel-parameters.txt: update 'isolcpus' kernel option

cpuset can be used to move a process onto or off an isolated CPU.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofbdev: fix fb_compat_ioctl() deadlocks
Geert Uytterhoeven [Thu, 6 Nov 2008 20:53:37 +0000 (12:53 -0800)] 
fbdev: fix fb_compat_ioctl() deadlocks

commit 3e680aae4e53ab54cdbb0c29257dae0cbb158e1c ("fb: convert
lock/unlock_kernel() into local fb mutex") introduced several deadlocks
in the fb_compat_ioctl() path, as mutex_lock() doesn't allow recursion,
unlike lock_kernel().  This broke frame buffer applications on 64-bit
systems with a 32-bit userland.

commit 120a37470c2831fea49fdebaceb5a7039f700ce6 ("framebuffer compat_ioctl
deadlock") fixed one of the deadlocks.

This patch fixes the remaining deadlocks:
  - Revert commit 120a37470c2831fea49fdebaceb5a7039f700ce6,
  - Extract the core logic of fb_ioctl() into a new function do_fb_ioctl(),
  - Change all callsites of fb_ioctl() where info->lock is already held to
    call do_fb_ioctl() instead,
  - Add sparse annotations to all routines that take info->lock.

Signed-off-by: Geert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Cc: Mikulas Patocka <mpatocka@redhat.com>
Cc: Krzysztof Helt <krzysztof.h1@wp.pl>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomemory hotplug: fix page_zone() calculation in test_pages_isolated()
Gerald Schaefer [Thu, 6 Nov 2008 20:53:36 +0000 (12:53 -0800)] 
memory hotplug: fix page_zone() calculation in test_pages_isolated()

My last bugfix here (adding zone->lock) introduced a new problem: Using
page_zone(pfn_to_page(pfn)) to get the zone after the for() loop is wrong.
 pfn will then be >= end_pfn, which may be in a different zone or not
present at all.  This may lead to an addressing exception in page_zone()
or spin_lock_irqsave().

Now I use __first_valid_page() again after the loop to find a valid page
for page_zone().

Signed-off-by: Gerald Schaefer <gerald.schaefer@de.ibm.com>
Acked-by: Nathan Fontenot <nfont@austin.ibm.com>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoext3: wait on all pending commits in ext3_sync_fs
Arthur Jones [Thu, 6 Nov 2008 20:53:35 +0000 (12:53 -0800)] 
ext3: wait on all pending commits in ext3_sync_fs

In ext3_sync_fs, we only wait for a commit to finish if we started it, but
there may be one already in progress which will not be synced.

In the case of a data=ordered umount with pending long symlinks which are
delayed due to a long list of other I/O on the backing block device, this
causes the buffer associated with the long symlinks to not be moved to the
inode dirty list in the second phase of fsync_super.  Then, before they
can be dirtied again, kjournald exits, seeing the UMOUNT flag and the
dirty pages are never written to the backing block device, causing long
symlink corruption and exposing new or previously freed block data to
userspace.

This can be reproduced with a script created
by Eric Sandeen <sandeen@redhat.com>:

#!/bin/bash

umount /mnt/test2
mount /dev/sdb4 /mnt/test2
rm -f /mnt/test2/*
dd if=/dev/zero of=/mnt/test2/bigfile bs=1M count=512
touch
/mnt/test2/thisisveryveryveryveryveryveryveryveryveryveryveryveryveryveryveryverylongfilename
ln -s
/mnt/test2/thisisveryveryveryveryveryveryveryveryveryveryveryveryveryveryveryverylongfilename
/mnt/test2/link
umount /mnt/test2
mount /dev/sdb4 /mnt/test2
ls /mnt/test2/
umount /mnt/test2

To ensure all commits are synced, we flush all journal commits now when
sync_fs'ing ext3.

Signed-off-by: Arthur Jones <ajones@riverbed.com>
Cc: Eric Sandeen <sandeen@redhat.com>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: <linux-ext4@vger.kernel.org>
Cc: <stable@kernel.org> [2.6.everything]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm/oom_kill.c: fix badness() kerneldoc
Qinghuang Feng [Thu, 6 Nov 2008 20:53:34 +0000 (12:53 -0800)] 
mm/oom_kill.c: fix badness() kerneldoc

Paramter @mem has been removed since v2.6.26, now delete it's comment.

Signed-off-by: Qinghuang Feng <qhfeng.kernel@gmail.com>
Acked-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agocciss: add P700m to list of supported controllers
dann frazier [Thu, 6 Nov 2008 20:53:34 +0000 (12:53 -0800)] 
cciss: add P700m to list of supported controllers

P700m support was added in:
  9cff3b383dad193b0762c27278a16237e10b53dc

Update cciss.txt to match.

Signed-off-by: dann frazier <dannf@hp.com>
Acked-by: Mike Miller <mike.miller@hp.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoDocumentation/email-clients.txt: add some info about gmail
Tim Hockin [Thu, 6 Nov 2008 20:53:33 +0000 (12:53 -0800)] 
Documentation/email-clients.txt: add some info about gmail

Signed-off-by: Tim Hockin <thockin@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agocgroups: fix invalid cgrp->dentry before cgroup has been completely removed
Li Zefan [Thu, 6 Nov 2008 20:53:32 +0000 (12:53 -0800)] 
cgroups: fix invalid cgrp->dentry before cgroup has been completely removed

This fixes an oops when reading /proc/sched_debug.

