Russell King [Thu, 11 Dec 2008 17:20:44 +0000 (17:20 +0000)]
[ARM] fix xm_x2xx_defconfig build errors
drivers/built-in.o: In function `pci_set_consistent_dma_mask':
hid-quirks.c:(.text+0x2664): multiple definition of `pci_set_consistent_dma_mask'
drivers/built-in.o: In function `pci_set_dma_mask':
hid-quirks.c:(.text+0x42c4): multiple definition of `pci_set_dma_mask'
because drivers/pci/pci.c was not seeing the definition disabling these
functions.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Hartley Sweeten [Mon, 8 Dec 2008 16:57:22 +0000 (17:57 +0100)]
[ARM] 5342/1: ep93xx: platform init cleanup
Use EP93XX_*_PHYS_BASE and SZ_* defines in ep93xx platform inits.
The following patch changes the flash memory hard-coded resource
addresses and MACHINE_START boot_params to EP93XX_*_PHYS_BASE and
SZ_* defines to improve readability. Also some minor whitespace
cleanup resulting from previous patches.
Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Russell King [Mon, 20 Oct 2008 10:18:40 +0000 (11:18 +0100)]
[ARM] dma: correct dma_supported() implementation
dma_supported() is supposed to indicate whether the system can support
the DMA mask it was passed, which depends on the maximal address which
can be returned for DMA allocations. If the mask is smaller than that,
we are unable to guarantee that the driver can reliably obtain suitable
memory.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Nicolas Pitre [Thu, 4 Dec 2008 06:03:57 +0000 (01:03 -0500)]
[ARM] Feroceon: pass proper -mtune argument to gcc
Compilation for the Feroceon core should use -mtune=marvell-f. This is
available in Code Sourcery's 2008Q3 release at the moment. Otherwise
fall back to -mtune-=xscale.
Ronen Shitrit [Sun, 19 Oct 2008 21:19:07 +0000 (23:19 +0200)]
[ARM] Kirkwood: properly handle the WAN port on newer RD88F6281 boards
On newer versions of the RD88F6281 board, the WAN port is connected to
its own ethernet port on the CPU, via a separate PHY, whereas on older
versions of the board, it is connected to one of the PHYs in the
ethernet switch. In the RD8F6281 setup code, detect which version of
the board we are running on, and instantiate the ethernet ports and
switch driver accordingly.
Ronen Shitrit [Sun, 19 Oct 2008 21:10:14 +0000 (23:10 +0200)]
[ARM] Kirkwood: allow instantiating the second ethernet port
The 88f6192 and 88f6281 Kirkwood SoCs support two ethernet ports.
Add the platform glue that will allow board support files to
instantiate the second ethernet port.
The 88F5182 found in the DNS-323 rev B1 (and some other devices, such
as the CH3SNAS) require different initialisation of the SATA
controller and MPP registers.
Matt Palmer [Sat, 18 Oct 2008 21:15:09 +0000 (08:15 +1100)]
[ARM] DNS323: Read MAC address from flash
Based on similar code from the tsx09 series of machines, just rips the MAC
address out of flash and stuffs it into the NIC. Tested on a DNS323 rev B1.
It's possible (though unlikely) that an A1 will have the MAC in a different
location in flash.
[ARM] Orion: add the option to support different ehci phy initialization
The Orion ehci driver serves the Orion, kirkwood and DD Soc families.
Since each of those integrate a different USB phy we should have the
ability to use few initialization sequences or to leave the boot loader
phy settings as is.
Anton Vorontsov [Thu, 23 Oct 2008 19:06:56 +0000 (21:06 +0200)]
[ARM] pxa: add basic support for HP iPAQ h5000
This patch adds HP iPAQ h5000's (h5400, h5500) basic definitions.
Kernel will able to boot, work via serial console, mount filesystems
placed on flashes and run USB gadgets (g_ether by default).
Other device drivers (frame buffer, LCD, touchscreen, backlight,
bluetooth, w1/battery, ...) are depend on SAMCOP and MediaQ
SoCs/MFDs, drivers to which will be submitted too, after massive
cleanups.
This machine will be used as "real user" for these new drivers.
This is an updated version of the patch, which contains fixes proposed on
linux-arm-kernel mailing list.
Signed-off-by: Anton Vorontsov <cbouatmailru@gmail.com> Signed-off-by: Milan Plzik <milan.plzik@gmail.com> Signed-off-by: Eric Miao <eric.miao@marvell.com>
Robert Jarzmik [Sat, 15 Nov 2008 15:09:59 +0000 (16:09 +0100)]
[ARM] pxa/MioA701: improve power supply sources
Take advantage of the newly created wm97xx battery driver
and remove useless code in mioa701 board code. Add also the
ac connection detect capability after the matching gpio was
discovered.
Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr> Signed-off-by: Eric Miao <eric.miao@marvell.com>
Robert Jarzmik [Tue, 18 Nov 2008 19:23:32 +0000 (20:23 +0100)]
[ARM] pxa/MioA701: discovered new gpio definitions.
The charger enable gpio is straight (1 means draw from USB
Vbus, 0 mean do not draw). The USB Vbus sensing is inverted
(1 means no Vbus voltage sensed, 0 means Vbus voltage
present).
Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr> Signed-off-by: Eric Miao <eric.miao@marvell.com>
Robert Jarzmik [Sat, 15 Nov 2008 15:09:58 +0000 (16:09 +0100)]
[ARM] pxa/MioA701: change reset function to preserve RTC.
Change the halt and reboot method from gpio based to "jump
to ROM IPL beginning". This gives control back to IPL, which
without PowerOn key pressed, will put the device into deep
sleep until PowerOn is pressed for 1 second.
But this has the benefit of keeping the RTC registers across
reboots, which is good for OS change.
Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr> Signed-off-by: Eric Miao <eric.miao@marvell.com>
Robert Jarzmik [Sat, 15 Nov 2008 15:09:54 +0000 (16:09 +0100)]
[ARM] pxa/MioA701: add camera support for Mio A701 board.
Add GPIO configuration and platform specific declarations
to make Mitac Mio A701 camera chip work. The chip is a
Micron MT9M111 CMOS sensor, based on PXA QIF interface and
I2C bus for sensor control.
Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr> Signed-off-by: Eric Miao <eric.miao@marvell.com>
Since mioa701 board has migrated to the mfp architecture,
low power gpio setup is now correctly handled even when gpio
led driver is not loaded, and leds and vibrator don't stay
activated in suspend mode (especially vibrator).
Remove the not needed anymore dependency.
Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr> Signed-off-by: Eric Miao <eric.miao@marvell.com>
Philipp Zabel [Sun, 29 Jun 2008 14:53:34 +0000 (16:53 +0200)]
[ARM] pxa: cpufreq-pxa2xx: sdram_rows detection support
This patch implements Eric Miao's idea to detect the correct value of
sdram_rows by inspecting the MDCNFG register settings.
It is only tested on two pxa27x devices with 64MB RAM (magician and
hx4700) so far.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com> Signed-off-by: Eric Miao <eric.miao@marvell.com>
Marc Zyngier [Tue, 4 Nov 2008 12:33:25 +0000 (13:33 +0100)]
[ARM] pxa: cpufreq-pxa2xx: allow frequency table selection
Following the removal of the "->policy" usage for PXA255 in patch 459fc208abd1b365fa013c17d433dfb5b4bc1e3a (cpufreq: remove
policy->governor setting in drivers initialization), this patch
introduces an option (called "pxa255_turbo_table") to select either
the "run" or "turbo" frequency table.
It also cures the runtime warning that was printed each time the
frequency was changed.
Got rid of all references to CPUFREQ_POLICY_* for pxa255, and sticked
with the run/turbo thing.
