Alastair Watts wrote:
OK... Please ignore this :) Was a combination of a whole pile of things that I finally sorted out (like the malformed release marker file and a few other things that I missed!)
This has been a persistant, but intermittant problem in other archs in the "testing" dist since last Aug. or so. It seems like if your first attempt fails, the release file is kept and re-used on subsequent attempts rather than being renewed. There was lots of discussion earlier on the Debian-testing and Debian-user lists. About the quickest way to clear it was to re-start the install from scratch, unless you know what you are doing and looking for.
Did you have a chance to try compiling the 2.4.18 kernel yet? I have done LOTs of trials since my last message... all failures. I have tried compiling with the 3.0 gcc, tried the kernel-source from kernel.org and even tried the 2.4.17 kernel. I have changed about every config option I can think of. I must have done 10 - 15 different compiles at least.
The PERSISTANT and COMMON problem is that the ncr53c8xx SCSI driver properly detects the SCSI controller card and starts to initialize it, but the driver fails the cache test and disconnects the controller. This happens fairly early in the kernel execution. Subsequent disk accesses fail with a "Kernel Panic: VFS: Unable to mount root fs on 08:13".
There are some earlier messages during the controller init about "CACHE TEST FAILED: DMA error (dstat=0xa0). <3> CACHE Incorrectly Configured." followed a couple of line later by "CACHE TEST FAILED: script execution failed. Start 800F9414, pc=800F9414, end 800F9440". Strange thing is that the source code for this particular driver indicates it hasn't been changed in several years! It looks like something might be messed up in the DMA setup routines... or maybe one of the other files linked to this driver's source. I wish I could read and understand the source code better, but I am afraid it is beyond me. I would be very interested in your experiences with getting this kernle to run. At this point I really don't know if it is something I am doing wrong or if there is something awry in the kernel source code for the XLT.
Cheers, -Don Spoon- -- To UNSUBSCRIBE, email to debian-alpha-request@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org