On Tue, Jan 26, 2021 at 08:46:48AM -0800, James Bottomley wrote: > On Tue, 2021-01-26 at 16:46 +0100, Łukasz Majczak wrote: > > Hi Jarkko, Guenter > > > > Yes, here are the logs when failure occurs - > > https://gist.github.com/semihalf-majczak-lukasz/1575461f585f1e7fb1e9366b8eceaab9 > > Look for a phrase "TPM returned invalid status" > > We've had other reports of this: > > https://lore.kernel.org/linux-integrity/ghsgagsnag.fsf@xxxxxxxxxxx/ > https://lore.kernel.org/linux-integrity/374e918c-f167-9308-2bea-ae6bc6a3d2e3@elloe.vision/ > > The problem is some TIS TPMs don't begin in the correct locality so we > have to set it. When I proposed the check, I also proposed a fix for > this problem: > > https://lore.kernel.org/linux-integrity/20201001180925.13808-5-James.Bottomley@xxxxxxxxxxxxxxxxxxxxx/ > > But it's part of a series that never went upstream. Part of the reason > was Jarkko proposed the get/put patch to fix this instead, but that > never went upstream either. We need to decide an approach and apply > one or other fixes. Can you remind me what I proposed? I remember only proposing removing interrupt code. Can you pick up just 1/5 and 2/5 from that serieis and send them as a mini series? I had one remark for 1/5, which can be found here: https://lore.kernel.org/linux-integrity/20201024120744.GA32607@xxxxxxxxxx/ I don't think there was never argument on locality changes. /Jarkko
