Hi,

> But I've never seen something like that on a filesystem built with
> NetBSD's newfs ...

Well, it happens on a NetBSD newfs'd file system too for me. I am beginning
to suspect that my really old gcc-2.2.2 is generating bad code somewhere,
since NetBSD-current libc built but has some weird problems too.

> I've seen this "disklabel corrupted" or "no disk label" messages ever since 
> the bootblocks work on uVAX. But since xxboot finds and loads /boot even if 
> there's no label at all (and usually I do test new bootblocks without having 
> a label in them), I've never looked for the reason ...
> 
> I've just checked a few things which make me think that xxboot really 
> corrupts the in-memory-copy of the disklabel. I'll look for the reason
> of that in the next few days. 
> (If I can't find it, then I'll remove the warning message :-)))

No big deal. I have been trying to figure out a simple way to get a miniroot
booted and wanted the bootblocks to grab /boot out of the b partition. I
just added an offset to the block number to romstrategy() to do the same
thing. Since I noticed some mention of TK50 in comments, have you been doing
anything about bootstrapping a root fs onto disk from tk50?
(My current thought is just a little program that copies bootblocks and
 a miniroot image (dd) onto disk from memory. ie. this program + bootblocks +
 the miniroot would be concatenated together and then loaded into memory by
 booting it off tk50.)

Unless I here that you have something else going, I'll do this and let you
know how it goes.

Have fun and thanks for the info, rick