RE: 2.3.99-pre7: intermittent PCMCIA crashes on loading yenta.o

From: J.W. Hoogervorst (jeroenho@ic.uva.nl)
Date: Tue May 16 2000 - 04:02:52 EST


Hello, all!

Yesterday I already wrote some about my quest to get the yenta code
working again on my Dell I7500. Here's an update, so people won't be
trying to all do the same ;-)
Well, I looked a bit more into it, and placed some debug code in
yenta_interrupt. I made the setup code put an magic number in
socket->private[0], and cheched for that to make really sure the
structure was what the driver expected.
Still an lockup.

Then I put some printk's into it (at the beginning and the end of
yenta_interrupt), and they rapidly scrolled off the screen.

Conclusion: the system isn't really hanging, just really busy with
those interrupts, and because the things done in the interrupt code are
also done in other places where there hasn't been an interrupt, the
TI1225 isn't hanging the system either.

Now if I only could find some docs on the ATI Rage mobility to disable
the interrupts, I could at least get systems using that chip to work...
Anybody got those?

But an better solution would be to find out why the kernel can't cope
with the interrupt frequency, the system should be able to cope with it
in X anyways... I'll leave that to the more skilled than me, 'cause I
really wouldn't know where even to start.

And of course losedows works on it.

Or are X and windoze disabling the interrupts, and - where necesary -
polling the ATI chip?

Well, anyway, for now I'm stuck. I tried some code from XFree 4.0,
that claims to disable the interrupts (although only for the IBM 8415
support, it seems), but that doesn't work.

Maybe I should figure out what the video bios is doing, but that won't
be before the weekend. Tonight I'll try adding some checks first to
check if the TI1225 really generated the interrupt, and see what that
does.

With regards,
Jeroen Hoogervorst

> I have a Dell Inspiron 7500, and have this problem EACH time
> pci_socket gets loaded
> (after yenta.o), since 2.3.99pre7-pre6
>
> After some days of searching and fiddeling with the code, I
> came to the conclusion
> the changes in vga16fb.c are indirectly the cause of this:
>
> Because vga16fb started working in 2.3.99pre7-pre6, the
> display goes graphical,
> and this seems to enable the video interrupt, that is shared
> by the PCMCIA controller.
> It seems the PCMCIA interrupt code can't cope with this many
> interrupts, or the code doesn't
> like the interrupts from the videocontroller (maybe the
> system locks up because the controller gets
> read when there is nothing to be read.
>
> For now, i placed an "return -1;" in the place of the old
> return code from 2.3.99pre7-pre5 in vga16fb.c,
> and all seems to be ok for now.
>
> I'll be looking in to it, but I don't really what to look
> for, so more knowing minds are asked not to wait for me ;-)
>
> P.S. I sometimes had this in earlier kernels too, guess this
> was due to X or similar.
>
> With regards,
> Jeroen Hoogervorst
>
> > > It seems that 2.3.99-pre7 has intermittent problems when
> > > loading PCMCIA modules
> > > at boot time again: the system goes into a hard lockup after
> > > loading the yenta
> > > module in random cases.

-
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 : Tue May 23 2000 - 21:00:10 EST