[Unix-jun72] mount woes

[Unix-jun72] mount woes

[Brad Parker]

I wrote this:

  exit  = 1.
  mount = 21.

	sys mount; rk0; usr
	sys exit

  rk0:    </dev/rk0\0>
  usr:    </usr\0>

And tried to assemble it with /bin/as, but it's a 407 exec and it didn't
work (core dump, created odd files)

So, I built it with apout & as like this:

  apout /backup/raid2/pdp11/v7/bin/as ./mount.s

  # reconsruct v1 0405 a.out header
  echo -e '\005' >mount
  dd if=a.out of=mount bs=1 skip=1 seek=1 count=11
  dd if=a.out of=mount bs=1 skip=16 seek=12

And it runs, but nothing happens.  I have not debugged it much.  Wish we
had a working v1 as.

-brad

[Unix-jun72] mount woes (fixed)

[Brad Parker]

Turns out not linking my a.out was a bad idea; I fixed it by doing this:

exit  = 1.
mount = 21.

	sys mount; rk0+040014; usr+040014
	sys exit

rk0:    </dev/rk0\0>
usr:    </usr\0>

This seems to work.  I think the rk fs image is not right, but Warren seems
to have a grip on that and I'll try his latest mkfs next.

-brad

[Unix-jun72] mount woes (fixed)

[Tim Newsham]

On Sat, 3 May 2008, Brad Parker wrote:
> 	sys mount; rk0+040014; usr+040014
> 	sys exit

We may not need mount afterall.  The init process mounts /dev/rk0 on
/usr during startup (after the second "1:" label on e12-01).

Btw, how does one manipulate the consoel switches in simh?  The
init process goes into single user mode (top of init on e12-01)
if the console switches are set appropriately..  I've stepped through
this code to verify that it is not doing that with our config
(if it was, it should go straight to shell with no login prompt).

> -brad

Tim Newsham
http://www.thenewsh.com/~newsham/

[Unix-jun72] mount woes (fixed)

[Tim Newsham]

> We may not need mount afterall.  The init process mounts /dev/rk0 on
> /usr during startup (after the second "1:" label on e12-01).

Oh, for what its worth, the init sources say its an "rk05" disk ("rko5" 
actually ;-) I think Warren was asking about which model to use earlier.

Tim Newsham
http://www.thenewsh.com/~newsham/

[Unix-jun72] rk03 vs rk05

[Warren Toomey]

On Sat, May 03, 2008 at 08:16:23AM -1000, Tim Newsham wrote:
> > We may not need mount afterall.  The init process mounts /dev/rk0 on
> > /usr during startup (after the second "1:" label on e12-01).
> 
> Oh, for what its worth, the init sources say its an "rk05" disk ("rko5" 
> actually ;-) I think Warren was asking about which model to use earlier.

Confusingly, the 1e manual for /dev/rk0 says its an RK03. I'll have to go
read the DEC manuals, but I would guess that the RK11 drive and interface is
the same, but the RK03 and RK05 removable disk packs were of different
storage capacity.

	Warren

[Unix-jun72] Dealing with 0407 a.outs in the V1 kernel

[Warren Toomey]

On Sat, May 03, 2008 at 10:35:47AM -0400, Brad Parker wrote:
> And it runs, but nothing happens.  I have not debugged it much.  Wish we
> had a working v1 as.

I think getting the V1 kernel to grok 0407 a.outs should be easier. Here are
snippets from the V1 and V2 a.out man pages:

1st Edition                               2nd Edition
-----------                               -----------
The header always contains 6 words:       The header always contains 8 words:
 1  a "br .+14" instruction [205(8)]       1  a "br .+20" instruction [207(8)]
 2  the size of the program text           2 the size of the program text
 3  the size of the symbol table           3 the size of the initialised data
 4  the size of the relocation bits area   4 the size of the uninit (bss) data
 5  the size of a data area                5 the size of the symbol table
 6  a zero word, unused at present         6 the entry location, always 0
                                           7 stack size required, always 0
                                           8 a flag indicating relocation bits
                                             have been suppressed


When I wrote the Apout support for V1 and V2, I found that all of an 0405
a.out was loaded into memory including the header, but the 0407 a.out is
loaded into memory MINUS the 8-word header.

So, I'm guessing, in the kernel's sysexec: code, where it does

        cmp     core,$405       / br .+14 is first instruction if file is
                                / standard a.out format

we have to add a section that recognises 0407, adds up the size of text,
data and bss (words 2, 3, 4), saves that in u.count, and loads the
rest of the a.out into memory but with the 8-word header removed.

Cheers,
	Warren