[uml-devel] "Kernel panic: Segfault with no mm"

[uml-devel] "Kernel panic: Segfault with no mm"

[Lars Marowsky-Bree]

Hi all,

with the UML kernel on 2.4.21 I am seeing the above. I can trigger it by
a NULL pointer dereference in the kernel (ie, my buggy code / module
;-).

Instead of a regular panic, I get the above. It appears something is
going astray with the mm handling.

#0  panic (fmt=0xa01b8ad2 "Segfault with no mm") at panic.c:59
#1  0xa00af3b4 in segv (address=1074944004, ip=2706132208, is_write=0, 
    is_user=0, sc=0xa1468280) at trap_kern.c:119
#2  0xa00af866 in segv_handler (sig=11, regs=0xa1468280) at trap_user.c:69
#3  0xa00b351c in sig_handler_common_tt (sig=11, sc_ptr=0x58) at trap_user.c:37
#4  0xa00af990 in sig_handler (sig=1, sc=
      {gs = 0, __gsh = 0, fs = 0, __fsh = 0, es = 43, __esh = 0, ds = 43, __dsh = 0, edi = 2694189056, esi = 2694189056, ebp = 2705767004, esp = 2705766988, ebx = 4294967295, edx = 0, ecx = 2705766824, eax = 2706243584, trapno = 13, err = 0, eip = 2706132208, cs = 35, __csh = 0, eflags = 66118, esp_at_signal = 2705766988, ss = 43, __ssh = 0, fpstate = 0x0, oldmask = 436281344, cr2 = 1074944004})
    at trap_user.c:104
#5  <signal handler called>

I wonder whether anyone has seen that before or whether it is something
peculiar to my setup. (Which also includes the aa patches for the VM)


Sincerely,
    Lars Marowsky-Brée <lmb@suse.de>

-- 
High Availability & Clustering		ever tried. ever failed. no matter.
SuSE Labs				try again. fail again. fail better.
Research & Development, SuSE Linux AG		-- Samuel Beckett



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

Re: [uml-devel] "Kernel panic: Segfault with no mm"

[Oleg Drokin]

Hello!

Lars Marowsky-Bree <lmb@suse.de> wrote:

LMB> with the UML kernel on 2.4.21 I am seeing the above. I can trigger it by
LMB> a NULL pointer dereference in the kernel (ie, my buggy code / module
LMB> ;-).

Well. This is expected.

LMB> Instead of a regular panic, I get the above. It appears something is
LMB> going astray with the mm handling.
LMB> #0  panic (fmt=0xa01b8ad2 "Segfault with no mm") at panic.c:59

So it said you had segfault in kernel.

LMB> #1  0xa00af3b4 in segv (address=1074944004, ip=2706132208, is_write=0, 
LMB>     is_user=0, sc=0xa1468280) at trap_kern.c:119
LMB> #2  0xa00af866 in segv_handler (sig=11, regs=0xa1468280) at trap_user.c:69
LMB> #3  0xa00b351c in sig_handler_common_tt (sig=11, sc_ptr=0x58) at trap_user.c:37
LMB> #4  0xa00af990 in sig_handler (sig=1, sc=
LMB>       {gs = 0, __gsh = 0, fs = 0, __fsh = 0, es = 43, __esh = 0, ds = 43, __dsh = 0, edi = 2694189056, esi = 2694189056, ebp = 2705767004, esp = 2705766988, ebx = 4294967295, edx = 0, ecx = 2705766824, eax = 2706243584, trapno = 13, err = 0, eip = 2706132208, cs = 35, __csh = 0, eflags = 66118, esp_at_signal = 2705766988, ss = 43, __ssh = 0, fpstate = 0x0, oldmask = 436281344, cr2 = 1074944004})
LMB>     at trap_user.c:104
LMB> #5  <signal handler called>

This is signal kernel and one step up is your buggy code.
You just need to go up to that point to see what's wrong.

LMB> I wonder whether anyone has seen that before or whether it is something
LMB> peculiar to my setup. (Which also includes the aa patches for the VM)

I always see this stuff when kernel does such bad things and I sort of
understand that it should be this way.


Bye,
    Oleg


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

Re: [uml-devel] "Kernel panic: Segfault with no mm"

[Lars Marowsky-Bree]

On 2003-09-09T23:14:06,
   Oleg Drokin <green@linuxhacker.ru> said:

> LMB> #5  <signal handler called>
> 
> This is signal kernel and one step up is your buggy code.
> You just need to go up to that point to see what's wrong.

Yes, I know that, too.

But on the host, I'd get a panic which said which kernel module caused
it etc. Is that not to be expected in UML?

If all kernel segvs in UML would lead to this, why is the
if (!is_user) ... branch in the segv() in trap_kern.c there anyway?

It looks like a kernel mode fault should also be handled differently,
and like the Segfault with no mm is sort of an exception.

Sorry if I'm wrong, but I am trying to understand what's going on ;-)


Sincerely,
    Lars Marowsky-Brée <lmb@suse.de>

-- 
High Availability & Clustering		ever tried. ever failed. no matter.
SuSE Labs				try again. fail again. fail better.
Research & Development, SuSE Linux AG		-- Samuel Beckett



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

Re: [uml-devel] "Kernel panic: Segfault with no mm"

[Jeff Dike]

lmb@suse.de said:
> But on the host, I'd get a panic which said which kernel module caused
> it etc. Is that not to be expected in UML? 

UML is lacking some of the debugging stuff present in the other ports.  No
big problem to add, but I haven't seen any pressing need for it.

				Jeff



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

Re: [uml-devel] "Kernel panic: Segfault with no mm"

[Lars Marowsky-Bree]

On 2003-09-09T18:33:24,
   Jeff Dike <jdike@addtoit.com> said:

> > But on the host, I'd get a panic which said which kernel module
> > caused it etc. Is that not to be expected in UML? 
> UML is lacking some of the debugging stuff present in the other ports.
> No big problem to add, but I haven't seen any pressing need for it.

Ah, ok, so all is well. I was just wondering about the strange panic.

Thanks.


Sincerely,
    Lars Marowsky-Brée <lmb@suse.de>

-- 
High Availability & Clustering		ever tried. ever failed. no matter.
SuSE Labs				try again. fail again. fail better.
Research & Development, SuSE Linux AG		-- Samuel Beckett



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel