Previous size thresholds were guessed from various user space benchmarks
using a kernel with and without the alternative uaccess option. This
is however not as precise as a kernel based test to measure the real
speed of each method.
This adds a simple test bench to show the time needed for each method.
With this, the optimal size treshold for the alternative implementation
can be determined with more confidence. It appears that the optimal
threshold for both copy_to_user and clear_user is around 64 bytes. This
is not a surprise knowing that the memcpy and memset implementations
need at least 64 bytes to achieve maximum throughput.
One might suggest that such test be used to determine the optimal
threshold at run time instead, but results are near enough to 64 on
tested targets concerned by this alternative copy_to_user implementation,
so adding some overhead associated with a variable threshold is probably
not worth it for now.
Signed-off-by: Nicolas Pitre <nico@marvell.com>
Because the alternate copy_to_user implementation has a higher setup cost
than the standard implementation, the size of the memory area to copy
is tested and the standard implementation invoked instead when that size
is too small. Still, that test is made after the processor has preserved
a bunch of registers on the stack which have to be reloaded right away
needlessly in that case, causing a measurable performance regression
compared to plain usage of the standard implementation only.
To make the size test overhead negligible, let's factorize it out of
the alternate copy_to_user function where it is clear to the compiler
that no stack frame is needed. Thanks to CONFIG_ARM_UNWIND allowing
for frame pointers to be disabled and tail call optimization to kick in,
the overhead in the small copy case becomes only 3 assembly instructions.
A similar trick is applied to clear_user as well.
Signed-off-by: Nicolas Pitre <nico@marvell.com>
This implements {copy_to,clear}_user() by faulting in the userland
pages and then using the regular kernel mem{cpy,set}() to copy the
data (while holding the page table lock). This is a win if the regular
mem{cpy,set}() implementations are faster than the user copy functions,
which is the case e.g. on Feroceon, where 8-word STMs (which memcpy()
uses under the right conditions) give significantly higher memory write
throughput than a sequence of individual 32bit stores.
Here are numbers for page sized buffers on some Feroceon cores:
- copy_to_user on Orion5x goes from 51 MB/s to 83 MB/s
- clear_user on Orion5x goes from 89MB/s to 314MB/s
- copy_to_user on Kirkwood goes from 240 MB/s to 356 MB/s
- clear_user on Kirkwood goes from 367 MB/s to 1108 MB/s
- copy_to_user on Disco-Duo goes from 248 MB/s to 398 MB/s
- clear_user on Disco-Duo goes from 328 MB/s to 1741 MB/s
Because the setup cost is non negligible, this is worthwhile only if
the amount of data to copy is large enough. The operation falls back
to the standard implementation when the amount of data is below a certain
threshold. This threshold was determined empirically, however some targets
could benefit from a lower runtime determined value for optimal results
eventually.
In the copy_from_user() case, this technique does not provide any
worthwhile performance gain due to the fact that any kind of read access
allocates the cache and subsequent 32bit loads are just as fast as the
equivalent 8-word LDM.
Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: Nicolas Pitre <nico@marvell.com>
Tested-by: Martin Michlmayr <tbm@cyrius.com>
This allows for optional alternative implementations of __copy_to_user
and __clear_user, with a possible runtime fallback to the standard
version when the alternative provides no gain over that standard
version. This is done by making the standard __copy_to_user into a weak
alias for the symbol __copy_to_user_std. Same thing for __clear_user.
Those two functions are particularly good candidates to have alternative
implementations for, since they rely on the STRT instruction which has
lower performances than STM instructions on some CPU cores such as
the ARM1176 and Marvell Feroceon.
Signed-off-by: Nicolas Pitre <nico@marvell.com>
Small change (mostly formatting) to limit lookup based open calls to
file create only.
After discussion yesteday on samba-technical about the posix lookup
regression, and looking at a problem with cifs posix open to one
particular Samba version, Jeff and JRA realized that Samba server's
behavior changed in this area (posix open behavior on files vs.
directories). To make this behavior consistent, JRA just made a
fix to Samba server to alter how it handles open of directories (now
returning the equivalent of EISDIR instead of success). Since we don't
know at lookup time whether the inode is a directory or file (and
thus whether posix open will succeed with most current Samba server),
this change avoids the posix open code on lookup open (just issues
posix open on creates). This gets the semantic benefits we want
(atomicity, posix byte range locks, improved write semantics on newly
created files) and file create still is fast, and we avoid the problem
that Jeff noticed yesterday with "openat" (and some open directory
calls) of non-cached directories to one version of Samba server, and
will work with future Samba versions (which include the fix jra just
pushed into Samba server). I confirmed this approach with jra
yesterday and with Shirish today.
Posix open is only called (at lookup time) for file create now.
