platform/kernel/linux-arm64.git
12 years agoMerge tags 'dt-for-linus', 'gpio-for-linus' and 'spi-for-linus' of git://git.secretla...
Linus Torvalds [Thu, 20 Dec 2012 04:26:16 +0000 (20:26 -0800)]
Merge tags 'dt-for-linus', 'gpio-for-linus' and 'spi-for-linus' of git://git.secretlab.ca/git/linux-2.6

Pull devicetree, gpio and spi bugfixes from Grant Likely:
 "Device tree v3.8 bug fix:
   - Fixes an undefined struct device build error and a missing symbol
     export.

  GPIO device driver bug fixes:
   - gpio/mvebu-gpio: Make mvebu-gpio depend on OF_CONFIG
   - gpio/ich: Add missing spinlock init

  SPI device driver bug fixes:
   - Most of this is bug fixes to the core code and the sh-hspi and
     s3c64xx device drivers.

   - There is also a patch here to add DT support to the Atmel driver.
     This one should have been in the first round, but I missed it.
     It's a low risk change contained within a single driver and the
     Atmel maintainer has requested it."

* tag 'dt-for-linus' of git://git.secretlab.ca/git/linux-2.6:
  of: define struct device in of_platform.h if !OF_DEVICE and !OF_ADDRESS
  of: Fix export of of_find_matching_node_and_match()

* tag 'gpio-for-linus' of git://git.secretlab.ca/git/linux-2.6:
  gpio/mvebu-gpio: Make mvebu-gpio depend on OF_CONFIG
  gpio/ich: Add missing spinlock init

* tag 'spi-for-linus' of git://git.secretlab.ca/git/linux-2.6:
  spi/sh-hspi: fix return value check in hspi_probe().
  spi: fix tegra SPI binding examples
  spi/atmel: add DT support
  of/spi: Fix SPI module loading by using proper "spi:" modalias prefixes.
  spi: Change FIFO flush operation and spi channel off
  spi: Keep chipselect assertion during one message

12 years agoMerge branch 'drm-next' of git://people.freedesktop.org/~airlied/linux
Linus Torvalds [Thu, 20 Dec 2012 04:24:25 +0000 (20:24 -0800)]
Merge branch 'drm-next' of git://people.freedesktop.org/~airlied/linux

Pull drm bugfix from Dave Airlie:
 "Just a single urgent regression fix, seeing a few wierd behaviours I'd
  like not to persist."

* 'drm-next' of git://people.freedesktop.org/~airlied/linux:
  drm/ttm: fix delayed ttm_bo_cleanup_refs_and_unlock delayed handling

12 years agoMerge tag 'random_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso...
Linus Torvalds [Thu, 20 Dec 2012 04:23:37 +0000 (20:23 -0800)]
Merge tag 'random_for_linus' of git://git./linux/kernel/git/tytso/random

Pull random updates from Ted Ts'o:
 "A few /dev/random improvements for the v3.8 merge window."

* tag 'random_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/random:
  random: Mix cputime from each thread that exits to the pool
  random: prime last_data value per fips requirements
  random: fix debug format strings
  random: make it possible to enable debugging without rebuild

12 years agogpio/mvebu-gpio: Make mvebu-gpio depend on OF_CONFIG
Gregory CLEMENT [Wed, 19 Dec 2012 11:31:11 +0000 (12:31 +0100)]
gpio/mvebu-gpio: Make mvebu-gpio depend on OF_CONFIG

When building without device tree support (for the mv78x00 SoCs for
example), the build failed because of_irq_count is undeclared. However
mvebu-gpio is not designed to build without device tree support. So
make it depends on OF_CONFIG, remove the #ifdef OF_CONFIG line and the
platform_device_id.

Tested on RD-78x00-mASA, DB-78460-BP, DB-88F6710-BP-DDR3,
DB-MV784MP-GP, Mirabox and OpenBlocks AX3.

Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
12 years agogpio/ich: Add missing spinlock init
Jean Delvare [Sun, 16 Dec 2012 20:31:40 +0000 (21:31 +0100)]
gpio/ich: Add missing spinlock init

As reported by CONFIG_DEBUG_SPINLOCK=y.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Acked-by: Peter Tyser <ptyser@xes-inc.com>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
12 years agodrm/ttm: fix delayed ttm_bo_cleanup_refs_and_unlock delayed handling
Maarten Lankhorst [Wed, 19 Dec 2012 17:21:10 +0000 (18:21 +0100)]
drm/ttm: fix delayed ttm_bo_cleanup_refs_and_unlock delayed handling

Fix regression introduced by 85b144f860176
"drm/ttm: call ttm_bo_cleanup_refs with reservation and lru lock held, v3"

Slowpath ttm_bo_cleanup_refs_and_unlock accidentally tried to increase
refcount on &bo->sync_obj instead of bo->sync_obj.

The compiler didn't complain since sync_obj_ref takes a void pointer,
so it was still valid c.

This could result in lockups, memory corruptions, and warnings like
these when graphics card VRAM usage is high:

------------[ cut here ]------------
WARNING: at include/linux/kref.h:42 radeon_fence_ref+0x2c/0x40()
Hardware name: System Product Name
Pid: 157, comm: X Not tainted 3.7.0-rc7-00520-g85b144f-dirty #174
Call Trace:
[<ffffffff81058c84>] ? warn_slowpath_common+0x74/0xb0
[<ffffffff8129273c>] ? radeon_fence_ref+0x2c/0x40
[<ffffffff8125e95c>] ? ttm_bo_cleanup_refs_and_unlock+0x18c/0x2d0
[<ffffffff8125f17c>] ? ttm_mem_evict_first+0x1dc/0x2a0
[<ffffffff81264452>] ? ttm_bo_man_get_node+0x62/0xb0
[<ffffffff8125f4ce>] ? ttm_bo_mem_space+0x28e/0x340
[<ffffffff8125fb0c>] ? ttm_bo_move_buffer+0xfc/0x170
[<ffffffff810de172>] ? kmem_cache_alloc+0xb2/0xc0
[<ffffffff8125fc15>] ? ttm_bo_validate+0x95/0x110
[<ffffffff8125ff7c>] ? ttm_bo_init+0x2ec/0x3b0
[<ffffffff8129419a>] ? radeon_bo_create+0x18a/0x200
[<ffffffff81293e80>] ? radeon_bo_clear_va+0x40/0x40
[<ffffffff812a5342>] ? radeon_gem_object_create+0x92/0x160
[<ffffffff812a575c>] ? radeon_gem_create_ioctl+0x6c/0x150
[<ffffffff812a529f>] ? radeon_gem_object_free+0x2f/0x40
[<ffffffff81246b60>] ? drm_ioctl+0x420/0x4f0
[<ffffffff812a56f0>] ? radeon_gem_pwrite_ioctl+0x20/0x20
[<ffffffff810f53a4>] ? do_vfs_ioctl+0x2e4/0x4e0
[<ffffffff810e5588>] ? vfs_read+0x118/0x160
[<ffffffff810f55ec>] ? sys_ioctl+0x4c/0xa0
[<ffffffff810e5851>] ? sys_read+0x51/0xa0
[<ffffffff814b0612>] ? system_call_fastpath+0x16/0x1b

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Reported-by: Markus Trippelsdorf <markus@trippelsdorf.de>
Acked-by: Paul Menzel <paulepanter@users.sourceforge.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
12 years agoMerge tag 'xtensa-20121218' of git://github.com/czankel/xtensa-linux
Linus Torvalds [Wed, 19 Dec 2012 21:05:22 +0000 (13:05 -0800)]
Merge tag 'xtensa-20121218' of git://github.com/czankel/xtensa-linux

Pull Xtensa patchset from Chris Zankel:
 "This contains support of device trees, many fixes, and code clean-ups"

* tag 'xtensa-20121218' of git://github.com/czankel/xtensa-linux: (33 commits)
  xtensa: don't try to build DTB when OF is disabled
  xtensa: set the correct ethernet address for xtfpga
  xtensa: clean up files to make them code-style compliant
  xtensa: provide endianness macro for sparse
  xtensa: fix RASID SR initialization
  xtensa: initialize CPENABLE SR when core has one
  xtensa: reset all timers on initialization
  Use for_each_compatible_node() macro.
  xtensa: add XTFPGA DTS
  xtensa: add support for the XTFPGA boards
  xtensa: add device trees support
  xtensa: add IRQ domains support
  xtensa: add U-Boot image support (uImage).
  xtensa: clean up boot make rules
  xtensa: fix mb and wmb definitions
  xtensa: add s32c1i-based spinlock implementations
  xtensa: add s32c1i-based bitops implementations
  xtensa: add s32c1i-based atomic ops implementations
  xtensa: add s32c1i sanity check
  xtensa: add trap_set_handler function
  ...

12 years agoMerge branch 'x86/nuke386' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Linus Torvalds [Wed, 19 Dec 2012 21:02:23 +0000 (13:02 -0800)]
Merge branch 'x86/nuke386' of git://git./linux/kernel/git/tip/tip

Pull one final 386 removal patch from Peter Anvin.

IRQ 13 FPU error handling is gone.  That was not one of the proudest
moments in PC history.

* 'x86/nuke386' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86, 386 removal: Remove support for IRQ 13 FPU error reporting

12 years agoMerge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Wed, 19 Dec 2012 20:56:42 +0000 (12:56 -0800)]
Merge branch 'x86-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull small x86 fixes from Peter Anvin:
 "A collection of very small fixes, mostly pure documentation."

