Skip to contentSkip to conversation

net: dsa: mv88e6xxx: features to handle network storms

PATCH In review Robert Beckett 7 participants 42 messages Sep 10, 2019 – Sep 14, 2019

Subsystems (5)
lore
Share & tools
lore
#0 lore raw

cover net: dsa: mv88e6xxx: features to handle network storms

This patch-set adds support for some features of the Marvell switch chips that can be used to handle packet storms.

The rationale for this was a setup that requires the ability to receive traffic from one port, while a packet storm is occuring on another port (via an external switch with a deliberate loop). This is needed to ensure vital data delivery from a specific port, while mitigating any loops or DoS that a user may introduce on another port (can't guarantee sensible users).

[patch 1/7] configures auto negotiation for CPU ports connected with phys to enable pause frame propogation.

[patch 2/7] allows setting of port's default output queue priority for any ingressing packets on that port.

[patch 3/7] dt-bindings for patch 2.

[patch 4/7] allows setting of a port's queue scheduling so that it can prioritise egress of traffic routed from high priority ports.

[patch 5/7] dt-bindings for patch 4.

[patch 6/7] allows ports to rate limit their egress. This can be used to stop the host CPU from becoming swamped by packet delivery and exhasting descriptors.

[patch 7/7] dt-bindings for patch 6.

Robert Beckett (7):
net/dsa: configure autoneg for CPU port
net: dsa: mv88e6xxx: add ability to set default queue priorities per
port
dt-bindings: mv88e6xxx: add ability to set default queue priorities
per port
net: dsa: mv88e6xxx: add ability to set queue scheduling
dt-bindings: mv88e6xxx: add ability to set queue scheduling
net: dsa: mv88e6xxx: add egress rate limiting
dt-bindings: mv88e6xxx: add egress rate limiting

 .../devicetree/bindings/net/dsa/marvell.txt   |  38 +++++
 drivers/net/dsa/mv88e6xxx/chip.c              | 122 ++++++++++++---
 drivers/net/dsa/mv88e6xxx/chip.h              |   5 +-
 drivers/net/dsa/mv88e6xxx/port.c              | 140 +++++++++++++++++-
 drivers/net/dsa/mv88e6xxx/port.h              |  24 ++-
 include/dt-bindings/net/dsa-mv88e6xxx.h       |  22 +++
 net/dsa/port.c                                |  10 ++
 7 files changed, 327 insertions(+), 34 deletions(-)
 create mode 100644 include/dt-bindings/net/dsa-mv88e6xxx.h
Signature · 1 line
2.18.0
#1 lore raw

PATCH 1/7 net/dsa: configure autoneg for CPU port

Configure autoneg for phy connected CPU ports.
This allows us to use autoneg between the CPU port's phy and the link partner's phy.
This enables us to negoatiate pause frame transmission to prioritise packet delivery over throughput.

---
 net/dsa/port.c | 10 ++++++++++
 1 file changed, 10 insertions(+)
net/dsa/port.c +10 -0
index f071acf2842b..1b6832eac2c5 100644
@@ -538,10 +538,20 @@ static int dsa_port_setup_phy_of(struct dsa_port *dp, bool enable)
 		return PTR_ERR(phydev);
 
 	if (enable) {
+		phydev->supported = PHY_GBIT_FEATURES | SUPPORTED_MII |
+				    SUPPORTED_AUI | SUPPORTED_FIBRE |
+				    SUPPORTED_BNC | SUPPORTED_Pause |
+				    SUPPORTED_Asym_Pause;
+		phydev->advertising = phydev->supported;
+
 		err = genphy_config_init(phydev);
 		if (err < 0)
 			goto err_put_dev;
 
+		err = genphy_config_aneg(phydev);
+		if (err < 0)
+			goto err_put_dev;
+
 		err = genphy_resume(phydev);
 		if (err < 0)
 			goto err_put_dev;
Signature · 1 line
2.18.0
#2 lore raw

Re: [PATCH 1/7] net/dsa: configure autoneg for CPU port

Hi Robert,

Quoted 16 lines · Robert Beckett

Configure autoneg for phy connected CPU ports.
This allows us to use autoneg between the CPU port's phy and the link partner's phy.
This enables us to negoatiate pause frame transmission to prioritise packet delivery over throughput.

---
 net/dsa/port.c | 10 ++++++++++
 1 file changed, 10 insertions(+)
net/dsa/port.c +0 -0
index f071acf2842b..1b6832eac2c5 100644

net/dsa/port.c static int dsa_port_setup_phy_of(struct dsa_port *dp, bool enable) · line 546

Quoted 9 lines
net/dsa/port.c +6 -0
@@ -538,10 +538,20 @@ static int dsa_port_setup_phy_of(struct dsa_port *dp, bool enable)
 		return PTR_ERR(phydev);
 
 	if (enable) {
+		phydev->supported = PHY_GBIT_FEATURES | SUPPORTED_MII |
+				    SUPPORTED_AUI | SUPPORTED_FIBRE |
+				    SUPPORTED_BNC | SUPPORTED_Pause |
+				    SUPPORTED_Asym_Pause;
+		phydev->advertising = phydev->supported;
+

This seems a bit intruisive to me. I'll get back to you.

Quoted 11 lines
net/dsa/port.c +4 -0
 		err = genphy_config_init(phydev);
 		if (err < 0)
 			goto err_put_dev;
 
+		err = genphy_config_aneg(phydev);
+		if (err < 0)
+			goto err_put_dev;
+
 		err = genphy_resume(phydev);
 		if (err < 0)
 			goto err_put_dev;
#3 lore raw

Re: [PATCH 1/7] net/dsa: configure autoneg for CPU port

In reply to Robert Beckett
Quoted 5 lines · Robert Beckett

Configure autoneg for phy connected CPU ports.
This allows us to use autoneg between the CPU port's phy and the link partner's phy.
This enables us to negoatiate pause frame transmission to prioritise packet delivery over throughput.

s/autoneg/auto-negotiation/
s/phy/PHY/
s/negoatiate/negotiate/
s/prioritise/prioritize/ (maybe the latter is just my US english dictionary tripping up)

Also the subject should be net: dsa: Configure auto-negotiation for CPU port to match previous submissions done to that file.

Fixing up that code path sounds reasonable, but are you not hitting the PHYLINK code path instead?

Quoted 32 lines
---
 net/dsa/port.c | 10 ++++++++++
 1 file changed, 10 insertions(+)
net/dsa/port.c +10 -0
index f071acf2842b..1b6832eac2c5 100644
@@ -538,10 +538,20 @@ static int dsa_port_setup_phy_of(struct dsa_port *dp, bool enable)
 		return PTR_ERR(phydev);
 
 	if (enable) {
+		phydev->supported = PHY_GBIT_FEATURES | SUPPORTED_MII |
+				    SUPPORTED_AUI | SUPPORTED_FIBRE |
+				    SUPPORTED_BNC | SUPPORTED_Pause |
+				    SUPPORTED_Asym_Pause;
+		phydev->advertising = phydev->supported;
+
 		err = genphy_config_init(phydev);
 		if (err < 0)
 			goto err_put_dev;
 
+		err = genphy_config_aneg(phydev);
+		if (err < 0)
+			goto err_put_dev;
+
 		err = genphy_resume(phydev);
 		if (err < 0)
 			goto err_put_dev;
Signature · 1 line
Florian
maintainer
#4 lore raw

Re: [PATCH 1/7] net/dsa: configure autoneg for CPU port

In reply to Robert Beckett
Quoted 2 lines · Robert Beckett

This enables us to negoatiate pause frame transmission to prioritise packet delivery over throughput.

I don't think we can unconditionally enable this. It is a big behaviour change, and it is likely to break running systems. It has affects on QoS, packet prioritisation, etc.

I think there needs to be a configuration knob. But unfortunately, i don't know of a good place to put this knob. The switch CPU port is not visible in any way.

Andrew

#5 lore raw

Re: [PATCH 1/7] net/dsa: configure autoneg for CPU port

Quoted 11 lines · Andrew Lunn
Quoted 2 lines · Robert Beckett

This enables us to negoatiate pause frame transmission to prioritise packet delivery over throughput.

I don't think we can unconditionally enable this. It is a big behaviour change, and it is likely to break running systems. It has affects on QoS, packet prioritisation, etc.

I think there needs to be a configuration knob. But unfortunately, i don't know of a good place to put this knob. The switch CPU port is not visible in any way.

Broadcast storm suppression is to be solved at ingress, not on the CPU port, once this lands on the CPU port, it's game over already.

Signature · 1 line
Florian
#6 lore raw

Re: [PATCH 1/7] net/dsa: configure autoneg for CPU port

Quoted 10 lines · Florian Fainelli
Quoted 9 lines · Andrew Lunn
Quoted 3 lines · Robert Beckett

This enables us to negoatiate pause frame transmission to prioritise
packet delivery over throughput.

I don't think we can unconditionally enable this. It is a big behaviour change, and it is likely to break running systems. It has affects on QoS, packet prioritisation, etc.

Quoted 8 lines
Quoted 4 lines

I think there needs to be a configuration knob. But unfortunately, i
don't know of a good place to put this knob. The switch CPU port is not visible in any way.

Broadcast storm suppression is to be solved at ingress, not on the CPU
port, once this lands on the CPU port, it's game over already.

It is not just for broadcast storm protection. The original issue that made me look in to all of this turned out to be rx descritor ring buffer exhaustion due to the CPU not being able to keep up with packet reception.

Although the simple repro case for it is a broadcast storm, this could happen with many legitimate small packets, and the correct way to handle it seems to be pause frames, though I am not traditionally a network programmer, so my knowledge may be incorrect. Please advise if you know of a better way to handle that.

Fundamentally, with a phy to phy CPU connection, the CPU MAC may well wish to enable pause frames for various reasons, so we should strive to handle that I think.

#7 lore raw

Re: [PATCH 1/7] net/dsa: configure autoneg for CPU port

Quoted 38 lines · Robert Beckett
Quoted 18 lines · Florian Fainelli
Quoted 10 lines · Andrew Lunn
Quoted 3 lines · Robert Beckett

This enables us to negoatiate pause frame transmission to prioritise
packet delivery over throughput.

I don't think we can unconditionally enable this. It is a big behaviour change, and it is likely to break running systems. It has
affects on QoS, packet prioritisation, etc.

Quoted 6 lines

I think there needs to be a configuration knob. But unfortunately,
i
don't know of a good place to put this knob. The switch CPU port is
not visible in any way.

Quoted 3 lines

Broadcast storm suppression is to be solved at ingress, not on the CPU
port, once this lands on the CPU port, it's game over already.

