linux-kernel - Re: [PATCH] soundwire: bus: Fix wrong port number in sdw_handle_slave

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: <633a7b75-91f8-9a54-c042-87079d27cc8c@opensource.cirrus.com>
Date:   Tue, 20 Sep 2022 10:17:28 +0100
From:   Richard Fitzgerald <rf@...nsource.cirrus.com>
To:     Pierre-Louis Bossart <pierre-louis.bossart@...ux.intel.com>,
        <vkoul@...nel.org>, <yung-chuan.liao@...ux.intel.com>,
        <sanyog.r.kale@...el.com>
CC:     <patches@...nsource.cirrus.com>, <alsa-devel@...a-project.org>,
        <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] soundwire: bus: Fix wrong port number in
 sdw_handle_slave_alerts()

On 19/09/2022 09:44, Pierre-Louis Bossart wrote:
> 
> 
> On 9/17/22 16:02, Richard Fitzgerald wrote:
>> for_each_set_bit() gives the bit-number counting from 0 (LSbit==0).
>> When processing INTSTAT2, bit 0 is DP4 so the port number is (bit + 4).
>> Likewise for INTSTAT3 bit 0 is DP11 so port number is (bit + 11).
>>
>> Signed-off-by: Richard Fitzgerald <rf@...nsource.cirrus.com>
> 
> Another thing that never worked, but no one has used port interrupts so
> far. I only used it for PRBS tests in early enabling some 3 years ago,
> and it was for port 2 IIRC.
> 

Except us because Cirrus Soundwire requires a full DP prepare sequence,
it doesn't support simplified DP_SM.

There's still the problem of the code deadlocking while waiting for the
prepare interrupt, because it holds the bus lock while waiting for the
completion, so the alert handler can't run. That's something I need to
look at fixing. But it's not a blocker, just means that the code has to
wait the whole timeout for each amp. (That's another thing, couldn't we
issue all the DP prepares first so the amps prepare in parallel, then
wait for them to ack?)