On Thu, 24 Feb 2000, Ingo Molnar wrote:
> > > something like:
> > >
> > > + if (current->need_resched) {
> > > + INODE(tmp)->i_count++;
> > > + spin_unlock(&inode_lock);
> > > + schedule();
> > > + iput(tmp);
> > > + goto repeat;
> > >
> > > is there any problem with this?
> > >
I fixed __free_inodes as suggested by Ingo, all seemed fine during
10 hours of testing. Then I got a crash/lockup and had to reboot.
Last entry in syslog:
Feb 26 04:41:19 vmebox kernel: iput: device 00:00 inode -2119956477 still has aliases!
Feb 26 04:41:19 vmebox kernel: iput: Aieee, semaphore in use inode 00:00/-2119956477, count=0
Feb 26 04:41:19 vmebox kernel: iput: Aieee, atomic write semaphore in use inode 00:00/-2119956477, count=0
Feb 26 04:41:19 vmebox kernel: iput: device 00:00 inode -2119956477 still has aliases!
Feb 26 04:41:19 vmebox kernel: iput: Aieee, semaphore in use inode 00:00/-2119956477, count=0
Feb 26 04:41:19 vmebox kernel: iput: Aieee, atomic write semaphore in use inode 00:00/-2119956477, count=0
Feb 26 04:41:19 vmebox kernel: iput: device 00:00 inode -2119956477 still has aliases!
Feb 26 04:41:19 vmebox kernel: iput: Aieee, semaphore in use inode 00:00/-2119956477, count=0
Feb 26 04:41:19 vmebox kernel: iput: Aieee, atomic write semaphore in use inode 00:00/-2119956477, count=0
Feb 26 04:41:19 vmebox kernel: iput: device 00:00 inode -2119956477 still has aliases!
I think I mangled something, the code now looks like this:
-----------------------------------------
int found = 0, count = 128;
int repeat_count = 1;
repeat:
entry = inode_in_use.next;
while (entry != &inode_in_use) {
struct list_head *tmp = entry;
/*
* Tough one to time-limit ...
*/
entry = entry->next;
if (!CAN_UNUSE(INODE(tmp))) {
if (current->need_resched) {
INODE(tmp)->i_count++;
spin_unlock(&inode_lock);
schedule();
iput(tmp);
if (repeat_count) {
repeat_count--;
goto repeat;
}
}
continue;
}
list_del(tmp);
list_del(&INODE(tmp)->i_hash);
INIT_LIST_HEAD(&INODE(tmp)->i_hash);
list_add(tmp, freeable);
list_entry(tmp, struct inode, i_list)->i_state = I_FREEING;
found++;
if (!--count)
break;
}
return found;
}
------------------------
It appears Andrea has put in a lowlatency enhancement already in this
code by only allowing 128 inodes to be freed at a time. Perhaps we
dont need the conditional schedule at all? Any rough idea about how long
it takes to free 128 inodes?
Wm
-
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/
This archive was generated by hypermail 2b29 : Tue Feb 29 2000 - 21:00:19 EST