The current mutex spinning code (with MUTEX_SPIN_ON_OWNER option
turned on) allow multiple tasks to spin on a single mutex
concurrently. A potential problem with the current approach is
that when the mutex becomes available, all the spinning tasks
will try to acquire the mutex more or less simultaneously. As a
result, there will be a lot of cacheline bouncing especially on
systems with a large number of CPUs.
This patch tries to reduce this kind of contention by putting
the mutex spinners into a queue so that only the first one in
the queue will try to acquire the mutex. This will reduce
contention and allow all the tasks to move forward faster.
The queuing of mutex spinners is done using an MCS lock based
implementation which will further reduce contention on the mutex
cacheline than a similar ticket spinlock based implementation.
This patch will add a new field into the mutex data structure
for holding the MCS lock. This expands the mutex size by 8 bytes
for 64-bit system and 4 bytes for 32-bit system. This overhead
will be avoid if the MUTEX_SPIN_ON_OWNER option is turned off.
The following table shows the jobs per minute (JPM) scalability
data on an 8-node 80-core Westmere box with a 3.7.10 kernel. The
numactl command is used to restrict the running of the fserver
workloads to 1/2/4/8 nodes with hyperthreading off.
+-----------------+-----------+-----------+-------------+----------+
| Configuration | Mean JPM | Mean JPM | Mean JPM | % Change |
| | w/o patch | patch 1 | patches 1&2 | 1->1&2 |
+-----------------+------------------------------------------------+
| | User Range 1100 - 2000 |
+-----------------+------------------------------------------------+
| 8 nodes, HT off | 227972 | 227237 | 305043 | +34.2% |
| 4 nodes, HT off | 393503 | 381558 | 394650 | +3.4% |
| 2 nodes, HT off | 334957 | 325240 | 338853 | +4.2% |
| 1 node , HT off | 198141 | 197972 | 198075 | +0.1% |
+-----------------+------------------------------------------------+
| | User Range 200 - 1000 |
+-----------------+------------------------------------------------+
| 8 nodes, HT off | 282325 | 312870 | 332185 | +6.2% |
| 4 nodes, HT off | 390698 | 378279 | 393419 | +4.0% |
| 2 nodes, HT off | 336986 | 326543 | 340260 | +4.2% |
| 1 node , HT off | 197588 | 197622 | 197582 | 0.0% |
+-----------------+-----------+-----------+-------------+----------+
At low user range 10-100, the JPM differences were within +/-1%.
So they are not that interesting.
The fserver workload uses mutex spinning extensively. With just
the mutex change in the first patch, there is no noticeable
change in performance. Rather, there is a slight drop in
performance. This mutex spinning patch more than recovers the
lost performance and show a significant increase of +30% at high
user load with the full 8 nodes. Similar improvements were also
seen in a 3.8 kernel.
The table below shows the %time spent by different kernel
functions as reported by perf when running the fserver workload
at 1500 users with all 8 nodes.
+-----------------------+-----------+---------+-------------+
| Function | % time | % time | % time |
| | w/o patch | patch 1 | patches 1&2 |
+-----------------------+-----------+---------+-------------+
| __read_lock_failed | 34.96% | 34.91% | 29.14% |
| __write_lock_failed | 10.14% | 10.68% | 7.51% |
| mutex_spin_on_owner | 3.62% | 3.42% | 2.33% |
| mspin_lock | N/A | N/A | 9.90% |
| __mutex_lock_slowpath | 1.46% | 0.81% | 0.14% |
| _raw_spin_lock | 2.25% | 2.50% | 1.10% |
+-----------------------+-----------+---------+-------------+
The fserver workload for an 8-node system is dominated by the
contention in the read/write lock. Mutex contention also plays a
role. With the first patch only, mutex contention is down (as
shown by the __mutex_lock_slowpath figure) which help a little
bit. We saw only a few percents improvement with that.
By applying patch 2 as well, the single mutex_spin_on_owner
figure is now split out into an additional mspin_lock figure.
