All of lore.kernel.org
 help / color / mirror / Atom feed
From: Arnd Bergmann <arnd@arndb.de>
To: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Cc: Christoph Hellwig <hch@infradead.org>,
	David Miller <davem@davemloft.net>,
	netdev@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-mm@kvack.org
Subject: Re: [PATCH 1/1] network memory allocator.
Date: Wed, 16 Aug 2006 13:27:02 +0200	[thread overview]
Message-ID: <200608161327.02826.arnd@arndb.de> (raw)
In-Reply-To: <20060816090028.GA25476@2ka.mipt.ru>

On Wednesday 16 August 2006 11:00, Evgeniy Polyakov wrote:
> There is drawback here - if data was allocated on CPU wheere NIC is
> "closer" and then processed on different CPU it will cost more than 
> in case where buffer was allocated on CPU where it will be processed.
> 
> But from other point of view, most of the adapters preallocate set of
> skbs, and with msi-x help there will be a possibility to bind irq and
> processing to the CPU where data was origianlly allocated.
> 
> So I would like to know how to determine which node should be used for
> allocation. Changes of __get_user_pages() to alloc_pages_node() are
> trivial.

There are two separate memory areas here: Your own metadata used by the
allocator and the memory used for skb data.

avl_node_array[cpu] and avl_container_array[cpu] are only designed to
be accessed only by the local cpu, so these should be done like

avl_node_array[cpu] = kmalloc_node(AVL_NODE_PAGES * sizeof(void *),
			GFP_KERNEL, cpu_to_node(cpu));

or you could make the whole array DEFINE_PER_CPU(void *, which would
waste some space in the kernel object file.

Now for the actual pages you get with __get_free_pages(), doing the
same (alloc_pages_node), will help accessing your avl_container 
members, but may not be the best solution for getting the data
next to the network adapter.

	Arnd <><

WARNING: multiple messages have this Message-ID (diff)
From: Arnd Bergmann <arnd@arndb.de>
To: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Cc: Christoph Hellwig <hch@infradead.org>,
	David Miller <davem@davemloft.net>,
	netdev@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-mm@kvack.org
Subject: Re: [PATCH 1/1] network memory allocator.
Date: Wed, 16 Aug 2006 13:27:02 +0200	[thread overview]
Message-ID: <200608161327.02826.arnd@arndb.de> (raw)
In-Reply-To: <20060816090028.GA25476@2ka.mipt.ru>

On Wednesday 16 August 2006 11:00, Evgeniy Polyakov wrote:
> There is drawback here - if data was allocated on CPU wheere NIC is
> "closer" and then processed on different CPU it will cost more than 
> in case where buffer was allocated on CPU where it will be processed.
> 
> But from other point of view, most of the adapters preallocate set of
> skbs, and with msi-x help there will be a possibility to bind irq and
> processing to the CPU where data was origianlly allocated.
> 
> So I would like to know how to determine which node should be used for
> allocation. Changes of __get_user_pages() to alloc_pages_node() are
> trivial.

There are two separate memory areas here: Your own metadata used by the
allocator and the memory used for skb data.

avl_node_array[cpu] and avl_container_array[cpu] are only designed to
be accessed only by the local cpu, so these should be done like

avl_node_array[cpu] = kmalloc_node(AVL_NODE_PAGES * sizeof(void *),
			GFP_KERNEL, cpu_to_node(cpu));

or you could make the whole array DEFINE_PER_CPU(void *, which would
waste some space in the kernel object file.

Now for the actual pages you get with __get_free_pages(), doing the
same (alloc_pages_node), will help accessing your avl_container 
members, but may not be the best solution for getting the data
next to the network adapter.

	Arnd <><

--
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>

  parent reply	other threads:[~2006-08-16 11:27 UTC|newest]