A cgroup won't be removed completely until finishing cgroup_diput(), so we
shouldn't invalidate cgrp->dentry in cgroup_rmdir().  Otherwise, when a
group is being removed while cgroup_path() gets called, we may trigger
NULL dereference BUG.

The bug can be reproduced:

 # cat test.sh
 #!/bin/sh
 mount -t cgroup -o cpu xxx /mnt
 for (( ; ; ))
 {
mkdir /mnt/sub
rmdir /mnt/sub
 }
 # ./test.sh &
 # cat /proc/sched_debug

BUG: unable to handle kernel NULL pointer dereference at 00000038
IP: [<c045a47f>] cgroup_path+0x39/0x90
...
Call Trace:
 [<c0420344>] ? print_cfs_rq+0x6e/0x75d
 [<c0421160>] ? sched_debug_show+0x72d/0xc1e
...

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Acked-by: Paul Menage <menage@google.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: <stable@kernel.org> [2.6.26.x, 2.6.27.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agovmemmap: warn about page_structs with remote distance
David Rientjes [Thu, 6 Nov 2008 20:53:31 +0000 (12:53 -0800)] 
vmemmap: warn about page_structs with remote distance

It's insufficient to simply compare node ids when warning about offnode
page_structs since it's possible to still have local affinity.

Acked-by: Christoph Lameter <cl@linux-foundation.org>
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: move migrate_prep out from under mmap_sem
Christoph Lameter [Thu, 6 Nov 2008 20:53:30 +0000 (12:53 -0800)] 
mm: move migrate_prep out from under mmap_sem

Move the migrate_prep outside the mmap_sem for the following system calls

1. sys_move_pages
2. sys_migrate_pages
3. sys_mbind()

It really does not matter when we flush the lru.  The system is free to
add pages onto the lru even during migration which will make the page
migration either skip the page (mbind, migrate_pages) or return a busy
state (move_pages).

Fixes this lockdep warning (and potential deadlock):

Some VM place has
      mmap_sem -> kevent_wq via lru_add_drain_all()

net/core/dev.c::dev_ioctl()  has
     rtnl_lock  ->  mmap_sem        (*) the ioctl has copy_from_user() and it can do page fault.

linkwatch_event has
     kevent_wq -> rtnl_lock

Signed-off-by: Christoph Lameter <cl@linux-foundation.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reported-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Hugh Dickins <hugh@veritas.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofbdev: add new framebuffer driver for Fujitsu MB862xx GDCs
Anatolij Gustschin [Thu, 6 Nov 2008 20:53:29 +0000 (12:53 -0800)] 
fbdev: add new framebuffer driver for Fujitsu MB862xx GDCs

Add a framebuffer driver for the Fujitsu Carmine/Coral-P(A)/Lime graphics
controllers.  Lime GDC support is known to work on PPC440EPx based lwmon5
and MPC8544E based socrates embedded boards, both equipped with Lime GDC.
Carmine/Coral-P PCI GDC support is known to work on PPC440EPx based
Sequoia board and also on x86 platform.

Signed-off-by: Anatolij Gustschin <agust@denx.de>
Cc: Dmitry Baryshkov <dbaryshkov@gmail.com>
Cc: Anton Vorontsov <avorontsov@ru.mvista.com>
Cc: Matteo Fortini <m.fortini@selcomgroup.com>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agooom: do not dump task state for non thread group leaders
David Rientjes [Thu, 6 Nov 2008 20:53:29 +0000 (12:53 -0800)] 
oom: do not dump task state for non thread group leaders

When /proc/sys/vm/oom_dump_tasks is enabled, it's only necessary to dump
task state information for thread group leaders.  The kernel log gets
quickly overwhelmed on machines with a massive number of threads by
dumping non-thread group leaders.

Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMAINTAINERS: make IOAT easier to find
Andrew Morton [Thu, 6 Nov 2008 20:53:28 +0000 (12:53 -0800)] 
MAINTAINERS: make IOAT easier to find

Searching MAINTAINERS for "ioat" comes up empty.  Fix this.

Cc: "Dan Williams" <dan.j.williams@intel.com>
Cc: "Sosnowski, Maciej" <maciej.sosnowski@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agohugetlb: pull gigantic page initialisation out of the default path
Andy Whitcroft [Thu, 6 Nov 2008 20:53:27 +0000 (12:53 -0800)] 
hugetlb: pull gigantic page initialisation out of the default path

As we can determine exactly when a gigantic page is in use we can optimise
the common regular page cases by pulling out gigantic page initialisation
into its own function.  As gigantic pages are never released to buddy we
do not need a destructor.  This effectivly reverts the previous change to
the main buddy allocator.  It also adds a paranoid check to ensure we
never release gigantic pages from hugetlbfs to the main buddy.

Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Cc: Jon Tollefson <kniht@linux.vnet.ibm.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: <stable@kernel.org> [2.6.27.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agohugetlbfs: handle pages higher order than MAX_ORDER
Andy Whitcroft [Thu, 6 Nov 2008 20:53:26 +0000 (12:53 -0800)] 
hugetlbfs: handle pages higher order than MAX_ORDER

When working with hugepages, hugetlbfs assumes that those hugepages are
smaller than MAX_ORDER.  Specifically it assumes that the mem_map is
contigious and uses that to optimise access to the elements of the mem_map
that represent the hugepage.  Gigantic pages (such as 16GB pages on
powerpc) by definition are of greater order than MAX_ORDER (larger than
MAX_ORDER_NR_PAGES in size).  This means that we can no longer make use of
the buddy alloctor guarentees for the contiguity of the mem_map, which
ensures that the mem_map is at least contigious for maximmally aligned
areas of MAX_ORDER_NR_PAGES pages.

This patch adds new mem_map accessors and iterator helpers which handle
any discontiguity at MAX_ORDER_NR_PAGES boundaries.  It then uses these to
implement gigantic page versions of copy_huge_page and clear_huge_page,
and to allow follow_hugetlb_page handle gigantic pages.

Signed-off-by: Andy Whitcroft <apw@shadowen.org>
Cc: Jon Tollefson <kniht@linux.vnet.ibm.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: <stable@kernel.org> [2.6.27.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agocciss: fix regression firmware not displayed in procfs
Mike Miller [Thu, 6 Nov 2008 20:53:25 +0000 (12:53 -0800)] 
cciss: fix regression firmware not displayed in procfs

This regression was introduced by commit
6ae5ce8e8d4de666f31286808d2285aa6a50fa40 ("cciss: remove redundant code").

This patch fixes a regression where the controller firmware version is not
displayed in procfs.  The previous patch would be called anytime something
changed.  This will get called only once for each controller.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Cc: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Jens Axboe <jens.axboe@oracle.com>
Cc: <stable@kernel.org> [2.6.27.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agocciss: fix sysfs broken symlink regression
Mike Miller [Thu, 6 Nov 2008 20:53:24 +0000 (12:53 -0800)] 
cciss: fix sysfs broken symlink regression

Regression introduced by commit 6ae5ce8e8d4de666f31286808d2285aa6a50fa40
("cciss: remove redundant code").

This patch fixes a broken symlink in sysfs that was introduced by the
above commit.  We broke it in 2.6.27-rc on or about 20080804.  Some
installers are broken if this symlink does not exist and they may not
detect the logical drives configured on the controller.  It does not
require being backported into 2.6.26.x or earlier kernels.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Cc: <stable@kernel.org> [2.6.27.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: collect version check return
Ian Kent [Thu, 6 Nov 2008 20:53:23 +0000 (12:53 -0800)] 
autofs4: collect version check return

The function check_dev_ioctl_version() returns an error code upon fail but
it isn't captured and returned in validate_dev_ioctl() as it should be.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Ian Kent <raven@themaw.net>
Signed-off-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoautofs4: correct offset mount expire check
Ian Kent [Thu, 6 Nov 2008 20:53:22 +0000 (12:53 -0800)] 
autofs4: correct offset mount expire check

When checking a directory tree in autofs_tree_busy() we can incorrectly
decide that the tree isn't busy.  This happens for the case of an active
offset mount as autofs4_follow_mount() follows past the active offset
mount, which has an open file handle used for expires, causing the file
handle not to count toward the busyness check.

Signed-off-by: Ian Kent <raven@themaw.net>
Signed-off-by: Jeff Moyer <jmoyer@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agohwmon: applesmc: add support for iMac 8
Henrik Rydberg [Thu, 6 Nov 2008 20:53:22 +0000 (12:53 -0800)] 
hwmon: applesmc: add support for iMac 8

Add temperature sensor support for iMac 8.

Signed-off-by: Henrik Rydberg <rydberg@euromail.se>
Tested-by: Klaus Doblmann <klaus.doblmann@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agohwmon: applesmc: add support for Macbook Pro 5
Henrik Rydberg [Thu, 6 Nov 2008 20:53:21 +0000 (12:53 -0800)] 
hwmon: applesmc: add support for Macbook Pro 5

Add accelerometer, backlight and temperature sensor support for the new
unibody Macbook Pro 5.

Signed-off-by: Henrik Rydberg <rydberg@euromail.se>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agohwmon: applesmc: add support for Macbook 5
Henrik Rydberg [Thu, 6 Nov 2008 20:53:20 +0000 (12:53 -0800)] 
hwmon: applesmc: add support for Macbook 5

Add accelerometer, backlight and temperature sensor support for the new
unibody Macbook 5.

Signed-off-by: Henrik Rydberg <rydberg@euromail.se>
Tested-by: David M. Lary <dmlary@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agohwmon: applesmc: add support for iMac 5
Henrik Rydberg [Thu, 6 Nov 2008 20:53:19 +0000 (12:53 -0800)] 
hwmon: applesmc: add support for iMac 5

Add temperature sensor support for iMac 5.

Signed-off-by: Henrik Rydberg <rydberg@euromail.se>
Tested-by: Ricky Campbell <johnrcampbell@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc: fix handling of missing tm_year data when reading alarms
Mark Brown [Thu, 6 Nov 2008 20:53:18 +0000 (12:53 -0800)] 
rtc: fix handling of missing tm_year data when reading alarms

When fixing up invalid years rtc_read_alarm() was calling rtc_valid_tm()
as a boolean but rtc_valid_tm() returns zero on success or a negative
number if the time is not valid so the test was inverted.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Alessandro Zummo <a.zummo@towertech.it>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agovt: incomplete initialization of vc_tab_stop
Wolfgang Kroworsch [Thu, 6 Nov 2008 20:53:16 +0000 (12:53 -0800)] 
vt: incomplete initialization of vc_tab_stop

Problem 1 (see patch below):
  vc_tab_stop is declared as an array of 8 unsigned ints in struct
  vc_data in include/linux/console_struct.h .
  In drivers/char/vt.c only 5 of these 8 unsigned ints get initialized
  leading to unintended tabulator placement on displays with more than
  160 columns text.

