l4-linux booted ! [patch3]

[Harald Fritzsche]

Hello Micheal, Hello Hacker,

Michael Hohmuth wrote:

> > Other times the reboot command  causes a seg fault (null pointer
> > derefenrence) after the message "Send all processes kill sig" (This is
> > reported earlier).
>
> I haven't ever seen this.  Do you get a register/stack dump from
> L4Linux?  In this case it should be possible to find the location of
> the page fault (with System.map and disassembly output
> ("objdump -ld vmlinux")).

this is a copy of the message i have got during reboot (with some history):

[booting]

login:
Strg/Alt/Del
Runing /sbin/init.d/halt ....
...
killall5 -9
[now the error message]
root: rcvd 1c, 421098 [Hmm, Strange .... pte: 67 and eip: 421098]
pte: 67
page fault rcv error = 0x40
KDB: rcv from pager failed
Dump: of trap_state at 0xc0141f30:

EAX: 00000000  EBX: 00000040  ECX: f000c000  EDX: 00003200
ESI: c0141800  EDI: c0141800  EBP: c0141fa8  ESP: c0141f74
EIP: f00066be  EFLAG: 0000324b
CS:  0010  SS: 0018  DS: 002b  ES: 002b   FS: 002b   GS: 002b
trap no 3, error 00000000, from kernel mode

f000c040 f000c1be 00000004 0001f498 c0141800 c0141fa0 c0141800 0000001c
c014f000 0000000f 00421098 00000082 00000040 c0141fcc f000464b c0141800
0000001c 00000004 00000000 0001f498 00000000 f0011a80 c0141fe8 f000876d
0000001c 00000004 004c64a0 0001f018 0001f018 0001efe8 00420a70 00000023

(probably some wrong numbers)

The address 421098 is pointing to line 309 in funtion do_signal() in signal.c

 /* Did we come from a system call? */
309: if (regs->orig_eax >= 0) {
  /* Restart the system call - no handlers present */
  if (regs->eax == -ERESTARTNOHAND ||
      regs->eax == -ERESTARTSYS ||
      regs->eax == -ERESTARTNOINTR) {
   regs->eax = regs->orig_eax;
   regs->eip -= 2;
  }
 }
 return 0;

objdump:

/volumes/v1/SRC1/L4/l4-981012/l4linux-2.0.21/arch/l4-i386/kernel/signal.c:309
  421095:       8b 7d 0c        movl   0xc(%ebp),%edi
->421098:       8b 57 1c        movl   0x1c(%edi),%edx
  42109b:       85 d2           testl  %edx,%edx
  42109d:       7c 22           jl     4210c1 <do_signal+0x1e1>

I hope this helps to catch the problem.
It was  mentioned earlier that this error occures only occasionaly.

With my best regards
Harald Fritzsche