2.4.28 Oops in fs/locks.c:time_out_leases()
got an Oops the same time for 9 days, at the same EIP:
ksymoops 2.4.9 on i686 2.4.28-x97. Options used
-V (default)
-k /proc/ksyms (default)
-l /proc/modules (default)
-o /lib/modules/2.4.28-x97/ (default)
-m /boot/System.map-2.4.28-x97 (default)
kernel: <1>Unable to handle kernel NULL pointer dereference at virtual address 0000003c kernel: c0153cb8 kernel: *pde = 00000000 kernel: Oops: 0000 kernel: CPU: 0 kernel: EIP: 0010:[time_out_leases+24/128] Not tainted kernel: EFLAGS: 00010202 kernel: eax: c1b42a14 ebx: 00000010 ecx: 00000000 edx: c349139c kernel: esi: c1b42ad0 edi: c06f6000 ebp: c06f7f0c esp: c06f7f04 kernel: ds: 0018 es: 0018 ss: 0018 kernel: Process find (pid: 17476, stackpage=c06f7000) kernel: Stack: c1b42a14 00018801 c06f7f38 c0153d79 c1b42a14 00000000 00000000 c06f7f38 kernel: 00000000 c349139c ffffffff 00018801 c1b42a14 c06f7f64 c014d8c7 c1b42a14 kernel: 00018801 00000000 00000000 00000004 c2f51898 00018800 c160f000 080665cb kernel: Call Trace: [__get_lease+89/640] [open_namei+487/1376] [filp_open+47/80] [sys_open+61/160] [system_call+51/64] kernel: Code: f6 43 2c 20 74 23 f6 43 2d 10 74 1d 8b 53 50 85 d2 75 1d 89
Using defaults from ksymoops -t elf32-i386 -a i386
Quoted 6 lines
Quoted 6 lines
eax;
c1b42a14<_end+164147c/4347ac8>
edx;c349139c<_end+2f8fe04/4347ac8>
esi;c1b42ad0<_end+1641538/4347ac8>
edi;c06f6000<_end+1f4a68/4347ac8>
ebp;c06f7f0c<_end+1f6974/4347ac8>
esp;c06f7f04<_end+1f696c/4347ac8>
Code; 00000000 Before first symbol
00000000 <_EIP>:
Code; 00000000 Before first symbol
0: f6 43 2c 20 testb $0x20,0x2c(%ebx)
Code; 00000004 Before first symbol
4: 74 23 je 29 <_EIP+0x29>
Code; 00000006 Before first symbol
6: f6 43 2d 10 testb $0x10,0x2d(%ebx)
Code; 0000000a Before first symbol
a: 74 1d je 29 <_EIP+0x29>
Code; 0000000c Before first symbol
c: 8b 53 50 mov 0x50(%ebx),%edx
Code; 0000000f Before first symbol
f: 85 d2 test %edx,%edx
Code; 00000011 Before first symbol
11: 75 1d jne 30 <_EIP+0x30>
Code; 00000013 Before first symbol
13: 89 00 mov %eax,(%eax)
Details:
- compiled with gcc version 3.3.4 (Debian 1:3.3.4-3)
- only ext3 and NFSv3 mounts (and automounter).
- The "find" causing the oops appears to be started from cron.daily
and only touches local filesystems.
