platform/adaptation/renesas_rcar/renesas_kernel.git
14 years agosfc: QT202x: Remove unreliable MMD check at initialisation
Matthew Slattery [Mon, 18 Jan 2010 05:47:16 +0000 (05:47 +0000)]
sfc: QT202x: Remove unreliable MMD check at initialisation

Checking the PHY XS MMD here is unnecessary and can give false negatives.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoucc_geth: Fix full TX queue processing
Jiajun Wu [Mon, 18 Jan 2010 05:47:50 +0000 (05:47 +0000)]
ucc_geth: Fix full TX queue processing

commit 7583605b6d29f1f7f6fc505b883328089f3485ad ("ucc_geth: Fix empty
TX queue processing") fixed empty TX queue mishandling, but didn't
account another corner case: when TX queue becomes full.

Without this patch the driver will stop transmiting when TX queue
becomes full since 'bd == ugeth->txBd[txQ]' actually checks for
two things: queue empty or full.

Let's better check for NULL skb, which unambiguously signals an empty
queue.

Signed-off-by: Jiajun Wu <b06378@freescale.com>
Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Cc: Stable <stable@vger.kernel.org> [2.6.32]
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agophylib: Move workqueue initialization to a proper place
Anton Vorontsov [Mon, 18 Jan 2010 05:37:16 +0000 (05:37 +0000)]
phylib: Move workqueue initialization to a proper place

commit 541cd3ee00a4fe975b22fac6a3bc846bacef37f7 ("phylib: Fix deadlock
on resume") caused TI DaVinci EMAC ethernet driver to oops upon resume:

 PM: resume of devices complete after 237.098 msecs
 Restarting tasks ... done.
 kernel BUG at kernel/workqueue.c:354!
 Unable to handle kernel NULL pointer dereference at virtual address 00000000
 [...]
 Backtrace:
 [<c002c598>] (__bug+0x0/0x2c) from [<c0052a54>] (queue_delayed_work_on+0x74/0xf8)
 [<c00529e0>] (queue_delayed_work_on+0x0/0xf8) from [<c0052b30>] (queue_delayed_work+0x2c/0x30)

The oops pops up because TI DaVinci EMAC driver detaches PHY on
suspend and attaches it back on resume. Attaching makes phylib call
phy_start_machine() that initializes a workqueue. On the other hand,
PHY's resume routine will call phy_start_machine() again, and that
will cause the oops since we just destroyed the already scheduled
workqueue.

This patch fixes the issue by moving workqueue initialization to
phy_device_create().

p.s. We don't see this oops with ucc_geth and gianfar drivers because
they perform a fine-grained suspend, i.e. they just stop the PHYs
without detaching.

Reported-by: Sekhar Nori <nsekhar@ti.com>
Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agodccp: fix dccp rmmod when kernel configured to use slub
Neil Horman [Sun, 17 Jan 2010 17:16:12 +0000 (17:16 +0000)]
dccp: fix dccp rmmod when kernel configured to use slub

Hey all-
I was tinkering with dccp recently and noticed that I BUG halted the
kernel when I rmmod-ed the dccp module.  The bug halt occured because the page
that I passed to kfree failed the PageCompound and PageSlab test in the slub
implementation of kfree.  I tracked the problem down to the following set of
events:

1) dccp, unlike all other uses of kmem_cache_create, allocates a string
dynamically when registering a slab cache.  This allocated string is freed when
the cache is destroyed.

2) Normally, (1) is not an issue, but when Slub is in use, it is possible that
caches are 'merged'.  This process causes multiple caches of simmilar
configuration to use the same cache data structure.  When this happens, the new
name of the cache is effectively dropped.

3) (2) results in kmem_cache_name returning an ambigous value (i.e.
ccid_kmem_cache_destroy, which uses this fuction to retrieve the name pointer
for freeing), is no longer guaranteed that the string it assigned is what is
returned.

4) If such merge event occurs, ccid_kmem_cache_destroy frees the wrong pointer,
which trips over the BUG in the slub implementation of kfree (since its likely
not a slab allocation, but rather a pointer into the static string table
section.

So, what to do about this.  At first blush this is pretty clearly a leak in the
information that slub owns, and as such a slub bug.  Unfortunately, theres no
really good way to fix it, without exposing slub specific implementation details
to the generic slab interface.  Also, even if we could fix this in slub cleanly,
I think the RCU free option would force us to do lots of string duplication, not
only in slub, but in every slab allocator.  As such, I'd like to propose this
solution.  Basically, I just move the storage for the kmem cache name to the
ccid_operations structure.  In so doing, we don't have to do the kstrdup or
kfree when we allocate/free the various caches for dccp, and so we avoid the
problem, by storing names with static memory, rather than heap, the way all
other calls to kmem_cache_create do.

I've tested this out myself here, and it solves the problem quite well.

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoipv4: don't remove /proc/net/rt_acct
Alexey Dobriyan [Sun, 17 Jan 2010 03:32:50 +0000 (03:32 +0000)]
ipv4: don't remove /proc/net/rt_acct

/proc/net/rt_acct is not created if NET_CLS_ROUTE=n.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoax25: netrom: rose: Fix timer oopses
Jarek Poplawski [Sat, 16 Jan 2010 09:04:04 +0000 (01:04 -0800)]
ax25: netrom: rose: Fix timer oopses

Wrong ax25_cb refcounting in ax25_send_frame() and by its callers can
cause timer oopses (first reported with 2.6.29.6 kernel).

Fixes: http://bugzilla.kernel.org/show_bug.cgi?id=14905

Reported-by: Bernard Pidoux <bpidoux@free.fr>
Tested-by: Bernard Pidoux <bpidoux@free.fr>
Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoappletalk:: da.s_net not copied but assigned to itself in aarp_rcv()
Roel Kluin [Fri, 15 Jan 2010 09:49:28 +0000 (01:49 -0800)]
appletalk:: da.s_net not copied but assigned to itself in aarp_rcv()

da.s_net was not copied but assigned to itself.

Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoMAINTAINERS: transfer maintainership of I/OAT
Dan Williams [Fri, 15 Jan 2010 09:47:37 +0000 (01:47 -0800)]
MAINTAINERS: transfer maintainership of I/OAT

Dan Williams takes over I/OAT from Maciej Sosnowski

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Maciej Sosnowski <maciej.sosnowski@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agodccp_probe: Fix module load dependencies between dccp and dccp_probe
Neil Horman [Fri, 15 Jan 2010 09:40:55 +0000 (01:40 -0800)]
dccp_probe: Fix module load dependencies between dccp and dccp_probe

This was just recently reported to me.  When built as modules, the
dccp_probe module has a silent dependency on the dccp module.  This
stems from the fact that the module_init routine of dccp_probe
registers a jprobe on the dccp_sendmsg symbol.  Since the symbol is
only referenced as a text string (the .symbol_name field in the jprobe
struct) rather than the address of the symbol itself, depmod never
picks this dependency up, and so if you load the dccp_probe module
without the dccp module loaded, the register_jprobe call fails with an
-EINVAL, and the whole module load fails.

The fix is pretty easy, we can just wrap the register_jprobe call in a
try_then_request_module call, which forces the dependency to get
satisfied prior to the probe registration.

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet: fix build erros with CONFIG_BUG=n, CONFIG_GENERIC_BUG=n
Octavian Purdila [Thu, 14 Jan 2010 02:10:36 +0000 (18:10 -0800)]
net: fix build erros with CONFIG_BUG=n, CONFIG_GENERIC_BUG=n

Fixed build errors introduced by commit 7ad6848c (ip: fix mc_loop
checks for tunnels with multicast outer addresses)

Signed-off-by: Octavian Purdila <opurdila@ixiacom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoipv6: skb_dst() can be NULL in ipv6_hop_jumbo().
David S. Miller [Thu, 14 Jan 2010 01:27:37 +0000 (17:27 -0800)]
ipv6: skb_dst() can be NULL in ipv6_hop_jumbo().

This fixes CERT-FI FICORA #341748

Discovered by Olli Jarva and Tuomo Untinen from the CROSS
project at Codenomicon Ltd.

Just like in CVE-2007-4567, we can't rely upon skb_dst() being
non-NULL at this point.  We fixed that in commit
e76b2b2567b83448c2ee85a896433b96150c92e6 ("[IPV6]: Do no rely on
skb->dst before it is assigned.")

