Skip to contentSkip to conversation

scsi: add non-sleeping variant of scsi_device_put() and use it in alua

PATCH In review [email protected] 2 participants 4 messages Jan 24, 2023 SCSI

lore
Share & tools
lore
#0 lore raw

PATCH v3 1/1 scsi: add non-sleeping variant of scsi_device_put() and use it in alua

From: Martin Wilck <[email protected]>

Since the might_sleep() annotation was added in scsi_device_put() and alua_rtpg_queue(), we have seen repeated reports of "BUG: sleeping function called from invalid context" [1], [2]. alua_rtpg_queue() is always called from contexts where the caller must hold at least one reference to the scsi device in question. This means that the reference taken by alua_rtpg_queue() itself can't be the last one, and thus can be dropped without entering the code path in which scsi_device_put() might actually sleep.

Add a new helper function, scsi_device_put_nosleep() for cases like this, where a device reference is put from atomic context, and at the same time it is certain that this reference is not the last one, and use it from alua_rtpg_queue().

[1] https://lore.kernel.org/linux-scsi/[email protected]/
[2] https://lore.kernel.org/linux-scsi/[email protected]/

Fixes f93ed747e2c7 ("scsi: core: Release SCSI devices synchronously")
Cc Bart Van Assche <[email protected]>
Cc Hannes Reinecke <[email protected]>
Cc Sachin Sant <[email protected]>
Cc Benjamin Block <[email protected]>

----
Changes v2 -> v3:
- clarified the comment on scsi_device_put_nosleep() (Bart van Assche)

Changes v1 -> v2:
- rebased onto the scsi-fixes branch.
- fixed several mistakes pointed out by Steffen Maier.

---
 drivers/scsi/device_handler/scsi_dh_alua.c |  2 +-
 drivers/scsi/scsi.c                        | 24 ++++++++++++++++++----
 include/scsi/scsi_device.h                 |  1 +
 3 files changed, 22 insertions(+), 5 deletions(-)
drivers/scsi/device_handler/scsi_dh_alua.c +1 -1
index 29a2865b8e2e..8f3aac9e6a50 100644
@@ -1032,7 +1032,7 @@ static bool alua_rtpg_queue(struct alua_port_group *pg,
 			kref_put(&pg->kref, release_port_group);
 	}
 	if (sdev)
-		scsi_device_put(sdev);
+		scsi_device_put_nosleep(sdev);
 
 	return true;
 }
drivers/scsi/scsi.c +20 -4
index 1426b9b03612..fb7befacd831 100644
@@ -576,6 +576,25 @@ int scsi_device_get(struct scsi_device *sdev)
 }
 EXPORT_SYMBOL(scsi_device_get);
 
+/**
+ * scsi_device_put_nosleep  -  release a reference to a scsi_device
+ * @sdev:	device to release a reference on.
+ *
+ * Description: Release a reference to the scsi_device and decrements the use
+ * count of the underlying LLDD module. This function may only be called from
+ * a call context where it is certain that the reference dropped is not the
+ * last one. Exception: calling this from scsi_device_put() is legal, because
+ * scsi_device_put() has a might_sleep() annotation.
+ */
+void scsi_device_put_nosleep(struct scsi_device *sdev)
+{
+	struct module *mod = sdev->host->hostt->module;
+
+	put_device(&sdev->sdev_gendev);
+	module_put(mod);
+}
+EXPORT_SYMBOL(scsi_device_put_nosleep);
+
 /**
  * scsi_device_put  -  release a reference to a scsi_device
  * @sdev:	device to release a reference on.
@@ -586,12 +605,9 @@ EXPORT_SYMBOL(scsi_device_get);
  */
 void scsi_device_put(struct scsi_device *sdev)
 {
-	struct module *mod = sdev->host->hostt->module;
-
 	might_sleep();
 