Tested on an Arcom/Eurotech Viper.
Signed-off-by: Marc Zyngier <maz@misterjones.org> Acked-by: Dominik Brodowski <linux@dominikbrodowski.net> Signed-off-by: Eric Miao <eric.miao@marvell.com>
Eric Miao [Sat, 29 Nov 2008 13:42:39 +0000 (21:42 +0800)]
[ARM] pxa: add base PXA935 support due to CPUID change
PXA935 has changed its implementor ID from Intel to Marvell, this
patch modifies arch/arm/boot/compressed/head.S and proc-xsc3.S to
support a smooth bootup.
Eric Miao [Fri, 28 Nov 2008 06:19:33 +0000 (14:19 +0800)]
[ARM] pxa: move AC97 register definitions into dedicated regs-ac97.h
The optimal change would be to move the AC97 register definitions into
the AC97 driver, unfortunately, the registers are shared between several
files. Move them into a dedicated regs-ac97.h first.
Eric Miao [Fri, 28 Nov 2008 05:38:03 +0000 (13:38 +0800)]
[ARM] pxa: remove the now unused IMPMCR/IMPMSR register definitions
There two are internal registers that are used to control the power
management of the Internal Memory (i.e. Internal SRAM). They are
referenced nowhere and removed here to simplify pxa-regs.h a bit.
Eric Miao [Thu, 27 Nov 2008 09:08:42 +0000 (17:08 +0800)]
[ARM] pxa: remove unused PWM register definitions, use generic PWM API
We now have generic PWM API for PXA, the PWM registers definitions are
now used nowhere, and it is not encouraged to manipulate them directly
by driver code.
Eric Miao [Sun, 30 Nov 2008 12:58:42 +0000 (20:58 +0800)]
[ARM] pxa: removed unused declarations of pxa_gpio_* in hardware.h
pxa_gpio_{get,set}_value() are not used anymore, remove them from hardware.h.
Declaration of pxa_gpio_mode() is still being referenced and thus moved into
pxa2xx-gpio.h
Eric Miao [Wed, 26 Nov 2008 10:12:04 +0000 (18:12 +0800)]
[ARM] pxa: add support for additional GPIOs on PXA26x
Original patch from Marek Vasut, the problems with PXA26x are:
1. there are additional 4 GPIOs 86,87,88,89 have their direction bits
inverted in GPDR2, as well as their alternate function bits being
'1' for their GPIO functionality in GAFRx
2. there is no easy way to decide if the processor is a pxa26x or a
pxa250/pxa255 at run-time, so the assumption here is the pxa26x
will be treated as one of the pxa25x variants, and board code
should have a better knowledge of the processor it is featured
Introduce pxa26x_init_irq() for the second purpose, and treat the
additional GPIOs > 85 on PXA25x specially.
Kconfig option CONFIG_CPU_PXA26x is introduced to optimize the code
a bit when PXA26x support isn't needed. Board config options have
to select this to enable the support for PXA26x.
__gpio_is_inverted() will be optimized way when CONFIG_CPU_PXA26x
isn't selected.
Signed-off-by: Marek Vasut <marek.vasut@gmail.com> Signed-off-by: Eric Miao <eric.miao@marvell.com>
Linus Torvalds [Tue, 2 Dec 2008 03:56:34 +0000 (19:56 -0800)]
Merge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-2.6
* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-2.6: (25 commits)
em28xx: remove backward compat macro added on a previous fix
V4L/DVB (9748): em28xx: fix compile warning
V4L/DVB (9743): em28xx: fix oops audio
V4L/DVB (9742): em28xx-alsa: implement another locking schema
V4L/DVB (9732): sms1xxx: use new firmware for Hauppauge WinTV MiniStick
V4L/DVB (9691): gspca: Move the video device to a separate area.
V4L/DVB (9690): gspca: Lock the subdrivers via module_get/put.
V4L/DVB (9689): gspca: Memory leak when disconnect while streaming.
