[openib-general] Re: [PATCH] fix race in mad.c

[Michael S. Tsirkin]

Quoting Sean Hefty <sean.hefty at intel.com>:
> There should be some way to fix this that doesn't involve walking a list on
> every completion.  Can't the cleanup be changed?

I guess we could set some kind of flag. Is this better?
And we still have to take a spinlock across the entire operation.

> Either move destroying the QP after the workqueue flush or transition it to
> the error state before flushing.

How would this help? As far as I know neither will flush completion events.
You'd have to destroy the CQ for this.



-- 
MST