Network folks reported that directing all MSI-X vectors of their multi
queue NICs to a single core can cause interrupt stack overflows when
enough interrupts fire at the same time.
This is caused by the fact that we run interrupt handlers by default
with interrupts enabled unless the driver reuqests the interrupt with
the IRQF_DISABLED set. The NIC handlers do not set this flag, so
simultaneous interrupts can nest unlimited and cause the stack
overflow.
The only safe counter measure is to run the interrupt handlers with
interrupts disabled. We can't switch to this mode in general right
now, but it is safe to do so for MSI interrupts.
Force IRQF_DISABLED for MSI interrupt handlers.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Cc: Andi Kleen <andi@firstfloor.org> Cc: Linus Torvalds <torvalds@osdl.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Alan Cox <alan@lxorguk.ukuu.org.uk> Cc: David Miller <davem@davemloft.net> Cc: Greg Kroah-Hartman <gregkh@suse.de> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
These just represent the secondary and further heads attached to the
card, and they have different sets of PCI bar registers to map.
So don't try to drive them in the main driver.
Reported-by: Frans van Berckel <fberckel@xs4all.nl> Tested-by: Frans van Berckel <fberckel@xs4all.nl> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When there is a need to restart/reconfig hw, tear down all the
aggregation queues and let the mac80211 and driver get in-sync to have
the opportunity to re-establish the aggregation queues again.
Need to wait until driver re-establish all the station information before tear
down the aggregation queues, driver(at least iwlwifi driver) will reject the
stop aggregation queue request if station is not ready. But also need to make
sure the aggregation queues are tear down before waking up the queues, so
mac80211 will not sending frames with aggregation bit set.
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
"mac80211: fix skb buffering issue" still left a race
between enabling the hardware queues and the virtual
interface queues. In hindsight it's totally obvious
that enabling the netdev queues for a hardware queue
when the hardware queue is enabled is wrong, because
it could well possible that we can fill the hw queue
with packets we already have pending. Thus, we must
only enable the netdev queues once all the pending
packets have been processed and sent off to the device.
In testing, I haven't been able to trigger this race
condition, but it's clearly there, possibly only when
aggregation is being enabled.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
1st) a PREQ should only be processed, if it has the same SN and better
metric (instead of better or equal).
2nd) next_hop[ETH_ALEN] now actually used to buffer
mpath->next_hop->sta.addr for use out of lock.
Signed-off-by: Marco Porsch <marco.porsch@siemens.com> Acked-by: Javier Cardona <javier@cozybit.com> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
An int urb is constructed but we fill it in with a bulk pipe type.
Commit f661c6f8c67bd55e93348f160d590ff9edf08904 implemented a pipe type
check when CONFIG_USB_DEBUG is enabled. The check failed for all the ar9170
usb transfers and the driver could not configure the wifi dongle.
This went unnoticed until now because most people don't have
CONFIG_USB_DEBUG enabled.
Signed-off-by: Valentin Longchamp <valentin.longchamp@epfl.ch> Acked-by: Christian Lamparter <chunkeey@googlemail.com> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
IWL_RATE_COUNT is 13 and IWL_RATE_COUNT_LEGACY is 12.
IWL_RATE_COUNT_LEGACY is the right one here because iwl3945_rates
doesn't support 60M and also that's how "rates" is defined in
iwlcore_init_geos() from drivers/net/wireless/iwlwifi/iwl-core.c.
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> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When the cgroup freezer is used to freeze tasks we do not want to thaw
those tasks during resume. Currently we test the cgroup freezer
state of the resuming tasks to see if the cgroup is FROZEN. If so
then we don't thaw the task. However, the FREEZING state also indicates
that the task should remain frozen.
This also avoids a problem pointed out by Oren Ladaan: the freezer state
transition from FREEZING to FROZEN is updated lazily when userspace reads
or writes the freezer.state file in the cgroup filesystem. This means that
resume will thaw tasks in cgroups which should be in the FROZEN state if
there is no read/write of the freezer.state file to trigger this
transition before suspend.
NOTE: Another "simple" solution would be to always update the cgroup
freezer state during resume. However it's a bad choice for several reasons:
Updating the cgroup freezer state is somewhat expensive because it requires
walking all the tasks in the cgroup and checking if they are each frozen.
Worse, this could easily make resume run in N^2 time where N is the number
of tasks in the cgroup. Finally, updating the freezer state from this code
path requires trickier locking because of the way locks must be ordered.
Instead of updating the freezer state we rely on the fact that lazy
updates only manage the transition from FREEZING to FROZEN. We know that
a cgroup with the FREEZING state may actually be FROZEN so test for that
state too. This makes sense in the resume path even for partially-frozen
cgroups -- those that really are FREEZING but not FROZEN.
Reported-by: Oren Ladaan <orenl@cs.columbia.edu> Signed-off-by: Matt Helsley <matthltc@us.ibm.com> Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
It is required to call hw_breakpoint_init() on an attr before using it
in any other calls. This fixes the problem where kgdb will sometimes
fail to initialize on x86_64.
We could be failing/stopping a connection due to libiscsi starting
recovery/cleanup, but the xmit path or scsi eh thread path
could be dropping the connection at the same time.
As a result the session->state gets set to failed instead of in
recovery. We end up not blocking the session
and so the replacement timeout never gets started and we only end up
failing the IO when scsi_softirq_done sees that the
cmd has been running for (cmd->allowed + 1) * rq->timeout secs.
We used to fail the IO right away so users are seeing a long
delay when using dm-multipath. This problem was added in
2.6.28.
