I'm getting an Oops in recent devel kernels. To say it up front,
this Oops only occurs when kernels since test1-ac18 are compiled with
2.95.2; the recommended compilers are fine. If the bug is with the
compiler, I hope to get some info that could be useful to the egcs team.
Executing the program 'pidof' with any arguments (or none) causes
an Oops and a segmentation fault. I have seen this behavior on two
different machines running two different distribs (RedHat 6.2 and
Mandrake 7.1) with two different compilers (gcc / pgcc 2.95.3). Both
these distribs use 'pidof' in their init scripts, spewing many Oopsen
with each boot. I build most of the alpha kernels, and have saw it
first with t1-ac18.
Both distribs use the latest 2.78 version of SysVinit (which
contains the 'pidof' program.)
The Oops follows (this one from Mandrake 7.1 and gcc 2.95.3,
test3-pre2):
ksymoops 2.3.3 on i586 2.4.0-test3. Options used
-v /usr/src/linux/vmlinux (specified)
-k /proc/ksyms (default)
-l /proc/modules (default)
-o /lib/modules/2.4.0-test3/ (default)
-m /usr/src/linux/System.map (default)
Jun 30 02:55:17 hermes kernel: Unable to handle kernel NULL pointer
dereference at virtual address 00000008
Jun 30 02:55:17 hermes kernel: c0148804
Jun 30 02:55:17 hermes kernel: *pde = 00000000
Jun 30 02:55:17 hermes kernel: Oops: 0000
Jun 30 02:55:17 hermes kernel: CPU: 0
Jun 30 02:55:17 hermes kernel: EIP: 0010:[proc_permission+100/244]
Jun 30 02:55:17 hermes kernel: EFLAGS: 00010207
Jun 30 02:55:17 hermes kernel: eax: 00000000 ebx: 00000000 ecx:
c10edac0 edx: 00000000
Jun 30 02:55:17 hermes kernel: esi: c1f75000 edi: c10be360 ebp:
c172dfa4 esp: c172df00
Jun 30 02:55:17 hermes kernel: ds: 0018 es: 0018 ss: 0018
Jun 30 02:55:17 hermes kernel: Process pidof (pid: 2294,
stackpage=c172d000)
Jun 30 02:55:17 hermes kernel: Stack: c1d38460 c1a348e0 c1a116e0
00000000 00000000 c0148dbf c1a348e0 00000001
Jun 30 02:55:18 hermes kernel: c172dfa4 c172c000 c1a348e0
c1a116e0 c172dfa4 c013a979 c1a116e0 c172dfa4
Jun 30 02:55:18 hermes kernel: c1a348e0 c1292000 00000000
c172dfa4 bffff6cc 00000001 bffff6cc c172dfa8
Jun 30 02:55:18 hermes kernel: Call Trace: [proc_pid_follow_link+39/120]
[path_walk+1541/1804] [__user_walk+58/124] [sys_newstat+21/108]
[system_call+52/64]
Jun 30 02:55:18 hermes kernel: Code: 8b 50 08 39 d0 74 4c 8b 00 89 54 24
10 39 f2 75 eb 57 50 e8
Using defaults from ksymoops -t elf32-i386 -a i386
Code; 00000000 Before first symbol
00000000 <_EIP>:
Code; 00000000 Before first symbol
0: 8b 50 08 mov 0x8(%eax),%edx
Code; 00000003 Before first symbol
3: 39 d0 cmp %edx,%eax
Code; 00000005 Before first symbol
5: 74 4c je 53 <_EIP+0x53> 00000053 Before
first symbol
Code; 00000007 Before first symbol
7: 8b 00 mov (%eax),%eax
Code; 00000009 Before first symbol
9: 89 54 24 10 mov %edx,0x10(%esp,1)
Code; 0000000d Before first symbol
d: 39 f2 cmp %esi,%edx
Code; 0000000f Before first symbol
f: 75 eb jne fffffffc <_EIP+0xfffffffc>
fffffffc <END_OF_CODE+3d5bf081/????>
Code; 00000011 Before first symbol
11: 57 push %edi
Code; 00000012 Before first symbol
12: 50 push %eax
Code; 00000013 Before first symbol
13: e8 00 00 00 00 call 18 <_EIP+0x18> 00000018 Before
first symbol
If someone could determine if and basically where the problem is
with gcc 2.95.x, I will try to contact someone on the egcs team.
Please CC: me, I couldn't handle the traffic on this list. Thanks!
-- Joseph Fannin fannin.30@osu.edu"That's all I have to say about that." -- Forrest Gump
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/