Re: Problem in log_do_checkpoint()?

2005-04-12 Thread Stephen C. Tweedie
Hi, On Mon, 2005-04-11 at 12:36, Jan Kara wrote: > > The prevention of multiple writes in this case should also improve > > performance a little. > > > > That ought to be pretty straightforward, I think. The existing cases > > where we remove buffers from a checkpoint shouldn't have to care abo

Re: Problem in log_do_checkpoint()?

2005-04-11 Thread Stephen C. Tweedie
Hi, On Fri, 2005-04-08 at 18:14, Badari Pulavarty wrote: > I get OOPs in log_do_checkpoint() while using ext3 quotas. > Is this anyway related to what you are working on ? > > Unable to handle kernel NULL pointer dereference at virtual address > Doesn't look like it, no. If we underst

Re: Problem in log_do_checkpoint()?

2005-04-11 Thread Jan Kara
Hello, > On Mon, 2005-04-04 at 10:04, Jan Kara wrote: > > > In log_do_checkpoint() we go through the t_checkpoint_list of a > > transaction and call __flush_buffer() on each buffer. Suppose there is > > just one buffer on the list and it is dirty. __flush_buffer() sees it and > > puts it to a

Re: Problem in log_do_checkpoint()?

2005-04-11 Thread Jan Kara
> I get OOPs in log_do_checkpoint() while using ext3 quotas. > Is this anyway related to what you are working on ? Nope, it does not seem to be the same problem. In theory it could be a bug Stephen fixed some time ago - could you try to reproduce the problem with 2.6.12-rc2 (it contains the fix)

Re: Problem in log_do_checkpoint()?

2005-04-08 Thread Badari Pulavarty
I get OOPs in log_do_checkpoint() while using ext3 quotas. Is this anyway related to what you are working on ? Unable to handle kernel NULL pointer dereference at virtual address printing eip: 801aeee1 *pde = 52b31001 Oops: 0002 [#1] PREEMPT SMP Modules linked in: CPU:3 EIP:006

Re: Problem in log_do_checkpoint()?

2005-04-08 Thread Stephen C. Tweedie
Hi, On Mon, 2005-04-04 at 10:04, Jan Kara wrote: > In log_do_checkpoint() we go through the t_checkpoint_list of a > transaction and call __flush_buffer() on each buffer. Suppose there is > just one buffer on the list and it is dirty. __flush_buffer() sees it and > puts it to an array of buffer

Problem in log_do_checkpoint()?

2005-04-04 Thread Jan Kara
Hello, I've been looking through the JBD code when trying to understand the assertion failure in log_do_checkpoint() (it was on old SUSE 2.6.5 kernel though the reporter claims to be able to get the failure even with the Stephen's patch fixing a race with journal_put_journal_head()) and I've s