EDK and GDB - Code does not run

nexys2 edk

After a few hours of brute-force mouse clicking I have successfully built and added auto-generated test code to a MicroBlaze for my Spartan3E 500 board.

I generated the MB with hardware debug support as I could not get XMD to connect to a software stub over the JTAG3 cable.

Now, XMD connects to the MB and GDB connects to the XMD server.

However, the code does not seem to want to run or step . I thought that it was time to ask for help as I am now seemingly getting nowhere.

I am trying to run the TestApp_Peripheral code (tried Executable and XmdStub selected), the custom linker script which XPS generated and Hardware selected in the XMD Debug Options. I also tried the GPIO test example (including UART RS232) but that doesn't seem to do much either.

Although I can't find an exact example for what I want to do I followed some others such as from http://www.eecg.toronto.edu/~pc/courses/edk/node4.html. My settings seem correct to me in Software Platform Settings, clock frequency is correct, debug_module is selected, RS232 is selected in OS and Libraries, Drivers all contain something.

Many thanks for any ideas of what to look for,

Rob

edk gdb debug options

your first problem was probebly due to jtag chain.

for your current problem :

1. have you put the start address - normally you put 0x100 - don't let field to be empty !.

2. also of course you should mark the booloop for bram initialisation <green>, and your application should not be markd <x mark over> !

3. cleen sw and regenerate your libraries and build.

Is the BootLoop the only thing that should be marked for BRAM initialisation?

I have a XMDStub in there as well for some reason.

I was trusting the linker script generator to place the program but where do you mean to set the address manually? Is it the "Use Default Linker Script" section under the "Set Compiler Options" - "Environment" tab?

I'm using 9.1i if that's any help.

its the only one to be marked after that you must do the update bitsream command.

the program start address is when defaul linkerscript is marked.

you can also add stach, heap size but this is the most important.

Just been trying the project using the default linker script and manual location of the program and it yielded the same result, after clean builds.

Could this be an issue? I don't have what looks like a reset pin allocated in the UCF.

# Net sys_rst_pin LOC=;

Memory reads using XMD seem to be coming back as 0x00000000, even at 0x100... That doesn't look correct at all .

i think it is good to have a reset pin.

don't know how you cope without it.

I will be moving onto the Spartan 3E Starter Kit this week.

I think itt has better support within EDK, I really did not want to try making a memory controller for the NEXYS2 board from the ground up anyway.

I should get that early this week so my next question somewhere will probably involve getting Core Generator modules and ISE modules into the EDK .

Thanks.

Rob