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86, doc: Document that bootloader ID 4 is used also by iPXE
  x86, doc: Add a formal bootloader ID for kexec-tools
  x86, 8042: Enable A20 using KBC to fix S3 resume on some MSI laptops

12 years agoMerge tag 'for-linus-20121219' of git://git.infradead.org/linux-mtd
Linus Torvalds [Wed, 19 Dec 2012 20:47:41 +0000 (12:47 -0800)]
Merge tag 'for-linus-20121219' of git://git.infradead.org/linux-mtd

Pull MTD updates from David Woodhouse:
 - Various cleanups especially in NAND tests
 - Add support for NAND flash on BCMA bus
 - DT support for sh_flctl and denali NAND drivers
 - Kill obsolete/superceded drivers (fortunet, nomadik_nand)
 - Fix JFFS2 locking bug in ENOMEM failure path
 - New SPI flash chips, as usual
 - Support writing in 'reliable mode' for DiskOnChip G4
 - Debugfs support in nandsim

* tag 'for-linus-20121219' of git://git.infradead.org/linux-mtd: (96 commits)
  mtd: nand: typo in nand_id_has_period() comments
  mtd: nand/gpio: use io{read,write}*_rep accessors
  mtd: block2mtd: throttle writes by calling balance_dirty_pages_ratelimited.
  mtd: nand: gpmi: reset BCH earlier, too, to avoid NAND startup problems
  mtd: nand/docg4: fix and improve read of factory bbt
  mtd: nand/docg4: reserve bb marker area in ecclayout
  mtd: nand/docg4: add support for writing in reliable mode
  mtd: mxc_nand: reorder part_probes to let cmdline override other sources
  mtd: mxc_nand: fix unbalanced clk_disable() in error path
  mtd: nandsim: Introduce debugfs infrastructure
  mtd: physmap_of: error checking to prevent a NULL pointer dereference
  mtg: docg3: potential divide by zero in doc_write_oob()
  mtd: bcm47xxnflash: writing support
  mtd: tests/read: initialize buffer for whole next page
  mtd: at91: atmel_nand: return bit flips for the PMECC read_page()
  mtd: fix recovery after failed write-buffer operation in cfi_cmdset_0002.c
  mtd: nand: onfi need to be probed in 8 bits mode
  mtd: nand: add NAND_BUSWIDTH_AUTO to autodetect bus width
  mtd: nand: print flash size during detection
  mted: nand_wait_ready timeout fix
  ...

12 years agowatchdog: Fix disable/enable regression
Bjørn Mork [Wed, 19 Dec 2012 19:51:31 +0000 (20:51 +0100)]
watchdog: Fix disable/enable regression

Commit 8d4516904b39 ("watchdog: Fix CPU hotplug regression") causes an
oops or hard lockup when doing

 echo 0 > /proc/sys/kernel/nmi_watchdog
 echo 1 > /proc/sys/kernel/nmi_watchdog

and the kernel is booted with nmi_watchdog=1 (default)

Running laptop-mode-tools and disconnecting/connecting AC power will
cause this to trigger, making it a common failure scenario on laptops.

Instead of bailing out of watchdog_disable() when !watchdog_enabled we
can initialize the hrtimer regardless of watchdog_enabled status.  This
makes it safe to call watchdog_disable() in the nmi_watchdog=0 case,
without the negative effect on the enabled => disabled => enabled case.

All these tests pass with this patch:
- nmi_watchdog=1
  echo 0 > /proc/sys/kernel/nmi_watchdog
  echo 1 > /proc/sys/kernel/nmi_watchdog

- nmi_watchdog=0
  echo 0 > /sys/devices/system/cpu/cpu1/online

- nmi_watchdog=0
  echo mem > /sys/power/state

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=51661

Cc: <stable@vger.kernel.org> # v3.7
Cc: Norbert Warmuth <nwarmuth@t-online.de>
Cc: Joseph Salisbury <joseph.salisbury@canonical.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoMerge tag 'for-3.8-rc1' of git://gitorious.org/linux-pwm/linux-pwm
Linus Torvalds [Wed, 19 Dec 2012 16:19:07 +0000 (08:19 -0800)]
Merge tag 'for-3.8-rc1' of git://gitorious.org/linux-pwm/linux-pwm

Pull pwm changes from Thierry Reding:
 "A new driver has been added for the SPEAr platform and the
  TWL4030/6030 driver has been replaced by two drivers that control the
  regular PWMs and the PWM driven LEDs provided by the chips.

  The vt8500, tiecap, tiehrpwm, i.MX, LPC32xx and Samsung drivers have
  all been improved and the device tree bindings now support the PWM
  signal polarity."

Fix up trivial conflicts due to __devinit/exit removal.

* tag 'for-3.8-rc1' of git://gitorious.org/linux-pwm/linux-pwm: (21 commits)
  pwm: samsung: add missing s3c->pwm_id assignment
  pwm: lpc32xx: Set the chip base for dynamic allocation
  pwm: lpc32xx: Properly disable the clock on device removal
  pwm: lpc32xx: Fix the PWM polarity
  pwm: i.MX: eliminate build warning
  pwm: Export of_pwm_xlate_with_flags()
  pwm: Remove pwm-twl6030 driver
  pwm: New driver to support PWM driven LEDs on TWL4030/6030 series of PMICs
  pwm: New driver to support PWMs on TWL4030/6030 series of PMICs
  pwm: pwm-tiehrpwm: pinctrl support
  pwm: tiehrpwm: Add device-tree binding
  pwm: pwm-tiehrpwm: Adding TBCLK gating support.
  pwm: pwm-tiecap: pinctrl support
  pwm: tiecap: Add device-tree binding
  pwm: Add TI PWM subsystem driver
  pwm: Device tree support for PWM polarity
  pwm: vt8500: Ensure PWM clock is enabled during pwm_config
  pwm: vt8500: Fix build error
  pwm: spear: Staticize spear_pwm_config()
  pwm: Add SPEAr PWM chip driver support
  ...

12 years agoof: define struct device in of_platform.h if !OF_DEVICE and !OF_ADDRESS
Jonas Gorski [Wed, 19 Dec 2012 15:10:09 +0000 (09:10 -0600)]
of: define struct device in of_platform.h if !OF_DEVICE and !OF_ADDRESS

Fixes the following warning:

include/linux/of_platform.h:106:13: warning: 'struct device' declared
inside parameter list [enabled by default]
include/linux/of_platform.h:106:13: warning: its scope is only this
definition or declaration, which is probably not what you want [enabled
by default]

Signed-off-by: Jonas Gorski <jogo@openwrt.org>
Signed-off-by: Rob Herring <robherring2@gmail.com>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
12 years agoMerge tag 'for-v3.8-part2' of git://git.infradead.org/battery-2.6
Linus Torvalds [Wed, 19 Dec 2012 16:14:08 +0000 (08:14 -0800)]
Merge tag 'for-v3.8-part2' of git://git.infradead.org/battery-2.6

Pull battery update, part 2, from Anton Vorontsov:
 "These are left overs that I didn't have time to review/apply before
  the merge window opened.  I didn't want to "spoil" the first pull
  request with these late patches, so they were not included:

   - A small patch for the RX51 OMAP board (Nokia N900 phone), the patch
     creates a battery monitor device instance, so that it can be
     probed.  It was acked by the OMAP maintainer;

   - A couple of late bug fixes for the charger-manager: corrects corner
     cases for the battery full handling."

* tag 'for-v3.8-part2' of git://git.infradead.org/battery-2.6:
  charger-manager: Fix bug when check dropped voltage after fullbatt event
  charger-manager: Fix bug related to checking fully charged state of battery
  ARM: OMAP: rx51: Register platform device for rx51_battery

12 years agoMerge tag 'modules-next-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 19 Dec 2012 15:55:08 +0000 (07:55 -0800)]
Merge tag 'modules-next-for-linus' of git://git./linux/kernel/git/rusty/linux

Pull module update from Rusty Russell:
 "Nothing all that exciting; a new module-from-fd syscall for those who
  want to verify the source of the module (ChromeOS) and/or use standard
  IMA on it or other security hooks."

* tag 'modules-next-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux:
  MODSIGN: Fix kbuild output when using default extra_certificates
  MODSIGN: Avoid using .incbin in C source
  modules: don't hand 0 to vmalloc.
  module: Remove a extra null character at the top of module->strtab.
  ASN.1: Use the ASN1_LONG_TAG and ASN1_INDEFINITE_LENGTH constants
  ASN.1: Define indefinite length marker constant
  moduleparam: use __UNIQUE_ID()
  __UNIQUE_ID()
  MODSIGN: Add modules_sign make target
  powerpc: add finit_module syscall.
  ima: support new kernel module syscall
  add finit_module syscall to asm-generic
  ARM: add finit_module syscall to ARM
  security: introduce kernel_module_from_file hook
  module: add flags arg to sys_finit_module()
  module: add syscall to load module from fd

12 years agoMerge tag 'byteswap-for-linus-20121219' of git://git.infradead.org/users/dwmw2/byteswap
Linus Torvalds [Wed, 19 Dec 2012 15:52:48 +0000 (07:52 -0800)]
Merge tag 'byteswap-for-linus-20121219' of git://git.infradead.org/users/dwmw2/byteswap

Pull preparatory gcc intrisics bswap patch from David Woodhouse:
 "This single patch is effectively a no-op for now.  It enables
  architectures to opt in to using GCC's __builtin_bswapXX() intrinsics
  for byteswapping, and if we merge this now then the architecture
  maintainers can enable it for their arch during the next cycle without
  dependency issues.

