Skip to contentSkip to conversation

net: netif_set_real_num_rx_queues may cap num_rx_queues at init time

PATCH Applied John Fastabend 5 participants 16 messages Oct 4, 2010 – Oct 7, 2010

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

PATCH 1/1 net: netif_set_real_num_rx_queues may cap num_rx_queues at init time

The logic for netif_set_real_num_rx_queues is the following,

netif_set_real_num_rx_queues(dev, rxq)

{
	...
	if (dev->reg_state == NETREG_REGISTERED) {
		...
	} else {
		dev->num_rx_queues = rxq;
	}

	dev->real_num_rx_queues = rxq;
	return 0;
}

Some drivers init path looks like the following,

alloc_etherdev_mq(priv_sz, max_num_queues_ever);
...
netif_set_real_num_rx_queues(dev, queues_to_use_now);
...
register_netdev(dev);
...

Because netif_set_real_num_rx_queues sets num_rx_queues if the reg state is not NETREG_REGISTERED we end up with the incorrect max number of rx queues. This patch proposes to remove the else clause above so this does not occur. Also just reading the function set_real_num it seems a bit unexpected that num_rx_queues gets set.

CC Ben Hutchings <[email protected]>

---

 net/core/dev.c |    2 --
 1 files changed, 0 insertions(+), 2 deletions(-)
net/core/dev.c +0 -2
index a313bab..f78d996 100644
@@ -1592,8 +1592,6 @@ int netif_set_real_num_rx_queues(struct net_device *dev, unsigned int rxq)
 						  rxq);
 		if (rc)
 			return rc;
-	} else {
-		dev->num_rx_queues = rxq;
 	}
 
 	dev->real_num_rx_queues = rxq;
#1 lore raw
Quoted 24 lines · John Fastabend

The logic for netif_set_real_num_rx_queues is the following,

netif_set_real_num_rx_queues(dev, rxq)

{
	...
	if (dev->reg_state == NETREG_REGISTERED) {
		...
	} else {
		dev->num_rx_queues = rxq;
	}

	dev->real_num_rx_queues = rxq;
	return 0;
}

Some drivers init path looks like the following,

alloc_etherdev_mq(priv_sz, max_num_queues_ever);
...
netif_set_real_num_rx_queues(dev, queues_to_use_now);
...
register_netdev(dev);
...

Quoted 7 lines

Because netif_set_real_num_rx_queues sets num_rx_queues if the reg state is not NETREG_REGISTERED we end up with the incorrect max number of rx queues. This patch proposes to remove the else clause above so this does not occur. Also just reading the function set_real_num it seems a bit unexpected that num_rx_queues gets set.

You dont tell why its "incorrect".

Why should we keep num_rx_queues > real_num_rx_queues ?

It creates /sys files, and Qdisc stuff for nothing...

Quoted 23 lines
CC Ben Hutchings <[email protected]>

---

 net/core/dev.c |    2 --
 1 files changed, 0 insertions(+), 2 deletions(-)
net/core/dev.c +0 -3
index a313bab..f78d996 100644
@@ -1592,8 +1592,6 @@ int netif_set_real_num_rx_queues(struct net_device *dev, unsigned int rxq)
 						  rxq);
 		if (rc)
 			return rc;
-	} else {
-		dev->num_rx_queues = rxq;
 	}
 
 	dev->real_num_rx_queues = rxq;
 
--
#2 lore raw
Quoted 33 lines · Eric Dumazet
Quoted 24 lines · John Fastabend

The logic for netif_set_real_num_rx_queues is the following,

netif_set_real_num_rx_queues(dev, rxq)

{
	...
	if (dev->reg_state == NETREG_REGISTERED) {
		...
	} else {
		dev->num_rx_queues = rxq;
	}

	dev->real_num_rx_queues = rxq;
	return 0;
}

Some drivers init path looks like the following,

alloc_etherdev_mq(priv_sz, max_num_queues_ever);
...
netif_set_real_num_rx_queues(dev, queues_to_use_now);
...
register_netdev(dev);
...

Quoted 7 lines

Because netif_set_real_num_rx_queues sets num_rx_queues if the reg state is not NETREG_REGISTERED we end up with the incorrect max number of rx queues. This patch proposes to remove the else clause above so this does not occur. Also just reading the function set_real_num it seems a bit unexpected that num_rx_queues gets set.

Quoted 2 lines

You dont tell why its "incorrect".

