Hi Marcel:
Bug1118(https://bugzilla.moblin.org/show_bug.cgi?id=1118) is a show stop issue
and need to fix right away, so I spent some time to work out a patch to fix it.
The patch now is maintained in RPM package.
I found the issue comes from:
connman inserts unmodified "identifier" to harsh table to identify the network
(see src/device.c connman_device_add_network())
However,
next time connman will used modified(using '_' substitute "none number and
letter" char) "identifier" to search the hash table, so the registered network
will be registered second time. (at src/device.c connman_device_get_network())
I find that the network->element->name is modified. Why we need to modify the
name?
I think there are two methods to fix the issue, one is to add
network->element->name to harsh table to identify the network, the other one is
use unmodified "identifier" to search harsh table. I prefer the last one. Below
is the patch.
Can you let me know you comments? :)
Further question, what if the two AP has the same essid? I think the same essid
leads to the same identify and may cause the same issue with bug 1118. What do
you think of it?
diff -Nurp connman-0.11-orig/src/device.c connman-0.11/src/device.c
--- connman-0.11-orig/src/device.c 2009-03-06 13:33:10.000000000 +0800
+++ connman-0.11/src/device.c 2009-03-06 13:34:28.000000000 +0800
@@ -1451,28 +1451,8 @@ int connman_device_add_network(struct co
struct connman_network *connman_device_get_network(struct connman_device
*device,
const char *identifier)
{
- struct connman_network *network;
- char *temp;
- unsigned int i;
-
DBG("device %p identifier %s", device, identifier);
-
- temp = g_strdup(identifier);
- if (temp == NULL)
- return NULL;
-
- for (i = 0; i < strlen(temp); i++) {
- char tmp = temp[i];
- if ((tmp < '0' || tmp > '9') && (tmp < 'A' || tmp > 'Z') &&
- (tmp < 'a' || tmp > 'z'))
- temp[i] = '_';
- }
-
- network = g_hash_table_lookup(device->networks, temp);
-
- g_free(temp);
-
- return network;
+ return g_hash_table_lookup(device->networks, identifier);
}
_______________________________________________
connman mailing list
[email protected]
https://lists.moblin.org/mailman/listinfo/connman