V4L/DVB (9668): em28xx: fix a race condition with hald
V4L/DVB (9664): af9015: don't reconnect device in USB-bus
V4L/DVB (9647): em28xx: void having two concurrent control URB's
V4L/DVB (9646): em28xx: avoid allocating/dealocating memory on every control urb
V4L/DVB (9645): em28xx: Avoid memory leaks if registration fails
V4L/DVB (9639): Make dib0700 remote control support work with firmware v1.20
V4L/DVB (9635): v4l: s2255drv fix firmware test on big-endian
V4L/DVB (9634): Make sure the i2c gate is open before powering down tuner
V4L/DVB (9632): make em28xx aux audio input work
V4L/DVB (9631): Make s2api work for ATSC support
V4L/DVB (9627): em28xx: Avoid i2c register error for boards without eeprom
V4L/DVB (9608): Fix section mismatch warning for dm1105 during make
...
Andrew Morton [Mon, 1 Dec 2008 21:14:08 +0000 (13:14 -0800)]
drivers/gpu/drm/i915/i915_irq.c: fix warning
drivers/gpu/drm/i915/i915_irq.c: In function 'i915_disable_pipestat':
drivers/gpu/drm/i915/i915_irq.c:101: warning: control may reach end of non-void function 'i915_pipestat' being inlined
Cc: Dave Airlie <airlied@linux.ie> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Jarkko Lavinen [Mon, 1 Dec 2008 21:14:08 +0000 (13:14 -0800)]
i82875p_edac: fix module remove
Fix module removal bugs of i82875p_edac. Also i82975x_edac code seems to
have the same module removal bugs as in i82875p_edac.
The problems were:
1. In module removal i82875p_remove_one() is never called.
Variable i82875p_registered is newer changed from 1, which
guarantees i82875p_remove_one() is not called (and even if it were
called, it would be called in wrong order).
As a result, the edac_mc workque is not stopped and keeps probing.
If kernel debugging options are not enabled, user may not notice
anything going wrong.
if debugging options are enabled and I do "rmmod i82875p_edac", I
get:
Fix for this is to get rid of needles variable i82875p_registered
altogether and run i82875p_remove_one() *before*
pci_unregister_driver().
2. edac_mc_del_mc() uses mci after freeing mci
edac_mc_del_mc() calls calls edac_remove_sysfs_mci_device(). The
kobject refcount of mci drops to 0 and mci is freed. After this
mci is accessed via debug print and i82875p_remove_one() still
uses mci->pvt and tries to free mci again with edac_mc_free().
The fix for this is add kobject_get(&mci->edac_mci_kobj) after
edac_mc_alloc(). Then the mci is still available after returning
from edac_mc_del_mc() with refcount 1, and mci->pvt is still
available. When i82875p_remove_one() finally calls edac_mc_free(),
this will cause kobject_put() and mci is released properly.
Signed-off-by: Jarkko Lavinen <jlavi@iki.fi> Cc: Doug Thompson <norsk5@yahoo.com> Cc: Alan Cox <alan@lxorguk.ukuu.org.uk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Jarkko Lavinen [Mon, 1 Dec 2008 21:14:06 +0000 (13:14 -0800)]
i82875p_edac: fix overflow device resource setup
When I do "modprobe i82875p_edac" on my Asus P4C800 MB on kernels 2.6.26
or later, the module load fails due to BAR 0 collision. On 2.6.25 the
module loads just fine.
The overflow device on the MB seems to be hidden and its resources are not
allocated at normal PCI bus init. Log shows the missing resource problem:
EDAC DEBUG: i82875p_probe1()
PCI: 0000:00:06.0 reg 10 32bit mmio: [fecf0000, fecf0fff]
pci 0000:00:06.0: device not available because of BAR 0
[0xfecf0000-0xfecf0fff] collisions
EDAC i82875p: i82875p_setup_overfl_dev(): Failed to enable overflow
device
The patch below fixes this by calling pci_bus_assign_resources() after
the overflow device is revealed and added to the bus. With this patch
I am again able to load and use the module.
