[PATCH 2/2] DRI2: Add error message when working around driver bug

[Pauli Nieminen]

On 07/10/10 00:35 +0200, ext Mario Kleiner wrote:
> On Oct 6, 2010, at 1:05 PM, Pauli Nieminen wrote:
> 
> > There isn't API that allows application atomically query for msc  
> > changes
> > and schedule swaps. If msc changes dramatically between query and
> > scheduling application would schedule swap to happen at wrong time.
> >
> > Because of API limitations driver has to make msc increment for each
> > vblank affecting the drawable.
> >
> 
> Hi Paul,
> 
> this is not a driver bug which should be reported as an error, but  
> simply handling of what happens if a drawable is moved by the user  
> from one display head to a different display head, which likely has a  
> different msc count because it runs at a different refresh rate or  
> was enabled later than the first crtc.
> 

I understand.

Same problem may happen for application using OML_sync_control. That why I
think it is important to notify that we are hitting work around.

> Therefore i don't think this error printout should be added. It is  
> arguably not perfect, but probably good enough for the most common  
> cases. The correct solution would be to virtualize the msc counter  
> for each drawable and keep track of crtc changes and compensate for  
> those at each change. I have some ideas on how to do this properly,  
> just didn't have the time to test them.
> 

Do you add some callback to common code that DDX has to call when drawable
moves to different CRTC? Parameters could be msc for old and new crtc. Then
common code could add msc offset to drawable that driver can use to schedule
swaps at correct time.

> best,
> -mario
> 
> > Signed-off-by: Pauli Nieminen <ext-pauli.nieminen at nokia.com>
> > ---
> >  hw/xfree86/dri2/dri2.c |    7 +++++--
> >  1 files changed, 5 insertions(+), 2 deletions(-)
> >
> > diff --git a/hw/xfree86/dri2/dri2.c b/hw/xfree86/dri2/dri2.c
> > index d9b9d57..d70c115 100644
> > --- a/hw/xfree86/dri2/dri2.c
> > +++ b/hw/xfree86/dri2/dri2.c
> > @@ -860,9 +860,12 @@ DRI2SwapBuffers(ClientPtr client, DrawablePtr  
> > pDraw, CARD64 target_msc,
> >  	    if (!(*ds->GetMSC)(pDraw, &ust, &current_msc))
> >  		pPriv->last_swap_target = 0;
> >
> > -	    if (current_msc < pPriv->last_swap_target)
> > +	    if (current_msc < pPriv->last_swap_target) {
> >  		pPriv->last_swap_target = current_msc;
> > -
> > +		xf86DrvMsg(pScreen->myNum, X_ERROR,
> > +			"[DRI2] %s: GetMSC returned swap count that is in "
> > +			"past. Working around driver bug.\n", __func__);
> > +	    }
> >  	}
> >
> >  	/*
> > -- 
> > 1.7.0.4
> >
> > _______________________________________________
> > xorg-devel at lists.x.org: X.Org development
> > Archives: http://lists.x.org/archives/xorg-devel
> > Info: http://lists.x.org/mailman/listinfo/xorg-devel
> 
> *********************************************************************
> Mario Kleiner
> Max Planck Institute for Biological Cybernetics
> Spemannstr. 38
> 72076 Tuebingen
> Germany
> 
> e-mail: mario.kleiner at tuebingen.mpg.de
> office: +49 (0)7071/601-1623
> fax:    +49 (0)7071/601-616
> www:    http://www.kyb.tuebingen.mpg.de/~kleinerm
> *********************************************************************
> "For a successful technology, reality must take precedence
> over public relations, for Nature cannot be fooled."
> (Richard Feynman)