GitHub/moto-9609/android_kernel_motorola_exynos9610.git
11 years agoMerge branch 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Wed, 18 Sep 2013 16:23:32 +0000 (11:23 -0500)]
Merge branch 'sched-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull scheduler fixes from Ingo Molnar:
 "Misc fixes"

* 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  sched: Fix comment for sched_info_depart
  sched/Documentation: Update sched-design-CFS.txt documentation
  sched/debug: Take PID namespace into account
  sched/fair: Fix small race where child->se.parent,cfs_rq might point to invalid ones

11 years agoMerge branch 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Wed, 18 Sep 2013 16:22:53 +0000 (11:22 -0500)]
Merge branch 'perf-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull perf fixes from Ingo Molnar:
 "Two small fixes"

* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf: Fix UAPI export of PERF_EVENT_IOC_ID
  perf/x86/intel: Fix Silvermont offcore masks

11 years agoperf: Fix UAPI export of PERF_EVENT_IOC_ID
Vince Weaver [Tue, 17 Sep 2013 18:53:41 +0000 (14:53 -0400)]
perf: Fix UAPI export of PERF_EVENT_IOC_ID

Without the following patch I have problems compiling code using
the new PERF_EVENT_IOC_ID ioctl().  It looks like u64 was used
instead of __u64

Signed-off-by: Vince Weaver <vincent.weaver@maine.edu>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Link: http://lkml.kernel.org/r/alpine.DEB.2.10.1309171450380.11444@vincent-weaver-1.um.maine.edu
Signed-off-by: Ingo Molnar <mingo@kernel.org>
11 years agoMerge branch 'fixes' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Wed, 18 Sep 2013 02:20:30 +0000 (22:20 -0400)]
Merge branch 'fixes' of git://git./virt/kvm/kvm

Pull KVM fixes from Gleb Natapov.

* 'fixes' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: VMX: set "blocked by NMI" flag if EPT violation happens during IRET from NMI
  kvm: free resources after canceling async_pf
  KVM: nEPT: reset PDPTR register cache on nested vmentry emulation
  KVM: mmu: allow page tables to be in read-only slots
  KVM: x86 emulator: emulate RETF imm

11 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid
Linus Torvalds [Wed, 18 Sep 2013 01:54:05 +0000 (21:54 -0400)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jikos/hid

Pull HID updates from Jiri Kosina:
 "Fixes for CVE-2013-2897, CVE-2013-2895, CVE-2013-2897, CVE-2013-2894,
  CVE-2013-2893, CVE-2013-2891, CVE-2013-2890, CVE-2013-2889.

  All the bugs are triggerable only by specially crafted evil-on-purpose
  HW devices.  Fixes by Kees Cook and Benjamin Tissoires"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid:
  HID: lenovo-tpkbd: fix leak if tpkbd_probe_tp fails
  HID: multitouch: validate indexes details
  HID: logitech-dj: validate output report details
  HID: validate feature and input report details
  HID: lenovo-tpkbd: validate output report details
  HID: LG: validate HID output report details
  HID: steelseries: validate output report details
  HID: sony: validate HID output report details
  HID: zeroplus: validate output report details
  HID: provide a helper for validating hid reports

11 years agotty: disassociate_ctty() sends the extra SIGCONT
Oleg Nesterov [Sun, 15 Sep 2013 15:50:26 +0000 (17:50 +0200)]
tty: disassociate_ctty() sends the extra SIGCONT

Starting from v3.10 (probably commit f91e2590410b: "tty: Signal
foreground group processes in hangup") disassociate_ctty() sends SIGCONT
if tty && on_exit.  This breaks LSB test-suite, in particular test8 in
_exit.c and test40 in sigcon5.c.

Put the "!on_exit" check back to restore the old behaviour.

Review by Peter Hurley:
 "Yes, this regression was introduced by me in that commit.  The effect
  of the regression is that ptys will receive a SIGCONT when, in similar
  circumstances, ttys would not.

  The fact that two test vectors accidentally tripped over this
  regression suggests that some other apps may as well.

  Thanks for catching this"

Cc: stable@vger.kernel.org # v3.10+
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Reported-by: Karel Srot <ksrot@redhat.com>
Reviewed-by: Peter Hurley <peter@hurleysoftware.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoKVM: VMX: set "blocked by NMI" flag if EPT violation happens during IRET from NMI
Gleb Natapov [Sun, 15 Sep 2013 08:07:23 +0000 (11:07 +0300)]
KVM: VMX: set "blocked by NMI" flag if EPT violation happens during IRET from NMI

Set "blocked by NMI" flag if EPT violation happens during IRET from NMI
otherwise NMI can be called recursively causing stack corruption.

Signed-off-by: Gleb Natapov <gleb@redhat.com>
11 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf/linux-tile
Linus Torvalds [Tue, 17 Sep 2013 15:40:49 +0000 (11:40 -0400)]
Merge git://git./linux/kernel/git/cmetcalf/linux-tile

Pull more tile architecture updates from Chris Metcalf:
 "This second batch of changes is just cleanup of various kinds from
  doing some tidying work in the sources.

  Some dead code is removed, comment typos fixed, whitespace and style
  issues cleaned up, and some header updates from our internal
  "upstream" architecture team"

* git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf/linux-tile:
  tile: remove stray blank space
  tile: <arch/> header updates from upstream
  tile: improve gxio iorpc autogenerated code style
  tile: double default VMALLOC space
  tile: remove stale arch/tile/kernel/futex_64.S
  tile: remove HUGE_VMAP dead code
  tile: use pmd_pfn() instead of casting via pte_t
  tile: fix typos in comment in arch/tile/kernel/unaligned.c

11 years agokvm: free resources after canceling async_pf
Radim Krčmář [Wed, 4 Sep 2013 20:32:23 +0000 (22:32 +0200)]
kvm: free resources after canceling async_pf

When we cancel 'async_pf_execute()', we should behave as if the work was
never scheduled in 'kvm_setup_async_pf()'.
Fixes a bug when we can't unload module because the vm wasn't destroyed.

Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
11 years agoKVM: nEPT: reset PDPTR register cache on nested vmentry emulation
Gleb Natapov [Mon, 2 Sep 2013 12:25:28 +0000 (15:25 +0300)]
KVM: nEPT: reset PDPTR register cache on nested vmentry emulation

After nested vmentry stale cache can be used to reload L2 PDPTR pointers
which will cause L2 guest to fail. Fix it by invalidating cache on nested
vmentry emulation.

https://bugzilla.kernel.org/show_bug.cgi?id=60830

Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
11 years agoKVM: mmu: allow page tables to be in read-only slots
Paolo Bonzini [Mon, 9 Sep 2013 11:52:33 +0000 (13:52 +0200)]
KVM: mmu: allow page tables to be in read-only slots

Page tables in a read-only memory slot will currently cause a triple
fault because the page walker uses gfn_to_hva and it fails on such a slot.

OVMF uses such a page table; however, real hardware seems to be fine with
that as long as the accessed/dirty bits are set.  Save whether the slot
is readonly, and later check it when updating the accessed and dirty bits.

Reviewed-by: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Reviewed-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
11 years agoKVM: x86 emulator: emulate RETF imm
Bruce Rogers [Mon, 9 Sep 2013 15:40:20 +0000 (09:40 -0600)]
KVM: x86 emulator: emulate RETF imm

Opcode CA

This gets used by a DOS based NetWare guest.

Signed-off-by: Bruce Rogers <brogers@suse.com>
Reviewed-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
11 years agoLinux 3.12-rc1
Linus Torvalds [Mon, 16 Sep 2013 20:17:51 +0000 (16:17 -0400)]
Linux 3.12-rc1

11 years agoMerge branch 'timers/core' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Linus Torvalds [Mon, 16 Sep 2013 20:10:26 +0000 (16:10 -0400)]
Merge branch 'timers/core' of git://git./linux/kernel/git/tip/tip

Pull timer code update from Thomas Gleixner:
 - armada SoC clocksource overhaul with a trivial merge conflict
 - Minor improvements to various SoC clocksource drivers

* 'timers/core' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  clocksource: armada-370-xp: Add detailed clock requirements in devicetree binding
  clocksource: armada-370-xp: Get reference fixed-clock by name
  clocksource: armada-370-xp: Replace WARN_ON with BUG_ON
  clocksource: armada-370-xp: Fix device-tree binding
  clocksource: armada-370-xp: Introduce new compatibles
  clocksource: armada-370-xp: Use CLOCKSOURCE_OF_DECLARE
  clocksource: armada-370-xp: Simplify TIMER_CTRL register access
  clocksource: armada-370-xp: Use BIT()
  ARM: timer-sp: Set dynamic irq affinity
  ARM: nomadik: add dynamic irq flag to the timer
  clocksource: sh_cmt: 32-bit control register support
  clocksource: em_sti: Convert to devm_* managed helpers

11 years agotile: remove stray blank space
Chris Metcalf [Mon, 16 Sep 2013 18:41:21 +0000 (14:41 -0400)]
tile: remove stray blank space

The compat sys_llseek() definition addition added a bogus space
on an otherwise-blank line.

Signed-off-by: Chris Metcalf <cmetcalf@tilera.com>
11 years agotile: <arch/> header updates from upstream
Chris Metcalf [Mon, 16 Sep 2013 18:18:21 +0000 (14:18 -0400)]
tile: <arch/> header updates from upstream

The hardware architecture descriptor headers have been updated, in
particular to reflect some larger MMIO fields on the mPIPE shims for
controlling the network hardware, from the recent Gx72 release.

Signed-off-by: Chris Metcalf <cmetcalf@tilera.com>
11 years agotile: improve gxio iorpc autogenerated code style
Chris Metcalf [Mon, 16 Sep 2013 17:52:45 +0000 (13:52 -0400)]
tile: improve gxio iorpc autogenerated code style

Fix some whitespace style issues in some auto-generated files.

Signed-off-by: Chris Metcalf <cmetcalf@tilera.com>
11 years agotile: double default VMALLOC space
Chris Metcalf [Mon, 16 Sep 2013 17:02:57 +0000 (13:02 -0400)]
tile: double default VMALLOC space

With per-cpu data as well as loaded kernel modules coming from
the vmalloc arena, we get close to the line all the time and
occasionally need more than we had, so just double it up by default.

Signed-off-by: Chris Metcalf <cmetcalf@tilera.com>
11 years agoMerge branch 'for-next' of git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Mon, 16 Sep 2013 19:39:21 +0000 (15:39 -0400)]
Merge branch 'for-next' of git://git.samba.org/sfrench/cifs-2.6

Pull CIFS fixes from Steve French:
 "Two minor cifs fixes and a minor documentation cleanup for cifs.txt"

* 'for-next' of git://git.samba.org/sfrench/cifs-2.6:
  cifs: update cifs.txt and remove some outdated infos
  cifs: Avoid calling unlock_page() twice in cifs_readpage() when using fscache
  cifs: Do not take a reference to the page in cifs_readpage_worker()

11 years agoMerge tag 'upstream-3.12-rc1' of git://git.infradead.org/linux-ubi
Linus Torvalds [Mon, 16 Sep 2013 19:37:52 +0000 (15:37 -0400)]
Merge tag 'upstream-3.12-rc1' of git://git.infradead.org/linux-ubi

Pull UBI fixes from Artem Bityutskiy:
 "Just a single fastmap fix plus a regression fix"

* tag 'upstream-3.12-rc1' of git://git.infradead.org/linux-ubi:
  UBI: Fix invalidate_fastmap()
  UBI: Fix PEB leak in wear_leveling_worker()

11 years agoMerge tag 'upstream-3.12-rc1' of git://git.infradead.org/linux-ubifs
Linus Torvalds [Mon, 16 Sep 2013 19:36:55 +0000 (15:36 -0400)]
Merge tag 'upstream-3.12-rc1' of git://git.infradead.org/linux-ubifs

Pull ubifs fix from Artem Bityutskiy:
 "Just one patch which fixes the power-cut recovery testing mode.

  I'll start using a single UBI/UBIFS tree instead of 2 trees from now
  on.  So in the future you'll get 1 small pull request instead of 2
  tiny ones"

* tag 'upstream-3.12-rc1' of git://git.infradead.org/linux-ubifs:
  UBIFS: remove invalid warn msg with tst_recovery enabled

11 years agosched: Fix comment for sched_info_depart
Michael S. Tsirkin [Mon, 16 Sep 2013 08:30:36 +0000 (11:30 +0300)]
sched: Fix comment for sched_info_depart

sched_info_depart seems to be only called from
sched_info_switch(), so only on involuntary task switch.

Fix the comment to match.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Link: http://lkml.kernel.org/r/20130916083036.GA1113@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
11 years agoMerge branch 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus
Linus Torvalds [Sun, 15 Sep 2013 21:45:52 +0000 (17:45 -0400)]
Merge branch 'upstream' of git://git.linux-mips.org/ralf/upstream-linus

Pull MIPS fixes from Ralf Baechle:
 "These are four patches for three construction sites:

   - Fix register decoding for the combination of multi-core processors
     and multi-threading.

   - Two more fixes that are part of the ongoing DECstation resurrection
     work.  One of these touches a DECstation-only network driver.

   - Finally Markos' trivial build fix for the AP/SP support.

  (With this applied now all MIPS defconfigs are building again)"

* 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus:
  MIPS: kernel: vpe: Make vpe_attrs an array of pointers.
  MIPS: Fix SMP core calculations when using MT support.
  MIPS: DECstation I/O ASIC DMA interrupt handling fix
  MIPS: DECstation HRT initialization rearrangement

11 years agoMerge branch 'for_linus' of git://cavan.codon.org.uk/platform-drivers-x86
Linus Torvalds [Sun, 15 Sep 2013 21:42:59 +0000 (17:42 -0400)]
Merge branch 'for_linus' of git://cavan.codon.org.uk/platform-drivers-x86

Pull x86 platform updates from Matthew Garrett:
 "Nothing amazing here, almost entirely cleanups and minor bugfixes and
  one bit of hardware enablement in the amilo-rfkill driver"

* 'for_linus' of git://cavan.codon.org.uk/platform-drivers-x86:
  platform/x86: panasonic-laptop: reuse module_acpi_driver
  samsung-laptop: fix config build error
  platform: x86: remove unnecessary platform_set_drvdata()
  amilo-rfkill: Enable using amilo-rfkill with the FSC Amilo L1310.
  wmi: parse_wdg() should return kernel error codes
  hp_wmi: Fix unregister order in hp_wmi_rfkill_setup()
  platform: replace strict_strto*() with kstrto*()
  x86: irst: use module_acpi_driver to simplify the code
  x86: smartconnect: use module_acpi_driver to simplify the code
  platform samsung-q10: use ACPI instead of direct EC calls
  thinkpad_acpi: add the ability setting TPACPI_LED_NONE by quirk
  thinkpad_acpi: return -NODEV while operating uninitialized LEDs

11 years agoMerge tag 'scsi-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Linus Torvalds [Sun, 15 Sep 2013 21:41:30 +0000 (17:41 -0400)]
Merge tag 'scsi-misc' of git://git./linux/kernel/git/jejb/scsi

Pull misc SCSI driver updates from James Bottomley:
 "This patch set is a set of driver updates (megaraid_sas, fnic, lpfc,
  ufs, hpsa) we also have a couple of bug fixes (sd out of bounds and
  ibmvfc error handling) and the first round of esas2r checker fixes and
  finally the much anticipated big endian additions for megaraid_sas"

* tag 'scsi-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi: (47 commits)
  [SCSI] fnic: fnic Driver Tuneables Exposed through CLI
  [SCSI] fnic: Kernel panic while running sh/nosh with max lun cfg
  [SCSI] fnic: Hitting BUG_ON(io_req->abts_done) in fnic_rport_exch_reset
  [SCSI] fnic: Remove QUEUE_FULL handling code
  [SCSI] fnic: On system with >1.1TB RAM, VIC fails multipath after boot up
  [SCSI] fnic: FC stat param seconds_since_last_reset not getting updated
  [SCSI] sd: Fix potential out-of-bounds access
  [SCSI] lpfc 8.3.42: Update lpfc version to driver version 8.3.42
  [SCSI] lpfc 8.3.42: Fixed issue of task management commands having a fixed timeout
  [SCSI] lpfc 8.3.42: Fixed inconsistent spin lock usage.
  [SCSI] lpfc 8.3.42: Fix driver's abort loop functionality to skip IOs already getting aborted
  [SCSI] lpfc 8.3.42: Fixed failure to allocate SCSI buffer on PPC64 platform for SLI4 devices
  [SCSI] lpfc 8.3.42: Fix WARN_ON when driver unloads
  [SCSI] lpfc 8.3.42: Avoided making pci bar ioremap call during dual-chute WQ/RQ pci bar selection
  [SCSI] lpfc 8.3.42: Fixed driver iocbq structure's iocb_flag field running out of space
  [SCSI] lpfc 8.3.42: Fix crash on driver load due to cpu affinity logic
  [SCSI] lpfc 8.3.42: Fixed logging format of setting driver sysfs attributes hard to interpret
  [SCSI] lpfc 8.3.42: Fixed back to back RSCNs discovery failure.
  [SCSI] lpfc 8.3.42: Fixed race condition between BSG I/O dispatch and timeout handling
  [SCSI] lpfc 8.3.42: Fixed function mode field defined too small for not recognizing dual-chute mode
  ...

11 years agoMerge branch 'slab/next' of git://git.kernel.org/pub/scm/linux/kernel/git/penberg...
Linus Torvalds [Sun, 15 Sep 2013 11:15:06 +0000 (07:15 -0400)]
Merge branch 'slab/next' of git://git./linux/kernel/git/penberg/linux

Pull SLAB update from Pekka Enberg:
 "Nothing terribly exciting here apart from Christoph's kmalloc
  unification patches that brings sl[aou]b implementations closer to
  each other"

* 'slab/next' of git://git.kernel.org/pub/scm/linux/kernel/git/penberg/linux:
  slab: Use correct GFP_DMA constant
  slub: remove verify_mem_not_deleted()
  mm/sl[aou]b: Move kmallocXXX functions to common code
  mm, slab_common: add 'unlikely' to size check of kmalloc_slab()
  mm/slub.c: beautify code for removing redundancy 'break' statement.
  slub: Remove unnecessary page NULL check
  slub: don't use cpu partial pages on UP
  mm/slub: beautify code for 80 column limitation and tab alignment
  mm/slub: remove 'per_cpu' which is useless variable

11 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Linus Torvalds [Sun, 15 Sep 2013 11:13:39 +0000 (07:13 -0400)]
Merge branch 'for-linus' of git://git./linux/kernel/git/dtor/input

Pull input update from Dmitry Torokhov:
 "The only change is David Hermann's new EVIOCREVOKE evdev ioctl that
  allows safely passing file descriptors to input devices to session
  processes and later being able to stop delivery of events through
  these fds so that inactive sessions will no longer receive user input
  that does not belong to them"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: evdev - add EVIOCREVOKE ioctl

11 years agovfs: fix typo in comment in recent dentry work
Linus Torvalds [Sun, 15 Sep 2013 11:11:01 +0000 (07:11 -0400)]
vfs: fix typo in comment in recent dentry work

Sedat points out that I transposed some letters in "LRU" and wrote "RLU"
instead in one of the new comments explaining the flow.  Let's just fix
it.

Reported-by: Sedat Dilek <sedat.dilek@jpberlin.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agopartitions/efi: loosen check fot pmbr size in lba
Davidlohr Bueso [Fri, 13 Sep 2013 22:02:22 +0000 (15:02 -0700)]
partitions/efi: loosen check fot pmbr size in lba

Matt found that commit 27a7c642174e ("partitions/efi: account for pmbr
size in lba") caused his GPT formatted eMMC device not to boot.  The
reason is that this commit enforced Linux to always check the lesser of
the whole disk or 2Tib for the pMBR size in LBA.  While most disk
partitioning tools out there create a pMBR with these characteristics,
Microsoft does not, as it always sets the entry to the maximum 32-bit
limitation - even though a drive may be smaller than that[1].

