profile/ivi/kernel-x86-ivi.git
13 years agoMerge branch 'common/fbdev' of master.kernel.org:/pub/scm/linux/kernel/git/lethal...
Paul Mundt [Tue, 29 Mar 2011 07:06:21 +0000 (16:06 +0900)]
Merge branch 'common/fbdev' of /linux/kernel/git/lethal/sh-2.6

13 years agofbdev: sh_mobile_lcdcfb: fix module lock acquisition
Alexandre Courbot [Wed, 23 Feb 2011 08:41:50 +0000 (08:41 +0000)]
fbdev: sh_mobile_lcdcfb: fix module lock acquisition

Whenever the LCDC is to be started or stopped, a board callback is
checked for existence and invoked. Prior to the invokation, the
callback's module lock is also acquired, to be released once the
callback returns. However, the order of testing makes it possible for
the lock to be acquired and not released in case the callback does not
exist. This patch reorders the tests to prevent this particular case.

Signed-off-by: Alexandre Courbot <gnurou@gmail.com>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
13 years agofbdev: sh_mobile_lcdcfb: add blanking support
Alexandre Courbot [Wed, 23 Feb 2011 08:36:30 +0000 (08:36 +0000)]
fbdev: sh_mobile_lcdcfb: add blanking support

Add a blanking callback to the LCDC driver in order to support both
FBIOBLANK and TIOCLINUX blanking ioctls. LCDC clocks are also released
if the requested blanking level is superior to FB_BLANK_NORMAL, to allow
runtime PM to disable the clocks if possible.

Signed-off-by: Alexandre Courbot <gnurou@gmail.com>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
13 years agoMerge branch 'viafb-next' of git://github.com/schandinat/linux-2.6
Paul Mundt [Thu, 24 Mar 2011 22:52:38 +0000 (07:52 +0900)]
Merge branch 'viafb-next' of git://github.com/schandinat/linux-2.6

13 years agoMerge branch 'for-2.6.39/core' of git://git.kernel.dk/linux-2.6-block
Linus Torvalds [Thu, 24 Mar 2011 17:16:26 +0000 (10:16 -0700)]
Merge branch 'for-2.6.39/core' of git://git.kernel.dk/linux-2.6-block

* 'for-2.6.39/core' of git://git.kernel.dk/linux-2.6-block: (65 commits)
  Documentation/iostats.txt: bit-size reference etc.
  cfq-iosched: removing unnecessary think time checking
  cfq-iosched: Don't clear queue stats when preempt.
  blk-throttle: Reset group slice when limits are changed
  blk-cgroup: Only give unaccounted_time under debug
  cfq-iosched: Don't set active queue in preempt
  block: fix non-atomic access to genhd inflight structures
  block: attempt to merge with existing requests on plug flush
  block: NULL dereference on error path in __blkdev_get()
  cfq-iosched: Don't update group weights when on service tree
  fs: assign sb->s_bdi to default_backing_dev_info if the bdi is going away
  block: Require subsystems to explicitly allocate bio_set integrity mempool
  jbd2: finish conversion from WRITE_SYNC_PLUG to WRITE_SYNC and explicit plugging
  jbd: finish conversion from WRITE_SYNC_PLUG to WRITE_SYNC and explicit plugging
  fs: make fsync_buffers_list() plug
  mm: make generic_writepages() use plugging
  blk-cgroup: Add unaccounted time to timeslice_used.
  block: fixup plugging stubs for !CONFIG_BLOCK
  block: remove obsolete comments for blkdev_issue_zeroout.
  blktrace: Use rq->cmd_flags directly in blk_add_trace_rq.
  ...

Fix up conflicts in fs/{aio.c,super.c}

13 years agoMerge branch 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux...
Linus Torvalds [Thu, 24 Mar 2011 17:07:50 +0000 (10:07 -0700)]
Merge branch 'next' of git://git./linux/kernel/git/dhowells/linux-2.6-mn10300

* 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-2.6-mn10300:
  MN10300: gcc 4.6 vs am33 inline assembly
  MN10300: Deprecate gdbstub
  MN10300: Allow KGDB to use the MN10300 serial ports
  MN10300: Emulate single stepping in KGDB on MN10300
  MN10300: Generalise kernel debugger kernel halt, reboot or power off hook
  KGDB: Notify GDB of machine halt, reboot or power off
  MN10300: Use KGDB
  MN10300: Create generic kernel debugger hooks
  MN10300: Create general kernel debugger cache flushing
  MN10300: Introduce a general config option for kernel debugger hooks
  MN10300: The icache invalidate functions should disable the icache first
  MN10300: gdbstub: Restrict single-stepping to non-preemptable non-SMP configs

13 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6
Linus Torvalds [Thu, 24 Mar 2011 17:05:23 +0000 (10:05 -0700)]
Merge branch 'release' of git://git./linux/kernel/git/aegl/linux-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6:
  pstore: cleanups to pstore_dump()
  [IA64] New syscalls for 2.6.39

13 years agoMerge branch 'rmobile-latest' of git://git.kernel.org/pub/scm/linux/kernel/git/lethal...
Linus Torvalds [Thu, 24 Mar 2011 17:04:59 +0000 (10:04 -0700)]
Merge branch 'rmobile-latest' of git://git./linux/kernel/git/lethal/sh-2.6

* 'rmobile-latest' of git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6:
  mmc: Add MMC_PROGRESS_*
  mmc, ARM: Rename SuperH Mobile ARM zboot helpers
  ARM: mach-shmobile: add coherent DMA mask to CEU camera devices
  ARM: mach-shmobile: Dynamic backlight control for Mackerel

13 years agoMerge branch 'sh-latest' of git://git.kernel.org/pub/scm/linux/kernel/git/lethal...
Linus Torvalds [Thu, 24 Mar 2011 17:04:05 +0000 (10:04 -0700)]
Merge branch 'sh-latest' of git://git./linux/kernel/git/lethal/sh-2.6

* 'sh-latest' of git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6:
  sh: Fix build alloc_thread_info_node function
  sh: Fix ptrace hw_breakpoint handling
  sh: Fix ptrace fpu state initialisation
  sh: Re-enable GENERIC_HARDIRQS_NO_DEPRECATED.
  sh: pmb: Use struct syscore_ops instead of sysdevs
  sh: Use struct syscore_ops instead of sysdevs
  sh: Conver to asm-generic/sizes.h.
  sh: wire up sys_syncfs.

13 years agoMerge branch 'usb-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Thu, 24 Mar 2011 17:02:55 +0000 (10:02 -0700)]
Merge branch 'usb-linus' of git://git./linux/kernel/git/gregkh/usb-2.6

* 'usb-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6:
  USB: cdc-acm: fix potential null-pointer dereference on disconnect
  USB: cdc-acm: fix potential null-pointer dereference
  USB: cdc-acm: fix memory corruption / panic
  USB: Fix 'bad dma' problem on WDM device disconnect
  usb: wwan: fix compilation without CONFIG_PM_RUNTIME
  USB: uss720 fixup refcount position
  usb: musb: blackfin: fix typo in new bfin_musb_vbus_status func
  usb: musb: blackfin: fix typo in new dev_pm_ops struct
  usb: musb: blackfin: fix typo in platform driver name
  usb: musb: Fix for merge issue
  ehci-hcd: Bug fix: don't set a QH's Halt bit
  USB: Do not pass negative length to snoop_urb()

13 years agoMerge branch 'v4l_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab...
Linus Torvalds [Thu, 24 Mar 2011 16:50:13 +0000 (09:50 -0700)]
Merge branch 'v4l_for_linus' of git://git./linux/kernel/git/mchehab/linux-2.6

* 'v4l_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-2.6: (442 commits)
  [media] videobuf2-dma-contig: make cookie() return a pointer to dma_addr_t
  [media] sh_mobile_ceu_camera: Do not call vb2's mem_ops directly
  [media] V4L: soc-camera: explicitly require V4L2_BUF_TYPE_VIDEO_CAPTURE
  [media] v4l: soc-camera: Store negotiated buffer settings
  [media] rc: interim support for 32-bit NEC-ish scancodes
  [media] mceusb: topseed 0x0011 needs gen3 init for tx to work
  [media] lirc_zilog: error out if buffer read bytes != chunk size
  [media] lirc: silence some compile warnings
  [media] hdpvr: use same polling interval as other OS
  [media] ir-kbd-i2c: pass device code w/key in hauppauge case
  [media] rc/keymaps: Remove the obsolete rc-rc5-tv keymap
  [media] remove the old RC_MAP_HAUPPAUGE_NEW RC map
  [media] rc/keymaps: Rename Hauppauge table as rc-hauppauge
  [media] rc-rc5-hauppauge-new: Fix Hauppauge Grey mapping
  [media] rc-rc5-hauppauge-new: Add support for the old Black RC
  [media] rc-rc5-hauppauge-new: Add the old control to the table
  [media] rc-winfast: Fix the keycode tables
  [media] a800: Fix a few wrong IR key assignments
  [media] opera1: Use multimedia keys instead of an app-specific mapping
  [media] dw2102: Use multimedia keys instead of an app-specific mapping
  ...