Problem 2 (open):
  Upcoming displays will have more than 256 columns of text leading to
  invalid memory access in drivers/char/vt.c during tabulator
  calculations:
    if (vc->vc_tab_stop[vc->vc_x >> 5] & (1 << (vc->vc_x & 31)))
break;

Signed-off-by: Wolfgang Kroworsch <wolfgang@kroworsch.de>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agocpqarry: fix return value of cpqarray_init()
Andrey Borzenkov [Thu, 6 Nov 2008 20:53:15 +0000 (12:53 -0800)] 
cpqarry: fix return value of cpqarray_init()

As reported by Dick Gevers on Compaq ProLiant:

Oct 13 18:06:51 dvgcpl kernel: Compaq SMART2 Driver (v 2.6.0)
Oct 13 18:06:51 dvgcpl kernel: sys_init_module: 'cpqarray'->init
suspiciously returned 1, it should follow 0/-E convention
Oct 13 18:06:51 dvgcpl kernel: sys_init_module: loading module anyway...
Oct 13 18:06:51 dvgcpl kernel: Pid: 315, comm: modprobe Not tainted
2.6.27-desktop-0.rc8.2mnb #1
Oct 13 18:06:51 dvgcpl kernel:  [<c0380612>] ? printk+0x18/0x1e
Oct 13 18:06:51 dvgcpl kernel:  [<c0158f85>] sys_init_module+0x155/0x1c0
Oct 13 18:06:51 dvgcpl kernel:  [<c0103f06>] syscall_call+0x7/0xb
Oct 13 18:06:51 dvgcpl kernel:  =======================

Make it return 0 on success and -ENODEV if no array was found.

Reported-by: Dick Gevers <dvgevers@xs4all.nl>
Signed-off-by: Andrey Borzenkov <arvidjaar@mail.ru>
Cc: Jens Axboe <jens.axboe@oracle.com>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agocciss: new hardware support
Mike Miller [Thu, 6 Nov 2008 20:53:14 +0000 (12:53 -0800)] 
cciss: new hardware support

Add support for 2 new SAS/SATA controllers.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Cc: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agonet: Fix recursive descent in __scm_destroy().
David Miller [Thu, 6 Nov 2008 08:37:40 +0000 (00:37 -0800)] 
net: Fix recursive descent in __scm_destroy().

__scm_destroy() walks the list of file descriptors in the scm_fp_list
pointed to by the scm_cookie argument.

Those, in turn, can close sockets and invoke __scm_destroy() again.

There is nothing which limits how deeply this can occur.

The idea for how to fix this is from Linus.  Basically, we do all of
the fput()s at the top level by collecting all of the scm_fp_list
objects hit by an fput().  Inside of the initial __scm_destroy() we
keep running the list until it is empty.

Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Tue, 4 Nov 2008 16:30:12 +0000 (08:30 -0800)] 
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6:
  xfrm: Fix xfrm_policy_gc_lock handling.
  niu: Use pci_ioremap_bar().
  bnx2x: Version Update
  bnx2x: Calling netif_carrier_off at the end of the probe
  bnx2x: PCI configuration bug on big-endian
  bnx2x: Removing the PMF indication when unloading
  mv643xx_eth: fix SMI bus access timeouts
  net: kconfig cleanup
  fs_enet: fix polling
  XFRM: copy_to_user_kmaddress() reports local address twice
  SMC91x: Fix compilation on some platforms.
  udp: Fix the SNMP counter of UDP_MIB_INERRORS
  udp: Fix the SNMP counter of UDP_MIB_INDATAGRAMS
  drivers/net/smc911x.c: Fix lockdep warning on xmit.

16 years agoMerge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzi...
Linus Torvalds [Tue, 4 Nov 2008 16:19:01 +0000 (08:19 -0800)] 
Merge branch 'upstream-linus' of git://git./linux/kernel/git/jgarzik/libata-dev

* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/libata-dev:
  libata: mask off DET when restoring SControl for detach
  libata: implement ATA_HORKAGE_ATAPI_MOD16_DMA and apply it
  libata: Fix a potential race condition in ata_scsi_park_show()
  sata_nv: fix generic, nf2/3 detection regression
  sata_via: restore vt*_prepare_host error handling
  sata_promise: add ATA engine reset to reset ops

16 years agodrivers: remove duplicated #include
Jianjun Kong [Tue, 4 Nov 2008 13:47:07 +0000 (21:47 +0800)] 
drivers: remove duplicated #include

Signed-off-by: Jianjun Kong <jianjun@zeuux.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agolibata: mask off DET when restoring SControl for detach
Tejun Heo [Mon, 3 Nov 2008 10:27:07 +0000 (19:27 +0900)] 
libata: mask off DET when restoring SControl for detach

libata restores SControl on detach; however, trying to restore
non-zero DET can cause undeterministic behavior including PMP device
going offline till power cycling.  Mask off DET when restoring
SControl.

Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agolibata: implement ATA_HORKAGE_ATAPI_MOD16_DMA and apply it
Tejun Heo [Mon, 3 Nov 2008 10:01:09 +0000 (19:01 +0900)] 
libata: implement ATA_HORKAGE_ATAPI_MOD16_DMA and apply it

libata always uses PIO for ATAPI commands when the number of bytes to
transfer isn't multiple of 16 but quantum DAT72 chokes on odd bytes
PIO transfers.  Implement a horkage to skip the mod16 check and apply
it to the quantum device.

