diff mbox series

[v2] tpm, tpm_tis: Workaround failed command reception on Infineon devices

Message ID Z87Y69l5_GbzlLfp@earth.li (mailing list archive)
State New
Headers show
Series [v2] tpm, tpm_tis: Workaround failed command reception on Infineon devices | expand

Commit Message

Jonathan McDowell March 10, 2025, 12:19 p.m. UTC 
From: Jonathan McDowell <noodles@meta.com>

Some Infineon devices have a issue where the status register will get
stuck with a quick REQUEST_USE / COMMAND_READY sequence. This is not
simply a matter of requiring a longer timeout; the work around is to
retry the command submission. Add appropriate logic to do this in the
send path.

This is fixed in later firmware revisions, but those are not always
available, and cannot generally be easily updated from outside a
firmware environment.

Testing has been performed with a simple repeated loop of doing a
TPM2_CC_GET_CAPABILITY for TPM_CAP_PROP_MANUFACTURER using the Go code
at:

  https://the.earth.li/~noodles/tpm-stuff/timeout-reproducer-simple.go

It can take several hours to reproduce, and several million operations.

Signed-off-by: Jonathan McDowell <noodles@meta.com>
---
v2: Rename flag to TPM_TIS_STATUS_VALID_RETRY

  drivers/char/tpm/tpm_tis_core.c | 17 ++++++++++++++---
  drivers/char/tpm/tpm_tis_core.h |  1 +
  include/linux/tpm.h             |  1 +
  3 files changed, 16 insertions(+), 3 deletions(-)

Comments

Paul Menzel March 10, 2025, 2:12 p.m. UTC | #1 
Dear Jonathan,


Some nits, should you resend. Feel free to ignore. The verb *work 
around* is spelled with a space.

Am 10.03.25 um 13:19 schrieb Jonathan McDowell:
> From: Jonathan McDowell <noodles@meta.com>
> 
> Some Infineon devices have a issue where the status register will get
> stuck with a quick REQUEST_USE / COMMAND_READY sequence. This is not
> simply a matter of requiring a longer timeout; the work around is to

The noun without. ;-)

> retry the command submission. Add appropriate logic to do this in the
> send path.

Does the workaround have downsides?

> This is fixed in later firmware revisions, but those are not always
> available, and cannot generally be easily updated from outside a
> firmware environment.

Please mention the affected revisions. Is there an errata for tis.

> Testing has been performed with a simple repeated loop of doing a
> TPM2_CC_GET_CAPABILITY for TPM_CAP_PROP_MANUFACTURER using the Go code
> at:
> 
>   https://the.earth.li/~noodles/tpm-stuff/timeout-reproducer-simple.go

Awesome. Thank you for sharing.

> It can take several hours to reproduce, and several million operations.
> 
> Signed-off-by: Jonathan McDowell <noodles@meta.com>
> ---
> v2: Rename flag to TPM_TIS_STATUS_VALID_RETRY
> 
>   drivers/char/tpm/tpm_tis_core.c | 17 ++++++++++++++---
>   drivers/char/tpm/tpm_tis_core.h |  1 +
>   include/linux/tpm.h             |  1 +
>   3 files changed, 16 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/char/tpm/tpm_tis_core.c b/drivers/char/tpm/ 
> tpm_tis_core.c
> index c969a1793184..4ab69c3e103c 100644
> --- a/drivers/char/tpm/tpm_tis_core.c
> +++ b/drivers/char/tpm/tpm_tis_core.c
> @@ -463,7 +463,10 @@ static int tpm_tis_send_data(struct tpm_chip *chip, 
> const u8 *buf, size_t len)
> 
>           if (wait_for_tpm_stat(chip, TPM_STS_VALID, chip->timeout_c,
>                       &priv->int_queue, false) < 0) {
> -            rc = -ETIME;
> +            if (test_bit(TPM_TIS_STATUS_VALID_RETRY, &priv->flags))
> +                rc = -EAGAIN;
> +            else
> +                rc = -ETIME;

I’d use a ternary operator as the same variable is assigned to.

