Skip to contentSkip to conversation

staging: unisys: uislib: uisqueue.c: rewrite of do_locked_client_insert

PATCH Superseded Sudip Mukherjee 3 participants 9 messages Sep 2, 2014 – Sep 3, 2014 Staging

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 staging: unisys: uislib: uisqueue.c: rewrite of do_locked_client_insert

From: Sudip Mukherjee <[email protected]>

removed unused variables
fixed sparse warning of context imbalance in 'do_locked_client_insert'
different lock contexts for basic block

---

v1 of the patch of the patch just fixed the sparse warning.
On suggestion of Dan Carpenter v2 is the total rewrite of the function.
Two of the function arguments (interruptHandle,channelId) are also not used. Wanted to remove them as well , but then thought maybe the original author have planned for some use of those variables.

 drivers/staging/unisys/uislib/uisqueue.c | 37 ++++++++------------------------
 1 file changed, 9 insertions(+), 28 deletions(-)
drivers/staging/unisys/uislib/uisqueue.c +9 -28
index 63dbe7c..acfa46d 100644
@@ -78,39 +78,20 @@ do_locked_client_insert(struct uisqueue_info *queueinfo,
 			u64 interruptHandle, u8 *channelId)
 {
 	unsigned long flags;
-	unsigned char queueWasEmpty;
-	unsigned int locked = 0;
-	unsigned int acquired = 0;
 	u8 rc = 0;
 
 	spin_lock_irqsave(lock, flags);
-	locked = 1;
-
 	if (!ULTRA_CHANNEL_CLIENT_ACQUIRE_OS(queueinfo->chan, channelId, NULL))
-		goto Away;
-
-	acquired = 1;
-
-	queueWasEmpty = visor_signalqueue_empty(queueinfo->chan, whichqueue);
-	if (!visor_signal_insert(queueinfo->chan, whichqueue, pSignal))
-		goto Away;
-	ULTRA_CHANNEL_CLIENT_RELEASE_OS(queueinfo->chan, channelId, NULL);
-	acquired = 0;
-
-	queueinfo->packets_sent++;
-
-	rc = 1;
-Away:
-	if (acquired) {
-		ULTRA_CHANNEL_CLIENT_RELEASE_OS(queueinfo->chan, channelId,
-						NULL);
-		acquired = 0;
-	}
-	if (locked) {
-		spin_unlock_irqrestore((spinlock_t *) lock, flags);
-		locked = 0;
+		goto unlock;
+	visor_signalqueue_empty(queueinfo->chan, whichqueue);
+	/*visor_signal_insert() only return 0 or 1 */
+	if (visor_signal_insert(queueinfo->chan, whichqueue, pSignal) == 1) {
+		queueinfo->packets_sent++;
+		rc = 1;
 	}
-
+	ULTRA_CHANNEL_CLIENT_RELEASE_OS(queueinfo->chan, channelId, NULL);
+unlock:
+	spin_unlock_irqrestore((spinlock_t *)lock, flags);
 	return rc;
 }
 
Signature · 1 line
1.8.1.2
#1 lore raw
Quoted 49 lines · Sudip Mukherjee

From: Sudip Mukherjee <[email protected]>

removed unused variables
fixed sparse warning of context imbalance in 'do_locked_client_insert'
different lock contexts for basic block

---

v1 of the patch of the patch just fixed the sparse warning.
On suggestion of Dan Carpenter v2 is the total rewrite of the function.
Two of the function arguments (interruptHandle,channelId) are also not used. Wanted to remove them as well , but then thought maybe the original author have planned for some use of those variables.

