[ofa-general] Re: [PATCH] IB/mthca: change token on command timeout

Roland Dreier rdreier at cisco.com
Tue Mar 27 11:57:27 PDT 2007


 > Command token is currently only updated on command
 > event. This means that on command timeout, the same token
 > will be reused for new command, which results in a mess
 > if the timed out command *is* eventually completed.

OK, I guess this is a theoretical problem -- but with our current 60
second timeout, is there really much chance of this?

And do we have much chance of recovering from a command completing
after we gave up on it, given that we probably free the mailbox
(meaning the HCA can scribble on memory that's now been reused for
something else, etc)?

I guess I could queue this for 2.6.22 since it probably doesn't hurt,
but I don't see any reason to put it in 2.6.21.

 - R.



More information about the general mailing list