Loosen this check and only verify that the size is either the whole disk
or 0xFFFFFFFF.  No tool in its right mind would set it to any value
other than these.

[1] http://thestarman.pcministry.com/asm/mbr/GPT.htm#GPTPT

Reported-and-tested-by: Matt Porter <matt.porter@linaro.org>
Signed-off-by: Davidlohr Bueso <davidlohr@hp.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoMerge tag 'writeback-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/wfg...
Linus Torvalds [Sat, 14 Sep 2013 03:06:40 +0000 (23:06 -0400)]
Merge tag 'writeback-fixes' of git://git./linux/kernel/git/wfg/linux

Pull writeback fix from Wu Fengguang:
 "A trivial writeback fix"

* tag 'writeback-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/wfg/linux:
  writeback: Do not sort b_io list only because of block device inode

11 years agovfs: fix dentry LRU list handling and nr_dentry_unused accounting
Linus Torvalds [Sat, 14 Sep 2013 02:55:10 +0000 (22:55 -0400)]
vfs: fix dentry LRU list handling and nr_dentry_unused accounting

The LRU list changes interacted badly with our nr_dentry_unused
accounting, and even worse with the new DCACHE_LRU_LIST bit logic.

This introduces helper functions to make sure everything follows the
proper dcache d_lru list rules: the dentry cache is complicated by the
fact that some of the hotpaths don't even want to look at the LRU list
at all, and the fact that we use the same list entry in the dentry for
both the LRU list and for our temporary shrinking lists when removing
things from the LRU.

The helper functions temporarily have some extra sanity checking for the
flag bits that have to match the current LRU state of the dentry.  We'll
remove that before the final 3.12 release, but considering how easy it
is to get wrong, this first cleanup version has some very particular
sanity checking.

Acked-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agocifs: update cifs.txt and remove some outdated infos
Björn Jacke [Tue, 10 Sep 2013 08:28:38 +0000 (10:28 +0200)]
cifs: update cifs.txt and remove some outdated infos

Acked-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Björn JACKE <bj@sernet.de>
Signed-off-by: Steve French <smfrench@gmail.com>
11 years agocifs: Avoid calling unlock_page() twice in cifs_readpage() when using fscache
Sachin Prabhu [Fri, 13 Sep 2013 13:11:57 +0000 (14:11 +0100)]
cifs: Avoid calling unlock_page() twice in cifs_readpage() when using fscache

When reading a single page with cifs_readpage(), we make a call to
fscache_read_or_alloc_page() which once done, asynchronously calls
the completion function cifs_readpage_from_fscache_complete(). This
completion function unlocks the page once it has been populated from
cache. The module then attempts to unlock the page a second time in
cifs_readpage() which leads to warning messages.

In case of a successful call to fscache_read_or_alloc_page() we should skip
the second unlock_page() since this will be called by the
cifs_readpage_from_fscache_complete() once the page has been populated by
fscache.

With the modifications to cifs_readpage_worker(), we will need to re-grab the
page lock in cifs_write_begin().

The problem was first noticed when testing new fscache patches for cifs.
https://bugzilla.redhat.com/show_bug.cgi?id=1005737

Signed-off-by: Sachin Prabhu <sprabhu@redhat.com>
Reviewed-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <smfrench@gmail.com>
11 years agocifs: Do not take a reference to the page in cifs_readpage_worker()
Sachin Prabhu [Fri, 13 Sep 2013 13:11:56 +0000 (14:11 +0100)]
cifs: Do not take a reference to the page in cifs_readpage_worker()

We do not need to take a reference to the pagecache in
cifs_readpage_worker() since the calling function will have already
taken one before passing the pointer to the page as an argument to the
function.

Signed-off-by: Sachin Prabhu <sprabhu@redhat.com>
Reviewed-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <smfrench@gmail.com>
11 years agoMerge tag 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck...
Linus Torvalds [Fri, 13 Sep 2013 17:58:41 +0000 (10:58 -0700)]
Merge tag 'hwmon-for-linus' of git://git./linux/kernel/git/groeck/linux-staging

Pull hwmon fixes from Guenter Roeck:
 "Some more low risk cleanup patches:

   - Remove unnecessary pci_set_drvdata in k10temp driver from Jingoo Han
   - Fix return values in several drivers from Sachin Kamat
   - Remove redundant break in amc6821 driver from Sachin Kamat"

* tag 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
  hwmon: (k10temp) remove unnecessary pci_set_drvdata()
  hwmon: (tmp421) Fix return value
  hwmon: (amc6821) Remove redundant break
  hwmon: (amc6821) Fix return value
  hwmon: (ibmaem) Fix return value
  hwmon: (emc2103) Fix return value

11 years agoMerge tag 'xtensa-next-20130912' of git://github.com/czankel/xtensa-linux
Linus Torvalds [Fri, 13 Sep 2013 17:57:48 +0000 (10:57 -0700)]
Merge tag 'xtensa-next-20130912' of git://github.com/czankel/xtensa-linux

Pull Xtensa updates from Chris Zankel.

* tag 'xtensa-next-20130912' of git://github.com/czankel/xtensa-linux:
  xtensa: Fix broken allmodconfig build
  xtensa: remove CCOUNT_PER_JIFFY
  xtensa: fix !CONFIG_XTENSA_CALIBRATE_CCOUNT build failure
  xtensa: don't use echo -e needlessly
  xtensa: new fast_alloca handler
  xtensa: keep a3 and excsave1 on entry to exception handlers
  xtensa: enable kernel preemption
  xtensa: check thread flags atomically on return from user exception

11 years agoMerge git://git.kvack.org/~bcrl/aio-next
Linus Torvalds [Fri, 13 Sep 2013 17:55:58 +0000 (10:55 -0700)]
Merge git://git.kvack.org/~bcrl/aio-next

Pull aio changes from Ben LaHaise:
 "First off, sorry for this pull request being late in the merge window.
  Al had raised a couple of concerns about 2 items in the series below.
  I addressed the first issue (the race introduced by Gu's use of
  mm_populate()), but he has not provided any further details on how he
  wants to rework the anon_inode.c changes (which were sent out months
  ago but have yet to be commented on).

  The bulk of the changes have been sitting in the -next tree for a few
  months, with all the issues raised being addressed"

* git://git.kvack.org/~bcrl/aio-next: (22 commits)
  aio: rcu_read_lock protection for new rcu_dereference calls
  aio: fix race in ring buffer page lookup introduced by page migration support
  aio: fix rcu sparse warnings introduced by ioctx table lookup patch
  aio: remove unnecessary debugging from aio_free_ring()
  aio: table lookup: verify ctx pointer
  staging/lustre: kiocb->ki_left is removed
  aio: fix error handling and rcu usage in "convert the ioctx list to table lookup v3"
  aio: be defensive to ensure request batching is non-zero instead of BUG_ON()
  aio: convert the ioctx list to table lookup v3
  aio: double aio_max_nr in calculations
  aio: Kill ki_dtor
  aio: Kill ki_users
  aio: Kill unneeded kiocb members
  aio: Kill aio_rw_vect_retry()
  aio: Don't use ctx->tail unnecessarily
  aio: io_cancel() no longer returns the io_event
  aio: percpu ioctx refcount
  aio: percpu reqs_available
  aio: reqs_active -> reqs_available
  aio: fix build when migration is disabled
  ...

11 years agotile: remove stale arch/tile/kernel/futex_64.S
Chris Metcalf [Mon, 9 Sep 2013 16:18:54 +0000 (12:18 -0400)]
tile: remove stale arch/tile/kernel/futex_64.S

This should have been removed with commit 47d632f9f8f3, but
it was overlooked.

Signed-off-by: Chris Metcalf <cmetcalf@tilera.com>
11 years agotile: remove HUGE_VMAP dead code
Chris Metcalf [Wed, 11 Sep 2013 17:57:15 +0000 (13:57 -0400)]
tile: remove HUGE_VMAP dead code

A config option to allow a variant vmap() using huge pages that was never
upstreamed had some bits of code related to it scattered around the tile
architecture; the config option was removed downstream and this commit
cleans up the scattered evidence of it from the upstream as well.

Signed-off-by: Chris Metcalf <cmetcalf@tilera.com>
11 years agotile: use pmd_pfn() instead of casting via pte_t
Chris Metcalf [Fri, 13 Sep 2013 15:14:25 +0000 (11:14 -0400)]
tile: use pmd_pfn() instead of casting via pte_t

Signed-off-by: Chris Metcalf <cmetcalf@tilera.com>
11 years agotile: fix typos in comment in arch/tile/kernel/unaligned.c
Chris Metcalf [Fri, 13 Sep 2013 14:57:54 +0000 (10:57 -0400)]
tile: fix typos in comment in arch/tile/kernel/unaligned.c

Signed-off-by: Chris Metcalf <cmetcalf@tilera.com>
11 years agoMerge branch 'genirq' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux
Linus Torvalds [Fri, 13 Sep 2013 14:31:38 +0000 (07:31 -0700)]
Merge branch 'genirq' of git://git./linux/kernel/git/s390/linux

Pull generic hardirq option removal from Martin Schwidefsky:
 "All architectures now use generic hardirqs, s390 has been last to
  switch.

  With that the code under !CONFIG_GENERIC_HARDIRQS and the related
  HAVE_GENERIC_HARDIRQS and GENERIC_HARDIRQS config options can be
  removed.  Yay!"

* 'genirq' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
  Remove GENERIC_HARDIRQ config option

11 years agoMerge branch 'kconfig' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild
Linus Torvalds [Fri, 13 Sep 2013 14:30:17 +0000 (07:30 -0700)]
Merge branch 'kconfig' of git://git./linux/kernel/git/mmarek/kbuild

Pull kconfig fix from Michal Marek:
 "This is a fix for a regression caused by my previous pull request.

  A sed command in scripts/config that used colons as separator was
  accidentally changed to use slashes, which fails when you use slashes
  in a value.  Changing it back to colons is of course not a proper fix,
  but at least it will be broken in the same way it had been for four
  years.  A proper fix is pending"

* 'kconfig' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild:
  scripts/config: fix variable substitution command

11 years agoMerge tag 'blackfin-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/realm...
Linus Torvalds [Fri, 13 Sep 2013 14:23:49 +0000 (07:23 -0700)]
Merge tag 'blackfin-for-linus' of git://git./linux/kernel/git/realmz6/blackfin-linux

Pull blackfin updates from Steven Miao.

