Comment 16 for bug 22544

Revision history for this message
Grondr (grondr) wrote :

I just got zapped by this, too, and I've some additional debugging data, from
two different machine; let's call them WHITE and BLACK below. They're both AMD
Athlons, but with different motherboards.

On WHITE, anything that attempts to restart (phase two of the installer, or
shutdown from the GUI, or even a restart from the LiveCD) will instead turn off
the machine. This happens w/the Colony 5 images, but did -not- happen when the
machine was running a two-year-old version of Mandrake, nor Windows 98, nor even
the Knoppix 3.7 disk I just tried while debugging this. I haven't tried any
Breezy's except Colony 5, and this machine hasn't run any other Ubuntus.

On BLACK, on the other hand, reboot from the LiveCD correctly restarts the
machine. BLACK is running Hoary, and restarting that has always worked fine.
If it's important to debugging, I can do an install on BLACK and see what
happens; I've got a spare disk on which I can do a clean install without losing
my Hoary.

I can't figure out how to try Ben Collins' reboot=b|h tests below. Typing
"linux" at Grub isn't a valid command. I tried editing the boot line inside
Grub's built-in editor to say either "boot reboot=b" or "boot reboot=h" and then
booting; neither one made any difference & I don't even know if they're noticed
on that line in the first place.

WHITE (the failing machine) has an MSI K7T Turbo 2 motherboard; BLACK (the
working machine) has an MSI K7N2 Delta-L motherboard. They also have vastly
different video hardware, FWIW, but I'm not going to go into the rest of the
machine configurations unless someone thinks it's important for debugging.

I'll attach /var/log/dmesg and the output of dmidecode for each machine in one
large attachment after I submit this.

Note that this behavior is pretty much a showstopper for me---WHITE is slated to
become an unattended server, but I can't put Breezy on it if this shutdown
behavior isn't fixed. Even if I had remote-power-cycle hardware available for
it (which I don't), the fact that the machine is commanding itself off means
that removing and resupplying power won't turn it back on, unless the correct
bit it set in the BIOS. I can do that on this particular motherboard, but it's
a fragile solution, and anyway won't work without remote power cycle, which
almost no one has. I'd therefore argue that the severity of this bug should be
increased.