]> git.proxmox.com Git - mirror_ubuntu-disco-kernel.git/log
mirror_ubuntu-disco-kernel.git
14 years agox86, UV: Make kdump avoid stack dumps
Cliff Wickman [Tue, 20 Jul 2010 23:09:05 +0000 (18:09 -0500)]
x86, UV: Make kdump avoid stack dumps

UV NMI callback's should not write stack dumps when a kdump is to be written.

When invoking the crash kernel to write a dump, kdump_nmi_shootdown_cpus()
uses NMI's to get all the cpu's to save their register context and halt.

But the NMI interrupt handler runs a callback list.  This patch sets a flag
to prevent any of those callbacks from interfering with the halt of the cpu.

For UV, which currently has the only callback to which this is relevant, the
uv_handle_nmi() callback should not do dumping of stacks.

The 'in_crash_kexec' flag is defined as an extern in kdebug.h firstly
because x2apic_uv_x.c includes it.  Secondly because some future callback
might need the flag to know that it should not enter the debugger.
(Such a scenario was in fact present in the 2.6.32 kernel, SuSE distribution,
 where a call to kdb needed to be avoided.)

Signed-off-by: Cliff Wickman <cpw@sgi.com>
LKML-Reference: <E1ObLvt-0005UZ-Va@eag09.americas.sgi.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
14 years agox86, UV: Initialize BAU MMRs only on hubs with cpus
Cliff Wickman [Fri, 16 Jul 2010 15:11:21 +0000 (10:11 -0500)]
x86, UV: Initialize BAU MMRs only on hubs with cpus

Remove the initialization of MMRs
UVH_LB_BAU_SB_ACTIVATION_CONTROL and UVH_BAU_DATA_BROADCAST on
UV hubs that have no active cpus. Such initialization on hubs
with no active cpus would result in a kernel page fault.

This is not of real high priority, because we don't have any
such systems (with UV hubs that have no active cpus).  But they
will be coming.

Signed-off-by: Cliff Wickman <cpw@sgi.com>
LKML-Reference: <E1OZmZN-0006cW-RC@eag09.americas.sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agox86, UV: Modularize BAU send and wait
Cliff Wickman [Wed, 2 Jun 2010 21:22:02 +0000 (16:22 -0500)]
x86, UV: Modularize BAU send and wait

Streamline the large uv_flush_send_and_wait() function by use of
a couple of helper functions.

And remove some excess comments.

Signed-off-by: Cliff Wickman <cpw@sgi.com>
Cc: gregkh@suse.de
LKML-Reference: <E1OJvNy-0004ay-IH@eag09.americas.sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agox86, UV: BAU broadcast to the local hub
Cliff Wickman [Wed, 2 Jun 2010 21:22:02 +0000 (16:22 -0500)]
x86, UV: BAU broadcast to the local hub

Make the Broadcast Assist Unit driver use the BAU for TLB
shootdowns of cpu's on the local uvhub.

It was previously thought that IPI might be faster to the cpu's
on the local hub.  But the IPI operation would have to follow
the completion of the BAU broadcast anyway.  So we broadcast to
the local uvhub in all cases except when the current cpu was the
only local cpu in the mask.

This simplifies uv_flush_send_and_wait() in that it returns
either all shootdowns complete, or none.

Adjust the statistics to account for shootdowns on the local
uvhub.

Signed-off-by: Cliff Wickman <cpw@sgi.com>
Cc: gregkh@suse.de
LKML-Reference: <E1OJvNy-0004aq-G7@eag09.americas.sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agox86, UV: Correct BAU regular message type
Cliff Wickman [Wed, 2 Jun 2010 21:22:02 +0000 (16:22 -0500)]
x86, UV: Correct BAU regular message type

The Broadcast Assist Unit messages have a regular or retry
message type. The regular type was not being set, but needs to
be, because the lack of a message type is sometimes used to
identify an unused entry in the message queue.
Also removing some excess comments.

Signed-off-by: Cliff Wickman <cpw@sgi.com>
Cc: gregkh@suse.de
LKML-Reference: <E1OJvNy-0004ak-Dy@eag09.americas.sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agox86, UV: Remove BAU check for stay-busy
Cliff Wickman [Wed, 2 Jun 2010 21:22:02 +0000 (16:22 -0500)]
x86, UV: Remove BAU check for stay-busy

Remove a faulty assumption that a long running BAU request has
encountered a hardware problem and will never finish.

Numalink congestion can make a request appear to have
encountered such a problem, but it is not safe to cancel the
request.  If such a cancel is done but a reply is later received
we can miss a TLB shootdown.

We depend upon the max_bau_concurrent 'throttle' to prevent the
stay-busy case from happening.

Signed-off-by: Cliff Wickman <cpw@sgi.com>
Cc: gregkh@suse.de
LKML-Reference: <E1OJvNy-0004ad-BV@eag09.americas.sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agox86, UV: Correct BAU discovery of hubs and sockets
Cliff Wickman [Wed, 2 Jun 2010 21:22:02 +0000 (16:22 -0500)]
x86, UV: Correct BAU discovery of hubs and sockets

Correct the initialization-time assumption of contigous blade
numbers and of sockets numbered from zero.

There may be hubs present with no cpu's enabled.
There may be disabled sockets such that the active socket is not
number zero.

And assign a 'socket master' by assuming that a socket is a
node. (it is not safe to extract socket number from an apicid)

Signed-off-by: Cliff Wickman <cpw@sgi.com>
Cc: gregkh@suse.de
LKML-Reference: <E1OJvNy-0004aW-9S@eag09.americas.sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agox86, UV: Correct BAU software acknowledge
Cliff Wickman [Wed, 2 Jun 2010 21:22:02 +0000 (16:22 -0500)]
x86, UV: Correct BAU software acknowledge

Correct the acknowledgment and the reset of a BAU
software-acknowledged message.

A retry message should be testing only for timed-out resources
(mask << 8). (And we delete a log message that might cause
unnecessary concern) The acknowledge MMR is
|--timed-out--|---pending--|,  each is 8 bits.

The IPI-driven reset of software acknowledge resources frees
both timed out and pending resources.

Signed-off-by: Cliff Wickman <cpw@sgi.com>
Cc: gregkh@suse.de
LKML-Reference: <E1OJvNy-0004aP-7O@eag09.americas.sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agox86, UV: BAU structure rearranging
Cliff Wickman [Wed, 2 Jun 2010 21:22:02 +0000 (16:22 -0500)]
x86, UV: BAU structure rearranging

Move some structure definitions from the C code to the BAU
header file, and change the organization of that header file a
little.

Signed-off-by: Cliff Wickman <cpw@sgi.com>
Cc: gregkh@suse.de
LKML-Reference: <E1OJvNy-0004aI-54@eag09.americas.sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agox86, UV: Shorten access to BAU statistics structure
Cliff Wickman [Wed, 2 Jun 2010 21:22:02 +0000 (16:22 -0500)]
x86, UV: Shorten access to BAU statistics structure

Use a pointer from the per-cpu BAU control structure to the
per-cpu BAU statistics structure.
We nearly always know the first before needing the second.

Signed-off-by: Cliff Wickman <cpw@sgi.com>
Cc: gregkh@suse.de
LKML-Reference: <E1OJvNy-0004aB-2k@eag09.americas.sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agox86, UV: Disable BAU on network congestion
Cliff Wickman [Wed, 2 Jun 2010 21:22:02 +0000 (16:22 -0500)]
x86, UV: Disable BAU on network congestion

The numalink network can become so congested that TLB shootdown
using the Broadcast Assist Unit becomes slower than using IPI's.

In that case, disable the use of the BAU for a period of time.
The period is tunable.  When the period expires the use of the
BAU is re-enabled. A count of these actions is added to the
statistics file.

Signed-off-by: Cliff Wickman <cpw@sgi.com>
Cc: gregkh@suse.de
LKML-Reference: <E1OJvNy-0004a4-0a@eag09.americas.sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agox86, UV: BAU tunables into a debugfs file
Cliff Wickman [Wed, 2 Jun 2010 21:22:01 +0000 (16:22 -0500)]
x86, UV: BAU tunables into a debugfs file

Make the Broadcast Assist Unit driver's nine tuning values variable by
making them accessible through a read/write debugfs file.

The file will normally be mounted as
/sys/kernel/debug/sgi_uv/bau_tunables. The tunables are kept in each
cpu's per-cpu BAU structure.

The patch also does a little name improvement, and corrects the reset of
two destination timeout counters.

