Skip to contentSkip to conversation

xc4000: Fix atomicity violation in xc4000_get_frequency

PATCH In review Gui-Dong Han 1 participant 2 messages Dec 22, 2023 – Jan 11, 2024

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 xc4000: Fix atomicity violation in xc4000_get_frequency

In xc4000_get_frequency():

	*freq = priv->freq_hz + priv->freq_offset;

The code accesses priv->freq_hz and priv->freq_offset without holding any lock.

In xc4000_set_params():

	// Code that updates priv->freq_hz and priv->freq_offset
	...

xc4000_get_frequency() and xc4000_set_params() may execute concurrently, risking inconsistent reads of priv->freq_hz and priv->freq_offset. Since these related data may update during reading, it can result in incorrect frequency calculation, leading to atomicity violations.

This possible bug is found by an experimental static analysis tool developed by our team, BassCheck[1]. This tool analyzes the locking APIs to extract function pairs that can be concurrently executed, and then analyzes the instructions in the paired functions to identify possible concurrency bugs including data races and atomicity violations. The above possible bug is reported when our tool analyzes the source code of Linux 6.2.

To address this issue, it is proposed to add a mutex lock pair in xc4000_get_frequency() to ensure atomicity. With this patch applied, our tool no longer reports the possible bug, with the kernel configuration allyesconfig for x86_64. Due to the lack of associated hardware, we cannot test the patch in runtime testing, and just verify it according to the code logic.

[1] https://sites.google.com/view/basscheck/

Fixes 4c07e32884ab6 ("[media] xc4000: Fix get_frequency()")

---
v2:
* In this patch v2, we've added some information of the static analysis tool used, as per the researcher guidelines. Also, we've added a cc in the signed-off-by area, according to the stable-kernel-rules. Thank Greg KH for helpful advice.

---
 drivers/media/tuners/xc4000.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/media/tuners/xc4000.c +2 -2
index 57ded9ff3f04..29bc63021c5a 100644
@@ -1515,10 +1515,10 @@ static int xc4000_get_frequency(struct dvb_frontend *fe, u32 *freq)
 {
 	struct xc4000_priv *priv = fe->tuner_priv;
 
+	mutex_lock(&priv->lock);
 	*freq = priv->freq_hz + priv->freq_offset;
 
 	if (debug) {
-		mutex_lock(&priv->lock);
 		if ((priv->cur_fw.type
 		     & (BASE | FM | DTV6 | DTV7 | DTV78 | DTV8)) == BASE) {
 			u16	snr = 0;
@@ -1529,8 +1529,8 @@ static int xc4000_get_frequency(struct dvb_frontend *fe, u32 *freq)
 				return 0;
 			}
 		}
-		mutex_unlock(&priv->lock);
 	}
+	mutex_unlock(&priv->lock);
 
 	dprintk(1, "%s()\n", __func__);
 
Signature · 1 line
2.34.1
#1 lore raw

Dear All:

I hope this email finds you well. I hope you haven't missed my previous email, as I understand that everyone has a busy schedule. I just wanted to follow up on my previous message sent.
I understand that you may be occupied with other tasks or priorities. However, I would greatly appreciate it if you could spare a few moments to check the patch in my previous email. Your prompt response would be highly valuable to me. Thank you for your attention to this matter, and I look forward to hearing from you soon.

Thanks,
Han

Quoted 71 lines · Gui-Dong Han

In xc4000_get_frequency():

	*freq = priv->freq_hz + priv->freq_offset;

The code accesses priv->freq_hz and priv->freq_offset without holding any lock.

In xc4000_set_params():

	// Code that updates priv->freq_hz and priv->freq_offset
	...

xc4000_get_frequency() and xc4000_set_params() may execute concurrently, risking inconsistent reads of priv->freq_hz and priv->freq_offset. Since these related data may update during reading, it can result in incorrect frequency calculation, leading to atomicity violations.

This possible bug is found by an experimental static analysis tool developed by our team, BassCheck[1]. This tool analyzes the locking APIs to extract function pairs that can be concurrently executed, and then analyzes the instructions in the paired functions to identify possible concurrency bugs including data races and atomicity violations. The above possible bug is reported when our tool analyzes the source code of Linux 6.2.

To address this issue, it is proposed to add a mutex lock pair in xc4000_get_frequency() to ensure atomicity. With this patch applied, our tool no longer reports the possible bug, with the kernel configuration allyesconfig for x86_64. Due to the lack of associated hardware, we cannot test the patch in runtime testing, and just verify it according to the code logic.

[1] https://sites.google.com/view/basscheck/

Fixes 4c07e32884ab6 ("[media] xc4000: Fix get_frequency()")

---
v2:
* In this patch v2, we've added some information of the static analysis tool used, as per the researcher guidelines. Also, we've added a cc in the signed-off-by area, according to the stable-kernel-rules. Thank Greg KH for helpful advice.

---
  drivers/media/tuners/xc4000.c | 4 ++--
  1 file changed, 2 insertions(+), 2 deletions(-)
drivers/media/tuners/xc4000.c +2 -2
index 57ded9ff3f04..29bc63021c5a 100644
@@ -1515,10 +1515,10 @@ static int xc4000_get_frequency(struct dvb_frontend *fe, u32 *freq)
  {
  	struct xc4000_priv *priv = fe->tuner_priv;
 
+	mutex_lock(&priv->lock);
  	*freq = priv->freq_hz + priv->freq_offset;
 
  	if (debug) {
-		mutex_lock(&priv->lock);
  		if ((priv->cur_fw.type
  		     & (BASE | FM | DTV6 | DTV7 | DTV78 | DTV8)) == BASE) {
  			u16	snr = 0;
@@ -1529,8 +1529,8 @@ static int xc4000_get_frequency(struct dvb_frontend *fe, u32 *freq)
  				return 0;
  			}
  		}
-		mutex_unlock(&priv->lock);
  	}
+	mutex_unlock(&priv->lock);
 
  	dprintk(1, "%s()\n", __func__);