Say that reading the code found stage 3's two races, and how a host changes apps

The use after free in local_done and the owner's lock made after another
app's outlet used it were found reading the code, not by the check,
which passed with both before it was made to change senders (115, 017).
The usage now says changed lines change in place unless their store
moved; and 115 names the path no check takes: a module refused while
changing in place.

1 files changed, 2 insertions, 1 deletions

diff --git a/serve/serve.c b/serve/serve.c
index fc1a063..6f7220a 100644
--- a/serve/serve.c
+++ b/serve/serve.c
@@ -4289,7 +4289,8 @@ usage:
           "      NAME MODULE.pmod STORE [memory=BYTES] [work=N] [queue=N] [send=OUTPUT=URL ...] [seal=OUTPUT=KEY ...]\n"
           "      [reply=OUTPUT=INPUT ...]; its routes start /NAME/, its sends' sources are NAME.OUTPUT, and a\n"
           "      --link-port takes its inputs as NAME/INPUT. On SIGHUP the host reads FILE again: the apps of lines\n"
-          "      gone stop, those of new or changed lines start (again), from their stores; a line out says what it did\n"
+          "      gone stop, those of new lines start, those of changed lines change in place (their states kept in\n"
+          "      memory) unless their store moved; a line out says what it did\n"
           "  --app-slots N: the apps a host of apps holds at most, those started later included\n"
           "  --workers N: the threads the apps' owners take turns on (1 for one app; for apps, one a CPU at most)\n"
           "  --send OUTPUT=pure://host:port/INPUT: to a Pure node's --link-port, the way between nodes (UDP, no\n"