It is not just for broadcast storm protection. The original issue that
made me look in to all of this turned out to be rx descritor ring buffer exhaustion due to the CPU not being able to keep up with packet
reception.

Although the simple repro case for it is a broadcast storm, this could
happen with many legitimate small packets, and the correct way to handle it seems to be pause frames, though I am not traditionally a network programmer, so my knowledge may be incorrect. Please advise if
you know of a better way to handle that.

Quoted 5 lines

Fundamentally, with a phy to phy CPU connection, the CPU MAC may well wish to enable pause frames for various reasons, so we should strive to
handle that I think.

As an aside, do any of you have experience of trying to enable PIRL on the Marvell switches? The first thing I tried was configuring it for packet number based (rather than byte count based) input rate limiting, but it never seemed to have any effect even at extreme values that should in theory have greatly limited the number of packets allowed to ingress.

After investigating the root cause and finding it was due to the CPU's inability to process the received packets quickly enough, pause frames and port prioritization seemed like the correct fix anyway.

maintainer
#8 lore raw

Re: [PATCH 1/7] net/dsa: configure autoneg for CPU port

In reply to Robert Beckett
Quoted 4 lines

It is not just for broadcast storm protection. The original issue that made me look in to all of this turned out to be rx descritor ring buffer exhaustion due to the CPU not being able to keep up with packet reception.

Pause frames does not really solve this problem. The switch will at some point fill its buffers, and start throwing packets away. Or it needs to send pause packets it its peers. And then your whole switch throughput goes down. Packets will always get thrown away, so you need QoS in your network to give the network hints about which frames is should throw away first.

..

Quoted 3 lines

Fundamentally, with a phy to phy CPU connection, the CPU MAC may well wish to enable pause frames for various reasons, so we should strive to handle that I think.

It actually has nothing to do with PHY to PHY connections. You can use pause frames with direct MAC to MAC connections. PHY auto-negotiation is one way to indicate both ends support it, but there are also other ways. e.g.

ethtool -A|--pause devname [autoneg on|off] [rx on|off] [tx on|off]

on the SoC you could do

ethtool --pause eth0 autoneg off rx on tx on

to force the SoC to send and process pause frames. Ideally i would prefer a solution like this, since it is not a change of behaviour for everybody else.

Andrew

#9 lore raw

Re: [PATCH 1/7] net/dsa: configure autoneg for CPU port

Quoted 7 lines · Andrew Lunn
Quoted 6 lines

It is not just for broadcast storm protection. The original issue that
made me look in to all of this turned out to be rx descritor ring buffer exhaustion due to the CPU not being able to keep up with packet
reception.

Quoted 8 lines

Pause frames does not really solve this problem. The switch will at some point fill its buffers, and start throwing packets away. Or it needs to send pause packets it its peers. And then your whole switch throughput goes down. Packets will always get thrown away, so you need
QoS in your network to give the network hints about which frames is should throw away first.

Indeed. This is the understanding I was working with.
This patch series enables pause frames, output queue prriority and strict scheduling to egress the high priority queues first. This means that when the switch starts dropping frames, it drops from the lowest priority as the highest ones are delivered at line speed without issue.

Quoted 16 lines

..

Quoted 5 lines

Fundamentally, with a phy to phy CPU connection, the CPU MAC may well
wish to enable pause frames for various reasons, so we should strive to
handle that I think.

It actually has nothing to do with PHY to PHY connections. You can use
pause frames with direct MAC to MAC connections. PHY auto-negotiation is one way to indicate both ends support it, but there are also other ways. e.g.

ethtool -A|--pause devname [autoneg on|off] [rx on|off] [tx on|off]

Quoted 8 lines

on the SoC you could do

ethtool --pause eth0 autoneg off rx on tx on

to force the SoC to send and process pause frames. Ideally i would prefer a solution like this, since it is not a change of behaviour for
everybody else.

Good point, well made.
The reason for using autoneg in this series was due to having no netdev to run ethtool against for the CPU port.
If we go down the route of creating a netdev for the CPU port, then we could indeed force pause frames at both ends.

However, given that the phy on the marvell switch is capable of autoneg , is it not reasonable to setup the advertisement and let autoneg take care of it if using phy to phy connection?

Quoted 2 lines

Andrew

maintainer
#10 lore raw

Re: [PATCH 1/7] net/dsa: configure autoneg for CPU port

Quoted 12 lines
Quoted 12 lines

It actually has nothing to do with PHY to PHY connections. You can use
pause frames with direct MAC to MAC connections. PHY auto-negotiation is one way to indicate both ends support it, but there are also other ways. e.g.

ethtool -A|--pause devname [autoneg on|off] [rx on|off] [tx on|off]

on the SoC you could do

ethtool --pause eth0 autoneg off rx on tx on

Quoted 8 lines
Quoted 4 lines

to force the SoC to send and process pause frames. Ideally i would prefer a solution like this, since it is not a change of behaviour for
everybody else.

Good point, well made.
The reason for using autoneg in this series was due to having no netdev to run ethtool against for the CPU port.

Do you need one? It is the IMX which is the bottle neck. It is the one which needs to send pause frames. You have a netdev for that. Have you checked if the switch will react on pause frames without your change. Play with the command i give above on the master interface. It looks like the FEC driver fully supports synchronous pause configuration.

Quoted 3 lines

However, given that the phy on the marvell switch is capable of autoneg , is it not reasonable to setup the advertisement and let autoneg take care of it if using phy to phy connection?

Most designs don't use back to back PHYs for the CPU port. They save the cost and connect MACs back to back using RGMII, or maybe SERDES. If we are going for a method which can configure pause between the CPU and the switch, it needs to be generic and work for both setups.

Andrew

#11 lore raw

Re: [PATCH 1/7] net/dsa: configure autoneg for CPU port

In reply to Robert Beckett
Thank you for the patch! Yet something to improve:

Hi Robert,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on linus/master]
[cannot apply to v5.3-rc8 next-20190904]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]

url: https://github.com/0day-ci/linux/commits/Robert-Beckett/net-dsa-mv88e6xxx-features-to-handle-network-storms/20190911-142233
config: mips-allmodconfig (attached as .config)
compiler: mips-linux-gcc (GCC) 7.4.0
reproduce:
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross
# save the attached .config to linux build tree
GCC_VERSION=7.4.0 make.cross ARCH=mips

If you fix the issue, kindly add following tag

All errors (new ones prefixed by >>):

net//dsa/port.c: In function 'dsa_port_setup_phy_of':

>> net//dsa/port.c:541:41: error: invalid operands to binary | (have 'long unsigned int *' and 'long unsigned int')
      phydev->supported = PHY_GBIT_FEATURES | SUPPORTED_MII |
                                            ^
>> net//dsa/port.c:545:23: error: assignment to expression with array type
      phydev->advertising = phydev->supported;
                          ^

vim +541 net//dsa/port.c

525
526 static int dsa_port_setup_phy_of(struct dsa_port *dp, bool enable)
527 {
528 struct dsa_switch *ds = dp->ds;
529 struct phy_device *phydev;
530 int port = dp->index;
531 int err = 0;
532
533 phydev = dsa_port_get_phy_device(dp);
534 if (!phydev)
535 return 0;
536
537 if (IS_ERR(phydev))
538 return PTR_ERR(phydev);
539
540 if (enable) {

 > 541			phydev->supported = PHY_GBIT_FEATURES | SUPPORTED_MII |
   542					    SUPPORTED_AUI | SUPPORTED_FIBRE |
   543					    SUPPORTED_BNC | SUPPORTED_Pause |
   544					    SUPPORTED_Asym_Pause;
 > 545			phydev->advertising = phydev->supported;

546
547 err = genphy_config_init(phydev);
548 if (err < 0)
549 goto err_put_dev;
550
551 err = genphy_config_aneg(phydev);
552 if (err < 0)
553 goto err_put_dev;
554
555 err = genphy_resume(phydev);
556 if (err < 0)
557 goto err_put_dev;
558
559 err = genphy_read_status(phydev);
560 if (err < 0)
561 goto err_put_dev;
562 } else {
563 err = genphy_suspend(phydev);
564 if (err < 0)
565 goto err_put_dev;
566 }
567
568 if (ds->ops->adjust_link)
569 ds->ops->adjust_link(ds, port, phydev);
570
571 dev_dbg(ds->dev, "enabled port's phy: %s", phydev_name(phydev));
572
573 err_put_dev:
574 put_device(&phydev->mdio.dev);
575 return err;
576 }
577

---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation

#12 lore raw

Re: [PATCH 1/7] net/dsa: configure autoneg for CPU port

In reply to Robert Beckett
Thank you for the patch! Perhaps something to improve:

Hi Robert,

Thank you for the patch! Perhaps something to improve:

[auto build test WARNING on linus/master]
[cannot apply to v5.3-rc8 next-20190904]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]

url: https://github.com/0day-ci/linux/commits/Robert-Beckett/net-dsa-mv88e6xxx-features-to-handle-network-storms/20190911-142233
reproduce:

        # apt-get install sparse
        # sparse version: v0.6.1-rc1-7-g2b96cd8-dirty
        make ARCH=x86_64 allmodconfig
        make C=1 CF='-fdiagnostic-prefix -D__CHECK_ENDIAN__'

If you fix the issue, kindly add following tag

sparse warnings: (new ones prefixed by >>)

include/linux/sched.h:609:43: sparse: sparse: bad integer constant expression include/linux/sched.h:609:73: sparse: sparse: invalid named zero-width bitfield value' include/linux/sched.h:610:43: sparse: sparse: bad integer constant expression include/linux/sched.h:610:67: sparse: sparse: invalid named zero-width bitfield bucket_id'

>> net/dsa/port.c:541:55: sparse: sparse: incompatible types for operation (|)
>> net/dsa/port.c:541:55: sparse:    left side has type unsigned long *
>> net/dsa/port.c:541:55: sparse:    right side has type unsigned long

vim +541 net/dsa/port.c

525
526 static int dsa_port_setup_phy_of(struct dsa_port *dp, bool enable)
527 {
528 struct dsa_switch *ds = dp->ds;
529 struct phy_device *phydev;
530 int port = dp->index;
531 int err = 0;
532
533 phydev = dsa_port_get_phy_device(dp);
534 if (!phydev)
535 return 0;
536
537 if (IS_ERR(phydev))
538 return PTR_ERR(phydev);
539
540 if (enable) {

 > 541			phydev->supported = PHY_GBIT_FEATURES | SUPPORTED_MII |
   542					    SUPPORTED_AUI | SUPPORTED_FIBRE |
   543					    SUPPORTED_BNC | SUPPORTED_Pause |
   544					    SUPPORTED_Asym_Pause;

545 phydev->advertising = phydev->supported;
546
547 err = genphy_config_init(phydev);
548 if (err < 0)
549 goto err_put_dev;
550
551 err = genphy_config_aneg(phydev);
552 if (err < 0)
553 goto err_put_dev;
554
555 err = genphy_resume(phydev);
556 if (err < 0)
557 goto err_put_dev;
558
559 err = genphy_read_status(phydev);
560 if (err < 0)
561 goto err_put_dev;
562 } else {
563 err = genphy_suspend(phydev);
564 if (err < 0)
565 goto err_put_dev;
566 }
567
568 if (ds->ops->adjust_link)
569 ds->ops->adjust_link(ds, port, phydev);
570
571 dev_dbg(ds->dev, "enabled port's phy: %s", phydev_name(phydev));
572
573 err_put_dev:
574 put_device(&phydev->mdio.dev);
575 return err;
576 }
577

---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation

#13 lore raw

PATCH 2/7 net: dsa: mv88e6xxx: add ability to set default queue priorities per port

In reply to Robert Beckett

Add code to set DefQPri for any port that specifies "defqpri" in their device tree node.
This allows setting the default output queue priority for all packets entering the switch via the port that uses this, which is useful for prioritizing traffic based on port.