Fix up trivial conflicts (remove/modify and some real conflicts) in:
arch/arm/mach-omap2/devices.c
drivers/staging/Kconfig
drivers/staging/Makefile
drivers/staging/dabusb/dabusb.c
drivers/staging/dabusb/dabusb.h
drivers/staging/easycap/easycap_ioctl.c
drivers/staging/usbvideo/usbvideo.c
drivers/staging/usbvideo/vicam.c

13 years agoMerge branch 'for-linus' of git://android.git.kernel.org/kernel/tegra
Linus Torvalds [Thu, 24 Mar 2011 16:33:14 +0000 (09:33 -0700)]
Merge branch 'for-linus' of git://android.git./kernel/tegra

* 'for-linus' of git://android.git.kernel.org/kernel/tegra:
  ARM: tegra: harmony: initialize the TPS65862 PMIC
  ARM: tegra: update defconfig
  ARM: tegra: harmony: update PCI-e initialization sequence
  ARM: tegra: trimslice: enable MMC/SD slots
  ARM: tegra: enable new drivers in defconfig
  ARM: tegra: Add Toshiba AC100 support
  ARM: tegra: harmony: Set WM8903 gpio_base
  ARM: tegra: harmony: I2C-related portions of audio support
  ARM: tegra: harmony: register i2c devices
  ARM: tegra: seaboard: register i2c devices
  ARM: tegra: harmony: Beginnings of audio support
  ARM: tegra: create defines for SD-related GPIO names
  ARM: tegra: add devices.c entries for audio

13 years agoMerge branch 'devicetree/merge' of git://git.secretlab.ca/git/linux-2.6
Linus Torvalds [Thu, 24 Mar 2011 16:30:20 +0000 (09:30 -0700)]
Merge branch 'devicetree/merge' of git://git.secretlab.ca/git/linux-2.6

* 'devicetree/merge' of git://git.secretlab.ca/git/linux-2.6:
  spi/pl022: Add loopback support for the SPI on 5500
  spi/omap_mcspi: Fix broken last word xfer
  of/flattree: minor cleanups
  dt: eliminate OF_NO_DEEP_PROBE and test for NULL match table
  dt: protect against NULL matches passed to of_match_node()
  dt: Refactor of_platform_bus_probe()

13 years agommc: Add MMC_PROGRESS_*
Simon Horman [Thu, 24 Mar 2011 07:04:38 +0000 (07:04 +0000)]
mmc: Add MMC_PROGRESS_*

This is my second attempt to make this enum generally available.
The first attempt added MMCIF_PROGRESS_* to include/linux/mmc/sh_mmcif.h.
However this is not sufficiently generic as the enum will be
used by SDHI boot code.

Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
13 years agommc, ARM: Rename SuperH Mobile ARM zboot helpers
Simon Horman [Thu, 24 Mar 2011 07:04:37 +0000 (07:04 +0000)]
mmc, ARM: Rename SuperH Mobile ARM zboot helpers

These headers and helpers will also be used for SDHI boot
so the mmcif name will start to make a lot less sense.

Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
13 years agoMerge branch 'idle-release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb...
Linus Torvalds [Thu, 24 Mar 2011 15:25:53 +0000 (08:25 -0700)]
Merge branch 'idle-release' of git://git./linux/kernel/git/lenb/linux-idle-2.6

* 'idle-release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-idle-2.6:
  intel_idle: Rename cpuidle states

13 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux...
Linus Torvalds [Thu, 24 Mar 2011 15:25:15 +0000 (08:25 -0700)]
Merge branch 'release' of git://git./linux/kernel/git/lenb/linux-acpi-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6: (42 commits)
  ACPI: minor printk format change in acpi_pad
  ACPI: make acpi_pad /sys output more readable
  ACPICA: Update version to 20110316
  ACPICA: Header support for SLIC table
  ACPI: Make sure the FADT is at least rev 2 before using the reset register
  ACPI: Bug compatibility for Windows on the ACPI reboot vector
  ACPICA: Fix access width for reset vector
  ACPI battery: fribble sysfs files from a resume notifier
  ACPI button: remove unused procfs I/F
  ACPI, APEI, Add PCIe AER error information printing support
  PCIe, AER, use pre-generated prefix in error information printing
  ACPI, APEI, Add ERST record ID cache
  ACPI: Use syscore_ops instead of sysdev class and sysdev
  ACPI: Remove the unused EC sysdev class
  ACPI: use __cpuinit for the acpi_processor_set_pdc() call tree
  ACPI: use __init where possible in processor driver
  Thermal_Framework-Fix_crash_during_hwmon_unregister
  ACPICA: Update version to 20110211.
  ACPICA: Add mechanism to defer _REG methods for some installed handlers
  ACPICA: Add support for FunctionalFixedHW in acpi_ut_get_region_name
  ...

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/vapier...
Linus Torvalds [Thu, 24 Mar 2011 15:24:28 +0000 (08:24 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/vapier/blackfin

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/vapier/blackfin:
  Blackfin: bf54x: re-enable anomaly 05000353 for all revs
  Blackfin: enable atomic64_t support
  Blackfin: wire up new syncfs syscall
  Blackfin: SMP: flush CoreB cache when shutting down

13 years agoMerge branch 'for-linus' of git://git.infradead.org/ubi-2.6
Linus Torvalds [Thu, 24 Mar 2011 15:22:34 +0000 (08:22 -0700)]
Merge branch 'for-linus' of git://git.infradead.org/ubi-2.6

* 'for-linus' of git://git.infradead.org/ubi-2.6:
  UBIFS: fix assertion warning and refine comments
  UBIFS: kill CONFIG_UBIFS_FS_DEBUG_CHKS
  UBIFS: use GFP_NOFS properly
  UBI: use GFP_NOFS properly

13 years agoMerge branch 'for-2.6.39' of git://linux-nfs.org/~bfields/linux
Linus Torvalds [Thu, 24 Mar 2011 15:20:39 +0000 (08:20 -0700)]
Merge branch 'for-2.6.39' of git://linux-nfs.org/~bfields/linux

* 'for-2.6.39' of git://linux-nfs.org/~bfields/linux:
  SUNRPC: Remove resource leak in svc_rdma_send_error()
  nfsd: wrong index used in inner loop
  nfsd4: fix comment and remove unused nfsd4_file fields
  nfs41: make sure nfs server return right ca_maxresponsesize_cached
  nfsd: fix compile error
  svcrpc: fix bad argument in unix_domain_find
  nfsd4: fix struct file leak
  nfsd4: minor nfs4state.c reshuffling
  svcrpc: fix rare race on unix_domain creation
  nfsd41: modify the members value of nfsd4_op_flags
  nfsd: add proc file listing kernel's gss_krb5 enctypes
  gss:krb5 only include enctype numbers in gm_upcall_enctypes
  NFSD, VFS: Remove dead code in nfsd_rename()
  nfsd: kill unused macro definition
  locks: use assign_type()

13 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/pkl/squashfs-linus
Linus Torvalds [Thu, 24 Mar 2011 15:02:21 +0000 (08:02 -0700)]
Merge git://git./linux/kernel/git/pkl/squashfs-linus

* git://git.kernel.org/pub/scm/linux/kernel/git/pkl/squashfs-linus:
  Squashfs: Use vmalloc rather than kmalloc for zlib workspace
  Squashfs: handle corruption of directory structure
  Squashfs: wrap squashfs_mount() definition
  Squashfs: xz_wrapper doesn't need to include squashfs_fs_i.h anymore
  Squashfs: Update documentation to include compression options
  Squashfs: Update Kconfig help text to include xz compression
  Squashfs: add compression options support to xz decompressor
  Squashfs: extend decompressor framework to handle compression options

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland...
Linus Torvalds [Thu, 24 Mar 2011 14:59:46 +0000 (07:59 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/roland/infiniband

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
  IB: Increase DMA max_segment_size on Mellanox hardware
  IB/mad: Improve an error message so error code is included
  RDMA/nes: Don't print success message at level KERN_ERR
  RDMA/addr: Fix return of uninitialized ret value
  IB/srp: try to use larger FMR sizes to cover our mappings
  IB/srp: add support for indirect tables that don't fit in SRP_CMD
  IB/srp: rework mapping engine to use multiple FMR entries
  IB/srp: allow sg_tablesize to be set for each target
  IB/srp: move IB CM setup completion into its own function
  IB/srp: always avoid non-zero offsets into an FMR

13 years agoMerge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/sameo/mfd-2.6
Linus Torvalds [Thu, 24 Mar 2011 14:59:01 +0000 (07:59 -0700)]
Merge branch 'for-next' of git://git./linux/kernel/git/sameo/mfd-2.6

* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/sameo/mfd-2.6: (90 commits)
  mfd: Push byte swaps out of wm8994 bulk read path
  mfd: Rename ab8500 gpadc header
  mfd: Constify WM8994 write path
  mfd: Push byte swap out of WM8994 bulk I/O
  mfd: Avoid copying data in WM8994 I2C write
  mfd: Remove copy from WM831x I2C write function
  mfd: Staticise WM8994 PM ops
  regulator: Add a subdriver for TI TPS6105x regulator portions v2
  mfd: Add a core driver for TI TPS61050/TPS61052 chips v2
  gpio: Add Tunnel Creek support to sch_gpio
  mfd: Add Tunnel Creek support to lpc_sch
  pci_ids: Add Intel Tunnel Creek LPC Bridge device ID.
  regulator: MAX8997/8966 support
  mfd: Add WM8994 bulk register write operation
  mfd: Append additional read write on 88pm860x
  mfd: Adopt mfd_data in 88pm860x input driver
  mfd: Adopt mfd_data in 88pm860x regulator
  mfd: Adopt mfd_data in 88pm860x led
  mfd: Adopt mfd_data in 88pm860x backlight
  mfd: Fix MAX8997 Kconfig entry typos
  ...

13 years agoMerge branch 'for-linus' of git://git.open-osd.org/linux-open-osd
Linus Torvalds [Thu, 24 Mar 2011 14:57:38 +0000 (07:57 -0700)]
Merge branch 'for-linus' of git://git.open-osd.org/linux-open-osd

* 'for-linus' of git://git.open-osd.org/linux-open-osd:
  exofs: deprecate the commands pending counter
  exofs: Write sbi->s_nextid as part of the Create command
  exofs: Add option to mount by osdname
  exofs: Override read-ahead to align on stripe_size
  exofs: simple fsync race fix
  exofs: Optimize read_4_write
  exofs: Trivial: fix some indentation and debug prints
  exofs: Remove redundant unlikely()

13 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/lethal/fbdev-2.6
Linus Torvalds [Thu, 24 Mar 2011 14:56:52 +0000 (07:56 -0700)]
Merge git://git./linux/kernel/git/lethal/fbdev-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/lethal/fbdev-2.6: (140 commits)
  MAINTAINERS: de-orphan fbdev.
  MAINTAINERS: Add file pattern for fb dt bindings.
  video: Move sm501fb devicetree binding documentation to a better place.
  fbcon: fix situation where fbcon gets deinitialised and can't reinit.
  video, sm501: add OF binding to support SM501
  video, sm501: add edid and commandline support
  video, sm501: add I/O functions for use on powerpc
  video: Fix EDID macros H_SYNC_WIDTH and H_SYNC_OFFSET
  fbcon: Bugfix soft cursor detection in Tile Blitting
  video: add missing framebuffer_release in error path
  video: metronomefb: add __devexit_p around reference to metronomefb_remove
  video: hecubafb: add __devexit_p around reference to hecubafb_remove
  drivers:video:aty:radeon_base Fix typo occationally to occasionally
  atmel_lcdfb: add fb_blank function
  atmel_lcdfb: implement inverted contrast pwm
  video: s3c-fb: return proper error if clk_get fails
  uvesafb,vesafb: create WC or WB PAT-entries
  video: ffb: fix ffb_probe error path
  radeonfb: Let hwmon driver probe the "monid" I2C bus
  fbdev: sh_mobile_lcdc: checking NULL instead of IS_ERR()
  ...

13 years agoUBIFS: fix assertion warning and refine comments
Artem Bityutskiy [Wed, 23 Mar 2011 08:32:58 +0000 (10:32 +0200)]
UBIFS: fix assertion warning and refine comments

This patch fixes the following UBIFS assertion warning:

UBIFS assert failed in do_readpage at 115 (pid 199)
[<b00321b8>] (unwind_backtrace+0x0/0xdc) from [<af025118>]
(do_readpage+0x108/0x594 [ubifs])
[<af025118>] (do_readpage+0x108/0x594 [ubifs]) from [<af025764>]
(ubifs_write_end+0x1c0/0x2e8 [ubifs])
[<af025764>] (ubifs_write_end+0x1c0/0x2e8 [ubifs]) from
[<b00a0164>] (generic_file_buffered_write+0x18c/0x270)
[<b00a0164>] (generic_file_buffered_write+0x18c/0x270) from
[<b00a08d4>] (__generic_file_aio_write+0x478/0x4c0)
[<b00a08d4>] (__generic_file_aio_write+0x478/0x4c0) from
[<b00a0984>] (generic_file_aio_write+0x68/0xc8)
[<b00a0984>] (generic_file_aio_write+0x68/0xc8) from
[<af024a78>] (ubifs_aio_write+0x178/0x1d8 [ubifs])
[<af024a78>] (ubifs_aio_write+0x178/0x1d8 [ubifs]) from
[<b00d104c>] (do_sync_write+0xb0/0x100)
[<b00d104c>] (do_sync_write+0xb0/0x100) from [<b00d1abc>]
(vfs_write+0xac/0x154)
[<b00d1abc>] (vfs_write+0xac/0x154) from [<b00d1c10>]
(sys_write+0x3c/0x68)
[<b00d1c10>] (sys_write+0x3c/0x68) from [<b002d9a0>]
(ret_fast_syscall+0x0/0x2c)

The 'PG_checked' flag is used to indicate that the page does not
supposedly exist on the media (e.g., a hole or a page beyond the
inode size), so it requires slightly bigger budget, because we have
to account the indexing size increase. And this flag basically
tells that the budget for this page has to be "new page budget".
The "new page budget" is slightly bigger than the "existing page
budget".

The 'do_readpage()' function has the following assertion which
sometimes is hit: 'ubifs_assert(!PageChecked(page))'. Obviously,
the meaning of this assertion is: "I should not be asked to read
a page which does not exist on the media".

However, in 'ubifs_write_begin()' we have a small "trick". Notice,
that VFS may write pages which were not read yet, so the page data
were not loaded from the media to the page cache yet. If VFS tells
that it is going to change only some part of the page, we obviously
have to load it from the media. However, if VFS tells that it is
going to change whole page, we do not read it from the media for
optimization purposes.

However, since we do not read it, we do not know if it exists on
the media or not (a hole, etc). So we set the 'PG_checked' flag
to this page to force bigger budget, just in case.

So 'ubifs_write_begin()' sets 'PG_checked'. Then we are in
'ubifs_write_end()'. And VFS tells us: "hey, for some reasons I
changed my mind and did not change whole page". Frankly, I do not
know why this happens, but I hit this somehow on an ARM platform.
And this is extremely rare.

So in this case UBIFS does the following:

1. Cancels allocated budget.
2. Loads the page from the media by calling 'do_readpage()'.
3. Asks VFS to repeat the whole write operation from the very
   beginning (call '->write_begin() again, etc).

And the assertion warning is hit at the step 2 - remember we have
the 'PG_checked' set for this page, and 'do_readpage()' does not
like this. So this patch fixes the problem by adding step 1.5 and
cleaning the 'PG_checked' before calling 'do_readpage()'.

All in all, this patch does not fix any functionality issue, but it
silences UBIFS false positive warning which may happen in very very
rare cases.

And while on it, this patch also improves a commentary which explains
the reasons of setting the 'PG_checked' flag for the page. The old
commentary was a bit difficult to understand.

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
13 years agoUBIFS: kill CONFIG_UBIFS_FS_DEBUG_CHKS
Artem Bityutskiy [Mon, 21 Mar 2011 14:16:29 +0000 (16:16 +0200)]
UBIFS: kill CONFIG_UBIFS_FS_DEBUG_CHKS

Simplify UBIFS configuration menu and kill the option to enable self-check
compile-time. We do not really need this because we can do this run-time
using the module parameters or the corresponding sysfs interfaces. And
there is a value in simplifying the kernel configuration menu which becomes
increasingly large.

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
13 years agoUBIFS: use GFP_NOFS properly
Artem Bityutskiy [Thu, 24 Mar 2011 14:14:26 +0000 (16:14 +0200)]
UBIFS: use GFP_NOFS properly

This patch fixes a brown-paperbag bug which was introduced by me:
I used incorrect "GFP_KERNEL | GFP_NOFS" allocation flags to make
sure my allocations do not cause write-back. But the correct form
is "GFP_NOFS".

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
13 years agoUBI: use GFP_NOFS properly
Artem Bityutskiy [Thu, 24 Mar 2011 14:09:56 +0000 (16:09 +0200)]
UBI: use GFP_NOFS properly

This patch fixes a brown-paperbag bug which was introduced by me:
I used incorrect "GFP_KERNEL | GFP_NOFS" allocation flags to make
sure my allocations do not cause write-back. But the correct form
is "GFP_NOFS".

Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
13 years agosh: Fix build alloc_thread_info_node function
Nobuhiro Iwamatsu [Thu, 24 Mar 2011 05:47:40 +0000 (05:47 +0000)]
sh: Fix build alloc_thread_info_node function

By commit b6a84016bd2598e35ead635147fa53619982648d,
alloc_thread_info was replaced by alloc_thread_info_node.
However, the change of the function name and the addition of the argument
were incomplete.

Signed-off-by: Nobuhiro Iwamatsu <nobuhiro.iwamatsu.yj@renesas.com>
CC: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
13 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux...
Paul Mundt [Thu, 24 Mar 2011 06:17:25 +0000 (15:17 +0900)]
Merge branch 'master' of git://git./linux/kernel/git/torvalds/linux-2.6 into sh-latest

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6
Linus Torvalds [Thu, 24 Mar 2011 03:51:42 +0000 (20:51 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/viro/vfs-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6:
  deal with races in /proc/*/{syscall,stack,personality}
  proc: enable writing to /proc/pid/mem
  proc: make check_mem_permission() return an mm_struct on success
  proc: hold cred_guard_mutex in check_mem_permission()
  proc: disable mem_write after exec
  mm: implement access_remote_vm
  mm: factor out main logic of access_process_vm
  mm: use mm_struct to resolve gate vma's in __get_user_pages
  mm: arch: rename in_gate_area_no_task to in_gate_area_no_mm
  mm: arch: make in_gate_area take an mm_struct instead of a task_struct
  mm: arch: make get_gate_vma take an mm_struct instead of a task_struct
  x86: mark associated mm when running a task in 32 bit compatibility mode
  x86: add context tag to mark mm when running a task in 32-bit compatibility mode
  auxv: require the target to be tracable (or yourself)
  close race in /proc/*/environ
  report errors in /proc/*/*map* sanely
  pagemap: close races with suid execve
  make sessionid permissions in /proc/*/task/* match those in /proc/*
  fix leaks in path_lookupat()

Fix up trivial conflicts in fs/proc/base.c

13 years agoMerge branch 'devel' of master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Thu, 24 Mar 2011 03:37:26 +0000 (20:37 -0700)]
Merge branch 'devel' of /home/rmk/linux-2.6-arm

* 'devel' of master.kernel.org:/home/rmk/linux-2.6-arm: (35 commits)
  ARM: Update (and cut down) mach-types
  ARM: 6771/1: vexpress: add support for multiple core tiles
  ARM: 6797/1: hw_breakpoint: Fix newlines in WARNings
  ARM: 6751/1: vexpress: select applicable errata workarounds in Kconfig
  ARM: 6753/1: omap4: Enable ARM local timers with OMAP4430 es1.0 exception
  ARM: 6759/1: smp: Select local timers vs broadcast timer support runtime
  ARM: pgtable: add pud-level code
  ARM: 6673/1: LPAE: use phys_addr_t instead of unsigned long for start of membanks
  ARM: Use long long format when printing meminfo physical addresses
  ARM: integrator: add Integrator/CP sched_clock support
  ARM: realview/vexpress: consolidate SMP bringup code
  ARM: realview/vexpress: consolidate localtimer support
  ARM: integrator/versatile: consolidate FPGA IRQ handling code
  ARM: rationalize versatile family Kconfig/Makefile
  ARM: realview: remove old AMBA device DMA definitions
  ARM: versatile: remove old AMBA device DMA definitions
  ARM: vexpress: use new init_early for clock tree and sched_clock init
  ARM: realview: use new init_early for clock tree and sched_clock init
  ARM: versatile: use new init_early for clock tree and sched_clock init
  ARM: integrator: use new init_early for clock tree init
  ...

13 years agospi/pl022: Add loopback support for the SPI on 5500
Philippe Langlais [Wed, 23 Mar 2011 10:05:16 +0000 (11:05 +0100)]
spi/pl022: Add loopback support for the SPI on 5500

Extend the vendor data with a loopback field, and add new
amba-pl022 vendor data for the DB5500 pl023, as the pl023
on db8500 and db5500 vary.

Signed-off-by: Prajadevi H <prajadevi.h@stericsson.com>
Signed-off-by: Philippe Langlais <philippe.langlais@linaro.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
13 years agocrash_dump: export is_kdump_kernel to modules, consolidate elfcorehdr_addr, setup_elf...
Olaf Hering [Wed, 23 Mar 2011 23:43:29 +0000 (16:43 -0700)]
crash_dump: export is_kdump_kernel to modules, consolidate elfcorehdr_addr, setup_elfcorehdr and saved_max_pfn

The Xen PV drivers in a crashed HVM guest can not connect to the dom0
backend drivers because both frontend and backend drivers are still in
connected state.  To run the connection reset function only in case of a
crashdump, the is_kdump_kernel() function needs to be available for the PV
driver modules.

Consolidate elfcorehdr_addr, setup_elfcorehdr and saved_max_pfn into
kernel/crash_dump.c Also export elfcorehdr_addr to make is_kdump_kernel()
usable for modules.

Leave 'elfcorehdr' as early_param().  This changes powerpc from __setup()
to early_param().  It adds an address range check from x86 also on ia64
and powerpc.

[akpm@linux-foundation.org: additional #includes]
[akpm@linux-foundation.org: remove elfcorehdr_addr export]
[akpm@linux-foundation.org: fix for Tejun's mm/nobootmem.c changes]
Signed-off-by: Olaf Hering <olaf@aepfle.de>
Cc: Russell King <rmk@arm.linux.org.uk>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoremove dma64_addr_t
FUJITA Tomonori [Wed, 23 Mar 2011 23:43:28 +0000 (16:43 -0700)]
remove dma64_addr_t

There is no user now.

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: David Miller <davem@davemloft.net>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Richard Henderson <rth@twiddle.net>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Cc: Matt Turner <mattst88@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agotaskstats: use appropriate printk priority level
Mandeep Singh Baines [Wed, 23 Mar 2011 23:43:27 +0000 (16:43 -0700)]
taskstats: use appropriate printk priority level

printk()s without a priority level default to KERN_WARNING.  To reduce
noise at KERN_WARNING, this patch set the priority level appriopriately
for unleveled printks()s.  This should be useful to folks that look at
dmesg warnings closely.

Signed-off-by: Mandeep Singh Baines <msb@chromium.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agouserns: rename is_owner_or_cap to inode_owner_or_capable
Serge E. Hallyn [Wed, 23 Mar 2011 23:43:26 +0000 (16:43 -0700)]
userns: rename is_owner_or_cap to inode_owner_or_capable

And give it a kernel-doc comment.

[akpm@linux-foundation.org: btrfs changed in linux-next]
Signed-off-by: Serge E. Hallyn <serge.hallyn@canonical.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Daniel Lezcano <daniel.lezcano@free.fr>
Acked-by: David Howells <dhowells@redhat.com>
Cc: James Morris <jmorris@namei.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agouserns: userns: check user namespace for task->file uid equivalence checks
Serge E. Hallyn [Wed, 23 Mar 2011 23:43:25 +0000 (16:43 -0700)]
userns: userns: check user namespace for task->file uid equivalence checks

Cheat for now and say all files belong to init_user_ns.  Next step will be
to let superblocks belong to a user_ns, and derive inode_userns(inode)
from inode->i_sb->s_user_ns.  Finally we'll introduce more flexible
arrangements.

Changelog:
Feb 15: make is_owner_or_cap take const struct inode
Feb 23: make is_owner_or_cap bool

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Serge E. Hallyn <serge.hallyn@canonical.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Acked-by: Daniel Lezcano <daniel.lezcano@free.fr>
Acked-by: David Howells <dhowells@redhat.com>
Cc: James Morris <jmorris@namei.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agouserns: user namespaces: convert several capable() calls
Serge E. Hallyn [Wed, 23 Mar 2011 23:43:24 +0000 (16:43 -0700)]
userns: user namespaces: convert several capable() calls

CAP_IPC_OWNER and CAP_IPC_LOCK can be checked against current_user_ns(),
because the resource comes from current's own ipc namespace.

setuid/setgid are to uids in own namespace, so again checks can be against
current_user_ns().

Changelog:
Jan 11: Use task_ns_capable() in place of sched_capable().
Jan 11: Use nsown_capable() as suggested by Bastian Blank.
Jan 11: Clarify (hopefully) some logic in futex and sched.c
Feb 15: use ns_capable for ipc, not nsown_capable
Feb 23: let copy_ipcs handle setting ipc_ns->user_ns
Feb 23: pass ns down rather than taking it from current

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Serge E. Hallyn <serge.hallyn@canonical.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Acked-by: Daniel Lezcano <daniel.lezcano@free.fr>
Acked-by: David Howells <dhowells@redhat.com>
Cc: James Morris <jmorris@namei.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agouserns: add a user namespace owner of ipc ns
Serge E. Hallyn [Wed, 23 Mar 2011 23:43:23 +0000 (16:43 -0700)]
userns: add a user namespace owner of ipc ns

Changelog:
Feb 15: Don't set new ipc->user_ns if we didn't create a new
ipc_ns.
Feb 23: Move extern declaration to ipc_namespace.h, and group
fwd declarations at top.

Signed-off-by: Serge E. Hallyn <serge.hallyn@canonical.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Acked-by: Daniel Lezcano <daniel.lezcano@free.fr>
Acked-by: David Howells <dhowells@redhat.com>
Cc: James Morris <jmorris@namei.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agouserns: user namespaces: convert all capable checks in kernel/sys.c
Serge E. Hallyn [Wed, 23 Mar 2011 23:43:22 +0000 (16:43 -0700)]
userns: user namespaces: convert all capable checks in kernel/sys.c

This allows setuid/setgid in containers.  It also fixes some corner cases
where kernel logic foregoes capability checks when uids are equivalent.
The latter will need to be done throughout the whole kernel.

Changelog:
Jan 11: Use nsown_capable() as suggested by Bastian Blank.
Jan 11: Fix logic errors in uid checks pointed out by Bastian.
Feb 15: allow prlimit to current (was regression in previous version)
Feb 23: remove debugging printks, uninline set_one_prio_perm and
make it bool, and document its return value.

Signed-off-by: Serge E. Hallyn <serge.hallyn@canonical.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Acked-by: Daniel Lezcano <daniel.lezcano@free.fr>
Acked-by: David Howells <dhowells@redhat.com>
Cc: James Morris <jmorris@namei.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agouserns: make has_capability* into real functions
Serge E. Hallyn [Wed, 23 Mar 2011 23:43:21 +0000 (16:43 -0700)]
userns: make has_capability* into real functions

So we can let type safety keep things sane, and as a bonus we can remove
the declaration of init_user_ns in capability.h.

Signed-off-by: Serge E. Hallyn <serge.hallyn@canonical.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Daniel Lezcano <daniel.lezcano@free.fr>
Cc: David Howells <dhowells@redhat.com>
Cc: James Morris <jmorris@namei.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agouserns: allow ptrace from non-init user namespaces
Serge E. Hallyn [Wed, 23 Mar 2011 23:43:20 +0000 (16:43 -0700)]
userns: allow ptrace from non-init user namespaces

ptrace is allowed to tasks in the same user namespace according to the
usual rules (i.e.  the same rules as for two tasks in the init user
namespace).  ptrace is also allowed to a user namespace to which the
current task the has CAP_SYS_PTRACE capability.

Changelog:
Dec 31: Address feedback by Eric:
. Correct ptrace uid check
. Rename may_ptrace_ns to ptrace_capable
. Also fix the cap_ptrace checks.
Jan  1: Use const cred struct
Jan 11: use task_ns_capable() in place of ptrace_capable().
Feb 23: same_or_ancestore_user_ns() was not an appropriate
check to constrain cap_issubset.  Rather, cap_issubset()
only is meaningful when both capsets are in the same
user_ns.

Signed-off-by: Serge E. Hallyn <serge.hallyn@canonical.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Acked-by: Daniel Lezcano <daniel.lezcano@free.fr>
Acked-by: David Howells <dhowells@redhat.com>
Cc: James Morris <jmorris@namei.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agouserns: allow killing tasks in your own or child userns
Serge E. Hallyn [Wed, 23 Mar 2011 23:43:19 +0000 (16:43 -0700)]
userns: allow killing tasks in your own or child userns

Changelog:
Dec  8: Fixed bug in my check_kill_permission pointed out by
        Eric Biederman.
Dec 13: Apply Eric's suggestion to pass target task into kill_ok_by_cred()
        for clarity
Dec 31: address comment by Eric Biederman:
don't need cred/tcred in check_kill_permission.
Jan  1: use const cred struct.
Jan 11: Per Bastian Blank's advice, clean up kill_ok_by_cred().
Feb 16: kill_ok_by_cred: fix bad parentheses
Feb 23: per akpm, let compiler inline kill_ok_by_cred

Signed-off-by: Serge E. Hallyn <serge.hallyn@canonical.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Acked-by: Daniel Lezcano <daniel.lezcano@free.fr>
Acked-by: David Howells <dhowells@redhat.com>
Cc: James Morris <jmorris@namei.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agouserns: allow sethostname in a container
Serge E. Hallyn [Wed, 23 Mar 2011 23:43:18 +0000 (16:43 -0700)]
userns: allow sethostname in a container

Changelog:
Feb 23: let clone_uts_ns() handle setting uts->user_ns
To do so we need to pass in the task_struct who'll
get the utsname, so we can get its user_ns.
Feb 23: As per Oleg's coment, just pass in tsk, instead of two
of its members.

Signed-off-by: Serge E. Hallyn <serge.hallyn@canonical.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Acked-by: Daniel Lezcano <daniel.lezcano@free.fr>
Acked-by: David Howells <dhowells@redhat.com>
Cc: James Morris <jmorris@namei.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agouserns: security: make capabilities relative to the user namespace
Serge E. Hallyn [Wed, 23 Mar 2011 23:43:17 +0000 (16:43 -0700)]
userns: security: make capabilities relative to the user namespace

- Introduce ns_capable to test for a capability in a non-default
  user namespace.
- Teach cap_capable to handle capabilities in a non-default
  user namespace.

The motivation is to get to the unprivileged creation of new
namespaces.  It looks like this gets us 90% of the way there, with
only potential uid confusion issues left.

I still need to handle getting all caps after creation but otherwise I
think I have a good starter patch that achieves all of your goals.

Changelog:
11/05/2010: [serge] add apparmor
12/14/2010: [serge] fix capabilities to created user namespaces
Without this, if user serge creates a user_ns, he won't have
capabilities to the user_ns he created.  THis is because we
were first checking whether his effective caps had the caps
he needed and returning -EPERM if not, and THEN checking whether
he was the creator.  Reverse those checks.
12/16/2010: [serge] security_real_capable needs ns argument in !security case
01/11/2011: [serge] add task_ns_capable helper
01/11/2011: [serge] add nsown_capable() helper per Bastian Blank suggestion
02/16/2011: [serge] fix a logic bug: the root user is always creator of
    init_user_ns, but should not always have capabilities to
    it!  Fix the check in cap_capable().
02/21/2011: Add the required user_ns parameter to security_capable,
    fixing a compile failure.
02/23/2011: Convert some macros to functions as per akpm comments.  Some
    couldn't be converted because we can't easily forward-declare
    them (they are inline if !SECURITY, extern if SECURITY).  Add
    a current_user_ns function so we can use it in capability.h
    without #including cred.h.  Move all forward declarations
    together to the top of the #ifdef __KERNEL__ section, and use
    kernel-doc format.
02/23/2011: Per dhowells, clean up comment in cap_capable().
02/23/2011: Per akpm, remove unreachable 'return -EPERM' in cap_capable.

(Original written and signed off by Eric;  latest, modified version
acked by him)

[akpm@linux-foundation.org: fix build]
[akpm@linux-foundation.org: export current_user_ns() for ecryptfs]
[serge.hallyn@canonical.com: remove unneeded extra argument in selinux's task_has_capability]
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Serge E. Hallyn <serge.hallyn@canonical.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Acked-by: Daniel Lezcano <daniel.lezcano@free.fr>
Acked-by: David Howells <dhowells@redhat.com>
Cc: James Morris <jmorris@namei.org>
Signed-off-by: Serge E. Hallyn <serge.hallyn@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agouserns: add a user_namespace as creator/owner of uts_namespace
Serge E. Hallyn [Wed, 23 Mar 2011 23:43:16 +0000 (16:43 -0700)]
userns: add a user_namespace as creator/owner of uts_namespace

The expected course of development for user namespaces targeted
capabilities is laid out at https://wiki.ubuntu.com/UserNamespace.

Goals:

- Make it safe for an unprivileged user to unshare namespaces.  They
  will be privileged with respect to the new namespace, but this should
  only include resources which the unprivileged user already owns.

- Provide separate limits and accounting for userids in different
  namespaces.

Status:

  Currently (as of 2.6.38) you can clone with the CLONE_NEWUSER flag to
  get a new user namespace if you have the CAP_SYS_ADMIN, CAP_SETUID, and
  CAP_SETGID capabilities.  What this gets you is a whole new set of
  userids, meaning that user 500 will have a different 'struct user' in
  your namespace than in other namespaces.  So any accounting information
  stored in struct user will be unique to your namespace.

  However, throughout the kernel there are checks which

  - simply check for a capability.  Since root in a child namespace
    has all capabilities, this means that a child namespace is not
    constrained.

  - simply compare uid1 == uid2.  Since these are the integer uids,
    uid 500 in namespace 1 will be said to be equal to uid 500 in
    namespace 2.

  As a result, the lxc implementation at lxc.sf.net does not use user
  namespaces.  This is actually helpful because it leaves us free to
  develop user namespaces in such a way that, for some time, user
  namespaces may be unuseful.

Bugs aside, this patchset is supposed to not at all affect systems which
are not actively using user namespaces, and only restrict what tasks in
child user namespace can do.  They begin to limit privilege to a user
namespace, so that root in a container cannot kill or ptrace tasks in the
parent user namespace, and can only get world access rights to files.
Since all files currently belong to the initila user namespace, that means
that child user namespaces can only get world access rights to *all*
files.  While this temporarily makes user namespaces bad for system
containers, it starts to get useful for some sandboxing.

I've run the 'runltplite.sh' with and without this patchset and found no
difference.

This patch:

copy_process() handles CLONE_NEWUSER before the rest of the namespaces.
So in the case of clone(CLONE_NEWUSER|CLONE_NEWUTS) the new uts namespace
will have the new user namespace as its owner.  That is what we want,
since we want root in that new userns to be able to have privilege over
it.

Changelog:
Feb 15: don't set uts_ns->user_ns if we didn't create
a new uts_ns.
Feb 23: Move extern init_user_ns declaration from
init/version.c to utsname.h.

Signed-off-by: Serge E. Hallyn <serge.hallyn@canonical.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Acked-by: Daniel Lezcano <daniel.lezcano@free.fr>
Acked-by: David Howells <dhowells@redhat.com>
Cc: James Morris <jmorris@namei.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoprocfs: kill the global proc_mnt variable
Oleg Nesterov [Wed, 23 Mar 2011 23:43:14 +0000 (16:43 -0700)]
procfs: kill the global proc_mnt variable

After the previous cleanup in proc_get_sb() the global proc_mnt has no
reasons to exists, kill it.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Daniel Lezcano <daniel.lezcano@free.fr>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Acked-by: Serge E. Hallyn <serge@hallyn.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agopidns: call pid_ns_prepare_proc() from create_pid_namespace()
Eric W. Biederman [Wed, 23 Mar 2011 23:43:13 +0000 (16:43 -0700)]
pidns: call pid_ns_prepare_proc() from create_pid_namespace()

Reorganize proc_get_sb() so it can be called before the struct pid of the
first process is allocated.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Daniel Lezcano <daniel.lezcano@free.fr>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Acked-by: Serge E. Hallyn <serge@hallyn.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agopid: remove the child_reaper special case in init/main.c
Eric W. Biederman [Wed, 23 Mar 2011 23:43:12 +0000 (16:43 -0700)]
pid: remove the child_reaper special case in init/main.c

This patchset is a cleanup and a preparation to unshare the pid namespace.
These prerequisites prepare for Eric's patchset to give a file descriptor
to a namespace and join an existing namespace.

This patch:

It turns out that the existing assignment in copy_process of the
child_reaper can handle the initial assignment of child_reaper we just
need to generalize the test in kernel/fork.c

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Daniel Lezcano <daniel.lezcano@free.fr>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Acked-by: Serge E. Hallyn <serge@hallyn.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agosysctl: restrict write access to dmesg_restrict
Richard Weinberger [Wed, 23 Mar 2011 23:43:11 +0000 (16:43 -0700)]
sysctl: restrict write access to dmesg_restrict

When dmesg_restrict is set to 1 CAP_SYS_ADMIN is needed to read the kernel
ring buffer.  But a root user without CAP_SYS_ADMIN is able to reset
dmesg_restrict to 0.

This is an issue when e.g.  LXC (Linux Containers) are used and complete
user space is running without CAP_SYS_ADMIN.  A unprivileged and jailed
root user can bypass the dmesg_restrict protection.

With this patch writing to dmesg_restrict is only allowed when root has
CAP_SYS_ADMIN.

Signed-off-by: Richard Weinberger <richard@nod.at>
Acked-by: Dan Rosenberg <drosenberg@vsecurity.com>
Acked-by: Serge E. Hallyn <serge@hallyn.com>
Cc: Eric Paris <eparis@redhat.com>
Cc: Kees Cook <kees.cook@canonical.com>
Cc: James Morris <jmorris@namei.org>
Cc: Eugene Teo <eugeneteo@kernel.org>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agosysctl: add some missing input constraint checks
Petr Holasek [Wed, 23 Mar 2011 23:43:09 +0000 (16:43 -0700)]
sysctl: add some missing input constraint checks

Add boundaries of allowed input ranges for: dirty_expire_centisecs,
drop_caches, overcommit_memory, page-cluster and panic_on_oom.

Signed-off-by: Petr Holasek <pholasek@redhat.com>
Acked-by: Dave Young <hidave.darkstar@gmail.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agosysctl_check: drop dead code
Denis Kirjanov [Wed, 23 Mar 2011 23:43:08 +0000 (16:43 -0700)]
sysctl_check: drop dead code

Drop dead code.

Signed-off-by: Denis Kirjanov <dkirjanov@kernel.org>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agosysctl_check: drop table->procname checks
Denis Kirjanov [Wed, 23 Mar 2011 23:43:08 +0000 (16:43 -0700)]
sysctl_check: drop table->procname checks

Since the for loop checks for the table->procname drop useless
table->procname checks inside the loop body

Signed-off-by: Denis Kirjanov <dkirjanov@kernel.org>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agorapidio: fix potential null deref on failure path
Dan Carpenter [Wed, 23 Mar 2011 23:43:07 +0000 (16:43 -0700)]
rapidio: fix potential null deref on failure path

If rio is not a switch then "rswitch" is null.

Signed-off-by: Dan Carpenter <error27@gmail.com>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agorapidio: remove mport resource reservation from common RIO code
Alexandre Bounine [Wed, 23 Mar 2011 23:43:06 +0000 (16:43 -0700)]
rapidio: remove mport resource reservation from common RIO code

Removes resource reservation from the common sybsystem initialization code
and make it part of mport driver initialization.  This resolves conflict
with resource reservation by device specific mport drivers.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Kumar Gala <galak@kernel.crashing.org>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Li Yang <leoli@freescale.com>
Cc: Thomas Moll <thomas.moll@sysgo.com>
Cc: Micha Nelissen <micha@neli.hopto.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agorapidio: modify mport ID assignment
Alexandre Bounine [Wed, 23 Mar 2011 23:43:05 +0000 (16:43 -0700)]
rapidio: modify mport ID assignment

Changes mport ID and host destination ID assignment to implement unified
method common to all mport drivers.  Makes "riohdid=" kernel command line
parameter common for all architectures with support for more that one host
destination ID assignment.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Kumar Gala <galak@kernel.crashing.org>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Li Yang <leoli@freescale.com>
Cc: Thomas Moll <thomas.moll@sysgo.com>
Cc: Micha Nelissen <micha@neli.hopto.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agorapidio: modify subsystem and driver initialization sequence
Alexandre Bounine [Wed, 23 Mar 2011 23:43:04 +0000 (16:43 -0700)]
rapidio: modify subsystem and driver initialization sequence

Subsystem initialization sequence modified to support presence of multiple
RapidIO controllers in the system.  The new sequence is compatible with
initialization of PCI devices.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Kumar Gala <galak@kernel.crashing.org>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Li Yang <leoli@freescale.com>
Cc: Thomas Moll <thomas.moll@sysgo.com>
Cc: Micha Nelissen <micha@neli.hopto.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agorapidio: modify configuration to support PCI-SRIO controller
Alexandre Bounine [Wed, 23 Mar 2011 23:43:03 +0000 (16:43 -0700)]
rapidio: modify configuration to support PCI-SRIO controller

1. Add an option to include RapidIO support if the PCI is available.
2. Add FSL_RIO configuration option to enable controller selection.
3. Add RapidIO support option into x86 and MIPS architectures.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Acked-by: Kumar Gala <galak@kernel.crashing.org>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Li Yang <leoli@freescale.com>
Cc: Thomas Moll <thomas.moll@sysgo.com>
Cc: Micha Nelissen <micha@neli.hopto.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agorapidio: add architecture specific callbacks
Alexandre Bounine [Wed, 23 Mar 2011 23:43:02 +0000 (16:43 -0700)]
rapidio: add architecture specific callbacks

This set of patches eliminates RapidIO dependency on PowerPC architecture
and makes it available to other architectures (x86 and MIPS).  It also
enables support of new platform independent RapidIO controllers such as
PCI-to-SRIO and PCI Express-to-SRIO.

This patch:

Extend number of mport callback functions to eliminate direct linking of
architecture specific mport operations.

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Kumar Gala <galak@kernel.crashing.org>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Li Yang <leoli@freescale.com>
Cc: Thomas Moll <thomas.moll@sysgo.com>
Cc: Micha Nelissen <micha@neli.hopto.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agorapidio: add RapidIO documentation
Alexandre Bounine [Wed, 23 Mar 2011 23:43:00 +0000 (16:43 -0700)]
rapidio: add RapidIO documentation

Add RapidIO documentation files as it was discussed earlier (see thread
http://marc.info/?l=linux-kernel&m=129202338918062&w=2)

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Kumar Gala <galak@kernel.crashing.org>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agorapidio: add new sysfs attributes
Alexandre Bounine [Wed, 23 Mar 2011 23:42:59 +0000 (16:42 -0700)]
rapidio: add new sysfs attributes

Add new sysfs attributes.

1. Routing information required to to reach the RIO device:
destid - device destination ID (real for for endpoint, route for switch)
hopcount - hopcount for maintenance requests (switches only)

2. device linking information:
lprev - name of device that precedes the given device in the enumeration
        or discovery order (displayed along with of the port to which it
        is attached).
lnext - names of devices (with corresponding port numbers) that are
        attached to the given device as next in the enumeration or
        discovery order (switches only)

Signed-off-by: Alexandre Bounine <alexandre.bounine@idt.com>
Cc: Kumar Gala <galak@kernel.crashing.org>
Cc: Matt Porter <mporter@kernel.crashing.org>
Cc: Li Yang <leoli@freescale.com>
Cc: Thomas Moll <thomas.moll@sysgo.com>
Cc: Micha Nelissen <micha@neli.hopto.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agodrivers/char/mem.c: clean up the code
Changli Gao [Wed, 23 Mar 2011 23:42:58 +0000 (16:42 -0700)]
drivers/char/mem.c: clean up the code

Reduce the lines of code and simplify the logic.

Signed-off-by: Changli Gao <xiaosuo@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agodrivers/staging/tty/specialix.c: convert func_enter to func_exit
Julia Lawall [Wed, 23 Mar 2011 23:42:57 +0000 (16:42 -0700)]
drivers/staging/tty/specialix.c: convert func_enter to func_exit

Convert calls to func_enter on leaving a function to func_exit.

The semantic patch that fixes this problem is as follows:
(http://coccinelle.lip6.fr/)

// <smpl>
@@
@@

- func_enter();
+ func_exit();
  return...;
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Cc: Roger Wolff <R.E.Wolff@BitWizard.nl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agodrivers/tty/bfin_jtag_comm.c: avoid calling put_tty_driver on NULL
Julia Lawall [Wed, 23 Mar 2011 23:42:56 +0000 (16:42 -0700)]
drivers/tty/bfin_jtag_comm.c: avoid calling put_tty_driver on NULL

put_tty_driver calls tty_driver_kref_put on its argument, and then
tty_driver_kref_put calls kref_put on the address of a field of this
argument.  kref_put checks for NULL, but in this case the field is likely
to have some offset and so the result of taking its address will not be
NULL.  Labels are added to be able to skip over the call to put_tty_driver
when the argument will be NULL.

The semantic match that finds this problem is as follows:
(http://coccinelle.lip6.fr/)

// <smpl>
@@
expression *x;
@@

*if (x == NULL)
{ ...
* put_tty_driver(x);
  ...
  return ...;
}
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Cc: Torben Hohn <torbenh@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agodrivers/char: add MSM smd_pkt driver
Niranjana Vishwanathapura [Wed, 23 Mar 2011 23:42:55 +0000 (16:42 -0700)]
drivers/char: add MSM smd_pkt driver

Add smd_pkt driver which provides device interface to smd packet ports.

Signed-off-by: Niranjana Vishwanathapura <nvishwan@codeaurora.org>
Cc: Brian Swetland <swetland@google.com>
Cc: Greg KH <gregkh@suse.de>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: David Brown <davidb@codeaurora.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agodrivers/char/ipmi/ipmi_si_intf.c: fix cleanup_one_si section mismatch
Sergey Senozhatsky [Wed, 23 Mar 2011 23:42:54 +0000 (16:42 -0700)]
drivers/char/ipmi/ipmi_si_intf.c: fix cleanup_one_si section mismatch

commit d2478521afc2022 ("char/ipmi: fix OOPS caused by
pnp_unregister_driver on unregistered driver") introduced a section
mismatch by calling __exit cleanup_ipmi_si from __devinit init_ipmi_si.

Remove __exit annotation from cleanup_ipmi_si.

Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Acked-by: Corey Minyard <cminyard@mvista.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoproc: protect mm start_code/end_code in /proc/pid/stat
Kees Cook [Wed, 23 Mar 2011 23:42:53 +0000 (16:42 -0700)]
proc: protect mm start_code/end_code in /proc/pid/stat

While mm->start_stack was protected from cross-uid viewing (commit
f83ce3e6b02d5 ("proc: avoid information leaks to non-privileged
processes")), the start_code and end_code values were not.  This would
allow the text location of a PIE binary to leak, defeating ASLR.

Note that the value "1" is used instead of "0" for a protected value since
"ps", "killall", and likely other readers of /proc/pid/stat, take
start_code of "0" to mean a kernel thread and will misbehave.  Thanks to
Brad Spengler for pointing this out.

Addresses CVE-2011-0726

Signed-off-by: Kees Cook <kees.cook@canonical.com>
Cc: <stable@kernel.org>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Eugene Teo <eugeneteo@kernel.sg>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Brad Spengler <spender@grsecurity.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoproc: make struct proc_dir_entry::namelen unsigned int
Alexey Dobriyan [Wed, 23 Mar 2011 23:42:52 +0000 (16:42 -0700)]
proc: make struct proc_dir_entry::namelen unsigned int

1. namelen is declared "unsigned short" which hints for "maybe space savings".
   Indeed in 2.4 struct proc_dir_entry looked like:

        struct proc_dir_entry {
                unsigned short low_ino;
                unsigned short namelen;

   Now, low_ino is "unsigned int", all savings were gone for a long time.
   "struct proc_dir_entry" is not that countless to worry about it's size,
   anyway.

2. converting from unsigned short to int/unsigned int can only create
   problems, we better play it safe.

Space is not really conserved, because of natural alignment for the next
field.  sizeof(struct proc_dir_entry) remains the same.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoprocfs: fix some wrong error code usage
Jovi Zhang [Wed, 23 Mar 2011 23:42:51 +0000 (16:42 -0700)]
procfs: fix some wrong error code usage

[root@wei 1]# cat /proc/1/mem
cat: /proc/1/mem: No such process

error code -ESRCH is wrong in this situation.  Return -EPERM instead.

Signed-off-by: Jovi Zhang <bookjovi@gmail.com>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoprocfs: fix /proc/<pid>/maps heap check
Aaro Koskinen [Wed, 23 Mar 2011 23:42:50 +0000 (16:42 -0700)]
procfs: fix /proc/<pid>/maps heap check

The current code fails to print the "[heap]" marking if the heap is split
into multiple mappings.

Fix the check so that the marking is displayed in all possible cases:
1. vma matches exactly the heap
2. the heap vma is merged e.g. with bss
3. the heap vma is splitted e.g. due to locked pages

Test cases. In all cases, the process should have mapping(s) with
[heap] marking:

(1) vma matches exactly the heap

#include <stdio.h>
#include <unistd.h>
#include <sys/types.h>

int main (void)
{
if (sbrk(4096) != (void *)-1) {
printf("check /proc/%d/maps\n", (int)getpid());
while (1)
sleep(1);
}
return 0;
}

# ./test1
check /proc/553/maps
[1] + Stopped                    ./test1
# cat /proc/553/maps | head -4
00008000-00009000 r-xp 00000000 01:00 3113640    /test1
00010000-00011000 rw-p 00000000 01:00 3113640    /test1
00011000-00012000 rw-p 00000000 00:00 0          [heap]
4006f000-40070000 rw-p 00000000 00:00 0

(2) the heap vma is merged

#include <stdio.h>
#include <unistd.h>
#include <sys/types.h>

char foo[4096] = "foo";
char bar[4096];

int main (void)
{
if (sbrk(4096) != (void *)-1) {
printf("check /proc/%d/maps\n", (int)getpid());
while (1)
sleep(1);
}
return 0;
}

# ./test2
check /proc/556/maps
[2] + Stopped                    ./test2
# cat /proc/556/maps | head -4
00008000-00009000 r-xp 00000000 01:00 3116312    /test2
00010000-00012000 rw-p 00000000 01:00 3116312    /test2
00012000-00014000 rw-p 00000000 00:00 0          [heap]
4004a000-4004b000 rw-p 00000000 00:00 0

(3) the heap vma is splitted (this fails without the patch)

#include <stdio.h>
#include <unistd.h>
#include <sys/mman.h>
#include <sys/types.h>

int main (void)
{
if ((sbrk(4096) != (void *)-1) && !mlockall(MCL_FUTURE) &&
    (sbrk(4096) != (void *)-1)) {
printf("check /proc/%d/maps\n", (int)getpid());
while (1)
sleep(1);
}
return 0;
}

# ./test3
check /proc/559/maps
[1] + Stopped                    ./test3
# cat /proc/559/maps|head -4
00008000-00009000 r-xp 00000000 01:00 3119108    /test3
00010000-00011000 rw-p 00000000 01:00 3119108    /test3
00011000-00012000 rw-p 00000000 00:00 0          [heap]
00012000-00013000 rw-p 00000000 00:00 0          [heap]

It looks like the bug has been there forever, and since it only results in
some information missing from a procfile, it does not fulfil the -stable
"critical issue" criteria.

Signed-off-by: Aaro Koskinen <aaro.koskinen@nokia.com>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoproc: hide kernel addresses via %pK in /proc/<pid>/stack
Konstantin Khlebnikov [Wed, 23 Mar 2011 23:42:48 +0000 (16:42 -0700)]
proc: hide kernel addresses via %pK in /proc/<pid>/stack

This file is readable for the task owner.  Hide kernel addresses from
unprivileged users, leave them function names and offsets.

Signed-off-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
Acked-by: Kees Cook <kees.cook@canonical.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agocpuset: hold callback_mutex in cpuset_post_clone()
Li Zefan [Wed, 23 Mar 2011 23:42:48 +0000 (16:42 -0700)]
cpuset: hold callback_mutex in cpuset_post_clone()

Chaning cpuset->mems/cpuset->cpus should be protected under
callback_mutex.

cpuset_clone() doesn't follow this rule. It's ok because it's
called when creating and initializing a cgroup, but we'd better
hold the lock to avoid subtil break in the future.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Acked-by: Paul Menage <menage@google.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agocpuset: fix unchecked calls to NODEMASK_ALLOC()
Li Zefan [Wed, 23 Mar 2011 23:42:47 +0000 (16:42 -0700)]
cpuset: fix unchecked calls to NODEMASK_ALLOC()

Those functions that use NODEMASK_ALLOC() can't propagate errno
to users, but will fail silently.

Fix it by using a static nodemask_t variable for each function, and
those variables are protected by cgroup_mutex;

[akpm@linux-foundation.org: fix comment spelling, strengthen cgroup_lock comment]
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agocpuset: remove unneeded NODEMASK_ALLOC() in cpuset_attach()
Li Zefan [Wed, 23 Mar 2011 23:42:46 +0000 (16:42 -0700)]
cpuset: remove unneeded NODEMASK_ALLOC() in cpuset_attach()

oldcs->mems_allowed is not modified during cpuset_attach(), so we don't
have to copy it to a buffer allocated by NODEMASK_ALLOC().  Just pass it
to cpuset_migrate_mm().

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Cc: Paul Menage <menage@google.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agocpuset: remove unneeded NODEMASK_ALLOC() in cpuset_sprintf_memlist()
Li Zefan [Wed, 23 Mar 2011 23:42:45 +0000 (16:42 -0700)]
cpuset: remove unneeded NODEMASK_ALLOC() in cpuset_sprintf_memlist()

It's not necessary to copy cpuset->mems_allowed to a buffer allocated by
NODEMASK_ALLOC().  Just pass it to nodelist_scnprintf().

As spotted by Paul, a side effect is we fix a bug that the function can
return -ENOMEM but the caller doesn't expect negative return value.
Therefore change the return value of cpuset_sprintf_cpulist() and
cpuset_sprintf_memlist() from int to size_t.

Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Acked-by: Paul Menage <menage@google.com>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: give current access to memory reserves if it's trying to die
David Rientjes [Wed, 23 Mar 2011 23:42:44 +0000 (16:42 -0700)]
memcg: give current access to memory reserves if it's trying to die

When a memcg is oom and current has already received a SIGKILL, then give
it access to memory reserves with a higher scheduling priority so that it
may quickly exit and free its memory.

This is identical to the global oom killer and is done even before
checking for panic_on_oom: a pending SIGKILL here while panic_on_oom is
selected is guaranteed to have come from userspace; the thread only needs
access to memory reserves to exit and thus we don't unnecessarily panic
the machine until the kernel has no last resort to free memory.

Signed-off-by: David Rientjes <rientjes@google.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: fix leak on wrong LRU with FUSE
KAMEZAWA Hiroyuki [Wed, 23 Mar 2011 23:42:42 +0000 (16:42 -0700)]
memcg: fix leak on wrong LRU with FUSE

fs/fuse/dev.c::fuse_try_move_page() does

   (1) remove a page by ->steal()
   (2) re-add the page to page cache
   (3) link the page to LRU if it was not on LRU at (1)

This implies the page is _on_ LRU when it's added to radix-tree.  So, the
page is added to memory cgroup while it's on LRU.  because LRU is lazy and
no one flushs it.

This is the same behavior as SwapCache and needs special care as
 - remove page from LRU before overwrite pc->mem_cgroup.
 - add page to LRU after overwrite pc->mem_cgroup.

And we need to taking care of pagevec.

If PageLRU(page) is set before we add PCG_USED bit, the page will not be
added to memcg's LRU (in short period).  So, regardlress of PageLRU(page)
value before commit_charge(), we need to check PageLRU(page) after
commit_charge().

Addresses https://bugzilla.kernel.org/show_bug.cgi?id=30432

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Miklos Szeredi <miklos@szeredi.hu>
Cc: Balbir Singh <balbir@in.ibm.com>
Reported-by: Daniel Poelzleithner <poelzi@poelzi.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: page_cgroup array is never stored on reserved pages
Michal Hocko [Wed, 23 Mar 2011 23:42:41 +0000 (16:42 -0700)]
memcg: page_cgroup array is never stored on reserved pages

KAMEZAWA Hiroyuki noted that free_pages_cgroup doesn't have to check for
PageReserved because we never store the array on reserved pages (neither
alloc_pages_exact nor vmalloc use those pages).

So we can replace the check by a BUG_ON.

Signed-off-by: Michal Hocko <mhocko@suse.cz>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Balbir Singh <balbir@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agopage_cgroup: reduce allocation overhead for page_cgroup array for CONFIG_SPARSEMEM
Michal Hocko [Wed, 23 Mar 2011 23:42:40 +0000 (16:42 -0700)]
page_cgroup: reduce allocation overhead for page_cgroup array for CONFIG_SPARSEMEM

Currently we are allocating a single page_cgroup array per memory section
(stored in mem_section->base) when CONFIG_SPARSEMEM is selected.  This is
correct but memory inefficient solution because the allocated memory
(unless we fall back to vmalloc) is not kmalloc friendly:

        - 32b - 16384 entries (20B per entry) fit into 327680B so the
          524288B slab cache is used
        - 32b with PAE - 131072 entries with 2621440B fit into 4194304B
        - 64b - 32768 entries (40B per entry) fit into 2097152 cache

This is ~37% wasted space per memory section and it sumps up for the whole
memory.  On a x86_64 machine it is something like 6MB per 1GB of RAM.

