I downloaded the VM and followed the instructions in Running_SW_on_FPGA_board.txt. With the default load on the board (as shipped) I was able to load linux and ping around.

Then I tried to load linux instead of orpmon but that failed. I have not verified, maybe the SPI isn't large enough so I tried to restore the board to defaults but that's not working. At this point I have very little confidence in uploading my own hardware to the board if I can't even make it work with the pre-compiled known good build. The problem is that the load seems to complete and succeed but then the console is non responsive (please note that /dev/ttyUSB2 worked initially).

Does anyone have any suggestions on how to debug what's going on?

I'm taking the liberty to post the log of the programming:


root@openrisc-VirtualBox:/home/openrisc/program-spi-flash# jtag ./program_spi.jtag
Connected to libftdi driver.
IR length: 10
Chain length: 1
Device Id: 00000010000011110011000011011101 (0x00000000020F30DD)
Filename: /home/openrisc/fpga_dev_board/ordb2a-ep4ce22/EP4CE22F17C6_pre.bsd
Warning svf: unimplemented mode 'ABSENT' for TRST
Parsing 5700/5704 ( 99%)
Scanned device output matched expected TDO values.

root@openrisc-VirtualBox:/home/openrisc/program-spi-flash# ./spiflash/spiflash-program -e -p /home/openrisc/fpga_dev_board/ordb2a-ep4ce22/output_file.rbf -a 0xc0000 -P /home/openrisc/program-spi-flash/orpmon.or32.sizebin
Manufacturer ff type 20 capacity 20
Enabling write
Sending chip erase
Waiting for busy to clear.
Manufacturer 02 type 20 capacity 20
Writing 256 bytes at 0 (718569 left)
Waiting for busy to clear.
Writing 256 bytes at 0x100 (718313 left)
Waiting for busy to clear...
Writing 256 bytes at 0x200 (718057 left)
Waiting for busy to clear.
[...]
Writing 233 bytes at 0xaf600 (233 left)
Waiting for busy to clear.
Address 0xc0000 (786432)
Manufacturer 03 type 20 capacity 20
3 operations done.

Then reboot and minicom but there's nobody on /dev/ttyUSB2

Any help would be appreciated.

PS: It's also not very clear how to use Quartus to generate the required binaries but I'm looking at the Makefiles, I'm all I need is in there. This said, if anyone wants to post an *updated* documentation, it would be much appreciated. I would gladly volunteer to help writing a doc for the ordb2a-ep4ce22. Everything out there is SO generic or not applicable.

Sorry, known bug in spiflash-program. The thread "Geting started with the ordb2a-ep4ce22 board, Ubuntu" at http://opencores.org/forum,OpenRISC,0,4666,0 explains this issue. Chip erase should take several seconds to complete, but early versions of spiflash-program sent the wrong command (works on Winbond chips, not ST). Change the chip erase command to 0xC7.

Thank you, I changed the command, recompiled and it worked like a charm.

Then I installed Quartus and tried the same thing as "iztok" on the other thread. I too was unable to get the compiled version to run. I'm thinking I'm doing something wrong in generating the output_file.rbf

I am surprised at the rather large footprint though. When I purchased the board, I was looking for a processor core that could run Linux (I need a TCP/IP stack) and co-habitate with another FPGA design of mine. Quartus claims that the project (as is with ethernet MAC) requires 17091 LEs out of the 22320 available. Does that sound about right? (I'm afraid I won't be able to merge that with my design of 12K LEs).

Thanks for the help though :-)

I have replaced the original Spiflash with an 8MB Winbond part. However, I'm having problems getting a match between the downloaded FPGA image file and the data read back from the flash. I checked to ensure that the data reads back as all 0xff after erase, so I know that works.
I have attached the original FPGA programming file, as provided with the Ubuntu virtual machine, and another file with the data read back from the spiflash after erasing and programming.
Has anybody else got the 8MB part to work?

Never mind. Bad soldering to blame for unreliable spiflash. 8MB spiflash works great.