 drivers/staging/unisys/uislib/uisqueue.c | 37 ++++++++------------------------
 1 file changed, 9 insertions(+), 28 deletions(-)
drivers/staging/unisys/uislib/uisqueue.c +0 -20
index 63dbe7c..acfa46d 100644
@@ -78,39 +78,20 @@ do_locked_client_insert(struct uisqueue_info *queueinfo,
 			u64 interruptHandle, u8 *channelId)
 {
 	unsigned long flags;
-	unsigned char queueWasEmpty;
-	unsigned int locked = 0;
-	unsigned int acquired = 0;
 	u8 rc = 0;
 
 	spin_lock_irqsave(lock, flags);
-	locked = 1;
-
 	if (!ULTRA_CHANNEL_CLIENT_ACQUIRE_OS(queueinfo->chan, channelId, NULL))
-		goto Away;
-
-	acquired = 1;
-
-	queueWasEmpty = visor_signalqueue_empty(queueinfo->chan, whichqueue);
-	if (!visor_signal_insert(queueinfo->chan, whichqueue, pSignal))
-		goto Away;
-	ULTRA_CHANNEL_CLIENT_RELEASE_OS(queueinfo->chan, channelId, NULL);
-	acquired = 0;
-
-	queueinfo->packets_sent++;
-
-	rc = 1;
-Away:
-	if (acquired) {

drivers/staging/unisys/uislib/uisqueue.c do_locked_client_insert(struct uisqueue_info *queueinfo, · line 87

Quoted 10 lines
drivers/staging/unisys/uislib/uisqueue.c +3 -7
-		ULTRA_CHANNEL_CLIENT_RELEASE_OS(queueinfo->chan, channelId,
-						NULL);
-		acquired = 0;
-	}
-	if (locked) {
-		spin_unlock_irqrestore((spinlock_t *) lock, flags);
-		locked = 0;
+		goto unlock;
+	visor_signalqueue_empty(queueinfo->chan, whichqueue);
+	/*visor_signal_insert() only return 0 or 1 */

Odd comment style (hint, you need a ' ' at the beginning...)

And why comment this at all?

thanks,

greg k-h

#2 lore raw
Quoted 63 lines · Greg Kroah-Hartman
Quoted 49 lines · Sudip Mukherjee

From: Sudip Mukherjee <[email protected]>

removed unused variables
fixed sparse warning of context imbalance in 'do_locked_client_insert'
different lock contexts for basic block

---

v1 of the patch of the patch just fixed the sparse warning.
On suggestion of Dan Carpenter v2 is the total rewrite of the function.
Two of the function arguments (interruptHandle,channelId) are also not used. Wanted to remove them as well , but then thought maybe the original author have planned for some use of those variables.

 drivers/staging/unisys/uislib/uisqueue.c | 37 ++++++++------------------------
 1 file changed, 9 insertions(+), 28 deletions(-)
drivers/staging/unisys/uislib/uisqueue.c +0 -20
index 63dbe7c..acfa46d 100644
@@ -78,39 +78,20 @@ do_locked_client_insert(struct uisqueue_info *queueinfo,
 			u64 interruptHandle, u8 *channelId)
 {
 	unsigned long flags;
-	unsigned char queueWasEmpty;
-	unsigned int locked = 0;
-	unsigned int acquired = 0;
 	u8 rc = 0;
 
 	spin_lock_irqsave(lock, flags);
-	locked = 1;
-
 	if (!ULTRA_CHANNEL_CLIENT_ACQUIRE_OS(queueinfo->chan, channelId, NULL))
-		goto Away;
-
-	acquired = 1;
-
-	queueWasEmpty = visor_signalqueue_empty(queueinfo->chan, whichqueue);
-	if (!visor_signal_insert(queueinfo->chan, whichqueue, pSignal))
-		goto Away;
-	ULTRA_CHANNEL_CLIENT_RELEASE_OS(queueinfo->chan, channelId, NULL);
-	acquired = 0;
-
-	queueinfo->packets_sent++;
-
-	rc = 1;
-Away:
-	if (acquired) {

drivers/staging/unisys/uislib/uisqueue.c do_locked_client_insert(struct uisqueue_info *queueinfo, · line 87

Quoted 10 lines
drivers/staging/unisys/uislib/uisqueue.c +3 -7
-		ULTRA_CHANNEL_CLIENT_RELEASE_OS(queueinfo->chan, channelId,
-						NULL);
-		acquired = 0;
-	}
-	if (locked) {
-		spin_unlock_irqrestore((spinlock_t *) lock, flags);
-		locked = 0;
+		goto unlock;
+	visor_signalqueue_empty(queueinfo->chan, whichqueue);
+	/*visor_signal_insert() only return 0 or 1 */

Odd comment style (hint, you need a ' ' at the beginning...)

