linux-kernel - Re: [PATCH] drivers/cros

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:   Tue, 3 Jan 2023 18:19:48 -0700
From:   Rob Barnes <robbarnes@...gle.com>
To:     Prashant Malani <pmalani@...omium.org>
Cc:     Guenter Roeck <groeck@...omium.org>,
        chrome-platform@...ts.linux.dev, linux-kernel@...r.kernel.org,
        dtor@...omium.org, Benson Leung <bleung@...omium.org>
Subject: Re: [PATCH] drivers/cros_ec: Handle CrOS EC panics

On Tue, Jan 3, 2023 at 4:28 PM Prashant Malani <pmalani@...omium.org> wrote:
>
> On Tue, Jan 3, 2023 at 3:15 PM Rob Barnes <robbarnes@...gle.com> wrote:
> >
> > On Wed, Dec 21, 2022 at 5:38 PM Prashant Malani <pmalani@...omium.org> wrote:
> > >
> > > On Wed, Dec 21, 2022 at 3:55 PM Rob Barnes <robbarnes@...gle.com> wrote:
> > > >
> > > > On Wed, Dec 21, 2022 at 12:23 PM Prashant Malani <pmalani@...omium.org> wrote:
> > > > >
> >
> > >
> > > It's just I find having a notifier for a single use case overkill(even
> > > 2 would be fine); one could get away with exposing a method
> > > in cros_typec_debugfs via a local .h file (it can compile to a stub if
> > > cros_typec_debugfs is not compiled to the kernel);
> > > the LPC code can then just call that method instead of invoking a notifier.
> >
> > My first implementation did make a direct call to cros_ec_debugfs.c,
> > but an internal reviewer recommended using an event notifier instead.
> > So I'm histent to go back to a direct call.
> >
> > There may be other sub drivers that want to handle EC panics. So I
> > think keeping this as a separate notifier makes sense given the
> > constraints.
>
> The issue with that reasoning vis-à-vis your implementation is that
> the panic notifier is tied to cros_ec_debugfs. What if another
> (sub)-driver wants to use the
> panic notifier to do something, but that system doesn't have CONFIG_DEBUGFS
> enabled?

There isn't a dependency between cros_ec_debugfs.c and
`panic_notifier` inside `cros_ec_proto.h`. So (sub)-drivers can handle
EC panics when CONFIG_DEBUGFS is not enabled.

>
> Having a direct/explicit dependency avoids that issue; LPC depends on
> debugfs being compiled
> to have the log printed out (a stub is used when debugfs is not
> enabled), but nothing else
> relies on debugfs for a panic notifier.
>
> BR,
>
> -Prashant