chip@ateng.com (Chip Salzenberg) (11/21/89)
I've found the answer! For those who've forgotten, here is the problem I reported: According to chip@ateng.com (Chip Salzenberg): >We're running SCO Xenix/386 2.3 on a 16MHz Wyse 386 with 4M RAM, a WD1007- >WA2 ESDI controller and two CDC Wren V 300M drives. The BIOS is by Phoenix. > >When we boot this large kernel, the Wyse acts as if we did a hardware reset >-- it checks the memory and returns to the "Boot" prompt. However, if we >keep trying, eventually it works! > > Why is this big kernel killing the Wyse??? And the answer is: Because we didn't install SCO patch #xnx126! Excerpt from the documentation for this fix: CAUSE: The kernel has grown larger that 640K and /boot is trying to load into extended memory. Due to a timing problem on certain machines, pin A20 on the cpu is being dragged low, and the address /boot is passing to the cpu is incorrect. SOLUTION: By adding additional delays into the /boot binary, pin A20 goes high, and the correct address is passed to the cpu. The new /boot binary is available as a Support Level Supplement (SLS) called xnx126. This SLS is available from the Customer Services Department of The Santa Cruz Operation, Inc. Thanks to all (especially Paul Zola at SCO) who responded with the answer. -- You may redistribute this article only to those who may freely do likewise. Chip Salzenberg at A T Engineering; <chip@ateng.com> or <uunet!ateng!chip> "Did I ever tell you the Jim Gladding story about the binoculars?"