It compiles.
The usage output appears correct.
The man page output appears properly formatted (although, someone else
will probably be the final judge of that :-) ).
This sequence produces identical files:
root@:/usr/obj/usr/src/usr.sbin/pmcstat # pmcstat -D /var/tmp/gprof-new -R /var/tmp/sample.out -e -g
CONVERSION STATISTICS:
#exec/elf 2
#samples/total 4542
root@:/usr/obj/usr/src/usr.sbin/pmcstat # pmcstat -D /var/tmp/gprof-old -R /var/tmp/sample.out -g
CONVERSION STATISTICS:
#exec/elf 2
#samples/total 4542
root@:/usr/src # cd /var/tmp
root@:/var/tmp # gprof /boot/kernel/kernel gprof-new/CLOCK.PROF/kernel.gmon > gprof-new/gprof.out
time is in ticks, not seconds
root@:/var/tmp # gprof /boot/kernel/kernel gprof-old/CLOCK.PROF/kernel.gmon > gprof-old/gprof.out
time is in ticks, not seconds
root@:/var/tmp # diff gprof-???/gprof.out
root@:/var/tmp #
Running hd(1) on the gmon files shows the expected types in the gmon
header.
Finally, I see no complaints about overflows when presenting pmcstat
with a large data set and using the -e option.