GitHub/LineageOS/G12/android_kernel_amlogic_linux-4.9.git
9 years agoMerge branch 'for-4.1/core-noarch' into for-linus
Jiri Kosina [Mon, 13 Apr 2015 21:57:20 +0000 (23:57 +0200)]
Merge branch 'for-4.1/core-noarch' into for-linus

9 years agolivepatch: Fix subtle race with coming and going modules
Petr Mladek [Thu, 12 Mar 2015 11:55:13 +0000 (12:55 +0100)]
livepatch: Fix subtle race with coming and going modules

There is a notifier that handles live patches for coming and going modules.
It takes klp_mutex lock to avoid races with coming and going patches but
it does not keep the lock all the time. Therefore the following races are
possible:

  1. The notifier is called sometime in STATE_MODULE_COMING. The module
     is visible by find_module() in this state all the time. It means that
     new patch can be registered and enabled even before the notifier is
     called. It might create wrong order of stacked patches, see below
     for an example.

   2. New patch could still see the module in the GOING state even after
      the notifier has been called. It will try to initialize the related
      object structures but the module could disappear at any time. There
      will stay mess in the structures. It might even cause an invalid
      memory access.

This patch solves the problem by adding a boolean variable into struct module.
The value is true after the coming and before the going handler is called.
New patches need to be applied when the value is true and they need to ignore
the module when the value is false.

Note that we need to know state of all modules on the system. The races are
related to new patches. Therefore we do not know what modules will get
patched.

Also note that we could not simply ignore going modules. The code from the
module could be called even in the GOING state until mod->exit() finishes.
If we start supporting patches with semantic changes between function
calls, we need to apply new patches to any still usable code.
See below for an example.

Finally note that the patch solves only the situation when a new patch is
registered. There are no such problems when the patch is being removed.
It does not matter who disable the patch first, whether the normal
disable_patch() or the module notifier. There is nothing to do
once the patch is disabled.

Alternative solutions:
======================

+ reject new patches when a patched module is coming or going; this is ugly

+ wait with adding new patch until the module leaves the COMING and GOING
  states; this might be dangerous and complicated; we would need to release
  kgr_lock in the middle of the patch registration to avoid a deadlock
  with the coming and going handlers; also we might need a waitqueue for
  each module which seems to be even bigger overhead than the boolean

+ stop modules from entering COMING and GOING states; wait until modules
  leave these states when they are already there; looks complicated; we would
  need to ignore the module that asked to stop the others to avoid a deadlock;
  also it is unclear what to do when two modules asked to stop others and
  both are in COMING state (situation when two new patches are applied)

+ always register/enable new patches and fix up the potential mess (registered
  patches order) in klp_module_init(); this is nasty and prone to regressions
  in the future development

+ add another MODULE_STATE where the kallsyms are visible but the module is not
  used yet; this looks too complex; the module states are checked on "many"
  locations

Example of patch stacking breakage:
===================================

The notifier could _not_ _simply_ ignore already initialized module objects.
For example, let's have three patches (P1, P2, P3) for functions a() and b()
where a() is from vmcore and b() is from a module M. Something like:

a() b()
P1 a1() b1()
P2 a2() b2()
P3 a3() b3(3)

If you load the module M after all patches are registered and enabled.
The ftrace ops for function a() and b() has listed the functions in this
order:

ops_a->func_stack -> list(a3,a2,a1)
ops_b->func_stack -> list(b3,b2,b1)

, so the pointer to b3() is the first and will be used.

Then you might have the following scenario. Let's start with state when patches
P1 and P2 are registered and enabled but the module M is not loaded. Then ftrace
ops for b() does not exist. Then we get into the following race:

CPU0 CPU1

load_module(M)

  complete_formation()

  mod->state = MODULE_STATE_COMING;
  mutex_unlock(&module_mutex);

klp_register_patch(P3);
klp_enable_patch(P3);

# STATE 1

  klp_module_notify(M)
    klp_module_notify_coming(P1);
    klp_module_notify_coming(P2);
    klp_module_notify_coming(P3);

# STATE 2

The ftrace ops for a() and b() then looks:

  STATE1:

ops_a->func_stack -> list(a3,a2,a1);
ops_b->func_stack -> list(b3);

  STATE2:
ops_a->func_stack -> list(a3,a2,a1);
ops_b->func_stack -> list(b2,b1,b3);

therefore, b2() is used for the module but a3() is used for vmcore
because they were the last added.

Example of the race with going modules:
=======================================

CPU0 CPU1

delete_module()  #SYSCALL

   try_stop_module()
     mod->state = MODULE_STATE_GOING;

   mutex_unlock(&module_mutex);

klp_register_patch()
klp_enable_patch()

#save place to switch universe

b()     # from module that is going
  a()   # from core (patched)

   mod->exit();

Note that the function b() can be called until we call mod->exit().

If we do not apply patch against b() because it is in MODULE_STATE_GOING,
it will call patched a() with modified semantic and things might get wrong.

[jpoimboe@redhat.com: use one boolean instead of two]
Signed-off-by: Petr Mladek <pmladek@suse.cz>
Acked-by: Josh Poimboeuf <jpoimboe@redhat.com>
Acked-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
9 years agolivepatch: remove unnecessary call to klp_find_object_module()
Josh Poimboeuf [Tue, 3 Mar 2015 23:02:21 +0000 (17:02 -0600)]
livepatch: remove unnecessary call to klp_find_object_module()

klp_find_object_module() is called from both the klp register and enable
paths.  Only the call from the register path is necessary because the
module notifier will let us know if the patched module gets loaded or
unloaded.

Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Reviewed-by: Petr Mladek <pmladek@suse.cz>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
9 years agolivepatch: fix RCU usage in klp_find_external_symbol()
Peter Zijlstra [Sat, 28 Feb 2015 21:24:48 +0000 (22:24 +0100)]
livepatch: fix RCU usage in klp_find_external_symbol()

While one must hold RCU-sched (aka. preempt_disable) for find_symbol()
one must equally hold it over the use of the object returned.

The moment you release the RCU-sched read lock, the object can be dead
and gone.

[jkosina@suse.cz: change subject line to be aligned with other patches]
Cc: Seth Jennings <sjenning@redhat.com>
Cc: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Miroslav Benes <mbenes@suse.cz>
Cc: Petr Mladek <pmladek@suse.cz>
Cc: Jiri Kosina <jkosina@suse.cz>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Acked-by: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
9 years agolivepatch: RCU protect struct klp_func all the time when used in klp_ftrace_handler()
Petr Mladek [Wed, 18 Feb 2015 17:02:13 +0000 (18:02 +0100)]
livepatch: RCU protect struct klp_func all the time when used in klp_ftrace_handler()

func->new_func has been accessed after rcu_read_unlock() in klp_ftrace_handler()
and therefore the access was not protected.

Signed-off-by: Petr Mladek <pmladek@suse.cz>
Acked-by: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
9 years agolivepatch: simplify disable error path
Josh Poimboeuf [Mon, 9 Feb 2015 17:31:13 +0000 (11:31 -0600)]
livepatch: simplify disable error path

If registering the function with ftrace has previously succeeded,
unregistering will almost never fail.  Even if it does, it's not a fatal
error.  We can still carry on and disable the klp_func from being used
by removing it from the klp_ops func stack.

Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Reviewed-by: Miroslav Benes <mbenes@suse.cz>
Reviewed-by: Petr Mladek <pmladek@suse.cz>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
9 years agolivepatch: remove extern specifier from header files
Miroslav Benes [Wed, 18 Feb 2015 14:21:07 +0000 (15:21 +0100)]
livepatch: remove extern specifier from header files

Storage-class specifier 'extern' is redundant in front of the function
declaration. According to the C specification it has the same meaning as
if not present at all. So remove it.

Signed-off-by: Miroslav Benes <mbenes@suse.cz>
Acked-by: Josh Poimboeuf <jpoimboe@redhat.com>
Reviewed-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
9 years agolivepatch: fix format string in kobject_init_and_add()
Jiri Kosina [Sun, 15 Feb 2015 09:03:20 +0000 (10:03 +0100)]
livepatch: fix format string in kobject_init_and_add()

kobject_init_and_add() takes expects format string for a name, so we
better provide it in order to avoid infoleaks if modules craft their
mod->name in a special way.

Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Reported-by: Kees Cook <keescook@chromium.org>
Acked-by: Seth Jennings <sjenning@redhat.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/livep...
Linus Torvalds [Wed, 11 Feb 2015 02:35:40 +0000 (18:35 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jikos/livepatching

Pull live patching infrastructure from Jiri Kosina:
 "Let me provide a bit of history first, before describing what is in
  this pile.

  Originally, there was kSplice as a standalone project that implemented
  stop_machine()-based patching for the linux kernel.  This project got
  later acquired, and the current owner is providing live patching as a
  proprietary service, without any intentions to have their
  implementation merged.

  Then, due to rising user/customer demand, both Red Hat and SUSE
  started working on their own implementation (not knowing about each
  other), and announced first versions roughly at the same time [1] [2].

  The principle difference between the two solutions is how they are
  making sure that the patching is performed in a consistent way when it
  comes to different execution threads with respect to the semantic
  nature of the change that is being introduced.

  In a nutshell, kPatch is issuing stop_machine(), then looking at
  stacks of all existing processess, and if it decides that the system
  is in a state that can be patched safely, it proceeds insterting code
  redirection machinery to the patched functions.

  On the other hand, kGraft provides a per-thread consistency during one
  single pass of a process through the kernel and performs a lazy
  contignuous migration of threads from "unpatched" universe to the
  "patched" one at safe checkpoints.

  If interested in a more detailed discussion about the consistency
  models and its possible combinations, please see the thread that
  evolved around [3].

  It pretty quickly became obvious to the interested parties that it's
  absolutely impractical in this case to have several isolated solutions
  for one task to co-exist in the kernel.  During a dedicated Live
  Kernel Patching track at LPC in Dusseldorf, all the interested parties
  sat together and came up with a joint aproach that would work for both
  distro vendors.  Steven Rostedt took notes [4] from this meeting.

  And the foundation for that aproach is what's present in this pull
  request.

  It provides a basic infrastructure for function "live patching" (i.e.
  code redirection), including API for kernel modules containing the
  actual patches, and API/ABI for userspace to be able to operate on the
  patches (look up what patches are applied, enable/disable them, etc).

