Here's an updated patch with proper headers. :)
Also, for background on the solution, see:
http://blog.netherlabs.nl/articles/2009/01/18/the-ultimate-so_linger-page-or-why-is-my-tcp-not-reliable
--
Kees Cook @debian.org
Description: it is possible for the client side of the socket to miss data
when the server uses close() immediately after the last write(). To avoid
this, shutdown down the write side of the socket and wait for the client
to close the connection (0 byte read) before closing the server side.
Author: Kees Cook <k...@debian.org>
Index: cyrus-sasl2-2.1.26.dfsg1/saslauthd/ipc_unix.c
===================================================================
--- cyrus-sasl2-2.1.26.dfsg1.orig/saslauthd/ipc_unix.c 2012-01-27 15:31:36.000000000 -0800
+++ cyrus-sasl2-2.1.26.dfsg1/saslauthd/ipc_unix.c 2015-02-07 02:09:41.432110469 -0800
@@ -217,6 +217,7 @@
int rc;
int conn_fd;
+ unsigned char dummy;
while(1) {
@@ -261,6 +262,8 @@
}
do_request(conn_fd);
+ shutdown(conn_fd, SHUT_WR);
+ while (read(conn_fd, &dummy, 1) > 0) { }
close(conn_fd);
if(flags & DETACH_TTY) {
@@ -275,6 +278,8 @@
* Normal prefork mode.
*************************************************************/
do_request(conn_fd);
+ shutdown(conn_fd, SHUT_WR);
+ while (read(conn_fd, &dummy, 1) > 0) { }
close(conn_fd);
}