However commit 483a47d2fe794328d29950fe00ce26dd405d9437 ("ipv6: added
net argument to IP6_INC_STATS_BH") put a new version of the same bug
into this function.

Complicating analysis further, this bug can only trigger when network
namespaces are enabled in the build.  When namespaces are turned off,
the dev_net() does not evaluate it's argument, so the dereference
would not occur.

So, for a long time, namespaces couldn't be turned on unless SYSFS was
disabled.  Therefore, this code has largely been disabled except by
people turning it on explicitly for namespace development.

With help from Eugene Teo <eugene@redhat.com>

Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agotg3: Update copyright and driver version
Matt Carlson [Tue, 12 Jan 2010 10:11:40 +0000 (10:11 +0000)]
tg3: Update copyright and driver version

This patch updates the copyright notice for 2010 and updates the version
number to 3.106.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agotg3: Disable 5717 serdes and B0 support
Matt Carlson [Tue, 12 Jan 2010 10:11:39 +0000 (10:11 +0000)]
tg3: Disable 5717 serdes and B0 support

The B0 revision of the 5717 will not get enough testing by the time
2.6.33 ships.  Since the kernel is already at RC3, serdes support
will require too many patches to fix.  For these reasons, this patch
disables 5717 serdes support and will refuse to attach to all 5717
devices that are later than an A0 revision.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agotg3: Add reliable serdes detection for 5717 A0
Matt Carlson [Tue, 12 Jan 2010 10:11:38 +0000 (10:11 +0000)]
tg3: Add reliable serdes detection for 5717 A0

The serdes status bit does not work as intended for the 5717 A0.
This patch implements an alternative detection scheme that will only be
valid for A0 revisions.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agotg3: Fix std rx prod ring handling
Matt Carlson [Tue, 12 Jan 2010 10:11:37 +0000 (10:11 +0000)]
tg3: Fix std rx prod ring handling

There are some tg3 devices that require the driver to post new rx
buffers in smaller increments.  Commit
4361935afe3abc3e5a93006b99197fac1fabbd50, "tg3: Consider
rx_std_prod_idx a hw mailbox" changed how the driver tracks the rx
producer ring updates, but it does not make any special considerations
for the above-mentioned devices.  For those devices, it is possible for
the driver to hit the special case path, which updates the hardware
mailbox register but skips updating the shadow software mailbox member.
If the special case path represents the final mailbox update for this
ISR iteration, the hardware and software mailbox values will be out of
sync.  Ultimately, this will cause the driver to use a stale mailbox
value on the next iteration, which will appear to the hardware as a
large rx buffer update.  Bad things ensue.

The fix is to update the software shadow mailbox member when the special
case path is taken.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reported-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agotg3: Fix std prod ring nicaddr for 5787 and 57765
Matt Carlson [Tue, 12 Jan 2010 10:11:36 +0000 (10:11 +0000)]
tg3: Fix std prod ring nicaddr for 5787 and 57765

Commit 87668d352aa8d135bd695a050f18bbfc7b50b506, titled "tg3: Don't
touch RCB nic addresses", tried to avoid assigning the nic address of
the standard producer ring.  Unfortunately, the default nic address is
not correct for the 5787, the 5755M, or the 57765.  This patch
reenables the old behavior and opts out of the assignment only
for the 5717.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Tested-by: Chow Loong Jin <hyperair@ubuntu.com>
Tested-by: Dmitry Torokhov <dtor@mail.ru>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Fix conditions for MDIO self-test
Ben Hutchings [Wed, 13 Jan 2010 10:59:13 +0000 (10:59 +0000)]
sfc: Fix conditions for MDIO self-test

The MDIO self-test should not be run on boards without an MDIO PHY,
such as SFN5122F-R3 and later revisions.  It should also not try to
address a specific MMD in an MDIO clause 22 PHY.  Check the
mode_support field to decide which mode to use, if any.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosfc: Fix polling for slow MCDI operations
Ben Hutchings [Wed, 13 Jan 2010 04:34:25 +0000 (04:34 +0000)]
sfc: Fix polling for slow MCDI operations

When the interface is down and we are using polled mode for MCDI
operations, we busy-wait for completion for approximately 1 jiffy
using udelay() and then back off to schedule().  But the completion
will not wake the task, since we are using polled mode!  We must use
schedule_timeout_uninterruptible() instead.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoe1000e: workaround link issues on busy hub in half duplex on 82577/82578
Bruce Allan [Wed, 13 Jan 2010 01:53:08 +0000 (01:53 +0000)]
e1000e: workaround link issues on busy hub in half duplex on 82577/82578

This patch removes a delay in hardware after every received packet allowing
more time for transmitted packets to go out in between received packets in
half duplex.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoe1000e: MDIO slow mode should always be done for 82577
Bruce Allan [Wed, 13 Jan 2010 01:52:49 +0000 (01:52 +0000)]
e1000e: MDIO slow mode should always be done for 82577

A previous 82577 workaround that set the MDIO access speed to slow mode for
every PHY register read/write when the cable is unplugged should instead
set the access mode to always be slow before any PHY register access.
Since the mode bit gets cleared when the PHY is reset, set the mode after
every PHY reset.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoixgbe: update copyright dates
Shannon Nelson [Wed, 13 Jan 2010 01:49:34 +0000 (01:49 +0000)]
ixgbe: update copyright dates

Signed-off-by: Shannon Nelson <shannon.nelson@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoixgbe: Do not attempt to perform interrupts in netpoll when down
Alexander Duyck [Wed, 13 Jan 2010 01:49:13 +0000 (01:49 +0000)]
ixgbe: Do not attempt to perform interrupts in netpoll when down

This patch resolves issues seen when running netconsole and rebooting via
reboot -f.  The issue was due to the fact that we were attempting to
perform interrupt actions when the q_vectors and rings had already been
freed via the ixgbe_shutdown routines.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Acked-by: Mallikarjuna R Chilakala <mallikarjuna.chilakala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wirel...
David S. Miller [Wed, 13 Jan 2010 05:33:49 +0000 (21:33 -0800)]
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless-2.6

14 years agosky2: Fix oops in sky2_xmit_frame() after TX timeout
Jarek Poplawski [Mon, 4 Jan 2010 08:48:41 +0000 (08:48 +0000)]
sky2: Fix oops in sky2_xmit_frame() after TX timeout

During TX timeout procedure dev could be awoken too early, e.g. by
sky2_complete_tx() called from sky2_down(). Then sky2_xmit_frame()
can run while buffers are freed causing an oops. This patch fixes it
by adding netif_device_present() test in sky2_tx_complete().

Fixes: http://bugzilla.kernel.org/show_bug.cgi?id=14925

With debugging by: Mike McCormack <mikem@ring3k.org>

Reported-by: Berck E. Nash <flyboy@gmail.com>
Tested-by: Berck E. Nash <flyboy@gmail.com>
Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agocfg80211: fix refcount imbalance when wext is disabled
Johannes Berg [Mon, 11 Jan 2010 15:14:57 +0000 (16:14 +0100)]
cfg80211: fix refcount imbalance when wext is disabled

When CONFIG_CFG80211_WEXT is not set, there is
a refcount imbalance with rdev->opencount, fix
that by moving it out of the ifdef.

Reported-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agomac80211: fix queue selection for data frames on monitor interfaces
Felix Fietkau [Mon, 11 Jan 2010 05:47:00 +0000 (06:47 +0100)]
mac80211: fix queue selection for data frames on monitor interfaces

When ieee80211_monitor_select_queue encounters data frames, it selects
the WMM AC based on skb->priority and assumes that skb->priority
contains a valid 802.1d tag. However this assumption is incorrect, since
ieee80211_select_queue has not been called at this point.
If skb->priority > 7, an array overrun occurs, which could lead to
invalid values, resulting in crashes in the tx path.
Fix this by setting skb->priority based on the 802.11 header for QoS
frames and using the default AC for all non-QoS frames.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agoDocumentation/3c509: document ethtool support
Ben Hutchings [Mon, 11 Jan 2010 23:53:45 +0000 (15:53 -0800)]
Documentation/3c509: document ethtool support

3c509 was changed to support ethtool in 2002, making the 'xcvr' module
parameter obsolete in most cases.  More recently 3c509 was converted
to the modern driver model and this parameter was removed.  Fix the
documentation to refer to ethtool rather than the module parameter.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoaf_packet: Don't use skb after dev_queue_xmit()
Jarek Poplawski [Sun, 10 Jan 2010 22:04:19 +0000 (22:04 +0000)]
af_packet: Don't use skb after dev_queue_xmit()

tpacket_snd() can change and kfree an skb after dev_queue_xmit(),
which is illegal.

