Re: iBook G3 stopped booting after last testing update

[Martin Kuball <martinkuball@web.de>]

Am Wednesday 02 January 2013 schrieb Martin Kuball:
> Am Wednesday 02 January 2013 schrieb Michel Dänzer:
> > On Mit, 2013-01-02 at 14:40 +0100, Martin Kuball wrote: 
> > > Am Wednesday 02 January 2013 schrieb Michel Dänzer:
> > > > On Die, 2013-01-01 at 21:52 +0100, Martin Kuball wrote: 
> > > > > Hello,
> > > > > 
> > > > > I made some progress. After switching update to sequential boot I saw
> > > > > the following exception:
> > > > > 
> > > > > machine check in kernel mode
> > > > > caused by (from SRR1=49030): Transfer error ack signal
> > > > > oops: machine check, sig: 7 [#1]
> > > > > 
> > > > > the top entry of the call trace is:
> > > > > gem_ioctl+0x60/0x94 [sungem]
> > > > > 
> > > > > But this seems really odd to me. If the error is in the kernel module,
> > > > > why does it still occur after switching back to the old kernel?
> > > > 
> > > > Maybe one of the userspace packages you upgraded is now using the broken
> > > > kernel functionality but wasn't before.
> > > > 
> > > But sungem ist the ethernet driver, isn't it? And indeed networking is
> > > broken.
> > 
> > Right, so the most likely trigger was the upgrade of anything network
> > related, or maybe some kind of boot/init related package which is now
> > setting networking parameters it wasn't before.
> > 
> > But a machine check leaves little doubt that the root cause is a bug in
> > the sungem kernel driver (or a hardware problem).
> > 
> > 
> > > Running ping gives "network unreachable".
> > > ifconfig simply hangs (and with it the shell/terminal from which it
> > > was started).
> > 
> > Given the subject says 'stopped booting', how are you able to test
> > that? :) 
> 
> See the top of my posting. I configured upstart (oh, my bad - I wrote update) to run the init jobs sequentially. After that i saw the error message and got a login prompt. But no networking. And no X.
> 
> The next things I will try is disabling eth0 and reverting every updated package to its previous versions. 
> 
> I don't think it's a hardware problem. The rescue disk works just fine - including newtwork.
> 
I made a step forward by doing something I should have done right away: disabling the init script that run just before the kernel produced the error message. And it is: laptop-mode. But still no clue as to what actually triggered the error. laptop-mode-tools was not updated.