On 06/20/2018 10:26 PM, James Bottomley wrote: >> Yes, it does polling: >> http://docs.libuv.org/en/v1.x/design.html#the-i-o-loop > But that's for networking. You'll be talking to the TPM RM over the > file descriptor so that follows the thread pool model in > > http://docs.libuv.org/en/v1.x/design.html#file-i-o > > This precisely describes the current file descriptor abstraction we'd > use for the TPM. That is for the file IO that doesn't support non-blocking, because there is no need for it as the operations are "fast". Operations on the TPM would fall under the io loop model. > >> Regardless of how it actually might be used, I'm happy that we agree >> on that this *is* the right thing to do. > I didn't say that. I think using a single worker thread queue is the > correct abstraction for the TPM. If there's a legacy use case for > poll(), I don't see why not since the code seems to be fairly small and > self contained, but I don't really see it as correct or necessary to do > it that way. This discussion starts to go around the circle. You don't agree, but you also don't disagree? Is this what you are saying? Thanks, -- Tadeusz
