[wellylug] Install Suggestions

Ewen McNeill wellylug at ewen.mcneill.gen.nz
Sun Apr 3 16:25:42 NZST 2011


On 2011-04-03 15:25 , Grant McLean wrote:
> I burnt a 10.10 (Maverick) i386 image onto a USB drive and later onto a
> CDROM.  In both cases the initial signs were promising and the Ubuntu
> boot screen appeared with the logo at the bottom but then it dropped
> back to a text screen with a mass of messages scrolling by at very high
> speed.

My hunch from your description and your "screen capture" is that the act 
of printing out the kernel error message (which appears to be trying to 
include a stack trace) is, itself, crashing and that is triggering the 
same cycle of messages to be printed out and the same crash.  Repeat ad 
infinitum.  (Presumably it is somehow managing to avoid triggering a 
multiple fault in the same, which I'd expect to stop in a more static way.)

> It's my understanding that the numbers in the first column indicate the
> time (milliseconds?) since the kernel started running, but in this case
> the numbers are not increasing (even after leaving the messages to
> scroll by for several minutes).

Yes, that's my understanding too.  And makes me think that there's a 
very large volume of messages being buffered in the same period of time. 
  Modern computers can get quite a bit done in a millisecond.  (It's 
also possible that the messages and/or interrupts are sufficiently 
disabling the incrementing of the time-since-boot counter that it'd show 
the same time forever.)

> Does anyone have any suggestions of what I could try next?

If you haven't already run memtest86 (at least a couple of passes over 
the memory) it'd be worth trying that.  RAM errors can cause all sorts 
of weird symptoms.  (At a guess that'll take 2-4 hours for modern CPUs 
with modern amounts of RAM -- I usually just leave it running overnight 
assuming it doesn't show problems in the first 10 minutes.)

Beyond that: different distro (with different kernel version), or try 
the various "safe mode" boot parameters (disabling ACPI, APIC, etc) and 
see if they help.  (I have, in the past, seen various machines which had 
APM/ACPI/etc issues that prevented them from booting properly. 
Microsoft Windows works around a remarkable number of sins, so typically 
Linux ends up needing to work around those same sins too.)

Ewen



More information about the wellylug mailing list