linux-kernel - Re: [PATCH 00/18] i2c: remove printout on handled timeouts

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

Date: Wed, 24 Apr 2024 11:00:53 +0200
From: Wolfram Sang <wsa+renesas@...g-engineering.com>
To: Andy Shevchenko <andy.shevchenko@...il.com>
Cc: linux-i2c@...r.kernel.org, linux-arm-kernel@...ts.infradead.org, 
	linux-arm-msm@...r.kernel.org, linux-kernel@...r.kernel.org, linux-omap@...r.kernel.org, 
	linux-renesas-soc@...r.kernel.org, linux-rockchip@...ts.infradead.org, 
	linux-rpi-kernel@...ts.infradead.org, linux-tegra@...r.kernel.org
Subject: Re: [PATCH 00/18] i2c: remove printout on handled timeouts

Hi Andy,

On Wed, Apr 24, 2024 at 03:08:14AM +0300, Andy Shevchenko wrote:
> Wed, Apr 10, 2024 at 01:24:14PM +0200, Wolfram Sang kirjoitti:
> > While working on another cleanup series, I stumbled over the fact that
> > some drivers print an error on I2C or SMBus related timeouts. This is
> > wrong because it may be an expected state. The client driver on top
> > knows this, so let's keep error handling on this level and remove the
> > prinouts from controller drivers.
> > 
> > Looking forward to comments,
> 
> I do not see an equivalent change in I²C core.

There shouldn't be. The core neither knows if it is okay or not. The
client driver knows.

> IIRC in our case (DW or i801 or iSMT) we often have this message as the only

Often? How often?

> one that points to the issues (on non-debug level), it will be much harder to
> debug for our customers with this going away.

The proper fix is to print the error in the client driver. Maybe this
needs a helper for client drivers which they can use like:

	i2c_report_error(client, retval, flags);

The other thing which is also more helpful IMO is that we have
trace_events for __i2c_transfer. There, you can see what was happening
on the bus before the timeout. It can easily be that, if device X
times out, it was because of the transfer before to device Y which locks
up the bus. A simple "Bus timed out" message will not help you a lot
there.

And, keep in mind the false positives I mentioned in the coverletter.

All the best,

   Wolfram

Download attachment "signature.asc" of type "application/pgp-signature" (834 bytes)