UML kernel failed to start betvee 2.6.8-rc8 and current git sources

UML kernel failed to start betvee 2.6.8-rc8 and current git sources

[Toralf Förster]

[-- Attachment #1: Type: text/plain, Size: 2281 bytes --]

Hm,

from v2.6.26-rc8 to v2.6.26-rc9-56-g6329d30 sth. affected UML, b/c my user mode linux kernel produces this :

...
VFS: Mounted root (ext3 filesystem) readonly.
Registers -
        0       0x0
        1       0x0
        2       0x0
        3       0x0
        4       0x0
        5       0x0
        6       0x0
        7       0x0
        8       0x0
        9       0x0
        10      0x0
        11      0x0
        12      0x0
        13      0x0
        14      0x0
        15      0x0
        16      0x0
Kernel panic - not syncing: do_syscall_stub : PTRACE_SETREGS failed, errno = 5


EIP: 0073:[<b7eef424>] CPU: 0 Not tainted ESP: 007b:bfc0acf0 EFLAGS: 00200246
    Not tainted
EAX: 00000000 EBX: 00002e8f ECX: 00000013 EDX: 00002e8f
ESI: 00002e8b EDI: 00000000 EBP: 00000002 DS: 007b ES: 007b
17c6bd00:  [<080927c3>] notifier_call_chain+0x43/0x90
17c6bd24:  [<0809286e>] atomic_notifier_call_chain+0x2e/0x40
17c6bd3c:  [<08078eb6>] panic+0x76/0x110
17c6bd5c:  [<0806e4d3>] run_syscall_stub+0x2f3/0x300
17c6bd90:  [<0807044d>] write_ldt_entry+0x1ad/0x1c0
17c6bdd8:  [<08070c0e>] init_new_ldt+0x18e/0x3d0
17c6bdf0:  [<080a55a7>] __alloc_pages_internal+0x87/0x450
17c6be1c:  [<0806f105>] start_userspace+0x1a5/0x1f0
17c6be50:  [<0805ede1>] init_new_context+0x91/0x140
17c6be6c:  [<080c6789>] bprm_mm_init+0x49/0x1c0
17c6be78:  [<0806c485>] set_signals+0x25/0x30
17c6be80:  [<080be18a>] kmem_cache_alloc+0x4a/0xa0
17c6be98:  [<080c7e73>] do_execve+0x83/0x1e0
17c6bebc:  [<0805a685>] execve1+0x35/0x60
17c6bedc:  [<0805a714>] um_execve+0x14/0x50
17c6bee0:  [<080ce4cb>] dupfd+0xfb/0x130
17c6beec:  [<0805ce8f>] kernel_execve+0x3f/0x50
17c6bf08:  [<0805a302>] run_init_process+0x22/0x30
17c6bf18:  [<0805a3c6>] init_post+0xb6/0x100
17c6bf2c:  [<080496e1>] kernel_init+0x251/0x2b0
17c6bf54:  [<08055400>] tcp_congestion_default+0x0/0x20
17c6bfa4:  [<08075b36>] finish_task_switch+0x26/0x80
17c6bfbc:  [<0806b0b3>] run_kernel_thread+0x43/0x50
17c6bfd8:  [<0806b08f>] run_kernel_thread+0x1f/0x50
17c6bfe4:  [<0805be11>] new_thread_handler+0x61/0x90
17c6bfe8:  [<08049490>] kernel_init+0x0/0x2b0

Terminated

-- 
MfG/Sincerely

Toralf Förster
pgp finger print: 7B1A 07F4 EC82 0F90 D4C2 8936 872A E508 7DB6 9DA3

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 197 bytes --]

Re: UML kernel failed to start betvee 2.6.8-rc8 and current git sources

[Jeff Dike]

On Thu, Jul 10, 2008 at 04:19:51PM +0200, Toralf Förster wrote:
> from v2.6.26-rc8 to v2.6.26-rc9-56-g6329d30 sth. affected UML, b/c my user mode linux kernel produces this :
> 
> ...
> VFS: Mounted root (ext3 filesystem) readonly.
> Registers -
>         0       0x0
>         1       0x0
>         2       0x0
>         3       0x0
>         4       0x0
>         5       0x0
>         6       0x0
>         7       0x0
>         8       0x0
>         9       0x0
>         10      0x0
>         11      0x0
>         12      0x0
>         13      0x0
>         14      0x0
>         15      0x0
>         16      0x0
> Kernel panic - not syncing: do_syscall_stub : PTRACE_SETREGS failed, errno = 5
> 

I have 2.6.26-rc9-00010-g3bc5ab9 and it seems fine.  I'll pull the latest git
and see if that's still OK.

What's the host?

				Jeff

-- 
Work email - jdike at linux dot intel dot com

Re: [uml-devel] UML kernel failed to start betvee 2.6.8-rc8 and current git sources

[Jeff Dike]

On Thu, Jul 10, 2008 at 04:19:51PM +0200, Toralf Förster wrote:
> from v2.6.26-rc8 to v2.6.26-rc9-56-g6329d30 sth. affected UML, b/c my user mode linux kernel produces this :

That UML runs fine here.  Did you also upgrade the host kernel between
rc8 and now?

			Jeff

-- 
Work email - jdike at linux dot intel dot com

Re: [uml-devel] UML kernel failed to start betvee 2.6.8-rc8 and current git sources

[Toralf Förster]

[-- Attachment #1: Type: text/plain, Size: 664 bytes --]

At Thursday 10 July 2008 18:58:11 Jeff Dike wrote :
> On Thu, Jul 10, 2008 at 04:19:51PM +0200, Toralf Förster wrote:
> > from v2.6.26-rc8 to v2.6.26-rc9-56-g6329d30 sth. affected UML, b/c my user mode linux kernel produces this :
> 
> That UML runs fine here.  Did you also upgrade the host kernel between
> rc8 and now?
> 
Yes, but that's not the reason. The UML crashes both under host kernel
2.6.24-gentoo-r8 and 2.6.25-gentoo-r6. The last (available) working UML kernel
is linux-v2.6.26-rc8-227 (tested under host kernel 2.6.25-gentoo-r6).

-- 
MfG/Sincerely

Toralf Förster
pgp finger print: 7B1A 07F4 EC82 0F90 D4C2 8936 872A E508 7DB6 9DA3

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 197 bytes --]

Re: [uml-devel] UML kernel failed to start betvee 2.6.8-rc8 and current git sources

[Jeff Dike]

On Thu, Jul 10, 2008 at 07:34:29PM +0200, Toralf Förster wrote:
> At Thursday 10 July 2008 18:58:11 Jeff Dike wrote :
> > On Thu, Jul 10, 2008 at 04:19:51PM +0200, Toralf Förster wrote:
> > > from v2.6.26-rc8 to v2.6.26-rc9-56-g6329d30 sth. affected UML, b/c my user mode linux kernel produces this :
> > 
> > That UML runs fine here.  Did you also upgrade the host kernel between
> > rc8 and now?
> > 
> Yes, but that's not the reason. The UML crashes both under host kernel
> 2.6.24-gentoo-r8 and 2.6.25-gentoo-r6. The last (available) working UML kernel
> is linux-v2.6.26-rc8-227 (tested under host kernel 2.6.25-gentoo-r6).

Can you send me your crashing UML binary?

Also, can you bisect this and see where something broke?

      	      	     	  Jeff

-- 
Work email - jdike at linux dot intel dot com

Re: UML kernel failed to start betvee 2.6.8-rc8 and current git sources

[Jeff Dike]

On Thu, Jul 10, 2008 at 08:09:59PM +0200, Toralf Förster wrote:
> At Thursday 10 July 2008 18:35:15 Jeff Dike wrote :
> 
> > What's the host?
> 
> tfoerste@n22 ~ $ uname -a
> Linux n22 2.6.25-gentoo-r6 #4 Thu Jul 10 19:49:36 CEST 2008 i686 Intel(R) Pentium(R) M processor 1700MHz GenuineIntel GNU/Linux
> 
> BTW,  this works fine and shows only the config :
> n22 ~ # linux-v2.6.26-rc8-227 --showconfig 
> 
> whereas this failed:
> 
> n22 ~ # linux-v2.6.26-rc9-56 --showconfig
> Locating the bottom of the address space ... 0x0
> Locating the top of the address space ... 0xc0000000

OK, I believe you're seeing the same bug as Uli saw here:

    http://marc.info/?l=linux-kernel&m=121011518003727&w=2

More precise symptoms are here:

     http://marc.info/?l=linux-kernel&m=121011722806093&w=2

I never did figure that one out.  I had the same kernel version, same
toolchain, same everything as far as I could see, and I couldn't
reproduce it, except with a binary that he gave me.

You're seeing it on i386, whereas he saw it on x86_64.

The underlying problem is that somehow the UML initcalls aren't being
run, which is why you're seeing all zeros in the register dump.

If you bisect this, I bet you end up at the no-unit-at-a-time patch
that he ended up at.  And I have no idea what that has to do with
anything.

				Jeff

-- 
Work email - jdike at linux dot intel dot com

Re: UML kernel failed to start betvee 2.6.8-rc8 and current git sources

[Toralf Förster]

[-- Attachment #1: Type: text/plain, Size: 2079 bytes --]

At Thursday 10 July 2008 21:30:19 Jeff Dike wrote :
> On Thu, Jul 10, 2008 at 08:09:59PM +0200, Toralf Förster wrote:
> > At Thursday 10 July 2008 18:35:15 Jeff Dike wrote :
> > 
> > > What's the host?
> > 
> > tfoerste@n22 ~ $ uname -a
> > Linux n22 2.6.25-gentoo-r6 #4 Thu Jul 10 19:49:36 CEST 2008 i686 Intel(R) Pentium(R) M processor 1700MHz GenuineIntel GNU/Linux
> > 
> > BTW,  this works fine and shows only the config :
> > n22 ~ # linux-v2.6.26-rc8-227 --showconfig 
> > 
> > whereas this failed:
> > 
> > n22 ~ # linux-v2.6.26-rc9-56 --showconfig
> > Locating the bottom of the address space ... 0x0
> > Locating the top of the address space ... 0xc0000000
> 
> OK, I believe you're seeing the same bug as Uli saw here:
> 
>     http://marc.info/?l=linux-kernel&m=121011518003727&w=2
> 
> More precise symptoms are here:
> 
>      http://marc.info/?l=linux-kernel&m=121011722806093&w=2
> 
> I never did figure that one out.  I had the same kernel version, same
> toolchain, same everything as far as I could see, and I couldn't
> reproduce it, except with a binary that he gave me.
> 
> You're seeing it on i386, whereas he saw it on x86_64.
> 
> The underlying problem is that somehow the UML initcalls aren't being
> run, which is why you're seeing all zeros in the register dump.
> 
> If you bisect this, I bet you end up at the no-unit-at-a-time patch
> that he ended up at.  And I have no idea what that has to do with
> anything.
> 
You're right :

tfoerste@n22 ~/devel/linux-2.6 $ git bisect good
4f81c5350b44bcc501ab6f8a089b16d064b4d2f6 is first bad commit
commit 4f81c5350b44bcc501ab6f8a089b16d064b4d2f6
Author: Jeff Dike <jdike@addtoit.com>
Date:   Mon Jul 7 13:36:56 2008 -0400

    [UML] fix gcc ICEs and unresolved externs

    There are various constraints on the use of unit-at-a-time:
     - i386 uses no-unit-at-a-time for pre-4.0 (not 4.3)
     - x86_64 uses unit-at-a-time always

...


-- 
MfG/Sincerely

Toralf Förster
pgp finger print: 7B1A 07F4 EC82 0F90 D4C2 8936 872A E508 7DB6 9DA3

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 197 bytes --]