Re: [PATCH v2] tpm: Fix suspend/shutdown on some boards by preserving chip Locality

["Jarkko Sakkinen" <jarkko@kernel.org>]

On Thu Mar 14, 2024 at 6:31 PM EET, Adam Alves wrote:
> Hi Jarkko,
>
> I have an update here. I would like you to check if it makes sense
> before I submit a patch.
>
> The problem might be related to the chip itself which leaves the idle
> state whenever the locality is relinquished.

There's no real discrete chip. It is a TPM living in the firmware,
namely management engine.

So first thing to check would be to update the BIOS entirely.

> I probed the chip while operating and noted that the
> TPM_CRB_CTRL_STS_0.tpmIdle bit (located at `regs_t->ctrl_sts` on
> `crb_priv` structure in tpm_crb.c) is always cleared whenever the
> locality is relinquished.
>
> The result is that this chip never becomes idle since after going idle
> the locality is relinquished (`tpm_chip_stop` function in tpm-chip.c).

This can be verified as tpm_crb does print "goIdle timed out" if it does
not go to idle. Should be visible in klog.

BR, Jarkko