  It's relatively simple and minimalistic, as it's making use of
  existing kernel infrastructure (namely ftrace) as much as possible.
  It's also self-contained, in a sense that it doesn't hook itself in
  any other kernel subsystem (it doesn't even touch any other code).
  It's now implemented for x86 only as a reference architecture, but
  support for powerpc, s390 and arm is already in the works (adding
  arch-specific support basically boils down to teaching ftrace about
  regs-saving).

  Once this common infrastructure gets merged, both Red Hat and SUSE
  have agreed to immediately start porting their current solutions on
  top of this, abandoning their out-of-tree code.  The plan basically is
  that each patch will be marked by flag(s) that would indicate which
  consistency model it is willing to use (again, the details have been
  sketched out already in the thread at [3]).

  Before this happens, the current codebase can be used to patch a large
  group of secruity/stability problems the patches for which are not too
  complex (in a sense that they don't introduce non-trivial change of
  function's return value semantics, they don't change layout of data
  structures, etc) -- this corresponds to LEAVE_FUNCTION &&
  SWITCH_FUNCTION semantics described at [3].

  This tree has been in linux-next since December.

    [1] https://lkml.org/lkml/2014/4/30/477
    [2] https://lkml.org/lkml/2014/7/14/857
    [3] https://lkml.org/lkml/2014/11/7/354
    [4] http://linuxplumbersconf.org/2014/wp-content/uploads/2014/10/LPC2014_LivePatching.txt

  [ The core code is introduced by the three commits authored by Seth
    Jennings, which got a lot of changes incorporated during numerous
    respins and reviews of the initial implementation.  All the followup
    commits have materialized only after public tree has been created,
    so they were not folded into initial three commits so that the
    public tree doesn't get rebased ]"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/livepatching:
  livepatch: add missing newline to error message
  livepatch: rename config to CONFIG_LIVEPATCH
  livepatch: fix uninitialized return value
  livepatch: support for repatching a function
  livepatch: enforce patch stacking semantics
  livepatch: change ARCH_HAVE_LIVE_PATCHING to HAVE_LIVE_PATCHING
  livepatch: fix deferred module patching order
  livepatch: handle ancient compilers with more grace
  livepatch: kconfig: use bool instead of boolean
  livepatch: samples: fix usage example comments
  livepatch: MAINTAINERS: add git tree location
  livepatch: use FTRACE_OPS_FL_IPMODIFY
  livepatch: move x86 specific ftrace handler code to arch/x86
  livepatch: samples: add sample live patching module
  livepatch: kernel: add support for live patching
  livepatch: kernel: add TAINT_LIVEPATCH

9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid
Linus Torvalds [Wed, 11 Feb 2015 02:29:38 +0000 (18:29 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jikos/hid

Pull HID updates from Jiri Kosina:
 "Updates for HID code

   - improveements of Logitech HID++ procotol implementation, from
     Benjamin Tissoires

   - support for composite RMI devices, from Andrew Duggan

   - new driver for BETOP controller, from Huang Bo

   - fixup for conflicting mapping in HID core between PC-101/103/104
     and PC-102/105 keyboards from David Herrmann

   - new hardware support and fixes in Wacom driver, from Ping Cheng

   - assorted small fixes and device ID additions all over the place"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid: (33 commits)
  HID: wacom: add support for Cintiq 27QHD and 27QHD touch
  HID: wacom: consolidate input capability settings for pen and touch
  HID: wacom: make sure touch arbitration is applied consistently
  HID: pidff: Fix initialisation forMicrosoft Sidewinder FF Pro 2
  HID: hyperv: match wait_for_completion_timeout return type
  HID: wacom: Report ABS_MISC event for Cintiq Companion Hybrid
  HID: Use Kbuild idiom in Makefiles
  HID: do not bind to Microchip Pick16F1454
  HID: hid-lg4ff: use DEVICE_ATTR_RW macro
  HID: hid-lg4ff: fix sysfs attribute permission
  HID: wacom: peport In Range event according to the spec
  HID: wacom: process invalid Cintiq and Intuos data in wacom_intuos_inout()
  HID: rmi: Add support for the touchpad in the Razer Blade 14 laptop
  HID: rmi: Support touchpads with external buttons
  HID: rmi: Use hid_report_len to compute the size of reports
  HID: logitech-hidpp: store the name of the device in struct hidpp
  HID: microsoft: add support for Japanese Surface Type Cover 3
  HID: fixup the conflicting keyboard mappings quirk
  HID: apple: fix battery support for the 2009 ANSI wireless keyboard
  HID: fix Kconfig text
  ...

9 years agosata_dwc_460ex: disable COMPILE_TEST again
Linus Torvalds [Wed, 11 Feb 2015 02:24:57 +0000 (18:24 -0800)]
sata_dwc_460ex: disable COMPILE_TEST again

Commit 84683a7e081f ("sata_dwc_460ex: enable COMPILE_TEST for the
driver") enabled this driver for non-ppc460-ex platforms, but it was
then disabled for ARM and ARM64 by commit 2de5a9c004e9 ("sata_dwc_460ex:
disable compilation on ARM and ARM64") because it's too noisy and
broken.

This disabled is entirely, because it's too noisy on x86-64 too, and
there's no point in disabling architectures one by one.  At a minimum,
the code isn't 64-bit clean, and even on 32-bit it is questionable
whether it makes sense.

Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoMerge branch 'akpm' (patches from Andrew)
Linus Torvalds [Wed, 11 Feb 2015 00:45:56 +0000 (16:45 -0800)]
Merge branch 'akpm' (patches from Andrew)

Merge misc updates from Andrew Morton:
 "Bite-sized chunks this time, to avoid the MTA ratelimiting woes.

   - fs/notify updates

   - ocfs2

   - some of MM"

That laconic "some MM" is mainly the removal of remap_file_pages(),
which is a big simplification of the VM, and which gets rid of a *lot*
of random cruft and special cases because we no longer support the
non-linear mappings that it used.

From a user interface perspective, nothing has changed, because the
remap_file_pages() syscall still exists, it's just done by emulating the
old behavior by creating a lot of individual small mappings instead of
one non-linear one.

The emulation is slower than the old "native" non-linear mappings, but
nobody really uses or cares about remap_file_pages(), and simplifying
the VM is a big advantage.

* emailed patches from Andrew Morton <akpm@linux-foundation.org>: (78 commits)
  memcg: zap memcg_slab_caches and memcg_slab_mutex
  memcg: zap memcg_name argument of memcg_create_kmem_cache
  memcg: zap __memcg_{charge,uncharge}_slab
  mm/page_alloc.c: place zone_id check before VM_BUG_ON_PAGE check
  mm: hugetlb: fix type of hugetlb_treat_as_movable variable
  mm, hugetlb: remove unnecessary lower bound on sysctl handlers"?
  mm: memory: merge shared-writable dirtying branches in do_wp_page()
  mm: memory: remove ->vm_file check on shared writable vmas
  xtensa: drop _PAGE_FILE and pte_file()-related helpers
  x86: drop _PAGE_FILE and pte_file()-related helpers
  unicore32: drop pte_file()-related helpers
  um: drop _PAGE_FILE and pte_file()-related helpers
  tile: drop pte_file()-related helpers
  sparc: drop pte_file()-related helpers
  sh: drop _PAGE_FILE and pte_file()-related helpers
  score: drop _PAGE_FILE and pte_file()-related helpers
  s390: drop pte_file()-related helpers
  parisc: drop _PAGE_FILE and pte_file()-related helpers
  openrisc: drop _PAGE_FILE and pte_file()-related helpers
  nios2: drop _PAGE_FILE and pte_file()-related helpers
  ...

9 years agoMerge tag 'gfs2-merge-window' of git://git.kernel.org/pub/scm/linux/kernel/git/steve...
Linus Torvalds [Wed, 11 Feb 2015 00:20:49 +0000 (16:20 -0800)]
Merge tag 'gfs2-merge-window' of git://git./linux/kernel/git/steve/gfs2-3.0-nmw

Pull gfs2 updates from Steven Whitehouse:
 "This time we have mostly clean ups.  There is a bug fix for a NULL
  dereference relating to ACLs, and another which improves (but does not
  fix entirely) an allocation fall-back code path.  The other three
  patches are small clean ups"

* tag 'gfs2-merge-window' of git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-3.0-nmw:
  GFS2: Fix crash during ACL deletion in acl max entry check in gfs2_set_acl()
  GFS2: use __vmalloc GFP_NOFS for fs-related allocations.
  GFS2: Eliminate a nonsense goto
  GFS2: fix sprintf format specifier
  GFS2: Eliminate __gfs2_glock_remove_from_lru

9 years agoMerge tag 'xfs-for-linus-3.20-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 11 Feb 2015 00:15:17 +0000 (16:15 -0800)]
Merge tag 'xfs-for-linus-3.20-rc1' of git://git./linux/kernel/git/dgc/linux-xfs

Pull xfs update from Dave Chinner:
 "This update contains:

   - RENAME_EXCHANGE support

   - Rework of the superblock logging infrastructure

   - Rework of the XFS_IOCTL_SETXATTR implementation
       * enables use inside user namespaces
       * fixes inconsistencies setting extent size hints

   - fixes for missing buffer type annotations used in log recovery

   - more consolidation of libxfs headers

   - preparation patches for block based PNFS support

   - miscellaneous bug fixes and cleanups"

* tag 'xfs-for-linus-3.20-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/dgc/linux-xfs: (37 commits)
  xfs: only trace buffer items if they exist
  xfs: report proper f_files in statfs if we overshoot imaxpct
  xfs: fix panic_mask documentation
  xfs: xfs_ioctl_setattr_check_projid can be static
  xfs: growfs should use synchronous transactions
  xfs: fix behaviour of XFS_IOC_FSSETXATTR on directories
  xfs: factor projid hint checking out of xfs_ioctl_setattr
  xfs: factor extsize hint checking out of xfs_ioctl_setattr
  xfs: XFS_IOCTL_SETXATTR can run in user namespaces
  xfs: kill xfs_ioctl_setattr behaviour mask
  xfs: disaggregate xfs_ioctl_setattr
  xfs: factor out xfs_ioctl_setattr transaciton preamble
  xfs: separate xflags from xfs_ioctl_setattr
  xfs: FSX_NONBLOCK is not used
  xfs: don't allocate an ioend for direct I/O completions
  xfs: change kmem_free to use generic kvfree()
  xfs: factor out a xfs_update_prealloc_flags() helper
  xfs: remove incorrect error negation in attr_multi ioctl
  xfs: set superblock buffer type correctly
  xfs: set buf types when converting extent formats
  ...

9 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs
Linus Torvalds [Tue, 10 Feb 2015 23:52:38 +0000 (15:52 -0800)]
Merge branch 'for_linus' of git://git./linux/kernel/git/jack/linux-fs

Pull quota interface unification and misc cleanups from Jan Kara:
 "The first part of the series unifying XFS and VFS quota interfaces.

  This part unifies turning quotas on and off so quota-tools and
  xfs_quota can be used to manage any filesystem.  This is useful so
  that userspace doesn't have to distinguish which filesystem it is
  working with.  As a result we can then easily reuse tests for project
  quotas in XFS for ext4.

  This also contains minor cleanups and fixes for udf, isofs, and ext3"

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs: (23 commits)
  udf: remove bool assignment to 0/1
  udf: use bool for done
  quota: Store maximum space limit in bytes
  quota: Remove quota_on_meta callback
  ocfs2: Use generic helpers for quotaon and quotaoff
  ext4: Use generic helpers for quotaon and quotaoff
  quota: Add ->quota_{enable,disable} callbacks for VFS quotas
  quota: Wire up ->quota_{enable,disable} callbacks into Q_QUOTA{ON,OFF}
  quota: Split ->set_xstate callback into two
  xfs: Remove some pointless quota checks
  xfs: Remove some useless flags tests
  xfs: Remove useless test
  quota: Verify flags passed to Q_SETINFO
  quota: Cleanup flags definitions
  ocfs2: Move OLQF_CLEAN flag out of generic quota flags
  quota: Don't store flags for v2 quota format
  jbd: drop jbd_ENOSYS debug
  udf: destroy sbi mutex in put_super
  udf: Check length of extended attributes and allocation descriptors
  udf: Remove repeated loads blocksize
  ...

9 years agoMerge tag 'locks-v3.20-1' of git://git.samba.org/jlayton/linux
Linus Torvalds [Tue, 10 Feb 2015 23:34:42 +0000 (15:34 -0800)]
Merge tag 'locks-v3.20-1' of git://git.samba.org/jlayton/linux

Pull file locking related changes #1 from Jeff Layton:
 "This patchset contains a fairly major overhaul of how file locks are
  tracked within the inode.  Rather than a single list, we now create a
  per-inode "lock context" that contains individual lists for the file
  locks, and a new dedicated spinlock for them.

  There are changes in other trees that are based on top of this set so
  it may be easiest to pull this in early"

* tag 'locks-v3.20-1' of git://git.samba.org/jlayton/linux:
  locks: update comments that refer to inode->i_flock
  locks: consolidate NULL i_flctx checks in locks_remove_file
  locks: keep a count of locks on the flctx lists
  locks: clean up the lm_change prototype
  locks: add a dedicated spinlock to protect i_flctx lists
  locks: remove i_flock field from struct inode
  locks: convert lease handling to file_lock_context
  locks: convert posix locks to file_lock_context
  locks: move flock locks to file_lock_context
  ceph: move spinlocking into ceph_encode_locks_to_buffer and ceph_count_locks
  locks: add a new struct file_locking_context pointer to struct inode
  locks: have locks_release_file use flock_lock_file to release generic flock locks
  locks: add new struct list_head to struct file_lock

9 years agoMerge tag 'pm+acpi-3.20-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Tue, 10 Feb 2015 23:09:41 +0000 (15:09 -0800)]
Merge tag 'pm+acpi-3.20-rc1' of git://git./linux/kernel/git/rafael/linux-pm

Pull ACPI and power management updates from Rafael Wysocki:
 "We have a few new features this time, including a new SFI-based
  cpufreq driver, a new devfreq driver for Tegra Activity Monitor, a new
  devfreq class for providing its governors with raw utilization data
  and a new ACPI driver for AMD SoCs.

  Still, the majority of changes here are reworks of existing code to
  make it more straightforward or to prepare it for implementing new
  features on top of it.  The primary example is the rework of ACPI
  resources handling from Jiang Liu, Thomas Gleixner and Lv Zheng with
  support for IOAPIC hotplug implemented on top of it, but there is
  quite a number of changes of this kind in the cpufreq core, ACPICA,
  ACPI EC driver, ACPI processor driver and the generic power domains
  core code too.

  The most active developer is Viresh Kumar with his cpufreq changes.

  Specifics:

   - Rework of the core ACPI resources parsing code to fix issues in it
     and make using resource offsets more convenient and consolidation
     of some resource-handing code in a couple of places that have grown
     analagous data structures and code to cover the the same gap in the
     core (Jiang Liu, Thomas Gleixner, Lv Zheng).

   - ACPI-based IOAPIC hotplug support on top of the resources handling
     rework (Jiang Liu, Yinghai Lu).

   - ACPICA update to upstream release 20150204 including an interrupt
     handling rework that allows drivers to install raw handlers for
     ACPI GPEs which then become entirely responsible for the given GPE
     and the ACPICA core code won't touch it (Lv Zheng, David E Box,
     Octavian Purdila).

   - ACPI EC driver rework to fix several concurrency issues and other
     problems related to events handling on top of the ACPICA's new
     support for raw GPE handlers (Lv Zheng).

   - New ACPI driver for AMD SoCs analogous to the LPSS (Low-Power
     Subsystem) driver for Intel chips (Ken Xue).

   - Two minor fixes of the ACPI LPSS driver (Heikki Krogerus, Jarkko
     Nikula).

   - Two new blacklist entries for machines (Samsung 730U3E/740U3E and
     510R) where the native backlight interface doesn't work correctly
     while the ACPI one does (Hans de Goede).

   - Rework of the ACPI processor driver's handling of idle states to
     make the code more straightforward and less bloated overall (Rafael
     J Wysocki).

   - Assorted minor fixes related to ACPI and SFI (Andreas Ruprecht,
     Andy Shevchenko, Hanjun Guo, Jan Beulich, Rafael J Wysocki, Yaowei
     Bai).

   - PCI core power management modification to avoid resuming (some)
     runtime-suspended devices during system suspend if they are in the
     right states already (Rafael J Wysocki).

   - New SFI-based cpufreq driver for Intel platforms using SFI
     (Srinidhi Kasagar).

   - cpufreq core fixes, cleanups and simplifications (Viresh Kumar,
     Doug Anderson, Wolfram Sang).

   - SkyLake CPU support and other updates for the intel_pstate driver
     (Kristen Carlson Accardi, Srinivas Pandruvada).

   - cpufreq-dt driver cleanup (Markus Elfring).

   - Init fix for the ARM big.LITTLE cpuidle driver (Sudeep Holla).

   - Generic power domains core code fixes and cleanups (Ulf Hansson).

   - Operating Performance Points (OPP) core code cleanups and kernel
     documentation update (Nishanth Menon).

   - New dabugfs interface to make the list of PM QoS constraints
     available to user space (Nishanth Menon).

   - New devfreq driver for Tegra Activity Monitor (Tomeu Vizoso).

   - New devfreq class (devfreq_event) to provide raw utilization data
     to devfreq governors (Chanwoo Choi).

   - Assorted minor fixes and cleanups related to power management
     (Andreas Ruprecht, Krzysztof Kozlowski, Rickard Strandqvist, Pavel
     Machek, Todd E Brandt, Wonhong Kwon).

   - turbostat updates (Len Brown) and cpupower Makefile improvement
     (Sriram Raghunathan)"

* tag 'pm+acpi-3.20-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (151 commits)
  tools/power turbostat: relax dependency on APERF_MSR
  tools/power turbostat: relax dependency on invariant TSC
  Merge branch 'pci/host-generic' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci into acpi-resources
  tools/power turbostat: decode MSR_*_PERF_LIMIT_REASONS
  tools/power turbostat: relax dependency on root permission
  ACPI / video: Add disable_native_backlight quirk for Samsung 510R
  ACPI / PM: Remove unneeded nested #ifdef
  USB / PM: Remove unneeded #ifdef and associated dead code
  intel_pstate: provide option to only use intel_pstate with HWP
  ACPI / EC: Add GPE reference counting debugging messages
  ACPI / EC: Add query flushing support
  ACPI / EC: Refine command storm prevention support
  ACPI / EC: Add command flushing support.
  ACPI / EC: Introduce STARTED/STOPPED flags to replace BLOCKED flag
  ACPI: add AMD ACPI2Platform device support for x86 system
  ACPI / table: remove duplicate NULL check for the handler of acpi_table_parse()
  ACPI / EC: Update revision due to raw handler mode.
  ACPI / EC: Reduce ec_poll() by referencing the last register access timestamp.
  ACPI / EC: Fix several GPE handling issues by deploying ACPI_GPE_DISPATCH_RAW_HANDLER mode.
  ACPICA: Events: Enable APIs to allow interrupt/polling adaptive request based GPE handling model
  ...

9 years agoMerge tag 'pci-v3.20-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaa...
Linus Torvalds [Tue, 10 Feb 2015 22:31:28 +0000 (14:31 -0800)]
Merge tag 'pci-v3.20-changes' of git://git./linux/kernel/git/helgaas/pci

Pull PCI changes from Bjorn Helgaas:
 "Enumeration
    - Move domain assignment from arm64 to generic code (Lorenzo Pieralisi)
    - ARM: Remove artificial dependency on pci_sys_data domain (Lorenzo Pieralisi)
    - ARM: Move to generic PCI domains (Lorenzo Pieralisi)
    - Generate uppercase hex for modalias var in uevent (Ricardo Ribalda Delgado)
    - Add and use generic config accessors on ARM, PowerPC (Rob Herring)

  Resource management
    - Free resources on failure in of_pci_get_host_bridge_resources() (Lorenzo Pieralisi)
    - Fix infinite loop with ROM image of size 0 (Michel Dänzer)

  PCI device hotplug
    - Handle surprise add even if surprise removal isn't supported (Bjorn Helgaas)

  Virtualization
    - Mark AMD/ATI VGA devices that don't reset on D3hot->D0 transition (Alex Williamson)
    - Add DMA alias quirk for Adaptec 3405 (Alex Williamson)
    - Add Wellsburg (X99) to Intel PCH root port ACS quirk (Alex Williamson)
    - Add ACS quirk for Emulex NICs (Vasundhara Volam)

  MSI
    - Fail MSI-X mappings if there's no space assigned to MSI-X BAR (Yijing Wang)

  Freescale Layerscape host bridge driver
    - Fix platform_no_drv_owner.cocci warnings (Julia Lawall)

