Message-ID: <20040908041307.34927.qmail@web11908.mail.yahoo.com>
Date: Tue, 7 Sep 2004 21:13:07 -0700 (PDT)
From: Mike Mestnik <cheako911@yahoo.com>
Subject: Re: [BUG] r200 dri driver deadlocks
To: Dave Airlie <airlied@gmail.com>, arjanv@redhat.com,
       Roland Scheidegger <rscheidegger_lists@hispeed.ch>
Cc: =?ISO-8859-1?Q?=20=22Felix=5FK=FChling=22?= <fxkuehl@gmx.de>,
       Lee Revell <rlrevell@joe-job.com>, diablod3@gmail.com,
       dri-devel@lists.sourceforge.net, linux-kernel@vger.kernel.org
In-Reply-To: <21d7e997040907013071ebb60d@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 3200
Lines: 85

Sorry, I don't know why we are cross posting and including subscribers in
CC.  This belongs on the DRI list, as it is only with 3rd party DRI-client
code that the problem exists.

--- Dave Airlie <airlied@gmail.com> wrote:

> On Tue, 07 Sep 2004 09:07:11 +0200, Arjan van de Ven <arjanv@redhat.com>
> wrote:
> > On Tue, 2004-09-07 at 08:54, Dave Airlie wrote:
> > 
> > > Feel free to implement it and profile it, but there are so many ways
> > > to lock up a radeon chip it is scary, the above was just one
> example,
> > > some days if you look at it funny it can lockup :-), it is accepted
> > > that userland can crap out 3D chips, the Intel ones are fairly easy
> to
> > > hangup also..
> > 
> > 
> > hmmm.. I thought the entire reason for having part of DRM in the
> kernel
> > was to be able to prevent such events from happening....
> 
> only one reason...
> http://dri.sourceforge.net/doc/drm_low_level.html
> 
> But to be honest the chips are entirely capable of locking up on what
> the docs say are valid things, writing enough workarounds and test
> would bloat the drm considerably,
> at the moment we try and have it so a valid OpenGL application doesn't
> lock it up, but someone writing directly to the DRM would be able to
> lockup a fair few chips in many interesting ways....
> 
> Dave.
> 

--- Roland Scheidegger <rscheidegger_lists@hispeed.ch> wrote:
> 
> I seriously doubt this is doable. Unless you put the whole driver in the
> 
> kernel, which of course nobody wants. I frequently caused gpu lockups by
> 
> experimental driver changes (for instance, wrong vertex setup). I think 
> the consensus was that it's ok for the driver to lock up the gpu, but it
> 
> should not lock up the kernel.
> It might be possible to prevent lockups by a watchdog, resetting the gpu
> 
> if a lockup is detected. This is how ATI deals with lockups in windows 
> (dubbed "VPU Recover"), and there is a patch floating around for DRI too
> 
> (though it is not exactly for that, and doesn't always work).
> 
> Roland
> 

It's a simple matter of enforcing 3rd party(this means every DRM user)
clients to use DRI's *dialect or style*.  If the DRM see activities that
are not expected to be generated by pure DRI-clients, action should be
taken to prevent a posible lockup.  This means that even valid activities
should be treated as invalid IF the DRM can clerly detect a deviation from
pure DRI-client activities.

For example, pure DRI-clients emit state changing commands is a vary
specific order.  The DRM could easily spot if these cmds where out of any
knowen/used order or if any other cmds where also inserted into the
expected order.  This should be denied"."  Only DRI-clients(any client)
using the DRI supplied order(the one used by pure DRI-clients) should be
allowed to access the hardware.


__________________________________
Do you Yahoo!?
New and Improved Yahoo! Mail - Send 10MB messages!
http://promotions.yahoo.com/new_mail 
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/