[patch] syslog.3 - LOG_PID cannot be disabled

[fernape]

Summary

Document this FreeBSD peculiarity.

PR: 255664
Reported by: [e-mail]

Test Plan

We say it in the code (lib/libc/gen/syslog.c):

...

(void)fprintf(fp, "%s ", LogTag == NULL ? NILVALUE : LogTag);
      /*
       * Provide the process ID regardless of whether LOG_PID has been
       * specified, as it provides valuable information. Many
       * applications tend not to use this, even though they should.
       */

...

[gbe]

OK, for the manpage side of things, but I am not sure if this is true, since the code fragment from libc doesn't point out that this option can't be disabled?

@imp @jilles what do you think?

[fernape]

In D30188#678418, @gbe wrote:

OK, for the manpage side of things, but I am not sure if this is true, since the code fragment from libc doesn't point out that this option can't be disabled?

@imp @jilles what do you think?

Here is a small repro:

#include <stdio.h>
#include <syslog.h>
#include <unistd.h>

int main(int argc, char **argv)
{
        printf("I am process %d\n", getpid());

        openlog("myprefix", LOG_PID, LOG_USER);
        syslog(LOG_WARNING, "hello world with pid");
        closelog();

        openlog("myotherprefix", 0, LOG_USER);
        syslog(LOG_WARNING, "hello world without pid");
        closelog();

        return (0);
}

$ clang -Wall main.c && ./a.out && grep "hello world" /var/log/messages
I am process 10010
May 11 17:06:37 beastie myprefix[10010]: hello world with pid
May 11 17:06:37 beastie myotherprefix[10010]: hello world without pid

In syslog.h

#define LOG_PID         0x01    /* log the pid with each message */

I will not commit anything yet :-)

[jilles]

The comment is correct. LOG_PID has no effect and the process ID is always included. This change was made along with the move from RFC 3164 to RFC 5424 log messages.

[fernape]

@gbe are we good to go?

[gbe]

In D30188#679256, @fernape wrote:

@gbe are we good to go?

Sure, this differential is definitely LGTM.

[fernape]

Thank you both!

[danfe]

Note that logger(1) manpage is also incorrect now, as it mentions -i switch used to get the PID to be logged with each line, but now is no-op.

[fernape]

In D30188#783319, @danfe wrote:

Note that logger(1) manpage is also incorrect now, as it mentions -i switch used to get the PID to be logged with each line, but now is no-op.

Thanks for reporting this. I created a new revision