public inbox for netdev@vger.kernel.org 
 help / color / mirror / Atom feed
From: Jesper Dangaard Brouer <brouer@redhat•com>
To: David Miller <davem@davemloft•net>
Cc: ttoukan.linux@gmail•com, edumazet@google•com,
	alexander.duyck@gmail•com, netdev@vger•kernel.org,
	tariqt@mellanox•com, kafai@fb•com, saeedm@mellanox•com,
	willemb@google•com, bblanco@plumgrid•com, ast@kernel•org,
	eric.dumazet@gmail•com, linux-mm@kvack•org, brouer@redhat•com
Subject: Re: [PATCH v3 net-next 08/14] mlx4: use order-0 pages for RX
Date: Tue, 14 Feb 2017 20:38:22 +0100	[thread overview]
Message-ID: <20170214203822.72d41268@redhat.com> (raw)
In-Reply-To: <20170214.120426.2032015522492111544.davem@davemloft.net>

On Tue, 14 Feb 2017 12:04:26 -0500 (EST)
David Miller <davem@davemloft•net> wrote:

> From: Tariq Toukan <ttoukan.linux@gmail•com>
> Date: Tue, 14 Feb 2017 16:56:49 +0200
> 
> > Internally, I already implemented "dynamic page-cache" and
> > "page-reuse" mechanisms in the driver, and together they totally
> > bridge the performance gap.  

It sounds like you basically implemented a page_pool scheme...

> I worry about a dynamically growing page cache inside of drivers
> because it is invisible to the rest of the kernel.

Exactly, that is why I wanted a separate standardized thing, I call the
page_pool, which is part of the MM-tree and interacts with the page
allocator.  E.g. it must implement/support a way the page allocator can
reclaim pages from it (admit I didn't implement this in RFC patches).


> It responds only to local needs.

Generally true, but a side effect of recycling these pages, result in
less fragmentation of the page allocator/buddy system.


> The price of the real page allocator comes partly because it can
> respond to global needs.
> 
> If a driver consumes some unreasonable percentage of system memory, it
> is keeping that memory from being used from other parts of the system
> even if it would be better for networking to be slightly slower with
> less cache because that other thing that needs memory is more
> important.

(That is why I want to have OOM protection at device level, with the
recycle feedback from page pool we have this knowledge, and further I
wanted to allow blocking a specific RX queue[1])
[1] https://prototype-kernel.readthedocs.io/en/latest/vm/page_pool/design/memory_model_nic.html#userspace-delivery-and-oom


> I think this is one of the primary reasons that the MM guys severely
> chastise us when we build special purpose local caches into networking
> facilities.
> 
> And the more I think about it the more I think they are right.

+1
 
> One path I see around all of this is full integration.  Meaning that
> we can free pages into the page allocator which are still DMA mapped.
> And future allocations from that device are prioritized to take still
> DMA mapped objects.

I like this idea.  Are you saying that this should be done per DMA
engine or per device?

If this is per device, it is almost the page_pool idea.  

 
> Yes, we still need to make the page allocator faster, but this kind of
> work helps everyone not just 100GB ethernet NICs.

True.  And Mel already have some generic improvements to the page
allocator queued for the next merge.  And I have the responsibility to
get the bulking API into shape.

-- 
Best regards,
  Jesper Dangaard Brouer
  MSc.CS, Principal Kernel Engineer at Red Hat
  LinkedIn: http://www.linkedin.com/in/brouer

--
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:[~2017-02-14 19:38 UTC|newest]

