[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

kernel 2.6.15, PPP, keyboard failure and freezing



Hi, there.

Recently I got myself some new hardware, which included an Asus A8N-SLI and an 
Athlon 64 X2 3800. After much confusion and loss of data (and wailing and 
gnashing of teeth[1]), I got Etch beta AMD64 netinstall up and running, 
connecting to the net through a dialup line on another computer over an 
ethernet. (Much downloading ensued, punctuated by frequent cursing at the 
speed of the local dialup line, which is particularly poor where I am at the 
moment.)

I had a couple of problems, though. The first was getting the X2 part of my 
chip running. I installed the k8-smp 2.6.12 kernel and found the timing 
issues which I'm sure has been discussed here already. So away to unstable I 
went to get 2.6.15-3 k8-smp image, and a happy little camper I became.

Until I tried getting dialup to work. Through KPPP (KDE 3.5 from 
alioth/unstable) I connect to my ISP, and this seems to cause my keyboard 
(PS/2) to stop working. I can switch windows using my mouse (USB) to, say, an 
open Konsole, and no matter how much I mash my keyboard nothing comes up. 
Ctrl+Alt+Fn doesn't work, either. KPPP seems to get lost in the logging in 
stage -- I don't think I've seen it actually connect.

After a minute or two the system freezes hard. I have to hit the reset button 
on my case to start it up again.

2.6.12 doesn't trigger these symptoms, so I'm left with no reason to suspect 
something like KPPP as the cause. Something in the kernel, perhaps?

Or it may be some other weirdness. I have an NVidia card and 8178 drivers, and 
I guess there is a possibility that my problem is related to this one: 
http://lists.debian.org/debian-amd64/2006/01/msg00665.html
Though I have not experienced the almost random freezing evident there. Mine 
has only ever frozen when dialling-up under 2.6.15.

I don't know what's going on. As such, I'm asking if anyone has seen similar 
behaviour with PPP, or has any ideas how I might attack this. I'm not on the 
list, so please please CC me.

Thanks,
Brendon


[1] Long story. A comedy of errors. What happened, if you care to know, is 
that I had also bought a new harddrive (SATA 300), and was in the process of 
copying an old Sid system over from my old harddrive when something broke. 
The reiserfs partition it was on (which I was trying to resize for some 
reason) puked, and lost its tree. I had to regenerate it. Most of my home 
files still seemed to be in the right places, and some others I knew I'd be 
able to track down without too much trouble. But lost+found was quite large, 
full of system files so it seems, and now my Sid refused to start.

After attempting, and failing, to rescussitate it, I gave up and instead tried 
installing the Sarge netinstall I had lying around. Start the system from 
scratch, was the idea. I planned to copy the lost+found and other stuff onto 
this new system then pick out the bits that I wanted to keep -- ie, the stuff 
in, or that was supposed to be in, /home.

It didn't work. I couldn't install it. The 2.4 kernel couldn't detect the new 
drive, and the 2.6 kernel couldn't generate the initrd image due to some sort 
of device labelling stupidity.

So after carefully copying my old data into another partition on my new 
harddrive using an RIP CD (the same one that happened to see the end of the 
reiser tree in the first place), I made a new install of Sarge on my old 
harddrive, copied that onto space on my new harddrive, copied the backed up 
stuff over from another partition into that one, started it up, and sure 
enough it was working.

Enter the villain -- Windows. My new machine is also used for games now and 
then, and as games these days are still very OSist, it remains necessary for 
me to have Windows installed. (Any actual work I do in Debian, though. I use 
Windows purely as a toy OS. I think that's appropriate.)

I already had a partition in place ready for Windows to please itself in. I 
was also ready, having just finished downloading the amd64 netinstall, to 
install AMD64 Etch. I did Windows first. That seemed to go alright, so I did 
Etch next, and toyed around in that for a while.

It wasn't until one or two days after when I discovered my original Sarge i386 
install had somehow become unmountable. Something had completely destroyed it 
-- bad superblock, and none of the backup superblocks worked.

I was left angry and bemused, until I read that Windows has a nasty habit of 
"stamping" other partitions during installation; one of the reasons to 
install it first. Now I was just angry. (Honestly, how can MS possibly 
justify this as being anything other than monopolistic lock-in tactic and an 
abuse of their position? In a just society (or so we're told), how can they 
keep getting away with stuff like this?)

Let this sorry saga be a lesson to anyone thinking they'll be alright just 
transferring data around and installing stuff on seperate partitions without 
making a backup first. I was a bit lucky in that this is the off-season for 
me and I didn't have much data that wasn't still on my latest backup -- from 
two months ago.

That's it. I gave up on the i386 partition, so here's hoping AMD64 Debian 
remains as stable and as useful as the i386 Debian has been to me in the 
past. I'd hate to have to download it all again. ;-)

Oh, and in case you forgot, I did have a question up above somewhere. 
Something about a kernel and dialup, I think. I told you it was a long story.

Attachment: pgpl6iOS7wTAd.pgp
Description: PGP signature


Reply to: