GitHub/LineageOS/android_kernel_motorola_exynos9610.git
11 years agoclockevents: Switch into oneshot mode even if broadcast registered late
Stephen Boyd [Wed, 17 Apr 2013 17:26:06 +0000 (10:26 -0700)]
clockevents: Switch into oneshot mode even if broadcast registered late

tick_oneshot_notify() is used to notify a particular CPU to try
to switch into oneshot mode after a oneshot capable tick device
is registered and tick_clock_notify() is used to notify all CPUs
to try to switch into oneshot mode after a high res clocksource
is registered. There is one caveat; if the tick devices suffer
from FEAT_C3_STOP we don't try to switch into oneshot mode unless
we have a oneshot capable broadcast device already registered.

If the broadcast device is registered after the tick devices that
have FEAT_C3_STOP we'll never try to switch into oneshot mode
again, causing us to be stuck in periodic mode forever. Avoid
this scenario by calling tick_clock_notify() after we register
the broadcast device so that we try to switch into oneshot mode
on all CPUs one more time.

[ tglx: Adopted to timers/core and added a comment ]

Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Link: http://lkml.kernel.org/r/1366219566-29783-1-git-send-email-sboyd@codeaurora.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
11 years agotimer_list: Convert timer list to be a proper seq_file
Nathan Zimmer [Wed, 27 Mar 2013 00:56:30 +0000 (19:56 -0500)]
timer_list: Convert timer list to be a proper seq_file

When running with 4096 cores attemping to read /proc/timer_list will fail
with an ENOMEM condition.  On a sufficantly large systems the total amount
of data is more then 4mb, so it won't fit into a single buffer.  The
failure can also occur on smaller systems when memory fragmentation is
high as reported by Dave Jones.

Convert /proc/timer_list to a proper seq_file with its own iterator.  This
is a little more complex given that we have to make two passes with two
separate headers.

sysrq_timer_list_show also needed to be updated to reflect the fact that
now timer_list_show only does one cpu at at time.

Signed-off-by: Nathan Zimmer <nzimmer@sgi.com>
Reported-by: Dave Jones <davej@redhat.com>
Cc: John Stultz <johnstul@us.ibm.com>
Cc: Stephen Boyd <sboyd@codeaurora.org>
Link: http://lkml.kernel.org/r/1364345790-14577-3-git-send-email-nzimmer@sgi.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
11 years agotimer_list: Split timer_list_show_tickdevices
Nathan Zimmer [Wed, 27 Mar 2013 00:56:29 +0000 (19:56 -0500)]
timer_list: Split timer_list_show_tickdevices

Split timer_list_show_tickdevices() into the header printout and pull
the rest up to timer_list_show. This is a preparatory patch for
converting timer_list to a proper seqfile with its own iterator

Signed-off-by: Nathan Zimmer <nzimmer@sgi.com>
Reported-by: Dave Jones <davej@redhat.com>
Cc: John Stultz <johnstul@us.ibm.com>
Cc: Stephen Boyd <sboyd@codeaurora.org>
Link: http://lkml.kernel.org/r/1364345790-14577-2-git-send-email-nzimmer@sgi.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
11 years agoposix-timers: Show sigevent info in proc file
Pavel Emelyanov [Mon, 11 Mar 2013 09:13:08 +0000 (13:13 +0400)]
posix-timers: Show sigevent info in proc file

Previous patch added proc file to list posix timers created by task.
Expand the information provided in this file by adding info about
notification method, with which timers were created. I.e. after
the "ID:" line there go

1. "signal:" line, that shows signal number and sigval bits;
2. "notify:" line, that shows the timer notification method.

Thus the timer entry would looke like this:

ID: 123
signal: 14/0000000000b005d0
notify: signal/pid.732

This information is enough to understand how timer_create() was called
for each particular timer.

Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Cc: Matthew Helsley <matt.helsley@gmail.com>
Link: http://lkml.kernel.org/r/513DA024.80404@parallels.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
11 years agoposix-timers: Introduce /proc/PID/timers file
Pavel Emelyanov [Mon, 11 Mar 2013 09:12:45 +0000 (13:12 +0400)]
posix-timers: Introduce /proc/PID/timers file

Currently kernel doesn't provide any API for getting info about what
posix timers are configured by processes. It's implied, that a process
which configured some timers, knows what it did. However, for external
tools it's impossible to get this information. In particular, this is
critical for checkpoint-restore project to have this info.

Introduce a per-pid proc file with information about posix
timers. Since these timers are shared between threads, this file is
present on tgid level only, no such thing in tid subdirs.

The file format is expected to be the "/proc/<pid>/smaps"-like,
i.e. each timer will occupy seveal lines to allow for future
extending.

Each new timer entry starts with the

ID: <number>

line which is added by this patch.

Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Cc: Matthew Helsley <matt.helsley@gmail.com>
Link: http://lkml.kernel.org/r/513DA00D.6070009@parallels.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
11 years agoposix timers: Allocate timer id per process (v2)
Pavel Emelyanov [Mon, 11 Mar 2013 09:12:21 +0000 (13:12 +0400)]
posix timers: Allocate timer id per process (v2)

Currently kernel generates IDs for posix timers in a global manner --
there's a kernel-wide IDR tree from which IDs are created. This makes
it impossible to recreate a timer with a desired ID (in particular
this is done by the CRIU checkpoint-restore project) -- since these
IDs are global it may happen, that at the time we recreate a timer, the
ID we want for it is already busy by some other timer.

In order to address this, replace the IDR tree with a global hash
table for timers and makes timer IDs unique per signal_struct (to
which timers are linked anyway). With this, two timers belonging to
different processes may have equal IDs and we can recreate either of
them with the ID we want.

Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Cc: Matthew Helsley <matt.helsley@gmail.com>
Link: http://lkml.kernel.org/r/513D9FF5.9010004@parallels.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
11 years agotimekeeping: Make sure to notify hrtimers when TAI offset changes
John Stultz [Wed, 10 Apr 2013 19:41:49 +0000 (12:41 -0700)]
timekeeping: Make sure to notify hrtimers when TAI offset changes

Now that we have CLOCK_TAI timers, make sure we notify hrtimer
code when TAI offset is changed.

Signed-off-by: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/1365622909-953-1-git-send-email-john.stultz@linaro.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
11 years agohrtimer: Fix ktime_add_ns() overflow on 32bit architectures
David Engraf [Tue, 19 Mar 2013 12:29:55 +0000 (13:29 +0100)]
hrtimer: Fix ktime_add_ns() overflow on 32bit architectures

One can trigger an overflow when using ktime_add_ns() on a 32bit
architecture not supporting CONFIG_KTIME_SCALAR.

When passing a very high value for u64 nsec, e.g. 7881299347898368000
the do_div() function converts this value to seconds (7881299347) which
is still to high to pass to the ktime_set() function as long. The result
in is a negative value.

The problem on my system occurs in the tick-sched.c,
tick_nohz_stop_sched_tick() when time_delta is set to
timekeeping_max_deferment(). The check for time_delta < KTIME_MAX is
valid, thus ktime_add_ns() is called with a too large value resulting in
a negative expire value. This leads to an endless loop in the ticker code:

time_delta: 7881299347898368000
expires = ktime_add_ns(last_update, time_delta)
expires: negative value

This fix caps the value to KTIME_MAX.

This error doesn't occurs on 64bit or architectures supporting
CONFIG_KTIME_SCALAR (e.g. ARM, x86-32).

Cc: stable@vger.kernel.org
Signed-off-by: David Engraf <david.engraf@sysgo.com>
[jstultz: Minor tweaks to commit message & header]
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agohrtimer: Add expiry time overflow check in hrtimer_interrupt
Prarit Bhargava [Mon, 8 Apr 2013 12:47:15 +0000 (08:47 -0400)]
hrtimer: Add expiry time overflow check in hrtimer_interrupt

The settimeofday01 test in the LTP testsuite effectively does

        gettimeofday(current time);
        settimeofday(Jan 1, 1970 + 100 seconds);
        settimeofday(current time);

This test causes a stack trace to be displayed on the console during the
setting of timeofday to Jan 1, 1970 + 100 seconds:

