[623] in linux-net channel archive
Re: Kernel panics with 1.2.10
daemon@ATHENA.MIT.EDU (Theodore Ts'o)
Thu Jul 6 23:36:02 1995
Date: Thu, 6 Jul 1995 12:29:31 +0500
From: "Theodore Ts'o" <tytso@mit.edu>
To: dhollis@hq.jcic.org
Cc: gpg109@rsphy1.anu.edu.au, linux-net@vger.rutgers.edu
In-Reply-To: Daniel Hollis's message of Wed, 5 Jul 1995 02:13:17 -0700 (PDT),
<m0sTQWD-000xApC@hq.jcic.org>
From: dhollis@hq.jcic.org (Daniel Hollis)
Date: Wed, 5 Jul 1995 02:13:17 -0700 (PDT)
The rocketport supposedly does't use interrupts at all. At least, that's
what the driver, documentation, and hardware manual claim.
I did move the wd card to several other irq's, with the same results --
crash every time. It doesn't seem to be related with the irq number at all.
Is it possible the rocketport driver is stomping on internal kernel
structures? Or simply interacting with the network drivers in a strange way?
I didn't see the original bug report, so it's a bit hard for me to
comment about what specifically is going on, but ----
It is correct that the Rocketport doesn't use interrupts at all. It
does tie itself into an unused timer slot, and while any port is active,
it polls on every clock tick. (This turns out to be more efficient in
just about every single case than an interrupt driven setup.)
I've been using a Rocketport in the same system as a NE2000 compatible
card, and have had absolutely no problem. It's possible that there
might be some sort of bug where the Rocketport driver is stomping on
kernel structures, but I judge it unlikely, since I should have gotten
many more bug reports than I have received to date.
- Ted