This is reported by John Clark in the following thread.

  http://thread.gmane.org/gmane.linux.ide/34748

Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: John Clark <clarkjc@runbox.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agolibata: Fix a potential race condition in ata_scsi_park_show()
Elias Oltmanns [Mon, 3 Nov 2008 10:01:08 +0000 (19:01 +0900)] 
libata: Fix a potential race condition in ata_scsi_park_show()

Peter Moulder has pointed out that there is a slight chance that a
negative value might be passed to jiffies_to_msecs() in
ata_scsi_park_show(). This is fixed by saving the value of jiffies in a
local variable, thus also reducing code since the volatile variable
jiffies is accessed only once.

Signed-off-by: Elias Oltmanns <eo@nebensachen.de>
Signed-off-by: Tejun Heo <tj.kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agosata_nv: fix generic, nf2/3 detection regression
Tejun Heo [Mon, 3 Nov 2008 03:37:49 +0000 (12:37 +0900)] 
sata_nv: fix generic, nf2/3 detection regression

All three flavors of sata_nv's are different in how their hardreset
behaves.

* generic: Hardreset is not reliable.  Link often doesn't come online
  after hardreset.

* nf2/3: A little bit better - link comes online with longer debounce
  timing.  However, nf2/3 can't reliable wait for the first D2H
  Register FIS, so it can't wait for device readiness or classify the
  device after hardreset.  Follow-up SRST required.

* ck804: Hardreset finally works.

The core layer change to prefer hardreset and follow up changes
exposed the above issues and caused various detection regressions for
all three flavors.  This patch, hopefully, fixes all the known issues
and should make sata_nv error handling more reliable.

Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agosata_via: restore vt*_prepare_host error handling
Marcin Slusarz [Sun, 2 Nov 2008 21:18:52 +0000 (22:18 +0100)] 
sata_via: restore vt*_prepare_host error handling

commit b9d5b89b487517cbd4cb4702da829e07ef9e4432 (sata_via: fix support
for 5287) accidently (?) removed vt*_prepare_host error handling - restore it

catched by gcc:
drivers/ata/sata_via.c: In function 'svia_init_one':
drivers/ata/sata_via.c:567: warning: 'host' may be used uninitialized in this function

Signed-off-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Joseph Chan <JosephChan@via.com.tw>
Cc: Jeff Garzik <jgarzik@redhat.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agosata_promise: add ATA engine reset to reset ops
Mikael Pettersson [Fri, 31 Oct 2008 07:03:55 +0000 (08:03 +0100)] 
sata_promise: add ATA engine reset to reset ops

Promise ATA engines need to be reset when errors occur.
That's currently done for errors detected by sata_promise itself,
but it's not done for errors like timeouts detected outside of
the low-level driver.

The effect of this omission is that a timeout tends to result
in a sequence of failed COMRESETs after which libata EH gives
up and disables the port. At that point the port's ATA engine
hangs and even reloading the driver will not resume it.

To fix this, make sata_promise override ->hardreset on SATA
ports with code which calls pdc_reset_port() on the port in
question before calling libata's hardreset. PATA ports don't
use ->hardreset, so for those we override ->softreset instead.

Signed-off-by: Mikael Pettersson <mikpe@it.uu.se>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agoxfrm: Fix xfrm_policy_gc_lock handling.
Alexey Dobriyan [Tue, 4 Nov 2008 03:11:29 +0000 (19:11 -0800)] 
xfrm: Fix xfrm_policy_gc_lock handling.

From: Alexey Dobriyan <adobriyan@gmail.com>

Based upon a lockdep trace by Simon Arlott.

xfrm_policy_kill() can be called from both BH and
non-BH contexts, so we have to grab xfrm_policy_gc_lock
with BH disabling.

Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoniu: Use pci_ioremap_bar().
David S. Miller [Tue, 4 Nov 2008 01:05:16 +0000 (17:05 -0800)] 
niu: Use pci_ioremap_bar().

Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: Version Update
Eilon Greenstein [Tue, 4 Nov 2008 00:46:58 +0000 (16:46 -0800)] 
bnx2x: Version Update

Updating the version

Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: Calling netif_carrier_off at the end of the probe
Eilon Greenstein [Tue, 4 Nov 2008 00:46:40 +0000 (16:46 -0800)] 
bnx2x: Calling netif_carrier_off at the end of the probe

netif_carrier_off was called too early at the probe. In case of failure
or simply bad timing, this can cause a fatal error since linkwatch_event
might run too soon.

Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: PCI configuration bug on big-endian
Eilon Greenstein [Tue, 4 Nov 2008 00:46:19 +0000 (16:46 -0800)] 
bnx2x: PCI configuration bug on big-endian

The current code read nothing but zeros on big-endian (wrong part of the
32bits). This caused poor performance on big-endian machines. Though this
issue did not cause the system to crash, the performance is significantly
better with the fix so I view it as critical bug fix.

Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: Removing the PMF indication when unloading
Eilon Greenstein [Tue, 4 Nov 2008 00:45:55 +0000 (16:45 -0800)] 
bnx2x: Removing the PMF indication when unloading

When the PMF flag is set, the driver can access the HW freely. When the
driver is unloaded, it should not access the HW. The problem caused fatal
errors when "ethtool -i" was called after the calling instance was unloaded
and another instance was already loaded

Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agomv643xx_eth: fix SMI bus access timeouts
Lennert Buytenhek [Sat, 1 Nov 2008 05:32:20 +0000 (06:32 +0100)] 
mv643xx_eth: fix SMI bus access timeouts

The mv643xx_eth mii bus implementation uses wait_event_timeout() to
wait for SMI completion interrupts.

If wait_event_timeout() would return zero, mv643xx_eth would conclude
that the SMI access timed out, but this is not necessarily true --
wait_event_timeout() can also return zero in the case where the SMI
completion interrupt did happen in time but where it took longer than
the requested timeout for the process performing the SMI access to be
scheduled again.  This would lead to occasional SMI access timeouts
when the system would be under heavy load.

The fix is to ignore the return value of wait_event_timeout(), and
to re-check the SMI done bit after wait_event_timeout() returns to
determine whether or not the SMI access timed out.

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agonet: kconfig cleanup
Jeff Kirsher [Mon, 3 Nov 2008 04:30:33 +0000 (20:30 -0800)] 
net: kconfig cleanup

The bool kconfig option added to ixgbe and myri10ge for DCA is ambigous,
so this patch adds a description to the kconfig option.

Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agofs_enet: fix polling
Alexey Dobriyan [Sun, 2 Nov 2008 14:49:59 +0000 (17:49 +0300)] 
fs_enet: fix polling

1. compile fix for irqreturn_t type change
2. restore ->poll_controller after CONFIG_PPC_CPM_NEW_BINDING transition

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6
Linus Torvalds [Mon, 3 Nov 2008 19:43:59 +0000 (11:43 -0800)] 
Merge git://git./linux/kernel/git/sfrench/cifs-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6:
  cifs: fix renaming one hardlink on top of another
  [CIFS] fix error in smb_send2
  [CIFS] Reduce number of socket retries in large write path

16 years agocifs: fix renaming one hardlink on top of another
Jeff Layton [Mon, 3 Nov 2008 19:05:08 +0000 (14:05 -0500)] 
cifs: fix renaming one hardlink on top of another

cifs: fix renaming one hardlink on top of another

POSIX says that renaming one hardlink on top of another to the same
inode is a no-op. We had the logic mostly right, but forgot to clear
the return code.

Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>
16 years agoMerge branch 'tracing-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 3 Nov 2008 18:21:40 +0000 (10:21 -0800)] 
Merge branch 'tracing-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'tracing-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  tracing, ring-buffer: add paranoid checks for loops
  ftrace: use kretprobe trampoline name to test in output
  tracing, alpha: undefined reference to `save_stack_trace'

16 years agoMerge branch 'io-mappings-for-linus-2' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 3 Nov 2008 18:15:40 +0000 (10:15 -0800)] 
Merge branch 'io-mappings-for-linus-2' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'io-mappings-for-linus-2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  io mapping: clean up #ifdefs
  io mapping: improve documentation
  i915: use io-mapping interfaces instead of a variety of mapping kludges
  resources: add io-mapping functions to dynamically map large device apertures
  x86: add iomap_atomic*()/iounmap_atomic() on 32-bit using fixmaps

16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
Linus Torvalds [Mon, 3 Nov 2008 18:14:59 +0000 (10:14 -0800)] 
Merge branch 'for-linus' of git://git./linux/kernel/git/tiwai/sound-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6:
  ALSA: hda: make a STAC_DELL_EQ option
  ALSA: emu10k1 - Add more invert_shared_spdif flag to Audigy models
  ALSA: hda - Add a quirk for another Acer Aspire (1025:0090)
  ALSA: remove direct access of dev->bus_id in sound/isa/*
  sound: struct device - replace bus_id with dev_name(), dev_set_name()
  ALSA: Fix PIT lockup on some chipsets when using the PC-Speaker
  ALSA: rawmidi - Add open check in rawmidi callbacks
  ALSA: hda - Add digital-mic for ALC269 auto-probe mode
  ALSA: hda - Disable broken mic auto-muting in Realtek codes

16 years agoMerge branch 'proc-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/adobriyan...
Linus Torvalds [Mon, 3 Nov 2008 17:59:01 +0000 (09:59 -0800)] 
Merge branch 'proc-linus' of git://git./linux/kernel/git/adobriyan/proc

* 'proc-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/adobriyan/proc:
  proc: revert /proc/uptime to ->read_proc hook

16 years agoMerge branch 'drm-next' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Mon, 3 Nov 2008 17:58:40 +0000 (09:58 -0800)] 
Merge branch 'drm-next' of git://git./linux/kernel/git/airlied/drm-2.6

* 'drm-next' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6:
  i915: Add GEM ioctl to get available aperture size.
  drm/radeon: fixup further bus mastering confusion.
  build fix: CONFIG_DRM_I915=y && CONFIG_ACPI=n

16 years agoio mapping: clean up #ifdefs
Keith Packard [Mon, 3 Nov 2008 17:21:45 +0000 (18:21 +0100)] 
io mapping: clean up #ifdefs

Impact: cleanup

clean up ifdefs: change #ifdef CONFIG_X86_32/64 to
CONFIG_HAVE_ATOMIC_IOMAP.

flip around the #ifdef sections to clean up the structure.

Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agoio mapping: improve documentation
Keith Packard [Mon, 3 Nov 2008 17:21:44 +0000 (18:21 +0100)] 
io mapping: improve documentation

Impact: add documentation

Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agoMerge branches 'topic/fix/misc' and 'topic/fix/hda' into for-linus
Takashi Iwai [Mon, 3 Nov 2008 15:28:24 +0000 (16:28 +0100)] 
Merge branches 'topic/fix/misc' and 'topic/fix/hda' into for-linus

16 years agoALSA: hda: make a STAC_DELL_EQ option
Matthew Ranostay [Mon, 3 Nov 2008 13:12:43 +0000 (08:12 -0500)] 
ALSA: hda: make a STAC_DELL_EQ option

Add support for explicitly enabling the EQ distortion hack for
systems without software biquad support.

