Re: [PATCH 2/2] reset: Add Infineon SLB9670 TPM reset driver

[Lukas Wunner <lukas@wunner.de>]

On Wed, Nov 22, 2023 at 12:33:58AM +0100, Francesco Dolcini wrote:
> On Tue, Sep 26, 2023 at 09:09:36PM +0200, Lukas Wunner wrote:
> > Normally the platform firmware is responsible for taking a Trusted
> > Platform Module out of reset on boot and storing measurements into it.
> > 
> > However if the platform firmware is incapable of doing that -- as is the
> > case on the Raspberry Pi -- then the onus is on the kernel to take the
> > TPM out of reset before trying to attach a driver to it.
> > 
> > Provide a reset driver for such platforms.
> > 
> > The Infineon SLB9670 TPM requires a specific reset sequence on its RST#
> > pin which is documented in sections 5.4 and 5.5 of the datasheet:
> 
> Are you really sure that this change is required?
> I have seen the RST# Timing diagram in the datasheet, however I wonder
> if a reset is required at all during power-up, for example.

If the RST# pin is not toggled at all upon a warm reset (reboot),
the TPM will remain in whatever state it was during the previous boot.

Also, the pin controller connected to RST# might be reset upon a reboot
(think of a SoC internal pin controller setting all its registers to 0)
and RST# might be asserted as a result.  It is then necessary to take
the TPM out of reset.


> Not to mention that I would have expected some firmware to implement
> such reset timing and I was not able to find any (I looked at
> arm/arm64), if this is really required I the driver can work at all?
> Which platform firmware implements such reset sequence?

I can't answer how a TPM is reset by firmware on arm/arm64, you'd have
to ask an FAE at ARM.  Normally I'd expect firmware in ROM do that so
all subsequently executed code which is mutable (EFI, bootloader, kernel)
can be measured.  Again, on simple platforms such as the Raspberry Pi
there's no support to reset a TPM in ROM.


> Not to mention that I was able to see the driver probe succeed in a
> similar setup to the one you are describing in the commit message
> (different board, arm64, but nothing done by the platform firmware).

Hm, is the RST# pin even connected on that board?

Thanks,

Lukas