We can reduce the internal fragmentation by using alloc_pages_exact which
allocates PAGE_SIZE aligned blocks so we will get down to <4kB wasted
memory per section which is much better.

We still need a fallback to vmalloc because we have no guarantees that we
will have a continuous memory of that size (order-10) later on during the
hotplug events.

[hannes@cmpxchg.org: do not define unused free_page_cgroup() without memory hotplug]
Signed-off-by: Michal Hocko <mhocko@suse.cz>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Balbir Singh <balbir@in.ibm.com>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomm/memcontrol.c: suppress uninitialized-var warning with older gcc's
Andrew Morton [Wed, 23 Mar 2011 23:42:39 +0000 (16:42 -0700)]
mm/memcontrol.c: suppress uninitialized-var warning with older gcc's

mm/memcontrol.c: In function 'mem_cgroup_force_empty':
mm/memcontrol.c:2280: warning: 'flags' may be used uninitialized in this function

It's a false positive.

Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Greg Thelen <gthelen@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: use native word page statistics counters
Johannes Weiner [Wed, 23 Mar 2011 23:42:38 +0000 (16:42 -0700)]
memcg: use native word page statistics counters

The statistic counters are in units of pages, there is no reason to make
them 64-bit wide on 32-bit machines.

Make them native words.  Since they are signed, this leaves 31 bit on
32-bit machines, which can represent roughly 8TB assuming a page size of
4k.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Greg Thelen <gthelen@google.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: break out event counters from other stats
Johannes Weiner [Wed, 23 Mar 2011 23:42:37 +0000 (16:42 -0700)]
memcg: break out event counters from other stats

For increasing and decreasing per-cpu cgroup usage counters it makes sense
to use signed types, as single per-cpu values might go negative during
updates.  But this is not the case for only-ever-increasing event
counters.

All the counters have been signed 64-bit so far, which was enough to count
events even with the sign bit wasted.

This patch:
- divides s64 counters into signed usage counters and unsigned
  monotonically increasing event counters.
- converts unsigned event counters into 'unsigned long' rather than
  'u64'.  This matches the type used by the /proc/vmstat event counters.

The next patch narrows the signed usage counters type (on 32-bit CPUs,
that is).

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Greg Thelen <gthelen@google.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: unify charge/uncharge quantities to units of pages
Johannes Weiner [Wed, 23 Mar 2011 23:42:36 +0000 (16:42 -0700)]
memcg: unify charge/uncharge quantities to units of pages

There is no clear pattern when we pass a page count and when we pass a
byte count that is a multiple of PAGE_SIZE.

We never charge or uncharge subpage quantities, so convert it all to page
counts.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: convert uncharge batching from bytes to page granularity
Johannes Weiner [Wed, 23 Mar 2011 23:42:35 +0000 (16:42 -0700)]
memcg: convert uncharge batching from bytes to page granularity

We never uncharge subpage quantities.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: convert per-cpu stock from bytes to page granularity
Johannes Weiner [Wed, 23 Mar 2011 23:42:34 +0000 (16:42 -0700)]
memcg: convert per-cpu stock from bytes to page granularity

We never keep subpage quantities in the per-cpu stock.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: keep only one charge cancelling function
Johannes Weiner [Wed, 23 Mar 2011 23:42:33 +0000 (16:42 -0700)]
memcg: keep only one charge cancelling function

We have two charge cancelling functions: one takes a page count, the other
a page size.  The second one just divides the parameter by PAGE_SIZE and
then calls the first one.  This is trivial, no need for an extra function.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: remove memcg->reclaim_param_lock
Johannes Weiner [Wed, 23 Mar 2011 23:42:32 +0000 (16:42 -0700)]
memcg: remove memcg->reclaim_param_lock

The reclaim_param_lock is only taken around single reads and writes to
integer variables and is thus superfluous.  Drop it.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: charged pages always have valid per-memcg zone info
Johannes Weiner [Wed, 23 Mar 2011 23:42:31 +0000 (16:42 -0700)]
memcg: charged pages always have valid per-memcg zone info

page_cgroup_zoneinfo() will never return NULL for a charged page, remove
the check for it in mem_cgroup_get_reclaim_stat_from_page().

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: remove direct page_cgroup-to-page pointer
Johannes Weiner [Wed, 23 Mar 2011 23:42:30 +0000 (16:42 -0700)]
memcg: remove direct page_cgroup-to-page pointer

In struct page_cgroup, we have a full word for flags but only a few are
reserved.  Use the remaining upper bits to encode, depending on
configuration, the node or the section, to enable page_cgroup-to-page
lookups without a direct pointer.

This saves a full word for every page in a system with memory cgroups
enabled.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: condense page_cgroup-to-page lookup points
Johannes Weiner [Wed, 23 Mar 2011 23:42:29 +0000 (16:42 -0700)]
memcg: condense page_cgroup-to-page lookup points

The per-cgroup LRU lists string up 'struct page_cgroup's.  To get from
those structures to the page they represent, a lookup is required.
Currently, the lookup is done through a direct pointer in struct
page_cgroup, so a lot of functions down the callchain do this lookup by
themselves instead of receiving the page pointer from their callers.

The next patch removes this pointer, however, and the lookup is no longer
that straight-forward.  In preparation for that, this patch only leaves
the non-optional lookups when coming directly from the LRU list and passes
the page down the stack.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: fold __mem_cgroup_move_account into caller
Johannes Weiner [Wed, 23 Mar 2011 23:42:28 +0000 (16:42 -0700)]
memcg: fold __mem_cgroup_move_account into caller

It is one logical function, no need to have it split up.

Also, get rid of some checks from the inner function that ensured the
sanity of the outer function.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: change page_cgroup_zoneinfo signature
Johannes Weiner [Wed, 23 Mar 2011 23:42:27 +0000 (16:42 -0700)]
memcg: change page_cgroup_zoneinfo signature

Instead of passing a whole struct page_cgroup to this function, let it
take only what it really needs from it: the struct mem_cgroup and the
page.

This has the advantage that reading pc->mem_cgroup is now done at the same
place where the ordering rules for this pointer are enforced and
explained.

It is also in preparation for removing the pc->page backpointer.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: no uncharged pages reach page_cgroup_zoneinfo
Johannes Weiner [Wed, 23 Mar 2011 23:42:26 +0000 (16:42 -0700)]
memcg: no uncharged pages reach page_cgroup_zoneinfo

This patch series removes the direct page pointer from struct page_cgroup,
which saves 20% of per-page memcg memory overhead (Fedora and Ubuntu
enable memcg per default, openSUSE apparently too).

The node id or section number is encoded in the remaining free bits of
pc->flags which allows calculating the corresponding page without the
extra pointer.

I ran, what I think is, a worst-case microbenchmark that just cats a large
sparse file to /dev/null, because it means that walking the LRU list on
behalf of per-cgroup reclaim and looking up pages from page_cgroups is
happening constantly and at a high rate.  But it made no measurable
difference.  A profile reported a 0.11% share of the new
lookup_cgroup_page() function in this benchmark.

This patch:

All callsites check PCG_USED before passing pc->mem_cgroup, so the latter
is never NULL.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Acked-by: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: add memcg sanity checks at allocating and freeing pages
Daisuke Nishimura [Wed, 23 Mar 2011 23:42:25 +0000 (16:42 -0700)]
memcg: add memcg sanity checks at allocating and freeing pages

Add checks at allocating or freeing a page whether the page is used (iow,
charged) from the view point of memcg.

This check may be useful in debugging a problem and we did similar checks
before the commit 52d4b9ac(memcg: allocate all page_cgroup at boot).

This patch adds some overheads at allocating or freeing memory, so it's
enabled only when CONFIG_DEBUG_VM is enabled.

Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: remove NULL check from lookup_page_cgroup() result
Johannes Weiner [Wed, 23 Mar 2011 23:42:24 +0000 (16:42 -0700)]
memcg: remove NULL check from lookup_page_cgroup() result

The page_cgroup array is set up before even fork is initialized.  I
seriously doubt that this code executes before the array is alloc'd.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: remove impossible conditional when committing
Johannes Weiner [Wed, 23 Mar 2011 23:42:23 +0000 (16:42 -0700)]
memcg: remove impossible conditional when committing

No callsite ever passes a NULL pointer for a struct mem_cgroup * to the
committing function.  There is no need to check for it.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: remove unused page flag bitfield defines
Johannes Weiner [Wed, 23 Mar 2011 23:42:22 +0000 (16:42 -0700)]
memcg: remove unused page flag bitfield defines

These definitions have been unused since '4b3bde4 memcg: remove the
overhead associated with the root cgroup'.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomemcg: simplify the way memory limits are checked
Johannes Weiner [Wed, 23 Mar 2011 23:42:21 +0000 (16:42 -0700)]
memcg: simplify the way memory limits are checked

Since transparent huge pages, checking whether memory cgroups are below
their limits is no longer enough, but the actual amount of chargeable
space is important.

To not have more than one limit-checking interface, replace
memory_cgroup_check_under_limit() and memory_cgroup_check_margin() with a
single memory_cgroup_margin() that returns the chargeable space and leaves
the comparison to the callsite.

Soft limits are now checked the other way round, by using the already
existing function that returns the amount by which soft limits are
exceeded: res_counter_soft_limit_excess().

Also remove all the corresponding functions on the res_counter side that
are now no longer used.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Acked-by: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Minchan Kim <minchan.kim@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>