Skip to contentSkip to conversation

scsi: scsi_dh_alua: Use common code in alua_rtpg()

PATCH New Markus Elfring 2 participants 2 messages Jun 18

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

PATCH 1/1 scsi: scsi_dh_alua: Use common code in alua_rtpg()

From: Markus Elfring <[email protected]>
Date: Thu, 18 Jun 2026 14:54:13 +0200

Use an additional label so that a bit of common code can be better reused at the end of this function implementation.

This issue was detected by using the Coccinelle software.

---
 drivers/scsi/device_handler/scsi_dh_alua.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)
drivers/scsi/device_handler/scsi_dh_alua.c +6 -6
index 80ab0ff921d4..016150701064 100644
@@ -582,8 +582,8 @@ static int alua_rtpg(struct scsi_device *sdev, struct alua_port_group *pg)
 			sdev_printk(KERN_INFO, sdev,
 				    "%s: ignoring rtpg result %d\n",
 				    ALUA_DH_NAME, retval);
-			kfree(buff);
-			return SCSI_DH_OK;
+			err = SCSI_DH_OK;
+			goto free_buff;
 		}
 		if (retval < 0 || !scsi_sense_valid(&sense_hdr)) {
 			sdev_printk(KERN_INFO, sdev,
@@ -632,15 +632,14 @@ static int alua_rtpg(struct scsi_device *sdev, struct alua_port_group *pg)
 			sdev_printk(KERN_ERR, sdev, "%s: rtpg retry\n",
 				    ALUA_DH_NAME);
 			scsi_print_sense_hdr(sdev, ALUA_DH_NAME, &sense_hdr);
-			kfree(buff);
-			return err;
+			goto free_buff;
 		}
 		sdev_printk(KERN_ERR, sdev, "%s: rtpg failed\n",
 			    ALUA_DH_NAME);
 		scsi_print_sense_hdr(sdev, ALUA_DH_NAME, &sense_hdr);
-		kfree(buff);
 		pg->expiry = 0;
-		return SCSI_DH_IO;
+		err = SCSI_DH_IO;
+		goto free_buff;
 	}
 
 	len = get_unaligned_be32(&buff[0]) + 4;
@@ -770,6 +769,7 @@ static int alua_rtpg(struct scsi_device *sdev, struct alua_port_group *pg)
 		break;
 	}
 	spin_unlock_irqrestore(&pg->lock, flags);
+free_buff:
 	kfree(buff);
 	return err;
 }
Signature · 1 line
2.54.0
#1 lore raw
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] A race condition in alua_rtpg can cause ALUA port group state updates to be permanently dropped. --

commit a4fc4f8579d1bb0ef6314519e4902a1bf335fb2b
Author: Markus Elfring <[email protected]>
Subject: scsi: scsi_dh_alua: Use common code in alua_rtpg()

This commit introduces a goto label to consolidate the memory freeing and return logic in alua_rtpg(). It replaces multiple redundant kfree() calls throughout the error handling paths of the function.

Quoted 4 lines
drivers/scsi/device_handler/scsi_dh_alua.c +0 -0
index 80ab0ff921d43..0161507010643 100644

[ ... ]

Quoted 9 lines
drivers/scsi/device_handler/scsi_dh_alua.c +1 -2
@@ -632,15 +632,14 @@ static int alua_rtpg(struct scsi_device *sdev, struct alua_port_group *pg)
 			sdev_printk(KERN_ERR, sdev, "%s: rtpg retry\n",
 				    ALUA_DH_NAME);
 			scsi_print_sense_hdr(sdev, ALUA_DH_NAME, &sense_hdr);
-			kfree(buff);
-			return err;
+			goto free_buff;
 		}
 		sdev_printk(KERN_ERR, sdev, "%s: rtpg failed\n",

drivers/scsi/device_handler/scsi_dh_alua.c static int alua_rtpg(struct scsi_device *sdev, struct alua_port_group *pg) · line 645

Quoted 10 lines
drivers/scsi/device_handler/scsi_dh_alua.c +2 -2
 			    ALUA_DH_NAME);
 		scsi_print_sense_hdr(sdev, ALUA_DH_NAME, &sense_hdr);
-		kfree(buff);
 		pg->expiry = 0;
-		return SCSI_DH_IO;
+		err = SCSI_DH_IO;
+		goto free_buff;
 	}
 
 	len = get_unaligned_be32(&buff[0]) + 4;

[Severity: High]
This is a pre-existing issue, but was not introduced by the patch under review. While we are consolidating the return paths in alua_rtpg(), does this function still have a race condition when applying the new access state?

Looking at the loop that processes the descriptors further down in alua_rtpg():

    if (spin_trylock_irqsave(&tmp_pg->lock, flags)) {
        if ((tmp_pg == pg) ||
            !(tmp_pg->flags & ALUA_PG_RUNNING)) {
            struct alua_dh_data *h;

            tmp_pg->state = desc[0] & 0x0f;

If this trylock fails because the lock is held by a function that doesn't set the ALUA_PG_RUN_RTPG flag (like alua_bus_detach() or alua_set_params()), the new ALUA access state is silently discarded without queuing a retry.

Could this cause the port group's access state to become permanently stale (e.g., stuck in SCSI_ACCESS_STATE_STANDBY despite the hardware transitioning to an active state), ultimately failing all I/O to non-optimal paths and causing data unavailability?

Quoted 8 lines
drivers/scsi/device_handler/scsi_dh_alua.c +1 -0
@@ -770,6 +769,7 @@ static int alua_rtpg(struct scsi_device *sdev, struct alua_port_group *pg)
 		break;
 	}
 	spin_unlock_irqrestore(&pg->lock, flags);
+free_buff:
 	kfree(buff);
 	return err;
 }
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1