OK that is a poor description.

Quoted 2 lines

Why should we keep num_rx_queues > real_num_rx_queues ?

If we do not ever need them then we should not keep them I agree. But having netif_set_real_num_rx_queues set something other then 'real_num_rx_queues' does not seem right to me at least. Also netif_set_real_num_tx_queues and netif_set_real_num_rx_queues have different behavior. It would be nice if this weren't the case but they allocate queues in two places.

The drivers Ben modified seem to be split between these two flows

alloc_etherdev_mq(priv_sz, max_num_queues_ever); register_netdev(dev); netif_set_real_num_rx_queues(dev, queues_to_use_now);

and

alloc_etherdev_mq(priv_sz, max_num_queues_ever); netif_set_real_num_rx_queues(dev, queues_to_use_now); register_netdev(dev);

In the first case we never set 'num_rx_queues' because its after the register so that leaves the second case. All the remaining drivers except ixgbe, igb, and myri10ge know or could easily find out the number of rx queues at alloc_etherdev_mq and are really trying to explicitly set the num_rx_queues. Adding a num_rx_queues param to alloc_etherdev_mq might make more sense in this case.

The myri10ge is querying the firmware which seems to be tangled up with the netdev priv space, but otherwise isn't using any new knowledge. And ixgbe/igb may end up capping the max number of queues because it is trying to set the number of queues it intends to use now not the max it will ever consume.

My point with this patch is I do not see any cases where we really do not know the max number rx queues until after alloc_etherdev_mq() but before register_netdev(). The piece I missed is if a driver wants to set rx queues and tx queues separately they either need to explicitly set rx_num_queues or use netif_set_real_num_rx_queues before registering the netdev. This syntax seems error prone to me, and it would be better to set this in alloc_etherdev_mq(). But, maybe you disagree?

So I can either go rework the ixgbe/igb init flow so it does what I want. Or add a parameter to alloc_etherdev_mq for rx queues, remove the netif_set_real_num_rx_queues where it is not needed and modify the drivers to use this interface. I like the second case because it makes the netif_set_real_num_[rx|tx}_queues() behave the same but could do the first just as easily.

Thanks,
John

Quoted 29 lines

It creates /sys files, and Qdisc stuff for nothing...

Quoted 23 lines
CC Ben Hutchings <[email protected]>

---

 net/core/dev.c |    2 --
 1 files changed, 0 insertions(+), 2 deletions(-)
net/core/dev.c +0 -3
index a313bab..f78d996 100644
@@ -1592,8 +1592,6 @@ int netif_set_real_num_rx_queues(struct net_device *dev, unsigned int rxq)
 						  rxq);
 		if (rc)
 			return rc;
-	} else {
-		dev->num_rx_queues = rxq;
 	}
 
 	dev->real_num_rx_queues = rxq;
 
--
#3 lore raw
Quoted 42 lines · John Fastabend
Quoted 33 lines · Eric Dumazet
Quoted 24 lines · John Fastabend

The logic for netif_set_real_num_rx_queues is the following,

netif_set_real_num_rx_queues(dev, rxq)

{
	...
	if (dev->reg_state == NETREG_REGISTERED) {
		...
	} else {
		dev->num_rx_queues = rxq;
	}

	dev->real_num_rx_queues = rxq;
	return 0;
}

Some drivers init path looks like the following,

alloc_etherdev_mq(priv_sz, max_num_queues_ever);
...
netif_set_real_num_rx_queues(dev, queues_to_use_now);
...
register_netdev(dev);
...

Quoted 7 lines

Because netif_set_real_num_rx_queues sets num_rx_queues if the reg state is not NETREG_REGISTERED we end up with the incorrect max number of rx queues. This patch proposes to remove the else clause above so this does not occur. Also just reading the function set_real_num it seems a bit unexpected that num_rx_queues gets set.

Quoted 2 lines

You dont tell why its "incorrect".

OK that is a poor description.

Quoted 2 lines

Why should we keep num_rx_queues > real_num_rx_queues ?

Quoted 6 lines

If we do not ever need them then we should not keep them I agree. But having netif_set_real_num_rx_queues set something other then 'real_num_rx_queues' does not seem right to me at least. Also netif_set_real_num_tx_queues and netif_set_real_num_rx_queues have different behavior. It would be nice if this weren't the case but they allocate queues in two places.

[...]

