Subject: Re: booting an 11/750
To: None <port-vax@NetBSD.ORG>
From: Bertram Barth <email@example.com>
Date: 06/28/1995 13:48:48
> Hmm, bootstrapping a 11/750 is an interesting question and it would be really
> nice if someone with the hardware could figure out how.
> A couple of thoughts/comments:
> - the only tape driver in the current kernel supports unibus/qbus tape mscp
> (ie. if you 9 track is anything else, there isn't a driver yet)
I suppose it's TS11 or TS05 (TU80?) which needs the ts-driver.
If anybody is working on a ts-driver, please contact me.
[I've a TSV05/TS05 and some docs. Thus I started to write tsprobe()
but it's far from being of any interest ... ]
> - I don't know what it takes to get "racopy" to work properly once it is
> loaded into memory. (I suspect it uses stuff set up by vmb, so does that
> mean that racopy has to be loaded off the console tu58?)
racopy uses ra-routines to access mscp/ra-devices and rom-routines for
anything else. rom-routines are different for 11/750 and MicroVAX II:
- on uVAX vmb loads the device-driver for the boot-device (and only that)
into memory. Then the driver is accessible via rpb. Thus uVAXen must
be booted from a device to make it's rom-driver accessible.
- on 11/750 there are 4(?) types of devices supported by rom-drivers.
Current rom-routines access only the boot-device. But if I remember
correctly, then it's possible to access any of these 4 rom-drivers
since the addresses of the drivers can be calculated.
It might be possible to use the existing boot.fs / miniroot.fs (evtl. split
into different tu58 volumes) to install miniroot onto 11/750 if booted
from tu58. Maybe you want to try that first ...
> - I could create a "dd" image of a root fs (ie dearchive rootfs.cpio and
> then dd the disk partition) and then you would "just" have to get that
> image onto the first N blocks of the RAnn disk. Since you mentioned VMS,
> you might know how to do that using VMS QIOs to a disk mounted foreign
> - In the 4.3BSD installation guide, there were little toggle in at the
> console programs that read a tape block into memory to start bootstrapping.
> A similar program that read an entire file into memory could get the
> miniroot.fs into memory, but I am not sure if it could be booted at that
> point. (ie. what has to be done so that the vmb console routines are
> available to the boot code at this point?)
I'm not sure if the rom-routines for 11/750 are still accessible in
racopy (I don't have the sources here). If yes, then it could work
without changes. If no, then it should be easy to change racopy and
rom-routines/conf.c to support loading from tu58.