Signed-off-by: Cliff Wickman <cpw@sgi.com>
Cc: gregkh@suse.de
LKML-Reference: <E1OJvNx-0004Zx-Uo@eag09.americas.sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agox86, UV: Calculate BAU destination timeout
Cliff Wickman [Wed, 2 Jun 2010 21:22:01 +0000 (16:22 -0500)]
x86, UV: Calculate BAU destination timeout

Calculate the Broadcast Assist Unit's destination timeout period from the
values in the relevant MMR's.

Store it in each cpu's per-cpu BAU structure so that a destination
timeout can be differentiated from a 'plugged' situation in which all
software ack resources are already allocated and a timeout is pending.
That case returns an immediate destination error.

Signed-off-by: Cliff Wickman <cpw@sgi.com>
Cc: gregkh@suse.de
LKML-Reference: <E1OJvNx-0004Zq-RK@eag09.americas.sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agoMerge git://git.infradead.org/~dwmw2/mtd-2.6.35
Linus Torvalds [Tue, 8 Jun 2010 00:10:06 +0000 (17:10 -0700)]
Merge git://git.infradead.org/~dwmw2/mtd-2.6.35

* git://git.infradead.org/~dwmw2/mtd-2.6.35:
  jffs2: update ctime when changing the file's permission by setfacl
  jffs2: Fix NFS race by using insert_inode_locked()
  jffs2: Fix in-core inode leaks on error paths
  mtd: Fix NAND submenu
  mtd/r852: update card detect early.
  mtd/r852: Fixes in case of DMA timeout
  mtd/r852: register IRQ as last step
  drivers/mtd: Use memdup_user
  docbook: make mtd nand module init static

14 years agoMerge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzi...
Linus Torvalds [Tue, 8 Jun 2010 00:09:03 +0000 (17:09 -0700)]
Merge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/libata-dev

* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/libata-dev:
  ahci: redo stopping DMA engines on empty ports
  sata_sil24: fix kernel panic on ARM caused by unaligned access in sata_sil24
  ahci: add pci quirk for JMB362
  sata_via: explain the magic fix

14 years agoahci: redo stopping DMA engines on empty ports
Tejun Heo [Mon, 7 Jun 2010 13:15:08 +0000 (15:15 +0200)]
ahci: redo stopping DMA engines on empty ports

Commit 96d60303fd (ahci: Turn off DMA engines when there's no device)
implemented stopping DMA engines on empty ports but it used single
sampling of status registers to determine device presence which led to
disabling of DMA engines on occupied ports.  Do it after all EH
actions are complete using device presence state determined by EH.
This avoids spurious disabling of DMA engines and simplifies the code.

Signed-off-by: Tejun Heo <tj@kernel.org>
Tested-by: Marc Dionne <marc.c.dionne@gmail.com>
Cc: Matthew Garrett <mjg@redhat.com>
Cc: Robert Hancock <hancockrwd@gmail.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
14 years agosata_sil24: fix kernel panic on ARM caused by unaligned access in sata_sil24
Colin Tuckley [Fri, 4 Jun 2010 14:19:51 +0000 (16:19 +0200)]
sata_sil24: fix kernel panic on ARM caused by unaligned access in sata_sil24

The sata_sil24 driver has six 16-bit registers that are initialised with
32-bit writes. This cause a kernel panic on ARM due to the unaligned
accesses which result.

This patch changes the accesses to the correct 16-bit ones.

Signed-off-by: Colin Tuckley <colin.tuckley@arm.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
14 years agoahci: add pci quirk for JMB362
Tejun Heo [Thu, 3 Jun 2010 09:57:04 +0000 (11:57 +0200)]
ahci: add pci quirk for JMB362

JMB362 is a new variant of jmicron controller which is similar to
JMB360 but has two SATA ports instead of one.  As there is no PATA
port, single function AHCI mode can be used as in JMB360.  Add pci
quirk for JMB362.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Aries Lee <arieslee@jmicron.com>
Cc: stable@kernel.org
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
14 years agosata_via: explain the magic fix
Tejun Heo [Thu, 3 Jun 2010 09:35:03 +0000 (11:35 +0200)]
sata_via: explain the magic fix

Add Joseph Chan's explanation of the problem and workaround to the
VT6421 magic fix.

Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Joseph Chan <JosephChan@via.com.tw>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
14 years ago[PATCH 2/11] drivers/watchdog: Eliminate a NULL pointer dereference
Julia Lawall [Thu, 27 May 2010 12:32:24 +0000 (14:32 +0200)]
[PATCH 2/11] drivers/watchdog: Eliminate a NULL pointer dereference

At the point of the call to dev_err, wm8350 is NULL.

A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)

// <smpl>
@r exists@
expression E,E1;
identifier f;
statement S1,S2,S3;
@@

if ((E == NULL && ...) || ...)
{
  ... when != if (...) S1 else S2
      when != E = E1
* E->f
  ... when any
  return ...;
}
else S3
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
14 years agoRevert "tty: fix a little bug in scrup, vt.c"
Linus Torvalds [Mon, 7 Jun 2010 03:44:04 +0000 (20:44 -0700)]
Revert "tty: fix a little bug in scrup, vt.c"

This reverts commit 962400e8fd29981a7b166e463dd143b6ac6a3e76, which was
entirely bogus.

The code used to multiply the character offset by "vc->vc_cols", and
that's actually correct, because 'd' itself is an 'unsigned short'.  So
the pointer arithmetic already takes the size of a VGA character into
account.  Changing it to use vc_size_row (which is just "vc_cols"
shifted up to take the size of the character into account) ends up
multiplying with the VGA character size twice.

This got reported as bugs for various other subsystems, because what it
actually results in is writing the 16-bit vc_video_erase_char pattern
(usually 0x0720: 0x07 is the default attribute, 0x20 is ASCII space)
into some random other allocation.

So Markus ended up reporting this as a ext4 bug, while to Torsten Kaiser
it looked like a problem with KMS or libata.  Jeff Chua saw it in
different places.

And finally - Justin Mattock had slab poisoning enabled, and saw it as a
slab poison overwritten.  And bisected and reverted this to verify the
buggy commit.

Reported-by: Markus Trippelsdorf <markus@trippelsdorf.de>
Reported-by: Torsten Kaiser <just.for.lkml@googlemail.com>
Reported-by: Jeff Chua <jeff.chua.linux@gmail.com>
Reported-by: Justin P. Mattock <justinmattock@gmail.com>
Reported-bisected-and-tested-by: Justin P. Mattock <justinmattock@gmail.com>
Acked-by: Dave Airlie <airlied@redhat.com>
Cc: Frank Pan <frankpzh@gmail.com>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agojffs2: update ctime when changing the file's permission by setfacl
Jan Kara [Fri, 4 Jun 2010 15:07:55 +0000 (17:07 +0200)]
jffs2: update ctime when changing the file's permission by setfacl

jffs2 didn't update the ctime of the file when its permission was changed.

Steps to reproduce:
 # touch aaa
 # stat -c %Z aaa
 1275289822
 # setfacl -m  'u::x,g::x,o::x' aaa
 # stat -c %Z aaa
 1275289822                         <- unchanged

But, according to the spec of the ctime, jffs2 must update it.

Port of ext3 patch by Miao Xie <miaox@cn.fujitsu.com>.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
14 years agoLinux 2.6.35-rc2
Linus Torvalds [Sun, 6 Jun 2010 03:43:24 +0000 (20:43 -0700)]
Linux 2.6.35-rc2

14 years agodrm/i915: Move non-phys cursors into the GTT
Chris Wilson [Wed, 2 Jun 2010 07:30:48 +0000 (08:30 +0100)]
drm/i915: Move non-phys cursors into the GTT

Cursors need to be in the GTT domain when being accessed by the GPU.
Previously this was a fortuitous byproduct of userspace using pwrite()
to upload the image data into the cursor. The redundant clflush was
removed in commit 9b8c4a and so the image was no longer being flushed
out of the caches into main memory. One could also devise a scenario
where the cursor was rendered by the GPU, prior to being attached as the
cursor, resulting in similar corruption due to the missing MI_FLUSH.

