netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Jakub Kicinski <kuba@kernel.org>
To: Aurelien Aptel <aaptel@nvidia.com>
Cc: linux-nvme@lists.infradead.org, netdev@vger.kernel.org,
	sagi@grimberg.me, hch@lst.de, kbusch@kernel.org, axboe@fb.com,
	chaitanyak@nvidia.com, davem@davemloft.net,
	aurelien.aptel@gmail.com, smalin@nvidia.com, malin1024@gmail.com,
	ogerlitz@nvidia.com, yorayz@nvidia.com, borisp@nvidia.com
Subject: Re: [PATCH v9 03/25] net/ethtool: add ULP_DDP_{GET,SET} operations for caps and stats
Date: Mon, 23 Jan 2023 14:38:30 -0800	[thread overview]
Message-ID: <20230123143830.60f436ef@kernel.org> (raw)
In-Reply-To: <253o7qprtcq.fsf@mtr-vdi-124.i-did-not-set--mail-host-address--so-tickle-me>

On Mon, 23 Jan 2023 20:36:21 +0200 Aurelien Aptel wrote:
> >> Compact statistics are nested as follows:
> >>
> >>     STATS (nest)
> >>         COUNT (u32)
> >>         COMPACT_VALUES (array of u64)  
> >
> > That's not how other per-cmd stats work, why are you inventing
> > new ways..  
> 
> As we commented in patch 2, dynamic strings are used for ethtool
> forward-compability (being able to list future stats, which we are
> planning) without updating or recompiling.

But this is not how they should be carried.

The string set is retrieved by a separate command, then you request
a string based on the attribute ID (global_stringset() + get_string() 
in ethtool CLI code).

That way long running code or code dumping muliple interfaces can load
strings once and dumps are kept smaller.

> >> +     int     (*get_ulp_ddp_stats)(struct net_device *dev, struct ethtool_ulp_ddp_stats *stats);
> >> +     int     (*set_ulp_ddp_capabilities)(struct net_device *dev, unsigned long *bits);  
> >
> > Why are these two callbacks not in struct ulp_ddp_dev_ops?  
> 
> We were trying to implement these callbacks in alignment with the
> existing ethtool commands, for this reason we implemented it in the
> ethtool API.

ethtool commands mostly talk to HW, note that the feature configuration
(ethtool -k/-K) does not use ethtool ops either.

> > Why does the ethtool API not expose limits?  
> 
> Originally, and before we started adding the netlink interface, we were
> not planning to include the ability to modify the limits as part of this
> series.  We do agree that it now makes sense, but we will add, some
> limits reflect hardware limitations while other could be tweaked by
> users.  Those limits will be per-device and per-protocol. We will
> suggest how to design it.

Alright, I was mostly curious, it's not a requirement for initial
support.

  reply	other threads:[~2023-01-23 22:38 UTC|newest]

Thread overview: 40+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-01-17 15:35 [PATCH v9 00/25] nvme-tcp receive offloads Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 01/25] net: Introduce direct data placement tcp offload Aurelien Aptel
2023-01-19 10:57   ` Paolo Abeni
2023-01-19 19:25     ` Shai Malin
2023-01-20  8:52   ` Paolo Abeni
2023-01-26  9:47     ` Shai Malin
2023-01-26 12:53       ` Paolo Abeni
2023-01-17 15:35 ` [PATCH v9 02/25] net/ethtool: add new stringset ETH_SS_ULP_DDP_{CAPS,STATS} Aurelien Aptel
2023-01-20  2:36   ` Jakub Kicinski
2023-01-23 18:31     ` Aurelien Aptel
2023-01-23 22:32       ` Jakub Kicinski
2023-01-24 12:04         ` Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 03/25] net/ethtool: add ULP_DDP_{GET,SET} operations for caps and stats Aurelien Aptel
2023-01-20  2:41   ` Jakub Kicinski
2023-01-23 18:36     ` Aurelien Aptel
2023-01-23 22:38       ` Jakub Kicinski [this message]
2023-01-24 12:07         ` Aurelien Aptel
2023-01-24 19:55           ` Jakub Kicinski
2023-01-17 15:35 ` [PATCH v9 04/25] Documentation: document netlink ULP_DDP_GET/SET messages Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 05/25] iov_iter: skip copy if src == dst for direct data placement Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 06/25] net/tls,core: export get_netdev_for_sock Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 07/25] nvme-tcp: Add DDP offload control path Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 08/25] nvme-tcp: Add DDP data-path Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 09/25] nvme-tcp: RX DDGST offload Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 10/25] nvme-tcp: Deal with netdevice DOWN events Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 11/25] nvme-tcp: Add modparam to control the ULP offload enablement Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 12/25] Documentation: add ULP DDP offload documentation Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 13/25] net/mlx5e: Rename from tls to transport static params Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 14/25] net/mlx5e: Refactor ico sq polling to get budget Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 15/25] net/mlx5e: Have mdev pointer directly on the icosq structure Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 16/25] net/mlx5e: Refactor doorbell function to allow avoiding a completion Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 17/25] net/mlx5: Add NVMEoTCP caps, HW bits, 128B CQE and enumerations Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 18/25] net/mlx5e: NVMEoTCP, offload initialization Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 19/25] net/mlx5e: TCP flow steering for nvme-tcp acceleration Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 20/25] net/mlx5e: NVMEoTCP, use KLM UMRs for buffer registration Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 21/25] net/mlx5e: NVMEoTCP, queue init/teardown Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 22/25] net/mlx5e: NVMEoTCP, ddp setup and resync Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 23/25] net/mlx5e: NVMEoTCP, async ddp invalidation Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 24/25] net/mlx5e: NVMEoTCP, data-path for DDP+DDGST offload Aurelien Aptel
2023-01-17 15:35 ` [PATCH v9 25/25] net/mlx5e: NVMEoTCP, statistics Aurelien Aptel

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=20230123143830.60f436ef@kernel.org \
    --to=kuba@kernel.org \
    --cc=aaptel@nvidia.com \
    --cc=aurelien.aptel@gmail.com \
    --cc=axboe@fb.com \
    --cc=borisp@nvidia.com \
    --cc=chaitanyak@nvidia.com \
    --cc=davem@davemloft.net \
    --cc=hch@lst.de \
    --cc=kbusch@kernel.org \
    --cc=linux-nvme@lists.infradead.org \
    --cc=malin1024@gmail.com \
    --cc=netdev@vger.kernel.org \
    --cc=ogerlitz@nvidia.com \
    --cc=sagi@grimberg.me \
    --cc=smalin@nvidia.com \
    --cc=yorayz@nvidia.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).