Skip to contentSkip to conversation

staging: unisys: use common return path

PATCH In review Sudip Mukherjee 2 participants 4 messages Dec 1, 2015 – Feb 10, 2016 Staging

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 staging: unisys: use common return path

The success path and the error path both are first doing spin_unlock_irqrestore() before returning. Use that in the common path and return the success/error value.

---

v2: changed label name

It is dependent on the patch series sent by Benjamin (Nov 30th).

 drivers/staging/unisys/visornic/visornic_main.c | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)
drivers/staging/unisys/visornic/visornic_main.c +7 -5
index 77fd1ef..342c0bf 100644
@@ -357,15 +357,16 @@ visornic_serverdown(struct visornic_devdata *devdata,
 		    visorbus_state_complete_func complete_func)
 {
 	unsigned long flags;
+	int retval = 0;
 
 	spin_lock_irqsave(&devdata->priv_lock, flags);
 	if (!devdata->server_down && !devdata->server_change_state) {
 		if (devdata->going_away) {
-			spin_unlock_irqrestore(&devdata->priv_lock, flags);
 			dev_dbg(&devdata->dev->device,
 				"%s aborting because device removal pending\n",
 				__func__);
-			return -ENODEV;
+			retval = -ENODEV;
+			goto error;
 		}
 		devdata->server_change_state = true;
 		devdata->server_down_complete_func = complete_func;
@@ -374,11 +375,12 @@ visornic_serverdown(struct visornic_devdata *devdata,
 	} else if (devdata->server_change_state) {
 		dev_dbg(&devdata->dev->device, "%s changing state\n",
 			__func__);
-		spin_unlock_irqrestore(&devdata->priv_lock, flags);
-		return -EINVAL;
+		retval = -EINVAL;
 	}
+
+unlock:
 	spin_unlock_irqrestore(&devdata->priv_lock, flags);
-	return 0;
+	return retval;
 }
 
 /**
Signature · 1 line
1.9.1
#1 lore raw
Quoted 10 lines · Sudip Mukherjee

The success path and the error path both are first doing spin_unlock_irqrestore() before returning. Use that in the common path and return the success/error value.

---

v2: changed label name

It is dependent on the patch series sent by Benjamin (Nov 30th).

Which didn't get applied, so can you resend this after that does get accepted?

thanks,

greg k-h

#2 lore raw
Quoted 8 lines · Greg Kroah-Hartman
Quoted 7 lines · Sudip Mukherjee

The success path and the error path both are first doing spin_unlock_irqrestore() before returning. Use that in the common path and return the success/error value.

---

Quoted 6 lines
Quoted 3 lines

v2: changed label name

It is dependent on the patch series sent by Benjamin (Nov 30th).

Which didn't get applied, so can you resend this after that does get accepted?

oops, i dont think i have the patch anymore. But looking at the commit message, i think i just sent the same suggestion to Benjamin for his [PATCH v3 13/14] staging: unisys: fix else statement in visornic_main.c

If he is not changing then I will create the patch and send you.

regards
sudip

#3 lore raw
Quoted 13 lines · Sudip Mukherjee
Quoted 12 lines · Greg Kroah-Hartman
Quoted 10 lines · Sudip Mukherjee

The success path and the error path both are first doing spin_unlock_irqrestore() before returning. Use that in the common path and return the success/error value.

---

v2: changed label name

It is dependent on the patch series sent by Benjamin (Nov 30th).

Quoted 8 lines
Quoted 2 lines

Which didn't get applied, so can you resend this after that does get accepted?

oops, i dont think i have the patch anymore. But looking at the commit message, i think i just sent the same suggestion to Benjamin for his [PATCH v3 13/14] staging: unisys: fix else statement in visornic_main.c

If he is not changing then I will create the patch and send you.

Hi Greg,
I will not be resending this patch as Ben is against to using goto.

regards
sudip