Signed-off-by: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: James Bottomley <James.Bottomley@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The error label releases the lock before calling journal_release_error,
but it requires that the lock be held. do_journal_release unlocks and
retakes it. When it releases it without it held, we trigger a BUG().
The error_alloc label skips the unlock since the lock isn't held yet
but none of the other conditions that are clean up exist yet either.
This patch returns immediately after the kzalloc failure and moves
the reiserfs_write_unlock after the journal_release_error call.
This was reported in https://bugzilla.novell.com/show_bug.cgi?id=591807
Reported-by: Thomas Siedentopf <thomas.siedentopf@novell.com> Signed-off-by: Jeff Mahoney <jeffm@suse.com> Cc: Thomas Siedentopf <thomas.siedentopf@novell.com> Cc: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
For the boot, enable_mmu() is called from setup_arch() but we don't call
setup_arch() for any of the other cpus. So turn on the non-boot cpu's
mmu inside of start_secondary().
I noticed this bug on an SMP board when trying to map I/O memory
(smsc911x registers) into the kernel address space. Since the Address
Translation bit in MMUCR wasn't set, accessing the virtual address where
the smsc911x registers were supposedly mapped actually performed a
physical address access.
Signed-off-by: Matt Fleming <matt@console-pimps.org> Signed-off-by: Paul Mundt <lethal@linux-sh.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Replace a BUG_ON with an error code in the event that the inode mapping
changes between calls to drm_open. This may happen for instance if udev
is loaded subsequent to the original opening of the device:
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
drm/radeon/kms: Fix NULL pointer dereference if memory allocation failed in a simple way
> From: Pauli Nieminen <suokkos@gmail.com>
> Date: Fri, 19 Mar 2010 07:44:33 +0000
> Subject: drm/radeon/kms: Fix NULL pointer dereference if memory allocation failed.
>
> From: Pauli Nieminen <suokkos@gmail.com>
>
> commit fcbc451ba1948fba967198bd150ecbd10bbb7075 upstream.
>
> When there is allocation failure in radeon_cs_parser_relocs parser->nrelocs
> is not cleaned. This causes NULL pointer defeference in radeon_cs_parser_fini
> when clean up code is trying to loop over the relocation array and free the
> objects.
>
> Fix adds a check for a possible NULL pointer in clean up code.
[...]
This patch breaks compiling kernel 2.6.33 + the current stable queue:
CC [M] drivers/gpu/drm/radeon/radeon_cs.o
/tmp/buildd/linux-sidux-2.6-2.6.33/debian/build/source_amd64_none/drivers/gpu/drm/radeon/radeon_cs.c: In function 'radeon_cs_parser_fini':
/tmp/buildd/linux-sidux-2.6-2.6.33/debian/build/source_amd64_none/drivers/gpu/drm/radeon/radeon_cs.c:200: error: implicit declaration of function 'drm_gem_object_unreference_unlocked'
make[6]: *** [drivers/gpu/drm/radeon/radeon_cs.o] Error 1
as it depends on the introduction of drm_gem_object_unreference_unlocked()
in:
This patch introduces the drm_gem_object_unreference_unlocked
and drm_gem_object_handle_unreference_unlocked functions that
do not require holding struct_mutex.
drm_gem_object_unreference_unlocked calls the new
->gem_free_object_unlocked entry point if available, and
otherwise just takes struct_mutex and just calls ->gem_free_object
Use drm_gem_object_[handle_]unreference_unlocked where possible
Mostly obvious simplifications.
The i915 pread/pwrite ioctls, intel_overlay_put_image and
nouveau_gem_new were incorrectly using the locked versions
without locking: this is also fixed in this patch.
which don't really look like candidates for 2.6.33-stable.
> --- a/drivers/gpu/drm/radeon/radeon_cs.c
> +++ b/drivers/gpu/drm/radeon/radeon_cs.c
> @@ -193,11 +193,13 @@ static void radeon_cs_parser_fini(struct
> radeon_bo_list_fence(&parser->validated, parser->ib->fence);
> }
> radeon_bo_list_unreserve(&parser->validated);
> - for (i = 0; i < parser->nrelocs; i++) {
> - if (parser->relocs[i].gobj) {
> - mutex_lock(&parser->rdev->ddev->struct_mutex);
> - drm_gem_object_unreference(parser->relocs[i].gobj);
> - mutex_unlock(&parser->rdev->ddev->struct_mutex);
> + if (parser->relocs != NULL) {
^ the only important part, the rest merely covers the new indentation
level
> + for (i = 0; i < parser->nrelocs; i++) {
> + if (parser->relocs[i].gobj) {
> + mutex_lock(&parser->rdev->ddev->struct_mutex);
> + drm_gem_object_unreference_unlocked(parser->relocs[i].gobj);
^ drm_gem_object_unreference_unlocked() doesn't exist in 2.6.33, yet
we can use drm_gem_object_unreference() instead.
As a consequence, I'd suggest to merely backport the NULL pointer check,
while ignoring the simplification of using the newly introduced
drm_gem_object_unreference_unlocked() from 2.6.34:
Signed-off-by: Stefan Lippers-Hollmann <s.l-h@gmx.de> Cc: Pauli Nieminen <suokkos@gmail.com> Cc: Dave Airlie <airlied@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
RS4xx+ IGP chips use an internal gart, however,
some of them have the agp cap bits set in their pci
configs. Make sure to clear the AGP flag as AGP will
not work with them.
Should fix fdo bug 27225
Signed-off-by: Alex Deucher <alexdeucher@gmail.com> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
proc_oom_score(task) has a reference to task_struct, but that is all.
