Hi, this patch reorganizes the hw filter in skystar2.c to avoid many similar functions. I also tried to raise the artificial limit of 0x27 simulated filters to 60 (there is a #define now). I'm not still sure about how to handle the group/mask filter, hence a couple of "#ifdef 0" in the code. I don't see any regression, and I expect some improvements because of the 0x27->60 change (but "scan" behaviour is not improved too much). This patch is not ready for CVS. I only want some feedback about it (do you experience problems?). I would like to know if there is any specific reason to mask pid numbers with 0x3fff or 0xffff (instead of 0x1fff). I converted everything to 0x1fff, otherwise there are problems with spurious high bits not cleared in the hardware registers (no packets match). The patch is not very human-readable because diff didn't do a good job, sorry. -- Roberto Ragusa r.ragusa at libero.it
Attachment:
skystar2.c_02FILTER.diff
Description: Binary data