With debugging by: Stephen Hemminger <shemminger@vyatta.com>

Reported-by: Michael Breuer <mbreuer@majjas.com>
With help from: David S. Miller <davem@davemloft.net>
Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
Tested-by: Michael Breuer<mbreuer@majjas.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoiwlwifi: silence buffer overflow warning
Dan Carpenter [Sat, 9 Jan 2010 08:41:48 +0000 (11:41 +0300)]
iwlwifi: silence buffer overflow warning

Smatch (and presumably other static checkers) complain that MAX_TID_COUNT is
past the end of the array.  In the resulting discussion, Zhu Yi pointed out
that this value is not used in real life and the assignment was only there to
silence a gcc warning.

If there were a bug in the surrounding code and the value were used, the
WARN_ON(!qc) would print a warning before the crash.

Signed-off-by: Dan Carpenter <error27@gmail.com>
Acked-by: Zhu Yi <yi.zhu@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agovxge: use pci_dma_mapping_error to test return value
Denis Kirjanov [Sun, 10 Jan 2010 21:40:10 +0000 (13:40 -0800)]
vxge: use pci_dma_mapping_error to test return value

pci_dma_mapping_error should be used to test return value of
pci_map_single or pci_map_page.

Signed-off-by: Denis Kirjanov <kirjanov@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/kaber/nf-2.6
David S. Miller [Fri, 8 Jan 2010 21:15:47 +0000 (13:15 -0800)]
Merge branch 'master' of git://git./linux/kernel/git/kaber/nf-2.6

14 years agoiwlwifi: disable tx on beacon update notification
Abhijeet Kolekar [Fri, 8 Jan 2010 18:04:31 +0000 (10:04 -0800)]
iwlwifi: disable tx on beacon update notification

On beacon change update notification from mac we are not disabling
the tx in adhoc mode. Mac sends BSS_CHANGED_BEACON_ENABLED when
station leaves IBSS. Driver should indicate uCode to not to send
anything on receiving this notification.

Functionality to indicate uCode is duplicated across
two notifications so created a common function called iwl_set_no_assoc.

Fix the issue at
http://bugzilla.intellinuxwireless.org/show_bug.cgi?id=2133.

Signed-off-by: Abhijeet Kolekar <abhijeet.kolekar@intel.com>
Tested-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agoiwlwifi: fix iwl_queue_used bug when read_ptr == write_ptr
Zhu Yi [Fri, 8 Jan 2010 18:04:30 +0000 (10:04 -0800)]
iwlwifi: fix iwl_queue_used bug when read_ptr == write_ptr

When txq read_ptr equals to write_ptr, iwl_queue_used should
always return false. Because there is no used TFD in this case.

This is a complementary fix to the fix already included in commit "iwl3945:
fix panic in iwl3945 driver". Both fixes are needed to address the panic
below.

This problem was discussed on linux-wireless in
http://thread.gmane.org/gmane.linux.kernel.wireless.general/43568

<1>[ 7290.414172] IP: [<ffffffffa0dd53a1>] iwl3945_rx_reply_tx+0xc1/0x450 [iwl3945]
<4>[ 7290.414205] PGD 0
<1>[ 7290.414214] Thread overran stack, or stack corrupted
<0>[ 7290.414229] Oops: 0002 [#1] PREEMPT SMP
<0>[ 7290.414246] last sysfs file: /sys/devices/platform/coretemp.1/temp1_input
<4>[ 7290.414265] CPU 0
<4>[ 7290.414274] Modules linked in: af_packet nfsd usb_storage usb_libusual cpufreq_powersave exportfs cpufreq_conservative iwl3945 nfs cpufreq_userspace snd_hda_codec_realtek acpi_cpufreq uvcvideo lockd iwlcore snd_hda_intel joydev coretemp nfs_acl videodev snd_hda_codec mac80211 v4l1_compat snd_hwdep sbp2 v4l2_compat_ioctl32 uhci_hcd psmouse auth_rpcgss ohci1394 cfg80211 ehci_hcd video ieee1394 snd_pcm serio_raw battery ac nvidia(P) usbcore output sunrpc evdev lirc_ene0100 snd_page_alloc rfkill tg3 libphy fuse lzo lzo_decompress lzo_compress
<6>[ 7290.414486] Pid: 0, comm: swapper Tainted: P           2.6.32-rc8-wl #213 Aspire 5720
<6>[ 7290.414507] RIP: 0010:[<ffffffffa0dd53a1>]  [<ffffffffa0dd53a1>] iwl3945_rx_reply_tx+0xc1/0x450 [iwl3945]
<6>[ 7290.414541] RSP: 0018:ffff880002203d60  EFLAGS: 00010246
<6>[ 7290.414557] RAX: 000000000000004f RBX: ffff880064c11600 RCX: 0000000000000013
<6>[ 7290.414576] RDX: ffffffffa0ddcf20 RSI: ffff8800512b7008 RDI: 0000000000000038
<6>[ 7290.414596] RBP: ffff880002203dd0 R08: 0000000000000000 R09: 0000000000000100
<6>[ 7290.414616] R10: 0000000000000001 R11: 0000000000000000 R12: 00000000000000a0
<6>[ 7290.414635] R13: 0000000000000002 R14: 0000000000000013 R15: 0000000000020201
<6>[ 7290.414655] FS:  0000000000000000(0000) GS:ffff880002200000(0000) knlGS:0000000000000000
<6>[ 7290.414677] CS:  0010 DS: 0018 ES: 0018 CR0: 000000008005003b
<6>[ 7290.414693] CR2: 0000000000000041 CR3: 0000000001001000 CR4: 00000000000006f0
<6>[ 7290.414712] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
<6>[ 7290.414732] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
<4>[ 7290.414752] Process swapper (pid: 0, threadinfo ffffffff81524000, task ffffffff81528b60)
<0>[ 7290.414772] Stack:
<4>[ 7290.414780]  ffff880002203da0 0000000000000046 0000000000000000 0000000000000046
<4>[ 7290.414804] <0> 0000000000000282 0000000000000282 0000000000000282 ffff880064c12010
<4>[ 7290.414830] <0> ffff880002203db0 ffff880064c11600 ffff880064c12e50 ffff8800512b7000
<0>[ 7290.414858] Call Trace:
<0>[ 7290.414867]  <IRQ>
<4>[ 7290.414884]  [<ffffffffa0dc8c47>] iwl3945_irq_tasklet+0x657/0x1740 [iwl3945]
<4>[ 7290.414910]  [<ffffffff8138fc60>] ? _spin_unlock+0x30/0x60
<4>[ 7290.414931]  [<ffffffff81049a21>] tasklet_action+0x101/0x110
<4>[ 7290.414950]  [<ffffffff8104a3d0>] __do_softirq+0xc0/0x160
<4>[ 7290.414968]  [<ffffffff8100d01c>] call_softirq+0x1c/0x30
<4>[ 7290.414986]  [<ffffffff8100eff5>] do_softirq+0x75/0xb0
<4>[ 7290.415003]  [<ffffffff81049ee5>] irq_exit+0x95/0xa0
<4>[ 7290.415020]  [<ffffffff8100e547>] do_IRQ+0x77/0xf0
<4>[ 7290.415038]  [<ffffffff8100c7d3>] ret_from_intr+0x0/0xf
<0>[ 7290.415052]  <EOI>
<4>[ 7290.415067]  [<ffffffff81234efa>] ? acpi_idle_enter_bm+0x270/0x2a5
<4>[ 7290.415087]  [<ffffffff81234f04>] ? acpi_idle_enter_bm+0x27a/0x2a5
<4>[ 7290.415107]  [<ffffffff81234efa>] ? acpi_idle_enter_bm+0x270/0x2a5
<4>[ 7290.415130]  [<ffffffff812c11f3>] ? cpuidle_idle_call+0x93/0xf0
<4>[ 7290.415149]  [<ffffffff8100b0d7>] ? cpu_idle+0xa7/0x110
<4>[ 7290.415168]  [<ffffffff8137b3d5>] ? rest_init+0x75/0x80
<4>[ 7290.415187]  [<ffffffff8158cd0a>] ? start_kernel+0x3a7/0x3b3
<4>[ 7290.415206]  [<ffffffff8158c315>] ? x86_64_start_reservations+0x125/0x129
<4>[ 7290.415227]  [<ffffffff8158c3fd>] ? x86_64_start_kernel+0xe4/0xeb
<0>[ 7290.415243] Code: 00 41 39 ce 0f 8d e8 01 00 00 48 8b 47 40 48 63 d2 48 69 d2 98 00 00 00 4c 8b 04 02 48 c7 c2 20 cf dd a0 49 8d 78 38 49 8d 40 4f <c6> 47 09 00 c6 47 0c 00 c6 47 0f 00 c6 47 12 00 c6 47 15 00 49
<1>[ 7290.415382] RIP  [<ffffffffa0dd53a1>] iwl3945_rx_reply_tx+0xc1/0x450 [iwl3945]
<4>[ 7290.415410]  RSP <ffff880002203d60>
<0>[ 7290.415421] CR2: 0000000000000041
<4>[ 7290.415436] ---[ end trace ec46807277caa515 ]---
<0>[ 7290.415450] Kernel panic - not syncing: Fatal exception in interrupt
<4>[ 7290.415468] Pid: 0, comm: swapper Tainted: P      D    2.6.32-rc8-wl #213
<4>[ 7290.415486] Call Trace:
<4>[ 7290.415495]  <IRQ>  [<ffffffff8138c040>] panic+0x7d/0x13a
<4>[ 7290.415519]  [<ffffffff8101071a>] oops_end+0xda/0xe0
<4>[ 7290.415538]  [<ffffffff8102e1ea>] no_context+0xea/0x250
<4>[ 7290.415557]  [<ffffffff81038991>] ? select_task_rq_fair+0x511/0x780
<4>[ 7290.415578]  [<ffffffff8102e475>] __bad_area_nosemaphore+0x125/0x1e0
<4>[ 7290.415597]  [<ffffffff81038d0c>] ? __enqueue_entity+0x7c/0x80
<4>[ 7290.415616]  [<ffffffff81039201>] ? enqueue_task_fair+0x111/0x150
<4>[ 7290.415636]  [<ffffffff8102e53e>] bad_area_nosemaphore+0xe/0x10
<4>[ 7290.415656]  [<ffffffff8102e8fa>] do_page_fault+0x26a/0x320
<4>[ 7290.415674]  [<ffffffff813905df>] page_fault+0x1f/0x30
<4>[ 7290.415697]  [<ffffffffa0dd53a1>] ? iwl3945_rx_reply_tx+0xc1/0x450 [iwl3945]
<4>[ 7290.415723]  [<ffffffffa0dc8c47>] iwl3945_irq_tasklet+0x657/0x1740 [iwl3945]
<4>[ 7290.415746]  [<ffffffff8138fc60>] ? _spin_unlock+0x30/0x60
<4>[ 7290.415764]  [<ffffffff81049a21>] tasklet_action+0x101/0x110
<4>[ 7290.415783]  [<ffffffff8104a3d0>] __do_softirq+0xc0/0x160
<4>[ 7290.415801]  [<ffffffff8100d01c>] call_softirq+0x1c/0x30
<4>[ 7290.415818]  [<ffffffff8100eff5>] do_softirq+0x75/0xb0
<4>[ 7290.415835]  [<ffffffff81049ee5>] irq_exit+0x95/0xa0
<4>[ 7290.415852]  [<ffffffff8100e547>] do_IRQ+0x77/0xf0
<4>[ 7290.415869]  [<ffffffff8100c7d3>] ret_from_intr+0x0/0xf
<4>[ 7290.415883]  <EOI>  [<ffffffff81234efa>] ? acpi_idle_enter_bm+0x270/0x2a5
<4>[ 7290.415911]  [<ffffffff81234f04>] ? acpi_idle_enter_bm+0x27a/0x2a5
<4>[ 7290.415931]  [<ffffffff81234efa>] ? acpi_idle_enter_bm+0x270/0x2a5
<4>[ 7290.415952]  [<ffffffff812c11f3>] ? cpuidle_idle_call+0x93/0xf0
<4>[ 7290.415971]  [<ffffffff8100b0d7>] ? cpu_idle+0xa7/0x110
<4>[ 7290.415989]  [<ffffffff8137b3d5>] ? rest_init+0x75/0x80
<4>[ 7290.416007]  [<ffffffff8158cd0a>] ? start_kernel+0x3a7/0x3b3
<4>[ 7290.416026]  [<ffffffff8158c315>] ? x86_64_start_reservations+0x125/0x129
<4>[ 7290.416047]  [<ffffffff8158c3fd>] ? x86_64_start_kernel+0xe4/0xeb

Reported-by: Maxim Levitsky <maximlevitsky@gmail.com>
Tested-by: Maxim Levitsky <maximlevitsky@gmail.com>
Signed-off-by: Zhu Yi <yi.zhu@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
CC: stable@kernel.org
Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agomac80211: fix endian error
Johannes Berg [Fri, 8 Jan 2010 18:00:00 +0000 (19:00 +0100)]
mac80211: fix endian error

I forgot to convert the radiotap length to
CPU endian, which sparse found thankfully.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Cc: stable@kernel.org
Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agomac80211: add missing sanity checks for action frames
Felix Fietkau [Thu, 7 Jan 2010 19:23:53 +0000 (20:23 +0100)]
mac80211: add missing sanity checks for action frames

Various missing sanity checks caused rejected action frames to be
interpreted as channel switch announcements, which can cause a client
mode interface to switch away from its operating channel, thereby losing
connectivity. This patch ensures that only spectrum management action
frames are processed by the CSA handling function and prevents rejected
action frames from getting processed by the MLME code.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Cc: stable@kernel.org
Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agomac80211: fix queue selection for packets injected via monitor interface
Lennert Buytenhek [Thu, 7 Jan 2010 14:01:42 +0000 (15:01 +0100)]
mac80211: fix queue selection for packets injected via monitor interface

Commit 'mac80211: fix skb buffering issue' added an ->ndo_select_queue()
for monitor interfaces which can end up dereferencing ieee802_1d_to_ac[]
beyond the end of the array for injected data packets (as skb->priority
isn't guaranteed to be zero or within [0:7]), which then triggers the
WARN_ON in net/core/dev.c:dev_cap_txqueue().  Fix this by always setting
the priority to zero on injected data frames.

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Cc: stable@kernel.org
Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agonetfilter: ebtables: enforce CAP_NET_ADMIN
Florian Westphal [Fri, 8 Jan 2010 16:31:24 +0000 (17:31 +0100)]
netfilter: ebtables: enforce CAP_NET_ADMIN

normal users are currently allowed to set/modify ebtables rules.
Restrict it to processes with CAP_NET_ADMIN.

Note that this cannot be reproduced with unmodified ebtables binary
because it uses SOCK_RAW.

Signed-off-by: Florian Westphal <fwestphal@astaro.com>
Cc: stable@kernel.org
Signed-off-by: Patrick McHardy <kaber@trash.net>
14 years agoe1000e: fix and commonize code for setting the receive address registers
Bruce Allan [Thu, 7 Jan 2010 16:32:13 +0000 (16:32 +0000)]
e1000e: fix and commonize code for setting the receive address registers

Fix e1000e_rar_set() to flush consecutive register writes to avoid write
combining which some parts cannot handle.  Update e1000e_init_rx_addrs()
to call the fixed e1000e_rar_set() instead of duplicating code.

