While testing the performance of different receive interrupt
coalescing settings on a single stream TCP benchmark, I noticed two
very different results. With rx-usecs=50, most of the time a
connection would hit 8280 Mbps but once in a while it would hit
9330 Mbps.
It turns out we are only applying the interrupt coalescing settings
to the first queue and whenever the rx hash would direct us onto
that queue we ran faster.
With this patch applied and rx-usecs=50, I get 9330 Mbps
consistently.
Signed-off-by: Anton Blanchard <anton@samba.org>
Acked-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When analysing performance of the cxgb3 on a ppc64 box I noticed that
we weren't doing much GRO merging. It turns out we are limited by the
number of SKB frags:
#define MAX_SKB_FRAGS (65536/PAGE_SIZE + 2)
With a 4kB page size we have 18 frags, but with a 64kB page size we
only have 3 frags.
I ran a single stream TCP bandwidth test to compare the performance of
Signed-off-by: David S. Miller <davem@davemloft.net>
We incorrectly returned -EINVAL when none of the devices in the array
had an integrity profile. This in turn prevented mdadm from starting
the metadevice. Fix this so we only return errors on mismatched
profiles and memory allocation failures.
Reported-by: Giacomo Catenazzi <cate@cateee.net>
Reported-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
MFD changes in 4ec1b54c ('mfd: mfd_cell is now implicitly available to
mc13xxx drivers') changed the mc13xxx_platform_data struct layout.
At the time all users were changed, but this driver was introduced in
another tree at the same time. This updates the mc13xxx_platform_data
user, fixing a build error.
Signed-off-by: Andres Salomon <dilinger@queued.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'irq-cleanup-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
vlynq: Convert irq functions
* 'irq-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
genirq; Fix cleanup fallout
genirq: Fix typo and remove unused variable
genirq: Fix new kernel-doc warnings
genirq: Add setter for AFFINITY_SET in irq_data state
genirq: Provide setter inline for IRQD_IRQ_INPROGRESS
genirq: Remove handle_IRQ_event
arm: Ns9xxx: Remove private irq flow handler
powerpc: cell: Use the core flow handler
genirq: Provide edge_eoi flow handler
genirq: Move INPROGRESS, MASKED and DISABLED state flags to irq_data
genirq: Split irq_set_affinity() so it can be called with lock held.
genirq: Add chip flag for restricting cpu_on/offline calls
genirq: Add chip hooks for taking CPUs on/off line.
genirq: Add irq disabled flag to irq_data state
genirq: Reserve the irq when calling irq_set_chip()
I missed the CONFIG_GENERIC_PENDING_IRQ dependency in the affinity
related functions and the IRQ_LEVEL propagation into irq_data
state. Did not pop up on my main test platforms. :(
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: David Daney <ddaney@caviumnetworks.com>
Commit 6f5317e730 introduced a bug in the
handling of userspace object classes that is causing breakage for Xorg
when XSELinux is enabled. Fix the bug by changing map_class() to return
SECCLASS_NULL when the class cannot be mapped to a kernel object class.
Reported-by: "Justin P. Mattock" <justinmattock@gmail.com>
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
Signed-off-by: James Morris <jmorris@namei.org>
Fixes this build error:
drivers/memstick/host/r592.c:26: error: 'enable_dma' redeclared as different kind of symbol
arch/powerpc/include/asm/dma.h:189: note: previous definition of 'enable_dma' was here
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Alex Dubov <oakad@yahoo.com>
Acked-by: Maxim Levitsky <maximlevitsky@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'for-linus' of git://oss.sgi.com/xfs/xfs:
xfs: stop using the page cache to back the buffer cache
xfs: register the inode cache shrinker before quotachecks
xfs: xfs_trans_read_buf() should return an error on failure
xfs: introduce inode cluster buffer trylocks for xfs_iflush
vmap: flush vmap aliases when mapping fails
xfs: preallocation transactions do not need to be synchronous
Fix up trivial conflicts in fs/xfs/linux-2.6/xfs_buf.c due to plug removal.
Commit da48524eb2 ("Prevent rt_sigqueueinfo and rt_tgsigqueueinfo
from spoofing the signal code") made the check on si_code too strict.
There are several legitimate places where glibc wants to queue a
negative si_code different from SI_QUEUE:
- This was first noticed with glibc's aio implementation, which wants
to queue a signal with si_code SI_ASYNCIO; the current kernel
causes glibc's tst-aio4 test to fail because rt_sigqueueinfo()
fails with EPERM.
- Further examination of the glibc source shows that getaddrinfo_a()
wants to use SI_ASYNCNL (which the kernel does not even define).
