On Mon, 5 Oct 2026 at 11:20, Michael Haener <[email protected]> wrote: > > Hi Ilias, > > On Mon, 5 Oct 2026 at 10:48, Ilias Apalodimas <[email protected]> > wrote: > > Do you have a real world use case where an OS wants a TPM with zero > > preboot measurements? > > Yes. On some of our i.MX8 based devices Linux uses the TPM for disk > encryption and for device certificates. Platform integrity is > established by other means, so nothing is sealed against PCRs and there > are no preboot measurements. Disabling the TPM in the DT is therefore > not an option for us, as Linux needs it.
Thanks Michael that helps. > > What I am really after is that Linux finds the TPM in the same state on > all our products, whichever SoC they use and whether or not they use > measurements, as it does on UEFI platforms where the firmware always > starts the TPM. Today that depends on the boot path in U-Boot: EFI or > MEASURED_BOOT start it, a plain booti/bootm does not, and Linux then > logs "A TPM error (256) occurred attempting the self test" and starts > it itself. > > How U-Boot does that, in probe or from an event, is less important to > me. I am happy to implement whichever you prefer for v2. Yeah, I'm not particularly glued on "let's make it on probe" either. I am fine with the event as well. I just need a proper justification for it, because initializing it during probe seems much simpler. Thanks /Ilias > > Regards, > Michael
