Skip to contentSkip to conversation

Refuse activation of tpm_tis driver if TPM is not working correctly

PATCH In review Stefan Berger 2 participants 3 messages Oct 10, 2011 TPM device driver

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Refuse activation of tpm_tis driver if TPM is not working correctly

Refuse activation of tpm_tis driver if the TPM is found not to be working correctly. This prevents suspend/resume problems for me on Lenovo W500 and T61p.

---
 drivers/char/tpm/tpm_tis.c |   13 +++++++++++++
 1 files changed, 13 insertions(+), 0 deletions(-)
drivers/char/tpm/tpm_tis.c +13 -0
index 3f4051a..7f2761d 100644
@@ -549,6 +549,7 @@ static int tpm_tis_init(struct device *dev, resource_size_t start,
 	u32 vendor, intfcaps, intmask;
 	int rc, i, irq_s, irq_e;
 	struct tpm_chip *chip;
+	u8 digest[TPM_DIGEST_SIZE];
 
 	if (!(chip = tpm_register_hardware(dev, &tpm_tis)))
 		return -ENODEV;
@@ -616,6 +617,18 @@ static int tpm_tis_init(struct device *dev, resource_size_t start,
 	/* get the timeouts before testing for irqs */
 	tpm_get_timeouts(chip);
 
+	/* test for correctly working TPM; refuse driver if not working 
+	   to prevent suspend/resume problems */
+	rc = tpm_pcr_read(chip->dev_num, 0, digest);
+	if (rc != 0) {
+		dev_err(dev, "Could not read PCR 0. "
+			"TPM is not working correctly.\n");
+		dev_err(dev, "Was machine previously suspended without TPM "
+			"driver present?\n");
+		rc = -ENODEV;
+		goto out_err;
+	}
+
 	/* INTERRUPT Setup */
 	init_waitqueue_head(&chip->vendor.read_queue);
 	init_waitqueue_head(&chip->vendor.int_queue);
Signature · 1 line
1.7.6.4
#1 lore raw
Quoted 5 lines · Stefan Berger

Refuse activation of tpm_tis driver if the TPM is found not to be working correctly. This prevents suspend/resume problems for me on Lenovo W500 and T61p.

Testing and trying to reproduce it here.

Rajiv

Quoted 36 lines
---
  drivers/char/tpm/tpm_tis.c |   13 +++++++++++++
  1 files changed, 13 insertions(+), 0 deletions(-)
drivers/char/tpm/tpm_tis.c +13 -0
index 3f4051a..7f2761d 100644
@@ -549,6 +549,7 @@ static int tpm_tis_init(struct device *dev, resource_size_t start,
  	u32 vendor, intfcaps, intmask;
  	int rc, i, irq_s, irq_e;
  	struct tpm_chip *chip;
+	u8 digest[TPM_DIGEST_SIZE];
 
  	if (!(chip = tpm_register_hardware(dev,&tpm_tis)))
  		return -ENODEV;
@@ -616,6 +617,18 @@ static int tpm_tis_init(struct device *dev, resource_size_t start,
  	/* get the timeouts before testing for irqs */
  	tpm_get_timeouts(chip);
 
+	/* test for correctly working TPM; refuse driver if not working
+	   to prevent suspend/resume problems */
+	rc = tpm_pcr_read(chip->dev_num, 0, digest);
+	if (rc != 0) {
+		dev_err(dev, "Could not read PCR 0. "
+			"TPM is not working correctly.\n");
+		dev_err(dev, "Was machine previously suspended without TPM "
+			"driver present?\n");
+		rc = -ENODEV;
+		goto out_err;
+	}
+
  	/* INTERRUPT Setup */
  	init_waitqueue_head(&chip->vendor.read_queue);
  	init_waitqueue_head(&chip->vendor.int_queue);
#2 lore raw
Quoted 7 lines · Rajiv Andrade
Quoted 5 lines · Stefan Berger

Refuse activation of tpm_tis driver if the TPM is found not to be working correctly. This prevents suspend/resume problems for me on Lenovo W500 and T61p.

Testing and trying to reproduce it here.

Yes, for some yet unknown reason the TPM doesn't respond after a suspend/resume in case it wasn't loaded before that. With this change/check, suspends work fine in this scenario for me as well. Applied to:

github.com/srajiv/tpm.git next-tpm-fixes

Thanks for the fix Stefan.
Rajiv