Date:   Thu, 20 Apr 2023 06:39:09 +0000
From:   <Parthiban.Veerasooran@...rochip.com>
To:     <andrew@...n.ch>
CC:     <ramon.nordin.rodriguez@...roamp.se>, <hkallweit1@...il.com>,
        <linux@...linux.org.uk>, <davem@...emloft.net>,
        <edumazet@...gle.com>, <kuba@...nel.org>, <pabeni@...hat.com>,
        <netdev@...r.kernel.org>, <Jan.Huber@...rochip.com>,
        <Horatiu.Vultur@...rochip.com>, <Woojung.Huh@...rochip.com>
Subject: Re: [PATCH] drivers/net/phy: add driver for Microchip LAN867x
 10BASE-T1S PHY

On 20/04/23 11:03 am, Parthiban Veerasooran wrote:
> On 19/04/23 8:40 pm, Andrew Lunn wrote:
>> EXTERNAL EMAIL: Do not click links or open attachments unless you know 
>> the content is safe
>>
>> On Wed, Apr 19, 2023 at 02:40:29PM +0000, 
>> Parthiban.Veerasooran@...rochip.com wrote:
>>> Hi Ramon,
>>>
>>> Good day...! This is Parthiban from Microchip.
>>>
>>> Thanks for your patches for the Microchip LAN867x 10BASE-T1S PHY. We
>>> really appreciate your effort on this.
>>>
>>> For your kind information, we are already working for the driver which
>>> supports all the 10BASE-T1S PHYs from Microchip and doing internal
>>> review of those driver patches to mainline. These patches are going to
>>> reach mainline in couple of days. It is very unfortunate that we two are
>>> working on the same task at the same time without knowing each other.
>>>
>>> The architecture of your patch is similar to our current implementation.
>>> However to be able to support also the upcoming 10BASE-T1S products
>>> e.g., the LAN865x 10BASE-T1S MAC-PHY, additional functionalities have to
>>> be implemented. In order to avoid unnecessary/redundant work on both
>>> sides, we would like to collaborate with you on this topic and have a
>>> sync outside of this mailing list before going forward.
>>
>> Hi Parthiban
>>
>> Please review version 2 of the patch which was posted today.  Is there
>> anything in that patch which is actually wrong?
>>
>> I don't like the idea of dropping a patch, because a vendor comes out
>> of, maybe unintentional, stealth mode, and asks for their version to
>> be used, not somebody else's. For me this is especially important for
>> a new contributor.
>>
>> My preferred way forward is to merge Ramon's code, and then you can
>> build on it with additional features to support other family
>> members.
>>
>> Please don't get me wrong, i find it great you are supporting your own
>> devices. Not many vendors do. But Linux is a community, we have to
>> respect each others work, other members of the community.
>>
>>          Andrew
>>
>> FYI: Do you have any other drivers in the pipeline you want to
>> announce, just to avoid this happening again.
> 
> Hi Andrew,
> 
> Thanks a lot for your reply and clarification.
> 
> Sure I will also participate in reviewing the patches including v2.
> 
> I fully agree with you and also I really appreciate Ramon's effort on 
> this which shows how much interest he has on our product and driver 
> development.
> 
>      1. Add support for LAN8650/1 10BASE-T1S MAC-PHY's PHY in the 
> microchip_t1s.c driver which is being mainlined by Ramon.
>      2. Add generic driver support for the OPEN Alliance 10BASE-T1x 
> MAC-PHY Serial Interface.
>      3. Add driver support for LAN8650/1 10BASE-T1S MAC-PHY.
> 
> Note: 2nd and 3rd will be in a single patch series.
> 
> Above product link: https://www.microchip.com/en-us/product/lan8650
> 
> As I communicated before in the below email, we have the above drivers 
> in the pipeline to mainline.
Hi Andrew,

Sorry, forgot to add one more task in the above pipeline. Please find 
the updated list below,

	1. Add support for Microchip LAN8650/1 Rev.B0 10BASE-T1S MAC-PHY's PHY 
in the microchip_t1s.c driver which is being mainlined by Ramon (which 
will be used by the below LAN8650/1 Rev.B0 10BASE-T1S MAC-PHY driver).
	2. Add support for Microchip LAN8670/1/2 Rev.C1 10BASE-T1S PHY in the 
microchip_t1s.c driver which is being mainlined by Ramon.
	3. Add generic driver support for the OPEN Alliance 10BASE-T1x MAC-PHY 
Serial Interface (which will be used by the below LAN8650/1 10BASE-T1S 
MAC-PHY driver).
	4. Add driver support for LAN8650/1 Rev.B0 10BASE-T1S MAC-PHY.

We are already working on the above drivers and going for internal 
review to get the drivers mainlined in the upstream.

Best Regards,
Parthiban V
> 
> Hi Andrew,
> 
> Thanks a lot for your support. I will check with our colleagues
> suggested by you and get back to mainline again.
> 
> Best Regards,
> Parthiban V
> On 11/03/23 11:15 pm, Andrew Lunn wrote:
>  > EXTERNAL EMAIL: Do not click links or open attachments unless you 
> know the content is safe
>  >
>  > Hi Allan
>  >
>  > It has been a long time since we talked, maybe 2019 at the Linux
>  > Plumbers conference.... And then PTP discussions etc.
>  >
>  > It seems like Sparx5 is going well, along with felix, seville, etc.
>  >
>  > On Fri, Mar 10, 2023 at 11:13:23AM +0000, 
> Parthiban.Veerasooran@...rochip.com wrote:
>  >> Hi All,
>  >>
>  >> I would like to add Microchip's LAN865x 10BASE-T1S MAC-PHY driver
>  >> support to Linux kernel.
>  >> (Product link: https://www.microchip.com/en-us/product/LAN8650)
>  >>
>  >> The LAN8650 combines a Media Access Controller (MAC) and an Ethernet 
> PHY
>  >> to access 10BASE‑T1S networks. The common standard Serial Peripheral
>  >> Interface (SPI) is used so that the transfer of Ethernet packets and
>  >> LAN8650 control/status commands are performed over a single, serial
>  >> interface.
>  >>
>  >> Ethernet packets are segmented and transferred over the serial 
> interface
>  >> according to the OPEN Alliance 10BASE‑T1x MAC‑PHY Serial Interface
>  >> specification designed by TC6.
>  >> (link: https://www.opensig.org/Automotive-Ethernet-Specifications/)
>  >> The serial interface protocol can simultaneously transfer both transmit
>  >> and receive packets between the host and the LAN8650.
>  >>
>  >> Basically the driver comprises of two parts. One part is to interface
>  >> with networking subsystem and SPI subsystem. The other part is a TC6
>  >> state machine which implements the Ethernet packets segmentation
>  >> according to OPEN Alliance 10BASE‑T1x MAC‑PHY Serial Interface
>  >> specification.
>  >>
>  >> The idea behind the TC6 state machine implementation is to make it as a
>  >> generic library and platform independent. A set of API's provided by
>  >> this TC6 state machine library can be used by the 10BASE-T1x MAC-PHY
>  >> drivers to segment the Ethernet packets according to the OPEN Alliance
>  >> 10BASE‑T1x MAC‑PHY Serial Interface specification.
>  >>
>  >> With the above information, kindly provide your valuable feedback on my
>  >> below queries.
>  >>
>  >> Can we keep this TC6 state machine within the LAN865x driver or as a
>  >> separate generic library accessible for other 10BASE-T1x MAC-PHY 
> drivers
>  >> as well?
>  >>
>  >> If you recommend to have that as a separate generic library then could
>  >> you please advice on what is the best way to do that in kernel?
>  >
>  > Microchip is getting more and more involved in mainline. Jakub
>  > publishes some developers statistics for netdev:
>  >
>  > https://lwn.net/Articles/918007/
>  >
>  > It shows Microchip are near the top for code contributions. Which is
>  > great. However, as a reviewer, i see the quality really varies. Given
>  > how active Microchip is within Linux, the netdev community, and to
>  > some extent Linux as a whole, expects a company like Microchip to
>  > build up its internal resources to offer training and Mentoring to
>  > mainline developers, rather than expect the community to do that
>  > work. Does such a thing exist within Microchip? Could you point
>  > Parthiban towards a mentor who can help guide the work adding generic
>  > support for the OPEN Alliance 10BASE-T1x MAC-PHY Serial Interface and
>  > the LAN8650/1 specific bits? If not, could Steen Hegelund or Horatiu
>  > Vultur make some time available to be a mentor?
>  >
>  > Thanks
>  >          Andrew
> 

Powered by blists - more mailing lists