Skip to contentSkip to conversation

staging: most: remove always true comparison

PATCH In review Sudip Mukherjee 3 participants 9 messages Sep 4, 2015 – Sep 8, 2015

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

PATCH 1/5 staging: most: remove always true comparison

channel->dev has already been checked for NULL and if it was NULL then we have returned with -EPIPE. So at this point it can not be NULL.

---
 drivers/staging/most/aim-cdev/cdev.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/staging/most/aim-cdev/cdev.c +1 -1
index b0a9a4a..1a17e2a 100644
@@ -166,7 +166,7 @@ static ssize_t aim_write(struct file *filp, const char __user *buf,
 
 	mbo = most_get_mbo(channel->iface, channel->channel_id);
 
-	if (!mbo && channel->dev) {
+	if (!mbo) {
 		if ((filp->f_flags & O_NONBLOCK))
 			return -EAGAIN;
 		if (wait_event_interruptible(
Signature · 1 line
1.9.1
#1 lore raw

PATCH 2/5 staging: most: return NULL instead of integer

The return type of get_aim_dev() is a pointer but we were returning 0 incase of failure.

---
 drivers/staging/most/aim-v4l2/video.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/staging/most/aim-v4l2/video.c +1 -1
index d968791..8333245 100644
@@ -430,7 +430,7 @@ static struct most_video_dev *get_aim_dev(
 		}
 	}
 	spin_unlock(&list_lock);
-	return 0;
+	return NULL;
 }
 
 static int aim_rx_data(struct mbo *mbo)
Signature · 1 line
1.9.1
#2 lore raw

PATCH 3/5 staging: most: remove driver owner

In reply to Sudip Mukherjee

The platform driver core will set the owner value, we do not need to do it in the module.

---
 drivers/staging/most/hdm-dim2/dim2_hdm.c | 1 -
 1 file changed, 1 deletion(-)
drivers/staging/most/hdm-dim2/dim2_hdm.c +0 -1
index 5b0a588..4481a0b 100644
@@ -923,7 +923,6 @@ static struct platform_driver dim2_driver = {
 	.id_table = dim2_id,
 	.driver = {
 		.name = "hdm_dim2",
-		.owner = THIS_MODULE,
 	},
 };
 
Signature · 1 line
1.9.1
#3 lore raw

Re: [PATCH 3/5] staging: most: remove driver owner

Quoted 13 lines · Sudip Mukherjee

The platform driver core will set the owner value, we do not need to do it in the module.

---
 drivers/staging/most/hdm-dim2/dim2_hdm.c | 1 -
 1 file changed, 1 deletion(-)
drivers/staging/most/hdm-dim2/dim2_hdm.c +0 -0
index 5b0a588..4481a0b 100644

drivers/staging/most/hdm-dim2/dim2_hdm.c static struct platform_driver dim2_driver = { · line 926

Quoted 4 lines
drivers/staging/most/hdm-dim2/dim2_hdm.c +0 -1
@@ -923,7 +923,6 @@ static struct platform_driver dim2_driver = {
 	.id_table = dim2_id,
 	.driver = {
 		.name = "hdm_dim2",
-		.owner = THIS_MODULE,

I cannot accept this.

This change is not significant for current kernel, but we still have customers using kernels <= 3.10, where auto assignment of .owner does not exist.

Quoted 10 lines
drivers/staging/most/hdm-dim2/dim2_hdm.c +0 -0
 	},
 };
 
Signature · 6 lines
1.9.1

_______________________________________________
devel mailing list
[email protected]
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
#4 lore raw

Re: [PATCH 3/5] staging: most: remove driver owner

Quoted 20 lines · Andrey Shvetsov
Quoted 13 lines · Sudip Mukherjee

The platform driver core will set the owner value, we do not need to do it in the module.

---
 drivers/staging/most/hdm-dim2/dim2_hdm.c | 1 -
 1 file changed, 1 deletion(-)
drivers/staging/most/hdm-dim2/dim2_hdm.c +0 -0
index 5b0a588..4481a0b 100644

drivers/staging/most/hdm-dim2/dim2_hdm.c static struct platform_driver dim2_driver = { · line 926

Quoted 4 lines
drivers/staging/most/hdm-dim2/dim2_hdm.c +0 -1
@@ -923,7 +923,6 @@ static struct platform_driver dim2_driver = {
 	.id_table = dim2_id,
 	.driver = {
 		.name = "hdm_dim2",
-		.owner = THIS_MODULE,

I cannot accept this.

Quoted 2 lines

This change is not significant for current kernel, but we still have customers using kernels <= 3.10, where auto assignment of .owner does not exist.