Signed-off-by: Jarkko Lavinen <jlavi@iki.fi> Cc: Doug Thompson <norsk5@yahoo.com> Cc: Jesse Barnes <jbarnes@virtuousgeek.org> Cc: Alan Cox <alan@lxorguk.ukuu.org.uk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The commit aef7db4bd5a3b6068dfa05919a3d685199eed116 fixed the problem with
recursive locking in fb blanking code if blank is caused by user setting
the /sys/class/graphics/fb*/blank. However this broke the fbcon timeout
blanking.
If you use a driver that defines ->fb_blank operation and at the same time
that driver relies on other driver (e.g. backlight or lcd class) to blank
the screen, when the fbcon times out and tries to blank the fb, it will
call only fb driver blanker and won't notify the other driver. Thus FB
output is disabled, but the screen isn't blanked.
Restore fbcon blanking and at the same time apply the proper fix for the
above problem: if fbcon_blank is called with FBINFO_FLAG_USEREVENT, we are
already called through notification from fb_blank, thus we don't have to
blank the fb again.
Randy Dunlap [Mon, 1 Dec 2008 21:14:04 +0000 (13:14 -0800)]
ntfs: don't fool kernel-doc
kernel-doc handles macros now (it has for quite some time), so change the
ntfs_debug() macro's kernel-doc to be just before the macro instead of
before a phony function prototype.
[akpm@linux-foundation.org: coding-style fixes] Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com> Cc: Anton Altaparmakov <aia21@cantab.net> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Randy Dunlap [Mon, 1 Dec 2008 21:14:03 +0000 (13:14 -0800)]
kernel-doc: handle varargs cleanly
The method for listing varargs in kernel-doc notation is:
* @...: these arguments are printed by the @fmt argument
but scripts/kernel-doc is confused: it always lists varargs as:
... variable arguments
and ignores the @...: line's description, but then prints that
line after the list of function parameters as though it's
not part of the function parameters.
This patch makes kernel-doc print the supplied @... description if it is
present; otherwise a boilerplate "variable arguments" is printed.
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
David Howells [Mon, 1 Dec 2008 21:14:00 +0000 (13:14 -0800)]
frv: fix mmap2 error handling
Fix the error handling in sys_mmap2(). Currently, if the pgoff check
fails, fput() might have to be called (which it isn't), so do the pgoff
check first, before fget() is called.
Signed-off-by: David Howells <dhowells@redhat.com> Reported-by: Julia Lawall <julia@diku.dk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The description for 'D' was missing in the comment... (causing me a
minute of WTF followed by looking at more of the code)
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
radeonfb: fix problem with color expansion & alignment
The engine on some radeon variants locks up if color expansion is called
for non aligned source data. This patch enables a feature of the core
fbdev to request aligned input pixmaps and uses the HW clipping engine to
clip the output to the requested size
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Tested-by: James Cloos <cloos@jhcloos.com> Cc: "Rafael J. Wysocki" <rjw@sisk.pl> Cc: "David S. Miller" <davem@davemloft.net> Cc: Krzysztof Helt <krzysztof.h1@poczta.fm> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Ben Dooks [Mon, 1 Dec 2008 21:13:56 +0000 (13:13 -0800)]
spi: fix spi_s3c24xx_gpio device handle lookup
The spidev_to_sg() call in spi_s3c24xx_gpio.c was using the wrong method
to convert the spi device into the private data for the driver. Fix this
by using spi_master_get_devdata.
Signed-off-by: Ben Dooks <ben-linux@fluff.org> Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Jan Nikitenko [Mon, 1 Dec 2008 21:13:56 +0000 (13:13 -0800)]
spi: au1550_spi full duplex dma fix
Fix unsafe order in dma mapping operation: always flush data from the
cache *BEFORE* invalidating it, to allow full duplex transfers where the
same buffer may be used for both writes and reads. Tested with mmc-spi.
