Re: [PATCH] fbdefio: add set_page_dirty handler to deferred IO FB

[Markus Armbruster <armbru@redhat.com>]

"Jaya Kumar" <jayakumar.lkml@gmail.com> writes:

> On Tue, Aug 19, 2008 at 2:38 AM, Andrew Morton
> <akpm@linux-foundation.org> wrote:
>> On Tue, 19 Aug 2008 07:02:45 +0100 Ian Campbell <ijc@hellion.org.uk> wrote:
>>
>>> +static int fb_deferred_io_set_page_dirty(struct page *page)
>>> +{
>>> +     if (!PageDirty(page))
>>> +             SetPageDirty(page);
>>> +     return 0;
>>> +}
>>
>> <searches, finds the thread.  "kernel BUG at lib/radix-tree.c:473!">
>>
>> Is there actually any benefit in setting these pages dirty?  Or should
>> this be an empty function?  I see claims in the above thread that this
>> driver uses PG_dirty for optimising writeback but I can't immediately
>> locate any code which actually does that.
>
> Hi Andrew,
>
> I hope I have understood your question. You are right that PG_dirty
> isn't used directly in defio. The defio portion does use each
> page_mkwrite callback to build a list of the pages of the framebuffer
> that were written to and then passes that list to pvfb (in this case).
> pvfb then optimizes writeback by interpreting that list according to
> its framebuffer and sending it to its actual destination. I think
> Markus's code in xenfb_deferred_io() and xenfb_send* is doing the
> latter.

Exactly.  xenfb_deferred_io() is the callback that receives the list
of pages that have been dirtied from fb_deferred_io_work().  It
computes a rectangle covering all these pages, and passes it to
xenfb_refresh().  Which takes care of sending it to the backend.

[...]