Thread overview: 71+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-02-13 19:58 [PATCH v3 net-next 00/14] mlx4: order-0 allocations and page recycling Eric Dumazet
2017-02-13 19:58 ` [PATCH v3 net-next 01/14] mlx4: use __skb_fill_page_desc() Eric Dumazet
2017-02-13 19:58 ` [PATCH v3 net-next 02/14] mlx4: dma_dir is a mlx4_en_priv attribute Eric Dumazet
2017-02-13 19:58 ` [PATCH v3 net-next 03/14] mlx4: remove order field from mlx4_en_frag_info Eric Dumazet
2017-02-13 19:58 ` [PATCH v3 net-next 04/14] mlx4: get rid of frag_prefix_size Eric Dumazet
2017-02-13 19:58 ` [PATCH v3 net-next 05/14] mlx4: rx_headroom is a per port attribute Eric Dumazet
2017-02-13 19:58 ` [PATCH v3 net-next 06/14] mlx4: reduce rx ring page_cache size Eric Dumazet
2017-02-13 19:58 ` [PATCH v3 net-next 07/14] mlx4: removal of frag_sizes[] Eric Dumazet
2017-02-13 19:58 ` [PATCH v3 net-next 08/14] mlx4: use order-0 pages for RX Eric Dumazet
2017-02-13 20:51   ` Alexander Duyck
2017-02-13 21:09     ` Eric Dumazet
2017-02-13 23:16       ` Alexander Duyck
2017-02-13 23:22         ` Eric Dumazet
2017-02-13 23:26           ` Alexander Duyck
2017-02-13 23:29             ` Eric Dumazet
2017-02-13 23:47               ` Alexander Duyck
2017-02-14  0:22                 ` Eric Dumazet
2017-02-14  0:34                   ` Alexander Duyck
2017-02-14  0:46                     ` Eric Dumazet
2017-02-14  0:47                       ` Eric Dumazet
2017-02-14  0:57                       ` Eric Dumazet
2017-02-14  1:32                         ` Alexander Duyck
2017-02-14 12:12         ` Jesper Dangaard Brouer
2017-02-14 13:45           ` Eric Dumazet
2017-02-14 14:12             ` Eric Dumazet
2017-02-14 14:56             ` Tariq Toukan
2017-02-14 15:51               ` Eric Dumazet
2017-02-14 16:03                 ` Eric Dumazet
2017-02-14 17:29                 ` Tom Herbert
2017-02-15 16:42                   ` Tariq Toukan
2017-02-15 16:57                     ` Eric Dumazet
2017-02-16 13:08                       ` Tariq Toukan
2017-02-16 15:47                         ` Eric Dumazet
2017-02-16 17:05                         ` Tom Herbert
2017-02-16 17:11                           ` Eric Dumazet
2017-02-16 20:49                             ` Saeed Mahameed
2017-02-16 19:03                           ` David Miller
2017-02-16 21:06                             ` Saeed Mahameed
2017-02-14 17:04               ` David Miller
2017-02-14 17:17                 ` David Laight
2017-02-14 17:22                   ` David Miller
2017-02-14 19:38                 ` Jesper Dangaard Brouer [this message]
2017-02-14 19:59                   ` David Miller
2017-02-14 17:29               ` Alexander Duyck
2017-02-14 18:46                 ` Jesper Dangaard Brouer
2017-02-14 19:02                   ` Eric Dumazet
2017-02-14 20:02                     ` Jesper Dangaard Brouer
2017-02-14 21:56                       ` Eric Dumazet
2017-02-14 19:06                   ` Alexander Duyck
2017-02-14 19:50                     ` Jesper Dangaard Brouer
2017-02-22 16:22   ` Eric Dumazet
2017-02-22 17:23     ` Alexander Duyck
2017-02-22 17:58       ` David Laight
2017-02-22 18:21       ` Eric Dumazet
2017-02-23  1:08         ` Alexander Duyck
2017-02-23  2:06           ` Eric Dumazet
2017-02-23  2:18             ` Alexander Duyck
2017-02-23 14:02               ` Tariq Toukan
2017-02-24  9:42             ` Jesper Dangaard Brouer
2017-03-12 14:57             ` Eric Dumazet
2017-03-12 15:29               ` Eric Dumazet
2017-03-12 15:49                 ` Saeed Mahameed
2017-03-12 16:49                   ` Eric Dumazet
2017-03-13  9:20                     ` Saeed Mahameed
2017-02-13 19:58 ` [PATCH v3 net-next 09/14] mlx4: add page recycling in receive path Eric Dumazet
2017-02-13 19:58 ` [PATCH v3 net-next 10/14] mlx4: add rx_alloc_pages counter in ethtool -S Eric Dumazet
2017-02-13 19:58 ` [PATCH v3 net-next 11/14] mlx4: do not access rx_desc from mlx4_en_process_rx_cq() Eric Dumazet
2017-02-13 19:58 ` [PATCH v3 net-next 12/14] mlx4: factorize page_address() calls Eric Dumazet
2017-02-13 19:58 ` [PATCH v3 net-next 13/14] mlx4: make validate_loopback() more generic Eric Dumazet
2017-02-13 19:58 ` [PATCH v3 net-next 14/14] mlx4: remove duplicate code in mlx4_en_process_rx_cq() Eric Dumazet
2017-02-17 16:00 ` [PATCH v3 net-next 00/14] mlx4: order-0 allocations and page recycling David Miller

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=20170214203822.72d41268@redhat.com \
    --to=brouer@redhat$(echo .)com \
    --cc=alexander.duyck@gmail$(echo .)com \
    --cc=ast@kernel$(echo .)org \
    --cc=bblanco@plumgrid$(echo .)com \
    --cc=davem@davemloft$(echo .)net \
    --cc=edumazet@google$(echo .)com \
    --cc=eric.dumazet@gmail$(echo .)com \
    --cc=kafai@fb$(echo .)com \
    --cc=linux-mm@kvack$(echo .)org \
    --cc=netdev@vger$(echo .)kernel.org \
    --cc=saeedm@mellanox$(echo .)com \
    --cc=tariqt@mellanox$(echo .)com \
    --cc=ttoukan.linux@gmail$(echo .)com \
    --cc=willemb@google$(echo .)com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html
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