2007/12/18 bruno randolf <[EMAIL PROTECTED]>: > On Tuesday 18 December 2007 01:24:07 Nick Kossifidis wrote: >> 2007/12/17, bruno randolf <[EMAIL PROTECTED]>: >> > hi nick! >> > >> > thanks for the explanations, that makes it a bit more transparent how you >> > figure that stuff out... >> > >> > On Monday 17 December 2007 15:02:05 Nick Kossifidis wrote: >> > > Bruno can you redo this dump with 802.11b link (CCK) ? It might be >> > > easier to spot what these registers do (and verify if i'm right about >> > > the part that disables ofdm weak signal detection). >> > >> > sure! here you go: >> > http://br1.einfach.org/ath/ani2.tgz >> > >> > - traces of a 5212 in b and g mode. it also includes traces from 5414 in >> > all modes (a, b, g). all done in adhoc mode using the current HAL >> > (0.9.30.13). >> > >> > please bear in mind that turning ANI on and off has resulted in different >> > experiences in different setups: turning it *off*, i.e. no ANI has been >> > reported to work very well and improve thruput in point-to-point links in >> > a countryside environment, and also in my lab tests using 2 nodes next to >> > each other thruput increases (11M -> 20M). >> > >> > other people report that turning ANI *off* results in bad performance in >> > urban and multihop environments, especially with omnidirectional >> > antennas. that kind of makes sense, but it's not really clear (yet?) what >> > the other factors are. in any case it would be good to have the option to >> > use ANI or not. >> > >> > bruno >> >> Dumps are the same for b and g, did you use iwpriv ath0 mode 2 to force b >> mode ? > > hi nick! > > interresting, the dumps for 5414 b/g don't differ in the relevant parts > ('5414_new_adhoc_b.txt' and '5414_new_adhoc_g.txt') but the ones for 5212 do > differ, imho ('new_adhoc_b.txt' and 'new_adhoc_g.txt'). i just re-did the > 5414 dumps, with the same results, to make sure i didn't make a mistake. of > course i used iwpriv ath0 mode 2 to force b mode. > > bruno >
Hello ppl ;-) I have some code for you to test with some initial ANI support, i 'm sure i miss something because it doesn't work after this (no rx) http://www.kernel.org/pub/linux/kernel/people/mickflemm/ath5k-ani.tar.bz2 Known bugs: a) On non-5212 hw it'll result a segfault because i don't check for chip version durring ani_reset but i do check durring attach (well most of you have 5212 chips so you won't hit this i guess). b) During attach i do a bad memory allocation ani_stats = kzalloc(sizeof(struct ath5k_ani_state), GFP_KERNEL); just change it to ani_stats = kzalloc(sizeof(struct ath5k_ani_stats), GFP_KERNEL); problem seems to be in the initial settings used, i'm working on it but the more people they check it out the faster we can make it work ;-) algo is the one described in the patent doc, not the one used on ath9k -- GPG ID: 0xD21DB2DB As you read this post global entropy rises. Have Fun ;-) Nick _______________________________________________ ath5k-devel mailing list ath5k-devel@lists.ath5k.org https://lists.ath5k.org/mailman/listinfo/ath5k-devel