On Sun, 07 Dec 2003 23:04:48 -0500, Lewis G Rosenthal wrote:

Hi all,

 

I realize. What's happening is solely on the client side.

   

The signal strength indicator on WifiState drops to yellow and
the system locks up and/or becomes sluggish until I get WifiState closed and
then the system is responsive again.

     

That's interesting to note. With WiFiState closed, if you start a transfer, does it behave normally?

Thinking about this, I recall a note about a PRISM 2.5 hardware error on the linux developer list. Transmit/Receive is using the BAP0 (Base Access Pointer or something like this) register, and IOCTL is using BAP1. I have no real influence on this unless I'd redesign the HCF library. If an access to BAP0 and BAP1 occurs within 5 microseconds, the chip can lock up one or both of the BAP registers (AFAIR) and nothing goes from there. Since 5 microseconds is a pretty short time, but an awfully long time to let the CPU do nothing in between, I don't have a real fix for this. This could happen more often if you have lots of simultaneous transfers going on.