Also change e1000e_rar_set() to _not_ set the Address Valid bit if the MAC
address is all zeros.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoe1000e: e1000e_enable_tx_pkt_filtering() returns wrong value
Bruce Allan [Thu, 7 Jan 2010 16:31:54 +0000 (16:31 +0000)]
e1000e: e1000e_enable_tx_pkt_filtering() returns wrong value

e1000e_enable_tx_pkt_filtering() will return a non-zero value if the
driver fails to enable the manageability interface on the host for
any reason; instead it should retun zero to indicate filtering has been
disabled.  Also provide a single exit point for the function.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoe1000e: perform 10/100 adaptive IFS only on parts that support it
Bruce Allan [Thu, 7 Jan 2010 16:31:35 +0000 (16:31 +0000)]
e1000e: perform 10/100 adaptive IFS only on parts that support it

Adaptive IFS which involves writing to the Adaptive IFS Throttle register
was being done for all devices supported by the driver even though it is
not supported (i.e. the register doesn't even exist) on some devices.  The
feature is supported on 8257x/82583 and ICH/PCH based devices, but not
on ESB2.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoe1000e: don't accumulate PHY statistics on PHY read failure
Bruce Allan [Thu, 7 Jan 2010 16:31:16 +0000 (16:31 +0000)]
e1000e: don't accumulate PHY statistics on PHY read failure

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoe1000e: call pci_save_state() after pci_restore_state()
Bruce Allan [Thu, 7 Jan 2010 16:30:56 +0000 (16:30 +0000)]
e1000e: call pci_save_state() after pci_restore_state()

Due to a change in pci_restore_state()[1] which clears the saved_state
flag, the driver should call pci_save_state() to set the flag once again
to avoid issues with EEH (same fix that recently was submitted for ixgbe).

[1] commmit 4b77b0a2ba27d64f58f16d8d4d48d8319dda36ff

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonetxen: update version to 4.0.72
Amit Kumar Salecha [Thu, 7 Jan 2010 22:10:17 +0000 (22:10 +0000)]
netxen: update version to 4.0.72

Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonetxen: fix set mac addr
Amit Kumar Salecha [Thu, 7 Jan 2010 22:10:16 +0000 (22:10 +0000)]
netxen: fix set mac addr

o If tx and rx resources are not available, during set mac request.
  Then this request wont be passed to firmware and it will be added to
  driver mac list and will never make it to firmware.
  So if resources are not available, don't add it to driver mac list.

Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonetxen: fix smatch warning
Amit Kumar Salecha [Thu, 7 Jan 2010 22:10:15 +0000 (22:10 +0000)]
netxen: fix smatch warning

o Fix pointless assignments

Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonetxen: fix tx ring memory leak
Amit Kumar Salecha [Thu, 7 Jan 2010 22:10:14 +0000 (22:10 +0000)]
netxen: fix tx ring memory leak

o While unloading driver or resetting the context, tx ring was not
  getting free.

Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agotcp: update the netstamp_needed counter when cloning sockets
Octavian Purdila [Fri, 8 Jan 2010 08:00:09 +0000 (00:00 -0800)]
tcp: update the netstamp_needed counter when cloning sockets

This fixes a netstamp_needed accounting issue when the listen socket
has SO_TIMESTAMP set:

    s = socket(AF_INET, SOCK_STREAM, 0);
    setsockopt(s, SOL_SOCKET, SO_TIMESTAMP, 1); -> netstamp_needed = 1
    bind(s, ...);
    listen(s, ...);
    s2 = accept(s, ...); -> netstamp_needed = 1
    close(s2); -> netstamp_needed = 0
    close(s); -> netstamp_needed = -1

Signed-off-by: Octavian Purdila <opurdila@ixiacom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoTI DaVinci EMAC: Handle emac module clock correctly.
Sriram [Thu, 7 Jan 2010 00:22:37 +0000 (00:22 +0000)]
TI DaVinci EMAC: Handle emac module clock correctly.

In the driver probe function the emac module clock needs to
be enabled before calling register_netdev(). As soon as the
device is registered the driver get_stats function can be invoked
by the core - the module clock must be switched on to be able to
read from stats registers. Also explicitly call matching clk_disable
for failure conditions in probe function.

Signed-off-by: Sriramakrishnan <srk@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agodmfe/tulip: Let dmfe handle DM910x except for SPARC on-board chips
Ben Hutchings [Thu, 7 Jan 2010 02:41:51 +0000 (02:41 +0000)]
dmfe/tulip: Let dmfe handle DM910x except for SPARC on-board chips

The Davicom DM9100 and DM9102 chips are used on the motherboards of
some SPARC systems (supported by the tulip driver) and also in PCI
expansion cards (supported by the dmfe driver).  There is no
difference in the PCI device ids for the two different configurations,
so these drivers both claim the device ids.  However, it is possible
to distinguish the two configurations by the presence of Open Firmware
properties for them, so we do that.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Grant Grundler <grundler@parisc-linux.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoixgbe: Fix compiler warning about variable being used uninitialized
PJ Waskiewicz [Wed, 6 Jan 2010 17:50:29 +0000 (17:50 +0000)]
ixgbe: Fix compiler warning about variable being used uninitialized

tc is still throwing a warning that is could be used
uninitialized.  This fixes it, and properly formats the device ID
checks for the use of this variable.

Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonetfilter: nf_ct_ftp: fix out of bounds read in update_nl_seq()
Patrick McHardy [Thu, 7 Jan 2010 17:33:18 +0000 (18:33 +0100)]
netfilter: nf_ct_ftp: fix out of bounds read in update_nl_seq()

As noticed by Dan Carpenter <error27@gmail.com>, update_nl_seq()
currently contains an out of bounds read of the seq_aft_nl array
when looking for the oldest sequence number position.

Fix it to only compare valid positions.

Cc: stable@kernel.org
Signed-off-by: Patrick McHardy <kaber@trash.net>
14 years agomv643xx_eth: don't include cache padding in rx desc buffer size
Saeed Bishara [Tue, 5 Jan 2010 09:15:32 +0000 (09:15 +0000)]
mv643xx_eth: don't include cache padding in rx desc buffer size

If NET_SKB_PAD is not a multiple of the cache line size, mv643xx_eth
allocates a couple of extra bytes at the start of each receive buffer
to make the data payload end up on a cache line boundary.

These extra bytes are skb_reserve()'d before DMA mapping, so they
should not be included in the DMA map byte count (as the mapping is
done starting at skb->data), nor should they be included in the
receive descriptor buffer size field, or the hardware can end up
DMAing beyond the end of the buffer, which can happen if someone
sends us a larger-than-MTU sized packet.

This problem was introduced in commit 7fd96ce47ff ("mv643xx_eth:
rework receive skb cache alignment", May 6 2009), but hasn't appeared
to be problematic so far, probably as the main users of mv643xx_eth
all have NET_SKB_PAD == L1_CACHE_BYTES.

Signed-off-by: Saeed Bishara <saeed@marvell.com>
Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoNET: atlx, fix memory leak
Jiri Slaby [Wed, 6 Jan 2010 06:54:16 +0000 (06:54 +0000)]
NET: atlx, fix memory leak

Stanse found a memory leak in atl2_get_eeprom. eeprom_buff is not
freed/assigned on all paths. Fix that.

Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Cc: Jay Cliburn <jcliburn@gmail.com>
Cc: Chris Snook <chris.snook@gmail.com>
Cc: Jie Yang <jie.yang@atheros.com>
Cc: atl1-devel@lists.sourceforge.net
Cc: "David S. Miller" <davem@davemloft.net>
Cc: netdev@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agopcmcia: ncmlan_cs: remove odd bracket
Alexander Beregalov [Wed, 6 Jan 2010 13:05:46 +0000 (13:05 +0000)]
pcmcia: ncmlan_cs: remove odd bracket

Signed-off-by: Alexander Beregalov <a.beregalov@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agocs89x0: Always report failure to request interrupt
Mark Brown [Thu, 7 Jan 2010 08:53:05 +0000 (00:53 -0800)]
cs89x0: Always report failure to request interrupt

A failure on request_irq() is always fatal but unlike other fatal
errors it's only reported to the user if net_debug is set. Make the
diagnostic unconditional and raise the priority so that errors are
more obvious to the user.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agohso: fixed missing newlines
Jan Dumon [Tue, 5 Jan 2010 04:53:00 +0000 (04:53 +0000)]
hso: fixed missing newlines

Fixed missing newlines in calls to dev_warn & dev_err.

Signed-off-by: Jan Dumon <j.dumon@option.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agohso: Fix for 5 sec timeouts with v2.x firmware
Jan Dumon [Tue, 5 Jan 2010 04:52:42 +0000 (04:52 +0000)]
hso: Fix for 5 sec timeouts with v2.x firmware

Don't send flow control settings to any port other than the modem port.
Older firmware ignored this request but did sent a reply. Newer firmware just
ignores it without reply and causes a 5 second timeout every time a port
(except for the modem port) is opened or if tiocm settings are changed.

Signed-off-by: Jan Dumon <j.dumon@option.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agohso: Attempt to recover from usb bus errors
Jan Dumon [Tue, 5 Jan 2010 04:52:13 +0000 (04:52 +0000)]
hso: Attempt to recover from usb bus errors

Attempt to reset the usb device when we receive usb bus errors.

