[openib-general] RE: [PATCH] Opensm - handling immediate error in vendor_send new

[Hal Rosenstock]

Hi again Yael,

On Wed, 2005-10-12 at 03:53, Yael Kalka wrote:
> > > I decided to remove the checking for zero in the atomic_dec after
> all,
> > > since as I mentioned before - clearing it is not a fix, and we
> will
> > > see the value in other infos in the log file.
> >
> > But there is danger is these counters wrap, right ?
> >
> There is still some danger - as you noted - the counters can wrap.
> This will happen if there is some problem in the lower layer.
> For example - if we get the same mad twice, and we allocated it
> already for another request (after getting the first answer).
> It shouldn't happen if the lower layer is functioning correctly.

I think that it's more than lower layer malfunction that can cause this
to occur.
 
-- Hal