On Tuesday 19 May 2009, Adam wrote:
> I'm stumped with this one. Any ideas?  Gateway mini-desktop from
> 2001, AMI BIOS, Ubuntu 8.04.2 has GRUB in MBR, 384 MB RAM
> (128+256).  When BIOS has "Quick Boot Enabled" (quick RAM test,
> etc.), no problem -- can run any OS from GRUB menu, all tests pass.
>  When "Quick Boot Disabled" in BIOS (more thorough RAM test and
> maybe more tests), GRUB menu appears, but choosing anything
> produces "Grub Error 28: Selected item could not fit into memory".

That's very interesting.  That sounds like the more extended BIOS 
memory test reserves a section of RAM that GRUB requires.

> I don't understand why a more thorough test at powerup would let
> GRUB load but then cause problems with the following step.  What
> I've tried so far: memtest86+ v2.xx (no errors found after 11
> hours), various combinations of DIMMs, moved jumper to reset BIOS,
> all no effect -- problem consistent, repeatable and reproducible. 
> Battery assumed OK, because it keeps time when power off.  Has
> anyone ever encountered anything like this?

I might have run into this error on a buggy machine once, though I'm 
not positive.

> Any guesses?

Well, I'm sure there's a way to track down specifically what's 
happening because the behavior is consistent.  But off the top-of-my-
head this sounds more likely to be a buggy BIOS issue, because you get 
a change in behavior based on a change in that alone.

> Or should I just leave the BIOS at "Quick Boot Enabled" and not
> worry about it?

Obviously you need to leave it in the setting that works, but of 
course it's also still of mild concern because you have to avoid 
something that should be an available feature.

> Thanks very much in advance for any ideas!
>
> BTW there seems to be a bug in Memtest86+ v1.70 (as shipped with
> Ubuntu 8.04 and 8.04.2) -- less than a minute into the test (time
> proportional to amount of RAM), it locks up with keyboard LEDs
> flashing.  As v2.xx runs uneventfully, I'm assuming that's a
> Memtest86+ bug and the RAM is OK.

There's no hard conclusion to draw concerning Memtest86+ v1.70 based 
on the above alone.  For instance, if your BIOS uses an undocumented 
memory area only when Quick Boot is disabled and Memtest86+ v1.70 
attempts to test that area then it's not clear which to blame.  But if 
you quickly scan through a list of bugs fixed in newer versions of 
Memtest, there's a chance you might find it was a Memtest "bug" that 
fits this description -- and a "bug" that's listed for Memtest might 
just be a work-around for a particular buggy BIOS...

  -- Chris

-- 

Chris Knadle
[email protected]

_______________________________________________
Mid-Hudson Valley Linux Users Group                  http://mhvlug.org
http://mhvlug.org/cgi-bin/mailman/listinfo/mhvlug
Upcoming Meetings (6pm - 8pm)                         MHVLS Auditorium
  Jun 3 - TBD
  Jul 1 - TBD
  Aug 5 - TBD

Reply via email to