  It's worth making it a par-arch opt-in, because although in *theory*
  the compiler should never do worse than hand-coded assembler (and of
  course it also ought to do a lot better on platforms like Atom and
  PowerPC which have load-and-swap or store-and-swap instructions), that
  isn't always the case.  See

     http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46453

  for example."

* tag 'byteswap-for-linus-20121219' of git://git.infradead.org/users/dwmw2/byteswap:
  byteorder: allow arch to opt to use GCC intrinsics for byteswapping

12 years agoblk: avoid divide-by-zero with zero discard granularity
Linus Torvalds [Wed, 19 Dec 2012 15:18:35 +0000 (07:18 -0800)]
blk: avoid divide-by-zero with zero discard granularity

Commit 8dd2cb7e880d ("block: discard granularity might not be power of
2") changed a couple of 'binary and' operations into modulus operations.
Which turned the harmless case of a zero discard_granularity into a
possible divide-by-zero.

The code also had a much more subtle bug: it was doing the modulus of a
value in bytes using 'sector_t'.  That was always conceptually wrong,
but didn't actually matter back when the code assumed a power-of-two
granularity: we only looked at the low bits anyway.

But with potentially arbitrary sector numbers, using a 'sector_t' to
express bytes is very very wrong: depending on configuration it limits
the starting offset of the device to just 32 bits, and any overflow
would result in a wrong value if the modulus wasn't a power-of-two.

So re-write the code to not only protect against the divide-by-zero, but
to do the starting sector arithmetic in sectors, and using the proper
types.

[ For any mathematicians out there: it also looks monumentally stupid to
  do the 'modulo granularity' operation *twice*, never mind having a "+
  granularity" in the second modulus op.

  But that's the easiest way to avoid negative values or overflow, and
  it is how the original code was done. ]

Reported-by: Ingo Molnar <mingo@kernel.org>
Reported-by: Doug Anderson <dianders@chromium.org>
Cc: Neil Brown <neilb@suse.de>
Cc: Shaohua Li <shli@fusionio.com>
Acked-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agospi/sh-hspi: fix return value check in hspi_probe().
Cyril Roelandt [Wed, 12 Dec 2012 00:24:54 +0000 (01:24 +0100)]
spi/sh-hspi: fix return value check in hspi_probe().

According to its documentation, clk_get() returns a "valid IS_ERR() condition
containing errno", so we should call IS_ERR() rather than a NULL check.

Signed-off-by: Cyril Roelandt <tipecaml@gmail.com>
Acked-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
12 years agospi: fix tegra SPI binding examples
Allen Martin [Fri, 14 Dec 2012 19:05:12 +0000 (11:05 -0800)]
spi: fix tegra SPI binding examples

Fix name of slink binding and address of sflash example to make it
self consistent.

Signed-off-by: Allen Martin <amartin@nvidia.com>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
12 years agoof: Fix export of of_find_matching_node_and_match()
Grant Likely [Wed, 19 Dec 2012 10:45:36 +0000 (10:45 +0000)]
of: Fix export of of_find_matching_node_and_match()

Commit 50c8af4cf9, "of: introduce for_each_matching_node_and_match()"
renamed of_find_matching_node() to of_find_matching_node_and_match() and
created a new static inline of_find_matching_node() wrapper around the
new name. However, the change neglected to change the EXPORT_SYMBOL()
reference causing build errors for modules.

This patch fixes the EXPORT_SYMBOL() statement. Discovered on a PowerPC
Efika build with the mpc52xx_uart driver being built as a module.

Reported-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
12 years agoxtensa: don't try to build DTB when OF is disabled
Max Filippov [Mon, 10 Dec 2012 21:26:21 +0000 (01:26 +0400)]
xtensa: don't try to build DTB when OF is disabled

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: set the correct ethernet address for xtfpga
Chris Zankel [Thu, 29 Nov 2012 00:52:09 +0000 (16:52 -0800)]
xtensa: set the correct ethernet address for xtfpga

The last byte of the mac address is determined by a DIP switch, so
update the OF property with that address.

Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: clean up files to make them code-style compliant
Chris Zankel [Thu, 29 Nov 2012 00:53:51 +0000 (16:53 -0800)]
xtensa: clean up files to make them code-style compliant

Remove heading and trailing spaces, trim trailing lines, and wrap lines
that are longer than 80 characters.

Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: provide endianness macro for sparse
Max Filippov [Mon, 10 Dec 2012 21:26:25 +0000 (01:26 +0400)]
xtensa: provide endianness macro for sparse

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: fix RASID SR initialization
Max Filippov [Mon, 10 Dec 2012 21:26:24 +0000 (01:26 +0400)]
xtensa: fix RASID SR initialization

set_rasid_register accepts new RASID SR value, but ASID_USER_FIRST is
ASID value for the ring 1; RASID value is made by ASID_INSERT macro.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: initialize CPENABLE SR when core has one
Max Filippov [Mon, 10 Dec 2012 21:26:23 +0000 (01:26 +0400)]
xtensa: initialize CPENABLE SR when core has one

XCHAL_CP_NUM is defined in variant/tie.h and it is not included by
head.S, leaving CPENABLE register uninitialised. XCHAL_HAVE_CP is
defined in variant/core.h to 1 when core has CPENABLE SR.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: reset all timers on initialization
Max Filippov [Mon, 10 Dec 2012 21:26:22 +0000 (01:26 +0400)]
xtensa: reset all timers on initialization

There are XCHAL_NUM_TIMERS, reset them all.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoUse for_each_compatible_node() macro.
Wei Yongjun [Mon, 3 Dec 2012 08:35:11 +0000 (00:35 -0800)]
Use for_each_compatible_node() macro.

Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Acked-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: add XTFPGA DTS
Max Filippov [Sat, 3 Nov 2012 08:57:52 +0000 (12:57 +0400)]
xtensa: add XTFPGA DTS

Add common XTFPGA parts as *.dtsi (base board, flash) and DTS for LX60
and for ML605.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: add support for the XTFPGA boards
Max Filippov [Mon, 5 Nov 2012 03:37:14 +0000 (07:37 +0400)]
xtensa: add support for the XTFPGA boards

The Avnet LX60/LX110/LX200 board is an FPGA board that can be configured with
an Xtensa processor and an OpenCores Ethernet device.

Signed-off-by: Chris Zankel <chris@zankel.net>
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: add device trees support
Max Filippov [Sat, 3 Nov 2012 20:30:13 +0000 (00:30 +0400)]
xtensa: add device trees support

Device trees allow specification of hardware topology and device
parameters at runtime instead of hard-coding them in platform setup
code. This allows running single binary kernel on a range of compatible
boards.

New boot parameters tag BP_TAG_FDT is allocated and a pointer to flat
device tree is passed in it.

Note that current interrupt mapping scheme uses single cell for
interrupt identification. That means that IRQ numbers used in DTS must
be CPU internal IRQ numbers, not external. It is possible to extend
interrupt identification to two cells, and use second cell to tell
external IRQ numbers form internal. That would allow to use single DTS
on multiple boards with different mapping of external IRQ numbers.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: add IRQ domains support
Max Filippov [Sat, 3 Nov 2012 20:29:12 +0000 (00:29 +0400)]
xtensa: add IRQ domains support

IRQ domains provide a mechanism for conversion of linux IRQ numbers to
hardware IRQ numbers and vice versus. It is used by OpenFirmware for
linking device tree objects to their respective interrupt controllers.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: add U-Boot image support (uImage).
Max Filippov [Mon, 5 Nov 2012 03:36:00 +0000 (07:36 +0400)]
xtensa: add U-Boot image support (uImage).

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: clean up boot make rules
Max Filippov [Mon, 5 Nov 2012 05:10:00 +0000 (09:10 +0400)]
xtensa: clean up boot make rules

- remove duplicate rules for binary and packed image
- use predefined macros for ld/objcopy/gzip
- remove build-id section from bootable elf image

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: fix mb and wmb definitions
Max Filippov [Sat, 24 Nov 2012 02:22:08 +0000 (06:22 +0400)]
xtensa: fix mb and wmb definitions

Define mb and wmb as memw to force memory barrier.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: add s32c1i-based spinlock implementations
Max Filippov [Sun, 11 Nov 2012 03:01:21 +0000 (07:01 +0400)]
xtensa: add s32c1i-based spinlock implementations

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: add s32c1i-based bitops implementations
Max Filippov [Sun, 11 Nov 2012 01:47:25 +0000 (05:47 +0400)]
xtensa: add s32c1i-based bitops implementations

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: add s32c1i-based atomic ops implementations
Max Filippov [Sat, 10 Nov 2012 23:30:02 +0000 (03:30 +0400)]
xtensa: add s32c1i-based atomic ops implementations

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: add s32c1i sanity check
Max Filippov [Wed, 28 Nov 2012 07:33:02 +0000 (11:33 +0400)]
xtensa: add s32c1i sanity check

Add a brief sanity test of S32C1I functionality.  This instruction
is needed by the kernel and userland as part of the base ABI
(including GCC atomic builtins, certain threading packages, future
atomic support in the C++ standard, etc).  However, correct operation
of this instruction requires some cooperation by hardware external to
the processor (such as bus bridge, bus fabric, or memory controller).
Minimally exercising this mechanism and reporting explicit status
early in the boot process is helpful to chip vendors using the Linux
kernel as a benchmark of correctness of hardware.