[  131.066751] ------------[ cut here ]------------
[  131.096448] WARNING: at kernel/time/clockevents.c:209 clockevents_program_event+0x135/0x140()
[  131.104935] Hardware name: Dinar
[  131.108150] Modules linked in: sg nfsv3 nfs_acl nfsv4 auth_rpcgss nfs dns_resolver fscache lockd sunrpc nf_conntrack_netbios_ns nf_conntrack_broadcast ipt_MASQUERADE ip6table_mangle ip6t_REJECT nf_conntrack_ipv6 nf_defrag_ipv6 iptable_nat nf_nat_ipv4 nf_nat iptable_mangle ipt_REJECT nf_conntrack_ipv4 nf_defrag_ipv4 xt_conntrack nf_conntrack ebtable_filter ebtables ip6table_filter ip6_tables iptable_filter ip_tables kvm_amd kvm sp5100_tco bnx2 i2c_piix4 crc32c_intel k10temp fam15h_power ghash_clmulni_intel amd64_edac_mod pcspkr serio_raw edac_mce_amd edac_core microcode xfs libcrc32c sr_mod sd_mod cdrom ata_generic crc_t10dif pata_acpi radeon i2c_algo_bit drm_kms_helper ttm drm ahci pata_atiixp libahci libata usb_storage i2c_core dm_mirror dm_region_hash dm_log dm_mod
[  131.176784] Pid: 0, comm: swapper/28 Not tainted 3.8.0+ #6
[  131.182248] Call Trace:
[  131.184684]  <IRQ>  [<ffffffff810612af>] warn_slowpath_common+0x7f/0xc0
[  131.191312]  [<ffffffff8106130a>] warn_slowpath_null+0x1a/0x20
[  131.197131]  [<ffffffff810b9fd5>] clockevents_program_event+0x135/0x140
[  131.203721]  [<ffffffff810bb584>] tick_program_event+0x24/0x30
[  131.209534]  [<ffffffff81089ab1>] hrtimer_interrupt+0x131/0x230
[  131.215437]  [<ffffffff814b9600>] ? cpufreq_p4_target+0x130/0x130
[  131.221509]  [<ffffffff81619119>] smp_apic_timer_interrupt+0x69/0x99
[  131.227839]  [<ffffffff8161805d>] apic_timer_interrupt+0x6d/0x80
[  131.233816]  <EOI>  [<ffffffff81099745>] ? sched_clock_cpu+0xc5/0x120
[  131.240267]  [<ffffffff814b9ff0>] ? cpuidle_wrap_enter+0x50/0xa0
[  131.246252]  [<ffffffff814b9fe9>] ? cpuidle_wrap_enter+0x49/0xa0
[  131.252238]  [<ffffffff814ba050>] cpuidle_enter_tk+0x10/0x20
[  131.257877]  [<ffffffff814b9c89>] cpuidle_idle_call+0xa9/0x260
[  131.263692]  [<ffffffff8101c42f>] cpu_idle+0xaf/0x120
[  131.268727]  [<ffffffff815f8971>] start_secondary+0x255/0x257
[  131.274449] ---[ end trace 1151a50552231615 ]---

When we change the system time to a low value like this, the value of
timekeeper->offs_real will be a negative value.

It seems that the WARN occurs because an hrtimer has been started in the time
between the releasing of the timekeeper lock and the IPI call (via a call to
on_each_cpu) in clock_was_set() in the do_settimeofday() code.  The end result
is that a REALTIME_CLOCK timer has been added with softexpires = expires =
KTIME_MAX.  The hrtimer_interrupt() fires/is called and the loop at
kernel/hrtimer.c:1289 is executed.  In this loop the code subtracts the
clock base's offset (which was set to timekeeper->offs_real in
do_settimeofday()) from the current hrtimer_cpu_base->expiry value (which
was KTIME_MAX):

KTIME_MAX - (a negative value) = overflow

A simple check for an overflow can resolve this problem.  Using KTIME_MAX
instead of the overflow value will result in the hrtimer function being run,
and the reprogramming of the timer after that.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@vger.kernel.org
Reviewed-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Prarit Bhargava <prarit@redhat.com>
[jstultz: Tweaked commit subject]
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agotimekeeping: Shorten seq_count region
Thomas Gleixner [Thu, 21 Feb 2013 22:51:40 +0000 (22:51 +0000)]
timekeeping: Shorten seq_count region

Shorten the seqcount write hold region to the actual update of the
timekeeper and the related data (e.g vsyscall).

On a contemporary x86 system this reduces the maximum latencies on
Preempt-RT from 8us to 4us on the non-timekeeping cores.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agotimekeeping: Implement a shadow timekeeper
Thomas Gleixner [Thu, 21 Feb 2013 22:51:40 +0000 (22:51 +0000)]
timekeeping: Implement a shadow timekeeper

Use the shadow timekeeper to do the update_wall_time() adjustments and
then copy it over to the real timekeeper.

Keep the shadow timekeeper in sync when updating stuff outside of
update_wall_time().

This allows us to limit the timekeeper_seq hold time to the update of
the real timekeeper and the vsyscall data in the next patch.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agotimekeeping: Delay update of clock->cycle_last
Thomas Gleixner [Thu, 21 Feb 2013 22:51:39 +0000 (22:51 +0000)]
timekeeping: Delay update of clock->cycle_last

For calculating the new timekeeper values store the new cycle_last
value in the timekeeper and update the clock->cycle_last just when we
actually update the new values.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agotimekeeping: Store cycle_last value in timekeeper struct as well
Thomas Gleixner [Thu, 21 Feb 2013 22:51:38 +0000 (22:51 +0000)]
timekeeping: Store cycle_last value in timekeeper struct as well

For implementing a shadow timekeeper and a split calculation/update
region we need to store the cycle_last value in the timekeeper and
update the value in the clocksource struct only in the update region.

Add the extra storage to the timekeeper.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agontp: Remove ntp_lock, using the timekeeping locks to protect ntp state
John Stultz [Fri, 22 Mar 2013 18:52:03 +0000 (11:52 -0700)]
ntp: Remove ntp_lock, using the timekeeping locks to protect ntp state

In order to properly handle the NTP state in future changes to the
timekeeping lock management, this patch moves the management of
all of the ntp state under the timekeeping locks.

This allows us to remove the ntp_lock.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Richard Cochran <richardcochran@gmail.com>
Cc: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agotimekeeping: Simplify tai updating from do_adjtimex
John Stultz [Fri, 22 Mar 2013 21:20:03 +0000 (14:20 -0700)]
timekeeping: Simplify tai updating from do_adjtimex

Since we are taking the timekeeping locks, just go ahead
and update any tai change directly, rather then dropping
the lock and calling a function that will just take it again.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Richard Cochran <richardcochran@gmail.com>
Cc: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agotimekeeping: Hold timekeepering locks in do_adjtimex and hardpps
John Stultz [Fri, 22 Mar 2013 18:37:28 +0000 (11:37 -0700)]
timekeeping: Hold timekeepering locks in do_adjtimex and hardpps

In moving the NTP state to be protected by the timekeeping locks,
be sure to acquire the timekeeping locks prior to calling
ntp functions.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Richard Cochran <richardcochran@gmail.com>
Cc: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agotimekeeping: Move ADJ_SETOFFSET to top level do_adjtimex()
John Stultz [Fri, 22 Mar 2013 22:04:13 +0000 (15:04 -0700)]
timekeeping: Move ADJ_SETOFFSET to top level do_adjtimex()

Since ADJ_SETOFFSET adjusts the timekeeping state, process
it as part of the top level do_adjtimex() function in
timekeeping.c.

This avoids deadlocks that could occur once we change the
ntp locking rules.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Richard Cochran <richardcochran@gmail.com>
Cc: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agontp: Rework do_adjtimex to take timespec and tai arguments
John Stultz [Fri, 22 Mar 2013 19:28:15 +0000 (12:28 -0700)]
ntp: Rework do_adjtimex to take timespec and tai arguments

In order to change the locking rules, we need to provide
the timespec and tai values rather then having the ntp
logic acquire these values itself.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Richard Cochran <richardcochran@gmail.com>
Cc: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agontp: Move timex validation to timekeeping do_adjtimex call.
John Stultz [Fri, 22 Mar 2013 19:08:52 +0000 (12:08 -0700)]
ntp: Move timex validation to timekeeping do_adjtimex call.

Move logic that does not need the ntp state to be done
in the timekeeping do_adjtimex() call.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Richard Cochran <richardcochran@gmail.com>
Cc: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agontp: Move do_adjtimex() and hardpps() functions to timekeeping.c
John Stultz [Fri, 22 Mar 2013 18:31:29 +0000 (11:31 -0700)]
ntp: Move do_adjtimex() and hardpps() functions to timekeeping.c

In preparation for changing the ntp locking rules, move
do_adjtimex and hardpps accessor functions to timekeeping.c,
but keep the code logic in ntp.c.

This patch also introduces a ntp_internal.h file so timekeeping
specific interfaces of ntp.c can be more limitedly shared with
timekeeping.c.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Richard Cochran <richardcochran@gmail.com>
Cc: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agontp: Split out timex validation from do_adjtimex
John Stultz [Fri, 22 Mar 2013 18:59:04 +0000 (11:59 -0700)]
ntp: Split out timex validation from do_adjtimex

Split out the timex validation done in do_adjtimex into a separate
function. This will help simplify logic in following patches.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Richard Cochran <richardcochran@gmail.com>
Cc: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agoMerge branch 'fortglx/3.10/time' of git://git.linaro.org/people/jstultz/linux into...
Thomas Gleixner [Wed, 3 Apr 2013 10:27:29 +0000 (12:27 +0200)]
Merge branch 'fortglx/3.10/time' of git://git.linaro.org/people/jstultz/linux into timers/core

11 years agoARM: bcm281xx: Add timer driver (driver portion)
Christian Daudt [Wed, 13 Mar 2013 21:27:27 +0000 (14:27 -0700)]
ARM: bcm281xx: Add timer driver (driver portion)

This adds support for the Broadcom timer, used in the following SoCs:
BCM11130, BCM11140, BCM11351, BCM28145, BCM28155

Updates from V6:
- Split DT portion into a separate patch

Updates from V5:
- Rebase to latest arm-soc/for-next

Updates from V4:
- Switch code to use CLOCKSOURCE_OF_DECLARE

Updates from V3:
- Migrate to 3.9 timer framework updates

Updates from V2:
- prepend static fns + fields with kona_

Updates from V1:
- Rename bcm_timer.c to bcm_kona_timer.c
- Pull .h into bcm_kona_timer.c
- Make timers static
- Clean up comment block
- Switched to using clockevents_config_and_register
- Added an error to the get_timer loop if it repeats too much
- Added to Documentation/devicetree/bindings/arm/bcm/bcm,kona-timer.txt
- Added missing readl to timer_disable_and_clear

Note: bcm,kona-timer was kept as the 'compatible' field to make it
specific enough for when there are multiple bcm timers (bcm,timer is
too generic).

