]> git.karo-electronics.de Git - karo-tx-linux.git/log
karo-tx-linux.git
11 years agopartitions-add-aix-lvm-partition-support-files-v2
Philippe De Muyter [Wed, 19 Jun 2013 00:08:34 +0000 (10:08 +1000)]
partitions-add-aix-lvm-partition-support-files-v2

Fix a problem in the discovering of small (1 pp) partitions in presence of
discontiguous partitions.

Signed-off-by: Philippe De Muyter <phdm@macqel.be>
Cc: Karel Zak <kzak@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agopartitions: add aix lvm partition support files
Philippe De Muyter [Wed, 19 Jun 2013 00:08:33 +0000 (10:08 +1000)]
partitions: add aix lvm partition support files

Add partitions/aix.h and partitions/aix.c.

AIX LVM permits to make "logical volumes" which are made of multiple
slices of multiple disks.  The new code allows only access to the "logical
volumes" which are made of one slice on the probed disk, a slice being a
contiguous disk area.  The code also detects "logical volumes" made of
multiple slices on the probed disk, but can not describe them to the
partition layer, because the partition layer generic code does not support
that.  When such non-contiguous "logical volumes" are detected, a
diagnostic message is printed.

Signed-off-by: Philippe De Muyter <phdm@macqel.be>
Cc: Karel Zak <kzak@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agopartitions/msdos.c: end-of-line whitespace and semicolon cleanup
Philippe De Muyter [Wed, 19 Jun 2013 00:08:33 +0000 (10:08 +1000)]
partitions/msdos.c: end-of-line whitespace and semicolon cleanup

Signed-off-by: Philippe De Muyter <phdm@macqel.be>
Cc: Karel Zak <kzak@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agomwave: fix info leak in mwave_ioctl()
Dan Carpenter [Wed, 19 Jun 2013 00:08:33 +0000 (10:08 +1000)]
mwave: fix info leak in mwave_ioctl()

Smatch complains that on 64 bit systems, there is a hole in the
MW_ABILITIES struct between ->component_count and ->component_list[].  It
leaks stack information from the mwave_ioctl() function.

I've added a memset() to initialize the struct to zero.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Greg KH <greg@kroah.com>
Cc: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc/sem.c: rename try_atomic_semop() to perform_atomic_semop(), docu update
Manfred Spraul [Wed, 19 Jun 2013 00:08:32 +0000 (10:08 +1000)]
ipc/sem.c: rename try_atomic_semop() to perform_atomic_semop(), docu update

Cleanup: Some minor points that I noticed while writing the
previous patches

1) The name try_atomic_semop() is misleading: The function performs the
   operation (if it is possible).

2) Some documentation updates.

No real code change, a rename and documentation changes.

Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc/sem.c: replace shared sem_otime with per-semaphore value
Manfred Spraul [Wed, 19 Jun 2013 00:08:32 +0000 (10:08 +1000)]
ipc/sem.c: replace shared sem_otime with per-semaphore value

sem_otime contains the time of the last semaphore operation that completed
successfully.  Every operation updates this value, thus access from
multiple cpus can cause thrashing.

Therefore the patch replaces the variable with a per-semaphore variable.
The per-array sem_otime is only calculated when required.

No performance improvement on a single-socket i3 - only important
for larger systems.

Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc/sem.c: always use only one queue for alter operations
Manfred Spraul [Wed, 19 Jun 2013 00:08:32 +0000 (10:08 +1000)]
ipc/sem.c: always use only one queue for alter operations

There are two places that can contain alter operations:
- the global queue: sma->pending_alter
- the per-semaphore queues: sma->sem_base[].pending_alter.

Since one of the queues must be processed first, this causes an odd
priorization of the wakeups:
Right now, complex operations have priority over simple ops.

The patch restores the behavior of linux <=3.0.9: The longest
waiting operation has the highest priority.

This is done by using only one queue:
- if there are complex ops, then sma->pending_alter is used.
- otherwise, the per-semaphore queues are used.

As a side effect, do_smart_update_queue() becomes much simpler:
No more goto logic.

Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc/sem.c: scan complex wait-for-zero after undefined updates
Manfred Spraul [Wed, 19 Jun 2013 00:08:31 +0000 (10:08 +1000)]
ipc/sem.c: scan complex wait-for-zero after undefined updates

After an update of a semaphore array that does not use struct sembuf,
complex wait-for-zero operations were not checked if they are able to
proceed.

Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc/sem: separate wait-for-zero and alter tasks into seperate queues
Manfred Spraul [Wed, 19 Jun 2013 00:08:31 +0000 (10:08 +1000)]
ipc/sem: separate wait-for-zero and alter tasks into seperate queues

Introduce separate queues for operations that do not modify the semaphore
values.  Advantages:

- Simpler logic in check_restart().
- Faster update_queue(): Right now, all wait-for-zero operations
  are always tested, even if the semaphore value is not 0.
- wait-for-zero gets again priority, as in linux <=3.0.9

Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc/sem.c: cacheline align the semaphore structures
Manfred Spraul [Wed, 19 Jun 2013 00:08:31 +0000 (10:08 +1000)]
ipc/sem.c: cacheline align the semaphore structures

As now each semaphore has its own spinlock and parallel operations are
possible, give each semaphore its own cacheline.

On a i3 laptop, this gives up to 28% better performance:

#semscale 10 | grep "interleave 2"
- before:
Cpus 1, interleave 2 delay 0: 36109234 in 10 secs
Cpus 2, interleave 2 delay 0: 55276317 in 10 secs
Cpus 3, interleave 2 delay 0: 62411025 in 10 secs
Cpus 4, interleave 2 delay 0: 81963928 in 10 secs

-after:
Cpus 1, interleave 2 delay 0: 35527306 in 10 secs
Cpus 2, interleave 2 delay 0: 70922909 in 10 secs <<< + 28%
Cpus 3, interleave 2 delay 0: 80518538 in 10 secs
Cpus 4, interleave 2 delay 0: 89115148 in 10 secs <<< + 8.7%

i3, with 2 cores and with hyperthreading enabled.  Interleave 2 in order
use first the full cores.  HT partially hides the delay from cacheline
trashing, thus the improvement is "only" 8.7% if 4 threads are running.

Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc-utilc-ipc_rcu_alloc-cacheline-align-allocation-checkpatch-fixes
Andrew Morton [Wed, 19 Jun 2013 00:08:30 +0000 (10:08 +1000)]
ipc-utilc-ipc_rcu_alloc-cacheline-align-allocation-checkpatch-fixes

ERROR: "(foo*)" should be "(foo *)"
#47: FILE: ipc/util.c:493:
+ struct ipc_rcu *p = ((struct ipc_rcu*)ptr)-1;

ERROR: "(foo*)" should be "(foo *)"
#58: FILE: ipc/util.c:509:
+ struct ipc_rcu *p = ((struct ipc_rcu*)ptr)-1;

total: 2 errors, 0 warnings, 34 lines checked

./patches/ipc-utilc-ipc_rcu_alloc-cacheline-align-allocation.patch has style problems, please review.

If any of these errors are false positives, please report
them to the maintainer, see CHECKPATCH in MAINTAINERS.

Please run checkpatch prior to sending patches

Cc: Manfred Spraul <manfred@colorfullife.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc/util.c, ipc_rcu_alloc: cacheline align allocation
Manfred Spraul [Wed, 19 Jun 2013 00:08:30 +0000 (10:08 +1000)]
ipc/util.c, ipc_rcu_alloc: cacheline align allocation

Enforce that ipc_rcu_alloc returns a cacheline aligned pointer on SMP.

Rationale:
The SysV sem code tries to move the main spinlock into a seperate cacheline
(____cacheline_aligned_in_smp). This works only if ipc_rcu_alloc returns
cacheline aligned pointers.
vmalloc and kmalloc return cacheline algined pointers, the implementation
of ipc_rcu_alloc breaks that.

Signed-off-by: Manfred Spraul <manfred@colorfullife.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Davidlohr Bueso <davidlohr.bueso@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc: remove unused functions
Davidlohr Bueso [Wed, 19 Jun 2013 00:08:30 +0000 (10:08 +1000)]
ipc: remove unused functions

We can now drop the msg_lock and msg_lock_check functions along with a
bogus comment introduced previously in semctl_down.

Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc,msg: shorten critical region in msgrcv
Davidlohr Bueso [Wed, 19 Jun 2013 00:08:30 +0000 (10:08 +1000)]
ipc,msg: shorten critical region in msgrcv

do_msgrcv() is the last msg queue function that abuses the ipc lock Take
it only when needed when actually updating msq.

Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc,msg: shorten critical region in msgsnd
Davidlohr Bueso [Wed, 19 Jun 2013 00:08:29 +0000 (10:08 +1000)]
ipc,msg: shorten critical region in msgsnd

do_msgsnd() is another function that does too many things with the ipc
object lock acquired.  Take it only when needed when actually updating
msq.

Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc,msg: make msgctl_nolock lockless
Davidlohr Bueso [Wed, 19 Jun 2013 00:08:29 +0000 (10:08 +1000)]
ipc,msg: make msgctl_nolock lockless

While the INFO cmd doesn't take the ipc lock, the STAT commands do acquire
it unnecessarily.  We can do the permissions and security checks only
holding the rcu lock.

This function now mimics semctl_nolock().

Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc,msg: introduce lockless functions to obtain the ipc object
Davidlohr Bueso [Wed, 19 Jun 2013 00:08:29 +0000 (10:08 +1000)]
ipc,msg: introduce lockless functions to obtain the ipc object

Add msq_obtain_object() and msq_obtain_object_check(), which will allow us
to get the ipc object without acquiring the lock.  Just as with
semaphores, these functions are basically wrappers around
ipc_obtain_object*().

Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc,msg: introduce msgctl_nolock
Davidlohr Bueso [Wed, 19 Jun 2013 00:08:28 +0000 (10:08 +1000)]
ipc,msg: introduce msgctl_nolock

Similar to semctl, when calling msgctl, the *_INFO and *_STAT commands can
be performed without acquiring the ipc object.

Add a msgctl_nolock() function and move the logic of *_INFO and *_STAT out
of msgctl().  This change still takes the lock and it will be properly
lockless in the next patch

Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc,msg: shorten critical region in msgctl_down
Davidlohr Bueso [Wed, 19 Jun 2013 00:08:28 +0000 (10:08 +1000)]
ipc,msg: shorten critical region in msgctl_down

Instead of holding the ipc lock for the entire function, use the
ipcctl_pre_down_nolock and only acquire the lock for specific commands:
RMID and SET.

Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc: move locking out of ipcctl_pre_down_nolock
Davidlohr Bueso [Wed, 19 Jun 2013 00:08:28 +0000 (10:08 +1000)]
ipc: move locking out of ipcctl_pre_down_nolock

This function currently acquires both the rw_mutex and the rcu lock on
successful lookups, leaving the callers to explicitly unlock them,
creating another two level locking situation.

Make the callers (including those that still use ipcctl_pre_down())
explicitly lock and unlock the rwsem and rcu lock.

Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc: close open coded spin lock calls
Davidlohr Bueso [Wed, 19 Jun 2013 00:08:27 +0000 (10:08 +1000)]
ipc: close open coded spin lock calls

Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc: introduce ipc object locking helpers
Davidlohr Bueso [Wed, 19 Jun 2013 00:08:27 +0000 (10:08 +1000)]
ipc: introduce ipc object locking helpers

Simple helpers around the (kern_ipc_perm *)->lock spinlock.

Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc: restore rcu locking in ipc_addid
Davidlohr Bueso [Wed, 19 Jun 2013 00:08:27 +0000 (10:08 +1000)]
ipc: restore rcu locking in ipc_addid

Fengguang reported the following trinity triggered issue:

[   51.524946]
[   51.525983] ===============================
[   51.532875] [ INFO: suspicious RCU usage. ]
[   51.535385] 3.10.0-rc4-next-20130606 #6 Not tainted
[   51.538304] -------------------------------
[   51.540937] include/linux/rcupdate.h:471 Illegal context switch in RCU read-side critical section!
[   51.548110]
[   51.548110] other info that might help us debug this:
[   51.548110]
[   51.553055]
[   51.553055] rcu_scheduler_active = 1, debug_locks = 1
[   51.557199] 2 locks held by trinity/1107:
[   51.560168]  #0:  (&ids->rw_mutex){+.+.+.}, at: [<ffffffff811e15ee>] ipcget+0x38/0x2b3
[   51.566465]  #1:  (rcu_read_lock){.+.+..}, at: [<ffffffff811e7698>] newseg+0x19d/0x3fd
[   51.572413]
[   51.572413] stack backtrace:
[   51.574761] CPU: 0 PID: 1107 Comm: trinity Not tainted 3.10.0-rc4-next-20130606 #6
[   51.579331] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2007
[   51.583068]  0000000000000001 ffff880004a07d88 ffffffff817b1f5c ffff880004a07db8
[   51.592119]  ffffffff810f2f1d ffffffff81b78569 00000000000001a8 0000000000000000
[   51.596726]  0000000000000000 ffff880004a07de8 ffffffff810ded5e ffff880004a07fd8
[   51.605189] Call Trace:
[   51.606409]  [<ffffffff817b1f5c>] dump_stack+0x19/0x1b
[   51.609632]  [<ffffffff810f2f1d>] lockdep_rcu_suspicious+0xeb/0xf4
[   51.612905]  [<ffffffff810ded5e>] __might_sleep+0x59/0x1dc
[   51.618614]  [<ffffffff81238623>] idr_preload+0x9b/0x142
[   51.621939]  [<ffffffff811e0e56>] ipc_addid+0x3d/0x193
[   51.624373]  [<ffffffff811e771c>] newseg+0x221/0x3fd
[   51.626596]  [<ffffffff811e7698>] ? newseg+0x19d/0x3fd
[   51.630177]  [<ffffffff811e1774>] ipcget+0x1be/0x2b3
[   51.633174]  [<ffffffff817bc094>] ? retint_swapgs+0x13/0x1b
[   51.636356]  [<ffffffff811e7a5a>] SyS_shmget+0x59/0x5d
[   51.639576]  [<ffffffff811e74fb>] ? shm_try_destroy_orphaned+0xbf/0xbf
[   51.643673]  [<ffffffff811e6ce5>] ? shm_get_unmapped_area+0x20/0x20
[   51.647321]  [<ffffffff811e6cf0>] ? shm_security+0xb/0xb
[   51.650831]  [<ffffffff817bcb27>] system_call_fastpath+0x16/0x1b

The issue was caused because we were allocating memory in GFP_KERNEL
context after calling rcu_read_lock.  This patch restores the
rcu_read_lock call into ipc_addid() and thus maintains the original
behavior.

Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Reported-by: Wu Fengguang <fengguang.wu@intel.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc: move rcu lock out of ipc_addid
Davidlohr Bueso [Wed, 19 Jun 2013 00:08:26 +0000 (10:08 +1000)]
ipc: move rcu lock out of ipc_addid

This patchset continues the work that began in the sysv ipc semaphore
scaling series: https://lkml.org/lkml/2013/3/20/546

Just like semaphores used to be, sysv shared memory and msg queues also
abuse the ipc lock, unnecessarily holding it for operations such as
permission and security checks.  This patchset mostly deals with mqueues,
and while shared mem can be done in a very similar way, I want to get
these patches out in the open first.  It also does some pending cleanups,
mostly focused on the two level locking we have in ipc code, taking care
of ipc_addid() and ipcctl_pre_down_nolock() - yes there are still
functions that need to be updated as well.

This patch:

Make all callers explicitly take and release the RCU read lock.