As it turns out, S32C1I is not exercised by the kernel and by uClibc
based userland as of early June 2008.  This is expected to change
soon as both incorporate more recent open source developments.

Signed-off-by: Marc Gauthier <marc@tensilica.com>
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: add trap_set_handler function
Max Filippov [Mon, 19 Nov 2012 04:30:15 +0000 (08:30 +0400)]
xtensa: add trap_set_handler function

trap_set_handler sets new C-handler in the exception table and returns
previous handler.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: initialize atomctl SR
Max Filippov [Mon, 19 Nov 2012 03:00:41 +0000 (07:00 +0400)]
xtensa: initialize atomctl SR

In order to use S32C1I instruction on cores with ATOMCTL SR the register
must be properly initialized.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: save and restore scompare1 SR on kernel entry
Max Filippov [Thu, 15 Nov 2012 02:25:48 +0000 (06:25 +0400)]
xtensa: save and restore scompare1 SR on kernel entry

Although scompare1 may be saved/restored by xchal_ncp_{load,store}
macros, explicit save/restore of registers manipulated by the kernel
itself is considered more correct.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: display s32c1i feature flag in cpuinfo
Max Filippov [Sun, 11 Nov 2012 00:44:22 +0000 (04:44 +0400)]
xtensa: display s32c1i feature flag in cpuinfo

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: fix CPU cache flags formatting
Max Filippov [Sat, 10 Nov 2012 21:29:10 +0000 (01:29 +0400)]
xtensa: fix CPU cache flags formatting

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: properly fix missing compiler barrier in simcall
Max Filippov [Mon, 22 Oct 2012 04:42:39 +0000 (08:42 +0400)]
xtensa: properly fix missing compiler barrier in simcall

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: fix build warning for arch/xtensa/mm/tlb.c
Max Filippov [Mon, 5 Nov 2012 03:44:03 +0000 (07:44 +0400)]
xtensa: fix build warning for arch/xtensa/mm/tlb.c

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: provide DMA_ERROR_CODE definition
Max Filippov [Thu, 1 Nov 2012 14:38:27 +0000 (18:38 +0400)]
xtensa: provide DMA_ERROR_CODE definition

This fixes the following allmodconfig build error:

drivers/uio/uio_dmem_genirq.c:95:18: error: 'DMA_ERROR_CODE' undeclared (first use in this function)
drivers/uio/uio_dmem_genirq.c:238:18: error: 'DMA_ERROR_CODE' undeclared (first use in this function)
make[3]: *** [drivers/uio/uio_dmem_genirq.o] Error 1

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: ISS: add BASE_BAUD definition to serial.h
Max Filippov [Wed, 31 Oct 2012 08:37:04 +0000 (12:37 +0400)]
xtensa: ISS: add BASE_BAUD definition to serial.h

This fixes the following build error in allyesconfig:

drivers/tty/serial/8250/8250_early.c: In function 'parse_options':
drivers/tty/serial/8250/8250_early.c:160:18: error: 'BASE_BAUD' undeclared (first use in this function)

Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: provide proper assembler function boundaries with ENDPROC()
Chris Zankel [Sat, 17 Nov 2012 00:16:20 +0000 (16:16 -0800)]
xtensa: provide proper assembler function boundaries with ENDPROC()

Use ENDPROC() to mark the end of assembler functions.

Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: make DoubleExceptionVector literals fit the gap
Max Filippov [Thu, 9 Aug 2012 01:25:35 +0000 (05:25 +0400)]
xtensa: make DoubleExceptionVector literals fit the gap

Manually load references to exc_table from the explicit literal in order
to fit DoubleExceptionVector.literals into the available 16-byte gap
before DoubleExceptionVector.text in the absence of link time
relaxation. Without this fix DoubleExceptionVector.literal section
overlaps DoubleExceptionVector.text section in the linked vmlinux image.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: add config option to disable linker relaxation
Chris Zankel [Tue, 13 Nov 2012 23:16:36 +0000 (15:16 -0800)]
xtensa: add config option to disable linker relaxation

The default linker behavior is to optimize identical literal values and
remove unnecessary overhead from assembler-generated "longcall" sequences
to reduce code size. Provide an option to disable this behavior to improve
compile time.

Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa: unbalanced parentheses
Nicolas Kaiser [Sat, 7 May 2011 18:55:42 +0000 (20:55 +0200)]
xtensa: unbalanced parentheses

Signed-off-by: Nicolas Kaiser <nikai@nikai.net>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoxtensa:fix the incompatible pointer type warning in time.c
Wanlong Gao [Wed, 1 Jun 2011 14:37:43 +0000 (22:37 +0800)]
xtensa:fix the incompatible pointer type warning in time.c

Fix the definition of the function ccount_read to be compatible
to the member read of the structure clocksource.

Signed-off-by: Wanlong Gao <wanlong.gao@gmail.com>
Signed-off-by: Chris Zankel <chris@zankel.net>
12 years agoMerge branch 'i2c-embedded/for-next' of git://git.pengutronix.de/git/wsa/linux
Linus Torvalds [Wed, 19 Dec 2012 00:51:10 +0000 (16:51 -0800)]
Merge branch 'i2c-embedded/for-next' of git://git.pengutronix.de/git/wsa/linux

Pull i2c-embedded changes from Wolfram Sang:
 - CBUS driver (an I2C variant)
 - continued rework of the omap driver
 - s3c2410 gets lots of fixes and gains pinctrl support
 - at91 gains DMA support
 - the GPIO muxer gains devicetree probing
 - typical fixes and additions all over

* 'i2c-embedded/for-next' of git://git.pengutronix.de/git/wsa/linux: (45 commits)
  i2c: omap: Remove the OMAP_I2C_FLAG_RESET_REGS_POSTIDLE flag
  i2c: at91: add dma support
  i2c: at91: change struct members indentation
  i2c: at91: fix compilation warning
  i2c: mxs: Do not disable the I2C SMBus quick mode
  i2c: mxs: Handle i2c DMA failure properly
  i2c: s3c2410: Remove recently introduced performance overheads
  i2c: ocores: Move grlib set/get functions into #ifdef CONFIG_OF block
  i2c: s3c2410: Add fix for i2c suspend/resume
  i2c: s3c2410: Fix code to free gpios
  i2c: i2c-cbus-gpio: introduce driver
  i2c: ocores: Add support for the GRLIB port of the controller and use function pointers for getreg and setreg functions
  i2c: ocores: Add irq support for sparc
  i2c: omap: Move the remove constraint
  ARM: dts: cfa10049: Add the i2c muxer buses to the CFA-10049
  i2c: s3c2410: do not special case HDMIPHY stuck bus detection
  i2c: s3c2410: use exponential back off while polling for bus idle
  i2c: s3c2410: do not generate STOP for QUIRK_HDMIPHY
  i2c: s3c2410: grab adapter lock while changing i2c clock
  i2c: s3c2410: Add support for pinctrl
  ...

12 years agoMerge branch 'akpm' (more patches from Andrew)
Linus Torvalds [Tue, 18 Dec 2012 23:08:12 +0000 (15:08 -0800)]
Merge branch 'akpm' (more patches from Andrew)

Merge patches from Andrew Morton:
 "Most of the rest of MM, plus a few dribs and drabs.

  I still have quite a few irritating patches left around: ones with
  dubious testing results, lack of review, ones which should have gone
  via maintainer trees but the maintainers are slack, etc.

  I need to be more activist in getting these things wrapped up outside
  the merge window, but they're such a PITA."

* emailed patches from Andrew Morton <akpm@linux-foundation.org>: (48 commits)
  mm/vmscan.c: avoid possible deadlock caused by too_many_isolated()
  vmscan: comment too_many_isolated()
  mm/kmemleak.c: remove obsolete simple_strtoul
  mm/memory_hotplug.c: improve comments
  mm/hugetlb: create hugetlb cgroup file in hugetlb_init
  mm/mprotect.c: coding-style cleanups
  Documentation: ABI: /sys/devices/system/node/
  slub: drop mutex before deleting sysfs entry
  memcg: add comments clarifying aspects of cache attribute propagation
  kmem: add slab-specific documentation about the kmem controller
  slub: slub-specific propagation changes
  slab: propagate tunable values
  memcg: aggregate memcg cache values in slabinfo
  memcg/sl[au]b: shrink dead caches
  memcg/sl[au]b: track all the memcg children of a kmem_cache
  memcg: destroy memcg caches
  sl[au]b: allocate objects from memcg cache
  sl[au]b: always get the cache from its page in kmem_cache_free()
  memcg: skip memcg kmem allocations in specified code regions
  memcg: infrastructure to match an allocation to the right cache
  ...

12 years agoMerge tag 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck...
Linus Torvalds [Tue, 18 Dec 2012 23:06:11 +0000 (15:06 -0800)]
Merge tag 'hwmon-for-linus' of git://git./linux/kernel/git/groeck/linux-staging

Pull hwmon fixlet from Guenter Roeck:
 "Fix fallout from __devexit removal"

* tag 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
  hwmon: (twl4030-madc-hwmon) Fix warning message caused by removal of __devexit

12 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf/linux-tile
Linus Torvalds [Tue, 18 Dec 2012 23:05:30 +0000 (15:05 -0800)]
Merge git://git./linux/kernel/git/cmetcalf/linux-tile

Pull tile updates from Chris Metcalf:
 "These are a smattering of minor changes from Tilera and other folks,
  mostly in the ptrace area."