Signed-off-by: Jan Nikitenko <jan.nikitenko@gmail.com> Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Davide Libenzi [Mon, 1 Dec 2008 21:13:55 +0000 (13:13 -0800)]
epoll: introduce resource usage limits
It has been thought that the per-user file descriptors limit would also
limit the resources that a normal user can request via the epoll
interface. Vegard Nossum reported a very simple program (a modified
version attached) that can make a normal user to request a pretty large
amount of kernel memory, well within the its maximum number of fds. To
solve such problem, default limits are now imposed, and /proc based
configuration has been introduced. A new directory has been created,
named /proc/sys/fs/epoll/ and inside there, there are two configuration
points:
max_user_instances = Maximum number of devices - per user
max_user_watches = Maximum number of "watched" fds - per user
The current default for "max_user_watches" limits the memory used by epoll
to store "watches", to 1/32 of the amount of the low RAM. As example, a
256MB 32bit machine, will have "max_user_watches" set to roughly 90000.
That should be enough to not break existing heavy epoll users. The
default value for "max_user_instances" is set to 128, that should be
enough too.
This also changes the userspace, because a new error code can now come out
from EPOLL_CTL_ADD (-ENOSPC). The EMFILE from epoll_create() was already
listed, so that should be ok.
[akpm@linux-foundation.org: use get_current_user()] Signed-off-by: Davide Libenzi <davidel@xmailserver.org> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Cc: <stable@kernel.org> Cc: Cyrill Gorcunov <gorcunov@gmail.com> Reported-by: Vegard Nossum <vegardno@ifi.uio.no> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Stefano Babic [Mon, 1 Dec 2008 21:13:53 +0000 (13:13 -0800)]
spi: mpc52xx_psc_spi chipselect bugfix
According to the manual the "tdfOnExit" flag must be set on the last byte
we want to send. The PSC controller holds SS low until the flag is set.
However, the flag was set always on the last byte of the FIFO,
independently if it is the last byte of the transfer. This generates
spurious toggling of the SS signals that breaks the protocol of some
peripherals. Fix.
Signed-off-by: Stefano Babic <sbabic@denx.de> Acked-by: Grant Likely <grant.likely@secretlab.ca> Signed-off-by: David Brownell <dbrownell@users.sourceforge.net> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Takashi Iwai [Mon, 1 Dec 2008 21:13:49 +0000 (13:13 -0800)]
parport_serial: fix array overflow
The netmos_9xx5_combo type assumes that PCI SSID provides always the
correct value for the number of parallel and serial ports, but there are
indeed broken devices with wrong numbers, which may result in Oops.
This patch simply adds the check of the array range.
While memory hotplug allocate/free memmap, page_cgroup doesn't free
page_cgroup at OFFLINE when page_cgroup is allocated via bootomem.
(Because freeing bootmem requires special care.)
Then, if page_cgroup is allocated by bootmem and memmap is freed/allocated
by memory hotplug, page_cgroup->page == page is no longer true.
But current MEM_ONLINE handler doesn't check it and update
page_cgroup->page if it's not necessary to allocate page_cgroup. (This
was not found because memmap is not freed if SPARSEMEM_VMEMMAP is y.)
And I noticed that MEM_ONLINE can be called against "part of section".
So, freeing page_cgroup at CANCEL_ONLINE will cause trouble. (freeing
used page_cgroup) Don't rollback at CANCEL.
One more, current memory hotplug notifier is stopped by slub because it
sets NOTIFY_STOP_MASK to return vaule. So, page_cgroup's callback never
be called. (low priority than slub now.)
I think this slub's behavior is not intentional(BUG). and fixes it.
Another way to be considered about page_cgroup allocation:
- free page_cgroup at OFFLINE even if it's from bootmem
and remove specieal handler. But it requires more changes.
