linux-kernel - Re: [PATCH] ASoC: mediatek: Allow setting readable driver names through kconfig

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]

Date: Wed, 26 Jun 2024 08:44:01 -0400
From: Nícolas F. R. A. Prado <nfraprado@...labora.com>
To: AngeloGioacchino Del Regno <angelogioacchino.delregno@...labora.com>
Cc: Liam Girdwood <lgirdwood@...il.com>, Mark Brown <broonie@...nel.org>,
	Jaroslav Kysela <perex@...ex.cz>, Takashi Iwai <tiwai@...e.com>,
	Matthias Brugger <matthias.bgg@...il.com>, kernel@...labora.com,
	linux-sound@...r.kernel.org, linux-kernel@...r.kernel.org,
	linux-arm-kernel@...ts.infradead.org,
	linux-mediatek@...ts.infradead.org
Subject: Re: [PATCH] ASoC: mediatek: Allow setting readable driver names
 through kconfig

On Wed, Jun 26, 2024 at 09:49:55AM +0200, AngeloGioacchino Del Regno wrote:
> Il 25/06/24 22:23, Nícolas F. R. A. Prado ha scritto:
> > Commit c8d18e440225 ("ASoC: core: clarify the driver name
> > initialization") introduced an error message when driver names are
> > automatically generated and result in truncated names.
> > 
> > For example, this error message is printed on mt8195-cherry-tomato-rev2:
> > 
> >    mt8195_mt6359 mt8195-sound: ASoC: driver name too long 'sof-mt8195_r1019_5682' -> 'sof-mt8195_r101'
> > 
> > Since that truncated driver name is already used by userspace (eg UCM),
> > it can't be unconditionally updated.
> > 
> > As suggested by Jaroslav, update the driver name but hide it behind a
> > kernel config option, which Linux distributions can enable once the
> > corresponding support in userspace audio configuration software lands.
> > This ensures that audio doesn't regress, while still allowing the error
> > to be fixed.
> 
> I can propose the following plan of action for such a rename:
>  1. Temporarily have both UCMs
>  2. Wait until distros update their packages

It's not possible to monitor the package version for every single distro, so I
guess you mean tracking only the big distros and hoping that no small distro
still uses the old version.

Also, is UCM the only userspace tool that might use this name? There might be
others too that we would need to care about.

>  3. Change the name in kernel without extra config options
>  4. Remove the now deprecated UCMs
> 
> ...that's requiring a bit of time, yes, but I think this is the best way to
> ensure that no breakage happens in the meantime for any users.

My worry is that we don't know how long it would take, if a small distro decides
to keep using the old version. But even in the best case it might take years for
the old version to be phased out, I'm not sure I'll be around for that, so maybe
we'll just keep using the old name forever.

The nice thing about the config is that distros and users can opt-in immediately
to get rid of the error. For example KernelCI, which is monitoring such errors,
can enable this config to fix this error right now.

Thanks,
Nícolas