Eclipse GDB

Started by Jonathan Masters August 18, 2009
Hi All,

I am following up on my post of a week or so ago. I have spent the
entire day trying to get Eclipse+GDB to work. Everything appears to work
except nothing works.

To be less obscure, the init script runs in gdb without any errors or
problems. However, after loading the program, the script sets a
breakpoint at main(). There is nothing in the disassembly window, and
there is no indication that the debugger is communicating with Eclipse
to show where the target pc is. The step icons in the debug window are
also inactive.

If I type 'x $pc', the debugger responds with:

0x40000050 <_start>: 0xe5df00f8

Which is precisely what I would expect from the linker and startup files
that I have. If I then type: 'step' the debugger responds with:

'Cannot find bounds of current function'

The disassembly window shows that the processor is trying to run at
0x3fffd2dc which is obviously not sane. 'x $pc' still prints the same
output as above.

On a different tack, if I type 'continue' after loading the program I
get a number of different results: it either runs to the breakpoint at
main() or crashes. If it does run to the breakpoint, it might step once
or twice before crashing. If it does step, OpenOCD reports that a memory
read caused a data abort at the location just stepped.

Incidently, OpenOCD reports the same data abort error when trying to
read memoryt at 0xffffffff during the loading phase. I haven;t
determined exactly where yet.

I have the same results on two pieces of hardware.

I expected the learning curve on GDB to be hard, but this is out of this
world! I have alot of experience with Eclipse (none with CDT) and I
don't know where to turn now. This is the second installation of
Eclipse+GDB I have done and I have followed Micheal Fischers tutorial
fairly closely. Can anyone help steer my ship in the right direction?

Jonathan Masters

An Engineer's Guide to the LPC2100 Series

Thanks all for listening.

Sometimes putting it all down on paper for all to see makes the problem
much clearer. I have reduced the JTAG speed and now the whole thing
seems to be reliable. Thanks for the much earlier responses about my new
board which suggested a lower jtag speed.