Skip to contentSkip to conversation

scsi: hisi_sas: Fix potential deadlock on &hisi_hba->lock

PATCH In review Chengfeng Ye 2 participants 3 messages Jun 28, 2023 – Jul 5, 2023

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

PATCH 1/1 scsi: hisi_sas: Fix potential deadlock on &hisi_hba->lock

As &hisi_hba->lock is acquired by hard irq int_abnormal_v1_hw(), other acquisition of the same lock under process context should disable irq, otherwise deadlock could happen if the
irq preempt the execution while the lock is held in process context on the same CPU.

[Interrupt]: int_abnormal_v1_hw()

    -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_v1_hw.c:1447
    -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:2050
    -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:1079
    -->spin_lock_irqsave(&hisi_hba->lock, flags);

[Process Context]: hisi_sas_clear_nexus_ha()

    -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:1932
    -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:1135
    -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:1116
    -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:1105
    -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:166
    -->spin_lock(&hisi_hba->lock);

[Process Context]: hisi_sas_dev_found()

    -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:665
    -->spin_lock(&hisi_hba->lock);

[Process Context]: hisi_sas_queue_command()

    -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:188
    -->spin_lock(&hisi_hba->lock);

This flaw was found by an experimental static analysis tool I am developing for irq-related deadlock, which reported the above warning when analyzing the linux kernel 6.4-rc7 release.

The tentative patch fix the potential deadlock by spin_lock_irqsave().

---
 drivers/scsi/hisi_sas/hisi_sas_main.c | 17 ++++++++++-------
 1 file changed, 10 insertions(+), 7 deletions(-)
drivers/scsi/hisi_sas/hisi_sas_main.c +10 -7
index 412431c901a7..47c5062a732e 100644
@@ -161,11 +161,12 @@ static void hisi_sas_slot_index_clear(struct hisi_hba *hisi_hba, int slot_idx)
 
 static void hisi_sas_slot_index_free(struct hisi_hba *hisi_hba, int slot_idx)
 {
+	unsigned long flags;
 	if (hisi_hba->hw->slot_index_alloc ||
 	    slot_idx < HISI_SAS_RESERVED_IPTT) {
-		spin_lock(&hisi_hba->lock);
+		spin_lock_irqsave(&hisi_hba->lock, flags);
 		hisi_sas_slot_index_clear(hisi_hba, slot_idx);
-		spin_unlock(&hisi_hba->lock);
+		spin_unlock_irqrestore(&hisi_hba->lock, flags);
 	}
 }
 
@@ -181,11 +182,12 @@ static int hisi_sas_slot_index_alloc(struct hisi_hba *hisi_hba,
 {
 	int index;
 	void *bitmap = hisi_hba->slot_index_tags;
+	unsigned long flags;
 
 	if (rq)
 		return rq->tag + HISI_SAS_RESERVED_IPTT;
 
-	spin_lock(&hisi_hba->lock);
+	spin_lock_irqsave(&hisi_hba->lock, flags);
 	index = find_next_zero_bit(bitmap, HISI_SAS_RESERVED_IPTT,
 				   hisi_hba->last_slot_index + 1);
 	if (index >= HISI_SAS_RESERVED_IPTT) {
@@ -193,13 +195,13 @@ static int hisi_sas_slot_index_alloc(struct hisi_hba *hisi_hba,
 				HISI_SAS_RESERVED_IPTT,
 				0);
 		if (index >= HISI_SAS_RESERVED_IPTT) {
-			spin_unlock(&hisi_hba->lock);
+			spin_unlock_irqrestore(&hisi_hba->lock, flags);
 			return -SAS_QUEUE_FULL;
 		}
 	}
 	hisi_sas_slot_index_set(hisi_hba, index);
 	hisi_hba->last_slot_index = index;
-	spin_unlock(&hisi_hba->lock);
+	spin_unlock_irqrestore(&hisi_hba->lock, flags);
 
 	return index;
 }
@@ -658,11 +660,12 @@ static struct hisi_sas_device *hisi_sas_alloc_dev(struct domain_device *device)
 {
 	struct hisi_hba *hisi_hba = dev_to_hisi_hba(device);
 	struct hisi_sas_device *sas_dev = NULL;
+	unsigned long flags;
 	int last = hisi_hba->last_dev_id;
 	int first = (hisi_hba->last_dev_id + 1) % HISI_SAS_MAX_DEVICES;
 	int i;
 