Signed-off-by: Jan Dumon <j.dumon@option.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agohso: don't change the state of a closed port
Jan Dumon [Tue, 5 Jan 2010 04:51:28 +0000 (04:51 +0000)]
hso: don't change the state of a closed port

Don't change the state of a port if it's not open. This fixes an issue where a
port sometimes has to be opened twice before data can be received.

Signed-off-by: Jan Dumon <j.dumon@option.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agohso: Fix for endian issues on big endian machines
Jan Dumon [Tue, 5 Jan 2010 04:51:02 +0000 (04:51 +0000)]
hso: Fix for endian issues on big endian machines

Some fields are always little endian and have to be converted on big endian
machines.

Signed-off-by: Jan Dumon <j.dumon@option.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agohso: Add Vendor/Product ID's for new devices
Jan Dumon [Tue, 5 Jan 2010 04:50:31 +0000 (04:50 +0000)]
hso: Add Vendor/Product ID's for new devices

Add product ID's for new devices.

Signed-off-by: Jan Dumon <j.dumon@option.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoclaw: use "claw" as root device name
Ursula Braun [Mon, 4 Jan 2010 03:14:45 +0000 (03:14 +0000)]
claw: use "claw" as root device name

Claw module cannot be loaded together with qeth, because "qeth" has
been errorneously used as root device name. It is changed into "claw".

Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agopcnet_cs: add cis of KTI PE520 pcmcia network card
Ken Kawasaki [Thu, 7 Jan 2010 04:37:58 +0000 (20:37 -0800)]
pcnet_cs: add cis of KTI PE520 pcmcia network card

pcnet_cs,serial_cs:

add cis of KTI PE520 pcmcia network card,
and serial card(Sierra Wireless AC860).

Signed-off-by: Ken Kawasaki <ken_kawasaki@spring.nifty.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoip: fix mc_loop checks for tunnels with multicast outer addresses
Octavian Purdila [Thu, 7 Jan 2010 04:37:01 +0000 (20:37 -0800)]
ip: fix mc_loop checks for tunnels with multicast outer addresses

When we have L3 tunnels with different inner/outer families
(i.e. IPV4/IPV6) which use a multicast address as the outer tunnel
destination address, multicast packets will be loopbacked back to the
sending socket even if IP*_MULTICAST_LOOP is set to disabled.

The mc_loop flag is present in the family specific part of the socket
(e.g. the IPv4 or IPv4 specific part).  setsockopt sets the inner
family mc_loop flag. When the packet is pushed through the L3 tunnel
it will eventually be processed by the outer family which if different
will check the flag in a different part of the socket then it was set.

Signed-off-by: Octavian Purdila <opurdila@ixiacom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoucc_geth: Fix the wrong the Rx/Tx FIFO size
Dave Liu [Thu, 7 Jan 2010 04:32:38 +0000 (20:32 -0800)]
ucc_geth: Fix the wrong the Rx/Tx FIFO size

current the Rx/Tx FIFO size settings cause problem
when four UEC ethernets work simultaneously.

eg: GETH1, UEM-J15, GETH2, UEC-J5 on 8569MDS board

 $ ifconfig eth0 10.193.20.166
 $ ifconfig eth1 10.193.20.167
 $ ifconfig eth2 10.193.20.168
 then
 $ ifconfig eth3 10.193.20.169

The fourth ethernet will cause all of interface broken,
you cann't ping successfully any more.
The patch fix this issue for MPC8569 Rev1.0 and Rev2.0

Signed-off-by: Dave Liu <daveliu@freescale.com>
Acked-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoRevert "mac80211: replace netif_tx_{start,stop,wake}_all_queues"
John W. Linville [Wed, 6 Jan 2010 20:39:39 +0000 (15:39 -0500)]
Revert "mac80211: replace netif_tx_{start,stop,wake}_all_queues"

This reverts commit 53623f1a09c7a7d23b74f0f7d93dba0ebde1006b.

This was inadvertantly missed in "mac80211: fix skb buffering issue",
and is required with that patch to restore proper queue operation.

Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agomac80211: fix-up build breakage in 2.6.33
John W. Linville [Wed, 6 Jan 2010 20:35:49 +0000 (15:35 -0500)]
mac80211: fix-up build breakage in 2.6.33

"mac80211: fix skb buffering issue" is based on what will become 2.6.34,
so it includes an incompatible usage of sta_info_get.  This patch will
need to be effectively reverted when merging for 2.6.34.

Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agort2x00: Properly request tx headroom for alignment operations.
Gertjan van Wingerde [Wed, 30 Dec 2009 10:36:30 +0000 (11:36 +0100)]
rt2x00: Properly request tx headroom for alignment operations.

Current rt2x00 drivers may result in a "ieee80211_tx_status: headroom too
small" error message when a frame needs to be properly aligned before
transmitting it.
This is because the space needed to ensure proper alignment isn't
requested from mac80211.
Fix this by adding sufficient amount of alignment space to the amount
of headroom requested for TX frames.

Reported-by: David Ellingsworth <david@identd.dyndns.org>
Signed-off-by: Gertjan van Wingerde <gwingerde@gmail.com>
Acked-by: Ivo van Doorn <ivdoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agomac80211: fix skb buffering issue
Johannes Berg [Tue, 5 Jan 2010 17:00:58 +0000 (18:00 +0100)]
mac80211: fix skb buffering issue

Since I removed the master netdev, we've been
keeping internal queues only, and even before
that we never told the networking stack above
the virtual interfaces about congestion. This
means that packets are queued in mac80211 and
the upper layers never know, possibly leading
to memory exhaustion and other problems.

This patch makes all interfaces multiqueue and
uses ndo_select_queue to put the packets into
queues per AC. Additionally, when the driver
stops a queue, we now stop all corresponding
queues for the virtual interfaces as well.

The injection case will use VO by default for
non-data frames, and BE for data frames, but
downgrade any data frames according to ACM. It
needs to be fleshed out in the future to allow
chosing the queue/AC in radiotap.

Reported-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Cc: stable@kernel.org [2.6.32]
Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agort2x00: Fix LED configuration setting for rt2800.
Gertjan van Wingerde [Wed, 30 Dec 2009 10:36:33 +0000 (11:36 +0100)]
rt2x00: Fix LED configuration setting for rt2800.

rt2800_blink_set uses an illegal value to set the LED_CFG_G_LED_MODE
field of the LED_CFG register. This field is only 2 bits large, so
should be initialized with value that fits. Use default value from
the vendor driver.

Signed-off-by: Gertjan van Wingerde <gwingerde@gmail.com>
Acked-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agort2x00: use correct headroom for transmission
Pavel Roskin [Wed, 30 Dec 2009 10:36:29 +0000 (11:36 +0100)]
rt2x00: use correct headroom for transmission

Use rt2x00dev->ops->extra_tx_headroom, not rt2x00dev->hw->extra_tx_headroom
in the tx code, as the later may include other headroom not to be used in
the chipset driver.

Signed-off-by: Pavel Roskin <proski@gnu.org>
Signed-off-by: Gertjan van Wingerde <gwingerde@gmail.com>
Acked-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agocxgb3i: Fix flags test.
David S. Miller [Mon, 4 Jan 2010 22:36:40 +0000 (14:36 -0800)]
cxgb3i: Fix flags test.

As noticed by Stephen Rothwell.

Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agocfg80211: fix syntax error on user regulatory hints
Luis R. Rodriguez [Mon, 4 Jan 2010 16:37:39 +0000 (11:37 -0500)]
cfg80211: fix syntax error on user regulatory hints

This fixes a syntax error when setting up the user regulatory
hint. This change yields the same exact binary object though
so it ends up just being a syntax typo fix, fortunately.

Cc: stable@kernel.org
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agoath5k: Fix eeprom checksum check for custom sized eeproms
Luis R. Rodriguez [Mon, 4 Jan 2010 15:40:39 +0000 (10:40 -0500)]
ath5k: Fix eeprom checksum check for custom sized eeproms

Commit 8bf3d79bc401ca417ccf9fc076d3295d1a71dbf5 enabled EEPROM
checksum checks to avoid bogus bug reports but failed to address
updating the code to consider devices with custom EEPROM sizes.
Devices with custom sized EEPROMs have the upper limit size stuffed
in the EEPROM. Use this as the upper limit instead of the static
default size. In case of a checksum error also provide back the
max size and whether or not this was the default size or a custom
one. If the EEPROM is busted we add a failsafe check to ensure
we don't loop forever or try to read bogus areas of hardware.

