[Mimedefang] MIMEDefang 2.73-BETA-2 is available

Philip Prindeville philipp_subx at redfish-solutions.com
Fri Jan 20 01:11:09 EST 2012


On 1/19/12 6:49 PM, David F. Skoll wrote:
> Right now, you can work around it by putting:
> 
>       umask(007);
> 
> near the top of your mimedefang-filter file, outside any subroutine definition.


If we do that, then the -G flag will be in force in mimedefang-filter whether it's specified or not.

Here's a patch I'm proposing: have activateSlave() change its umask() after the fork() and before the exec().

-Philip
-------------- next part --------------
--- mimedefang-2.73-BETA-2/mimedefang-multiplexor.c.orig	2011-12-21 12:35:56.000000000 -0700
+++ mimedefang-2.73-BETA-2/mimedefang-multiplexor.c	2012-01-19 18:55:38.415153722 -0700
@@ -207,6 +207,8 @@ static HistoryBucket hourly_history[NUM_
 /* Pipe written on reception of SIGCHLD */
 static int Pipe[2] = {-1, -1};
 
+static mode_t my_umask = 077;
+
 #ifndef HAVE_SIG_ATOMIC_T
 #define sig_atomic_t int
 #endif
@@ -553,8 +555,6 @@ main(int argc, char *argv[], char **env)
     char *options;
     int facility = LOG_MAIL;
 
-    mode_t my_umask = 077;
-
     EventSelector *es;
     struct sigaction act;
     struct timeval t;
@@ -2549,6 +2549,9 @@ activateSlave(Slave *s, char const *reas
 	pname = Settings.progPath;
     }
 
+    /* pass umask (less group write) to child */
+    umask(my_umask | 020);
+
 #ifdef EMBED_PERL
     if (Settings.useEmbeddedPerl) {
 	run_embedded_filter();


More information about the MIMEDefang mailing list