  NVIDIA Tegra host bridge driver
    - Remove unnecessary tegra_pcie_fixup_bridge() (Lucas Stach)

  Renesas R-Car host bridge driver
    - Fix error handling of irq_of_parse_and_map() (Dmitry Torokhov)

  TI Keystone host bridge driver
    - Fix error handling of irq_of_parse_and_map() (Dmitry Torokhov)
    - Fix misspelling of current function in debug output (Julia Lawall)

  Xilinx AXI host bridge driver
    - Fix harmless format string warning (Arnd Bergmann)

  Miscellaneous
    - Use standard parsing functions for ASPM sysfs setters (Chris J Arges)
    - Add pci_device_to_OF_node() stub for !CONFIG_OF (Kevin Hao)
    - Delete unnecessary NULL pointer checks (Markus Elfring)
    - Add and use defines for PCIe Max_Read_Request_Size (RafaÅ‚ MiÅ‚ecki)
    - Include clk.h instead of clk-private.h (Stephen Boyd)"

* tag 'pci-v3.20-changes' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci: (48 commits)
  PCI: Add pci_device_to_OF_node() stub for !CONFIG_OF
  PCI: xilinx: Convert to use generic config accessors
  PCI: xgene: Convert to use generic config accessors
  PCI: tegra: Convert to use generic config accessors
  PCI: rcar: Convert to use generic config accessors
  PCI: generic: Convert to use generic config accessors
  powerpc/powermac: Convert PCI to use generic config accessors
  powerpc/fsl_pci: Convert PCI to use generic config accessors
  ARM: ks8695: Convert PCI to use generic config accessors
  ARM: sa1100: Convert PCI to use generic config accessors
  ARM: integrator: Convert PCI to use generic config accessors
  PCI: versatile: Add DT-based ARM Versatile PB PCIe host driver
  ARM: dts: versatile: add PCI controller binding
  of/pci: Free resources on failure in of_pci_get_host_bridge_resources()
  PCI: versatile: Add DT docs for ARM Versatile PB PCIe driver
  PCI: Fail MSI-X mappings if there's no space assigned to MSI-X BAR
  r8169: use PCI define for Max_Read_Request_Size
  [SCSI] esas2r: use PCI define for Max_Read_Request_Size
  tile: use PCI define for Max_Read_Request_Size
  rapidio/tsi721: use PCI define for Max_Read_Request_Size
  ...

9 years agomemcg: zap memcg_slab_caches and memcg_slab_mutex
Vladimir Davydov [Tue, 10 Feb 2015 22:11:47 +0000 (14:11 -0800)]
memcg: zap memcg_slab_caches and memcg_slab_mutex

mem_cgroup->memcg_slab_caches is a list of kmem caches corresponding to
the given cgroup.  Currently, it is only used on css free in order to
destroy all caches corresponding to the memory cgroup being freed.  The
list is protected by memcg_slab_mutex.  The mutex is also used to protect
kmem_cache->memcg_params->memcg_caches arrays and synchronizes
kmem_cache_destroy vs memcg_unregister_all_caches.

However, we can perfectly get on without these two.  To destroy all caches
corresponding to a memory cgroup, we can walk over the global list of kmem
caches, slab_caches, and we can do all the synchronization stuff using the
slab_mutex instead of the memcg_slab_mutex.  This patch therefore gets rid
of the memcg_slab_caches and memcg_slab_mutex.

Apart from this nice cleanup, it also:

 - assures that rcu_barrier() is called once at max when a root cache is
   destroyed or a memory cgroup is freed, no matter how many caches have
   SLAB_DESTROY_BY_RCU flag set;

 - fixes the race between kmem_cache_destroy and kmem_cache_create that
   exists, because memcg_cleanup_cache_params, which is called from
   kmem_cache_destroy after checking that kmem_cache->refcount=0,
   releases the slab_mutex, which gives kmem_cache_create a chance to
   make an alias to a cache doomed to be destroyed.

Signed-off-by: Vladimir Davydov <vdavydov@parallels.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@suse.cz>
Acked-by: Christoph Lameter <cl@linux.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomemcg: zap memcg_name argument of memcg_create_kmem_cache
Vladimir Davydov [Tue, 10 Feb 2015 22:11:44 +0000 (14:11 -0800)]
memcg: zap memcg_name argument of memcg_create_kmem_cache

Instead of passing the name of the memory cgroup which the cache is
created for in the memcg_name_argument, let's obtain it immediately in
memcg_create_kmem_cache.

Signed-off-by: Vladimir Davydov <vdavydov@parallels.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomemcg: zap __memcg_{charge,uncharge}_slab
Vladimir Davydov [Tue, 10 Feb 2015 22:11:41 +0000 (14:11 -0800)]
memcg: zap __memcg_{charge,uncharge}_slab

They are simple wrappers around memcg_{charge,uncharge}_kmem, so let's
zap them and call these functions directly.

Signed-off-by: Vladimir Davydov <vdavydov@parallels.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm/page_alloc.c: place zone_id check before VM_BUG_ON_PAGE check
Weijie Yang [Tue, 10 Feb 2015 22:11:39 +0000 (14:11 -0800)]
mm/page_alloc.c: place zone_id check before VM_BUG_ON_PAGE check

If the freeing page and its buddy page are not at the same zone, the
current holding zone->lock for the freeing page cann't prevent buddy page
getting allocated, this could trigger VM_BUG_ON_PAGE in page_is_buddy() at
a very tiny chance, such as:

cpu 0: cpu 1:
hold zone_1 lock
check page and it buddy
PageBuddy(buddy) is true hold zone_2 lock
page_order(buddy) == order is true alloc buddy
trigger VM_BUG_ON_PAGE(page_count(buddy) != 0)

zone_1->lock prevents the freeing page getting allocated
zone_2->lock prevents the buddy page getting allocated
they are not the same zone->lock.

If we can't remove the zone_id check statement, it's better handle this
rare race.  This patch fixes this by placing the zone_id check before the
VM_BUG_ON_PAGE check.

Signed-off-by: Weijie Yang <weijie.yang@samsung.com>
Acked-by: Mel Gorman <mgorman@suse.de>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm: hugetlb: fix type of hugetlb_treat_as_movable variable
Andrey Ryabinin [Tue, 10 Feb 2015 22:11:36 +0000 (14:11 -0800)]
mm: hugetlb: fix type of hugetlb_treat_as_movable variable

hugetlb_treat_as_movable declared as unsigned long, but
proc_dointvec() used for parsing it:

static struct ctl_table vm_table[] = {
...
        {
                .procname = "hugepages_treat_as_movable",
                .data = &hugepages_treat_as_movable,
                .maxlen = sizeof(int),
                .mode = 0644,
                .proc_handler = proc_dointvec,
        },

This seems harmless, but it's better to use int type here.

Signed-off-by: Andrey Ryabinin <a.ryabinin@samsung.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Manfred Spraul <manfred@colorfullife.com>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm, hugetlb: remove unnecessary lower bound on sysctl handlers"?
Andrey Ryabinin [Tue, 10 Feb 2015 22:11:33 +0000 (14:11 -0800)]
mm, hugetlb: remove unnecessary lower bound on sysctl handlers"?

Commit ed4d4902ebdd ("mm, hugetlb: remove hugetlb_zero and
hugetlb_infinity") replaced 'unsigned long hugetlb_zero' with 'int zero'
leading to out-of-bounds access in proc_doulongvec_minmax().  Use
'.extra1 = NULL' instead of '.extra1 = &zero'.  Passing NULL is
equivalent to passing minimal value, which is 0 for unsigned types.

Fixes: ed4d4902ebdd ("mm, hugetlb: remove hugetlb_zero and hugetlb_infinity")
Signed-off-by: Andrey Ryabinin <a.ryabinin@samsung.com>
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Suggested-by: Manfred Spraul <manfred@colorfullife.com>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm: memory: merge shared-writable dirtying branches in do_wp_page()
Johannes Weiner [Tue, 10 Feb 2015 22:11:30 +0000 (14:11 -0800)]
mm: memory: merge shared-writable dirtying branches in do_wp_page()

Whether there is a vm_ops->page_mkwrite or not, the page dirtying is
pretty much the same.  Make sure the page references are the same in both
cases, then merge the two branches.

It's tempting to go even further and page-lock the !page_mkwrite case, to
get it in line with everybody else setting the page table and thus further
simplify the model.  But that's not quite compelling enough to justify
dropping the pte lock, then relocking and verifying the entry for
filesystems without ->page_mkwrite, which notably includes tmpfs.  Leave
it for now and lock the page late in the !page_mkwrite case.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm: memory: remove ->vm_file check on shared writable vmas
Johannes Weiner [Tue, 10 Feb 2015 22:11:28 +0000 (14:11 -0800)]
mm: memory: remove ->vm_file check on shared writable vmas

Shared anonymous mmaps are implemented with shmem files, so all VMAs with
shared writable semantics also have an underlying backing file.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Reviewed-by: Jan Kara <jack@suse.cz>
Acked-by: 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>
9 years agoxtensa: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:11:25 +0000 (14:11 -0800)]
xtensa: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agox86: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:11:22 +0000 (14:11 -0800)]
x86: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agounicore32: drop pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:11:20 +0000 (14:11 -0800)]
unicore32: drop pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Guan Xuetao <gxt@mprc.pku.edu.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoum: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:11:17 +0000 (14:11 -0800)]
um: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Jeff Dike <jdike@addtoit.com>
Cc: Richard Weinberger <richard@nod.at>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agotile: drop pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:11:14 +0000 (14:11 -0800)]
tile: drop pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Chris Metcalf <cmetcalf@ezchip.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agosparc: drop pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:11:12 +0000 (14:11 -0800)]
sparc: drop pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

This patch also increase number of bits availble for swap offset.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agosh: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:11:09 +0000 (14:11 -0800)]
sh: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: 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>
9 years agoscore: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:11:06 +0000 (14:11 -0800)]
score: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.
Nobody creates non-linear mapping anymore.

This patch also increase number of bits availble for swap offset.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Chen Liqin <liqin.linux@gmail.com>
Cc: Lennox Wu <lennox.wu@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agos390: drop pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:11:04 +0000 (14:11 -0800)]
s390: drop pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoparisc: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:11:01 +0000 (14:11 -0800)]
parisc: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: "James E.J. Bottomley" <jejb@parisc-linux.org>
Cc: Helge Deller <deller@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoopenrisc: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:58 +0000 (14:10 -0800)]
openrisc: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Jonas Bonn <jonas@southpole.se>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agonios2: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:56 +0000 (14:10 -0800)]
nios2: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Ley Foon Tan <lftan@altera.com>
Reviewed-by: Tobias Klauser <tklauser@distanz.ch>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomn10300: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:53 +0000 (14:10 -0800)]
mn10300: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

This patch also increases the number of bits availble for swap offset.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Koichi Yasutake <yasutake.koichi@jp.panasonic.com>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomips: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:50 +0000 (14:10 -0800)]
mips: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomicroblaze: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:47 +0000 (14:10 -0800)]
microblaze: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Michal Simek <monstr@monstr.eu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agometag: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:45 +0000 (14:10 -0800)]
metag: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: James Hogan <james.hogan@imgtec.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agom68k: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:41 +0000 (14:10 -0800)]
m68k: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agom32r: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:39 +0000 (14:10 -0800)]
m32r: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: 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>
9 years agoia64: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:36 +0000 (14:10 -0800)]
ia64: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

This patch also increase number of bits availble for swap offset.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agohexagon: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:33 +0000 (14:10 -0800)]
hexagon: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

This patch also increase number of bits availble for swap offset.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Richard Kuo <rkuo@codeaurora.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agofrv: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:31 +0000 (14:10 -0800)]
frv: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

This patch also increase number of bits availble for swap offset.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agocris: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:28 +0000 (14:10 -0800)]
cris: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Mikael Starvik <starvik@axis.com>
Cc: Jesper Nilsson <jesper.nilsson@axis.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoc6x: drop pte_file()
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:25 +0000 (14:10 -0800)]
c6x: drop pte_file()

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Mark Salter <msalter@redhat.com>
Cc: Aurelien Jacquiot <a-jacquiot@ti.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoblackfin: drop pte_file()
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:23 +0000 (14:10 -0800)]
blackfin: drop pte_file()

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Steven Miao <realmz6@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoavr32: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:20 +0000 (14:10 -0800)]
avr32: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Haavard Skinnemoen <hskinnemoen@gmail.com>
Acked-by: Hans-Christian Egtvedt <egtvedt@samfundet.no>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoarm: drop L_PTE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:17 +0000 (14:10 -0800)]
arm: drop L_PTE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

This patch also adjust __SWP_TYPE_SHIFT, effectively increase size of
possible swap file to 128G.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Russell King <linux@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoarm64: drop PTE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:15 +0000 (14:10 -0800)]
arm64: drop PTE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

This patch also adjust __SWP_TYPE_SHIFT and increase number of bits
availble for swap offset.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoarc: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:12 +0000 (14:10 -0800)]
arc: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Vineet Gupta <vgupta@synopsys.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoalpha: drop _PAGE_FILE and pte_file()-related helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:09 +0000 (14:10 -0800)]
alpha: drop _PAGE_FILE and pte_file()-related helpers

We've replaced remap_file_pages(2) implementation with emulation.  Nobody
creates non-linear mapping anymore.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
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>
9 years agoasm-generic: drop unused pte_file* helpers
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:07 +0000 (14:10 -0800)]
asm-generic: drop unused pte_file* helpers

All users are gone.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm: remove rest usage of VM_NONLINEAR and pte_file()
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:04 +0000 (14:10 -0800)]
mm: remove rest usage of VM_NONLINEAR and pte_file()

One bit in ->vm_flags is unused now!

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Michal Hocko <mhocko@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm: replace vma->sharead.linear with vma->shared
Kirill A. Shutemov [Tue, 10 Feb 2015 22:10:02 +0000 (14:10 -0800)]
mm: replace vma->sharead.linear with vma->shared

After removing vma->shared.nonlinear we have only one member of
vma->shared union, which doesn't make much sense.

This patch drops the union and move struct vma->shared.linear to
vma->shared.

Signed-off-by: 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>
9 years agormap: drop support of non-linear mappings
Kirill A. Shutemov [Tue, 10 Feb 2015 22:09:59 +0000 (14:09 -0800)]
rmap: drop support of non-linear mappings

We don't create non-linear mappings anymore.  Let's drop code which
handles them in rmap.

Signed-off-by: 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>
9 years agoproc: drop handling non-linear mappings
Kirill A. Shutemov [Tue, 10 Feb 2015 22:09:57 +0000 (14:09 -0800)]
proc: drop handling non-linear mappings

We have to handle non-linear mappings for /proc/PID/{smaps,clear_refs}
which is unused now.  Let's drop it.

Signed-off-by: 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>
9 years agomm: drop vm_ops->remap_pages and generic_file_remap_pages() stub
Kirill A. Shutemov [Tue, 10 Feb 2015 22:09:54 +0000 (14:09 -0800)]
mm: drop vm_ops->remap_pages and generic_file_remap_pages() stub

Nobody uses it anymore.

[akpm@linux-foundation.org: fix filemap_xip.c]
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm: drop support of non-linear mapping from fault codepath
Kirill A. Shutemov [Tue, 10 Feb 2015 22:09:51 +0000 (14:09 -0800)]
mm: drop support of non-linear mapping from fault codepath

We don't create non-linear mappings anymore.  Let's drop code which
handles them on page fault.

Signed-off-by: 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>
9 years agomm: drop support of non-linear mapping from unmap/zap codepath
Kirill A. Shutemov [Tue, 10 Feb 2015 22:09:49 +0000 (14:09 -0800)]
mm: drop support of non-linear mapping from unmap/zap codepath

We have remap_file_pages(2) emulation in -mm tree for few release cycles
and we plan to have it mainline in v3.20. This patchset removes rest of
VM_NONLINEAR infrastructure.

Patches 1-8 take care about generic code. They are pretty
straight-forward and can be applied without other of patches.

Rest patches removes pte_file()-related stuff from architecture-specific
code. It usually frees up one bit in non-present pte. I've tried to reuse
that bit for swap offset, where I was able to figure out how to do that.

For obvious reason I cannot test all that arch-specific code and would
like to see acks from maintainers.

In total, remap_file_pages(2) required about 1.4K lines of not-so-trivial
kernel code. That's too much for functionality nobody uses.

Tested-by: Felipe Balbi <balbi@ti.com>
This patch (of 38):

We don't create non-linear mappings anymore. Let's drop code which
handles them on unmap/zap.

Signed-off-by: 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>
9 years agomm: replace remap_file_pages() syscall with emulation
Kirill A. Shutemov [Tue, 10 Feb 2015 22:09:46 +0000 (14:09 -0800)]
mm: replace remap_file_pages() syscall with emulation

remap_file_pages(2) was invented to be able efficiently map parts of
huge file into limited 32-bit virtual address space such as in database
workloads.

Nonlinear mappings are pain to support and it seems there's no
legitimate use-cases nowadays since 64-bit systems are widely available.

Let's drop it and get rid of all these special-cased code.

The patch replaces the syscall with emulation which creates new VMA on
each remap_file_pages(), unless they it can be merged with an adjacent
one.

I didn't find *any* real code that uses remap_file_pages(2) to test
emulation impact on.  I've checked Debian code search and source of all
packages in ALT Linux.  No real users: libc wrappers, mentions in
strace, gdb, valgrind and this kind of stuff.

There are few basic tests in LTP for the syscall.  They work just fine
with emulation.

To test performance impact, I've written small test case which
demonstrate pretty much worst case scenario: map 4G shmfs file, write to
begin of every page pgoff of the page, remap pages in reverse order,
read every page.

The test creates 1 million of VMAs if emulation is in use, so I had to
set vm.max_map_count to 1100000 to avoid -ENOMEM.

Before: 23.3 ( +-  4.31% ) seconds
After: 43.9 ( +-  0.85% ) seconds
Slowdown: 1.88x

I believe we can live with that.

Test case:

        #define _GNU_SOURCE
        #include <assert.h>
        #include <stdlib.h>
        #include <stdio.h>
        #include <sys/mman.h>

        #define MB (1024UL * 1024)
        #define SIZE (4096 * MB)

        int main(int argc, char **argv)
        {
                unsigned long *p;
                long i, pass;

                for (pass = 0; pass < 10; pass++) {
                        p = mmap(NULL, SIZE, PROT_READ|PROT_WRITE,
                                        MAP_SHARED | MAP_ANONYMOUS, -1, 0);
                        if (p == MAP_FAILED) {
                                perror("mmap");
                                return -1;
                        }

                        for (i = 0; i < SIZE / 4096; i++)
                                p[i * 4096 / sizeof(*p)] = i;

                        for (i = 0; i < SIZE / 4096; i++) {
                                if (remap_file_pages(p + i * 4096 / sizeof(*p), 4096,
                                                0, (SIZE - 4096 * (i + 1)) >> 12, 0)) {
                                        perror("remap_file_pages");
                                        return -1;
                                }
                        }

                        for (i = SIZE / 4096 - 1; i >= 0; i--)
                                assert(p[i * 4096 / sizeof(*p)] == SIZE / 4096 - i - 1);

                        munmap(p, SIZE);
                }

                return 0;
        }

[akpm@linux-foundation.org: fix spello]
[sasha.levin@oracle.com: initialize populate before usage]
[sasha.levin@oracle.com: grab file ref to prevent race while mmaping]
Signed-off-by: "Kirill A. Shutemov" <kirill@shutemov.name>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Dave Jones <davej@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Armin Rigo <arigo@tunes.org>
Signed-off-by: Sasha Levin <sasha.levin@oracle.com>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm/vmstat.c: fix/cleanup ifdefs
Andrew Morton [Tue, 10 Feb 2015 22:09:43 +0000 (14:09 -0800)]
mm/vmstat.c: fix/cleanup ifdefs

CONFIG_COMPACTION=y, CONFIG_DEBUG_FS=n:

  mm/vmstat.c:690: warning: 'frag_start' defined but not used
  mm/vmstat.c:702: warning: 'frag_next' defined but not used
  mm/vmstat.c:710: warning: 'frag_stop' defined but not used
  mm/vmstat.c:715: warning: 'walk_zones_in_node' defined but not used

It's all a bit of a tangly mess and it's unclear why CONFIG_COMPACTION
figures in there at all.  Move frag_start/frag_next/frag_stop and
migratetype_names[] into the existing CONFIG_PROC_FS block.

walk_zones_in_node() gets a special ifdef.

Also move the #include lines up to where #include lines live.

