[PATCH] tpm-dev: Require response to be read only if there actually is a response

If the tpm_transmit call in tpm_common_write fails for any reason, there is
no response that could be read. Therefore, do not require the application
to issue a read call before sending further commands. This restores the
behavior from before support for partial reads was introduced.

Signed-off-by: Alexander Steffen <redacted>
---
Feel free to merge this into Tadeusz' original commit.

 drivers/char/tpm/tpm-dev-common.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/char/tpm/tpm-dev-common.c b/drivers/char/tpm/tpm-dev-common.c
index 344739223451..5eecad233ea1 100644
--- a/drivers/char/tpm/tpm-dev-common.c
+++ b/drivers/char/tpm/tpm-dev-common.c

@@ -140,7 +140,8 @@ ssize_t tpm_common_write(struct file *file, const char __user *buf,
 	 * tpm_read or a user_read_timer timeout. This also prevents split
 	 * buffered writes from blocking here.
 	 */
-	if (!priv->response_read || priv->command_enqueued) {
+	if ((!priv->response_read && priv->response_length) ||
+	    priv->command_enqueued) {
 		ret = -EBUSY;
 		goto out;
 	}

-- 
2.17.1

On Thu, Dec 06, 2018 at 06:58:01PM +0100, Alexander Steffen wrote:

If the tpm_transmit call in tpm_common_write fails for any reason, there is
no response that could be read. Therefore, do not require the application
to issue a read call before sending further commands. This restores the
behavior from before support for partial reads was introduced.

Signed-off-by: Alexander Steffen <redacted>
---
Feel free to merge this into Tadeusz' original commit.

Thank you, good catch.

My PR is a bit late becuse just came from Oregon back to Finland so
I'll follow your advice and squash this. Thanks again.

/Jarkko