On Fri, 19 Jul 1996, Mark Brinicombe wrote:

> Ok thought I had fixed this one (it used to happen on faster processors) The
> problem is the code is hanging in the keyboard reset routine and continually
> restting the keyboard. I'll look into it.

The hanging only happened to me once (and I booted bsd-4444 about 5 times).
I have an ARM610 which I clocked up to 40 MHz.
 
> >-Copying to DOS floppies results in unreadable files - both when mounting
> > or using the mtools. Normally it did work with the mtools up to atapi-4339.
> Hmm odd, works for me ;-)

Really? It never worked properly when mounting with mount -t msdos ... 
but used to do with the mtools. This doesn't seem to be true with 
bsd-4444 any more.

> The panic would be printed beyond screen memory causing another page etc. etc.
> I'll have to look into the columns thing.

This would be nice. I don't even know what else I could try. At least I 
can use xterms do log in.

> Can you send me details, I have not had a problem with the daily script.

Jul 18 00:10:09 emw4maba /netbsd: Data abort: 'Permission error (page)' 
status = 00f address = 00000000 PC = 00002bb4
Jul 18 00:10:09 emw4maba /netbsd: 2350 bytes of undefined stack fill 
pattern out of 2350 bytes
Jul 18 00:10:09 emw4maba /netbsd: 2556 bytes of svc stack fill pattern 
out of 4096 bytes
Jul 18 00:10:10 emw4maba /netbsd: Process = f14e1600 pid = 1134 comm = join
Jul 18 00:10:11 emw4maba /netbsd: CPSR=60000013 Traceback info
Jul 18 00:10:11 emw4maba /netbsd: fp->lr=f00d7e28 fp->pc=f00e5f18
Jul 18 00:10:11 emw4maba /netbsd: Trapframe PC = 00002bb4
Jul 18 00:10:12 emw4maba /netbsd: Trapframe SPSR = 60000010

This has happened multiple times and it only differs in the number of 
bytes that fill the svc stack, process and pid.

> 4444 should have ATAPI support (if it was built with the PROTAGONIST config
> file (cannot remember off hand)

Yes it is.

> The source for the 4444 kernel will be checked into NetBSD-current shortly.

That's good! I prefer to compile my own kernels because they boot faster 
(not checking for non-existent IDE drives) and I can use my optimised 
mode definitions.

> Further more RiscBSD 1.2 I want to support the A7000. The latest kernels will
> but the NetBSD kernel will not.
> 
> So as I see it there will be a NetBSD 1.2 base kernel available but 
> RiscBSD 1.2 will be the NetBSD 1.2 distribution with a newer kernel 
> with A7000 support and the new interrupt system etc.

I wouldn't start this because it would confuse people if NetBSD 1.2 and 
RiscBSD 1.2 have different kernels.
Can't the 1.2 Release kernel get new features afterwards because it still 
has it's flaws? Strictly speaking this does not introduce new features 
but cures bugs like the problems when writing to floppies, silo overflows 
etc.
And the 1.2 Release kernel (which number does it correspond to?) will not 
necessarily be more stable than more recent kernels.

MfG Markus