GitHub/exynos8895/android_kernel_samsung_universal8895.git
16 years agoclaw: fix memory leak in claw_probe.
Martin Schwidefsky [Thu, 21 Aug 2008 15:10:22 +0000 (17:10 +0200)]
claw: fix memory leak in claw_probe.

probe_error() frees memory only, if cgdev->dev.driver_data refers
to the claw_privbk structure. Move forward its setting in claw_probe()
to ensure proper freeing of claw_privbk allocations.

Cc: Daniel <danielm77@spray.se>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years ago[netdrvr] fix build issue: undefined reference to `NS8390p_init'
Alan Cox [Fri, 22 Aug 2008 18:24:15 +0000 (19:24 +0100)]
[netdrvr] fix build issue: undefined reference to `NS8390p_init'

Signed-off-by: Alan 'pass the paper bags' Cox <alan@redhat.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agofs_enet: Fix SCC Ethernet on CPM2, and crash in fs_enet_rx_napi()
Heiko Schocher [Tue, 26 Aug 2008 01:20:53 +0000 (20:20 -0500)]
fs_enet: Fix SCC Ethernet on CPM2, and crash in fs_enet_rx_napi()

Signed-off-by: Heiko Schocher <hs@denx.de>
Signed-off-by: Vitaly Bordug <vitb@kernel.crashing.org>
Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agoigb: fix setting the number of tx queues
Alexander Duyck [Tue, 26 Aug 2008 11:25:13 +0000 (04:25 -0700)]
igb: fix setting the number of tx queues

The real_num_tx_queues was not being set when in MSI-X only mode.  This patch
corrects that path so all interrupt types are correctly configured.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agoigb: ethtool -d reads EICR which is incorrect as it is read on clear
Alexander Duyck [Tue, 26 Aug 2008 11:25:05 +0000 (04:25 -0700)]
igb: ethtool -d reads EICR which is incorrect as it is read on clear

Ethtool -d is reading the EICR and ICR registers which is currently
clearing these registers and masking off interrupts.  To prevent this we
read the EICS and ICS equivilents as they can be read without clearing or
masking.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agoigb: force all queues to interrupt once every 2 seconds
Alexander Duyck [Tue, 26 Aug 2008 11:25:03 +0000 (04:25 -0700)]
igb: force all queues to interrupt once every 2 seconds

Set the EICS bit for each of the RX queues at least once every 2 seconds to
prevent the rx queues from stalling.

Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agor8169: balance pci_map / pci_unmap pair
Francois Romieu [Tue, 26 Aug 2008 19:56:06 +0000 (21:56 +0200)]
r8169: balance pci_map / pci_unmap pair

The leak hurts with swiotlb and jumbo frames.

Fix http://bugzilla.kernel.org/show_bug.cgi?id=9468.

Heavily hinted by Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Tested-by: Alistair John Strachan <alistair@devzero.co.uk>
Tested-by: Timothy J Fontaine <tjfontaine@atxconsulting.com>
Cc: Edward Hsu <edward_hsu@realtek.com.tw>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agomyri10ge: update version string to 1.4.3-1.358
Brice Goglin [Tue, 26 Aug 2008 21:29:12 +0000 (23:29 +0200)]
myri10ge: update version string to 1.4.3-1.358

Update myri10ge version string to 1.4.3-1.358.

Signed-off-by: Brice Goglin <brice@myri.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agoixgbe: fix vlan filtering
Alexander Duyck [Wed, 27 Aug 2008 01:30:04 +0000 (18:30 -0700)]
ixgbe: fix vlan filtering

VLAN filtering is broken, due to reading the incorrect register for
the VLAN filtering settings.  Fixed by reading/writing the correct
register.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agoBlackfin EMAC Driver: the BF526 also supports the MAC,
Mike Frysinger [Wed, 27 Aug 2008 03:47:57 +0000 (11:47 +0800)]
Blackfin EMAC Driver: the BF526 also supports the MAC,
so update things accordingly

Signed-off-by: Mike Frysinger <vapier.adi@gmail.com>
Signed-off-by: Bryan Wu <cooloney@kernel.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agoMerge branch 'for-2.6.27' of git://git.marvell.com/mv643xx_eth into upstream-fixes
Jeff Garzik [Wed, 27 Aug 2008 09:10:25 +0000 (05:10 -0400)]
Merge branch 'for-2.6.27' of git://git.marvell.com/mv643xx_eth into upstream-fixes

16 years agobnx2x: Version update
Eilon Greenstein [Mon, 25 Aug 2008 22:30:39 +0000 (15:30 -0700)]
bnx2x: Version update

Version update

Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: Multi Queue
Yitchak Gertner [Mon, 25 Aug 2008 22:27:06 +0000 (15:27 -0700)]
bnx2x: Multi Queue

The multi queue support is still disabled by default for the bnx2x
(needs some more testing and validation), but there are 2 obvious bug in
it which are fixed in this patch

Signed-off-by: Yitchak Gertner <gertner@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: NAPI and interrupts enable/disable
Yitchak Gertner [Mon, 25 Aug 2008 22:26:24 +0000 (15:26 -0700)]
bnx2x: NAPI and interrupts enable/disable

Fixing the order of enabling and disabling NAPI and the interrupts

Signed-off-by: Yitchak Gertner <gertner@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: NIC load failure cleanup
Yitchak Gertner [Mon, 25 Aug 2008 22:25:45 +0000 (15:25 -0700)]
bnx2x: NIC load failure cleanup

Load failures were not handled correctly

Signed-off-by: Yitchak Gertner <gertner@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: Initialization structure
Yitchak Gertner [Mon, 25 Aug 2008 22:24:21 +0000 (15:24 -0700)]
bnx2x: Initialization structure

The TPA initialization is part of the FW internal memory initialization
and so it is moved to the appropriate function

Signed-off-by: Yitchak Gertner <gertner@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: HW lock timeout
Eilon Greenstein [Mon, 25 Aug 2008 22:23:30 +0000 (15:23 -0700)]
bnx2x: HW lock timeout

Increasing the lock timeout to 5 seconds instead of 1 second to minimize
the chance of failures due to timeout

Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: Minimize lock time
Eilon Greenstein [Mon, 25 Aug 2008 22:22:46 +0000 (15:22 -0700)]
bnx2x: Minimize lock time

After iSCSI boot, the HW lock should only protect the flag so only the
first function will reset the chip and not then entire chip reset
process

Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: Fan failure mechanism on additional design
Eilon Greenstein [Mon, 25 Aug 2008 22:20:48 +0000 (15:20 -0700)]
bnx2x: Fan failure mechanism on additional design

The A1021G board is also using the fan failure mechanism in the same way
the A1022G board does

Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: Rx work check
Eilon Greenstein [Mon, 25 Aug 2008 22:19:17 +0000 (15:19 -0700)]
bnx2x: Rx work check

The has Rx work check was wrong: when the FW was at the end of the page,
the driver was already at the beginning of the next page. Since the
check only validated that both driver and FW are pointing to the same
place, it concluded that there is still work to be done. This caused
some serious issues including long latency results on ping-pong test and
lockups while unloading the driver in that condition.

Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoipv6: sysctl fixes
Al Viro [Mon, 25 Aug 2008 22:18:15 +0000 (15:18 -0700)]
ipv6: sysctl fixes

Braino: net.ipv6 in ipv6 skeleton has no business in rotable
class

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoipv4: sysctl fixes
Al Viro [Mon, 25 Aug 2008 22:17:44 +0000 (15:17 -0700)]
ipv4: sysctl fixes

net.ipv4.neigh should be a part of skeleton to avoid ordering problems

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agosctp: add verification checks to SCTP_AUTH_KEY option
Vlad Yasevich [Mon, 25 Aug 2008 22:16:19 +0000 (15:16 -0700)]
sctp: add verification checks to SCTP_AUTH_KEY option

The structure used for SCTP_AUTH_KEY option contains a
length that needs to be verfied to prevent buffer overflow
conditions.  Spoted by Eugene Teo <eteo@redhat.com>.

Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agomv643xx_eth: bump version to 1.3
Lennert Buytenhek [Sun, 24 Aug 2008 05:33:05 +0000 (07:33 +0200)]
mv643xx_eth: bump version to 1.3

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
16 years agomv643xx_eth: enforce multiple-of-8-bytes receive buffer size restriction
Lennert Buytenhek [Sun, 24 Aug 2008 01:00:20 +0000 (03:00 +0200)]
mv643xx_eth: enforce multiple-of-8-bytes receive buffer size restriction

The mv643xx_eth hardware ignores the lower three bits of the buffer
size field in receive descriptors, causing the reception of full-sized
packets to fail at some MTUs.  Fix this by rounding the size of
allocated receive buffers up to a multiple of eight bytes.

While we are at it, add a bit of extra space to each receive buffer so
that we can handle multiple vlan tags on ingress.

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
16 years agomv643xx_eth: fix NULL pointer dereference in rxq_process()
Lennert Buytenhek [Sun, 24 Aug 2008 00:33:47 +0000 (02:33 +0200)]
mv643xx_eth: fix NULL pointer dereference in rxq_process()

When we are low on memory, the assumption that every descriptor in the
receive ring will have an skbuff associated with it does not hold.

rxq_process() was assuming that if the receive descriptor it is working
on is not owned by the hardware, it can safely be processed and handed
to the networking stack.  But a descriptor in the receive ring not being
owned by the hardware can also happen when we are low on memory and did
not manage to refill the receive ring fully.

This patch changes rxq_process()'s bailout condition from "the first
receive descriptor to be processed is owned by the hardware" to "the
first receive descriptor to be processed is owned by the hardware OR
the number of valid receive descriptors in the ring is zero".

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
16 years agomv643xx_eth: fix inconsistent lock semantics
Lennert Buytenhek [Sun, 24 Aug 2008 00:30:42 +0000 (02:30 +0200)]
mv643xx_eth: fix inconsistent lock semantics

Nicolas Pitre noted that mv643xx_eth_poll was incorrectly using
non-IRQ-safe locks while checking whether to wake up the netdevice's
transmit queue.  Convert the locking to *_irq() variants, since we
are running from softirq context where interrupts are enabled.

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
16 years agomv643xx_eth: fix double add_timer() on the receive oom timer
Lennert Buytenhek [Sat, 23 Aug 2008 23:59:05 +0000 (01:59 +0200)]
mv643xx_eth: fix double add_timer() on the receive oom timer

Commit 12e4ab79cd828563dc090d2117dc8626b344bc8f ("mv643xx_eth: be
more agressive about RX refill") changed the condition for the receive
out-of-memory timer to be scheduled from "the receive ring is empty"
to "the receive ring is not full".

This can lead to a situation where the receive out-of-memory timer is
pending because a previous rxq_refill() didn't manage to refill the
receive ring entirely as a result of being out of memory, and
rxq_refill() is then called again as a side effect of a packet receive
interrupt, and that rxq_refill() call then again does not succeed to
refill the entire receive ring with fresh empty skbuffs because we are
still out of memory, and then tries to call add_timer() on the already
scheduled out-of-memory timer.

This patch fixes this issue by changing the add_timer() call in
rxq_refill() to a mod_timer() call.  If the OOM timer was not already
scheduled, this will behave as before, whereas if it was already
scheduled, this patch will push back its firing time a bit, which is
safe because we've (unsuccessfully) attempted to refill the receive
ring just before we do this.

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
16 years agomv643xx_eth: fix NAPI 'rotting packet' issue
Lennert Buytenhek [Sun, 24 Aug 2008 00:45:32 +0000 (02:45 +0200)]
mv643xx_eth: fix NAPI 'rotting packet' issue

When a receive interrupt occurs, mv643xx_eth would first process the
receive descriptors and then ACK the receive interrupt, instead of the
other way round.

This would leave a small race window between processing the last
receive descriptor and clearing the receive interrupt status in which
a new packet could come in, which would then 'rot' in the receive
ring until the next receive interrupt would come in.

Fix this by ACKing (clearing) the receive interrupt condition before
processing the receive descriptors.

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
16 years agoipv6: protocol for address routes
Stephen Hemminger [Sat, 23 Aug 2008 12:16:46 +0000 (05:16 -0700)]
ipv6: protocol for address routes

This fixes a problem spotted with zebra, but not sure if it is
necessary a kernel problem.  With IPV6 when an address is added to an
interface, Zebra creates a duplicate RIB entry, one as a connected
route, and other as a kernel route.

When an address is added to an interface the RTN_NEWADDR message
causes Zebra to create a connected route. In IPV4 when an address is
added to an interface a RTN_NEWROUTE message is set to user space with
the protocol RTPROT_KERNEL. Zebra ignores these messages, because it
already has the connected route.

The problem is that route created in IPV6 has route protocol ==
RTPROT_BOOT.  Was this a design decision or a bug? This fixes it. Same
patch applies to both net-2.6 and stable.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoicmp: icmp_sk() should not use smp_processor_id() in preemptible code
Denis V. Lunev [Sat, 23 Aug 2008 11:43:33 +0000 (04:43 -0700)]
icmp: icmp_sk() should not use smp_processor_id() in preemptible code

Pass namespace into icmp_xmit_lock, obtain socket inside and return
it as a result for caller.

Thanks Alexey Dobryan for this report:

Steps to reproduce:

CONFIG_PREEMPT=y
CONFIG_DEBUG_PREEMPT=y
tracepath <something>

BUG: using smp_processor_id() in preemptible [00000000] code: tracepath/3205
caller is icmp_sk+0x15/0x30
Pid: 3205, comm: tracepath Not tainted 2.6.27-rc4 #1

Call Trace:
 [<ffffffff8031af14>] debug_smp_processor_id+0xe4/0xf0
 [<ffffffff80409405>] icmp_sk+0x15/0x30
 [<ffffffff8040a17b>] icmp_send+0x4b/0x3f0
 [<ffffffff8025a415>] ? trace_hardirqs_on_caller+0xd5/0x160
 [<ffffffff8025a4ad>] ? trace_hardirqs_on+0xd/0x10
 [<ffffffff8023a475>] ? local_bh_enable_ip+0x95/0x110
 [<ffffffff804285b9>] ? _spin_unlock_bh+0x39/0x40
 [<ffffffff8025a26c>] ? mark_held_locks+0x4c/0x90
 [<ffffffff8025a4ad>] ? trace_hardirqs_on+0xd/0x10
 [<ffffffff8025a415>] ? trace_hardirqs_on_caller+0xd5/0x160
 [<ffffffff803e91b4>] ip_fragment+0x8d4/0x900
 [<ffffffff803e7030>] ? ip_finish_output2+0x0/0x290
 [<ffffffff803e91e0>] ? ip_finish_output+0x0/0x60
 [<ffffffff803e6650>] ? dst_output+0x0/0x10
 [<ffffffff803e922c>] ip_finish_output+0x4c/0x60
 [<ffffffff803e92e3>] ip_output+0xa3/0xf0
 [<ffffffff803e68d0>] ip_local_out+0x20/0x30
 [<ffffffff803e753f>] ip_push_pending_frames+0x27f/0x400
 [<ffffffff80406313>] udp_push_pending_frames+0x233/0x3d0
 [<ffffffff804067d1>] udp_sendmsg+0x321/0x6f0
 [<ffffffff8040d155>] inet_sendmsg+0x45/0x80
 [<ffffffff803b967f>] sock_sendmsg+0xdf/0x110
 [<ffffffff8024a100>] ? autoremove_wake_function+0x0/0x40
 [<ffffffff80257ce5>] ? validate_chain+0x415/0x1010
 [<ffffffff8027dc10>] ? __do_fault+0x140/0x450
 [<ffffffff802597d0>] ? __lock_acquire+0x260/0x590
 [<ffffffff803b9e55>] ? sockfd_lookup_light+0x45/0x80
 [<ffffffff803ba50a>] sys_sendto+0xea/0x120
 [<ffffffff80428e42>] ? _spin_unlock_irqrestore+0x42/0x80
 [<ffffffff803134bc>] ? __up_read+0x4c/0xb0
 [<ffffffff8024e0c6>] ? up_read+0x26/0x30
 [<ffffffff8020b8bb>] system_call_fastpath+0x16/0x1b

icmp6_sk() is similar.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agopkt_sched: Fix qdisc list locking
Jarek Poplawski [Fri, 22 Aug 2008 10:24:05 +0000 (03:24 -0700)]
pkt_sched: Fix qdisc list locking

Since some qdiscs call qdisc_tree_decrease_qlen() (so qdisc_lookup())
without rtnl_lock(), adding and deleting from a qdisc list needs
additional locking. This patch adds global spinlock qdisc_list_lock
and wrapper functions for modifying the list. It is considered as a
temporary solution until hfsc_dequeue(), netem_dequeue() and
tbf_dequeue() (or qdisc_tree_decrease_qlen()) are redone.

With feedback from Herbert Xu and David S. Miller.

Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agopkt_sched: Fix qdisc_watchdog() vs. dev_deactivate() race
Jarek Poplawski [Thu, 21 Aug 2008 12:11:14 +0000 (05:11 -0700)]
pkt_sched: Fix qdisc_watchdog() vs. dev_deactivate() race

dev_deactivate() can skip rescheduling of a qdisc by qdisc_watchdog()
or other timer calling netif_schedule() after dev_queue_deactivate().
We prevent this checking aliveness before scheduling the timer. Since
during deactivation the root qdisc is available only as qdisc_sleeping
additional accessor qdisc_root_sleeping() is created.

With feedback from Herbert Xu <herbert@gondor.apana.org.au>

Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agosctp: fix potential panics in the SCTP-AUTH API.
Vlad Yasevich [Thu, 21 Aug 2008 10:34:25 +0000 (03:34 -0700)]
sctp: fix potential panics in the SCTP-AUTH API.

All of the SCTP-AUTH socket options could cause a panic
if the extension is disabled and the API is envoked.

Additionally, there were some additional assumptions that
certain pointers would always be valid which may not
always be the case.

This patch hardens the API and address all of the crash
scenarios.

Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoLinux v2.6.27-rc4
Linus Torvalds [Thu, 21 Aug 2008 02:35:56 +0000 (19:35 -0700)]
Linux v2.6.27-rc4

16 years agocramfs: fix named-pipe handling
Al Viro [Wed, 20 Aug 2008 21:09:24 +0000 (14:09 -0700)]
cramfs: fix named-pipe handling

After commit a97c9bf33f4612e2aed6f000f6b1d268b6814f3c (fix cramfs
making duplicate entries in inode cache) in kernel 2.6.14, named-pipe
on cramfs does not work properly.

It seems the commit make all named-pipe on cramfs share their inode
(and named-pipe buffer).

Make ..._test() refuse to merge inodes with ->i_ino == 1, take inode setup
back to get_cramfs_inode() and make ->drop_inode() evict ones with ->i_ino
== 1 immediately.

Reported-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: <stable@kernel.org> [2.6.14 and later]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofbdefio: add set_page_dirty handler to deferred IO FB
Ian Campbell [Wed, 20 Aug 2008 21:09:23 +0000 (14:09 -0700)]
fbdefio: add set_page_dirty handler to deferred IO FB

Fixes kernel BUG at lib/radix-tree.c:473.

Previously the handler was incidentally provided by tmpfs but this was
removed with:

  commit 14fcc23fdc78e9d32372553ccf21758a9bd56fa1
  Author: Hugh Dickins <hugh@veritas.com>
  Date:   Mon Jul 28 15:46:19 2008 -0700

    tmpfs: fix kernel BUG in shmem_delete_inode

relying on this behaviour was incorrect in any case and the BUG also
appeared when the device node was on an ext3 filesystem.

v2: override a_ops at open() time rather than mmap() time to minimise
races per AKPM's concerns.

Signed-off-by: Ian Campbell <ijc@hellion.org.uk>
Cc: Jaya Kumar <jayakumar.lkml@gmail.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Hugh Dickins <hugh@veritas.com>
Cc: Johannes Weiner <hannes@saeurebad.de>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Kel Modderman <kel@otaku42.de>
Cc: Markus Armbruster <armbru@redhat.com>
Cc: Krzysztof Helt <krzysztof.h1@poczta.fm>
Cc: <stable@kernel.org> [14fcc23fd is in 2.6.25.14 and 2.6.26.1]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc: rtc-ds1374: fix 'no irq' case handling
Anton Vorontsov [Wed, 20 Aug 2008 21:09:21 +0000 (14:09 -0700)]
rtc: rtc-ds1374: fix 'no irq' case handling

On a PowerPC board with ds1374 RTC I'm getting this error while RTC tries
to probe:

rtc-ds1374 0-0068: unable to request IRQ

This happens because I2C probing code (drivers/of/of_i2c.c) is specifying
IRQ0 for 'no irq' case, which is correct.

The driver handles this incorrectly, though. This patch fixes it.

Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Cc: David Brownell <david-b@pacbell.net>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Acked-by: Peter Korsgaard <jacmet@sunsite.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: xip/ext2 fix block allocation race
Nick Piggin [Wed, 20 Aug 2008 21:09:20 +0000 (14:09 -0700)]
mm: xip/ext2 fix block allocation race

XIP can call into get_xip_mem concurrently with the same file,offset with
create=1.  This usually maps down to get_block, which expects the page
lock to prevent such a situation.  This causes ext2 to explode for one
reason or another.

Serialise those calls for the moment.  For common usages today, I suspect
get_xip_mem rarely is called to create new blocks.  In future as XIP
technologies evolve we might need to look at which operations require
scalability, and rework the locking to suit.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Cc: Jared Hulbert <jaredeh@gmail.com>
Acked-by: Carsten Otte <cotte@freenet.de>
Cc: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: xip fix fault vs sparse page invalidate race
Nick Piggin [Wed, 20 Aug 2008 21:09:20 +0000 (14:09 -0700)]
mm: xip fix fault vs sparse page invalidate race

XIP has a race between sparse pages being inserted into page tables, and
sparse pages being zapped when its time to put a non-sparse page in.

What can happen is that a process can be left with a dangling sparse page
in a MAP_SHARED mapping, while the rest of the world sees the non-sparse
version.  Ie.  data corruption.

Guard these operations with a seqlock, making fault-in-sparse-pages the
slowpath, and try-to-unmap-sparse-pages the fastpath.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Cc: Jared Hulbert <jaredeh@gmail.com>
Acked-by: Carsten Otte <cotte@freenet.de>
Cc: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: dirty page tracking race fix
Nick Piggin [Wed, 20 Aug 2008 21:09:18 +0000 (14:09 -0700)]
mm: dirty page tracking race fix

There is a race with dirty page accounting where a page may not properly
be accounted for.

clear_page_dirty_for_io() calls page_mkclean; then TestClearPageDirty.

page_mkclean walks the rmaps for that page, and for each one it cleans and
write protects the pte if it was dirty.  It uses page_check_address to
find the pte.  That function has a shortcut to avoid the ptl if the pte is
not present.  Unfortunately, the pte can be switched to not-present then
back to present by other code while holding the page table lock -- this
should not be a signal for page_mkclean to ignore that pte, because it may
be dirty.

For example, powerpc64's set_pte_at will clear a previously present pte
before setting it to the desired value.  There may also be other code in
core mm or in arch which do similar things.

The consequence of the bug is loss of data integrity due to msync, and
loss of dirty page accounting accuracy.  XIP's __xip_unmap could easily
also be unreliable (depending on the exact XIP locking scheme), which can
lead to data corruption.

Fix this by having an option to always take ptl to check the pte in
page_check_address.

It's possible to retain this optimization for page_referenced and
try_to_unmap.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Cc: Jared Hulbert <jaredeh@gmail.com>
Cc: Carsten Otte <cotte@freenet.de>
Cc: Hugh Dickins <hugh@veritas.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agofix setpriority(PRIO_PGRP) thread iterator breakage
Ken Chen [Wed, 20 Aug 2008 21:09:17 +0000 (14:09 -0700)]
fix setpriority(PRIO_PGRP) thread iterator breakage

When user calls sys_setpriority(PRIO_PGRP ...) on a NPTL style multi-LWP
process, only the task leader of the process is affected, all other
sibling LWP threads didn't receive the setting.  The problem was that the
iterator used in sys_setpriority() only iteartes over one task for each
process, ignoring all other sibling thread.

