mbox series

[net-next,0/2] qed*: SmartAN query support

Message ID 20190207142012.4521-1-skalluru@marvell.com
Headers show
Series qed*: SmartAN query support | expand

Message

Sudarsana Reddy Kalluru Feb. 7, 2019, 2:20 p.m. UTC 
SmartAN feature detects the peer/cable capabilities and establishes the
link in the best possible configuration.
The patch series adds support for querying the capability. Please consider
applying it net-next.

Sudarsana Reddy Kalluru (2):
  qed: Add API for SmartAN query.
  qede: Add ethtool interface for SmartAN query.

 drivers/net/ethernet/qlogic/qed/qed_hsi.h       |  1 +
 drivers/net/ethernet/qlogic/qed/qed_main.c      |  2 ++
 drivers/net/ethernet/qlogic/qed/qed_mcp.c       |  6 ++++++
 drivers/net/ethernet/qlogic/qed/qed_mcp.h       | 10 ++++++++++
 drivers/net/ethernet/qlogic/qede/qede_ethtool.c | 11 ++++++++++-
 include/linux/qed/qed_if.h                      |  1 +
 6 files changed, 30 insertions(+), 1 deletion(-)

Comments

Jakub Kicinski Feb. 7, 2019, 7:54 p.m. UTC | #1 
On Thu, 7 Feb 2019 06:20:10 -0800, Sudarsana Reddy Kalluru wrote:
> SmartAN feature detects the peer/cable capabilities and establishes the
> link in the best possible configuration.

It sounds familiar, I need to check with FW team, but I think we may be
doing a similar thing, and adding a common API rather than ethtool flag
would be preferable.

Could you please share a little bit more detail?  What are the
configurations this would choose between?
David Miller Feb. 7, 2019, 9:22 p.m. UTC | #2 
From: Sudarsana Reddy Kalluru <skalluru@marvell.com>
Date: Thu, 7 Feb 2019 06:20:10 -0800

> SmartAN feature detects the peer/cable capabilities and establishes the
> link in the best possible configuration.
> The patch series adds support for querying the capability. Please consider
> applying it net-next.

Series applied, thanks.
David Miller Feb. 7, 2019, 9:32 p.m. UTC | #3 
From: David Miller <davem@davemloft.net>
Date: Thu, 07 Feb 2019 13:22:35 -0800 (PST)

> From: Sudarsana Reddy Kalluru <skalluru@marvell.com>
> Date: Thu, 7 Feb 2019 06:20:10 -0800
> 
>> SmartAN feature detects the peer/cable capabilities and establishes the
>> link in the best possible configuration.
>> The patch series adds support for querying the capability. Please consider
>> applying it net-next.
> 
> Series applied, thanks.

Nevermind, I reverted before pushing out as I see Jakub has some questions.
Sudarsana Reddy Kalluru Feb. 8, 2019, 11:32 a.m. UTC | #4 
>-----Original Message-----
>From: Jakub Kicinski [mailto:jakub.kicinski@netronome.com]
>Sent: 08 February 2019 01:24
>To: Sudarsana Reddy Kalluru <skalluru@marvell.com>
>Cc: davem@davemloft.net; netdev@vger.kernel.org; Ariel Elior
><aelior@marvell.com>; Michal Kalderon <mkalderon@marvell.com>
>Subject: [EXT] Re: [PATCH net-next 0/2] qed*: SmartAN query support
>
>External Email
>
>----------------------------------------------------------------------
>On Thu, 7 Feb 2019 06:20:10 -0800, Sudarsana Reddy Kalluru wrote:
>> SmartAN feature detects the peer/cable capabilities and establishes
>> the link in the best possible configuration.
>
>It sounds familiar, I need to check with FW team, but I think we may be doing
>a similar thing, and adding a common API rather than ethtool flag would be
>preferable.
>
>Could you please share a little bit more detail?  What are the configurations
>this would choose between?

Jakub,
  Following doc provides detailed information on this feature. We simply need a flag to display whether the feature is enabled in the hardware or not, hence adding it to "ethtool --show-priv-flags".
https://www.cavium.com/Dell/Documents/Converged/TB_Establishing_Adaptive_Links_with_SmartAN_Dell.pdf
****
When an administrator first plugs a device (discrete optical module or Active Optical Cable assembly or DAC) into the Cavium FastLinQ adapter SFP/QSFP interface, the Cavium SmartAN technology reads the device type (discrete optics or AOC or DAC), what its speed rating is, special optics data (such as what FEC mode is required or if it is multi-speed capable), DAC CA-x type, and DAC length.
Armed with this information, the FastLinQ adapter attempts each possible mode (supported by that device) until it secures a link with the connected link partner (switch), without input from the end user.
****

Thanks,
Sudarsana
Jakub Kicinski Feb. 8, 2019, 4:04 p.m. UTC | #5 
On Fri, 8 Feb 2019 11:32:14 +0000, Sudarsana Reddy Kalluru wrote:
> >On Thu, 7 Feb 2019 06:20:10 -0800, Sudarsana Reddy Kalluru wrote:  
> >> SmartAN feature detects the peer/cable capabilities and establishes
> >> the link in the best possible configuration.  
> >
> >It sounds familiar, I need to check with FW team, but I think we may be doing
> >a similar thing, and adding a common API rather than ethtool flag would be
> >preferable.
> >
> >Could you please share a little bit more detail?  What are the configurations
> >this would choose between?  
> 
> Jakub,
>   Following doc provides detailed information on this feature. We simply need a flag to display whether the feature is enabled in the hardware or not, hence adding it to "ethtool --show-priv-flags".
> https://www.cavium.com/Dell/Documents/Converged/TB_Establishing_Adaptive_Links_with_SmartAN_Dell.pdf

Thanks!  Yes, that's exactly the same thing.  In a nutshell trying
different serdes and FEC speeds.  I guess I'll just put adding an API
for this down as a TODO, and add it once we have ethtool netlink.  And
keep IOCTL ethtool frozen until then.

So I think the priv flag is fine for now.
David Miller Feb. 9, 2019, 6:55 a.m. UTC | #6 
From: Jakub Kicinski <jakub.kicinski@netronome.com>
Date: Fri, 8 Feb 2019 08:04:29 -0800

> Thanks!  Yes, that's exactly the same thing.  In a nutshell trying
> different serdes and FEC speeds.  I guess I'll just put adding an API
> for this down as a TODO, and add it once we have ethtool netlink.  And
> keep IOCTL ethtool frozen until then.
> 
> So I think the priv flag is fine for now.

Series applied to net-next.