For opens (rather than creates), because we do not know if it
is a file or directory yet, and current Samba no longer allows
us to do posix open on dirs, we could end up wasting an open call
on what turns out to be a dir. For file opens, we wait to call posix
open till cifs_open. It could be added here (lookup) in the future
but the performance tradeoff of the extra network request when EISDIR
or EACCES is returned would have to be weighed against the 50%
reduction in network traffic in the other paths.
Reviewed-by: Shirish Pargaonkar <shirishp@us.ibm.com>
Tested-by: Jeff Layton <jlayton@redhat.com>
CC: Jeremy Allison <jra@samba.org>
Signed-off-by: Steve French <sfrench@us.ibm.com>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6:
ALSA: Kill truncate warning by shortening Sigmatel-specific AC97 control name
ALSA: hda - fix audio on HP TX25xx series notebooks
ALSA: pcsp - fix printk format warning again
ALSA sound/core/control.c:232: Control name 'Sigmatel Surround Phase
Inversion Playback Switch' truncated to 'Sigmatel Surround Phase
Inversion Playback ' bootup message by omitting weird Sigmatel prefix
in this case; also fix up the related ca0106 mixer control removal
part by using identical naming there.
Signed-off-by: Andreas Mohr <andi@lisas.de>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
MIPS: IP32: Remove unnecessary if not even harmful volatile keywords.
MIPS: IP32: Fix build error due to uninitialized variable.
MIPS: Fix sparse warning in incompatiable argument type of clear_user.
* 'sh/for-2.6.30' of git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6:
video: stop sh_mobile_lcdcfb only if started
sh: ap325 camera without i2c driver fix
Instead of queuing IPMB messages before channel initialization, just
throw them away. Nobody will be listening for them at this point,
anyway, and they will clog up the queue and nothing will be delivered
if we queue them.
Also set the current channel to the number of channels, as this value
is used to tell if the channel information has been initialized.
Signed-off-by: Corey Minyard <cminyard@mvista.com>
Cc: Ferenc Wagner <wferi@niif.hu>
Cc: Dan Frazier <dannf@hp.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This patch adds the PCI Device ID 0xc409 to the PCI ID table of via82cxxx.c,
as well as the 0x8409 south bridge ID.
This is required to make the IDE driver work on the VX855/VX875 integrated
chipset.
Signed-off-by: Harald Welte <HaraldWelte@viatech.com>
Cc: Joseph Chan <JosephChan@via.com.tw>
Cc: Bruce Chang <BruceChang@via.com.tw>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Nowadays we (almost) always store the currently executing command
in hwif->cmd so we can use it for the failed opcode reporting.
Cc: Martin Lottermoser <Martin.Lottermoser@t-online.de>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
On Tuesday 19 May 2009 20:29:28 Martin Lottermoser wrote:
> hdc: cdrom_decode_status: error=0x40 <3>{ LastFailedSense=0x04 }
> ide: failed opcode was: unknown
> hdc: DMA disabled
> ------------[ cut here ]------------
> kernel BUG at drivers/ide/ide-io.c:872!
It is possible for ide-cd to ignore ide_error()'s return value under
some circumstances. Workaround it in ide_intr() and ide_timer_expiry()
by checking if there is a device/port reset pending currently.
Fixes bug #13345:
http://bugzilla.kernel.org/show_bug.cgi?id=13345
Reported-by: Martin Lottermoser <Martin.Lottermoser@t-online.de>
Reported-and-tested-by: Modestas Vainius <modestas@vainius.eu>
Cc: Borislav Petkov <petkovbb@gmail.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Since 2.6.26 we support UDMA66 on ATAPI devices requiring IVB quirk:
commit 8588a2b732
("ide: add SH-S202J to ivb_list[]")
We also later added support for more such devices in:
commit e97564f362
("ide: More TSST drives with broken cable detection")
and in:
commit 3ced5c49bd
("ide: add TSSTcorp CDDVDW SH-S202H to ivb_list[]")
It turns out that such devices lack cable detection altogether
(which in turn results in incorrect detection of 40-wire cables
by our current cable detection strategy) so always handle them
by trusting host-side cable detection only.
v2:
Model detection fixup from Martin.
Reported-and-tested-by: Martin Lottermoser <Martin.Lottermoser@t-online.de>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
They are unneeded and as the issue fixed in lmo commit
63f7ec59053e3f850ab67a9938e631bcba64c6ce shows even harmful.
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
CC arch/mips/sgi-ip32/ip32-reset.o
cc1: warnings being treated as errors
arch/mips/sgi-ip32/ip32-reset.c: In function 'debounce':
arch/mips/sgi-ip32/ip32-reset.c:97: error: 'reg_a' is used uninitialized in this function
The issues is old but due to the volatile keyword gcc older than 4.4 did
not warn about this obvious bug.
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
The type of the second argument of access_ok should be (void __user *).
The unnecessary conversion of the clear_user address argument was causing
sparse to emit warnings on the __chk_user_ptr check.
