Re: [uml-devel] Strange UML Kernel Panic

[Alexandra W. <cha...@ho...>]

>From: Jeff Dike <jd...@ad...>

>No.  It's a bug.  You can track it down by getting a stack trace from the
>point at which a /proc/mm file descriptor is being closed.  A breakpoint
>in close or __libc_close or something, with a command that does a 'bt' and
>'cont' should be enough.
>
>Also put a breakpoint on panic, and when that happens, look at the file
>descriptor in question, and go back through the gdb output to see where it
>got closed.
>
>Then send me the stack trace.
>
>				Jeff

It crashed again, It's a different Kernel Panic though. 'Munmap failed = 
errno 9'

Gdb:
-----


Breakpoint 3, panic (fmt=0xa01e8270 "munmap failed, errno = %d\n")
    at panic.c:52
52              bust_spinlocks(1);
(gdb) bt
#0  panic (fmt=0xa01e8270 "munmap failed, errno = %d\n") at panic.c:52
(gdb) continue
Continuing.

Program received signal SIGSEGV, Segmentation fault.
close_chan (chans=0xa0235220) at chan_kern.c:180
180                     if(!chan->opened) continue;
(gdb) bt
#0  close_chan (chans=0xa0235220) at chan_kern.c:180
#1  0xa012eb15 in close_lines (lines=0xa02351c0, nlines=-1) at line.c:297
#2  0xa012d48f in console_exit () at stdio_console.c:236
#3  0xa00fc8e2 in do_uml_exitcalls () at process_kern.c:322
#4  0xa00fe277 in machine_power_off () at reboot.c:52
#5  0xa01046ff in panic_exit (self=0xa022e2f8, unused1=0, 
unused2=0xa02451a0)
    at um_arch.c:390

(gdb) enable 1
(gdb) continue
Continuing.

Breakpoint 3, panic (fmt=0xa01e80c8 "Segfault with no mm") at panic.c:52
52              bust_spinlocks(1);
(gdb) bt
#0  panic (fmt=0xa01e80c8 "Segfault with no mm") at panic.c:52
(gdb) continue
Continuing.


Breakpoint 1, 0xa01a18d0 in close ()
(gdb) bt
#0  0xa01a18d0 in close ()
#1  0xa012f1ec in winch_cleanup () at line.c:550
#2  0xa00fc8e2 in do_uml_exitcalls () at process_kern.c:322
#3  0xa00fe277 in machine_power_off () at reboot.c:52
#4  0xa01046ff in panic_exit (self=0xa022e2f8, unused1=0, 
unused2=0xa02451a0)
    at um_arch.c:390
(gdb) continue
Continuing.
x 5

Breakpoint 1, 0xa01a18d0 in close ()
(gdb) bt
#0  0xa01a18d0 in close ()
#1  0xa012d773 in close_chan (chans=0xa0235278) at chan_kern.c:182
#2  0xa012deba in chan_interrupt (chans=0xa0235278, task=0xa02352ac,
    tty=0xa1ca3000, irq=2, dev=0xa0235270) at chan_kern.c:486
#3  0xa012e42f in line_interrupt (irq=2, data=0x4b54, unused=0xa0324274)
    at line.c:27
#4  0xa00f973c in handle_IRQ_event (irq=2, regs=0xa0324274, 
action=0xa1ce8244)
    at irq.c:159
(gdb) continue

Breakpoint 1, 0xa01a18d0 in close ()
(gdb) bt
#0  0xa01a18d0 in close ()
#1  0xa012d773 in close_chan (chans=0xa0235278) at chan_kern.c:182
#2  0xa012eb15 in close_lines (lines=0xa02351c0, nlines=0) at line.c:297
#3  0xa012d48f in console_exit () at stdio_console.c:236
#4  0xa00fc8e2 in do_uml_exitcalls () at process_kern.c:322
#5  0xa00fe277 in machine_power_off () at reboot.c:52
#6  0xa01046ff in panic_exit (self=0xa022e2f8, unused1=0, 
unused2=0xa02451a0)
    at um_arch.c:390
(gdb) continue


Program received signal SIGSEGV, Segmentation fault.
close_chan (chans=0xa0235278) at chan_kern.c:180
180                     if(!chan->opened) continue;
(gdb) bt
#0  close_chan (chans=0xa0235278) at chan_kern.c:180
#1  0xa012eb15 in close_lines (lines=0xa02351c0, nlines=-1) at line.c:297
#2  0xa012d48f in console_exit () at stdio_console.c:236
#3  0xa00fc8e2 in do_uml_exitcalls () at process_kern.c:322
#4  0xa00fe277 in machine_power_off () at reboot.c:52
#5  0xa01046ff in panic_exit (self=0xa022e2f8, unused1=0, 
unused2=0xa02451a0)
    at um_arch.c:390
(gdb) continue
Continuing.

Breakpoint 3, panic (fmt=0xa01e80c8 "Segfault with no mm") at panic.c:52
52              bust_spinlocks(1);
#0  panic (fmt=0xa01e80c8 "Segfault with no mm") at panic.c:52




... (This entire process repats itself a few times, including the close( ) 
and panic.)



AWalther

_________________________________________________________________
The new MSN 8: advanced junk mail protection and 2 months FREE* 
http://join.msn.com/?page=features/junkmail