[PATCH v2 3/3] tpm: allocate tpm event log if missing

[Ilias Apalodimas]

On Tue, 16 Jun 2026 at 13:34, Ludwig Nussel <ludwig.nussel at siemens.com> wrote:
>
> On 6/16/26 10:40, Ilias Apalodimas wrote:
> > On Tue, 16 Jun 2026 at 11:12, Ludwig Nussel <ludwig.nussel at siemens.com> wrote:
> >>
> >> On 6/11/26 11:57, Ilias Apalodimas wrote:
> >>> On Wed, 10 Jun 2026 at 13:52, Ludwig Nussel <ludwig.nussel at siemens.com> wrote:
> >>>> [...]
> >>>>    /**
> >>>> diff --git a/include/tpm_tcg2.h b/include/tpm_tcg2.h
> >>>> index eb6afe49e77..db3f6f486e9 100644
> >>>> --- a/include/tpm_tcg2.h
> >>>> +++ b/include/tpm_tcg2.h
> >>>> @@ -162,12 +162,14 @@ struct tcg_efi_spec_id_event {
> >>>>     * @log_position:      Current entry position
> >>>>     * @log_size:          Log space available
> >>>>     * @found:             Boolean indicating if an existing log was discovered
> >>>> + * @allocated:         Boolean indicating that the log was allocated by u-boot
> >>>>     */
> >>>>    struct tcg2_event_log {
> >>>>           u8 *log;
> >>>>           u32 log_position;
> >>>>           u32 log_size;
> >>>>           bool found;
> >>>> +       bool allocated;
> >>>>    };
> >>>
> >>> I think it makes more sense to define an eventlog state with a u32 and
> >>> replace the found boolean with it. e.g
> >>>
> >>> #define ELOG_FOUND BIT(0)
> >>> #define ELOG_ALLOCATED BIT(1) etc
> >>
> >> Is there a specific reason why explicit low level bit operations would
> >> be preferred? Since the structure is only used as internal data
> >> structure of u-boot, what about declaring bits like this:
> >>
> >> +       bool found:1;
> >> +       bool allocated:1;
> >>
> >> Looks like that style is also used within U-Boot.
> >
> > With the patches I pointed you at there are a bunch of booleans before
> > this as well, so I doubt declaring bitflelds will make the struct
> > smaller in the end. Padding/alignment will probably give you the same
> > size.
> >
> > OTOH the patches introduce more states that are conceptually relevant
> > to the eventlog.
> > The updated struct looks like
> > struct tcg2_event_log {
> > void *log;
> > u32 log_size;
> > u32 log_position;
> > void *final_buffer;
> > u32 final_position;
> > u32 last_event_size;
> > bool get_event_called;
> > bool ebs_called;
> > bool truncated;
> > bool found;
> > bool allocated;
> > };
> >
> > The TCG2 code behaves different if GetEvent or ExitBootServices has
> > been called as far as the EventLog is concerned. So I think it would
> > make more sense to encode the last four booleans to a 'log_state' (or
> > something along those lines) variable.
>
> To clarify, you are asking me to change the code like this, right?
>
> - !elog.found
> + !(elog.log_state & ELOG_FOUND)
>
> - elog.found = true
> + elog.log_state |= ELOG_FOUND
>
> - elog->found = false
> + elog->log_state &= ~ELOG_FOUND
>
> If that's what's preventing the patches from getting in I will do it of
> course.

Nop that's not preventing the patches from getting in. That's a matter
of taste mostly. I was trying to see if we have cases we need to check
or modify two states at once, in which case I'd prefer the u32.
Otherwise, I don't mind.

Cheers
/Ilias

>
> cu
> Ludwig
>
> --
> Ludwig Nussel
> Siemens AG
> www.siemens.com