The time increases from 3.42% to 11.23%. It shows a great
reduction in contention among the spinners leading to a 30%
improvement. The time ratio 9.9/2.33=4.3 indicates that there
are on average 4+ spinners waiting in the spin_lock loop for
each spinner in the mutex_spin_on_owner loop. Contention in
other locking functions also go down by quite a lot.
The table below shows the performance change of both patches 1 &
2 over patch 1 alone in other AIM7 workloads (at 8 nodes,
hyperthreading off).
+--------------+---------------+----------------+-----------------+
| Workload | mean % change | mean % change | mean % change |
| | 10-100 users | 200-1000 users | 1100-2000 users |
+--------------+---------------+----------------+-----------------+
| alltests | 0.0% | -0.8% | +0.6% |
| five_sec | -0.3% | +0.8% | +0.8% |
| high_systime | +0.4% | +2.4% | +2.1% |
| new_fserver | +0.1% | +14.1% | +34.2% |
| shared | -0.5% | -0.3% | -0.4% |
| short | -1.7% | -9.8% | -8.3% |
+--------------+---------------+----------------+-----------------+
The short workload is the only one that shows a decline in
performance probably due to the spinner locking and queuing
overhead.
Signed-off-by: Waiman Long <Waiman.Long@hp.com>
Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Acked-by: Rik van Riel <riel@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Chandramouleeswaran Aswin <aswin@hp.com>
Cc: Norton Scott J <scott.norton@hp.com>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Clark Williams <williams@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1366226594-5506-4-git-send-email-Waiman.Long@hp.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
In the __mutex_lock_common() function, an initial entry into
the lock slow path will cause two atomic_xchg instructions to be
issued. Together with the atomic decrement in the fast path, a
total of three atomic read-modify-write instructions will be
issued in rapid succession. This can cause a lot of cache
bouncing when many tasks are trying to acquire the mutex at the
same time.
This patch will reduce the number of atomic_xchg instructions
used by checking the counter value first before issuing the
instruction. The atomic_read() function is just a simple memory
read. The atomic_xchg() function, on the other hand, can be up
to 2 order of magnitude or even more in cost when compared with
atomic_read(). By using atomic_read() to check the value first
before calling atomic_xchg(), we can avoid a lot of unnecessary
cache coherency traffic. The only downside with this change is
that a task on the slow path will have a tiny bit less chance of
getting the mutex when competing with another task in the fast
path.
The same is true for the atomic_cmpxchg() function in the
mutex-spin-on-owner loop. So an atomic_read() is also performed
before calling atomic_cmpxchg().
The mutex locking and unlocking code for the x86 architecture
can allow any negative number to be used in the mutex count to
indicate that some tasks are waiting for the mutex. I am not so
sure if that is the case for the other architectures. So the
default is to avoid atomic_xchg() if the count has already been
set to -1. For x86, the check is modified to include all
negative numbers to cover a larger case.
The following table shows the jobs per minutes (JPM) scalability
data on an 8-node 80-core Westmere box with a 3.7.10 kernel. The
numactl command is used to restrict the running of the
high_systime workloads to 1/2/4/8 nodes with hyperthreading on
and off.
+-----------------+-----------+------------+----------+
| Configuration | Mean JPM | Mean JPM | % Change |
| | w/o patch | with patch | |
+-----------------+-----------------------------------+
| | User Range 1100 - 2000 |
+-----------------+-----------------------------------+
| 8 nodes, HT on | 36980 | 148590 | +301.8% |
| 8 nodes, HT off | 42799 | 145011 | +238.8% |
| 4 nodes, HT on | 61318 | 118445 | +51.1% |
| 4 nodes, HT off | 158481 | 158592 | +0.1% |
| 2 nodes, HT on | 180602 | 173967 | -3.7% |
| 2 nodes, HT off | 198409 | 198073 | -0.2% |
| 1 node , HT on | 149042 | 147671 | -0.9% |
| 1 node , HT off | 126036 | 126533 | +0.4% |
+-----------------+-----------------------------------+
| | User Range 200 - 1000 |
+-----------------+-----------------------------------+
| 8 nodes, HT on | 41525 | 122349 | +194.6% |
| 8 nodes, HT off | 49866 | 124032 | +148.7% |
| 4 nodes, HT on | 66409 | 106984 | +61.1% |
| 4 nodes, HT off | 119880 | 130508 | +8.9% |
| 2 nodes, HT on | 138003 | 133948 | -2.9% |
| 2 nodes, HT off | 132792 | 131997 | -0.6% |
| 1 node , HT on | 116593 | 115859 | -0.6% |
| 1 node , HT off | 104499 | 104597 | +0.1% |
+-----------------+------------+-----------+----------+
At low user range 10-100, the JPM differences were within +/-1%.
