Re: netbsd on vax 11/730; booting in sim

[Johnny Billquist <bqt%softjar.se@localhost>]

John Klos wrote:
> > > Yes, it's nice to see activity, plus it's nice to have NetBSD 5 
> > > working as well as it does.
> >
> > ??? What do you mean "well"? I thing I've said it a number of times by 
> > now. It don't. Atleast not on VAXen. And it hasn't been for a year and 
> > a half. Except nobody tries to test it, so nobody sees it.
>
> What I mean is that NetBSD 5 is working VERY well compared with NetBSD 
> 4. Shared libraries work, paging works, many more things compile 
> properly, and, perhaps more importantly run properly, than with NetBSD 4.

Actually, since 4.0 atleast works, while 5.0 don't, I disagree.
Sure, 4.0 are missing some features, but atleast it don't hang.

> > The 86x0 support was broken a while back (I think it was a rototill of 
> > the code by Matt Thomas), but the hanging system thing was introduced 
> > with the yamt-idlewp import already at 4.99.20 about 18 months ago. 
> > NetBSD/vax haven't worked right since.
>
> I don't have an 86x0, and I'm not sure I've run into the yamt-idlewp 
> problem. I've use my VAXstation 4000/60 with NetBSD-current to do bulk 
> package builds for months at a time throughout 2008 without any crashes. 
> I'm about to start one with NetBSD 5, and if it took a day with your 
> 4000/90, then I expect I'd probably see something in the first week or so.

Well let me know how it goes.
It could be that I have some odd hardware problem as well, even though 
VMS can run forever (it seems) without crashing.
The worst is if it's some timing dependant bug, which means it can take 
a long time, and maybe not even show up in some machines depending on 
the speed of them.

> > And then we have the ever present crashes during build.sh when 
> > building groff (pic crashes).
>
> Is that the NaN issue? Or something completely different?

Something completely different. And there is even a TR on that one, but 
the problem have been around for atleast 10 years now, if I remember right.

> > So NetBSD 5 is not something I'd try to use on a VAX. :-(
>
> ...unless you're moving from NetBSD 4, I'd qualify. If people try it, 
> report broken things, and try to help where they can, it might be 
> releasable. At the moment, I'm running BIND 9.6, Apache 2.2.11, the 
> latest sendmail, pine, imap-uw, and all sorts of other things. It's not 
> perfect, but it's definitely usable and definitely better than NetBSD 4.

Since 4.0 atleast worked for me, while 5.0 will hang after a while, I 
see it the other way.

> I have hope because things are (perhaps slowly, but nonetheless) getting 
> better, even though they're certainly not perfect. Big problems, such as 
> keeping up with gcc, are inevitable. If it weren't for the work of 
> people on the VAX NetBSD port, my guess is that VAX support in gcc 
> would've been depricated already.

The VAX support in gcc is probably better now that it have been in quite 
a while. Big kudos to the people (I suspect Matt Thomas) for getting 
that done. However, gcc itself is getting so slow it's just not fun to 
even joke about anymore.
But apart from that, it's my view that the kernel and various subsystems 
around that have degenerated since 4.0.

> On the other hand, if someone integrates a lightweight compiler suite, 
> embedded and low memory systems might make very good use of it. I 
> vaguely remember some discussion about getting NetBSD to compile with 
> one of those lightweight compilers... pcc?

Yeah, pcc would be very nice to get in the game.

> All in all, I think NetBSD/VAX keeps many of the other ports more 
> honest. We're not just an i386/amd64 OS with some interesting side 
> projects, and I hope it stays that way.

One can always hope. But I somehow doubt it. This last year or two have 
really seen an acceleration towards getting more and more "new" stuff 
into NetBSD. Often just because "others have it".

        Johnny