I only did this to satisfy Eric's desire to reduce memory usage. However, I believe that there are currently no drivers that dynamically increase numbers of RX or TX queues. Until there are, there is not much point in removing this assignment to num_rx_queues.

Ben.

Signature · 3 lines
Ben Hutchings, Senior Software Engineer, Solarflare Communications
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
#4 lore raw
Quoted 49 lines · Ben Hutchings
Quoted 42 lines · John Fastabend
Quoted 33 lines · Eric Dumazet
Quoted 24 lines · John Fastabend

The logic for netif_set_real_num_rx_queues is the following,

netif_set_real_num_rx_queues(dev, rxq)

{
	...
	if (dev->reg_state == NETREG_REGISTERED) {
		...
	} else {
		dev->num_rx_queues = rxq;
	}

	dev->real_num_rx_queues = rxq;
	return 0;
}

Some drivers init path looks like the following,

alloc_etherdev_mq(priv_sz, max_num_queues_ever);
...
netif_set_real_num_rx_queues(dev, queues_to_use_now);
...
register_netdev(dev);
...

Quoted 7 lines

Because netif_set_real_num_rx_queues sets num_rx_queues if the reg state is not NETREG_REGISTERED we end up with the incorrect max number of rx queues. This patch proposes to remove the else clause above so this does not occur. Also just reading the function set_real_num it seems a bit unexpected that num_rx_queues gets set.

Quoted 2 lines

You dont tell why its "incorrect".

OK that is a poor description.

Quoted 2 lines

Why should we keep num_rx_queues > real_num_rx_queues ?

Quoted 6 lines

If we do not ever need them then we should not keep them I agree. But having netif_set_real_num_rx_queues set something other then 'real_num_rx_queues' does not seem right to me at least. Also netif_set_real_num_tx_queues and netif_set_real_num_rx_queues have different behavior. It would be nice if this weren't the case but they allocate queues in two places.

Quoted 9 lines

[...]

I only did this to satisfy Eric's desire to reduce memory usage. However, I believe that there are currently no drivers that dynamically increase numbers of RX or TX queues. Until there are, there is not much point in removing this assignment to num_rx_queues.

Ben.

ixgbe increases the real_num_[rx|tx]_queues when FCoE or DCB is enabled. Also many of the drivers could increase the number of queues if they were given more interrupt vectors at some point.

But, it is easy enough to patch ixgbe to not set the number of RX queues currently in use until after the device is registered. Which brings it inline with many of the other drivers. And then the drivers that are using it to explicitly set num_rx_queues do not need to change.

Thanks,
John

#5 lore raw
Quoted 57 lines · John Fastabend
Quoted 56 lines · Ben Hutchings
Quoted 42 lines · John Fastabend
Quoted 33 lines · Eric Dumazet
Quoted 24 lines · John Fastabend

The logic for netif_set_real_num_rx_queues is the following,

netif_set_real_num_rx_queues(dev, rxq)

{
	...
	if (dev->reg_state == NETREG_REGISTERED) {
		...
	} else {
		dev->num_rx_queues = rxq;
	}

	dev->real_num_rx_queues = rxq;
	return 0;
}

Some drivers init path looks like the following,

alloc_etherdev_mq(priv_sz, max_num_queues_ever);
...
netif_set_real_num_rx_queues(dev, queues_to_use_now);
...
register_netdev(dev);
...

Quoted 7 lines

Because netif_set_real_num_rx_queues sets num_rx_queues if the reg state is not NETREG_REGISTERED we end up with the incorrect max number of rx queues. This patch proposes to remove the else clause above so this does not occur. Also just reading the function set_real_num it seems a bit unexpected that num_rx_queues gets set.

Quoted 2 lines

You dont tell why its "incorrect".

OK that is a poor description.

Quoted 2 lines

Why should we keep num_rx_queues > real_num_rx_queues ?

Quoted 6 lines

If we do not ever need them then we should not keep them I agree. But having netif_set_real_num_rx_queues set something other then 'real_num_rx_queues' does not seem right to me at least. Also netif_set_real_num_tx_queues and netif_set_real_num_rx_queues have different behavior. It would be nice if this weren't the case but they allocate queues in two places.

[...]

I only did this to satisfy Eric's desire to reduce memory usage. However, I believe that there are currently no drivers that dynamically increase numbers of RX or TX queues. Until there are, there is not much point in removing this assignment to num_rx_queues.

