credential-cache--daemon: clarify "exit" action semantics

When this code was originally written, there wasn't much
thought given to the timing between a client asking for
"exit", the daemon signaling that the action is done (with
EOF), and the actual cleanup of the socket.

However, we need to care about this so that our test scripts
do not end up racy (e.g., by asking for an exit and checking
that the socket was cleaned up). The code that is already
there happens to behave very reasonably; let's add a comment
to make it clear that any changes should retain the same
behavior.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

1 files changed, 10 insertions, 1 deletions

diff --git a/credential-cache--daemon.c b/credential-cache--daemon.c
index 9365f2ce5..8a3d0b7e2 100644
--- a/credential-cache--daemon.c
+++ b/credential-cache--daemon.c
@@ -126,8 +126,17 @@ static void serve_one_client(FILE *in, FILE *out)
 			fprintf(out, "password=%s\n", e->item.password);
 		}
 	}
-	else if (!strcmp(action.buf, "exit"))
+	else if (!strcmp(action.buf, "exit")) {
+		/*
+		 * It's important that we clean up our socket first, and then
+		 * signal the client only once we have finished the cleanup.
+		 * Calling exit() directly does this, because we clean up in
+		 * our atexit() handler, and then signal the client when our
+		 * process actually ends, which closes the socket and gives
+		 * them EOF.
+		 */
 		exit(0);
+	}
 	else if (!strcmp(action.buf, "erase"))
 		remove_credential(&c);
 	else if (!strcmp(action.buf, "store")) {