[PATCH 3/3] daemon.c: make sure kill_some_child() really kills somebody
- From
Junio C Hamano <gitster@pobox.com>
- Date
- Aug 24, 2008, 20:33 UTC
- Message-ID
- <7vk5e6kvaw.fsf_-_@gitster.siamese.dyndns.org>
- In-Reply-To
- <7vwsi6kvow.fsf@gitster.siamese.dyndns.org>
We used to kill nobody if there is no existing connection from the same address the new connection we are trying to handle, and dropped the new connection. Make sure we at least kill one.
Signed-off-by: Junio C Hamano <gitster@pobox.com> --- * I am not sure about this one, but it may be more in spirit with the old behaviour that made sure at max connection limit we killed some to handle new ones.
Actually I do think this is probably a bad idea. What we really want to do is to detect an old one that is not making any progress instead. "old" we can detect by looking at its position in the queue (or we could even add an explicit timestamp to the child structure), but it is harder to measure "not making any progress", especially without going too platform specific, e.g. monitoring rusage or somesuch, which we would want to avoid.
daemon.c | 5 ++++- 1 files changed, 4 insertions(+), 1 deletions(-)
diff --git a/daemon.c b/daemon.c index 8d2755a..a0d8f65 100644 --- a/daemon.c +++ b/daemon.c @@ -641,8 +641,11 @@ static void kill_some_child(void) if (!memcmp(&blanket->address, &next->address, sizeof(next->address))) { kill(blanket->pid, SIGTERM); - break; + return; } + + /* Nobody from the same address? Kill the youngest one, then. */ + kill(firstborn->pid, SIGTERM); } static void check_dead_children(void)
-- 1.6.0.129.ge10d2