The /proc/ide/hd*/geometry reports different logical values for the same drive between the 2.2 and 2.4 kernel. [root@localhost /root]# cat /proc/ide/hda/geometry physical 12556/16/63 logical 787/255/63 [root@localhost /root]# uname -a Linux localhost.localdomain 2.2.16-22 #1 Tue Aug 22 15:48:40 EDT 2000 i686 unkn\own [root@localhost /root]# cat /proc/ide/hda/geometry physical 12556/16/63 logical 12556/16/63 [root@localhost /root]# uname -a Linux localhost.localdomain 2.4.0-0.43.12 #1 Wed Dec 27 18:40:11 EST 2000 i686 \unknown [root@localhost /root]#
a side effect of this procedure is that drives partitioned with linux in 2.2 space used the 2.2 logical numbers; but running fdisk with 2.4 will use the 2.4 logical numbers (different). As a result, fdisk will display errors about "partitions do not end on cylinder boundaries" ...
Another side-effect ocurred here, ist that my lilo stops with 'L' (without any hexnumber behind).
verified this issue is fixed in the qa0118.1 test tree ... in my example above, the 2.4 kernel used DOES report the same geometry as the 2.2 kernel did ...