[axel.lin@ingics.com: fix build error when !CONFIG_PROC_FS]
Signed-off-by: Axel Lin <axel.lin@ingics.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Tested-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm/slab_common.c: use kmem_cache_free()
Vaishali Thakkar [Tue, 10 Feb 2015 22:09:40 +0000 (14:09 -0800)]
mm/slab_common.c: use kmem_cache_free()

Here, free memory is allocated using kmem_cache_zalloc.  So, use
kmem_cache_free instead of kfree.

This is done using Coccinelle and semantic patch used
is as follows:

@@
expression x,E,c;
@@

 x = \(kmem_cache_alloc\|kmem_cache_zalloc\|kmem_cache_alloc_node\)(c,...)
 ... when != x = E
     when != &x
?-kfree(x)
+kmem_cache_free(c,x)

Signed-off-by: Vaishali Thakkar <vthakkar1994@gmail.com>
Acked-by: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@kernel.org>
Acked-by: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm/slub.c: fix typo in comment
Kim Phillips [Tue, 10 Feb 2015 22:09:37 +0000 (14:09 -0800)]
mm/slub.c: fix typo in comment

Signed-off-by: Kim Phillips <kim.phillips@freescale.com>
Acked-by: Christoph Lameter <cl@linux.com>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm: don't use compound_head() in virt_to_head_page()
Joonsoo Kim [Tue, 10 Feb 2015 22:09:35 +0000 (14:09 -0800)]
mm: don't use compound_head() in virt_to_head_page()

compound_head() is implemented with assumption that there would be race
condition when checking tail flag.  This assumption is only true when we
try to access arbitrary positioned struct page.

The situation that virt_to_head_page() is called is different case.  We
call virt_to_head_page() only in the range of allocated pages, so there
is no race condition on tail flag.  In this case, we don't need to
handle race condition and we can reduce overhead slightly.  This patch
implements compound_head_fast() which is similar with compound_head()
except tail flag race handling.  And then, virt_to_head_page() uses this
optimized function to improve performance.

I saw 1.8% win in a fast-path loop over kmem_cache_alloc/free, (14.063
ns -> 13.810 ns) if target object is on tail page.

Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Acked-by: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agomm/slub: optimize alloc/free fastpath by removing preemption on/off
Joonsoo Kim [Tue, 10 Feb 2015 22:09:32 +0000 (14:09 -0800)]
mm/slub: optimize alloc/free fastpath by removing preemption on/off

We had to insert a preempt enable/disable in the fastpath a while ago in
order to guarantee that tid and kmem_cache_cpu are retrieved on the same
cpu.  It is the problem only for CONFIG_PREEMPT in which scheduler can
move the process to other cpu during retrieving data.

Now, I reach the solution to remove preempt enable/disable in the
fastpath.  If tid is matched with kmem_cache_cpu's tid after tid and
kmem_cache_cpu are retrieved by separate this_cpu operation, it means
that they are retrieved on the same cpu.  If not matched, we just have
to retry it.

With this guarantee, preemption enable/disable isn't need at all even if
CONFIG_PREEMPT, so this patch removes it.

I saw roughly 5% win in a fast-path loop over kmem_cache_alloc/free in
CONFIG_PREEMPT.  (14.821 ns -> 14.049 ns)

Below is the result of Christoph's slab_test reported by Jesper Dangaard
Brouer.

* Before

 Single thread testing
 =====================
 1. Kmalloc: Repeatedly allocate then free test
 10000 times kmalloc(8) -> 49 cycles kfree -> 62 cycles
 10000 times kmalloc(16) -> 48 cycles kfree -> 64 cycles
 10000 times kmalloc(32) -> 53 cycles kfree -> 70 cycles
 10000 times kmalloc(64) -> 64 cycles kfree -> 77 cycles
 10000 times kmalloc(128) -> 74 cycles kfree -> 84 cycles
 10000 times kmalloc(256) -> 84 cycles kfree -> 114 cycles
 10000 times kmalloc(512) -> 83 cycles kfree -> 116 cycles
 10000 times kmalloc(1024) -> 81 cycles kfree -> 120 cycles
 10000 times kmalloc(2048) -> 104 cycles kfree -> 136 cycles
 10000 times kmalloc(4096) -> 142 cycles kfree -> 165 cycles
 10000 times kmalloc(8192) -> 238 cycles kfree -> 226 cycles
 10000 times kmalloc(16384) -> 403 cycles kfree -> 264 cycles
 2. Kmalloc: alloc/free test
 10000 times kmalloc(8)/kfree -> 68 cycles
 10000 times kmalloc(16)/kfree -> 68 cycles
 10000 times kmalloc(32)/kfree -> 69 cycles
 10000 times kmalloc(64)/kfree -> 68 cycles
 10000 times kmalloc(128)/kfree -> 68 cycles
 10000 times kmalloc(256)/kfree -> 68 cycles
 10000 times kmalloc(512)/kfree -> 74 cycles
 10000 times kmalloc(1024)/kfree -> 75 cycles
 10000 times kmalloc(2048)/kfree -> 74 cycles
 10000 times kmalloc(4096)/kfree -> 74 cycles
 10000 times kmalloc(8192)/kfree -> 75 cycles
 10000 times kmalloc(16384)/kfree -> 510 cycles

* After

 Single thread testing
 =====================
 1. Kmalloc: Repeatedly allocate then free test
 10000 times kmalloc(8) -> 46 cycles kfree -> 61 cycles
 10000 times kmalloc(16) -> 46 cycles kfree -> 63 cycles
 10000 times kmalloc(32) -> 49 cycles kfree -> 69 cycles
 10000 times kmalloc(64) -> 57 cycles kfree -> 76 cycles
 10000 times kmalloc(128) -> 66 cycles kfree -> 83 cycles
 10000 times kmalloc(256) -> 84 cycles kfree -> 110 cycles
 10000 times kmalloc(512) -> 77 cycles kfree -> 114 cycles
 10000 times kmalloc(1024) -> 80 cycles kfree -> 116 cycles
 10000 times kmalloc(2048) -> 102 cycles kfree -> 131 cycles
 10000 times kmalloc(4096) -> 135 cycles kfree -> 163 cycles
 10000 times kmalloc(8192) -> 238 cycles kfree -> 218 cycles
 10000 times kmalloc(16384) -> 399 cycles kfree -> 262 cycles
 2. Kmalloc: alloc/free test
 10000 times kmalloc(8)/kfree -> 65 cycles
 10000 times kmalloc(16)/kfree -> 66 cycles
 10000 times kmalloc(32)/kfree -> 65 cycles
 10000 times kmalloc(64)/kfree -> 66 cycles
 10000 times kmalloc(128)/kfree -> 66 cycles
 10000 times kmalloc(256)/kfree -> 71 cycles
 10000 times kmalloc(512)/kfree -> 72 cycles
 10000 times kmalloc(1024)/kfree -> 71 cycles
 10000 times kmalloc(2048)/kfree -> 71 cycles
 10000 times kmalloc(4096)/kfree -> 71 cycles
 10000 times kmalloc(8192)/kfree -> 65 cycles
 10000 times kmalloc(16384)/kfree -> 511 cycles

Most of the results are better than before.

Note that this change slightly worses performance in !CONFIG_PREEMPT,
roughly 0.3%.  Implementing each case separately would help performance,
but, since it's so marginal, I didn't do that.  This would help
maintanance since we have same code for all cases.

Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Acked-by: Christoph Lameter <cl@linux.com>
Tested-by: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: Jesper Dangaard Brouer <brouer@redhat.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agofsioctl.c: make generic_block_fiemap() signal-tolerant
Dmitry Monakhov [Tue, 10 Feb 2015 22:09:29 +0000 (14:09 -0800)]
fsioctl.c: make generic_block_fiemap() signal-tolerant

__generic_block_fiemap may spin very long time for large sparse files.

Without this patch an unprivileged user may abuse system resources simply
by spawning a vast number of unkilable busyloops (works on ext2/ext3):

  truncate --size 1T test
  for ((i=0;i<1024;i++))
  do
         filefrag test > /dev/null &
  done

Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoo2dlm: fix NULL pointer dereference in o2dlm_blocking_ast_wrapper
Srinivas Eeda [Tue, 10 Feb 2015 22:09:26 +0000 (14:09 -0800)]
o2dlm: fix NULL pointer dereference in o2dlm_blocking_ast_wrapper

A tiny race between BAST and unlock message causes the NULL dereference.

A node sends an unlock request to master and receives a response.  Before
processing the response it receives a BAST from the master.  Since both
requests are processed by different threads it creates a race.  While the
BAST is being processed, lock can get freed by unlock code.

This patch makes bast to return immediately if lock is found but unlock is
pending.  The code should handle this race.  We also have to fix master
node to skip sending BAST after receiving unlock message.

Below is the crash stack

    BUG: unable to handle kernel NULL pointer dereference at 0000000000000048
    IP: o2dlm_blocking_ast_wrapper+0xd/0x16
    dlm_do_local_bast+0x8e/0x97 [ocfs2_dlm]
    dlm_proxy_ast_handler+0x838/0x87e [ocfs2_dlm]
    o2net_process_message+0x395/0x5b8 [ocfs2_nodemanager]
    o2net_rx_until_empty+0x762/0x90d [ocfs2_nodemanager]
    worker_thread+0x14d/0x1ed

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Srinivas Eeda <srinivas.eeda@oracle.com>
Reviewed-by: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Joseph Qi <joseph.qi@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoocfs2: prune the dcache before deleting the dentry of directory
alex chen [Tue, 10 Feb 2015 22:09:23 +0000 (14:09 -0800)]
ocfs2: prune the dcache before deleting the dentry of directory

In ocfs2_dentry_convert_worker, we should prune the dcache before deleting
the dentry of directory, otherwise, in the following cases the inode of
directory will still remain in orphan directory until the device being
umounted.

Mount point: /mnt/ocfs2
Node A                              Node B
mkdir /mnt/ocfs2/testdir
  ocfs2_mkdir
  ->ocfs2_mknod
  ->ocfs2_dentry_attach_lock
  ->ocfs2_dentry_lock(dentry, 0)
  ... ...
touch /mnt/ocfs2/testdir/testfile
                                    unlink /mnt/test/testdir/testfile
                                    rmdir /mnt/ocfs2/testdir
                                      ocfs2_unlink
                                      ->ocfs2_remote_dentry_delete
                                      ->ocfs2_dentry_lock(dentry, 1)
                                      ... ...
... ...
ocfs2_downconvert_thread
->ocfs2_unblock_lock
->ocfs2_dentry_convert_worker
->ocfs2_find_local_alias
  ->dget_dlock
->d_delete
Here the dentry can not be
released because the children's
dentry is negative but still exist.
Finally, this inode will still remain
in orphan directory until its children
are destroyed.

So before deleting dentry of directory, we should prune the dcache to
remove unused children of the parent dentry by shrink_dcache_parent().

Signed-off-by: Alex Chen <alex.chen@huawei.com>
Reviewed-by: Joseph Qi <joseph.qi@huawei.com>
Reviewed-by: joyce.xue <xuejiufei@huawei.com>
Reviewed-by: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoocfs2: make resv_lock spinlock static
Fabian Frederick [Tue, 10 Feb 2015 22:09:20 +0000 (14:09 -0800)]
ocfs2: make resv_lock spinlock static

resv_lock is only used in reservations.c

Signed-off-by: Fabian Frederick <fabf@skynet.be>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoocfs2: remove unreachable code in __ocfs2_recovery_thread()
Daeseok Youn [Tue, 10 Feb 2015 22:09:18 +0000 (14:09 -0800)]
ocfs2: remove unreachable code in __ocfs2_recovery_thread()

Signed-off-by: Daeseok Youn <daeseok.youn@gmail.com>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoocfs2: removes mlog_errno() call twice in ocfs2_find_dir_space_el()
Daeseok Youn [Tue, 10 Feb 2015 22:09:15 +0000 (14:09 -0800)]
ocfs2: removes mlog_errno() call twice in ocfs2_find_dir_space_el()

mlog_errno() is called twice when some functions are failed.

Signed-off-by: Daeseok Youn <daeseok.youn@gmail.com>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoocfs2: remove unreachable code
Daeseok Youn [Tue, 10 Feb 2015 22:09:12 +0000 (14:09 -0800)]
ocfs2: remove unreachable code

Signed-off-by: Daeseok Youn <daeseok.youn@gmail.com>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoocfs2: o2net: silence uninitialized variable warning
Dan Carpenter [Tue, 10 Feb 2015 22:09:10 +0000 (14:09 -0800)]
ocfs2: o2net: silence uninitialized variable warning

Smatch complains that, if o2net_tx_can_proceed() returns false, then "sc"
and "ret" are uninialized or maybe we are re-using the data from previous
iteration.  I do not know if we can hit this bug in real life but checking
the return value is harmless and we may as well silence the static checker
warning.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoocfs2: remove pointless assignment from ocfs2_calc_refcount_meta_credits()
Jan Kara [Tue, 10 Feb 2015 22:09:07 +0000 (14:09 -0800)]
ocfs2: remove pointless assignment from ocfs2_calc_refcount_meta_credits()

The assigned value is never used.

Coverity-id 1226847.

Signed-off-by: Jan Kara <jack@suse.cz>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoocfs2: add a mount option journal_async_commit on ocfs2 filesystem
alex chen [Tue, 10 Feb 2015 22:09:04 +0000 (14:09 -0800)]
ocfs2: add a mount option journal_async_commit on ocfs2 filesystem

Add a mount option to support JBD2 feature:

JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT.  When this feature is opened, journal
commit block can be written to disk without waiting for descriptor blocks,
which can improve journal commit performance.  This option will enable
'journal_checksum' internally.

Using the fs_mark benchmark, using journal_async_commit shows a 50%
improvement, the files per second go up from 215.2 to 317.5.

test script:
fs_mark  -d  /mnt/ocfs2/  -s  10240  -n  1000

default:
FSUse%        Count         Size    Files/sec     App Overhead
     0         1000        10240        215.2            17878

with journal_async_commit option:
FSUse%        Count         Size    Files/sec     App Overhead
     0         1000        10240        317.5            17881

Signed-off-by: Alex Chen <alex.chen@huawei.com>
Signed-off-by: Weiwei Wang <wangww631@huawei.comm>
Reviewed-by: Joseph Qi <joseph.qi@huawei.com>
Reviewed-by: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoocfs2: fix journal commit deadlock in ocfs2_convert_inline_data_to_extents
alex chen [Tue, 10 Feb 2015 22:09:02 +0000 (14:09 -0800)]
ocfs2: fix journal commit deadlock in ocfs2_convert_inline_data_to_extents

Similar to ocfs2_write_end_nolock() which is metioned at commit
136f49b91710 ("ocfs2: fix journal commit deadlock"), we should unlock
pages before ocfs2_commit_trans() in ocfs2_convert_inline_data_to_extents.

Otherwise, it will cause a deadlock with journal commit threads.

Signed-off-by: Alex Chen <alex.chen@huawei.com>
Reviewed-by: Joseph Qi <joseph.qi@huawei.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Mark Fasheh <mfasheh@suse.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoocfs2: dlm: dlmdomain: remove unused function
Rickard Strandqvist [Tue, 10 Feb 2015 22:08:59 +0000 (14:08 -0800)]
ocfs2: dlm: dlmdomain: remove unused function

Remove dlm_joined() that is not used anywhere.

This was partially found by using a static code analysis program called
cppcheck.

Signed-off-by: Rickard Strandqvist <rickard_strandqvist@spectrumdigital.se>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoocfs2: quota_local: remove unused function
Rickard Strandqvist [Tue, 10 Feb 2015 22:08:56 +0000 (14:08 -0800)]
ocfs2: quota_local: remove unused function

Remove ol_dqblk_file_block() that is not used anywhere.

This was partially found by using a static code analysis program called
cppcheck.

Signed-off-by: Rickard Strandqvist <rickard_strandqvist@spectrumdigital.se>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoocfs2: xattr: remove unused function
Rickard Strandqvist [Tue, 10 Feb 2015 22:08:54 +0000 (14:08 -0800)]
ocfs2: xattr: remove unused function

Remove ocfs2_xattr_bucket_get_val() that is not used anywhere.

This was partially found by using a static code analysis program called
cppcheck.

Signed-off-by: Rickard Strandqvist <rickard_strandqvist@spectrumdigital.se>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoocfs2: fix snprintf format specifier in dlmdebug.c
alex chen [Tue, 10 Feb 2015 22:08:51 +0000 (14:08 -0800)]
ocfs2: fix snprintf format specifier in dlmdebug.c

Use snprintf format specifier "%lu" instead of "%ld" for argument of type
'unsigned long'.

Signed-off-by: Alex Chen <alex.chen@huawei.com>
Reviewed-by: Joseph Qi <joseph.qi@huawei.com>
Reviewed-by: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoocfs2: fix wrong comment
Junxiao Bi [Tue, 10 Feb 2015 22:08:48 +0000 (14:08 -0800)]
ocfs2: fix wrong comment

O2NET_CONN_IDLE_DELAY is not defined, connection attempts will not be
canceled due to timeout.

Signed-off-by: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoocfs2: fix uninitialized variable access
Junxiao Bi [Tue, 10 Feb 2015 22:08:46 +0000 (14:08 -0800)]
ocfs2: fix uninitialized variable access

Variable "why" is not yet initialized at line 615, fix it.

Signed-off-by: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoocfs2: remove unnecessary else in ocfs2_set_acl()
Fabian Frederick [Tue, 10 Feb 2015 22:08:43 +0000 (14:08 -0800)]
ocfs2: remove unnecessary else in ocfs2_set_acl()

else is unnecessary after return.

Signed-off-by: Fabian Frederick <fabf@skynet.be>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoocfs2/dlm: add missing dlm_lock_put() when recovery master down
Xue jiufei [Tue, 10 Feb 2015 22:08:40 +0000 (14:08 -0800)]
ocfs2/dlm: add missing dlm_lock_put() when recovery master down

When the recovery master is down, the owner of $RECOVERY calls
dlm_do_local_recovery_cleanup() to prune any $RECOVERY entries for dead
nodes.  The lock is in the granted list and the refcount must be 2.  We
should put twice to remove this lock.  Otherwise, it will lead to a memory
leak.

Signed-off-by: joyce.xue <xuejiufei@huawei.com>
Reported-by: yangwenfang <vicky.yangwenfang@huawei.com>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agosh: eliminate unused irq_reg_{readl,writel} accessors
Kevin Cernekee [Tue, 10 Feb 2015 22:08:38 +0000 (14:08 -0800)]
sh: eliminate unused irq_reg_{readl,writel} accessors

Defining these macros way down in arch/sh/.../irq.c doesn't cause
kernel/irq/generic-chip.c to use them.  As far as I can tell this code has
no effect.

Fixes: 332fd7c4fef5f3b1 ("genirq: Generic chip: Change irq_reg_{readl,writel} arguments")
Signed-off-by: Kevin Cernekee <cernekee@gmail.com>
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> (cpp/asm comparison)
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Jason Cooper <jason@lakedaemon.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agosh: build superh without CONFIG_EXPERT
Rob Landley [Tue, 10 Feb 2015 22:08:35 +0000 (14:08 -0800)]
sh: build superh without CONFIG_EXPERT

What sh4 actually wants is HAVE_PATA_PLATFORM, so select that instead.

Signed-off-by: Rob Landley <rob@landley.net>
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Acked-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agofsnotify: fix handling of renames in audit
Jan Kara [Tue, 10 Feb 2015 22:08:32 +0000 (14:08 -0800)]
fsnotify: fix handling of renames in audit

Commit e9fd702a58c4 ("audit: convert audit watches to use fsnotify
instead of inotify") broke handling of renames in audit.  Audit code
wants to update inode number of an inode corresponding to watched name
in a directory.  When something gets renamed into a directory to a
watched name, inotify previously passed moved inode to audit code
however new fsnotify code passes directory inode where the change
happened.  That confuses audit and it starts watching parent directory
instead of a file in a directory.

This can be observed for example by doing:

  cd /tmp
  touch foo bar
  auditctl -w /tmp/foo
  touch foo
  mv bar foo
  touch foo

In audit log we see events like:

  type=CONFIG_CHANGE msg=audit(1423563584.155:90): auid=1000 ses=2 op="updated rules" path="/tmp/foo" key=(null) list=4 res=1
  ...
  type=PATH msg=audit(1423563584.155:91): item=2 name="bar" inode=1046884 dev=08:0 2 mode=0100644 ouid=0 ogid=0 rdev=00:00 nametype=DELETE
  type=PATH msg=audit(1423563584.155:91): item=3 name="foo" inode=1046842 dev=08:0 2 mode=0100644 ouid=0 ogid=0 rdev=00:00 nametype=DELETE
  type=PATH msg=audit(1423563584.155:91): item=4 name="foo" inode=1046884 dev=08:0 2 mode=0100644 ouid=0 ogid=0 rdev=00:00 nametype=CREATE
  ...

and that's it - we see event for the first touch after creating the
audit rule, we see events for rename but we don't see any event for the
last touch.  However we start seeing events for unrelated stuff
happening in /tmp.