-	spin_lock(&hisi_hba->lock);
+	spin_lock_irqsave(&hisi_hba->lock, flags);
 	for (i = first; i != last; i %= HISI_SAS_MAX_DEVICES) {
 		if (hisi_hba->devices[i].dev_type == SAS_PHY_UNUSED) {
 			int queue = i % hisi_hba->queue_count;
@@ -682,7 +685,7 @@ static struct hisi_sas_device *hisi_sas_alloc_dev(struct domain_device *device)
 		i++;
 	}
 	hisi_hba->last_dev_id = i;
-	spin_unlock(&hisi_hba->lock);
+	spin_unlock_irqrestore(&hisi_hba->lock, flags);
 
 	return sas_dev;
 }
Signature · 1 line
2.17.1
#1 lore raw

Hi,

Quoted 25 lines · Chengfeng Ye

As &hisi_hba->lock is acquired by hard irq int_abnormal_v1_hw(), other acquisition of the same lock under process context should disable irq, otherwise deadlock could happen if the
irq preempt the execution while the lock is held in process context on the same CPU.

[Interrupt]: int_abnormal_v1_hw()

     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_v1_hw.c:1447
     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:2050
     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:1079
     -->spin_lock_irqsave(&hisi_hba->lock, flags);

[Process Context]: hisi_sas_clear_nexus_ha()

     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:1932
     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:1135
     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:1116
     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:1105
     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:166
     -->spin_lock(&hisi_hba->lock);

[Process Context]: hisi_sas_dev_found()

     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:665
     -->spin_lock(&hisi_hba->lock);

[Process Context]: hisi_sas_queue_command()

Quoted 8 lines
     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:188
     -->spin_lock(&hisi_hba->lock);

This flaw was found by an experimental static analysis tool I am developing for irq-related deadlock, which reported the above warning when analyzing the linux kernel 6.4-rc7 release.

The tentative patch fix the potential deadlock by spin_lock_irqsave().

Thank you for reporting it.
But we consider about removing hisi_hba->lock in function hisi_sas_port_notify_formed()
which is called by int_abnormal_v1_hw(), as we think it is not necessary to add hisi_hba->lock in this function. So please ignore it and still thank you for pointing out the issue.

Thanks,
Shawn

Quoted 78 lines
---
  drivers/scsi/hisi_sas/hisi_sas_main.c | 17 ++++++++++-------
  1 file changed, 10 insertions(+), 7 deletions(-)
drivers/scsi/hisi_sas/hisi_sas_main.c +10 -7
index 412431c901a7..47c5062a732e 100644
@@ -161,11 +161,12 @@ static void hisi_sas_slot_index_clear(struct hisi_hba *hisi_hba, int slot_idx)
 
  static void hisi_sas_slot_index_free(struct hisi_hba *hisi_hba, int slot_idx)
  {
+	unsigned long flags;
  	if (hisi_hba->hw->slot_index_alloc ||
  	    slot_idx < HISI_SAS_RESERVED_IPTT) {
-		spin_lock(&hisi_hba->lock);
+		spin_lock_irqsave(&hisi_hba->lock, flags);
  		hisi_sas_slot_index_clear(hisi_hba, slot_idx);
-		spin_unlock(&hisi_hba->lock);
+		spin_unlock_irqrestore(&hisi_hba->lock, flags);
  	}
  }
 
@@ -181,11 +182,12 @@ static int hisi_sas_slot_index_alloc(struct hisi_hba *hisi_hba,
  {
  	int index;
  	void *bitmap = hisi_hba->slot_index_tags;
+	unsigned long flags;
 
  	if (rq)
  		return rq->tag + HISI_SAS_RESERVED_IPTT;
 
