[Mesa3d-dev] Re: GLX indirect rendering broken

[Alex Deucher]

On 9/23/05, Brian Paul <brian.paul at tungstengraphics.com> wrote:
> Alex Deucher wrote:
> > On 9/21/05, Brian Paul <brian.paul at tungstengraphics.com> wrote:
> >
> >>Brian Paul wrote:
> >>
> >>>I was going to look into bug 4087 so I updated my CVS tree and did a
> >>>fresh build.  Indirect rendering seems to be quite broken.  Almost
> >>>everything I try running produces a protocol error:
> >>>
> >>>$ ./gears
> >>>X Error of failed request:  BadLength (poly request too large or
> >>>internal Xlib length error)
> >>>  Major opcode of failed request:  143 (GLX)
> >>>  Minor opcode of failed request:  1 (X_GLXRender)
> >>>  Serial number of failed request:  37
> >>>  Current serial number in output stream:  38
> >>>
> >>>
> >>>If I set a breakpoint in _XError() I get this stack trace:
> >>>
> >>>(gdb) where
> >>>#0  0x009ead66 in _XError () from /usr/X11R6/lib/libX11.so.6
> >>>#1  0x009eb557 in _XReply () from /usr/X11R6/lib/libX11.so.6
> >>>#2  0xf6ef7e35 in __glXReadReply (dpy=0x8212050, size=0, dest=0x0,
> >>>    reply_is_always_array=0 '\0') at indirect.c:70
> >>>#3  0xf6ef8a9e in __indirect_glGenLists (range=1) at indirect.c:400
> >>>#4  0xf6f20427 in glGenLists (range=1) at glapitemp.h:100
> >>>#5  0x0804a948 in init (argc=1, argv=0xfef6d0e4) at gears.c:318
> >>>#6  0x0804ac9a in main (argc=1, argv=0xfef6d0e4) at gears.c:369
> >>>
> >>>Other programs generate the BadLength error from different GL calls.
> >>>
> >>>Is anyone else seeing this?
> >>>
> >>>I'm pretty sure it's a server-side problem and not in libGL.
> >>
> >>Can anyone confirm this?  Anyone?
> >>
> >
> >
> > I'm not getting craches, but glxgears cuases my display to blank if I
> > move the mouse over the window with both direct and indirect
> > rendering.
>
> Which driver?  The whole screen just goes blank?  I haven't heard of
> that happening before.
>

On a radeon x800 (r430) with or without direct rendering enabled. the
whole screen goes blank as if dpms had been triggered or the mode is
changing.  moving the mouse out of the window brings it back.

>
> > Also, on IRC, Michel also mentioned that moving the
> > glxgears window with indirect rendering locks the Xserver, but I
> > haven't tried that yet myself.
>
> I've fixed a memory corruption problem in Mesa that _might_ explain
> this.  I'll try to verify the fix this weekend.
>

I'll try again with latest cvs this weekend too.

Alex

> -Brian
>