If you wonder why top(1) on your macppc claims 99% intr and 0% idle,
it's because I broke it in October 2024 (locore.S r1.66).  I changed
ci_idepth from -1 to 0 when no interrupts, but I forgot to change
CLKF_INTR to match.  The (frame)->depth in this macro is a copy of
curcpu()->ci_idepth saved by "stw %r5,20(%r1)" in INTRENTER in
locore.S.  This depth is now 1 if the clock interrupted an idle cpu,
so CLKF_INTR should check > 1, as it does on most archs.

May I get an ok?

Index: sys/arch/powerpc/include/cpu.h
===================================================================
RCS file: /cvs/src/sys/arch/powerpc/include/cpu.h,v
diff -u -p -r1.79 cpu.h
--- sys/arch/powerpc/include/cpu.h	22 Oct 2024 12:51:56 -0000	1.79
+++ sys/arch/powerpc/include/cpu.h	5 Apr 2025 00:21:05 -0000
@@ -155,7 +155,7 @@ extern struct cpu_info cpu_info[PPC_MAXP
 
 #define	CLKF_USERMODE(frame)	(((frame)->srr1 & PSL_PR) != 0)
 #define	CLKF_PC(frame)		((frame)->srr0)
-#define	CLKF_INTR(frame)	((frame)->depth != 0)
+#define	CLKF_INTR(frame)	((frame)->depth > 1)
 
 extern int ppc_cpuidle;
 extern int ppc_proc_is_64b;