[wpa_supplicant] essid with non-ascii characters

[Jouni Malinen]

On Mon, Jun 25, 2012 at 10:37:05PM +0200, LuX wrote:
> I'm using version 0.7.3-5 of wpa_supplicant (1.0 is not yet available
> in 'core' repository of Arch linux at the moment) and I have
> unsuccessfully tried to connect to a public wifi spot whose essid
> (correctly detected by 'iwlist wlan0 scan') was "Wifi Ch?teau
> d'Olonne". I haven't seen anything concerning essid with non-ascii
> characters in the change-log of 1.0 thus I guess that the problem is
> unsolved yet.

SSID is actually any arbitrary binary data and non-ASCII characters may
end up getting encoding in many different ways.. wpa_supplicant allows
you to set the ssid parameter to any value, so I don't think there is
really a problem within wpa_supplicant. That said, I would agree that
there is a problem in handling manual configuration of SSIDs in many
cases, but that is not really specific to any implementation.

> And below is the output of the corresponding wpa_supplicant command
> (which I have interrupted after a moment). As one can see, "Wifi
> Ch?teau d'Olonne" has been changed to "Wifi Ch_teau d'Olonne".

This is just for printing out debug information. The actual SSID value
is maintained internally. Plese note that '?' in your configuration file
was printed as '__' while '?' in the scan results was printed as '_'
which indicate that they used different number of octets to encode the
character.

> Question: Is that problem already known? Should I report a bug for it?

It is not a problem with wpa_supplicant implementation, but in the way
the APs can be configured using different encoding.. Some operating
systems uses Latin 1, some use UTF-8, some something else.. Unless you
copy the SSID from scan results as-is (i.e., as binary data), you can
get a mismatch here..

In theory, wpa_supplicant could be extended to do matching with multiple
different encoding combinations (i.e., guess what encoding is used in
the SSID from the AP), but I'm not sure whether this would be worth the
effort taken into account that most configuration cases nowadays copy
the SSID automatically without manual entry and with advanced use cases
may prefer to avoid inexact matches.

-- 
Jouni Malinen                                            PGP id EFC895FA