On 17 Feb 2022 03:05, Hans-Peter Nilsson wrote:
> > On 15 Feb 2022 00:06, Hans-Peter Nilsson via Gdb-patches wrote:
> > > --- a/sim/common/sim-memopt.c
> > > +++ b/sim/common/sim-memopt.c
> > >
> > > +/* Convenience function for use when debugging the simulator.  */
> > 
> > can you clarify that it's meant for an attached debugger to call vs a
> > manual call from within the sim itself ?  i know your commit message
> > said that, but the code comment would help too.
> > 
> > i vaguely recall seeing another such hook point in the
> > tree.
> 
> That *could* have been in gdbinit.in a.k.a. .gdbinit.

hrm, maybe.  i want to say it was real code.  maybe in one of the ports.
at any rate, prob not worth diving deeper for now.

> Speaking of that, I just noticed there's no .gdbinit in
> sim/<any> anymore.  They aren't build from sim/common/gdbinit.in
> apparently due to some mixup of @SIM_COMMON_BUILD_FALSE@ and/or
> @SIM_COMMON_BUILD_TRUE@ which seems to have been introduced in
> your 36bb57e40c506.  (No, not planning on going there.)

thanks for pointing this out.  it was left behind because it has a configure
dependency on cgen ports and i hadn't figured out how to unwind that.  but
from playing around a bit, i think we can just always set the breakpoint.
if the symbol doesn't exist, gdb will warn, but that's it.  if you know how
to get it to not complain, i'm all ears.
-mike