-	spin_lock(&hisi_hba->lock);
+	spin_lock_irqsave(&hisi_hba->lock, flags);
  	index = find_next_zero_bit(bitmap, HISI_SAS_RESERVED_IPTT,
  				   hisi_hba->last_slot_index + 1);
  	if (index >= HISI_SAS_RESERVED_IPTT) {
@@ -193,13 +195,13 @@ static int hisi_sas_slot_index_alloc(struct hisi_hba *hisi_hba,
  				HISI_SAS_RESERVED_IPTT,
  				0);
  		if (index >= HISI_SAS_RESERVED_IPTT) {
-			spin_unlock(&hisi_hba->lock);
+			spin_unlock_irqrestore(&hisi_hba->lock, flags);
  			return -SAS_QUEUE_FULL;
  		}
  	}
  	hisi_sas_slot_index_set(hisi_hba, index);
  	hisi_hba->last_slot_index = index;
-	spin_unlock(&hisi_hba->lock);
+	spin_unlock_irqrestore(&hisi_hba->lock, flags);
 
  	return index;
  }
@@ -658,11 +660,12 @@ static struct hisi_sas_device *hisi_sas_alloc_dev(struct domain_device *device)
  {
  	struct hisi_hba *hisi_hba = dev_to_hisi_hba(device);
  	struct hisi_sas_device *sas_dev = NULL;
+	unsigned long flags;
  	int last = hisi_hba->last_dev_id;
  	int first = (hisi_hba->last_dev_id + 1) % HISI_SAS_MAX_DEVICES;
  	int i;
 
-	spin_lock(&hisi_hba->lock);
+	spin_lock_irqsave(&hisi_hba->lock, flags);
  	for (i = first; i != last; i %= HISI_SAS_MAX_DEVICES) {
  		if (hisi_hba->devices[i].dev_type == SAS_PHY_UNUSED) {
  			int queue = i % hisi_hba->queue_count;
@@ -682,7 +685,7 @@ static struct hisi_sas_device *hisi_sas_alloc_dev(struct domain_device *device)
  		i++;
  	}
  	hisi_hba->last_dev_id = i;
-	spin_unlock(&hisi_hba->lock);
+	spin_unlock_irqrestore(&hisi_hba->lock, flags);
 
  	return sas_dev;
  }
#2 lore raw

Thanks for the notice!

Yes, removing the lock acquisition inside hisi_sas_port_notify_formed() can avoid the deadlock problem by the way.

Best Regards,
Chengfeng

Quoted 128 lines · chenxiang (M) <[email protected]> 于2023年7月4日周二 13:55写道:

Hi,

Quoted 25 lines · Chengfeng Ye

As &hisi_hba->lock is acquired by hard irq int_abnormal_v1_hw(), other acquisition of the same lock under process context should disable irq, otherwise deadlock could happen if the
irq preempt the execution while the lock is held in process context on the same CPU.

[Interrupt]: int_abnormal_v1_hw()

     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_v1_hw.c:1447
     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:2050
     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:1079
     -->spin_lock_irqsave(&hisi_hba->lock, flags);

[Process Context]: hisi_sas_clear_nexus_ha()

     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:1932
     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:1135
     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:1116
     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:1105
     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:166
     -->spin_lock(&hisi_hba->lock);

[Process Context]: hisi_sas_dev_found()

     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:665
     -->spin_lock(&hisi_hba->lock);

[Process Context]: hisi_sas_queue_command()

Quoted 8 lines
     -->/root/linux/drivers/scsi/hisi_sas/hisi_sas_main.c:188
     -->spin_lock(&hisi_hba->lock);

This flaw was found by an experimental static analysis tool I am developing for irq-related deadlock, which reported the above warning when analyzing the linux kernel 6.4-rc7 release.

The tentative patch fix the potential deadlock by spin_lock_irqsave().

