20 Apr
2014
20 Apr
'14
8:49 a.m.
On 20/04/14 01:15, Andrew Warkentin wrote:
I have now released my fixes for FreeBee to allow Unix
to install and
boot to multi-user mode. They are against revision d91346487fe9 of
FreeBee (the last to use the original version of the MMU emulation) and
can be imported with hg import.
Nice! Thanks for these.
The SSL certificate for dev.philpem.me.uk has expired. Frankly I can't
be bothered maintaining repositories in two places, so I've moved
FreeBee to Bitbucket:
https://bitbucket.org/philpem/freebee/commits/all
Pushing and pulling works the same as ever, but now you can create your
own forks, commit to them and send pull requests, which makes patch
management quite a bit easier. I'll be removing the dev.philpem.me.uk
repository shortly and updating all the links to it.
Your Mercurial username doesn't appear to be set (it's supposed to be of
the form "Real Name <email address>"), so I've patched the patches to
set it. If you have a Bitbucket account, you can add the email address
to it and "claim" the patches, which will also hide your email address
from the web interface.
Most of the changes seem to be easy to cherry-pick. I'll look at merging
the bus-error patch across to MMU_V2 and see if it improves anything.
Otherwise we continue on MMU_V1 and I'll close the V2 branch.
The installer complains about the
floppies not unmounting properly, but that doesn't seem to affect
anything.
It's probably a bug in the floppy emulation; it may be locked in a "read
only" mode of some kind.
Also, Unix doesn't reliably boot to the login
prompt unless
you remove .lpstartsched from /etc/rc (it's easy enough to do this with
a hex editor). I'm not quite sure what the problem is. Starting the
print server from the command line doesn't cause a hang, and the system
never seems to access the parallel port when the print server is started
(be it from /etc/rc or the command line). I think the hang might be
occurring somewhere after the print server is started.
It might be worth scanning the I/O traffic before it happens... perhaps
it's waiting for the printer to go ready, or waiting for some state
we're not setting.
I did notice that the kernel thinks the motherboard is a P5.1; I don't
recall adding any logic for the P5.1 detect bits, so that's a bug.
There's undoubtedly other brokenness.
Another issue is
that the ROM won't boot the hard disk at the moment (which I think might
have something to do with the head select).
I'll get that on the bug list :)
There's also an issue with one of the hard disk patches. The hard disk
*does* have a 17th sector -- that's correct. From what I've been told,
UNIX normally only accesses the first 16 sectors, unless there's a bad
block on the disk. In that instance, the 17th sector is then
(allegedly!) used as a "spare" sector.
Disclaimer: this is stated on comp.sys.3b1 somewhere, and I haven't
checked it.
Thanks,
--
Phil.
philpem at philpem.me.uk
http://www.philpem.me.uk/