---
 drivers/net/dsa/mv88e6xxx/chip.c | 25 +++++++++++++++++++++++++
 drivers/net/dsa/mv88e6xxx/chip.h |  1 +
 drivers/net/dsa/mv88e6xxx/port.c | 19 +++++++++++++++++++
 drivers/net/dsa/mv88e6xxx/port.h |  4 ++++
 4 files changed, 49 insertions(+)
drivers/net/dsa/mv88e6xxx/chip.c +25 -0
index d0a97eb73a37..5005a35493e3 100644
@@ -2086,6 +2086,23 @@ static int mv88e6xxx_setup_upstream_port(struct mv88e6xxx_chip *chip, int port)
 	return 0;
 }
 
+static int mv88e6xxx_set_port_defqpri(struct mv88e6xxx_chip *chip, int port)
+{
+	struct dsa_switch *ds = chip->ds;
+	struct device_node *dn = ds->ports[port].dn;
+	int err;
+	u32 pri;
+
+	if (!dn || !chip->info->ops->port_set_defqpri)
+		return 0;
+
+	err = of_property_read_u32(dn, "defqpri", &pri);
+	if (err < 0)
+		return 0;
+
+	return chip->info->ops->port_set_defqpri(chip, port, (u16)pri);
+}
+
 static int mv88e6xxx_setup_port(struct mv88e6xxx_chip *chip, int port)
 {
 	struct dsa_switch *ds = chip->ds;
@@ -2176,6 +2193,10 @@ static int mv88e6xxx_setup_port(struct mv88e6xxx_chip *chip, int port)
 			return err;
 	}
 
+	err = mv88e6xxx_set_port_defqpri(chip, port);
+	if (err)
+		return err;
+
 	/* Port Association Vector: when learning source addresses
 	 * of packets, add the address to the address database using
 	 * a port bitmap that has only the bit for this port set and
@@ -3107,6 +3128,7 @@ static const struct mv88e6xxx_ops mv88e6172_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
+	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
 	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
@@ -3190,6 +3212,7 @@ static const struct mv88e6xxx_ops mv88e6176_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
+	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
 	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
@@ -3407,6 +3430,7 @@ static const struct mv88e6xxx_ops mv88e6240_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
+	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
 	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
@@ -3750,6 +3774,7 @@ static const struct mv88e6xxx_ops mv88e6352_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
+	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
 	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
drivers/net/dsa/mv88e6xxx/chip.h +1 -0
index 4646e46d47f2..2d2c24f5a79d 100644
@@ -383,6 +383,7 @@ struct mv88e6xxx_ops {
 				   u16 etype);
 	int (*port_set_jumbo_size)(struct mv88e6xxx_chip *chip, int port,
 				   size_t size);
+	int (*port_set_defqpri)(struct mv88e6xxx_chip *chip, int port, u16 pri);
 
 	int (*port_egress_rate_limiting)(struct mv88e6xxx_chip *chip, int port);
 	int (*port_pause_limit)(struct mv88e6xxx_chip *chip, int port, u8 in,
drivers/net/dsa/mv88e6xxx/port.c +19 -0
index 04309ef0a1cc..3a45fcd5cd9c 100644
@@ -1147,6 +1147,25 @@ int mv88e6165_port_set_jumbo_size(struct mv88e6xxx_chip *chip, int port,
 	return mv88e6xxx_port_write(chip, port, MV88E6XXX_PORT_CTL2, reg);
 }
 
+int mv88e6xxx_port_set_defqpri(struct mv88e6xxx_chip *chip, int port, u16 pri)
+{
+	u16 reg;
+	int err;
+
+	if (pri > 3)
+		return -EINVAL;
+
+	err = mv88e6xxx_port_read(chip, port, MV88E6XXX_PORT_CTL2, &reg);
+	if (err)
+		return err;
+
+	reg &= ~MV88E6XXX_PORT_CTL2_DEFQPRI_MASK;
+	reg |= pri << MV88E6XXX_PORT_CTL2_DEFQPRI_SHIFT;
+	reg |= MV88E6XXX_PORT_CTL2_USE_DEFQPRI;
+
+	return mv88e6xxx_port_write(chip, port, MV88E6XXX_PORT_CTL2, reg);
+}
+
 /* Offset 0x09: Port Rate Control */
 
 int mv88e6095_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port)
drivers/net/dsa/mv88e6xxx/port.h +4 -0
index 8d5a6cd6fb19..03884bbaa762 100644
@@ -197,6 +197,9 @@
 #define MV88E6XXX_PORT_CTL2_DEFAULT_FORWARD		0x0040
 #define MV88E6XXX_PORT_CTL2_EGRESS_MONITOR		0x0020
 #define MV88E6XXX_PORT_CTL2_INGRESS_MONITOR		0x0010
+#define MV88E6XXX_PORT_CTL2_USE_DEFQPRI		0x0008
+#define MV88E6XXX_PORT_CTL2_DEFQPRI_MASK		0x0006
+#define MV88E6XXX_PORT_CTL2_DEFQPRI_SHIFT		1
 #define MV88E6095_PORT_CTL2_CPU_PORT_MASK		0x000f
 
 /* Offset 0x09: Egress Rate Control */
@@ -326,6 +329,7 @@ int mv88e6xxx_port_set_message_port(struct mv88e6xxx_chip *chip, int port,
 				    bool message_port);
 int mv88e6165_port_set_jumbo_size(struct mv88e6xxx_chip *chip, int port,
 				  size_t size);
+int mv88e6xxx_port_set_defqpri(struct mv88e6xxx_chip *chip, int port, u16 pri);
 int mv88e6095_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port);
 int mv88e6097_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port);
 int mv88e6097_port_pause_limit(struct mv88e6xxx_chip *chip, int port, u8 in,
Signature · 1 line
2.18.0
#14 lore raw

Re: [PATCH 2/7] net: dsa: mv88e6xxx: add ability to set default queue priorities per port

Hi Robert,

Quoted 32 lines · Robert Beckett
drivers/net/dsa/mv88e6xxx/chip.c +21 -0
+static int mv88e6xxx_set_port_defqpri(struct mv88e6xxx_chip *chip, int port)
+{
+	struct dsa_switch *ds = chip->ds;
+	struct device_node *dn = ds->ports[port].dn;
+	int err;
+	u32 pri;
+
+	if (!dn || !chip->info->ops->port_set_defqpri)
+		return 0;
+
+	err = of_property_read_u32(dn, "defqpri", &pri);
+	if (err < 0)
+		return 0;
+
+	return chip->info->ops->port_set_defqpri(chip, port, (u16)pri);
+}
+
 static int mv88e6xxx_setup_port(struct mv88e6xxx_chip *chip, int port)
 {
 	struct dsa_switch *ds = chip->ds;
@@ -2176,6 +2193,10 @@ static int mv88e6xxx_setup_port(struct mv88e6xxx_chip *chip, int port)
 			return err;
 	}
 
+	err = mv88e6xxx_set_port_defqpri(chip, port);
+	if (err)
+		return err;
+
 	/* Port Association Vector: when learning source addresses
 	 * of packets, add the address to the address database using
 	 * a port bitmap that has only the bit for this port set and

drivers/net/dsa/mv88e6xxx/chip.c static const struct mv88e6xxx_ops mv88e6172_ops = { · line 3131

Quoted 4 lines
drivers/net/dsa/mv88e6xxx/chip.c +1 -0
@@ -3107,6 +3128,7 @@ static const struct mv88e6xxx_ops mv88e6172_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
+	.port_set_defqpri = mv88e6xxx_port_set_defqpri,

Please use a reference model, like mv88e6352_port_set_defqpri to avoid confusion with a generic wrapper or implementation.

Quoted 32 lines
drivers/net/dsa/mv88e6xxx/chip.c +3 -0
 	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
@@ -3190,6 +3212,7 @@ static const struct mv88e6xxx_ops mv88e6176_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
+	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
 	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
@@ -3407,6 +3430,7 @@ static const struct mv88e6xxx_ops mv88e6240_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
+	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
 	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
@@ -3750,6 +3774,7 @@ static const struct mv88e6xxx_ops mv88e6352_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
+	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
 	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
drivers/net/dsa/mv88e6xxx/chip.h +0 -0
index 4646e46d47f2..2d2c24f5a79d 100644

drivers/net/dsa/mv88e6xxx/chip.h struct mv88e6xxx_ops { · line 386

Quoted 4 lines
drivers/net/dsa/mv88e6xxx/chip.h +1 -0
@@ -383,6 +383,7 @@ struct mv88e6xxx_ops {
 				   u16 etype);
 	int (*port_set_jumbo_size)(struct mv88e6xxx_chip *chip, int port,
 				   size_t size);
+	int (*port_set_defqpri)(struct mv88e6xxx_chip *chip, int port, u16 pri);

The default queue priority seems to be an integer in the [0:3] range, not a register mask or value per-se. So an unsigned int seems more appropriate.