Signed-off-by: Wu Zhangjin <wuzhangjin@gmail.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
This fixes a new memory leak problem in garbage collection. The
problem was brought by the bugfix patch ("nilfs2: fix lock order
reversal in nilfs_clean_segments ioctl").
Thanks to Kentaro Suzuki for finding this problem.
Reported-by: Kentaro Suzuki <k_suzuki@ms.sylc.co.jp>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Something in the HW or FW setup is busted and MSIs aren't working with
IPR on Bimini, so until we figure out exaxtly what's up, we quirk them
out
Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
This patch fixes the LCDC driver to avoid calling the
function sh_mobile_lcdc_start_stop(priv, 0) unless the
same function has been called before to start the LCDC
hardware.
Triggered when sh_mobile_lcdcfb.c failed to probe() due to
missing MSTP clocks.
Signed-off-by: Magnus Damm <damm@igel.co.jp>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
This patch fixes the ap325rxa ncm03j camera code to handle
the case where no i2c driver is present. Without this fix
i2c_transfer() may be passed NULL as adapter which results
in a crash.
Triggered when i2c-sh_mobile.c failed to probe() due to
missing MSTP clocks.
Signed-off-by: Magnus Damm <damm@igel.co.jp>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
Posix open code was not properly adding the file to the
list of open files. Fix allocating cifsFileInfo
more than once, and adding twice to flist and tlist.
Also fix mode setting to be done in one place in these
paths.
Signed-off-by: Steve French <sfrench@us.ibm.com>
Reviewed-by: Shirish Pargaonkar <shirishp@us.ibm.com>
Tested-by: Jeff Layton <jlayton@redhat.com>
Tested-by: Luca Tettamanti <kronos.it@gmail.com>
My old address will shut down in a few days time: remove it from the tree,
and add a tmpfs (shmem filesystem) maintainer entry with the new address.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/drm-2.6:
drm: Copy back ioctl data to userspace regardless of return code.
drm: Round size of SHM maps to PAGE_SIZE
* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
MIPS: 64-bit: Fix system lockup.
MIPS: IP28: Change to build with -mr10k-cache-barrier=store
MIPS: IP22: Fix hang in power button interrupt handler
MIPS: IP32: Fix hang on shutdown in power button interrupt handler.
The second argument of the probe method points to the amba_id
structure, so it's better passed with the correct type. None of the
current in-tree drivers uses the pointer, so they have only been
checked for a clean compile.
Change suggested by Russell King.
Signed-off-by: Alessandro Rubini <rubini@unipv.it>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Fixes a regression from commit 9d5b3ffc42
('drm: fixup some of the ioctl function exit paths'): The vblank ioctl
needs to update the userspace parameters when interrupted by a signal,
which was prevented by the return code check. This could cause the X
server to hang in drmWaitVBlank().
Signed-off-by: Michel Dänzer <daenzer@vmware.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
The address range size calculation inside local_flush_tlb_kernel_range()
is being truncated by a too small size variable holder on 64-bit systems.
The truncated size can result in an erroneous tlbsize check that means we
sit spinning inside a loop trying to flush a hige number of TLB entries.
This is for all intents and purposes a system hang. Fix by using an
appropriately sized valiable to hold the size.
[Ralf: Greg's original patch submission identified the issue and fixed one
instance in tlb-r4k.c but there there were several more. For consistency
I also modified tlb-r3k.c even though that file is only used on 32-bit.]
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Richard Sandiford's new code for inserting the cache-barriers, for GCC
4.3 and above and already incorporated in the current GCC-release, uses
a slightly different option-syntax.
Signed-off-by: peter fuerst <post@pfrst.de>
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
The hang was caused by the use of disable_irq() from the interrupt handler
itself. Fixed by the use of disable_irq_nosync(). The issue was
triggered by:
commit 3aa551c9b4
Author: Thomas Gleixner <tglx@linutronix.de>
Date: Mon Mar 23 18:28:15 2009 +0100
genirq: add threaded interrupt handler support
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
The hang was caused by the use of disable_irq() from the interrupt handler
itself. Fixed by the use of disable_irq_nosync(). The issue was
triggered by:
commit 3aa551c9b4
Author: Thomas Gleixner <tglx@linutronix.de>
Date: Mon Mar 23 18:28:15 2009 +0100
genirq: add threaded interrupt handler support
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* 'for-linus' of git://git.kernel.dk/linux-2.6-block:
cdrom: beyond ARRAY_SIZE of viocd_diskinfo
xen/blkfront: fix warning when deleting gendisk on unplug/shutdown
xen/blkfront: allow xenbus state transition to Closing->Closed when not Connected
The commit 5a641bcd63 changed the
printk format to '%lu', but the value passed seems to be dependent
on the architecture. On x86-64, I got a new warning now because an
int value is passed actaully.
As a workaround, just cast the value always to unsigned long.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
