[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <a1cd2239-d8b7-412e-bca6-ebcf3a19ed65@os.amperecomputing.com>
Date: Fri, 1 Dec 2023 18:47:30 +0700
From: Quan Nguyen <quan@...amperecomputing.com>
To: Jeremy Kerr <jk@...econstruct.com.au>,
Matt Johnston <matt@...econstruct.com.au>,
"David S . Miller" <davem@...emloft.net>,
Eric Dumazet <edumazet@...gle.com>,
Jakub Kicinski <kuba@...nel.org>,
Paolo Abeni <pabeni@...hat.com>, netdev@...r.kernel.org,
linux-kernel@...r.kernel.org, openbmc@...ts.ozlabs.org,
Open Source Submission <patches@...erecomputing.com>
Cc: Phong Vo <phong@...amperecomputing.com>,
Thang Nguyen <thang@...amperecomputing.com>,
Dung Cao <dung@...amperecomputing.com>
Subject: Re: [PATCH] mctp i2c: Requeue the packet when arbitration is lost
On 01/12/2023 15:38, Jeremy Kerr wrote:
> Hi Quan,
>
>> As per [1], __i2c_transfer() will retry for adap->retries times
>> consecutively (without any wait) within the amount of time specified
>> by adap->timeout.
>>
>> So as per my observation, once it loses the arbitration, the next
>> retry is most likely still lost as another controller who win the
>> arbitration may still be using the bus.
>
> In general (and specifically with your hardware setup), the controller
> should be waiting for a bus-idle state before attempting the
> retransmission. You may well hit another arbitration loss after that,
> but it won't be from the same bus activity.
>
Yes, that is the correct behaviour and I think that is why
__i2c_transfer() return -EAGAIN when arbitration loss.
One example I could find in kernel tree is here [a], and, it seems not
to totally rely on the retry mechanism implemented in i2c core.
[a]
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/video/fbdev/omap2/omapfb/displays/connector-dvi.c#n157
Anyway, it's been great to discuss on this topic. To answer your
questions I did have to dig around and have learn a lot. :-)
My sincere thanks to you,
- Quan
Powered by blists - more mailing lists