gdb(1) in the tree

[Peter Grehan]

> I just committed PowerPC support to gdb in the tree.

  Thankyou for that Marcel !

> There are various problems with gdb. One of the most annoying is that
> if you set a breakpoint, it's not always being hit. I suspect this
> is a kernel problem and not a gdb problem, because the breakpoint is
> sometimes hit.

  PowerPC has separate non-coherent I and D caches - writing to 
instruction memory requires a a flush of the D cache and an invalidate 
of the I cache for that particular memory address.

  This sounds like that type of problem. I'm assuming that gdb uses a 
ptrace write to do the breakpoint. There probably needs to be an MD 
callout after a PT_WRITE_I.

  I'll see if I can work out what NetBSD does.

later,

Peter.