Re: [PATCH v2] tpm: Don't make vendor check required for probe

[Jarkko Sakkinen]

On Tue Aug 22, 2023 at 10:50 PM EEST, Jerry Snitselaar wrote:
> On Tue, Aug 22, 2023 at 05:56:03PM +0300, Jarkko Sakkinen wrote:
> > On Tue Aug 22, 2023 at 5:05 PM EEST, Mario Limonciello wrote:
> > > On 8/22/2023 08:22, Jarkko Sakkinen wrote:
> > > > On Mon Aug 21, 2023 at 5:02 PM EEST, Mario Limonciello wrote:
> > > >> The vendor check introduced by commit 554b841d4703 ("tpm: Disable RNG for
> > > >> all AMD fTPMs") doesn't work properly on a number of Intel fTPMs.  On the
> > > >> reported systems the TPM doesn't reply at bootup and returns back the
> > > >> command code. This makes the TPM fail probe.
> > > >>
> > > >> As this isn't crucial for anything but AMD fTPM and AMD fTPM works, check
> > > >> the chip vendor and if it's not AMD don't run the checks.
> > > >>
> > > >> Cc: stable@xxxxxxxxxxxxxxx
> > > >> Fixes: 554b841d4703 ("tpm: Disable RNG for all AMD fTPMs")
> > > >> Reported-by: Todd Brandt <todd.e.brandt@xxxxxxxxx>
> > > >> Reported-by: Patrick Steinhardt <ps@xxxxxx>
> > > >> Reported-by: Ronan Pigott <ronan@xxxxxx>
> > > >> Reported-by: Raymond Jay Golo <rjgolo@xxxxxxxxx>
> > > >> Closes: https://bugzilla.kernel.org/show_bug.cgi?id=217804
> > > >> Signed-off-by: Mario Limonciello <mario.limonciello@xxxxxxx>
> > > >> ---
> > > >> v1->v2:
> > > >>   * Check x86 vendor for AMD
> > > >> ---
> > > >>   drivers/char/tpm/tpm_crb.c | 7 ++++++-
> > > >>   1 file changed, 6 insertions(+), 1 deletion(-)
> > > >>
> > > >> diff --git a/drivers/char/tpm/tpm_crb.c b/drivers/char/tpm/tpm_crb.c
> > > >> index 9eb1a18590123..7faf670201ccc 100644
> > > >> --- a/drivers/char/tpm/tpm_crb.c
> > > >> +++ b/drivers/char/tpm/tpm_crb.c
> > > >> @@ -465,8 +465,12 @@ static bool crb_req_canceled(struct tpm_chip *chip, u8 status)
> > > >>   
> > > >>   static int crb_check_flags(struct tpm_chip *chip)
> > > >>   {
> > > >> +	int ret = 0;

Oops I missed this. This adds unnecessary clutter to the diff.
> > > >> +#ifdef CONFIG_X86 u32 val;
> > > >> -	int ret;
> > > >> +
> > > >> +	if (boot_cpu_data.x86_vendor != X86_VENDOR_AMD)
> > > >> +		return ret;

No reason to use variable here. Should be just "return 0". It also
documents what is going on. Now this gives impression as the "ret"
could change.

I dropped the current version from my -next. I did not notice the
change in declarations earlier, sorry.

> > > >>   
> > > >>   	ret = crb_request_locality(chip, 0);
> > > >>   	if (ret)
> > > >> @@ -481,6 +485,7 @@ static int crb_check_flags(struct tpm_chip *chip)
> > > >>   
> > > >>   release:
> > > >>   	crb_relinquish_locality(chip, 0);
> > > >> +#endif
> > > > 
> > > > Looks much better but the main question here is that is this combination
> > > > possible:
> > > > 
> > > > 1. AMD CPU
> > > > 2. Non-AMD fTPM (i.e. manufacturer property differs)
> > > > 
> > > > BR, Jarkko
> > >
> > > Yes that combination is possible.
> > >
> > > Pluton TPM uses the tpm_crb driver.
> > 
> > Then I guess we'll go with this for now. Thanks for the effort.
> > 
> > Tested-by: Jarkko Sakkinen <jarkko@xxxxxxxxxx> # QEMU + swtpm
> > Reviewed-by: Jarkko Sakkinen <jarkko@xxxxxxxxxx>
> > 
> > I'm planning to send a pull request right after this with the fix so it
> > will land to v6.6-rc1 or v6.6-rc2:
> > https://lore.kernel.org/linux-integrity/20230817201935.31399-1-jarkko@xxxxxxxxxx/
> > 
> > BR, Jarkko
>
>
> Super minor nit that isn't this patch in particular so don't hold this
> up, but it seems like the function name for the earlier attempt to
> solve this issue that mentioned amd and ftpm was a clearer description
> of what was happening than crb_check_flags.

I posted an alternative: https://lkml.org/lkml/2023/8/22/1188

BR, Jarkko