[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Weird crash with dhcpcd and X11



Mark H. Mabry wrote:

> >>>>> "Brandon" == Brandon Mitchell <bhmit1@mail.wm.edu>
> >>>>> wrote the following on Tue, 7 Jul 1998 11:14:50 -0400 (EDT)
>
>   Brandon> On Tue, 7 Jul 1998, Mark H. Mabry wrote:
>   >> Running Debian 2.0beta and 2.0.34 with a cable-modem.  I'm in the
>   >> process of converting from modem ISP access to using this
>   >> cable-modem.
>   >>
>   >> I get a complete lock-up of Linux when I try to run both X11 and
>   >> dhcpcd.  I can run either one alone and things are fine.  But
>   >> when I start X, after dhcpcd, I see the X server spew stuff to
>   >> the screen, then the screen blanks, and SILENCE.  No X, no hard
>   >> disk activity, nothing.
>
>   Brandon> Maybe a hardware problem.  Check for irq conflicts with
>   Brandon> your ethernet card.
>
> Thanks!  I believe that you are correct.  As soon as I read your
> message, it rang true.
>
> Proving it wasn't as easy, though.  Looking in /proc/interrupts didn't
> show any problem, because it didn't show my graphics card's (Riva 128
> AGP) interrupt even when X was running.  Since I dual boot Win95 I
> took a look there.  It showed that my ethernet card and my video card
> share the same interrupt (IRQ 11).  Then I search DejaNews with the
> new info and found people who had the exact ethernet card (3C905) and
> video card and were having the same problem.  They suggesting moving
> the ethernet card to a different PCI slot.
>
> It was disappointing to see that Win95 can effectively share IRQ11 but
> Linux cannot.

Ah, but Mark both these cards are PCI. Interrupts are almost always shared for
PCI cards. PCI actually has its own interrupt levels A, B, C, & D. Actual
IRQ(s) is (are) assigned at runtime. Unlike ISA, PCI was designed to allow the
sharing of interrupts and provides an easy way to identify which device issued
the interrupte. Linux handles this just fine. The problem must have to do with
Linux getting confused about the devices since it isn't aware that your RIVA
128 exists. Switching the slots the cards are in might work indeed.

--
Jens B. Jorgensen
jjorgens@bdsinc.com



--  
Unsubscribe?  mail -s unsubscribe debian-user-request@lists.debian.org < /dev/null


Reply to: