From: Vlastimil Babka <vbabka@suse.cz>
To: js1304@gmail.com, Andrew Morton <akpm@linux-foundation.org>
Cc: mgorman@techsingularity.net, Minchan Kim <minchan@kernel.org>,
Alexander Potapenko <glider@google.com>,
Hugh Dickins <hughd@google.com>, Michal Hocko <mhocko@kernel.org>,
linux-kernel@vger.kernel.org, linux-mm@kvack.org,
Joonsoo Kim <iamjoonsoo.kim@lge.com>
Subject: Re: [PATCH v2 2/7] mm/page_owner: initialize page owner without holding the zone lock
Date: Fri, 3 Jun 2016 12:23:06 +0200 [thread overview]
Message-ID: <b548cad8-e7d1-b742-cb29-caf6263cc65d@suse.cz> (raw)
In-Reply-To: <1464230275-25791-2-git-send-email-iamjoonsoo.kim@lge.com>
On 05/26/2016 04:37 AM, js1304@gmail.com wrote:
> From: Joonsoo Kim <iamjoonsoo.kim@lge.com>
>
> It's not necessary to initialized page_owner with holding the zone lock.
> It would cause more contention on the zone lock although it's not
> a big problem since it is just debug feature. But, it is better
> than before so do it. This is also preparation step to use stackdepot
> in page owner feature. Stackdepot allocates new pages when there is no
> reserved space and holding the zone lock in this case will cause deadlock.
>
> Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
> ---
> mm/compaction.c | 3 +++
> mm/page_alloc.c | 2 --
> mm/page_isolation.c | 9 ++++++---
> 3 files changed, 9 insertions(+), 5 deletions(-)
>
> diff --git a/mm/compaction.c b/mm/compaction.c
> index 8e013eb..6043ef8 100644
> --- a/mm/compaction.c
> +++ b/mm/compaction.c
> @@ -20,6 +20,7 @@
> #include <linux/kasan.h>
> #include <linux/kthread.h>
> #include <linux/freezer.h>
> +#include <linux/page_owner.h>
> #include "internal.h"
>
> #ifdef CONFIG_COMPACTION
> @@ -80,6 +81,8 @@ static void map_pages(struct list_head *list)
> arch_alloc_page(page, order);
> kernel_map_pages(page, nr_pages, 1);
> kasan_alloc_pages(page, order);
> +
> + set_page_owner(page, order, __GFP_MOVABLE);
> if (order)
> split_page(page, order);
>
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index 5134f46..1b1ca57 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -2507,8 +2507,6 @@ int __isolate_free_page(struct page *page, unsigned int order)
> zone->free_area[order].nr_free--;
> rmv_page_order(page);
>
> - set_page_owner(page, order, __GFP_MOVABLE);
> -
> /* Set the pageblock if the isolated page is at least a pageblock */
> if (order >= pageblock_order - 1) {
> struct page *endpage = page + (1 << order) - 1;
> diff --git a/mm/page_isolation.c b/mm/page_isolation.c
> index 612122b..927f5ee 100644
> --- a/mm/page_isolation.c
> +++ b/mm/page_isolation.c
> @@ -7,6 +7,7 @@
> #include <linux/pageblock-flags.h>
> #include <linux/memory.h>
> #include <linux/hugetlb.h>
> +#include <linux/page_owner.h>
> #include "internal.h"
>
> #define CREATE_TRACE_POINTS
> @@ -108,8 +109,6 @@ static void unset_migratetype_isolate(struct page *page, unsigned migratetype)
> if (pfn_valid_within(page_to_pfn(buddy)) &&
> !is_migrate_isolate_page(buddy)) {
> __isolate_free_page(page, order);
> - kernel_map_pages(page, (1 << order), 1);
> - set_page_refcounted(page);
> isolated_page = page;
> }
> }
> @@ -128,8 +127,12 @@ static void unset_migratetype_isolate(struct page *page, unsigned migratetype)
> zone->nr_isolate_pageblock--;
> out:
> spin_unlock_irqrestore(&zone->lock, flags);
> - if (isolated_page)
> + if (isolated_page) {
> + kernel_map_pages(page, (1 << order), 1);
So why we don't need the other stuff done by e.g. map_pages()? For example
arch_alloc_page() and kasan_alloc_pages(). Maybe kasan_free_pages() (called
below via __free_pages() I assume) now doesn't check if the allocation part was
done. But maybe it will start doing that?
See how the multiple places doing similar stuff is fragile? :(
> + set_page_refcounted(page);
> + set_page_owner(page, order, __GFP_MOVABLE);
> __free_pages(isolated_page, order);
This mixing of "isolated_page" and "page" is not a bug, but quite ugly. Can't
isolated_page variable just be a bool?
> + }
> }
>
> static inline struct page *
>
next prev parent reply other threads:[~2016-06-03 10:23 UTC|newest]
Thread overview: 27+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-05-26 2:37 [PATCH v2 1/7] mm/compaction: split freepages without holding the zone lock js1304
2016-05-26 2:37 ` [PATCH v2 2/7] mm/page_owner: initialize page owner " js1304
2016-06-03 10:23 ` Vlastimil Babka [this message]
2016-06-03 12:47 ` Joonsoo Kim
2016-06-06 15:20 ` Vlastimil Babka
2016-05-26 2:37 ` [PATCH v2 3/7] mm/page_owner: copy last_migrate_reason in copy_page_owner() js1304
2016-06-06 13:31 ` Vlastimil Babka
2016-05-26 2:37 ` [PATCH v2 4/7] mm/page_owner: introduce split_page_owner and replace manual handling js1304
2016-05-26 2:37 ` [PATCH v2 5/7] tools/vm/page_owner: increase temporary buffer size js1304
2016-05-26 2:37 ` [PATCH v2 6/7] mm/page_owner: use stackdepot to store stacktrace js1304
2016-06-06 13:56 ` Michal Hocko
2016-06-17 7:25 ` Joonsoo Kim
2016-06-17 9:55 ` Michal Hocko
2016-06-20 6:58 ` Joonsoo Kim
2016-06-06 14:51 ` Vlastimil Babka
2016-06-20 13:04 ` Alexander Potapenko
2016-05-26 2:37 ` [PATCH v2 7/7] mm/page_alloc: introduce post allocation processing on page allocator js1304
2016-06-06 15:21 ` Vlastimil Babka
2016-06-17 7:55 ` Joonsoo Kim
2016-06-03 10:10 ` [PATCH v2 1/7] mm/compaction: split freepages without holding the zone lock Vlastimil Babka
2016-06-03 12:45 ` Joonsoo Kim
2016-06-06 15:19 ` Vlastimil Babka
2016-06-13 20:31 ` Sasha Levin
2016-06-14 5:52 ` Joonsoo Kim
2016-06-14 19:10 ` Sasha Levin
2016-06-15 2:27 ` Joonsoo Kim
2016-06-17 7:27 ` Joonsoo Kim
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=b548cad8-e7d1-b742-cb29-caf6263cc65d@suse.cz \
--to=vbabka@suse.cz \
--cc=akpm@linux-foundation.org \
--cc=glider@google.com \
--cc=hughd@google.com \
--cc=iamjoonsoo.kim@lge.com \
--cc=js1304@gmail.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=mgorman@techsingularity.net \
--cc=mhocko@kernel.org \
--cc=minchan@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).