Nick Piggin [Mon, 1 Dec 2008 21:13:47 +0000 (13:13 -0800)]
mm: vmalloc fix lazy unmapping cache aliasing
Jim Radford has reported that the vmap subsystem rewrite was sometimes
causing his VIVT ARM system to behave strangely (seemed like going into
infinite loops trying to fault in pages to userspace).
We determined that the problem was most likely due to a cache aliasing
issue. flush_cache_vunmap was only being called at the moment the page
tables were to be taken down, however with lazy unmapping, this can happen
after the page has subsequently been freed and allocated for something
else. The dangling alias may still have dirty data attached to it.
The fix for this problem is to do the cache flushing when the caller has
called vunmap -- it would be a bug for them to write anything else to the
mapping at that point.
That appeared to solve Jim's problems.
Reported-by: Jim Radford <radford@blackbean.org> Signed-off-by: Nick Piggin <npiggin@suse.de> Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Linus Torvalds [Tue, 2 Dec 2008 02:56:55 +0000 (18:56 -0800)]
Merge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mfasheh/ocfs2
* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mfasheh/ocfs2:
ocfs2: fix regression in ocfs2_read_blocks_sync()
ocfs2: fix return value set in init_dlmfs_fs()
ocfs2: Small documentation update
ocfs2: fix wake_up in unlock_ast
ocfs2: initialize stack_user lvbptr
ocfs2: comments typo fix
Mark Fasheh [Fri, 21 Nov 2008 22:06:55 +0000 (14:06 -0800)]
ocfs2: fix regression in ocfs2_read_blocks_sync()
We're panicing in ocfs2_read_blocks_sync() if a jbd-managed buffer is seen.
At first glance, this seems ok but in reality it can happen. My test case
was to just run 'exorcist'. A struct inode is being pushed out of memory but
is then re-read at a later time, before the buffer has been checkpointed by
jbd. This causes a BUG to be hit in ocfs2_read_blocks_sync().
Reviewed-by: Joel Becker <joel.becker@oracle.com> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
Coly Li [Mon, 17 Nov 2008 04:38:22 +0000 (12:38 +0800)]
ocfs2: fix return value set in init_dlmfs_fs()
In init_dlmfs_fs(), if calling kmem_cache_create() failed, the code will use return value from
calling bdi_init(). The correct behavior should be set status as -ENOMEM before going to "bail:".
Signed-off-by: Coly Li <coyli@suse.de> Acked-by: Sunil Mushran <sunil.mushran@oracle.com> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
Linus Torvalds [Mon, 1 Dec 2008 19:23:33 +0000 (11:23 -0800)]
Merge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/libata-dev
* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/libata-dev:
libata: blacklist Seagate drives which time out FLUSH_CACHE when used with NCQ
[libata] pata_rb532_cf: fix signature of the xfer function
[libata] pata_rb532_cf: fix and rename register definitions
ata_piix: add borked Tecra M4 to broken suspend list
Linus Torvalds [Mon, 1 Dec 2008 19:01:54 +0000 (11:01 -0800)]
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
IB/mlx4: Fix MTT leakage in resize CQ
IB/ehca: Fix problem with generated flush work completions
IB/ehca: Change misleading error message on memory hotplug
mlx4_core: Save/restore default port IB capability mask
Tejun Heo [Thu, 27 Nov 2008 04:36:48 +0000 (13:36 +0900)]
libata: blacklist Seagate drives which time out FLUSH_CACHE when used with NCQ
Some recent Seagate harddrives have firmware bug which causes FLUSH
CACHE to timeout under certain circumstances if NCQ is being used.
This can be worked around by disabling NCQ and fixed by updating the
firmware. Implement ATA_HORKAGE_FIRMWARE_UPDATE and blacklist these
devices.
The wiki page has been updated to contain information on this issue.
http://ata.wiki.kernel.org/index.php/Known_issues
Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
projects / linux-beck.git / log