Re: [PATCH -mm 2/2] mm: vmscan: init reclaim_state in do_try_to_free_pages

[Vladimir Davydov <vdavydov@parallels.com>]

On Mon, Jan 12, 2015 at 05:26:34PM -0500, Johannes Weiner wrote:
> On Mon, Jan 12, 2015 at 12:30:38PM +0300, Vladimir Davydov wrote:
> > All users of do_try_to_free_pages() want to have current->reclaim_state
> > set in order to account reclaimed slab pages. So instead of duplicating
> > the reclaim_state initialization code in each call site, let's do it
> > directly in do_try_to_free_pages().
> 
> Couldn't this be contained in shrink_slab() directly?

I had considered this possibility, but finally rejected it, because

 - some slab pages can be reclaimed from shrink_lruvec (e.g.
   buffer_head's); there shouldn't be too many of them though

 - struct reclaim_state looks to me as a generic placeholder for lots of
   reclaim-related stuff, though currently it is only used for counting
   reclaimed slab pages, so IMO it should be initialized before starting
   reclaim

Both arguments are not rock-solid as you can see, so if you think we can
neglect them, I'll do.

Thanks,
Vladimir

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>