I'd like to report an issue w/ a 4.1.0.1 [1] multihead setup, using an on-board
AGP NVidia Riva Ultra 64 and PCI STB S3 ViRGE/DX - basically the combination
of these two cards causes the system timer interrupt handling to be slowed
down by a factor of ~5:
1. install PCI card, boot ok
2. run X -configure, pretty much fine [2]
3. run X (via startx / gdm)
- everything seems fine except system clock slows right down
- interactive performance seems normal
- sleep 1 takes about 5s realtime
- cat /proc/interrupts shows int 0 (timer) increasing at around 20 Hz
The slowdown only begins once the S3 card is initialised, and stays even after
stopping X. Running X in single-head mode (NVidia only) is fine. I moved
the STB card up and down the PCI slots to no affect (i.e. to change IRQs -
but do video cards/X drivers even use interrupts?).
Switching the PCI card to a Diamond S3 ViRGE (86c325) resolves the issue
(using exactly the same X config as the STB). I also tested a Cirrus Logic
card and a plain S3 (non-virge) card, both using the vesa driver, these
both worked ok wrt. system interrupts.
If this issue is of interest to anyone, I can provide further details
(XF86Config, logs, etc).
[1] as included by debian testing (woody)
[2] Monitor1 was not fully detected - the section was present but not
filled in as per Monitor0:
Section "Monitor"
Identifier "Monitor1"
VendorName "Monitor Vendor"
ModelName "Monitor Model"
EndSection
--
Ben Low
"Vote early and vote often." - Al Capone (1899-1947)
_______________________________________________
Xpert mailing list
[EMAIL PROTECTED]
http://XFree86.Org/mailman/listinfo/xpert
