Hi folks, I am using a BDI2000 to try to debug a custom board with a 440GP on it as well as gigabit ethernet. I figure I will run U-Boot for my bootloader and then use that to start up a Linux kernel.
I am able to connect to the custom board and flash the bootloader into place, but then the fun starts. I am using the bdiGDB version of the BDI, but I am very new to GDB. I downloaded DDD to help me out a bit, and that seems to work better, but I am still having some troubles. First of all, I can set hardware breakpoints, and I know I am limited to 2 of them -- but if I use the 'c' command to 'continue', the BDI never responds back and when I finally stop the CPU over the telnet session -- the code has already executed something it didn't want to. Also, I have different sections where the code dies -- and the only common thing about the different places is the fact that they are ALL return statements. This makes me think that my stack is being blased somewhere, but I don't know how to read where the stack is pointing before I execute those return statements or to follow the stack in GDB/DDD. Lastly, in U-Boot, I have to define a setting to write to the debug registers to acknowledge the BDI and give the BDI control. Is this necessary if I end up deciding I don't want to use U-Boot for a bootloader? Does the BDI require those debug ports to be written to before it can take over and insert the instructions over the JTAG? Most of the times, the errors I get are 0x0700 or ox1400 -- which I have been told are wrong op codes or a tlb miss that isnt handled. This is my first board I am ever bringing up and this is my first time with such a sophisticated CPU so _any_ information that anyone can give to me about this subject would be great. I have been reading the documentation as much as I can, but there is so much in the 1666 page IBM document that I oftentimes feel overwhelmed. Thanks and everything is appreciated. Sincerely, Brian Padalino ** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/
