Re: Bug in kernel 2.6.31, Slow wb_kupdate writeout

[Wu Fengguang <fengguang.wu@intel.com>]

On Thu, Jul 30, 2009 at 09:28:07AM +0800, Martin Bligh wrote:
> On Wed, Jul 29, 2009 at 5:19 PM, Martin Bligh<mbligh@google.com> wrote:
> > BTW, can you explain this code at the bottom of generic_sync_sb_inodes
> > for me?
> >
> >                if (wbc->nr_to_write <= 0) {
> >                        wbc->more_io = 1;
> >                        break;
> >                }
> >
> > I don't understand why we are setting more_io here? AFAICS, more_io
> > means there's more stuff to write ... I would think we'd set this if
> > nr_to_write was > 0 ?
> >
> > Or just have the section below brought up above this
> > break check and do:
> >
> > if (!list_empty(&sb->s_more_io) || !list_empty(&sb->s_io))
> >        wbc->more_io = 1;
> >
> > Am I just misunderstanding the intent of more_io ?
> 
> I am thinking along the lines of:

On closer looks I found this line:

                if (inode_dirtied_after(inode, start))
                        break;

In this case "list_empty(&sb->s_io)" is not a good criteria:
here we are breaking away for some other reasons, and shall
not touch wbc.more_io.

So let's stick with the current code?

Thanks,
Fengguang

> @@ -638,13 +609,11 @@ sync_sb_inodes(struct super_block *sb, s
>                 iput(inode);
>                 cond_resched();
>                 spin_lock(&inode_lock);
> -               if (wbc->nr_to_write <= 0) {
> -                       wbc->more_io = 1;
> +               if (wbc->nr_to_write <= 0)
>                         break;
> -               }
> -               if (!list_empty(&sb->s_more_io))
> -                       wbc->more_io = 1;
>         }
> +       if (!list_empty(&sb->s_more_io) || !list_empty(&sb->s_io)
> +               wbc->more_io = 1;
>         return;         /* Leave any unwritten inodes on s_io */
>  }