pit_clockevent wants to replaced in the argument of the callback
function as well.
Reported-by; Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
For UP it's stupid to request an initialized cpumask for the clock
event devices. Though we need the mask set even on UP to avoid a
horrible ifdeffery especially in the broadcast code.
For SMP we can at least try to survive with a warning and set the
cpumask of the cpu we're running on. That gives a decent chance to
bring the machine up and retrieve the debug info.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Linus Walleij <linus.walleij@linaro.org
Cc: Lee Jones <lee.jones@linaro.org>
Cc: Russell King - ARM Linux <linux@arm.linux.org.uk>
Cc: Stephen Boyd <sboyd@codeaurora.org>
commit 9ec2690758 ("timerfd: Manage cancelable timers in timerfd")
introduced a CONFIG_HIGHRES_TIMERS (should be CONFIG_HIGH_RES_TIMERS)
typo, which caused applications depending on CLOCK_REALTIME timers to
become sluggy due to the fact that the time base of the realtime
timers was not updated when the wall clock time was set.
This causes anything from 100% CPU use for some applications to odd
delays and hickups.
Reported-bisected-and-tested-by: Anca Emanuel <anca.emanuel@gmail.com>
Tested-by: Linus Torvalds <torvalds@linux-foundation.org>
Fatfingered-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The ordering of the clock bases is historical due to the
CLOCK_REALTIME and CLOCK_MONOTONIC constants. Now the hrtimer bases
have their own enumeration due to the gap between CLOCK_MONOTONIC and
CLOCK_BOOTTIME. So we can be more clever as most timers end up on the
CLOCK_MONOTONIC base due to the virtue of POSIX declaring that
relative CLOCK_REALTIME timers are not affected by time changes. In
desktop environments this is slowly changing as applications switch to
absolute timers, but I've observed empty CLOCK_REALTIME bases often
enough. There is no performance penalty or overhead when
CLOCK_REALTIME timers are active, but in case they are not we don't
skip over a full cache line.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Peter Zijlstra <peterz@infradead.org>
Instead of iterating over all possible timer bases avoid it by marking
the active bases in the cpu base.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Peter Zijlstra <peterz@infradead.org>
In the HIGHRES=y case we access the members at the end of struct
hrtimer_cpu_base first and then the one at the beginning. Move the
hrtimer data to front, so we have linear progressing access.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Peter Zijlstra <peterz@infradead.org>
Peter is concerned about the extra scan of CLOCK_REALTIME_COS in the
timer interrupt. Yes, I did not think about it, because the solution
was so elegant. I didn't like the extra list in timerfd when it was
proposed some time ago, but with a rcu based list the list walk it's
less horrible than the original global lock, which was held over the
list iteration.
Requested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Peter Zijlstra <peterz@infradead.org>
generic_handle_irq() is missing a NULL pointer check for the result of
irq_to_desc. This was a not a big problem, but we want to expose it to
drivers, so we better have sanity checks in place. Add a return value
as well, which indicates that the irq number was valid and the handler
was invoked.
Based on the pure code move from Jonathan Cameron.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Jonathan Cameron <jic23@cam.ac.uk>
kernel/irq/ is only built when CONFIG_GENERIC_HARDIRQS=y. So making
code inside of kernel/irq/ conditional on CONFIG_GENERIC_HARDIRQS is
pointless.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
The first cpu which switches from periodic to oneshot mode switches
also the broadcast device into oneshot mode. The broadcast device
serves as a backup for per cpu timers which stop in deeper
C-states. To avoid starvation of the cpus which might be in idle and
depend on broadcast mode it marks the other cpus as broadcast active
and sets the brodcast expiry value of those cpus to the next tick.
The oneshot mode broadcast bit for the other cpus is sticky and gets
only cleared when those cpus exit idle. If a cpu was not idle while
the bit got set in consequence the bit prevents that the broadcast
device is armed on behalf of that cpu when it enters idle for the
first time after it switched to oneshot mode.
In most cases that goes unnoticed as one of the other cpus has usually
a timer pending which keeps the broadcast device armed with a short
timeout. Now if the only cpu which has a short timer active has the
bit set then the broadcast device will not be armed on behalf of that
cpu and will fire way after the expected timer expiry. In the case of
Christians bug report it took ~145 seconds which is about half of the
wrap around time of HPET (the limit for that device) due to the fact
that all other cpus had no timers armed which expired before the 145
seconds timeframe.
The solution is simply to clear the broadcast active bit
unconditionally when a cpu switches to oneshot mode after the first
cpu switched the broadcast device over. It's not idle at that point
otherwise it would not be executing that code.
