Re: [PATCH 2/6] writeback: introduce writeback_control.inodes_cleaned

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

On Wed, May 04, 2011 at 07:16:47PM +0800, Christoph Hellwig wrote:
> On Wed, May 04, 2011 at 07:11:30PM +0800, Wu Fengguang wrote:
> > On Wed, May 04, 2011 at 07:05:00PM +0800, Christoph Hellwig wrote:
> > > Same here - this has nothing to do with actual page writeback and really
> > > should stay internal to fs/fs-writeback.c
> > 
> > OK, I'll check how to constrain writeback_control to the minimal scope.
> 
> The basic idea would be to pass the struct wb_writeback_work all the
> way down to writeback_sb_inodes, and initialize the struct writeback_control
> there.

OK.

> This requires adding a few more fields like more_io and

We can in fact just kill more_io and test list_empty(b_more_io)
instead.

> older_than_this to struct wb_writeback_work, and redoing a lot of the
> tracing,

Yeah, older_than_this must still be set in wb_writeback().
The two trace points could be removed for now.

> but it immediately clean things up, e.g. suddently
> wbc.nr_to_write vs work->nr_pages starts to make sense, and instead
> of saving and restoring pages_skipped in writeback_sb_inodes it can
> always start with a clean zero value.

That's right.

Thanks,
Fengguang