[PATCH v6] tpm: add support for partial reads

[PATCH v6] tpm: add support for partial reads

[Tadeusz Struk]

Currently to read a response from the TPM device an application needs
provide big enough buffer for the whole response and read it in one go.
The application doesn't know how big the response it beforehand so it
always needs to maintain a 4K buffer and read the max (4K).
In case if the user of the TSS library doesn't provide big enough
buffer the TCTI spec says that the library should set the required
size and return TSS2_TCTI_RC_INSUFFICIENT_BUFFER error code so that the
application could allocate a bigger buffer and call receive again.
To make it possible in the TSS library, this requires being able to do
partial reads from the driver.
The library would read the 10 bytes header first to get the actual size
of the response from the header, and then read the rest of the response.

This patch adds support for partial reads, i.e. the user can read the
response in one or multiple reads, until the whole response is consumed.
The user can also read only part of the response and ignore
the rest by issuing a new write to send a new command.

Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
---
The usecase is implemented in this TSS commit:
https://github.com/tpm2-software/tpm2-tss/commit/ce982f67a67dc08e24683d30b05800648d8a264c

Changes in v6:
 - Remove needless comments in struct file_priv

Changes in v5:
 - Merge the two patches back into one.
 - Change variable 'partial_data' to a new flag 'response_read'.
 - Update the logic around partial_data to reflect the new flag.
 - Add a new variable response_length that keeps track of how much
   of the response is left.

Changes in v4:
 - Use unsigned type for response_pending as it will never be negative.
 - Rebased on top of name change data_pending to transmit_result patch.

Changes in v3:
 - Remove link to usecase implemented in TSS out of the commit message.
 - Update the conddition in tpm_common_poll() to take into account
   the partial_data also.

Changes in v2:
 - Allow writes after only partial response is consumed to maintain
   backwords compatibility.
---
 drivers/char/tpm/tpm-dev-common.c |   51 +++++++++++++++++++++++++------------
 drivers/char/tpm/tpm-dev.h        |    5 +---
 2 files changed, 37 insertions(+), 19 deletions(-)

diff --git a/drivers/char/tpm/tpm-dev-common.c b/drivers/char/tpm/tpm-dev-common.c
index 99b5133a9d05..344739223451 100644
--- a/drivers/char/tpm/tpm-dev-common.c
+++ b/drivers/char/tpm/tpm-dev-common.c
@@ -40,7 +40,7 @@ static void tpm_async_work(struct work_struct *work)
 
 	tpm_put_ops(priv->chip);
 	if (ret > 0) {
-		priv->data_pending = ret;
+		priv->response_length = ret;
 		mod_timer(&priv->user_read_timer, jiffies + (120 * HZ));
 	}
 	mutex_unlock(&priv->buffer_mutex);
@@ -63,7 +63,8 @@ static void tpm_timeout_work(struct work_struct *work)
 					      timeout_work);
 
 	mutex_lock(&priv->buffer_mutex);
-	priv->data_pending = 0;
+	priv->response_read = true;
+	priv->response_length = 0;
 	memset(priv->data_buffer, 0, sizeof(priv->data_buffer));
 	mutex_unlock(&priv->buffer_mutex);
 	wake_up_interruptible(&priv->async_wait);
@@ -74,6 +75,7 @@ void tpm_common_open(struct file *file, struct tpm_chip *chip,
 {
 	priv->chip = chip;
 	priv->space = space;
+	priv->response_read = true;
 
 	mutex_init(&priv->buffer_mutex);
 	timer_setup(&priv->user_read_timer, user_reader_timeout, 0);
@@ -90,22 +92,35 @@ ssize_t tpm_common_read(struct file *file, char __user *buf,
 	ssize_t ret_size = 0;
 	int rc;
 
-	del_singleshot_timer_sync(&priv->user_read_timer);
-	flush_work(&priv->timeout_work);
 	mutex_lock(&priv->buffer_mutex);
 
-	if (priv->data_pending) {
-		ret_size = min_t(ssize_t, size, priv->data_pending);
-		if (ret_size > 0) {
-			rc = copy_to_user(buf, priv->data_buffer, ret_size);
-			memset(priv->data_buffer, 0, priv->data_pending);
-			if (rc)
-				ret_size = -EFAULT;
+	if (priv->response_length) {
+		priv->response_read = true;
+
+		ret_size = min_t(ssize_t, size, priv->response_length);
+		if (!ret_size) {
+			priv->response_length = 0;
+			goto out;
 		}
 
-		priv->data_pending = 0;
+		rc = copy_to_user(buf, priv->data_buffer + *off, ret_size);
+		if (rc) {
+			memset(priv->data_buffer, 0, TPM_BUFSIZE);
+			priv->response_length = 0;
+			ret_size = -EFAULT;
+		} else {
+			memset(priv->data_buffer + *off, 0, ret_size);
+			priv->response_length -= ret_size;
+			*off += ret_size;
+		}
 	}
 
+out:
+	if (!priv->response_length) {
+		*off = 0;
+		del_singleshot_timer_sync(&priv->user_read_timer);
+		flush_work(&priv->timeout_work);
+	}
 	mutex_unlock(&priv->buffer_mutex);
 	return ret_size;
 }
@@ -125,7 +140,7 @@ 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->data_pending != 0 || priv->command_enqueued) {
+	if (!priv->response_read || priv->command_enqueued) {
 		ret = -EBUSY;
 		goto out;
 	}
@@ -150,6 +165,10 @@ ssize_t tpm_common_write(struct file *file, const char __user *buf,
 		goto out;
 	}
 
+	priv->response_length = 0;
+	priv->response_read = false;
+	*off = 0;
+
 	/*
 	 * If in nonblocking mode schedule an async job to send
 	 * the command return the size.
@@ -168,7 +187,7 @@ ssize_t tpm_common_write(struct file *file, const char __user *buf,
 	tpm_put_ops(priv->chip);
 
 	if (ret > 0) {
-		priv->data_pending = ret;
+		priv->response_length = ret;
 		mod_timer(&priv->user_read_timer, jiffies + (120 * HZ));
 		ret = size;
 	}
@@ -184,7 +203,7 @@ __poll_t tpm_common_poll(struct file *file, poll_table *wait)
 
 	poll_wait(file, &priv->async_wait, wait);
 
-	if (priv->data_pending)
+	if (!priv->response_read || priv->response_length)
 		mask = EPOLLIN | EPOLLRDNORM;
 	else
 		mask = EPOLLOUT | EPOLLWRNORM;
@@ -201,7 +220,7 @@ void tpm_common_release(struct file *file, struct file_priv *priv)
 	del_singleshot_timer_sync(&priv->user_read_timer);
 	flush_work(&priv->timeout_work);
 	file->private_data = NULL;
-	priv->data_pending = 0;
+	priv->response_length = 0;
 }
 
 int __init tpm_dev_common_init(void)
diff --git a/drivers/char/tpm/tpm-dev.h b/drivers/char/tpm/tpm-dev.h
index a126b575cb8c..1089fc0bb290 100644
--- a/drivers/char/tpm/tpm-dev.h
+++ b/drivers/char/tpm/tpm-dev.h
@@ -9,14 +9,13 @@ struct file_priv {
 	struct tpm_chip *chip;
 	struct tpm_space *space;
 
-	/* Holds the amount of data passed or an error code from async op */
-	ssize_t data_pending;
 	struct mutex buffer_mutex;
-
 	struct timer_list user_read_timer;      /* user needs to claim result */
 	struct work_struct timeout_work;
 	struct work_struct async_work;
 	wait_queue_head_t async_wait;
+	size_t response_length;
+	bool response_read;
 	bool command_enqueued;
 
 	u8 data_buffer[TPM_BUFSIZE];

Re: [PATCH v6] tpm: add support for partial reads

[Jarkko Sakkinen]

On Wed, Nov 21, 2018 at 06:50:37AM -0800, Tadeusz Struk wrote:
> Currently to read a response from the TPM device an application needs
> provide big enough buffer for the whole response and read it in one go.
> The application doesn't know how big the response it beforehand so it
> always needs to maintain a 4K buffer and read the max (4K).
> In case if the user of the TSS library doesn't provide big enough
> buffer the TCTI spec says that the library should set the required
> size and return TSS2_TCTI_RC_INSUFFICIENT_BUFFER error code so that the
> application could allocate a bigger buffer and call receive again.
> To make it possible in the TSS library, this requires being able to do
> partial reads from the driver.
> The library would read the 10 bytes header first to get the actual size
> of the response from the header, and then read the rest of the response.
> 
> This patch adds support for partial reads, i.e. the user can read the
> response in one or multiple reads, until the whole response is consumed.
> The user can also read only part of the response and ignore
> the rest by issuing a new write to send a new command.
> 
> Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>

Reviewed-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>

/Jarkko

Re: [PATCH v6] tpm: add support for partial reads

[Jarkko Sakkinen]

On Wed, Nov 21, 2018 at 05:18:38PM +0200, Jarkko Sakkinen wrote:
> On Wed, Nov 21, 2018 at 06:50:37AM -0800, Tadeusz Struk wrote:
> > Currently to read a response from the TPM device an application needs
> > provide big enough buffer for the whole response and read it in one go.
> > The application doesn't know how big the response it beforehand so it
> > always needs to maintain a 4K buffer and read the max (4K).
> > In case if the user of the TSS library doesn't provide big enough
> > buffer the TCTI spec says that the library should set the required
> > size and return TSS2_TCTI_RC_INSUFFICIENT_BUFFER error code so that the
> > application could allocate a bigger buffer and call receive again.
> > To make it possible in the TSS library, this requires being able to do
> > partial reads from the driver.
> > The library would read the 10 bytes header first to get the actual size
> > of the response from the header, and then read the rest of the response.
> > 
> > This patch adds support for partial reads, i.e. the user can read the
> > response in one or multiple reads, until the whole response is consumed.
> > The user can also read only part of the response and ignore
> > the rest by issuing a new write to send a new command.
> > 
> > Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
> 
> Reviewed-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>

Tested-by: Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>

Pushed.

/Jarkko