From: Jouni Malinen (jkmaline_at_cc.hut.fi)
Date: 2002-05-02 17:13:43 UTC
On Wed, May 01, 2002 at 07:06:00PM +0300, Jouni Malinen wrote:
> I fixed couple of potential concurrency problems with
> PRISM2_USE_CMD_COMPL_INTERRUPT yesterday, but there are still some
> problems on SMP platforms (that do not show up when using busy
> waiting). I haven't been able to produce any problems on uniprocessor
> setups, but I think I will still make one release with cmd compl
> interrupts "off" by default. When the issues on SMP platforms are
> fixed (and assuming no new problems are found with the next release),
> I'll change the default setting.
Hups.. I seem to have broken something else and both cmd comp interrupt and busy waiting produce some problems with SMP hosts. Most of these are handled by automatic re-reading of rxdesc whenever it looks invalid. I think I'll release current version today and take a closer look at locking later. I'll change the default to be cmd compl interrupt already in the next release so that it will get more testing.
-- Jouni Malinen PGP id EFC895FA