This closes bug 14874

http://bugzilla.kernel.org/show_bug.cgi?id=14874

Cc: stable@kernel.org
Cc: David Quan <david.quan@atheros.com>
Cc: Stephen Beahm <stephenbeahm@comcast.net>
Reported-by: Joshua Covington <joshuacov@googlemail.com>
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agozd1211rw: Fix multicast filtering.
Benoit Papillault [Sun, 3 Jan 2010 09:20:01 +0000 (10:20 +0100)]
zd1211rw: Fix multicast filtering.

If multicast parameter (as returned by zd_op_prepare_multicast) has
changed, no bit in changed_flags is set. To handle this situation, we do
not return if changed_flags is 0. If we do so, we will have some issue
with IPv6 which uses multicast for link layer address resolution.

Signed-off-by: Benoit Papillault <benoit.papillault@free.fr>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agoiwl: off by one bug
Dan Carpenter [Sun, 3 Jan 2010 09:19:35 +0000 (11:19 +0200)]
iwl: off by one bug

tid is used as an array offset.
agg = &priv->stations[sta_id].tid[tid].agg;
iwl4965_tx_status_reply_tx(priv, agg, tx_resp, txq_id, index);

It should be limitted to MAX_TID_COUNT - 1;
        struct iwl_tid_data tid[MAX_TID_COUNT];

regards,
dan carpenter

Signed-off-by: Dan Carpenter <error27@gmail.com>
CC: stable@kernel.org
Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agomwl8k: fix configure_filter() memory leak on error
Lennert Buytenhek [Sat, 2 Jan 2010 09:31:42 +0000 (10:31 +0100)]
mwl8k: fix configure_filter() memory leak on error

If there was an error acquiring the firmware lock in
mwl8k_configure_filter(), we would end up leaking the multicast
command packet prepared by mwl8k_prepare_multicast().

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agoath9k: fix ito64
Ming Lei [Tue, 29 Dec 2009 14:57:28 +0000 (22:57 +0800)]
ath9k: fix ito64

The unit of sizeof() is byte instead of bit, so fix it.
The patch can fix debug output of some dma_addr_t variables.

Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agoath9k: Fix Kconfig depends for ATH9K_DEBUGFS
Dominik Geyer [Tue, 29 Dec 2009 07:27:57 +0000 (08:27 +0100)]
ath9k: Fix Kconfig depends for ATH9K_DEBUGFS

Add missing DEBUG_FS dependency for ATH9K_DEBUGFS in ath9k's Kconfig.

Signed-off-by: Dominik D. Geyer <dominik.geyer@gmx.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
14 years agoipvs: Add boundary check on ioctl arguments
Arjan van de Ven [Mon, 4 Jan 2010 15:37:12 +0000 (16:37 +0100)]
ipvs: Add boundary check on ioctl arguments

The ipvs code has a nifty system for doing the size of ioctl command
copies; it defines an array with values into which it indexes the cmd
to find the right length.

Unfortunately, the ipvs code forgot to check if the cmd was in the
range that the array provides, allowing for an index outside of the
array, which then gives a "garbage" result into the length, which
then gets used for copying into a stack buffer.

Fix this by adding sanity checks on these as well as the copy size.

[ horms@verge.net.au: adjusted limit to IP_VS_SO_GET_MAX ]
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Acked-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Patrick McHardy <kaber@trash.net>
14 years agonetxen: fix ethtool link test
Sucheta Chakraborty [Sat, 2 Jan 2010 03:25:19 +0000 (03:25 +0000)]
netxen: fix ethtool link test

o Fix ethtool link test for NX3031 chip.
o Remove unused code from phy interrupt callback

Signed-off-by: Sucheta Chakraborty <sucheta.chakraborty@qlogic.com>
Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonetxen: fix ethtool register dump
Sucheta Chakraborty [Sat, 2 Jan 2010 03:25:18 +0000 (03:25 +0000)]
netxen: fix ethtool register dump

o Dump registers such as tx ring and rx ring counter, firmware state,
  niu regs, etc. which can be useful for debugging purpose.

Signed-off-by: Sucheta Chakraborty <sucheta.chakraborty@qlogic.com>
Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agorrunner: fix buffer overflow
Dan Carpenter [Sat, 2 Jan 2010 04:08:46 +0000 (04:08 +0000)]
rrunner: fix buffer overflow

tx_skbuff is define as:
        struct sk_buff          *tx_skbuff[TX_RING_ENTRIES];

EVT_RING_ENTRIES is 64 and TX_RING_ENTRIES is 32.

This function is in a error path so that's why it wasn't noticed.

Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agobroadcom: Fix &&/|| confusion in bcm54xx_adjust_rxrefclk()
roel kluin [Wed, 30 Dec 2009 06:43:06 +0000 (06:43 +0000)]
broadcom: Fix &&/|| confusion in bcm54xx_adjust_rxrefclk()

This always evaluates to true.

Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet: Test off by one in sh_eth_reset()
roel kluin [Wed, 30 Dec 2009 01:43:45 +0000 (01:43 +0000)]
net: Test off by one in sh_eth_reset()

If no break occurred, cnt reaches 0 after the loop.

Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoniu: timeout ignored in tcam_wait_bit()
roel kluin [Sun, 27 Dec 2009 04:10:59 +0000 (04:10 +0000)]
niu: timeout ignored in tcam_wait_bit()

With `while (--limit > 0)' i reaches 0 after the loop, so upon timeout the
error was not returned.

Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoatarilance: timeout ignored in lance_open()
roel kluin [Sun, 27 Dec 2009 03:26:12 +0000 (03:26 +0000)]
atarilance: timeout ignored in lance_open()

With `while (--i > 0)' i reaches 0 after the loop, so upon timeout the
error was not issued.

Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agodrivers/net : Correct the size argument to kzalloc
Julia Lawall [Tue, 29 Dec 2009 09:15:42 +0000 (09:15 +0000)]
drivers/net : Correct the size argument to kzalloc

lp->rx_skb has type struct sk_buff **, not struct sk_buff *, so the
elements of the array should have pointer type, not structure type.

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

// <smpl>
@disable sizeof_type_expr@
type T;
T **x;
@@

  x =
  <+...sizeof(
- T
+ *x
  )...+>
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agocxgb3i: Fix a login over vlan issue
Rakesh Ranjan [Sun, 27 Dec 2009 07:03:08 +0000 (12:33 +0530)]
cxgb3i: Fix a login over vlan issue

Fix a target login issue, when parent interface is vlan and we are using cxgb3i sepecific
private ip address in '/etc/iscsi/ifaces/' iface file.

Acked-by: Karen Xie <kxie@chelsio.com>
Signed-off-by: Rakesh Ranjan <rakesh@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agohamradio: avoid null deref v3
Dan Carpenter [Mon, 28 Dec 2009 06:54:55 +0000 (06:54 +0000)]
hamradio: avoid null deref v3

This should address the problems in version 1 (lazy) and version 2 (ugly).

Bump the stats on orig_dev not on the newly assigned NULL dev variable.

Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agousbnet: test off by one
roel kluin [Sun, 27 Dec 2009 11:22:08 +0000 (11:22 +0000)]
usbnet: test off by one

With `while (i++ < MII_TIMEOUT)' i reaches MII_TIMEOUT + 1 after the loop
This is probably unlikely a problem in practice.

Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agodrivers/net/can: Correct NULL test
Julia Lawall [Sun, 27 Dec 2009 11:27:44 +0000 (11:27 +0000)]
drivers/net/can: Correct NULL test

Test the just-allocated value for NULL rather than some other value.

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

// <smpl>
@@
expression x,y;
statement S;
@@

x = \(kmalloc\|kcalloc\|kzalloc\)(...);
(
if ((x) == NULL) S
|
if (
-   y
+   x
       == NULL)
 S
)
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Acked-by: Oliver Hartkopp <oliver@hartkopp.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agodrivers/isdn: eliminate duplicated test
Julia Lawall [Mon, 21 Dec 2009 14:25:32 +0000 (14:25 +0000)]
drivers/isdn: eliminate duplicated test

The code checked slot_rx twice.  Check slot_tx by analogy with the bank
case.

