I have a host which uses sensorsd which periodically stops running, which
prompted me to dig into the cause, it's one or both of these two issues:
(1) memory leak on configuration reload, lines 753-758:
if (cgetstr(buf, "command", &ebuf) < 0)
ebuf = NULL;
if (ebuf != NULL) {
p->command = ebuf;
ebuf = NULL;
}
cgetstr allocates a new ebuf and the previous pointer is overwritten
without freeing the old one.
Fix: insert free(p->command) before p->command = ebuf:
if (cgetstr(buf, "command", &ebuf) < 0)
ebuf = NULL;
if (ebuf != NULL) {
free(p->command);
p->command = ebuf;
ebuf = NULL;
}
(2) unnecessary exit in check_sdlim lines 367-368 for individual
sensor failure that is handled differently from the same code in
create_sdlim; sensorsd shouldn't be aborting because of a sysctl
failure on one sensor.
if (sysctl(mib, 5, &sensor, &len, NULL, 0) == -1)
err(1, "sysctl");
vs. create_sdlim, lines 240-243:
if (sysctl(mib, 5, &sensor, &slen, NULL, 0) == -1) {
if (errno != ENOENT)
warn("sysctl");
continue;
}
Fix: make check_sdlim code equivalent to create_sdlim:
if (sysctl(mib, 5, &sensor, &len, NULL, 0) == -1) {
if (errno != ENOENT)
warn("sysctl");
continue;
}
--
Jim Lippard [email protected] http://www.discord.org/
GPG Key ID: 0x99FD5CD6
