From: Nitin Gupta <ngupta@vflare.org>
To: Pekka Enberg <penberg@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>,
Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
Seth Jennings <sjenning@linux.vnet.ibm.com>,
Dan Magenheimer <dan.magenheimer@oracle.com>,
Andrew Morton <akpm@linux-foundation.org>,
linux-kernel@vger.kernel.org, linux-mm@kvack.org,
cl@linux-foundation.org
Subject: Re: [PATCH 4/4] zsmalloc: zsmalloc: align cache line size
Date: Mon, 07 May 2012 08:40:28 -0400 [thread overview]
Message-ID: <4FA7C2BC.2090400@vflare.org> (raw)
In-Reply-To: <alpine.LFD.2.02.1205071038090.2851@tux.localdomain>
On 5/7/12 3:41 AM, Pekka Enberg wrote:
> On Fri, 4 May 2012, Minchan Kim wrote:
>>>> It's a overkill to align pool size with PAGE_SIZE to avoid
>>>> false-sharing. This patch aligns it with just cache line size.
>>>>
>>>> Signed-off-by: Minchan Kim<minchan@kernel.org>
>>>> ---
>>>> drivers/staging/zsmalloc/zsmalloc-main.c | 6 +++---
>>>> 1 file changed, 3 insertions(+), 3 deletions(-)
>>>>
>>>> diff --git a/drivers/staging/zsmalloc/zsmalloc-main.c
>>>> b/drivers/staging/zsmalloc/zsmalloc-main.c
>>>> index 51074fa..3991b03 100644
>>>> --- a/drivers/staging/zsmalloc/zsmalloc-main.c
>>>> +++ b/drivers/staging/zsmalloc/zsmalloc-main.c
>>>> @@ -489,14 +489,14 @@ fail:
>>>>
>>>> struct zs_pool *zs_create_pool(const char *name, gfp_t flags)
>>>> {
>>>> - int i, error, ovhd_size;
>>>> + int i, error;
>>>> struct zs_pool *pool;
>>>>
>>>> if (!name)
>>>> return NULL;
>>>>
>>>> - ovhd_size = roundup(sizeof(*pool), PAGE_SIZE);
>>>> - pool = kzalloc(ovhd_size, GFP_KERNEL);
>>>> + pool = kzalloc(ALIGN(sizeof(*pool), cache_line_size()),
>>>> + GFP_KERNEL);
>>>
>>> a basic question:
>>> Is rounding off allocation size to cache_line_size enough to ensure
>>> that the object is cache-line-aligned? Isn't it possible that even
>>> though the object size is multiple of cache-line, it may still not be
>>> properly aligned and end up sharing cache line with some other
>>> read-mostly object?
>>
>> AFAIK, SLAB allocates object aligned cache-size so I think that problem cannot happen.
>> But needs double check.
>> Cced Pekka.
>
> The kmalloc(size) function only gives you the following guarantees:
>
> (1) The allocated object is _at least_ 'size' bytes.
>
> (2) The returned pointer is aligned to ARCH_KMALLOC_MINALIGN.
>
> Anything beyond that is implementation detail and probably will break if
> you switch between SLAB/SLUB/SLOB.
>
> Pekka
So, we can probably leave it as is (PAGE_SIZE aligned) or use
kmem_cache_create(...,SLAB_HWCACHE_ALIGN,...) for allocating 'struct
zs_pool's.
zcache can potentially create a lot of pools, so the latter will save
some memory.
Thanks,
Nitin
--
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/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
WARNING: multiple messages have this Message-ID (diff)
From: Nitin Gupta <ngupta@vflare.org>
To: Pekka Enberg <penberg@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>,
Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
Seth Jennings <sjenning@linux.vnet.ibm.com>,
Dan Magenheimer <dan.magenheimer@oracle.com>,
Andrew Morton <akpm@linux-foundation.org>,
linux-kernel@vger.kernel.org, linux-mm@kvack.org,
cl@linux-foundation.org
Subject: Re: [PATCH 4/4] zsmalloc: zsmalloc: align cache line size
Date: Mon, 07 May 2012 08:40:28 -0400 [thread overview]
Message-ID: <4FA7C2BC.2090400@vflare.org> (raw)
In-Reply-To: <alpine.LFD.2.02.1205071038090.2851@tux.localdomain>
On 5/7/12 3:41 AM, Pekka Enberg wrote:
> On Fri, 4 May 2012, Minchan Kim wrote:
>>>> It's a overkill to align pool size with PAGE_SIZE to avoid
>>>> false-sharing. This patch aligns it with just cache line size.
>>>>
>>>> Signed-off-by: Minchan Kim<minchan@kernel.org>
>>>> ---
>>>> drivers/staging/zsmalloc/zsmalloc-main.c | 6 +++---
>>>> 1 file changed, 3 insertions(+), 3 deletions(-)
>>>>
>>>> diff --git a/drivers/staging/zsmalloc/zsmalloc-main.c
>>>> b/drivers/staging/zsmalloc/zsmalloc-main.c
>>>> index 51074fa..3991b03 100644
>>>> --- a/drivers/staging/zsmalloc/zsmalloc-main.c
>>>> +++ b/drivers/staging/zsmalloc/zsmalloc-main.c
>>>> @@ -489,14 +489,14 @@ fail:
>>>>
>>>> struct zs_pool *zs_create_pool(const char *name, gfp_t flags)
>>>> {
>>>> - int i, error, ovhd_size;
>>>> + int i, error;
>>>> struct zs_pool *pool;
>>>>
>>>> if (!name)
>>>> return NULL;
>>>>
>>>> - ovhd_size = roundup(sizeof(*pool), PAGE_SIZE);
>>>> - pool = kzalloc(ovhd_size, GFP_KERNEL);
>>>> + pool = kzalloc(ALIGN(sizeof(*pool), cache_line_size()),
>>>> + GFP_KERNEL);
>>>
>>> a basic question:
>>> Is rounding off allocation size to cache_line_size enough to ensure
>>> that the object is cache-line-aligned? Isn't it possible that even
>>> though the object size is multiple of cache-line, it may still not be
>>> properly aligned and end up sharing cache line with some other
>>> read-mostly object?
>>
>> AFAIK, SLAB allocates object aligned cache-size so I think that problem cannot happen.
>> But needs double check.
>> Cced Pekka.
>
> The kmalloc(size) function only gives you the following guarantees:
>
> (1) The allocated object is _at least_ 'size' bytes.
>
> (2) The returned pointer is aligned to ARCH_KMALLOC_MINALIGN.
>
> Anything beyond that is implementation detail and probably will break if
> you switch between SLAB/SLUB/SLOB.
>
> Pekka
So, we can probably leave it as is (PAGE_SIZE aligned) or use
kmem_cache_create(...,SLAB_HWCACHE_ALIGN,...) for allocating 'struct
zs_pool's.
zcache can potentially create a lot of pools, so the latter will save
some memory.
Thanks,
Nitin
next prev parent reply other threads:[~2012-05-07 12:40 UTC|newest]
Thread overview: 104+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-05-03 6:40 [PATCH 1/4] zsmalloc: rename zspage_order with zspage_pages Minchan Kim
2012-05-03 6:40 ` Minchan Kim
2012-05-03 6:40 ` [PATCH 2/4] zsmalloc: add/fix function comment Minchan Kim
2012-05-03 6:40 ` Minchan Kim
2012-05-03 13:19 ` Nitin Gupta
2012-05-03 13:19 ` Nitin Gupta
2012-05-03 6:40 ` [PATCH 3/4] zsmalloc use zs_handle instead of void * Minchan Kim
2012-05-03 6:40 ` Minchan Kim
2012-05-03 13:32 ` Nitin Gupta
2012-05-03 13:32 ` Nitin Gupta
2012-05-03 15:23 ` Seth Jennings
2012-05-03 15:23 ` Seth Jennings
2012-05-04 2:24 ` Minchan Kim
2012-05-04 2:24 ` Minchan Kim
2012-05-07 15:01 ` Seth Jennings
2012-05-07 15:01 ` Seth Jennings
2012-05-09 20:19 ` Greg Kroah-Hartman
2012-05-09 20:19 ` Greg Kroah-Hartman
2012-05-10 2:03 ` Minchan Kim
2012-05-10 2:03 ` Minchan Kim
2012-05-10 14:02 ` Konrad Rzeszutek Wilk
2012-05-10 14:02 ` Konrad Rzeszutek Wilk
2012-05-10 14:12 ` Greg Kroah-Hartman
2012-05-10 14:12 ` Greg Kroah-Hartman
2012-05-10 14:47 ` Nitin Gupta
2012-05-10 14:47 ` Nitin Gupta
2012-05-10 15:00 ` Minchan Kim
2012-05-10 15:00 ` Minchan Kim
2012-05-10 15:11 ` Seth Jennings
2012-05-10 15:11 ` Seth Jennings
2012-05-10 15:19 ` Minchan Kim
2012-05-10 15:19 ` Minchan Kim
2012-05-10 15:19 ` Greg Kroah-Hartman
2012-05-10 15:19 ` Greg Kroah-Hartman
2012-05-10 16:29 ` Nitin Gupta
2012-05-10 16:29 ` Nitin Gupta
2012-05-10 16:44 ` Greg Kroah-Hartman
2012-05-10 16:44 ` Greg Kroah-Hartman
2012-05-10 17:24 ` Nitin Gupta
2012-05-10 17:24 ` Nitin Gupta
2012-05-10 17:33 ` Konrad Rzeszutek Wilk
2012-05-10 17:33 ` Konrad Rzeszutek Wilk
2012-05-10 23:24 ` Minchan Kim
2012-05-10 23:24 ` Minchan Kim
2012-05-10 23:50 ` Dan Magenheimer
2012-05-10 23:50 ` Dan Magenheimer
2012-05-11 0:14 ` Minchan Kim
2012-05-11 0:14 ` Minchan Kim
2012-05-11 16:31 ` Dan Magenheimer
2012-05-11 16:31 ` Dan Magenheimer
2012-05-11 19:29 ` Konrad Rzeszutek Wilk
2012-05-11 19:29 ` Konrad Rzeszutek Wilk
2012-05-11 21:49 ` Seth Jennings
2012-05-11 21:49 ` Seth Jennings
2012-05-14 2:26 ` Minchan Kim
2012-05-14 2:26 ` Minchan Kim
2012-05-11 19:28 ` Konrad Rzeszutek Wilk
2012-05-11 19:28 ` Konrad Rzeszutek Wilk
2012-05-14 2:18 ` Minchan Kim
2012-05-14 2:18 ` Minchan Kim
2012-05-15 1:57 ` Nitin Gupta
2012-05-15 1:57 ` Nitin Gupta
2012-05-15 2:21 ` Minchan Kim
2012-05-15 2:21 ` Minchan Kim
2012-05-15 15:04 ` Konrad Rzeszutek Wilk
2012-05-15 15:04 ` Konrad Rzeszutek Wilk
2012-05-16 1:36 ` Minchan Kim
2012-05-16 1:36 ` Minchan Kim
2012-05-16 11:06 ` Konrad Rzeszutek Wilk
2012-05-16 11:06 ` Konrad Rzeszutek Wilk
2012-05-03 6:40 ` [PATCH 4/4] zsmalloc: zsmalloc: align cache line size Minchan Kim
2012-05-03 6:40 ` Minchan Kim
2012-05-03 13:58 ` Nitin Gupta
2012-05-03 13:58 ` Nitin Gupta
2012-05-04 2:27 ` Minchan Kim
2012-05-04 2:27 ` Minchan Kim
2012-05-07 7:41 ` Pekka Enberg
2012-05-07 7:41 ` Pekka Enberg
2012-05-07 12:40 ` Nitin Gupta [this message]
2012-05-07 12:40 ` Nitin Gupta
2012-05-08 1:34 ` Minchan Kim
2012-05-08 1:34 ` Minchan Kim
2012-05-08 14:00 ` Dan Magenheimer
2012-05-08 14:00 ` Dan Magenheimer
2012-05-09 0:58 ` Minchan Kim
2012-05-09 0:58 ` Minchan Kim
2012-05-09 3:08 ` Dan Magenheimer
2012-05-09 3:08 ` Dan Magenheimer
2012-05-09 4:07 ` Minchan Kim
2012-05-09 4:07 ` Minchan Kim
2012-05-11 0:03 ` Dan Magenheimer
2012-05-11 0:03 ` Dan Magenheimer
2012-05-11 0:25 ` Minchan Kim
2012-05-11 0:25 ` Minchan Kim
2012-05-11 19:06 ` Konrad Rzeszutek Wilk
2012-05-11 19:06 ` Konrad Rzeszutek Wilk
2012-05-14 1:55 ` Minchan Kim
2012-05-14 1:55 ` Minchan Kim
2012-05-15 15:18 ` Konrad Rzeszutek Wilk
2012-05-15 15:18 ` Konrad Rzeszutek Wilk
2012-05-16 1:44 ` Minchan Kim
2012-05-16 1:44 ` Minchan Kim
2012-05-03 13:18 ` [PATCH 1/4] zsmalloc: rename zspage_order with zspage_pages Nitin Gupta
2012-05-03 13:18 ` Nitin Gupta
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=4FA7C2BC.2090400@vflare.org \
--to=ngupta@vflare.org \
--cc=akpm@linux-foundation.org \
--cc=cl@linux-foundation.org \
--cc=dan.magenheimer@oracle.com \
--cc=gregkh@linuxfoundation.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=minchan@kernel.org \
--cc=penberg@kernel.org \
--cc=sjenning@linux.vnet.ibm.com \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.