diff mbox series

bus: mhi: host: Fix race between unprepare and queue_buf

Message ID 20250306172913.856982-1-jeff.hugo@oss.qualcomm.com
State New
Headers show
Series bus: mhi: host: Fix race between unprepare and queue_buf | expand

Commit Message

Jeff Hugo March 6, 2025, 5:29 p.m. UTC
From: Jeffrey Hugo <quic_jhugo@quicinc.com>

A client driver may use mhi_unprepare_from_transfer() to quiesce
incoming data during the client driver's tear down. The client driver
might also be processing data at the same time, resulting in a call to
mhi_queue_buf() which will invoke mhi_gen_tre(). If mhi_gen_tre() runs
after mhi_unprepare_from_transfer() has torn down the channel, a panic
will occur due to an invalid dereference leading to a page fault.

This occurs because mhi_gen_tre() does not verify the channel state
after locking it. Fix this by having mhi_gen_tre() confirm the channel
state is valid, or return error to avoid accessing deinitialized data.

Fixes: b89b6a863dd5 ("bus: mhi: host: Add spinlock to protect WP access when queueing TREs")
Signed-off-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
Signed-off-by: Jeff Hugo <jeff.hugo@oss.qualcomm.com>
---
 drivers/bus/mhi/host/main.c | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)

Comments

Krishna Chaitanya Chundru March 13, 2025, 4:36 a.m. UTC | #1
On 3/6/2025 10:59 PM, Jeff Hugo wrote:
> From: Jeffrey Hugo <quic_jhugo@quicinc.com>
> 
> A client driver may use mhi_unprepare_from_transfer() to quiesce
> incoming data during the client driver's tear down. The client driver
> might also be processing data at the same time, resulting in a call to
> mhi_queue_buf() which will invoke mhi_gen_tre(). If mhi_gen_tre() runs
> after mhi_unprepare_from_transfer() has torn down the channel, a panic
> will occur due to an invalid dereference leading to a page fault.
> 
> This occurs because mhi_gen_tre() does not verify the channel state
> after locking it. Fix this by having mhi_gen_tre() confirm the channel
> state is valid, or return error to avoid accessing deinitialized data.
> 
> Fixes: b89b6a863dd5 ("bus: mhi: host: Add spinlock to protect WP access when queueing TREs")
> Signed-off-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
> Signed-off-by: Jeff Hugo <jeff.hugo@oss.qualcomm.com>
Reviewed-by: Krishna Chaitanya Chundru <krishna.chundru@oss.qualcomm.com>
> ---
>   drivers/bus/mhi/host/main.c | 16 ++++++++++------
>   1 file changed, 10 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/bus/mhi/host/main.c b/drivers/bus/mhi/host/main.c
> index 4de75674f193..aa8a0ef697c7 100644
> --- a/drivers/bus/mhi/host/main.c
> +++ b/drivers/bus/mhi/host/main.c
> @@ -1207,11 +1207,16 @@ int mhi_gen_tre(struct mhi_controller *mhi_cntrl, struct mhi_chan *mhi_chan,
>   	struct mhi_ring_element *mhi_tre;
>   	struct mhi_buf_info *buf_info;
>   	int eot, eob, chain, bei;
> -	int ret;
> +	int ret = 0;
>   
>   	/* Protect accesses for reading and incrementing WP */
>   	write_lock_bh(&mhi_chan->lock);
>   
> +	if (mhi_chan->ch_state != MHI_CH_STATE_ENABLED) {
> +		ret = -ENODEV;
> +		goto out;
> +	}
> +
>   	buf_ring = &mhi_chan->buf_ring;
>   	tre_ring = &mhi_chan->tre_ring;
>   
> @@ -1229,10 +1234,8 @@ int mhi_gen_tre(struct mhi_controller *mhi_cntrl, struct mhi_chan *mhi_chan,
>   
>   	if (!info->pre_mapped) {
>   		ret = mhi_cntrl->map_single(mhi_cntrl, buf_info);
> -		if (ret) {
> -			write_unlock_bh(&mhi_chan->lock);
> -			return ret;
> -		}
> +		if (ret)
> +			goto out;
>   	}
>   
>   	eob = !!(flags & MHI_EOB);
> @@ -1250,9 +1253,10 @@ int mhi_gen_tre(struct mhi_controller *mhi_cntrl, struct mhi_chan *mhi_chan,
>   	mhi_add_ring_element(mhi_cntrl, tre_ring);
>   	mhi_add_ring_element(mhi_cntrl, buf_ring);
>   
> +out:
>   	write_unlock_bh(&mhi_chan->lock);
>   
> -	return 0;
> +	return ret;
>   }
>   
>   int mhi_queue_buf(struct mhi_device *mhi_dev, enum dma_data_direction dir,
Manivannan Sadhasivam March 14, 2025, 5:29 a.m. UTC | #2
On Thu, Mar 06, 2025 at 10:29:13AM -0700, Jeff Hugo wrote:
> From: Jeffrey Hugo <quic_jhugo@quicinc.com>
> 
> A client driver may use mhi_unprepare_from_transfer() to quiesce
> incoming data during the client driver's tear down. The client driver
> might also be processing data at the same time, resulting in a call to
> mhi_queue_buf() which will invoke mhi_gen_tre(). If mhi_gen_tre() runs
> after mhi_unprepare_from_transfer() has torn down the channel, a panic
> will occur due to an invalid dereference leading to a page fault.
> 
> This occurs because mhi_gen_tre() does not verify the channel state
> after locking it. Fix this by having mhi_gen_tre() confirm the channel
> state is valid, or return error to avoid accessing deinitialized data.
> 
> Fixes: b89b6a863dd5 ("bus: mhi: host: Add spinlock to protect WP access when queueing TREs")
> Signed-off-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
> Signed-off-by: Jeff Hugo <jeff.hugo@oss.qualcomm.com>

Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>

- Mani

> ---
>  drivers/bus/mhi/host/main.c | 16 ++++++++++------
>  1 file changed, 10 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/bus/mhi/host/main.c b/drivers/bus/mhi/host/main.c
> index 4de75674f193..aa8a0ef697c7 100644
> --- a/drivers/bus/mhi/host/main.c
> +++ b/drivers/bus/mhi/host/main.c
> @@ -1207,11 +1207,16 @@ int mhi_gen_tre(struct mhi_controller *mhi_cntrl, struct mhi_chan *mhi_chan,
>  	struct mhi_ring_element *mhi_tre;
>  	struct mhi_buf_info *buf_info;
>  	int eot, eob, chain, bei;
> -	int ret;
> +	int ret = 0;
>  
>  	/* Protect accesses for reading and incrementing WP */
>  	write_lock_bh(&mhi_chan->lock);
>  
> +	if (mhi_chan->ch_state != MHI_CH_STATE_ENABLED) {
> +		ret = -ENODEV;
> +		goto out;
> +	}
> +
>  	buf_ring = &mhi_chan->buf_ring;
>  	tre_ring = &mhi_chan->tre_ring;
>  
> @@ -1229,10 +1234,8 @@ int mhi_gen_tre(struct mhi_controller *mhi_cntrl, struct mhi_chan *mhi_chan,
>  
>  	if (!info->pre_mapped) {
>  		ret = mhi_cntrl->map_single(mhi_cntrl, buf_info);
> -		if (ret) {
> -			write_unlock_bh(&mhi_chan->lock);
> -			return ret;
> -		}
> +		if (ret)
> +			goto out;
>  	}
>  
>  	eob = !!(flags & MHI_EOB);
> @@ -1250,9 +1253,10 @@ int mhi_gen_tre(struct mhi_controller *mhi_cntrl, struct mhi_chan *mhi_chan,
>  	mhi_add_ring_element(mhi_cntrl, tre_ring);
>  	mhi_add_ring_element(mhi_cntrl, buf_ring);
>  
> +out:
>  	write_unlock_bh(&mhi_chan->lock);
>  
> -	return 0;
> +	return ret;
>  }
>  
>  int mhi_queue_buf(struct mhi_device *mhi_dev, enum dma_data_direction dir,
> -- 
> 2.34.1
>
Manivannan Sadhasivam March 14, 2025, 5:47 a.m. UTC | #3
On Thu, Mar 06, 2025 at 10:29:13AM -0700, Jeff Hugo wrote:
> From: Jeffrey Hugo <quic_jhugo@quicinc.com>
> 
> A client driver may use mhi_unprepare_from_transfer() to quiesce
> incoming data during the client driver's tear down. The client driver
> might also be processing data at the same time, resulting in a call to
> mhi_queue_buf() which will invoke mhi_gen_tre(). If mhi_gen_tre() runs
> after mhi_unprepare_from_transfer() has torn down the channel, a panic
> will occur due to an invalid dereference leading to a page fault.
> 
> This occurs because mhi_gen_tre() does not verify the channel state
> after locking it. Fix this by having mhi_gen_tre() confirm the channel
> state is valid, or return error to avoid accessing deinitialized data.
> 
> Fixes: b89b6a863dd5 ("bus: mhi: host: Add spinlock to protect WP access when queueing TREs")
> Signed-off-by: Jeffrey Hugo <quic_jhugo@quicinc.com>
> Signed-off-by: Jeff Hugo <jeff.hugo@oss.qualcomm.com>

Applied to mhi-next!

- Mani

> ---
>  drivers/bus/mhi/host/main.c | 16 ++++++++++------
>  1 file changed, 10 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/bus/mhi/host/main.c b/drivers/bus/mhi/host/main.c
> index 4de75674f193..aa8a0ef697c7 100644
> --- a/drivers/bus/mhi/host/main.c
> +++ b/drivers/bus/mhi/host/main.c
> @@ -1207,11 +1207,16 @@ int mhi_gen_tre(struct mhi_controller *mhi_cntrl, struct mhi_chan *mhi_chan,
>  	struct mhi_ring_element *mhi_tre;
>  	struct mhi_buf_info *buf_info;
>  	int eot, eob, chain, bei;
> -	int ret;
> +	int ret = 0;
>  
>  	/* Protect accesses for reading and incrementing WP */
>  	write_lock_bh(&mhi_chan->lock);
>  
> +	if (mhi_chan->ch_state != MHI_CH_STATE_ENABLED) {
> +		ret = -ENODEV;
> +		goto out;
> +	}
> +
>  	buf_ring = &mhi_chan->buf_ring;
>  	tre_ring = &mhi_chan->tre_ring;
>  
> @@ -1229,10 +1234,8 @@ int mhi_gen_tre(struct mhi_controller *mhi_cntrl, struct mhi_chan *mhi_chan,
>  
>  	if (!info->pre_mapped) {
>  		ret = mhi_cntrl->map_single(mhi_cntrl, buf_info);
> -		if (ret) {
> -			write_unlock_bh(&mhi_chan->lock);
> -			return ret;
> -		}
> +		if (ret)
> +			goto out;
>  	}
>  
>  	eob = !!(flags & MHI_EOB);
> @@ -1250,9 +1253,10 @@ int mhi_gen_tre(struct mhi_controller *mhi_cntrl, struct mhi_chan *mhi_chan,
>  	mhi_add_ring_element(mhi_cntrl, tre_ring);
>  	mhi_add_ring_element(mhi_cntrl, buf_ring);
>  
> +out:
>  	write_unlock_bh(&mhi_chan->lock);
>  
> -	return 0;
> +	return ret;
>  }
>  
>  int mhi_queue_buf(struct mhi_device *mhi_dev, enum dma_data_direction dir,
> -- 
> 2.34.1
>
diff mbox series

Patch

diff --git a/drivers/bus/mhi/host/main.c b/drivers/bus/mhi/host/main.c
index 4de75674f193..aa8a0ef697c7 100644
--- a/drivers/bus/mhi/host/main.c
+++ b/drivers/bus/mhi/host/main.c
@@ -1207,11 +1207,16 @@  int mhi_gen_tre(struct mhi_controller *mhi_cntrl, struct mhi_chan *mhi_chan,
 	struct mhi_ring_element *mhi_tre;
 	struct mhi_buf_info *buf_info;
 	int eot, eob, chain, bei;
-	int ret;
+	int ret = 0;
 
 	/* Protect accesses for reading and incrementing WP */
 	write_lock_bh(&mhi_chan->lock);
 
+	if (mhi_chan->ch_state != MHI_CH_STATE_ENABLED) {
+		ret = -ENODEV;
+		goto out;
+	}
+
 	buf_ring = &mhi_chan->buf_ring;
 	tre_ring = &mhi_chan->tre_ring;
 
@@ -1229,10 +1234,8 @@  int mhi_gen_tre(struct mhi_controller *mhi_cntrl, struct mhi_chan *mhi_chan,
 
 	if (!info->pre_mapped) {
 		ret = mhi_cntrl->map_single(mhi_cntrl, buf_info);
-		if (ret) {
-			write_unlock_bh(&mhi_chan->lock);
-			return ret;
-		}
+		if (ret)
+			goto out;
 	}
 
 	eob = !!(flags & MHI_EOB);
@@ -1250,9 +1253,10 @@  int mhi_gen_tre(struct mhi_controller *mhi_cntrl, struct mhi_chan *mhi_chan,
 	mhi_add_ring_element(mhi_cntrl, tre_ring);
 	mhi_add_ring_element(mhi_cntrl, buf_ring);
 
+out:
 	write_unlock_bh(&mhi_chan->lock);
 
-	return 0;
+	return ret;
 }
 
 int mhi_queue_buf(struct mhi_device *mhi_dev, enum dma_data_direction dir,