linux-kernel - Re: [alsa-devel] unload Audio drivers while playback stream is active case kernel crash

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: <20150114120228.GT4160@sirena.org.uk>
Date:	Wed, 14 Jan 2015 12:02:28 +0000
From:	Mark Brown <broonie@...nel.org>
To:	Takashi Iwai <tiwai@...e.de>
Cc:	Lars-Peter Clausen <lars@...afoo.de>,
	"alsa-devel@...a-project.org" <alsa-devel@...a-project.org>,
	"Wang, Jiada (ESD)" <Jiada_Wang@...tor.com>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	"lgirdwood@...il.com" <lgirdwood@...il.com>,
	"Frkuska, Joshua" <Joshua_Frkuska@...tor.com>
Subject: Re: [alsa-devel] unload Audio drivers while playback stream is
 active case kernel crash

On Wed, Jan 14, 2015 at 11:50:48AM +0100, Takashi Iwai wrote:
> Lars-Peter Clausen wrote:

> > > OK, so it's not about active stream.  From the reporter's description,
> > > I supposed that the module gets unloaded while playing a stream, which
> > > shouldn't be allowed.

> > Well one of the ways to trigger this is to remove the module while the 
> > stream is active. But it is not exclusively a problem module unload problem. 
> > E.g. the same happens if you hot-unplug the ASoC card.

> Yes, unbinding can trigger a similar problem, ends up the same bad
> code path.

Right, which was my point - we need to be able to cope with the driver
being removed while in use, unbind is just one path where this could
happen and the issue isn't specific to unbind.

> > I don't think that we need to prevent module unload when a stream is active. 
> >  From a framework point of view is not different from hot-unplug. I don't 
> > see a reason why we'd jump through hoops to actively forbid removing the 
> > module once it works just fine.

> Well, the module unload means a more drastic cleanup.  Even if you
> unbind, the code and data are still there while module unload may
> clean them up all.

> Above all, disallowing the module unload while using is the common
> behavior of any other drivers.  Why do we have to be a rebel against
> all civil manner? :)

That's not true for everything and for ASoC I'd tend to assume that the
user knows what they're doing and has a good reason for it; it's
certainly something that can be helpful in development.

Download attachment "signature.asc" of type "application/pgp-signature" (474 bytes)