* tag 'blackfin-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/realmz6/blackfin-linux:
  blackfin: Ignore generated uImages
  blackfin: Add STMMAC platform data to enable dwmac1000 driver on BF60x.
  bf609: adv7343: add S-Video and Component output support
  bf609: add adv7343 video encoder support
  clock: add stmmac clock for ethernet driver
  blackfin: scb: Add SCB1 to SCB9 config options and data.
  blackfin: scb: Add system crossbar init code.

11 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6
Linus Torvalds [Fri, 13 Sep 2013 14:11:14 +0000 (07:11 -0700)]
Merge git://git./linux/kernel/git/herbert/crypto-2.6

Pull crypto fixes from Herbert Xu:
 "This fixes a 7+ year race condition in the crypto API that causes
  sporadic crashes when multiple threads load the same algorithm.

  It also fixes the crct10dif algorithm again to prevent boot failures
  on systems where the initramfs tool ignores module softdeps"

* git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  crypto: crct10dif - Add fallback for broken initrds
  crypto: api - Fix race condition in larval lookup

11 years agoHID: lenovo-tpkbd: fix leak if tpkbd_probe_tp fails
Benjamin Tissoires [Wed, 11 Sep 2013 19:56:59 +0000 (21:56 +0200)]
HID: lenovo-tpkbd: fix leak if tpkbd_probe_tp fails

If tpkbd_probe_tp() bails out, the probe() function return an error,
but hid_hw_stop() is never called.

fixes:
https://bugzilla.redhat.com/show_bug.cgi?id=1003998

Cc: stable@vger.kernel.org
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
11 years agoHID: multitouch: validate indexes details
Benjamin Tissoires [Wed, 11 Sep 2013 19:56:58 +0000 (21:56 +0200)]
HID: multitouch: validate indexes details

When working on report indexes, always validate that they are in bounds.
Without this, a HID device could report a malicious feature report that
could trick the driver into a heap overflow:

[  634.885003] usb 1-1: New USB device found, idVendor=0596, idProduct=0500
...
[  676.469629] BUG kmalloc-192 (Tainted: G        W   ): Redzone overwritten

Note that we need to change the indexes from s8 to s16 as they can
be between -1 and 255.

CVE-2013-2897

Cc: stable@vger.kernel.org
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Acked-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
11 years agoHID: logitech-dj: validate output report details
Kees Cook [Wed, 11 Sep 2013 19:56:56 +0000 (21:56 +0200)]
HID: logitech-dj: validate output report details

A HID device could send a malicious output report that would cause the
logitech-dj HID driver to leak kernel memory contents to the device, or
trigger a NULL dereference during initialization:

[  304.424553] usb 1-1: New USB device found, idVendor=046d, idProduct=c52b
...
[  304.780467] BUG: unable to handle kernel NULL pointer dereference at 0000000000000028
[  304.781409] IP: [<ffffffff815d50aa>] logi_dj_recv_send_report.isra.11+0x1a/0x90

CVE-2013-2895

Signed-off-by: Kees Cook <keescook@chromium.org>
Cc: stable@vger.kernel.org
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
11 years agoHID: validate feature and input report details
Benjamin Tissoires [Wed, 11 Sep 2013 19:56:57 +0000 (21:56 +0200)]
HID: validate feature and input report details

When dealing with usage_index, be sure to properly use unsigned instead of
int to avoid overflows.

When working on report fields, always validate that their report_counts are
in bounds.
Without this, a HID device could report a malicious feature report that
could trick the driver into a heap overflow:

[  634.885003] usb 1-1: New USB device found, idVendor=0596, idProduct=0500
...
[  676.469629] BUG kmalloc-192 (Tainted: G        W   ): Redzone overwritten

CVE-2013-2897

Cc: stable@vger.kernel.org
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Acked-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
11 years agoHID: lenovo-tpkbd: validate output report details
Kees Cook [Wed, 11 Sep 2013 19:56:55 +0000 (21:56 +0200)]
HID: lenovo-tpkbd: validate output report details

A HID device could send a malicious output report that would cause the
lenovo-tpkbd HID driver to write just beyond the output report allocation
during initialization, causing a heap overflow:

[   76.109807] usb 1-1: New USB device found, idVendor=17ef, idProduct=6009
...
[   80.462540] BUG kmalloc-192 (Tainted: G        W   ): Redzone overwritten

CVE-2013-2894

Signed-off-by: Kees Cook <keescook@chromium.org>
Cc: stable@vger.kernel.org
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
11 years agoMIPS: kernel: vpe: Make vpe_attrs an array of pointers.
Markos Chandras [Wed, 11 Sep 2013 12:17:52 +0000 (13:17 +0100)]
MIPS: kernel: vpe: Make vpe_attrs an array of pointers.

Commit 567b21e973ccf5b0d13776e408d7c67099749eb8
"mips: convert vpe_class to use dev_groups"

broke the build on MIPS since vpe_attrs should be an array
of 'struct device_attribute' pointers.

Fixes the following build problem:
arch/mips/kernel/vpe.c:1372:2: error: missing braces around initializer
[-Werror=missing-braces]
arch/mips/kernel/vpe.c:1372:2: error: (near initialization for 'vpe_attrs[0]')
[-Werror=missing-braces]

Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: John Crispin <blogic@openwrt.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Markos Chandras <markos.chandras@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/5819/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
11 years agoHID: LG: validate HID output report details
Kees Cook [Wed, 11 Sep 2013 19:56:54 +0000 (21:56 +0200)]
HID: LG: validate HID output report details

A HID device could send a malicious output report that would cause the
lg, lg3, and lg4 HID drivers to write beyond the output report allocation
during an event, causing a heap overflow:

[  325.245240] usb 1-1: New USB device found, idVendor=046d, idProduct=c287
...
[  414.518960] BUG kmalloc-4096 (Not tainted): Redzone overwritten

Additionally, while lg2 did correctly validate the report details, it was
cleaned up and shortened.

CVE-2013-2893

Signed-off-by: Kees Cook <keescook@chromium.org>
Cc: stable@vger.kernel.org
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
11 years agoHID: steelseries: validate output report details
Kees Cook [Wed, 11 Sep 2013 19:56:53 +0000 (21:56 +0200)]
HID: steelseries: validate output report details

A HID device could send a malicious output report that would cause the
steelseries HID driver to write beyond the output report allocation
during initialization, causing a heap overflow:

[  167.981534] usb 1-1: New USB device found, idVendor=1038, idProduct=1410
...
[  182.050547] BUG kmalloc-256 (Tainted: G        W   ): Redzone overwritten

CVE-2013-2891

Signed-off-by: Kees Cook <keescook@chromium.org>
Cc: stable@vger.kernel.org
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
11 years agoHID: sony: validate HID output report details
Kees Cook [Wed, 11 Sep 2013 19:56:52 +0000 (21:56 +0200)]
HID: sony: validate HID output report details

This driver must validate the availability of the HID output report and
its size before it can write LED states via buzz_set_leds(). This stops
a heap overflow that is possible if a device provides a malicious HID
output report:

[  108.171280] usb 1-1: New USB device found, idVendor=054c, idProduct=0002
...
[  117.507877] BUG kmalloc-192 (Not tainted): Redzone overwritten

CVE-2013-2890

Signed-off-by: Kees Cook <keescook@chromium.org>
Cc: stable@vger.kernel.org #3.11
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
11 years agoHID: zeroplus: validate output report details
Kees Cook [Wed, 11 Sep 2013 19:56:51 +0000 (21:56 +0200)]
HID: zeroplus: validate output report details

The zeroplus HID driver was not checking the size of allocated values
in fields it used. A HID device could send a malicious output report
that would cause the driver to write beyond the output report allocation
during initialization, causing a heap overflow:

[ 1442.728680] usb 1-1: New USB device found, idVendor=0c12, idProduct=0005
...
[ 1466.243173] BUG kmalloc-192 (Tainted: G        W   ): Redzone overwritten

CVE-2013-2889

Signed-off-by: Kees Cook <keescook@chromium.org>
Cc: stable@vger.kernel.org
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
11 years agoHID: provide a helper for validating hid reports
Kees Cook [Wed, 11 Sep 2013 19:56:50 +0000 (21:56 +0200)]
HID: provide a helper for validating hid reports

Many drivers need to validate the characteristics of their HID report
during initialization to avoid misusing the reports. This adds a common
helper to perform validation of the report exisitng, the field existing,
and the expected number of values within the field.

Signed-off-by: Kees Cook <keescook@chromium.org>
Cc: stable@vger.kernel.org
Reviewed-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
11 years agoRemove GENERIC_HARDIRQ config option
Martin Schwidefsky [Fri, 30 Aug 2013 07:39:53 +0000 (09:39 +0200)]
Remove GENERIC_HARDIRQ config option

After the last architecture switched to generic hard irqs the config
options HAVE_GENERIC_HARDIRQS & GENERIC_HARDIRQS and the related code
for !CONFIG_GENERIC_HARDIRQS can be removed.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
11 years agoscripts/config: fix variable substitution command
Clement Chauplannaz [Fri, 13 Sep 2013 08:45:13 +0000 (10:45 +0200)]
scripts/config: fix variable substitution command

Commit 229455bc02b87f7128f190c4491b4ceffff38648 accidentally changed the
separator between sed `s' command and its parameters from ':' to '/'.

Revert this change.

Reported-and-tested-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Clement Chauplannaz <chauplac@gmail.com>
Signed-off-by: Michal Marek <mmarek@suse.cz>
11 years agoMIPS: Fix SMP core calculations when using MT support.
Leonid Yegoshin [Wed, 11 Sep 2013 19:17:47 +0000 (14:17 -0500)]
MIPS: Fix SMP core calculations when using MT support.

The TCBIND register is only available if the core has MT support. It
should not be read otherwise. Secondly, the number of TCs (siblings)
are calculated differently depending on if the kernel is configured
as SMVP or SMTC.

Signed-off-by: Leonid Yegoshin <Leonid.Yegoshin@imgtec.com>
Signed-off-by: Steven J. Hill <Steven.Hill@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/5822/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
11 years agoMIPS: DECstation I/O ASIC DMA interrupt handling fix
Maciej W. Rozycki [Thu, 12 Sep 2013 11:14:31 +0000 (12:14 +0100)]
MIPS: DECstation I/O ASIC DMA interrupt handling fix

This change complements commit d0da7c002f7b2a93582187a9e3f73891a01d8ee4
and brings clear_ioasic_irq back, renaming it to clear_ioasic_dma_irq at
the same time, to make I/O ASIC DMA interrupts functional.

Unlike ordinary I/O ASIC interrupts DMA interrupts need to be deasserted
by software by writing 0 to the respective bit in I/O ASIC's System
Interrupt Register (SIR), similarly to how CP0.Cause.IP0 and CP0.Cause.IP1
bits are handled in the CPU (the difference is SIR DMA interrupt bits are
R/W0C so there's no need for an RMW cycle).  Otherwise the handler is
reentered over and over again.

The only current user is the DEC LANCE Ethernet driver and its extremely
uncommon DMA memory error handler that does not care when exactly the
interrupt is cleared.  Anticipating the use of DMA interrupts by the Zilog
SCC driver this change however exports clear_ioasic_dma_irq for device
drivers to choose the right application-specific sequence to clear the
request explicitly rather than calling it implicitly in the .irq_eoi
handler of `struct irq_chip'.  Previously these interrupts were cleared in
the .end handler of the said structure, before it was removed.

Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/5826/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
11 years agoMIPS: DECstation HRT initialization rearrangement
Maciej W. Rozycki [Thu, 12 Sep 2013 11:01:53 +0000 (12:01 +0100)]
MIPS: DECstation HRT initialization rearrangement

Not all I/O ASIC versions have the free-running counter implemented, an
early revision used in the 5000/1xx models aka 3MIN and 4MIN did not have
it.  Therefore we cannot unconditionally use it as a clock source.
Fortunately if not implemented its register slot has a fixed value so it
is enough if we check for the value at the end of the calibration period
being the same as at the beginning.

This also means we need to look for another high-precision clock source on
the systems affected.  The 5000/1xx can have an R4000SC processor
installed where the CP0 Count register can be used as a clock source.
Unfortunately all the R4k DECstations suffer from the missed timer
interrupt on CP0 Count reads erratum, so we cannot use the CP0 timer as a
clock source and a clock event both at a time.  However we never need an
R4k clock event device because all DECstations have a DS1287A RTC chip
whose periodic interrupt can be used as a clock source.

This gives us the following four configuration possibilities for I/O ASIC
DECstations:

1. No I/O ASIC counter and no CP0 timer, e.g. R3k 5000/1xx (3MIN).

2. No I/O ASIC counter but the CP0 timer, i.e. R4k 5000/150 (4MIN).

3. The I/O ASIC counter but no CP0 timer, e.g. R3k 5000/240 (3MAX+).

4. The I/O ASIC counter and the CP0 timer, e.g. R4k 5000/260 (4MAX+).

For #1 and #2 this change stops the I/O ASIC free-running counter from
being installed as a clock source of a 0Hz frequency.  For #2 it also
arranges for the CP0 timer to be used as a clock source rather than a
clock event device, because having an accurate wall clock is more
important than a high-precision interval timer.  For #3 there is no
change.  For #4 the change makes the I/O ASIC free-running counter
installed as a clock source so that the CP0 timer can be used as a clock
event device.

Unfortunately the use of the CP0 timer as a clock event device relies on a
succesful completion of c0_compare_interrupt.  That never happens, because
while waiting for a CP0 Compare interrupt to happen the function spins in
a loop reading the CP0 Count register.  This makes the CP0 Count erratum
trigger reliably causing the interrupt waited for to be lost in all cases.
As a result #4 resorts to using the CP0 timer as a clock source as well,
just as #2.  However we want to keep this separate arrangement in case
(hope) c0_compare_interrupt is eventually rewritten such that it avoids
the erratum.

Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/5825/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
11 years agoblackfin: Ignore generated uImages
Mark Brown [Thu, 29 Aug 2013 10:23:08 +0000 (11:23 +0100)]
blackfin: Ignore generated uImages

We have the build infrastructure to generate uImages so we should ignore
the resulting generated files.

Signed-off-by: Mark Brown <broonie@linaro.org>
Acked-by: Mike Frysinger <vapier@gentoo.org>
11 years agoblackfin: Add STMMAC platform data to enable dwmac1000 driver on BF60x.
Sonic Zhang [Thu, 15 Aug 2013 06:08:05 +0000 (14:08 +0800)]
blackfin: Add STMMAC platform data to enable dwmac1000 driver on BF60x.

- Enable GMAC
- Set propler DMA PBL
- Disable DMA store and forward mode
- Select PTP input clock from MII
clock.

Signed-off-by: Sonic Zhang <sonic.zhang@analog.com>
Signed-off-by: Steven Miao <realmz6@gmail.com>
11 years agobf609: adv7343: add S-Video and Component output support
Scott Jiang [Thu, 8 Aug 2013 22:48:22 +0000 (18:48 -0400)]
bf609: adv7343: add S-Video and Component output support

Signed-off-by: Scott Jiang <scott.jiang.linux@gmail.com>
11 years agobf609: add adv7343 video encoder support
Scott Jiang [Wed, 7 Aug 2013 06:30:43 +0000 (02:30 -0400)]
bf609: add adv7343 video encoder support

Signed-off-by: Scott Jiang <scott.jiang.linux@gmail.com>
11 years agoclock: add stmmac clock for ethernet driver
Steven Miao [Fri, 5 Jul 2013 06:36:09 +0000 (14:36 +0800)]
clock: add stmmac clock for ethernet driver

Signed-off-by: Steven Miao <realmz6@gmail.com>
11 years agoblackfin: scb: Add SCB1 to SCB9 config options and data.
Sonic Zhang [Thu, 15 Nov 2012 04:32:26 +0000 (12:32 +0800)]
blackfin: scb: Add SCB1 to SCB9 config options and data.

Signed-off-by: Sonic Zhang <sonic.zhang@analog.com>
11 years agoblackfin: scb: Add system crossbar init code.
Steven Miao [Thu, 12 Sep 2013 08:36:16 +0000 (16:36 +0800)]
blackfin: scb: Add system crossbar init code.

If SCB exists in select blackfin cpu, developer can change the SCB
priority in kernel configuration.

Signed-off-by: Sonic Zhang <sonic.zhang@analog.com>
Signed-off-by: Steven Miao <realmz6@gmail.com>
11 years agoMerge branch 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus
Linus Torvalds [Thu, 12 Sep 2013 23:14:49 +0000 (16:14 -0700)]
Merge branch 'upstream' of git://git.linux-mips.org/ralf/upstream-linus

Pull MIPS updates from Ralf Baechle:
 "This has been sitting in -next for a while with no objections and all
  MIPS defconfigs except one are building fine; that one platform got
  broken by another patch in your tree and I'm going to submit a patch
  separately.

   - a handful of fixes that didn't make 3.11
   - a few bits of Octeon 3 support with more to come for a later
     release
   - platform enhancements for Octeon, ath79, Lantiq, Netlogic and
     Ralink SOCs
   - a GPIO driver for the Octeon
   - some dusting off of the DECstation code
   - the usual dose of cleanups"

* 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus: (65 commits)
  MIPS: DMA: Fix BUG due to smp_processor_id() in preemptible code
  MIPS: kexec: Fix random crashes while loading crashkernel
  MIPS: kdump: Skip walking indirection page for crashkernels
  MIPS: DECstation HRT calibration bug fixes
  MIPS: Export copy_from_user_page() (needed by lustre)
  MIPS: Add driver for the built-in PCI controller of the RT3883 SoC
  MIPS: DMA: For BMIPS5000 cores flush region just like non-coherent R10000
  MIPS: ralink: Add support for reset-controller API
  MIPS: ralink: mt7620: Add cpu-feature-override header
  MIPS: ralink: mt7620: Add spi clock definition
  MIPS: ralink: mt7620: Add wdt clock definition
  MIPS: ralink: mt7620: Improve clock frequency detection
  MIPS: ralink: mt7620: This SoC has EHCI and OHCI hosts
  MIPS: ralink: mt7620: Add verbose ram info
  MIPS: ralink: Probe clocksources from OF
  MIPS: ralink: Add support for systick timer found on newer ralink SoC
  MIPS: ralink: Add support for periodic timer irq
  MIPS: Netlogic: Built-in DTB for XLP2xx SoC boards
  MIPS: Netlogic: Add support for USB on XLP2xx
  MIPS: Netlogic: XLP2xx update for I2C controller
  ...

11 years agoMerge tag 'xfs-for-linus-v3.12-rc1-2' of git://oss.sgi.com/xfs/xfs
Linus Torvalds [Thu, 12 Sep 2013 23:13:41 +0000 (16:13 -0700)]
Merge tag 'xfs-for-linus-v3.12-rc1-2' of git://oss.sgi.com/xfs/xfs

Pull xfs update #2 from Ben Myers:
 "Here we have defrag support for v5 superblock, a number of bugfixes
  and a cleanup or two.

