This can be taken verbatim as an undeadly article. Thanks for the hard work and this update.
On Tue, Mar 22, 2011 at 3:17 PM, Kenneth R Westerback <[email protected]> wrote: > A recent flurry of commits have substantially modified the support > for MBR Extended Partitions. We hope in a positive direction. > > The two most visible changes are > > 1) We now correctly calculate the location of the next EBR when > traversing the list of EBRs. > > 2) Up to 256 Extended Boot Records are now traversed when looking for > the OpenBSD partition. The previous limit of 8 was a result of > conflating the two tasks of looking for the OpenBSD partition and > spoofing partitions we found during the search. > > 3) There is a hard limit of 2 ^ 28 - 1 disk sectors within which the > /boot program must fit. i.e. 128GB with 512-byte sectors. This is > synonymous with the first few sectors of the OpenBSD partition lying > below this limit. > > 4) The install script now warns when it fails to install the /boot > program and thus make the disk bootable. > > The partitions are also searched in a slightly different order than > previously, to make sure the install process, the boot process and > the kernel all agree on the 'first' OpenBSD partition. > > These changes could break some existing systems that managed, despite > previous bugs, to install on a partition above the 2 ^ 28 - 1 limit. > > We are VERY interested in such systems! If nothing else to find out > how many there are. (So far I have heard of 1). > > To get such a system working again, one can simply change the define > in sys/sys/disklabel.h, BOOTBIOS_MAXSEC. If anyone knows BIOS, EDD, > assembler and gcc assembler syntax and has such a system we would > be very interested in diffs enable reliable detection of systems > where it is safe to relax the 2 ^ 28 - 1 limit. > > Thanks again to ucsav for starthing this off with a diff. > > .... Ken
