lkml.org 
[lkml]   [2013]   [Jan]   [18]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: linux-next: Tree for Jan 18 [ BROKEN suspend: jbd2|acpi|pm? ]
Date
On Saturday, January 19, 2013 12:41:11 AM Sedat Dilek wrote:
> On Sat, Jan 19, 2013 at 12:39 AM, Rafael J. Wysocki <rjw@sisk.pl> wrote:
> > On Saturday, January 19, 2013 12:28:55 AM Sedat Dilek wrote:
> >> On Sat, Jan 19, 2013 at 12:25 AM, Rafael J. Wysocki <rjw@sisk.pl> wrote:
> >> > On Friday, January 18, 2013 11:56:53 PM Sedat Dilek wrote:
> >> >> On Fri, Jan 18, 2013 at 11:35 PM, Sedat Dilek <sedat.dilek@gmail.com> wrote:
> >> >> > On Fri, Jan 18, 2013 at 11:20 PM, Rafael J. Wysocki <rjw@sisk.pl> wrote:
> >> >> >> On Friday, January 18, 2013 11:11:07 PM Sedat Dilek wrote:
> >> >> >>> On Fri, Jan 18, 2013 at 5:37 AM, Stephen Rothwell <sfr@canb.auug.org.au> wrote:
> >> >> >>> > Hi all,
> >> >> >>> >
> >> >> >>> > Changes since 20130117:
> >> >> >>> >
> >> >> >>> > Undropped tree: samung
> >> >> >>> >
> >> >> >>> > The powerpc tree still had a build failure.
> >> >> >>> >
> >> >> >>> > The driver-core tree gained a build failure for which I applied a merge
> >> >> >>> > fix patch.
> >> >> >>> >
> >> >> >>> > The gpio-lw tree gained a build failure so I used the version from
> >> >> >>> > next-20130117.
> >> >> >>> >
> >> >> >>> > The samsung tree lost the majority of its conflicts but gained more
> >> >> >>> > against the arm-soc and slave-dma tree.
> >> >> >>> >
> >> >> >>> > ----------------------------------------------------------------------------
> >> >> >>> >
> >> >> >>>
> >> >> >>> From my dmesg diff-file:
> >> >> >>>
> >> >> >>> +[ 288.730849] PM: Syncing filesystems ... done.
> >> >> >>> +[ 294.050498] Freezing user space processes ... (elapsed 0.04 seconds) done.
> >> >> >>> +[ 294.097024] Freezing remaining freezable tasks ...
> >> >> >>> +[ 314.098849] Freezing of tasks failed after 20.01 seconds (1 tasks
> >> >> >>> refusing to freeze, wq_busy=0):
> >> >> >>> +[ 314.098862] jbd2/loop0-8 D ffffffff8180d780 0 297 2 0x00000000
> >> >> >>> +[ 314.098865] ffff880117ec5b68 0000000000000046 ffff880117ec5b08
> >> >> >>> ffffffff81044c29
> >> >> >>> +[ 314.098868] ffff88011829dc80 ffff880117ec5fd8 ffff880117ec5fd8
> >> >> >>> ffff880117ec5fd8
> >> >> >>> +[ 314.098871] ffff880119b34560 ffff88011829dc80 ffff880117ec5b68
> >> >> >>> ffff88011fad4738
> >> >> >>> +[ 314.098873] Call Trace:
> >> >> >>> +[ 314.098881] [<ffffffff81044c29>] ? default_spin_lock_flags+0x9/0x10
> >> >> >>> +[ 314.098885] [<ffffffff811c63e0>] ? __wait_on_buffer+0x30/0x30
> >> >> >>> +[ 314.098888] [<ffffffff816b4b59>] schedule+0x29/0x70
> >> >> >>> +[ 314.098890] [<ffffffff816b4c2f>] io_schedule+0x8f/0xd0
> >> >> >>> +[ 314.098892] [<ffffffff811c63ee>] sleep_on_buffer+0xe/0x20
> >> >> >>> +[ 314.098896] [<ffffffff816b342f>] __wait_on_bit+0x5f/0x90
> >> >> >>> +[ 314.098898] [<ffffffff811c5aa1>] ? submit_bh+0x121/0x1e0
> >> >> >>> +[ 314.098900] [<ffffffff811c63e0>] ? __wait_on_buffer+0x30/0x30
> >> >> >>> +[ 314.098903] [<ffffffff816b34dc>] out_of_line_wait_on_bit+0x7c/0x90
> >> >> >>> +[ 314.098906] [<ffffffff8107eb00>] ? autoremove_wake_function+0x40/0x40
> >> >> >>> +[ 314.098909] [<ffffffff811c63de>] __wait_on_buffer+0x2e/0x30
> >> >> >>> +[ 314.098913] [<ffffffff8128a6a1>]
> >> >> >>> jbd2_journal_commit_transaction+0x1791/0x1960
> >> >> >>> +[ 314.098917] [<ffffffff8109269d>] ? sched_clock_cpu+0xbd/0x110
> >> >> >>> +[ 314.098920] [<ffffffff8107eac0>] ? add_wait_queue+0x60/0x60
> >> >> >>> +[ 314.098923] [<ffffffff81069fbf>] ? try_to_del_timer_sync+0x4f/0x70
> >> >> >>> +[ 314.098925] [<ffffffff8128e4e8>] kjournald2+0xb8/0x240
> >> >> >>> +[ 314.098927] [<ffffffff8107eac0>] ? add_wait_queue+0x60/0x60
> >> >> >>> +[ 314.098929] [<ffffffff8128e430>] ? commit_timeout+0x10/0x10
> >> >> >>> +[ 314.098931] [<ffffffff8107ded0>] kthread+0xc0/0xd0
> >> >> >>> +[ 314.098933] [<ffffffff8107de10>] ? flush_kthread_worker+0xb0/0xb0
> >> >> >>> +[ 314.098936] [<ffffffff816be52c>] ret_from_fork+0x7c/0xb0
> >> >> >>> +[ 314.098938] [<ffffffff8107de10>] ? flush_kthread_worker+0xb0/0xb0
> >> >> >>> +[ 314.098969]
> >> >> >>> +[ 314.098970] Restarting kernel threads ... done.
> >> >> >>> +[ 314.099052] Restarting tasks ... done.
> >> >> >>>
> >> >> >>> Please, have a lot at it.
> >> >> >>
> >> >> >> This is a freezer failure while freezing kernel threads, so I don't think it's
> >> >> >> related to ACPI or PM directly.
> >> >> >>
> >> >> >> Does it happen on every suspend?
> >> >> >>
> >> >> >
> >> >> > No, I only did one S/R.
> >> >> >
> >> >> > I have built a 2nd new kernel where I pulled-in latest pm.git#linux-next.
> >> >> > With this kernel two S/Rs were fine - but that says not much.
> >> >> >
> >> >>
> >> >> After several S/Rs on the "buggy" -1 kernel I know see in my syslogs:
> >> >>
> >> >> Jan 18 23:50:02 fambox kernel: [ 141.853828] Disabling non-boot CPUs ...
> >> >> Jan 18 23:50:02 fambox kernel: [ 141.956943] smpboot: CPU 1 is now offline
> >> >> Jan 18 23:50:02 fambox kernel: [ 141.957438] NOHZ: local_softirq_pending 02
> >> >> Jan 18 23:50:02 fambox kernel: [ 141.957454] NOHZ: local_softirq_pending 02
> >> >> Jan 18 23:50:02 fambox kernel: [ 142.060830] smpboot: CPU 2 is now offline
> >> >> Jan 18 23:50:02 fambox kernel: [ 142.164639] smpboot: CPU 3 is now offline
> >> >
> >> > Are you worried about the "local_softirq_pending" messages?
> >> >
> >>
> >> That's the only new messages I have seen after several S/Rs.
> >
> > They are kind of unusual.
> >
> > Anyway, they seem to be related to CPU hotplug (CPU offline), so you can try
> > if you can trigger them through the sysfs CPU offline/online interface.
> >
>
> Can you explain that a bit clearer or give some sample lines for testing?

There is a sysfs file

/sys/devices/system/cpu/cpuX/online

(where X=0,1,2,3,...) for each CPU core in the system. The value read from it
indicates whether or not the given core is online (1 means online). Writing 0
to it means that the given core should be put offline. Writing 1 means to put
it back online. You can simply write first 0s and than 1s to those files
for CPUs > 0 multiple times in a row and see if that triggers messages like the
above. If it does, that may mean there's been a change in kernel/cpu.c, for
example, that causes it to appear. The change may have been made somewhere in
arch/x86 too, though.

> >> If you have a testcase for me to reproduce it here, I would be happy.
> >
> > Do you mean the freezer-related issue?
> >
>
> Any one as I am still stepping in the dark.
> I checked my disc-space as I built a lot of software today and run
> once out of space.
> But 1.7GiB should be enough on / for testing.
> I wanted to run the new LTP version I built the last days.
> Let's see what I get...

Stress-testing the freezer is rather easy and doesn't require disk space.
All it takes is to echo "freezer" to /sys/power/pm_test and then do
"echo mem > /sys/power/state && sleep 1" in a loop. This is described in
Documentation/power/basic-pm-debugging.txt IIRC.

Thanks,
Rafael


--
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.


\
 
 \ /
  Last update: 2013-01-19 01:21    [W:0.113 / U:0.092 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site