linux-kernel - Re: [PATCH v2 5/6] tpm: add driver for cr50 on SPI

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]

Message-ID: <20190717183645.GM12119@ziepe.ca>
Date:   Wed, 17 Jul 2019 15:36:45 -0300
From:   Jason Gunthorpe <jgg@...pe.ca>
To:     Guenter Roeck <groeck@...gle.com>
Cc:     Stephen Boyd <swboyd@...omium.org>,
        Alexander Steffen <Alexander.Steffen@...ineon.com>,
        Peter Huewe <peterhuewe@....de>,
        Jarkko Sakkinen <jarkko.sakkinen@...ux.intel.com>,
        Andrey Pronin <apronin@...omium.org>,
        linux-kernel <linux-kernel@...r.kernel.org>,
        Arnd Bergmann <arnd@...db.de>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        linux-integrity@...r.kernel.org,
        Duncan Laurie <dlaurie@...omium.org>,
        Guenter Roeck <groeck@...omium.org>
Subject: Re: [PATCH v2 5/6] tpm: add driver for cr50 on SPI

On Wed, Jul 17, 2019 at 11:30:42AM -0700, Guenter Roeck wrote:
> On Wed, Jul 17, 2019 at 11:21 AM Stephen Boyd <swboyd@...omium.org> wrote:
> >
> > Quoting Jason Gunthorpe (2019-07-17 10:25:44)
> > > On Wed, Jul 17, 2019 at 10:22:20AM -0700, Stephen Boyd wrote:
> > > > Quoting Jason Gunthorpe (2019-07-17 10:12:16)
> > > > > On Wed, Jul 17, 2019 at 10:05:52AM -0700, Stephen Boyd wrote:
> > > > > >
> > > > > > Yes. The space savings comes from having the extra module 'cr50.ko' that
> > > > > > holds almost nothing at all when the two drivers are modules.
> > > > >
> > > > > I'm not sure it is an actual savings, there is alot of minimum
> > > > > overhead and alignment to have a module in the first place.
> > > > >
> > > >
> > > > Yeah. I'm pretty sure that's why it's a bool and not a tristate for this
> > > > symbol. A module has overhead that is not necessary for these little
> > > > helpers.
> > >
> > > Linking driver stuff like that to the kernel is pretty hacky, IMHO
> > >
> >
> > So combine lines?
> >
> >         obj-$(CONFIG_...) += cr50.o cr50_spi.o
> >
> > Sounds great.
> >
> 
> Please keep in mind that cr50.c exports symbols. If cr50.o is added to
> two modules, those symbols will subsequently available from both
> modules. To avoid that, you might want to consider removing the
> EXPORT_SYMBOL() declarations from cr50.c.

Yep

> I don't know what happens if those two modules are both built into the
> kernel (as happens, for example, with allyesconfig). Does the linker
> try to load cr50.o twice, resulting in duplicate symbols ?

Hum. Looks like it uses --whole-archive here and would probably
break? Maybe not, hns recently sent a patch doing this, but maybe they
never tested it too.

Jason