But this patch is for 4.4-rc1, and it is not marked for stable. Then how is it going to affect your customers who are still using <=3.10?

regards
sudip

#5 lore raw

Re: [PATCH 3/5] staging: most: remove driver owner

Quoted 21 lines · Sudip Mukherjee
Quoted 20 lines · Andrey Shvetsov
Quoted 13 lines · Sudip Mukherjee

The platform driver core will set the owner value, we do not need to do it in the module.

---
 drivers/staging/most/hdm-dim2/dim2_hdm.c | 1 -
 1 file changed, 1 deletion(-)
drivers/staging/most/hdm-dim2/dim2_hdm.c +0 -0
index 5b0a588..4481a0b 100644

drivers/staging/most/hdm-dim2/dim2_hdm.c static struct platform_driver dim2_driver = { · line 926

Quoted 4 lines
drivers/staging/most/hdm-dim2/dim2_hdm.c +0 -1
@@ -923,7 +923,6 @@ static struct platform_driver dim2_driver = {
 	.id_table = dim2_id,
 	.driver = {
 		.name = "hdm_dim2",
-		.owner = THIS_MODULE,

I cannot accept this.

Quoted 4 lines
Quoted 2 lines

This change is not significant for current kernel, but we still have customers using kernels <= 3.10, where auto assignment of .owner does not exist.

But this patch is for 4.4-rc1, and it is not marked for stable. Then how is it going to affect your customers who are still using <=3.10?

The customers will get MOST Linux Driver of 4.x and compile it with 3.10. They'll not even see the warning, but missing owner is definitely not what they want to have. This patch prepares wasted time of cusomers expiriencing the issue and ours supporting resolving of some problem. But what we get instead?

Signature · 2 lines
regards
andy
#6 lore raw

Re: [PATCH 3/5] staging: most: remove driver owner

Quoted 22 lines · Andrey Shvetsov
Quoted 21 lines · Sudip Mukherjee
Quoted 20 lines · Andrey Shvetsov
Quoted 13 lines · Sudip Mukherjee

The platform driver core will set the owner value, we do not need to do it in the module.

---
 drivers/staging/most/hdm-dim2/dim2_hdm.c | 1 -
 1 file changed, 1 deletion(-)
drivers/staging/most/hdm-dim2/dim2_hdm.c +0 -0
index 5b0a588..4481a0b 100644

drivers/staging/most/hdm-dim2/dim2_hdm.c static struct platform_driver dim2_driver = { · line 926

Quoted 4 lines
drivers/staging/most/hdm-dim2/dim2_hdm.c +0 -1
@@ -923,7 +923,6 @@ static struct platform_driver dim2_driver = {
 	.id_table = dim2_id,
 	.driver = {
 		.name = "hdm_dim2",
-		.owner = THIS_MODULE,

I cannot accept this.

Quoted 5 lines
Quoted 4 lines
Quoted 2 lines

This change is not significant for current kernel, but we still have customers using kernels <= 3.10, where auto assignment of .owner does not exist.

But this patch is for 4.4-rc1, and it is not marked for stable. Then how is it going to affect your customers who are still using <=3.10?

The customers will get MOST Linux Driver of 4.x and compile it with 3.10.

Really? If they do that they are on their own and odds are, it will break.

Quoted 3 lines

They'll not even see the warning, but missing owner is definitely not what they want to have. This patch prepares wasted time of cusomers expiriencing the issue and ours supporting resolving of some problem. But what we get instead?

We do not support people taking a kernel driver from one kernel and putting it into another one. That's never been something that anyone can ever do with Linux, sorry.

The patch is correct.

greg k-h

#7 lore raw

PATCH 4/5 staging: most: remove unneeded NULL check

In reply to Sudip Mukherjee

The loop cursor of list_for_each_entry_safe() can never be NULL.

---
 drivers/staging/most/mostcore/core.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)
drivers/staging/most/mostcore/core.c +2 -4
index 7bb16db..87f950f 100644
@@ -190,8 +190,7 @@ static void flush_channel_fifos(struct most_c_obj *c)
 	list_for_each_entry_safe(mbo, tmp, &c->fifo, list) {
 		list_del(&mbo->list);
 		spin_unlock_irqrestore(&c->fifo_lock, flags);
-		if (likely(mbo))
-			most_free_mbo_coherent(mbo);
+		most_free_mbo_coherent(mbo);
 		spin_lock_irqsave(&c->fifo_lock, flags);
 	}
 	spin_unlock_irqrestore(&c->fifo_lock, flags);
@@ -200,8 +199,7 @@ static void flush_channel_fifos(struct most_c_obj *c)
 	list_for_each_entry_safe(mbo, tmp, &c->halt_fifo, list) {
 		list_del(&mbo->list);
 		spin_unlock_irqrestore(&c->fifo_lock, hf_flags);
-		if (likely(mbo))
-			most_free_mbo_coherent(mbo);
+		most_free_mbo_coherent(mbo);
 		spin_lock_irqsave(&c->fifo_lock, hf_flags);
 	}
 	spin_unlock_irqrestore(&c->fifo_lock, hf_flags);
Signature · 1 line
1.9.1
#8 lore raw

PATCH 5/5 staging: most: style of bool comparison

In reply to Sudip Mukherjee

BOOLEAN tests do not need any comparison to TRUE or FALSE.

---
 drivers/staging/most/hdm-usb/hdm_usb.c | 10 +++++-----
 drivers/staging/most/mostcore/core.c   |  4 ++--
 2 files changed, 7 insertions(+), 7 deletions(-)
drivers/staging/most/hdm-usb/hdm_usb.c +5 -5
index 34843b0..9bbefaa 100644
@@ -308,7 +308,7 @@ static int hdm_poison_channel(struct most_interface *iface, int channel)
 
 	mutex_lock(&mdev->io_mutex);
 	free_anchored_buffers(mdev, channel);
-	if (mdev->padding_active[channel] == true)
+	if (mdev->padding_active[channel])
 		mdev->padding_active[channel] = false;
 
 	if (mdev->conf[channel].data_type == MOST_CH_ASYNC) {
@@ -411,7 +411,7 @@ static void hdm_write_completion(struct urb *urb)
 	dev = &mdev->usb_device->dev;
 
 	if ((urb->status == -ENOENT) || (urb->status == -ECONNRESET) ||
-	    (mdev->is_channel_healthy[channel] == false)) {
+	    (!mdev->is_channel_healthy[channel])) {
 		complete(&anchor->urb_compl);
 		return;
 	}
@@ -576,7 +576,7 @@ static void hdm_read_completion(struct urb *urb)
 	dev = &mdev->usb_device->dev;
 
 	if ((urb->status == -ENOENT) || (urb->status == -ECONNRESET) ||
-	    (mdev->is_channel_healthy[channel] == false)) {
+	    (!mdev->is_channel_healthy[channel])) {
 		complete(&anchor->urb_compl);
 		return;
 	}
@@ -605,7 +605,7 @@ static void hdm_read_completion(struct urb *urb)
 		}
 	} else {
 		mbo->processed_length = urb->actual_length;
-		if (mdev->padding_active[channel] == false) {
+		if (!mdev->padding_active[channel]) {
 			mbo->status = MBO_SUCCESS;
 		} else {
 			if (hdm_remove_padding(mdev, channel, mbo)) {
@@ -685,7 +685,7 @@ static int hdm_enqueue(struct most_interface *iface, int channel, struct mbo *mb
 	list_add_tail(&anchor->list, &mdev->anchor_list[channel]);
 	spin_unlock_irqrestore(&mdev->anchor_list_lock[channel], flags);
 
-	if ((mdev->padding_active[channel] == true) &&
+	if ((mdev->padding_active[channel]) &&
 	    (conf->direction & MOST_CH_TX))
 		if (hdm_add_padding(mdev, channel, mbo)) {
 			retval = -EIO;
drivers/staging/most/mostcore/core.c +2 -2
index 87f950f..eb4e159 100644
@@ -1339,7 +1339,7 @@ static void most_write_completion(struct mbo *mbo)
 	c = mbo->context;
 	if (mbo->status == MBO_E_INVAL)
 		pr_info("WARN: Tx MBO status: invalid\n");
-	if (unlikely((c->is_poisoned == true) || (mbo->status == MBO_E_CLOSE)))
+	if (unlikely(c->is_poisoned || (mbo->status == MBO_E_CLOSE)))
 		trash_mbo(mbo);
 	else
 		arm_mbo(mbo);
@@ -1444,7 +1444,7 @@ static void most_read_completion(struct mbo *mbo)
 	struct most_c_obj *c;
 
 	c = mbo->context;
-	if (unlikely((c->is_poisoned == true) || (mbo->status == MBO_E_CLOSE)))
+	if (unlikely(c->is_poisoned || (mbo->status == MBO_E_CLOSE)))
 		goto release_mbo;
 
 	if (mbo->status == MBO_E_INVAL) {
Signature · 1 line
1.9.1