So they are not that interesting.
AIM7 benchmark run has a pretty large run-to-run variance due to
random nature of the subtests executed. So a difference of less
than +-5% may not be really significant.
This patch improves high_systime workload performance at 4 nodes
and up by maintaining transaction rates without significant
drop-off at high node count. The patch has practically no
impact on 1 and 2 nodes system.
The table below shows the percentage time (as reported by perf
record -a -s -g) spent on the __mutex_lock_slowpath() function
by the high_systime workload at 1500 users for 2/4/8-node
configurations with hyperthreading off.
+---------------+-----------------+------------------+---------+
| Configuration | %Time w/o patch | %Time with patch | %Change |
+---------------+-----------------+------------------+---------+
| 8 nodes | 65.34% | 0.69% | -99% |
| 4 nodes | 8.70% | 1.02% | -88% |
| 2 nodes | 0.41% | 0.32% | -22% |
+---------------+-----------------+------------------+---------+
It is obvious that the dramatic performance improvement at 8
nodes was due to the drastic cut in the time spent within the
__mutex_lock_slowpath() function.
The table below show the improvements in other AIM7 workloads
(at 8 nodes, hyperthreading off).
+--------------+---------------+----------------+-----------------+
| Workload | mean % change | mean % change | mean % change |
| | 10-100 users | 200-1000 users | 1100-2000 users |
+--------------+---------------+----------------+-----------------+
| alltests | +0.6% | +104.2% | +185.9% |
| five_sec | +1.9% | +0.9% | +0.9% |
| fserver | +1.4% | -7.7% | +5.1% |
| new_fserver | -0.5% | +3.2% | +3.1% |
| shared | +13.1% | +146.1% | +181.5% |
| short | +7.4% | +5.0% | +4.2% |
+--------------+---------------+----------------+-----------------+
Signed-off-by: Waiman Long <Waiman.Long@hp.com>
Reviewed-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Reviewed-by: Rik van Riel <riel@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Chandramouleeswaran Aswin <aswin@hp.com>
Cc: Norton: Scott J <scott.norton@hp.com>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Clark Williams <williams@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1366226594-5506-3-git-send-email-Waiman.Long@hp.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
As mentioned by Ingo, the SCHED_FEAT_OWNER_SPIN scheduler
feature bit was really just an early hack to make with/without
mutex-spinning testable. So it is no longer necessary.
This patch removes the SCHED_FEAT_OWNER_SPIN feature bit and
move the mutex spinning code from kernel/sched/core.c back to
kernel/mutex.c which is where they should belong.
Signed-off-by: Waiman Long <Waiman.Long@hp.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Chandramouleeswaran Aswin <aswin@hp.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Norton Scott J <scott.norton@hp.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Clark Williams <williams@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1366226594-5506-2-git-send-email-Waiman.Long@hp.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Pull user-namespace fixes from Andy Lutomirski.
* 'userns-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/luto/linux:
userns: Changing any namespace id mappings should require privileges
userns: Check uid_map's opener's fsuid, not the current fsuid
userns: Don't let unprivileged users trick privileged users into setting the id_map
Pull x86 platform driver revert from Matthew Garrett:
"It turns out that one of the hp-wmi patches this cycle breaks some
other HP laptops. I think we have a good idea how to work on it for
3.10, but it's safer to just revert it for now."
* 'for_linus' of git://cavan.codon.org.uk/platform-drivers-x86:
Revert "hp-wmi: Add support for SMBus hotkeys"
Alex Efros reported rpfilter module doesn't match following packets:
IN=br.qemu SRC=192.168.2.1 DST=192.168.2.255 [ .. ]
(netfilter bugzilla #814).
Problem is that network stack arranges for the locally generated broadcasts
to appear on the interface they were sent out, so the IFF_LOOPBACK check
doesn't trigger.
As -m rpfilter is restricted to PREROUTING, we can check for existing
rtable instead, it catches locally-generated broad/multicast case, too.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
This reverts commit fabf85e3ca which breaks
hotkey support on some other HP laptops. We'll try doing this differently
in 3.10.
Signed-off-by: Matthew Garrett <matthew.garrett@nebula.com>
The type when timeout support was enabled, could not list all elements,
just the first ones which could fit into one netlink message: it just
did not continue listing after the first message.
Reported-by: Yoann JUET <yoann.juet@univ-nantes.fr>
Signed-off-by: Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>
Tested-by: Yoann JUET <yoann.juet@univ-nantes.fr>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Since commit 6b923cb718 (bonding: support for IPv6 transmit hashing)
bonding doesn't properly hash traffic in forwarding setups.
Vitaly V. Bursov diagnosed that skb_network_header_len() returned 0 in
this case.
More generally, the transport header might not be in the skb head.
Use pskb_may_pull() & skb_header_pointer() to get it right, and use
proto_ports_offset() in bond_xmit_hash_policy_l34() to get support for
more protocols than TCP and UDP.
Reported-by: Vitaly V. Bursov <vitalyb@telenet.dn.ua>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Jay Vosburgh <fubar@us.ibm.com>
Cc: Andy Gospodarek <andy@greyhouse.net>
Cc: John Eaglesham <linux@8192.net>
Tested-by: Vitaly V. Bursov <vitalyb@telenet.dn.ua>
Signed-off-by: David S. Miller <davem@davemloft.net>
This fixes 2 issues regarding bnx2x's status blocks:
1. ethtool -c caused corruption of status blocks in FW RAM.
2. when using multi-CoS, the configuration of the timeout values of
status blocks is incorrect, harming the coalescing of interrupts
for such CoSs.
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When loading after UNDI (e.g., Boot from SAN) the UNDI does not
gracefully yield its resources; The bnx2x driver handles that release
itself.
During the manipulation required to release those resources, it's possible
for the UNDI to try and write to memory regions which are no longer accessible,
causing the PCI bus to prevent further writes from the chip.
This would in turn cause DMAE timeouts later on in the driver, as the driver
will be unable to use the chip's DMA engines.
This patch prevents the chip from actually writing through the PCI bus
in said scenario, thus allowing the release without the unfortunate by-product.
Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Shahed Shaikh says:
====================
This patch series contains bug fixes for -
* Loopback test failure while traffic is running.
* Tx timeout and subsequent firmware reset by removing check for
'(adapter->netdev->features & (NETIF_F_TSO | NETIF_F_TSO6)' from tx fast
path, as per Eric's suggestion.
* Typo in logs.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
o Debug logs were not matching with code functionality.
o Changed dev_info to netdev_err
Signed-off-by: Shahed Shaikh <shahed.shaikh@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When driver receives a packet with gso size > 0 and when TSO is disabled,
it should be transmitted as a TSO packet to prevent Tx timeout and subsequent
firmware reset.
Signed-off-by: Sritej Velaga <sritej.velaga@qlogic.com>
Signed-off-by: Shahed Shaikh <shahed.shaikh@qlogic.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Before conducting loopback test by sending packets, driver should stop transmit
queue and turn off carrier.
