12-01-2009 03:55 AM
I am running ISE webpack in Linux, and often use picoblaze and KCPSM3 for my work lately.
Looking around online it seemed DOSbox was the way to run KCPSM3.EXE from Linux.. a bit slow, but it works.
However I've noticed that all of my files get saved with all caps filenames, and ISE does not seem to like files where the .VHD extension is capitals.
Originally I was keeping my picoblaze and main vhdl code in separate working directories, and copying over the picoblaze .vhd ROM file after each compilation, but I decided I would like to have everything in my project directory so I could try using the utility to load the configuration straight into the bit-stream so I wouldn't have to wait for my whole project to re-synthesize (that did not work, DOSBox claimed that a label was missing in the batch file, which was not the case.. looks like I could manually enter the commands, but once again I might get issues with the caps filenames anyway).
When I have everything in the same directory, the project loses track of the ROM vhdl file when it gets overwritten with a new file whose name is all caps, and even if I rename it back, it doesn't work, I have to change it to a totally different name and add it to the project again
Has anybody found any solution for this (aside from using a different filename, which I have just started doing)?
12-01-2009 05:20 AM
12-01-2009 06:53 AM
I also had a hard time putting KCPSM.EXE (and KCPSM3.EXE) in the DOSbox.
My solution was to let the tools work in a separated directory.
I wrote some shell scripts and *.bat files to shuffle the files around.
So in your case after the latest assembler run you would have a line similar to this:
cp ./UPPERCASE.VHD ../lowercase.vhd
and everything should be fine.
Have a nice synthesis
12-01-2009 07:50 AM
The undefined symbols are from a standard POSIX library. add
(or something like it) to the file
picoasm is merely an assembler. If you want a whole IDE with otehr capabilties, use kpicoasm. However, I have not experience with it.
Ken's toolchain updates the BRAM image in the FPGA image by using the Xilinx utility data2MEM:
I use this same technique for my projects. I will respond further this evening when I get home. Note: the Xilinx data2MEM documentation is incomprehensible: the link above instead.
12-01-2009 10:36 PM
Cool, thanks for the replies!
Yeah, looking around last night, I found that
#include <string> probably needs to be replaced with #include <string.h>
I've been busy debugging my code so I didn't get around to messing with it just yet... I'm almost done with my current project anyway, so I'll look into switching before I start coding up my next thing.