[PATCH] [RFC] TPM: be silent if disabled or deactivated

[PATCH] [RFC] TPM: be silent if disabled or deactivated

[Paul Bolle]

Since v3.3 the TPM security chip on a laptop I use prints two messages
at every boot and every resume:
    tpm_tis 00:0a: A TPM error (6) occurred attempting to read a pcr value
    tpm_tis 00:0a: TPM is disabled/deactivated (0x6)

The second message is just an informational message, indicating that
this chip is deactivated (0x6 is TPM_ERR_DEACTIVATED). That doesn't
bother me. The first message is printed at KERN_ERR level. To me it
seems that it is not an error if a security chip is deactivated or
disabled. So suppress that error message in those two cases.

Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
---
0) Tested against v3.3. (This laptop is currently tracking the Fedora 16
kernel, which is now v3.3 based.) Applies cleanly to v3.4-rc2.

1) Sent as an RFC because I don't actually use any TPM functionality.
(At least, that's what I think. I'm entirely unfamiliar with TPM.)
Moreover, there are a number of code paths that hit this messages, and
for some of those this patch might hide this error where people still
would like to see it. But my usage probably doesn't trigger those code
paths.

 drivers/char/tpm/tpm.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/char/tpm/tpm.c b/drivers/char/tpm/tpm.c
index 32362cf..40a09b5 100644
--- a/drivers/char/tpm/tpm.c
+++ b/drivers/char/tpm/tpm.c
@@ -473,7 +473,7 @@ static ssize_t transmit_cmd(struct tpm_chip *chip, struct tpm_cmd_t *cmd,
 		return -EFAULT;
 
 	err = be32_to_cpu(cmd->header.out.return_code);
-	if (err != 0)
+	if (err != 0 && err != TPM_ERR_DISABLED && err != TPM_ERR_DEACTIVATED)
 		dev_err(chip->dev, "A TPM error (%d) occurred %s\n", err, desc);
 
 	return err;
-- 
1.7.7.6

Re: [PATCH] [RFC] TPM: be silent if disabled or deactivated

[Rajiv Andrade]

On Thu, 12 Apr 2012, Paul Bolle wrote:

> Since v3.3 the TPM security chip on a laptop I use prints two messages
> at every boot and every resume:
>     tpm_tis 00:0a: A TPM error (6) occurred attempting to read a pcr value
>     tpm_tis 00:0a: TPM is disabled/deactivated (0x6)
> 
> The second message is just an informational message, indicating that
> this chip is deactivated (0x6 is TPM_ERR_DEACTIVATED). That doesn't
> bother me. The first message is printed at KERN_ERR level. To me it
> seems that it is not an error if a security chip is deactivated or
> disabled. So suppress that error message in those two cases.
> 
> Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
> ---
> 0) Tested against v3.3. (This laptop is currently tracking the Fedora 16
> kernel, which is now v3.3 based.) Applies cleanly to v3.4-rc2.
> 
> 1) Sent as an RFC because I don't actually use any TPM functionality.
> (At least, that's what I think. I'm entirely unfamiliar with TPM.)
> Moreover, there are a number of code paths that hit this messages, and
> for some of those this patch might hide this error where people still
> would like to see it. But my usage probably doesn't trigger those code
> paths.
>

Thanks for pointing this out Paul. We indeed don't want this being
thrown out as an error if the TPM is in this state. Although your patch
fits well and solves this neatly, transmit_cmd(), as you mentioned, is
called by essentially all functions attempting to send a command to the
TPM. Therefore, we do want to trigger this as an error in case a faulty
hardware claims to be functional after the tpm_do_selftest(), but decides
to return this error code when already registered. I'll modify
tpm_do_selftest() to handle these two scenarios. 

Rajiv

>  drivers/char/tpm/tpm.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/drivers/char/tpm/tpm.c b/drivers/char/tpm/tpm.c
> index 32362cf..40a09b5 100644
> --- a/drivers/char/tpm/tpm.c
> +++ b/drivers/char/tpm/tpm.c
> @@ -473,7 +473,7 @@ static ssize_t transmit_cmd(struct tpm_chip *chip, struct tpm_cmd_t *cmd,
>  		return -EFAULT;
> 
>  	err = be32_to_cpu(cmd->header.out.return_code);
> -	if (err != 0)
> +	if (err != 0 && err != TPM_ERR_DISABLED && err != TPM_ERR_DEACTIVATED)
>  		dev_err(chip->dev, "A TPM error (%d) occurred %s\n", err, desc);
> 
>  	return err;
> -- 
> 1.7.7.6
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 

Re: [PATCH] [RFC] TPM: be silent if disabled or deactivated

[Paul Bolle]

On Tue, 2012-04-24 at 11:47 -0300, Rajiv Andrade wrote:
> [...], we do want to trigger this as an error in case a faulty
> hardware claims to be functional after the tpm_do_selftest(), but decides
> to return this error code when already registered. I'll modify
> tpm_do_selftest() to handle these two scenarios. 

If you think your patch needs some testing (against v3.3.x) feel free to
send something my way.

By the way, would anyone know whether a patch that suppresses unneeded
error messages qualify for stable (once it is in Linus' tree, of
course)?


Paul Bolle