Skip to contentSkip to conversation

tpm: don't return bool from update_timeouts

RFC PATCH Superseded Jerry Snitselaar 2 participants 9 messages Jan 29, 2019 – Jan 30, 2019

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

cover tpm: don't return bool from update_timeouts

Sorry, I've been tied up tracking down some iommu issues. I just took a quick stab at this. It looks like the only failure case is the tpm read of the did/vid failing for the spi driver. My question is do we want to fail the probe in that case, spit out a message, or just continue on? Thoughts?

Regards,
Jerry

#1 lore raw

RFC PATCH 1/1 tpm: don't return bool from update_timeouts

Set tpm_chip->timeouts_adjusted directly in the update_timeouts code instead of returning bool. Return rc of tpm read instead, which can be return an error in the case of the spi driver.

---
 drivers/char/tpm/tpm1-cmd.c     | 3 +--
 drivers/char/tpm/tpm_tis_core.c | 8 ++++----
 include/linux/tpm.h             | 2 +-
 3 files changed, 6 insertions(+), 7 deletions(-)
drivers/char/tpm/tpm1-cmd.c +1 -2
index 6f306338953b..bda9a16b44f6 100644
@@ -380,8 +380,7 @@ int tpm1_get_timeouts(struct tpm_chip *chip)
 	 * of misreporting.
 	 */
 	if (chip->ops->update_timeouts)
-		chip->timeout_adjusted =
-			chip->ops->update_timeouts(chip, timeout_eff);
+		chip->ops->update_timeouts(chip, timeout_eff);
 
 	if (!chip->timeout_adjusted) {
 		/* Restore default if chip reported 0 */
drivers/char/tpm/tpm_tis_core.c +4 -4
index bf7e49cfa643..57c5b6712563 100644
@@ -521,13 +521,15 @@ static const struct tis_vendor_timeout_override vendor_timeout_overrides[] = {
 			(TIS_SHORT_TIMEOUT*1000), (TIS_SHORT_TIMEOUT*1000) } },
 };
 
-static bool tpm_tis_update_timeouts(struct tpm_chip *chip,
+static int tpm_tis_update_timeouts(struct tpm_chip *chip,
 				    unsigned long *timeout_cap)
 {
 	struct tpm_tis_data *priv = dev_get_drvdata(&chip->dev);
 	int i, rc;
 	u32 did_vid;
 
+	chip->timeout_adjusted = false;
+
 	if (chip->ops->clk_enable != NULL)
 		chip->ops->clk_enable(chip, true);
 
@@ -540,11 +542,9 @@ static bool tpm_tis_update_timeouts(struct tpm_chip *chip,
 			continue;
 		memcpy(timeout_cap, vendor_timeout_overrides[i].timeout_us,
 		       sizeof(vendor_timeout_overrides[i].timeout_us));
-		rc = true;
+		chip->timeout_adjusted = true;
 	}
 
-	rc = false;
-
 out:
 	if (chip->ops->clk_enable != NULL)
 		chip->ops->clk_enable(chip, false);
include/linux/tpm.h +1 -1
index b49a55cf775f..d18df4b6247f 100644
@@ -41,7 +41,7 @@ struct tpm_class_ops {
 	int (*send) (struct tpm_chip *chip, u8 *buf, size_t len);
 	void (*cancel) (struct tpm_chip *chip);
 	u8 (*status) (struct tpm_chip *chip);
-	bool (*update_timeouts)(struct tpm_chip *chip,
+	int (*update_timeouts)(struct tpm_chip *chip,
 				unsigned long *timeout_cap);
 	int (*go_idle)(struct tpm_chip *chip);
 	int (*cmd_ready)(struct tpm_chip *chip);
Signature · 1 line
2.20.1.98.gecbdaf0899
#2 lore raw
Quoted 5 lines · Jerry Snitselaar

Set tpm_chip->timeouts_adjusted directly in the update_timeouts code instead of returning bool. Return rc of tpm read instead, which can be return an error in the case of the spi driver.

Couple of remarks:

1. Shouldn't it then be just void?
2. Why do we even need timeout_adjusted in struct tpm_chip [1]?

[1] Only used locally in tpm1_get_timeouts().

/Jarkko

#3 lore raw
Quoted 10 lines · Jarkko Sakkinen
Quoted 5 lines · Jerry Snitselaar

Set tpm_chip->timeouts_adjusted directly in the update_timeouts code instead of returning bool. Return rc of tpm read instead, which can be return an error in the case of the spi driver.

Couple of remarks:

1. Shouldn't it then be just void?

Yes I think if we don't really care about the tpm read failing it can just be void, and print out a warning when the read fail occurs.

Quoted 1 line

2. Why do we even need timeout_adjusted in struct tpm_chip [1]?

It gets used by timeouts_show in tpm-sysfs.c as well.

Quoted 4 lines

[1] Only used locally in tpm1_get_timeouts().

/Jarkko

#4 lore raw
Quoted 8 lines · Jerry Snitselaar
Quoted 7 lines · Jarkko Sakkinen
Quoted 5 lines · Jerry Snitselaar

Set tpm_chip->timeouts_adjusted directly in the update_timeouts code instead of returning bool. Return rc of tpm read instead, which can be return an error in the case of the spi driver.