Signed-off-by: Jitendra Kalsaria <jitendra.kalsaria@qlogic.com>
Signed-off-by: Shahed Shaikh <shahed.shaikh@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The PF driver does not check if the administrator has already set a VF
VLAN via the PF driver before setting the new VLAN. This results in
the following scenario:
A) Administrator sets VF <n> to VLAN 100
B) Administrator sets VF <x> to VLAN 100
C) Administrator sets VF <n> to VLAN 200
D) The VF <n> driver continues to be able to receive traffic on VLAN
100 because the VLVFB pool enable bit for that VF was left set
instead of being cleared as it should be.
This fix ensures that the old VLAN filter for VF <n> is first removed
and the pool bit enable for VF <n> is cleared so that it no longer
receives traffic on VLAN 100.
Signed-off-by: Greg Rose <gregory.v.rose@intel.com>
Tested-by: Sibai Li <sibai.li@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch actually reverts:
igb: Support using build_skb in the case that jumbo frames are disabled
The reason for reverting this patch is that it can lead to data corruption.
The following flow was pointed out by Ben Hutchings:
1. skb is forwarded to another device
2. Packet headers are modified and it's put into a queue
3. Second packet is received into the other half of this page
4. Page cannot be reused, so is DMA-unmapped
5. The DMA mapping was non-coherent, so unmap copies or invalidates
cache
The headers added in step 2 get trashed in step 5.
Reported-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
- fix MAC address check in case of multiple mesh interfaces
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.19 (GNU/Linux)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=oL+V
-----END PGP SIGNATURE-----
Merge tag 'batman-adv-fix-for-davem' of git://git.open-mesh.org/linux-merge
Included changes:
- fix MAC address check in case of multiple mesh interfaces
Signed-off-by: David S. Miller <davem@davemloft.net>
Previously we used PCI_MSI_FLAGS to locate a register in the MSI-X
capability. This did work because the MSI and MSI-X flags happen
to be at the same offsets, but was confusing.
PCI_MSIX_FLAGS_ENABLE is already defined in include/uapi/linux/pci_regs.h,
so no need to define it again.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Adam Radford <aradford@gmail.com>
The L1 data cache flush needs to be after highbank_set_cpu_jump call which
pollutes the cache with the l2x0_lock. This causes other cores to deadlock
waiting for the l2x0_lock. Moving the flush of the entire data cache after
highbank_set_cpu_jump fixes the problem. Use flush_cache_louis instead of
flush_cache_all are that is sufficient to flush only the L1 data cache.
flush_cache_louis did not exist when highbank_cpu_die was originally
written.
With PL310 errata 769419 enabled, a wmb is inserted into idle which takes
the l2x0_lock. This makes the problem much more easily hit and causes
reset to hang.
Reported-by: Paolo Pisati <p.pisati@gmail.com>
Signed-off-by: Rob Herring <rob.herring@calxeda.com>
Signed-off-by: Olof Johansson <olof@lixom.net>
This reverts commit 3a366e614d.
Wanlong Gao reports that it causes a kernel panic on his machine several
minutes after boot. Reverting it removes the panic.
Jens says:
"It's not quite clear why that is yet, so I think we should just revert
the commit for 3.9 final (which I'm assuming is pretty close).
The wifi is crap at the LSF hotel, so sending this email instead of
queueing up a revert and pull request."
Reported-by: Wanlong Gao <gaowanlong@cn.fujitsu.com>
Requested-by: Jens Axboe <axboe@kernel.dk>
Cc: Tejun Heo <tj@kernel.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Install the Hyper-V specific interrupt handler only when needed. This would
permit us to get rid of the Xen check. Note that when the vmbus drivers invokes
the call to register its handler, we are sure to be running on Hyper-V.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Link: http://lkml.kernel.org/r/1366299886-6399-1-git-send-email-kys@microsoft.com
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Fix a double locking bug caused when debug.kprobe-optimization=0.
While the proc_kprobes_optimization_handler locks kprobe_mutex,
wait_for_kprobe_optimizer locks it again and that causes a double lock.
To fix the bug, this introduces different mutex for protecting
sysctl parameter and locks it in proc_kprobes_optimization_handler.
Of course, since we need to lock kprobe_mutex when touching kprobes
resources, that is done in *optimize_all_kprobes().