This addresses the two level locking seen in newary(), newseg() and
newqueue().  For the last two, explicitly unlock the ipc object and the
rcu lock, instead of calling the custom shm_unlock and msg_unlock
functions.  The next patch will deal with the open coded locking for
->perm.lock

Signed-off-by: Davidlohr Bueso <davidlohr.bueso@hp.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoipc/shmc.c: eliminate ugly 80-col tricks
Andrew Morton [Wed, 19 Jun 2013 00:08:26 +0000 (10:08 +1000)]
ipc/shmc.c: eliminate ugly 80-col tricks

Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoidr-print-a-stack-dump-after-ida_remove-warning-fix
Andrew Morton [Wed, 19 Jun 2013 00:08:26 +0000 (10:08 +1000)]
idr-print-a-stack-dump-after-ida_remove-warning-fix

convert the open-coded printk+dump_stack into WARN()

Cc: Jean Delvare <jdelvare@suse.de>
Cc: Takashi Iwai <tiwai@suse.de>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoidr: print a stack dump after ida_remove warning
Jean Delvare [Wed, 19 Jun 2013 00:08:25 +0000 (10:08 +1000)]
idr: print a stack dump after ida_remove warning

We print a dump stack after idr_remove warning.  This is useful to find
the faulty piece of code.  Let's do the same for ida_remove, as it would
be equally useful there.

Signed-off-by: Jean Delvare <jdelvare@suse.de>
Cc: Tejun Heo <tj@kernel.org>
Cc: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agos390: remove setting for saved_max_pfn
Zhang Yanfei [Wed, 19 Jun 2013 00:08:25 +0000 (10:08 +1000)]
s390: remove setting for saved_max_pfn

The only user of saved_max_pfn in s390 is read_oldmem interface but we
have removed that interface, so saved_max_pfn is now unneeded in s390, and
we needn't set it anymore.

Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Dave Hansen <dave@sr71.net>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: Matt Fleming <matt.fleming@intel.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoia64: remove setting for saved_max_pfn
Zhang Yanfei [Wed, 19 Jun 2013 00:08:25 +0000 (10:08 +1000)]
ia64: remove setting for saved_max_pfn

The only user of saved_max_pfn in ia64 is read_oldmem interface but we
have removed that interface, so saved_max_pfn is now unneeded in ia64, and
we needn't set it anymore.

Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Cc: Matt Fleming <matt.fleming@intel.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Dave Hansen <dave@sr71.net>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agopowerpc: Remove savemaxmem parameter setup
Zhang Yanfei [Wed, 19 Jun 2013 00:08:25 +0000 (10:08 +1000)]
powerpc: Remove savemaxmem parameter setup

saved_max_pfn is used to know the amount of memory that the previous
kernel used.  And for powerpc, we set saved_max_pfn by passing the kernel
commandline parameter "savemaxmem=".

The only user of saved_max_pfn in powerpc is read_oldmem interface.  Since
we have removed read_oldmem, we don't need this parameter anymore.

Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Dave Hansen <dave@sr71.net>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Matt Fleming <matt.fleming@intel.com>
Cc: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agomips: remove savemaxmem parameter setup
Zhang Yanfei [Wed, 19 Jun 2013 00:08:24 +0000 (10:08 +1000)]
mips: remove savemaxmem parameter setup

saved_max_pfn is used to know the amount of memory that the previous
kernel used.  And for powerpc, we set saved_max_pfn by passing the kernel
commandline parameter "savemaxmem=".

The only user of saved_max_pfn in mips is read_oldmem interface.  Since we
have removed read_oldmem, so we don't need this parameter anymore.

Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Dave Hansen <dave@sr71.net>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Matt Fleming <matt.fleming@intel.com>
Cc: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoDocumentation/kdump/kdump.txt: remove /dev/oldmem description
Zhang Yanfei [Wed, 19 Jun 2013 00:08:24 +0000 (10:08 +1000)]
Documentation/kdump/kdump.txt: remove /dev/oldmem description

Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Cc: Vivek Goyal <vgoyal@redhat.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Dave Hansen <dave@sr71.net>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Matt Fleming <matt.fleming@intel.com>
Cc: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Tony Luck <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodev-oldmem-remove-the-interface-fix
Andrew Morton [Wed, 19 Jun 2013 00:08:24 +0000 (10:08 +1000)]
dev-oldmem-remove-the-interface-fix

Mention /dev/oldmem obsolescence in devices.txt

Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Dave Hansen <dave@sr71.net>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Matt Fleming <matt.fleming@intel.com>
Cc: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Vivek Goyal <vgoyal@redhat.com>
Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years ago/dev/oldmem: Remove the interface
Zhang Yanfei [Wed, 19 Jun 2013 00:08:23 +0000 (10:08 +1000)]
/dev/oldmem: Remove the interface

/dev/oldmem provides the interface for us to access the "old memory" in
the dump-capture kernel.  Unfortunately, no one actually uses this
interface.

And this interface could actually cause some real problems if used on ia64
where the cached/uncached accesses are mixed.  See the discussion from the
link: https://lkml.org/lkml/2013/4/12/386.
So Eric suggested that we should remove /dev/oldmem as an unused piece of
code.

Suggested-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Cc: Vivek Goyal <vgoyal@redhat.com>
Cc: Dave Hansen <dave@sr71.net>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Matt Fleming <matt.fleming@intel.com>
Cc: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Tony Luck <tony.luck@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agofs/exec.c:de_thread: mt-exec should update ->real_start_time
Oleg Nesterov [Wed, 19 Jun 2013 00:08:23 +0000 (10:08 +1000)]
fs/exec.c:de_thread: mt-exec should update ->real_start_time

924b42d5 ("Use boot based time for process start time and boot time in
/proc") updated copy_process/do_task_stat but forgot about de_thread().
This breaks "ps axOT" if a sub-thread execs.

Note: I think that task->start_time should die.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Acked-by: John Stultz <johnstul@us.ibm.com>
Cc: Tomas Janousek <tjanouse@redhat.com>
Cc: Tomas Smetana <tsmetana@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agofs/exec.c: do_execve_common(): use current_user()
Oleg Nesterov [Wed, 19 Jun 2013 00:08:23 +0000 (10:08 +1000)]
fs/exec.c: do_execve_common(): use current_user()

Trivial cleanup.  do_execve_common() can use current_user() and avoid the
unnecessary "struct cred *cred" var.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: Vasiliy Kulikov <segoon@openwall.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agokernel/fork.c:copy_process(): consolidate the lockless CLONE_THREAD checks
Oleg Nesterov [Wed, 19 Jun 2013 00:08:22 +0000 (10:08 +1000)]
kernel/fork.c:copy_process(): consolidate the lockless CLONE_THREAD checks

copy_process() does a lot of "chaotic" initializations and checks
CLONE_THREAD twice before it takes tasklist.  In particular it sets
"p->group_leader = p" and then changes it again under tasklist if
!thread_group_leader(p).

This looks a bit confusing, lets create a single "if (CLONE_THREAD)" block
which initializes ->exit_signal, ->group_leader, and ->tgid.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pavel Emelyanov <xemul@parallels.com>
Cc: Sergey Dyasly <dserrg@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agokernel/fork.c:copy_process(): don't add the uninitialized child to thread/task/pid...
Oleg Nesterov [Wed, 19 Jun 2013 00:08:22 +0000 (10:08 +1000)]
kernel/fork.c:copy_process(): don't add the uninitialized child to thread/task/pid lists

copy_process() adds the new child to thread_group/init_task.tasks list and
then does attach_pid(child, PIDTYPE_PID).  This means that the lockless
next_thread() or next_task() can see this thread with the wrong pid.  Say,
"ls /proc/pid/task" can list the same inode twice.

We could move attach_pid(child, PIDTYPE_PID) up, but in this case
find_task_by_vpid() can find the new thread before it was fully
initialized.

And this is already true for PIDTYPE_PGID/PIDTYPE_SID, With this patch
copy_process() initializes child->pids[*].pid first, then calls
attach_pid() to insert the task into the pid->tasks list.

attach_pid() no longer need the "struct pid*" argument, it is always
called after pid_link->pid was already set.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pavel Emelyanov <xemul@parallels.com>
Cc: Sergey Dyasly <dserrg@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agokernel/fork.c:copy_process(): unify CLONE_THREAD-or-thread_group_leader code
Oleg Nesterov [Wed, 19 Jun 2013 00:08:22 +0000 (10:08 +1000)]
kernel/fork.c:copy_process(): unify CLONE_THREAD-or-thread_group_leader code

Cleanup and preparation for the next changes.

Move the "if (clone_flags & CLONE_THREAD)" code down under "if
(likely(p->pid))" and turn it into into the "else" branch.  This makes the
process/thread initialization more symmetrical and removes one check.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pavel Emelyanov <xemul@parallels.com>
Cc: Sergey Dyasly <dserrg@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agofork: reorder permissions when violating number of processes limits
Eric Paris [Wed, 19 Jun 2013 00:08:21 +0000 (10:08 +1000)]
fork: reorder permissions when violating number of processes limits

When a task is attempting to violate the RLIMIT_NPROC limit we have a
check to see if the task is sufficiently priviledged.  The check first
looks at CAP_SYS_ADMIN, then CAP_SYS_RESOURCE, then if the task is uid=0.

A result is that tasks which are allowed by the uid=0 check are first
checked against the security subsystem.  This results in the security
subsystem auditting a denial for sys_admin and sys_resource and then the
task passing the uid=0 check.

This patch rearranges the code to first check uid=0, since if we pass that
we shouldn't hit the security system at all.  We then check sys_resource,
since it is the smallest capability which will solve the problem.  Lastly
we check the fallback everything cap_sysadmin.  We don't want to give this
capability many places since it is so powerful.

This will eliminate many of the false positive/needless denial messages we
get when a root task tries to violate the nproc limit.  (note that
kthreads count against root, so on a sufficiently large machine we can
actually get past the default limits before any userspace tasks are
launched.)

Signed-off-by: Eric Paris <eparis@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agofs/proc/uptime.c:uptime_proc_show(): use get_monotonic_boottime()
Oleg Nesterov [Wed, 19 Jun 2013 00:08:21 +0000 (10:08 +1000)]
fs/proc/uptime.c:uptime_proc_show(): use get_monotonic_boottime()

Change uptime_proc_show() to use get_monotonic_boottime() instead of
do_posix_clock_monotonic_gettime() + monotonic_to_bootbased().

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Acked-by: John Stultz <johnstul@us.ibm.com>
Cc: Tomas Janousek <tjanouse@redhat.com>
Cc: Tomas Smetana <tsmetana@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoexit.c: unexport __set_special_pids()
Oleg Nesterov [Wed, 19 Jun 2013 00:08:21 +0000 (10:08 +1000)]
exit.c: unexport __set_special_pids()

Move __set_special_pids() from exit.c to sys.c close to its single caller
and make it static.

And rename it to set_special_pids(), another helper with this name has
gone away.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agomove-exit_task_namespaces-outside-of-exit_notify-fix
Andrew Morton [Wed, 19 Jun 2013 00:08:20 +0000 (10:08 +1000)]
move-exit_task_namespaces-outside-of-exit_notify-fix

make fput() comment more truthful

Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Andrey Vagin <avagin@openvz.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agofs/exec.c:de_thread(): use change_pid() rather than detach_pid/attach_pid
Oleg Nesterov [Wed, 19 Jun 2013 00:08:20 +0000 (10:08 +1000)]
fs/exec.c:de_thread(): use change_pid() rather than detach_pid/attach_pid

de_thread() can use change_pid() instead of detach + attach.  This looks
better and this ensures that, say, next_thread() can never see a task with
->pid == NULL.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: Michal Hocko <mhocko@suse.cz>
Cc: Pavel Emelyanov <xemul@parallels.com>
Cc: Sergey Dyasly <dserrg@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agocoredump: avoid the uninitialized cn->corename if core_pattern is empty
Oleg Nesterov [Wed, 19 Jun 2013 00:08:20 +0000 (10:08 +1000)]
coredump: avoid the uninitialized cn->corename if core_pattern is empty

If core_pattern is "" or "|", cn->corename is used uninitialized by
filp_open() or call_usermodehelper_exec().

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Colin Walters <walters@verbum.org>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Jiri Slaby <jslaby@suse.cz>
Cc: Lennart Poettering <mzxreary@0pointer.de>
Cc: Lucas De Marchi <lucas.de.marchi@gmail.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agocoredump: '% at the end' shouldn't bypass core_uses_pid logic
Oleg Nesterov [Wed, 19 Jun 2013 00:08:19 +0000 (10:08 +1000)]
coredump: '% at the end' shouldn't bypass core_uses_pid logic

"goto end" should not bypass the "Backward compatibility with
core_uses_pid" code, move this label up.

While at it,

- It is ugly to copy '|' into cn->corename and then inc
  the pointer for argv_split().

  Change format_corename() to increment pat_ptr instead.

- Remove the dead "if (*pat_ptr == 0)" in format_corename(),
  we already checked it is not zero.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Colin Walters <walters@verbum.org>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Jiri Slaby <jslaby@suse.cz>
Cc: Lennart Poettering <mzxreary@0pointer.de>
Cc: Lucas De Marchi <lucas.de.marchi@gmail.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agocoredump: kill call_count, add core_name_size
Oleg Nesterov [Wed, 19 Jun 2013 00:08:19 +0000 (10:08 +1000)]
coredump: kill call_count, add core_name_size

Imho, "atomic_t call_count" is ugly and should die.  It buys nothing and
in fact it can grow more than necessary, expand doesn't check if it was
already incremented by another task.

Kill it, and introduce "static int core_name_size" updated by
expand_corename().  This is obviously racy too but harmless, and
core_name_size never grows for no reason.

We do not bother to to calculate the "right" new size, we simply do
kmalloc(size_we_need) and use ksize() to rely on kmalloc_index's decision.

Finally change format_corename() to use expand_corename(), krealloc(NULL)
is fine.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Colin Walters <walters@verbum.org>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Jiri Slaby <jslaby@suse.cz>
Cc: Lennart Poettering <mzxreary@0pointer.de>
Cc: Lucas De Marchi <lucas.de.marchi@gmail.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agocoredump: kill cn_escape(), introduce cn_esc_printf()
Oleg Nesterov [Wed, 19 Jun 2013 00:08:19 +0000 (10:08 +1000)]
coredump: kill cn_escape(), introduce cn_esc_printf()

The usage of cn_escape() looks really annoying, imho this sequence needs a
wrapper.  And it is buggy.  If cn_printf() does expand_corename()
cn_escape() writes to the freed memory.

Introduce cn_esc_printf() which hopefully does this all right.  It records
the index before cn_vprintf(), not "char *" which is no longer valid (in
general) after krealloc().

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Colin Walters <walters@verbum.org>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Jiri Slaby <jslaby@suse.cz>
Cc: Lennart Poettering <mzxreary@0pointer.de>
Cc: Lucas De Marchi <lucas.de.marchi@gmail.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agocoredump: cn_vprintf() has no reason to call vsnprintf() twice
Oleg Nesterov [Wed, 19 Jun 2013 00:08:18 +0000 (10:08 +1000)]
coredump: cn_vprintf() has no reason to call vsnprintf() twice

cn_vprintf() looks really overcomplicated and sub-optimal.  We do not need
vsnprintf(NULL) to calculate the size we need, we can simply try to print
into the current buffer and expand/retry only if necessary.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Colin Walters <walters@verbum.org>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Jiri Slaby <jslaby@suse.cz>
Cc: Lennart Poettering <mzxreary@0pointer.de>
Cc: Lucas De Marchi <lucas.de.marchi@gmail.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agocoredump: introduce cn_vprintf()
Oleg Nesterov [Wed, 19 Jun 2013 00:08:18 +0000 (10:08 +1000)]
coredump: introduce cn_vprintf()

Turn cn_printf(...) into cn_vprintf(va_list args), reintroduce
cn_printf() as a trivial wrapper.

This simplifies the next change and cn_vprintf() will have more
callers.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Colin Walters <walters@verbum.org>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Jiri Slaby <jslaby@suse.cz>
Cc: Lennart Poettering <mzxreary@0pointer.de>
Cc: Lucas De Marchi <lucas.de.marchi@gmail.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agocoredump: format_corename() can leak cn->corename
Oleg Nesterov [Wed, 19 Jun 2013 00:08:18 +0000 (10:08 +1000)]
coredump: format_corename() can leak cn->corename

do_coredump() assumes that format_corename() can only fail if
expand_corename() fails and frees cn->corename.  This is not true, for
example cn_print_exe_file() can fail and in this case nobody frees
cn->corename.

