Jerome Glisse | 1 Jul 10:56 2006
Picon

Re: Moving to git ?

Just wanted to add, even if i am sure most people here also
follow xorg list :). Xorg are thinking to move to "real" git organisation.
If i understand well, the git official repo would be accessible to developer
and each developer would have is git repo. Thus to push something
to official repo log on the machine go to the repo and do a pull from
your repo. You could also publish unofficial repo from your account.

This solve the way branch are handled when having shared repository
and thus make things easier and cleaner (no more need to edit .git file).
I think, if we move to git, that this approach should be taken.

best,
Jerome Glisse

Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
Martijn van Oosterhout | 1 Jul 14:43 2006
Picon

Re: Small patches for r128

On 6/30/06, Michel Dänzer <michel <at> tungstengraphics.com> wrote:
> On Fri, 2006-06-30 at 11:23 -0600, Brian Paul wrote:
> >
> > 2. Use 'else if' since x1 can't be both < 0 and >= fbWidth
>
> Actually, this points out the possible source of the remaining problems:
> x2 and y2 also need clamping to fbWidth-1 and fbHeight-1, respectively,
> and if (x1 >= x2 || y1 >= y2), the cliprect is empty and should be
> skipped.

*smacks forehead* well duh, ofcourse. Once I clamped x2 and y2 aswell,
all the wrapping problems are solved. It's still drawing over the top
of other windows though, but we're definitly closer.

Have a nice day,
--

-- 
Martijn van Oosterhout <kleptog <at> gmail.com> http://svana.org/kleptog/

Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
Michel Dänzer | 1 Jul 15:07 2006

Re: Small patches for r128

On Sat, 2006-07-01 at 14:43 +0200, Martijn van Oosterhout wrote:
> On 6/30/06, Michel Dänzer <michel <at> tungstengraphics.com> wrote:
> > On Fri, 2006-06-30 at 11:23 -0600, Brian Paul wrote:
> > >
> > > 2. Use 'else if' since x1 can't be both < 0 and >= fbWidth
> >
> > Actually, this points out the possible source of the remaining problems:
> > x2 and y2 also need clamping to fbWidth-1 and fbHeight-1, respectively,
> > and if (x1 >= x2 || y1 >= y2), the cliprect is empty and should be
> > skipped.

Make that '[xy]2 should be clamped against fb{Width,Height}-1, not
[xy]1, and if (x1 > x2 || y1 > y2) ...'.

--

-- 
Earthling Michel Dänzer           |          http://tungstengraphics.com
Libre software enthusiast         |          Debian, X and DRI developer

Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
Mesa3d-dev mailing list
Mesa3d-dev <at> lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mesa3d-dev
Martijn van Oosterhout | 1 Jul 16:50 2006
Picon

Re: Small patches for r128

On 7/1/06, Michel Dänzer <michel <at> tungstengraphics.com> wrote:
> > > Actually, this points out the possible source of the remaining problems:
> > > x2 and y2 also need clamping to fbWidth-1 and fbHeight-1, respectively,
> > > and if (x1 >= x2 || y1 >= y2), the cliprect is empty and should be
> > > skipped.
>
> Make that '[xy]2 should be clamped against fb{Width,Height}-1, not
> [xy]1, and if (x1 > x2 || y1 > y2) ...'.

I'm kinda hoping that the xserver isn't going to send windows
coordinates with negative width/height (ie assume drawable->[wh] > 0).
And that if the window is entirely offscreen, an unmap event will
cause nothing at all to be drawn (rmesa->driDrawable == NULL).

Have a nice day,
--

-- 
Martijn van Oosterhout <kleptog <at> gmail.com> http://svana.org/kleptog/

Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
Keith Whitwell | 1 Jul 17:00 2006

Re: Small patches for r128

Martijn van Oosterhout wrote:
> On 7/1/06, Michel Dänzer <michel <at> tungstengraphics.com> wrote:
>>>> Actually, this points out the possible source of the remaining problems:
>>>> x2 and y2 also need clamping to fbWidth-1 and fbHeight-1, respectively,
>>>> and if (x1 >= x2 || y1 >= y2), the cliprect is empty and should be
>>>> skipped.
>> Make that '[xy]2 should be clamped against fb{Width,Height}-1, not
>> [xy]1, and if (x1 > x2 || y1 > y2) ...'.
> 
> I'm kinda hoping that the xserver isn't going to send windows
> coordinates with negative width/height (ie assume drawable->[wh] > 0).
> And that if the window is entirely offscreen, an unmap event will
> cause nothing at all to be drawn (rmesa->driDrawable == NULL).
> 

You can't rely on nothing being drawn when the window is not visible, 
I'm afraid.

Keith

Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
Roland Scheidegger | 1 Jul 17:38 2006
Picon

Re: glPointSize in radeon driver

Brian Paul wrote:
>> I still can't quite see how you'd do that so it works no matter
>> what alpha blend function is used (unless you have hw which
>> supports texkill, which r300 for instance doesn't even support
>> natively). And, for drivers which support hw tcl, you probably have
>> to fall back to sw tcl or do some other hacks.
> 
> Are you concerned about emitting fragments outside the radius of the
>  large/smooth point?
Yes.

> I think we could use the alpha test to cull them.
Probably.

