diff options
| author | Jukka Rissanen <jukka.rissanen@linux.intel.com> | 2013-07-25 16:31:27 +0300 |
|---|---|---|
| committer | Zhang zhengguang <zhengguang.zhang@intel.com> | 2013-08-15 09:08:40 +0800 |
| commit | bb057ae7cf1cb028af07ee5a17127e86bcf4a12c (patch) | |
| tree | 43a8c665af40694ae08828627fbac546204e0a5c /src/technology.c | |
| parent | 4a631b073fc92bc422569d336791c842c389b092 (diff) | |
| download | connman-bb057ae7cf1cb028af07ee5a17127e86bcf4a12c.tar.gz connman-bb057ae7cf1cb028af07ee5a17127e86bcf4a12c.tar.bz2 connman-bb057ae7cf1cb028af07ee5a17127e86bcf4a12c.zip | |
gsupplicant: Only add interface when wpa_supplicant is in ready state
I noticed following scenario:
- hardblock the device
- turn on offline mode
- turn off hardblock, as offline mode is still on connman will
set softblock on
- wpa_supplicant generated these state signals in this order:
1. INACTIVE
2. DISABLED
- wifi.c cannot handle these events because the wifi pointer
is NULL at this point
- gsupplicant will then call wifi.c:interface_added(), this will cause
the scanning to start but that will fail because we are still
in softblocked state (wpa_supplicant returns error)
- wifi.c goes now to forever loop i.e., it tries to start scan
constantly. Only way to recover is to unblock wifi by turning
offline mode off.
The fix to this issue is simple, if the interface is not yet ready
then gsupplicant will not call wifi.c:interface_added() callback as
that is useless anyway.
Upstream commit id:70445458847721186622677e2810febbf2f2f454
Diffstat (limited to 'src/technology.c')
0 files changed, 0 insertions, 0 deletions