Change do_coredump() to always do kfree(cn->corename) after it calls
format_corename() (NULL is fine), change expand_corename() to do nothing
if kmalloc() fails.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Colin Walters <walters@verbum.org>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Jiri Slaby <jslaby@suse.cz>
Cc: Lennart Poettering <mzxreary@0pointer.de>
Cc: Lucas De Marchi <lucas.de.marchi@gmail.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agousermodehelper: kill the sub_info->path[0] check
Oleg Nesterov [Wed, 19 Jun 2013 00:08:17 +0000 (10:08 +1000)]
usermodehelper: kill the sub_info->path[0] check

call_usermodehelper_exec() does nothing but returns success if path[0] ==
0.  The only user which needs this strange feature is request_module(), it
can check modprobe_path[0] itself like other users do if they want to
detect the "disabled by admin" case.

Kill it.  Not only it looks strange, it can confuse other callers.  And
this allows us to revert 264b83c0 ("usermodehelper: check
subprocess_info->path != NULL"), do_execve(NULL) is safe.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Lucas De Marchi <lucas.de.marchi@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoptrace-add-ability-to-get-set-signal-blocked-mask-fix
Andrew Morton [Wed, 19 Jun 2013 00:08:17 +0000 (10:08 +1000)]
ptrace-add-ability-to-get-set-signal-blocked-mask-fix

be consistent with brace layout

Cc: Andrey Vagin <avagin@openvz.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoptrace: add ability to get/set signal-blocked mask
Andrey Vagin [Wed, 19 Jun 2013 00:08:17 +0000 (10:08 +1000)]
ptrace: add ability to get/set signal-blocked mask

crtools uses a parasite code for dumping processes.  The parasite code is
injected into a process with help PTRACE_SEIZE.

Currently crtools blocks signals from a parasite code.  If a process has
pending signals, crtools wait while a process handles these signals.

This method is not suitable for stopped tasks.  A stopped task can have a
few pending signals, when we will try to execute a parasite code, we will
need to drop SIGSTOP, but all other signals must remain pending, because a
state of processes must not be changed during checkpointing.

This patch adds two ptrace commands to set/get signal-blocked mask.

I think gdb can use this commands too.

Signed-off-by: Andrey Vagin <avagin@openvz.org>
Reviewed-by: Oleg Nesterov <oleg@redhat.com>
Cc: Roland McGrath <roland@redhat.com>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Cc: Pavel Emelyanov <xemul@parallels.com>
Cc: Cyrill Gorcunov <gorcunov@openvz.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agox86: kill TIF_DEBUG
Oleg Nesterov [Wed, 19 Jun 2013 00:08:16 +0000 (10:08 +1000)]
x86: kill TIF_DEBUG

Because it is not used.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jan Kratochvil <jan.kratochvil@redhat.com>
Cc: Michael Neuling <mikey@neuling.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Prasad <prasad@linux.vnet.ibm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoptrace/x86: flush_ptrace_hw_breakpoint() shoule clear the virtual debug registers
Oleg Nesterov [Wed, 19 Jun 2013 00:08:16 +0000 (10:08 +1000)]
ptrace/x86: flush_ptrace_hw_breakpoint() shoule clear the virtual debug registers

flush_ptrace_hw_breakpoint() destroys the counters set by ptrace, but
"leaks" ->debugreg6 and ->ptrace_dr7.

The problem is minor, but still it doesn't look right and flush_thread()
did this until 66cb5917 ("hw-breakpoints: use the new wrapper routines to
access debug registers in process/thread code").  Now that PTRACE_DETACH
does flush_ too this makes even more sense.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jan Kratochvil <jan.kratochvil@redhat.com>
Cc: Michael Neuling <mikey@neuling.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Prasad <prasad@linux.vnet.ibm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoptrace: PTRACE_DETACH should do flush_ptrace_hw_breakpoint(child)
Oleg Nesterov [Wed, 19 Jun 2013 00:08:16 +0000 (10:08 +1000)]
ptrace: PTRACE_DETACH should do flush_ptrace_hw_breakpoint(child)

Change ptrace_detach() to call flush_ptrace_hw_breakpoint(child).  This
frees the slots for non-ptrace PERF_TYPE_BREAKPOINT users, and this
ensures that the tracee won't be killed by SIGTRAP triggered by the active
breakpoints.

Test-case:

unsigned long encode_dr7(int drnum, int enable, unsigned int type, unsigned int len)
{
unsigned long dr7;

dr7 = ((len | type) & 0xf)
<< (DR_CONTROL_SHIFT + drnum * DR_CONTROL_SIZE);
if (enable)
dr7 |= (DR_GLOBAL_ENABLE << (drnum * DR_ENABLE_SIZE));

return dr7;
}

int write_dr(int pid, int dr, unsigned long val)
{
return ptrace(PTRACE_POKEUSER, pid,
offsetof (struct user, u_debugreg[dr]),
val);
}

void func(void)
{
}

int main(void)
{
int pid, stat;
unsigned long dr7;

pid = fork();
if (!pid) {
assert(ptrace(PTRACE_TRACEME, 0,0,0) == 0);
kill(getpid(), SIGHUP);

func();
return 0x13;
}

assert(pid == waitpid(-1, &stat, 0));
assert(WSTOPSIG(stat) == SIGHUP);

assert(write_dr(pid, 0, (long)func) == 0);
dr7 = encode_dr7(0, 1, DR_RW_EXECUTE, DR_LEN_1);
assert(write_dr(pid, 7, dr7) == 0);

assert(ptrace(PTRACE_DETACH, pid, 0,0) == 0);
assert(pid == waitpid(-1, &stat, 0));
assert(stat == 0x1300);

return 0;
}

Before this patch the child is killed after PTRACE_DETACH.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jan Kratochvil <jan.kratochvil@redhat.com>
Cc: Michael Neuling <mikey@neuling.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Prasad <prasad@linux.vnet.ibm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoptrace/x86: cleanup ptrace_set_debugreg()
Oleg Nesterov [Wed, 19 Jun 2013 00:08:16 +0000 (10:08 +1000)]
ptrace/x86: cleanup ptrace_set_debugreg()

ptrace_set_debugreg() is trivial but looks horrible.  Kill the unnecessary
goto's and return's to cleanup the code.

This matches ptrace_get_debugreg() which also needs the trivial whitespace
cleanups.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jan Kratochvil <jan.kratochvil@redhat.com>
Cc: Michael Neuling <mikey@neuling.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Prasad <prasad@linux.vnet.ibm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoptrace/x86: ptrace_write_dr7() should create bp if !disabled
Oleg Nesterov [Wed, 19 Jun 2013 00:08:15 +0000 (10:08 +1000)]
ptrace/x86: ptrace_write_dr7() should create bp if !disabled

24f1e32c ("hw-breakpoints: Rewrite the hw-breakpoints layer on top of perf
events") introduced the minor regression.  Before this commit

PTRACE_POKEUSER DR7, enableDR0
PTRACE_POKEUSER DR0, address

was perfectly valid, now PTRACE_POKEUSER(DR7) fails if DR0 was not
previously initialized by PTRACE_POKEUSER(DR0).

Change ptrace_write_dr7() to do ptrace_register_breakpoint(addr => 0) if
!bp && !disabled.  This fixes watchpoint-zeroaddr from ptrace-tests, see
https://bugzilla.redhat.com/show_bug.cgi?id=660204.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Reported-by: Jan Kratochvil <jan.kratochvil@redhat.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Michael Neuling <mikey@neuling.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Prasad <prasad@linux.vnet.ibm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoptrace/x86: introduce ptrace_register_breakpoint()
Oleg Nesterov [Wed, 19 Jun 2013 00:08:15 +0000 (10:08 +1000)]
ptrace/x86: introduce ptrace_register_breakpoint()

No functional changes, preparation.

Extract the "register breakpoint" code from ptrace_get_debugreg() into the
new/generic helper, ptrace_register_breakpoint().  It will have more
users.

The patch also adds another simple helper, ptrace_fill_bp_fields(), to
factor out the arch_bp_generic_fields() logic in register/modify.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jan Kratochvil <jan.kratochvil@redhat.com>
Cc: Michael Neuling <mikey@neuling.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Prasad <prasad@linux.vnet.ibm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoptrace/x86: dont delay "disable" till second pass in ptrace_write_dr7()
Oleg Nesterov [Wed, 19 Jun 2013 00:08:15 +0000 (10:08 +1000)]
ptrace/x86: dont delay "disable" till second pass in ptrace_write_dr7()

ptrace_write_dr7() skips ptrace_modify_breakpoint(disabled => true) unless
second_pass, this buys nothing but complicates the code and means that we
always do the main loop twice even if "disabled" was never true.

The comment says:

Don't unregister the breakpoints right-away,
unless all register_user_hw_breakpoint()
requests have succeeded.

Firstly, we do not do register_user_hw_breakpoint(), it was removed by
24f1e32c ("hw-breakpoints: Rewrite the hw-breakpoints layer on top of perf
events").

We are going to restore register_user_hw_breakpoint() (see the next patch)
but this doesn't matter, after 44234adc "hw-breakpoints: Modify
breakpoints without unregistering them" perf_event_disable() can not hurt,
hw_breakpoint_del() does not free the slot.

Remove the "second_pass" check from the main loop and simplify the code.
Since we have to check "bp != NULL" anyway, the patch also removes the
same check in ptrace_modify_breakpoint() and moves the comment into
ptrace_write_dr7().

With this patch the second pass is only needed to restore the saved
old_dr7.  This should never fail, so the patch adds WARN_ON() to catch the
potential problems as Frederic suggested.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jan Kratochvil <jan.kratochvil@redhat.com>
Cc: Michael Neuling <mikey@neuling.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Prasad <prasad@linux.vnet.ibm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoptrace/x86: simplify the "disable" logic in ptrace_write_dr7()
Oleg Nesterov [Wed, 19 Jun 2013 00:08:14 +0000 (10:08 +1000)]
ptrace/x86: simplify the "disable" logic in ptrace_write_dr7()

ptrace_write_dr7() looks unnecessarily overcomplicated.  We can factor out
ptrace_modify_breakpoint() and do not do "continue" twice, just we need to
pass the proper "disabled" argument to ptrace_modify_breakpoint().

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jan Kratochvil <jan.kratochvil@redhat.com>
Cc: Michael Neuling <mikey@neuling.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Prasad <prasad@linux.vnet.ibm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoptrace: revert "Prepare to fix racy accesses on task breakpoints"
Oleg Nesterov [Wed, 19 Jun 2013 00:08:14 +0000 (10:08 +1000)]
ptrace: revert "Prepare to fix racy accesses on task breakpoints"

This reverts commit bf26c018490c2fce ("Prepare to fix racy accesses on
task breakpoints").

The patch was fine but we can no longer race with SIGKILL after 9899d11f
("ptrace: ensure arch_ptrace/ptrace_request can never race with SIGKILL"),
the __TASK_TRACED tracee can't be woken up and ->ptrace_bps[] can't go
away.

Now that ptrace_get_breakpoints/ptrace_put_breakpoints have no callers, we
can kill them and remove task->ptrace_bp_refcnt.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Michael Neuling <mikey@neuling.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jan Kratochvil <jan.kratochvil@redhat.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Prasad <prasad@linux.vnet.ibm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoptrace/sh: revert "hw_breakpoints: Fix racy access to ptrace breakpoints"
Oleg Nesterov [Wed, 19 Jun 2013 00:08:14 +0000 (10:08 +1000)]
ptrace/sh: revert "hw_breakpoints: Fix racy access to ptrace breakpoints"

This reverts commit e0ac8457d020c ("hw_breakpoints: Fix racy access to
ptrace breakpoints").

The patch was fine but we can no longer race with SIGKILL after 9899d11f
("ptrace: ensure arch_ptrace/ptrace_request can never race with SIGKILL"),
the __TASK_TRACED tracee can't be woken up and ->ptrace_bps[] can't go
away.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jan Kratochvil <jan.kratochvil@redhat.com>
Cc: Michael Neuling <mikey@neuling.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Prasad <prasad@linux.vnet.ibm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoptrace/arm: revert "hw_breakpoints: Fix racy access to ptrace breakpoints"
Oleg Nesterov [Wed, 19 Jun 2013 00:08:13 +0000 (10:08 +1000)]
ptrace/arm: revert "hw_breakpoints: Fix racy access to ptrace breakpoints"

This reverts commit bf0b8f4b55e ("hw_breakpoints: Fix racy access to
ptrace breakpoints").

The patch was fine but we can no longer race with SIGKILL after 9899d11f
("ptrace: ensure arch_ptrace/ptrace_request can never race with SIGKILL"),
the __TASK_TRACED tracee can't be woken up and ->ptrace_bps[] can't go
away.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Will Deacon <will.deacon@arm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jan Kratochvil <jan.kratochvil@redhat.com>
Cc: Michael Neuling <mikey@neuling.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Prasad <prasad@linux.vnet.ibm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoptrace/powerpc: revert "hw_breakpoints: Fix racy access to ptrace breakpoints"
Oleg Nesterov [Wed, 19 Jun 2013 00:08:13 +0000 (10:08 +1000)]
ptrace/powerpc: revert "hw_breakpoints: Fix racy access to ptrace breakpoints"

This reverts commit 07fa7a0a8a586 ("hw_breakpoints: Fix racy access to
ptrace breakpoints") and removes ptrace_get/put_breakpoints() added by
other commits.

The patch was fine but we can no longer race with SIGKILL after 9899d11f
("ptrace: ensure arch_ptrace/ptrace_request can never race with SIGKILL"),
the __TASK_TRACED tracee can't be woken up and ->ptrace_bps[] can't go
away.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Michael Neuling <mikey@neuling.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jan Kratochvil <jan.kratochvil@redhat.com>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Prasad <prasad@linux.vnet.ibm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoptrace/x86: revert "hw_breakpoints: Fix racy access to ptrace breakpoints"
Oleg Nesterov [Wed, 19 Jun 2013 00:08:13 +0000 (10:08 +1000)]
ptrace/x86: revert "hw_breakpoints: Fix racy access to ptrace breakpoints"

This reverts commit 87dc669ba257 ("hw_breakpoints: Fix racy access to
ptrace breakpoints").

The patch was fine but we can no longer race with SIGKILL after 9899d11f
("ptrace: ensure arch_ptrace/ptrace_request can never race with SIGKILL"),
the __TASK_TRACED tracee can't be woken up and ->ptrace_bps[] can't go
away.

The patch only removes ptrace_get_breakpoints/ptrace_put_breakpoints and
does a couple of "while at it" cleanups, it doesn't remove other changes
from the reverted commit.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Acked-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Jan Kratochvil <jan.kratochvil@redhat.com>
Cc: Michael Neuling <mikey@neuling.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Prasad <prasad@linux.vnet.ibm.com>
Cc: Russell King <linux@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodocbook: add futexes to kernel-locking docbook
Randy Dunlap [Wed, 19 Jun 2013 00:08:12 +0000 (10:08 +1000)]
docbook: add futexes to kernel-locking docbook

Add Fast User Mutexes (futexes) to kernel-locking docbook.

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Acked-by: Rob Landley <rob@landley.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agoDocumentation/CodingStyle: allow multiple return statements per function
Dan Carpenter [Wed, 19 Jun 2013 00:08:12 +0000 (10:08 +1000)]
Documentation/CodingStyle: allow multiple return statements per function

A surprising number of newbies interpret this section to mean that only
one return statement is allowed per function.  Part of the problem is that
the "one return statement per function" rule is an actual style guideline
that people are used to from other projects.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Eduardo Valentin <eduardo.valentin@ti.com>
Cc: Rob Landley <rob@landley.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agofat-additions-to-support-fat_fallocate-fix
Andrew Morton [Wed, 19 Jun 2013 00:08:12 +0000 (10:08 +1000)]
fat-additions-to-support-fat_fallocate-fix

fix min() warning

Cc: Amit Sahrawat <a.sahrawat@samsung.com>
Cc: Namjae Jeon <namjae.jeon@samsung.com>
Cc: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Cc: Ravishankar N <ravi.n1@samsung.com>
Reported-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agofat: additions to support fat_fallocate
Namjae Jeon [Wed, 19 Jun 2013 00:08:11 +0000 (10:08 +1000)]
fat: additions to support fat_fallocate

Implement preallocation via the fallocate syscall on VFAT partitions.

With FALLOC_FL_KEEP_SIZE, there is no way to distinguish if the mismatch
between i_size and no.  of clusters allocated is a consequence of
fallocate or just plain corruption.  When a non fallocate aware (old)
linux fat driver tries to write to such a file, it throws an error.Also,
fsck detects this as inconsistency and truncates the prealloc'd blocks.

To avoid this, as suggested by OGAWA, remove changes that make fallocate
persistent across mounts and restrict lifetime of blocks from fallocate(2)
to file release.

Signed-off-by: Namjae Jeon <namjae.jeon@samsung.com>
Signed-off-by: Ravishankar N <ravi.n1@samsung.com>
Signed-off-by: Amit Sahrawat <a.sahrawat@samsung.com>
Cc: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agofs/fat: use fat_msg() to replace printk() in __fat_fs_error()
Gu Zheng [Wed, 19 Jun 2013 00:08:11 +0000 (10:08 +1000)]
fs/fat: use fat_msg() to replace printk() in __fat_fs_error()

Signed-off-by: Gu Zheng <guz.fnst@cn.fujitsu.com>
Acked-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years ago] nilfs2: use atomic64_t type for inodes_count and blocks_count fields in nilfs_root...
Vyacheslav Dubeyko [Wed, 19 Jun 2013 00:08:11 +0000 (10:08 +1000)]
] nilfs2: use atomic64_t type for inodes_count and blocks_count fields in nilfs_root struct

The cp_inodes_count and cp_blocks_count are represented as __le64 type in
on-disk structure (struct nilfs_checkpoint).  But analogous fields in
in-core structure (struct nilfs_root) are represented by atomic_t type.

This patch replaces atomic_t on atomic64_t type in representation of
inodes_count and blocks_count fields in struct nilfs_root.

Signed-off-by: Vyacheslav Dubeyko <slava@dubeyko.com>
Acked-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Acked-by: Joern Engel <joern@logfs.org>
Cc: Clemens Eisserer <linuxhippy@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agonilfs2: implement calculation of free inodes count
Vyacheslav Dubeyko [Wed, 19 Jun 2013 00:08:10 +0000 (10:08 +1000)]
nilfs2: implement calculation of free inodes count

Currently, NILFS2 returns 0 as free inodes count (f_ffree) and current
used inodes count as total file nodes in file system (f_files):

df -i
Filesystem      Inodes  IUsed   IFree IUse% Mounted on
/dev/loop0           2      2       0  100% /mnt/nilfs2

This patch implements real calculation of free inodes count.  First of
all, it is calculated total file nodes in file system as
(desc_blocks_count * groups_per_desc_block * entries_per_group).  Then, it
is calculated free inodes count as difference the total file nodes and
used inodes count.  As a result, we have such output for NILFS2:

df -i
Filesystem       Inodes   IUsed    IFree IUse% Mounted on
/dev/loop0      4194304 2114701  2079603   51% /mnt/nilfs2

Reported-by: Clemens Eisserer <linuxhippy@gmail.com>
Signed-off-by: Vyacheslav Dubeyko <slava@dubeyko.com>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Tested-by: Vyacheslav Dubeyko <slava@dubeyko.com>
Cc: Joern Engel <joern@logfs.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agominix: bug widening a binary "not" operation
Dan Carpenter [Wed, 19 Jun 2013 00:08:10 +0000 (10:08 +1000)]
minix: bug widening a binary "not" operation

"chunk_size" is an unsigned int and "pos" is an unsigned long.  The
"& ~(chunk_size-1)" operation clears the high 32 bits unintentionally.

The ALIGN() macro does the correct thing.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/class: convert from Legacy pm ops to dev_pm_ops
Shuah Khan [Wed, 19 Jun 2013 00:08:10 +0000 (10:08 +1000)]
drivers/rtc/class: convert from Legacy pm ops to dev_pm_ops

Convert drivers/rtc/class to use dev_pm_ops for power management and
remove Legacy PM ops hooks.  With this change, rtc class registers
suspend/resume callbacks via class->pm (dev_pm_ops) instead of Legacy
class->suspend/resume.  When __device_suspend() runs call-backs, it will
find class->pm ops for the rtc class.

Signed-off-by: Shuah Khan <shuah.kh@samsung.com>
Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Cc: Shuah Khan <shuahkhan@gmail.com>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-pcf2123.c: replace strict_strtoul() with kstrtoul()
Jingoo Han [Wed, 19 Jun 2013 00:08:09 +0000 (10:08 +1000)]
drivers/rtc/rtc-pcf2123.c: replace strict_strtoul() with kstrtoul()

The usage of strict_strtoul() is not preferred, because strict_strtoul()
is obsolete.  Thus, kstrtoul() should be used.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/interface.c: return -EBUSY, not -EACCES when device is busy
Chris Brand [Wed, 19 Jun 2013 00:08:09 +0000 (10:08 +1000)]
drivers/rtc/interface.c: return -EBUSY, not -EACCES when device is busy

If rtc->irq_task is non-NULL and task is NULL, they always
rtc_irq_set_freq(), whenever err is set to -EBUSY it will then immediately
be set to -EACCES, misleading the caller as to the underlying problem.

Signed-off-by: Chris Brand <chris.brand@broadcom.com>
Acked-by: Alessandro Zummo <a.zummo@towertech.it>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-twl.c: cleanup with module_platform_driver() conversion
Peter Ujfalusi [Wed, 19 Jun 2013 00:08:09 +0000 (10:08 +1000)]
drivers/rtc/rtc-twl.c: cleanup with module_platform_driver() conversion

Use module_platform_driver() to register the platform driver.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Acked-by: Kevin Hilman <khilman@linaro.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-twl.c: fix rtc_reg_map initialization
Peter Ujfalusi [Wed, 19 Jun 2013 00:08:08 +0000 (10:08 +1000)]
drivers/rtc/rtc-twl.c: fix rtc_reg_map initialization

Initialize the rtc_reg_map in platform_driver's probe function instead at
module_init time.  This way we can make sure that the twl-core has been
already probed and initialized (twl_priv->twl_id is valid) since the
platform device for the RTC driver will be created by the twl-core after
it finished its init.

Reported-by: Christoph Fritz <chf.fritz@googlemail.com>
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: Kevin Hilman <khilman@linaro.org>
Tested-by: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-cmos.c: work around bios clearing rtc control
Derek Basehore [Wed, 19 Jun 2013 00:08:08 +0000 (10:08 +1000)]
drivers/rtc/rtc-cmos.c: work around bios clearing rtc control

The bios may clear the rtc control register when resuming the system. Since the
cmos interrupt handler may now be run before the rtc_cmos is resumed, this can
cause the interrupt handler to ignore an alarm since the alarm bit is not set in
the rtc control register. To work around this, check if the rtc_cmos is
suspended and use the stored value for the rtc control register.

Signed-off-by: Derek Basehore <dbasehore@chromium.org>
Reviewed-by: Sameer Nanda <snanda@chromium.org>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Cc: Jingoo Han <jg1.han@samsung.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: John Stultz <john.stultz@linaro.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-twl.c: ensure IRQ is wakeup enabled
Kevin Hilman [Wed, 19 Jun 2013 00:08:08 +0000 (10:08 +1000)]
drivers/rtc/rtc-twl.c: ensure IRQ is wakeup enabled

Currently, the RTC IRQ is never wakeup-enabled so is not capable of
bringing the system out of suspend.

On OMAP platforms, we have gotten by without this because the TWL RTC is
on an I2C-connected chip which is capable of waking up the OMAP via the IO
ring when the OMAP is in low-power states.

However, if the OMAP suspends without hitting the low-power states (and
the IO ring is not enabled), RTC wakeups will not work because the IRQ is
not wakeup enabled.