Quoted 15 lines
drivers/net/dsa/mv88e6xxx/chip.h +0 -0
 	int (*port_egress_rate_limiting)(struct mv88e6xxx_chip *chip, int port);
 	int (*port_pause_limit)(struct mv88e6xxx_chip *chip, int port, u8 in,
drivers/net/dsa/mv88e6xxx/port.c +4 -0
index 04309ef0a1cc..3a45fcd5cd9c 100644
@@ -1147,6 +1147,25 @@ int mv88e6165_port_set_jumbo_size(struct mv88e6xxx_chip *chip, int port,
 	return mv88e6xxx_port_write(chip, port, MV88E6XXX_PORT_CTL2, reg);
 }
 
+int mv88e6xxx_port_set_defqpri(struct mv88e6xxx_chip *chip, int port, u16 pri)
+{
+	u16 reg;
+	int err;

drivers/net/dsa/mv88e6xxx/port.c int mv88e6165_port_set_jumbo_size(struct mv88e6xxx_chip *chip, int port, · line 1163

Quoted 10 lines
drivers/net/dsa/mv88e6xxx/port.c +10 -0
+
+	if (pri > 3)
+		return -EINVAL;
+
+	err = mv88e6xxx_port_read(chip, port, MV88E6XXX_PORT_CTL2, &reg);
+	if (err)
+		return err;
+
+	reg &= ~MV88E6XXX_PORT_CTL2_DEFQPRI_MASK;
+	reg |= pri << MV88E6XXX_PORT_CTL2_DEFQPRI_SHIFT;
                      __bf_shf(MV88E6XXX_PORT_CTL2_DEFQPRI_MASK)
Quoted 13 lines
drivers/net/dsa/mv88e6xxx/port.c +5 -0
+	reg |= MV88E6XXX_PORT_CTL2_USE_DEFQPRI;
+
+	return mv88e6xxx_port_write(chip, port, MV88E6XXX_PORT_CTL2, reg);
+}
+
 /* Offset 0x09: Port Rate Control */
 
 int mv88e6095_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port)
drivers/net/dsa/mv88e6xxx/port.h +0 -0
index 8d5a6cd6fb19..03884bbaa762 100644

drivers/net/dsa/mv88e6xxx/port.h line 202

Quoted 6 lines
drivers/net/dsa/mv88e6xxx/port.h +3 -0
@@ -197,6 +197,9 @@
 #define MV88E6XXX_PORT_CTL2_DEFAULT_FORWARD		0x0040
 #define MV88E6XXX_PORT_CTL2_EGRESS_MONITOR		0x0020
 #define MV88E6XXX_PORT_CTL2_INGRESS_MONITOR		0x0010
+#define MV88E6XXX_PORT_CTL2_USE_DEFQPRI		0x0008
+#define MV88E6XXX_PORT_CTL2_DEFQPRI_MASK		0x0006
+#define MV88E6XXX_PORT_CTL2_DEFQPRI_SHIFT		1

No shift macro needed, MV88E6XXX_PORT_CTL2_DEFQPRI_MASK is enough.

drivers/net/dsa/mv88e6xxx/port.h int mv88e6xxx_port_set_message_port(struct mv88e6xxx_chip *chip, int port, · line 335

Quoted 11 lines
drivers/net/dsa/mv88e6xxx/port.h +1 -0
 #define MV88E6095_PORT_CTL2_CPU_PORT_MASK		0x000f
 
 /* Offset 0x09: Egress Rate Control */
@@ -326,6 +329,7 @@ int mv88e6xxx_port_set_message_port(struct mv88e6xxx_chip *chip, int port,
 				    bool message_port);
 int mv88e6165_port_set_jumbo_size(struct mv88e6xxx_chip *chip, int port,
 				  size_t size);
+int mv88e6xxx_port_set_defqpri(struct mv88e6xxx_chip *chip, int port, u16 pri);
 int mv88e6095_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port);
 int mv88e6097_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port);
 int mv88e6097_port_pause_limit(struct mv88e6xxx_chip *chip, int port, u8 in,

Thanks,

Vivien

#15 lore raw

PATCH 3/7 dt-bindings: mv88e6xxx: add ability to set default queue priorities per port

In reply to Robert Beckett

Document a new setting for Marvell switch chips to set the default queue priorities per port.

---
 Documentation/devicetree/bindings/net/dsa/marvell.txt | 4 ++++
 1 file changed, 4 insertions(+)
Documentation/devicetree/bindings/net/dsa/marvell.txt +4 -0
index 6f9538974bb9..e097c3c52eac 100644
@@ -47,6 +47,10 @@ Optional properties:
 			  bus. The node must contains a compatible string of
 			  "marvell,mv88e6xxx-mdio-external"
 
+Optional properties for ports:
+- defqpri=<n>		: Enforced default queue priority for the given port.
+			  Valid range is 0..3
+
 Example:
 
 	mdio {
Signature · 1 line
2.18.0
#16 lore raw

Re: [PATCH 3/7] dt-bindings: mv88e6xxx: add ability to set default queue priorities per port

Quoted 13 lines · Robert Beckett

Document a new setting for Marvell switch chips to set the default queue priorities per port.

---
 Documentation/devicetree/bindings/net/dsa/marvell.txt | 4 ++++
 1 file changed, 4 insertions(+)
Documentation/devicetree/bindings/net/dsa/marvell.txt +0 -0
index 6f9538974bb9..e097c3c52eac 100644

Documentation/devicetree/bindings/net/dsa/marvell.txt Optional properties: · line 52

Quoted 6 lines
Documentation/devicetree/bindings/net/dsa/marvell.txt +3 -0
@@ -47,6 +47,10 @@ Optional properties:
 			  bus. The node must contains a compatible string of
 			  "marvell,mv88e6xxx-mdio-external"
 
+Optional properties for ports:
+- defqpri=<n>		: Enforced default queue priority for the given port.
+			  Valid range is 0..3

This is a vendor specific driver/property, marvell,default-queue-priority (which be cheapskate on words) would be more readable. But still, I have some more fundamental issues with the general approach, see my response in the cover letter.

Signature · 1 line
Florian
#17 lore raw

Re: [PATCH 3/7] dt-bindings: mv88e6xxx: add ability to set default queue priorities per port

Hi Robert,

Quoted 4 lines · Florian Fainelli

This is a vendor specific driver/property, marvell,default-queue-priority (which be cheapskate on words) would be more readable. But still, I have some more fundamental issues with the general approach, see my response in the cover letter.

As Florian said, the DT is unlikely to welcome vendor specific nodes for configuration which may be generic through standard network userspace tools.

Thanks,

Vivien

maintainer
#18 lore raw

Re: [PATCH 3/7] dt-bindings: mv88e6xxx: add ability to set default queue priorities per port

Hi guys,

Quoted 9 lines · Vivien Didelot

Hi Robert,

Quoted 4 lines · Florian Fainelli

This is a vendor specific driver/property, marvell,default-queue-priority (which be cheapskate on words) would be more readable. But still, I have some more fundamental issues with the general approach, see my response in the cover letter.

Quoted 9 lines

As Florian said, the DT is unlikely to welcome vendor specific nodes for configuration which may be generic through standard network userspace tools.

Thanks,

Vivien

While I do agree that the DT bindings are a big no-no for QoS settings, the topic is interesting.
What is the user space knob for configuring port-default priority (say RX queue)?
Something like this maybe? (a very forced "matchall" with rxnfc) ethtool --config-nfc eth0 flow-type ether src 00:00:00:00:00:00 m 00:00:00:00:00:00 action 5

Regards,
-Vladimir

#19 lore raw

PATCH 4/7 net: dsa: mv88e6xxx: add ability to set queue scheduling

In reply to Robert Beckett

Add code to set Schedule for any port that specifies "schedule" in their device tree node.
This allows port prioritization in conjunction with port default queue priorities or packet priorities.

---
 drivers/net/dsa/mv88e6xxx/chip.c | 25 +++++++++++++++++++++++++
 drivers/net/dsa/mv88e6xxx/chip.h |  1 +
 drivers/net/dsa/mv88e6xxx/port.c | 21 +++++++++++++++++++++
 drivers/net/dsa/mv88e6xxx/port.h |  6 ++++++
 4 files changed, 53 insertions(+)
drivers/net/dsa/mv88e6xxx/chip.c +25 -0
index 5005a35493e3..2bc22c59200c 100644
@@ -2103,6 +2103,23 @@ static int mv88e6xxx_set_port_defqpri(struct mv88e6xxx_chip *chip, int port)
 	return chip->info->ops->port_set_defqpri(chip, port, (u16)pri);
 }
 
+static int mv88e6xxx_set_port_sched(struct mv88e6xxx_chip *chip, int port)
+{
+	struct dsa_switch *ds = chip->ds;
+	struct device_node *dn = ds->ports[port].dn;
+	int err;
+	u32 sched;
+
+	if (!dn || !chip->info->ops->port_set_sched)
+		return 0;
+
+	err = of_property_read_u32(dn, "schedule", &sched);
+	if (err < 0)
+		return 0;
+
+	return chip->info->ops->port_set_sched(chip, port, (u16)sched);
+}
+
 static int mv88e6xxx_setup_port(struct mv88e6xxx_chip *chip, int port)
 {
 	struct dsa_switch *ds = chip->ds;
@@ -2218,6 +2235,10 @@ static int mv88e6xxx_setup_port(struct mv88e6xxx_chip *chip, int port)
 	if (err)
 		return err;
 
+	err = mv88e6xxx_set_port_sched(chip, port);
+	if (err)
+		return err;
+
 	if (chip->info->ops->port_pause_limit) {
 		err = chip->info->ops->port_pause_limit(chip, port, 0, 0);
 		if (err)
@@ -3130,6 +3151,7 @@ static const struct mv88e6xxx_ops mv88e6172_ops = {
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
 	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
 	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_set_sched = mv88e6xxx_port_set_sched,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3214,6 +3236,7 @@ static const struct mv88e6xxx_ops mv88e6176_ops = {
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
 	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
 	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_set_sched = mv88e6xxx_port_set_sched,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3432,6 +3455,7 @@ static const struct mv88e6xxx_ops mv88e6240_ops = {
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
 	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
 	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_set_sched = mv88e6xxx_port_set_sched,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3776,6 +3800,7 @@ static const struct mv88e6xxx_ops mv88e6352_ops = {
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
 	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
 	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_set_sched = mv88e6xxx_port_set_sched,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
drivers/net/dsa/mv88e6xxx/chip.h +1 -0
index 2d2c24f5a79d..ff3e35eceee0 100644
@@ -386,6 +386,7 @@ struct mv88e6xxx_ops {
 	int (*port_set_defqpri)(struct mv88e6xxx_chip *chip, int port, u16 pri);
 
 	int (*port_egress_rate_limiting)(struct mv88e6xxx_chip *chip, int port);
+	int (*port_set_sched)(struct mv88e6xxx_chip *chip, int port, u16 sched);
 	int (*port_pause_limit)(struct mv88e6xxx_chip *chip, int port, u8 in,
 				u8 out);
 	int (*port_disable_learn_limit)(struct mv88e6xxx_chip *chip, int port);
drivers/net/dsa/mv88e6xxx/port.c +21 -0
index 3a45fcd5cd9c..236732fc598d 100644
@@ -1180,6 +1180,27 @@ int mv88e6097_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port)
 				    0x0001);
 }
 
