linux-kernel - RE: Fw: Arm CryptoCell driver -- default Y, even on machines where it is obviously useless

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Sun, 26 Apr 2020 12:49:57 +0000
From:   Hadar Gat <Hadar.Gat@....com>
To:     Ard Biesheuvel <ardb@...nel.org>,
        Gilad Ben-Yossef <gilad@...yossef.com>
CC:     Herbert Xu <herbert@...dor.apana.org.au>,
        kernel list <linux-kernel@...r.kernel.org>,
        Linux Crypto Mailing List <linux-crypto@...r.kernel.org>,
        nd <nd@....com>
Subject: RE: Fw: Arm CryptoCell driver -- default Y, even on machines where it
 is obviously useless

Hi Ard,

> -----Original Message-----
> From: Ard Biesheuvel <ardb@...nel.org>
> Sent: Sunday, 19 April 2020 11:55
> 
> > > > -----Original Message-----
> > > > From: Pavel Machek <pavel@....cz>
> > > > Sent: Saturday, 18 April 2020 13:44
> > > >
> > > > Hi!
> > > >
> > > > I'm configuring kernel for x86, and I get offered
> HW_RANDOM_CCTRNG
> > > > with default=Y, and help text suggesting I should enable it.
> > > >
> > > > That's... two wrong suggestions, right?
> > > >
> > > > Best regards,
> > > > Pavel
> > ...
> > > ________________________________________
> > > From: Hadar Gat <Hadar.Gat@....com>
> > > Sent: Saturday, April 18, 2020 11:31 PM
> > >
> > > Hi Pavel,
> > > I think you got it right..
> > > Indeed, Arm CryptoCell CCTRNG driver couldn't be used and obviously
> useless if the Arm CryptoCell HW does not exist in the system.
> >
> > There's a delicate point here though - CryptoCell is an independent
> > hardware block, it is not tied to a particular CPU architecture.
> > There are SoCs with none-Arm architecture CPU using it.
> >
> > So I would say whatever the answer is, it should be the same for any
> > generic embedded style HW block.
> >
> > And the help text is not architecture specific anyway, is it not..?
> >
> 
> Both the default y and and the help text are indeed incorrect. This should be
> fixed. We don't enable device drivers by default, and definitely not as as
> builtins. A conditional default m could be acceptable if the condition is
> sufficiently narrow.

On one hand I totally agree with that and think the default should be N.
On the other hand, most of the HW_RANDOM drivers set the default to HW_RANDOM
and it doesn't make sense to me to do something different than almost every other HW RANDOM device.
Do I miss something here?

> 
> While at it, could we add a depends on CONFIG_OF since this code is
> definitely unusable on non-DT systems.