Issue on original Rabbit 2000

[spth]

I am unable (both using the 0.2.3 release and the current version from the repo) to write the original Rabbit 2000 (i.e. not a Rabbit 2000A or later) on the Rabbit 2000 TCP/IP development kit:

philipp@notebook6:~/sdcc-trunk/sdcc-extra/historygraphs$ /tmp/OpenRabbit-0.2.3/src/openrabbitfu --verbose --verbose --run --serialout /tmp/OpenRabbit-0.2.3/coldboot/coldload.bin /tmp/OpenRabbit-0.2.3/coldboot/pilot.bin ~/OpenRabbit/examples/hello/hello-RCM2200.bin /dev/ttyUSB3 
Reset Rabbit.
loaded 831 bytes from /tmp/OpenRabbit-0.2.3/coldboot/coldload.bin
Sending initial loader.
set baudrate to 2400
sending 276 initial loader triplets
set baudrate to 57600
loaded 5517 bytes from /tmp/OpenRabbit-0.2.3/coldboot/pilot.bin
Secondary loader format detected as Dynamic C 9
Sending secondary loader.
sending 5517 secondary loader bytes.
Negotiating baudrate.
set baudrate to 460800
set baudrate to 57600
warning: received NAK for subtype 0x46
warning: received NAK for subtype 0x46
set baudrate to 115200
parsed 140 bytes
CPU:  0x0000 (Rabbit 2000)
Freq: 0
sectorSize: 0x0000
numSectors: 0x0000
flashSize:  0x0000
Flash speed: 0 ns
RAM speed: 0 ns
loaded 1166 bytes from /home/philipp/OpenRabbit/examples/hello/hello-RCM2200.bin
warning: received NAK for subtype 0x43les/hello/hello-RCM2200.bin... 43% (bps: in=0, out=0)

It always stops at exactly 43%. I haven't looked into this more closely yet. It might even be a hardware problem on the Rabbit 2000 TCP/IP development kit. The problem is not the power supply, though (I've tried with a lab power supply instead of the one from the kit - exact same result).

[tomlogic]

Based on Appendix B of the Rabbit 2000 Product Manual, there are three differences between the original chip and the 2000A:

1. Certain instructions did not function correctly as described in Technical Note TN302, Rabbit 2000 Instruction Bug. The Dynamic C compiler corrects this situation automatically.
2. The external interrupt inputs had to be tied together with a resistor as described in Technical Note TN301, Rabbit 2000 Microprocessor Interrupt Problem.
3. Wait states did not function properly when used to access code in slower memories because certain instructions failed in these circumstances. This bug is fixed in the Rabbit 2000A through Rabbit 2000C revisions.

I'm not sure about details on which instructions failed with slow memories (item 3), but it could be involved here.

TN302 describes an issue when "an I/O instruction (prefix IOI or IOE) is followed by one of 12 single-byte opcode instructions that use HL as an index register."

Is the OpenRabbitRFU using the loaders from Dynamic C? If so, they should work correctly with that hardware. You could run a test with Dynamic C 9.62A to verify that the hardware is good.

If you're using some of your own code (either assembly or code compiled with sdcc), you'd want to verify that it doesn't include any of those instructions sequences.

[spth]

1. Should be fine: I use the pilot/secondary loader from Dynamic C 9. The primary loader is similar to the Dynamic C one (and doesn't use any of the problematic instructions from TN302). SDCC is not used for either one, only for the example program hello-RCM2200.bin, that I was trying to put in the flash (and SDCC works around the TN302 issue anyway).