If this task was already released before we take tasklist_lock
- we can't use task->group_leader, it points to nowhere
- it is not safe to call badness() even if this task is
->group_leader, has_intersects_mems_allowed() assumes
it is safe to iterate over ->thread_group list.
- even worse, badness() can hit ->signal == NULL
Add the pid_alive() check to ensure __unhash_process() was not called.
Also, use "task" instead of task->group_leader. badness() should return
the same result for any sub-thread. Currently this is not true, but
this should be changed anyway.
When using the string representation of a random counter as part of the base
name, ensure that it is no longer than 4 bytes.
Since we are repeatedly decrementing the counter in a loop until we have found a
unique base name, the counter may wrap around zero; therefore, it is not enough
to mask its higher bits before entering the loop, this must be done inside the
loop.
Now the EDID property will be updated when the corresponding EDID can be
obtained from the external display device. But after the external device
is plugged-out, the EDID property is not updated. In such case we still
get the corresponding EDID property although it is already detected as
disconnected.
gfs2_lock() will skip locks on file which have mode set to 02666. This is a problem in cases where the mode of the file is changed after a process has obtained a lock on the file. Such a lock will be skipped and will result in a BUG in locks_remove_flock().
gfs2_lock() should skip the check for mandatory locks when unlocking a file.
Instead of a MODULE_DEVICE_TABLE for every acpi_driver ids table, we
create a table containing all ids to export to get a module alias for
each one.
This will fix automatic loading of the driver when one of the ACPI
devices is not present (like the accelerometer, which is not present in
some models).
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@holoscopio.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
On UV systems, the TSC is not synchronized across blades. The
sched_clock_cpu() function is returning values that can go
backwards (I've seen as much as 8 seconds) when switching
between cpus.
As each cpu comes up, early_init_intel() will currently set the
sched_clock_stable flag true. When mark_tsc_unstable() runs, it
clears the flag, but this only occurs once (the first time a cpu
comes up whose TSC is not synchronized with cpu 0). After this,
early_init_intel() will set the flag again as the next cpu comes
up.
Only set sched_clock_stable if tsc has not been marked unstable.
The TKIP key update callback is called from the RX path, where the driver
mutex is already locked. This results in a circular locking bug.
Avoid this by removing the lock.
Johannes noted that there is a separate bug: The callback still breaks on SDIO
hardware, because SDIO hardware access needs to sleep, but we are not allowed
to sleep in the callback due to mac80211's RCU locking.
Signed-off-by: Michael Buesch <mb@bu3sch.de> Tested-by: Larry Finger <Larry.Finger@lwfinger.net> Reported-by: kecsa@kutfo.hit.bme.hu Cc: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Added very simple check that req buffer has enough space to
fit configuration parameters. Shall be enough to reject packets
with configuration size more than req buffer.
Crash trace below
[ 6069.659393] Unable to handle kernel paging request at virtual address 02000205
[ 6069.673034] Internal error: Oops: 805 [#1] PREEMPT
...
[ 6069.727172] PC is at l2cap_add_conf_opt+0x70/0xf0 [l2cap]
[ 6069.732604] LR is at l2cap_recv_frame+0x1350/0x2e78 [l2cap]
...
[ 6070.030303] Backtrace:
[ 6070.032806] [<bf1c2880>] (l2cap_add_conf_opt+0x0/0xf0 [l2cap]) from
[<bf1c6624>] (l2cap_recv_frame+0x1350/0x2e78 [l2cap])
[ 6070.043823] r8:dc5d3100 r7:df2a91d6 r6:00000001 r5:df2a8000 r4:00000200
[ 6070.050659] [<bf1c52d4>] (l2cap_recv_frame+0x0/0x2e78 [l2cap]) from
[<bf1c8408>] (l2cap_recv_acldata+0x2bc/0x350 [l2cap])
[ 6070.061798] [<bf1c814c>] (l2cap_recv_acldata+0x0/0x350 [l2cap]) from
[<bf0037a4>] (hci_rx_task+0x244/0x478 [bluetooth])
[ 6070.072631] r6:dc647700 r5:00000001 r4:df2ab740
[ 6070.077362] [<bf003560>] (hci_rx_task+0x0/0x478 [bluetooth]) from
[<c006b9fc>] (tasklet_action+0x78/0xd8)
[ 6070.087005] [<c006b984>] (tasklet_action+0x0/0xd8) from [<c006c160>]
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@nokia.com> Acked-by: Gustavo F. Padovan <gustavo@padovan.org> Signed-off-by: Marcel Holtmann <marcel@holtmann.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When creating a high number of Bluetooth sockets (L2CAP, SCO
and RFCOMM) it is possible to scribble repeatedly on arbitrary
pages of memory. Ensure that the content of these sysfs files is
always less than one page. Even if this means truncating. The
files in question are scheduled to be moved over to debugfs in
the future anyway.
Based on initial patches from Neil Brown and Linus Torvalds
Reported-by: Neil Brown <neilb@suse.de> Signed-off-by: Marcel Holtmann <marcel@holtmann.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
HP is recycling both DMI_PRODUCT_NAME and DMI_BIOS_VERSION making
ahci_broken_suspend() trigger for later products which are not
affected by the original problems. Match BIOS date instead of version
and add references to bko's so that full information can be found
easier later.
This fixes http://bugzilla.kernel.org/show_bug.cgi?id=15462
If a delayed-allocation write happens before quota is enabled, the
kernel spits out a warning:
WARNING: at fs/quota/dquot.c:988 dquot_claim_space+0x77/0x112()
because the fact that user has some delayed allocation is not recorded
in quota structure.
Make dquot_initialize() update amount of reserved space for user if it sees
inode has some space reserved. Also make sure that reserved quota space does
not go negative and we warn about the filesystem bug just once.
Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Since we implemented generic reserved space management interface,
then it is possible to account reserved space even when quota
is not active (similar to i_blocks/i_bytes).
Without this patch following testcase result in massive comlain from
WARN_ON in dquot_claim_space()
Commit 4957faad (TCPCT part 1g: Responder Cookie => Initiator), part
of TCP_COOKIE_TRANSACTION implementation, forgot to correctly size
synack skb in case user data must be included.
Many thanks to Mika Pentillä for spotting this error.
Reported-by: Penttillä Mika <mika.penttila@ixonos.com> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
A packet is marked as lost in case packets == 0, although nothing should be done.
This results in a too early retransmitted packet during recovery in some cases.
This small patch fixes this issue by returning immediately.
Signed-off-by: Lennart Schulte <lennart.schulte@nets.rwth-aachen.de> Signed-off-by: Arnd Hannemann <hannemann@nets.rwth-aachen.de> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The dma map fields in the skb_shared_info structure no longer has any users
and can be dropped since it is making the skb_shared_info unecessarily larger.
Running slabtop show that we were using 4K slabs for the skb->head on x86_64 w/
an allocation size of 1522. It turns out that the dma_head and dma_maps array
made skb_shared large enough that we had crossed over the 2k boundary with
standard frames and as such we were using 4k blocks of memory for all skbs.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com> Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com> Acked-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When doing "ifenslave -d bond0 eth0", there is chance to get NULL
dereference in netif_receive_skb(), because dev->master suddenly becomes
NULL after we tested it.
We should use ACCESS_ONCE() to avoid this (or rcu_dereference())
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Official patch to fix the r8169 frame length check error.
Based on this initial thread:
http://marc.info/?l=linux-netdev&m=126202972828626&w=1
This is the official patch to fix the frame length problems in the r8169
driver. As noted in the previous thread, while this patch incurs a performance
hit on the driver, its possible to improve performance dynamically by updating
the mtu and rx_copybreak values at runtime to return performance to what it was
for those NICS which are unaffected by the ideosyncracy (if there are any).
Summary:
A while back Eric submitted a patch for r8169 in which the proper
allocated frame size was written to RXMaxSize to prevent the NIC from dmaing too
much data. This was done in commit fdd7b4c3302c93f6833e338903ea77245eb510b4. A
long time prior to that however, Francois posted 126fa4b9ca5d9d7cb7d46f779ad3bd3631ca387c, which expiclitly disabled the MaxSize
setting due to the fact that the hardware behaved in odd ways when overlong
frames were received on NIC's supported by this driver. This was mentioned in a
security conference recently:
http://events.ccc.de/congress/2009/Fahrplan//events/3596.en.html
It seems that if we can't enable frame size filtering, then, as Eric correctly
noticed, we can find ourselves DMA-ing too much data to a buffer, causing
corruption. As a result is seems that we are forced to allocate a frame which
is ready to handle a maximally sized receive.
This obviously has performance issues with it, so to mitigate that issue, this
patch does two things:
1) Raises the copybreak value to the frame allocation size, which should force
appropriately sized packets to get allocated on rx, rather than a full new 16k
buffer.
2) This patch only disables frame filtering initially (i.e., during the NIC
open), changing the MTU results in ring buffer allocation of a size in relation
to the new mtu (along with a warning indicating that this is dangerous).
Because of item (2), individuals who can't cope with the performance hit (or can
otherwise filter frames to prevent the bug), or who have hardware they are sure
is unaffected by this issue, can manually lower the copybreak and reset the mtu
such that performance is restored easily.
Signed-off-by: Neil Horman <nhorman@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch fixes a bug that allows to lose events when reliable
event delivery mode is used, ie. if NETLINK_BROADCAST_SEND_ERROR
and NETLINK_RECV_NO_ENOBUFS socket options are set.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Currently, ENOBUFS errors are reported to the socket via
netlink_set_err() even if NETLINK_RECV_NO_ENOBUFS is set. However,
that should not happen. This fixes this problem and it changes the
prototype of netlink_set_err() to return the number of sockets that
have set the NETLINK_RECV_NO_ENOBUFS socket option. This return
value is used in the next patch in these bugfix series.
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Under NET_DMA, data transfer can grind to a halt when userland issues a
large read on a socket with a high RCVLOWAT (i.e., 512 KB for both).
This appears to be because the NET_DMA design queues up lots of memcpy
operations, but doesn't issue or wait for them (and thus free the
associated skbs) until it is time for tcp_recvmesg() to return.
The socket hangs when its TCP window goes to zero before enough data is
available to satisfy the read.
Periodically issue asynchronous memcpy operations, and free skbs for ones
that have completed, to prevent sockets from going into zero-window mode.
Signed-off-by: Steven J. Magnani <steve@digidescorp.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This is ipv6 variant of the commit 5e016cbf6.. ("ipv4: Don't drop
redirected route cache entry unless PTMU actually expired")
by Guenter Roeck <guenter.roeck@ericsson.com>.
Remove cache route entry in ipv6_negative_advice() only if
the timer is expired.
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
TCP sessions over IPv4 can get stuck if routers between endpoints
do not fragment packets but implement PMTU instead, and we are using
those routers because of an ICMP redirect.
Setup is as follows
MTU1 MTU2 MTU1
A--------B------C------D
with MTU1 > MTU2. A and D are endpoints, B and C are routers. B and C
implement PMTU and drop packets larger than MTU2 (for example because
DF is set on all packets). TCP sessions are initiated between A and D.
There is packet loss between A and D, causing frequent TCP
retransmits.
