Change from v2:
Added more description in commit message of pacthes.
Deepak Kumar Singh (2):
rpmsg: glink: Add lock to avoid race when rpmsg device is released
rpmsg: glink: Add lock to rpmsg_ctrldev_remove
drivers/rpmsg/rpmsg_char.c | 8 ++++++++
drivers/rpmsg/rpmsg_ctrl.c | 2 ++
2 files changed, 10 insertions(+)
--
2.7.4
When remote host goes down glink char device channel is freed and
associated rpdev is destroyed through rpmsg_chrdev_eptdev_destroy(),
At the same time user space apps can still try to open/poll rpmsg
char device which will result in calling rpmsg_create_ept()/rpmsg_poll().
These functions will try to reference rpdev which has already been freed
through rpmsg_chrdev_eptdev_destroy().
File operation functions and device removal function must be protected
with lock. This patch adds existing ept lock in remove function as well.
Signed-off-by: Deepak Kumar Singh <[email protected]>
---
drivers/rpmsg/rpmsg_char.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/drivers/rpmsg/rpmsg_char.c b/drivers/rpmsg/rpmsg_char.c
index 4f21891..5500dc0 100644
--- a/drivers/rpmsg/rpmsg_char.c
+++ b/drivers/rpmsg/rpmsg_char.c
@@ -75,6 +75,7 @@ int rpmsg_chrdev_eptdev_destroy(struct device *dev, void *data)
struct rpmsg_eptdev *eptdev = dev_to_eptdev(dev);
mutex_lock(&eptdev->ept_lock);
+ eptdev->rpdev = NULL;
if (eptdev->ept) {
rpmsg_destroy_ept(eptdev->ept);
eptdev->ept = NULL;
@@ -126,6 +127,11 @@ static int rpmsg_eptdev_open(struct inode *inode, struct file *filp)
return -EBUSY;
}
+ if (!eptdev->rpdev) {
+ mutex_unlock(&eptdev->ept_lock);
+ return -ENETRESET;
+ }
+
get_device(dev);
/*
@@ -277,7 +283,9 @@ static __poll_t rpmsg_eptdev_poll(struct file *filp, poll_table *wait)
if (!skb_queue_empty(&eptdev->queue))
mask |= EPOLLIN | EPOLLRDNORM;
+ mutex_lock(&eptdev->ept_lock);
mask |= rpmsg_poll(eptdev->ept, filp, wait);
+ mutex_unlock(&eptdev->ept_lock);
return mask;
}
--
2.7.4
Call to rpmsg_ctrldev_ioctl() and rpmsg_ctrldev_remove() must be synchronized.
In present code rpmsg_ctrldev_remove() is not protected with lock, therefore
new char device creation can succeed through rpmsg_ctrldev_ioctl() call. At the
same time call to rpmsg_ctrldev_remove() funtion for ctrl device removal will
free associated rpdev device. As char device creation already succeeded, user
space is free to issue open() call which maps to rpmsg_create_ept() in kernel.
rpmsg_create_ept() function tries to reference rpdev which has already been
freed through rpmsg_ctrldev_remove(). Issue is predominantly seen in aggressive
reboot tests where rpmsg_ctrldev_ioctl() and rpmsg_ctrldev_remove() can race with
each other.
Adding lock in rpmsg_ctrldev_remove() avoids any new char device creation
throught rpmsg_ctrldev_ioctl() while remove call is already in progress.
Signed-off-by: Deepak Kumar Singh <[email protected]>
---
drivers/rpmsg/rpmsg_ctrl.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/rpmsg/rpmsg_ctrl.c b/drivers/rpmsg/rpmsg_ctrl.c
index 107da70..4332538 100644
--- a/drivers/rpmsg/rpmsg_ctrl.c
+++ b/drivers/rpmsg/rpmsg_ctrl.c
@@ -194,10 +194,12 @@ static void rpmsg_ctrldev_remove(struct rpmsg_device *rpdev)
struct rpmsg_ctrldev *ctrldev = dev_get_drvdata(&rpdev->dev);
int ret;
+ mutex_lock(&ctrldev->ctrl_lock);
/* Destroy all endpoints */
ret = device_for_each_child(&ctrldev->dev, NULL, rpmsg_chrdev_eptdev_destroy);
if (ret)
dev_warn(&rpdev->dev, "failed to nuke endpoints: %d\n", ret);
+ mutex_unlock(&ctrldev->ctrl_lock);
cdev_device_del(&ctrldev->cdev, &ctrldev->dev);
put_device(&ctrldev->dev);
--
2.7.4
The commit subject should start with "rpmsg: char:"
no glink driver update...
