From: Matthew T. R. <rus...@sp...> - 2002-12-02 23:05:58
|
On Sunday, December 1, 2002, at 05:47 PM, Andrew Gallatin wrote: > If this is not a common problem caused by pilot error on my part, then > my #1 suspicion is that there is a race condition somewhere which gets > worse with a faster CPU. (ie, this box is brand new, and probably > faster than the box the driver was developed on). Perhaps a missing > IODelay() that was never noticed on older machines? I've found at least one race condition -- if you call setNetworkName (which is not gated), the network name variable is set, then the card is reconfigured with the new network name. If a timer interrupt occurs between those points, the network name variable could get reset before it takes effect. |