Re: [PATCH v3 0/4] Fix multiple race condition vulnerabilities in dvb-core and device driver

[Mauro Carvalho Chehab <mchehab@kernel.org>]

Em Sat, 13 May 2023 19:09:01 +0100
Mauro Carvalho Chehab <mchehab@kernel.org> escreveu:

> Hi Lee,
> 
> Em Thu, 9 Mar 2023 17:17:14 +0000
> Lee Jones <lee@kernel.org> escreveu:
> 
> > > > Still nothing heard from the author or any maintainer.    
> > >
> > > We're currently lacking a sub-maintainer for dvb. Changes at the
> > > DVB mutexes have been problematic and require tests on some
> > > devices, specially on those with multiple frontends.
> > >
> > > I'll try to find some time to review and test those patches.    
> > 
> > Thank you Mauro, I fully appreciate the struggles and the effort.  
> 
> It took more time than I originally anticipated, as I had to setup
> a way to test it here with some DVB-T devices, but I reviewed the
> code and tested it.
> 
> I'm placing the patches I picked at this series:
> 
> 	https://lore.kernel.org/lkml/53558de2b5c4f4ee6bfcfbe34e27071c2d0073d5.1684000646.git.mchehab@kernel.org/T/#t
> 
> They seemed to work fine. I tested using two devices:
> 
> 	USB ID 2013:0246 PCTV Systems PCTV 74E
> 	USB ID 2040:5200 Hauppauge NovaT 500Stick
> 
> The second one has two DVB-T independent devices on it.
> 
> I did some tests of removing and re-inserting them with the
> devices closed and with the device opened and streamed. I didn't
> find any regressions. I didn't try to use kmemleak or KASAN to
> detect UAF conditions, though.

Did some tests today: neither KASAN nor kmemleak pointed any issue.

I found a few issues on one of the patches: some mutex unbalance
at dvb_frontend at the error path. Already updated the patches to
fix it. Should be merging at the media subsystem today and send
a PR upstream along the next week with the fixes.

Regards,
Mauro