This bug was introduced by commit ad72b3bea7 ("kprobes: fix
wait_for_kprobe_optimizer()")
Signed-off-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Acked-by: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The BUG_ON() directive is triggered probably due to a latency
modification following inclusion of commit c10d73671a ("softirq:
reduce latencies"). This condition has not been met before 3.9-rc1 and
doesn't trigger without this patch.
We now make sure that DMA channel is idle before calling
atc_complete_all() which makes the BUG_ON() "protection" useless.
Signed-off-by: Ludovic Desroches <ludovic.desroches@atmel.com>
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Acked-by: Vinod Koul <vinod.koul@intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Fix an issue when probing pinctrl-abx500 with no
platform data.
Signed-off-by: Patrice Chotard <patrice.chotard@st.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Jesse Gross says:
====================
Two small bug fixes for net/3.9 including the issue previously
discussed where allocation of netlink notifications can fail after
changes have been committed.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull idle patches from Len Brown:
"A pair of small patches for 3.9-rc7.
This CPU-id should have been included in the ones that we updated
earlier in 3.9. This pair of patches will allow this flavor of
Haswell to behave like the other flavors."
* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux:
tools/power turbostat: additional Haswell CPU-id
intel_idle: additional Haswell CPU-id
This fixes a kernel memory contents leak via the tkill and tgkill syscalls
for compat processes.
This is visible in the siginfo_t->_sifields._rt.si_sigval.sival_ptr field
when handling signals delivered from tkill.
The place of the infoleak:
int copy_siginfo_to_user32(compat_siginfo_t __user *to, siginfo_t *from)
{
...
put_user_ex(ptr_to_compat(from->si_ptr), &to->si_ptr);
...
}
Signed-off-by: Emese Revfy <re.emese@gmail.com>
Reviewed-by: PaX Team <pageexec@freemail.hu>
Signed-off-by: Kees Cook <keescook@chromium.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Serge Hallyn <serge.hallyn@canonical.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Fix the error return value in kswapd_run(). The bug was introduced by
commit d5dc0ad928 ("mm/vmscan: fix error number for failed kthread").
Signed-off-by: Xishi Qiu <qiuxishi@huawei.com>
Reviewed-by: Wanpeng Li <liwanp@linux.vnet.ibm.com>
Reviewed-by: Rik van Riel <riel@redhat.com>
Reported-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
fixed the following compile error when use avr32 atstk1006_defconfig:
drivers/mtd/nand/atmel_nand.c: In function 'pmecc_err_location':
drivers/mtd/nand/atmel_nand.c:639: error: implicit declaration of function 'writel_relaxed'
which was introduced by commit 1c7b874d33 ("mtd: at91: atmel_nand: add
Programmable Multibit ECC controller support"). The PMECC for nand
flash code uses writel_relaxed(). But in avr32, there is no macro
"writel_relaxed" defined.
This patch add writex_relaxed macro definitions.
Signed-off-by: Josh Wu <josh.wu@atmel.com>
Acked-by: Havard Skinnemoen <havard@skinnemoen.net>
Acked-by: Hans-Christian Egtvedt <egtvedt@samfundet.no>
Cc: David Woodhouse <David.Woodhouse@intel.com>
Cc: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
With applying the previous patch "hugetlbfs: stop setting VM_DONTDUMP in
initializing vma(VM_HUGETLB)" to reenable hugepage coredump, if a memory
error happens on a hugepage and the affected processes try to access the
error hugepage, we hit VM_BUG_ON(atomic_read(&page->_count) <= 0) in
get_page().
The reason for this bug is that coredump-related code doesn't recognise
"hugepage hwpoison entry" with which a pmd entry is replaced when a memory
error occurs on a hugepage.
In other words, physical address information is stored in different bit
layout between hugepage hwpoison entry and pmd entry, so
follow_hugetlb_page() which is called in get_dump_page() returns a wrong
page from a given address.
