Description of problem: The driver for intel integrated graphics in RHEL 5.x does not support 32-bit colors (?) with DRI and (at least) Q35 chipset, most OpenGL applications work fine but the error message libGL warning: 3D driver claims to not support visual 0x5b is printed on the console. However, at least MATLAB R2008a does not work properly. Consider the .m file below: function testmesh % if interval >= 1.0 plot works % if interval <= 0.5 does not work interval = 0.5; [X,Y] = meshgrid(1:interval:10, 1:interval:10); Z = sin(X/3).*Y - X; mesh(X,Y,Z); If the interval variable is modified so that it's > 0.5, the above function works correctly, but for interval <= 0.5, the plot area in the window remains empty. I have tried this with both Driver "i810" and Driver "intel" in xorg.conf. Both drivers have the same problem. Version-Release number of selected component (if applicable): xorg-x11-drv-i810-1.6.5-9.13.el5 I haven't tested xorg-x11-drv-i810-1.6.5-9.18.el5 in RHEL 5.3 beta, but based on the version number I suspect it's not any big change. AFAICT it seems this problem has been fixed in newer versions of the intel/i810 driver, so it would be nice if the driver in RHEL 5.3 could be rebased to a working version.
I'd like to add my vote to this, and some additional info. There seems to have been a regression (in the mesa driver?), since RH4.x machines are working, but 5.x machines do not. for example [X,Y,Z]=peaks(30); surf(X,Y,Z); does not work (produces a blank image) It currently only works for small number of points e.g. [X,Y,Z]=peaks(5); surf(X,Y,Z); The glxinfo command produces the libGL warning: 3D driver claims to not support visual 0x5b error on the 5.x machines. The result of this is that 5.x users are required to use software rendering instead of direct hardware rendering. I am required to keep the users on RH4.x until this is fixed, therefore is a show stopper for me. I believe this is also causing crashes on IDL software too. Hardware is Intel 82945G/GZ Integrated graphics running glxinfo on 5.3 gives 0x5b 32 tc 0 32 0 r . . 8 8 8 8 0 0 0 0 0 0 0 0 0 Ncon which does not appear on 4.x, however this may not be related to the matlab regression.
The crashes on IDL software can be mitigated by switching to software rendering and IDL produced updates in the meantime to try and work around the fault, but the driver definitely needs rebasing as it has a regression: The current EL5 driver also causes X to crash on the Q35 chipset when using Firefox on specific web pages. The Q35 boards themselves are RH hardware certified. Other (older and newer) Intel chipsets are not affected. https://hardware.redhat.com/show.cgi?id=383961 https://hardware.redhat.com/show.cgi?id=383941 Backtrace: 0: /usr/bin/X(xf86SigHandler+0x81) [0x80d1bb1] 1: [0xb64420] 2: /usr/lib/xorg/modules/libfb.so(fbCopyAreammx+0x21e) [0xf6b6ae] 3: /usr/lib/xorg/modules/libfb.so(fbCompositeCopyAreammx+0x66) [0xf6b776] 4: /usr/lib/xorg/modules/libfb.so(fbComposite+0x57b) [0xf5c4fb] 5: /usr/lib/xorg/modules/libxaa.so(XAAComposite+0x1db) [0x556eeb] 6: /usr/lib/xorg/modules/drivers/intel_drv.so(i830_xaa_composite+0x148) [0x211da8] 7: /usr/bin/X [0x816aa46] 8: /usr/bin/X [0x81673b6] 9: /usr/bin/X(CompositePicture+0x153) [0x8154503] 10: /usr/bin/X [0x815a34f] 11: /usr/bin/X [0x81576c5] 12: /usr/bin/X(Dispatch+0x19a) [0x80894fa] 13: /usr/bin/X(main+0x485) [0x8070755] 14: /lib/libc.so.6(__libc_start_main+0xdc) [0x2c1e9c] 15: /usr/bin/X(FontFileCompleteXLFD+0x1fd) [0x806fa51] Fatal server error: Caught signal 11. Server aborting
Note that https://access.redhat.com/kb/docs/DOC-7722 discusses X panicing on exit in Xen kernels with more than 2Gb ram installed and claimed the issue is fixed. The issue noted in comment 2 is with 4-8Gb 32bit PAE kernels and happens in mid-session.
No further hardware enablement updates are planned for RHEL5's X stack. Please use RHEL6 for up-to-date intel graphics support.