linux-kernel - Re: [PATCH 00/33] fbcon notifier begone!

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: <20190527071729.GM21222@phenom.ffwll.local>
Date:   Mon, 27 May 2019 09:17:29 +0200
From:   Daniel Vetter <daniel@...ll.ch>
To:     Sam Ravnborg <sam@...nborg.org>
Cc:     Daniel Vetter <daniel.vetter@...ll.ch>,
        LKML <linux-kernel@...r.kernel.org>,
        Intel Graphics Development <intel-gfx@...ts.freedesktop.org>,
        DRI Development <dri-devel@...ts.freedesktop.org>
Subject: Re: [PATCH 00/33] fbcon notifier begone!

On Sat, May 25, 2019 at 07:19:28PM +0200, Sam Ravnborg wrote:
> Hi Daniel.
> 
> Good work, nice cleanup all over.
> 
> A few comments to a few patches - not something that warrant a
> new series to be posted as long as it is fixed before the patches are
> applied.

Hm yeah good idea, I'll add that to the next version.

> > btw for future plans: I think this is tricky enough (it's old code and all
> > that) that we should let this soak for 2-3 kernel releases. I think the
> > following would be nice subsequent cleanup/fixes:
> > 
> > - push the console lock completely from fbmem.c to fbcon.c. I think we're
> >   mostly there with prep, but needs to pondering of corner cases.
> I wonder - should this code consistently use __acquire() etc so we could
> get a little static analysis help for the locking?
> 
> I have not played with this for several years and I do not know the
> maturity of this today.

Ime these sparse annotations are pretty useless because too inflexible. I
tend to use runtime locking checks based on lockdep. Those are more
accurate, and work even when the place the lock is taken is very far away
from where you're checking, without having to annoate all functions in
between. We need that for something like console_lock which is a very big
lock. Only downside is that only paths you hit at runtime are checked.

> > - move fbcon.c from using indices for tracking fb_info (and accessing
> >   registered_fbs without proper locking all the time) to real fb_info
> >   pointers with the right amount of refcounting. Mostly motivated by the
> >   fun I had trying to simplify fbcon_exit().
> > 
> > - make sure that fbcon call lock/unlock_fb when it calls fbmem.c
> >   functions, and sprinkle assert_lockdep_held around in fbmem.c. This
> >   needs the console_lock cleanups first.
> > 
> > But I think that's material for maybe next year or so.
> Or maybe after next kernel release.
> Could we put this nice plan into todo.rst or similar so we do not have
> to hunt it down by asking google?
> 
> For the whole series you can add my:
> Reviewed-by: Sam Ravnborg <sam@...nborg.org>
> 
> Some parts are reviewed as "this looks entirely correct", other parts
> I would claim that I actually understood.
> And after having spend some hours on this a r-b seems in order.

Thanks, Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch