fix timeout of age_callout_queue

[scholarchen]

timeout->tv_sec = (secs > 3) ? 3 : secs; // aimwang: set max timeout
pselect return max timeout is 3 seconds not secs when Rt==0

[pali]

IIRC updating timeout value of pselect is Linux specific. So it would not work on other platforms correctly.

[pali]

Really manpage for pselect says:

On Linux, select() modifies timeout to reflect the amount of time not slept; most other implementations do not do this. (POSIX.1 permits either behavior.) This causes problems both when Linux code which reads timeout is ported to other operating systems, and when code is ported to Linux that reuses a struct timeval for multiple select()s in a loop without reinitializing it. Consider timeout to be undefined after select() returns.

So this patch is Linux-specific, but igmpproxy is used on more *BSD platforms.

[pali]

Really manpage for pselect says:

On Linux, select() modifies timeout to reflect the amount of time not slept; most other implementations do not do this. (POSIX.1 permits either behavior.) This causes problems both when Linux code which reads timeout is ported to other operating systems, and when code is ported to Linux that reuses a struct timeval for multiple select()s in a loop without reinitializing it. Consider timeout to be undefined after select() returns.

So this patch is Linux-specific, but igmpproxy is used on more *BSD platforms.

[zocker007]

I have found that issue too and fixed it like in this PR. On FreeBSD, this patch works like expected.

I think this is the same as in #59 . When the solution there works like expected, I think we should probably use it over this because of portability.

[pali]

I guess that instead of using undefined value timeout->tv_sec you can use (secs > 3) ? 3 : secs or call clock_gettime with CLOCK_MONOTONIC.