Four Ubuntu Installations -- some pointers

Paul Sladen ubuntu at
Mon Sep 12 07:56:57 UTC 2005

On Sat, 10 Sep 2005, Robert Bronson wrote:

Hello Bob,

> Unfortunately I can't get past the install. [..] I have an AMD 64 Athlon
> machine with 1GB RAM. The motherboard is MICRO-STAR MS-7145. This
> machine is not exotic. It's a relatively new eMachines T6520,

Okay, lets see we can help---or how we can help *you* with debugging the
problem, as this is going to affect other people aswell (one other person
has already mentioned they had a problem with the same or a similar

> [..] the problem I see with the v5.04 installations. Both the AMD64 and
> the i386 releases have the exact same hang behavior:
> ehci-hcd: already loaded
> ohci1394: already loaded
> pci [success]
> ohci_hcd 0000:00:13.1 Unlink after no-IRQ? Different ACPI or
> APIC settings may help.

That's a very detailed and useful report;  thank you!  'ACPI' is the
'Advanced Configuration and Power Interface' and in a modern PC computer
ACPI is used as a standard, abstract way for Operating Systems to talk to
hardware that varies on a manufactuer-manufactuer and model-model basis.

It is effectively part of the BIOS.

ACPI is also known to be buggy on huge numbers of systems---something we
know about and people keep on finding with new brands of computer (quite
often, some of these manufactueres release a BIOS/Flash upgrade that can be
applied to fix the problems)...

These bugs tend not to be found originally during testing, because, as you
say, the system has only ever been tested with a particular version of
Windows as that is all that is required to *sell* the machine to the general

Because of there being a known tendancy for ACPI issues, the first thing I
did was search for:

  eMachines linux acpi

and one of the first hits is a page that humorously begins with the title:

  ''So you bought one of these turkeys too?''

I suspect you can imagine some of the contents of the page without even
looking!  :)  Sadly, laughing and grinning doesn't necessarily help us get
Ubuntu past the (presumed) ACPI bugs on your machine and booting!

The next thing to do is start searching with Google for phrases like:

  linux acpi emachines T6520
  acpi micro-star MS-7145
  "debian on" t6520
  linux acpi emachines desktop

Unfortunately, this machine appears to be so new that nobody has worked out
what's wrong with it yet, or how to tell the Linux kernel to work around the
issues that are broken.  Some other similar models of eMachine's did pop up:

and this posts hints to try installing Ubuntu using:

  linux noacpi nolacpi acpi=off

from the CD/grub commandline.  Unfortunately this is crippling Linux from
talking to lots of your hardware (eg. power-saving, battery monitoring) just
because some ACPI tables are wrong or interupt routeing is broken in the
BIOS by the manufactuer.

If you manage to boot Ubuntu with those options, you may be able to grab the
'DSDT' file:

  sudo cat /proc/acpi/dsdt > emachines-6520.dsdt

and upload it with a bug-report to Bugzilla:

The reason this issue is showing up on all the versions of Ubuntu (or other
Linux-based distributions aswell) is because it's not actually not Ubuntu,
or Linux's, at fault ...sadly, a bug in Linux is much easier to fix!

If you can isolate what's wrong;  or get the information about what is
wrong, a workaround can probably be sorted out.

The summer is familiar here.  Edinburgh, GB

More information about the ubuntu-users mailing list