On Thu, May 17, 2018 at 04:33:24PM +0000, Hefty, Sean wrote: > > > Yes, that is where I was driving too as well, but I don't know why > > > this is try and fail with no lock today. Sean? Do you recall? > > > > I don't recall off the top of my head why the lock couldn't be held > > throughout the call. Guessing, I don't think all calling threads > > could block, and part of the scheme resulted from trying to handle > > device removal. > > Overall, I think the locks were trying to protect the state against > CM callbacks, not a kernel caller invoking functions willy-nilly. > It might be easier to focus the fix in the ucma and serialize calls > into the cma from there. Interesting, yes.. That would be simpler. So the rule for cma is that the caller must serialize all calls? I wonder if ULPs did that right, but I can live with that. Jason -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
