From: Brian Furtaw (brian++at++sgi.com)
Date: 05/02/2002 20:07:24
I've seen some very strange things happening with gluBuild2DMipmaps on
Linux. Turns out the Nvidia driver install does not install a new
version of libGLU.so on your machine. So you actually have the Mesa3D
version of it. I did investigate at one time what the exact problem was
that gluBuild2DMipmaps but I can't remeber what I found. The result was
that I got the SGI version of libGLU off a VWE-3.1 CD and used it
instead, this worked like a charm. Now VWE-3.1 CDs may be scarce these
days so your best bet is to go onto oss.sgi.com and get the source for
libGLU and build it yourself. Oops..I just checked on oss the libGLU
source is not there.
Brian
Marcin Romaszewicz wrote:
>
> It looks like gluBuild2DMipmaps is telling performer that it can't mipmap
> the texture you specify. Perhaps the texture has some format which GLU
> does not know how to deal with. The size passed to gluBuild2DMipmaps is
> not important, since images are always scaled to the nearest power of 2
> before mipmapping.
>
> I would start by writing a simple GL program which calls gluBuild2DMipmaps
> with the data in your image file and see if that crashes. If it does, send
> nVidia a bug report. This could be something like gluBuild2DMipmaps
> running out of memory, but it's probably not a hardware problem since GLU
> is usually not HW accelerated. If your simple GL program does work, maybe
> performer is sucking up so much memory that GLU does not have enough to
> build the mip levels - try decreasing the shared arena size. However, this
> is most likely a GLU bug, since no matter what data you send at it, it
> should not crash.
>
> -- Marcin
>
> On Thu, 2 May 2002, Ive Akkermans wrote:
>
> > Hi everyone,
> >
> > I'm having some very strange problems with my textures. Sometimes,
> > after Performer has loaded my .pfb models and their textures, I get the
> > following error for everyone of these textures:
> >
> > PF Warning/Internal(12): gluBuild2DMipmaps failed on wall.sgi
> > (id=2076)
> > PF Defaulting to bilinear.
> >
> > Although I'm sure my textures are properly formatted (2^n * 2^n, true
> > sgi format etc...) ! After these errors, my app immediately segfaults:
> >
> > Program received signal SIGSEGV, Segmentation fault.
> > [Switching to Thread 1024 (LWP 19205)]
> > 0x40c1cb00 in gluScaleImage () from /usr/lib/libGLU.so.1
> > (gdb) info stack
> > #0 0x40c1cb00 in gluScaleImage () from /usr/lib/libGLU.so.1
> > #1 0x00000005 in ?? ()
> > Cannot access memory at address 0x24a77d0
> >
> > (I can't make anything of this, looks like corrupted memory to me)
> >
> > But the weirdest thing is, I don't get this error always ! It only
> > happens from time to time... I noticed that, if I don't run my app for a
> > long time (let's say an hour or so), the error doesn't occur. But when
> > I start running it too much, the error re-occurs more and more often,
> > until it happens approx. 9 times out of 10. This is very annoying while
> > testing code.
> >
> > Could this be a sort of hardware failure ? (Other GL- or non-GL apps
> > never crash)
> >
> > Please, if anyone knows of this problem, or has an idea as how I can
> > solve this, I'd be very grateful !! This is driving me insane :(
> >
> > Hardware specs:
> > Intel Pentium III 730 MHz w/ 256kb
> > 384MB Ram
> > RedHat 7.1 (Seawolf)
> > NVidia GeForce2 GTS
> >
> > Software:
> > OpenGL Performer 2.5
> > Kernel 2.4.18
> > Nvidia 28.02 drivers
> > (from glxinfo:)
> > OpengL 1.3.1
> > glu version: 1.1 Mesa 3.4.2
> > glu extensions: GL_EXT_abgr
> >
> > Many thanks for any response,
> > ive.
> >
> > --
> > Ive Akkermans
> > K.U.Leuven - ESAT-PSI tel: +32-16-32.17.06 fax: +32-16-32.17.23
> > Kasteelpark Arenberg 10 e-mail:ive.akkermans++at++esat.kuleuven.ac.be
> > B-3001 Leuven-Heverlee http://www.esat.kuleuven.ac.be/psi/visics/
> >
--
----oOOo---- ----oOOo---- ----oOOo---- ----oOOo----
Brian Furtaw (brian++at++sgi.com) Mailstop: DSE-2612
Graphics Guru Office:(301)572-3293 Fax: (253)323-8182
12200-G Plum Orchard Drive OpenGL/Performer/OpenInventor/ImageVision
Silver Spring, Maryland 20904 Volumizer/Optimizer/React/PCI Device
Drivers
This archive was generated by hypermail 2b29 : Thu May 02 2002 - 20:20:35 PDT