Quoted 6 lines
Quoted 2 lines

Ben.

ixgbe increases the real_num_[rx|tx]_queues when FCoE or DCB is enabled. Also many of the drivers could increase the number of queues if they were given more interrupt vectors at some point.

If I update the handful drivers that use netif_set_real_num_rx_queues() before the netdevice is registered to explicitly set num_rx_queues this would address Eric's concerns and fix drivers that really only want to set real_num_rx_queue.

Any thoughts?

-- John

#6 lore raw
Quoted 4 lines · John Fastabend
Quoted 3 lines · John Fastabend
Quoted 2 lines · Ben Hutchings
Quoted 1 line · John Fastabend

On 10/4/2010 10:35 PM, Eric Dumazet wrote:

[...]

Quoted 24 lines
Quoted 16 lines
Quoted 16 lines
Quoted 9 lines
Quoted 2 lines

Why should we keep num_rx_queues > real_num_rx_queues ?

If we do not ever need them then we should not keep them I agree. But having netif_set_real_num_rx_queues set something other then 'real_num_rx_queues' does not seem right to me at least. Also netif_set_real_num_tx_queues and netif_set_real_num_rx_queues have different behavior. It would be nice if this weren't the case but they allocate queues in two places.

[...]

I only did this to satisfy Eric's desire to reduce memory usage. However, I believe that there are currently no drivers that dynamically increase numbers of RX or TX queues. Until there are, there is not much point in removing this assignment to num_rx_queues.

Quoted 6 lines
Quoted 2 lines

Ben.

ixgbe increases the real_num_[rx|tx]_queues when FCoE or DCB is enabled. Also many of the drivers could increase the number of queues if they were given more interrupt vectors at some point.

Quoted 6 lines

If I update the handful drivers that use netif_set_real_num_rx_queues() before the netdevice is registered to explicitly set num_rx_queues this would address Eric's concerns and fix drivers that really only want to set real_num_rx_queue.

Any thoughts?

Don't add assignments to num_rx_queues. If it's useful to increase the number of RX queues later then just remove the assignment to num_rx_queues from netif_set_real_num_rx_queues() and be done with it. The waste of memory is minimal now that we only allocate kobjects for real_num_rx_queues.

Ben.

Signature · 3 lines
Ben Hutchings, Senior Software Engineer, Solarflare Communications
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
#7 lore raw
Quoted 37 lines · Ben Hutchings
Quoted 4 lines · John Fastabend
Quoted 3 lines · John Fastabend
Quoted 2 lines · Ben Hutchings
Quoted 1 line · John Fastabend

On 10/4/2010 10:35 PM, Eric Dumazet wrote:

[...]

Quoted 24 lines
Quoted 16 lines
Quoted 16 lines
Quoted 9 lines
Quoted 2 lines

Why should we keep num_rx_queues > real_num_rx_queues ?

If we do not ever need them then we should not keep them I agree. But having netif_set_real_num_rx_queues set something other then 'real_num_rx_queues' does not seem right to me at least. Also netif_set_real_num_tx_queues and netif_set_real_num_rx_queues have different behavior. It would be nice if this weren't the case but they allocate queues in two places.

[...]

I only did this to satisfy Eric's desire to reduce memory usage. However, I believe that there are currently no drivers that dynamically increase numbers of RX or TX queues. Until there are, there is not much point in removing this assignment to num_rx_queues.

Quoted 6 lines
Quoted 2 lines

Ben.

ixgbe increases the real_num_[rx|tx]_queues when FCoE or DCB is enabled. Also many of the drivers could increase the number of queues if they were given more interrupt vectors at some point.

Quoted 6 lines

If I update the handful drivers that use netif_set_real_num_rx_queues() before the netdevice is registered to explicitly set num_rx_queues this would address Eric's concerns and fix drivers that really only want to set real_num_rx_queue.

Any thoughts?

Quoted 8 lines

Don't add assignments to num_rx_queues. If it's useful to increase the number of RX queues later then just remove the assignment to num_rx_queues from netif_set_real_num_rx_queues() and be done with it. The waste of memory is minimal now that we only allocate kobjects for real_num_rx_queues.

Ben.

OK Thanks Ben. I will get a better description on this and resend it.

John.

#8 lore raw
In reply to Ben Hutchings
Quoted 2 lines · Ben Hutchings

The waste of memory is minimal now that we only allocate kobjects for real_num_rx_queues.