- SMP kernel running on UP (pentium II)
Signature · 1 line
Frank
Quoted 25 lines · Frank van Maarseveen
got an Oops the same time for 9 days, at the same EIP:
ksymoops 2.4.9 on i686 2.4.28-x97. Options used
-V (default)
-k/proc/ksyms(default)
-l/proc/modules(default)
-o/lib/modules/2.4.28-x97/(default)
-m/boot/System.map-2.4.28-x97(default)kernel: <1>Unable to handle kernel NULL pointer dereference at virtual address 0000003c kernel: c0153cb8 kernel: *pde = 00000000 kernel: Oops: 0000 kernel: CPU: 0 kernel: EIP: 0010:[time_out_leases+24/128] Not tainted kernel: EFLAGS: 00010202 kernel: eax: c1b42a14 ebx: 00000010 ecx: 00000000 edx: c349139c kernel: esi: c1b42ad0 edi: c06f6000 ebp: c06f7f0c esp: c06f7f04 kernel: ds: 0018 es: 0018 ss: 0018 kernel: Process find (pid: 17476, stackpage=c06f7000) kernel: Stack: c1b42a14 00018801 c06f7f38 c0153d79 c1b42a14 00000000 00000000 c06f7f38 kernel: 00000000 c349139c ffffffff 00018801 c1b42a14 c06f7f64 c014d8c7 c1b42a14 kernel: 00018801 00000000 00000000 00000004 c2f51898 00018800 c160f000 080665cb kernel: Call Trace: [__get_lease+89/640] [open_namei+487/1376] [filp_open+47/80] [sys_open+61/160] [system_call+51/64] kernel: Code: f6 43 2c 20 74 23 f6 43 2d 10 74 1d 8b 53 50 85 d2 75 1d 89
Quoted 1 line
Using defaults from ksymoops -t elf32-i386 -a i386
Frank,
I strongly suspect you are hitting a physical memory problem:
Quoted 4 lines
Hi Frank,
Can you please do
gdb vmlinux
disassembletime_out_leases
(gdb) disas time_out_leases
Dump of assembler code for function time_out_leases:
0xc0153ca0 <time_out_leases+0>: push %ebp 0xc0153ca1 <time_out_leases+1>: mov %esp,%ebp 0xc0153ca3 <time_out_leases+3>: push %esi 0xc0153ca4 <time_out_leases+4>: push %ebx 0xc0153ca5 <time_out_leases+5>: mov 0x8(%ebp),%eax 0xc0153ca8 <time_out_leases+8>: mov 0xbc(%eax),%ebx 0xc0153cae <time_out_leases+14>: lea 0xbc(%eax),%esi 0xc0153cb4 <time_out_leases+20>: test %ebx,%ebx 0xc0153cb6 <time_out_leases+22>: je 0xc0153ce1 <time_out_leases+65> 0xc0153cb8 <time_out_leases+24>: testb $0x20,0x2c(%ebx) <---------- (1) 0xc0153cbc <time_out_leases+28>: je 0xc0153ce1 <time_out_leases+65> 0xc0153cbe <time_out_leases+30>: testb $0x10,0x2d(%ebx) <---------- (2) 0xc0153cc2 <time_out_leases+34>: je 0xc0153ce1 <time_out_leases+65> 0xc0153cc4 <time_out_leases+36>: mov 0x50(%ebx),%edx <--------- (OOPS) 0xc0153cc7 <time_out_leases+39>: test %edx,%edx
The ebx register holds the "struct file_lock *fl" pointer, which is accessed twice a few instructions before at (1) and (2).
Suddenly %ebx contains "00000010" (zero with fifth bit flipped) and the kernel crashes.
static void time_out_leases(struct inode *inode)
{
struct file_lock **before;
struct file_lock *fl;
before = &inode->i_flock;
while ((fl = *before) && (fl->fl_flags & FL_LEASE)
&& (fl->fl_type & F_INPROGRESS)) {
if ((fl->fl_break_time == 0)
|| time_before(jiffies, fl->fl_break_time)) {
before = &fl->fl_next;
continue;
}I suggest you to run memtest86 on this box.
Quoted 35 lines
Quoted 8 lines
Quoted 8 lines
eax;
c1b42a14<_end+164147c/4347ac8>
edx;c349139c<_end+2f8fe04/4347ac8>
esi;c1b42ad0<_end+1641538/4347ac8>
edi;c06f6000<_end+1f4a68/4347ac8>
ebp;c06f7f0c<_end+1f6974/4347ac8>
esp;c06f7f04<_end+1f696c/4347ac8>Code; 00000000 Before first symbol
00000000 <_EIP>:
Code; 00000000 Before first symbol
0: f6 43 2c 20 testb $0x20,0x2c(%ebx)
Code; 00000004 Before first symbol
4: 74 23 je 29 <_EIP+0x29>
Code; 00000006 Before first symbol
6: f6 43 2d 10 testb $0x10,0x2d(%ebx)
Code;0000000aBefore first symbol
a: 74 1d je 29 <_EIP+0x29>
Code;0000000cBefore first symbol
c: 8b 53 50 mov 0x50(%ebx),%edx
Code;0000000fBefore first symbol
f: 85 d2 test %edx,%edx
Code; 00000011 Before first symbol
11: 75 1d jne 30 <_EIP+0x30>
Code; 00000013 Before first symbol
13: 89 00 mov %eax,(%eax)Details:
(quoted diff) +0
-3- compiled with gcc version 3.3.4 (Debian 1:3.3.4-3) - only ext3 and NFSv3 mounts (and automounter). - The "find" causing the oops appears to be started from cron.dailyand only touches local filesystems.
(quoted diff) +0
-1- SMP kernel running on UP (pentium II)