* git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf/linux-tile:
  arch/tile: set CORE_DUMP_USE_REGSET on tile
  arch/tile: implement arch_ptrace using user_regset on tile
  arch/tile: implement user_regset interface on tile
  arch/tile: clean up tile-specific PTRACE_SETOPTIONS
  arch/tile: provide PT_FLAGS_COMPAT value in pt_regs
  tile/PCI: use for_each_pci_dev to simplify the code
  tilegx: remove __init from pci fixup hook

12 years agomm/vmscan.c: avoid possible deadlock caused by too_many_isolated()
Fengguang Wu [Tue, 18 Dec 2012 22:23:31 +0000 (14:23 -0800)]
mm/vmscan.c: avoid possible deadlock caused by too_many_isolated()

Neil found that if too_many_isolated() returns true while performing
direct reclaim we can end up waiting for other threads to complete their
direct reclaim.  If those threads are allowed to enter the FS or IO to
free memory, but this thread is not, then it is possible that those
threads will be waiting on this thread and so we get a circular deadlock.

some task enters direct reclaim with GFP_KERNEL
  => too_many_isolated() false
    => vmscan and run into dirty pages
      => pageout()
        => take some FS lock
          => fs/block code does GFP_NOIO allocation
            => enter direct reclaim again
              => too_many_isolated() true
                => waiting for others to progress, however the other
                   tasks may be circular waiting for the FS lock..

The fix is to let !__GFP_IO and !__GFP_FS direct reclaims enjoy higher
priority than normal ones, by lowering the throttle threshold for the
latter.

Allowing ~1/8 isolated pages in normal is large enough.  For example, for
a 1GB LRU list, that's ~128MB isolated pages, or 1k blocked tasks (each
isolates 32 4KB pages), or 64 blocked tasks per logical CPU (assuming 16
logical CPUs per NUMA node).  So it's not likely some CPU goes idle
waiting (when it could make progress) because of this limit: there are
much more sleeping reclaim tasks than the number of CPU, so the task may
well be blocked by some low level queue/lock anyway.

Now !GFP_IOFS reclaims won't be waiting for GFP_IOFS reclaims to progress.
 They will be blocked only when there are too many concurrent !GFP_IOFS
reclaims, however that's very unlikely because the IO-less direct reclaims
is able to progress much more faster, and they won't deadlock each other.
The threshold is raised high enough for them, so that there can be
sufficient parallel progress of !GFP_IOFS reclaims.

[akpm@linux-foundation.org: tweak comment]
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Cc: Torsten Kaiser <just.for.lkml@googlemail.com>
Tested-by: NeilBrown <neilb@suse.de>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: Rik van Riel <riel@redhat.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agovmscan: comment too_many_isolated()
Fengguang Wu [Tue, 18 Dec 2012 22:23:28 +0000 (14:23 -0800)]
vmscan: comment too_many_isolated()

Comment "Why it's doing so" rather than "What it does" as proposed by
Andrew Morton.

Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Reviewed-by: Rik van Riel <riel@redhat.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm/kmemleak.c: remove obsolete simple_strtoul
Abhijit Pawar [Tue, 18 Dec 2012 22:23:27 +0000 (14:23 -0800)]
mm/kmemleak.c: remove obsolete simple_strtoul

Replace the obsolete simple_strtoul() with kstrtoul().

Signed-off-by: Abhijit Pawar <abhi.c.pawar@gmail.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm/memory_hotplug.c: improve comments
Tang Chen [Tue, 18 Dec 2012 22:23:24 +0000 (14:23 -0800)]
mm/memory_hotplug.c: improve comments

Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Cc: Jiang Liu <jiang.liu@huawei.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Wen Congyang <wency@cn.fujitsu.com>
Cc: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm/hugetlb: create hugetlb cgroup file in hugetlb_init
Jianguo Wu [Tue, 18 Dec 2012 22:23:19 +0000 (14:23 -0800)]
mm/hugetlb: create hugetlb cgroup file in hugetlb_init

Build kernel with CONFIG_HUGETLBFS=y,CONFIG_HUGETLB_PAGE=y and
CONFIG_CGROUP_HUGETLB=y, then specify hugepagesz=xx boot option, system
will fail to boot.

This failure is caused by following code path:

  setup_hugepagesz
    hugetlb_add_hstate
      hugetlb_cgroup_file_init
        cgroup_add_cftypes
          kzalloc <--slab is *not available* yet

For this path, slab is not available yet, so memory allocated will be
failed, and cause WARN_ON() in hugetlb_cgroup_file_init().

So I move hugetlb_cgroup_file_init() into hugetlb_init().

[akpm@linux-foundation.org: tweak coding-style, remove pointless __init on inlined function]
[akpm@linux-foundation.org: fix warning]
Signed-off-by: Jianguo Wu <wujianguo@huawei.com>
Signed-off-by: Jiang Liu <jiang.liu@huawei.com>
Reviewed-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm/mprotect.c: coding-style cleanups
Andrew Morton [Tue, 18 Dec 2012 22:23:17 +0000 (14:23 -0800)]
mm/mprotect.c: coding-style cleanups

A few gremlins have recently crept in.

Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoDocumentation: ABI: /sys/devices/system/node/
Davidlohr Bueso [Tue, 18 Dec 2012 22:23:16 +0000 (14:23 -0800)]
Documentation: ABI: /sys/devices/system/node/

Describe NUMA node sysfs files/attributes.

Note that for the specific dates and contacts I couldn't find,
I left it as default for Oct 2002 and linux-mm.

Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoslub: drop mutex before deleting sysfs entry
Glauber Costa [Tue, 18 Dec 2012 22:23:13 +0000 (14:23 -0800)]
slub: drop mutex before deleting sysfs entry

Sasha Levin recently reported a lockdep problem resulting from the new
attribute propagation introduced by kmemcg series.  In short, slab_mutex
will be called from within the sysfs attribute store function.  This will
create a dependency, that will later be held backwards when a cache is
destroyed - since destruction occurs with the slab_mutex held, and then
calls in to the sysfs directory removal function.

In this patch, I propose to adopt a strategy close to what
__kmem_cache_create does before calling sysfs_slab_add, and release the
lock before the call to sysfs_slab_remove.  This is pretty much the last
operation in the kmem_cache_shutdown() path, so we could do better by
splitting this and moving this call alone to later on.  This will fit
nicely when sysfs handling is consistent between all caches, but will look
weird now.

Lockdep info:

  ======================================================
  [ INFO: possible circular locking dependency detected ]
  3.7.0-rc4-next-20121106-sasha-00008-g353b62f #117 Tainted: G        W
  -------------------------------------------------------
  trinity-child13/6961 is trying to acquire lock:
   (s_active#43){++++.+}, at:  sysfs_addrm_finish+0x31/0x60

  but task is already holding lock:
   (slab_mutex){+.+.+.}, at:  kmem_cache_destroy+0x22/0xe0

  which lock already depends on the new lock.

  the existing dependency chain (in reverse order) is:
  -> #1 (slab_mutex){+.+.+.}:
          lock_acquire+0x1aa/0x240
          __mutex_lock_common+0x59/0x5a0
          mutex_lock_nested+0x3f/0x50
          slab_attr_store+0xde/0x110
          sysfs_write_file+0xfa/0x150
          vfs_write+0xb0/0x180
          sys_pwrite64+0x60/0xb0
          tracesys+0xe1/0xe6
  -> #0 (s_active#43){++++.+}:
          __lock_acquire+0x14df/0x1ca0
          lock_acquire+0x1aa/0x240
          sysfs_deactivate+0x122/0x1a0
          sysfs_addrm_finish+0x31/0x60
          sysfs_remove_dir+0x89/0xd0
          kobject_del+0x16/0x40
          __kmem_cache_shutdown+0x40/0x60
          kmem_cache_destroy+0x40/0xe0
          mon_text_release+0x78/0xe0
          __fput+0x122/0x2d0
          ____fput+0x9/0x10
          task_work_run+0xbe/0x100
          do_exit+0x432/0xbd0
          do_group_exit+0x84/0xd0
          get_signal_to_deliver+0x81d/0x930
          do_signal+0x3a/0x950
          do_notify_resume+0x3e/0x90
          int_signal+0x12/0x17

  other info that might help us debug this:

   Possible unsafe locking scenario:

         CPU0                    CPU1
         ----                    ----
    lock(slab_mutex);
                                 lock(s_active#43);
                                 lock(slab_mutex);
    lock(s_active#43);

   *** DEADLOCK ***

  2 locks held by trinity-child13/6961:
   #0:  (mon_lock){+.+.+.}, at:  mon_text_release+0x25/0xe0
   #1:  (slab_mutex){+.+.+.}, at:  kmem_cache_destroy+0x22/0xe0

  stack backtrace:
  Pid: 6961, comm: trinity-child13 Tainted: G        W    3.7.0-rc4-next-20121106-sasha-00008-g353b62f #117
  Call Trace:
    print_circular_bug+0x1fb/0x20c
    __lock_acquire+0x14df/0x1ca0
    lock_acquire+0x1aa/0x240
    sysfs_deactivate+0x122/0x1a0
    sysfs_addrm_finish+0x31/0x60
    sysfs_remove_dir+0x89/0xd0
    kobject_del+0x16/0x40
    __kmem_cache_shutdown+0x40/0x60
    kmem_cache_destroy+0x40/0xe0
    mon_text_release+0x78/0xe0
    __fput+0x122/0x2d0
    ____fput+0x9/0x10
    task_work_run+0xbe/0x100
    do_exit+0x432/0xbd0
    do_group_exit+0x84/0xd0
    get_signal_to_deliver+0x81d/0x930
    do_signal+0x3a/0x950
    do_notify_resume+0x3e/0x90
    int_signal+0x12/0x17

Signed-off-by: Glauber Costa <glommer@parallels.com>
Reported-by: Sasha Levin <sasha.levin@oracle.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Pekka Enberg <penberg@kernel.org>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomemcg: add comments clarifying aspects of cache attribute propagation
Glauber Costa [Tue, 18 Dec 2012 22:23:10 +0000 (14:23 -0800)]
memcg: add comments clarifying aspects of cache attribute propagation

This patch clarifies two aspects of cache attribute propagation.

First, the expected context for the for_each_memcg_cache macro in
memcontrol.h.  The usages already in the codebase are safe.  In mm/slub.c,
it is trivially safe because the lock is acquired right before the loop.
In mm/slab.c, it is less so: the lock is acquired by an outer function a
few steps back in the stack, so a VM_BUG_ON() is added to make sure it is
indeed safe.

A comment is also added to detail why we are returning the value of the
parent cache and ignoring the children's when we propagate the attributes.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agokmem: add slab-specific documentation about the kmem controller
Glauber Costa [Tue, 18 Dec 2012 22:23:08 +0000 (14:23 -0800)]
kmem: add slab-specific documentation about the kmem controller

Signed-off-by: Glauber Costa <glommer@parallels.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.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>
12 years agoslub: slub-specific propagation changes
Glauber Costa [Tue, 18 Dec 2012 22:23:05 +0000 (14:23 -0800)]
slub: slub-specific propagation changes

SLUB allows us to tune a particular cache behavior with sysfs-based
tunables.  When creating a new memcg cache copy, we'd like to preserve any
tunables the parent cache already had.

This can be done by tapping into the store attribute function provided by
the allocator.  We of course don't need to mess with read-only fields.
Since the attributes can have multiple types and are stored internally by
sysfs, the best strategy is to issue a ->show() in the root cache, and
then ->store() in the memcg cache.

The drawback of that, is that sysfs can allocate up to a page in buffering
for show(), that we are likely not to need, but also can't guarantee.  To
avoid always allocating a page for that, we can update the caches at store
time with the maximum attribute size ever stored to the root cache.  We
will then get a buffer big enough to hold it.  The corolary to this, is
that if no stores happened, nothing will be propagated.

It can also happen that a root cache has its tunables updated during
normal system operation.  In this case, we will propagate the change to
all caches that are already active.

[akpm@linux-foundation.org: tweak code to avoid __maybe_unused]
Signed-off-by: Glauber Costa <glommer@parallels.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.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>
12 years agoslab: propagate tunable values
Glauber Costa [Tue, 18 Dec 2012 22:23:03 +0000 (14:23 -0800)]
slab: propagate tunable values

SLAB allows us to tune a particular cache behavior with tunables.  When
creating a new memcg cache copy, we'd like to preserve any tunables the
parent cache already had.

This could be done by an explicit call to do_tune_cpucache() after the
cache is created.  But this is not very convenient now that the caches are
created from common code, since this function is SLAB-specific.

Another method of doing that is taking advantage of the fact that
do_tune_cpucache() is always called from enable_cpucache(), which is
called at cache initialization.  We can just preset the values, and then
things work as expected.

It can also happen that a root cache has its tunables updated during
normal system operation.  In this case, we will propagate the change to
all caches that are already active.

This change will require us to move the assignment of root_cache in
memcg_params a bit earlier.  We need this to be already set - which
memcg_kmem_register_cache will do - when we reach __kmem_cache_create()

Signed-off-by: Glauber Costa <glommer@parallels.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.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>
12 years agomemcg: aggregate memcg cache values in slabinfo
Glauber Costa [Tue, 18 Dec 2012 22:23:01 +0000 (14:23 -0800)]
memcg: aggregate memcg cache values in slabinfo

When we create caches in memcgs, we need to display their usage
information somewhere.  We'll adopt a scheme similar to /proc/meminfo,
with aggregate totals shown in the global file, and per-group information
stored in the group itself.

For the time being, only reads are allowed in the per-group cache.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.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>
12 years agomemcg/sl[au]b: shrink dead caches
Glauber Costa [Tue, 18 Dec 2012 22:22:59 +0000 (14:22 -0800)]
memcg/sl[au]b: shrink dead caches

This means that when we destroy a memcg cache that happened to be empty,
those caches may take a lot of time to go away: removing the memcg
reference won't destroy them - because there are pending references, and
the empty pages will stay there, until a shrinker is called upon for any
reason.

In this patch, we will call kmem_cache_shrink() for all dead caches that
cannot be destroyed because of remaining pages.  After shrinking, it is
possible that it could be freed.  If this is not the case, we'll schedule
a lazy worker to keep trying.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.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>
12 years agomemcg/sl[au]b: track all the memcg children of a kmem_cache
Glauber Costa [Tue, 18 Dec 2012 22:22:55 +0000 (14:22 -0800)]
memcg/sl[au]b: track all the memcg children of a kmem_cache

This enables us to remove all the children of a kmem_cache being
destroyed, if for example the kernel module it's being used in gets
unloaded.  Otherwise, the children will still point to the destroyed
parent.

Signed-off-by: Suleiman Souhlal <suleiman@google.com>
Signed-off-by: Glauber Costa <glommer@parallels.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.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>
12 years agomemcg: destroy memcg caches
Glauber Costa [Tue, 18 Dec 2012 22:22:50 +0000 (14:22 -0800)]
memcg: destroy memcg caches

Implement destruction of memcg caches.  Right now, only caches where our
reference counter is the last remaining are deleted.  If there are any
other reference counters around, we just leave the caches lying around
until they go away.

When that happens, a destruction function is called from the cache code.
Caches are only destroyed in process context, so we queue them up for
later processing in the general case.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.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>
12 years agosl[au]b: allocate objects from memcg cache
Glauber Costa [Tue, 18 Dec 2012 22:22:48 +0000 (14:22 -0800)]
sl[au]b: allocate objects from memcg cache

We are able to match a cache allocation to a particular memcg.  If the
task doesn't change groups during the allocation itself - a rare event,
this will give us a good picture about who is the first group to touch a
cache page.

This patch uses the now available infrastructure by calling
memcg_kmem_get_cache() before all the cache allocations.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.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>
12 years agosl[au]b: always get the cache from its page in kmem_cache_free()
Glauber Costa [Tue, 18 Dec 2012 22:22:46 +0000 (14:22 -0800)]
sl[au]b: always get the cache from its page in kmem_cache_free()

struct page already has this information.  If we start chaining caches,
this information will always be more trustworthy than whatever is passed
into the function.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.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>
12 years agomemcg: skip memcg kmem allocations in specified code regions
Glauber Costa [Tue, 18 Dec 2012 22:22:42 +0000 (14:22 -0800)]
memcg: skip memcg kmem allocations in specified code regions

Create a mechanism that skip memcg allocations during certain pieces of
our core code.  It basically works in the same way as
preempt_disable()/preempt_enable(): By marking a region under which all
allocations will be accounted to the root memcg.

We need this to prevent races in early cache creation, when we
allocate data using caches that are not necessarily created already.

Signed-off-by: Glauber Costa <glommer@parallels.com>
yCc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.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>
12 years agomemcg: infrastructure to match an allocation to the right cache
Glauber Costa [Tue, 18 Dec 2012 22:22:40 +0000 (14:22 -0800)]
memcg: infrastructure to match an allocation to the right cache

The page allocator is able to bind a page to a memcg when it is
allocated.  But for the caches, we'd like to have as many objects as
possible in a page belonging to the same cache.

This is done in this patch by calling memcg_kmem_get_cache in the
beginning of every allocation function.  This function is patched out by
static branches when kernel memory controller is not being used.

It assumes that the task allocating, which determines the memcg in the
page allocator, belongs to the same cgroup throughout the whole process.
Misaccounting can happen if the task calls memcg_kmem_get_cache() while
belonging to a cgroup, and later on changes.  This is considered
acceptable, and should only happen upon task migration.

Before the cache is created by the memcg core, there is also a possible
imbalance: the task belongs to a memcg, but the cache being allocated from
is the global cache, since the child cache is not yet guaranteed to be
ready.  This case is also fine, since in this case the GFP_KMEMCG will not
be passed and the page allocator will not attempt any cgroup accounting.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.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>
12 years agomemcg: allocate memory for memcg caches whenever a new memcg appears
Glauber Costa [Tue, 18 Dec 2012 22:22:38 +0000 (14:22 -0800)]
memcg: allocate memory for memcg caches whenever a new memcg appears

Every cache that is considered a root cache (basically the "original"
caches, tied to the root memcg/no-memcg) will have an array that should be
large enough to store a cache pointer per each memcg in the system.

Theoreticaly, this is as high as 1 << sizeof(css_id), which is currently
in the 64k pointers range.  Most of the time, we won't be using that much.

What goes in this patch, is a simple scheme to dynamically allocate such
an array, in order to minimize memory usage for memcg caches.  Because we
would also like to avoid allocations all the time, at least for now, the
array will only grow.  It will tend to be big enough to hold the maximum
number of kmem-limited memcgs ever achieved.

We'll allocate it to be a minimum of 64 kmem-limited memcgs.  When we have
more than that, we'll start doubling the size of this array every time the
limit is reached.

