On Mon, Jan 20, 2003 at 12:23:23PM -0500, Don wrote: > > FreeBSD -CURRENT > XFree86 4.2.1 > > I have a system with 1 AGP Radeon 7000 and 2 PCI ATI Radeon 7000's. > > These cards are from Saphire and powered by ATI. They have a VGA output, a > DVI output and an S-video output. > > I can display on each monitor individually. They will not work at the same > time however. I have tried a number of variations on the configuration > file. I have tried disabling every option I can think of and nothing has > helped. > > The specific errors are: > (EE) Screen 1 deleted because of no matching config section. > and > (EE) RADEON(0): Cannot read V_BIOS (3) > > As far as I can tell, however, there is a perfectly valid config section > for Screen 1. > > If I alter the ServerLayout section to only reflect one screen, and I > specify only one card and test it, that card/monitor will work fine. > > I have tested all three in this way and each one works. > > I have tried disabling DRI, disabling DGA, etc. and nothing has helped.
The only thing that helped me was to hack the driver... :-\ My analysis of the problem went like this: The 7000 is aka Radeon-VE, which can support two heads on one card. The driver, recognizing the Radeon-VE, EXPECTS two heads to be configured on one card, and WILL NOT ALLOW only one head to be defined on it, when there is more than one "Device/Screen" section, regardless of the use of "BusID" option. I believe I actually tried a pair of 7000s with a config file that had 2 pairs of 2 screens as appropriate, and still had a problem (though I can't remember if it was the same as you noted above). Also, since I had PCI 7000s from ATI which had ONLY a VGA output (not DVI nor S-VID), I could configure a card with 2 heads, but could only see the primary one. IIRC, I had to hack two places to get the driver to allow only one screen per card when more than one card was present; one was in radeon_driver.c:RADEONPreInitConfig(), to prevent "HasCRTC2" from getting set for the RADEON_QY (the VE), and the other was in radeon_probe.c:RADEONProbe(), to prevent the RADEON_QY from being marked as "sharable". I don't know why the "HasCRTC2" hack was required, beyond the more obvious "sharable" hack, but it was the only way I could get it to work. >From the code, it also looked like there would be some other flavors of Radeon that would be susceptible to the same problem, so this may not be limited to a 7000/VE problem. --Jay++ ----------------------------------------------------------------------------- Jay A Estabrook HPTC - LINUX support Hewlett-Packard Company - MRO1-2/K15 (508) 467-2080 200 Forest Street, Marlboro MA 01752 [EMAIL PROTECTED] ----------------------------------------------------------------------------- _______________________________________________ Devel mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/devel