Quoted 5 lines

And why comment this at all?

thanks,

greg k-h

comment is actually not required. thought it will be easier to understant the code if the return values can be see. I will delete the comment and send it again.

thanks
sudip

#3 lore raw
In reply to Sudip Mukherjee
Quoted 1 line · Sudip Mukherjee

From: Sudip Mukherjee <[email protected]>

I really would prefer if you just figured out your email settings so this isn't needed. The From: header is mostly used for people forwarding patches from other people. We have allowed people to use the From header like this if they can't get their corporate email configured properly but I try to discorage it. If everyone starts using

Quoted 10 lines

From headers like this then it becomes a pain to deal with.

removed unused variables
fixed sparse warning of context imbalance in 'do_locked_client_insert'
different lock contexts for basic block

---

This patch is much better and more interesting, but I still want some more changes.

Quoted 4 lines

v1 of the patch of the patch just fixed the sparse warning.
On suggestion of Dan Carpenter v2 is the total rewrite of the function.
Two of the function arguments (interruptHandle,channelId) are also not used. Wanted to remove them as well , but then thought maybe the original author have planned for some use of those variables.

In the kernel we don't put code in until we are ready to use it. Don't worry about future changes. But on the other hand, don't remove the parameters in this patch because that is doing too many changes in one patch. It would have to be done in a follow on patch if you decide to do it.

drivers/staging/unisys/uislib/uisqueue.c do_locked_client_insert(struct uisqueue_info *queueinfo, · line 86

Quoted 5 lines
drivers/staging/unisys/uislib/uisqueue.c +2 -3
-	if (locked) {
-		spin_unlock_irqrestore((spinlock_t *) lock, flags);
-		locked = 0;
+		goto unlock;
+	visor_signalqueue_empty(queueinfo->chan, whichqueue);

Just remove this function. But mention it in the changelog in case there are side effects.

drivers/staging/unisys/uislib/uisqueue.c do_locked_client_insert(struct uisqueue_info *queueinfo, · line 87

Quoted 1 line
drivers/staging/unisys/uislib/uisqueue.c +1 -0
+	/*visor_signal_insert() only return 0 or 1 */

Don't put obvious comments like this. A normal reader will assume that this function is boolean based on how it is used.

drivers/staging/unisys/uislib/uisqueue.c do_locked_client_insert(struct uisqueue_info *queueinfo, · line 88

Quoted 1 line
drivers/staging/unisys/uislib/uisqueue.c +1 -0
+	if (visor_signal_insert(queueinfo->chan, whichqueue, pSignal) == 1) {

Don't put the == 1. In terms of English, 1 really is intended as "success" and not the number one. Also don't test for == true or == false.

	if (foo) {
	if (foo == true) {

These two statement *mean* the same thing in terms of English, but the first one is simpler and less wordy.

regards,
dan carpenter

#4 lore raw
Quoted 10 lines · Dan Carpenter
Quoted 1 line · Sudip Mukherjee

From: Sudip Mukherjee <[email protected]>

I really would prefer if you just figured out your email settings so this isn't needed. The From: header is mostly used for people forwarding patches from other people. We have allowed people to use the From header like this if they can't get their corporate email configured properly but I try to discorage it. If everyone starts using From headers like this then it becomes a pain to deal with.

I will configure the corporate mail. I am the server admin , so there should not be any problem in settings. :)

Quoted 5 lines
Quoted 5 lines

removed unused variables
fixed sparse warning of context imbalance in 'do_locked_client_insert'
different lock contexts for basic block

Quoted 7 lines
Quoted 3 lines

---

This patch is much better and more interesting, but I still want some more changes.

I have already sent v3 of the patch just before your mail , based on what greg k-h has suggested about the commnent. Please discard that.

Quoted 31 lines
Quoted 4 lines

v1 of the patch of the patch just fixed the sparse warning.
On suggestion of Dan Carpenter v2 is the total rewrite of the function.
Two of the function arguments (interruptHandle,channelId) are also not used. Wanted to remove them as well , but then thought maybe the original author have planned for some use of those variables.

In the kernel we don't put code in until we are ready to use it. Don't worry about future changes. But on the other hand, don't remove the parameters in this patch because that is doing too many changes in one patch. It would have to be done in a follow on patch if you decide to do it.

drivers/staging/unisys/uislib/uisqueue.c do_locked_client_insert(struct uisqueue_info *queueinfo, · line 86

Quoted 5 lines
drivers/staging/unisys/uislib/uisqueue.c +2 -3
-	if (locked) {
-		spin_unlock_irqrestore((spinlock_t *) lock, flags);
-		locked = 0;
+		goto unlock;
+	visor_signalqueue_empty(queueinfo->chan, whichqueue);

Just remove this function. But mention it in the changelog in case there are side effects.

drivers/staging/unisys/uislib/uisqueue.c do_locked_client_insert(struct uisqueue_info *queueinfo, · line 87

Quoted 1 line
drivers/staging/unisys/uislib/uisqueue.c +1 -0
+	/*visor_signal_insert() only return 0 or 1 */

Don't put obvious comments like this. A normal reader will assume that this function is boolean based on how it is used.

drivers/staging/unisys/uislib/uisqueue.c do_locked_client_insert(struct uisqueue_info *queueinfo, · line 88

Quoted 1 line
drivers/staging/unisys/uislib/uisqueue.c +1 -0
+	if (visor_signal_insert(queueinfo->chan, whichqueue, pSignal) == 1) {

Don't put the == 1. In terms of English, 1 really is intended as "success" and not the number one. Also don't test for == true or == false.

Quoted 8 lines
	if (foo) {
	if (foo == true) {

These two statement *mean* the same thing in terms of English, but the first one is simpler and less wordy.

regards,
dan carpenter

thanks
sudip

#5 lore raw
Quoted 5 lines · Sudip Mukherjee
Quoted 3 lines

This patch is much better and more interesting, but I still want some more changes.

I have already sent v3 of the patch just before your mail , based on what greg k-h has suggested about the commnent. Please discard that.

Greg has a million emails in his inbox so you need to reply to the patch itself if you want him to discard it. If it's in a different thread he can't keep track.

regards,
dan carpenter

#6 lore raw
In reply to Sudip Mukherjee
Quoted 11 lines · Sudip Mukherjee
Quoted 10 lines · Dan Carpenter
Quoted 1 line · Sudip Mukherjee

From: Sudip Mukherjee <[email protected]>

I really would prefer if you just figured out your email settings so this isn't needed. The From: header is mostly used for people forwarding patches from other people. We have allowed people to use the From header like this if they can't get their corporate email configured properly but I try to discorage it. If everyone starts using From headers like this then it becomes a pain to deal with.

Quoted 3 lines

I will configure the corporate mail. I am the server admin , so there should not be any problem in settings. :)

v4 of the patch was sent from the corporate mail. The settings were done. But the problem is coming in a different area. I have given strict DMARC check for the corporate mail server. DMARC = domain based message authentication. So the mail i sent reached all the list subscriber from a different server than our designated server , and as a result it has been marked as spam in many places. I have already received a few complaints regarding that.
Is there any other way that i send the patch from my personal account , and use my corporate mail in Signed-off-by ...

thanks
sudip

Quoted 57 lines
Quoted 5 lines
Quoted 5 lines

removed unused variables
fixed sparse warning of context imbalance in 'do_locked_client_insert'
different lock contexts for basic block

Quoted 7 lines
Quoted 3 lines

---

This patch is much better and more interesting, but I still want some more changes.

I have already sent v3 of the patch just before your mail , based on what greg k-h has suggested about the commnent. Please discard that.

Quoted 31 lines
Quoted 4 lines

v1 of the patch of the patch just fixed the sparse warning.
On suggestion of Dan Carpenter v2 is the total rewrite of the function.
Two of the function arguments (interruptHandle,channelId) are also not used. Wanted to remove them as well , but then thought maybe the original author have planned for some use of those variables.