Quoted 7 lines
Quoted 3 lines

Couple of remarks:

1. Shouldn't it then be just void?

Yes I think if we don't really care about the tpm read failing it can just be void, and print out a warning when the read fail occurs.

I don't think we care.

Quoted 3 lines
Quoted 1 line

2. Why do we even need timeout_adjusted in struct tpm_chip [1]?

It gets used by timeouts_show in tpm-sysfs.c as well.

Ugh, dejavu :-) I've made this same false remark at least once before.

/Jarkko

#5 lore raw
In reply to Jerry Snitselaar
Quoted 8 lines · Jerry Snitselaar

Sorry, I've been tied up tracking down some iommu issues. I just took a quick stab at this. It looks like the only failure case is the tpm read of the did/vid failing for the spi driver. My question is do we want to fail the probe in that case, spit out a message, or just continue on? Thoughts?

Regards,
Jerry

Probably printing warning would be enough. I mean do selftests and everything.

/Jarkko

#6 lore raw

RFC PATCH v2 1/1 tpm: don't return bool from update_timeouts

In reply to Jerry Snitselaar

Set tpm_chip->timeouts_adjusted directly in the update_timeouts code instead of returning bool. In case of tpm read failing print warning that the read failed and continue on.

---
v2: changed update_timeouts to void and print warning in case
of tpm read failing.

drivers/char/tpm/tpm1-cmd.c | 3 +--

 drivers/char/tpm/tpm_tis_core.c | 14 ++++++++------
 include/linux/tpm.h             |  2 +-
 3 files changed, 10 insertions(+), 9 deletions(-)
drivers/char/tpm/tpm1-cmd.c +1 -2
index 6f306338953b..bda9a16b44f6 100644
@@ -380,8 +380,7 @@ int tpm1_get_timeouts(struct tpm_chip *chip)
 	 * of misreporting.
 	 */
 	if (chip->ops->update_timeouts)
-		chip->timeout_adjusted =
-			chip->ops->update_timeouts(chip, timeout_eff);
+		chip->ops->update_timeouts(chip, timeout_eff);
 
 	if (!chip->timeout_adjusted) {
 		/* Restore default if chip reported 0 */
drivers/char/tpm/tpm_tis_core.c +8 -6
index bf7e49cfa643..aec57bd25dd2 100644
@@ -521,35 +521,37 @@ static const struct tis_vendor_timeout_override vendor_timeout_overrides[] = {
 			(TIS_SHORT_TIMEOUT*1000), (TIS_SHORT_TIMEOUT*1000) } },
 };
 
-static bool tpm_tis_update_timeouts(struct tpm_chip *chip,
+static void tpm_tis_update_timeouts(struct tpm_chip *chip,
 				    unsigned long *timeout_cap)
 {
 	struct tpm_tis_data *priv = dev_get_drvdata(&chip->dev);
 	int i, rc;
 	u32 did_vid;
 
+	chip->timeout_adjusted = false;
+
 	if (chip->ops->clk_enable != NULL)
 		chip->ops->clk_enable(chip, true);
 
 	rc = tpm_tis_read32(priv, TPM_DID_VID(0), &did_vid);
-	if (rc < 0)
+	if (rc < 0) {
+		dev_warn(&chip->dev, "%s: failed to read did_vid: %d\n", __func__, rc);
 		goto out;
+	}
 
 	for (i = 0; i != ARRAY_SIZE(vendor_timeout_overrides); i++) {
 		if (vendor_timeout_overrides[i].did_vid != did_vid)
 			continue;
 		memcpy(timeout_cap, vendor_timeout_overrides[i].timeout_us,
 		       sizeof(vendor_timeout_overrides[i].timeout_us));
-		rc = true;
+		chip->timeout_adjusted = true;
 	}
 
-	rc = false;
-
 out:
 	if (chip->ops->clk_enable != NULL)
 		chip->ops->clk_enable(chip, false);
 
-	return rc;
+	return;
 }
 
 /*
include/linux/tpm.h +1 -1
index b49a55cf775f..13563b8c0c3a 100644
@@ -41,7 +41,7 @@ struct tpm_class_ops {
 	int (*send) (struct tpm_chip *chip, u8 *buf, size_t len);
 	void (*cancel) (struct tpm_chip *chip);
 	u8 (*status) (struct tpm_chip *chip);
-	bool (*update_timeouts)(struct tpm_chip *chip,
+	void (*update_timeouts)(struct tpm_chip *chip,
 				unsigned long *timeout_cap);
 	int (*go_idle)(struct tpm_chip *chip);
 	int (*cmd_ready)(struct tpm_chip *chip);
Signature · 1 line
2.20.1.98.gecbdaf0899
#8 lore raw
Quoted 8 lines · Jarkko Sakkinen
Quoted 5 lines · Jerry Snitselaar

Set tpm_chip->timeouts_adjusted directly in the update_timeouts code instead of returning bool. In case of tpm read failing print warning that the read failed and continue on.

Hey, can you-re-send this without RFC and the earlier that I already applied to my tree, with CC to linux-kernel and linux-security-module?

Keep my reviewed-by in both.

Otherwise, I cannot include these into my next PR.

/Jarkko