On 1/5/2014 1:28 PM, Ryusuke Konishi wrote:
Oh no, I found this series has a critical issue after I applied them.
The current umount.nilfs2 (and mount.nilfs2 -o remount,nogc) uses pid
of cleanerd to shutdown it, and this series breaks the logic since the
fork of cleanerd changes its pid. That was the true reason why nofork
option is required.
We need to fix it by reverting the series or other means.
Regards,
Ryusuke Konishi
From what I've learned of writing daemons, good practice is:
* Allow an option for writing a PID file, but don't by default
(location is often distro-specific)
* To daemonize, double-fork, chdir("/"), and then redirect stdin,
stdout, stderr to /dev/null
* Allow users to request messages to syslog, since stdout is lost.
* If you want to capture the PID, have an option to print it on
stderr before redirecting to /dev/null. The caller can read a pipe to
find the final PID of the daemon.
* Allow the option to run the program in the foreground (i.e. do not
daemonize at all) for debugging or in case you want to run the daemon
from a monitoring program which wait()s on it. In this case do not
close stdout, so that debugging messages may be seen on the terminal, or
the messages can be piped to a logger.
Until now, I had thought "-n" was the "no-daemonize" option, which I
thought was strange for mount to use.
-Mike
--
To unsubscribe from this list: send the line "unsubscribe linux-nilfs" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
