perlmonger (![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
perlmonger) wrote2003-10-26 05:44 pm
perlmonger) wrote2003-10-26 05:44 pm
sad geekness
Well, I have a (mostly) working machine here again...
It started on Friday evening; I've an occasional X problem that manifests as an input device lockout - the mouse moves ('cos that's done by the video card), but neither keyboard nor mouse input will input. This has survived moving from Rage Pro to twinhead nVidia and a complete X rebuild, so ATM I'm just cursing and doing init 4; init 5.
Except on Friday, when this process cattled the box entirely and, on reboot, the nVidia drivers wouldn't. I switched X configs to the built-in XFree nv driver and continued single headed.
At this point I suffered a major brain aneurism and decided this would be an excellent time to try a 2.6 kernel build (I've been on 2.4.20 most of this year). Configuring the thing is nice - the new KDE-based xconfig is irritating and badly designed, but a huge improvement on the old Tk effort.
Building OTOH...
Failure one led to installation of up to date binutils.
Failure two meant a deep breath, a lot of crap clearing out from /home and a long wait for gcc 3.3.2 to bootstrap itself.
Coo. A booting kernel. And nv-based X even worked.
The nVidia drivers wouldn't though and, unsurprisingly, the VMware kernel module build didn't.
Goole led me to this page which saved me huge amounts of grief - thankyou Gil Thorner, wherever you are. His VMware patches weren't sufficient for my setup (2.6.0-test8 and VMware build 4.0.2 build 5592), but his link to the Czech Republic sorted that one.
Oh, and somewhere in all this xscreensaver cut in and I discovered that it refused point blank to accept my password. And whilst a kill -HUP stopped it quite happily and allowed me to switch back to X, X crashed as soon as I touched the mouse.
I thought it worth rebuilding X with the new compiler and kernel headers at this point. Which seemed to go fine until I logged out and found that not only were the nVidia drivers corpsed again (no shock there) but that xdm refused to run either. By now, this was Saturday night and I decided bed was the only sane response.
Sunday morning.
Well, it turns out I've been using kdm all these years without realising, and it's fscked now. Turns out gdm works fine, so that issue can go on hold indefinitely. Another Goole search found that my xscreensaver problem is actually a PAM problem and turned up a patch for the latest (0.77) version that left all as it should be. Can't get data on the second CPU from top, but an update to the latest procps fixed that.
Which all leaves me with a non-working USB webcam, and some sort of codepage problem with accessing smbfs shares on the W2K/AS server. These, I think, can wait for now.
Meanwhile, in other news, we have our first Panther install part way complete on the downstairs iMac...
It started on Friday evening; I've an occasional X problem that manifests as an input device lockout - the mouse moves ('cos that's done by the video card), but neither keyboard nor mouse input will input. This has survived moving from Rage Pro to twinhead nVidia and a complete X rebuild, so ATM I'm just cursing and doing init 4; init 5.
Except on Friday, when this process cattled the box entirely and, on reboot, the nVidia drivers wouldn't. I switched X configs to the built-in XFree nv driver and continued single headed.
At this point I suffered a major brain aneurism and decided this would be an excellent time to try a 2.6 kernel build (I've been on 2.4.20 most of this year). Configuring the thing is nice - the new KDE-based xconfig is irritating and badly designed, but a huge improvement on the old Tk effort.
Building OTOH...
Failure one led to installation of up to date binutils.
Failure two meant a deep breath, a lot of crap clearing out from /home and a long wait for gcc 3.3.2 to bootstrap itself.
Coo. A booting kernel. And nv-based X even worked.
The nVidia drivers wouldn't though and, unsurprisingly, the VMware kernel module build didn't.
Goole led me to this page which saved me huge amounts of grief - thankyou Gil Thorner, wherever you are. His VMware patches weren't sufficient for my setup (2.6.0-test8 and VMware build 4.0.2 build 5592), but his link to the Czech Republic sorted that one.
Oh, and somewhere in all this xscreensaver cut in and I discovered that it refused point blank to accept my password. And whilst a kill -HUP stopped it quite happily and allowed me to switch back to X, X crashed as soon as I touched the mouse.
I thought it worth rebuilding X with the new compiler and kernel headers at this point. Which seemed to go fine until I logged out and found that not only were the nVidia drivers corpsed again (no shock there) but that xdm refused to run either. By now, this was Saturday night and I decided bed was the only sane response.
Sunday morning.
Well, it turns out I've been using kdm all these years without realising, and it's fscked now. Turns out gdm works fine, so that issue can go on hold indefinitely. Another Goole search found that my xscreensaver problem is actually a PAM problem and turned up a patch for the latest (0.77) version that left all as it should be. Can't get data on the second CPU from top, but an update to the latest procps fixed that.
Which all leaves me with a non-working USB webcam, and some sort of codepage problem with accessing smbfs shares on the W2K/AS server. These, I think, can wait for now.
Meanwhile, in other news, we have our first Panther install part way complete on the downstairs iMac...
no subject