>> So, X 4.0 seems to be the problem. Any ideas as to how to fix it? How
>> does it manage to grab enough CPU time to cause other tasks to stall,
even
>> tasks that have been reniced to -20?
>
>My guess is XFree 4.0 is hitting the hardware in a way that is locking out
>the PCI DMA. Can you raise it with the XFree folks see what ideas they have
Quite a few VGA cards will force the PCI bus into "PCI retry" state when
the internal request queue in the VGA chip gets full and the PCI bus is
trying
to write another command to the queue. This is very very bad because in that
state the PCI bus is completely locked out from any device other than the
one
forcing the retry. Locking out the PCI bus also locks out the ISA bus since
it
is essentially a peripheral on the PCI bus. In particular, I know that
Matrox
and S3 cards exhibit this behavior. The fix is to have the driver ask the
VGA
chip whether there is space in its queue before writing the next command.
I'm
afraid I haven't looked at the drivers for X 4.0 so the above is speculation
on my part, but I have experienced this problem in Windoze and discussed it
at
some length with Matrox hardware engineers.
Bruce Cowan
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Thu Jun 15 2000 - 21:00:31 EST