seg fault when interface does not exist

Kel Modderman kelmo
Sun Dec 17 22:13:33 PST 2006

On Monday 18 December 2006 14:53, Jouni Malinen wrote:
> On Mon, Dec 18, 2006 at 07:15:41AM +1000, Kel Modderman wrote:
> > wpa_supplicant currently crashes hard if the interface to be used is
> > non-existant (or a typo is made).
> What do you mean with "currently"? Which version are you using? I cannot
> reproduce this with either the 0.5.x or 0.6.x branch..

Debian 0.5.6-1 binary. No modification to your tarball apart from a patch that 
points to specific doc's. The following config is used (madwifi headers are 
activated via another patch):

diff -Nru wpa_supplicant.orig/.config wpa_supplicant/.config
--- wpa_supplicant.orig/.config	1970-01-01 10:00:00.000000000 +1000
+++ wpa_supplicant/.config	2006-03-14 21:53:16.364723608 +1000
@@ -0,0 +1,41 @@
+#CONFIG_DRIVER_PRISM54=y	# Prism54 support broken upstream
+#CONFIG_DRIVER_NDIS=y		# Windows only
+#CONFIG_DRIVER_HERMES=y		# Non-GPL header required
+#CONFIG_DRIVER_BROADCOM=y	# Non-GPL header required
+#CONFIG_EAP_FAST=y		# Requires OpenSSL rebuild

> > Wouldn't it be nicer if it exited gracefully (with a hint) before trying
> > to configure the non-existant interface?
> Crashing is of course not the expected behavior. The driver wrappers
> could be modified to verify that the interface exists before trying to
> use it, but at least driver_wext output is quite clear on the interface
> really not being there no matter how hard one tries to use it.. ;-)

Yeah its clear enough for me. Just don't know what the segfault issue is 
caused by. Might try a more minimal build config to see if it exists there.

Thanks, Kel.

More information about the Hostap mailing list