On Mon, Oct 16, 2006 at 09:10:31PM -0700, Yunfeng Hou wrote:
> a sleep between each call will have the machine live longer, and finally it will hang for sure. In the real case, I wait 100ms between each call.
hm, I wouldn't sleep() for less than the max beacon time - I've seen
50..300ms so far - don't know your need but wouldn't go lower than say
300-500 ms. Whatever, you may have found a way to stress the driver enough
to shake out bugs.
On Wed, Oct 18, 2006 at 06:43:30AM -0700, Yunfeng Hou wrote:
> yes, we are trying to have madwifi working in a fast moving object, and we need to find ap and roaming with 100ms. Any idea? It is very easy to re-create the problem, the final message on the screen is something like CPU soft lock.
might be a kernel bug, not madwifi's? have you tried other (recent) k.vers?
Anyway, if you could report / post somewhere such screendump it'll surely
help others to think about it.