[Suspend2-devel] Re: [Suspend2-users] kswapd0 issue

[Sunil Kumar]

I actually had the kswapd0 oops on me yesterday after resume from
suspend-to-ram. One question I have is: how do the /etc/sysctl.conf
parameters work with suspend-resume? Are they loaded again from that file or
the values from before suspend are restored? It looks like the former and
that's what might have led to the oops below because I changed at least one
VM parameter thru /proc before suspend and its value was that from
/etc/sysctl.conf after resume.

I am not sure what the problem might be but here is the oops trace if
anyone's interested:

Unable to handle kernel paging request at 0000000b07000070 RIP:
<ffffffff8026c910>{_spin_lock+0}
PGD 0
Oops: 0002 [1] SMP
CPU 1
Modules linked in: vmnet vmmon snd_intel8x0 snd_ac97_codec snd_ac97_bus
abitugur
u i2c_nforce2 forcedeth button nfsd exportfs nvidia xfs ohci_hcd evdev tun
Pid: 207, comm: kswapd0 Tainted: PF     2.6.17-reiser4-r7 #1
RIP: 0010:[<ffffffff8026c910>] <ffffffff8026c910>{_spin_lock+0}
RSP: 0018:ffff81007f495d90  EFLAGS: 00010202
RAX: 0000000000000001 RBX: ffff810000050ed0 RCX: ffffffff80632bc0
RDX: 0000000000000001 RSI: 00000000000000d0 RDI: 0000000b07000070
RBP: ffff810000050d40 R08: ffff81007f495b18 R09: 0000000000000000
R10: 0000000000000010 R11: ffff81007f495bc8 R12: 0000000b07000000
R13: ffff81007f495db8 R14: 0000000000000031 R15: 0000000000000000
FS:  00002aefb69af3a0(0000) GS:ffff81007fccd4c0(0000) knlGS:00000000f7efc6b0
CS:  0010 DS: 0018 ES: 0018 CR0: 000000008005003b
CR2: 0000000b07000070 CR3: 0000000000201000 CR4: 00000000000006a0
Process kswapd0 (pid: 207, threadinfo ffff81007f494000, task
ffff81007f493420)
Stack: ffffffff802cadfd ffff810000050d40 0000000000000080 0000000000000031
       ffffffff8022fe84 ffff81000004e350 ffff81000003c990 ffff810001e70d90
       00000000000203a0 000000000000009d
Call Trace: <ffffffff802cadfd>{remove_inode_buffers+45}
       <ffffffff8022fe84>{shrink_icache_memory+276}
<ffffffff8024382d>{shrink_slab+253}
       <ffffffff802bcb91>{balance_pgdat+641} <ffffffff8025f167>{kswapd+311}
       <ffffffff802a2500>{autoremove_wake_function+0}
<ffffffff8026649a>{child_rip+8}
       <ffffffff8027f0d0>{physflat_send_IPI_mask+0}
<ffffffff8025f030>{kswapd+0}
       <ffffffff80266492>{child_rip+0}

Code: f0 ff 0f 0f 88 70 02 00 00 c3 66 66 90 66 66 90 48 89 f8 f0
RIP <ffffffff8026c910>{_spin_lock+0} RSP <ffff81007f495d90>
CR2: 0000000b07000070

I was restoring a vmware machine after resume when the oops happened. Both
vmware-vmx and X went into D state. Nothing would work locally. I could ssh
into machine and still use it. I couldn't shut it down clean (I think umount
hung) though and had to resort to sysreq ultimately.

I know vmware, nvidia taint the kernel, but can't help it, and that's why I
didn't even bother to report it. The same is probably true about reiser4.
But my setup has been rock stable for a very long time, and that's the
reason I haven't upgraded to latest kernel or suspend2.

Thanks,
-Sunil

On 11/1/06, Alon Bar-Lev <alon.barlev at gmail.com> wrote:
>
> Hello,
>
> Have anyone had a problem with kswapd and suspend2?
> Suspend was failed because kswapd0 was hung for ore than 20 seconds.
>
> I stated that in hibernate.log and in messages... I copied the
> messages into a different location to save it...
>
> And tried again, this time sync was hung during hibernate script
> operation... I have (Just in case...) the following in
> hibernate.conf:
> OnSuspend 55 sync
>
> I could not do anything with the machine, so I've turned it on... When
> turning it on, the previous message was gone from any of the files it
> was... So I guess the disk caches were not flushed.
>
> Sorry I don't have the exact messages... :)
>
> Best Regards,
> Alon Bar-Lev.
>
> _______________________________________________
> Suspend2-users mailing list
> Suspend2-users at lists.suspend2.net
> http://lists.suspend2.net/mailman/listinfo/suspend2-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.tuxonice.net/pipermail/tuxonice-devel/attachments/20061102/138619d6/attachment.html>