linux-kernel - Re: [PATCH v2] i2c-mlxbf: Add repeated start condition support

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID:
 <BN9PR12MB5196216AD3B780419AAB2B03C61F2@BN9PR12MB5196.namprd12.prod.outlook.com>
Date: Mon, 13 Jan 2025 16:39:04 +0000
From: Chris Babroski <cbabroski@...dia.com>
To: Asmaa Mnebhi <asmaa@...dia.com>, Khalil Blaiech <kblaiech@...dia.com>,
	Andi Shyti <andi.shyti@...nel.org>
CC: David Thompson <davthompson@...dia.com>, "linux-i2c@...r.kernel.org"
	<linux-i2c@...r.kernel.org>, "linux-kernel@...r.kernel.org"
	<linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v2] i2c-mlxbf: Add repeated start condition support

Hi Andi,

Thank you for reviewing. I will send an updated V3 patch with the changes you have requested soon.

> Can you please explain better how the stop bit affects the timing
> values?

There were two changes needed to get I2C communication working with a specific I2C device on BlueField 3 and are contained within this patch:
1. Support repeated start conditions for SMBus I2C block read. This is implemented by removing the stop bit for the first write of a block read transaction.
2. Increase the time that the clock can be held low before an I2C transaction is aborted (the "timeout" value in the bus timing configuration). Without this change the bus will time out attempting a block read with this device.

While debugging to find #2 above we noticed that the "timeout" value, along with several other timing values, did not match our ATF I2C driver which contains the latest bus timing configuration and has been fully verified by SW and HW teams. When I updated the "timeout" value I also updated any other timing values that were incorrect to ensure our Linux kernel driver uses the complete tested bus timing configuration.

Testing this bug fix requires both the repeated start condition and I2C bus timing changes which is why I have included them in a single patch. If you would prefer for the timing changes to be separated into a different patch then I can do that instead.

Thanks,
Chris