2.6.13-rc5-git2 does not boot on (my) amd64

2.6.13-rc5-git2 does not boot on (my) amd64

[Danny ter Haar]

This is as far as it comes:


Freeing unused kernel memory: 248k freed
VM: killing process hotplug
VM: killing process hotplug
VM: killing process hotplug
VM: killing process hotplug
Unable to handle kernel paging request at fffffff28017b5be RIP:
[<fffffff28017b5be>]
PGD 103027 PUD 0
Oops: 0010 [1]
CPU 0
Modules linked in:
Pid: 1, comm: init Not tainted 2.6.13-rc5-git2
RIP: 0010:[<fffffff28017b5be>] [<fffffff28017b5be>]
RSP: 0000:ffff81007ffa3d38  EFLAGS: 00010296
RAX: 0000000000000000 RBX: ffff81007f450005 RCX: 0000000000000013
RDX: ffff81007ffe0ec0 RSI: ffff81007f42e560 RDI: ffff81007ffa3d48
RBP: 0000000000000000 R08: 0000000000000000 R09: 0000000000000000
R10: ffff81007f463a90 R11: ffff81007f454130 R12: ffff81007fc639e8
R13: ffff81007ffa3ed8 R14: fffffff27f450000 R15: 0000000000000310
FS:  00002aaaaae00640(0000) GS:ffffffff80485800(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: fffffff28017b5be CR3: 000000007fc77000 CR4: 00000000000006e0
Process init (pid: 1, threadinfo ffff81007ffa2000, task ffff81007ffa14a0)
Stack: 000000030029ab98 ffff81007f450001 ffff81007ffe0ec0 ffff81007f42e560
       0000000000000001 ffff81007ffe0ec0 ffff81007f450000 ffff8100033a8900
       ffff81007ffa3ed8 ffff81007f450000
Call Trace:<ffffffff8017bff7>{link_path_walk+135} <ffffffff8016bd2a>{get_unused_ fd+90}
       <ffffffff8017c57c>{path_lookup+380} <ffffffff8017d7ed>{open_namei+205}
       <ffffffff8015f68b>{__vma_link+75} <ffffffff8016cb1d>{filp_open+45}
       <ffffffff8016bd2a>{get_unused_fd+90} <ffffffff8016cbd4>{sys_open+84}
       <ffffffff8010d91e>{system_call+126}

Code:  Bad RIP value.
RIP [<fffffff28017b5be>] RSP <ffff81007ffa3d38>
CR2: fffffff28017b5be
 <0>Kernel panic - not syncing: Attempted to kill init!
 <0>Rebooting in 30 seconds..

2.6.13-rc5 (vanilla) does work/boot but gave me scsi error after 24+ hours 
(log/config file is @ http://newsgate.newsserver.nl/kernel  )

Danny

Re: 2.6.13-rc5-git2 does not boot on (my) amd64

[Andi Kleen]

dth@picard.cistron.nl (Danny ter Haar) writes:

> This is as far as it comes:
> 
> 
> Freeing unused kernel memory: 248k freed
> VM: killing process hotplug
> VM: killing process hotplug
> VM: killing process hotplug
> VM: killing process hotplug
> Unable to handle kernel paging request at fffffff28017b5be RIP:
> [<fffffff28017b5be>]

Looks weird. Just to make sure can you do a make distclean and try
again? It might be a bad compile.

-Andi

Re: 2.6.13-rc5-git2 does not boot on (my) amd64

[Hugh Dickins]

On Thu, 5 Aug 2005, Andi Kleen wrote:
> dth@picard.cistron.nl (Danny ter Haar) writes:
> > 
> > Freeing unused kernel memory: 248k freed
> > VM: killing process hotplug
> > VM: killing process hotplug
> > VM: killing process hotplug
> > VM: killing process hotplug
> > Unable to handle kernel paging request at fffffff28017b5be RIP:
> > [<fffffff28017b5be>]
> 
> Looks weird. Just to make sure can you do a make distclean and try
> again? It might be a bad compile.

No, like Pavel's and Martin's reports, this is just an effect
of the not-quite-fully-baked do_wp_page/get_user_pages race fix in
2.6.13-rc5-git2, which AlexN reported earlier.  Should now be fully
fixed in Linus' current git, and in the 2.6.13-rc6 akpm prophesies
to be coming soon - please all test that.

Hugh

Re: 2.6.13-rc5-git2 does not boot on (my) amd64

[Martin J. Bligh]

--Hugh Dickins <hugh@veritas.com> wrote (on Friday, August 05, 2005 00:06:25 +0100):

> On Thu, 5 Aug 2005, Andi Kleen wrote:
>> dth@picard.cistron.nl (Danny ter Haar) writes:
>> > 
>> > Freeing unused kernel memory: 248k freed
>> > VM: killing process hotplug
>> > VM: killing process hotplug
>> > VM: killing process hotplug
>> > VM: killing process hotplug
>> > Unable to handle kernel paging request at fffffff28017b5be RIP:
>> > [<fffffff28017b5be>]
>> 
>> Looks weird. Just to make sure can you do a make distclean and try
>> again? It might be a bad compile.
> 
> No, like Pavel's and Martin's reports, this is just an effect
> of the not-quite-fully-baked do_wp_page/get_user_pages race fix in
> 2.6.13-rc5-git2, which AlexN reported earlier.  Should now be fully
> fixed in Linus' current git, and in the 2.6.13-rc6 akpm prophesies
> to be coming soon - please all test that.

OK, nightly builds tests should auto-appear on http://test.kernel.org
tommorow morning. I'll try to remember to look for it; if you remember
before I wake up Hugh, should be published there ... ;-)

Thanks for the feedback,

M

Re: 2.6.13-rc5-git2 does not boot on (my) amd64

[Andi Kleen]

On Fri, Aug 05, 2005 at 12:06:25AM +0100, Hugh Dickins wrote:
> On Thu, 5 Aug 2005, Andi Kleen wrote:
> > dth@picard.cistron.nl (Danny ter Haar) writes:
> > > 
> > > Freeing unused kernel memory: 248k freed
> > > VM: killing process hotplug
> > > VM: killing process hotplug
> > > VM: killing process hotplug
> > > VM: killing process hotplug
> > > Unable to handle kernel paging request at fffffff28017b5be RIP:
> > > [<fffffff28017b5be>]
> > 
> > Looks weird. Just to make sure can you do a make distclean and try
> > again? It might be a bad compile.
> 
> No, like Pavel's and Martin's reports, this is just an effect
> of the not-quite-fully-baked do_wp_page/get_user_pages race fix in
> 2.6.13-rc5-git2, which AlexN reported earlier.  Should now be fully
> fixed in Linus' current git, and in the 2.6.13-rc6 akpm prophesies
> to be coming soon - please all test that.

Yes. I didn't read the mail fully before replying to that one.

-Andi

Re: 2.6.13-rc5-git2 does not boot on (my) amd64

[Martin J. Bligh]

>>> > Freeing unused kernel memory: 248k freed
>>> > VM: killing process hotplug
>>> > VM: killing process hotplug
>>> > VM: killing process hotplug
>>> > VM: killing process hotplug
>>> > Unable to handle kernel paging request at fffffff28017b5be RIP:
>>> > [<fffffff28017b5be>]
>>> 
>>> Looks weird. Just to make sure can you do a make distclean and try
>>> again? It might be a bad compile.
>> 
>> No, like Pavel's and Martin's reports, this is just an effect
>> of the not-quite-fully-baked do_wp_page/get_user_pages race fix in
>> 2.6.13-rc5-git2, which AlexN reported earlier.  Should now be fully
>> fixed in Linus' current git, and in the 2.6.13-rc6 akpm prophesies
>> to be coming soon - please all test that.
> 
> OK, nightly builds tests should auto-appear on http://test.kernel.org
> tommorow morning. I'll try to remember to look for it; if you remember
> before I wake up Hugh, should be published there ... ;-)

-git3 works! wheeeeeeeee! thanks guys.

M.

Re: 2.6.13-rc5-git2 does not boot on (my) amd64

[Danny ter Haar]

Martin J. Bligh <mbligh@mbligh.org> wrote:
>>>> > VM: killing process hotplug
>>>> > Unable to handle kernel paging request at fffffff28017b5be RIP:
>>>> > [<fffffff28017b5be>]
>>>> 
>-git3 works! wheeeeeeeee! thanks guys.

ACK!

The newsgateway is pounding on git3 as well ;-)

# procinfo
Linux 2.6.13-rc5-git3 (root@newsgate) 

Memory:      Total        Used        Free      Shared     Buffers
Mem:       2058040     2041800       16240           0         464
Swap:            0           0           0

Bootup: Fri Aug  5 16:54:42 2005    Load average: 3.10 3.29 3.31 1/65 9610

user  :       0:10:28.91   8.7%  page in :        0
nice  :       0:01:18.71   1.1%  page out:        0
system:       0:31:07.22  26.0%  swap in :        0
idle  :       0:04:34.04   3.8%  swap out:        0
uptime:       1:59:55.23         context : 18646886

irq  0:   1795085 timer                 irq 12:         3
irq  1:         8 i8042                 irq 24:   1981124 aic79xx
irq  2:         0 cascade [4]           irq 25:  17270208 aic79xx, eth3
irq  4:       771 serial                irq 28:  34999920 acenic


Let's see if this survives longer than 30 hours ! ;-)

Danny