aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorPhil Blundell <philb@gnu.org>1998-11-14 10:37:04 +0000
committerPhil Blundell <philb@gnu.org>1998-11-14 10:37:04 +0000
commit3d2c31233e3cb25a905fe11a8dd0e6c9f441a333 (patch)
treec210699181c8a475db01835bf632a0050f8d9326 /lib/econet.c
parentUpdate for snapshot (diff)
downloadnet-tools-3d2c31233e3cb25a905fe11a8dd0e6c9f441a333.tar.gz
net-tools-3d2c31233e3cb25a905fe11a8dd0e6c9f441a333.tar.bz2
net-tools-3d2c31233e3cb25a905fe11a8dd0e6c9f441a333.zip
Rework socket handling again. Rather than
issuing ioctls to a random socket and just hoping, we do our best to pick the right one for the address family in use. This should fix a bug reported for 1.47 where a command like "ifconfig eth0 broadcast 1.0.0.255" picked on the AF_APPLETALK socket and so didn't work. Quite an invasive fix but hopefully the right one. Andi, can you take a look since this is stuff you've been working on recently.
Diffstat (limited to 'lib/econet.c')
-rw-r--r--lib/econet.c2
1 files changed, 2 insertions, 0 deletions
diff --git a/lib/econet.c b/lib/econet.c
index 4ba7668..faf0246 100644
--- a/lib/econet.c
+++ b/lib/econet.c
@@ -78,6 +78,8 @@ ec_input(int type, char *bufp, struct sockaddr *sap)
struct aftype ec_aftype = {
"ec", NULL, AF_ECONET, 0,
ec_print, ec_sprint, ec_input, NULL,
+ NULL, NULL, NULL,
+ -1,
NULL
};