>               goto out_err;
>           }
>           status = tpm_tis_status(chip);
> @@ -480,7 +483,10 @@ static int tpm_tis_send_data(struct tpm_chip *chip, 
> const u8 *buf, size_t len)
> 
>       if (wait_for_tpm_stat(chip, TPM_STS_VALID, chip->timeout_c,
>                   &priv->int_queue, false) < 0) {
> -        rc = -ETIME;
> +        if (test_bit(TPM_TIS_STATUS_VALID_RETRY, &priv->flags))
> +            rc = -EAGAIN;
> +        else
> +            rc = -ETIME;
>           goto out_err;
>       }
>       status = tpm_tis_status(chip);
> @@ -545,9 +551,11 @@ static int tpm_tis_send_main(struct tpm_chip *chip, 
> const u8 *buf, size_t len)
>           if (rc >= 0)
>               /* Data transfer done successfully */
>               break;
> -        else if (rc != -EIO)
> +        else if (rc != EAGAIN && rc != -EIO)
>               /* Data transfer failed, not recoverable */
>               return rc;
> +
> +        usleep_range(priv->timeout_min, priv->timeout_max);
>       }
> 
>       /* go and do it */
> @@ -1143,6 +1151,9 @@ int tpm_tis_core_init(struct device *dev, struct 
> tpm_tis_data *priv, int irq,
>           priv->timeout_max = TIS_TIMEOUT_MAX_ATML;
>       }
> 
> +    if (priv->manufacturer_id == TPM_VID_IFX)
> +        set_bit(TPM_TIS_STATUS_VALID_RETRY, &priv->flags);
> +
>       if (is_bsw()) {
>           priv->ilb_base_addr = ioremap(INTEL_LEGACY_BLK_BASE_ADDR,
>                       ILB_REMAP_SIZE);
> diff --git a/drivers/char/tpm/tpm_tis_core.h b/drivers/char/tpm/ 
> tpm_tis_core.h
> index 690ad8e9b731..970d02c337c7 100644
> --- a/drivers/char/tpm/tpm_tis_core.h
> +++ b/drivers/char/tpm/tpm_tis_core.h
> @@ -89,6 +89,7 @@ enum tpm_tis_flags {
>       TPM_TIS_INVALID_STATUS        = 1,
>       TPM_TIS_DEFAULT_CANCELLATION    = 2,
>       TPM_TIS_IRQ_TESTED        = 3,
> +    TPM_TIS_STATUS_VALID_RETRY    = 4,
>   };
> 
>   struct tpm_tis_data {
> diff --git a/include/linux/tpm.h b/include/linux/tpm.h
> index 20a40ade8030..6c3125300c00 100644
> --- a/include/linux/tpm.h
> +++ b/include/linux/tpm.h
> @@ -335,6 +335,7 @@ enum tpm2_cc_attrs {
>   #define TPM_VID_WINBOND  0x1050
>   #define TPM_VID_STM      0x104A
>   #define TPM_VID_ATML     0x1114
> +#define TPM_VID_IFX      0x15D1
> 
>   enum tpm_chip_flags {
>       TPM_CHIP_FLAG_BOOTSTRAPPED        = BIT(0),
Jarkko Sakkinen March 11, 2025, 9:46 a.m. UTC | #2 
On Mon, Mar 10, 2025 at 12:19:55PM +0000, Jonathan McDowell wrote:
> From: Jonathan McDowell <noodles@meta.com>
> 
> Some Infineon devices have a issue where the status register will get
> stuck with a quick REQUEST_USE / COMMAND_READY sequence. This is not
> simply a matter of requiring a longer timeout; the work around is to
> retry the command submission. Add appropriate logic to do this in the
> send path.
> 
> This is fixed in later firmware revisions, but those are not always
> available, and cannot generally be easily updated from outside a
> firmware environment.
> 
> Testing has been performed with a simple repeated loop of doing a
> TPM2_CC_GET_CAPABILITY for TPM_CAP_PROP_MANUFACTURER using the Go code
