[PATCH] tpm: Don't make log failures fatal

[PATCH] tpm: Don't make log failures fatal

[Matthew Garrett]

If a TPM is in disabled state, it's reasonable for it to have an empty
log. Bailing out of probe in this case means that the PPI interface
isn't available, so there's no way to then enable the TPM from the OS.
In general it seems reasonable to ignore log errors - they shouldn't
itnerfere with any other TPM functionality.

Signed-off-by: Matthew Garrett <mjg59@google.com>
Cc: stable@vger.kernel.org
---
 drivers/char/tpm/tpm-chip.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/drivers/char/tpm/tpm-chip.c b/drivers/char/tpm/tpm-chip.c
index 3d6d394a8661..58073836b555 100644
--- a/drivers/char/tpm/tpm-chip.c
+++ b/drivers/char/tpm/tpm-chip.c
@@ -596,9 +596,7 @@ int tpm_chip_register(struct tpm_chip *chip)
 
 	tpm_sysfs_add_device(chip);
 
-	rc = tpm_bios_log_setup(chip);
-	if (rc != 0 && rc != -ENODEV)
-		return rc;
+	tpm_bios_log_setup(chip);
 
 	tpm_add_ppi(chip);
 
-- 
2.24.1.735.g03f4e72817-goog

Re: [PATCH] tpm: Don't make log failures fatal

[Jarkko Sakkinen]

On Fri, 2019-12-13 at 14:57 -0800, Matthew Garrett wrote:
> If a TPM is in disabled state, it's reasonable for it to have an empty
> log. Bailing out of probe in this case means that the PPI interface
> isn't available, so there's no way to then enable the TPM from the OS.
> In general it seems reasonable to ignore log errors - they shouldn't
> itnerfere with any other TPM functionality.
> 
> Signed-off-by: Matthew Garrett <mjg59@google.com>
> Cc: stable@vger.kernel.org

Otherwise looks great but maybe it would make sense to change
tpm_bios_log_setup() as void as part of the change?

> ---
>  drivers/char/tpm/tpm-chip.c | 4 +---
>  1 file changed, 1 insertion(+), 3 deletions(-)
> 
> diff --git a/drivers/char/tpm/tpm-chip.c b/drivers/char/tpm/tpm-chip.c
> index 3d6d394a8661..58073836b555 100644
> --- a/drivers/char/tpm/tpm-chip.c
> +++ b/drivers/char/tpm/tpm-chip.c
> @@ -596,9 +596,7 @@ int tpm_chip_register(struct tpm_chip *chip)
>  
>  	tpm_sysfs_add_device(chip);
>  
> -	rc = tpm_bios_log_setup(chip);
> -	if (rc != 0 && rc != -ENODEV)
> -		return rc;
> +	tpm_bios_log_setup(chip);
>  
>  	tpm_add_ppi(chip);
>  

/Jarkko

Re: [PATCH] tpm: Don't make log failures fatal

[Paul Menzel]

Dear Matthew,


Am 13.12.19 um 23:57 schrieb Matthew Garrett:
> If a TPM is in disabled state, it's reasonable for it to have an empty
> log. Bailing out of probe in this case means that the PPI interface
> isn't available, so there's no way to then enable the TPM from the OS.
> In general it seems reasonable to ignore log errors - they shouldn't
> itnerfere with any other TPM functionality.

interfere

Can this be tested with QEMU somehow?

> Signed-off-by: Matthew Garrett <mjg59@google.com>
> Cc: stable@vger.kernel.org
> ---
>   drivers/char/tpm/tpm-chip.c | 4 +---
>   1 file changed, 1 insertion(+), 3 deletions(-)
> 
> diff --git a/drivers/char/tpm/tpm-chip.c b/drivers/char/tpm/tpm-chip.c
> index 3d6d394a8661..58073836b555 100644
> --- a/drivers/char/tpm/tpm-chip.c
> +++ b/drivers/char/tpm/tpm-chip.c
> @@ -596,9 +596,7 @@ int tpm_chip_register(struct tpm_chip *chip)
>   
>   	tpm_sysfs_add_device(chip);
>   
> -	rc = tpm_bios_log_setup(chip);
> -	if (rc != 0 && rc != -ENODEV)
> -		return rc;
> +	tpm_bios_log_setup(chip);
>   
>   	tpm_add_ppi(chip);

Acked-by: Paul Menzel <pmenzel@molgen.mpg.de>


Kind regards,

Paul

Re: [PATCH] tpm: Don't make log failures fatal

[Matthew Garrett]

On Tue, Dec 17, 2019 at 4:15 AM Paul Menzel <pmenzel@molgen.mpg.de> wrote:
> Am 13.12.19 um 23:57 schrieb Matthew Garrett:
> > If a TPM is in disabled state, it's reasonable for it to have an empty
> > log. Bailing out of probe in this case means that the PPI interface
> > isn't available, so there's no way to then enable the TPM from the OS.
> > In general it seems reasonable to ignore log errors - they shouldn't
> > itnerfere with any other TPM functionality.
>
> interfere
>
> Can this be tested with QEMU somehow?

I think you could hack ovmf to disable the event log, but I can't
think of a trivial test otherwise.