Introduce a new macro do_each_pid_thread / while_each_pid_thread to walk
each thread of a process.  Convert 4 call sites in {set/get}priority and
ioprio_{set/get}.

Signed-off-by: Ken Chen <kenchen@google.com>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Roland McGrath <roland@redhat.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
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 agoVideo/Framebuffer: add fuctional power management support to Blackfin BF54x LQ043...
Michael Hennerich [Wed, 20 Aug 2008 21:09:16 +0000 (14:09 -0700)]
Video/Framebuffer: add fuctional power management support to Blackfin BF54x LQ043 framebuffer driver

Fix bug: does nor properply resume after suspend mem
Fix for PM_SUSPEND_MEM: Save and restore peripheral base and DMA registers

Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Bryan Wu <cooloney@kernel.org>
Acked-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agobootmem: fix aligning of node-relative indexes and offsets
Johannes Weiner [Wed, 20 Aug 2008 21:09:15 +0000 (14:09 -0700)]
bootmem: fix aligning of node-relative indexes and offsets

Absolute alignment requirements may never be applied to node-relative
offsets.  Andreas Herrmann spotted this flaw when a bootmem allocation on
an unaligned node was itself not aligned because the combination of an
unaligned node with an aligned offset into that node is not garuanteed to
be aligned itself.

This patch introduces two helper functions that align a node-relative
index or offset with respect to the node's starting address so that the
absolute PFN or virtual address that results from combining the two
satisfies the requested alignment.

Then all the broken ALIGN()s in alloc_bootmem_core() are replaced by these
helpers.

Signed-off-by: Johannes Weiner <hannes@saeurebad.de>
Reported-by: Andreas Herrmann <andreas.herrmann3@amd.com>
Debugged-by: Andreas Herrmann <andreas.herrmann3@amd.com>
Reviewed-by: Andreas Herrmann <andreas.herrmann3@amd.com>
Tested-by: Andreas Herrmann <andreas.herrmann3@amd.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agodrivers/char/ipmi/ipmi_si_intf.c:default_find_bmc(): fix leak
Andrew Morton [Wed, 20 Aug 2008 21:09:14 +0000 (14:09 -0700)]
drivers/char/ipmi/ipmi_si_intf.c:default_find_bmc(): fix leak

If check_legacy_ioport() returns true, we leak *info.

Addresses http://bugzilla.kernel.org/show_bug.cgi?id=11362

Reported-by: Daniel Marjamki <danielm77@spray.se>
Cc: Christian Krafft <krafft@de.ibm.com>
Cc: Michael Ellerman <michael@ellerman.id.au>
Cc: Corey Minyard <minyard@acm.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agopm2fb: free cmap memory on module remove
Krzysztof Helt [Wed, 20 Aug 2008 21:09:11 +0000 (14:09 -0700)]
pm2fb: free cmap memory on module remove

Release cmap memory allocated in the probe function.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc: fix double lock on UIE emulation
Atsushi Nemoto [Wed, 20 Aug 2008 21:09:11 +0000 (14:09 -0700)]
rtc: fix double lock on UIE emulation

With commit 5ad31a575157147b43fa84ef1e21471661653878 ("rtc: remove BKL
for ioctl()"), RTC_UIE_ON ioctl cause double lock on rtc->ops_lock.
The ops_lock must not be held while set_uie() calls rtc_read_time()
which takes the lock.  Also clear_uie() does not need ops_lock.  This
patch fixes return value of RTC_UIE_OFF ioctl too.

Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Cc: David Brownell <david-b@pacbell.net>
Cc: Russell King <rmk@arm.linux.org.uk>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agobinfmt_misc: fix false -ENOEXEC when coupled with other binary handlers
Pavel Emelyanov [Wed, 20 Aug 2008 21:09:10 +0000 (14:09 -0700)]
binfmt_misc: fix false -ENOEXEC when coupled with other binary handlers

In case the binfmt_misc binary handler is registered *before* the e.g.
script one (when for example being compiled as a module) the following
situation may occur:

1. user launches a script, whose interpreter is a misc binary;
2. the load_misc_binary sets the misc_bang and returns -ENOEVEC,
   since the binary is a script;
3. the load_script_binary loads one and calls for search_binary_hander
   to run the interpreter;
4. the load_misc_binary is called again, but refuses to load the
   binary due to misc_bang bit set.

The fix is to move the misc_bang setting lower - prior to the actual
call to the search_binary_handler.

Caused by the commit 3a2e7f47 (binfmt_misc.c: avoid potential kernel
stack overflow)

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Reported-by: Kirill A. Shutemov <kirill@shutemov.name>
Tested-by: Kirill A. Shutemov <kirill@shutemov.name>
Cc: <stable@kernel.org> [2.6.26.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agobrd: fix name argument of unregister_blkdev()
Akinobu Mita [Wed, 20 Aug 2008 21:09:09 +0000 (14:09 -0700)]
brd: fix name argument of unregister_blkdev()

The name of brd block device is "ramdisk", it's not "brd".
(The block device is registered by register_blkdev(RAMDISK_MAJOR, "ramdisk")
So it should be unregistered by unregister_blkdev(RAMDISK_MAJOR, "ramdisk")

Signed-off-by: Akinobu Mita <akinobu.mita@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>
16 years agonbd: fix memory leak of nbd_dev array
Sven Wegener [Wed, 20 Aug 2008 21:09:07 +0000 (14:09 -0700)]
nbd: fix memory leak of nbd_dev array

We leak the memory allocated for the nbd_dev array at multiple places.
Fix them by either adding a kfree() or by rearranging code to return
before we allocate the memory.

Signed-off-by: Sven Wegener <sven.wegener@stealer.net>
Cc: Paul Clements <paul.clements@steeleye.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: mminit_loglevel cannot be __meminitdata anymore
Marcin Slusarz [Wed, 20 Aug 2008 21:09:06 +0000 (14:09 -0700)]
mm: mminit_loglevel cannot be __meminitdata anymore

mminit_loglevel is now used from mminit_verify_zonelist <- build_all_zonelists <-

1. online_pages <- memory_block_action <- memory_block_change_state <- store_mem_state (sys handler)
2. numa_zonelist_order_handler (proc handler)

so it cannot be annotated __meminit - drop it

fixes following section mismatch warning:
WARNING: vmlinux.o(.text+0x71628): Section mismatch in reference from the function mminit_verify_zonelist() to the variable .meminit.data:mminit_loglevel
The function mminit_verify_zonelist() references
the variable __meminitdata mminit_loglevel.
This is often because mminit_verify_zonelist lacks a __meminitdata
annotation or the annotation of mminit_loglevel is wrong.

Signed-off-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: show free swap as signed
Hugh Dickins [Wed, 20 Aug 2008 21:09:05 +0000 (14:09 -0700)]
mm: show free swap as signed

Adjust <Alt><SysRq>m show_swap_cache_info() to show "Free swap" as a
signed long: the signed format is preferable, because during swapoff
nr_swap_pages can legitimately go negative, so makes more sense thus
(it used to be shown redundantly, once as signed and once as unsigned).

Signed-off-by: Hugh Dickins <hugh@veritas.com>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agomm: page_remove_rmap comments on PageAnon
Hugh Dickins [Wed, 20 Aug 2008 21:09:04 +0000 (14:09 -0700)]
mm: page_remove_rmap comments on PageAnon

Add a comment to s390's page_test_dirty/page_clear_dirty/page_set_dirty
dance in page_remove_rmap(): I was wrong to think the PageSwapCache test
could be avoided, and would like a comment in there to remind me.  And
mention s390, to help us remember that this block is not really common.

Also move down the "It would be tidy to reset PageAnon" comment: it does
not belong to s390's block, and it would be unwise to reset PageAnon
before we're done with testing it.