The expected behavior is like this:
absent is_swap_pte FOLL_DUMP Expected behavior
-------------------------------------------------------------------
true false false hugetlb_fault
false true false hugetlb_fault
false false false return page
true false true skip page (to avoid allocation)
false true true hugetlb_fault
false false true return page
With this patch, we can call hugetlb_fault() and take proper actions (we
wait for migration entries, fail with VM_FAULT_HWPOISON_LARGE for
hwpoisoned entries,) and as the result we can dump all hugepages except
for hwpoisoned ones.
Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Rik van Riel <riel@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Cc: HATAYAMA Daisuke <d.hatayama@jp.fujitsu.com>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: <stable@vger.kernel.org> [2.6.34+?]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Documentation/filesystems/proc.txt says about coredump_filter bitmask,
Note bit 0-4 doesn't effect any hugetlb memory. hugetlb memory are only
effected by bit 5-6.
However current code can go into the subsequent flag checks of bit 0-4
for vma(VM_HUGETLB). So this patch inserts 'return' and makes it work
as written in the document.
Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Reviewed-by: Rik van Riel <riel@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Reviewed-by: HATAYAMA Daisuke <d.hatayama@jp.fujitsu.com>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: <stable@vger.kernel.org> [3.7+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Currently we fail to include any data on hugepages into coredump,
because VM_DONTDUMP is set on hugetlbfs's vma. This behavior was
recently introduced by commit 314e51b985 ("mm: kill vma flag
VM_RESERVED and mm->reserved_vm counter").
This looks to me a serious regression, so let's fix it.
Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Acked-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
Acked-by: Michal Hocko <mhocko@suse.cz>
Reviewed-by: Rik van Riel <riel@redhat.com>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: <stable@vger.kernel.org> [3.7+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Fix checkpatch misreporting defect with stringification macros
ERROR: Macros with complex values should be enclosed in parenthesis
#27: FILE: arch/arm/include/asm/kgdb.h:41:
+#define ___to_string(X) #X
Signed-off-by: Joe Perches <joe@perches.com>
Reported-by: Vincent Stehlé <v-stehle@ti.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Found with coccinelle.
Signed-off-by: Andrei Epure <epure.andrei@gmail.com>
Cc: Florian Tobias Schandinat <FlorianSchandinat@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
hotkey_kthread() does try_to_freeze() under hotkey_thread_mutex.
We can simply kill this mutex, hotkey_poll_stop_sync() does not need to
serialize with hotkey_kthread(). When kthread_stop() returns the thread
is already dead, it called do_exit()->complete_vfork_done().
Reported-by: Artem Savkov <artem.savkov@gmail.com>
Reported-by: Maciej Rutecki <maciej.rutecki@gmail.com>
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
Cc: Matthew Garrett <matthew.garrett@nebula.com>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Reviewed-by: Mandeep Singh Baines <msb@chromium.org>
Cc: Aaron Lu <aaron.lu@intel.com>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
On a multi-mesh node (a node running more than one batman-adv
virtual interface) batadv_is_my_mac() has to check MAC
addresses of hard interfaces belonging to the current mesh
only.
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Pull Ceph fix from Sage Weil:
"It's a simple fix for a hard to hit race, but low-risk and clearly
correct"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client:
rbd: do a safe list traversal in rbd_img_request_submit()
In the very unlikely event where a guest would be foolish enough to
*read* from a write-only cache maintainance register, we end up
with preemption disabled, due to a misplaced get_cpu().
Just move the "is_write" test outside of the critical section.
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Christoffer Dall <cdall@cs.columbia.edu>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We can extend kexec-tools to support multiple "Crash kernel" in /proc/iomem
instead.
So we can use "Crash kernel" instead of "Crash kernel low" in /proc/iomem.
Suggested-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Link: http://lkml.kernel.org/r/1366089828-19692-3-git-send-email-yinghai@kernel.org
Acked-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Per hpa, use crashkernel=X,high crashkernel=Y,low instead of
crashkernel_hign=X crashkernel_low=Y. As that could be extensible.