> at:
> 
>  https://the.earth.li/~noodles/tpm-stuff/timeout-reproducer-simple.go
> 
> It can take several hours to reproduce, and several million operations.
> 
> Signed-off-by: Jonathan McDowell <noodles@meta.com>
> ---
> v2: Rename flag to TPM_TIS_STATUS_VALID_RETRY
> 
>  drivers/char/tpm/tpm_tis_core.c | 17 ++++++++++++++---
>  drivers/char/tpm/tpm_tis_core.h |  1 +
>  include/linux/tpm.h             |  1 +
>  3 files changed, 16 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/char/tpm/tpm_tis_core.c b/drivers/char/tpm/tpm_tis_core.c
> index c969a1793184..4ab69c3e103c 100644
> --- a/drivers/char/tpm/tpm_tis_core.c
> +++ b/drivers/char/tpm/tpm_tis_core.c
> @@ -463,7 +463,10 @@ static int tpm_tis_send_data(struct tpm_chip *chip, const u8 *buf, size_t len)
>  		if (wait_for_tpm_stat(chip, TPM_STS_VALID, chip->timeout_c,
>  					&priv->int_queue, false) < 0) {
> -			rc = -ETIME;
> +			if (test_bit(TPM_TIS_STATUS_VALID_RETRY, &priv->flags))
> +				rc = -EAGAIN;
> +			else
> +				rc = -ETIME;
>  			goto out_err;
>  		}
>  		status = tpm_tis_status(chip);
> @@ -480,7 +483,10 @@ static int tpm_tis_send_data(struct tpm_chip *chip, const u8 *buf, size_t len)
>  	if (wait_for_tpm_stat(chip, TPM_STS_VALID, chip->timeout_c,
>  				&priv->int_queue, false) < 0) {
> -		rc = -ETIME;
> +		if (test_bit(TPM_TIS_STATUS_VALID_RETRY, &priv->flags))
> +			rc = -EAGAIN;
> +		else
> +			rc = -ETIME;
>  		goto out_err;
>  	}
>  	status = tpm_tis_status(chip);
> @@ -545,9 +551,11 @@ static int tpm_tis_send_main(struct tpm_chip *chip, const u8 *buf, size_t len)
>  		if (rc >= 0)
>  			/* Data transfer done successfully */
>  			break;
> -		else if (rc != -EIO)
> +		else if (rc != EAGAIN && rc != -EIO)
>  			/* Data transfer failed, not recoverable */
>  			return rc;
> +
> +		usleep_range(priv->timeout_min, priv->timeout_max);
>  	}
>  	/* go and do it */
> @@ -1143,6 +1151,9 @@ int tpm_tis_core_init(struct device *dev, struct tpm_tis_data *priv, int irq,
>  		priv->timeout_max = TIS_TIMEOUT_MAX_ATML;
>  	}
> +	if (priv->manufacturer_id == TPM_VID_IFX)
> +		set_bit(TPM_TIS_STATUS_VALID_RETRY, &priv->flags);
> +
>  	if (is_bsw()) {
>  		priv->ilb_base_addr = ioremap(INTEL_LEGACY_BLK_BASE_ADDR,
>  					ILB_REMAP_SIZE);
> diff --git a/drivers/char/tpm/tpm_tis_core.h b/drivers/char/tpm/tpm_tis_core.h
> index 690ad8e9b731..970d02c337c7 100644
> --- a/drivers/char/tpm/tpm_tis_core.h
> +++ b/drivers/char/tpm/tpm_tis_core.h
> @@ -89,6 +89,7 @@ enum tpm_tis_flags {
>  	TPM_TIS_INVALID_STATUS		= 1,
>  	TPM_TIS_DEFAULT_CANCELLATION	= 2,
>  	TPM_TIS_IRQ_TESTED		= 3,
> +	TPM_TIS_STATUS_VALID_RETRY	= 4,
>  };
>  struct tpm_tis_data {
> diff --git a/include/linux/tpm.h b/include/linux/tpm.h
> index 20a40ade8030..6c3125300c00 100644
> --- a/include/linux/tpm.h
> +++ b/include/linux/tpm.h
> @@ -335,6 +335,7 @@ enum tpm2_cc_attrs {
>  #define TPM_VID_WINBOND  0x1050
>  #define TPM_VID_STM      0x104A
>  #define TPM_VID_ATML     0x1114
> +#define TPM_VID_IFX      0x15D1
>  enum tpm_chip_flags {
>  	TPM_CHIP_FLAG_BOOTSTRAPPED		= BIT(0),
> -- 
> 2.48.1
> 

Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>

BR, Jarkko
Jonathan McDowell March 21, 2025, 4:49 p.m. UTC | #3 
Jarkko, I've realised I've somehow introduced a typo in the patch below 
that means it doesn't fire correctly; I'm not sure how this happened as 
my local copy I was testing on is definitely correct. Would you like a 
one line fix up patch, or can you manually fix it up in your tree?

This hunk:

>@@ -545,9 +551,11 @@ static int tpm_tis_send_main(struct tpm_chip *chip, const u8 *buf, size_t len)
> 		if (rc >= 0)
> 			/* Data transfer done successfully */
> 			break;
>-		else if (rc != -EIO)
>+		else if (rc != EAGAIN && rc != -EIO)
> 			/* Data transfer failed, not recoverable */
> 			return rc;
>+
>+		usleep_range(priv->timeout_min, priv->timeout_max);
> 	}
> 	/* go and do it */

should be "rc != -EAGAIN" - the "-" sign has somehow been lost.

Apologies for this, let me know what's easiest for you in terms of 
resolving it.

On Mon, Mar 10, 2025 at 12:19:55PM +0000, Jonathan McDowell wrote:
>From: Jonathan McDowell <noodles@meta.com>
>
>Some Infineon devices have a issue where the status register will get
>stuck with a quick REQUEST_USE / COMMAND_READY sequence. This is not
>simply a matter of requiring a longer timeout; the work around is to
>retry the command submission. Add appropriate logic to do this in the
>send path.
>
>This is fixed in later firmware revisions, but those are not always
>available, and cannot generally be easily updated from outside a
>firmware environment.
>
>Testing has been performed with a simple repeated loop of doing a
>TPM2_CC_GET_CAPABILITY for TPM_CAP_PROP_MANUFACTURER using the Go code
>at:
>
> https://the.earth.li/~noodles/tpm-stuff/timeout-reproducer-simple.go
>
>It can take several hours to reproduce, and several million operations.
>
>Signed-off-by: Jonathan McDowell <noodles@meta.com>
>---
>v2: Rename flag to TPM_TIS_STATUS_VALID_RETRY
>
> drivers/char/tpm/tpm_tis_core.c | 17 ++++++++++++++---
> drivers/char/tpm/tpm_tis_core.h |  1 +
> include/linux/tpm.h             |  1 +
> 3 files changed, 16 insertions(+), 3 deletions(-)
>
>diff --git a/drivers/char/tpm/tpm_tis_core.c b/drivers/char/tpm/tpm_tis_core.c
>index c969a1793184..4ab69c3e103c 100644
>--- a/drivers/char/tpm/tpm_tis_core.c
>+++ b/drivers/char/tpm/tpm_tis_core.c
>@@ -463,7 +463,10 @@ static int tpm_tis_send_data(struct tpm_chip *chip, const u8 *buf, size_t len)
> 		if (wait_for_tpm_stat(chip, TPM_STS_VALID, chip->timeout_c,
> 					&priv->int_queue, false) < 0) {
>-			rc = -ETIME;
>+			if (test_bit(TPM_TIS_STATUS_VALID_RETRY, &priv->flags))
>+				rc = -EAGAIN;
>+			else
>+				rc = -ETIME;
> 			goto out_err;
> 		}
> 		status = tpm_tis_status(chip);
>@@ -480,7 +483,10 @@ static int tpm_tis_send_data(struct tpm_chip *chip, const u8 *buf, size_t len)
> 	if (wait_for_tpm_stat(chip, TPM_STS_VALID, chip->timeout_c,
> 				&priv->int_queue, false) < 0) {
>-		rc = -ETIME;
>+		if (test_bit(TPM_TIS_STATUS_VALID_RETRY, &priv->flags))
>+			rc = -EAGAIN;
>+		else
>+			rc = -ETIME;
> 		goto out_err;
> 	}
> 	status = tpm_tis_status(chip);
>@@ -545,9 +551,11 @@ static int tpm_tis_send_main(struct tpm_chip *chip, const u8 *buf, size_t len)
> 		if (rc >= 0)
> 			/* Data transfer done successfully */
> 			break;
>-		else if (rc != -EIO)
>+		else if (rc != EAGAIN && rc != -EIO)
> 			/* Data transfer failed, not recoverable */
> 			return rc;
>+
>+		usleep_range(priv->timeout_min, priv->timeout_max);
> 	}
> 	/* go and do it */
>@@ -1143,6 +1151,9 @@ int tpm_tis_core_init(struct device *dev, struct tpm_tis_data *priv, int irq,
> 		priv->timeout_max = TIS_TIMEOUT_MAX_ATML;
> 	}
>+	if (priv->manufacturer_id == TPM_VID_IFX)
>+		set_bit(TPM_TIS_STATUS_VALID_RETRY, &priv->flags);
>+
> 	if (is_bsw()) {
> 		priv->ilb_base_addr = ioremap(INTEL_LEGACY_BLK_BASE_ADDR,
> 					ILB_REMAP_SIZE);
>diff --git a/drivers/char/tpm/tpm_tis_core.h b/drivers/char/tpm/tpm_tis_core.h
>index 690ad8e9b731..970d02c337c7 100644
>--- a/drivers/char/tpm/tpm_tis_core.h
>+++ b/drivers/char/tpm/tpm_tis_core.h
>@@ -89,6 +89,7 @@ enum tpm_tis_flags {
> 	TPM_TIS_INVALID_STATUS		= 1,
> 	TPM_TIS_DEFAULT_CANCELLATION	= 2,
> 	TPM_TIS_IRQ_TESTED		= 3,
>+	TPM_TIS_STATUS_VALID_RETRY	= 4,
> };
> struct tpm_tis_data {
>diff --git a/include/linux/tpm.h b/include/linux/tpm.h
>index 20a40ade8030..6c3125300c00 100644
>--- a/include/linux/tpm.h
>+++ b/include/linux/tpm.h
>@@ -335,6 +335,7 @@ enum tpm2_cc_attrs {
> #define TPM_VID_WINBOND  0x1050
> #define TPM_VID_STM      0x104A
> #define TPM_VID_ATML     0x1114
>+#define TPM_VID_IFX      0x15D1
> enum tpm_chip_flags {
> 	TPM_CHIP_FLAG_BOOTSTRAPPED		= BIT(0),
>-- 
>2.48.1
>
>