+/* Offset 0x0A: Egress Rate Control 2 */
+int mv88e6xxx_port_set_sched(struct mv88e6xxx_chip *chip, int port, u16 sched)
+{
+	u16 reg;
+	int err;
+
+	if (sched > MV88E6XXX_PORT_SCHED_STRICT_ALL)
+		return -EINVAL;
+
+	err = mv88e6xxx_port_read(chip, port, MV88E6XXX_PORT_EGRESS_RATE_CTL2,
+				  &reg);
+	if (err)
+		return err;
+
+	reg &= ~MV88E6XXX_PORT_SCHED_MASK;
+	reg |= sched << MV88E6XXX_PORT_SCHED_SHIFT;
+
+	return mv88e6xxx_port_write(chip, port, MV88E6XXX_PORT_EGRESS_RATE_CTL2,
+				    reg);
+}
+
 /* Offset 0x0C: Port ATU Control */
 
 int mv88e6xxx_port_disable_learn_limit(struct mv88e6xxx_chip *chip, int port)
drivers/net/dsa/mv88e6xxx/port.h +6 -0
index 03884bbaa762..710d6eccafae 100644
@@ -11,6 +11,7 @@
 #ifndef _MV88E6XXX_PORT_H
 #define _MV88E6XXX_PORT_H
 
+#include <dt-bindings/net/dsa-mv88e6xxx.h>
 #include "chip.h"
 
 /* Offset 0x00: Port Status Register */
@@ -207,6 +208,10 @@
 
 /* Offset 0x0A: Egress Rate Control 2 */
 #define MV88E6XXX_PORT_EGRESS_RATE_CTL2		0x0a
+#define MV88E6XXX_PORT_SCHED_SHIFT		12
+#define MV88E6XXX_PORT_SCHED_MASK \
+	(0x3 << MV88E6XXX_PORT_SCHED_SHIFT)
+/* see MV88E6XXX_PORT_SCHED_* in include/dt-bindings/net/dsa-mv88e6xxx.h */
 
 /* Offset 0x0B: Port Association Vector */
 #define MV88E6XXX_PORT_ASSOC_VECTOR			0x0b
@@ -332,6 +337,7 @@ int mv88e6165_port_set_jumbo_size(struct mv88e6xxx_chip *chip, int port,
 int mv88e6xxx_port_set_defqpri(struct mv88e6xxx_chip *chip, int port, u16 pri);
 int mv88e6095_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port);
 int mv88e6097_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port);
+int mv88e6xxx_port_set_sched(struct mv88e6xxx_chip *chip, int port, u16 sched);
 int mv88e6097_port_pause_limit(struct mv88e6xxx_chip *chip, int port, u8 in,
 			       u8 out);
 int mv88e6390_port_pause_limit(struct mv88e6xxx_chip *chip, int port, u8 in,
Signature · 1 line
2.18.0
#20 lore raw

Re: [PATCH 4/7] net: dsa: mv88e6xxx: add ability to set queue scheduling

Hi Robert,

Quoted 152 lines · Robert Beckett

Add code to set Schedule for any port that specifies "schedule" in their device tree node.
This allows port prioritization in conjunction with port default queue priorities or packet priorities.

---
 drivers/net/dsa/mv88e6xxx/chip.c | 25 +++++++++++++++++++++++++
 drivers/net/dsa/mv88e6xxx/chip.h |  1 +
 drivers/net/dsa/mv88e6xxx/port.c | 21 +++++++++++++++++++++
 drivers/net/dsa/mv88e6xxx/port.h |  6 ++++++
 4 files changed, 53 insertions(+)
drivers/net/dsa/mv88e6xxx/chip.c +25 -0
index 5005a35493e3..2bc22c59200c 100644
@@ -2103,6 +2103,23 @@ static int mv88e6xxx_set_port_defqpri(struct mv88e6xxx_chip *chip, int port)
 	return chip->info->ops->port_set_defqpri(chip, port, (u16)pri);
 }
 
+static int mv88e6xxx_set_port_sched(struct mv88e6xxx_chip *chip, int port)
+{
+	struct dsa_switch *ds = chip->ds;
+	struct device_node *dn = ds->ports[port].dn;
+	int err;
+	u32 sched;
+
+	if (!dn || !chip->info->ops->port_set_sched)
+		return 0;
+
+	err = of_property_read_u32(dn, "schedule", &sched);
+	if (err < 0)
+		return 0;
+
+	return chip->info->ops->port_set_sched(chip, port, (u16)sched);
+}
+
 static int mv88e6xxx_setup_port(struct mv88e6xxx_chip *chip, int port)
 {
 	struct dsa_switch *ds = chip->ds;
@@ -2218,6 +2235,10 @@ static int mv88e6xxx_setup_port(struct mv88e6xxx_chip *chip, int port)
 	if (err)
 		return err;
 
+	err = mv88e6xxx_set_port_sched(chip, port);
+	if (err)
+		return err;
+
 	if (chip->info->ops->port_pause_limit) {
 		err = chip->info->ops->port_pause_limit(chip, port, 0, 0);
 		if (err)
@@ -3130,6 +3151,7 @@ static const struct mv88e6xxx_ops mv88e6172_ops = {
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
 	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
 	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_set_sched = mv88e6xxx_port_set_sched,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3214,6 +3236,7 @@ static const struct mv88e6xxx_ops mv88e6176_ops = {
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
 	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
 	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_set_sched = mv88e6xxx_port_set_sched,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3432,6 +3455,7 @@ static const struct mv88e6xxx_ops mv88e6240_ops = {
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
 	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
 	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_set_sched = mv88e6xxx_port_set_sched,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3776,6 +3800,7 @@ static const struct mv88e6xxx_ops mv88e6352_ops = {
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
 	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
 	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_set_sched = mv88e6xxx_port_set_sched,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
drivers/net/dsa/mv88e6xxx/chip.h +1 -0
index 2d2c24f5a79d..ff3e35eceee0 100644
@@ -386,6 +386,7 @@ struct mv88e6xxx_ops {
 	int (*port_set_defqpri)(struct mv88e6xxx_chip *chip, int port, u16 pri);
 
 	int (*port_egress_rate_limiting)(struct mv88e6xxx_chip *chip, int port);
+	int (*port_set_sched)(struct mv88e6xxx_chip *chip, int port, u16 sched);
 	int (*port_pause_limit)(struct mv88e6xxx_chip *chip, int port, u8 in,
 				u8 out);
 	int (*port_disable_learn_limit)(struct mv88e6xxx_chip *chip, int port);
drivers/net/dsa/mv88e6xxx/port.c +21 -0
index 3a45fcd5cd9c..236732fc598d 100644
@@ -1180,6 +1180,27 @@ int mv88e6097_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port)
 				    0x0001);
 }
 
+/* Offset 0x0A: Egress Rate Control 2 */
+int mv88e6xxx_port_set_sched(struct mv88e6xxx_chip *chip, int port, u16 sched)
+{
+	u16 reg;
+	int err;
+
+	if (sched > MV88E6XXX_PORT_SCHED_STRICT_ALL)
+		return -EINVAL;
+
+	err = mv88e6xxx_port_read(chip, port, MV88E6XXX_PORT_EGRESS_RATE_CTL2,
+				  &reg);
+	if (err)
+		return err;
+
+	reg &= ~MV88E6XXX_PORT_SCHED_MASK;
+	reg |= sched << MV88E6XXX_PORT_SCHED_SHIFT;
+
+	return mv88e6xxx_port_write(chip, port, MV88E6XXX_PORT_EGRESS_RATE_CTL2,
+				    reg);
+}
+
 /* Offset 0x0C: Port ATU Control */
 
 int mv88e6xxx_port_disable_learn_limit(struct mv88e6xxx_chip *chip, int port)
drivers/net/dsa/mv88e6xxx/port.h +5 -0
index 03884bbaa762..710d6eccafae 100644
@@ -11,6 +11,7 @@
 #ifndef _MV88E6XXX_PORT_H
 #define _MV88E6XXX_PORT_H
 
+#include <dt-bindings/net/dsa-mv88e6xxx.h>
 #include "chip.h"
 
 /* Offset 0x00: Port Status Register */
@@ -207,6 +208,10 @@
 
 /* Offset 0x0A: Egress Rate Control 2 */
 #define MV88E6XXX_PORT_EGRESS_RATE_CTL2		0x0a
+#define MV88E6XXX_PORT_SCHED_SHIFT		12
+#define MV88E6XXX_PORT_SCHED_MASK \
+	(0x3 << MV88E6XXX_PORT_SCHED_SHIFT)
+/* see MV88E6XXX_PORT_SCHED_* in include/dt-bindings/net/dsa-mv88e6xxx.h */
 
 /* Offset 0x0B: Port Association Vector */
 #define MV88E6XXX_PORT_ASSOC_VECTOR			0x0b

drivers/net/dsa/mv88e6xxx/port.h int mv88e6165_port_set_jumbo_size(struct mv88e6xxx_chip *chip, int port, · line 343

Quoted 7 lines
drivers/net/dsa/mv88e6xxx/port.h +1 -0
@@ -332,6 +337,7 @@ int mv88e6165_port_set_jumbo_size(struct mv88e6xxx_chip *chip, int port,
 int mv88e6xxx_port_set_defqpri(struct mv88e6xxx_chip *chip, int port, u16 pri);
 int mv88e6095_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port);
 int mv88e6097_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port);
+int mv88e6xxx_port_set_sched(struct mv88e6xxx_chip *chip, int port, u16 sched);
 int mv88e6097_port_pause_limit(struct mv88e6xxx_chip *chip, int port, u8 in,
 			       u8 out);
 int mv88e6390_port_pause_limit(struct mv88e6xxx_chip *chip, int port, u8 in,

Same comments applied as for the other patches adding implementations in port.c.

Thanks,

Vivien

#21 lore raw

PATCH 5/7 dt-bindings: mv88e6xxx: add ability to set queue scheduling

In reply to Robert Beckett

Document port queue scheduling settings.
Add definitions for specific valid values.