Signed-off-by: Hugh Dickins <hugh@veritas.com>
Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoBlackfin RTC Driver: dont let RTC programming in bootloaders randomly cause ~5 second...
Mike Frysinger [Wed, 20 Aug 2008 21:09:03 +0000 (14:09 -0700)]
Blackfin RTC Driver: dont let RTC programming in bootloaders randomly cause ~5 second boot delays

Signed-off-by: Mike Frysinger <vapier.adi@gmail.com>
Signed-off-by: Bryan Wu <cooloney@kernel.org>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoBlackfin RTC Driver: BF561 not have on-chip RTC
Graf Yang [Wed, 20 Aug 2008 21:09:02 +0000 (14:09 -0700)]
Blackfin RTC Driver: BF561 not have on-chip RTC

Signed-off-by: Graf Yang <graf.yang@analog.com>
Signed-off-by: Bryan Wu <cooloney@kernel.org>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoBlackfin RTC Driver: do all initialization before we register the rtc and make it...
Mike Frysinger [Wed, 20 Aug 2008 21:09:02 +0000 (14:09 -0700)]
Blackfin RTC Driver: do all initialization before we register the rtc and make it available

Signed-off-by: Mike Frysinger <vapier.adi@gmail.com>
Signed-off-by: Bryan Wu <cooloney@kernel.org>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoBlackfin RTC Driver: move irq request/free out of open/release and into probe/remove...
Mike Frysinger [Wed, 20 Aug 2008 21:09:01 +0000 (14:09 -0700)]
Blackfin RTC Driver: move irq request/free out of open/release and into probe/remove so that the non-dev interfaces (like sysfs) work as expected

Signed-off-by: Mike Frysinger <vapier.adi@gmail.com>
Signed-off-by: Bryan Wu <cooloney@kernel.org>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years ago/proc/self/maps doesn't display the real file offset
Clement Calmels [Wed, 20 Aug 2008 21:09:00 +0000 (14:09 -0700)]
/proc/self/maps doesn't display the real file offset

This addresses

http://bugzilla.kernel.org/show_bug.cgi?id=11318

In function show_map (file: fs/proc/task_mmu.c), if vma->vm_pgoff > 2^20
than (vma->vm_pgoff << PAGE_SIZE) is greater than 2^32 (with PAGE_SIZE
equal to 4096 (i.e.  2^12).  The next seq_printf use an unsigned long for
the conversion of (vma->vm_pgoff << PAGE_SIZE), as a result the offset
value displayed in /proc/self/maps is truncated if the page offset is
greater than 2^20.

A test that shows this issue:

#define _GNU_SOURCE
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/mman.h>
#include <stdlib.h>
#include <stdio.h>
#include <fcntl.h>
#include <unistd.h>
#include <string.h>

#define PAGE_SIZE (getpagesize())

#if __i386__
#   define U64_STR "%llx"
#elif __x86_64
#   define U64_STR "%lx"
#else
#   error "Architecture Unsupported"
#endif

int main(int argc, char *argv[])
{
int fd;
char *addr;
off64_t offset = 0x10000000;
char *filename = "/dev/zero";

fd = open(filename, O_RDONLY);
if (fd < 0) {
perror("open");
return 1;
}

offset *= 0x10;
printf("offset = " U64_STR "\n", offset);

addr = (char*)mmap64(NULL, PAGE_SIZE, PROT_READ, MAP_PRIVATE, fd,
     offset);
if ((void*)addr == MAP_FAILED) {
perror("mmap64");
return 1;
}

{
FILE *fmaps;
char *line = NULL;
size_t len = 0;
ssize_t read;
size_t filename_len = strlen(filename);

fmaps = fopen("/proc/self/maps", "r");
if (!fmaps) {
perror("fopen");
return 1;
}
while ((read = getline(&line, &len, fmaps)) != -1) {
if ((read > filename_len + 1)
    && (strncmp(&line[read - filename_len - 1], filename, filename_len) == 0))
printf("%s", line);
}

if (line)
free(line);

fclose(fmaps);
}

close(fd);
return 0;
}

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Clement Calmels <cboulte@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agodocumentation: describe bootmem_debug kernel parameter
Andreas Herrmann [Wed, 20 Aug 2008 21:08:58 +0000 (14:08 -0700)]
documentation: describe bootmem_debug kernel parameter

"bootmem_debug" is not mentioned in kernel-parameters.txt. Recently I
had to use that kernel option and I think it should be documented.

Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Johannes Weiner <hannes@saeurebad.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoeeepc-laptop: fix use after free
Matthew Garrett [Wed, 20 Aug 2008 21:08:57 +0000 (14:08 -0700)]
eeepc-laptop: fix use after free

eeepc-laptop uses the hwmon struct after unregistering the device, causing
an oops on module unload.  Flip the ordering to fix.

Signed-off-by: Matthew Garrett <mjg@redhat.com>
Cc: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
Cc: Corentin Chary <corentincj@iksaif.net>
Cc: Karol Kozimor <sziwan@users.sourceforge.net>
Cc: <stable@kernel.org> [2.6.26.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoFRV: Provide ioremap_wc() for FRV
David Howells [Wed, 20 Aug 2008 15:04:22 +0000 (16:04 +0100)]
FRV: Provide ioremap_wc() for FRV

Provide ioremap_wc() for FRV.

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMN10300: Supply ioremap_wc() for MN10300
David Howells [Wed, 20 Aug 2008 14:53:33 +0000 (15:53 +0100)]
MN10300: Supply ioremap_wc() for MN10300

Supply ioremap_wc() for MN10300.

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoReserve NFS fileid values for btrfs
David Woodhouse [Wed, 20 Aug 2008 13:58:23 +0000 (14:58 +0100)]
Reserve NFS fileid values for btrfs

Purely cosmetic for now, but we might as well get it merged ASAP.

Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoReduce brokenness of CRIS headers_install
David Woodhouse [Wed, 20 Aug 2008 13:44:45 +0000 (14:44 +0100)]
Reduce brokenness of CRIS headers_install

I won't say 'fix', because they still look broken, although this will at
least allow 'make ARCH=CRIS headers_install' to _complete_.

For headers which are exported, we should probably choose between
asm/arch-v10 and asm/arch-v32 by something that GCC defines -- we can't
rely on a generated symlink. And we certainly can't export an arch/
directory which doesn't even exist.

And the only thing that we seem to include from the arch/ directory is
<asm/arch/ptrace.h> from <asm/ptrace.h> ... and that isn't exported in
either arch-v10 or arch-v32 _anyway_.

Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMerge branch 'sh/for-2.6.27' of git://git.kernel.org/pub/scm/linux/kernel/git/lethal...
Linus Torvalds [Wed, 20 Aug 2008 15:46:11 +0000 (08:46 -0700)]
Merge branch 'sh/for-2.6.27' of git://git./linux/kernel/git/lethal/sh-2.6

* 'sh/for-2.6.27' of git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6:
  sh: Provide a FLAT_PLAT_INIT() definition.
  binfmt_flat: Stub in a FLAT_PLAT_INIT().
  video: export sh_mobile_lcdc panel size
  sh: select memchunk size using kernel cmdline
  sh: export sh7723 VEU as VEU2H
  input: migor_ts compile and detection fix
  sh: remove MSTPCR defines from Migo-R header file
  sh: Update sh7763rdp defconfig
  sh: Add support sh7760fb to sh7763rdp board
  sh: Add support sh_eth to sh7763rdp board
  sh: Disable 64kB hugetlbpage size when using 64kB PAGE_SIZE.
  sh: Don't export __{s,u}divsi3_i4i from SH-2 libgcc.
  fix SH7705_CACHE_32KB compilation
  sh: mach-x3proto: Fix up smc91x platform data.