J.
Jarkko Sakkinen March 22, 2025, 9:10 p.m. UTC | #4 
On Fri, Mar 21, 2025 at 04:49:15PM +0000, Jonathan McDowell wrote:
> Jarkko, I've realised I've somehow introduced a typo in the patch below that
> means it doesn't fire correctly; I'm not sure how this happened as my local
> copy I was testing on is definitely correct. Would you like a one line fix
> up patch, or can you manually fix it up in your tree?
> 
> This hunk:
> 
> > @@ -545,9 +551,11 @@ static int tpm_tis_send_main(struct tpm_chip *chip, const u8 *buf, size_t len)
> > 		if (rc >= 0)
> > 			/* Data transfer done successfully */
> > 			break;
> > -		else if (rc != -EIO)
> > +		else if (rc != EAGAIN && rc != -EIO)
> > 			/* Data transfer failed, not recoverable */
> > 			return rc;
> > +
> > +		usleep_range(priv->timeout_min, priv->timeout_max);
> > 	}
> > 	/* go and do it */
> 
> should be "rc != -EAGAIN" - the "-" sign has somehow been lost.
> 
> Apologies for this, let me know what's easiest for you in terms of resolving
> it.

NP, I missed it too so we're in the same boat ;-)

I did:

$ git -P diff
diff --git a/drivers/char/tpm/tpm_tis_core.c b/drivers/char/tpm/tpm_tis_core.c
index 4ab69c3e103c..ed0d3d8449b3 100644
--- a/drivers/char/tpm/tpm_tis_core.c
+++ b/drivers/char/tpm/tpm_tis_core.c
@@ -551,7 +551,7 @@ static int tpm_tis_send_main(struct tpm_chip *chip, const u8 *buf, size_t len)
 		if (rc >= 0)
 			/* Data transfer done successfully */
 			break;
-		else if (rc != EAGAIN && rc != -EIO)
+		else if (rc != -EAGAIN && rc != -EIO)
 			/* Data transfer failed, not recoverable */
 			return rc;

Ping, if anything else.