---
 .../devicetree/bindings/net/dsa/marvell.txt     | 12 ++++++++++++
 include/dt-bindings/net/dsa-mv88e6xxx.h         | 17 +++++++++++++++++
 2 files changed, 29 insertions(+)
 create mode 100644 include/dt-bindings/net/dsa-mv88e6xxx.h
Documentation/devicetree/bindings/net/dsa/marvell.txt +12 -0
index e097c3c52eac..7de90929c3c9 100644
@@ -50,6 +50,18 @@ Optional properties:
 Optional properties for ports:
 - defqpri=<n>		: Enforced default queue priority for the given port.
 			  Valid range is 0..3
+- schedule=<n>		: Set ports scheduling mode. Valid values are:
+			  MV88E6XXX_PORT_SCHED_ROUND_ROBIN - All output queues
+			  use a weighter round robin scheme.
+			  MV88E6XXX_PORT_SCHED_STRICT_3 - Output queue 3 uses
+			  a strict scheme, where any packets in queue 3 will be
+			  egressed first, followed by weighted round robin for
+			  the other ports.
+			  MV88E6XXX_PORT_SCHED_STRICT_3_2 - Output queue's 2
+			  and 3 use strict, other use weighted round robin.
+			  MV88E6XXX_PORT_SCHED_STRICT_ALL - All queues use
+			  strict priority, where queues drain in descending
+			  queue number order.
 
 Example:
 
include/dt-bindings/net/dsa-mv88e6xxx.h new file +17 -0
new file mode 100644
index 000000000000..3f62003841ce
@@ -0,0 +1,17 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Device Tree constants for Marvell 88E6xxx Switch Port Registers
+ *
+ * Copyright (c) 2019, Collabora Ltd.
+ * Copyright (c) 2019, General Electric Company
+ */
+
+#ifndef _DT_BINDINGS_MV88E6XXX_H
+#define _DT_BINDINGS_MV88E6XXX_H
+
+#define MV88E6XXX_PORT_SCHED_ROUND_ROBIN	0
+#define MV88E6XXX_PORT_SCHED_STRICT_3		1
+#define MV88E6XXX_PORT_SCHED_STRICT_3_2		2
+#define MV88E6XXX_PORT_SCHED_STRICT_ALL		3
+
+#endif /* _DT_BINDINGS_MV88E6XXX_H */
Signature · 1 line
2.18.0
#22 lore raw

PATCH 6/7 net: dsa: mv88e6xxx: add egress rate limiting

In reply to Robert Beckett

Add code for specifying egress rate limiting per port. The rate can be specified as ethernet frames or bits per second.

---
 drivers/net/dsa/mv88e6xxx/chip.c |  72 ++++++++++++++-------
 drivers/net/dsa/mv88e6xxx/chip.h |   3 +-
 drivers/net/dsa/mv88e6xxx/port.c | 106 ++++++++++++++++++++++++++++---
 drivers/net/dsa/mv88e6xxx/port.h |  14 +++-
 4 files changed, 158 insertions(+), 37 deletions(-)
drivers/net/dsa/mv88e6xxx/chip.c +48 -24
index 2bc22c59200c..8c116496ab2f 100644
@@ -2120,6 +2120,32 @@ static int mv88e6xxx_set_port_sched(struct mv88e6xxx_chip *chip, int port)
 	return chip->info->ops->port_set_sched(chip, port, (u16)sched);
 }
 
+static int mv88e6xxx_set_port_egress_rate_limiting(struct mv88e6xxx_chip *chip,
+						   int port)
+{
+	struct dsa_switch *ds = chip->ds;
+	struct device_node *dn = ds->ports[port].dn;
+	int err;
+	u32 mode, count;
+
+	if (!dn || !chip->info->ops->port_egress_rate_limiting)
+		return 0;
+
+	err = of_property_read_u32(dn, "egress-limit-mode", &mode);
+	if (err < 0)
+		goto disable;
+
+	err = of_property_read_u32(dn, "egress-limit-count", &count);
+	if (err < 0)
+		goto disable;
+
+	return chip->info->ops->port_egress_rate_limiting(chip, port, count,
+							  mode);
+
+disable:
+	return chip->info->ops->port_egress_rate_limiting(chip, port, 0, 0);
+}
+
 static int mv88e6xxx_setup_port(struct mv88e6xxx_chip *chip, int port)
 {
 	struct dsa_switch *ds = chip->ds;
@@ -2263,11 +2289,9 @@ static int mv88e6xxx_setup_port(struct mv88e6xxx_chip *chip, int port)
 			return err;
 	}
 
-	if (chip->info->ops->port_egress_rate_limiting) {
-		err = chip->info->ops->port_egress_rate_limiting(chip, port);
-		if (err)
-			return err;
-	}
+	err = mv88e6xxx_set_port_egress_rate_limiting(chip, port);
+	if (err)
+		return err;
 
 	err = mv88e6xxx_setup_message_port(chip, port);
 	if (err)
@@ -2809,7 +2833,7 @@ static const struct mv88e6xxx_ops mv88e6085_ops = {
 	.port_set_frame_mode = mv88e6351_port_set_frame_mode,
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -2879,7 +2903,7 @@ static const struct mv88e6xxx_ops mv88e6097_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6095_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -2951,7 +2975,7 @@ static const struct mv88e6xxx_ops mv88e6131_ops = {
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_upstream_port = mv88e6095_port_set_upstream_port,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_set_pause = mv88e6185_port_set_pause,
 	.port_link_state = mv88e6352_port_link_state,
@@ -2994,7 +3018,7 @@ static const struct mv88e6xxx_ops mv88e6141_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3034,7 +3058,7 @@ static const struct mv88e6xxx_ops mv88e6161_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3108,7 +3132,7 @@ static const struct mv88e6xxx_ops mv88e6171_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3150,7 +3174,7 @@ static const struct mv88e6xxx_ops mv88e6172_ops = {
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
 	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_set_sched = mv88e6xxx_port_set_sched,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
@@ -3193,7 +3217,7 @@ static const struct mv88e6xxx_ops mv88e6175_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3235,7 +3259,7 @@ static const struct mv88e6xxx_ops mv88e6176_ops = {
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
 	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_set_sched = mv88e6xxx_port_set_sched,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
@@ -3275,7 +3299,7 @@ static const struct mv88e6xxx_ops mv88e6185_ops = {
 	.port_set_speed = mv88e6185_port_set_speed,
 	.port_set_frame_mode = mv88e6085_port_set_frame_mode,
 	.port_set_egress_floods = mv88e6185_port_set_egress_floods,
-	.port_egress_rate_limiting = mv88e6095_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_set_upstream_port = mv88e6095_port_set_upstream_port,
 	.port_set_pause = mv88e6185_port_set_pause,
 	.port_link_state = mv88e6185_port_link_state,
@@ -3454,7 +3478,7 @@ static const struct mv88e6xxx_ops mv88e6240_ops = {
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
 	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_set_sched = mv88e6xxx_port_set_sched,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
@@ -3587,7 +3611,7 @@ static const struct mv88e6xxx_ops mv88e6320_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3630,7 +3654,7 @@ static const struct mv88e6xxx_ops mv88e6321_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3673,7 +3697,7 @@ static const struct mv88e6xxx_ops mv88e6341_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3716,7 +3740,7 @@ static const struct mv88e6xxx_ops mv88e6350_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3755,7 +3779,7 @@ static const struct mv88e6xxx_ops mv88e6351_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3799,7 +3823,7 @@ static const struct mv88e6xxx_ops mv88e6352_ops = {
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
 	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_set_sched = mv88e6xxx_port_set_sched,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
@@ -3851,7 +3875,7 @@ static const struct mv88e6xxx_ops mv88e6390_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6390_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3900,7 +3924,7 @@ static const struct mv88e6xxx_ops mv88e6390x_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6390_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
drivers/net/dsa/mv88e6xxx/chip.h +2 -1
index ff3e35eceee0..75fbd5df4aae 100644
@@ -385,7 +385,8 @@ struct mv88e6xxx_ops {
 				   size_t size);
 	int (*port_set_defqpri)(struct mv88e6xxx_chip *chip, int port, u16 pri);
 
-	int (*port_egress_rate_limiting)(struct mv88e6xxx_chip *chip, int port);
+	int (*port_egress_rate_limiting)(struct mv88e6xxx_chip *chip, int port,
+					 u32 count, u32 mode);
 	int (*port_set_sched)(struct mv88e6xxx_chip *chip, int port, u16 sched);
 	int (*port_pause_limit)(struct mv88e6xxx_chip *chip, int port, u8 in,
 				u8 out);
drivers/net/dsa/mv88e6xxx/port.c +96 -10

117 lines not on this page · raw message

drivers/net/dsa/mv88e6xxx/port.h +12 -2
index 710d6eccafae..724f839c570a 100644
@@ -205,13 +205,23 @@
 
 /* Offset 0x09: Egress Rate Control */
 #define MV88E6XXX_PORT_EGRESS_RATE_CTL1		0x09
+#define MV88E6XXX_PORT_EGRESS_DEC_SHIFT		0
+#define MV88E6XXX_PORT_EGRESS_DEC_MASK		0x7f
 
 /* Offset 0x0A: Egress Rate Control 2 */
 #define MV88E6XXX_PORT_EGRESS_RATE_CTL2		0x0a
+#define MV88E6XXX_PORT_EGRESS_COUNT_MODE_SHIFT	14
+#define MV88E6XXX_PORT_EGRESS_COUNT_MODE_MASK	\
+	(0x3 << MV88E6XXX_PORT_EGRESS_COUNT_MODE_SHIFT)
+/* see MV88E6XXX_PORT_EGRESS_COUNT_* in
+ * include/dt-bindings/net/dsa-mv88e6xxx.h
+ */
 #define MV88E6XXX_PORT_SCHED_SHIFT		12
 #define MV88E6XXX_PORT_SCHED_MASK \
 	(0x3 << MV88E6XXX_PORT_SCHED_SHIFT)
 /* see MV88E6XXX_PORT_SCHED_* in include/dt-bindings/net/dsa-mv88e6xxx.h */
+#define MV88E6XXX_PORT_EGRESS_RATE_SHIFT	0
+#define MV88E6XXX_PORT_EGRESS_RATE_MASK		0xfff
 
 /* Offset 0x0B: Port Association Vector */
 #define MV88E6XXX_PORT_ASSOC_VECTOR			0x0b
@@ -335,8 +345,8 @@ int mv88e6xxx_port_set_message_port(struct mv88e6xxx_chip *chip, int port,
 int mv88e6165_port_set_jumbo_size(struct mv88e6xxx_chip *chip, int port,
 				  size_t size);
 int mv88e6xxx_port_set_defqpri(struct mv88e6xxx_chip *chip, int port, u16 pri);
-int mv88e6095_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port);
-int mv88e6097_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port);
+int mv88e6xxx_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port,
+					u32 count, u32 mode);
 int mv88e6xxx_port_set_sched(struct mv88e6xxx_chip *chip, int port, u16 sched);
 int mv88e6097_port_pause_limit(struct mv88e6xxx_chip *chip, int port, u8 in,
 			       u8 out);
Signature · 1 line
2.18.0
#23 lore raw

Re: [PATCH 6/7] net: dsa: mv88e6xxx: add egress rate limiting

Hi Robert,

Quoted 376 lines · Robert Beckett

Add code for specifying egress rate limiting per port. The rate can be specified as ethernet frames or bits per second.

