lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAOCHtYi01NVp1j=MX+0-z7ygW5tJuoswn8eWTQp+0Z5mMGdeQw@mail.gmail.com>
Date:	Mon, 13 Feb 2012 12:57:58 -0600
From:	Robert Nelson <robertcnelson@...il.com>
To:	Marek Szyprowski <m.szyprowski@...sung.com>
Cc:	linux-kernel@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
	linux-media@...r.kernel.org, linux-mm@...ck.org,
	linaro-mm-sig@...ts.linaro.org, Ohad Ben-Cohen <ohad@...ery.com>,
	Daniel Walker <dwalker@...eaurora.org>,
	Russell King <linux@....linux.org.uk>,
	Arnd Bergmann <arnd@...db.de>,
	Jonathan Corbet <corbet@....net>, Mel Gorman <mel@....ul.ie>,
	Michal Nazarewicz <mina86@...a86.com>,
	Dave Hansen <dave@...ux.vnet.ibm.com>,
	Jesse Barker <jesse.barker@...aro.org>,
	Kyungmin Park <kyungmin.park@...sung.com>,
	Andrew Morton <akpm@...ux-foundation.org>,
	Rob Clark <rob.clark@...aro.org>,
	KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>
Subject: Re: [Linaro-mm-sig] [PATCHv21 12/16] mm: trigger page reclaim in
 alloc_contig_range() to stabilise watermarks

On Fri, Feb 10, 2012 at 11:32 AM, Marek Szyprowski
<m.szyprowski@...sung.com> wrote:
> alloc_contig_range() performs memory allocation so it also should keep
> track on keeping the correct level of memory watermarks. This commit adds
> a call to *_slowpath style reclaim to grab enough pages to make sure that
> the final collection of contiguous pages from freelists will not starve
> the system.
>
> Signed-off-by: Marek Szyprowski <m.szyprowski@...sung.com>
> Signed-off-by: Kyungmin Park <kyungmin.park@...sung.com>
> CC: Michal Nazarewicz <mina86@...a86.com>
> Tested-by: Rob Clark <rob.clark@...aro.org>
> Tested-by: Ohad Ben-Cohen <ohad@...ery.com>
> Tested-by: Benjamin Gaignard <benjamin.gaignard@...aro.org>
> ---
>  include/linux/mmzone.h |    9 +++++++
>  mm/page_alloc.c        |   62 ++++++++++++++++++++++++++++++++++++++++++++++++
>  2 files changed, 71 insertions(+), 0 deletions(-)
>
> diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
> index 82f4fa5..6a6c2cc 100644
> --- a/include/linux/mmzone.h
> +++ b/include/linux/mmzone.h
> @@ -63,8 +63,10 @@ enum {
>
>  #ifdef CONFIG_CMA
>  #  define is_migrate_cma(migratetype) unlikely((migratetype) == MIGRATE_CMA)
> +#  define cma_wmark_pages(zone)        zone->min_cma_pages
>  #else
>  #  define is_migrate_cma(migratetype) false
> +#  define cma_wmark_pages(zone) 0
>  #endif
>
>  #define for_each_migratetype_order(order, type) \
> @@ -371,6 +373,13 @@ struct zone {
>        /* see spanned/present_pages for more description */
>        seqlock_t               span_seqlock;
>  #endif
> +#ifdef CONFIG_CMA
> +       /*
> +        * CMA needs to increase watermark levels during the allocation
> +        * process to make sure that the system is not starved.
> +        */
> +       unsigned long           min_cma_pages;
> +#endif
>        struct free_area        free_area[MAX_ORDER];
>
>  #ifndef CONFIG_SPARSEMEM
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index 793c4e4..2fedd36 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -5035,6 +5035,11 @@ static void __setup_per_zone_wmarks(void)
>
>                zone->watermark[WMARK_LOW]  = min_wmark_pages(zone) + (tmp >> 2);
>                zone->watermark[WMARK_HIGH] = min_wmark_pages(zone) + (tmp >> 1);
> +
> +               zone->watermark[WMARK_MIN] += cma_wmark_pages(zone);
> +               zone->watermark[WMARK_LOW] += cma_wmark_pages(zone);
> +               zone->watermark[WMARK_HIGH] += cma_wmark_pages(zone);
> +
>                setup_zone_migrate_reserve(zone);
>                spin_unlock_irqrestore(&zone->lock, flags);
>        }
> @@ -5637,6 +5642,56 @@ static int __alloc_contig_migrate_range(unsigned long start, unsigned long end)
>        return ret > 0 ? 0 : ret;
>  }
>
> +/*
> + * Update zone's cma pages counter used for watermark level calculation.
> + */
> +static inline void __update_cma_wmark_pages(struct zone *zone, int count)
> +{
> +       unsigned long flags;
> +       spin_lock_irqsave(&zone->lock, flags);
> +       zone->min_cma_pages += count;
> +       spin_unlock_irqrestore(&zone->lock, flags);
> +       setup_per_zone_wmarks();
> +}
> +
> +/*
> + * Trigger memory pressure bump to reclaim some pages in order to be able to
> + * allocate 'count' pages in single page units. Does similar work as
> + *__alloc_pages_slowpath() function.
> + */
> +static int __reclaim_pages(struct zone *zone, gfp_t gfp_mask, int count)
> +{
> +       enum zone_type high_zoneidx = gfp_zone(gfp_mask);
> +       struct zonelist *zonelist = node_zonelist(0, gfp_mask);
> +       int did_some_progress = 0;
> +       int order = 1;
> +       unsigned long watermark;
> +
> +       /*
> +        * Increase level of watermarks to force kswapd do his job
> +        * to stabilise at new watermark level.
> +        */
> +       __modify_min_cma_pages(zone, count);

Hi Marek,   This ^^^ function doesn't seem to exist in this patchset,
is it in another set posted to lkml?

While (build error)testing this patchset on v3.3-rc3 with the
beagle/panda omapdrm driver..

mm/page_alloc.c: In function ‘__reclaim_pages’:
mm/page_alloc.c:5674:2: error: implicit declaration of function
‘__modify_min_cma_pages’ [-Werror=implicit-function-declaration]
cc1: some warnings being treated as errors

make[1]: *** [mm/page_alloc.o] Error 1
make: *** [mm] Error 2
make: *** Waiting for unfinished jobs....

:/KERNEL$ grep -R "modify_min_cma_pages" ./*
./mm/page_alloc.c:      __modify_min_cma_pages(zone, count);
./mm/page_alloc.c:      __modify_min_cma_pages(zone, -count);

Regards,

-- 
Robert Nelson
http://www.rcn-ee.com/
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