Red Hat Bugzilla – Bug 55052
XFree4 glint driver draws corrupted pixmaps
Last modified: 2007-04-18 12:37:45 EDT
From Bugzilla Helper: User-Agent: Mozilla/4.78 [en] (X11; U; Linux 2.4.9-7 i686) Description of problem: The hardware acceleration when using the "glint" driver in XFree86 4.1.0 doesn't draw graphics correctly. When using netscape, for example, large sections of the page will have corrupted images. (See attached example file) Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1. Install default RH 7.2, with updates as of 10/24/2001 2. Run "Xconfigurator"--accept default values 3. Launch a program with graphics (i.e. netscape) Actual Results: xterm applications seem normal, but graphics programs seem to exhibit the corruption. Expected Results: I should be able to >read< the web pages... :-( Additional info: When running Xconfigurator, it displays the following: PCI Entry: Texas Instruments|TVP4020 [Permedia 2] X Server: XF86_3DLabs XFree4 driver: glint (default) NOTE: I can install XF86_3DLabs-3.3.6-42 RPM, and the accelerated mode works-AS LONG AS I use the option "no_pixmap_cache". However, I can't find an equivalent to the "no_pixmap_cache" option under XFree 4.1.0
Created attachment 34961 [details] What netscape looks like with acceleration enabled...
More testing shows that this problem only manifests itself depending on the resolution of the video: 1024x768: OK. Problem does not appear 1280x1024: OK. Problem does not appear 1600x1200: FAILS. Screen is corrupted. Haven't tried other resolutions yet. My monitor is a Compaq V1100.
To force the usage of XFree86 4.x, use "Xconfigurator --preferxf4" Please attach your XFree86 log file, and config file using the link below. Then we can proceed to troubleshoot this problem, and see what can be done.
Created attachment 36061 [details] XFree86 4.1.0 configuration file
Created attachment 36062 [details] XFree86 4.1.0 log file
Can you try Option "XaaNoPixmapCache" in your Device section.
I tried adding option "XaaNoPixmapCache" as requested and unfortunately I get the same results as before. I'm attaching both the XFree86 config file and the log file in case anyone needs to see it...
Created attachment 36641 [details] Xfree86 4.1.0 log file (with XaaNoPixmapCache)
Created attachment 36642 [details] Xfree86 4.1.0 configuration file (with XaaNoPixmapCache)
Can you do an 'lspci -vn' for me.
Can you also try this VideoRam 4096 in your Device section. You can remove the XaaNoPixmapCache now too. I suspect the driver could be misdetecting your board as an 8MB one where it really only has 4MB.
Adding "VideoRam 4096" seemed to clear the problem up. No longer seeing any corruption. I'm attaching the output of "lspci -vn".
Created attachment 36662 [details] Output of "lspci -vn"
O.k. Your board is an ELSA Gloria Synergy. Has it got a white socket without anything plugged into it ? If not, your board really has only 4MB and you probably need to try and upgrade your VideoBIOS - not for the fainthearted. If it does have something plugged into it then it could be defective memory.
I have 2 systems that both exhibited this problem. They're both Compaq AP350 workstations, and each one has a video card with a label that says "Compaq Gloria Synergy+". Neither of the cards have anything plugged into the white sockets, so according to what I can tell, each host has 4MB of video memory. From my standpoint, I'm not sure if it's worth upgrading the VideoBios, when an easy workaround exists. If it really is the fault of the Video BIOS, then is there the possibility of implementing a workaround in the probing routines? Or at the very least, a "gotcha" in the GLINT(4x) manpage? Thanks again very much for the help!
Closing bug as WONTFIX due to it being a hardware bug. I don't see any simple configuration workaround that can be made automatically. If possible it would be nice if the driver could auto detect this card, and work around it somehow.