-	put_device(&sdev->sdev_gendev);
-	module_put(mod);
+	scsi_device_put_nosleep(sdev);
 }
 EXPORT_SYMBOL(scsi_device_put);
 
include/scsi/scsi_device.h +1 -0
index 3642b8e3928b..f1ad48c9c601 100644
@@ -365,6 +365,7 @@ void scsi_attach_vpd(struct scsi_device *sdev);
 
 extern struct scsi_device *scsi_device_from_queue(struct request_queue *q);
 extern int __must_check scsi_device_get(struct scsi_device *);
+void scsi_device_put_nosleep(struct scsi_device *sdev);
 extern void scsi_device_put(struct scsi_device *);
 extern struct scsi_device *scsi_device_lookup(struct Scsi_Host *,
 					      uint, uint, u64);
Signature · 1 line
2.39.0
#1 lore raw
Quoted 11 lines · On 1/24/23 07:49, [email protected] wrote:

From: Martin Wilck <[email protected]>

Since the might_sleep() annotation was added in scsi_device_put() and alua_rtpg_queue(), we have seen repeated reports of "BUG: sleeping function called from invalid context" [1], [2]. alua_rtpg_queue() is always called from contexts where the caller must hold at least one reference to the scsi device in question. This means that the reference taken by alua_rtpg_queue() itself can't be the last one, and thus can be dropped without entering the code path in which scsi_device_put() might actually sleep.

Quoted 4 lines

Add a new helper function, scsi_device_put_nosleep() for cases like this, where a device reference is put from atomic context, and at the same time it is certain that this reference is not the last one, and use it from alua_rtpg_queue().

Something I should have asked earlier, has this alternative been considered? This alternative has the advantage that no new functions are introduced.

drivers/scsi/scsi.c +0 -2
index 1426b9b03612..9feb0323bc44 100644
@@ -588,8 +588,6 @@ void scsi_device_put(struct scsi_device *sdev)
 {
 	struct module *mod = sdev->host->hostt->module;
 
-	might_sleep();
-
 	put_device(&sdev->sdev_gendev);
 	module_put(mod);
 }
drivers/scsi/scsi_sysfs.c +2 -0
index 981d1bab2120..8ef9a5494340 100644
@@ -452,5 +451,7 @@ static void scsi_device_dev_release(struct device
 	unsigned long flags;
 
+	might_sleep();
+
 	scsi_dh_release_device(sdev);
 
 	parent = sdev->sdev_gendev.parent;

Thanks,

Bart.

#2 lore raw
Quoted 26 lines · Bart Van Assche
Quoted 17 lines · On 1/24/23 07:49, [email protected] wrote:

From: Martin Wilck <[email protected]>

Since the might_sleep() annotation was added in scsi_device_put() and
alua_rtpg_queue(), we have seen repeated reports of "BUG: sleeping function
called from invalid context" [1], [2]. alua_rtpg_queue() is always called
from contexts where the caller must hold at least one reference to the scsi
device in question. This means that the reference taken by alua_rtpg_queue() itself can't be the last one, and thus can be dropped
without entering the code path in which scsi_device_put() might actually
sleep.

Quoted 7 lines

Add a new helper function, scsi_device_put_nosleep() for cases like this,
where a device reference is put from atomic context, and at the same time
it is certain that this reference is not the last one, and use it from
alua_rtpg_queue().

Quoted 4 lines

Something I should have asked earlier, has this alternative been considered? This alternative has the advantage that no new functions are
introduced.

Looks good to me, too. No, I didn't have this idea before.

Martin

Quoted 30 lines
drivers/scsi/scsi.c +0 -2
index 1426b9b03612..9feb0323bc44 100644
@@ -588,8 +588,6 @@ void scsi_device_put(struct scsi_device *sdev)
   {
         struct module *mod = sdev->host->hostt->module;
 
-       might_sleep();
-
         put_device(&sdev->sdev_gendev);
         module_put(mod);
   }
drivers/scsi/scsi_sysfs.c +2 -0
index 981d1bab2120..8ef9a5494340 100644
@@ -452,5 +451,7 @@ static void scsi_device_dev_release(struct device
         unsigned long flags;
 
+       might_sleep();
+
         scsi_dh_release_device(sdev);
 
         parent = sdev->sdev_gendev.parent;

Thanks,

Bart.

#3 lore raw
Quoted 6 lines · Martin Wilck
Quoted 4 lines

Something I should have asked earlier, has this alternative been considered? This alternative has the advantage that no new functions are
introduced.

Looks good to me, too. No, I didn't have this idea before.

Hi Martin,

Do you plan to post my suggestion as a patch or do you perhaps expect me to do that?

Thanks,

Bart.