Video subdevices, like cameras, decoders, connect to video bridges over
specialised busses. Data is being transferred over these busses in various
formats, which only loosely correspond to fourcc codes, describing how video
data is stored in RAM. This is not a one-to-one correspondence, therefore we
cannot use fourcc codes to configure subdevice output data formats. This patch
adds codes for several such on-the-bus formats and an API, similar to the
familiar .s_fmt(), .g_fmt(), .try_fmt(), .enum_fmt() API for configuring those
codes. After all users of the old API in struct v4l2_subdev_video_ops are
converted, it will be removed. Also add helper routines to support generic
pass-through mode for the soc-camera framework.
create mode 100644 drivers/media/video/soc_mediabus.c
create mode 100644 include/media/soc_mediabus.h
create mode 100644 include/media/v4l2-mediabus.h
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Acked-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
After this change drivers can be further extended to not fail, if they don't
get platform data, but to use defaults.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Up to now, if a client driver needed platform data apart from those contained
in struct soc_camera_link, it had to embed the struct into its own object. This
makes the use of such a driver in configurations other than soc-camera
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The 16-bit monochrome fourcc code has been previously abused for a 10-bit
format, add a new 10-bit code instead. Also add missing 8- and 10-bit Bayer
fourcc codes for completeness.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Introduce new v4l2-subdev sensor operations, move .enum_framesizes() and
.enum_frameintervals() methods to it, add a new .g_skip_top_lines() method
and switch soc-camera to use it instead of .y_skip_top soc_camera_device
member, which can now be removed.
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Reviewed-by: Hans Verkuil <hverkuil@xs4all.nl>
Reviewed-by: Sergio Aguirre <saaguirre@ti.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This patch adds a helper function to get description of a digital
video preset added by the video timing API. This will be useful for drivers
implementing the above API.
Signed-off-by: Muralidharan Karicheri <m-karicheri2@ti.com>
Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
We'll need to register a sysfs class for the IR devices. As such, the better
is to have the input_register_device()/input_unregister_device() inside
the ir register/unregister functions.
Also, solves a naming problem with V4L ir_input_init() function, that were,
in fact, registering a device.
While here, do a few cleanups at budget-ci IR logic.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Now, ir_input_free does more than just freeing the keytab. Better to
rename it as ir_input_unregister.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Move non-V4L specific stuff from ir-functions ir_input_init() into
a new function to register ir devices: ir_input_register().
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Split the ir-common into two separate modules:
- ir-core: it is the IR-independent functions;
- ir-common: has the common part used by V4L drivers.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This adds the above APIs to the v4l2 core. This is based on version v1.2
of the RFC titled "V4L - Support for video timings at the input/output interface"
Following new ioctls are added:-
- VIDIOC_ENUM_DV_PRESETS
- VIDIOC_S_DV_PRESET
- VIDIOC_G_DV_PRESET
- VIDIOC_QUERY_DV_PRESET
- VIDIOC_S_DV_TIMINGS
- VIDIOC_G_DV_TIMINGS
Please refer to the RFC for the details. This code was tested using vpfe
capture driver on TI's DM365. Following is the test configuration used :-
Blu-Ray HD DVD source -> TVP7002 -> DM365 (VPFE) ->DDR
A draft version of the TVP7002 driver (currently being reviewed in the mailing
list) was used that supports V4L2_DV_1080I60 & V4L2_DV_720P60 presets.
A loopback video capture application was used for testing these APIs. This calls
following IOCTLS :-
- verify the new v4l2_input capabilities flag added
- Enumerate available presets using VIDIOC_ENUM_DV_PRESETS
- Set one of the supported preset using VIDIOC_S_DV_PRESET
- Get current preset using VIDIOC_G_DV_PRESET
- Detect current preset using VIDIOC_QUERY_DV_PRESET
- Using stub functions in tvp7002, verify VIDIOC_S_DV_TIMINGS
and VIDIOC_G_DV_TIMINGS ioctls are received at the sub device.
- Tested on 64bit platform by Hans Verkuil
Signed-off-by: Muralidharan Karicheri <m-karicheri2@ti.com>
Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl>
Reviewed-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
The video_device::minor field is used where it shouldn't, either to
- test for error conditions that can't happen anymore with the current
v4l-dvb core,
- store the value in a driver private field that isn't used anymore,
- check the video device type where video_device::vfl_type should be
used, or
- create the name of a kernel thread that should get a stable name.
Remove or fix those use cases.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Replace the video_is_unregistered function by a video_is_registered
function. The V4L2_FL_UNREGISTERED flag is replaced by a
V4L2_FL_REGISTERED flag.
This change makes the video_is_registered function return coherent
results when called on an initialize but not yet registered video_device
instance. The function can now be used instead of checking
video_device::minor.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Many drivers access the device number (video_device::v4l2_devnode::num)
in order to print the video device node name. Add and use a helper
function to retrieve the video_device node name.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
This IR uses NEC protocol, with address=0x14. This keymap is similar
to the existing Terratec Cinergy XS, except that:
- it contains the full address/command code;
- the Music button were mapped as KEY_RADIO;
- some keycodes from the previous entry were wrong, IMHO.
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
* git://git.kernel.org/pub/scm/linux/kernel/git/agk/linux-2.6-dm: (80 commits)
dm snapshot: use merge origin if snapshot invalid
dm snapshot: report merge failure in status
dm snapshot: merge consecutive chunks together
dm snapshot: trigger exceptions in remaining snapshots during merge
dm snapshot: delay merging a chunk until writes to it complete
dm snapshot: queue writes to chunks being merged
dm snapshot: add merging
dm snapshot: permit only one merge at once
dm snapshot: support barriers in snapshot merge target
dm snapshot: avoid allocating exceptions in merge
dm snapshot: rework writing to origin
dm snapshot: add merge target
dm exception store: add merge specific methods
dm snapshot: create function for chunk_is_tracked wait
dm snapshot: make bio optional in __origin_write
dm mpath: reject messages when device is suspended
dm: export suspended state to targets
dm: rename dm_suspended to dm_suspended_md
dm: swap target postsuspend call and setting suspended flag
dm crypt: add plain64 iv
...
* 'core-locking-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (26 commits)
clockevents: Convert to raw_spinlock
clockevents: Make tick_device_lock static
debugobjects: Convert to raw_spinlocks
perf_event: Convert to raw_spinlock
hrtimers: Convert to raw_spinlocks
genirq: Convert irq_desc.lock to raw_spinlock
smp: Convert smplocks to raw_spinlocks
rtmutes: Convert rtmutex.lock to raw_spinlock
sched: Convert pi_lock to raw_spinlock
sched: Convert cpupri lock to raw_spinlock
sched: Convert rt_runtime_lock to raw_spinlock
sched: Convert rq->lock to raw_spinlock
plist: Make plist debugging raw_spinlock aware
bkl: Fixup core_lock fallout
locking: Cleanup the name space completely
locking: Further name space cleanups
alpha: Fix fallout from locking changes
locking: Implement new raw_spinlock
locking: Convert raw_rwlock functions to arch_rwlock
locking: Convert raw_rwlock to arch_rwlock
...
* git://git.infradead.org/battery-2.6:
power_supply_sysfs: Handle -ENODATA in a special way
wm831x_backup: Remove unused variables
gta02: Set pcf50633 charger_reference_current_ma
pcf50633: Query charger status directly
pcf50633: Properly reenable charging when the supply conditions change
pcf50633: Get rid of charging restart software auto-triggering
pcf50633: introduces battery charging current control
pcf50633: Add ac power supply class to the charger
wm831x: Factor out WM831x backup battery charger
Implement selftest feature as specified by chip manufacturer. Control:
read selftest sysfs entry
Response: "OK x y z" or "FAIL x y z"
where x, y, and z are difference between selftest mode and normal mode.
Test is passed when values are within acceptance limit values.
Acceptance limits are provided via platform data. See chip spesifications
for acceptance limits. If limits are not properly set, OK / FAIL decision
is meaningless. However, userspace application can still make decision
based on the numeric x, y, z values.
Selftest is meant for HW diagnostic purposes. It is not meant to be
called during normal use of the chip. It may cause false interrupt
events. Selftest mode delays polling of the normal results but it doesn't
cause wrong values. Chip must be in static state during selftest. Any
acceration during the test causes most probably failure.
Signed-off-by: Samu Onkalo <samu.p.onkalo@nokia.com>
Acked-by: Éric Piel <Eric.Piel@tremplin-utc.net>
Cc: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Add the possibility to remap axes via platform data. Function pointers
for resource setup and release purposes
Signed-off-by: Samu Onkalo <samu.p.onkalo@nokia.com>
Acked-by: Éric Piel <eric.piel@tremplin-utc.net>
Cc: Pavel Machek <pavel@ucw.cz>
Cc: Jean Delvare <khali@linux-fr.org>
Cc: "Trisal, Kalhan" <kalhan.trisal@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Allow the use of another DMA controller driver in atmel-mci sd/mmc driver.
This adds a generic dma_slave pointer to the mci platform structure where
we can store DMA controller information. In atmel-mci we use information
provided by this structure to initialize the driver (with new helper
functions that are architecture dependant).
This also adds at32/avr32 chip modifications to cope with this new access
method.
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Cc: Haavard Skinnemoen <hskinnemoen@atmel.com>
Cc: <linux-mmc@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Recently, We marked strstrip() as must_check. because it was frequently
misused and it should be checked. However, we found one exception.
scsi/ipr.c intentionally ignore return value of strstrip. Because it
wishes to keep the whitespace at the beginning.
Thus we need to keep with and without checked whitespace trim function.
This patch adds a new strim() and changes ipr.c to use it.
[akpm@linux-foundation.org: coding-style fixes]
Suggested-by: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
On the following sentence:
while (*s && isspace(*s))
s++;
If *s == 0, isspace() evaluates to ((_ctype[*s] & 0x20) != 0), which
evaluates to ((0x08 & 0x20) != 0) which equals to 0 as well.
If *s == 1, we depend on isspace() result anyway. In other words,
"a char equals zero is never a space", so remove this check.
Also, *s != 0 is most common case (non-null string).
Fixed const return as noticed by Jan Engelhardt and James Bottomley.
Fixed unnecessary extra cast on strstrip() as noticed by Jan Engelhardt.
Signed-off-by: André Goddard Rosa <andre.goddard@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
While at it, use tabs to indent the comments.
Signed-off-by: André Goddard Rosa <andre.goddard@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The kernel offers with TIOCL_GETKMSGREDIRECT ioctl() the possibility to
redirect the kernel messages to a specific console.
However, since it's not possible to switch to the kernel message console
after a panic(), it would be nice if the kernel would print the panic
message on the current console.
This patch series adds a new interface to access the global kmsg_redirect
variable by a function to be able to use it in code where
CONFIG_VT_CONSOLE is not set (kernel/panic.c).
This patch:
Instead of using and exporting a global value kmsg_redirect, introduce a
function vt_kmsg_redirect() that both can set and return the console where
messages are printed.
Change all users of kmsg_redirect (the VT code itself and kernel/power.c)
to the new interface.
The main advantage is that vt_kmsg_redirect() can also be used when
CONFIG_VT_CONSOLE is not set.
Signed-off-by: Bernhard Walle <bernhard@bwalle.de>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
..and include them in the lxfb/gxfb drivers rather than asm/geode.h (where
possible).
Signed-off-by: Andres Salomon <dilinger@collabora.co.uk>
Cc: Jordan Crouse <jordan@cosmicpenguin.net>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: john stultz <johnstul@us.ibm.com>
Cc: Chris Ball <cjb@laptop.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The only thing that uses this is the reboot_fixups code.
Signed-off-by: Andres Salomon <dilinger@collabora.co.uk>
Cc: Jordan Crouse <jordan@cosmicpenguin.net>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: john stultz <johnstul@us.ibm.com>
Cc: Chris Ball <cjb@laptop.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This is based on the old code on arch/x86/kernel/mfgpt_32.c, except it's
not x86 specific, it's modular, and it makes use of a PCI BAR rather than
a random MSR. Currently module unloading is not supported; it's uncertain
whether or not it can be made work with the hardware.
[akpm@linux-foundation.org: add X86 dependency]
Signed-off-by: Andres Salomon <dilinger@collabora.co.uk>
Cc: Jordan Crouse <jordan@cosmicpenguin.net>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: john stultz <johnstul@us.ibm.com>
Cc: Chris Ball <cjb@laptop.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This creates a CS5535/CS5536 GPIO driver which uses a gpio_chip backend
(allowing GPIO users to use the generic GPIO API if desired) while also
allowing architecture-specific users directly (via the cs5535_gpio_*
functions).
Tested on an OLPC machine. Some Leemotes also use CS5536 (with a mips
cpu), which is why this is in drivers/gpio rather than arch/x86.
Currently, it conflicts with older geode GPIO support; once MFGPT support
is reworked to also be more generic, the older geode code will be removed.