Because we are only considering kmem limited memcgs, a natural point for
this to happen is when we write to the limit.  At that point, we already
have set_limit_mutex held, so that will become our natural synchronization
mechanism.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.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>
12 years agoslab/slub: consider a memcg parameter in kmem_create_cache
Glauber Costa [Tue, 18 Dec 2012 22:22:34 +0000 (14:22 -0800)]
slab/slub: consider a memcg parameter in kmem_create_cache

Allow a memcg parameter to be passed during cache creation.  When the slub
allocator is being used, it will only merge caches that belong to the same
memcg.  We'll do this by scanning the global list, and then translating
the cache to a memcg-specific cache

Default function is created as a wrapper, passing NULL to the memcg
version.  We only merge caches that belong to the same memcg.

A helper is provided, memcg_css_id: because slub needs a unique cache name
for sysfs.  Since this is visible, but not the canonical location for slab
data, the cache name is not used, the css_id should suffice.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.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>
12 years agoslab: annotate on-slab caches nodelist locks
Glauber Costa [Tue, 18 Dec 2012 22:22:31 +0000 (14:22 -0800)]
slab: annotate on-slab caches nodelist locks

We currently provide lockdep annotation for kmalloc caches, and also
caches that have SLAB_DEBUG_OBJECTS enabled.  The reason for this is that
we can quite frequently nest in the l3->list_lock lock, which is not
something trivial to avoid.

My proposal with this patch, is to extend this to caches whose slab
management object lives within the slab as well ("on_slab").  The need for
this arose in the context of testing kmemcg-slab patches.  With such
patchset, we can have per-memcg kmalloc caches.  So the same path that led
to nesting between kmalloc caches will could then lead to in-memcg
nesting.  Because they are not annotated, lockdep will trigger.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.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>
12 years agoslab/slub: struct memcg_params
Glauber Costa [Tue, 18 Dec 2012 22:22:27 +0000 (14:22 -0800)]
slab/slub: struct memcg_params

For the kmem slab controller, we need to record some extra information in
the kmem_cache structure.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Signed-off-by: Suleiman Souhlal <suleiman@google.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.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>
12 years agomemcg: add documentation about the kmem controller
Glauber Costa [Tue, 18 Dec 2012 22:22:22 +0000 (14:22 -0800)]
memcg: add documentation about the kmem controller

Signed-off-by: Glauber Costa <glommer@parallels.com>
Acked-by: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.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>
12 years agofork: protect architectures where THREAD_SIZE >= PAGE_SIZE against fork bombs
Glauber Costa [Tue, 18 Dec 2012 22:22:18 +0000 (14:22 -0800)]
fork: protect architectures where THREAD_SIZE >= PAGE_SIZE against fork bombs

Because those architectures will draw their stacks directly from the page
allocator, rather than the slab cache, we can directly pass __GFP_KMEMCG
flag, and issue the corresponding free_pages.

This code path is taken when the architecture doesn't define
CONFIG_ARCH_THREAD_INFO_ALLOCATOR (only ia64 seems to), and has
THREAD_SIZE >= PAGE_SIZE.  Luckily, most - if not all - of the remaining
architectures fall in this category.

This will guarantee that every stack page is accounted to the memcg the
process currently lives on, and will have the allocations to fail if they
go over limit.

For the time being, I am defining a new variant of THREADINFO_GFP, not to
mess with the other path.  Once the slab is also tracked by memcg, we can
get rid of that flag.

Tested to successfully protect against :(){ :|:& };:

Signed-off-by: Glauber Costa <glommer@parallels.com>
Acked-by: Frederic Weisbecker <fweisbec@redhat.com>
Acked-by: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.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>
12 years agomemcg: execute the whole memcg freeing in free_worker()
Glauber Costa [Tue, 18 Dec 2012 22:22:13 +0000 (14:22 -0800)]
memcg: execute the whole memcg freeing in free_worker()

A lot of the initialization we do in mem_cgroup_create() is done with
softirqs enabled.  This include grabbing a css id, which holds
&ss->id_lock->rlock, and the per-zone trees, which holds
rtpz->lock->rlock.  All of those signal to the lockdep mechanism that
those locks can be used in SOFTIRQ-ON-W context.

This means that the freeing of memcg structure must happen in a
compatible context, otherwise we'll get a deadlock, like the one below,
caught by lockdep:

   free_accounted_pages+0x47/0x4c
   free_task+0x31/0x5c
   __put_task_struct+0xc2/0xdb
   put_task_struct+0x1e/0x22
   delayed_put_task_struct+0x7a/0x98
   __rcu_process_callbacks+0x269/0x3df
   rcu_process_callbacks+0x31/0x5b
   __do_softirq+0x122/0x277

This usage pattern could not be triggered before kmem came into play.
With the introduction of kmem stack handling, it is possible that we call
the last mem_cgroup_put() from the task destructor, which is run in an rcu
callback.  Such callbacks are run with softirqs disabled, leading to the
offensive usage pattern.

In general, we have little, if any, means to guarantee in which context
the last memcg_put will happen.  The best we can do is test it and try to
make sure no invalid context releases are happening.  But as we add more
code to memcg, the possible interactions grow in number and expose more
ways to get context conflicts.  One thing to keep in mind, is that part of
the freeing process is already deferred to a worker, such as vfree(), that
can only be called from process context.

For the moment, the only two functions we really need moved away are:

  * free_css_id(), and
  * mem_cgroup_remove_from_trees().

But because the later accesses per-zone info,
free_mem_cgroup_per_zone_info() needs to be moved as well.  With that, we
are left with the per_cpu stats only.  Better move it all.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Tested-by: Greg Thelen <gthelen@google.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.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>
12 years agomemcg: allow a memcg with kmem charges to be destructed
Glauber Costa [Tue, 18 Dec 2012 22:22:11 +0000 (14:22 -0800)]
memcg: allow a memcg with kmem charges to be destructed

Because the ultimate goal of the kmem tracking in memcg is to track slab
pages as well, we can't guarantee that we'll always be able to point a
page to a particular process, and migrate the charges along with it -
since in the common case, a page will contain data belonging to multiple
processes.

Because of that, when we destroy a memcg, we only make sure the
destruction will succeed by discounting the kmem charges from the user
charges when we try to empty the cgroup.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Acked-by: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.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>
12 years agomemcg: use static branches when code not in use
Glauber Costa [Tue, 18 Dec 2012 22:22:09 +0000 (14:22 -0800)]
memcg: use static branches when code not in use

We can use static branches to patch the code in or out when not used.

Because the _ACTIVE bit on kmem_accounted is only set after the increment
is done, we guarantee that the root memcg will always be selected for kmem
charges until all call sites are patched (see memcg_kmem_enabled).  This
guarantees that no mischarges are applied.

Static branch decrement happens when the last reference count from the
kmem accounting in memcg dies.  This will only happen when the charges
drop down to 0.

When that happens, we need to disable the static branch only on those
memcgs that enabled it.  To achieve this, we would be forced to complicate
the code by keeping track of which memcgs were the ones that actually
enabled limits, and which ones got it from its parents.

It is a lot simpler just to do static_key_slow_inc() on every child
that is accounted.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Acked-by: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Suleiman Souhlal <suleiman@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomemcg: kmem accounting lifecycle management
Glauber Costa [Tue, 18 Dec 2012 22:22:07 +0000 (14:22 -0800)]
memcg: kmem accounting lifecycle management

Because kmem charges can outlive the cgroup, we need to make sure that we
won't free the memcg structure while charges are still in flight.  For
reviewing simplicity, the charge functions will issue mem_cgroup_get() at
every charge, and mem_cgroup_put() at every uncharge.

This can get expensive, however, and we can do better.  mem_cgroup_get()
only really needs to be issued once: when the first limit is set.  In the
same spirit, we only need to issue mem_cgroup_put() when the last charge
is gone.

We'll need an extra bit in kmem_account_flags for that:
KMEM_ACCOUNTED_DEAD.  it will be set when the cgroup dies, if there are
charges in the group.  If there aren't, we can proceed right away.

Our uncharge function will have to test that bit every time the charges
drop to 0.  Because that is not the likely output of res_counter_uncharge,
this should not impose a big hit on us: it is certainly much better than a
reference count decrease at every operation.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Acked-by: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Suleiman Souhlal <suleiman@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agores_counter: return amount of charges after res_counter_uncharge()
Glauber Costa [Tue, 18 Dec 2012 22:22:04 +0000 (14:22 -0800)]
res_counter: return amount of charges after res_counter_uncharge()

It is useful to know how many charges are still left after a call to
res_counter_uncharge.  While it is possible to issue a res_counter_read
after uncharge, this can be racy.

If we need, for instance, to take some action when the counters drop down
to 0, only one of the callers should see it.  This is the same semantics
as the atomic variables in the kernel.

Since the current return value is void, we don't need to worry about
anything breaking due to this change: nobody relied on that, and only
users appearing from now on will be checking this value.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Acked-by: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Suleiman Souhlal <suleiman@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: allocate kernel pages to the right memcg
Glauber Costa [Tue, 18 Dec 2012 22:22:00 +0000 (14:22 -0800)]
mm: allocate kernel pages to the right memcg

When a process tries to allocate a page with the __GFP_KMEMCG flag, the
page allocator will call the corresponding memcg functions to validate
the allocation.  Tasks in the root memcg can always proceed.

To avoid adding markers to the page - and a kmem flag that would
necessarily follow, as much as doing page_cgroup lookups for no reason,
whoever is marking its allocations with __GFP_KMEMCG flag is responsible
for telling the page allocator that this is such an allocation at
free_pages() time.  This is done by the invocation of
__free_accounted_pages() and free_accounted_pages().