Thats strange, here with tg3 (and mono queue device) :

0a:04.0 Ethernet controller: Broadcom Corporation NetXtreme BCM5715S Gigabit Ethernet (rev a3)

grep . /sys/class/net/eth2/queues/rx-*/rps_cpus
/sys/class/net/eth2/queues/rx-0/rps_cpus:00000000
/sys/class/net/eth2/queues/rx-1/rps_cpus:00000000
/sys/class/net/eth2/queues/rx-2/rps_cpus:00000000
/sys/class/net/eth2/queues/rx-3/rps_cpus:00000000
/sys/class/net/eth2/queues/rx-4/rps_cpus:00000000

#9 lore raw
Quoted 10 lines · Eric Dumazet
Quoted 2 lines · Ben Hutchings

The waste of memory is minimal now that we only allocate kobjects for real_num_rx_queues.

Thats strange, here with tg3 (and mono queue device) :

0a:04.0 Ethernet controller: Broadcom Corporation NetXtreme BCM5715S Gigabit Ethernet (rev a3)

Quoted 6 lines

grep . /sys/class/net/eth2/queues/rx-*/rps_cpus
/sys/class/net/eth2/queues/rx-0/rps_cpus:00000000
/sys/class/net/eth2/queues/rx-1/rps_cpus:00000000
/sys/class/net/eth2/queues/rx-2/rps_cpus:00000000
/sys/class/net/eth2/queues/rx-3/rps_cpus:00000000
/sys/class/net/eth2/queues/rx-4/rps_cpus:00000000

It looks like I missed a necessary call to netif_set_real_num_rx_queues() in tg3. I suggest that Matt should check and correct this since I got the numbers wrong last time.

Ben.

Signature · 3 lines
Ben Hutchings, Senior Software Engineer, Solarflare Communications
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.
#10 lore raw
Quoted 18 lines · Ben Hutchings
Quoted 10 lines · Eric Dumazet

Le mercredi 06 octobre 2010 ?? 16:07 +0100, Ben Hutchings a ??crit :

Quoted 2 lines

The waste of memory is minimal now that we only allocate kobjects for real_num_rx_queues.

Thats strange, here with tg3 (and mono queue device) :

0a:04.0 Ethernet controller: Broadcom Corporation NetXtreme BCM5715S Gigabit Ethernet (rev a3)

Quoted 6 lines

grep . /sys/class/net/eth2/queues/rx-*/rps_cpus
/sys/class/net/eth2/queues/rx-0/rps_cpus:00000000
/sys/class/net/eth2/queues/rx-1/rps_cpus:00000000
/sys/class/net/eth2/queues/rx-2/rps_cpus:00000000
/sys/class/net/eth2/queues/rx-3/rps_cpus:00000000
/sys/class/net/eth2/queues/rx-4/rps_cpus:00000000

Quoted 3 lines

It looks like I missed a necessary call to netif_set_real_num_rx_queues() in tg3. I suggest that Matt should check and correct this since I got the numbers wrong last time.

Yes. We were missing a call to this function in the legacy case.

[PATCH net-next] tg3: Set real_num_rx_queues for non-multiq devs

Commit 2ddaad397c47de012dfb956b0c05540da1a0dde5 entitled "tg3: Use netif_set_real_num_{rx,tx}_queues()" added a new call to netif_set_real_num_rx_queues in tg3_enable_msix(). This call also needs to be added to the legacy path to correctly reflect the actual number of rx queues.

---
 drivers/net/tg3.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
drivers/net/tg3.c +1 -0
index 16e1a95..e5b9ec5 100644
@@ -8906,6 +8906,7 @@ defcfg:
 		tp->irq_cnt = 1;
 		tp->napi[0].irq_vec = tp->pdev->irq;
 		netif_set_real_num_tx_queues(tp->dev, 1);
+		netif_set_real_num_rx_queues(tp->dev, 1);
 	}
 }
 
Signature · 1 line
1.7.2.2
Acked-by
#11 lore raw
Quoted 21 lines · Matt Carlson

Yes. We were missing a call to this function in the legacy case.

[PATCH net-next] tg3: Set real_num_rx_queues for non-multiq devs

Commit 2ddaad397c47de012dfb956b0c05540da1a0dde5 entitled "tg3: Use netif_set_real_num_{rx,tx}_queues()" added a new call to netif_set_real_num_rx_queues in tg3_enable_msix(). This call also needs to be added to the legacy path to correctly reflect the actual number of rx queues.