Fixes:

  Bug 28335 - Cursor corruption caused by commit 9b8c4a0b21
  https://bugs.freedesktop.org/show_bug.cgi?id=28335

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reported-and-tested-by: Jeff Chua <jeff.chua.linux@gmail.com>
Tested-by: Linus Torvalds <torvalds@linux-foundation.org>
Reported-by: Andy Isaacson <adi@hexapodia.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4
Linus Torvalds [Sat, 5 Jun 2010 17:07:25 +0000 (10:07 -0700)]
Merge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
  ext4: Fix remaining racy updates of EXT4_I(inode)->i_flags
  ext4: Make sure the MOVE_EXT ioctl can't overwrite append-only files

14 years agoext4: Fix remaining racy updates of EXT4_I(inode)->i_flags
Dmitry Monakhov [Sat, 5 Jun 2010 15:51:27 +0000 (11:51 -0400)]
ext4: Fix remaining racy updates of EXT4_I(inode)->i_flags

A few functions were still modifying i_flags in a racy manner.

Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
14 years agoMerge branch 'for-linus' of git://oss.sgi.com/xfs/xfs
Linus Torvalds [Sat, 5 Jun 2010 14:33:05 +0000 (07:33 -0700)]
Merge branch 'for-linus' of git://oss.sgi.com/xfs/xfs

* 'for-linus' of git://oss.sgi.com/xfs/xfs:
  xfs: improve xfs_isilocked
  xfs: skip writeback from reclaim context
  xfs: remove done roadmap item from xfs-delayed-logging-design.txt
  xfs: fix race in inode cluster freeing failing to stale inodes
  xfs: fix access to upper inodes without inode64
  xfs: fix might_sleep() warning when initialising per-ag tree
  fs/xfs/quota: Add missing mutex_unlock
  xfs: remove duplicated #include
  xfs: convert more trace events to DEFINE_EVENT
  xfs: xfs_trace.c: remove duplicated #include
  xfs: Check new inode size is OK before preallocating
  xfs: clean up xlog_align
  xfs: cleanup log reservation calculactions
  xfs: be more explicit if RT mount fails due to config
  xfs: replace E2BIG with EFBIG where appropriate

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Sat, 5 Jun 2010 14:32:21 +0000 (07:32 -0700)]
Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (30 commits)
  X25: remove duplicated #include
  tcp: use correct net ns in cookie_v4_check()
  rps: tcp: fix rps_sock_flow_table table updates
  ppp_generic: fix multilink fragment sizes
  syncookies: remove Kconfig text line about disabled-by-default
  ixgbe: only check pfc bits in hang logic if pfc is enabled
  net: check for refcount if pop a stacked dst_entry
  ixgbe: return IXGBE_ERR_RAR_INDEX when out of range
  act_pedit: access skb->data safely
  sfc: Store port number in net_device::dev_id
  epic100: Test __BIG_ENDIAN instead of (non-existent) CONFIG_BIG_ENDIAN
  tehuti: return -EFAULT on copy_to_user errors
  isdn/kcapi: return -EFAULT on copy_from_user errors
  e1000e: change logical negate to bitwise
  sfc: Get port number from CS_PORT_NUM, not PCI function number
  cls_u32: use skb_header_pointer() to dereference data safely
  TCP: tcp_hybla: Fix integer overflow in slow start increment
  act_nat: fix the wrong checksum when addr isn't in old_addr/mask
  net/fec: fix pm to survive to suspend/resume
  korina: count RX DMA OVR as rx_fifo_error
  ...

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ryusuke...
Linus Torvalds [Sat, 5 Jun 2010 14:31:13 +0000 (07:31 -0700)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ryusuke/nilfs2

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ryusuke/nilfs2:
  nilfs2: remove obsolete declarations of cache constructor and destructor
  nilfs2: fix style issue in nilfs_destroy_cachep

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6
Linus Torvalds [Sat, 5 Jun 2010 04:12:39 +0000 (21:12 -0700)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6:
  Minix: Clean up left over label
  fix truncate inode time modification breakage
  fix setattr error handling in sysfs, configfs
  fcntl: return -EFAULT if copy_to_user fails
  wrong type for 'magic' argument in simple_fill_super()
  fix the deadlock in qib_fs
  mqueue doesn't need make_bad_inode()

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-for-linus
Linus Torvalds [Sat, 5 Jun 2010 04:09:48 +0000 (21:09 -0700)]
Merge git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-for-linus

* git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-for-linus:
  module: fix bne2 "gave up waiting for init of module libcrc32c"
  module: verify_export_symbols under the lock
  module: move find_module check to end
  module: make locking more fine-grained.
  module: Make module sysfs functions private.
  module: move sysfs exposure to end of load_module
  module: fix kdb's illicit use of struct module_use.
  module: Make the 'usage' lists be two-way

14 years agomodule: fix bne2 "gave up waiting for init of module libcrc32c"
Rusty Russell [Sat, 5 Jun 2010 17:17:37 +0000 (11:17 -0600)]
module: fix bne2 "gave up waiting for init of module libcrc32c"

Problem: it's hard to avoid an init routine stumbling over a
request_module these days.  And it's not clear it's always a bad idea:
for example, a module like kvm with dynamic dependencies on kvm-intel
or kvm-amd would be neater if it could simply request_module the right
one.

In this particular case, it's libcrc32c:

libcrc32c_mod_init
 crypto_alloc_shash
  crypto_alloc_tfm
   crypto_find_alg
    crypto_alg_mod_lookup
     crypto_larval_lookup
      request_module

If another module is waiting inside resolve_symbol() for libcrc32c to
finish initializing (ie. bne2 depends on libcrc32c) then it does so
holding the module lock, and our request_module() can't make progress
until that is released.

Waiting inside resolve_symbol() without the lock isn't all that hard:
we just need to pass the -EBUSY up the call chain so we can sleep
where we don't hold the lock.  Error reporting is a bit trickier: we
need to copy the name of the unfinished module before releasing the
lock.

Other notes:
1) This also fixes a theoretical issue where a weak dependency would allow
   symbol version mismatches to be ignored.
2) We rename use_module to ref_module to make life easier for the only
   external user (the out-of-tree ksplice patches).

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Tim Abbot <tabbott@ksplice.com>
Tested-by: Brandon Philips <bphilips@suse.de>
14 years agomodule: verify_export_symbols under the lock
Rusty Russell [Sat, 5 Jun 2010 17:17:37 +0000 (11:17 -0600)]
module: verify_export_symbols under the lock

It disabled preempt so it was "safe", but nothing stops another module
slipping in before this module is added to the global list now we don't
hold the lock the whole time.

So we check this just after we check for duplicate modules, and just
before we put the module in the global list.

(find_symbol finds symbols in coming and going modules, too).

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
14 years agomodule: move find_module check to end
Linus Torvalds [Sat, 5 Jun 2010 17:17:36 +0000 (11:17 -0600)]
module: move find_module check to end

I think Rusty may have made the lock a bit _too_ finegrained there, and
didn't add it to some places that needed it. It looks, for example, like
PATCH 1/2 actually drops the lock in places where it's needed
("find_module()" is documented to need it, but now load_module() didn't
hold it at all when it did the find_module()).

Rather than adding a new "module_loading" list, I think we should be able
to just use the existing "modules" list, and just fix up the locking a
bit.

In fact, maybe we could just move the "look up existing module" a bit
later - optimistically assuming that the module doesn't exist, and then
just undoing the work if it turns out that we were wrong, just before
adding ourselves to the list.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
14 years agomodule: make locking more fine-grained.
Rusty Russell [Sat, 5 Jun 2010 17:17:36 +0000 (11:17 -0600)]
module: make locking more fine-grained.

Kay Sievers <kay.sievers@vrfy.org> reports that we still have some
contention over module loading which is slowing boot.

Linus also disliked a previous "drop lock and regrab" patch to fix the
bne2 "gave up waiting for init of module libcrc32c" message.

This is more ambitious: we only grab the lock where we need it.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Brandon Philips <brandon@ifup.org>
Cc: Kay Sievers <kay.sievers@vrfy.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
14 years agomodule: Make module sysfs functions private.
Rusty Russell [Sat, 5 Jun 2010 17:17:36 +0000 (11:17 -0600)]
module: Make module sysfs functions private.

These were placed in the header in ef665c1a06 to get the various
SYSFS/MODULE config combintations to compile.

That may have been necessary then, but it's not now.  These functions
are all local to module.c.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Randy Dunlap <randy.dunlap@oracle.com>
14 years agomodule: move sysfs exposure to end of load_module
Rusty Russell [Sat, 5 Jun 2010 17:17:36 +0000 (11:17 -0600)]
module: move sysfs exposure to end of load_module

This means a little extra work, but is more logical: we don't put
anything in sysfs until we're about to put the module into the
global list an parse its parameters.

