Re: [patch 3/3] mm: PageActive no testset

[Marcelo Tosatti <marcelo.tosatti@cyclades.com>]

On Thu, Jan 19, 2006 at 09:02:26PM +0100, Nick Piggin wrote:
> On Thu, Jan 19, 2006 at 02:52:22PM -0200, Marcelo Tosatti wrote:
> > On Thu, Jan 19, 2006 at 03:50:08PM +0100, Nick Piggin wrote:
> > 
> > > The test-set / test-clear operations also kind of imply that it is
> > > being used for locking or without other synchronisation (usually).
> > 
> > Non-atomic versions such as __ClearPageLRU()/__ClearPageActive() are 
> > not usable, though.
> > 
> 
> Correct. Although I was able to use them in a couple of other places
> in a subsequent patch in the series. I trust you don't see a problem
> with those usages?

Indeed, sorry. Would you mind adding a comment that page->flags must be
accessed atomically otherwise and that __ versions are special as to
when the page cannot be referenced anymore? (its really not obvious)

Also this comments on top of page-flags.h could be updated

 * During disk I/O, PG_locked is used. This bit is set before I/O and
 * reset when I/O completes. page_waitqueue(page) is a wait queue of all tasks
 * waiting for the I/O on this page to complete.

s/PG_locked/PG_writeback/

 * Note that the referenced bit, the page->lru list_head and the active,
 * inactive_dirty and inactive_clean lists are protected by the
 * zone->lru_lock, and *NOT* by the usual PG_locked bit!

inactive_dirty and inactive_clean do not exist anymore