kgdb core code. Handles the protocol and the arch details.
[ mingo@elte.hu: heavily modified, simplified and cleaned up. ]
[ xemul@openvz.org: use find_task_by_pid_ns ]
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Jan Kiszka <jan.kiszka@web.de>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Move documentation from semaphore.h to semaphore.c as requested by
Andrew Morton. Also reformat to kernel-doc style and add some more
notes about the implementation.
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
By removing the negative values of 'count' and relying on the wait_list to
indicate whether we have any waiters, we can simplify the implementation
by removing the protection against an unlikely race condition. Thanks to
David Howells for his suggestions.
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
ACPI currently emulates a timeout for semaphores with calls to
down_trylock and sleep. This produces horrible behaviour in terms of
fairness and excessive wakeups. Now that we have a unified semaphore
implementation, adding a real down_trylock is almost trivial.
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
Semaphores are no longer performance-critical, so a generic C
implementation is better for maintainability, debuggability and
extensibility. Thanks to Peter Zijlstra for fixing the lockdep
warning. Thanks to Harvey Harrison for pointing out that the
unlikely() was unnecessary.
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
This way it checks if the clocks are synchronized between CPUs too.
This might be able to detect slowly drifting TSCs which only
go wrong over longer time.
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Call
ts = &per_cpu(tick_cpu_sched, cpu);
and
cpu = smp_processor_id();
once instead of twice.
No functional change done, as changed code runs with local irq off.
Reduces source lines and text size (20bytes on x86_64).
[ akpm@linux-foundation.org: Build fix ]
Signed-off-by: Karsten Wiese <fzu@wemgehoertderstaat.de>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
In order to avoid the false positive from lockdep, each per-cpu base->lock has
the separate lock class and migrate_hrtimers() uses double_spin_lock().
This is overcomplicated: except for migrate_hrtimers() we never take 2 locks
at once, and migrate_hrtimers() can use spin_lock_nested().
Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Arjan van de Ven <arjan@linux.intel.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
In order to avoid the false positive from lockdep, each per-cpu base->lock has
the separate lock class and migrate_timers() uses double_spin_lock().
This all is overcomplicated: except for migrate_timers() we never take 2 locks
at once, and migrate_timers() can use spin_lock_nested().
Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Cc: Arjan van de Ven <arjan@linux.intel.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Fix sparse warnings like this:
kernel/posix-cpu-timers.c:1090:25: warning: symbol 't' shadows an earlier one
kernel/posix-cpu-timers.c:1058:21: originally declared here
Signed-off-by: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Add timer list annotations to workqueue.c so we can see the call site
in the timer stats.
Signed-off-by: Pavel Machek <Pavel@suse.cz>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Convert all the nanosleep related users of restart_block to the
new nanosleep specific restart_block fields.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
> Generic code is not supposed to include irq.h. Replace this include
> by linux/hardirq.h instead and add/replace an include of linux/irq.h
> in asm header files where necessary.
> This change should only matter for architectures that make use of
> GENERIC_CLOCKEVENTS.
> Architectures in question are mips, x86, arm, sh, powerpc, uml and sparc64.
>
> I did some cross compile tests for mips, x86_64, arm, powerpc and sparc64.
> This patch fixes also build breakages caused by the include replacement in
> tick-common.h.
I generally dislike adding optional linux/* includes in asm/* includes -
I'm nervous about this causing include loops.
However, there's a separate point to be discussed here.
That is, what interfaces are expected of every architecture in the kernel.
If generic code wants to be able to set the affinity of interrupts, then
that needs to become part of the interfaces listed in linux/interrupt.h
rather than linux/irq.h.
So what I suggest is this approach instead (against Linus' tree of a
couple of days ago) - we move irq_set_affinity() and irq_can_set_affinity()
to linux/interrupt.h, change the linux/irq.h includes to linux/interrupt.h
and include asm/irq_regs.h where needed (asm/irq_regs.h is supposed to be
rarely used include since not much touches the stacked parent context
registers.)
Build tested on ARM PXA family kernels and ARM's Realview platform
kernels which both use genirq.
[ tglx@linutronix.de: add GENERIC_HARDIRQ dependencies ]
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
When I cleaned up printk() and split up the printk locking logic in
commit 266c2e0abeca649fa6667a1a427ad1da507c6375 ("Make printk() console
semaphore accesses sensible") I had incorrectly moved the call to
have_callable_console() outside of the console semaphore.
That was buggy. The console semaphore protects the console_drivers list
that is used by have_callable_console().
Thanks go to Bongani Hlope who saw this as a hang on shutdown and reboot
and bisected the bug to the right commit, and tested this patch. See
http://lkml.org/lkml/2008/4/11/315
Bisected-and-tested-by: Bongani Hlope <bonganilinux@mweb.co.za>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
revert "sched: fix fair sleepers" (e22ecef1d2658ba54ed7d3fdb5d60829fb434c23),
because it is causing audio skipping, see:
http://bugzilla.kernel.org/show_bug.cgi?id=10428
the patch is correct and the real cause of the skipping is not
understood (tracing makes it go away), but time has run out so we'll
revert it and re-try in 2.6.26.
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Extend the /proc/<pid>/cgroup file to include the appropriate hierarchy ID on
each line.
Currently this ID isn't really needed since a hierarchy can be completely
identified by the set of subsystems bound to it, but this is likely to change
in the near future in order to support stateless subsystems and
merging/rebinding of subsystems. Getting this change into 2.6.25 reduces the
need for an API change later.
Signed-off-by: Paul Menage <menage@google.com>
Cc: Balbir Singh <balbir@in.ibm.com>
Cc: Pavel Emelyanov <xemul@openvz.org>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The prevent_tail_call() macro works around the problem of the compiler
clobbering argument words on the stack, which for asmlinkage functions
is the caller's (user's) struct pt_regs. The tail/sibling-call
optimization is not the only way that the compiler can decide to use
stack argument words as scratch space, which we have to prevent.
Other optimizations can do it too.
Until we have new compiler support to make "asmlinkage" binding on the
compiler's own use of the stack argument frame, we have work around all
the manifestations of this issue that crop up.
More cases seem to be prevented by also keeping the incoming argument
variables live at the end of the function. This makes their original
stack slots attractive places to leave those variables, so the compiler
tends not clobber them for something else. It's still no guarantee, but
it handles some observed cases that prevent_tail_call() did not.
Signed-off-by: Roland McGrath <roland@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The effects of cgroup_disable=foo are:
- foo isn't auto-mounted if you mount all cgroups in a single hierarchy
- foo isn't visible as an individually mountable subsystem
As a result there will only ever be one call to foo->create(), at init time;
all processes will stay in this group, and the group will never be mounted on
a visible hierarchy. Any additional effects (e.g. not allocating metadata)
are up to the foo subsystem.
This doesn't handle early_init subsystems (their "disabled" bit isn't set be,
but it could easily be extended to do so if any of the early_init systems
wanted it - I think it would just involve some nastier parameter processing
since it would occur before the command-line argument parser had been run.
Hugh said:
Ballpark figures, I'm trying to get this question out rather than
processing the exact numbers: CONFIG_CGROUP_MEM_RES_CTLR adds 15% overhead
to the affected paths, booting with cgroup_disable=memory cuts that back to
1% overhead (due to slightly bigger struct page).
I'm no expert on distros, they may have no interest whatever in
CONFIG_CGROUP_MEM_RES_CTLR=y; and the rest of us can easily build with or
without it, or apply the cgroup_disable=memory patches.
Unix bench's execl test result on x86_64 was
== just after boot without mounting any cgroup fs.==
mem_cgorup=off : Execl Throughput 43.0 3150.1 732.6
mem_cgroup=on : Execl Throughput 43.0 2932.6 682.0
==
[lizf@cn.fujitsu.com: fix boot option parsing]
Signed-off-by: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Paul Menage <menage@google.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Pavel Emelyanov <xemul@openvz.org>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Hugh Dickins <hugh@veritas.com>
Cc: Sudhir Kumar <skumar@linux.vnet.ibm.com>
Cc: YAMAMOTO Takashi <yamamoto@valinux.co.jp>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Markers do not mix well with CONFIG_PREEMPT_RCU because it uses
preempt_disable/enable() and not rcu_read_lock/unlock for minimal
intrusiveness. We would need call_sched and sched_barrier primitives.
Currently, the modification (connection and disconnection) of probes
from markers requires changes to the data structure done in RCU-style :
a new data structure is created, the pointer is changed atomically, a
quiescent state is reached and then the old data structure is freed.
