On Wed, 16 Jul 2003, Chris Tillman wrote: > On Wed, Jul 16, 2003 at 09:14:38PM -0400, Fred Heitkamp wrote: > > On Wed, 16 Jul 2003, Chris Tillman wrote: > > > > > On Wed, Jul 16, 2003 at 08:08:23AM -0400, Fred Heitkamp wrote: > > > > > > The root.bin is the same when you use the standard Woody > kernel to boot the installer, so it sounds like the kernel > itself isn't playing nice with the keyboard.
I think it's because I had the usb_ohci driver compiled as a module. Since I don't have access to the module, the keyboard doesn't work. I also can't fsck my drive because the e2fsck on the Woody disk is too old. In order to solve those two problems I have tried to compile a new kernel with a cross compiler. I have compiled three kernels, two benh from bitkeeper, a stable tarball, and so far they all crash with oops like: Machine check in kernel mode Caused by (from SRR1=49030): transfer error ack signed Oops: Machine check, sig 7 blah blah blah. If anyone is interested I will copy down the full Oops message and post it. I might post it anyway because a lot of the recent kernels I've compiled oops. Could it be from using a compiler with no Altivec support compiled in? My Mac has a 7410 PPC chip with altivec. BTW is there anyway to do a "print screen" on an Mac, or capture output on a serial port? I have a USB->serial converter. BTW2: How to you retrieve a specific version with bitkeeper or rsync? To make a long story short though, I got a kernel from penguinppc that works enough to get the ram disk going. I cross compiled a static e2fsck 1.33 on my PC so I could do the disk check (which found tons of errors and problems, so I am worried now.) and was able to get on my system again. I had to use the wget program to retrieve the e2fsck from my PC, which worked nice (Good work somebody.) I tried to use an nfs mount but there seems to be no inetd or some other missing files, on the ram disk. > > I'm tempted to mention the recent libc6 breakage, but I don't see how > that could affect you when you're booting root.bin. Also the symptoms > were errors during the init that prevented it from booting at all. Well it appears that the lack of usb_ohci natively in the kernel might have been the problem as far as just booting. I'm worried that the libc6 breakage has done some serious damage to my system. Anyone have an experience with that? My next kernels are going to have native usb, usb hid, usb_ohci and usb storage compiled in. It should save me some time. That way I can use my zip drive to transfer "rescue files" if I have to. Fred