To fix, ensure the RTC IRQ is wakeup enabled whenever the RTC alarm is
set.

Signed-off-by: Kevin Hilman <khilman@linaro.org>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Cc: Tony Lindgren <tony@atomide.com>
Cc: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-pcf8583.c: use PTR_RET()
Sachin Kamat [Wed, 19 Jun 2013 00:08:07 +0000 (10:08 +1000)]
drivers/rtc/rtc-pcf8583.c: use PTR_RET()

Use of PTR_RET() simplifies the code.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-pcf8563.c: use PTR_RET()
Sachin Kamat [Wed, 19 Jun 2013 00:08:07 +0000 (10:08 +1000)]
drivers/rtc/rtc-pcf8563.c: use PTR_RET()

Use of PTR_RET() simplifies the code.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-m48t35.c: use PTR_RET()
Sachin Kamat [Wed, 19 Jun 2013 00:08:07 +0000 (10:08 +1000)]
drivers/rtc/rtc-m48t35.c: use PTR_RET()

Use of PTR_RET() simplifies the code.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-isl12022.c: use PTR_RET()
Sachin Kamat [Wed, 19 Jun 2013 00:08:06 +0000 (10:08 +1000)]
drivers/rtc/rtc-isl12022.c: use PTR_RET()

Use of PTR_RET() simplifies the code.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Cc: Roman Fietze <roman.fietze@telemotive.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-da9052.c: use PTR_RET()
Sachin Kamat [Wed, 19 Jun 2013 00:08:06 +0000 (10:08 +1000)]
drivers/rtc/rtc-da9052.c: use PTR_RET()

Use of PTR_RET() simplifies the code.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Cc: David Dajun Chen <dchen@diasemi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-mpc5121.c: use platform_{get,set}_drvdata()
Jingoo Han [Wed, 19 Jun 2013 00:08:06 +0000 (10:08 +1000)]
drivers/rtc/rtc-mpc5121.c: use platform_{get,set}_drvdata()

Use the wrapper functions for getting and setting the driver data using
platform_device instead of using dev_{get,set}_drvdata() with &pdev->dev,
so we can directly pass a struct platform_device.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Cc: Grant Likely <grant.likely@linaro.org>
Cc: Rob Herring <rob.herring@calxeda.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-ds1302.c: handle write protection
Sergey Yanovich [Wed, 19 Jun 2013 00:08:06 +0000 (10:08 +1000)]
drivers/rtc/rtc-ds1302.c: handle write protection

This chip has a control register and can prevent altering saved clock.
Without this patch we could have:

(arm)root@pac14:~# date
Tue May 21 03:08:27 MSK 2013
(arm)root@pac14:~# /etc/init.d/hwclock.sh show
Tue May 21 11:13:58 2013  -0.067322 seconds
(arm)root@pac14:~# /etc/init.d/hwclock.sh stop
[info] Saving the system clock.
[info] Hardware Clock updated to Tue May 21 03:09:01 MSK 2013.
(arm)root@pac14:~# /etc/init.d/hwclock.sh show
Tue May 21 11:14:15 2013  -0.624272 seconds

The patch enables write access to rtc before the driver tries to write
time and re-disables when time data is written.

Signed-off-by: Sergey Yanovich <ynvich@gmail.com>
Acked-by: Marc Zyngier <maz@misterjones.org>
Cc: Alessandro Zummo <a.zummo@towertech.it>
Cc: Sachin Kamat <sachin.kamat@linaro.org>
Cc: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-ab8500.c: add second resolution to rtc driver
Alexandre Torgue [Wed, 19 Jun 2013 00:08:05 +0000 (10:08 +1000)]
drivers/rtc/rtc-ab8500.c: add second resolution to rtc driver

Android expects the RTC to have second resolution.  On ab8540 cut2 RTC
block has a new register which allows setting seconds for wakeup alarms.

Existing registers (minutes hi, mid and low) have seen their offsets
changed. Here is the new mapping:
  * AlarmSec     (A) 0x22
  * AlarmMinLow  (M) from 0x8 to 0x23
  * AlarmMinMid  (M) from 0x9 to 0x24
  * AlarmMinHigh (M) from 0xA to 0x25

Signed-off-by: Julien Delacou <julien.delacou@stericsson.com>
Signed-off-by: Alexandre Torgue <alexandre.torgue@st.com>
Acked-by: Lee Jones <lee.jones@linaro.org>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Cc: Samuel Ortiz <sameo@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-wm831x.c: remove empty function
Sachin Kamat [Wed, 19 Jun 2013 00:08:05 +0000 (10:08 +1000)]
drivers/rtc/rtc-wm831x.c: remove empty function

After the switch to devm_* functions and the removal of
rtc_device_unregister(), the 'remove' function does not do anything.
Delete it.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-tps80031.c: remove empty function
Sachin Kamat [Wed, 19 Jun 2013 00:08:05 +0000 (10:08 +1000)]
drivers/rtc/rtc-tps80031.c: remove empty function

After the switch to devm_* functions and the removal of
rtc_device_unregister(), the 'remove' function does not do anything.
Delete it.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Cc: Laxman Dewangan <ldewangan@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-sun4v.c: remove empty function
Sachin Kamat [Wed, 19 Jun 2013 00:08:04 +0000 (10:08 +1000)]
drivers/rtc/rtc-sun4v.c: remove empty function

After the switch to devm_* functions and the removal of
rtc_device_unregister(), the 'remove' function does not do anything.
Delete it.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-starfire.c: remove empty function
Sachin Kamat [Wed, 19 Jun 2013 00:08:04 +0000 (10:08 +1000)]
drivers/rtc/rtc-starfire.c: remove empty function

After the switch to devm_* functions and the removal of
rtc_device_unregister(), the 'remove' function does not do anything.
Delete it.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-snvs.c: remove empty function
Sachin Kamat [Wed, 19 Jun 2013 00:08:04 +0000 (10:08 +1000)]
drivers/rtc/rtc-snvs.c: remove empty function

After the switch to devm_* functions and the removal of
rtc_device_unregister(), the 'remove' function does not do anything.
Delete it.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-rx8581.c: remove empty function
Sachin Kamat [Wed, 19 Jun 2013 00:08:03 +0000 (10:08 +1000)]
drivers/rtc/rtc-rx8581.c: remove empty function

After the switch to devm_* functions and the removal of
rtc_device_unregister(), the 'remove' function does not do anything.
Delete it.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Cc: Martyn Welch <martyn.welch@ge.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-rs5c348.c: remove empty function
Sachin Kamat [Wed, 19 Jun 2013 00:08:03 +0000 (10:08 +1000)]
drivers/rtc/rtc-rs5c348.c: remove empty function

After the switch to devm_* functions and the removal of
rtc_device_unregister(), the 'remove' function does not do anything.
Delete it.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Acked-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-rx4581.c: remove empty function
Sachin Kamat [Wed, 19 Jun 2013 00:08:03 +0000 (10:08 +1000)]
drivers/rtc/rtc-rx4581.c: remove empty function

After the switch to devm_* functions and the removal of
rtc_device_unregister(), the 'remove' function does not do anything.
Delete it.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Cc: Torben Hohn <torbenh@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-rv3029c2.c: remove empty function
Sachin Kamat [Wed, 19 Jun 2013 00:08:02 +0000 (10:08 +1000)]
drivers/rtc/rtc-rv3029c2.c: remove empty function

After the switch to devm_* functions and the removal of
rtc_device_unregister(), the 'remove' function does not do anything.
Delete it.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Cc: Gregory Hermant <gregory.hermant@calao-systems.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
11 years agodrivers/rtc/rtc-rs5c313.c: remove empty function
Sachin Kamat [Wed, 19 Jun 2013 00:08:02 +0000 (10:08 +1000)]
drivers/rtc/rtc-rs5c313.c: remove empty function

After the switch to devm_* functions and the removal of
rtc_device_unregister(), the 'remove' function does not do anything.
Delete it.

Signed-off-by: Sachin Kamat <sachin.kamat@linaro.org>
Cc: Nobuhiro Iwamatsu <iwamatsu@nigauri.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>