From: Alexei Starovoitov <alexei.starovoitov@gmail•com>
To: Eric Dumazet <eric.dumazet@gmail•com>
Cc: Tariq Toukan <ttoukan.linux@gmail•com>,
Eric Dumazet <edumazet@google•com>,
"David S . Miller" <davem@davemloft•net>,
netdev <netdev@vger•kernel.org>,
Tariq Toukan <tariqt@mellanox•com>,
Martin KaFai Lau <kafai@fb•com>,
Willem de Bruijn <willemb@google•com>,
Jesper Dangaard Brouer <brouer@redhat•com>,
Brenden Blanco <bblanco@plumgrid•com>,
Alexei Starovoitov <ast@kernel•org>
Subject: Re: [PATCH net-next 0/9] mlx4: order-0 allocations and page recycling
Date: Tue, 7 Feb 2017 11:05:49 -0800 [thread overview]
Message-ID: <20170207190546.GA51444@ast-mbp.thefacebook.com> (raw)
In-Reply-To: <1486484783.7793.61.camel@edumazet-glaptop3.roam.corp.google.com>
On Tue, Feb 07, 2017 at 08:26:23AM -0800, Eric Dumazet wrote:
> On Tue, 2017-02-07 at 08:06 -0800, Eric Dumazet wrote:
>
Awesome that you've started working on this. I think it's correct approach
and mlx5 should be cleaned up in similar way.
Long term we should be able to move all page alloc/free out of the drivers
completely.
> > /*
> > * make sure we read the CQE after we read the ownership bit
> > */
> > dma_rmb();
> > + prefetch(frags[0].page);
>
> Note that I would like to instead do a prefetch(frags[1].page)
yeah, this two look weird:
+ prefetch(frags[0].page);
+ va = page_address(frags[0].page) + frags[0].page_offset;
on most archs page_addres() is just math (not a load from memory),
but the result != frags[0].page, so I'm missing what are you trying to prefetch?
prefetch(frags[1].page)
is even more confusing. what will it prefetch?
btw we had a patch that was doing prefetch of 'va' of next packet
and it was very helpful. Like this:
pref_index = (index + 1) & ring->size_mask;
pref = ring->rx_info + (pref_index << priv->log_rx_info);
prefetch(page_address(pref->page) + pref->page_offset);
but since you're redesigning rxing->rx_info... not sure how will it fit.
> So I will probably change how ring->rx_info is allocated
>
> wasting all that space and forcing vmalloc() is silly :
>
> tmp = size * roundup_pow_of_two(MLX4_EN_MAX_RX_FRAGS *
> sizeof(struct mlx4_en_rx_alloc));
I think you'd still need roundup_pow_of_two otherwise priv->log_rx_info
optimization won't work.
next prev parent reply other threads:[~2017-02-07 19:14 UTC|newest]
Thread overview: 26+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-02-07 3:02 [PATCH net-next 0/9] mlx4: order-0 allocations and page recycling Eric Dumazet
2017-02-07 3:02 ` [PATCH net-next 1/9] mlx4: use __skb_fill_page_desc() Eric Dumazet
2017-02-07 3:02 ` [PATCH net-next 2/9] mlx4: dma_dir is a mlx4_en_priv attribute Eric Dumazet
2017-02-07 3:02 ` [PATCH net-next 3/9] mlx4: remove order field from mlx4_en_frag_info Eric Dumazet
2017-02-07 3:02 ` [PATCH net-next 4/9] mlx4: get rid of frag_prefix_size Eric Dumazet
2017-02-09 12:28 ` Tariq Toukan
2017-02-09 14:06 ` Eric Dumazet
2017-02-07 3:02 ` [PATCH net-next 5/9] mlx4: rx_headroom is a per port attribute Eric Dumazet
2017-02-07 3:02 ` [PATCH net-next 6/9] mlx4: reduce rx ring page_cache size Eric Dumazet
2017-02-07 3:02 ` [PATCH net-next 7/9] mlx4: removal of frag_sizes[] Eric Dumazet
2017-02-07 3:02 ` [PATCH net-next 8/9] mlx4: use order-0 pages for RX Eric Dumazet
2017-02-07 3:02 ` [PATCH net-next 9/9] mlx4: add page recycling in receive path Eric Dumazet
2017-02-07 16:20 ` Tariq Toukan
2017-02-07 16:34 ` Eric Dumazet
2017-02-08 10:27 ` Tariq Toukan
2017-02-07 15:50 ` [PATCH net-next 0/9] mlx4: order-0 allocations and page recycling Tariq Toukan
2017-02-07 16:06 ` Eric Dumazet
2017-02-07 16:26 ` Eric Dumazet
2017-02-07 16:28 ` Eric Dumazet
2017-02-07 19:05 ` Alexei Starovoitov [this message]
2017-02-07 19:18 ` Eric Dumazet
2017-02-08 9:02 ` Tariq Toukan
2017-02-08 10:29 ` Tariq Toukan
2017-02-08 15:52 ` Eric Dumazet
2017-02-09 12:00 ` Tariq Toukan
2017-02-09 13:31 ` Eric Dumazet
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=20170207190546.GA51444@ast-mbp.thefacebook.com \
--to=alexei.starovoitov@gmail$(echo .)com \
--cc=ast@kernel$(echo .)org \
--cc=bblanco@plumgrid$(echo .)com \
--cc=brouer@redhat$(echo .)com \
--cc=davem@davemloft$(echo .)net \
--cc=edumazet@google$(echo .)com \
--cc=eric.dumazet@gmail$(echo .)com \
--cc=kafai@fb$(echo .)com \
--cc=netdev@vger$(echo .)kernel.org \
--cc=tariqt@mellanox$(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