BR, Jarkko
diff mbox series

Patch

diff --git a/drivers/char/tpm/tpm_tis_core.c b/drivers/char/tpm/tpm_tis_core.c
index c969a1793184..4ab69c3e103c 100644
--- a/drivers/char/tpm/tpm_tis_core.c
+++ b/drivers/char/tpm/tpm_tis_core.c
@@ -463,7 +463,10 @@  static int tpm_tis_send_data(struct tpm_chip *chip, const u8 *buf, size_t len)
  
  		if (wait_for_tpm_stat(chip, TPM_STS_VALID, chip->timeout_c,
  					&priv->int_queue, false) < 0) {
-			rc = -ETIME;
+			if (test_bit(TPM_TIS_STATUS_VALID_RETRY, &priv->flags))
+				rc = -EAGAIN;
+			else
+				rc = -ETIME;
  			goto out_err;
  		}
  		status = tpm_tis_status(chip);
@@ -480,7 +483,10 @@  static int tpm_tis_send_data(struct tpm_chip *chip, const u8 *buf, size_t len)
  
  	if (wait_for_tpm_stat(chip, TPM_STS_VALID, chip->timeout_c,
  				&priv->int_queue, false) < 0) {
-		rc = -ETIME;
+		if (test_bit(TPM_TIS_STATUS_VALID_RETRY, &priv->flags))
+			rc = -EAGAIN;
+		else
+			rc = -ETIME;
  		goto out_err;
  	}
  	status = tpm_tis_status(chip);
@@ -545,9 +551,11 @@  static int tpm_tis_send_main(struct tpm_chip *chip, const u8 *buf, size_t len)
  		if (rc >= 0)
  			/* Data transfer done successfully */
  			break;
-		else if (rc != -EIO)
+		else if (rc != EAGAIN && rc != -EIO)
  			/* Data transfer failed, not recoverable */
  			return rc;
+
+		usleep_range(priv->timeout_min, priv->timeout_max);
  	}
  
  	/* go and do it */
@@ -1143,6 +1151,9 @@  int tpm_tis_core_init(struct device *dev, struct tpm_tis_data *priv, int irq,
  		priv->timeout_max = TIS_TIMEOUT_MAX_ATML;
  	}
  
+	if (priv->manufacturer_id == TPM_VID_IFX)
+		set_bit(TPM_TIS_STATUS_VALID_RETRY, &priv->flags);
+
  	if (is_bsw()) {
  		priv->ilb_base_addr = ioremap(INTEL_LEGACY_BLK_BASE_ADDR,
  					ILB_REMAP_SIZE);
diff --git a/drivers/char/tpm/tpm_tis_core.h b/drivers/char/tpm/tpm_tis_core.h
index 690ad8e9b731..970d02c337c7 100644
--- a/drivers/char/tpm/tpm_tis_core.h
+++ b/drivers/char/tpm/tpm_tis_core.h
@@ -89,6 +89,7 @@  enum tpm_tis_flags {
  	TPM_TIS_INVALID_STATUS		= 1,
  	TPM_TIS_DEFAULT_CANCELLATION	= 2,
  	TPM_TIS_IRQ_TESTED		= 3,
+	TPM_TIS_STATUS_VALID_RETRY	= 4,
  };
  
  struct tpm_tis_data {
diff --git a/include/linux/tpm.h b/include/linux/tpm.h
index 20a40ade8030..6c3125300c00 100644
--- a/include/linux/tpm.h
+++ b/include/linux/tpm.h
@@ -335,6 +335,7 @@  enum tpm2_cc_attrs {
  #define TPM_VID_WINBOND  0x1050
  #define TPM_VID_STM      0x104A
  #define TPM_VID_ATML     0x1114
+#define TPM_VID_IFX      0x15D1
  
  enum tpm_chip_flags {
  	TPM_CHIP_FLAG_BOOTSTRAPPED		= BIT(0),