blob: e1d95e64279a10faa9541940bfa3c1dd56510919 (plain
# Quirks in the DY-4 DMV177
JTAG and Caching
If data or code caching is enabled on certain revisions of the PPC603e,
then the JTAG emulator interface become disfunctional. You can not
debug using the emulator on these chip revisions. On certain revisions,
it is so bad that when code caching is enabled, you can not even
download code reliably to the board.
Caching and Peripherals
When caching is enabled, care must be exercised to insure that all
peripheral addresses are still uncached.
Exar 88681 Clock
This board uses a different clock for the Exar 88681 DUART than is
documented in the Exar manual or the original MC68681 manual. This
resulted in the need for the the mc68681 libchip driver to support
BSP specific baud rate tables and the development of a DMV177
specific baud rate table.
In the end, this all works but you have a very limited range of
useful baud rates on the 88681 ports compared to what would have
been supported had DY-4 just followed the Exar or Motorola manual.
The full set of SCC addresses is not documented in the DY-4 manual
and they are not ordered as one would expect. Normally the four
SCC registers are ordered Control A, Data A, Control B, and Data B.
DY-4 orders them with B first.
This required extra time to debug.
SCV64 and the Foundation Firmware
DY-4 technical support did not offer code to determine which interrupt
sources were pending at the SCV64. They recommended calling into the
Foundation Firmware ROM monitor to figure this out. The Foundation
Firmware did not recognize enough interrupts on this board to be useful.
In the end, we gave up on their technical support's recommendation
and directly manipulated the SVC64. This is what we wanted to do in
the first place but we got no information from them to aid in this.
Luckily, the manual does document enough of DY-4's mapping of the specific
interrupt sources to make this work.
Z85C30 SCC Clock Speed
The Z85C30 SCC can be factory configured for 10 Mhz or 2.4616 Mhz. Code
had to be added to dynamically determine which clock was installed.
The board we had used a 10 Mhz clock. No testing was done with a 2.4616 Mhz
P2 Octopus Cable
DY-4's P2 breakout is large and a bit unwieldy. It was difficult to
fight into the VME cage we used. The SCSI connector comes off the
side and is very stiff thus making it difficult to route around
anything in the back of the cage. We gave up on trying to use
it in the first few slots of OAR's cage.