The quiescent state is reached once all the currently running
preempt_disable regions are done running. We use the call_rcu mechanism
to execute kfree() after such quiescent state has been reached.
However, the new CONFIG_PREEMPT_RCU version of call_rcu and rcu_barrier
does not guarantee that all preempt_disable code regions have finished,
hence the race.
The "proper" way to do this is to use rcu_read_lock/unlock, but we don't
want to use it to minimize intrusiveness on the traced system. (we do
not want the marker code to call into much of the OS code, because it
would quickly restrict what can and cannot be instrumented, such as the
scheduler).
The temporary fix, until we get call_rcu_sched and rcu_barrier_sched in
mainline, is to use synchronize_sched before each call_rcu calls, so we
wait for the quiescent state in the system call code path. It will slow
down batch marker enable/disable, but will make sure the race is gone.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Silence two kerneldoc warnings.
Warning(kernel/audit.c:1276): No description found for parameter 'string'
Warning(kernel/audit.c:1276): No description found for parameter 'len'
[also fix a typo for bonus points]
Signed-off-by: Dave Jones <davej@codemonkey.org.uk>
Acked-by: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The futex init function is called init(). This is a pain in the neck
when debugging when you code dies in ... init :-)
This renames it to futex_init().
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/tglx/linux-2.6-hrt:
NOHZ: reevaluate idle sleep length after add_timer_on()
clocksource: revert: use init_timer_deferrable for clocksource_watchdog
relay doesn't reference the pages it adds, however we need a non-NULL
hook or splice_to_pipe() can oops.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
I found that relay files can be read by pread(2). I fix it,
for relay files are not capable of seeking.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
add_timer_on() can add a timer on a CPU which is currently in a long
idle sleep, but the timer wheel is not reevaluated by the nohz code on
that CPU. So a timer can be delayed for quite a long time. This
triggered a false positive in the clocksource watchdog code.
To avoid this we need to wake up the idle CPU and enforce the
reevaluation of the timer wheel for the next timer event.
Add a function, which checks a given CPU for idle state, marks the
idle task with NEED_RESCHED and sends a reschedule IPI to notify the
other CPU of the change in the timer wheel.
Call this function from add_timer_on().
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Ingo Molnar <mingo@elte.hu>
Cc: stable@kernel.org
--
include/linux/sched.h | 6 ++++++
kernel/sched.c | 43 +++++++++++++++++++++++++++++++++++++++++++
kernel/timer.c | 10 +++++++++-
3 files changed, 58 insertions(+), 1 deletion(-)
Revert
commit 1077f5a917b7c630231037826b344b2f7f5b903f
Author: Parag Warudkar <parag.warudkar@gmail.com>
Date: Wed Jan 30 13:30:01 2008 +0100
clocksource.c: use init_timer_deferrable for clocksource_watchdog
clocksource_watchdog can use a deferrable timer - reduces wakeups from
idle per second.
The watchdog timer needs to run with the specified interval. Otherwise
it will miss the possible wrap of the watchdog clocksource.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@kernel.org
The printk() logic on when/how to get the console semaphore was
unreadable, this splits the code up into a few helper functions and
makes it easier to follow what is going on.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
In case we're accounting from a sub-namespace, the tgids reported will not
refer to the right namespace.
Save the pid_namespace we're accounting in on the acct_glbs and use it in
do_acct_process.
Two less :) places using the task_struct.tgid member.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This is minor, but dereferencing even current real_parent is not safe on debug
kernels, since the memory, this points to, can be unmapped - RCU protection is
required.
Besides, the tgid field is deprecated and is to be replaced with task_tgid_xxx
call (the 2nd patch), so RCU will be required anyway.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
As Paul pointed out, the ACCESS_ONCE are not needed because we already have
the explicit surrounding memory barriers.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Mike Mason <mmlnx@us.ibm.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: David Smith <dsmith@redhat.com>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Adrian Bunk <adrian.bunk@movial.fi>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Add comments requested by Andrew.