Signed-off-by: Christian Daudt <csd@broadcom.com>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: John Stultz <john.stultz@linaro.org>
Reviewed-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agotimekeeping: __timekeeping_set_tai_offset can be static
Fengguang Wu [Mon, 25 Mar 2013 19:24:24 +0000 (12:24 -0700)]
timekeeping: __timekeeping_set_tai_offset can be static

Yet again, the kbuild test robot saves the day, noting
I left out defining __timekeeping_set_tai_offset as
static. It even sent me this patch.

Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agotick: Change log level of NOHZ: local_softirq_pending message
Rado Vrbovsky [Fri, 8 Feb 2013 17:37:30 +0000 (12:37 -0500)]
tick: Change log level of NOHZ: local_softirq_pending message

The "NOHZ: local_softirq_pending" message is a largely informational
message.  This makes extra work for customers that have a policy of
investigating all kernel log messages logged at <= KERN_ERR log level.
This patch sets the message to a different log level.

[ tglx: Use pr_warn() ]

Signed-off-by: Rado Vrbovsky <rvrbovsk@redhat.com>
Cc: Don Zickus <dzickus@redhat.com>
Link: http://lkml.kernel.org/r/2037057938.893524.1360345050772.JavaMail.root@redhat.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
11 years agohrtimer/trivial: Fix comment text in hrtimer.c
David Daney [Wed, 13 Mar 2013 19:20:38 +0000 (12:20 -0700)]
hrtimer/trivial: Fix comment text in hrtimer.c

The comments mention HRTIMER_ABS and HRTIMER_REL, these symbols don't
exist, the proper names are HRTIMER_MODE_ABS and HRTIMER_MODE_REL.

Signed-off-by: David Daney <david.daney@cavium.com>
Cc: Jiri Kosina <trivial@kernel.org>
Link: http://lkml.kernel.org/r/1363202438-21234-1-git-send-email-ddaney.cavm@gmail.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
11 years agotimekeeping: Split timekeeper_lock into lock and seqcount
Thomas Gleixner [Thu, 21 Feb 2013 22:51:38 +0000 (22:51 +0000)]
timekeeping: Split timekeeper_lock into lock and seqcount

We want to shorten the seqcount write hold time. So split the seqlock
into a lock and a seqcount.

Open code the seqwrite_lock in the places which matter and drop the
sequence counter update where it's pointless.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
[jstultz: Merge fixups from CLOCK_TAI collisions]
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agotimekeeping: Move lock out of timekeeper struct
Thomas Gleixner [Thu, 21 Feb 2013 22:51:37 +0000 (22:51 +0000)]
timekeeping: Move lock out of timekeeper struct

Make the lock a separate entity. Preparatory patch for shadow
timekeeper structure.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
[Merged with CLOCK_TAI changes]
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agotimekeeping: Make jiffies_lock internal
Thomas Gleixner [Thu, 21 Feb 2013 22:51:36 +0000 (22:51 +0000)]
timekeeping: Make jiffies_lock internal

Nothing outside of the timekeeping core needs that lock.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agotimekeeping: Calc stuff once
Thomas Gleixner [Thu, 21 Feb 2013 22:51:36 +0000 (22:51 +0000)]
timekeeping: Calc stuff once

Calculate the cycle interval shifted value once. No functional change,
just makes the code more readable.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agohrtimer: Add hrtimer support for CLOCK_TAI
John Stultz [Tue, 22 Jan 2013 01:00:11 +0000 (17:00 -0800)]
hrtimer: Add hrtimer support for CLOCK_TAI

Add hrtimer support for CLOCK_TAI, as well as posix timer interfaces.

Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agotimekeeping: Add CLOCK_TAI clockid
John Stultz [Thu, 3 May 2012 19:43:40 +0000 (12:43 -0700)]
timekeeping: Add CLOCK_TAI clockid

This add a CLOCK_TAI clockid and the needed accessors.

CC: Thomas Gleixner <tglx@linutronix.de>
CC: Eric Dumazet <eric.dumazet@gmail.com>
CC: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agotimekeeping: Move TAI managment into timekeeping core from ntp
John Stultz [Thu, 3 May 2012 19:30:07 +0000 (12:30 -0700)]
timekeeping: Move TAI managment into timekeeping core from ntp

Currently NTP manages the TAI offset. Since there's plans for a
CLOCK_TAI clockid, push the TAI management into the timekeeping
core.

CC: Thomas Gleixner <tglx@linutronix.de>
CC: Eric Dumazet <eric.dumazet@gmail.com>
CC: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agoclockevents: Add missing tick_check_broadcast_expired() for CLOCKEVENTS=n
Thomas Gleixner [Fri, 22 Mar 2013 09:48:33 +0000 (10:48 +0100)]
clockevents: Add missing tick_check_broadcast_expired() for CLOCKEVENTS=n

Fengs build robot reports:

arch/arm/kernel/process.c: In function 'cpu_idle':
arch/arm/kernel/process.c:211:4: error: implicit declaration of function
'tick_check_broadcast_expired' [-Werror=implicit-function-declaration]

Add the missing inline function for non clockevent builds

Reported-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
11 years agotimekeeping: utilize the suspend-nonstop clocksource to count suspended time
Feng Tang [Tue, 12 Mar 2013 03:56:48 +0000 (11:56 +0800)]
timekeeping: utilize the suspend-nonstop clocksource to count suspended time

There are some new processors whose TSC clocksource won't stop during
suspend. Currently, after system resumes, kernel will use persistent
clock or RTC to compensate the sleep time, but with these nonstop
clocksources, we could skip the special compensation from external
sources, and just use current clocksource for time recounting.

This can solve some time drift bugs caused by some not-so-accurate or
error-prone RTC devices.

The current way to count suspended time is first try to use the persistent
clock, and then try the RTC if persistent clock can't be used. This
patch will change the trying order to:
suspend-nonstop clocksource -> persistent clock -> RTC

When counting the sleep time with nonstop clocksource, use an accurate way
suggested by Jason Gunthorpe to cover very large delta cycles.

Signed-off-by: Feng Tang <feng.tang@intel.com>
[jstultz: Small optimization, avoiding re-reading the clocksource]
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agox86: tsc: Add support for new S3_NONSTOP feature
Feng Tang [Tue, 12 Mar 2013 03:56:47 +0000 (11:56 +0800)]
x86: tsc: Add support for new S3_NONSTOP feature

Add support for new S3_NONSTOP feature

Signed-off-by: Feng Tang <feng.tang@intel.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agoclocksource: Add new feature flag CLOCK_SOURCE_SUSPEND_NONSTOP
Feng Tang [Tue, 12 Mar 2013 03:56:46 +0000 (11:56 +0800)]
clocksource: Add new feature flag CLOCK_SOURCE_SUSPEND_NONSTOP

Some x86 processors have a TSC clocksource, which continues to run
even when system is suspended. Also most OMAP platforms have a
32 KHz timer which has similar capability. Add a feature flag so that
it could be utilized.

Signed-off-by: Feng Tang <feng.tang@intel.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agox86: Add cpu capability flag X86_FEATURE_NONSTOP_TSC_S3
Feng Tang [Tue, 12 Mar 2013 03:56:45 +0000 (11:56 +0800)]
x86: Add cpu capability flag X86_FEATURE_NONSTOP_TSC_S3

On some new Intel Atom processors (Penwell and Cloverview), there is
a feature that the TSC won't stop in S3 state, say the TSC value
won't be reset to 0 after resume. This feature makes TSC a more reliable
clocksource and could benefit the timekeeping code during system
suspend/resume cycle, so add a flag for it.

Signed-off-by: Feng Tang <feng.tang@intel.com>
[jstultz: Fix checkpatch warning]
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agox86: Do full rtc synchronization with ntp
Prarit Bhargava [Thu, 14 Feb 2013 17:02:54 +0000 (12:02 -0500)]
x86: Do full rtc synchronization with ntp

Every 11 minutes ntp attempts to update the x86 rtc with the current
system time.  Currently, the x86 code only updates the rtc if the system
time is within +/-15 minutes of the current value of the rtc. This
was done originally to avoid setting the RTC if the RTC was in localtime
mode (common with Windows dualbooting).  Other architectures do a full
synchronization and now that we have better infrastructure to detect
when the RTC is in localtime, there is no reason that x86 should be
software limited to a 30 minute window.

This patch changes the behavior of the kernel to do a full synchronization
(year, month, day, hour, minute, and second) of the rtc when ntp requests
a synchronization between the system time and the rtc.

I've used the RTC library functions in this patchset as they do all the
required bounds checking.

Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: John Stultz <john.stultz@linaro.org>
Cc: x86@kernel.org
Cc: Matt Fleming <matt.fleming@intel.com>
Cc: David Vrabel <david.vrabel@citrix.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: linux-efi@vger.kernel.org
Signed-off-by: Prarit Bhargava <prarit@redhat.com>
[jstultz: Tweak commit message, fold in build fix found by fengguang
Also add select RTC_LIB to X86, per new dependency, as found by prarit]
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agotimekeeping: Use inject_offset in warp_clock
John Stultz [Fri, 22 Feb 2013 20:33:29 +0000 (12:33 -0800)]
timekeeping: Use inject_offset in warp_clock

When warping the clock (from a local time RTC), use
timekeeping_inject_offset() to atomically add the offset.

This avoids any minor time error caused by the delay between
reading the time, and then setting the adjusted time.

Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agotimekeeping: Avoid adjust kernel time once hwclock kept in UTC time
Dong Zhu [Thu, 6 Dec 2012 14:03:34 +0000 (22:03 +0800)]
timekeeping: Avoid adjust kernel time once hwclock kept in UTC time

If the Hardware Clock kept in local time,kernel will adjust the time
to be UTC time.But if Hardware Clock kept in UTC time,system will make
a dummy settimeofday call first (sys_tz.tz_minuteswest = 0) to make sure
the time is not shifted,so at this point I think maybe it is not necessary
to set the kernel time once the sys_tz.tz_minuteswest is zero.

Signed-off-by: Dong Zhu <bluezhudong@gmail.com>
[jstultz: Updated to merge with conflicting changes ]
Signed-off-by: John Stultz <john.stultz@linaro.org>
11 years agox86: Use tick broadcast expired check
Thomas Gleixner [Wed, 6 Mar 2013 11:18:37 +0000 (11:18 +0000)]
x86: Use tick broadcast expired check

Avoid going back into deep idle if the tick broadcast IPI is about to
fire.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Arjan van de Veen <arjan@infradead.org>
Cc: x86@kernel.org
Link: http://lkml.kernel.org/r/20130306111537.702278273@linutronix.de
11 years agoarm: Use tick broadcast expired check
Thomas Gleixner [Wed, 6 Mar 2013 11:18:37 +0000 (11:18 +0000)]
arm: Use tick broadcast expired check

Avoid going back into deep idle if the tick broadcast IPI is about to
fire.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: LAK <linux-arm-kernel@lists.infradead.org>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Arjan van de Veen <arjan@infradead.org>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Tested-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Cc: Jason Liu <liu.h.jason@gmail.com>
Link: http://lkml.kernel.org/r/20130306111537.640722922@linutronix.de
11 years agotick: Provide a check for a forced broadcast pending
Thomas Gleixner [Wed, 6 Mar 2013 11:18:36 +0000 (11:18 +0000)]
tick: Provide a check for a forced broadcast pending

On the CPU which gets woken along with the target CPU of the broadcast
the following happens:

  deep_idle()
<-- spurious wakeup
  broadcast_exit()
    set forced bit

  enable interrupts

<-- Nothing happens

  disable interrupts

  broadcast_enter()
<-- Here we observe the forced bit is set
  deep_idle()

Now after that the target CPU of the broadcast runs the broadcast
handler and finds the other CPU in both the broadcast and the forced
mask, sends the IPI and stuff gets back to normal.

So it's not actually harmful, just more evidence for the theory, that
hardware designers have access to very special drug supplies.

Now there is no point in going back to deep idle just to wake up again
right away via an IPI. Provide a check which allows the idle code to
avoid the deep idle transition.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: LAK <linux-arm-kernel@lists.infradead.org>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Arjan van de Veen <arjan@infradead.org>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Tested-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Cc: Jason Liu <liu.h.jason@gmail.com>
Link: http://lkml.kernel.org/r/20130306111537.565418308@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
11 years agotick: Handle broadcast wakeup of multiple cpus
Thomas Gleixner [Wed, 6 Mar 2013 11:18:35 +0000 (11:18 +0000)]
tick: Handle broadcast wakeup of multiple cpus

Some brilliant hardware implementations wake multiple cores when the
broadcast timer fires. This leads to the following interesting
problem:

CPU0 CPU1
wakeup from idle wakeup from idle

leave broadcast mode leave broadcast mode
 restart per cpu timer  restart per cpu timer
          go back to idle
handle broadcast
 (empty mask)
enter broadcast mode
programm broadcast device
enter broadcast mode
programm broadcast device

So what happens is that due to the forced reprogramming of the cpu
local timer, we need to set a event in the future. Now if we manage to
go back to idle before the timer fires, we switch off the timer and
arm the broadcast device with an already expired time (covered by
forced mode). So in the worst case we repeat the above ping pong
forever.

Unfortunately we have no information about what caused the wakeup, but
we can check current time against the expiry time of the local cpu. If
the local event is already in the past, we know that the broadcast
timer is about to fire and send an IPI. So we mark ourself as an IPI
target even if we left broadcast mode and avoid the reprogramming of
the local cpu timer.

This still leaves the possibility that a CPU which is not handling the
broadcast interrupt is going to reach idle again before the IPI
arrives. This can't be solved in the core code and will be handled in
follow up patches.

Reported-by: Jason Liu <liu.h.jason@gmail.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: LAK <linux-arm-kernel@lists.infradead.org>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Arjan van de Veen <arjan@infradead.org>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Tested-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Link: http://lkml.kernel.org/r/20130306111537.492045206@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
11 years agotick: Avoid programming the local cpu timer if broadcast pending
Thomas Gleixner [Wed, 6 Mar 2013 11:18:35 +0000 (11:18 +0000)]
tick: Avoid programming the local cpu timer if broadcast pending

If the local cpu timer stops in deep idle, we arm the broadcast device
and get woken by an IPI. Now when we return from deep idle we reenable
the local cpu timer unconditionally before handling the IPI. But
that's a pointless exercise: the timer is already expired and the IPI
is on the way. And it's an expensive exercise as we use the forced
reprogramming mode so that we do not lose a timer event. This forced
reprogramming will loop at least once in the retry.

To avoid this reprogramming, we mark the cpu in a pending bit mask
before we send the IPI. Now when the IPI target cpu wakes up, it will
see the pending bit set and skip the reprogramming. The reprogramming
of the cpu local timer will happen in the IPI handler which runs the
cpu local timer interrupt function.

Reported-by: Jason Liu <liu.h.jason@gmail.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: LAK <linux-arm-kernel@lists.infradead.org>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Arjan van de Veen <arjan@infradead.org>
Cc: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Tested-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Link: http://lkml.kernel.org/r/20130306111537.431082074@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
11 years agoMerge commit 'v3.9-rc2' into timers/core
Thomas Gleixner [Wed, 13 Mar 2013 10:39:02 +0000 (11:39 +0100)]
Merge commit 'v3.9-rc2' into timers/core

Fold in upstream fixes.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
11 years agoMerge branch 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 11 Mar 2013 14:54:29 +0000 (07:54 -0700)]
Merge branch 'perf-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull perf fixes from Ingo Molnar:
 "Misc minor fixes mostly related to tracing"

* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  s390: Fix a header dependencies related build error
  tracing: update documentation of snapshot utility
  tracing: Do not return EINVAL in snapshot when not allocated
  tracing: Add help of snapshot feature when snapshot is empty
  ftrace: Update the kconfig for DYNAMIC_FTRACE

11 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Mon, 11 Mar 2013 14:51:59 +0000 (07:51 -0700)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:

 1) Missing cancel of work items in mac80211 MLME, from Ben Greear.

 2) Fix DMA mapping handling in iwlwifi by using coherent DMA for
    command headers, from Johannes Berg.

 3) Decrease the amount of pressure on the page allocator by using order
    1 pages less in iwlwifi, from Emmanuel Grumbach.

 4) Fix mesh PS broadcast OOPS in mac80211, from Marco Porsch.

 5) Don't forget to recalculate idle state in mac80211 monitor
    interface, from Felix Fietkau.

 6) Fix varargs in netfilter conntrack handler, from Joe Perches.

 7) Need to reset entire chip when command queue fills up in iwlwifi,
    from Emmanuel Grumbach.

 8) The TX antenna value must be valid when calibrations are performed
    in iwlwifi, fix from Dor Shaish.

 9) Don't generate netfilter audit log entries when audit is disabled,
    from Gao Feng.

10) Deal with DMA unit hang on e1000e during power state transitions,
    from Bruce Allan.

11) Remove BUILD_BUG_ON check from igb driver, from Alexander Duyck.

12) Fix lockdep warning on i2c handling of igb driver, from Carolyn
    Wyborny.

13) Fix several TTY handling issues in IRDA ircomm tty driver, from
    Peter Hurley.

14) Several QFQ packet scheduler fixes from Paolo Valente.

15) When VXLAN encapsulates on transmit, we have to reset the netfilter
    state.  From Zang MingJie.

16) Fix jiffie check in net_rx_action() so that we really cap the
    processing at 2HZ.  From Eric Dumazet.

17) Fix erroneous trigger of IP option space exhaustion, when routers
    are pre-specified and we are looking to see if we can insert a
    timestamp, we will have the space.  From David Ward.

18) Fix various issues in benet driver wrt waiting for firmware to
    finish POST after resets or errors.  From Gavin Shan and Sathya
    Perla.

19) Fix TX locking in SFC driver, from Ben Hutchings.

20) Like the VXLAN fix above, when we encap in a TUN device we have to
    reset the netfilter state.  This should fix several strange crashes
    reported by Dave Jones and others.  From Eric Dumazet.

21) Don't forget to clean up MAC address resources when shutting down a
    port in mlx4 driver, from Yan Burman.

22) Fix divide by zero in vmxnet3 driver, from Bhavesh Davda.

23) Fix device statistic regression in tg3 when the driver is using
    phylib, from Nithin Sujir.

24) Fix info leak in several netlink handlers, from Mathias Krause.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (79 commits)
  6lowpan: Fix endianness issue in is_addr_link_local().
  rrunner.c: fix possible memory leak in rr_init_one()
  dcbnl: fix various netlink info leaks
  rtnl: fix info leak on RTM_GETLINK request for VF devices
  bridge: fix mdb info leaks
  tg3: Update link_up flag for phylib devices
  ipv6: stop multicast forwarding to process interface scoped addresses
  bridging: fix rx_handlers return code
  netlabel: fix build problems when CONFIG_IPV6=n
  drivers/isdn: checkng length to be sure not memory overflow
  net/rds: zero last byte for strncpy
  bnx2x: Fix SFP+ misconfiguration in iSCSI boot scenario
  bnx2x: Fix intermittent long KR2 link up time
  macvlan: Set IFF_UNICAST_FLT flag to prevent unnecessary promisc mode.
  team: unsyc the devices addresses when port is removed
  bridge: add missing vid to br_mdb_get()
  Fix: sparse warning in inet_csk_prepare_forced_close
  afkey: fix a typo
  MAINTAINERS: Update qlcnic maintainers list
  netlabel: correctly list all the static label mappings
  ...

11 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rw/uml
Linus Torvalds [Mon, 11 Mar 2013 14:50:39 +0000 (07:50 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/rw/uml

Pull UML fixes from Richard Weinberger:
 "This update brings various fixes.
  Nothing special...

  In my local queue I have some more fixes which will be sent later to
  you.  3.9 uncovered strange UML issues.  :("

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rw/uml:
  um: Use tty_port in SIGWINCH handler
  um: Use tty_port_operations->destruct
  um: fix build failure due to mess-up of sig_info protorype
  um: add missing declaration of 'getrlimit()' and friends
  net : enable tx time stamping in the vde driver.
  hostfs: fix a not needed double check

11 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Linus Torvalds [Mon, 11 Mar 2013 14:49:37 +0000 (07:49 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/dtor/input

Pull input updates from Dmitry Torokhov:
 "Except for the largish change to the ALPS driver adding "Dolphin V1"
  support and Wacom getting a new signature of yet another device, the
  rest are straightforward driver fixes."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: mms114 - Fix regulator enable and disable paths
  Input: ads7864 - check return value of regulator enable
  Input: tc3589x-keypad - fix keymap size
  Input: wacom - add support for 0x10d
  Input: ALPS - update documentation for recent touchpad driver mods
  Input: ALPS - add "Dolphin V1" touchpad support
  Input: ALPS - remove unused argument to alps_enter_command_mode()
  Input: cypress_ps2 - fix trackpadi found in Dell XPS12

11 years agos390: Fix a header dependencies related build error
Li Zefan [Thu, 7 Mar 2013 09:35:53 +0000 (17:35 +0800)]
s390: Fix a header dependencies related build error

Commit 877c685607925238e302cd3aa38788dca6c1b226
("perf: Remove include of cgroup.h from perf_event.h") caused
this build failure if PERF_EVENTS is enabled:

   In file included from arch/s390/include/asm/perf_event.h:9:0,
                    from include/linux/perf_event.h:24,
                    from kernel/events/ring_buffer.c:12:
   arch/s390/include/asm/cpu_mf.h: In function 'qctri':
   arch/s390/include/asm/cpu_mf.h:61:12: error: 'EINVAL' undeclared (first use in this function)

cpu_mf.h had an implicit errno.h dependency, which was added
indirectly via cgroups.h but not anymore. Add it explicitly.

Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Tested-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Li Zefan <lizefan@huawei.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Link: http://lkml.kernel.org/r/51385F79.7000106@huawei.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
11 years agoum: Use tty_port in SIGWINCH handler
Richard Weinberger [Mon, 11 Mar 2013 09:05:45 +0000 (10:05 +0100)]
um: Use tty_port in SIGWINCH handler

The tty below tty_port might get destroyed by the tty layer
while we hold a reference to it.
So we have to carry tty_port around...

Signed-off-by: Richard Weinberger <richard@nod.at>
11 years agoum: Use tty_port_operations->destruct
Richard Weinberger [Mon, 11 Mar 2013 09:03:42 +0000 (10:03 +0100)]
um: Use tty_port_operations->destruct

As we setup the SIGWINCH handler in tty_port_operations->activate
it makes sense to tear down it in ->destruct.

Signed-off-by: Richard Weinberger <richard@nod.at>
11 years agoum: fix build failure due to mess-up of sig_info protorype
Sergei Trofimovich [Sat, 29 Dec 2012 22:37:31 +0000 (01:37 +0300)]
um: fix build failure due to mess-up of sig_info protorype

arch/um/os-Linux/signal.c:18:8: error: conflicting types for 'sig_info'
In file included from /home/slyfox/linux-2.6/arch/um/os-Linux/signal.c:12:0:
arch/um/include/shared/as-layout.h:64:15: note: previous declaration of 'sig_info' was here

Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org>
CC: Jeff Dike <jdike@addtoit.com>
CC: Richard Weinberger <richard@nod.at>
CC: "Martin Pärtel" <martin.partel@gmail.com>
CC: Al Viro <viro@zeniv.linux.org.uk>
CC: user-mode-linux-devel@lists.sourceforge.net
CC: user-mode-linux-user@lists.sourceforge.net
CC: linux-kernel@vger.kernel.org
Signed-off-by: Richard Weinberger <richard@nod.at>
11 years agoum: add missing declaration of 'getrlimit()' and friends
Sergei Trofimovich [Sat, 29 Dec 2012 22:37:30 +0000 (01:37 +0300)]
um: add missing declaration of 'getrlimit()' and friends

arch/um/os-Linux/start_up.c: In function 'check_coredump_limit':
arch/um/os-Linux/start_up.c:338:16: error: storage size of 'lim' isn't known
arch/um/os-Linux/start_up.c:339:2: error: implicit declaration of function 'getrlimit' [-Werror=implicit-function-declaration]

Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org>
CC: Jeff Dike <jdike@addtoit.com>
CC: Richard Weinberger <richard@nod.at>
CC: Al Viro <viro@zeniv.linux.org.uk>
CC: user-mode-linux-devel@lists.sourceforge.net
CC: user-mode-linux-user@lists.sourceforge.net
CC: linux-kernel@vger.kernel.org
Signed-off-by: Richard Weinberger <richard@nod.at>
11 years agonet : enable tx time stamping in the vde driver.
Paul Chavent [Thu, 6 Dec 2012 15:25:05 +0000 (16:25 +0100)]
net : enable tx time stamping in the vde driver.

This new version moves the skb_tx_timestamp in the main uml
driver. This should avoid the need to call this function in each
transport (vde, slirp, tuntap, ...). It also add support for ethtool
get_ts_info.

Signed-off-by: Paul Chavent <paul.chavent@onera.fr>
Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Richard Weinberger <richard@nod.at>
11 years agohostfs: fix a not needed double check
Marco Stornelli [Sat, 20 Oct 2012 10:02:59 +0000 (12:02 +0200)]
hostfs: fix a not needed double check

With the commit 3be2be0a32c18b0fd6d623cda63174a332ca0de1 we removed vmtruncate,
but actaully there is no need to call inode_newsize_ok() because the checks are
already done in inode_change_ok() at the begin of the function.

Signed-off-by: Marco Stornelli <marco.stornelli@gmail.com>
Signed-off-by: Richard Weinberger <richard@nod.at>
11 years agoInput: mms114 - Fix regulator enable and disable paths
Mark Brown [Mon, 4 Mar 2013 04:21:30 +0000 (20:21 -0800)]
Input: mms114 - Fix regulator enable and disable paths

When it uses regulators the mms114 driver checks to see if it managed to
acquire regulators and ignores errors. This is not the intended usage and
not great style in general.

Since the driver already refuses to probe if it fails to allocate the
regulators simply make the enable and disable calls unconditional and
add appropriate error handling, including adding cleanup of the
regulators if setup_reg() fails.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Joonyoung Shim <jy0922.shim@samsung.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
11 years agoInput: ads7864 - check return value of regulator enable
Mark Brown [Mon, 4 Mar 2013 04:19:07 +0000 (20:19 -0800)]
Input: ads7864 - check return value of regulator enable

At least print a warning if we can't power the device up.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
11 years agoInput: tc3589x-keypad - fix keymap size
Rabin Vincent [Sun, 10 Mar 2013 00:17:20 +0000 (16:17 -0800)]
Input: tc3589x-keypad - fix keymap size

The keymap size used by tc3589x is too low, leading to the driver
overwriting other people's memory.  Fix this by making the driver
use the automatically allocated keymap provided by
matrix_keypad_build_keymap() instead of allocating one on its own.