Signed-off-by: Andres Salomon <dilinger@collabora.co.uk>
Cc: Takashi Iwai <tiwai@suse.de>
Cc: Jordan Crouse <jordan@cosmicpenguin.net>
Cc: David Brownell <david-b@pacbell.net>
Reviewed-by: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
There are quite a few instances in the kernel of checks of pointers both
against NULL and against the errno range, handling both cases identically.
This additional helper function would simplify such code.
[akpm@linux-foundation.org: build fix]
Signed-off-by: Phil Carmody <ext-phil.2.carmody@nokia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
journal_info in task_struct is used in journaling file system only. So
introduce CONFIG_FS_JOURNAL_INFO and make it conditional.
Signed-off-by: Hiroshi Shimamoto <h-shimamoto@ct.jp.nec.com>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Add a printk_ratelimited statement expression macro that uses a per-call
ratelimit_state so that multiple subsystems output messages are not
suppressed by a global __ratelimit state.
[akpm@linux-foundation.org: coding-style fixes]
[akpm@linux-foundation.org: s/_rl/_ratelimited/g]
Signed-off-by: Joe Perches <joe@perches.com>
Cc: Naohiro Ooiwa <nooiwa@miraclelinux.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Hiroshi Shimamoto <h-shimamoto@ct.jp.nec.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
According to feature-removal-schedule.txt, it is the time to remove
print_fn_descriptor_symbol().
And a quick grep shows that it no longer has any callers.
Signed-off-by: WANG Cong <amwang@redhat.com>
Cc: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
rwsem_is_locked() tests ->activity without locks, so we should always keep
->activity consistent. However, the code in __rwsem_do_wake() breaks this
rule, it updates ->activity after _all_ readers waken up, this may give
some reader a wrong ->activity value, thus cause rwsem_is_locked() behaves
wrong.
Quote from Andrew:
"
- we have one or more processes sleeping in down_read(), waiting for access.
- we wake one or more processes up without altering ->activity
- they start to run and they do rwsem_is_locked(). This incorrectly
returns "false", because the waker process is still crunching away in
__rwsem_do_wake().
- the waker now alters ->activity, but it was too late.
"
So we need get a spinlock to protect this. And rwsem_is_locked() should
not block, thus we use spin_trylock_irqsave().
[akpm@linux-foundation.org: simplify code]
Reported-by: Brian Behlendorf <behlendorf1@llnl.gov>
Cc: Ben Woodard <bwoodard@llnl.gov>
Cc: David Howells <dhowells@redhat.com>
Signed-off-by: WANG Cong <amwang@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Don't initialize __print_once. Invert the test to reduce initialized
data.
defconfig before: $size vmlinux
text data bss dec hex filename
6976022 679572 1359668 9015262 898fde vmlinux
defconfig after: $size vmlinux
text data bss dec hex filename
6976006 679508 1359700 9015214 898fae vmlinux
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
If CONFIG_DYNAMIC_DEBUG is enabled and a source file has:
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
#include <linux/kernel.h>
dynamic_debug.h will duplicate KBUILD_MODNAME
in the output string.
Remove the use of KBUILD_MODNAME from the
output format string generated by dynamic_debug.h
If CONFIG_DYNAMIC_DEBUG is not enabled, no compile-time
check is done to printk/dev_printk arguments.
Add it.
Signed-off-by: Joe Perches <joe@perches.com>
Cc: Jason Baron <jbaron@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Commit 7086745309 ("printk_once(): use bool
for boolean flag") changed printk_once() to use bool instead of int for
its guard variable. Do the same change to WARN_ONCE() and WARN_ON_ONCE(),
for the same reasons.
This resulted in a reduction of 1462 bytes on a x86-64 defconfig:
text data bss dec hex filename
8101271 1207116 992764 10301151 9d2edf vmlinux.before
8100553 1207148 991988 10299689 9d2929 vmlinux.after
Signed-off-by: Cesar Eduardo Barros <cesarb@cesarb.net>
Cc: Roland Dreier <rolandd@cisco.com>
Cc: Daniel Walker <dwalker@fifo99.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The NOMMU code currently clears all anonymous mmapped memory. While this
is what we want in the default case, all memory allocation from userspace
under NOMMU has to go through this interface, including malloc() which is
allowed to return uninitialized memory. This can easily be a significant
performance penalty. So for constrained embedded systems were security is
irrelevant, allow people to avoid clearing memory unnecessarily.
This also alters the ELF-FDPIC binfmt such that it obtains uninitialised
memory for the brk and stack region.