Signed-off-by: Glauber Costa <glommer@parallels.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Acked-by: Mel Gorman <mgorman@suse.de>
Acked-by: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Suleiman Souhlal <suleiman@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomemcg: kmem controller infrastructure
Glauber Costa [Tue, 18 Dec 2012 22:21:56 +0000 (14:21 -0800)]
memcg: kmem controller infrastructure

Introduce infrastructure for tracking kernel memory pages to a given
memcg.  This will happen whenever the caller includes the flag
__GFP_KMEMCG flag, and the task belong to a memcg other than the root.

In memcontrol.h those functions are wrapped in inline acessors.  The idea
is to later on, patch those with static branches, so we don't incur any
overhead when no mem cgroups with limited kmem are being used.

Users of this functionality shall interact with the memcg core code
through the following functions:

memcg_kmem_newpage_charge: will return true if the group can handle the
                           allocation. At this point, struct page is not
                           yet allocated.

memcg_kmem_commit_charge: will either revert the charge, if struct page
                          allocation failed, or embed memcg information
                          into page_cgroup.

memcg_kmem_uncharge_page: called at free time, will revert the charge.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Acked-by: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm: add a __GFP_KMEMCG flag
Glauber Costa [Tue, 18 Dec 2012 22:21:51 +0000 (14:21 -0800)]
mm: add a __GFP_KMEMCG flag

This flag is used to indicate to the callees that this allocation is a
kernel allocation in process context, and should be accounted to current's
memcg.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Rik van Riel <riel@redhat.com>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Acked-by: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Acked-by: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Suleiman Souhlal <suleiman@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: JoonSoo Kim <js1304@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomemcg: kmem accounting basic infrastructure
Glauber Costa [Tue, 18 Dec 2012 22:21:47 +0000 (14:21 -0800)]
memcg: kmem accounting basic infrastructure

Add the basic infrastructure for the accounting of kernel memory.  To
control that, the following files are created:

 * memory.kmem.usage_in_bytes
 * memory.kmem.limit_in_bytes
 * memory.kmem.failcnt
 * memory.kmem.max_usage_in_bytes

They have the same meaning of their user memory counterparts.  They
reflect the state of the "kmem" res_counter.

Per cgroup kmem memory accounting is not enabled until a limit is set for
the group.  Once the limit is set the accounting cannot be disabled for
that group.  This means that after the patch is applied, no behavioral
changes exists for whoever is still using memcg to control their memory
usage, until memory.kmem.limit_in_bytes is set for the first time.

We always account to both user and kernel resource_counters.  This
effectively means that an independent kernel limit is in place when the
limit is set to a lower value than the user memory.  A equal or higher
value means that the user limit will always hit first, meaning that kmem
is effectively unlimited.

People who want to track kernel memory but not limit it, can set this
limit to a very high number (like RESOURCE_MAX - 1page - that no one will
ever hit, or equal to the user memory)

[akpm@linux-foundation.org: MEMCG_MMEM only works with slab and slub]
Signed-off-by: Glauber Costa <glommer@parallels.com>
Acked-by: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomemcg: change defines to an enum
Glauber Costa [Tue, 18 Dec 2012 22:21:45 +0000 (14:21 -0800)]
memcg: change defines to an enum

This is just a cleanup patch for clarity of expression.  In earlier
submissions, people asked it to be in a separate patch, so here it is.

Signed-off-by: Glauber Costa <glommer@parallels.com>
Acked-by: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomemcg: reclaim when more than one page needed
Suleiman Souhlal [Tue, 18 Dec 2012 22:21:41 +0000 (14:21 -0800)]
memcg: reclaim when more than one page needed

mem_cgroup_do_charge() was written before kmem accounting, and expects
three cases: being called for 1 page, being called for a stock of 32
pages, or being called for a hugepage.  If we call for 2 or 3 pages (and
both the stack and several slabs used in process creation are such, at
least with the debug options I had), it assumed it's being called for
stock and just retried without reclaiming.

Fix that by passing down a minsize argument in addition to the csize.

And what to do about that (csize == PAGE_SIZE && ret) retry?  If it's
needed at all (and presumably is since it's there, perhaps to handle
races), then it should be extended to more than PAGE_SIZE, yet how far?
And should there be a retry count limit, of what?  For now retry up to
COSTLY_ORDER (as page_alloc.c does) and make sure not to do it if
__GFP_NORETRY.

v4: fixed nr pages calculation pointed out by Christoph Lameter.

Signed-off-by: Suleiman Souhlal <suleiman@google.com>
Signed-off-by: Glauber Costa <glommer@parallels.com>
Acked-by: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomemcg: make it possible to use the stock for more than one page
Suleiman Souhlal [Tue, 18 Dec 2012 22:21:36 +0000 (14:21 -0800)]
memcg: make it possible to use the stock for more than one page

We currently have a percpu stock cache scheme that charges one page at a
time from memcg->res, the user counter.  When the kernel memory controller
comes into play, we'll need to charge more than that.

This is because kernel memory allocations will also draw from the user
counter, and can be bigger than a single page, as it is the case with the
stack (usually 2 pages) or some higher order slabs.

[glommer@parallels.com: added a changelog ]
Signed-off-by: Suleiman Souhlal <suleiman@google.com>
Signed-off-by: Glauber Costa <glommer@parallels.com>
Acked-by: David Rientjes <rientjes@google.com>
Acked-by: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: Frederic Weisbecker <fweisbec@redhat.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: JoonSoo Kim <js1304@gmail.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Rik van Riel <riel@redhat.com>
Cc: Suleiman Souhlal <suleiman@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomemory-hotplug: document and enable CONFIG_MOVABLE_NODE
Tang Chen [Tue, 18 Dec 2012 22:21:33 +0000 (14:21 -0800)]
memory-hotplug: document and enable CONFIG_MOVABLE_NODE

Add help info for CONFIG_MOVABLE_NODE and permit its selection.

This option allows the user to online all memory of a node as movable
memory.  So that the whole node can be hotplugged.  Users who don't use
the hotplug feature are also fine with this option on since they won't
online memory as movable.

Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com>
Reviewed-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Wen Congyang <wency@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
[akpm@linux-foundation.org: tweak help text]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm/page_alloc.c: remove duplicate check
Gavin Shan [Tue, 18 Dec 2012 22:21:32 +0000 (14:21 -0800)]
mm/page_alloc.c: remove duplicate check

While allocating pages using buddy allocator, the compound page is
probably split up to free pages.  Under these circumstances, the compound
page should be destroyed by destroy_compound_page().  However, there is a
duplicate check to judge if the page is compound.

Remove the duplicate check since the compound_order() returns 0 when the
page doesn't have PG_head set in destroy_compound_page().  That is to say,
destroy_compound_page() needn't check PageHead().

Signed-off-by: Gavin Shan <shangw@linux.vnet.ibm.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agodrivers/message/fusion/mptscsih.c: missing break
Alan Cox [Tue, 18 Dec 2012 22:21:30 +0000 (14:21 -0800)]
drivers/message/fusion/mptscsih.c: missing break

This happens to do the right thing in all cases on fibre channel but not on
other media types

Signed-off-by: Alan Cox <alan@linux.intel.com>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Cc: Nagalakshmi Nandigama <nagalakshmi.nandigama@lsi.com>
Cc: Kashyap Desai <kashyap.desai@lsi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoh8300: select generic atomic64_t support
Fengguang Wu [Tue, 18 Dec 2012 22:21:29 +0000 (14:21 -0800)]
h8300: select generic atomic64_t support

Rationales from Eric:

So I just looked a little deeper and it appears architectures that do
not support atomic64_t are broken.

The generic atomic64 support came in 2009 to support the perf subsystem
with the expectation that all architectures would implement atomic64
support.

Furthermore upon inspection of the kernel atomic64_t is used in a fair
number of places beyond the performance counters:

block/blk-cgroup.c
drivers/acpi/apei/
drivers/block/rbd.c
drivers/crypto/nx/nx.h
drivers/gpu/drm/radeon/radeon.h
drivers/infiniband/hw/ipath/
drivers/infiniband/hw/qib/
drivers/staging/octeon/
fs/xfs/
include/linux/perf_event.h
include/net/netfilter/nf_conntrack_acct.h
kernel/events/
kernel/trace/
net/mac80211/key.h
net/rds/

The block control group, infiniband, xfs, crypto, 802.11, netfilter.
Nothing quite so fundamental as fs/namespace.c but definitely in
multiplatform-code that should work, and is already broken on those
architecutres.

Looking at the implementation of atomic64_add_return in lib/atomic64.c the
code looks as efficient as these kinds of things get.

Which leads me to the conclusion that we need atomic64 support on all
architectures.

Signed-off-by: Fengguang Wu <fengguang.wu@intel.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoCoccinelle: add api/d_find_alias.cocci
Cyril Roelandt [Tue, 18 Dec 2012 22:21:28 +0000 (14:21 -0800)]
Coccinelle: add api/d_find_alias.cocci

Ensure that calls to d_find_alias() have a corresponding dput().

Signed-off-by: Cyril Roelandt <tipecaml@gmail.com>
Cc: Julia Lawall <Julia.Lawall@lip6.fr>
Cc: Gilles Muller <Gilles.Muller@lip6.fr>
Cc: Nicolas Palix <nicolas.palix@imag.fr>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>