Re: Problem with running tests during installing coreutils-5.94

[Philip Rowlands]

[gdb displays nothing]
> > Or a sign that everything was configured with debug information
> > stripped and compilation took place with optimizations

Not on my machine...

$ cat segv.c
#include <string.h>
void main() {
memcpy(0, "hello", 5);
}
$ gcc -O3 -o segv segv.c
segv.c: In function `main':
segv.c:2: warning: return type of `main' is not `int'
$ strip ./segv
$ gdb -x <( printf "run\nbt" ) --args ./segv
GNU gdb 6.3-debian
This GDB was configured as "i386-linux"...(no debugging symbols found)
Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1".

(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)
(no debugging symbols found)

Program received signal SIGSEGV, Segmentation fault.
0x0804836e in ?? ()
#0  0x0804836e in ?? ()
#1  0xb7fc0020 in __libc_missing_32bit_uids () from /lib/ld-linux.so.2
#2  0x080483c8 in ?? ()
#3  0xbf8bebe8 in ?? ()
#4  0xb7e7e8c8 in __libc_start_main () from /lib/tls/i686/cmov/libc.so.6
#5  0xb7e7e8c8 in __libc_start_main () from /lib/tls/i686/cmov/libc.so.6
#6  0x080482b1 in ?? ()
(gdb)

It should *at least* display the program counter, no?

> > Unkillable processes stuck in state "D" sound like possible hardware
> > error. Perhaps there are obvious error messages in the kernel log
> > which could confirm this.
>
> This is the error message I have in kernel log, not very helpful :-):
> Apr 18 14:05:29 [kernel] Unable to handle kernel NULL pointer dereference at
> virtual address 00000018
> Could anything be done to help me not to rebooot the server every time?

This is still putting your whole system in doubt, I'm afraid. gdb and 
shred aren't magic unkillable processes; there's no reason they should 
be hanging on the second invocation. If there's a process unkillable in 
state "D", try this to see where it's waiting:

ps -eo pid,wchan=WIDE-WCHAN-COLUMN -o comm


Cheers,
Phil