---
 drivers/net/tg3.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
drivers/net/tg3.c +0 -0
index 16e1a95..e5b9ec5 100644

drivers/net/tg3.c defcfg: · line 8912

Quoted 7 lines
drivers/net/tg3.c +1 -0
@@ -8906,6 +8906,7 @@ defcfg:
 		tp->irq_cnt = 1;
 		tp->napi[0].irq_vec = tp->pdev->irq;
 		netif_set_real_num_tx_queues(tp->dev, 1);
+		netif_set_real_num_rx_queues(tp->dev, 1);
 	}
 }
 

Thanks, this fixed the thing for me, once device is UP

# insmod drivers/net/tg3.ko
# grep . /sys/class/net/eth3/queues/*/rps_cpus
/sys/class/net/eth3/queues/rx-0/rps_cpus:00000000
/sys/class/net/eth3/queues/rx-1/rps_cpus:00000000
/sys/class/net/eth3/queues/rx-2/rps_cpus:00000000
/sys/class/net/eth3/queues/rx-3/rps_cpus:00000000
/sys/class/net/eth3/queues/rx-4/rps_cpus:00000000
# echo ff >/sys/class/net/eth3/queues/rx-2/rps_cpus
# grep . /sys/class/net/eth3/queues/*/rps_cpus
/sys/class/net/eth3/queues/rx-0/rps_cpus:00000000
/sys/class/net/eth3/queues/rx-1/rps_cpus:00000000
/sys/class/net/eth3/queues/rx-2/rps_cpus:000000ff
/sys/class/net/eth3/queues/rx-3/rps_cpus:00000000
/sys/class/net/eth3/queues/rx-4/rps_cpus:00000000
# ifconfig eth3 up
# grep . /sys/class/net/eth3/queues/*/rps_cpus
00000000
maintainer Applied
#12 lore raw

From: Eric Dumazet <[email protected]>
Date: Wed, 06 Oct 2010 20:49:05 +0200

Quoted 14 lines · Matt Carlson
Quoted 13 lines

Yes. We were missing a call to this function in the legacy case.

[PATCH net-next] tg3: Set real_num_rx_queues for non-multiq devs

Commit 2ddaad397c47de012dfb956b0c05540da1a0dde5 entitled "tg3: Use netif_set_real_num_{rx,tx}_queues()" added a new call to netif_set_real_num_rx_queues in tg3_enable_msix(). This call also needs to be added to the legacy path to correctly reflect the actual number of rx queues.

...

Quoted 1 line

Applied, thanks everyone.

#13 lore raw
In reply to John Fastabend
Quoted 5 lines · John Fastabend

If I update the handful drivers that use netif_set_real_num_rx_queues() before the netdevice is registered to explicitly set num_rx_queues this would address Eric's concerns and fix drivers that really only want to set real_num_rx_queue.

John, it seems current API is not very clean/orthogonal.

If you believe a driver can increase real_num_rx_queue, then we should apply your patch, and refine API if necessary.

We now allocate rx queues in register_netdevice(), maybe we should do the same for tx queues.

Acked-by on v1 1/1
#14 lore raw
In reply to John Fastabend
Quoted 2 lines · John Fastabend

The logic for netif_set_real_num_rx_queues is the following,

...

Quoted 11 lines

Because netif_set_real_num_rx_queues sets num_rx_queues if the reg state is not NETREG_REGISTERED we end up with the incorrect max number of rx queues. This patch proposes to remove the else clause above so this does not occur. Also just reading the function set_real_num it seems a bit unexpected that num_rx_queues gets set.

CC Ben Hutchings <[email protected]>

---

maintainer Applied
#15 lore raw

From: Eric Dumazet <[email protected]>
Date: Thu, 07 Oct 2010 08:16:39 +0200

Quoted 14 lines · John Fastabend
Quoted 2 lines

The logic for netif_set_real_num_rx_queues is the following,

...

Quoted 7 lines

Because netif_set_real_num_rx_queues sets num_rx_queues if the reg state is not NETREG_REGISTERED we end up with the incorrect max number of rx queues. This patch proposes to remove the else clause above so this does not occur. Also just reading the function set_real_num it seems a bit unexpected that num_rx_queues gets set.

Quoted 6 lines
Quoted 4 lines
CC Ben Hutchings <[email protected]>

---

Applied, thanks everyone.