The timer_create() fallback code wants to queue signals with SI_TIMER.
As suggested by Oleg Nesterov <oleg@redhat.com>, loosen the check to
forbid only the problematic SI_TKILL case.
Reported-by: Klaus Dittrich <kladit@arcor.de>
Acked-by: Julien Tinnes <jln@google.com>
Cc: <stable@kernel.org>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'for-linus' of git://www.jni.nu/cris:
Correct auto-restart of syscalls via restartblock
CRISv10: Fix return before mutex_unlock in pcf8563
Drop the CRISv32 version of pcf8563
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ecryptfs/ecryptfs-2.6:
eCryptfs: write lock requested keys
eCryptfs: move ecryptfs_find_auth_tok_for_sig() call before mutex_lock
eCryptfs: verify authentication tokens before their use
eCryptfs: modified size of keysig in the ecryptfs_key_sig structure
eCryptfs: removed num_global_auth_toks from ecryptfs_mount_crypt_stat
eCryptfs: ecryptfs_keyring_auth_tok_for_sig() bug fix
eCryptfs: Unlock page in write_begin error path
ecryptfs: modify write path to encrypt page in writepage
eCryptfs: Remove ECRYPTFS_NEW_FILE crypt stat flag
eCryptfs: Remove unnecessary grow_file() function
* 'for-linus-unmerged' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable: (45 commits)
Btrfs: fix __btrfs_map_block on 32 bit machines
btrfs: fix possible deadlock by clearing __GFP_FS flag
btrfs: check link counter overflow in link(2)
btrfs: don't mess with i_nlink of unlocked inode in rename()
Btrfs: check return value of btrfs_alloc_path()
Btrfs: fix OOPS of empty filesystem after balance
Btrfs: fix memory leak of empty filesystem after balance
Btrfs: fix return value of setflags ioctl
Btrfs: fix uncheck memory allocations
btrfs: make inode ref log recovery faster
Btrfs: add btrfs_trim_fs() to handle FITRIM
Btrfs: adjust btrfs_discard_extent() return errors and trimmed bytes
Btrfs: make btrfs_map_block() return entire free extent for each device of RAID0/1/10/DUP
Btrfs: make update_reserved_bytes() public
btrfs: return EXDEV when linking from different subvolumes
Btrfs: Per file/directory controls for COW and compression
Btrfs: add datacow flag in inode flag
btrfs: use GFP_NOFS instead of GFP_KERNEL
Btrfs: check return value of read_tree_block()
btrfs: properly access unaligned checksum buffer
...
Fix up trivial conflicts in fs/btrfs/volumes.c due to plug removal in
the block layer.
* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mjg59/platform-drivers-x86: (81 commits)
xo15-ebook: Remove device.wakeup_count
ips: use interruptible waits in ips-monitor
acer-wmi: does not poll device status when WMI event is available
acer-wmi: does not set persistence state by rfkill_init_sw_state
platform-drivers: x86: fix common misspellings
acer-wmi: use pr_<level> for messages
asus-wmi: potential NULL dereference in show_call()
asus-wmi: signedness bug in read_brightness()
platform-driver-x86: samsung-laptop: make dmi_check_cb to return 1 instead of 0
platform-driver-x86: fix wrong merge for compal-laptop.c
msi-laptop: use pr_<level> for messages
Platform: add Samsung Laptop platform driver
acer-wmi: Fix WMI ID
acer-wmi: deactive mail led when power off
msi-laptop: send out touchpad on/off key
acer-wmi: set the touchpad toggle key code to KEY_TOUCHPAD_TOGGLE
platform-driver-x86: intel_mid_thermal: fix unterminated platform_device_id table
sony-laptop: potential null dereference
sony-laptop: handle allocation failures
sony-laptop: return negative on failure in sony_nc_add()
...
* 'for-torvalds' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-stericsson:
mach-ux500: configure board for the TPS61052 regulator v2
mach-ux500: provide ab8500 init vector
mach-ux500: board support for AB8500 GPIO driver
gpio: driver for 42 AB8500 GPIO pins
* 'for-linus' of git://gitserver.sunplusct.com/linux-2.6-score:
score: Use generic show_interrupts()
score: Convert to new irq function names
score: lost a semicolon in asm/irqflags.h
score: Select GENERIC_HARDIRQS_NO_DEPRECATED
score: Convert irq_chip to new functions
The acpi video driver attempts to explicitly create a sysfs link between
the acpi device and the associated PCI device. However, we're now also
doing this from the backlight core, which means that we get a backtrace
caused by a duplicate file. Remove the code and leave it up to the
backlight core.
