On Sat, 28 Apr 2007, Rhialto wrote:

> On Fri 27 Apr 2007 at 23:04:44 +0200, Rhialto wrote:
>> I'll update, rebuild, and report, then.
>
> Alas. This is a clean crossbuild from amd64, from sources updated soon
> after the above quoted mail, and sysinst still fails, though possibly a
> bit different this time:
>
>
> I found only one disk, sd0.
> Therefore I assume you want to upgrade NetBSD on it.
>
>
>
>
>
>
>
>                           +-----------------------+
>                           |uid 0, pid 13, command sysinst, on /: file system full                         +-----------------------+
>
> /: write failed, file pid 13 (sysinst): user write of 118788@0x17a000 at 95548 failed: 28
> system is full
> [1]   Illegal instruction     sysinst
> #
>
> i.e., compared to previous cases the Illegal instruction happens even a
> bit sooner. :-(

 	One intersting datapoint might be to build a debug sysinst
 	to test. You can build a debug sysinst on your build box
 	with:

 	    cd .../src/distrib/utils/sysinst
 	    make CFLAGS=-g LDFLAGS=-g

 	To easily run this on your VAX you'll need to manually
 	install on a partition (ideally on a second disk), or setup
 	your VAX to boot with its root filesystem on NFS (this is
 	likely to be much easier to retest different sysinsts, but
 	may interact badly with sysinst configuring the network
 	interface)

 	If you get a coredump it will be against the sysinst with
 	full debugging symbols so you should be able to run 'gdb
 	sysinst sysinst.core' and see where it crashed

 	Actually you can probably just run 'gdb sysinst', then
 	'run' to run it under the debugger.

-- 
 		David/absolute       -- www.NetBSD.org: No hype required --