This also gives us a logical place to put duplicate module detection
in the next patch.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
14 years agomodule: fix kdb's illicit use of struct module_use.
Rusty Russell [Sat, 5 Jun 2010 17:17:35 +0000 (11:17 -0600)]
module: fix kdb's illicit use of struct module_use.

Linus changed the structure, and luckily this didn't compile any more.

Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Jason Wessel <jason.wessel@windriver.com>
Cc: Martin Hicks <mort@sgi.com>
14 years agomodule: Make the 'usage' lists be two-way
Linus Torvalds [Mon, 31 May 2010 19:19:37 +0000 (12:19 -0700)]
module: Make the 'usage' lists be two-way

When adding a module that depends on another one, we used to create a
one-way list of "modules_which_use_me", so that module unloading could
see who needs a module.

It's actually quite simple to make that list go both ways: so that we
not only can see "who uses me", but also see a list of modules that are
"used by me".

In fact, we always wanted that list in "module_unload_free()": when we
unload a module, we want to also release all the other modules that are
used by that module.  But because we didn't have that list, we used to
first iterate over all modules, and then iterate over each "used by me"
list of that module.

By making the list two-way, we simplify module_unload_free(), and it
allows for some trivial fixes later too.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> (cleaned & rebased)
14 years agoX25: remove duplicated #include
Huang Weiyi [Fri, 4 Jun 2010 23:14:15 +0000 (16:14 -0700)]
X25: remove duplicated #include

Remove duplicated #include('s) in drivers/net/wan/x25_asy.c

Signed-off-by: Huang Weiyi <weiyi.huang@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agotcp: use correct net ns in cookie_v4_check()
Eric Dumazet [Thu, 3 Jun 2010 05:45:47 +0000 (05:45 +0000)]
tcp: use correct net ns in cookie_v4_check()

Its better to make a route lookup in appropriate namespace.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agorps: tcp: fix rps_sock_flow_table table updates
Eric Dumazet [Thu, 3 Jun 2010 09:03:58 +0000 (09:03 +0000)]
rps: tcp: fix rps_sock_flow_table table updates

I believe a moderate SYN flood attack can corrupt RFS flow table
(rps_sock_flow_table), making RPS/RFS much less effective.

Even in a normal situation, server handling short lived sessions suffer
from bad steering for the first data packet of a session, if another SYN
packet is received for another session.

We do following action in tcp_v4_rcv() :

sock_rps_save_rxhash(sk, skb->rxhash);

We should _not_ do this if sk is a LISTEN socket, as about each
packet received on a LISTEN socket has a different rxhash than
previous one.
 -> RPS_NO_CPU markers are spread all over rps_sock_flow_table.

Also, it makes sense to protect sk->rxhash field changes with socket
lock (We currently can change it even if user thread owns the lock
and might use rxhash)

This patch moves sock_rps_save_rxhash() to a sock locked section,
and only for non LISTEN sockets.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoppp_generic: fix multilink fragment sizes
Ben McKeegan [Wed, 2 Jun 2010 23:14:33 +0000 (23:14 +0000)]
ppp_generic: fix multilink fragment sizes

Fix bug in multilink fragment size calculation introduced by
commit 9c705260feea6ae329bc6b6d5f6d2ef0227eda0a
"ppp: ppp_mp_explode() redesign"

Signed-off-by: Ben McKeegan <ben@netservers.co.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosyncookies: remove Kconfig text line about disabled-by-default
Florian Westphal [Thu, 3 Jun 2010 00:42:30 +0000 (00:42 +0000)]
syncookies: remove Kconfig text line about disabled-by-default

syncookies default to on since
e994b7c901ded7200b525a707c6da71f2cf6d4bb
(tcp: Don't make syn cookies initial setting depend on CONFIG_SYSCTL).

Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoixgbe: only check pfc bits in hang logic if pfc is enabled
John Fastabend [Thu, 3 Jun 2010 17:03:45 +0000 (17:03 +0000)]
ixgbe: only check pfc bits in hang logic if pfc is enabled

Only check pfc bits in hang logic if PFC is enabled.  Previously,
if DCB was enabled but PFC was disabled the incorrect pause
bits would be checked.

Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Acked-by: Don Skidmore <donald.c.skidmore@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet: check for refcount if pop a stacked dst_entry
Steffen Klassert [Fri, 4 Jun 2010 01:57:38 +0000 (01:57 +0000)]
net: check for refcount if pop a stacked dst_entry

xfrm triggers a warning if dst_pop() drops a refcount
on a noref dst. This patch changes dst_pop() to
skb_dst_pop(). skb_dst_pop() drops the refcnt only
on a refcounted dst. Also we don't clone the child
dst_entry, so it is not refcounted and we can use
skb_dst_set_noref() in xfrm_output_one().

Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Linus Torvalds [Fri, 4 Jun 2010 22:42:30 +0000 (15:42 -0700)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: wacom - add Cintiq 21UX2 and Intuos4 WL
  Input: ads7846 - fix compiler warning in ads7846_probe()
  Input: tps6507x-ts - a couple work queue cleanups
  Input: s3c2410_ts - tone down logging
  Input: s3c2410_ts - fix build error due to ADC Kconfig rename

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6
Linus Torvalds [Fri, 4 Jun 2010 22:42:09 +0000 (15:42 -0700)]
Merge git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6: (23 commits)
  sh: Make intc messages consistent via pr_fmt.
  sh: make sure static declaration on ms7724se
  sh: make sure static declaration on mach-migor
  sh: make sure static declaration on mach-ecovec24
  sh: make sure static declaration on mach-ap325rxa
  clocksource: sh_cmt: compute mult and shift before registration
  clocksource: sh_tmu: compute mult and shift before registration
  sh: PIO disabling for x3proto and urquell.
  sh: mach-sdk7786: conditionally disable PIO support.
  sh: support for platforms without PIO.
  usb: r8a66597-hcd pio to mmio accessor conversion.
  usb: gadget: r8a66597-udc pio to mmio accessor conversion.
  usb: gadget: m66592-udc pio to mmio accessor conversion.
  sh: add romImage MMCIF boot for sh7724 and Ecovec V2
  sh: add boot code to MMCIF driver header
  sh: prepare MMCIF driver header file
  sh: allow romImage data between head.S and the zero page
  sh: Add support MMCIF for ecovec
  sh: remove duplicated #include
  input: serio: disable i8042 for non-cayman sh platforms.
  ...

14 years agoMerge branch 'linux_next' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab...
Linus Torvalds [Fri, 4 Jun 2010 22:39:54 +0000 (15:39 -0700)]
Merge branch 'linux_next' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/i7core

* 'linux_next' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/i7core: (83 commits)
  i7core_edac: Better describe the supported devices
  Add support for Westmere to i7core_edac driver
  i7core_edac: don't free on success
  i7core_edac: Add support for X5670
  Always call i7core_[ur]dimm_check_mc_ecc_err
  i7core_edac: fix memory leak of i7core_dev
  EDAC: add __init to i7core_xeon_pci_fixup
  i7core_edac: Fix wrong device id for channel 1 devices
  i7core: add support for Lynnfield alternate address
  i7core_edac: Add initial support for Lynnfield
  i7core_edac: do not export static functions
  edac: fix i7core build
  edac: i7core_edac produces undefined behaviour on 32bit
  i7core_edac: Use a more generic approach for probing PCI devices
  i7core_edac: PCI device is called NONCORE, instead of NOCORE
  i7core_edac: Fix ringbuffer maxsize
  i7core_edac: First store, then increment
  i7core_edac: Better parse "any" addrmask
  i7core_edac: Use a lockless ringbuffer
  edac: Create an unique instance for each kobj
  ...

14 years agoMerge branch 'v4l_for_2.6.35' of git://git.kernel.org/pub/scm/linux/kernel/git/mcheha...
Linus Torvalds [Fri, 4 Jun 2010 22:38:12 +0000 (15:38 -0700)]
Merge branch 'v4l_for_2.6.35' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-2.6

* 'v4l_for_2.6.35' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-2.6: (87 commits)
  V4L/DVB: ivtv: Timing tweaks and code re-order to try and improve stability
  V4L/DVB: ivtv: Avoid accidental video standard change
  V4L/DVB: ivtvfb : Module load / unload fixes
  V4L/DVB: cx2341x: Report correct temporal setting for log-status
  V4L/DVB: cx18, cx23885, v4l2 doc, MAINTAINERS: Update Andy Walls' email address
  V4L/DVB: drivers/media: Eliminate a NULL pointer dereference
  V4L/DVB: dvb-core: Fix ULE decapsulation bug
  V4L/DVB: Bug fix: make IR work again for dm1105
  V4L/DVB: media/IR: nec-decoder needs to select BITREV
  V4L/DVB: video/saa7134: change dprintk() to i2cdprintk()
  V4L/DVB: video/saa7134: remove duplicate break
  V4L/DVB: IR/imon: add auto-config for 0xffdc rf device
  V4L/DVB: IR/imon: clean up usage of bools
  V4L/DVB: em28xx: remove unneeded null checks
  V4L/DVB: ngene: remove unused #include <linux/version.h>
  V4L/DVB: ak881x needs slab.h
  V4L/DVB: FusionHDTV: Use quick reads for I2C IR device probing
  V4L/DVB: Technotrend S2-3200 ships with a TT 1500 remote
  V4L/DVB: drivers/media: Use kzalloc
  V4L/DVB: m920x: Select simple tuner
  ...

14 years agoMerge branch 'for-linus' of git://git.kernel.dk/linux-2.6-block
Linus Torvalds [Fri, 4 Jun 2010 22:37:44 +0000 (15:37 -0700)]
Merge branch 'for-linus' of git://git.kernel.dk/linux-2.6-block

* 'for-linus' of git://git.kernel.dk/linux-2.6-block: (27 commits)
  block: make blk_init_free_list and elevator_init idempotent
  block: avoid unconditionally freeing previously allocated request_queue
  pipe: change /proc/sys/fs/pipe-max-pages to byte sized interface
  pipe: change the privilege required for growing a pipe beyond system max
  pipe: adjust minimum pipe size to 1 page
  block: disable preemption before using sched_clock()
  cciss: call BUG() earlier
  Preparing 8.3.8rc2
  drbd: Reduce verbosity
  drbd: use drbd specific ratelimit instead of global printk_ratelimit
  drbd: fix hang on local read errors while disconnected
  drbd: Removed the now empty w_io_error() function
  drbd: removed duplicated #includes
  drbd: improve usage of MSG_MORE
  drbd: need to set socket bufsize early to take effect
  drbd: improve network latency, TCP_QUICKACK
  drbd: Revert "drbd: Create new current UUID as late as possible"
  brd: support discard
  Revert "writeback: fix WB_SYNC_NONE writeback from umount"
  Revert "writeback: ensure that WB_SYNC_NONE writeback with sb pinned is sync"
  ...

14 years agoMerge branch 'rc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuil...
Linus Torvalds [Fri, 4 Jun 2010 22:37:21 +0000 (15:37 -0700)]
Merge branch 'rc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild-2.6

* 'rc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild-2.6:
  gconfig: fix build failure on fedora 13

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging-2.6
Linus Torvalds [Fri, 4 Jun 2010 22:27:59 +0000 (15:27 -0700)]
Merge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging-2.6: (27 commits)
  Staging: sep: return -EFAULT on copy_to_user errors
  Staging: rc2860: return -EFAULT on copy_to_user errors
  Staging: Eliminate a NULL pointer dereference
  staging: Use GFP_ATOMIC when a lock is held
  Staging: comedi - correct parameter gainlkup for DAQCard-6024E in driver ni_mio_cs.c
  Staging: comedi: fixing ni_labpc to mite dependancy
  Staging: wlags49_h2, wlags49_h25: fixed Kconfig dependencies
  Staging: phison: depends on ATA_BMDMA
  Staging: iio-utils: fix memory overflow for dynamically allocateded memory to hold filename
  Staging: adis16255: add proper section markings to hotplug funcs
  Staging: adis16255: fix typo in Kconfig
  Staging: batman-adv: Don't allocate icmp packet with GFP_KERNEL
  Staging: batman-adv: Don't call free_netdev twice
  Staging: batman-adv: Call unregister_netdev on failures to get rtnl lock
  Staging: batman-adv: fix rogue packets on shutdown
  Staging: add MSM framebuffer driver
  Staging: comedi: fixing ni_tio to mite PCI dependancy
  Staging: comedi: fix 8255 and DAS08 Kconfig dependancies.
  Staging: comedi: For COMEDI_BUFINFO, check access to command
  Staging: comedi: COMEDI_BUFINFO with no async - report no bytes read or written
  ...

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core-2.6
Linus Torvalds [Fri, 4 Jun 2010 22:27:27 +0000 (15:27 -0700)]
Merge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core-2.6:
  fix setattr error handling in sysfs, configfs
  kobject: free memory if netlink_kernel_create() fails
  lib/kobject_uevent.c: fix CONIG_NET=n warning

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty-2.6
Linus Torvalds [Fri, 4 Jun 2010 22:23:07 +0000 (15:23 -0700)]
Merge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty-2.6:
  serial: add support for various Titan PCI cards
  vt_ioctl: return -EFAULT on copy_from_user errors
  serial: altera_uart: Proper section for altera_uart_remove
  tty: fix a little bug in scrup, vt.c
  altera_uart: Simplify altera_uart_console_putc
  altera_uart: Don't take spinlock in already protected functions
  TTY/n_gsm: potential double lock
  serial: bfin_5xx: fix typo in IER check
  serial: bfin_5xx: IRDA is not affected by anomaly 05000230
  serial_cs: add and sort IDs for serial and modem cards
  msm_serial: fix serial on trout

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6
Linus Torvalds [Fri, 4 Jun 2010 22:22:31 +0000 (15:22 -0700)]
Merge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6:
  USB: unbind all interfaces before rebinding them
  USB: serial: digi_acceleport: Eliminate a NULL pointer dereference
  usb: fix ehci_hcd build failure when both generic-OF and xilinx is selected
  USB: cdc-acm: fix resource reclaim in error path of acm_probe
  USB: ftdi_sio: fix DTR/RTS line modes
  USB: s3c-hsotg: Ensure FIFOs are fully flushed after layout
  USB: s3c-hsotg: SoftDisconnect minimum 3ms
  USB: s3c-hsotg: Ensure TX FIFO addresses setup when initialising FIFOs
  USB: s3c_hsotg: define USB_GADGET_DUALSPEED in Kconfig
  USB: s3c: Enable soft disconnect during initialization
  USB: xhci: Print NEC firmware version.
  USB: xhci: Wait for host to start running.
  USB: xhci: Wait for controller to be ready after reset.
  USB: isp1362: fix inw warning on Blackfin systems
  USB: mos7840: fix null-pointer dereference

14 years agoomap: remove BUG_ON for disabled interrupts
Cory Maccarrone [Fri, 4 Jun 2010 21:15:07 +0000 (14:15 -0700)]
omap: remove BUG_ON for disabled interrupts

Remove a BUG_ON for when interrupts are disabled during an MMC request.

During boot, interrupts can be disabled when a request is made, causing
this bug to be triggered.  In reality, there's no reason this should halt
the kernel, as the driver has proved reliable in spite of disabled
interrupts, and additionally, there's nothing in this code that would
require interrupts to be enabled.

The only setup I've managed to make it trigger on is on the HTC Herald
during bootup when the driver is built into the kernel (mostly because
that's all I have).  I believe it's related to the fact that on bootup I
get many timeout errors on "CMD5" while initializing the card.  Each CMD5
timeout triggers that bug (I changed it to a WARN_ON to get it to boot in)
due to the fact that part of the timeout code involves sending the request
again.  With interrupts turned off, that BUG would be triggered.

Signed-off-by: Cory Maccarrone <darkstar6262@gmail.com>
Acked-by: Tony Lindgren <tony@atomide.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agovmscan: fix do_try_to_free_pages() return value when priority==0 reclaim failure
KOSAKI Motohiro [Fri, 4 Jun 2010 21:15:05 +0000 (14:15 -0700)]
vmscan: fix do_try_to_free_pages() return value when priority==0 reclaim failure

Greg Thelen reported recent Johannes's stack diet patch makes kernel hang.
 His test is following.

  mount -t cgroup none /cgroups -o memory
  mkdir /cgroups/cg1
  echo $$ > /cgroups/cg1/tasks
  dd bs=1024 count=1024 if=/dev/null of=/data/foo
  echo $$ > /cgroups/tasks
  echo 1 > /cgroups/cg1/memory.force_empty

Actually, This OOM hard to try logic have been corrupted since following
two years old patch.

commit a41f24ea9fd6169b147c53c2392e2887cc1d9247
Author: Nishanth Aravamudan <nacc@us.ibm.com>
Date:   Tue Apr 29 00:58:25 2008 -0700

    page allocator: smarter retry of costly-order allocations

Original intention was "return success if the system have shrinkable zones
though priority==0 reclaim was failure".  But the above patch changed to
"return nr_reclaimed if .....".  Oh, That forgot nr_reclaimed may be 0 if
priority==0 reclaim failure.

