Created attachment 314683 [details] The ticket opened at the madwifi wiki Description of problem: The problem is related to madwifi, the ticket is here https://madwifi.org/ticket/2081. There, mrenzmann write: Well, the "original" issue as such was introduced by changes to the wireless extensions that went into 2.6.26. We have a fix for this in trunk for vanilla kernel 2.6.26 (see #2022). Fedora backported the WE-changes to their 2.6.25 kernel, causing the problem described in this ticket. It should be possible to backport the fix from trunk and make it apply to Fedora's 2.6.25 kernel. However, Fedora users struck by this issue would be required to manually apply the patch themselves; as explained above it's not easily possible to commit this patch to our repository, because it would break compatibility for all non-Fedora users. [...] Because of this problem madwifi cannot more be used on Fedora because it cannot more be compiled. ath5k is not a valid alternative because it do not work well in many installations were madwifi worked. Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
Sorry, I'm not introducing kernel churn just to please madwifi. FWIW, it looks like it wouldn't take much to hack the madwifi sources if you insist on using it. OTOH, I recommend that you use ath5k and open bugs for whatever issues you find. It (along w/ ath9k) is the future of Atheros support in the kernel.
You are probably right, John ! But now, I'm not more able to run the Atheros hardware and I will probably have to change it. Debugging the ath5k software is a too complex task for me. The problem I have encountred is not very simple.