16 years agoMerge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc
Linus Torvalds [Wed, 20 Aug 2008 15:44:33 +0000 (08:44 -0700)]
Merge branch 'merge' of git://git./linux/kernel/git/paulus/powerpc

* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc:
  powerpc: Fix vio_bus_probe oops on probe error
  powerpc/ibmebus: Restore "name" sysfs attribute on ibmebus devices
  powerpc: Fix /dev/oldmem interface for kdump
  powerpc/spufs: Remove invalid semicolon after if statement
  powerpc/spufs: reference context while dropping state mutex in scheduler
  powerpc/spufs: fix npc setting for NOSCHED contexts

16 years agoMerge branch 'tracehook' of git://git.kernel.org/pub/scm/linux/kernel/git/frob/linux...
Linus Torvalds [Wed, 20 Aug 2008 15:43:47 +0000 (08:43 -0700)]
Merge branch 'tracehook' of git://git./linux/kernel/git/frob/linux-2.6-utrace

* 'tracehook' of git://git.kernel.org/pub/scm/linux/kernel/git/frob/linux-2.6-utrace:
  tracehook: fix SA_NOCLDWAIT

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6
Linus Torvalds [Wed, 20 Aug 2008 15:42:53 +0000 (08:42 -0700)]
Merge git://git./linux/kernel/git/jejb/scsi-rc-fixes-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6: (22 commits)
  [SCSI] ibmvfc: Driver version 1.0.2
  [SCSI] ibmvfc: Add details to async event log
  [SCSI] ibmvfc: Sanitize response lengths
  [SCSI] ibmvfc: Fix for lost async events
  [SCSI] ibmvfc: Fixup host state during reinit
  [SCSI] ibmvfc: Fix another hang on module removal
  [SCSI] ibmvscsi: Fixup desired DMA value for shared memory partitions
  [SCSI] megaraid_sas: remove sysfs dbg_lvl world writeable permissions
  [SCSI] qla2xxx: Update version number to 8.02.01-k7.
  [SCSI] qla2xxx: Explicitly tear-down vports during PCI remove_one().
  [SCSI] qla2xxx: Reference proper ha during SBR handling.
  [SCSI] qla2xxx: Set npiv_supported flag for FCoE HBAs.
  [SCSI] qla2xxx: Don't leak SG-DMA mappings while aborting commands.
  [SCSI] qla2xxx: Correct vport-state management issues during ISP-ABORT.
  [SCSI] qla2xxx: Correct synchronization of software/firmware fcport states.
  [SCSI] scsi_dh: Initialize lun_state in check_ownership()
  [SCSI] scsi_dh: Do not use scsilun in rdac hardware handler
  [SCSI] megaraid_sas: version and Documentation Update
  [SCSI] megaraid_sas: add new controllers (0x78 0x79)
  [SCSI] megaraid_sas: add the shutdown DCMD cmd to driver shutdown routine
  ...

16 years agovfat: fix 'sync' mount deadlock due to BKL->lock_super conversion
Linus Torvalds [Wed, 20 Aug 2008 15:31:19 +0000 (08:31 -0700)]
vfat: fix 'sync' mount deadlock due to BKL->lock_super conversion

There was another FAT BKL conversion deadlock reported by Bart
Trojanowski due to the BKL being used as a recursive lock by FAT, which
was missed because it only triggers with 'sync' (or 'dirsync') mounts.

The recursion worked for the BKL, but after the conversion to lock_super
(which uses a mutex), it just deadlocks.

Thanks to Bart for debugging this and testing the fix.  The lock
debugging information from the original report:

  =============================================
  [ INFO: possible recursive locking detected ]
  2.6.27-rc3-bisect-00448-ga7f5aaf #16
  ---------------------------------------------
  mv/4020 is trying to acquire lock:
   (&type->s_lock_key#9){--..}, at: [<c01a90fe>] lock_super+0x1e/0x20

  but task is already holding lock:
   (&type->s_lock_key#9){--..}, at: [<c01a90fe>] lock_super+0x1e/0x20

  other info that might help us debug this:
  3 locks held by mv/4020:
   #0:  (&sb->s_type->i_mutex_key#9/1){--..}, at: [<c01b2336>] do_unlinkat+0x66/0x140
   #1:  (&sb->s_type->i_mutex_key#9){--..}, at: [<c01b0954>] vfs_unlink+0x84/0x110
   #2:  (&type->s_lock_key#9){--..}, at: [<c01a90fe>] lock_super+0x1e/0x20

  stack backtrace:
  Pid: 4020, comm: mv Not tainted 2.6.27-rc3-bisect-00448-ga7f5aaf #16
   [<c014e694>] validate_chain+0x984/0xea0
   [<c0108d70>] ? native_sched_clock+0x0/0xf0
   [<c014ee9c>] __lock_acquire+0x2ec/0x9b0
   [<c014f5cf>] lock_acquire+0x6f/0x90
   [<c01a90fe>] ? lock_super+0x1e/0x20
   [<c044e5fd>] mutex_lock_nested+0xad/0x300
   [<c01a90fe>] ? lock_super+0x1e/0x20
   [<c01a90fe>] ? lock_super+0x1e/0x20
   [<c01a90fe>] lock_super+0x1e/0x20
   [<f8b3a700>] fat_write_inode+0x60/0x2b0 [fat]
   [<c0450878>] ? _spin_unlock_irqrestore+0x48/0x80
   [<f8b3a953>] ? fat_sync_inode+0x3/0x20 [fat]
   [<f8b3a962>] fat_sync_inode+0x12/0x20 [fat]
   [<f8b37c7e>] fat_remove_entries+0xbe/0x120 [fat]
   [<f8b422ef>] vfat_unlink+0x5f/0x90 [vfat]
   [<f8b42290>] ? vfat_unlink+0x0/0x90 [vfat]
   [<c01b0968>] vfs_unlink+0x98/0x110
   [<c01b2400>] do_unlinkat+0x130/0x140
   [<c016a8f5>] ? audit_syscall_entry+0x105/0x150
   [<c01b253b>] sys_unlinkat+0x3b/0x40
   [<c01040d3>] sysenter_do_call+0x12/0x3f
   =======================

where the deadlock is due to the nesting of lock_super from vfat_unlink
to fat_write_inode:

 - do_unlinkat
   - vfs_unlink
     - vfat_unlink
       * lock_super
       - fat_remove_entries
         - fat_sync_inode
           - fat_write_inode
             * lock_super

and the fix is to simply remove the use of lock_super() in fat_write_inode.

The lock_super() there had been just an automatic conversion of the
kernel lock to the superblock lock, but no locking was actually needed
there, since the code in fat_write_inode already protected all relevant
accesses with a spinlock (sbi->inode_hash_lock to be exact).  The only
code inside the BKL (and thus the superblock lock) was accesses tp local
variables or calls to functions that have long been SMP-safe (i.e.
sb_bread, mark_buffe_dirty and brlese).

Bart reports:
 "Looks good.  I ran 10 parallel processes creating 1M files truncating
  them, writing to them again and then deleting them.  This patch fixes
  the issue I ran into.

Signed-off-by: Bart Trojanowski <bart@jukie.net>"
Reported-and-tested-by: Bart Trojanowski <bart@jukie.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agotracehook: fix SA_NOCLDWAIT
Roland McGrath [Wed, 20 Aug 2008 03:37:07 +0000 (20:37 -0700)]
tracehook: fix SA_NOCLDWAIT

I outwitted myself again in commit 2b2a1ff64afbadac842bbc58c5166962cf4f7664,
and broke the SA_NOCLDWAIT behavior so it leaks zombies.  This fixes it.

Reported-by: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Roland McGrath <roland@redhat.com>
16 years agopowerpc: Fix vio_bus_probe oops on probe error
Brian King [Tue, 12 Aug 2008 19:21:45 +0000 (05:21 +1000)]
powerpc: Fix vio_bus_probe oops on probe error

When CMO is enabled and booted on a non CMO system and the VIO
device's probe function fails, an oops can result since
vio_cmo_bus_remove is called when it should not.  This fixes it by
avoiding the vio_cmo_bus_remove call on platforms that don't implement
CMO.

cpu 0x0: Vector: 300 (Data Access) at [c00000000e13b3d0]
    pc: c000000000020d34: .vio_cmo_bus_remove+0xc0/0x1f4
    lr: c000000000020ca4: .vio_cmo_bus_remove+0x30/0x1f4
    sp: c00000000e13b650
   msr: 8000000000009032
   dar: 0
 dsisr: 40000000
  current = 0xc00000000e0566c0
  paca    = 0xc0000000006f9b80
    pid   = 2428, comm = modprobe
enter ? for help
[c00000000e13b6e0c000000000021d94 .vio_bus_probe+0x2f8/0x33c
[c00000000e13b7a0c00000000029fc88 .driver_probe_device+0x13c/0x200
[c00000000e13b830c00000000029fdac .__driver_attach+0x60/0xa4
[c00000000e13b8c0c00000000029f050 .bus_for_each_dev+0x80/0xd8
[c00000000e13b980c00000000029f9ec .driver_attach+0x28/0x40
[c00000000e13ba00c00000000029f630 .bus_add_driver+0xd4/0x284
[c00000000e13baa0c0000000002a01bc .driver_register+0xc4/0x198
[c00000000e13bb50c00000000002168c .vio_register_driver+0x40/0x5c
[c00000000e13bbe0d0000000003b3f1c .ibmvfc_module_init+0x70/0x109c [ibmvfc]
[c00000000e13bc70c0000000000acf08 .sys_init_module+0x184c/0x1a10
[c00000000e13be30c000000000008748 syscall_exit+0x0/0x40

Signed-off-by: Brian King <brking@linux.vnet.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
16 years agopowerpc/ibmebus: Restore "name" sysfs attribute on ibmebus devices
Joachim Fenkes [Tue, 5 Aug 2008 14:30:18 +0000 (00:30 +1000)]
powerpc/ibmebus: Restore "name" sysfs attribute on ibmebus devices

Recent of_platform changes made of_bus_type_init() overwrite the bus
type's .dev_attrs list, meaning that the "name" attribute that ibmebus
devices previously had is no longer present.  This is a user-visible
regression which breaks the userspace eHCA support, since the eHCA
userspace driver relies on the name attribute to check for valid
adapters.

This fixes it by providing the "name" attribute in the generic OF
device code instead.  Tested on POWER.

Signed-off-by: Joachim Fenkes <fenkes@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
16 years agopowerpc: Fix /dev/oldmem interface for kdump
Michael Ellerman [Thu, 31 Jul 2008 06:54:28 +0000 (16:54 +1000)]
powerpc: Fix /dev/oldmem interface for kdump

A change to __ioremap() broke reading /dev/oldmem because we're no
longer able to ioremap pfn 0 (d177c207, "[PATCH] powerpc: IOMMU: don't
ioremap null addresses").

We actually don't need to ioremap for anything that's part of the linear
mapping, so just read it directly.

Also make sure we're only reading one page or less at a time.

Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Sachin Sant <sachinp@in.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
16 years agoMerge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/jk/spufs into...
Paul Mackerras [Tue, 19 Aug 2008 23:18:20 +0000 (09:18 +1000)]
Merge branch 'merge' of git://git./linux/kernel/git/jk/spufs into merge

16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394...
Linus Torvalds [Tue, 19 Aug 2008 20:56:21 +0000 (13:56 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/ieee1394/linux1394-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ieee1394/linux1394-2.6:
  firewire: Kconfig help update
  ieee1394: sbp2: let nodemgr retry node updates during bus reset series
  ieee1394: don't drop nodes during bus reset series
  ieee1394: regression in 2.6.25: updates should happen before probes

16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes...
Linus Torvalds [Tue, 19 Aug 2008 20:55:47 +0000 (13:55 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jbarnes/pci-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci-2.6:
  PCI: add acpi_find_root_bridge_handle
  PCI: acpi_pcihp: run _OSC on a root bridge
  x86/PCI: irq and pci_ids patch for Intel Ibex Peak PCHs
  x86/PCI: allow scanning of 255 PCI busses
  x86, pci: detect end_bus_number according to acpi/e820 reserved, v2
  pci: debug extra pci bus resources
  pci: debug extra pci resources range

16 years agoRevert "[CPUFREQ][2/2] preregister support for powernow-k8"
Linus Torvalds [Tue, 19 Aug 2008 20:34:59 +0000 (13:34 -0700)]
Revert "[CPUFREQ][2/2] preregister support for powernow-k8"

This reverts commit 34ae7f35a21694aa5cb8829dc5142c39d73d6ba0, which has
been reported to cause a number of problems.  During suspend and resume,
it apparently causes a crash in a CPU hotplug notifier to happen,
although the exact details are sketchy because of the inability to get
good traces during the suspend sequence.

See buzilla entries

http://bugzilla.kernel.org/show_bug.cgi?id=11296
http://bugzilla.kernel.org/show_bug.cgi?id=11339

for more examples and details.

[ Mark: "Revert the patch for now.  I'm still looking into getting a
  reliable reproduction and I do not have a fix at this time." ]

Requested-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Mark Langsdorf <mark.langsdorf@amd.com>
Acked-by: Dave Jones <davej@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@inux-foundation.org>
16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Linus Torvalds [Tue, 19 Aug 2008 20:34:33 +0000 (13:34 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/dtor/input

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: evdev - fix printf() format for sizeof
  Input: remove version.h from drivers that don't need it
  Input: cobalt_btns - add missing MODULE_LICENSE

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Tue, 19 Aug 2008 16:59:02 +0000 (09:59 -0700)]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (94 commits)
  pkt_sched: Prevent livelock in TX queue running.
  Revert "pkt_sched: Add BH protection for qdisc_stab_lock."
  Revert "pkt_sched: Protect gen estimators under est_lock."
  pkt_sched: remove bogus block (cleanup)
  nf_nat: use secure_ipv4_port_ephemeral() for NAT port randomization
  netfilter: ctnetlink: sleepable allocation with spin lock bh
  netfilter: ctnetlink: fix sleep in read-side lock section
  netfilter: ctnetlink: fix double helper assignation for NAT'ed conntracks
  netfilter: ipt_addrtype: Fix matching of inverted destination address type
  dccp: Fix panic caused by too early termination of retransmission mechanism
  pkt_sched: Don't hold qdisc lock over qdisc_destroy().
  pkt_sched: Add lockdep annotation for qdisc locks
  pkt_sched: Never schedule non-root qdiscs.
  removed unused #include <version.h>
  rt2x00: Fix txdone_entry_desc_flags
  b43: Fix for another Bluetooth Coexistence SPROM Programming error for BCM4306
  mac80211: remove kdoc references to IEEE80211_HW_HOST_GEN_BEACON_TEMPLATE
  p54u: reset skb's data/tail pointer on requeue
  p54: move p54_vdcf_init to the right place.
  iwlwifi: fix printk newlines
  ...

16 years agofirewire: Kconfig help update
Stefan Richter [Sat, 16 Aug 2008 19:52:28 +0000 (21:52 +0200)]
firewire: Kconfig help update

Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
16 years agoieee1394: sbp2: let nodemgr retry node updates during bus reset series
Stefan Richter [Sat, 16 Aug 2008 11:39:26 +0000 (13:39 +0200)]
ieee1394: sbp2: let nodemgr retry node updates during bus reset series

sbp2 was too quick to report .update() to the ieee1394 core as failed.
(Logged as "Failed to reconnect to sbp2 device!".)  The core would then
unbind sbp2 from the device.

This is not justified if the .update() failed because another bus reset
happened.  We check this and tell the ieee1394 that .update() succeeded,
and the core will call sbp2's .update() for the new bus reset as well.

This improves reconnection/re-login especially on buses with several
disks as they may issue bus resets in close succession when they come
online.

Tested by Damien Benoist.

Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
16 years agoieee1394: don't drop nodes during bus reset series
Stefan Richter [Sat, 16 Aug 2008 11:38:11 +0000 (13:38 +0200)]
ieee1394: don't drop nodes during bus reset series

nodemgr_node_probe checked for generation increments too late and
therefore prematurely reported nodes as "suspended".

Fixes http://bugzilla.kernel.org/show_bug.cgi?id=11349.  Reported and
tested by Damien Benoist.

Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
16 years agoieee1394: regression in 2.6.25: updates should happen before probes
Stefan Richter [Sat, 16 Aug 2008 11:36:47 +0000 (13:36 +0200)]
ieee1394: regression in 2.6.25: updates should happen before probes

Regression since commit 73cf60232ef16e1f8a64defa97214a1722db1e6c,
"ieee1394: use class iteration api":  The two loops for (1.) driver
updates and (2.) driver probes were replaced by a single loop with
bogus needs_probe checks.  Hence updates and probes were now intermixed,
and especially sbp2 updates (reconnects) held up longer than necessary.

While we fix it, change the needs_probe flag to bool type for clarity.

Tested by Damien Benoist.

Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
16 years agoInput: evdev - fix printf() format for sizeof
Geert Uytterhoeven [Tue, 19 Aug 2008 15:28:23 +0000 (11:28 -0400)]
Input: evdev - fix printf() format for sizeof

commit f2afa7711f8585ffc088ba538b9a510e0d5dca12 ("Input: paper over a bug in
Synaptics X driver") introduced a compiler warning on 64-bit platforms, as
sizeof() returns a size_t, not an (unsigned) int:

| drivers/input/evdev.c: In function 'handle_eviocgbit':
| drivers/input/evdev.c:684: warning: format '%d' expects type 'int', but argument 3 has type 'long unsigned int'

Use the proper `z' modifier for size_t, and make the printf() formats for the
sizes unsigned while we're at it.

