First, I am unable to install in graphics mode: screen is badly corrupt, no GUI controls visible. Setup is able to run but it's impossible to operate it. Text mode setup of XFree86 recognizes Voodoo banshee and installs. When trying to run X aftewards, I am experiencing a number of bugs: 1. Corrupt display: GUI controls invisible, only bitmaps displayed correctly. Window border exists but not window content and not window buttons. 2. Switching resolution occurs but the next 10-15 seconds computer does not responds and finally hangs. No response to Ctrl-Alt-BkSpace. 3. X can even hang unpredictably upon startup. The bug was first experienced in RedHat 6.1, reported in 6.2 and never fixed. In fact I am unable to run X at all that's why high prioirity and high severity.
If you have a Creative Labs 3D Blaster Banshee AGP, there are some issues between the hardware (specifically the SGRAM) and the driver. There are working drivers on http://linux.3dfx.com/ but you cannot use the drivers there that are based on XFree86-4.x. The modifications that these drivers incorporate have not been incorporated back into XFree86 for whatever reason. If this is indeed the problem that you are having and would like some help in setting up the drivers from linux.3dfx.com since there aren't any instructions there for redhat-7.0 send me an email and I would be glad to help.
No, that was not Creative 3d blaster: Video board is ASUS AGP-V3200 with 16 MB SGRAM. CPU: Intel Celeron 333 at 300-333 MHz. ACorp 440LX motherboard. 64 MB PC100 RAM.
I too have had this problem on my Banshee's since RH 6.1 and have tried most of the custom X servers (3dfx and Creative drivers/servers) and none of them fixed the random resolution changes... XFree86 4.0.1 does not fix the bug and the bug also appears in UniX (as the bug is in the XFree86 server). The problem is "not" limited to the creative cards as some report, it appears that many of the generic cards have this problem. Unfortunatly this bug prevents the user from entering X at all..... this is a high priority bug and i look forward to its fix.... My system specs: ----------- Skywell and STB AGP Banshee Celeron 400 Kernel 2.2.17 XFree86 3.3.6 (also tried 4.0.1, 3.3.6, and Creative/3dfx servers) -----------
From what i can figure out the bug is in the XFree86 drivers because now that i have changed over to freebsd i still have the problem it seem to affect the video modes dose all sorts of things from shake video at hi rez of 1280x1024/70hz and i know its not the card of my monitor or vidcard because it works fine in Windows 2000 at that and higher rez. I think that the Severity of this bug is very high and will affect lots of new Redhat users and should be fixed a.s.a.p seeing that it has been a problem since Redhat 6.1 and so that people with a Banshee can enjoy Linux/Unix beyond the console. I have also experanced some of the problems of the other bug reporters. My System |-------------------------------------------------------| |Lightspeed 3300 STB AGP Banshee | |P3 500 Running at 560Mhz 112Mhz FSB | |256Mb Ram | |Abit BH6 MB | |Was running Redhat 6.2 now using FreeBsd 4.1.1 Stable | |XFree86 4.0.1 using the tdfx driver | |-------------------------------------------------------| THX Tristan
After more testing and upgrading kernels to 2.4 test 9 i have discovered that when the bug occurs (resolution incorrect/invalid) if one kills the X server CTRL+ALT+BACKSPACE and is returned to the console (screen does not return however, all is black) and one tries to re-enter X all returns to a working state (screen comes back from state of blackness)..... This is interesting in that the banshee can be recovered -previously untested (i think)- and shows that the X server must place the banshee in an incorrect mode/state which causes the card to hang..... However if the X server is reinitialised the screen returns from a state of darkness (however the error randomly reoccurs and one must repeat this step). I hope this helps people to correct the bug..... -n3m3sis
Noticed that there is dependancy between the CPU clock speed and degree of screen corruption. For my configuration minimum speed is 300 MHz CPU at 50 MHz bus at which more screen objects displayed correctly than at nominal 333 CPU and 66 bus. Overclocking further makes screen slightly worse but not proportionally to the clock speed: there is barely visual difference between onscreen display at 343 and 375 MHz (68 and 75 MHz bus).
Ok this might help some people this person claims to have fixed the SNOW problem (this does not seem to work for me... makes it worse but its at least he is trying) http://www.cs.sunysb.edu/~giri/xfree86-3.3.6.html This reportedly works on SOME banshees (not mine) however the creator is very helpful and responds quickly to emails so it shows good promise..... There is also an XF 4.0.1 fix so i am going to try that out very shortly. have a good one -n3m3sis
A patch to fix the banshee has been added to my latest build, and will appear in rawhide soon now (a week or so).
*** Bug 47343 has been marked as a duplicate of this bug. ***
*** Bug 39148 has been marked as a duplicate of this bug. ***
Ok everyone, we have an aparent fix for this longstanding banshee problem. It is in rawhide XFree86-4.1.0-0.9.0 and later. Can you all please try this release and update the bug report on wether or not the Banshee works ok for you now? I've gotten back a couple good reports from the community in email so far, so it looks good.
I got a banshee loaner, tested it, it works. I received no feedback from above, but have recieved feedback via email from people that it works. I assume silence == it works for everyone else too, and everyone is so happy that they're playing Quake III instead of bug report updating. <grin>
*** Bug 51252 has been marked as a duplicate of this bug. ***
*** Bug 17499 has been marked as a duplicate of this bug. ***
*** Bug 44907 has been marked as a duplicate of this bug. ***
*** Bug 52951 has been marked as a duplicate of this bug. ***