Regards,
Arnaud
On 9/12/22 19:05, Deepak Kumar Singh wrote:
> When remote host goes down glink char device channel is freed and
> associated rpdev is destroyed through rpmsg_chrdev_eptdev_destroy(),
> At the same time user space apps can still try to open/poll rpmsg
> char device which will result in calling rpmsg_create_ept()/rpmsg_poll().
> These functions will try to reference rpdev which has already been freed
> through rpmsg_chrdev_eptdev_destroy().
>
> File operation functions and device removal function must be protected
> with lock. This patch adds existing ept lock in remove function as well.
>
> Signed-off-by: Deepak Kumar Singh <[email protected]>
> ---
> drivers/rpmsg/rpmsg_char.c | 8 ++++++++
> 1 file changed, 8 insertions(+)
>
> diff --git a/drivers/rpmsg/rpmsg_char.c b/drivers/rpmsg/rpmsg_char.c
> index 4f21891..5500dc0 100644
> --- a/drivers/rpmsg/rpmsg_char.c
> +++ b/drivers/rpmsg/rpmsg_char.c
> @@ -75,6 +75,7 @@ int rpmsg_chrdev_eptdev_destroy(struct device *dev, void *data)
> struct rpmsg_eptdev *eptdev = dev_to_eptdev(dev);
>
> mutex_lock(&eptdev->ept_lock);
> + eptdev->rpdev = NULL;
> if (eptdev->ept) {
> rpmsg_destroy_ept(eptdev->ept);
> eptdev->ept = NULL;
> @@ -126,6 +127,11 @@ static int rpmsg_eptdev_open(struct inode *inode, struct file *filp)
> return -EBUSY;
> }
>
> + if (!eptdev->rpdev) {
> + mutex_unlock(&eptdev->ept_lock);
> + return -ENETRESET;
> + }
> +
> get_device(dev);
>
> /*
> @@ -277,7 +283,9 @@ static __poll_t rpmsg_eptdev_poll(struct file *filp, poll_table *wait)
> if (!skb_queue_empty(&eptdev->queue))
> mask |= EPOLLIN | EPOLLRDNORM;
>
> + mutex_lock(&eptdev->ept_lock);
> mask |= rpmsg_poll(eptdev->ept, filp, wait);
> + mutex_unlock(&eptdev->ept_lock);
>
> return mask;
> }
replace "glink" by "ctrl" in subject
On 9/12/22 19:05, Deepak Kumar Singh wrote:
> Call to rpmsg_ctrldev_ioctl() and rpmsg_ctrldev_remove() must be synchronized.
> In present code rpmsg_ctrldev_remove() is not protected with lock, therefore
> new char device creation can succeed through rpmsg_ctrldev_ioctl() call. At the
> same time call to rpmsg_ctrldev_remove() funtion for ctrl device removal will
> free associated rpdev device. As char device creation already succeeded, user
> space is free to issue open() call which maps to rpmsg_create_ept() in kernel.
> rpmsg_create_ept() function tries to reference rpdev which has already been
> freed through rpmsg_ctrldev_remove(). Issue is predominantly seen in aggressive
> reboot tests where rpmsg_ctrldev_ioctl() and rpmsg_ctrldev_remove() can race with
> each other.
>
> Adding lock in rpmsg_ctrldev_remove() avoids any new char device creation
> throught rpmsg_ctrldev_ioctl() while remove call is already in progress.
>
> Signed-off-by: Deepak Kumar Singh <[email protected]>
> ---
> drivers/rpmsg/rpmsg_ctrl.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/drivers/rpmsg/rpmsg_ctrl.c b/drivers/rpmsg/rpmsg_ctrl.c
> index 107da70..4332538 100644
> --- a/drivers/rpmsg/rpmsg_ctrl.c
> +++ b/drivers/rpmsg/rpmsg_ctrl.c
> @@ -194,10 +194,12 @@ static void rpmsg_ctrldev_remove(struct rpmsg_device *rpdev)
> struct rpmsg_ctrldev *ctrldev = dev_get_drvdata(&rpdev->dev);
> int ret;
>
> + mutex_lock(&ctrldev->ctrl_lock);
> /* Destroy all endpoints */
> ret = device_for_each_child(&ctrldev->dev, NULL, rpmsg_chrdev_eptdev_destroy);
> if (ret)
> dev_warn(&rpdev->dev, "failed to nuke endpoints: %d\n", ret);
> + mutex_unlock(&ctrldev->ctrl_lock);
>
> cdev_device_del(&ctrldev->cdev, &ctrldev->dev);
> put_device(&ctrldev->dev);