   - defrag support for CRC filesystems
   - fix endian worning in xlog_recover_get_buf_lsn
   - fixes for sparse warnings
   - fix for assert in xfs_dir3_leaf_hdr_from_disk
   - fix for log recovery of remote symlinks
   - fix for log recovery of btree root splits
   - fixes formemory allocation failures with ACLs
   - fix for assert in xfs_buf_item_relse
   - fix for assert in xfs_inode_buf_verify
   - fix an assignment in an assert that should be a test in
     xfs_bmbt_change_owner
   - remove dead code in xlog_recover_inode_pass2"

* tag 'xfs-for-linus-v3.12-rc1-2' of git://oss.sgi.com/xfs/xfs:
  xfs: remove dead code from xlog_recover_inode_pass2
  xfs: = vs == typo in ASSERT()
  xfs: don't assert fail on bad inode numbers
  xfs: aborted buf items can be in the AIL.
  xfs: factor all the kmalloc-or-vmalloc fallback allocations
  xfs: fix memory allocation failures with ACLs
  xfs: ensure we copy buffer type in da btree root splits
  xfs: set remote symlink buffer type for recovery
  xfs: recovery of swap extents operations for CRC filesystems
  xfs: swap extents operations for CRC filesystems
  xfs: check magic numbers in dir3 leaf verifier first
  xfs: fix some minor sparse warnings
  xfs: fix endian warning in xlog_recover_get_buf_lsn()

11 years agoMerge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/nab/target...
Linus Torvalds [Thu, 12 Sep 2013 23:11:45 +0000 (16:11 -0700)]
Merge branch 'for-next' of git://git./linux/kernel/git/nab/target-pending

Pull SCSI target updates from Nicholas Bellinger:
 "Lots of activity again this round for I/O performance optimizations
  (per-cpu IDA pre-allocation for vhost + iscsi/target), and the
  addition of new fabric independent features to target-core
  (COMPARE_AND_WRITE + EXTENDED_COPY).

  The main highlights include:

   - Support for iscsi-target login multiplexing across individual
     network portals
   - Generic Per-cpu IDA logic (kent + akpm + clameter)
   - Conversion of vhost to use per-cpu IDA pre-allocation for
     descriptors, SGLs and userspace page pointer list
   - Conversion of iscsi-target + iser-target to use per-cpu IDA
     pre-allocation for descriptors
   - Add support for generic COMPARE_AND_WRITE (AtomicTestandSet)
     emulation for virtual backend drivers
   - Add support for generic EXTENDED_COPY (CopyOffload) emulation for
     virtual backend drivers.
   - Add support for fast memory registration mode to iser-target (Vu)

  The patches to add COMPARE_AND_WRITE and EXTENDED_COPY support are of
  particular significance, which make us the first and only open source
  target to support the full set of VAAI primitives.

  Currently Linux clients are lacking upstream support to actually
  utilize these primitives.  However, with server side support now in
  place for folks like MKP + ZAB working on the client, this logic once
  reserved for the highest end of storage arrays, can now be run in VMs
  on their laptops"

* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/nab/target-pending: (50 commits)
  target/iscsi: Bump versions to v4.1.0
  target: Update copyright ownership/year information to 2013
  iscsi-target: Bump default TCP listen backlog to 256
  target: Fix >= v3.9+ regression in PR APTPL + ALUA metadata write-out
  iscsi-target; Bump default CmdSN Depth to 64
  iscsi-target: Remove unnecessary wait_for_completion in iscsi_get_thread_set
  iscsi-target: Add thread_set->ts_activate_sem + use common deallocate
  iscsi-target: Fix race with thread_pre_handler flush_signals + ISCSI_THREAD_SET_DIE
  target: remove unused including <linux/version.h>
  iser-target: introduce fast memory registration mode (FRWR)
  iser-target: generalize rdma memory registration and cleanup
  iser-target: move rdma wr processing to a shared function
  target: Enable global EXTENDED_COPY setup/release
  target: Add Third Party Copy (3PC) bit in INQUIRY response
  target: Enable EXTENDED_COPY setup in spc_parse_cdb
  target: Add support for EXTENDED_COPY copy offload emulation
  target: Avoid non-existent tg_pt_gp_mem in target_alua_state_check
  target: Add global device list for EXTENDED_COPY
  target: Make helpers non static for EXTENDED_COPY command setup
  target: Make spc_parse_naa_6h_vendor_specific non static
  ...

11 years agoMerge branch 'akpm' (patches from Andrew Morton)
Linus Torvalds [Thu, 12 Sep 2013 22:44:27 +0000 (15:44 -0700)]
Merge branch 'akpm' (patches from Andrew Morton)

Merge more patches from Andrew Morton:
 "The rest of MM.  Plus one misc cleanup"

* emailed patches from Andrew Morton <akpm@linux-foundation.org>: (35 commits)
  mm/Kconfig: add MMU dependency for MIGRATION.
  kernel: replace strict_strto*() with kstrto*()
  mm, thp: count thp_fault_fallback anytime thp fault fails
  thp: consolidate code between handle_mm_fault() and do_huge_pmd_anonymous_page()
  thp: do_huge_pmd_anonymous_page() cleanup
  thp: move maybe_pmd_mkwrite() out of mk_huge_pmd()
  mm: cleanup add_to_page_cache_locked()
  thp: account anon transparent huge pages into NR_ANON_PAGES
  truncate: drop 'oldsize' truncate_pagecache() parameter
  mm: make lru_add_drain_all() selective
  memcg: document cgroup dirty/writeback memory statistics
  memcg: add per cgroup writeback pages accounting
  memcg: check for proper lock held in mem_cgroup_update_page_stat
  memcg: remove MEMCG_NR_FILE_MAPPED
  memcg: reduce function dereference
  memcg: avoid overflow caused by PAGE_ALIGN
  memcg: rename RESOURCE_MAX to RES_COUNTER_MAX
  memcg: correct RESOURCE_MAX to ULLONG_MAX
  mm: memcg: do not trap chargers with full callstack on OOM
  mm: memcg: rework and document OOM waiting and wakeup
  ...

11 years agomm/Kconfig: add MMU dependency for MIGRATION.
Chen Gang [Thu, 12 Sep 2013 22:14:08 +0000 (15:14 -0700)]
mm/Kconfig: add MMU dependency for MIGRATION.

MIGRATION must depend on MMU, or allmodconfig for the nommu sh
architecture fails to build:

    CC      mm/migrate.o
  mm/migrate.c: In function 'remove_migration_pte':
  mm/migrate.c:134:3: error: implicit declaration of function 'pmd_trans_huge' [-Werror=implicit-function-declaration]
     if (pmd_trans_huge(*pmd))
     ^
  mm/migrate.c:149:2: error: implicit declaration of function 'is_swap_pte' [-Werror=implicit-function-declaration]
    if (!is_swap_pte(pte))
    ^
  ...

Also let CMA depend on MMU, or when NOMMU, if we select CMA, it will
select MIGRATION by force.

Signed-off-by: Chen Gang <gang.chen@asianux.com>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agokernel: replace strict_strto*() with kstrto*()
Jingoo Han [Thu, 12 Sep 2013 22:14:07 +0000 (15:14 -0700)]
kernel: replace strict_strto*() with kstrto*()

The usage of strict_strto*() is not preferred, because strict_strto*() is
obsolete.  Thus, kstrto*() should be used.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agomm, thp: count thp_fault_fallback anytime thp fault fails
David Rientjes [Thu, 12 Sep 2013 22:14:06 +0000 (15:14 -0700)]
mm, thp: count thp_fault_fallback anytime thp fault fails

Currently, thp_fault_fallback in vmstat only gets incremented if a
hugepage allocation fails.  If current's memcg hits its limit or the page
fault handler returns an error, it is incorrectly accounted as a
successful thp_fault_alloc.

Count thp_fault_fallback anytime the page fault handler falls back to
using regular pages and only count thp_fault_alloc when a hugepage has
actually been faulted.

Signed-off-by: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agothp: consolidate code between handle_mm_fault() and do_huge_pmd_anonymous_page()
Kirill A. Shutemov [Thu, 12 Sep 2013 22:14:05 +0000 (15:14 -0700)]
thp: consolidate code between handle_mm_fault() and do_huge_pmd_anonymous_page()

do_huge_pmd_anonymous_page() has copy-pasted piece of handle_mm_fault()
to handle fallback path.

Let's consolidate code back by introducing VM_FAULT_FALLBACK return
code.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Hillf Danton <dhillf@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Hugh Dickins <hughd@google.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Matthew Wilcox <willy@linux.intel.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agothp: do_huge_pmd_anonymous_page() cleanup
Kirill A. Shutemov [Thu, 12 Sep 2013 22:14:03 +0000 (15:14 -0700)]
thp: do_huge_pmd_anonymous_page() cleanup

Minor cleanup: unindent most code of the fucntion by inverting one
condition.  It's preparation for the next patch.

No functional changes.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Hillf Danton <dhillf@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Hugh Dickins <hughd@google.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Matthew Wilcox <willy@linux.intel.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agothp: move maybe_pmd_mkwrite() out of mk_huge_pmd()
Kirill A. Shutemov [Thu, 12 Sep 2013 22:14:01 +0000 (15:14 -0700)]
thp: move maybe_pmd_mkwrite() out of mk_huge_pmd()

It's confusing that mk_huge_pmd() has semantics different from mk_pte() or
mk_pmd().  I spent some time on debugging issue cased by this
inconsistency.

Let's move maybe_pmd_mkwrite() out of mk_huge_pmd() and adjust prototype
to match mk_pte().

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Hugh Dickins <hughd@google.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Matthew Wilcox <willy@linux.intel.com>
Cc: Hillf Danton <dhillf@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agomm: cleanup add_to_page_cache_locked()
Kirill A. Shutemov [Thu, 12 Sep 2013 22:13:59 +0000 (15:13 -0700)]
mm: cleanup add_to_page_cache_locked()

Make add_to_page_cache_locked() cleaner:

 - unindent most code of the function by inverting one condition;
 - streamline code no-error path;
 - move insert error path outside normal code path;
 - call radix_tree_preload_end() earlier;

No functional changes.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Hugh Dickins <hughd@google.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Matthew Wilcox <willy@linux.intel.com>
Cc: Hillf Danton <dhillf@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agothp: account anon transparent huge pages into NR_ANON_PAGES
Kirill A. Shutemov [Thu, 12 Sep 2013 22:13:58 +0000 (15:13 -0700)]
thp: account anon transparent huge pages into NR_ANON_PAGES

We use NR_ANON_PAGES as base for reporting AnonPages to user.  There's
not much sense in not accounting transparent huge pages there, but add
them on printing to user.

Let's account transparent huge pages in NR_ANON_PAGES in the first place.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Hugh Dickins <hughd@google.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Matthew Wilcox <willy@linux.intel.com>
Cc: Hillf Danton <dhillf@gmail.com>
Cc: Ning Qu <quning@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agotruncate: drop 'oldsize' truncate_pagecache() parameter
Kirill A. Shutemov [Thu, 12 Sep 2013 22:13:56 +0000 (15:13 -0700)]
truncate: drop 'oldsize' truncate_pagecache() parameter

truncate_pagecache() doesn't care about old size since commit
cedabed49b39 ("vfs: Fix vmtruncate() regression").  Let's drop it.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agomm: make lru_add_drain_all() selective
Chris Metcalf [Thu, 12 Sep 2013 22:13:55 +0000 (15:13 -0700)]
mm: make lru_add_drain_all() selective

make lru_add_drain_all() only selectively interrupt the cpus that have
per-cpu free pages that can be drained.

This is important in nohz mode where calling mlockall(), for example,
otherwise will interrupt every core unnecessarily.

This is important on workloads where nohz cores are handling 10 Gb traffic
in userspace.  Those CPUs do not enter the kernel and place pages into LRU
pagevecs and they really, really don't want to be interrupted, or they
drop packets on the floor.

Signed-off-by: Chris Metcalf <cmetcalf@tilera.com>
Reviewed-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agomemcg: document cgroup dirty/writeback memory statistics
Sha Zhengju [Thu, 12 Sep 2013 22:13:54 +0000 (15:13 -0700)]
memcg: document cgroup dirty/writeback memory statistics

Signed-off-by: Sha Zhengju <handai.szj@taobao.com>
Cc: Fengguang Wu <fengguang.wu@intel.com>
Cc: Greg Thelen <gthelen@google.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agomemcg: add per cgroup writeback pages accounting
Sha Zhengju [Thu, 12 Sep 2013 22:13:53 +0000 (15:13 -0700)]
memcg: add per cgroup writeback pages accounting

Add memcg routines to count writeback pages, later dirty pages will also
be accounted.

After Kame's commit 89c06bd52fb9 ("memcg: use new logic for page stat
accounting"), we can use 'struct page' flag to test page state instead
of per page_cgroup flag.  But memcg has a feature to move a page from a
cgroup to another one and may have race between "move" and "page stat
accounting".  So in order to avoid the race we have designed a new lock:

         mem_cgroup_begin_update_page_stat()
         modify page information        -->(a)
         mem_cgroup_update_page_stat()  -->(b)
         mem_cgroup_end_update_page_stat()

It requires both (a) and (b)(writeback pages accounting) to be pretected
in mem_cgroup_{begin/end}_update_page_stat().  It's full no-op for
!CONFIG_MEMCG, almost no-op if memcg is disabled (but compiled in), rcu
read lock in the most cases (no task is moving), and spin_lock_irqsave
on top in the slow path.

There're two writeback interfaces to modify: test_{clear/set}_page_writeback().
And the lock order is:
--> memcg->move_lock
  --> mapping->tree_lock

Signed-off-by: Sha Zhengju <handai.szj@taobao.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Reviewed-by: Greg Thelen <gthelen@google.com>
Cc: Fengguang Wu <fengguang.wu@intel.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.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>
11 years agomemcg: check for proper lock held in mem_cgroup_update_page_stat
Sha Zhengju [Thu, 12 Sep 2013 22:13:52 +0000 (15:13 -0700)]
memcg: check for proper lock held in mem_cgroup_update_page_stat

We should call mem_cgroup_begin_update_page_stat() before
mem_cgroup_update_page_stat() to get proper locks, however the latter
doesn't do any checking that we use proper locking, which would be hard.
Suggested by Michal Hock we could at least test for rcu_read_lock_held()
because RCU is held if !mem_cgroup_disabled().

Signed-off-by: Sha Zhengju <handai.szj@taobao.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Reviewed-by: Greg Thelen <gthelen@google.com>
Cc: Fengguang Wu <fengguang.wu@intel.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.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>
11 years agomemcg: remove MEMCG_NR_FILE_MAPPED
Sha Zhengju [Thu, 12 Sep 2013 22:13:50 +0000 (15:13 -0700)]
memcg: remove MEMCG_NR_FILE_MAPPED

While accounting memcg page stat, it's not worth to use
MEMCG_NR_FILE_MAPPED as an extra layer of indirection because of the
complexity and presumed performance overhead.  We can use
MEM_CGROUP_STAT_FILE_MAPPED directly.

Signed-off-by: Sha Zhengju <handai.szj@taobao.com>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Acked-by: Fengguang Wu <fengguang.wu@intel.com>
Reviewed-by: Greg Thelen <gthelen@google.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>
11 years agomemcg: reduce function dereference
Sha Zhengju [Thu, 12 Sep 2013 22:13:49 +0000 (15:13 -0700)]
memcg: reduce function dereference

This function dereferences res far too often, so optimize it.

Signed-off-by: Sha Zhengju <handai.szj@taobao.com>
Signed-off-by: Qiang Huang <h.huangqiang@huawei.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Jeff Liu <jeff.liu@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agomemcg: avoid overflow caused by PAGE_ALIGN
Sha Zhengju [Thu, 12 Sep 2013 22:13:48 +0000 (15:13 -0700)]
memcg: avoid overflow caused by PAGE_ALIGN

Since PAGE_ALIGN is aligning up(the next page boundary), so after
PAGE_ALIGN, the value might be overflow, such as write the MAX value to
*.limit_in_bytes.

  $ cat /cgroup/memory/memory.limit_in_bytes
  18446744073709551615

  # echo 18446744073709551615 > /cgroup/memory/memory.limit_in_bytes
  bash: echo: write error: Invalid argument

Some user programs might depend on such behaviours(like libcg, we read
the value in snapshot, then use the value to reset cgroup later), and
that will cause confusion.  So we need to fix it.

Signed-off-by: Sha Zhengju <handai.szj@taobao.com>
Signed-off-by: Qiang Huang <h.huangqiang@huawei.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Jeff Liu <jeff.liu@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agomemcg: rename RESOURCE_MAX to RES_COUNTER_MAX
Sha Zhengju [Thu, 12 Sep 2013 22:13:47 +0000 (15:13 -0700)]
memcg: rename RESOURCE_MAX to RES_COUNTER_MAX

RESOURCE_MAX is far too general name, change it to RES_COUNTER_MAX.

Signed-off-by: Sha Zhengju <handai.szj@taobao.com>
Signed-off-by: Qiang Huang <h.huangqiang@huawei.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Jeff Liu <jeff.liu@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agomemcg: correct RESOURCE_MAX to ULLONG_MAX
Sha Zhengju [Thu, 12 Sep 2013 22:13:46 +0000 (15:13 -0700)]
memcg: correct RESOURCE_MAX to ULLONG_MAX

Current RESOURCE_MAX is ULONG_MAX, but the value we used to set resource
limit is unsigned long long, so we can set bigger value than that which is
strange.  The XXX_MAX should be reasonable max value, bigger than that
should be overflow.

Notice that this change will affect user output of default *.limit_in_bytes:
before change:

  $ cat /cgroup/memory/memory.limit_in_bytes
  9223372036854775807

after change:

  $ cat /cgroup/memory/memory.limit_in_bytes
  18446744073709551615

But it doesn't alter the API in term of input - we can still use "echo -1
> *.limit_in_bytes" to reset the numbers to "unlimited".

Signed-off-by: Sha Zhengju <handai.szj@taobao.com>
Signed-off-by: Qiang Huang <h.huangqiang@huawei.com>
Acked-by: Michal Hocko <mhocko@suse.cz>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: Jeff Liu <jeff.liu@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agomm: memcg: do not trap chargers with full callstack on OOM
Johannes Weiner [Thu, 12 Sep 2013 22:13:44 +0000 (15:13 -0700)]
mm: memcg: do not trap chargers with full callstack on OOM

The memcg OOM handling is incredibly fragile and can deadlock.  When a
task fails to charge memory, it invokes the OOM killer and loops right
there in the charge code until it succeeds.  Comparably, any other task
that enters the charge path at this point will go to a waitqueue right
then and there and sleep until the OOM situation is resolved.  The problem
is that these tasks may hold filesystem locks and the mmap_sem; locks that
the selected OOM victim may need to exit.

For example, in one reported case, the task invoking the OOM killer was
about to charge a page cache page during a write(), which holds the
i_mutex.  The OOM killer selected a task that was just entering truncate()
and trying to acquire the i_mutex:

OOM invoking task:
  mem_cgroup_handle_oom+0x241/0x3b0
  mem_cgroup_cache_charge+0xbe/0xe0
  add_to_page_cache_locked+0x4c/0x140
  add_to_page_cache_lru+0x22/0x50
  grab_cache_page_write_begin+0x8b/0xe0
  ext3_write_begin+0x88/0x270
  generic_file_buffered_write+0x116/0x290
  __generic_file_aio_write+0x27c/0x480
  generic_file_aio_write+0x76/0xf0           # takes ->i_mutex
  do_sync_write+0xea/0x130
  vfs_write+0xf3/0x1f0
  sys_write+0x51/0x90
  system_call_fastpath+0x18/0x1d

OOM kill victim:
  do_truncate+0x58/0xa0              # takes i_mutex
  do_last+0x250/0xa30
  path_openat+0xd7/0x440
  do_filp_open+0x49/0xa0
  do_sys_open+0x106/0x240
  sys_open+0x20/0x30
  system_call_fastpath+0x18/0x1d

The OOM handling task will retry the charge indefinitely while the OOM
killed task is not releasing any resources.

A similar scenario can happen when the kernel OOM killer for a memcg is
disabled and a userspace task is in charge of resolving OOM situations.
In this case, ALL tasks that enter the OOM path will be made to sleep on
the OOM waitqueue and wait for userspace to free resources or increase
the group's limit.  But a userspace OOM handler is prone to deadlock
itself on the locks held by the waiting tasks.  For example one of the
sleeping tasks may be stuck in a brk() call with the mmap_sem held for
writing but the userspace handler, in order to pick an optimal victim,
may need to read files from /proc/<pid>, which tries to acquire the same
mmap_sem for reading and deadlocks.

This patch changes the way tasks behave after detecting a memcg OOM and
makes sure nobody loops or sleeps with locks held:

1. When OOMing in a user fault, invoke the OOM killer and restart the
   fault instead of looping on the charge attempt.  This way, the OOM
   victim can not get stuck on locks the looping task may hold.

