hostapd with wpa dont work on big endian systems

Jouni Malinen jkmaline at cc.hut.fi
Wed Sep 15 23:46:14 EDT 2004


On Thu, Sep 09, 2004 at 12:55:07PM +0200, Gunter Burchardt wrote:

> I tested hostap with wpa on a big endian system (IXP425). I think there
> is a problem in MIC calculation. Here a log of hostapd. I will try to
> find the problem.

I cannot reproduce this on my big endian test setup (MPC8260-based,
i.e., PowerPC 603e), so I do not think this is an endianness issue. I
was able to use x86 wpa_supplicant with both WPA-PSK and WPA-EAP when
hostapd was running on this big-endian board. ARM has other differences
that may have been missed somewhere.

Could you please try whether WPA-PSK works? Another verification would
be to compare hostapd debug log to supplicant log. It would be good to
verify that the session key from IEEE 802.1X/EAP matches and whether
PMK/PTK in WPA handshake get same values.

SHA-1 code has integrated test cases. I would be interested in seeing
those being run on the IXP425 board that had this issue with WPA-EAP.
You can compile this with

gcc -o sha1-test -DTEST_MAIN sha1.c

(well, gcc replaced with suitable cross-compiler, if needed)


PS.

Your debug log excerpt had truncated lines, so I was unable to verify
whether the reported PTK would have matched with the EAPOL-Key MIC.

-- 
Jouni Malinen                                            PGP id EFC895FA



More information about the HostAP mailing list