[PATCH 0/4] mailbox: A few fixes due for the v4.5-rc's

[PATCH 0/4] mailbox: A few fixes due for the v4.5-rc's

[Lee Jones]

Hi Jassi,

I assume you already have your pull-request set for the v4.5
merge window, so please take this and add them to your -fixes
branch.

Kind regards,
Lee

Lee Jones (4):
  mailbox: mailbox-test: Use more consistent format for calling
    copy_from_user()
  mailbox: mailbox-test: Remove unused variable
  mailbox: mailbox-test: Prevent memory leak
  mailbox: Stop using ENOSYS for anything other than unimplemented
    syscalls

 drivers/mailbox/mailbox-test.c | 16 +++++++++-------
 drivers/mailbox/mailbox.c      |  4 ++--
 2 files changed, 11 insertions(+), 9 deletions(-)

-- 
1.9.1

[PATCH 1/4] mailbox: mailbox-test: Use more consistent format for calling copy_from_user()

[Lee Jones]

While we're at it, ensure copy-to location is NULL'ed in the error path.

Suggested-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
---
 drivers/mailbox/mailbox-test.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/mailbox/mailbox-test.c b/drivers/mailbox/mailbox-test.c
index 684ae17..7d9b915 100644
--- a/drivers/mailbox/mailbox-test.c
+++ b/drivers/mailbox/mailbox-test.c
@@ -63,9 +63,9 @@ static ssize_t mbox_test_signal_write(struct file *filp,
 	if (!tdev->signal)
 		return -ENOMEM;
 
-	ret = copy_from_user(tdev->signal, userbuf, count);
-	if (ret) {
+	if (copy_from_user(tdev->signal, userbuf, count)) {
 		kfree(tdev->signal);
+		tdev->signal = NULL;
 		return -EFAULT;
 	}
 
-- 
1.9.1

[PATCH 2/4] mailbox: mailbox-test: Remove unused variable

[Lee Jones]

Even before it was unused, it always returned zero anyway.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
---
 drivers/mailbox/mailbox-test.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/mailbox/mailbox-test.c b/drivers/mailbox/mailbox-test.c
index 7d9b915..e370aa6 100644
--- a/drivers/mailbox/mailbox-test.c
+++ b/drivers/mailbox/mailbox-test.c
@@ -45,7 +45,6 @@ static ssize_t mbox_test_signal_write(struct file *filp,
 				       size_t count, loff_t *ppos)
 {
 	struct mbox_test_device *tdev = filp->private_data;
-	int ret;
 
 	if (!tdev->tx_channel) {
 		dev_err(tdev->dev, "Channel cannot do Tx\n");
@@ -69,7 +68,7 @@ static ssize_t mbox_test_signal_write(struct file *filp,
 		return -EFAULT;
 	}
 
-	return ret < 0 ? ret : count;
+	return count;
 }
 
 static const struct file_operations mbox_test_signal_ops = {
-- 
1.9.1

[PATCH 3/4] mailbox: mailbox-test: Prevent memory leak

[Lee Jones]

If we set the Signal twice or more, without using it as part of a message,
memory will be re-allocated and the pointer over-written.  Prevent this
potential leak by only allocating memory when there isn't any already.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
---
 drivers/mailbox/mailbox-test.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/drivers/mailbox/mailbox-test.c b/drivers/mailbox/mailbox-test.c
index e370aa6..539d475 100644
--- a/drivers/mailbox/mailbox-test.c
+++ b/drivers/mailbox/mailbox-test.c
@@ -58,9 +58,12 @@ static ssize_t mbox_test_signal_write(struct file *filp,
 		return -EINVAL;
 	}
 
-	tdev->signal = kzalloc(MBOX_MAX_SIG_LEN, GFP_KERNEL);
-	if (!tdev->signal)
-		return -ENOMEM;
+	/* Only allocate memory if we need to */
+	if (!tdev->signal) {
+		tdev->signal = kzalloc(MBOX_MAX_SIG_LEN, GFP_KERNEL);
+		if (!tdev->signal)
+			return -ENOMEM;
+	}
 
 	if (copy_from_user(tdev->signal, userbuf, count)) {
 		kfree(tdev->signal);
-- 
1.9.1

[PATCH 4/4] mailbox: Stop using ENOSYS for anything other than unimplemented syscalls

[Lee Jones]

In accordance with e15f431fe2d5 ("errno.h: Improve ENOSYS's comment") and
91c9afaf97ee ("checkpatch.pl: new instances of ENOSYS are errors") we're
converting from the old meaning of: ENOSYS "Function not implemented" to
a more standard EINVAL.

Reported-by: Seraphin Bonnaffe <seraphin.bonnaffe@st.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
---
 drivers/mailbox/mailbox.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/mailbox/mailbox.c b/drivers/mailbox/mailbox.c
index 6a4811f..4a36632 100644
--- a/drivers/mailbox/mailbox.c
+++ b/drivers/mailbox/mailbox.c
@@ -375,13 +375,13 @@ struct mbox_chan *mbox_request_channel_byname(struct mbox_client *cl,
 
 	if (!np) {
 		dev_err(cl->dev, "%s() currently only supports DT\n", __func__);
-		return ERR_PTR(-ENOSYS);
+		return ERR_PTR(-EINVAL);
 	}
 
 	if (!of_get_property(np, "mbox-names", NULL)) {
 		dev_err(cl->dev,
 			"%s() requires an \"mbox-names\" property\n", __func__);
-		return ERR_PTR(-ENOSYS);
+		return ERR_PTR(-EINVAL);
 	}
 
 	of_property_for_each_string(np, "mbox-names", prop, mbox_name) {
-- 
1.9.1

[PATCH 1/4] mailbox: mailbox-test: Use more consistent format for calling copy_from_user()

[Jassi Brar]

On Thu, Jan 14, 2016 at 12:41 PM, Lee Jones <lee.jones@linaro.org> wrote:
> While we're at it, ensure copy-to location is NULL'ed in the error path.
>
> Suggested-by: Dan Carpenter <dan.carpenter@oracle.com>
> Signed-off-by: Lee Jones <lee.jones@linaro.org>
> ---
>  drivers/mailbox/mailbox-test.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/mailbox/mailbox-test.c b/drivers/mailbox/mailbox-test.c
> index 684ae17..7d9b915 100644
> --- a/drivers/mailbox/mailbox-test.c
> +++ b/drivers/mailbox/mailbox-test.c
> @@ -63,9 +63,9 @@ static ssize_t mbox_test_signal_write(struct file *filp,
>         if (!tdev->signal)
>                 return -ENOMEM;
>
> -       ret = copy_from_user(tdev->signal, userbuf, count);
> -       if (ret) {
> +       if (copy_from_user(tdev->signal, userbuf, count)) {
>                 kfree(tdev->signal);
> +               tdev->signal = NULL;
>                 return -EFAULT;
>         }
>
The function is left to return based on an uninitialized variable
('ret' on stack), so please merge this and the patch 2/4.

Thanks.