lkml.org 
[lkml]   [2009]   [Jan]   [13]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
SubjectRe: deadlocks if use htb
From
On Thu, 18 Dec 2008, Jarek Poplawski wrote:
> On Thu, Dec 18, 2008 at 09:43:51AM +0300, Badalian Vyacheslav wrote:
> > Hello
> > result: Patch 2+3 = uptime 7 days without crashes.
> > May i revert patches and try single new patch?
>
> Here is my current opinion on this bug:
>
> 1) I'm almost sure it's not a htb, but hrtimers bug (some race),
>
> 2) the htb patches you've tested are not "the proper" way of fixing
> it; I see substantial changes in hrtimers code in the "-tip" tree
> (probably for 2.6.29), which, probably, you'll be advised by
> hrtimers maintainers to try, and I guess, it's not easy on a
> production system,
>
> So, it's up to you:
>
> 1) since these patches work for you, you can stop with testing and
> wait with these patched kernels until 2.6.29 (I can propose this
> #2 patch as a temporary fix then),
>
> 2) for curiosity you could try this patch #4 alone on one box first
> (after reverting at least patch #2), but again: if it works, it
> could be only treated as a temporary hack, and alternative of #2.

I think I am hitting the same issue discussed in this thread and:

http://bugzilla.kernel.org/show_bug.cgi?id=11718

and wanted to share my data.

My system:

2x Xeon @ 3.06ghz
CONFIG_HZ=250
CONFIG_HIGH_RES_TIMERS=y
CONFIG_NO_HZ=y
(please let me know if more details are useful)

With 2.6.27.10 after around 30 minutes or an hour, server tends to hang
with:

Pid: 0, comm: swapper Not tainted (2.6.27.10 #1)
EIP: 0060:[<c023199f>] EFLAGS: 00000286 CPU: 0
EIP is at run_hrtimer_pending+0x31/0xb8
EAX: f6c1d468 EBX: c039b2ae ECX: 00000002 EDX: 00000001
ESI: f6c1d468 EDI: c1807e20 EBP: c0587f28 ESP: c0587f18
DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068
CR0: 8005003b CR2: 09a36358 CR3: 36d64000 CR4: 000006d0
DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
DR6: ffff0ff0 DR7: 00000400
[<c0231a3c>] run_hrtimer_softirq+0x16/0x18
[<c0223e6d>] __do_softirq+0x64/0xcd
[<c0223f0b>] do_softirq+0x35/0x3a
[<c0224199>] irq_exit+0x38/0x6d
[<c020eefb>] smp_apic_timer_interrupt+0x71/0x82
[<c0203640>] apic_timer_interrupt+0x28/0x30
[<c0207f9d>] ? default_idle+0x2d/0x42
[<c0201946>] cpu_idle+0xca/0xea
[<c045194e>] rest_init+0x4e/0x50

One time with 2.6.27.10:

BUG: unable to handle kernel NULL pointer dereference at 000000
[<c0239b6b>] smp_call_function+0x12/0x14
[<c020df20>] native_smp_send_stop+0x1b/0x28
[<c0220304>] panic+0x47/0xdf
[<c0203ac2>] oops_end+0x5d/0x71
[<c0203f9f>] die+0x57/0x5f
[<c02129a7>] do_page_fault+0x474/0x529
[<c0212533>] ? do_page_fault+0x0/0x529
[<c045eb3a>] error_code+0x72/0x78
[<c024007b>] ? cgroup_get_sb+0x20/0x2b4
[<c02f223c>] ? rb_erase+0x118/0x241
[<c023177d>] __remove_hrtimer+0x5f/0x67
[<c039b2ae>] ? qdisc_watchdog+0x0/0x1c
[<c023199a>] run_hrtimer_pending+0x2c/0xb8
[<c0231a3c>] run_hrtimer_softirq+0x16/0x18
[<c0223e6d>] __do_softirq+0x64/0xcd
[<c0223f0b>] do_softirq+0x35/0x3a
[<c0224199>] irq_exit+0x38/0x6d
[<c020eefb>] smp_apic_timer_interrupt+0x71/0x82
[<c0203640>] apic_timer_interrupt+0x28/0x30
[<c0207f9d>] ? default_idle+0x2d/0x42
[<c0201946>] cpu_idle+0xca/0xea
[<c045194e>] rest_init+0x4e/0x50
=======================
---[ end trace 818de8d3237477a5 ]---
Rebooting in 10 seconds..

With 2.6.27.10 plus what is being called patches #2 and #3 in this thread,
the server ran for a day or so before hanging again:

Pid: 0, comm: swapper Not tainted (2.6.27.10 #3)
EIP: 0060:[<c023199f>] EFLAGS: 00000206 CPU: 0
EIP is at run_hrtimer_pending+0x31/0xb8
EAX: f73f8470 EBX: c039b312 ECX: 00000002 EDX: 00000001
ESI: f73f8470 EDI: c1807e20 EBP: c0589f20 ESP: c0589f10
DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068
CR0: 8005003b CR2: 08ca3358 CR3: 005ca000 CR4: 000006d0
DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
DR6: ffff0ff0 DR7: 00000400
[<c0231a3c>] run_hrtimer_softirq+0x16/0x18
[<c0223e6d>] __do_softirq+0x64/0xcd
[<c0223f0b>] do_softirq+0x35/0x3a
[<c0224199>] irq_exit+0x38/0x6d
[<c02052b3>] do_IRQ+0x5c/0x75
[<c0203553>] common_interrupt+0x23/0x28
[<c0207f9d>] ? default_idle+0x2d/0x42
[<c0201946>] cpu_idle+0xca/0xea
[<c045225e>] rest_init+0x4e/0x50

With 2.6.28 (without the patches from this thread) after about an hour the
system hung again. "Show Regs" indicated:

Pid: 0, comm: swapper Not tainted (2.6.28 #1)
EIP: 0060:[<c03977bd>] EFLAGS: 00000247 CPU: 0
EIP is at __netif_schedule+0xc/0x39
EAX: f6934600 EBX: 00000000 ECX: f6934600 EDX: f6864000
ESI: f6864488 EDI: c1807480 EBP: c05c1f08 ESP: c05c1f04
DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068
CR0: 8005003b CR2: 09968358 CR3: 365cd000 CR4: 000006d0
DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
DR6: ffff0ff0 DR7: 00000400
Call Trace:
[<c03a669d>] qdisc_watchdog+0x18/0x1c
[<c02335c3>] run_hrtimer_pending+0x56/0xda
[<c03a6685>] ? qdisc_watchdog+0x0/0x1c
[<c023365d>] run_hrtimer_softirq+0x16/0x18
[<c0225b1a>] __do_softirq+0x7a/0x11c
[<c0225bf1>] do_softirq+0x35/0x3a
[<c0225eb5>] irq_exit+0x38/0x6d
[<c020f719>] smp_apic_timer_interrupt+0x71/0x7f
[<c0203858>] apic_timer_interrupt+0x28/0x30
[<c0208360>] ? default_idle+0x2d/0x42
[<c0201938>] cpu_idle+0x6b/0x84
[<c046b292>] rest_init+0x4e/0x50

Per Jarek's suggestion in bugzilla, I ran 2.6.28 plus Peter Zijlstra's
"hrtimer: removing all ur callback modes" patches dated 2008-11-25,
2008-12-04 and 2008-12-08. Uptime was 2 days 22 hours before I hit what
appears to be an unrelated bug related to the IPv6 FIB. (Separately
reported with subject 'panic with 2.6.28 while doing "ip -6 route"'.)

Chris


\
 
 \ /
  Last update: 2009-01-14 03:53    [W:0.127 / U:0.428 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site