On Fri, Apr 15, 2011 at 09:23:54AM -0700, Badari Pulavarty wrote: > True. That brings up a different question - whether we are doing > enough testing on mainline QEMU :(
It seems you're clearly not doing enough testing on any qemu. Even the RHEL6 qemu has had preadv/pwritev since the first beta. > I asked Stefan to work on a QEMU patch - since my QEMU skills are > limited and Stefan is more appropriate person to deal with QEMU :) > > As mentioned earlier, we don't see performance with any temporary > solutions suggested so far (RPC table slot + ASYNC) comes anywhere > close to readv/writev support patch or linearized QEMU. So push harder for the real patch. I haven't seen a clear NAK from Trond, just that he'd like to avoid it. For extra brownie points get him release his read/write rework and integrate it with that.
