[ from advogato, entry #44 ]
No news is good news, I guess. Very quiet on the feedback front. Meanwhile, a compile few fixes with certain versions of wireless extensions (v11 in particular) and a few more GETs are in. Jean Tourrilhes (creator/maintainer of the wireless extension API) stepped in with some much-appreciated feedback and some clarifications. By themselves, they're not really worth releasing another build just yet.
Jean also pointed out why some people get intermittent crashes on hot-unplug with the USB driver. Basically, we have linked URBs present on unload. When they finish, their callbacks fire and hit freshly-unloaded driver code. Instant panic. Unfortunately, it's a bitter pill to swallow as TheRightWay(tm) will require a fair amount of internal hax0ring, thanks to some overly-complex driver internals. Fortunately, it only affects a very small subset of users, so I can take the time necessary to do this correctly.
Meanwhile, I'm beginning to wonder if I'm going to have to write a first-pass at the scan/profile scripts. While I have a good idea how I want it to be done, I really don't feel like working on it anytime soon. Come on, folks, contribute some time &| code..
[censored] project
The chipset-specific driver is rock-solid now. Memory leak fixed, WEP problems "correctly worked around" (It's a firmware bug), and it happily handles ping floods without batting an eye. Oh, the generic 802.11 layer now fully implements the 802.11 infrastructure state machine, handing roaming and other nifty things like that. I still need to implement "absence of beacon frame" detection, but that's not necessary for this project.
Comments