After the number of retransmits on a TCP session reaches tcp_retries1,
tcp calls dst_negative_advice() prior to each retransmit. This results
in route cache entries for the peer to be deleted in
ipv4_negative_advice() if the Path MTU is set.
If the outstanding data on an affected TCP session is larger than
MTU2, packets sent from the endpoints will be dropped by B or C, and
ICMP NEEDFRAG will be returned. A and D receive NEEDFRAG messages and
update PMTU.
Before the next retransmit, tcp will again call dst_negative_advice(),
causing the route cache entry (with correct PMTU) to be deleted. The
retransmitted packet will be larger than MTU2, causing it to be
dropped again.
This sequence repeats until the TCP session aborts or is terminated.
Problem is fixed by removing redirected route cache entries in
ipv4_negative_advice() only if the PMTU is expired.
Signed-off-by: Guenter Roeck <guenter.roeck@ericsson.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Xfrm_dst keeps a reference to ipv4 rtable entries on each
cached bundle. The only way to renew xfrm_dst when the underlying
route has changed, is to implement dst_check for this. This is
what ipv6 side does too.
The problems started after 87c1e12b5eeb7b30b4b41291bef8e0b41fc3dde9
("ipsec: Fix bogus bundle flowi") which fixed a bug causing xfrm_dst
to not get reused, until that all lookups always generated new
xfrm_dst with new route reference and path mtu worked. But after the
fix, the old routes started to get reused even after they were expired
causing pmtu to break (well it would occationally work if the rtable
gc had run recently and marked the route obsolete causing dst_check to
get called).
Signed-off-by: Timo Teras <timo.teras@iki.fi> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When I merged the bundle creation code, I introduced a bogus
flowi value in the bundle. Instead of getting from the caller,
it was instead set to the flow in the route object, which is
totally different.
The end result is that the bundles we created never match, and
we instead end up with an ever growing bundle list.
Thanks to Jamal for find this problem.
Reported-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Acked-by: Steffen Klassert <steffen.klassert@secunet.com> Acked-by: Jamal Hadi Salim <hadi@cyberus.ca> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Taking route's header_len into account, and updating gre device
needed_headroom will give better hints on upper bound of required
headroom. This is useful if the gre traffic is xfrm'ed.
Signed-off-by: Timo Teras <timo.teras@iki.fi> Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This introduced a problem that can happen if the TX ring size
is increased. We need to size the PCI pool using cbs->max
instead of the default cbs->count value.
Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When the register_netdevice() call fails, the newly allocated device is
not freed.
Signed-off-by: Patrick McHardy <kaber@trash.net> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Zhu Yi <yi.zhu@intel.com> Acked-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Make x25 adapt to the limited socket backlog change.
Cc: Andrew Hendry <andrew.hendry@gmail.com> Signed-off-by: Zhu Yi <yi.zhu@intel.com> Acked-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Make tipc adapt to the limited socket backlog change.
Cc: Jon Maloy <jon.maloy@ericsson.com> Cc: Allan Stephens <allan.stephens@windriver.com> Signed-off-by: Zhu Yi <yi.zhu@intel.com> Acked-by: Eric Dumazet <eric.dumazet@gmail.com> Acked-by: Allan Stephens <allan.stephens@windriver.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Make llc adapt to the limited socket backlog change.
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net> Signed-off-by: Zhu Yi <yi.zhu@intel.com> Acked-by: Eric Dumazet <eric.dumazet@gmail.com> Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Make udp adapt to the limited socket backlog change.
Cc: "David S. Miller" <davem@davemloft.net> Cc: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru> Cc: "Pekka Savola (ipv6)" <pekkas@netcore.fi> Cc: Patrick McHardy <kaber@trash.net> Signed-off-by: Zhu Yi <yi.zhu@intel.com> Acked-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Make tcp adapt to the limited socket backlog change.
Cc: "David S. Miller" <davem@davemloft.net> Cc: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru> Cc: "Pekka Savola (ipv6)" <pekkas@netcore.fi> Cc: Patrick McHardy <kaber@trash.net> Signed-off-by: Zhu Yi <yi.zhu@intel.com> Acked-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
We got system OOM while running some UDP netperf testing on the loopback
device. The case is multiple senders sent stream UDP packets to a single
receiver via loopback on local host. Of course, the receiver is not able
to handle all the packets in time. But we surprisingly found that these
packets were not discarded due to the receiver's sk->sk_rcvbuf limit.
Instead, they are kept queuing to sk->sk_backlog and finally ate up all
the memory. We believe this is a secure hole that a none privileged user
can crash the system.
The root cause for this problem is, when the receiver is doing
__release_sock() (i.e. after userspace recv, kernel udp_recvmsg ->
skb_free_datagram_locked -> release_sock), it moves skbs from backlog to
sk_receive_queue with the softirq enabled. In the above case, multiple
busy senders will almost make it an endless loop. The skbs in the
backlog end up eat all the system memory.
The issue is not only for UDP. Any protocols using socket backlog is
potentially affected. The patch adds limit for socket backlog so that
the backlog size cannot be expanded endlessly.
Reported-by: Alex Shi <alex.shi@intel.com> Cc: David Miller <davem@davemloft.net> Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net> Cc: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru> Cc: "Pekka Savola (ipv6)" <pekkas@netcore.fi> Cc: Patrick McHardy <kaber@trash.net> Cc: Vlad Yasevich <vladislav.yasevich@hp.com> Cc: Sridhar Samudrala <sri@us.ibm.com> Cc: Jon Maloy <jon.maloy@ericsson.com> Cc: Allan Stephens <allan.stephens@windriver.com> Cc: Andrew Hendry <andrew.hendry@gmail.com> Signed-off-by: Zhu Yi <yi.zhu@intel.com> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
route: Fix caught BUG_ON during rt_secret_rebuild_oneshot()
Call rt_secret_rebuild can cause BUG_ON(timer_pending(&net->ipv4.rt_secret_timer)) in
add_timer as there is not any synchronization for call rt_secret_rebuild_oneshot()
for the same net namespace.