And Johannes's patch 0aeb2339e54e ("vmscan: remove all_unreclaimable scan
control") made it more corrupt.  Originally, priority==0 reclaim failure
on memcg return 0, but this patch changed to return 1.  It totally
confused memcg.

This patch fixes it completely.

Reported-by: Greg Thelen <gthelen@google.com>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Tested-by: Greg Thelen <gthelen@google.com>
Acked-by: Balbir Singh <balbir@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agokernel/: fix BUG_ON checks for cpu notifier callbacks direct call
Akinobu Mita [Fri, 4 Jun 2010 21:15:04 +0000 (14:15 -0700)]
kernel/: fix BUG_ON checks for cpu notifier callbacks direct call

The commit 80b5184cc537718122e036afe7e62d202b70d077 ("kernel/: convert cpu
notifier to return encapsulate errno value") changed the return value of
cpu notifier callbacks.

Those callbacks don't return NOTIFY_BAD on failures anymore.  But there
are a few callbacks which are called directly at init time and checking
the return value.

I forgot to change BUG_ON checking by the direct callers in the commit.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agocgroups: alloc_css_id() increments hierarchy depth
Greg Thelen [Fri, 4 Jun 2010 21:15:03 +0000 (14:15 -0700)]
cgroups: alloc_css_id() increments hierarchy depth

Child groups should have a greater depth than their parents.  Prior to
this change, the parent would incorrectly report zero memory usage for
child cgroups when use_hierarchy is enabled.

test script:
  mount -t cgroup none /cgroups -o memory
  cd /cgroups
  mkdir cg1

  echo 1 > cg1/memory.use_hierarchy
  mkdir cg1/cg11

  echo $$ > cg1/cg11/tasks
  dd if=/dev/zero of=/tmp/foo bs=1M count=1

  echo
  echo CHILD
  grep cache cg1/cg11/memory.stat

  echo
  echo PARENT
  grep cache cg1/memory.stat

  echo $$ > tasks
  rmdir cg1/cg11 cg1
  cd /
  umount /cgroups

Using fae9c79, a recent patch that changed alloc_css_id() depth computation,
the parent incorrectly reports zero usage:
  root@ubuntu:~# ./test
  1+0 records in
  1+0 records out
  1048576 bytes (1.0 MB) copied, 0.0151844 s, 69.1 MB/s

  CHILD
  cache 1048576
  total_cache 1048576

  PARENT
  cache 0
  total_cache 0

With this patch, the parent correctly includes child usage:
  root@ubuntu:~# ./test
  1+0 records in
  1+0 records out
  1048576 bytes (1.0 MB) copied, 0.0136827 s, 76.6 MB/s

  CHILD
  cache 1052672
  total_cache 1052672

  PARENT
  cache 0
  total_cache 1052672

Signed-off-by: Greg Thelen <gthelen@google.com>
Acked-by: Paul Menage <menage@google.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Li Zefan <lizf@cn.fujitsu.com>
Cc: <stable@kernel.org> [2.6.34.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agolib: add s390 to atomic64_dec_if_positive archs
Heiko Carstens [Fri, 4 Jun 2010 21:15:02 +0000 (14:15 -0700)]
lib: add s390 to atomic64_dec_if_positive archs

Add s390 to list of architectures that have atomic64_dec_if_positive
implemented so we get rid of this warning:

lib/atomic64_test.c:129:2: warning: #warning Please implement
atomic64_dec_if_positive for your architecture, and add it to the IF above

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Luca Barbieri <luca@luca-barbieri.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agofbdev: fix frame buffer devices menu
Thadeu Lima de Souza Cascardo [Fri, 4 Jun 2010 21:15:00 +0000 (14:15 -0700)]
fbdev: fix frame buffer devices menu

Commit f601441916d1e19291d0b4f044b4a7551e2924d0 ("imxfb: add support for
i.MX25:) has inserted the symbol HAVE_FB_IMX, which does not depend on FB
after the menuconfig FB.  This breaks the menu, presenting most of the
drivers outside of it, when using menuconfig.

Moving the symbol to the start of the file, just like HAVE_FB_ATMEL, fixes
the problem without breaking it for iMX25 configurations (tested with
ARCH=arm, no build).

Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@holoscopio.com>
Cc: Sascha Hauer <s.hauer@pengutronix.de>
Acked-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Acked-by: Baruch Siach <baruch@tkos.co.il>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoarch/um: fix kunmap_atomic() call in skas/uaccess.c
Cesar Eduardo Barros [Fri, 4 Jun 2010 21:14:58 +0000 (14:14 -0700)]
arch/um: fix kunmap_atomic() call in skas/uaccess.c

kunmap_atomic() takes a pointer to within the page, not the struct page.

Signed-off-by: Cesar Eduardo Barros <cesarb@cesarb.net>
Cc: Jeff Dike <jdike@addtoit.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agosys_personality: change sys_personality() to accept "unsigned int" instead of u_long
Oleg Nesterov [Fri, 4 Jun 2010 21:14:58 +0000 (14:14 -0700)]
sys_personality: change sys_personality() to accept "unsigned int" instead of u_long

task_struct->pesonality is "unsigned int", but sys_personality() paths use
"unsigned long pesonality".  This means that every assignment or
comparison is not right.  In particular, if this argument does not fit
into "unsigned int" __set_personality() changes the caller's personality
and then sys_personality() returns -EINVAL.

Turn this argument into "unsigned int" and avoid overflows.  Obviously,
this is the user-visible change, we just ignore the upper bits.  But this
can't break the sane application.

There is another thing which can confuse the poorly written applications.
User-space thinks that this syscall returns int, not long.  This means
that the returned value can be negative and look like the error code.  But
note that libc won't be confused and thus errno won't be set, and with
this patch the user-space can never get -1 unless sys_personality() really
fails.  And, most importantly, the negative RET != -1 is only possible if
that app previously called personality(RET).

Pointed-out-by: Wenming Zhang <wezhang@redhat.com>
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agofb_defio: redo fix for non-dirty ptes
Albert Herranz [Fri, 4 Jun 2010 21:14:57 +0000 (14:14 -0700)]
fb_defio: redo fix for non-dirty ptes

As pointed by Nick Piggin, ->page_mkwrite provides a way to keep a page
locked until the associated PTE is marked dirty.

Re-implement the fix by using this mechanism.

Signed-off-by: Albert Herranz <albert_herranz@yahoo.es>
Acked-by: Jaya Kumar <jayakumar.lkml@gmail.com>
Acked-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoRevert "fb_defio: fix for non-dirty ptes"
Albert Herranz [Fri, 4 Jun 2010 21:14:56 +0000 (14:14 -0700)]
Revert "fb_defio: fix for non-dirty ptes"

This reverts commit 49bbd815fd8ba26d0354900b783b767c7f47c816 ("fb_defio:
fix for non-dirty ptes").

Although the fix provided is correct, it's been suggested to avoid the
underlying race in the same way as it is currently done in filesystems
like NFS, for maintainability.

A following patch "fb_defio: redo fix for non-dirty ptes" will provide
such an alternate fix.

Signed-off-by: Albert Herranz <albert_herranz@yahoo.es>
Cc: Jaya Kumar <jayakumar.lkml@gmail.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoflat: fix unmap len in load error path
Mike Frysinger [Fri, 4 Jun 2010 21:14:55 +0000 (14:14 -0700)]
flat: fix unmap len in load error path

The data chunk is mmaped with 'len' which remains unchanged, so use that
when unmapping in the error path rather than trying to recalculate (and
incorrectly so) the value used originally.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Acked-by: David McCullough <davidm@snapgear.com>
Acked-by: Greg Ungerer <gerg@uclinux.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Michal Simek <monstr@monstr.eu>
Cc: Hirokazu Takata <takata@linux-m32r.org>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agofs/binfmt_flat.c: split the stack & data alignments
Mike Frysinger [Fri, 4 Jun 2010 21:14:53 +0000 (14:14 -0700)]
fs/binfmt_flat.c: split the stack & data alignments

The stack and data have different alignment requirements, so don't force
them to wear the same shoe.  Increase the data alignment to match that
which the elf2flt linker script has always been using: 0x20 bytes.  Not
only does this bring the kernel loader in line with the toolchain, but it
also fixes a swath of gcc tests which try to force larger alignment values
but randomly fail when the FLAT loader fails to deliver.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: David Woodhouse <David.Woodhouse@intel.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Acked-by: David McCullough <davidm@snapgear.com>
Acked-by: Greg Ungerer <gerg@uclinux.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Tested-by: Michal Simek <monstr@monstr.eu>
Cc: Hirokazu Takata <takata@linux-m32r.org>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Jie Zhang <jie@codesourcery.com>
Cc: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agovmware balloon: clamp number of collected non-balloonable pages
Dmitry Torokhov [Fri, 4 Jun 2010 21:14:52 +0000 (14:14 -0700)]
vmware balloon: clamp number of collected non-balloonable pages

Limit number of accumulated non-balloonable pages during inflation cycle,
otherwise there is a chance we will be spinning and growing the list
forever.  This happens during torture tests when balloon target changes
while we are in the middle of inflation cycle and monitor starts refusing
to lock pages (since they are not needed anymore).