Signed-off-by: Jie Zhang <jie.zhang@analog.com>
Signed-off-by: Robin Getz <rgetz@blackfin.uclinux.org>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: David Howells <dhowells@redhat.com>
Acked-by: Paul Mundt <lethal@linux-sh.org>
Acked-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The check code for CONFIG_SWAP is redundant, because there is a
non-CONFIG_SWAP version for PageSwapCache() which just returns 0.
Signed-off-by: Huang Shijie <shijie8@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The previous patch enables page migration of ksm pages, but that soon gets
into trouble: not surprising, since we're using the ksm page lock to lock
operations on its stable_node, but page migration switches the page whose
lock is to be used for that. Another layer of locking would fix it, but
do we need that yet?
Do we actually need page migration of ksm pages? Yes, memory hotremove
needs to offline sections of memory: and since we stopped allocating ksm
pages with GFP_HIGHUSER, they will tend to be GFP_HIGHUSER_MOVABLE
candidates for migration.
But KSM is currently unconscious of NUMA issues, happily merging pages
from different NUMA nodes: at present the rule must be, not to use
MADV_MERGEABLE where you care about NUMA. So no, NUMA page migration of
ksm pages does not make sense yet.
So, to complete support for ksm swapping we need to make hotremove safe.
ksm_memory_callback() take ksm_thread_mutex when MEM_GOING_OFFLINE and
release it when MEM_OFFLINE or MEM_CANCEL_OFFLINE. But if mapped pages
are freed before migration reaches them, stable_nodes may be left still
pointing to struct pages which have been removed from the system: the
stable_node needs to identify a page by pfn rather than page pointer, then
it can safely prune them when MEM_OFFLINE.
And make NUMA migration skip PageKsm pages where it skips PageReserved.
But it's only when we reach unmap_and_move() that the page lock is taken
and we can be sure that raised pagecount has prevented a PageAnon from
being upgraded: so add offlining arg to migrate_pages(), to migrate ksm
page when offlining (has sufficient locking) but reject it otherwise.
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Izik Eidus <ieidus@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Chris Wright <chrisw@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
A side-effect of making ksm pages swappable is that they have to be placed
on the LRUs: which then exposes them to isolate_lru_page() and hence to
page migration.
Add rmap_walk() for remove_migration_ptes() to use: rmap_walk_anon() and
rmap_walk_file() in rmap.c, but rmap_walk_ksm() in ksm.c. Perhaps some
consolidation with existing code is possible, but don't attempt that yet
(try_to_unmap needs to handle nonlinears, but migration pte removal does
not).
rmap_walk() is sadly less general than it appears: rmap_walk_anon(), like
remove_anon_migration_ptes() which it replaces, avoids calling
page_lock_anon_vma(), because that includes a page_mapped() test which
fails when all migration ptes are in place. That was valid when NUMA page
migration was introduced (holding mmap_sem provided the missing guarantee
that anon_vma's slab had not already been destroyed), but I believe not
valid in the memory hotremove case added since.
For now do the same as before, and consider the best way to fix that
unlikely race later on. When fixed, we can probably use rmap_walk() on
hwpoisoned ksm pages too: for now, they remain among hwpoison's various
exceptions (its PageKsm test comes before the page is locked, but its
page_lock_anon_vma fails safely if an anon gets upgraded).
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Izik Eidus <ieidus@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Chris Wright <chrisw@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
For full functionality, page_referenced_one() and try_to_unmap_one() need
to know the vma: to pass vma down to arch-dependent flushes, or to observe
VM_LOCKED or VM_EXEC. But KSM keeps no record of vma: nor can it, since
vmas get split and merged without its knowledge.
Instead, note page's anon_vma in its rmap_item when adding to stable tree:
all the vmas which might map that page are listed by its anon_vma.
page_referenced_ksm() and try_to_unmap_ksm() then traverse the anon_vma,
first to find the probable vma, that which matches rmap_item's mm; but if
that is not enough to locate all instances, traverse again to try the
others. This catches those occasions when fork has duplicated a pte of a
ksm page, but ksmd has not yet come around to assign it an rmap_item.
But each rmap_item in the stable tree which refers to an anon_vma needs to
take a reference to it. Andrea's anon_vma design cleverly avoided a
reference count (an anon_vma was free when its list of vmas was empty),
but KSM now needs to add that. Is a 32-bit count sufficient? I believe
so - the anon_vma is only free when both count is 0 and list is empty.
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Izik Eidus <ieidus@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Chris Wright <chrisw@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
