commit ae0cca99d4999f67c98e04bf1f806bd1147d9721
from: Stefan Sperling <stsp@stsp.name>
date: Thu Feb 09 09:22:32 2023 UTC

make gotd parent dispatch handlers more robust during teardown

We have observed gotd exiting after "cannot find client for fd N" errors.
If this occurs then we are likely in the process of disconnecting a client
session while processing an event from a child process. Treat the above
error as non-fatal and stop processing more events from the child process.

commit - 602eda794f1d6ffa7bf52408686191499e971da3
commit + ae0cca99d4999f67c98e04bf1f806bd1147d9721
blob - e1f90c1960d3e783711dd5f2f29e41b254d35a57
blob + 51069af23874055261175459d6d4d942db02ac41
--- gotd/gotd.c
+++ gotd/gotd.c
@@ -1075,8 +1075,12 @@ gotd_dispatch_auth_child(int fd, short event, void *ar
 	int do_disconnect = 0;
 
 	client = find_client_by_proc_fd(fd);
-	if (client == NULL)
-		fatalx("cannot find client for fd %d", fd);
+	if (client == NULL) {
+		/* Can happen during process teardown. */
+		warnx("cannot find client for fd %d", fd);
+		shut = 1;
+		goto done;
+	}
 
 	if (client->auth == NULL)
 		fatalx("cannot find auth child process for fd %d", fd);
@@ -1218,8 +1222,12 @@ gotd_dispatch_client_session(int fd, short event, void
 	struct imsg imsg;
 
 	client = find_client_by_proc_fd(fd);
-	if (client == NULL)
-		fatalx("cannot find client for fd %d", fd);
+	if (client == NULL) {
+		/* Can happen during process teardown. */
+		warnx("cannot find client for fd %d", fd);
+		shut = 1;
+		goto done;
+	}
 
 	if (event & EV_READ) {
 		if ((n = imsg_read(ibuf)) == -1 && errno != EAGAIN)
@@ -1340,8 +1348,12 @@ gotd_dispatch_repo_child(int fd, short event, void *ar
 	struct imsg imsg;
 
 	client = find_client_by_proc_fd(fd);
-	if (client == NULL)
-		fatalx("cannot find client for fd %d", fd);
+	if (client == NULL) {
+		/* Can happen during process teardown. */
+		warnx("cannot find client for fd %d", fd);
+		shut = 1;
+		goto done;
+	}
 
 	if (event & EV_READ) {
 		if ((n = imsg_read(ibuf)) == -1 && errno != EAGAIN)