Thank you for reporting it.
But we consider about removing hisi_hba->lock in function hisi_sas_port_notify_formed()
which is called by int_abnormal_v1_hw(), as we think it is not necessary to add hisi_hba->lock in this function.
So please ignore it and still thank you for pointing out the issue.

Thanks,
Shawn

Quoted 78 lines
---
  drivers/scsi/hisi_sas/hisi_sas_main.c | 17 ++++++++++-------
  1 file changed, 10 insertions(+), 7 deletions(-)
drivers/scsi/hisi_sas/hisi_sas_main.c +10 -7
index 412431c901a7..47c5062a732e 100644
@@ -161,11 +161,12 @@ static void hisi_sas_slot_index_clear(struct hisi_hba *hisi_hba, int slot_idx)
 
  static void hisi_sas_slot_index_free(struct hisi_hba *hisi_hba, int slot_idx)
  {
+     unsigned long flags;
      if (hisi_hba->hw->slot_index_alloc ||
          slot_idx < HISI_SAS_RESERVED_IPTT) {
-             spin_lock(&hisi_hba->lock);
+             spin_lock_irqsave(&hisi_hba->lock, flags);
              hisi_sas_slot_index_clear(hisi_hba, slot_idx);
-             spin_unlock(&hisi_hba->lock);
+             spin_unlock_irqrestore(&hisi_hba->lock, flags);
      }
  }
 
@@ -181,11 +182,12 @@ static int hisi_sas_slot_index_alloc(struct hisi_hba *hisi_hba,
  {
      int index;
      void *bitmap = hisi_hba->slot_index_tags;
+     unsigned long flags;
 
      if (rq)
              return rq->tag + HISI_SAS_RESERVED_IPTT;
 
-     spin_lock(&hisi_hba->lock);
+     spin_lock_irqsave(&hisi_hba->lock, flags);
      index = find_next_zero_bit(bitmap, HISI_SAS_RESERVED_IPTT,
                                 hisi_hba->last_slot_index + 1);
      if (index >= HISI_SAS_RESERVED_IPTT) {
@@ -193,13 +195,13 @@ static int hisi_sas_slot_index_alloc(struct hisi_hba *hisi_hba,
                              HISI_SAS_RESERVED_IPTT,
                              0);
              if (index >= HISI_SAS_RESERVED_IPTT) {
-                     spin_unlock(&hisi_hba->lock);
+                     spin_unlock_irqrestore(&hisi_hba->lock, flags);
                      return -SAS_QUEUE_FULL;
              }
      }
      hisi_sas_slot_index_set(hisi_hba, index);
      hisi_hba->last_slot_index = index;
-     spin_unlock(&hisi_hba->lock);
+     spin_unlock_irqrestore(&hisi_hba->lock, flags);
 
      return index;
  }
@@ -658,11 +660,12 @@ static struct hisi_sas_device *hisi_sas_alloc_dev(struct domain_device *device)
  {
      struct hisi_hba *hisi_hba = dev_to_hisi_hba(device);
      struct hisi_sas_device *sas_dev = NULL;
+     unsigned long flags;
      int last = hisi_hba->last_dev_id;
      int first = (hisi_hba->last_dev_id + 1) % HISI_SAS_MAX_DEVICES;
      int i;
 
-     spin_lock(&hisi_hba->lock);
+     spin_lock_irqsave(&hisi_hba->lock, flags);
      for (i = first; i != last; i %= HISI_SAS_MAX_DEVICES) {
              if (hisi_hba->devices[i].dev_type == SAS_PHY_UNUSED) {
                      int queue = i % hisi_hba->queue_count;
@@ -682,7 +685,7 @@ static struct hisi_sas_device *hisi_sas_alloc_dev(struct domain_device *device)
              i++;
      }
      hisi_hba->last_dev_id = i;
-     spin_unlock(&hisi_hba->lock);
+     spin_unlock_irqrestore(&hisi_hba->lock, flags);
 
      return sas_dev;
  }