Also this issue affects to rt_secret_reschedule().
Thus use mod_timer enstead.
Signed-off-by: Vitaliy Gusev <vgusev@openvz.org> Acked-by: Neil Horman <nhorman@tuxdriver.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch fixes UDP socket refcnt bugs in the pppol2tp driver.
A bug can cause a kernel stack trace when a tunnel socket is closed.
A way to reproduce the issue is to prepare the UDP socket for L2TP (by
opening a tunnel pppol2tp socket) and then close it before any L2TP
sessions are added to it. The sequence is
Create UDP socket
Create tunnel pppol2tp socket to prepare UDP socket for L2TP
pppol2tp_connect: session_id=0, peer_session_id=0
L2TP SCCRP control frame received (tunnel_id==0)
pppol2tp_recv_core: sock_hold()
pppol2tp_recv_core: sock_put
L2TP ZLB control frame received (tunnel_id=nnn)
pppol2tp_recv_core: sock_hold()
pppol2tp_recv_core: sock_put
Close tunnel management socket
pppol2tp_release: session_id=0, peer_session_id=0
Close UDP socket
udp_lib_close: BUG
The addition of sock_hold() in pppol2tp_connect() solves the problem.
For data frames, two sock_put() calls were added to plug a refcnt leak
per received data frame. The ref that is grabbed at the top of
pppol2tp_recv_core() must always be released, but this wasn't done for
accepted data frames or data frames discarded because of bad UDP
checksums. This leak meant that any UDP socket that had passed L2TP
data traffic (i.e. L2TP data frames, not just L2TP control frames)
using pppol2tp would not be released by the kernel.
Signed-off-by: James Chapman <jchapman@katalix.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When transmitting L2TP frames, we derive the outgoing interface's UDP
checksum hardware assist capabilities from the tunnel dst dev. This
can sometimes be NULL, especially when routing protocols are used and
routing changes occur. This patch just checks for NULL dst or dev
pointers when checking for netdev hardware assist features.
Signed-off-by: James Chapman <jchapman@katalix.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: David S. Miller <davem@davemloft.net> Acked-by: Frans van Berckel <fberckel@xs4all.nl> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
For 32-bit processes, we save the full 64-bits of the regs in pt_regs.
But unlike when the userspace actually does load and store
instructions, the top 32-bits don't get automatically truncated by the
cpu in kernel mode (because the kernel doesn't execute with PSTATE_AM
address masking enabled).
So we have to do it by hand.
Reported-by: Frederic Weisbecker <fweisbec@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
qlogicpti driver registers its irq with a name containing slash.
This results in
[ 71.049735] WARNING: at fs/proc/generic.c:316 __xlate_proc_name+0xa8/0xb8()
[ 71.132815] name 'Qlogic/PTI'
because proc_mkdir with the name of the irq fails. Fix it by just
removing the slash from irq name. Discovered and tested on real hardware
(Sun Ultra 1).
Signed-off-by: Meelis Roos <mroos@linux.ee> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
On Sun, Jan 03, 2010 at 12:23:14AM +0000, Russell King wrote:
> - with IDE
> - locks the interrupt line, and makes the machine extremely painful -
> about an hour to get to the point of being able to unload the
> pdc202xx_old module.
Having manually bisected kernel versions, I've narrowed it down to some
change between 2.6.30 and 2.6.31. There's not much which has changed
between the two kernels, but one change stands out like a sore thumb:
+static int pdc202xx_test_irq(ide_hwif_t *hwif)
+{
+ struct pci_dev *dev = to_pci_dev(hwif->dev);
+ unsigned long high_16 = pci_resource_start(dev, 4);
+ u8 sc1d = inb(high_16 + 0x1d);
+
+ if (hwif->channel) {
+ /*
+ * bit 7: error, bit 6: interrupting,
+ * bit 5: FIFO full, bit 4: FIFO empty
+ */
+ return ((sc1d & 0x50) == 0x40) ? 1 : 0;
+ } else {
+ /*
+ * bit 3: error, bit 2: interrupting,
+ * bit 1: FIFO full, bit 0: FIFO empty
+ */
+ return ((sc1d & 0x05) == 0x04) ? 1 : 0;
+ }
+}
Reading the (documented as a 32-bit) system control register when the
interface is idle gives: 0x01da110c
So, the byte at 0x1d is 0x11, which is documented as meaning that the
primary and secondary FIFOs are empty.
The code above, which is trying to see whether an IRQ is pending, checks
for the IRQ bit to be one, and the FIFO bit to be zero - or in English,
to be non-empty.
Since during a BM-DMA read, the FIFOs will naturally be drained to the
PCI bus, the chance of us getting to the interface before this happens
are extremely small - and if we don't, it means we decide not to service
the interrupt. Hence, the screaming interrupt problem with drivers/ide.
Fix this by only indicating an interrupt is ready if both the interrupt
and FIFO empty bits are at '1'.
This bug only affects PDC20246/PDC20247 (Promise Ultra33) based cards,
and has been tested on 2.6.31 and 2.6.33-rc2.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Tested-by: Russell King <rmk+kernel@arm.linux.org.uk> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
As requested by David Fries. This makes CDROMs which are slave drives
on a ribbon without a master disappear and causes other similar kinds
of badness.
Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Bring back ->maskproc method since it is still needed for proper operation,
as noticed by Russell King:
> This change is bogus.
>
> writeb(0, base + ICS_ARCIN_V6_INTROFFSET_1);
> readb(base + ICS_ARCIN_V6_INTROFFSET_2);
>
> writeb(0, base + ICS_ARCIN_V6_INTROFFSET_2);
> readb(base + ICS_ARCIN_V6_INTROFFSET_1);
>
> This sequence of code does:
>
> 1. enable interrupt 1
> 2. disable interrupt 2
> 3. enable interrupt 2
> 4. disable interrupt 1
>
> which results in the interrupt for the second channel being enabled -
> leaving channel 1 blocked.
>
> Firstly, icside shares its two IDE channels with one DMA engine - so it's
> a simplex interface. IDE supports those (or did when the code was written)
> serializing requests between the two interfaces. libata does not.
>
> Secondly, the interrupt lines on icside float when there's no drive connected
> or when the drive has its NIEN bit set, which means that you get spurious
> screaming interrupts which can kill off all expansion card interrupts on
> the machine unless you disable the channel interrupt on the card.
>
> Since libata can not serialize the operation of the two channels like IDE
> can, the libata version of the icside driver does not contain the interrupt
> stearing logic. Instead, it looks at the status after reset, and if
> nothing was found on that channel, it masks the interrupt from that
> channel.
This patch reverts changes done in commit dff8817 (I became confused due to
non-standard & undocumented ->maskproc method, anyway sorry about that).
Noticed-by: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Vivek Natarajan <vnatarajan@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Even if the null data frame is not acked by the AP, mac80211
goes into power save. This might lead to loss of frames
from the AP.
Prevent this by restarting dynamic_ps_timer when ack is not
received for null data frames.
Cc: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: Vivek Natarajan <vnatarajan@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
beacons configuration SHOULD be done only if the STA is associated.
Signed-off-by: Senthil Balasubramanian <senthilkumar@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The TIM timer interrupt is enabled even before the ACK of nullqos
is received which is unnecessary.
Also clean up the CONF_PS part of config callback properly for
better readability.
Signed-off-by: Senthil Balasubramanian <senthilkumar@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
BugLink: https://launchpad.net/bugs/303789
This model needs both 'Headphone Jack Sense' and 'Line Jack Sense'
muted for audible audio, so just add its SSID to the blacklist and
don't enumerate the controls.
Signed-off-by: Daniel T Chen <crimsun@ubuntu.com> Signed-off-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
BugLink: https://launchpad.net/bugs/481058
The OR has verified that both 'Headphone Jack Sense' and 'Line Jack Sense'
need to be muted for sound to be audible, so just add the machine's SSID
to the ac97 jack sense blacklist.
Reported-by: Richard Gagne Tested-by: Richard Gagne Signed-off-by: Daniel T Chen <crimsun@ubuntu.com> Signed-off-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Add missing newline to dev_warn() message string. This is more of an issue
with older kernels that don't automatically add a newline if it was missing
from the end of the previous line.
Signed-off-by: Dean Nelson <dnelson@redhat.com> Signed-off-by: Jean Delvare <khali@linux-fr.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Currently c1e_idle returns true for all CPUs greater than or equal to
family 0xf model 0x40. This covers too many CPUs.
Meanwhile a respective erratum for the underlying problem was filed
(#400). This patch adds the logic to check whether erratum #400
applies to a given CPU.
Especially for CPUs where SMI/HW triggered C1e is not supported,
c1e_idle() doesn't need to be used. We can check this by looking at
the respective OSVW bit for erratum #400.
Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
LKML-Reference: <20100319110922.GA19614@alberich.amd.com> Signed-off-by: H. Peter Anvin <hpa@zytor.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Ever for 32-bit with sufficiently high NR_CPUS, and starting
with commit 789d03f584484af85dbdc64935270c8e45f36ef7 also for
64-bit, the statically allocated early fixmap page tables were
not covering FIX_OHCI1394_BASE, leading to a boot time crash
when "ohci1394_dma=early" was used. Despite this entry not being
a permanently used one, it needs to be moved into the permanent
range since it has to be close to FIX_DBGP_BASE and
FIX_EARLYCON_MEM_BASE.
Commit 8ccb92ad (netfilter: xt_recent: fix false match) fixed supposedly
false matches in rules using a zero hit_count. As it turns out there is
nothing false about these matches and people are actually using entries
with a hit_count of zero to make rules dependant on addresses inserted
manually through /proc.
Since this slipped past the eyes of three reviewers, instead of
reverting the commit in question, this patch explicitly checks
for a hit_count of zero to make the intentions more clear.
Reported-by: Thomas Jarosch <thomas.jarosch@intra2net.com> Tested-by: Thomas Jarosch <thomas.jarosch@intra2net.com> Signed-off-by: Patrick McHardy <kaber@trash.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This is in preference to disconnected. If there's no other outputs
connected this will cause LVDS to be programmed even with the lid
closed rather than having X fail to start because of no available
outputs.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Reported-by: maciej.rutecki@gmail.com Cc: stable@kernel.org Reviewed-by: Eric Anholt <eric@anholt.net> Signed-off-by: Eric Anholt <eric@anholt.net> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Ensure additions on touch_ts do not overflow. This can occur
when the top 32 bits of the TSC reach 0xffffffff causing
additions to touch_ts to overflow and this in turn generates
spurious softlockup warnings.