-v2: according to Vivek, change delimiter to ;
-v3: let hign and low only handle simple form and it conforms to
description in kernel-parameters.txt
still keep crashkernel=X override any crashkernel=X,high
crashkernel=Y,low
-v4: update get_last_crashkernel returning and add more strict
checking in parse_crashkernel_simple() found by HATAYAMA.
-v5: Change delimiter back to , according to HPA.
also separate parse_suffix from parse_simper according to vivek.
so we can avoid @pos in that path.
-v6: Tight the checking about crashkernel=X,highblahblah,high
found by HTYAYAMA.
Cc: HATAYAMA Daisuke <d.hatayama@jp.fujitsu.com>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Link: http://lkml.kernel.org/r/1366089828-19692-5-git-send-email-yinghai@kernel.org
Acked-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Vivek found old kexec-tools does not work new kernel anymore.
So change back crashkernel= back to old behavoir, and add crashkernel_high=
to let user decide if buffer could be above 4G, and also new kexec-tools will
be needed.
-v2: let crashkernel=X override crashkernel_high=
update description about _high will be ignored by crashkernel=X
-v3: update description about kernel-parameters.txt according to Vivek.
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Link: http://lkml.kernel.org/r/1366089828-19692-4-git-send-email-yinghai@kernel.org
Acked-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Chao said that kdump does does work well on his system on 3.8
without extra parameter, even iommu does not work with kdump.
And now have to append crashkernel_low=Y in first kernel to make
kdump work.
We have now modified crashkernel=X to allocate memory beyong 4G (if
available) and do not allocate low range for crashkernel if the user
does not specify that with crashkernel_low=Y. This causes regression
if iommu is not enabled. Without iommu, swiotlb needs to be setup in
first 4G and there is no low memory available to second kernel.
Set crashkernel_low automatically if the user does not specify that.
For system that does support IOMMU with kdump properly, user could
specify crashkernel_low=0 to save that 72M low ram.
-v3: add swiotlb_size() according to Konrad.
-v4: add comments what 8M is for according to hpa.
also update more crashkernel_low= in kernel-parameters.txt
-v5: update changelog according to Vivek.
-v6: Change description about swiotlb referring according to HATAYAMA.
Reported-by: WANG Chao <chaowang@redhat.com>
Tested-by: WANG Chao <chaowang@redhat.com>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Link: http://lkml.kernel.org/r/1366089828-19692-2-git-send-email-yinghai@kernel.org
Acked-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
It's possible that the reference to the object request dropped
inside the loop in rbd_img_request_submit() will be the last
one, in which case the content of the object pointer can't be
trusted.
Use a safe form of the object request list traversal to avoid
problems.
This resolves:
http://tracker.ceph.com/issues/4705
Signed-off-by: Alex Elder <elder@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
The latest r1p5-revision of the ARM PL011 UART has 32-byte FIFOs,
while all earlier ones have 16-byte FIFOs. This patch suggests
a way to set the FIFO-size correctly & flexibly by using a member
function named get_fifosize, rather than using the fifosize member
variable. The function takes the UARTPeriphID, and returns the
correct FIFO size.
Signed-off-by: Jongsung Kim <neidhard.kim@lge.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The comedi_subdevice 'private' variable is a void * that is available
for the subdevice to use in manner. It's common in comedi drivers for
the driver to allocate memory for a subdevice and store the pointer
to that memory in the 'private' variable. It's then the responsibility
of the driver to free that memory when the device is detached.
Due to how the attach/detach works in comedi, the drivers need to do
some sanity checking before they can free the allocated memory during
the detach.
Introduce a helper function, comedi_spriv_free(), to handle freeing
the private data allocated for a subdevice. This allows moving all the
sanity checks into the helper function and makes it safe to call
with any context. It also allows removing some of the boilerplate
code in the (*detach) functions.
Remove the subdev_8255_cleanup() export in the 8255 subdevice driver
as well as the addi_watchdog_cleanup() export in the addi_watchdog
driver and use the new helper instead.
The amplc_dio200_common driver uses a number of local helper functions
to free the private data for it's subdevices. Remove those as well and
use the new helper.
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
