Re: debuggers

It might be worth considering giving some compiler support to doing soft
data watchpoints, by providing extra debug information to support this
scenario:


Rather than single stepping every instruction executed, and checking the
location, the debugger can:


1. single step conditional flow-control operations
2. put breakpoints after the next instruction that writes to memory


Using these two is sufficient to reducing the number of single-step
operations to (effectively) the number of conditional branches and memory
writes, which I'd expect to be about an order of magnitude better.


If a compiler actually wants to directly support it, every memory write
could be made a function call, with the obvious cost in performance, or
every memory write could have a nop (or multiple nops as necessary) after
it which the debugger can backpatch with the same function call. This
costs some performance (the extra nops), and it would take quite a while
for the debugger to patch every write-nop in the program when the debugee
(?) sets a data watchpoint, but it'd be a bit faster.


Using the hardware is even better, but I'm not holding my breath, given
the ubiquitousness of portable GNU software over platform-specific tools.


Sean Barrett
--

Return to the comp.compilers page.
Search the comp.compilers archives again.