Signed-off-by: Geert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
16 years agopowerpc/spufs: Remove invalid semicolon after if statement
Ilpo Järvinen [Tue, 19 Aug 2008 05:48:57 +0000 (08:48 +0300)]
powerpc/spufs: Remove invalid semicolon after if statement

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
16 years agopkt_sched: Prevent livelock in TX queue running.
David S. Miller [Tue, 19 Aug 2008 11:00:36 +0000 (04:00 -0700)]
pkt_sched: Prevent livelock in TX queue running.

If dev_deactivate() is trying to quiesce the queue, it
is theoretically possible for another cpu to livelock
trying to process that queue.  This happens because
dev_deactivate() grabs the queue spinlock as it checks
the queue state, whereas net_tx_action() does a trylock
and reschedules the qdisc if it hits the lock.

This breaks the livelock by adding a check on
__QDISC_STATE_DEACTIVATED to net_tx_action() when
the trylock fails.

Based upon feedback from Herbert Xu and Jarek Poplawski.

Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/holtmann/bluet...
David S. Miller [Tue, 19 Aug 2008 08:33:25 +0000 (01:33 -0700)]
Merge branch 'master' of git://git./linux/kernel/git/holtmann/bluetooth-2.6

16 years agoRevert "pkt_sched: Add BH protection for qdisc_stab_lock."
David S. Miller [Tue, 19 Aug 2008 05:33:05 +0000 (22:33 -0700)]
Revert "pkt_sched: Add BH protection for qdisc_stab_lock."

This reverts commit 1cfa26661a85549063e369e2b40275eeaa7b923c.

qdisc_destroy() runs fully under RTNL again and not from softint any
longer, so this change is no longer needed.

Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoRevert "pkt_sched: Protect gen estimators under est_lock."
David S. Miller [Tue, 19 Aug 2008 05:32:10 +0000 (22:32 -0700)]
Revert "pkt_sched: Protect gen estimators under est_lock."

This reverts commit d4766692e72422f3b0f0e9ac6773d92baad07d51.

qdisc_destroy() now runs in RTNL fully again, so this
change is no longer needed.

Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agopkt_sched: remove bogus block (cleanup)
Ilpo Järvinen [Tue, 19 Aug 2008 05:30:01 +0000 (22:30 -0700)]
pkt_sched: remove bogus block (cleanup)

...Last block local var got just deleted.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agonf_nat: use secure_ipv4_port_ephemeral() for NAT port randomization
Stephen Hemminger [Tue, 19 Aug 2008 04:32:32 +0000 (21:32 -0700)]
nf_nat: use secure_ipv4_port_ephemeral() for NAT port randomization

Use incoming network tuple as seed for NAT port randomization.
This avoids concerns of leaking net_random() bits, and also gives better
port distribution. Don't have NAT server, compile tested only.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
[ added missing EXPORT_SYMBOL_GPL ]

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agonetfilter: ctnetlink: sleepable allocation with spin lock bh
Pablo Neira Ayuso [Tue, 19 Aug 2008 04:31:46 +0000 (21:31 -0700)]
netfilter: ctnetlink: sleepable allocation with spin lock bh

This patch removes a GFP_KERNEL allocation while holding a spin lock with
bottom halves disabled in ctnetlink_change_helper().

This problem was introduced in 2.6.23 with the netfilter extension
infrastructure.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agonetfilter: ctnetlink: fix sleep in read-side lock section
Pablo Neira Ayuso [Tue, 19 Aug 2008 04:31:24 +0000 (21:31 -0700)]
netfilter: ctnetlink: fix sleep in read-side lock section

Fix allocation with GFP_KERNEL in ctnetlink_create_conntrack() under
read-side lock sections.

This problem was introduced in 2.6.25.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agonetfilter: ctnetlink: fix double helper assignation for NAT'ed conntracks
Pablo Neira Ayuso [Tue, 19 Aug 2008 04:30:55 +0000 (21:30 -0700)]
netfilter: ctnetlink: fix double helper assignation for NAT'ed conntracks

If we create a conntrack that has NAT handlings and a helper, the helper
is assigned twice. This happens because nf_nat_setup_info() - via
nf_conntrack_alter_reply() - sets the helper before ctnetlink, which
indeed does not check if the conntrack already has a helper as it thinks that
it is a brand new conntrack.

The fix moves the helper assignation before the set of the status flags.
This avoids a bogus assertion in __nf_ct_ext_add (if netfilter assertions are
enabled) which checks that the conntrack must not be confirmed.

This problem was introduced in 2.6.23 with the netfilter extension
infrastructure.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Patrick McHardy <kaber@trash.net>
16 years agonetfilter: ipt_addrtype: Fix matching of inverted destination address type
Anders Grafström [Tue, 19 Aug 2008 04:29:57 +0000 (21:29 -0700)]
netfilter: ipt_addrtype: Fix matching of inverted destination address type

This patch fixes matching of inverted destination address type.

Signed-off-by: Anders Grafström <grfstrm@users.sourceforge.net>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wirel...
David S. Miller [Tue, 19 Aug 2008 04:15:44 +0000 (21:15 -0700)]
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless-2.6

16 years agodccp: Fix panic caused by too early termination of retransmission mechanism
Gerrit Renker [Tue, 19 Aug 2008 04:14:20 +0000 (21:14 -0700)]
dccp: Fix panic caused by too early termination of retransmission mechanism

Thanks is due to Wei Yongjun for the detailed analysis and description of this
bug at http://marc.info/?l=dccp&m=121739364909199&w=2

The problem is that invalid packets received by a client in state REQUEST cause
the retransmission timer for the DCCP-Request to be reset. This includes freeing
the Request-skb ( in dccp_rcv_request_sent_state_process() ). As a consequence,
 * the arrival of further packets cause a double-free, triggering a panic(),
 * the connection then may hang, since further retransmissions are blocked.

This patch changes the order of statements so that the retransmission timer is
reset, and the pending Request freed, only if a valid Response has arrived (or
the number of sysctl-retries has been exhausted).

Further changes:
----------------
To be on the safe side, replaced __kfree_skb with kfree_skb so that if due to
unexpected circumstances the sk_send_head is NULL the WARN_ON is used instead.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agopkt_sched: Don't hold qdisc lock over qdisc_destroy().
David S. Miller [Tue, 19 Aug 2008 04:03:15 +0000 (21:03 -0700)]
pkt_sched: Don't hold qdisc lock over qdisc_destroy().

Based upon reports by Denys Fedoryshchenko, and feedback
and help from Jarek Poplawski and Herbert Xu.

We always either:

1) Never made an external reference to this qdisc.

or

2) Did a dev_deactivate() which purged all asynchronous
   references.

So do not lock the qdisc when we call qdisc_destroy(),
it's illegal anyways as when we drop the lock this is
free'd memory.

Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agopkt_sched: Add lockdep annotation for qdisc locks
Jarek Poplawski [Tue, 19 Aug 2008 03:53:34 +0000 (20:53 -0700)]
pkt_sched: Add lockdep annotation for qdisc locks

Qdisc locks are initialized in the same function, qdisc_alloc(), so
lockdep can't distinguish tx qdisc lock from rx and reports "possible
recursive locking detected" when both these locks are taken eg. while
using act_mirred with ifb. This looks like a false positive. Anyway,
after this patch these locks will be reported more exactly.

Reported-by: Denys Fedoryshchenko <denys@visp.net.lb>
Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agopkt_sched: Never schedule non-root qdiscs.
David S. Miller [Tue, 19 Aug 2008 03:51:18 +0000 (20:51 -0700)]
pkt_sched: Never schedule non-root qdiscs.

Based upon initial discovery and patch by Jarek Poplawski.

The qdisc watchdogs can be attached to any qdisc, not just the root,
so make sure we schedule the correct one.

CBQ has a similar bug.

Signed-off-by: David S. Miller <davem@davemloft.net>