The semantic match that finds this problem is as follows:
(http://coccinelle.lip6.fr/)

// <smpl>
@@
expression E;
@@

(
*E && E
|
*E || E
)
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Cc: Karsten Keil <isdn@linux-pingi.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agonet/sctp/socket.c: squish warning
Andrew Morton [Mon, 21 Dec 2009 14:25:06 +0000 (14:25 +0000)]
net/sctp/socket.c: squish warning

net/sctp/socket.c: In function 'sctp_setsockopt_autoclose':
net/sctp/socket.c:2090: warning: comparison is always false due to limited range of data type

Cc: Andrei Pelinescu-Onciul <andrei@iptel.org>
Cc: Vlad Yasevich <vladislav.yasevich@hp.com>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agorose_loopback_timer sets VC number <= ROSE_DEFAULT_MAXVC
Bernard Pidoux F6BVP [Thu, 17 Dec 2009 05:25:18 +0000 (05:25 +0000)]
rose_loopback_timer sets VC number <= ROSE_DEFAULT_MAXVC

cat /proc/net/rose displayed a rose sockets abnormal lci value, i.e.
greater than maximum number of VCs per neighbour allowed.
This number prevents further test of lci value during rose operations.

Example (lines shortened) :
[bernard]# cat /proc/net/rose
dest_addr  dest_call src_addr   src_call  dev   lci neigh st vs vr va
*          *         2080175520 F6BVP-1   rose0 000 00000  0  0  0  0
2080175520 FPAD-0    2080175520 WP-0      rose0 FFE 00001  3  0  0  0

Here are the default parameters :

linux/include/net/rose.h:#define ROSE_DEFAULT_MAXVC 50 /* Maximum number of VCs per neighbour */
linux/net/rose/af_rose.c:int sysctl_rose_maximum_vcs = ROSE_DEFAULT_MAXVC;

With the following patch, rose_loopback_timer() attributes a VC number
within limits.

Signed-off-by: Bernard Pidoux <f6bvp@amsat.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agovia-velocity: Give RX descriptors to the NIC later on open or MTU change
Ben Hutchings [Mon, 14 Dec 2009 16:05:09 +0000 (16:05 +0000)]
via-velocity: Give RX descriptors to the NIC later on open or MTU change

velocity_open() calls velocity_give_many_rx_descs(), which gives RX
descriptors to the NIC, before installing an interrupt handler or
calling velocity_init_registers().  I think this is very unsafe and it
appears to explain the bug report <http://bugs.debian.org/508527>.

On MTU change, velocity_give_many_rx_descs() is again called before
velocity_init_registers().  I'm not sure whether this is unsafe but
it does look wrong.

Therefore, move the calls to velocity_give_many_rx_descs() after
request_irq() and velocity_init_registers().

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Tested-by: Jan Ceuleers <jan.ceuleers@computer.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agofsl_pq_mdio: Fix iomem unmapping for non-eTSEC2.0 controllers
Anton Vorontsov [Wed, 30 Dec 2009 08:23:34 +0000 (08:23 +0000)]
fsl_pq_mdio: Fix iomem unmapping for non-eTSEC2.0 controllers

We use a rather complicated logic to support eTSEC and eTSEC2.0
registers maps in a single driver. Currently, the code tries to
unmap 'regs', but for non-eTSEC2.0 controllers 'regs' doesn't
point to a mapping start, and this might cause badness on probe
failure or module removal:

 Freescale PowerQUICC MII Bus: probed
 Trying to vfree() nonexistent vm area (e107f000)
 ------------[ cut here ]------------
 Badness at c00a7754 [verbose debug info unavailable]
 NIP: c00a7754 LR: c00a7754 CTR: c02231ec
 [...]
 NIP [c00a7754] __vunmap+0xec/0xf4
 LR [c00a7754] __vunmap+0xec/0xf4
 Call Trace:
 [df827e50] [c00a7754] __vunmap+0xec/0xf4 (unreliable)
 [df827e70] [c001519c] iounmap+0x44/0x54
 [df827e80] [c028b924] fsl_pq_mdio_probe+0x1cc/0x2fc
 [df827eb0] [c02fb9b4] of_platform_device_probe+0x5c/0x84
 [df827ed0] [c0229928] really_probe+0x78/0x1a8
 [df827ef0] [c0229b20] __driver_attach+0xa4/0xa8

Fix this by introducing a proper priv structure (finally!), which
now holds 'regs' and 'map' fields separately.

Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoucc_geth: Fix netdev watchdog triggering on suspend
Anton Vorontsov [Wed, 30 Dec 2009 08:23:32 +0000 (08:23 +0000)]
ucc_geth: Fix netdev watchdog triggering on suspend

Sometimes ucc_geth fails to suspend with the following trace:

 ucc_geth e0103000.ucc: suspend
 ucc_geth e0102000.ucc: suspend
 NETDEV WATCHDOG: eth0 (ucc_geth): transmit queue 0 timed out
 ------------[ cut here ]------------
 Badness at net/sched/sch_generic.c:255
 NIP: c021cb5c LR: c021cb5c CTR: c01ab4b4
 [...]
 NIP [c021cb5c] dev_watchdog+0x298/0x2a8
 LR [c021cb5c] dev_watchdog+0x298/0x2a8
 Call Trace:
 [c0389da0] [c021cb5c] dev_watchdog+0x298/0x2a8 (unreliable)
 [c0389e00] [c0031ed8] run_timer_softirq+0x16c/0x1dc
 [c0389e50] [c002c638] __do_softirq+0xa4/0x11c
 [...]

This patch fixes the issue by properly detaching the device on
suspend, and attaching it back on resume.

Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agophylib: Properly reinitialize PHYs after hibernation
Anton Vorontsov [Wed, 30 Dec 2009 08:23:30 +0000 (08:23 +0000)]
phylib: Properly reinitialize PHYs after hibernation

Since hibernation assumes power loss, we should fully reinitialize
PHYs (including platform fixups), as if PHYs were just attached.

This patch factors phy_init_hw() out of phy_attach_direct(), then
converts mdio_bus to dev_pm_ops and adds an appropriate restore()
callback.

Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agophylib: Fix deadlock on resume
Anton Vorontsov [Wed, 30 Dec 2009 08:23:28 +0000 (08:23 +0000)]
phylib: Fix deadlock on resume

Sometimes kernel hangs on resume with the following trace:

 ucc_geth e0102000.ucc: resume
 INFO: task bash:1764 blocked for more than 120 seconds.
 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
 bash          D 0fecf43c     0  1764   1763 0x00000000
 Call Trace:
 [cf9a7c10] [c0012868] ret_from_except+0x0/0x14 (unreliable)
 --- Exception: cf9a7ce0 at __switch_to+0x4c/0x6c
     LR = 0xcf9a7cc0
 [cf9a7cd0] [c0008c14] __switch_to+0x4c/0x6c (unreliable)
 [cf9a7ce0] [c028bcfc] schedule+0x158/0x260
 [cf9a7d10] [c028c720] __mutex_lock_slowpath+0x80/0xd8
 [cf9a7d40] [c01cf388] phy_stop+0x20/0x70
 [cf9a7d50] [c01d514c] ugeth_resume+0x6c/0x13c
 [...]

Here is why.

On suspend:

- PM core starts suspending devices, ucc_geth_suspend gets called;

- ucc_geth calls phy_stop() on suspend. Note that phy_stop() is
  mostly asynchronous so it doesn't block ucc_geth's suspend routine,
  it just sets PHY_HALTED state and disables PHY's interrupts;

- Suddenly the state machine gets scheduled, it grabs the phydev->lock
  mutex and tries to process the PHY_HALTED state, so it calls
  phydev->adjust_link(phydev->attached_dev). In ucc_geth case
  adjust_link() calls msleep(), which reschedules the code flow back to
  PM core, which now finishes suspend and so we end up sleeping with
  phydev->lock mutex held.

On resume:

- PM core starts resuming devices (notice that nobody rescheduled
  the state machine yet, so the mutex is still held), the core calls
  ucc_geth's resume routine;

- ucc_geth_resume restarts the PHY with phy_stop()/phy_start()
  sequence, and the phy_*() calls are trying to grab the phydev->lock
  mutex. Here comes the deadlock.

This patch fixes the issue by stopping the state machine on suspend
and starting it again on resume.

Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: David S. Miller <davem@davemloft.net>