> Andy, I guess since direct modification of registers is not working
> you cannot call arbitrary functions in the kernel as of yet? I find
> it a life saver at times to be able to do
> (gdb) print show_free_areas()
> and similar ;-) Also, you can do things like:
> (gdb) print panic("gdb panics")
> but only if you provide a 'malloc()' function in the kernel for the
> stub to call to get a memory pool to store the string into. I
> basically on the sparc allocate a page for gdb to alloc from and wrote
> simple memory pool routines and a malloc() which is static in the stub
> file itself, once the page is used up malloc() returns null. I find
> this very useful personally (especially for debugging the mm code).
Well, it's probably not that much effort to make this working too.
However, I'm quite overloaded these days and probably don't find the
time. As I said, if demand is high I'll add it sooner. Otherwise later :-)