Thread overview: 104+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2006-08-14 11:04 [PATCH 1/1] network memory allocator Evgeniy Polyakov
2006-08-14 11:04 ` Evgeniy Polyakov
2006-08-14 11:22 ` David Miller
2006-08-14 11:22   ` David Miller, Evgeniy Polyakov
2006-08-14 11:32   ` Evgeniy Polyakov
2006-08-14 11:32     ` Evgeniy Polyakov
2006-08-14 11:40 ` Andi Kleen
2006-08-14 11:40   ` Andi Kleen
2006-08-14 11:46   ` Evgeniy Polyakov
2006-08-14 11:46     ` Evgeniy Polyakov
2006-08-14 12:07 ` Keith Owens
2006-08-14 12:07   ` Keith Owens
2006-08-14 12:20   ` Evgeniy Polyakov
2006-08-14 12:20     ` Evgeniy Polyakov
2006-08-14 17:42     ` Rick Jones
2006-08-14 17:42       ` Rick Jones
2006-08-14 20:15       ` David Miller
2006-08-14 20:15         ` David Miller, Rick Jones
2006-08-14 12:25 ` Peter Zijlstra
2006-08-14 12:25   ` Peter Zijlstra
2006-08-14 12:35   ` Evgeniy Polyakov
2006-08-14 12:35     ` Evgeniy Polyakov
2006-08-14 12:38     ` Evgeniy Polyakov
2006-08-14 12:38       ` Evgeniy Polyakov
2006-08-15 10:55     ` Peter Zijlstra
2006-08-15 10:55       ` Peter Zijlstra
2006-08-15 11:26       ` Evgeniy Polyakov
2006-08-15 11:26         ` Evgeniy Polyakov
2006-08-15 12:03         ` Peter Zijlstra
2006-08-15 12:03           ` Peter Zijlstra
2006-08-15 12:34           ` Evgeniy Polyakov
2006-08-15 12:34             ` Evgeniy Polyakov
2006-08-15 13:49             ` Peter Zijlstra
2006-08-15 13:49               ` Peter Zijlstra
2006-08-15 14:15               ` Evgeniy Polyakov
2006-08-15 14:15                 ` Evgeniy Polyakov
2006-08-15 14:48                 ` Peter Zijlstra
2006-08-15 14:48                   ` Peter Zijlstra
2006-08-15 15:05                   ` Evgeniy Polyakov
2006-08-15 15:05                     ` Evgeniy Polyakov
2006-08-15 15:07                     ` Evgeniy Polyakov
2006-08-15 15:07                       ` Evgeniy Polyakov
2006-08-15 17:42                     ` Peter Zijlstra
2006-08-15 17:42                       ` Peter Zijlstra
2006-08-15 17:49                       ` Evgeniy Polyakov
2006-08-15 17:49                         ` Evgeniy Polyakov
2006-08-16  2:52                 ` Bill Fink
2006-08-16  2:52                   ` Bill Fink
2006-08-16  5:38                   ` Evgeniy Polyakov
2006-08-16  5:38                     ` Evgeniy Polyakov
2006-08-14 17:46 ` Rick Jones
2006-08-14 17:46   ` Rick Jones
2006-08-14 19:42   ` Evgeniy Polyakov
2006-08-14 19:42     ` Evgeniy Polyakov
2006-08-15  7:27 ` Andrew Morton
2006-08-15  7:27   ` Andrew Morton
2006-08-15  8:08   ` Andi Kleen
2006-08-15  8:08     ` Andi Kleen
2006-08-15 10:02     ` Evgeniy Polyakov
2006-08-15 10:02       ` Evgeniy Polyakov
2006-08-15 10:27       ` David Miller
2006-08-15 10:27         ` David Miller, Evgeniy Polyakov
2006-08-15  9:20   ` Evgeniy Polyakov
2006-08-15  9:20     ` Evgeniy Polyakov
2006-08-15 20:21 ` Arnd Bergmann
2006-08-16  5:35   ` Evgeniy Polyakov
2006-08-16  8:48     ` Christoph Hellwig
2006-08-16  8:48       ` Christoph Hellwig
2006-08-16  9:00       ` Evgeniy Polyakov
2006-08-16  9:00         ` Evgeniy Polyakov
2006-08-16  9:05         ` David Miller
2006-08-16  9:05           ` David Miller, Evgeniy Polyakov
2006-08-16  9:10           ` Christoph Hellwig
2006-08-16  9:10             ` Christoph Hellwig
2006-08-16  9:32             ` Evgeniy Polyakov
2006-08-16  9:32               ` Evgeniy Polyakov
2006-08-16  9:38               ` Christoph Hellwig
2006-08-16  9:38                 ` Christoph Hellwig
2006-08-16  9:40                 ` David Miller
2006-08-16  9:40                   ` David Miller, Christoph Hellwig
2006-08-16  9:44                   ` Christoph Hellwig
2006-08-16  9:44                     ` Christoph Hellwig
2006-08-16  9:42         ` Christoph Hellwig
2006-08-16  9:42           ` Christoph Hellwig
2006-08-16 11:27         ` Arnd Bergmann [this message]
2006-08-16 11:27           ` Arnd Bergmann
2006-08-16 12:00           ` Evgeniy Polyakov
2006-08-16 12:00             ` Evgeniy Polyakov
2006-08-16 12:25       ` Andi Kleen
2006-08-16 12:25         ` Andi Kleen
2006-08-18  2:25         ` Christoph Lameter
2006-08-18  2:25           ` Christoph Lameter
2006-08-18  9:29           ` Andi Kleen
2006-08-18  9:29             ` Andi Kleen
2006-08-18  8:51             ` David Miller
2006-08-18  8:51               ` David Miller, Andi Kleen
2006-08-18 17:04             ` Christoph Lameter
2006-08-18 17:04               ` Christoph Lameter
2006-08-16  7:51 ` [PATCH2 " Evgeniy Polyakov
2006-08-16  7:51   ` Evgeniy Polyakov
2006-08-16 16:57   ` Stephen Hemminger
2006-08-16 16:57     ` Stephen Hemminger
2006-08-16 19:27     ` Evgeniy Polyakov
2006-08-16 19:27       ` Evgeniy Polyakov

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=200608161327.02826.arnd@arndb.de \
    --to=arnd@arndb.de \
    --cc=davem@davemloft.net \
    --cc=hch@infradead.org \
    --cc=johnpol@2ka.mipt.ru \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=netdev@vger.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 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.