On Mon, Nov 19, 2018 at 01:17:56PM +0000, Winkler, Tomas wrote: > > On Sun, Nov 18, 2018 at 10:52:46PM +0000, Winkler, Tomas wrote: > > > This is still NACK from my side > > > > Last time you spoke about tboot intervention but I don't see why as even > > sending a single command is not atomic in the true sense of the word i.e. if > > there was a problem that would already affect the existing code and would > > essentially mean that tboot itself is broken. > So I've consulted the issue, I wasn't not correct in the description. > Tboot cannot acquire the locality, unless the host driver relinquish > it, so the issue is opposite, driver is expected to relinquish the > locality for tboot to work correctly. This is current status, other > behavior will need a different implementation on both sides. The locality is reliquished after the command sequence (load and save TPM space + command in /dev/tpmX case). Is this really such a big issue? The intention is not keep the locality reserved by the driver. /Jarkko
