When writing a message out over tcp that's too large to fit in a single

write(2), make sure that we don't restart the message on the next write
out, but continue where we left of.

OK sthen@, kn@

diff --git a/usr.sbin/snmpd/snmpe.c b/usr.sbin/snmpd/snmpe.c
index 263e1a1..e0467b0 100644 (file)
--- a/usr.sbin/snmpd/snmpe.c
+++ b/usr.sbin/snmpd/snmpe.c
@@ -1,4 +1,4 @@
-/*     $OpenBSD: snmpe.c,v 1.85 2022/10/06 14:41:08 martijn Exp $      */
+/*     $OpenBSD: snmpe.c,v 1.86 2022/12/20 19:53:33 martijn Exp $      */
 
 /*
  * Copyright (c) 2007, 2008, 2012 Reyk Floeter <reyk@openbsd.org>
@@ -612,8 +612,7 @@ snmpe_writecb(int fd, short type, void *arg)
        if (type == EV_TIMEOUT)
                goto fail;
 
-       len = ber->br_wend - ber->br_wbuf;
-       ber->br_wptr = ber->br_wbuf;
+       len = ber->br_wend - ber->br_wptr;
 
        log_debug("%s: write fd %d len %zd", __func__, fd, len);
 
@@ -799,6 +798,7 @@ snmpe_response(struct snmp_message *msg)
 
        usm_finalize_digest(msg, ptr, len);
        if (msg->sm_sock_tcp) {
+               msg->sm_ber.br_wptr = msg->sm_ber.br_wbuf;
                event_del(&msg->sm_sockev);
                event_set(&msg->sm_sockev, msg->sm_sock, EV_WRITE,
                    snmpe_writecb, msg);