Fix the problem by passing moved inode as data in the FS_MOVED_FROM and
FS_MOVED_TO events instead of the directory where the change happens.
This doesn't introduce any new problems because noone besides
audit_watch.c cares about the passed value:

  fs/notify/fanotify/fanotify.c cares only about FSNOTIFY_EVENT_PATH events.
  fs/notify/dnotify/dnotify.c doesn't care about passed 'data' value at all.
  fs/notify/inotify/inotify_fsnotify.c uses 'data' only for FSNOTIFY_EVENT_PATH.
  kernel/audit_tree.c doesn't care about passed 'data' at all.
  kernel/audit_watch.c expects moved inode as 'data'.

Fixes: e9fd702a58c49db ("audit: convert audit watches to use fsnotify instead of inotify")
Signed-off-by: Jan Kara <jack@suse.cz>
Cc: Paul Moore <paul@paul-moore.com>
Cc: Eric Paris <eparis@redhat.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoinotify: update documentation to reflect code changes
Zhang Zhen [Tue, 10 Feb 2015 22:08:30 +0000 (14:08 -0800)]
inotify: update documentation to reflect code changes

The inotify interface has changed a lot.  The user interface was too
old, and the kernel interface was removed by Eric Paris in commit:
2dfc1ca inotify: remove inotify in kernel interface.

Signed-off-by: Zhang Zhen <zhenzhang.zhang@huawei.com>
Cc: Wang Kai <morgan.wang@huawei.com>
Cc: Eric Paris <eparis@parisplace.org>
Cc: Robert Love <robert.w.love@intel.com>
Cc: John McCutchan <john@johnmccutchan.com>
Cc: Heinrich Schuchardt <xypron.glpk@gmx.de>
Acked-by: Jan Kara <jack@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agofanotify: don't set FAN_ONDIR implicitly on a marks ignored mask
Lino Sanfilippo [Tue, 10 Feb 2015 22:08:27 +0000 (14:08 -0800)]
fanotify: don't set FAN_ONDIR implicitly on a marks ignored mask

Currently FAN_ONDIR is always set on a mark's ignored mask when the
event mask is extended without FAN_MARK_ONDIR being set.  This may
result in events for directories being ignored unexpectedly for call
sequences like

  fanotify_mark(fd, FAN_MARK_ADD, FAN_OPEN | FAN_ONDIR , AT_FDCWD, "dir");
  fanotify_mark(fd, FAN_MARK_ADD, FAN_CLOSE, AT_FDCWD, "dir");

Also FAN_MARK_ONDIR is only honored when adding events to a mark's mask,
but not for event removal.  Fix both issues by not setting FAN_ONDIR
implicitly on the ignore mask any more.  Instead treat FAN_ONDIR as any
other event flag and require FAN_MARK_ONDIR to be set by the user for
both event mask and ignore mask.  Furthermore take FAN_MARK_ONDIR into
account when set for event removal.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Lino Sanfilippo <LinoSanfilippo@gmx.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Eric Paris <eparis@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agofanotify: don't recalculate a marks mask if only the ignored mask changed
Lino Sanfilippo [Tue, 10 Feb 2015 22:08:24 +0000 (14:08 -0800)]
fanotify: don't recalculate a marks mask if only the ignored mask changed

If removing bits from a mark's ignored mask, the concerning
inodes/vfsmounts mask is not affected.  So don't recalculate it.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Lino Sanfilippo <LinoSanfilippo@gmx.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Eric Paris <eparis@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agofanotify: only destroy mark when both mask and ignored_mask are cleared
Lino Sanfilippo [Tue, 10 Feb 2015 22:08:21 +0000 (14:08 -0800)]
fanotify: only destroy mark when both mask and ignored_mask are cleared

In fanotify_mark_remove_from_mask() a mark is destroyed if only one of
both bitmasks (mask or ignored_mask) of a mark is cleared.  However the
other mask may still be set and contain information that should not be
lost.  So only destroy a mark if both masks are cleared.

Signed-off-by: Lino Sanfilippo <LinoSanfilippo@gmx.de>
Reviewed-by: Jan Kara <jack@suse.cz>
Cc: Eric Paris <eparis@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agohugetlb, x86: register 1G page size if we can allocate them at runtime
Kirill A. Shutemov [Tue, 10 Feb 2015 22:08:19 +0000 (14:08 -0800)]
hugetlb, x86: register 1G page size if we can allocate them at runtime

After commit 944d9fec8d7a ("hugetlb: add support for gigantic page
allocation at runtime") we can allocate 1G pages at runtime if CMA is
enabled.

Let's register 1G pages into hugetlb even if the user hasn't requested
them explicitly at boot time with hugepagesz=1G.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Reviewed-by: Luiz Capitulino <lcapitulino@redhat.com>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoMerge tag 'stable/for-linus-3.20-rc0-tag' of git://git.kernel.org/pub/scm/linux/kerne...
Linus Torvalds [Tue, 10 Feb 2015 21:56:56 +0000 (13:56 -0800)]
Merge tag 'stable/for-linus-3.20-rc0-tag' of git://git./linux/kernel/git/xen/tip

Pull xen features and fixes from David Vrabel:

 - Reworked handling for foreign (grant mapped) pages to simplify the
   code, enable a number of additional use cases and fix a number of
   long-standing bugs.

 - Prefer the TSC over the Xen PV clock when dom0 (and the TSC is
   stable).

 - Assorted other cleanup and minor bug fixes.

* tag 'stable/for-linus-3.20-rc0-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip: (25 commits)
  xen/manage: Fix USB interaction issues when resuming
  xenbus: Add proper handling of XS_ERROR from Xenbus for transactions.
  xen/gntdev: provide find_special_page VMA operation
  xen/gntdev: mark userspace PTEs as special on x86 PV guests
  xen-blkback: safely unmap grants in case they are still in use
  xen/gntdev: safely unmap grants in case they are still in use
  xen/gntdev: convert priv->lock to a mutex
  xen/grant-table: add a mechanism to safely unmap pages that are in use
  xen-netback: use foreign page information from the pages themselves
  xen: mark grant mapped pages as foreign
  xen/grant-table: add helpers for allocating pages
  x86/xen: require ballooned pages for grant maps
  xen: remove scratch frames for ballooned pages and m2p override
  xen/grant-table: pre-populate kernel unmap ops for xen_gnttab_unmap_refs()
  mm: add 'foreign' alias for the 'pinned' page flag
  mm: provide a find_special_page vma operation
  x86/xen: cleanup arch/x86/xen/mmu.c
  x86/xen: add some __init annotations in arch/x86/xen/mmu.c
  x86/xen: add some __init and static annotations in arch/x86/xen/setup.c
  x86/xen: use correct types for addresses in arch/x86/xen/setup.c
  ...

9 years agoMerge tag 'microblaze-3.20-rc1' of git://git.monstr.eu/linux-2.6-microblaze
Linus Torvalds [Tue, 10 Feb 2015 21:51:06 +0000 (13:51 -0800)]
Merge tag 'microblaze-3.20-rc1' of git://git.monstr.eu/linux-2.6-microblaze

Pull Microblaze pupdates from Michal Simek:
 - Remove various compilation errors
 - Various code cleanup patches
 - Add missing MB versions/architectures for autodetection

* tag 'microblaze-3.20-rc1' of git://git.monstr.eu/linux-2.6-microblaze:
  microblaze: Remove *.dtb files in make clean
  microblaze: whitespace fix
  microblaze/uaccess: fix sparse errors
  microblaze: intc: Reformat output
  microblaze: intc: Refactor DT sanity check
  microblaze: intc: Don't override error codes
  microblaze: Add target architecture
  microblaze: Add missing PVR version codes
  microblaze: Fix variable types to remove W=1 warning
  microblaze: Use unsigned type for limit comparison in cache.c
  microblaze: Use unsigned type for proper comparison in cpuinfo*.c
  microblaze: Use unsigned type for "for" loop because of comparison-kgdb.c
  microblaze: Change extern inline to static inline
  microblaze: Mark get_frame_size as static
  microblaze: Use unsigned return type in do_syscall_trace_enter
  microblaze: Declare microblaze_kgdb_break in header
  microblaze: Remove unused prom header from reset.c
  microblaze: Remove unused prom_parse.c
  microblaze: Wire-up execveat syscall
  microblaze: Use empty asm-generic/linkage.h

9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/geert/linux...
Linus Torvalds [Tue, 10 Feb 2015 21:27:24 +0000 (13:27 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/geert/linux-m68k

Pull m68k updates from Geert Uytterhoeven:
 "Summary:

  - switch to asm-generic/futex.h

  - various cleanups

  - defconfig updates"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/geert/linux-m68k:
  m68k/defconfig: Enable Ethernet bridging
  m68k/defconfig: Enable Atari EtherNAT and EtherNEC Ethernet support
  m68k/defconfig: Enable automounting of devtmpfs at /dev
  m68k/defconfig: Enable early printk support
  m68k/defconfig: Enable test modules
  m68k/defconfig: Refresh defconfigs for v3.16-rc1--v3.19-rc2
  m68k/atari: Remove obsolete IRQ_TYPE_*
  sound: dmasound_atari: Remove obsolete IRQ_TYPE_SLOW
  video: atafb: Remove obsolete IRQ_TYPE_PRIO
  parport: parport_atari: Remove obsolete IRQ_TYPE_SLOW
  net: atarilance: Remove obsolete IRQ_TYPE_PRIO
  m68k/atari: Remove obsolete keyboard_tasklet scheduling
  m68k/mac: Fix scsi_type for Mac LC and similar models
  m68k: Switch to asm-generic/futex.h
  m68k/mvme147: config.c - Remove unused functions
  m68k/atari: atakeyb.c - Remove some unused functions
  m68k/mvme16x: rtc - Don't use module_init in non-modular code

9 years agomicroblaze: Remove *.dtb files in make clean
Michal Simek [Tue, 20 Jan 2015 08:40:14 +0000 (09:40 +0100)]
microblaze: Remove *.dtb files in make clean

dts/Makefile is called only for simpleImage target
which is causing that *.dtb are not removed.
This patch fix it.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>