lkml.org 
[lkml]   [2013]   [Jun]   [5]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
SubjectRe: [v5][PATCH 5/6] mm: vmscan: batch shrink_page_list() locking operations
From
On Tue, Jun 4, 2013 at 1:07 PM, Minchan Kim <minchan@kernel.org> wrote:
> On Tue, Jun 04, 2013 at 09:17:26AM +0800, Hillf Danton wrote:
>> On Tue, Jun 4, 2013 at 4:02 AM, Dave Hansen <dave@sr71.net> wrote:
>> > +/*
>> > + * pages come in here (via remove_list) locked and leave unlocked
>> > + * (on either ret_pages or free_pages)
>> > + *
>> > + * We do this batching so that we free batches of pages with a
>> > + * single mapping->tree_lock acquisition/release. This optimization
>> > + * only makes sense when the pages on remove_list all share a
>> > + * page_mapping(). If this is violated you will BUG_ON().
>> > + */
>> > +static int __remove_mapping_batch(struct list_head *remove_list,
>> > + struct list_head *ret_pages,
>> > + struct list_head *free_pages)
>> > +{
>> > + int nr_reclaimed = 0;
>> > + struct address_space *mapping;
>> > + struct page *page;
>> > + LIST_HEAD(need_free_mapping);
>> > +
>> > + if (list_empty(remove_list))
>> > + return 0;
>> > +
>> > + mapping = page_mapping(lru_to_page(remove_list));
>> > + spin_lock_irq(&mapping->tree_lock);
>> > + while (!list_empty(remove_list)) {
>> > + page = lru_to_page(remove_list);
>> > + BUG_ON(!PageLocked(page));
>> > + BUG_ON(page_mapping(page) != mapping);
>> > + list_del(&page->lru);
>> > +
>> > + if (!__remove_mapping(mapping, page)) {
>> > + unlock_page(page);
>> > + list_add(&page->lru, ret_pages);
>> > + continue;
>> > + }
>> > + list_add(&page->lru, &need_free_mapping);
>> > + }
>> > + spin_unlock_irq(&mapping->tree_lock);
>> > +
>> While reclaiming pages, can we open ears upon IRQ controller,
>> if the page list length is over 10, or even 20?
>
> At the moment, it implicitly could be bounded by SWAP_CLUSTER_MAX and

Could we reclaim a THP currently?

> it's the value used by isolate_migratepages_ranges to enable IRQ.
> I have no idea it's proper value to give a chace to IRQ but at least,
> Dave's code doesn't break the rule.
> If we need a tune for that, it could be a another patch to investigate
> all of places on vmscan.c in near future.
>


\
 
 \ /
  Last update: 2013-06-05 10:01    [W:0.087 / U:0.308 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site