On Mon, 11 Mar 2002, Hamish Moffatt wrote:
> On Sun, Mar 10, 2002 at 12:43:15PM -0800, Mark Vojkovich wrote:
> > The NVIDIA driver supports multiple cards (I have a GeForce2 +
> > TNT2 system as my main development machine), but it sometimes
> > has a difficult time posting the second head properly so you
> > may have to Option "UseInt10Module" to the Section "Device"
> > for the secondary card (the one the console does not come up
> > on).
>
> Hmm, OK I'll try that. I noticed that /proc/nv/card0 and card1
> both show AGP as being enabled (using either agpgart or
> NVdriver's internal AGP support) which seems wrong. Could
> that be a problem too?
It might just be a problem with the reporting. Newer
drivers get this from /proc/driver/nvidia/cards instead.
>
> To complicate things further I think both cards are using the
> same IRQ. Is that a problem?
In theory it shouldn't be, but on my system they're on separate
IRQs. The NVIDIA driver definitely uses the interrupts.
>
> [nv driver]
> > Is the server eating CPU?
>
> I think so, but I'm not certain.
>
> > Did you build the server yourself? If it hasn't been
>
> No, I used the Debian packages.
Maybe they didn't strip out the debug info?
>
> > stripped you should attach with gdb and see where it's hung
> > at. From the sounds of it (the cursor not moving) it's
> > hung in core XFree86 input code. If it's not eating CPU
> > then it's probably stuck on select or something.
>
> I'll try to rebuild the server so I can gdb it.
>
> Once it's hung, I can only move the mouse within one screen,
> not between screens. And I can't click or type (nothing happens).
> The hang has only just started though (clock shows the current
> time, but never updates after the hang occurs).
>
Oh, I though you couldn't move the mouse. If you can move
the mouse it's probably something in the driver that's hosed -
hardware not responding, except for the cursor which is fairly
separate from the rest of the chip.
I assume you're refering to the "nv" driver here. I'm
not aware of any such problems with the driver itself, but
I believe there are some interactions with rivafb, so if you're
running the framebuffer console you might try without it. Alternatively
there might be something weird with APM, again maybe rivafb doing
something there. The nv driver itself doesn't do any power
management stuff other than shutting off the monitor.
Mark.
_______________________________________________
Xpert mailing list
[EMAIL PROTECTED]
http://XFree86.Org/mailman/listinfo/xpert
