[PATCH] Alter the scsi_add_device() API to conform to what users expect

[PATCH] Alter the scsi_add_device() API to conform to what users expect

[James Bottomley]

The original API returned either an ERR_PTR() or a refcounted sdev.
Unfortunately, if it's successful, you need to do a scsi_device_put() on
the sdev otherwise the refcounting is wrong.

Everyone seems to expect that scsi_add_device() should be callable
without doing the ref put, so alter the API so it is (we still have
__scsi_add_device with the original behaviour).

The only actual caller that needs altering is the one in firewire ...
not because it gets this right, but because it acts on the error if one
is returned.

James

diff --git a/drivers/ieee1394/sbp2.c b/drivers/ieee1394/sbp2.c
--- a/drivers/ieee1394/sbp2.c
+++ b/drivers/ieee1394/sbp2.c
@@ -790,7 +790,7 @@ static void sbp2_host_reset(struct hpsb_
 static int sbp2_start_device(struct scsi_id_instance_data *scsi_id)
 {
 	struct sbp2scsi_host_info *hi = scsi_id->hi;
-	struct scsi_device *sdev;
+	int error;
 
 	SBP2_DEBUG("sbp2_start_device");
 
@@ -939,10 +939,10 @@ alloc_fail:
 	sbp2_max_speed_and_size(scsi_id);
 
 	/* Add this device to the scsi layer now */
-	sdev = scsi_add_device(scsi_id->scsi_host, 0, scsi_id->ud->id, 0);
-	if (IS_ERR(sdev)) {
+	error = scsi_add_device(scsi_id->scsi_host, 0, scsi_id->ud->id, 0);
+	if (error) {
 		SBP2_ERR("scsi_add_device failed");
-		return PTR_ERR(sdev);
+		return error;
 	}
 
 	return 0;
diff --git a/drivers/scsi/scsi_scan.c b/drivers/scsi/scsi_scan.c
--- a/drivers/scsi/scsi_scan.c
+++ b/drivers/scsi/scsi_scan.c
@@ -1264,6 +1264,19 @@ struct scsi_device *__scsi_add_device(st
 }
 EXPORT_SYMBOL(__scsi_add_device);
 
+int scsi_add_device(struct Scsi_Host *host, uint channel,
+		    uint target, uint lun)
+{
+	struct scsi_device *sdev = 
+		__scsi_add_device(host, channel, target, lun, NULL);
+	if (IS_ERR(sdev))
+		return PTR_ERR(sdev);
+
+	scsi_device_put(sdev);
+	return 0;
+}
+EXPORT_SYMBOL(scsi_add_device);
+
 void scsi_rescan_device(struct device *dev)
 {
 	struct scsi_driver *drv;
diff --git a/include/scsi/scsi_device.h b/include/scsi/scsi_device.h
--- a/include/scsi/scsi_device.h
+++ b/include/scsi/scsi_device.h
@@ -178,8 +178,8 @@ static inline struct scsi_target *scsi_t
 
 extern struct scsi_device *__scsi_add_device(struct Scsi_Host *,
 		uint, uint, uint, void *hostdata);
-#define scsi_add_device(host, channel, target, lun) \
-	__scsi_add_device(host, channel, target, lun, NULL)
+extern int scsi_add_device(struct Scsi_Host *host, uint channel,
+			   uint target, uint lun);
 extern void scsi_remove_device(struct scsi_device *);
 extern int scsi_device_cancel(struct scsi_device *, int);
 

Re: [PATCH] Alter the scsi_add_device() API to conform to what users expect

[Stefan Richter]

James Bottomley wrote:
...
> -	struct scsi_device *sdev;
> +	int error;
...

Side note: The most recent sbp2 at svn.linux1394.org (which is not in 
Linus' tree yet and probably also not in -mm) needs a different approach 
because the sdev is required in sbp2's .remove hook later. I will submit 
an according update to linux1394-devel when the time is rife.
-- 
Stefan Richter
-=====-=-=-= =--= -=-=-
http://arcgraph.de/sr/