Signed-off-by: Dmitry Torokhov <dtor@vmware.com>
Acked-by: Bhavesh Davda <bhavesh@vmware.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoxtensa: invoke oom-killer from page fault
Nick Piggin [Fri, 4 Jun 2010 21:14:51 +0000 (14:14 -0700)]
xtensa: invoke oom-killer from page fault

As explained in commit 1c0fe6e3bd ("mm: invoke oom-killer from page
fault") , we want to call the architecture independent oom killer when
getting an unexplained OOM from handle_mm_fault, rather than simply
killing current.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Chris Zankel <chris@zankel.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agomn10300: invoke oom-killer from page fault
Nick Piggin [Fri, 4 Jun 2010 21:14:51 +0000 (14:14 -0700)]
mn10300: invoke oom-killer from page fault

As explained in commit 1c0fe6e3bd ("mm: invoke oom-killer from page
fault") , we want to call the architecture independent oom killer when
getting an unexplained OOM from handle_mm_fault, rather than simply
killing current.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Acked-by: David Rientjes <rientjes@google.com>
Acked-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agom32r: invoke oom-killer from page fault
Nick Piggin [Fri, 4 Jun 2010 21:14:49 +0000 (14:14 -0700)]
m32r: invoke oom-killer from page fault

As explained in commit 1c0fe6e3bd ("mm: invoke oom-killer from page
fault") , we want to call the architecture independent oom killer when
getting an unexplained OOM from handle_mm_fault, rather than simply
killing current.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Hirokazu Takata <takata@linux-m32r.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agofrv: invoke oom-killer from page fault
Nick Piggin [Fri, 4 Jun 2010 21:14:49 +0000 (14:14 -0700)]
frv: invoke oom-killer from page fault

As explained in commit 1c0fe6e3bd ("mm: invoke oom-killer from page
fault") , we want to call the architecture independent oom killer when
getting an unexplained OOM from handle_mm_fault, rather than simply
killing current.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Acked-by: David Howells <dhowells@redhat.com>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoramoops: add HAS_IOMEM dependency
Heiko Carstens [Fri, 4 Jun 2010 21:14:48 +0000 (14:14 -0700)]
ramoops: add HAS_IOMEM dependency

The driver fails to compile on s390:

drivers/char/ramoops.c: In function 'ramoops_init':
drivers/char/ramoops.c:122: error: implicit declaration of function 'ioremap'

Since we won't make use of the driver anyway on s390 just let it depend on
HAS_IOMEM.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Marco Stornelli <marco.stornelli@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agofs/compat_rw_copy_check_uvector: add missing compat_ptr call
Heiko Carstens [Fri, 4 Jun 2010 21:14:47 +0000 (14:14 -0700)]
fs/compat_rw_copy_check_uvector: add missing compat_ptr call

A call to access_ok is missing a compat_ptr conversion.  Introduced with
b83733639a494d5f42fa00a2506563fbd2d3015d "compat: factor out
compat_rw_copy_check_uvector from compat_do_readv_writev"

fs/compat.c: In function 'compat_rw_copy_check_uvector':
fs/compat.c:629: warning: passing argument 1 of '__access_ok' makes pointer from integer without a cast

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Reviewed-by: Jeff Moyer <jmoyer@redhat.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agortc: s3c: initialize s3c_rtc_cpu_type before using it
Maurus Cuelenaere [Fri, 4 Jun 2010 21:14:46 +0000 (14:14 -0700)]
rtc: s3c: initialize s3c_rtc_cpu_type before using it

Make sure s3c_rtc_cpu_type is initialised _before_ it's used in an if()
check.

Reported-by: Jiri Pinkava <jiri.pinkava@vscht.cz>
Signed-off-by: Maurus Cuelenaere <mcuelenaere@gmail.com>
Cc: Paul Gortmaker <p_gortmaker@yahoo.com>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Cc: Maurus Cuelenaere <mcuelenaere@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agortc: s3c: initialize driver data before using it
Maurus Cuelenaere [Fri, 4 Jun 2010 21:14:44 +0000 (14:14 -0700)]
rtc: s3c: initialize driver data before using it

s3c_rtc_setfreq() uses the platform driver data to derive struct rtc_device,
so make sure drvdata is set _before_ s3c_rtc_setfreq() is called.

Signed-off-by: Maurus Cuelenaere <mcuelenaere@gmail.com>
Cc: Paul Gortmaker <p_gortmaker@yahoo.com>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Cc: Maurus Cuelenaere <mcuelenaere@gmail.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoMinix: Clean up left over label
Andrew Hendry [Fri, 4 Jun 2010 12:51:24 +0000 (22:51 +1000)]
Minix: Clean up left over label

Remove a left over fail label.

Signed-off-by: Andrew Hendry <andrew.hendry@gmail.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
14 years agofix truncate inode time modification breakage
Nick Piggin [Thu, 3 Jun 2010 12:01:46 +0000 (22:01 +1000)]
fix truncate inode time modification breakage

mtime and ctime should be changed only if the file size has actually
changed. Patches changing ext2 and tmpfs from vmtruncate to new truncate
sequence has caused regressions where they always update timestamps.

There is some strange cases in POSIX where truncate(2) must not update
times unless the size has acutally changed, see 6e656be89.

This area is all still rather buggy in different ways in a lot of
filesystems and needs a cleanup and audit (ideally the vfs will provide
a simple attribute or call to direct all filesystems exactly which
attributes to change). But coming up with the best solution will take a
while and is not appropriate for rc anyway.

So fix recent regression for now.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
14 years agofix setattr error handling in sysfs, configfs
Nick Piggin [Mon, 31 May 2010 07:58:02 +0000 (17:58 +1000)]
fix setattr error handling in sysfs, configfs

sysfs and configfs setattr functions have error cases after the generic inode's
attributes have been changed. Fix consistency by changing the generic inode
attributes only when it is guaranteed to succeed.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
14 years agofcntl: return -EFAULT if copy_to_user fails
Dan Carpenter [Thu, 3 Jun 2010 10:35:42 +0000 (12:35 +0200)]
fcntl: return -EFAULT if copy_to_user fails

copy_to_user() returns the number of bytes remaining, but we want to
return -EFAULT.
ret = fcntl(fd, F_SETOWN_EX, NULL);
With the original code ret would be 8 here.

V2: Takuya Yoshikawa pointed out a similar issue in f_getown_ex()

Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
14 years agowrong type for 'magic' argument in simple_fill_super()
Roberto Sassu [Thu, 3 Jun 2010 09:58:28 +0000 (11:58 +0200)]
wrong type for 'magic' argument in simple_fill_super()

It's used to superblock ->s_magic, which is unsigned long.

Signed-off-by: Roberto Sassu <roberto.sassu@polito.it>
Reviewed-by: Mimi Zohar <zohar@us.ibm.com>
Signed-off-by: Eric Paris <eparis@redhat.com>
CC: stable@kernel.org
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
14 years agofix the deadlock in qib_fs
Al Viro [Sat, 29 May 2010 01:32:44 +0000 (21:32 -0400)]
fix the deadlock in qib_fs

get_sb_single() calls fill_super with superblock locked; calling
deactivate_super() will deadlock immedately.  Moreover, if fill_super
callback returns an error, get_sb_single() will release the reference
to superblock itself just fine.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
14 years agomqueue doesn't need make_bad_inode()
Al Viro [Fri, 28 May 2010 23:06:15 +0000 (19:06 -0400)]
mqueue doesn't need make_bad_inode()

It never hashes them anyway and does final iput() immediately
afterwards.  With ->drop_inode() being generic_delete_inode()...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
14 years agoStaging: sep: return -EFAULT on copy_to_user errors
Dan Carpenter [Fri, 4 Jun 2010 10:38:40 +0000 (12:38 +0200)]
Staging: sep: return -EFAULT on copy_to_user errors

copy_to_user() returns the number of bytes remaining but we want to
return a negative error code here.  These functions are used in the
ioctl handler and the error code gets returned to userspace.

Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoStaging: rc2860: return -EFAULT on copy_to_user errors
Dan Carpenter [Fri, 4 Jun 2010 10:39:51 +0000 (12:39 +0200)]
Staging: rc2860: return -EFAULT on copy_to_user errors

copy_to_user() returns the number of bytes remaining but we want to
return a negative error code.  This is in the ioctl handler and the
error code gets passed to userspace.

Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoStaging: Eliminate a NULL pointer dereference
Julia Lawall [Thu, 27 May 2010 12:37:19 +0000 (14:37 +0200)]
Staging: Eliminate a NULL pointer dereference

Eliminate a NULL or near NULL pointer dereference.

A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)

// <smpl>
@r exists@
expression E,E1;
identifier f;
statement S1,S2,S3;
@@

if ((E == NULL && ...) || ...)
{
  ... when != if (...) S1 else S2
      when != E = E1
* E->f
  ... when any
  return ...;
}
else S3
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agostaging: Use GFP_ATOMIC when a lock is held
Julia Lawall [Sun, 30 May 2010 20:27:46 +0000 (22:27 +0200)]
staging: Use GFP_ATOMIC when a lock is held

In each case, the containing function is only called from one place, where
a spin lock is held.

The semantic patch that makes this change is as follows:
(http://coccinelle.lip6.fr/)

// <smpl>
@gfp exists@
identifier fn;
position p;
@@

fn(...) {
... when != spin_unlock
    when any
  GFP_KERNEL@p
 ... when any
}

@locked@
identifier gfp.fn;
@@

spin_lock(...)
... when != spin_unlock
fn(...)

@depends on locked@
position gfp.p;
@@

- GFP_KERNEL@p
+ GFP_ATOMIC
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Cc: Jonathan Cameron <jic23@cam.ac.uk>
Cc: Marek Lindner <lindner_marek@yahoo.de>
Cc: Martyn Welch <martyn.welch@ge.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoStaging: comedi - correct parameter gainlkup for DAQCard-6024E in driver ni_mio_cs.c
Martin Homuth-Rosemann [Mon, 31 May 2010 20:33:04 +0000 (22:33 +0200)]
Staging: comedi - correct parameter gainlkup for DAQCard-6024E in driver ni_mio_cs.c

Correct at least one of the incorrect specs for a national instrument
data acquisition card DAQCard-6024E. This card has only four different
gain settings (+-10V, +-5V, +-0.5V, +-0.05V).

Signed-off-by: Martin Homuth-Rosemann <homuth-rosemann@gmx.net>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoStaging: comedi: fixing ni_labpc to mite dependancy
Alexander Kurz [Thu, 3 Jun 2010 11:50:24 +0000 (15:50 +0400)]
Staging: comedi: fixing ni_labpc to mite dependancy

the dependancy of ni_labpc on mite was missing,

Signed-off-by: Alexander Kurz <linux@kbdbabel.org>
Acked-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
index 0aa2b0d..79f5f2e 100644

14 years agoStaging: wlags49_h2, wlags49_h25: fixed Kconfig dependencies
Henk de Groot [Wed, 2 Jun 2010 19:30:00 +0000 (21:30 +0200)]
Staging: wlags49_h2, wlags49_h25: fixed Kconfig dependencies

Fixes Kconfig so the wlags49_h2 and wlags49_h25 drivers can be
selected from menuconfig without having to select another WLAN
driver first. Before it could only be selected when another driver
already selected WIRELESS_EXT. Also adds WEXT_PRIV on which
the driver also depends.

Align help text in Kconfig.

Signed-off-by: Henk de Groot <pe1dnn@amsat.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoStaging: phison: depends on ATA_BMDMA
Randy Dunlap [Wed, 26 May 2010 17:30:02 +0000 (10:30 -0700)]
Staging: phison: depends on ATA_BMDMA

phison uses interfaces and data that are built only when
ATA_BMDMA is enabled, so it should depend on that symbol.

drivers/staging/phison/phison.c:43: error: implicit declaration of function 'ATA_BMDMA_SHT'
drivers/staging/phison/phison.c:43: error: initializer element is not constant
drivers/staging/phison/phison.c:43: error: (near initialization for 'phison_sht.module')
drivers/staging/phison/phison.c:47: error: 'ata_bmdma_port_ops' undeclared here (not in a function)

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Cc: evan_ko@phison.com
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoStaging: iio-utils: fix memory overflow for dynamically allocateded memory to hold...
Barry Song [Tue, 25 May 2010 09:40:04 +0000 (17:40 +0800)]
Staging: iio-utils: fix memory overflow for dynamically allocateded memory to hold filename

Signed-off-by: Barry Song <21cnbao@gmail.com>
Acked-by: Jonathan Cameron <jic23@cam.ac.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoStaging: adis16255: add proper section markings to hotplug funcs
Mike Frysinger [Mon, 24 May 2010 04:55:42 +0000 (00:55 -0400)]
Staging: adis16255: add proper section markings to hotplug funcs

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoStaging: adis16255: fix typo in Kconfig
Mike Frysinger [Mon, 24 May 2010 04:55:41 +0000 (00:55 -0400)]
Staging: adis16255: fix typo in Kconfig

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoStaging: batman-adv: Don't allocate icmp packet with GFP_KERNEL
Sven Eckelmann [Sat, 22 May 2010 15:48:47 +0000 (17:48 +0200)]
Staging: batman-adv: Don't allocate icmp packet with GFP_KERNEL

A new buffer for a packet is created when a icmp packet is received.
This happens in a context with disabled irq. Thus we are not allowed to
sleep or call function which might sleep. kmalloc must be called with
GFP_ATOMIC instead of GFP_KERNEL to ensure that it does not sleep.

Signed-off-by: Sven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoStaging: batman-adv: Don't call free_netdev twice
Sven Eckelmann [Sat, 22 May 2010 15:48:46 +0000 (17:48 +0200)]
Staging: batman-adv: Don't call free_netdev twice

Free_netdev is registered as destructor in interface_setup for every
soft_device. This destructor is automatically called from
unregister_netdev and we must not call it again for the freed
net_device.

Signed-off-by: Sven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoStaging: batman-adv: Call unregister_netdev on failures to get rtnl lock
Sven Eckelmann [Sat, 22 May 2010 15:48:45 +0000 (17:48 +0200)]
Staging: batman-adv: Call unregister_netdev on failures to get rtnl lock

We must call unregister_netdev when we couldn't initialise the
batman-adv module and the soft_device was registered. There are two
version of the function which we can use:
 * unregister_netdevice - removes device
 * unregister_netdev - takes rtnl semaphore and remove device

We don't hold the semaphore in an error situation. So we must use
unregister_netdev.

Signed-off-by: Sven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoStaging: batman-adv: fix rogue packets on shutdown
Marek Lindner [Sat, 22 May 2010 15:48:44 +0000 (17:48 +0200)]
Staging: batman-adv: fix rogue packets on shutdown

On module shutdown batman-adv would purge the internal packet
queue by sending all remaining packets which could confuse
other nodes. Now, the packets are silently discarded.

Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Sven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
14 years agoStaging: add MSM framebuffer driver
Stepan Moskovchenko [Wed, 19 May 2010 18:03:30 +0000 (11:03 -0700)]
Staging: add MSM framebuffer driver

Qualcomm development of the MSM SOC framebuffer driver has
diverged significantly from the driver used by Android. This
is a snapshot of our current driver, in all it's agony. We are
putting this in staging to help with the process of converging
the two drivers.

At this point, the driver has been tested only in dumb
framebuffer mode.

Signed-off-by: Stepan Moskovchenko <stepanm@codeaurora.org>
Signed-off-by: David Brown <davidb@codeaurora.org>
Signed-off-by: Abhijeet Dharmapurikar <adharmap@codeaurora.org>
[dwalker@codeaurora.org: added a small compile fix and TODO.]
Signed-off-by: Daniel Walker <dwalker@codeaurora.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>