Signed-off-by: Rabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
11 years agoLinux 3.9-rc2
Linus Torvalds [Sun, 10 Mar 2013 23:54:19 +0000 (16:54 -0700)]
Linux 3.9-rc2

11 years ago6lowpan: Fix endianness issue in is_addr_link_local().
YOSHIFUJI Hideaki / 吉藤英明 [Sat, 9 Mar 2013 09:11:57 +0000 (09:11 +0000)]
6lowpan: Fix endianness issue in is_addr_link_local().

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agorrunner.c: fix possible memory leak in rr_init_one()
David Oostdyk [Fri, 8 Mar 2013 08:28:15 +0000 (08:28 +0000)]
rrunner.c: fix possible memory leak in rr_init_one()

In the event that register_netdev() failed, the rrpriv->evt_ring
allocation would have not been freed.

Signed-off-by: David Oostdyk <daveo@ll.mit.edu>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agodcbnl: fix various netlink info leaks
Mathias Krause [Sat, 9 Mar 2013 05:52:21 +0000 (05:52 +0000)]
dcbnl: fix various netlink info leaks

The dcb netlink interface leaks stack memory in various places:
* perm_addr[] buffer is only filled at max with 12 of the 32 bytes but
  copied completely,
* no in-kernel driver fills all fields of an IEEE 802.1Qaz subcommand,
  so we're leaking up to 58 bytes for ieee_ets structs, up to 136 bytes
  for ieee_pfc structs, etc.,
* the same is true for CEE -- no in-kernel driver fills the whole
  struct,

Prevent all of the above stack info leaks by properly initializing the
buffers/structures involved.

Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agortnl: fix info leak on RTM_GETLINK request for VF devices
Mathias Krause [Sat, 9 Mar 2013 05:52:20 +0000 (05:52 +0000)]
rtnl: fix info leak on RTM_GETLINK request for VF devices

Initialize the mac address buffer with 0 as the driver specific function
will probably not fill the whole buffer. In fact, all in-kernel drivers
fill only ETH_ALEN of the MAX_ADDR_LEN bytes, i.e. 6 of the 32 possible
bytes. Therefore we currently leak 26 bytes of stack memory to userland
via the netlink interface.

Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobridge: fix mdb info leaks
Mathias Krause [Sat, 9 Mar 2013 05:52:19 +0000 (05:52 +0000)]
bridge: fix mdb info leaks

The bridging code discloses heap and stack bytes via the RTM_GETMDB
netlink interface and via the notify messages send to group RTNLGRP_MDB
afer a successful add/del.

Fix both cases by initializing all unset members/padding bytes with
memset(0).

Cc: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm...
Linus Torvalds [Sun, 10 Mar 2013 00:51:13 +0000 (16:51 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/ebiederm/user-namespace

Pull namespace bugfixes from Eric Biederman:
 "This is three simple fixes against 3.9-rc1.  I have tested each of
  these fixes and verified they work correctly.

  The userns oops in key_change_session_keyring and the BUG_ON triggered
  by proc_ns_follow_link were found by Dave Jones.

  I am including the enhancement for mount to only trigger requests of
  filesystem modules here instead of delaying this for the 3.10 merge
  window because it is both trivial and the kind of change that tends to
  bit-rot if left untouched for two months."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace:
  proc: Use nd_jump_link in proc_ns_follow_link
  fs: Limit sys_mount to only request filesystem modules (Part 2).
  fs: Limit sys_mount to only request filesystem modules.
  userns: Stop oopsing in key_change_session_keyring

11 years agoInput: wacom - add support for 0x10d
Stephan Frank [Thu, 7 Mar 2013 22:08:50 +0000 (14:08 -0800)]
Input: wacom - add support for 0x10d

It is a Wacom device found in Fujitsu Lifebook T902.

Signed-off-by: Stephan Frank <sfrank@cs.tu-berlin.de>
Acked-by: Ping Cheng <pingc@wacom.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
11 years agoAtmel MXT touchscreen: increase reset timeouts
Linus Torvalds [Sat, 9 Mar 2013 18:31:01 +0000 (10:31 -0800)]
Atmel MXT touchscreen: increase reset timeouts

There is a more complete atmel patch-series out by Nick Dyer that fixes
this and other things, but in the meantime this is the minimal thing to
get the touchscreen going on (at least my) Pixel Chromebook.

Not that I want my dirty fingers near that beautiful screen, but it
seems that a non-initialized touchscreen will also end up being a
constant wakeup source, so you have to disable it to go to sleep.  And
it's easier to just fix the initialization sequence.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoproc: Use nd_jump_link in proc_ns_follow_link
Eric W. Biederman [Sat, 9 Mar 2013 08:14:45 +0000 (00:14 -0800)]
proc: Use nd_jump_link in proc_ns_follow_link

Update proc_ns_follow_link to use nd_jump_link instead of just
manually updating nd.path.dentry.

This fixes the BUG_ON(nd->inode != parent->d_inode) reported by Dave
Jones and reproduced trivially with mkdir /proc/self/ns/uts/a.

Sigh it looks like the VFS change to require use of nd_jump_link
happend while proc_ns_follow_link was baking and since the common case
of proc_ns_follow_link continued to work without problems the need for
making this change was overlooked.

Cc: stable@vger.kernel.org
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
11 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux...
Linus Torvalds [Sat, 9 Mar 2013 01:33:20 +0000 (17:33 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/mason/linux-btrfs

Pull btrfs fixes from Chris Mason:
 "These are scattered fixes and one performance improvement.  The
  biggest functional change is in how we throttle metadata changes.  The
  new code bumps our average file creation rate up by ~13% in fs_mark,
  and lowers CPU usage.

  Stefan bisected out a regression in our allocation code that made
  balance loop on extents larger than 256MB."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
  Btrfs: improve the delayed inode throttling
  Btrfs: fix a mismerge in btrfs_balance()
  Btrfs: enforce min_bytes parameter during extent allocation
  Btrfs: allow running defrag in parallel to administrative tasks
  Btrfs: avoid deadlock on transaction waiting list
  Btrfs: do not BUG_ON on aborted situation
  Btrfs: do not BUG_ON in prepare_to_reloc
  Btrfs: free all recorded tree blocks on error
  Btrfs: build up error handling for merge_reloc_roots
  Btrfs: check for NULL pointer in updating reloc roots
  Btrfs: fix unclosed transaction handler when the async transaction commitment fails
  Btrfs: fix wrong handle at error path of create_snapshot() when the commit fails
  Btrfs: use set_nlink if our i_nlink is 0

11 years agoPlatform: x86: chromeos_laptop : Add basic platform data for atmel devices
Benson Leung [Fri, 8 Mar 2013 03:43:34 +0000 (19:43 -0800)]
Platform: x86: chromeos_laptop : Add basic platform data for atmel devices

Add basic platform data to get the current upstream driver working
with the 224s touchpad and 1664s touchscreen.
We will be using NULL config so we will use the settings from the
devices' NVRAMs.

Signed-off-by: Benson Leung <bleung@chromium.org>
Tested-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoInput: atmel_mxt_ts - Support for touchpad variant
Daniel Kurtz [Fri, 8 Mar 2013 03:43:33 +0000 (19:43 -0800)]
Input: atmel_mxt_ts - Support for touchpad variant

This same driver can be used by atmel based touchscreens and touchpads
(buttonpads). Platform data may specify a device is a touchpad
using the is_tp flag.

This will cause the driver to perform some touchpad specific
initializations, such as:
  * register input device name "Atmel maXTouch Touchpad" instead of
  Touchscreen.
  * register BTN_LEFT & BTN_TOOL_* event types.
  * register axis resolution (as a fixed constant, for now)
  * register BUTTONPAD property
  * process GPIO buttons using reportid T19

Input event GPIO mapping is done by the platform data key_map array.

key_map[x] should contain the KEY or BTN code to send when processing
GPIOx from T19. To specify a GPIO as not an input source, populate
with KEY_RESERVED, or 0.

Signed-off-by: Daniel Kurtz <djkurtz@chromium.org>
Signed-off-by: Benson Leung <bleung@chromium.org>
Signed-off-by: Nick Dyer <nick.dyer@itdev.co.uk>
Tested-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoMerge branch 'for-next' of git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Fri, 8 Mar 2013 23:22:08 +0000 (15:22 -0800)]
Merge branch 'for-next' of git://git.samba.org/sfrench/cifs-2.6

Pull CIFS fixes from Steve French:
 "A small set of cifs fixes which includes one for a recent regression
  in the write path (pointed out by Anton), some fixes for rename
  problems and as promised for 3.9 removing the obsolete sockopt mount
  option (and the accompanying deprecation warning)."