Updated comments about synchronize_sched(). Since we use call_rcu and
rcu_barrier now, these comments were out of sync with the code.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Mike Mason <mmlnx@us.ibm.com>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: David Smith <dsmith@redhat.com>
Cc: "Paul E. McKenney" <paulmck@us.ibm.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Adrian Bunk <adrian.bunk@movial.fi>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The printk() can deadlock because it can wake up klogd(), and
task enqueueing will try to read the time in order to set a hrtimer.
Reported-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Debugged-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Will be called each time the scheduling domains are rebuild.
Needed for architectures that don't have a static cpu topology.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Needed so it can be called from outside of sched.c.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
TREE_AVG and APPROX_AVG are initial task placement policies that have been
disabled for a long while.. time to remove them.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
CC: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
From: Pavel Emelyanov <xemul@openvz.org>
This patch is based on the one from Thomas.
The kauditd_thread() calls the netlink_unicast() and passes
the audit_pid to it. The audit_pid, in turn, is received from
the user space and the tool (I've checked the audit v1.6.9)
uses getpid() to pass one in the kernel. Besides, this tool
doesn't bind the netlink socket to this id, but simply creates
it allowing the kernel to auto-bind one.
That's the preamble.
The problem is that netlink_autobind() _does_not_ guarantees
that the socket will be auto-bound to the current pid. Instead
it uses the current pid as a hint to start looking for a free
id. So, in case of conflict, the audit messages can be sent
to a wrong socket. This can happen (it's unlikely, but can be)
in case some task opens more than one netlink sockets and then
the audit one starts - in this case the audit's pid can be busy
and its socket will be bound to another id.
The proposal is to introduce an audit_nlk_pid in audit subsys,
that will point to the netlink socket to send packets to. It
will most often be equal to audit_pid. The socket id can be
got from the skb's netlink CB right in the audit_receive_msg.
The audit_nlk_pid reset to 0 is not required, since all the
decisions are taken based on audit_pid value only.
Later, if the audit tools will bind the socket themselves, the
kernel will have to provide a way to setup the audit_nlk_pid
as well.
A good side effect of this patch is that audit_pid can later
be converted to struct pid, as it is not longer safe to use
pid_t-s in the presence of pid namespaces. But audit code still
uses the tgid from task_struct in the audit_signal_info and in
the audit_filter_syscall.
Signed-off-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Acked-by: Eric Paris <eparis@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Revert commit 1ada5cba6a0318f90e45b38557e7b5206a9cba38 ("clocksource:
make clocksource watchdog cycle through online CPUs") due to the
regression reported by Gabriel C at
http://lkml.org/lkml/2008/2/24/281
(short vesion: it makes TSC be marked as always unstable on his
machine).
Cc: Andi Kleen <ak@suse.de>
Acked-by: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Robert Hancock <hancockr@shaw.ca>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: Gabriel C <nix.or.die@googlemail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Wakeup-buddy tasks are cache-hot - this makes it a bit harder
for the load-balancer to tear them apart. (but it's still possible,
if the load is sufficiently assymetric)
Signed-off-by: Ingo Molnar <mingo@elte.hu>
improve affine wakeups. Maintain the 'overlap' metric based on CFS's
sum_exec_runtime - which means the amount of time a task executes
after it wakes up some other task.
Use the 'overlap' for the wakeup decisions: if the 'overlap' is short,
it means there's strong workload coupling between this task and the
woken up task. If the 'overlap' is large then the workload is decoupled
and the scheduler will move them to separate CPUs more easily.
( Also slightly move the preempt_check within try_to_wake_up() - this has
no effect on functionality but allows 'early wakeups' (for still-on-rq
tasks) to be correctly accounted as well.)
Signed-off-by: Ingo Molnar <mingo@elte.hu>
split out the affine-wakeup bits.
No code changed:
kernel/sched.o:
text data bss dec hex filename
42521 2858 232 45611 b22b sched.o.before
42521 2858 232 45611 b22b sched.o.after
md5:
9d76738f1272aa82f0b7affd2f51df6b sched.o.before.asm
09b31c44e9aff8666f72773dc433e2df sched.o.after.asm
(the md5's changed because stack slots changed and some registers
get scheduled by gcc in a different order - but otherwise the before
and after assembly is instruction for instruction equivalent.)
Signed-off-by: Ingo Molnar <mingo@elte.hu>
If subbuf_pages was larger than the max number of pages the pipe
buffer will hold, subbuf_splice_actor() would happily go beyond
the array size.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
