From: Pavel Roskin <proski@gn...> - 2007-01-31 07:51:26
I have fixed most sparse warnings in all active branches (madwifi,
madwifi-old-openhal, dadwifi and dadwifi-openhal).
There are no sparse warnings reported in dadwifi and dadwifi-openhal.
The only remaining issue in madwifi and madwifi-old-openhal is
"potentially expensive pointer subtraction".
I don't see a simple non-intrusive fix to get rid of it, so I think we
can live with it. The right fix would be to pass all WEP keys by index,
not by pointer, so we can avoid subtracting the pointers. But it would
require some significant changes in net80211, so it's just not worth the
trouble, especially since this code is not going to the kernel.
Several fixes were actually bugfixes with possible effect on the users.
- kernel freezing under heavy load may be fixed
- *-openhal branches should be safe on big-endian systems
- EAPOL frame detection may be fixed in madwifi-old-openhal
- Other encryption problems may be fixed in madwifi-old-openhal
While testing, compile problems have been found and fixed:
- AMRR rate control compiles with madwifi-old-openhal
- madwifi-old-openhal compiles for Linux 2.4
Of course, it's always possible that some errors were introduced, so
testing will be appreciated. If you find known problems disappear,
please note that in Trac.
P.S. Quick testing shows that madwifi-old-openhal is working as an AP!
That's on x86_64. Testing on big-endian systems would be useful, as
they are most affected by the fixes.