2. When OOMing in a user fault but somebody else is handling it
   (either the kernel OOM killer or a userspace handler), don't go to
   sleep in the charge context.  Instead, remember the OOMing memcg in
   the task struct and then fully unwind the page fault stack with
   -ENOMEM.  pagefault_out_of_memory() will then call back into the
   memcg code to check if the -ENOMEM came from the memcg, and then
   either put the task to sleep on the memcg's OOM waitqueue or just
   restart the fault.  The OOM victim can no longer get stuck on any
   lock a sleeping task may hold.

Debugged by Michal Hocko.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reported-by: azurIt <azurit@pobox.sk>
Acked-by: Michal Hocko <mhocko@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: 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>
11 years agomm: memcg: rework and document OOM waiting and wakeup
Johannes Weiner [Thu, 12 Sep 2013 22:13:43 +0000 (15:13 -0700)]
mm: memcg: rework and document OOM waiting and wakeup

The memcg OOM handler open-codes a sleeping lock for OOM serialization
(trylock, wait, repeat) because the required locking is so specific to
memcg hierarchies.  However, it would be nice if this construct would be
clearly recognizable and not be as obfuscated as it is right now.  Clean
up as follows:

1. Remove the return value of mem_cgroup_oom_unlock()

2. Rename mem_cgroup_oom_lock() to mem_cgroup_oom_trylock().

3. Pull the prepare_to_wait() out of the memcg_oom_lock scope.  This
   makes it more obvious that the task has to be on the waitqueue
   before attempting to OOM-trylock the hierarchy, to not miss any
   wakeups before going to sleep.  It just didn't matter until now
   because it was all lumped together into the global memcg_oom_lock
   spinlock section.

4. Pull the mem_cgroup_oom_notify() out of the memcg_oom_lock scope.
   It is proctected by the hierarchical OOM-lock.

5. The memcg_oom_lock spinlock is only required to propagate the OOM
   lock in any given hierarchy atomically.  Restrict its scope to
   mem_cgroup_oom_(trylock|unlock).

6. Do not wake up the waitqueue unconditionally at the end of the
   function.  Only the lockholder has to wake up the next in line
   after releasing the lock.

   Note that the lockholder kicks off the OOM-killer, which in turn
   leads to wakeups from the uncharges of the exiting task.  But a
   contender is not guaranteed to see them if it enters the OOM path
   after the OOM kills but before the lockholder releases the lock.
   Thus there has to be an explicit wakeup after releasing the lock.

7. Put the OOM task on the waitqueue before marking the hierarchy as
   under OOM as that is the point where we start to receive wakeups.
   No point in listening before being on the waitqueue.

8. Likewise, unmark the hierarchy before finishing the sleep, for
   symmetry.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Michal Hocko <mhocko@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: azurIt <azurit@pobox.sk>
Cc: 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>
11 years agomm: memcg: enable memcg OOM killer only for user faults
Johannes Weiner [Thu, 12 Sep 2013 22:13:42 +0000 (15:13 -0700)]
mm: memcg: enable memcg OOM killer only for user faults

System calls and kernel faults (uaccess, gup) can handle an out of memory
situation gracefully and just return -ENOMEM.

Enable the memcg OOM killer only for user faults, where it's really the
only option available.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Michal Hocko <mhocko@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: azurIt <azurit@pobox.sk>
Cc: 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>
11 years agox86: finish user fault error path with fatal signal
Johannes Weiner [Thu, 12 Sep 2013 22:13:40 +0000 (15:13 -0700)]
x86: finish user fault error path with fatal signal

The x86 fault handler bails in the middle of error handling when the
task has a fatal signal pending.  For a subsequent patch this is a
problem in OOM situations because it relies on pagefault_out_of_memory()
being called even when the task has been killed, to perform proper
per-task OOM state unwinding.

Shortcutting the fault like this is a rather minor optimization that
saves a few instructions in rare cases.  Just remove it for
user-triggered faults.

Use the opportunity to split the fault retry handling from actual fault
errors and add locking documentation that reads suprisingly similar to
ARM's.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: azurIt <azurit@pobox.sk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoarch: mm: pass userspace fault flag to generic fault handler
Johannes Weiner [Thu, 12 Sep 2013 22:13:39 +0000 (15:13 -0700)]
arch: mm: pass userspace fault flag to generic fault handler

Unlike global OOM handling, memory cgroup code will invoke the OOM killer
in any OOM situation because it has no way of telling faults occuring in
kernel context - which could be handled more gracefully - from
user-triggered faults.

Pass a flag that identifies faults originating in user space from the
architecture-specific fault handlers to generic code so that memcg OOM
handling can be improved.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Cc: David Rientjes <rientjes@google.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: azurIt <azurit@pobox.sk>
Cc: 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>
11 years agoarch: mm: do not invoke OOM killer on kernel fault OOM
Johannes Weiner [Thu, 12 Sep 2013 22:13:38 +0000 (15:13 -0700)]
arch: mm: do not invoke OOM killer on kernel fault OOM

Kernel faults are expected to handle OOM conditions gracefully (gup,
uaccess etc.), so they should never invoke the OOM killer.  Reserve this
for faults triggered in user context when it is the only option.

Most architectures already do this, fix up the remaining few.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: azurIt <azurit@pobox.sk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoarch: mm: remove obsolete init OOM protection
Johannes Weiner [Thu, 12 Sep 2013 22:13:36 +0000 (15:13 -0700)]
arch: mm: remove obsolete init OOM protection

The memcg code can trap tasks in the context of the failing allocation
until an OOM situation is resolved.  They can hold all kinds of locks
(fs, mm) at this point, which makes it prone to deadlocking.

This series converts memcg OOM handling into a two step process that is
started in the charge context, but any waiting is done after the fault
stack is fully unwound.

Patches 1-4 prepare architecture handlers to support the new memcg
requirements, but in doing so they also remove old cruft and unify
out-of-memory behavior across architectures.

Patch 5 disables the memcg OOM handling for syscalls, readahead, kernel
faults, because they can gracefully unwind the stack with -ENOMEM.  OOM
handling is restricted to user triggered faults that have no other
option.

Patch 6 reworks memcg's hierarchical OOM locking to make it a little
more obvious wth is going on in there: reduce locked regions, rename
locking functions, reorder and document.

Patch 7 implements the two-part OOM handling such that tasks are never
trapped with the full charge stack in an OOM situation.

This patch:

Back before smart OOM killing, when faulting tasks were killed directly on
allocation failures, the arch-specific fault handlers needed special
protection for the init process.

Now that all fault handlers call into the generic OOM killer (see commit
609838cfed97: "mm: invoke oom-killer from remaining unconverted page
fault handlers"), which already provides init protection, the
arch-specific leftovers can be removed.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: David Rientjes <rientjes@google.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: azurIt <azurit@pobox.sk>
Acked-by: Vineet Gupta <vgupta@synopsys.com> [arch/arc bits]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agomemcg: trivial cleanups
Andrew Morton [Thu, 12 Sep 2013 22:13:35 +0000 (15:13 -0700)]
memcg: trivial cleanups

Clean up some mess made by the "Soft limit rework" series, and a few other
things.

Cc: Michal Hocko <mhocko@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agomemcg, vmscan: do not fall into reclaim-all pass too quickly
Michal Hocko [Thu, 12 Sep 2013 22:13:34 +0000 (15:13 -0700)]
memcg, vmscan: do not fall into reclaim-all pass too quickly

shrink_zone starts with soft reclaim pass first and then falls back to
regular reclaim if nothing has been scanned.  This behavior is natural
but there is a catch.  Memcg iterators, when used with the reclaim
cookie, are designed to help to prevent from over reclaim by
interleaving reclaimers (per node-zone-priority) so the tree walk might
miss many (even all) nodes in the hierarchy e.g.  when there are direct
reclaimers racing with each other or with kswapd in the global case or
multiple allocators reaching the limit for the target reclaim case.  To
make it even more complicated, targeted reclaim doesn't do the whole
tree walk because it stops reclaiming once it reclaims sufficient pages.
As a result groups over the limit might be missed, thus nothing is
scanned, and reclaim would fall back to the reclaim all mode.

This patch checks for the incomplete tree walk in shrink_zone.  If no
group has been visited and the hierarchy is soft reclaimable then we
must have missed some groups, in which case the __shrink_zone is called
again.  This doesn't guarantee there will be some progress of course
because the current reclaimer might be still racing with others but it
would at least give a chance to start the walk without a big risk of
reclaim latencies.

Signed-off-by: Michal Hocko <mhocko@suse.cz>
Cc: Balbir Singh <bsingharora@gmail.com>
Cc: Glauber Costa <glommer@openvz.org>
Cc: Greg Thelen <gthelen@google.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Michel Lespinasse <walken@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Ying Han <yinghan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agomemcg: track all children over limit in the root
Michal Hocko [Thu, 12 Sep 2013 22:13:32 +0000 (15:13 -0700)]
memcg: track all children over limit in the root

Children in soft limit excess are currently tracked up the hierarchy in
memcg->children_in_excess.  Nevertheless there still might exist tons of
groups that are not in hierarchy relation to the root cgroup (e.g.  all
first level groups if root_mem_cgroup->use_hierarchy == false).

As the whole tree walk has to be done when the iteration starts at
root_mem_cgroup the iterator should be able to skip the walk if there is
no child above the limit without iterating them.  This can be done
easily if the root tracks all children rather than only hierarchical
children.  This is done by this patch which updates root_mem_cgroup
children_in_excess if root_mem_cgroup->use_hierarchy == false so the
root knows about all children in excess.

Please note that this is not an issue for inner memcgs which have
use_hierarchy == false because then only the single group is visited so
no special optimization is necessary.

Signed-off-by: Michal Hocko <mhocko@suse.cz>
Cc: Balbir Singh <bsingharora@gmail.com>
Cc: Glauber Costa <glommer@openvz.org>
Cc: Greg Thelen <gthelen@google.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Michel Lespinasse <walken@google.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Ying Han <yinghan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>