Reported-by: Jeff Chua <jeff.chua.linux@gmail.com>
Signed-off-by: Matthew Garrett <mjg@redhat.com>
Acked-by: Rafael J. Wysocki <rjw@sisk.pl>
Tested-by: Alessandro Suardi <alessandro.suardi@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Fix new irq-related kernel-doc warnings in 2.6.38:
Warning(kernel/irq/manage.c:149): No description found for parameter 'mask'
Warning(kernel/irq/manage.c:149): Excess function parameter 'cpumask' description in 'irq_set_affinity'
Warning(include/linux/irq.h:161): No description found for parameter 'state_use_accessors'
Warning(include/linux/irq.h:161): Excess struct/union/enum/typedef member 'state_use_accessor' description in 'irq_data'
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
LKML-Reference: <20110318093356.b939558d.randy.dunlap@oracle.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Convert to the new irq_chip functions and the new namespace.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Florian Fainelli <florian@openwrt.org>
LKML-Reference: <alpine.LFD.2.00.1103252150180.31464@localhost6.localdomain6>
When a hole spans across page boundaries, the next write forces
a read of the block. This could end up reading existing garbage
data from the disk in ocfs2_map_page_blocks. This leads to
non-zero holes. In order to avoid this, mark the writes as new
when the holes span across page boundaries.
Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.de>
Signed-off-by: jlbec <jlbec@evilplan.org>
When CONFIG_DEBUG_FS=y and CONFIG_OCFS2_FS_STATS=n, we get the
following warning:
fs/ocfs2/cluster/tcp.c:213:16: warning: ‘o2net_get_func_run_time’
defined but not used
Since o2net_get_func_run_time is only called from
o2net_update_recv_stats, so move it under CONFIG_OCFS2_FS_STATS.
Signed-off-by: Rakib Mullick <rakib.mullick@gmail.com>
Signed-off-by: jlbec <jlbec@evilplan.org>
Some archs want to prevent the default affinity being set on their
chips in the reqeust_irq() path.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
handle_prio_irq is almost identical with handle_fasteoi_irq. The
subtle differences are
1) The handler checks for IRQ_DISABLED after the device handler has
been called. In case it's set it masks the interrupt.
2) When the handler sees IRQ_DISABLED on entry it masks the interupt
in the same way as handle_fastoei_irq, but does not set the
IRQ_PENDING flag.
3) Instead of gracefully handling a recursive interrupt it crashes the
kernel.
#1 is just relevant when a device handler calls disable_irq_nosync()
and it does not matter whether we mask the interrupt right away or
not. We handle lazy masking for disable_irq anyway, so there is no
real reason to have this extra mask in place.
#2 will prevent the resend of a pending interrupt, which can result in
lost interrupts for edge type interrupts. For level type interrupts
the resend is a noop in the generic code. According to the
datasheet all interrupts are level type, so marking them as such
will result in the exact same behaviour as the private
handle_prio_irq implementation.
#3 is just stupid. Crashing the kernel instead of handling a problem
gracefully is just wrong. With the current semantics- all handlers
run with interrupts disabled - this is even more wrong.
Rename ack to eoi, remove the unused mask_ack, switch to
handle_fasteoi_irq and remove the private function.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Acked-by: Uwe Kleine-Koenig <u.kleine-koenig@pengutronix.de>
Cc: linux-arm-kernel@lists.infradead.org
LKML-Reference: <20110202212552.299898447@linutronix.de>
This is a replacment for the cell flow handler which is in the way of
cleanups. Must be selected to avoid general bloat.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
We really need these flags for some of the interrupt chips. Move it
from internal state to irq_data and provide proper accessors.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: David Daney <ddaney@caviumnetworks.com>
* 'bugfixes' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6:
NFS: Ensure that rpc_release_resources_task() can be called twice.
NFS: Don't leak RPC clients in NFSv4 secinfo negotiation
NFS: Fix a hang in the writeback path
Stephen ran into the following build error:
drivers/mfd/cs5535-mfd.c:30:22: error: asm/olpc.h: No such file or directory
olpc.h exists only on x86 (and in the future, ARM). Rather than
wrapping the include in an #ifdef, just change cs5535-mfd to only build
on x86.
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andres Salomon <dilinger@queued.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In commit 95a0f10cdd ("drbd: store in-core bitmap little endian,
regardless of architecture") drbd had made the sane choice to use
little-endian bitmap functions everywhere. However, it used the
horrible old functions names from <asm-generic/bitops/le.h>, that were
never really meant to be exported.
In the meantime, things got cleaned up, and in commit c4945b9ed4
("asm-generic: rename generic little-endian bitops functions") we
renamed the LE bitops to something sane, exactly so that they could be
used in random code without people gouging their eyes out when seeing
the crazy jumble of letters that were the old internal names.
As a result the drbd thing merged cleanly (commit 8d49a77568d1: "Merge
branch 'for-2.6.39/drivers' of git://git.kernel.dk/linux-2.6-block"),
since there was no data conflict - but the end result obviously doesn't
actually compile.
Reported-and-tested-by: Ingo Molnar <mingo@elte.hu>
Cc: Jens Axboe <jaxboe@fusionio.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Fix build breakage on platforms, not providing readsw and writesw
functions, e.g., on x86(_64).
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: Chris Ball <cjb@laptop.org>
The percpu code requires more functions to be implemented in the mm core
which nommu currently does not provide. So add inline implementations
since these are largely meaningless on nommu systems.
Signed-off-by: Graf Yang <graf.yang@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: David Howells <dhowells@redhat.com>
Acked-by: Greg Ungerer <gerg@uclinux.org>
per_cpu_ptr_to_phys() uses VMALLOC_START and VMALLOC_END to determine if an
address is in the vmalloc() region or not. This is incorrect on NOMMU as
there is no real vmalloc() capability (vmalloc() is emulated by kmalloc()).
The correct way to do this is to use is_vmalloc_addr(). This encapsulates the
vmalloc() region test in MMU mode and just returns 0 in NOMMU mode.
On FRV in NOMMU mode, the percpu compilation fails without this patch:
mm/percpu.c: In function 'per_cpu_ptr_to_phys':
mm/percpu.c:1011: error: 'VMALLOC_START' undeclared (first use in this function)
mm/percpu.c:1011: error: (Each undeclared identifier is reported only once
mm/percpu.c:1011: error: for each function it appears in.)
mm/percpu.c:1012: error: 'VMALLOC_END' undeclared (first use in this function)
mm/percpu.c:1018: warning: control reaches end of non-void function
Signed-off-by: David Howells <dhowells@redhat.com>
This is what I intended to do since:
1) the driver handles variable waits just fine, and
2) interruptible waits aren't reported as load in the load avg.
Reported-and-tested-by: Andreas Hartmann <andihartmann@freenet.de>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Matthew Garrett <mjg@redhat.com>
Acer WMI hotkey event's result include current device status, just
need sync the status to killswitch after acer-wmi driver receive
hotkey event but not always poll device status. This is good for
performance.
But, if use EC raw mode, Acer BIOS will not emit wmi event and
leave EC to control device status. So, still startup polling job
when doesn't detect WMI event GUID or user choice to use ec_raw_mode.
Tested on Acer TravelMate 8572 notebook.
Cc: Carlos Corbacho <carlos@strangeworlds.co.uk>
Cc: Matthew Garrett <mjg@redhat.com>
Cc: Dmitry Torokhov <dtor@mail.ru>
Cc: Corentin Chary <corentincj@iksaif.net>
Cc: Thomas Renninger <trenn@suse.de>
Signed-off-by: Lee, Chun-Yi <jlee@novell.com>
Signed-off-by: Matthew Garrett <mjg@redhat.com>
Acer BIOS keeps devices state when system reboot, but reset to default
device states (Wlan on, Bluetooth off, wwan on) if system cold boot.
That means BIOS's initial state is not always real persistence.
So, removed rfkill_init_sw_state because it sets initial state to
persistence then replicate to other new killswitch when rfkill-input
enabled.
After removed it, acer-wmi set initial soft-block state after rfkill
register, and doesn't allow set_block until rfkill initial finished.
Reference: bko#31002
https://bugzilla.kernel.org/show_bug.cgi?id=31002
Cc: Carlos Corbacho <carlos@strangeworlds.co.uk>
Cc: Matthew Garrett <mjg@redhat.com>
Cc: Dmitry Torokhov <dtor@mail.ru>
Cc: Corentin Chary <corentincj@iksaif.net>
Cc: OldÅich JedliÄka <oldium.pro@seznam.cz>
Cc: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Lee, Chun-Yi <jlee@novell.com>
Signed-off-by: Matthew Garrett <mjg@redhat.com>
In the earlier check we assumed that "obj" could be NULL. I looked at
some of the other places that call evaluate_object() and they check
for NULL as well.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Matthew Garrett <mjg@redhat.com>
"err" needs to be signed for the error handling to work.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Matthew Garrett <mjg@redhat.com>
Anton Blanchard