* 'for-next' of git://git.samba.org/sfrench/cifs-2.6:
  CIFS: Fix missing of oplock_read value in smb30_values structure
  cifs: don't try to unlock pagecache page after releasing it
  cifs: remove the sockopt= mount option
  cifs: Check server capability before attempting silly rename
  cifs: Fix bug when checking error condition in cifs_rename_pending_delete()

11 years agoMerge branch 'akpm' (fixes from Andrew)
Linus Torvalds [Fri, 8 Mar 2013 23:05:42 +0000 (15:05 -0800)]
Merge branch 'akpm' (fixes from Andrew)

Merge misc fixes from Andrew Morton.

* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
  alpha: boot: fix build breakage introduced by system.h disintegration
  memcg: initialize kmem-cache destroying work earlier
  Randy has moved
  ksm: fix m68k build: only NUMA needs pfn_to_nid
  dmi_scan: fix missing check for _DMI_ signature in smbios_present()
  Revert parts of "hlist: drop the node parameter from iterators"
  idr: remove WARN_ON_ONCE() on negative IDs
  mm/mempolicy.c: fix sp_node_init() argument ordering
  mm/mempolicy.c: fix wrong sp_node insertion
  ipc: don't allocate a copy larger than max
  ipc: fix potential oops when src msg > 4k w/ MSG_COPY

11 years agoalpha: boot: fix build breakage introduced by system.h disintegration
Will Deacon [Fri, 8 Mar 2013 20:43:37 +0000 (12:43 -0800)]
alpha: boot: fix build breakage introduced by system.h disintegration

Commit ec2212088c42 ("Disintegrate asm/system.h for Alpha") removed the
system.h include from boot/head.S, which puts the PAL_* asm constants
out of scope.

Include <asm/pal.h> so we can get building again.

Signed-off-by: Will Deacon <will.deacon@arm.com>
Cc: David Rusling <david.rusling@linaro.org>
Cc: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agomemcg: initialize kmem-cache destroying work earlier
Konstantin Khlebnikov [Fri, 8 Mar 2013 20:43:36 +0000 (12:43 -0800)]
memcg: initialize kmem-cache destroying work earlier

Fix a warning from lockdep caused by calling cancel_work_sync() for
uninitialized struct work.  This path has been triggered by destructon
kmem-cache hierarchy via destroying its root kmem-cache.

  cache ffff88003c072d80
  obj ffff88003b410000 cache ffff88003c072d80
  obj ffff88003b924000 cache ffff88003c20bd40
  INFO: trying to register non-static key.
  the code is fine but needs lockdep annotation.
  turning off the locking correctness validator.
  Pid: 2825, comm: insmod Tainted: G           O 3.9.0-rc1-next-20130307+ #611
  Call Trace:
    __lock_acquire+0x16a2/0x1cb0
    lock_acquire+0x8a/0x120
    flush_work+0x38/0x2a0
    __cancel_work_timer+0x89/0xf0
    cancel_work_sync+0xb/0x10
    kmem_cache_destroy_memcg_children+0x81/0xb0
    kmem_cache_destroy+0xf/0xe0
    init_module+0xcb/0x1000 [kmem_test]
    do_one_initcall+0x11a/0x170
    load_module+0x19b0/0x2320
    SyS_init_module+0xc6/0xf0
    system_call_fastpath+0x16/0x1b

Example module to demonstrate:

  #include <linux/module.h>
  #include <linux/slab.h>
  #include <linux/mm.h>
  #include <linux/workqueue.h>

  int __init mod_init(void)
  {
   int size = 256;
   struct kmem_cache *cache;
   void *obj;
   struct page *page;

   cache = kmem_cache_create("kmem_cache_test", size, size, 0, NULL);
   if (!cache)
   return -ENOMEM;

   printk("cache %p\n", cache);

   obj = kmem_cache_alloc(cache, GFP_KERNEL);
   if (obj) {
   page = virt_to_head_page(obj);
   printk("obj %p cache %p\n", obj, page->slab_cache);
   kmem_cache_free(cache, obj);
   }

   flush_scheduled_work();

   obj = kmem_cache_alloc(cache, GFP_KERNEL);
   if (obj) {
   page = virt_to_head_page(obj);
   printk("obj %p cache %p\n", obj, page->slab_cache);
   kmem_cache_free(cache, obj);
   }

   kmem_cache_destroy(cache);

   return -EBUSY;
  }

  module_init(mod_init);
  MODULE_LICENSE("GPL");

Signed-off-by: Konstantin Khlebnikov <khlebnikov@openvz.org>
Cc: Glauber Costa <glommer@parallels.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoRandy has moved
Randy Dunlap [Fri, 8 Mar 2013 20:43:35 +0000 (12:43 -0800)]
Randy has moved

Update email address and CREDITS info.  xenotime.net is defunct.

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Cc: Harry Wei <harryxiyou@gmail.com>
Cc: Keiichi KII <k-keiichi@bx.jp.nec.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoksm: fix m68k build: only NUMA needs pfn_to_nid
Hugh Dickins [Fri, 8 Mar 2013 20:43:34 +0000 (12:43 -0800)]
ksm: fix m68k build: only NUMA needs pfn_to_nid

A CONFIG_DISCONTIGMEM=y m68k config gave

  mm/ksm.c: In function `get_kpfn_nid':
  mm/ksm.c:492: error: implicit declaration of function `pfn_to_nid'

linux/mmzone.h declares it for CONFIG_SPARSEMEM and CONFIG_FLATMEM, but
expects the arch's asm/mmzone.h to declare it for CONFIG_DISCONTIGMEM
(see arch/mips/include/asm/mmzone.h for example).

Or perhaps it is only expected when CONFIG_NUMA=y: too much of a maze,
and m68k got away without it so far, so fix the build in mm/ksm.c.

Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Petr Holasek <pholasek@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agodmi_scan: fix missing check for _DMI_ signature in smbios_present()
Ben Hutchings [Fri, 8 Mar 2013 20:43:32 +0000 (12:43 -0800)]
dmi_scan: fix missing check for _DMI_ signature in smbios_present()

Commit 9f9c9cbb6057 ("drivers/firmware/dmi_scan.c: fetch dmi version
from SMBIOS if it exists") hoisted the check for "_DMI_" into
dmi_scan_machine(), which means that we don't bother to check for
"_DMI_" at offset 16 in an SMBIOS entry.  smbios_present() may also call
dmi_present() for an address where we found "_SM_", if it failed further
validation.

Check for "_DMI_" in smbios_present() before calling dmi_present().

[akpm@linux-foundation.org: fix build]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Reported-by: Tim McGrath <tmhikaru@gmail.com>
Tested-by: Tim Mcgrath <tmhikaru@gmail.com>
Cc: Zhenzhong Duan <zhenzhong.duan@oracle.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoRevert parts of "hlist: drop the node parameter from iterators"
Arnd Bergmann [Fri, 8 Mar 2013 20:43:31 +0000 (12:43 -0800)]
Revert parts of "hlist: drop the node parameter from iterators"

Commit b67bfe0d42ca ("hlist: drop the node parameter from iterators")
did a lot of nice changes but also contains two small hunks that seem to
have slipped in accidentally and have no apparent connection to the
intent of the patch.

This reverts the two extraneous changes.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: Peter Senna Tschudin <peter.senna@gmail.com>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Sasha Levin <sasha.levin@oracle.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoidr: remove WARN_ON_ONCE() on negative IDs
Tejun Heo [Fri, 8 Mar 2013 20:43:30 +0000 (12:43 -0800)]
idr: remove WARN_ON_ONCE() on negative IDs

idr_find(), idr_remove() and idr_replace() used to silently ignore the
sign bit and perform lookup with the rest of the bits.  The weird behavior
has been changed such that negative IDs are treated as invalid.  As the
behavior change was subtle, WARN_ON_ONCE() was added in the hope of
determining who's calling idr functions with negative IDs so that they can
be examined for problems.

Up until now, all two reported cases are ID number coming directly from
userland and getting fed into idr_find() and the warnings seem to cause
more problems than being helpful.  Drop the WARN_ON_ONCE()s.

Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: <markus@trippelsdorf.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agomm/mempolicy.c: fix sp_node_init() argument ordering
KOSAKI Motohiro [Fri, 8 Mar 2013 20:43:29 +0000 (12:43 -0800)]
mm/mempolicy.c: fix sp_node_init() argument ordering

Currently, n_new is wrongly initialized.  start and end parameter are
inverted.  Let's fix it.

Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Hillf Danton <dhillf@gmail.com>
Cc: Sasha Levin <sasha.levin@oracle.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Dave Jones <davej@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agomm/mempolicy.c: fix wrong sp_node insertion
Hillf Danton [Fri, 8 Mar 2013 20:43:28 +0000 (12:43 -0800)]
mm/mempolicy.c: fix wrong sp_node insertion

n->end is accessed in sp_insert(). Thus it should be update
before calling sp_insert(). This mistake may make kernel panic.

Signed-off-by: Hillf Danton <dhillf@gmail.com>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Sasha Levin <sasha.levin@oracle.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Dave Jones <davej@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>
11 years agoipc: don't allocate a copy larger than max
Peter Hurley [Fri, 8 Mar 2013 20:43:27 +0000 (12:43 -0800)]
ipc: don't allocate a copy larger than max

When MSG_COPY is set, a duplicate message must be allocated for the copy
before locking the queue.  However, the copy could not be larger than was
sent which is limited to msg_ctlmax.

Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Acked-by: Stanislav Kinsbursky <skinsbursky@parallels.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoipc: fix potential oops when src msg > 4k w/ MSG_COPY
Peter Hurley [Fri, 8 Mar 2013 20:43:26 +0000 (12:43 -0800)]
ipc: fix potential oops when src msg > 4k w/ MSG_COPY

If the src msg is > 4k, then dest->next points to the
next allocated segment; resetting it just prior to dereferencing
is bad.

Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
Acked-by: Stanislav Kinsbursky <skinsbursky@parallels.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoMerge branch 'stable' of git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf/linux...
Linus Torvalds [Fri, 8 Mar 2013 22:46:04 +0000 (14:46 -0800)]
Merge branch 'stable' of git://git./linux/kernel/git/cmetcalf/linux-tile

Pull tile architecture fixes from Chris Metcalf:
 "This fixes the bug that Al Viro spotted with the compat llseek code.
  I also fixed the compat syscall definitions to use the new syscall
  define macros to properly sign-extend their arguments."

* 'stable' of git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf/linux-tile:
  tile: properly use COMPAT_SYSCALL_DEFINEx
  tile: work around bug in the generic sys_llseek

11 years agoMerge tag 'metag-for-v3.9-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/jhoga...
Linus Torvalds [Fri, 8 Mar 2013 22:45:17 +0000 (14:45 -0800)]
Merge tag 'metag-for-v3.9-rc2' of git://git./linux/kernel/git/jhogan/metag

Pull metag bugfixes from James Hogan:
 "A couple of fairly minor arch/metag integration fixes from v3.9-rc1:

   - remove SET_PERSONALITY(): use default definition like other arches
     now do.

   - inhibit NUMA balancing: like SH, NUMA is used for memories with
     different latencies.  ARCH_WANT_VARIABLE_LOCALITY has been added
     for this purpose."

* tag 'metag-for-v3.9-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/jhogan/metag:
  metag: Inhibit NUMA balancing.
  metag: remove SET_PERSONALITY()

11 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound
Linus Torvalds [Fri, 8 Mar 2013 22:43:33 +0000 (14:43 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/tiwai/sound

Pull sound fixes from Takashi Iwai:
 "All are boring small fixes in various parts:
   - A few possible NULL-dereference or zero-division Oops fixes
   - Fix vmaster slave volume notification
   - Add codec ID for ALC233
   - Various fixes in several ASoC WM codecs
   - ASoC tegra i2c fix

  Sorry if you wanted a thrilling adventure with huge sharks :)"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
  ALSA: ice1712: Initialize card->private_data properly
  ALSA: hda - Add support of new codec ALC233
  ALSA: hda/ca0132 - Avoid division by zero in dspxfr_one_seg()
  ALSA: hda - check NULL pointer when creating SPDIF PCM switch
  ALSA: hda - check NULL pointer when creating SPDIF controls
  ASoC: wm5102: Apply a SYSCLK patch for later revs
  ALSA: vmaster: Fix slave change notification
  ASoC: tegra: fix I2S bit count mask
  ALSA: seq: seq_oss_event: missing range checks
  ASoC: wm8350: Use jiffies rather than msecs in schedule_delayed_work()
  ASoC: wm5110: Correct OUT2/3 volume and switch names
  ASoC: wm5102: Correct OUT2 volume and switch names
  ASoC: wm8960: Fix ADC power bits
  ASoC: wm8960: Correct register 0 and 1 defaults

11 years agoMerge branch 'for-3.9/upstream-fixes' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 8 Mar 2013 22:42:52 +0000 (14:42 -0800)]
Merge branch 'for-3.9/upstream-fixes' of git://git./linux/kernel/git/jikos/hid

Pull HID fix from Jiri Kosina:
 "Bugfix for a long-standing bug in logitech-dj driver causing all sorts
  of random initialization problems, finally debugged by Benjamin
  Tissoires with help of Bob Bowles."

* 'for-3.9/upstream-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid:
  HID: logitech-dj: do not directly call hid_output_raw_report() during probe

11 years agoMerge tag 'random_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso...
Linus Torvalds [Fri, 8 Mar 2013 22:42:16 +0000 (14:42 -0800)]
Merge tag 'random_for_linus' of git://git./linux/kernel/git/tytso/random

Pull random fixes from Ted Ts'o:
 "Fix a circular locking dependency in random's collection of cputime
  used by a thread when it exits."

* tag 'random_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/random:
  random: fix locking dependency with the tasklist_lock

11 years agotg3: Update link_up flag for phylib devices
Nithin Sujir [Fri, 8 Mar 2013 08:01:24 +0000 (08:01 +0000)]
tg3: Update link_up flag for phylib devices

Commit f4a46d1f46a8fece34edd2023e054072b02e110d introduced a bug where
the ifconfig stats would remain 0 for phylib devices. This is due to
tp->link_up flag never becoming true causing tg3_periodic_fetch_stats()
to return.

The link_up flag was being updated in tg3_test_and_report_link_chg()
after setting up the phy. This function however, is not called for
phylib devices since the driver does not do the phy setup.

This patch moves the link_up flag update into the common
tg3_link_report() function that gets called for phylib devices as well
for non phylib devices when the link state changes.

To avoid updating link_up twice, we replace tg3_carrier_...() calls that
are followed by tg3_link_report(), with netif_carrier_...(). We can then
remove the unused tg3_carrier_on() function.

CC: <stable@vger.kernel.org>
Reported-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Nithin Nayak Sujir <nsujir@broadcom.com>
Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoipv6: stop multicast forwarding to process interface scoped addresses
Hannes Frederic Sowa [Fri, 8 Mar 2013 02:07:23 +0000 (02:07 +0000)]
ipv6: stop multicast forwarding to process interface scoped addresses

v2:
a) used struct ipv6_addr_props

v3:
a) reverted changes for ipv6_addr_props

v4:
a) do not use __ipv6_addr_needs_scope_id

Cc: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobridging: fix rx_handlers return code
Cristian Bercaru [Fri, 8 Mar 2013 07:03:38 +0000 (07:03 +0000)]
bridging: fix rx_handlers return code

The frames for which rx_handlers return RX_HANDLER_CONSUMED are no longer
counted as dropped. They are counted as successfully received by
'netif_receive_skb'.

This allows network interface drivers to correctly update their RX-OK and
RX-DRP counters based on the result of 'netif_receive_skb'.

Signed-off-by: Cristian Bercaru <B43982@freescale.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agovfs: don't BUG_ON() if following a /proc fd pseudo-symlink results in a symlink
Linus Torvalds [Fri, 8 Mar 2013 17:03:07 +0000 (09:03 -0800)]
vfs: don't BUG_ON() if following a /proc fd pseudo-symlink results in a symlink

It's "normal" - it can happen if the file descriptor you followed was
opened with O_NOFOLLOW.

Reported-by: Dave Jones <davej@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agonetlabel: fix build problems when CONFIG_IPV6=n
Paul Moore [Fri, 8 Mar 2013 14:45:39 +0000 (09:45 -0500)]
netlabel: fix build problems when CONFIG_IPV6=n

My last patch to solve a problem where the static/fallback labels were
not fully displayed resulted in build problems when IPv6 was disabled.
This patch resolves the IPv6 build problems; sorry for the screw-up.

Please queue for -stable or simply merge with the previous patch.

Reported-by: Kbuild Test Robot <fengguang.wu@intel.com>
Signed-off-by: Paul Moore <pmoore@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agotracing: update documentation of snapshot utility
Hiraku Toyooka [Fri, 8 Mar 2013 07:32:25 +0000 (16:32 +0900)]
tracing: update documentation of snapshot utility

Now, "snapshot" file returns success on a reset of snapshot buffer
even if the buffer wasn't allocated, instead of returning EINVAL.
This patch updates snapshot desctiption according to the change.

Link: http://lkml.kernel.org/r/51399409.4090207@hitachi.com
Signed-off-by: Hiraku Toyooka <hiraku.toyooka.gu@hitachi.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
11 years agodrivers/isdn: checkng length to be sure not memory overflow
Chen Gang [Thu, 7 Mar 2013 18:25:41 +0000 (18:25 +0000)]
drivers/isdn: checkng length to be sure not memory overflow

sizeof (cmd.parm.cmsg.para) is 50 (MAX_CAPI_PARA_LEN).
  sizeof (cmd.parm) is 80+, but less than 100.
  strlen(msg) may be more than 80+ (Modem-Commandbuffer, less than 255).
    isdn_tty_send_msg is called by isdn_tty_parse_at
    the relative parameter is m->mdmcmd (atemu *m)
    the relative command may be "+M..."

  so need check the length to be sure not memory overflow.
    cmd.parm is a union, and need keep original valid buffer length no touch

Signed-off-by: Chen Gang <gang.chen@asianux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet/rds: zero last byte for strncpy
Chen Gang [Thu, 7 Mar 2013 17:19:32 +0000 (17:19 +0000)]
net/rds: zero last byte for strncpy

for NUL terminated string, need be always sure '\0' in the end.

additional info:
  strncpy will pads with zeroes to the end of the given buffer.
  should initialise every bit of memory that is going to be copied to userland

Signed-off-by: Chen Gang <gang.chen@asianux.com>
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>