Signed-off-by: Colin Ian King <colin.king@canonical.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Eric Dumazet <eric.dumazet@gmail.com>
LKML-Reference: <1268994482.1798.6.camel@lenovo> Signed-off-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Lars-Peter pointed out that the oneshot threaded interrupt handler
code has the following race:
CPU0 CPU1
hande_level_irq(irq X)
mask_ack_irq(irq X)
handle_IRQ_event(irq X)
wake_up(thread_handler)
thread handler(irq X) runs
finalize_oneshot(irq X)
does not unmask due to
!(desc->status & IRQ_MASKED)
return from irq
does not unmask due to
(desc->status & IRQ_ONESHOT)
This leaves the interrupt line masked forever.
The reason for this is the inconsistent handling of the IRQ_MASKED
flag. Instead of setting it in the mask function the oneshot support
sets the flag after waking up the irq thread.
The solution for this is to set/clear the IRQ_MASKED status whenever
we mask/unmask an interrupt line. That's the easy part, but that
cleanup opens another race:
CPU0 CPU1
hande_level_irq(irq)
mask_ack_irq(irq)
handle_IRQ_event(irq)
wake_up(thread_handler)
thread handler(irq) runs
finalize_oneshot_irq(irq)
unmask(irq)
irq triggers again
handle_level_irq(irq)
mask_ack_irq(irq)
return from irq due to IRQ_INPROGRESS
return from irq
does not unmask due to
(desc->status & IRQ_ONESHOT)
This requires that we synchronize finalize_oneshot_irq() with the
primary handler. If IRQ_INPROGESS is set we wait until the primary
handler on the other CPU has returned before unmasking the interrupt
line again.
We probably have never seen that problem because it does not happen on
UP and on SMP the irqbalancer protects us by pinning the primary
handler and the thread to the same CPU.
According to the report from Andreas Beckmann (Message-ID:
<4BA54677.3090902@abeckmann.de>), nilfs in 2.6.33 kernel got stuck
after a disk full error.
This turned out to be a regression by log writer updates merged at
kernel 2.6.33. nilfs_segctor_abort_construction, which is a cleanup
function for erroneous cases, was skipping writeback completion for
some logs.
This fixes the bug and would resolve the hang issue.
When I initially stumbled upon sequence number problems with PAE frames
in ath9k, I submitted a patch to remove all special cases for PAE
frames and let them go through the normal transmit path.
Out of concern about crypto incompatibility issues, this change was
merged instead:
After a lot of testing, I'm able to reliably trigger a driver crash on
rekeying with current versions with this change in place.
It seems that the driver does not support sending out regular MPDUs with
the same TID while an A-MPDU session is active.
This leads to duplicate entries in the TID Tx buffer, which hits the
following BUG_ON in ath_tx_addto_baw():
I believe until we actually have a reproducible case of an
incompatibility with another AP using no PAE special cases, we should
simply get rid of this mess.
This patch completely fixes my crash issues in STA mode and makes it
stay connected without throughput drops or connectivity issues even
when the AP is configured to a very short group rekey interval.
Signed-off-by: Felix Fietkau <nbd@openwrt.org> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The PCIBIOS_ errors are returned from the PCI functions generated by the
PCI_OP_READ() and PCI_OP_WRITE() macros.
In a similar manner, pcix_set_mmrbc() also returns the PCIBIOS_ error values
returned from pci_read_config_[word|dword]() and pci_write_config_word().
Following pcix_get_max_mmrbc()'s example, the following patch simply returns
-EINVAL for all PCIBIOS_ errors encountered by pcix_get_mmrbc(), and -EINVAL
or -EIO for those encountered by pcix_set_mmrbc().
This simplification was chosen in light of the fact that none of the current
callers of these functions are interested in the specific type of error
encountered. In the future, should this change, one could simply create a
function that maps each PCIBIOS_ error to a corresponding unique errno value,
which could be called by pcix_get_max_mmrbc(), pcix_get_mmrbc(), and
pcix_set_mmrbc().
Additionally, this patch eliminates some unnecessary variables.
Signed-off-by: Dean Nelson <dnelson@redhat.com> Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
An e1000 driver on a system with a PCI-X bus was always being returned
a value of 135 from both pcix_get_mmrbc() and pcix_set_mmrbc(). This
value reflects an error return of PCIBIOS_BAD_REGISTER_NUMBER from
pci_bus_read_config_dword(,, cap + PCI_X_CMD,).
This is because for a dword, the following portion of the PCI_OP_READ()
macro:
if (PCI_##size##_BAD) return PCIBIOS_BAD_REGISTER_NUMBER;
expands to:
if (pos & 3) return PCIBIOS_BAD_REGISTER_NUMBER;
And is always true for 'cap + PCI_X_CMD', which is 0xe4 + 2 = 0xe6. ('cap' is
the result of calling pci_find_capability(, PCI_CAP_ID_PCIX).)
The same problem exists for pci_bus_write_config_dword(,, cap + PCI_X_CMD,).
In both cases, instead of calling _dword(), _word() should be called.
Signed-off-by: Dean Nelson <dnelson@redhat.com> Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
For the PCI_X_STATUS register, pcix_get_max_mmrbc() is returning an incorrect
value, which is based on:
(stat & PCI_X_STATUS_MAX_READ) >> 12
Valid return values are 512, 1024, 2048, 4096, which correspond to a 'stat'
(masked and right shifted by 21) of 0, 1, 2, 3, respectively.
A right shift by 11 would generate the correct return value when 'stat' (masked
and right shifted by 21) has a value of 1 or 2. But for a value of 0 or 3 it's
not possible to generate the correct return value by only right shifting.
Fix is based on pcix_get_mmrbc()'s similar dealings with the PCI_X_CMD register.
Signed-off-by: Dean Nelson <dnelson@redhat.com> Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
projects / karo-tx-linux.git / log