On 4 November 2010 05:05, Robert P. Goldman <rpgoldman@...> wrote:
> This happens when I run inside SLIME (inside Aquamacs) and at the
> terminal. fe[nl]ix said that the Darwin version of SBCL once used
> SIGUSR1 in connection with GC, but does not do so any more.
That's correct: we currently use SIGUSR2 -- or at least that's what
the sources say. Perhaps there is #define confusion here?
Do you use SBCL you've built yourself, or a binary package?
For me. on plain old Leopard:
/usr/include/sys/signal.h:#define SIGUSR1 30 /* user defined signal 1 */
/usr/include/sys/signal.h:#define SIGUSR2 31 /* user defined signal 2 */
(list sb-unix::sigusr1 sb-unix::sigusr2) => (30 31)
Does the same hold for you? Can you double-check that the SIGUSR1 you
install your handler for is consistent with the above / value of
The only other thing to pops into mind you should check -- if you
If you leave the program otherwise unchanged, but remove the SIGUSR1
handler, does it crash with an unhandled SIGUSR1?
(That's maybe the easiest way to figure out if there's another handler
out there -- presumably coupled with another source of SIGUSR1s.)
Oh, of course: OS X has Dtrace. I haven't ever used it in anger, but I
bet it can tell you where those signals are coming from.