signal c/r: don't play with zero timeouts

If it_virt_expires or it_prof_expires is zero, then don't subtract {virt,prof}_ticks(t) from it. This is the same thing do_getitimer() does, and it fixes my issues with the signals c/r patchset on s390.

Signed-off-by Serge E. Hallyn <serue-r/[email protected]>

---
 checkpoint/signal.c |    8 ++++++--
 1 files changed, 6 insertions(+), 2 deletions(-)

checkpoint/signal.c +6 -2

index 960356c..15b7cdc 100644
@@ -382,13 +382,17 @@ static int checkpoint_signal(struct ckpt_ctx *ctx, struct task_struct *t)
 	}
 	h->it_real_incr = ktime_to_ns(signal->it_real_incr);
 
-	cputime = cputime_sub(signal->it_virt_expires, virt_ticks(t));
+	cputime = signal->it_virt_expires;
+	if (!cputime_eq(cputime, cputime_zero))
+		cputime = cputime_sub(signal->it_virt_expires, virt_ticks(t));
 	cputime_to_timeval(cputime, &tval);
 	h->it_virt_value = timeval_to_ns(&tval);
 	cputime_to_timeval(signal->it_virt_incr, &tval);
 	h->it_virt_incr = timeval_to_ns(&tval);
 
-	cputime = cputime_sub(signal->it_prof_expires, prof_ticks(t));
+	cputime = signal->it_prof_expires;
+	if (!cputime_eq(cputime, cputime_zero))
+		cputime = cputime_sub(signal->it_prof_expires, prof_ticks(t));
 	cputime_to_timeval(cputime, &tval);
 	h->it_prof_value = timeval_to_ns(&tval);
 	cputime_to_timeval(signal->it_prof_incr, &tval);