This seems to have absolutely nothing to do with the number of heads on the hard disk drive. So far, it's only happened when I used the low rent fdisk to partition the disk prior to the Windoze install. (I've found that if I install Windoze first, Linux generally behaves pretty well. Windoze, on the other hand, does not play nicely with others. Besides I like grub.)
The low rent fdisk looks like it always wants to start on a cylinder boundary regardless of where the MBR ends. I've seen this not only on 98 but 2K as well. Fdisk leaves a few sectors of free space before the Windoze partition. Unfortunately, disk druid (or whatever the Fedora Core 3 installation program uses to partition) interprets these skipped sectors plus partition as some weird combination of free space and vfat. The free space and vfat occupies the same space on the disk according to the table that pops up.
The last time I ran into this I was trying to install 2K and FC3 on the same 160GB hard drive with a shared vfat partition. I thought it had to do with the extended partitioning I was doing. The problem didn't go away until I used Ranish Partition Manager to partition the disk prior to the Windoze install. I specified the start and end of every partition, not leaving any free space. After that, all I had to do during the install was format the partitions I'd created earlier.
This time I'm installing 98SE and FC3 on three separate hard disk drives. Next, I'm going to try to create tiny partitions out of the few free sectors. If that doesn't work, I guess I'll have to blow away Windoze, brute force partition, reinstall Windoze (UGH!), and then install FC3.
Code:
BUG
Assertion (heads < 256) at disk_dos.c:486 in function probe_partition_for_geom() failed.
is the first indication of a problem. If I click ignore, things proceed rapidly downhill. If I click cancel (over and over again), I can at least limp through to the format step, where things do finally go casters-up-and-spinning with an unhandled exception. Now I've got a diskette that some bug reporting feature created, but alas, it is not readable in Windoze, so it's a little inconvenient.
This appears to be the same as bug 138419 from what I can tell. (
https://bugzilla.redhat.com/bugzilla....cgi?id=138419 ) It appears closed, but I didn't see a fix before I got crosseyed reading page after page of mostly more identical bug reports. Aside from the usual response, what should I do with this diskette with the bug report on it? (It would really help if the exception reporter program would also suggest an URL where the newbie should send said bug report--maybe even give example mount commands for floppy drives!) I'd welcome any comments, commiseration, suggestions, etc. Thanks!