> There may be some cases that can't be easily handled (like textured 
> points with a particular texenv/blend mode) where the fragment alpha
> at the end can't be used for culling.  But I think the common cases
> for large/smooth point rendering could be handled.
Maybe. The fact that fglrx doesn't manage to do it correctly with
even the simple mesa demo made me think it might not be so easy.

> You're right about the sw tcl fallback.  But I think the loss due to
>  software tcl wouldn't be nearly as bad as the penalty for sw 
> rasterization of large/smooth points.
Certainly. But at the same time we'd be penalizing apps which do the 
right thing (i.e. query the max point size and if they need larger 
points, emit (textured) quads/tris on their own with the right transform 
on their own). Not to mention that the tcl fallbacks introduce their own 
set of problems (though in this case it probably shouldn't be a problem, 
as drawing the same primitive twice will always cause a fallback and 
(Continue reading)

Martijn van Oosterhout | 1 Jul 17:57 2006
Picon

Re: Small patches for r128

On 7/1/06, Keith Whitwell <keith <at> tungstengraphics.com> wrote:
> > I'm kinda hoping that the xserver isn't going to send windows
> > coordinates with negative width/height (ie assume drawable->[wh] > 0).
> > And that if the window is entirely offscreen, an unmap event will
> > cause nothing at all to be drawn (rmesa->driDrawable == NULL).
>
> You can't rely on nothing being drawn when the window is not visible,
> I'm afraid.

That's ok. It's just that we don't need to test x1 > x2 because the
window width won't be negative. If the window completely offscreen
it'll get completely clipped anyway...

Have a nice day,
--

-- 
Martijn van Oosterhout <kleptog <at> gmail.com> http://svana.org/kleptog/

Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
Michel Dänzer | 1 Jul 23:06 2006

Re: The mysteriously changing pClipRects

On Tue, 2006-06-27 at 15:30 +0200, Martijn van Oosterhout wrote: 
> On 6/27/06, Martijn van Oosterhout <kleptog <at> gmail.com> wrote:
> > > See r128FlushVerticesLocked(), e.g.
> >
> > The thing is there's this optimisation that avoid loading the
> > cliprects everytime if there are less than three. However, that relies
> > on people notifying eachother when they've changed.
> 
> And we have a winner! If I remove the optimisation that avoids
> uploading the cliprects everytime, the problem vanishes, completely.
> So the hardware cliprects are being blatted and my guess is that it's
> the Xserver doing it, given it's the only other thing accessing the
> hardware, and it doesn't do it via the DRM driver.
> 
> The question is, how to fix it. I've downloaded the source for
> xserver-xorg-video-ati but the only time it ever seems to do any
> locking is changing screen modes.
> 
> Where else in the xserver is the DRI locking dealt with? 

In DRIDo{Wakeup,Block}Handler(), in hw/xfree86/dri/dri.c. SwitchMode()
has to do it explicitly because it can be called outside of these.

> However, the Xserver doesn't fiddle the aux clip rects, only the primary, 
> which should be correctly restored.
> OTOH, this code seems very suspicious:
> 
>         info->sc_right    = INREG(R128_SC_RIGHT);
>         info->sc_top      = INREG(R128_SC_TOP);
>         info->sc_bottom   = INREG(R128_SC_BOTTOM);
(Continue reading)

Martijn van Oosterhout | 1 Jul 23:34 2006
Picon

Re: The mysteriously changing pClipRects

On 7/1/06, Michel Dänzer <michel <at> tungstengraphics.com> wrote:
> > Where else in the xserver is the DRI locking dealt with?
>
> In DRIDo{Wakeup,Block}Handler(), in hw/xfree86/dri/dri.c. SwitchMode()
> has to do it explicitly because it can be called outside of these.

Thanks for that, I'll see what I can find...

> >         info->aux_sc_cntl = INREG(R128_SC_BOTTOM);
> >
> > Shouldn't that last one be R128_AUX_SC_CNTL? Maybe it restores the wrong value?
>
> Maybe, have you tried it? :)

Yeah, doesn't make any difference. Mind you, I think that code is dead
anyway, several values in that structure are stored but never read...

Have a nice day,
--

-- 
Martijn van Oosterhout <kleptog <at> gmail.com> http://svana.org/kleptog/

Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
bugzilla-daemon | 2 Jul 12:10 2006

[Bug 7391] New: Memory leak detected in _mesa_HashInsert with cairo-dock

Please do not reply to this email: if you want to comment on the bug, go to    

the URL shown below and enter yourcomments there.     

https://bugs.freedesktop.org/show_bug.cgi?id=7391          

           Summary: Memory leak detected in _mesa_HashInsert with cairo-dock
           Product: Mesa
           Version: CVS
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Mesa core
        AssignedTo: mesa3d-dev <at> lists.sourceforge.net
        ReportedBy: popolon <at> popolon.org

Using last cairo-dock SVN :
http://www.gnome-dock.org/trac

I obtained a Memory leak detection (I use r300 driver) with this night Mesa cvs :

./cairo-dock                  
Mesa: CPU vendor: GenuineIntel
Mesa: MMX cpu detected.
Mesa: SSE cpu detected.
Mesa: Not testing OS support for SSE, leaving enabled.
Mesa 6.5.1 implementation error: Memory leak detected in _mesa_HashInsert
Please report at bugzilla.freedesktop.org
(Continue reading)


Gmane