Red Hat Bugzilla – Bug 29032
fisher wigs out on Emachines 333cs
Last modified: 2007-04-18 12:31:40 EDT
From Bugzilla Helper:
User-Agent: Mozilla/4.76 [en] (X11; U; Linux 2.4.1 i686)
I have a stock Emachines 333cs. Whenever I boot the machine under fisher,
about the time the kernel tries to mount swap (stuff scrolls by quickly, so
it is hard to say for sure), I get the following repeatedly written to the
Replace XX with seemingly random hex digits. There are eight of these per
line, and this continues until I power down. I have installed 6.1 and 6.2
with no problems on this box. I do not have a copy of 7.0, so I have not
Steps to Reproduce:
1. Install fisher
Actual Results: See description
Expected Results: The joy of running Linux.
Please try wolverine, our more recent beta release. It may well
fix this; several different bugs that could potentially have
created symptoms like this are fixed in that kernel.
We (Red Hat) should really try to resolve this before next release.
I tried wolverine, and I can't even manage to get it to install. I get the
initial install screen, type 'expert', and when the kernel tries to boot, it
gets to 'Checking if this processor honours the WP bit even in supervisor
mode...' and then hangs.
Is this problem likely to go away if I swap out the Cyrix MII in the 333cs with
an AMD K6-II? I don't have one handy, but I'll put some effort into tracking
one down if it's likely to fix it. I'll definitely keep the Cyrix around for
any testing you might want to do, but I would like to get this box running some
version of 2.4.x soon.
It might; there has been a fix recently for that particular bug on
Cyrix chips, and I think that fix is more recent than wolverine.
While I think it's in our latest rawhide kernel, that doesn't help
See bug #29143 for more information
Replacing the Cyrix MII fixed my problem. Wolverine now boots fine on the box.
We also have a Cyrix processor here to test with, so we can test
cyrix fixes locally.
This should be fixed for a while. If it doesn't work with recent kernels,
please reopen this bug
I'd like to reopen this bug. I, too, have an eMachines 333cs, and get largely
the same behavior. In my case, I get the same "output format" - eight groups
along the lines of "[<c01132f6>]" followed by a newline - but after a while the
scrolling stops and the machine is well and truly hung.
I tried videotaping the screen to see how far it the boot process got before
the unwelcome behavior set in, and the last line I can make out is "Activating
ISA DMA hang workarounds." The next frame of video shows an apparently cleared
screen filled with the objectionable output.
My version of RedHat is store-bought (7.1, deluxe workstation).
Is CPU replacement really the way to go, here?
could you try booting with "ide=nodma" to see if that helps ?
Tried it. No joy.
FWIW, I've uploaded a JPG of the last frame I caught before things went blooey.
It's at http://www.galexi.com/data/Frame323.jpg. It's a bit blurry, but maybe
someone can "see" something there.