Hi Eric,
Could you test with CONFIG_DEBUG_PREEMPT kernel option ?
You can turn on the option by checking the following config menu item:
$ make menuconfig
Kernel hacking --->
[*] Debug preemtible kernel
With this option, kernel will detect preemption count underflows, and
I think this would help to identify the true cause of the problem.
Regards,
Ryusuke Konishi
On Wed, 12 Aug 2009 10:30:22 -0700 (PDT), Eric A wrote:
> I ran into the problem again. It is definetely being caused by firefox. I
> generally use uzbl, a webkit base browser, but after 5min of using firefox
> today, my syslog was filled with over 10 million lines! Here's a sample:
>
> Aug 12 12:53:39 gentoo klogd: BUG: scheduling while atomic:
> segctord/1274/0xfffffffe
> Aug 12 12:53:39 gentoo klogd: Modules linked in:
> Aug 12 12:53:39 gentoo klogd: Pid: 1274, comm: segctord Not tainted
> 2.6.31-rc5-git7 #1
> Aug 12 12:53:39 gentoo klogd: Call Trace:
> Aug 12 12:53:39 gentoo klogd: [<ffffffff813c9821>] ?
> thread_return+0x4af/0x6ee
> Aug 12 12:53:39 gentoo klogd: [<ffffffff8114431c>] ?
> nilfs_segctor_thread+0x33c/0x350
> Aug 12 12:53:39 gentoo klogd: [<ffffffff81140760>] ?
> nilfs_construction_timeout+0x0/0x10
> Aug 12 12:53:39 gentoo klogd: [<ffffffff810276d0>] ?
> __wake_up_common+0x50/0x80
> Aug 12 12:53:39 gentoo klogd: [<ffffffff810481d0>] ?
> autoremove_wake_function+0x0/0x30
> Aug 12 12:53:39 gentoo klogd: [<ffffffff81143fe0>] ?
> nilfs_segctor_thread+0x0/0x350
> Aug 12 12:53:39 gentoo klogd: [<ffffffff81047e56>] ? kthread+0x96/0xa0
> Aug 12 12:53:39 gentoo klogd: [<ffffffff8100c38a>] ? child_rip+0xa/0x20
> Aug 12 12:53:39 gentoo klogd: [<ffffffff81047dc0>] ? kthread+0x0/0xa0
> Aug 12 12:53:39 gentoo klogd: [<ffffffff8100c380>] ? child_rip+0x0/0x20
> Aug 12 12:53:39 gentoo klogd: BUG: scheduling while aaint7 #1
> Aug 12 12:53:39 gentoo klogd: <0x4af [<ff350
> Aug 12 12:53:39 gentoo klogd: [<f0/0x10> [<ff0x30
> Aug 12 12:53:39 gentoo klogd: <0
> Aug 12 12:53:39 gentoo klogd: fff8100ffff8104ff8100c380hedule
> Aug 12 12:53:39 gentoo klogd: Modulaint7 #1
> Aug 12 12:53:39 gentoo klogd: <0x4af [<ff350
> Aug 12 12:53:39 gentoo klogd: [<f0/0x10
> Aug 12 12:53:39 gentoo klogd: > [<ff0x30
> Aug 12 12:53:39 gentoo klogd: <0
> Aug 12 12:53:39 gentoo klogd: fff8100ffff8104ff8100c380hedule
> Aug 12 12:53:39 gentoo klogd: Moduain7 #1
> Aug 12 12:53:39 gentoo klogd: <0x4a [<ff350
> Aug 12 12:53:39 gentoo klogd: [<f0/0x10
> Aug 12 12:53:39 gentoo klogd: > [<fff0x30
> Aug 12 12:53:39 gentoo klogd: <0
> Aug 12 12:53:39 gentoo klogd: fff8100ffff8104ff8100c38hedule
> Aug 12 12:53:39 gentoo klogd: Modulainte7 #1
> Aug 12 12:53:39 gentoo klogd: <0x4af [<ff350
> Aug 12 12:53:39 gentoo klogd: [<f0/0x10
>
> Aug 12 12:53:39 gentoo klogd: > [<ff0x30
>
> Aug 12 12:53:39 gentoo klogd: 0
>
> Aug 12 12:53:39 gentoo klogd: fff8100ffff810ff8100c38hedule
> Aug 12 12:53:39 gentoo klogd: Moduainte7 #1
> Aug 12 12:53:39 gentoo klogd: <0x4a [<ff350
> Aug 12 12:53:39 gentoo klogd: [<f0/0x10> [<fff0x30
> Aug 12 12:53:39 gentoo klogd: <0
> Aug 12 12:53:39 gentoo klogd: fff8100ffff8104ff8100c38hedule
>
> .....
>
> Aug 12 12:56:26 gentoo klogd: while <4>Pomm: Trace:ee
> Aug 12 12:56:26 gentoo klogd: [<<4> [< [<fff/0x10
> Aug 12 12:56:26 gentoo klogd: 4> [<f0
> Aug 12 12:56:26 gentoo klogd: read+0x0ead+0x96ld_riread+0x00
> Aug 12 12:56:26 gentoo klogd: while <4>Pomm: Trace:ee
> Aug 12 12:56:26 gentoo klogd: [<<4> [< [<fffff/0x104> [<ff0
> Aug 12 12:56:26 gentoo klogd: read+0x0ead+0x96/ld_riread+0x0/0
> Aug 12 12:56:26 gentoo klogd: <3while <4>Piomm: Trace:ee
> Aug 12 12:56:26 gentoo klogd: <4> [< [<ff/0x10
> Aug 12 12:56:26 gentoo klogd: 4> [<f0
> Aug 12 12:56:26 gentoo klogd: read+0xead+0x96/0ld_riread+0x0/00
> Aug 12 12:56:26 gentoo klogd: <while <4>Pomm: Trace:ee
> Aug 12 12:56:26 gentoo klogd: [<<4> [< [<ffff/0x10
> Aug 12 12:56:26 gentoo klogd: 4> [<f0
> Aug 12 12:56:26 gentoo klogd: read+0x0/ead+0x96/ld_riread+0x0/0
>
> ....
>
> Eric
>
>
>
> ________________________________
> From: Ryusuke Konishi <[email protected]>
> To: [email protected]; [email protected]
> Sent: Tuesday, August 11, 2009 11:28:45 PM
> Subject: Re: [NILFS users] segctord problem
>
> Hi!
> On Tue, 11 Aug 2009 18:28:05 -0700 (PDT), Eric A wrote:
> > I'm using the latest vanilla kernel (2.6.31-rc5-git7). My distro is Gentoo
> > with nilfs-utils-2.0.14.
> >
> > Eric
> >
> >
> >
> > ________________________________
> > From: Jiro SEKIBA <[email protected]>
> > To: NILFS Users mailing list <[email protected]>
> > Sent: Tuesday, August 11, 2009 8:46:45 PM
> > Subject: Re: [NILFS users] segctord problem
> >
> > Hi,
> >
> > > I have been using nilfs as my rootfs for several weeks now and have been
> > > experiencing random problems. X programs will be become unresponsive and
> > > top
> > > will show that segctord is using ~50%hi (100% of one core).
> >
> > Which version of kernel and distro are you using?
>
> According to your log, preemption count of the log-writer thread goes
> wrong; it must be equal or greater than 0, but your log shows it
> became -1 (0xffffffff) maybe due to some kind of bug in nilfs.
>
> > Aug 11 17:00:59 gentoo klogd: BUG: scheduling while atomic:
> > segctord/1312/0xffffffff
>
> I'm looking at this problem, but I don't have no clue why this
> happened.
>
> Is it reproducible after reboot?
>
> Thanks,
> Ryusuke Konishi
> _______________________________________________
> users mailing list
> [email protected]
> https://www.nilfs.org/mailman/listinfo/users
_______________________________________________
users mailing list
[email protected]
https://www.nilfs.org/mailman/listinfo/users
