Bug#655129: [pkg-wpa-devel] Bug#655129: hostapd: EAPOL reauthentication/rekeying timeout loop when using WMM
Hi On Sunday 08 January 2012, Henry-Nicolas Tourneur wrote: Package: hostapd Severity: wishlist Dear Maintainer, After installing hostapd on a Wheezy embdedded box, I tried to get the full bw of the card I have (it's a Atheros chip, 802.11N, TL-WN861N). Apparently the HT40- or HT40+ options of hostapd are not working with WMM enabled. This result in an EAPOL reauthentication loop. It seems this happens more often with Intel chipsets (on the client side). Obviously, not having the 40Mhz bw available largely reduce the effective throughput. Anyway, the bug is described here with details: https://dev.openwrt.org/ticket/8830 And a commit in the hostapd git fixes that bug: http://hostap.epitest.fi/gitweb/gitweb.cgi?p=hostap-1.git;a=commit;h=4378fc14ebfb355705e7674bf347ea659bcd77bc So it would be really great if you could backport that commit as a patch in the Debian package. Also, it seems that wpa_supplicant needs to be patched as well in order to get this fixed. […] Thanks a lot for investigating this issue so well, looking at the patch and subsequent fix-ups in hostapd-1.git, I fear it wouldn't be a good idea to cherry-pick with to the dormant 0.7.x branch, given that the patch as-is breaks QoS and would need rather large fixes down the line. I will check this a bit more thouroughly, but tend to err on the side of caution for the 0.7 branch at the moment. At this moment, it has not been decided yet, if wheezy will ship hostapd/ wpasupplicant 0.7.x or 1.0, as moving to the 1.x branch would usually require a 1.0.0 upstream release well in advance to wheezy's intended freeze date… Regards Stefan Lippers-Hollmann signature.asc Description: This is a digitally signed message part.
Bug#655129: [pkg-wpa-devel] Bug#655129: hostapd: EAPOL reauthentication/rekeying timeout loop when using WMM
Hello, […] Thanks a lot for investigating this issue so well, looking at the patch and subsequent fix-ups in hostapd-1.git, I fear it wouldn't be a good idea to cherry-pick with to the dormant 0.7.x branch, given that the patch as-is breaks QoS and would need rather large fixes down the line. I will check this a bit more thouroughly, but tend to err on the side of caution for the 0.7 branch at the moment. Thanks for your quick and clear answer :) At this moment, it has not been decided yet, if wheezy will ship hostapd/ wpasupplicant 0.7.x or 1.0, as moving to the 1.x branch would usually require a 1.0.0 upstream release well in advance to wheezy's intended freeze date… So now if I understand correctly the situation, this bug report is a bit stuck. Either we have a 1.0 released soon enough or it looks quite complicated to get it working with 0.7.3 and, reading you, I assume we are not going to see a 0.7.4 released (on git the 0.7 branch looks indeed quite dead). Anyway, thanks again for the answer and if you have any news, I'll be happy to get the update :) Regards Stefan Lippers-Hollmann Regards Henry-Nicolas Tourneur. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#655129: [pkg-wpa-devel] Bug#655129: hostapd: EAPOL reauthentication/rekeying timeout loop when using WMM
Hi On Sunday 08 January 2012, Henry-Nicolas Tourneur wrote: [...] At this moment, it has not been decided yet, if wheezy will ship hostapd/ wpasupplicant 0.7.x or 1.0, as moving to the 1.x branch would usually require a 1.0.0 upstream release well in advance to wheezy's intended freeze date… So now if I understand correctly the situation, this bug report is a bit stuck. Either we have a 1.0 released soon enough or it looks quite Basically, yes - and we need sufficient time for stabilization and integration (e.g. the DBus API employed by network-manager) as well before wheezy freezes (so it would be ideal to get 1.0.1 or 1.0.2 (or whatever version scheme they're going to employ for the 1.x branch) into wheezy before the freeze as well - in other words now'ish. complicated to get it working with 0.7.3 and, reading you, I assume we are not going to see a 0.7.4 released (on git the 0.7 branch looks indeed quite dead). Three patches merged post the 0.7.3-release, no activity in the 0.7.x branch for over a year. I don't have further knowledge, but I'd venture to pronounce it dead. On the other hand there aren't significant patches shipped on top of 0.7.3 by the big mainstream distros either, nor has any[1] switched to the 1.x branch yet. It's also less than ideal that no upstream tarballs have been released for hostap_1_0_rc1 either [2]. Anyway, thanks again for the answer and if you have any news, I'll be happy to get the update :) Regards Stefan Lippers-Hollmann [1] OpenWrt (for both backfire and attitude adjustment) being the notable exception, but they're following hostap.git (aka 2.x) instead. [2] http://lists.shmoo.com/pipermail/hostap/2011-November/024540.html signature.asc Description: This is a digitally signed message part.