In the kernel we don't put code in until we are ready to use it. Don't worry about future changes. But on the other hand, don't remove the parameters in this patch because that is doing too many changes in one patch. It would have to be done in a follow on patch if you decide to do it.

drivers/staging/unisys/uislib/uisqueue.c do_locked_client_insert(struct uisqueue_info *queueinfo, · line 86

Quoted 5 lines
drivers/staging/unisys/uislib/uisqueue.c +2 -3
-	if (locked) {
-		spin_unlock_irqrestore((spinlock_t *) lock, flags);
-		locked = 0;
+		goto unlock;
+	visor_signalqueue_empty(queueinfo->chan, whichqueue);

Just remove this function. But mention it in the changelog in case there are side effects.

drivers/staging/unisys/uislib/uisqueue.c do_locked_client_insert(struct uisqueue_info *queueinfo, · line 87

Quoted 1 line
drivers/staging/unisys/uislib/uisqueue.c +1 -0
+	/*visor_signal_insert() only return 0 or 1 */

Don't put obvious comments like this. A normal reader will assume that this function is boolean based on how it is used.

drivers/staging/unisys/uislib/uisqueue.c do_locked_client_insert(struct uisqueue_info *queueinfo, · line 88

Quoted 1 line
drivers/staging/unisys/uislib/uisqueue.c +1 -0
+	if (visor_signal_insert(queueinfo->chan, whichqueue, pSignal) == 1) {

Don't put the == 1. In terms of English, 1 really is intended as "success" and not the number one. Also don't test for == true or == false.

Quoted 8 lines
	if (foo) {
	if (foo == true) {

These two statement *mean* the same thing in terms of English, but the first one is simpler and less wordy.

regards,
dan carpenter

thanks
sudip

#7 lore raw
Quoted 15 lines · Sudip Mukherjee
Quoted 11 lines · Sudip Mukherjee
Quoted 10 lines · Dan Carpenter
Quoted 1 line · Sudip Mukherjee

From: Sudip Mukherjee <[email protected]>

I really would prefer if you just figured out your email settings so this isn't needed. The From: header is mostly used for people forwarding patches from other people. We have allowed people to use the From header like this if they can't get their corporate email configured properly but I try to discorage it. If everyone starts using From headers like this then it becomes a pain to deal with.

Quoted 3 lines

I will configure the corporate mail. I am the server admin , so there should not be any problem in settings. :)

Quoted 8 lines

v4 of the patch was sent from the corporate mail. The settings were done. But the problem is coming in a different area. I have given strict DMARC check for the corporate mail server. DMARC = domain based message authentication. So the mail i sent reached all the list subscriber from a different server than our designated server , and as a result it has been marked as spam in many places. I have already received a few complaints regarding that.
Is there any other way that i send the patch from my personal account , and use my corporate mail in Signed-off-by ...

I guess it's ok by me. It's up to Greg.

regards,
dan carpenter

#8 lore raw
Quoted 25 lines · Dan Carpenter
Quoted 15 lines · Sudip Mukherjee
Quoted 11 lines · Sudip Mukherjee
Quoted 10 lines · Dan Carpenter
Quoted 1 line · Sudip Mukherjee

From: Sudip Mukherjee <[email protected]>

I really would prefer if you just figured out your email settings so this isn't needed. The From: header is mostly used for people forwarding patches from other people. We have allowed people to use the From header like this if they can't get their corporate email configured properly but I try to discorage it. If everyone starts using From headers like this then it becomes a pain to deal with.

Quoted 3 lines

I will configure the corporate mail. I am the server admin , so there should not be any problem in settings. :)

Quoted 8 lines

v4 of the patch was sent from the corporate mail. The settings were done. But the problem is coming in a different area. I have given strict DMARC check for the corporate mail server. DMARC = domain based message authentication. So the mail i sent reached all the list subscriber from a different server than our designated server , and as a result it has been marked as spam in many places. I have already received a few complaints regarding that.
Is there any other way that i send the patch from my personal account , and use my corporate mail in Signed-off-by ...

Quoted 2 lines

I guess it's ok by me. It's up to Greg.

Greh K-H has already accepted some of my patches and some are still pending :( , where from was my personal email and Signed-off-by as my corporate email. So i guess its ok for him also. .

thanks
sudip

Quoted 3 lines

regards,
dan carpenter