Bug#59312: boot-floppies: dbootstrap aborts when using raid partitions
>>>>> "Lee" == Lee Maguire <lee-debian@hexkey.co.uk> writes:
Lee> Package: boot-floppies
Lee> Version: 2.2.7-2000-02-13
Lee> I've been attempting to use the "frozen" install disks to install
Lee> Debian on a machine that uses a Mylex RAID controller.
Lee> [ ... ]
Lee> dbootstrap: fdisk.c:544 : fdisk_add_partition: Assertion
Lee> 'partition->name' failed. Aborted
Lee> Then restarts dbootstrap - which aborts with the same message and
Lee> restarts...
Lee> This happens every time dbootstrap checks the disks and finds linux
Lee> partitions. If I flip out to a console I can quite happily mount/umount
Lee> the raid disks (eg mount /dev/rd/c0d0p1 /target) and whatever.
Lee> [ ... ]
Lee> Any ideas?
I hope you can read and write C. 8-)
Since most of us do not have access to that type of equipment for
testing, it would be Very Good of you to take the time to produce a
patch that Fixes The Problem. That is the Fastest Route to a bugfix
known to Open Source Computer Science.
Please use the current CVS for this. Do you have writes on
cvs.debian.org? If not, patches sent to the list (with nice
ChangeLog entries please) are quite acceptable. One of us will
integrate them for you.
cvs -z3 -d :pserver:anonymous@cvs.debian.org:/cvs/debian-boot login
cvs -z3 -d :pserver:anonymous@cvs.debian.org:/cvs/debian-boot checkout boot-floppies
The problem you are experiencing lies in utilities/{libfdisk,dbootstrap}.
It is possible to create a root.bin that contains an unstripped `-g'
version of "libfdisk" and "dbootstrap". Do you need help figuring
out how? If you then build a kernel with support for your NIC and
ifconfig &c. by hand, you can NFS mount `gdb', its libraries, and the
sources, export PATH and LD_LIBRARY_PATH, then use `gdb' to attach to
`dbootstrap' (patiently waiting for you at the release notice screen)
and Debug. Use `file /sbin/dbootstrap' then `attach PID' from vt2...
If you're clueless on how to fix it, but can at least get the gdb
thing to work and send us a stack dump, perhaps we can deduce what is
wrong from that static dump and fix it. Better if you can debug it
interactivly though. I really hope you can do that.
Reply to: