(gdb) l *io_schedule+0xe
0xc02aa84e is in io_schedule (include2/asm/atomic.h:110).
105 *
106 * Atomically decrements @v by 1.
107 */
108 static __inline__ void atomic_dec(atomic_t *v)
109 {
110 __asm__ __volatile__(
111 LOCK_PREFIX "decl %0"
112 :"+m" (v->counter));
113 }
Well, I was about to say that it is probably gone as it hit again. Now with=
=20
IDE, so we at least can rule out libata. But reiserfs is still in path, so =
I=20
Cc list. It is stock 2.6.22. I will switch to AS, but it took 2 weeks to=20
happen with CFS and in one week I will be off for a couple of weeks.
Here is hand-copied information about locks and blocked processes.
Showing all locks held in the system:
1 lock held by syslogd/2515
#0: (&inode->i_mutex){--..}, at: [<c02ab4d15>] mutex_lock+0x21/0x30
1 lock held by X/3800:
#0: (&mm->mmap_sem){----}, at: [<c02ae2e8>] do_page_fault+0x1e8/0x5e0
6 times migetty - 1 lock held by mingetty/3838:
#0: (&tty->atomic_read_lock){--..}, at: [<c02ab0915>]=20
mutex_lock_interruptible+0x21/0x30
2 times zsh - 1 lock held by zsh/4276:
#0: (&tty->atomic_read_lock){--..}, at: [<c02ab0915>]=20
mutex_lock_interruptible+0x21/0x30
1 lock held by consolehelper-g/21231:
#0: (&inode->i_mutex){--..}, at: [<c02ab4b1>] mutex_lock+0x21/0x30
1 lock held by kio_http/31282:
#0: (&inode->i_mutex){--..}, at: [<c02ab4b1>] mutex_lock+0x21/0x30
and list of blocked tasks:
syslogd
io_schedule+0xe/0x20
sync_buffer+0x35/0x40
__wait_on_bit+0x45/0x70
out_of_line_wait_on_bit+0x50/0x60
__wait_on_buffer+0x27/0x30
flush_commit_list+0x397/0x610 [reiserfs]
do_journal_end+0xadc/0xc90 [reiserfs]
journal_end_sync+0x5d/0x70 [reiserfs]
reiserfs_commit_for_inode+0x17e/0x1a0 [reiserfs]
reiserfs_sync_file+0x2d/0x70 [reiserfs]
do_fsync+0x28/0x40
sys_fsync+0xd/0x10
sysenter_past_esp+0x5f/0x99
X
io_schedule+0xe/0x20
sync_page+0x3a/0x50
__wait_on_bit_lock+0x3f/0x70
...