Signed-off-by: Matthew Ranostay <mranostay@embeddedalley.com>
Cc: stable@kernel.org
Signed-off-by: Takashi Iwai <tiwai@suse.de>
16 years agotracing, ring-buffer: add paranoid checks for loops
Steven Rostedt [Fri, 31 Oct 2008 13:58:35 +0000 (09:58 -0400)] 
tracing, ring-buffer: add paranoid checks for loops

While writing a new tracer, I had a bug where I caused the ring-buffer
to recurse in a bad way. The bug was with the tracer I was writing
and not the ring-buffer itself. But it took a long time to find the
problem.

This patch adds paranoid checks into the ring-buffer infrastructure
that will catch bugs of this nature.

Note: I put the bug back in the tracer and this patch showed the error
      nicely and prevented the lockup.

Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agoftrace: use kretprobe trampoline name to test in output
Steven Rostedt [Fri, 31 Oct 2008 19:44:07 +0000 (15:44 -0400)] 
ftrace: use kretprobe trampoline name to test in output

Impact: ia64+tracing build fix

When a function is kprobed, the return address is set to the
kprobe_trampoline, or something similar. This caused the output
of the trace to look confusing when the parent seemed to be this
"kprobe_trampoline" function.

To fix this, Abhishek Sagar added a test of the instruction pointer
of the parent to see if it matched the kprobe_trampoline. If it
did, the output would print a "[unknown/kretprobe'd]" instead.

Unfortunately, not all archs do this the same way, and the trampoline
function may not be exported, which causes failures in builds.

This patch will compare the name instead of the pointer to see
if it matches. This prevents us from depending on a function from
being exported, and should work on all archs. The worst that can
happen is that an arch might use a different name and then we
go back to the confusing output. At least the arch will still build.

Reported-by: Abhishek Sagar <sagar.abhishek@gmail.com>
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Tested-by: Abhishek Sagar <sagar.abhishek@gmail.com>
Acked-by: Abhishek Sagar <sagar.abhishek@gmail.com>
16 years agoXFRM: copy_to_user_kmaddress() reports local address twice
Arnaud Ebalard [Mon, 3 Nov 2008 09:30:23 +0000 (01:30 -0800)] 
XFRM: copy_to_user_kmaddress() reports local address twice

While adding support for MIGRATE/KMADDRESS in strongSwan (as specified
in draft-ebalard-mext-pfkey-enhanced-migrate-00), Andreas Steffen
noticed that XFRMA_KMADDRESS attribute passed to userland contains the
local address twice (remote provides local address instead of remote
one).

This bug in copy_to_user_kmaddress() affects only key managers that use
native XFRM interface (key managers that use PF_KEY are not affected).

For the record, the bug was in the initial changeset I posted which
added support for KMADDRESS (13c1d18931ebb5cf407cb348ef2cd6284d68902d
'xfrm: MIGRATE enhancements (draft-ebalard-mext-pfkey-enhanced-migrate)').

Signed-off-by: Arnaud Ebalard <arno@natisbad.org>
Reported-by: Andreas Steffen <andreas.steffen@strongswan.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoALSA: emu10k1 - Add more invert_shared_spdif flag to Audigy models
Takashi Iwai [Mon, 3 Nov 2008 09:21:36 +0000 (10:21 +0100)] 
ALSA: emu10k1 - Add more invert_shared_spdif flag to Audigy models

Reported in Novell bnc#440862:
    https://bugzilla.novell.com/show_bug.cgi?id=440862

Signed-off-by: Takashi Iwai <tiwai@suse.de>
16 years agotracing, alpha: undefined reference to `save_stack_trace'
Al Viro [Fri, 31 Oct 2008 19:50:41 +0000 (19:50 +0000)] 
tracing, alpha: undefined reference to `save_stack_trace'

Impact: build fix on !stacktrace architectures

only select STACKTRACE on architectures that have STACKTRACE_SUPPORT

... since we also need to ifdef out the guts of ftrace_trace_stack().
We also want to disallow setting TRACE_ITER_STACKTRACE in trace_flags
on such configs, but that can wait.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Acked-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agoALSA: hda - Add a quirk for another Acer Aspire (1025:0090)
Takashi Iwai [Mon, 3 Nov 2008 09:07:43 +0000 (10:07 +0100)] 
ALSA: hda - Add a quirk for another Acer Aspire (1025:0090)

Added a quirk for another Acer Aspier laptop (1025:0090) with ALC883
codec.  Reported in Novell bnc#426935:
    https://bugzilla.novell.com/show_bug.cgi?id=426935

Signed-off-by: Takashi Iwai <tiwai@suse.de>
16 years agoSMC91x: Fix compilation on some platforms.
David S. Miller [Mon, 3 Nov 2008 08:04:24 +0000 (00:04 -0800)] 
SMC91x: Fix compilation on some platforms.

This reverts 51ac3beffd4afaea4350526cf01fe74aaff25eff ('SMC91x: delete
unused local variable "lp"') and adds __maybe_unused markers to these
(potentially) unused variables.

The issue is that in some configurations SMC_IO_SHIFT evaluates
to '(lp->io_shift)', but in some others it's plain '0'.

Based upon a build failure report from Manuel Lauss.

Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoALSA: remove direct access of dev->bus_id in sound/isa/*
Takashi Iwai [Mon, 3 Nov 2008 07:51:33 +0000 (08:51 +0100)] 
ALSA: remove direct access of dev->bus_id in sound/isa/*

Removed the direct accesses of dev->bus_id in sound/isa/* by replacement
with dev_err() or dev_warn() functions.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
16 years agosound: struct device - replace bus_id with dev_name(), dev_set_name()
Kay Sievers [Sun, 2 Nov 2008 02:50:35 +0000 (03:50 +0100)] 
sound: struct device - replace bus_id with dev_name(), dev_set_name()

[stripped sound/isa/* changes, replaced with the next patch -- tiwai]

Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
16 years agoALSA: Fix PIT lockup on some chipsets when using the PC-Speaker
Zoltan Devai [Sun, 2 Nov 2008 23:30:28 +0000 (00:30 +0100)] 
ALSA: Fix PIT lockup on some chipsets when using the PC-Speaker

Fix PIT lockup on some chipsets when using the PC-Speaker.

Signed-off-by: Zoltan Devai <zdevai@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
16 years agoALSA: rawmidi - Add open check in rawmidi callbacks
Takashi Iwai [Mon, 3 Nov 2008 07:17:05 +0000 (08:17 +0100)] 
ALSA: rawmidi - Add open check in rawmidi callbacks

The drivers (e.g. mtpav) may call rawmidi functions in irq handlers
even though the streams are not opened.  This results in Oops or panic.

This patch adds the rawmidi state check before actually operating the
rawmidi buffers.

Tested-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
16 years agoudp: Fix the SNMP counter of UDP_MIB_INERRORS
Wei Yongjun [Sun, 2 Nov 2008 16:14:27 +0000 (16:14 +0000)] 
udp: Fix the SNMP counter of UDP_MIB_INERRORS

UDP packets received in udpv6_recvmsg() are not only IPv6 UDP packets, but
also have IPv4 UDP packets, so when do the counter of UDP_MIB_INERRORS in
udpv6_recvmsg(), we should check whether the packet is a IPv6 UDP packet
or a IPv4 UDP packet.

Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoudp: Fix the SNMP counter of UDP_MIB_INDATAGRAMS
Wei Yongjun [Sun, 2 Nov 2008 16:11:01 +0000 (16:11 +0000)] 
udp: Fix the SNMP counter of UDP_MIB_INDATAGRAMS

If UDP echo is sent to xinetd/echo-dgram, the UDP reply will be received
at the sender. But the SNMP counter of UDP_MIB_INDATAGRAMS will be not
increased, UDP6_MIB_INDATAGRAMS will be increased instead.

  Endpoint A                      Endpoint B
  UDP Echo request ----------->
  (IPv4, Dst port=7)
                   <----------    UDP Echo Reply
                                  (IPv4, Src port=7)

This bug is come from this patch cb75994ec311b2cd50e5205efdcc0696abd6675d.

It do counter UDP[6]_MIB_INDATAGRAMS until udp[v6]_recvmsg. Because
xinetd used IPv6 socket to receive UDP messages, thus, when received
UDP packet, the UDP6_MIB_INDATAGRAMS will be increased in function
udpv6_recvmsg() even if the packet is a IPv4 UDP packet.

This patch fixed the problem.

Signed-off-by: Wei Yongjun <yjwei@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoi915: Add GEM ioctl to get available aperture size.
Eric Anholt [Thu, 23 Oct 2008 04:40:13 +0000 (21:40 -0700)] 
i915: Add GEM ioctl to get available aperture size.

This will let userland know when to submit its batchbuffers, before they get
too big to fit in the aperture.

Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
16 years agoLinux v2.6.28-rc3 v2.6.28-rc3
Linus Torvalds [Sun, 2 Nov 2008 22:17:19 +0000 (14:17 -0800)] 
Linux v2.6.28-rc3

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/bart/ide-2.6
Linus Torvalds [Sun, 2 Nov 2008 21:32:26 +0000 (13:32 -0800)] 
Merge git://git./linux/kernel/git/bart/ide-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/bart/ide-2.6:
  ide-gd: re-get capacity on revalidate
  tx4938ide: Avoid underflow on calculation of a wait cycle
  tx4938ide: Do not call devm_ioremap for whole 128KB
  tx4938ide: Check minimum cycle time and SHWT range (v2)
  ide: Switch to a common address
  ide-cd: fix DMA alignment regression

16 years agoide-gd: re-get capacity on revalidate
Borislav Petkov [Sun, 2 Nov 2008 20:40:10 +0000 (21:40 +0100)] 
ide-gd: re-get capacity on revalidate

We need to re-get a removable media's capacity when revalidating the
disk so that its partitions get rescanned by the block layer.

Signed-off-by: Borislav Petkov <petkovbb@gmail.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: axboe@kernel.dk
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agotx4938ide: Avoid underflow on calculation of a wait cycle
Atsushi Nemoto [Sun, 2 Nov 2008 20:40:10 +0000 (21:40 +0100)] 
tx4938ide: Avoid underflow on calculation of a wait cycle

Make 'wt' variable signed while it can be negative during calculation.

Suggested-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Cc: sshtylyov@ru.mvista.com
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
16 years agotx4938ide: Do not call devm_ioremap for whole 128KB
Atsushi Nemoto [Sun, 2 Nov 2008 20:40:09 +0000 (21:40 +0100)] 
tx4938ide: Do not call devm_ioremap for whole 128KB

Call devm_ioremap() for CS0 and CS1 separetely.
And some style cleanups.

Suggested-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Cc: ralf@linux-mips.org
Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>