Matthew Garrett <mj...@coreos.com> wrote on 08/08/2016 03:43:57 PM:
>
> On Fri, Aug 5, 2016 at 8:56 PM, Stefan Berger <stef...@us.ibm.com>
wrote:
> > Matthew Garrett <mj...@coreos.com> wrote on 08/05/2016 07:17:12 PM:
> >> This version of the implementation depends on port io, but if there's
> >> interest I'll add mmio as well.
> >
> > Port io is x86 specific, right? I don't think it should stay an x86
specific
> > device.
>
> Not strictly, there are other architectures that implement it. But
> yes, most other platforms will want MMIO support. The intent is
> certainly to add that.
>
> >> - if (misc.tpm_version != TPM_VERSION_UNSPEC) {
> >> + if (misc.tpm_version != TPM_VERSION_UNSPEC ||
> >> misc.measurements_io_base) {
> >> acpi_add_table(table_offsets, tables_blob);
> >> build_tpm_tcpa(tables_blob, tables->linker,
tables->tcpalog);
> >>
> >
> > If this device is hitchhiking on the TPM's ACPI tables, then are you
also
> > making this device mutually exclusive with the TPM ? Of not please do
so.
>
> I'll look into the best way to do that.
>
What's the plan for the driver level in Linux? Are you going to have this
device's measurement log appear under another path than the TPM path
/sys/kernel/security/tpm0/{ascii|binary}_bios_measurements? Also, I hope
that not /dev/tpm0 will be made available. Basically I think it is a mess
to use another device's ACPI table while not being compatible at all.
Though a measurement list is necessary for the state of the PCRs to make
sense.
