From: Amir Mizinski <redacted>

This issue could occur when the TPM does not update TPM_STS register after
a locality request (TPM_STS Initial value = 0xFF) and a TPM_STS register
read occurs in the tpm_tis_status(chip) function call.

When a call to tpm_tis_send_data() function is made after a
request_locality() call, the condition
("if ((status & TPM_STS_COMMAND_READY) == 0)") is checked. At this moment
if the status value is 0xFF, then it is considered, wrongly, in “ready”
state (by checking only one bit). However, at this moment the TPM is, in
fact, in "Idle" state and remains in "Idle" state because
"tpm_tis_ready(chip);" was not executed.
Waiting for the condition TPM_STS.tpmGo == 0, will ensure that the TPM
status register has the correct value.

Suggested-by: Benoit Houyere <redacted>
Signed-off-by: Amir Mizinski <redacted>
---
 drivers/char/tpm/tpm_tis_core.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/drivers/char/tpm/tpm_tis_core.c b/drivers/char/tpm/tpm_tis_core.c
index 4145758..d527c43 100644
--- a/drivers/char/tpm/tpm_tis_core.c
+++ b/drivers/char/tpm/tpm_tis_core.c

@@ -177,8 +177,14 @@ static int request_locality(struct tpm_chip *chip, int l)
 	} else {
 		/* wait for burstcount */
 		do {
-			if (check_locality(chip, l))
+			if (check_locality(chip, l)) {
+				if (tpm_tis_wait_for_stat(chip, TPM_STS_GO, 0,
+							  chip->timeout_c,
+							  &priv->int_queue,
+							  false) < 0)
+					return -ETIME;
 				return l;
+			}
 			tpm_msleep(TPM_TIMEOUT);
 		} while (time_before(jiffies, stop));
 	}

-- 
2.7.4