[ I fundamentally hate that broadcast crap. Why the heck thought some
folks that when going into deep idle it's a brilliant concept to
switch off the last device which brings the cpu back from that
state? ]
Thanks to Christian for providing all the valuable debug information!
Reported-and-tested-by: Christian Hoffmann <email@christianhoffmann.info>
Cc: John Stultz <johnstul@us.ibm.com>
Link: http://lkml.kernel.org/r/%3Calpine.LFD.2.02.1105161105170.3078%40ionos%3E
Cc: stable@kernel.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
The core interrupt chip is a straight forward conversion. The gpio
chip is implemented with two instances of the irq_chip_type which can
be switched with the irq_set_type function. That allows us to use the
generic callbacks and avoids the conditionals in them.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Nicolas Pitre <nico@fluxnic.net>
Conflicts:
arch/ia64/kernel/cyclone.c
arch/mips/kernel/i8253.c
arch/x86/kernel/i8253.c
Reason: Resolve conflicts so further cleanups do not conflict further
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Simple conversion which simply uses the fact that the second irq chip
base address has offset 0x04 to the first one.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-and-Tested-by: Kevin Hilman <khilman@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
The SLUB allocator use of the cmpxchg_double logic was wrong: it
actually needs the irq-safe one.
That happens automatically when we use the native unlocked 'cmpxchg8b'
instruction, but when compiling the kernel for older x86 CPUs that do
not support that instruction, we fall back to the generic emulation
code.
And if you don't specify that you want the irq-safe version, the generic
code ends up just open-coding the cmpxchg8b equivalent without any
protection against interrupts or preemption. Which definitely doesn't
work for SLUB.
This was reported by Werner Landgraf <w.landgraf@ru.ru>, who saw
instability with his distro-kernel that was compiled to support pretty
much everything under the sun. Most big Linux distributions tend to
compile for PPro and later, and would never have noticed this problem.
This also fixes the prototypes for the irqsafe cmpxchg_double functions
to use 'bool' like they should.
[ Btw, that whole "generic code defaults to no protection" design just
sounds stupid - if the code needs no protection, there is no reason to
use "cmpxchg_double" to begin with. So we should probably just remove
the unprotected version entirely as pointless. - Linus ]
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reported-and-tested-by: werner <w.landgraf@ru.ru>
Acked-and-tested-by: Ingo Molnar <mingo@elte.hu>
Acked-by: Christoph Lameter <cl@linux.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Tejun Heo <tj@kernel.org>
Link: http://lkml.kernel.org/r/alpine.LFD.2.02.1105041539050.3005@ionos
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
class_find_device() takes a refcount on the rtc device. rtc_open()
takes another one, so we can drop it after the rtc_open() call.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: John Stultz <john.stultz@linaro.org>
alarmtimer_late_init() uses class_find_device() to find a alarm
capable rtc device. The match callback stores a pointer to the name in
the char pointer handed in from the call site. alarmtimer_late_init()
checks the char pointer for NULL, but the pointer is on the stack and
not initialized to NULL before the call. So it can have random content
when the match function did not identify a device, which leads to
random access in the following rtc_open() call where the pointer is
dereferenced
Instead of relying on the char pointer, check the return value of
class_find_device. If a device is found then the name pointer is valid
as well.
Reported-by: Ingo Molnar <mingo@elte.hu>
Cc: John Stultz <john.stultz@linaro.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Some applications must be aware of clock realtime being set
backward. A simple example is a clock applet which arms a timer for
the next minute display. If clock realtime is set backward then the
applet displays a stale time for the amount of time which the clock
was set backwards. Due to that applications poll the time because we
don't have an interface.
Extend the timerfd interface by adding a flag which puts the timer
onto a different internal realtime clock. All timers on this clock are
expired whenever the clock was set.
The timerfd core records the monotonic offset when the timer is
created. When the timer is armed, then the current offset is compared
to the previous recorded offset. When it has changed, then
timerfd_settime returns -ECANCELED. When a timer is read the offset is
compared and if it changed -ECANCELED returned to user space. Periodic
timers are not rearmed in the cancelation case.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: John Stultz <johnstul@us.ibm.com>
Cc: Chris Friesen <chris.friesen@genband.com>
Tested-by: Kay Sievers <kay.sievers@vrfy.org>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Davide Libenzi <davidel@xmailserver.org>
Reviewed-by: Alexander Shishkin <virtuoso@slind.org>
Link: http://lkml.kernel.org/r/%3Calpine.LFD.2.02.1104271359580.3323%40ionos%3E
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Make clock_was_set() unconditional and rename hres_timers_resume to
hrtimers_resume. This is a preparatory patch for hrtimers which are
cancelled when clock realtime was set.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
These callbacks are only called in the syscore suspend/resume code on
interrupt chips which have been registered via the generic irq chip
mechanism. Calling those callbacks per irq would be rather icky, but
with the generic irq chip mechanism we can call this per registered
chip.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-arm-kernel@lists.infradead.org
Some unnamed moron fatfingered the arguments of the irq chip callbacks
to irq_chip instead of irq_data.
While at it remove the nmi_count() print in arch_show_interrupts()
which has been broken before the irq conversion already.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
The irq_set_wake() function of the gpio irq_chip calls
enable/disable_irq_wake() on the demultiplex interrupt. That leads to
a lockdep warning "INFO: possible recursive locking detected" because
irq_set_type() is called under irq_desc->lock and the *_irq_wake()
calls take irq_desc->lock of the demux interrupt.
Tell lockdep that the gpio irqs are in a different lock class.
Documentation/SubmitChecklist:
15: All codepaths have been exercised with all lockdep features enabled.
That's a non-optional requirement, AFAICT.
Reported-and-tested-by: Arnaud Patard <arnaud.patard@rtp-net.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
LAKML-Reference: alpine.LFD.2.00.1104041416290.19945@localhost6.localdomain6
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
When I added the buslock/unlock mechanism to the pmic code in order to
get rid of the horrible work queue stuff, stupid me missed to add the
new callbacks to the irq_chip.
In consequence Andrew removed the unused functions, but I missed that.
Add them back and hook them up proper.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Matthew Garrett <mjg@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Matthew Garrett <mjg@redhat.com>
After we made debugobjects working again, we got the following:
WARNING: at lib/debugobjects.c:262 debug_print_object+0x8e/0xb0()
Hardware name: System Product Name
ODEBUG: free active (active state 0) object type: timer_list hint: hci_cmd_timer+0x0/0x60
Pid: 2125, comm: dmsetup Tainted: G W 2.6.38-06707-gc62b389 #110375
Call Trace:
[<ffffffff8104700a>] warn_slowpath_common+0x7a/0xb0
[<ffffffff810470b6>] warn_slowpath_fmt+0x46/0x50
[<ffffffff812d3a5e>] debug_print_object+0x8e/0xb0
[<ffffffff81bd8810>] ? hci_cmd_timer+0x0/0x60
[<ffffffff812d4685>] debug_check_no_obj_freed+0x125/0x230
[<ffffffff810f1063>] ? check_object+0xb3/0x2b0
[<ffffffff810f3630>] kfree+0x150/0x190
[<ffffffff81be4d06>] ? bt_host_release+0x16/0x20
[<ffffffff81be4d06>] bt_host_release+0x16/0x20
[<ffffffff813a1907>] device_release+0x27/0xa0
[<ffffffff812c519c>] kobject_release+0x4c/0xa0
[<ffffffff812c5150>] ? kobject_release+0x0/0xa0
[<ffffffff812c61f6>] kref_put+0x36/0x70
[<ffffffff812c4d37>] kobject_put+0x27/0x60
[<ffffffff813a21f7>] put_device+0x17/0x20
[<ffffffff81bda4f9>] hci_free_dev+0x29/0x30
[<ffffffff81928be6>] vhci_release+0x36/0x70
[<ffffffff810fb366>] fput+0xd6/0x1f0
[<ffffffff810f8fe6>] filp_close+0x66/0x90
[<ffffffff810f90a9>] sys_close+0x99/0xf0
[<ffffffff81d4c96b>] system_call_fastpath+0x16/0x1b
That timer was introduced with commit 6bd32326cda(Bluetooth: Use
proper timer for hci command timout)
Timer seems to be running when the thing is closed. Removing the timer
unconditionally fixes the problem. And yes, it needs to be fixed
before the HCI_UP check.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>
The lonely user of the internal interface was not in the coccinelle
script.
Reported-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Convert to new function names. Converted with coccinelle.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: David Howells <dhowells@redhat.com>
irq_chip.end got obsolete with the removal of __do_IRQ().
irq-mb93093.c even lacks an implementation, but nobody noticed that
it's broken since commit 88d6e1 in 2006.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: David Howells <dhowells@redhat.com>
This driver is broken in several aspects.
1) old style irq_chip functions. Sigh
2) Abuse of the unlock callback. That's not supposed to be a state
machine for evrything and some more.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Make use of the new features in genirq:
1) Set the chip flag IRCHIP_EOI_IF_HANDLED, which ensures in the
core code that irq_eoi() is only called when the interrupt was
handled. That removes the extra status check in the callback.
2) Use the preflow handler, which is called from the fasteoi core code
before the device handler. That avoids another status check and the
open coded handler redirection.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: sparclinux@vger.kernel.org
Thomas Gleixner