I partitioned a large 4G hard disk, (observing usual precautions like boot sector below cylinder 1023). I created an extended partition and made some logical partitions within that, trying first fdisk and then disk druid. Regardless of the tool used, when the table is verified with fdisk -v, warnings appear that each of the logical partitions within the extended partitions overlap although they do not according to the columns of start and end locations
WPA: next day - there is some war going on between the BIOS (Award 4.51pg on M-Technology MIG-R540 motherboard), hard disk (Seagate ST34310a installed as slave on first EIDE controller), and Linux (RH5.2). (Processor is AMD-K6 233 for the record). None of these appear on R.Hat compatibility black lists? HDD spec for non-dos is CHS=8354/16/63. Setting these in the BIOS and booting Linux, this geometry appears in the boot sequence, but "Partition check:" has hdb:[PTBL] [4177/32/63] i.e. it has translated the geometry. I tried using these settings in the BIOS setup, but this does not sort the fdisk -v warning, which appears when cylinders > 1023 are used. Ignoring the warning and proceeding with install bombs out normally (I have tried this many times), with "Invalid operand: 0000 ...blah, blah, blah" when starting to format the partitions, or else just a lockup (cannot use alt-f3 to see what is going on). On the occasions the install ran through, the resulting installation is unbootable, locking up just after "Partition check" Latest flash BIOS update made no difference. Is DMA anything to do with this.
Reformatted so I (for one) can read it... Q> I partitioned a large 4G hard disk, (observing usual precautions like boot Q> sector below cylinder 1023). I created an extended partition and made some Q> logical partitions within that, trying first fdisk and then disk druid. Q> Regardless of the tool used, when the table is verified with fdisk -v, Q> warnings appear that each of the logical partitions within the extended Q> partitions overlap although they do not according to the columns of start Q> and end locations Q> WPA: next day - there is some war going on between the BIOS (Award 4.51pg Q> on M-Technology MIG-R540 motherboard), hard disk (Seagate ST34310a Q> installed as slave on first EIDE controller), and Linux (RH5.2). Q> (Processor is AMD-K6 233 for the record). None of these appear on R.Hat Q> compatibility black lists? Q> HDD spec for non-dos is CHS=8354/16/63. Setting these in the BIOS and Q> booting Linux, this geometry appears in the boot sequence, but "Partition Q> check:" has hdb:[PTBL] [4177/32/63] i.e. it has translated the geometry. I Q> tried using these settings in the BIOS setup, but this does not sort the Q> fdisk -v warning, which appears when cylinders > 1023 are used. Have you tried setting a geometry of 524/255/63 (which is the LBA version of the parameters you quoted), or even 696/192/63 (an oddball geometry that works on SOME systems, but not on others) ??? Both of those drop the cylinders below 1023... Q> Ignoring the warning and proceeding with install bombs out normally (I have Q> tried this many times), with "Invalid operand: 0000 ...blah, blah, blah" Q> when starting to format the partitions, or else just a lockup (cannot use Q> alt-f3 to see what is going on). On the occasions the install ran through, Q> the resulting installation is unbootable, locking up just after "Partition Q> check" Q> Latest flash BIOS update made no difference. Q> Is DMA anything to do with this.