linux-kernel - Re: [PATCH 1/8] soundwire: intel: filter SoundWire controller device search

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]

Message-ID: <20190508050853.GT16052@vkoul-mobl>
Date:   Wed, 8 May 2019 10:38:53 +0530
From:   Vinod Koul <vkoul@...nel.org>
To:     Pierre-Louis Bossart <pierre-louis.bossart@...ux.intel.com>
Cc:     alsa-devel@...a-project.org, linux-kernel@...r.kernel.org,
        tiwai@...e.de, broonie@...nel.org, gregkh@...uxfoundation.org,
        liam.r.girdwood@...ux.intel.com, jank@...ence.com, joe@...ches.com,
        srinivas.kandagatla@...aro.org,
        Sanyog Kale <sanyog.r.kale@...el.com>
Subject: Re: [PATCH 1/8] soundwire: intel: filter SoundWire controller device
 search

On 07-05-19, 09:43, Pierre-Louis Bossart wrote:
> 
> 
> On 5/7/19 7:26 AM, Vinod Koul wrote:
> > On 03-05-19, 19:29, Pierre-Louis Bossart wrote:
> > > The convention is that the SoundWire controller device is a child of
> > > the HDAudio controller. However there can be more than one child
> > > exposed in the DSDT table, and the current namespace walk returns the
> > > last device.
> > > 
> > > Add a filter and terminate early when a valid _ADR is provided,
> > > otherwise keep iterating to find the next child.
> > 
> > So what are the other devices in DSDT here..
> 
> this is what I see:
> 
> Scope (HDAS)
>         {
>             Device (IDA)
>             {
>                 Name (_ADR, 0x00020001)  // _ADR: Address
>             }
>         }
> 
> I thought this was nonsense but your question triggered me to look into the
> Intel SST ACPI specs (not public I am afraid but shared with the OS who
> shall not be named).
> Using the same source of information as below, I *believe* this is HDaudio
> related, bits 31..16 mean HDaudio with codec SDI 2, and NodeId 1 for the
> function group. This would make sense as I believe there are two codecs on
> the board that can be pin-strapped to boot either in HDaudio or SoundWire
> mode- but this is a conjecture only.
> 
> At any rate, we need a hardware rework and mutual exclusion between HDaudio
> and SoundWire, so we have to ignore this one when SoundWire is enabled.

That is how I was expecting it to be...

> > > +	/*
> > > +	 * On some Intel platforms, multiple children of the HDAS
> > > +	 * device can be found, but only one of them is the SoundWire
> > > +	 * controller. The SNDW device is always exposed with
> > > +	 * Name(_ADR, 0x40000000) so filter accordingly
> > > +	 */
> > > +	if (adr != 0x40000000)
> > 
> > I do not recall if 4 corresponds to the links you have or soundwire
> > device type, is this number documented somewhere is HDA specs?
> 
> I thought it was a magic number, but I did check and for once it's
> documented and the values match the spec :-)
> I see in the ACPI docs bits 31..28 set to 4 indicate a SoundWire Link Type
> and bits 3..0 indicate the SoundWire controller instance, the rest is
> reserved to zero.

So in that case we should mask with bits 31..28 and match, who knows you
may have multiple controller instances in future
I had a vague recollection that this was documented in the spec, glad
that in turned out to be the case.

Btw was the update to HDA spec made public?

> > Also it might good to create a define for this
> 
> I will respin this one to add the documentation above, and only filter on
> the 4 ms-bits. Thanks for forcing me to RTFM :-)

-- 
~Vinod