Fix for 4.2 not seeing all vax memory?

[#Bill.Stewart.2G202.x0705]

In article <1361 at brl-tgr.ARPA> ted%bragg1 at braggfs.arpa writes:
>Hi folks,
>
>We are in the process of bringing a virgin 4.2 distribution tape Vax 780
>into the world.  We have 10 Meg of memory on it, but 4.2 doesn't see all
>of it.  I remember reading of this as a known bug and remember seeing a

I don't have the 4.2 fix but I've fixed this on a System V.  I'll talk
in generic names since the specifics may be different.  Cd to the
source directory (/usr/src/uts/vax) and look in the machine-language
stuff (.../ml) for a program that autosizes memory.  On System V R 2,
it's in assembler, but it's easy to read.  Look for constants around
8000; for some reason the value was <8192.  Look for a loop that keeps
incrementing by 64K, reading a block, and testing for failures.
The constant 8??? is the number of blocks it will try; you can increase
this up to 16K.  Above that, you may get messed up because the number
of memory pieces <blocks?> becomes too large to fit in a signed short,
and needs to be made unsigned or long to prevent addresses from
becoming negative (unsigned gets you to 32K memory; talk to DEC if you
really need more).

		(DEC and VAX are trademarks of DEC and/or VAX)

-- 
# Bill Stewart, AT&T Bell Labs 2G-202, Holmdel NJ 1-201-949-0705 ihnp4!ho95c!wcs