Hi,
I'm not sure if this is really a bug or if my fix is appropriate and it's
the first time I've ever contacted the OpenBSD project so apologies if I
make some error in my etiquette :-)
I noticed that invocations of swapctl in rc always result in a non-zero
exit code even when swap is successfully initialised from fstab. It doesn't
affect the operation of rc in normal circumstances but in my case I wanted
to examine the exit code in $? and noticed that it was always 1. Using
swapctl in the same way from the shell after the system is up gave what I
expected, 0 or success.
Looking at /usr/src/sbin/swapctl.c it appears the code tolerates an EBUSY
response from the swapctl syscall and only warns on other errors. I guess
the circumstances during rc execution cause this response and this is
known. But since the variable gotone never gets set in this case, the exit
code is always 1 even though swap was successfully initialised.
Here is the patch I created with 'diff -u' which works around this problem
for me...
--- swapctl.c Fri Oct 25 13:11:42 2013
+++ my_swapctl.c Fri Oct 25 13:08:51 2013
@@ -427,8 +427,9 @@
}
if (swapctl(SWAP_ON, spec, (int)priority) < 0) {
- if (errno != EBUSY)
+ if (errno != EBUSY) {
warn("%s", spec);
+ } else gotone = 1;
} else {
gotone = 1;
printf("%s: adding %s as swap device at priority
%d\n",
@@ -454,4 +455,3 @@
fprintf(stderr, " %s [[-l] | -s] [-k]\n", __progname);
exit(1);
}
-
Please feel free to do what you wish with it! Any guidance on the way I
should approach these kinds of issues would be welcome as there's another
little thing I spotted and a potential enhancement to dhclient which I'm
still thrashing out the detail/working on.
Best regards,
Matthew Harrodine.