---
 drivers/net/dsa/mv88e6xxx/chip.c |  72 ++++++++++++++-------
 drivers/net/dsa/mv88e6xxx/chip.h |   3 +-
 drivers/net/dsa/mv88e6xxx/port.c | 106 ++++++++++++++++++++++++++++---
 drivers/net/dsa/mv88e6xxx/port.h |  14 +++-
 4 files changed, 158 insertions(+), 37 deletions(-)
drivers/net/dsa/mv88e6xxx/chip.c +48 -24
index 2bc22c59200c..8c116496ab2f 100644
@@ -2120,6 +2120,32 @@ static int mv88e6xxx_set_port_sched(struct mv88e6xxx_chip *chip, int port)
 	return chip->info->ops->port_set_sched(chip, port, (u16)sched);
 }
 
+static int mv88e6xxx_set_port_egress_rate_limiting(struct mv88e6xxx_chip *chip,
+						   int port)
+{
+	struct dsa_switch *ds = chip->ds;
+	struct device_node *dn = ds->ports[port].dn;
+	int err;
+	u32 mode, count;
+
+	if (!dn || !chip->info->ops->port_egress_rate_limiting)
+		return 0;
+
+	err = of_property_read_u32(dn, "egress-limit-mode", &mode);
+	if (err < 0)
+		goto disable;
+
+	err = of_property_read_u32(dn, "egress-limit-count", &count);
+	if (err < 0)
+		goto disable;
+
+	return chip->info->ops->port_egress_rate_limiting(chip, port, count,
+							  mode);
+
+disable:
+	return chip->info->ops->port_egress_rate_limiting(chip, port, 0, 0);
+}
+
 static int mv88e6xxx_setup_port(struct mv88e6xxx_chip *chip, int port)
 {
 	struct dsa_switch *ds = chip->ds;
@@ -2263,11 +2289,9 @@ static int mv88e6xxx_setup_port(struct mv88e6xxx_chip *chip, int port)
 			return err;
 	}
 
-	if (chip->info->ops->port_egress_rate_limiting) {
-		err = chip->info->ops->port_egress_rate_limiting(chip, port);
-		if (err)
-			return err;
-	}
+	err = mv88e6xxx_set_port_egress_rate_limiting(chip, port);
+	if (err)
+		return err;
 
 	err = mv88e6xxx_setup_message_port(chip, port);
 	if (err)
@@ -2809,7 +2833,7 @@ static const struct mv88e6xxx_ops mv88e6085_ops = {
 	.port_set_frame_mode = mv88e6351_port_set_frame_mode,
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -2879,7 +2903,7 @@ static const struct mv88e6xxx_ops mv88e6097_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6095_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -2951,7 +2975,7 @@ static const struct mv88e6xxx_ops mv88e6131_ops = {
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_upstream_port = mv88e6095_port_set_upstream_port,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_set_pause = mv88e6185_port_set_pause,
 	.port_link_state = mv88e6352_port_link_state,
@@ -2994,7 +3018,7 @@ static const struct mv88e6xxx_ops mv88e6141_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3034,7 +3058,7 @@ static const struct mv88e6xxx_ops mv88e6161_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3108,7 +3132,7 @@ static const struct mv88e6xxx_ops mv88e6171_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3150,7 +3174,7 @@ static const struct mv88e6xxx_ops mv88e6172_ops = {
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
 	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_set_sched = mv88e6xxx_port_set_sched,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
@@ -3193,7 +3217,7 @@ static const struct mv88e6xxx_ops mv88e6175_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3235,7 +3259,7 @@ static const struct mv88e6xxx_ops mv88e6176_ops = {
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
 	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_set_sched = mv88e6xxx_port_set_sched,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
@@ -3275,7 +3299,7 @@ static const struct mv88e6xxx_ops mv88e6185_ops = {
 	.port_set_speed = mv88e6185_port_set_speed,
 	.port_set_frame_mode = mv88e6085_port_set_frame_mode,
 	.port_set_egress_floods = mv88e6185_port_set_egress_floods,
-	.port_egress_rate_limiting = mv88e6095_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_set_upstream_port = mv88e6095_port_set_upstream_port,
 	.port_set_pause = mv88e6185_port_set_pause,
 	.port_link_state = mv88e6185_port_link_state,
@@ -3454,7 +3478,7 @@ static const struct mv88e6xxx_ops mv88e6240_ops = {
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
 	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_set_sched = mv88e6xxx_port_set_sched,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
@@ -3587,7 +3611,7 @@ static const struct mv88e6xxx_ops mv88e6320_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3630,7 +3654,7 @@ static const struct mv88e6xxx_ops mv88e6321_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3673,7 +3697,7 @@ static const struct mv88e6xxx_ops mv88e6341_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3716,7 +3740,7 @@ static const struct mv88e6xxx_ops mv88e6350_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3755,7 +3779,7 @@ static const struct mv88e6xxx_ops mv88e6351_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3799,7 +3823,7 @@ static const struct mv88e6xxx_ops mv88e6352_ops = {
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
 	.port_set_defqpri = mv88e6xxx_port_set_defqpri,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_set_sched = mv88e6xxx_port_set_sched,
 	.port_pause_limit = mv88e6097_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
@@ -3851,7 +3875,7 @@ static const struct mv88e6xxx_ops mv88e6390_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6390_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
@@ -3900,7 +3924,7 @@ static const struct mv88e6xxx_ops mv88e6390x_ops = {
 	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
 	.port_set_ether_type = mv88e6351_port_set_ether_type,
 	.port_set_jumbo_size = mv88e6165_port_set_jumbo_size,
-	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
+	.port_egress_rate_limiting = mv88e6xxx_port_egress_rate_limiting,
 	.port_pause_limit = mv88e6390_port_pause_limit,
 	.port_disable_learn_limit = mv88e6xxx_port_disable_learn_limit,
 	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
drivers/net/dsa/mv88e6xxx/chip.h +2 -1
index ff3e35eceee0..75fbd5df4aae 100644
@@ -385,7 +385,8 @@ struct mv88e6xxx_ops {
 				   size_t size);
 	int (*port_set_defqpri)(struct mv88e6xxx_chip *chip, int port, u16 pri);
 
-	int (*port_egress_rate_limiting)(struct mv88e6xxx_chip *chip, int port);
+	int (*port_egress_rate_limiting)(struct mv88e6xxx_chip *chip, int port,
+					 u32 count, u32 mode);
 	int (*port_set_sched)(struct mv88e6xxx_chip *chip, int port, u16 sched);
 	int (*port_pause_limit)(struct mv88e6xxx_chip *chip, int port, u8 in,
 				u8 out);
drivers/net/dsa/mv88e6xxx/port.c +96 -10
index 236732fc598d..41418cfaca56 100644
@@ -1166,21 +1166,107 @@ int mv88e6xxx_port_set_defqpri(struct mv88e6xxx_chip *chip, int port, u16 pri)
 	return mv88e6xxx_port_write(chip, port, MV88E6XXX_PORT_CTL2, reg);
 }
 
-/* Offset 0x09: Port Rate Control */
+/* Offset 0x09: Port Rate Control
+ * Offset 0x0A: Egress Rate Control 2
+ */
 
-int mv88e6095_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port)
+#define Kb			1000
+#define Mb			(1000 * Kb)
+#define Gb			(1000ull * Mb)
+#define EGRESS_FRAME_RATE_MIN	7632
+#define EGRESS_FRAME_RATE_MAX	31250000
+#define EGRESS_BPS_RATE_MIN	(64 * Kb)
+#define EGRESS_BPS_RATE_MAX	(1 * Gb)
+#define EGRESS_RATE_PERIOD	32
+int mv88e6xxx_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port,
+					u32 count, u32 mode)
 {
-	return mv88e6xxx_port_write(chip, port, MV88E6XXX_PORT_EGRESS_RATE_CTL1,
-				    0x0000);
-}
+	u16 reg1, reg2;
+	int err;
 
-int mv88e6097_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port)
-{
-	return mv88e6xxx_port_write(chip, port, MV88E6XXX_PORT_EGRESS_RATE_CTL1,
-				    0x0001);
+	/* quick exit for disabling */
+	if (count == 0) {
+		err = mv88e6xxx_port_read(chip, port,
+					  MV88E6XXX_PORT_EGRESS_RATE_CTL2,
+					  &reg2);
+		if (err)
+			return err;
+		reg2 &= ~MV88E6XXX_PORT_EGRESS_RATE_MASK;
+		err =  mv88e6xxx_port_write(chip, port,
+					    MV88E6XXX_PORT_EGRESS_RATE_CTL2,
+					    reg2);
+		return err;
+	}
+
+	if (mode > MV88E6XXX_PORT_EGRESS_COUNT_MODE_L3)
+		return -EINVAL;
+
+	if (mode == MV88E6XXX_PORT_EGRESS_COUNT_MODE_FRAMES &&
+	    (count < EGRESS_FRAME_RATE_MIN || count > EGRESS_FRAME_RATE_MAX))
+		return -EINVAL;
+
+	if (mode != MV88E6XXX_PORT_EGRESS_COUNT_MODE_FRAMES &&
+	    (count < EGRESS_BPS_RATE_MIN || count > EGRESS_BPS_RATE_MAX))
+		return -EINVAL;
+
+	err = mv88e6xxx_port_read(chip, port, MV88E6XXX_PORT_EGRESS_RATE_CTL1,
+				  &reg1);
+	if (err)
+		return err;
+
+	err = mv88e6xxx_port_read(chip, port, MV88E6XXX_PORT_EGRESS_RATE_CTL2,
+				  &reg2);
+	if (err)
+		return err;
+
+	reg1 &= ~MV88E6XXX_PORT_EGRESS_DEC_MASK;
+	reg2 &= ~MV88E6XXX_PORT_EGRESS_COUNT_MODE_MASK;
+
+	if (mode == MV88E6XXX_PORT_EGRESS_COUNT_MODE_FRAMES) {
+		u32 val;
+
+		/* recommended to use dec of 1 for frame based */
+		reg1 |= 1 << MV88E6XXX_PORT_EGRESS_DEC_SHIFT;
+
+		reg2 |= mode << MV88E6XXX_PORT_EGRESS_COUNT_MODE_SHIFT;
+		reg2 &= ~MV88E6XXX_PORT_EGRESS_RATE_MASK;
+
+		val = NSEC_PER_SEC / (EGRESS_RATE_PERIOD * count);
+		if (NSEC_PER_SEC % (EGRESS_RATE_PERIOD * count))
+			val++;
+		reg2 |= (u16)(val << MV88E6XXX_PORT_EGRESS_RATE_SHIFT);
+	} else {
+		u16 egress_dec, egress_rate;
+		u64 dec_bytes, ns_bits;
+
+		if (count < (1 * Mb))
+			egress_dec = (u16)roundup(count, (64 * Kb));
+		else if (count < (100 * Mb))
+			egress_dec = (u16)roundup(count, (1 * Mb));
+		else
+			egress_dec = (u16)roundup(count, (10 * Mb));
+
+		reg1 |= egress_dec;
+
+		dec_bytes = 8ull * NSEC_PER_SEC * egress_dec;
+		ns_bits = 32ull * count;
+		egress_rate = (u16)div64_u64(dec_bytes, ns_bits);
+		reg2 |= egress_rate;
+	}
+
+	err =  mv88e6xxx_port_write(chip, port, MV88E6XXX_PORT_EGRESS_RATE_CTL1,
+				    reg1);
+	if (err)
+		return err;
+
+	err =  mv88e6xxx_port_write(chip, port, MV88E6XXX_PORT_EGRESS_RATE_CTL2,
+				    reg2);
+	if (err)
+		return err;
+
+	return 0;
 }
 
-/* Offset 0x0A: Egress Rate Control 2 */
 int mv88e6xxx_port_set_sched(struct mv88e6xxx_chip *chip, int port, u16 sched)
 {
 	u16 reg;
drivers/net/dsa/mv88e6xxx/port.h +0 -0
index 710d6eccafae..724f839c570a 100644
@@ -205,13 +205,23 @@
 

drivers/net/dsa/mv88e6xxx/port.h line 215

Quoted 10 lines
drivers/net/dsa/mv88e6xxx/port.h +5 -0
 /* Offset 0x09: Egress Rate Control */
 #define MV88E6XXX_PORT_EGRESS_RATE_CTL1		0x09
+#define MV88E6XXX_PORT_EGRESS_DEC_SHIFT		0
+#define MV88E6XXX_PORT_EGRESS_DEC_MASK		0x7f
 
 /* Offset 0x0A: Egress Rate Control 2 */
 #define MV88E6XXX_PORT_EGRESS_RATE_CTL2		0x0a
+#define MV88E6XXX_PORT_EGRESS_COUNT_MODE_SHIFT	14
+#define MV88E6XXX_PORT_EGRESS_COUNT_MODE_MASK	\
+	(0x3 << MV88E6XXX_PORT_EGRESS_COUNT_MODE_SHIFT)

No shift macros please, only 0x1234 masks and their values named as in the documentation. This way we see clearly how the 16-bit registers are organized.

Quoted 13 lines
drivers/net/dsa/mv88e6xxx/port.h +5 -0
+/* see MV88E6XXX_PORT_EGRESS_COUNT_* in
+ * include/dt-bindings/net/dsa-mv88e6xxx.h
+ */
 #define MV88E6XXX_PORT_SCHED_SHIFT		12
 #define MV88E6XXX_PORT_SCHED_MASK \
 	(0x3 << MV88E6XXX_PORT_SCHED_SHIFT)
 /* see MV88E6XXX_PORT_SCHED_* in include/dt-bindings/net/dsa-mv88e6xxx.h */
+#define MV88E6XXX_PORT_EGRESS_RATE_SHIFT	0
+#define MV88E6XXX_PORT_EGRESS_RATE_MASK		0xfff
 
 /* Offset 0x0B: Port Association Vector */
 #define MV88E6XXX_PORT_ASSOC_VECTOR			0x0b

drivers/net/dsa/mv88e6xxx/port.h int mv88e6xxx_port_set_message_port(struct mv88e6xxx_chip *chip, int port, · line 352

Quoted 10 lines
drivers/net/dsa/mv88e6xxx/port.h +2 -2
@@ -335,8 +345,8 @@ int mv88e6xxx_port_set_message_port(struct mv88e6xxx_chip *chip, int port,
 int mv88e6165_port_set_jumbo_size(struct mv88e6xxx_chip *chip, int port,
 				  size_t size);
 int mv88e6xxx_port_set_defqpri(struct mv88e6xxx_chip *chip, int port, u16 pri);
-int mv88e6095_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port);
-int mv88e6097_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port);
+int mv88e6xxx_port_egress_rate_limiting(struct mv88e6xxx_chip *chip, int port,
+					u32 count, u32 mode);
 int mv88e6xxx_port_set_sched(struct mv88e6xxx_chip *chip, int port, u16 sched);
 int mv88e6097_port_pause_limit(struct mv88e6xxx_chip *chip, int port, u8 in,
 			       u8 out);

This patch does not look good. Implementations in port.c must be simple functions ordered per Port register, implementing read write operations for them, and eventually checking unsupported values. No logic in them. You may abstract some values with an enum defined in chip.h if needed. (some models don't use the same values for various definitions for example.)

Thanks,

Vivien

#24 lore raw

Re: [PATCH 6/7] net: dsa: mv88e6xxx: add egress rate limiting

In reply to Robert Beckett
Thank you for the patch! Yet something to improve:

Hi Robert,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on linus/master]
[cannot apply to v5.3-rc8 next-20190904]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]

url: https://github.com/0day-ci/linux/commits/Robert-Beckett/net-dsa-mv88e6xxx-features-to-handle-network-storms/20190911-142233
config: x86_64-fedora-25 (attached as .config)
compiler: gcc-7 (Debian 7.4.0-11) 7.4.0 reproduce:

        # save the attached .config to linux build tree
        make ARCH=x86_64 

If you fix the issue, kindly add following tag

All errors (new ones prefixed by >>):

>> drivers/net//dsa/mv88e6xxx/chip.c:3529:31: error: 'mv88e6097_port_egress_rate_limiting' undeclared here (not in a function); did you mean 'mv88e6xxx_port_egress_rate_limiting'?

.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ mv88e6xxx_port_egress_rate_limiting

vim +3529 drivers/net//dsa/mv88e6xxx/chip.c

b3469dd8adade1 Vivien Didelot   2016-09-29  3510  
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3511  static const struct mv88e6xxx_ops mv88e6250_ops = {
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3512  	/* MV88E6XXX_FAMILY_6250 */
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3513  	.ieee_pri_map = mv88e6250_g1_ieee_pri_map,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3514  	.ip_pri_map = mv88e6085_g1_ip_pri_map,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3515  	.irl_init_all = mv88e6352_g2_irl_init_all,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3516  	.get_eeprom = mv88e6xxx_g2_get_eeprom16,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3517  	.set_eeprom = mv88e6xxx_g2_set_eeprom16,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3518  	.set_switch_mac = mv88e6xxx_g2_set_switch_mac,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3519  	.phy_read = mv88e6xxx_g2_smi_phy_read,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3520  	.phy_write = mv88e6xxx_g2_smi_phy_write,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3521  	.port_set_link = mv88e6xxx_port_set_link,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3522  	.port_set_duplex = mv88e6xxx_port_set_duplex,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3523  	.port_set_rgmii_delay = mv88e6352_port_set_rgmii_delay,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3524  	.port_set_speed = mv88e6250_port_set_speed,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3525  	.port_tag_remap = mv88e6095_port_tag_remap,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3526  	.port_set_frame_mode = mv88e6351_port_set_frame_mode,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3527  	.port_set_egress_floods = mv88e6352_port_set_egress_floods,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3528  	.port_set_ether_type = mv88e6351_port_set_ether_type,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04 @3529  	.port_egress_rate_limiting = mv88e6097_port_egress_rate_limiting,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3530  	.port_pause_limit = mv88e6097_port_pause_limit,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3531  	.port_disable_pri_override = mv88e6xxx_port_disable_pri_override,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3532  	.port_link_state = mv88e6250_port_link_state,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3533  	.stats_snapshot = mv88e6320_g1_stats_snapshot,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3534  	.stats_set_histogram = mv88e6095_g1_stats_set_histogram,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3535  	.stats_get_sset_count = mv88e6250_stats_get_sset_count,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3536  	.stats_get_strings = mv88e6250_stats_get_strings,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3537  	.stats_get_stats = mv88e6250_stats_get_stats,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3538  	.set_cpu_port = mv88e6095_g1_set_cpu_port,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3539  	.set_egress_port = mv88e6095_g1_set_egress_port,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3540  	.watchdog_ops = &mv88e6250_watchdog_ops,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3541  	.mgmt_rsvd2cpu = mv88e6352_g2_mgmt_rsvd2cpu,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3542  	.pot_clear = mv88e6xxx_g2_pot_clear,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3543  	.reset = mv88e6250_g1_reset,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3544  	.vtu_getnext = mv88e6250_g1_vtu_getnext,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3545  	.vtu_loadpurge = mv88e6250_g1_vtu_loadpurge,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3546  	.phylink_validate = mv88e6065_phylink_validate,
1f71836f5d96e4 Rasmus Villemoes 2019-06-04  3547  };

1f71836f5d96e4 Rasmus Villemoes 2019-06-04 3548

:::::: The code at line 3529 was first introduced by commit
:::::: 1f71836f5d96e4c87fad16db86d324bee47e1d30 net: dsa: mv88e6xxx: add support for mv88e6250

:::::: TO: Rasmus Villemoes <[email protected]> :::::: CC: David S. Miller <[email protected]>

---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation

#25 lore raw

PATCH 7/7 dt-bindings: mv88e6xxx: add egress rate limiting

In reply to Robert Beckett

Document port egress rate limiting settings. Add defines for specifying egress rate limiting mode.

---
 .../devicetree/bindings/net/dsa/marvell.txt   | 22 +++++++++++++++++++
 include/dt-bindings/net/dsa-mv88e6xxx.h       |  5 +++++
 2 files changed, 27 insertions(+)
Documentation/devicetree/bindings/net/dsa/marvell.txt +22 -0
index 7de90929c3c9..d33c1958f420 100644
@@ -62,6 +62,28 @@ Optional properties for ports:
 			  MV88E6XXX_PORT_SCHED_STRICT_ALL - All queues use
 			  strict priority, where queues drain in descending
 			  queue number order.
+- egress-limit-mode=<n>	: Set port egress rate limiting mode. Valid values are:
+			  MV88E6XXX_PORT_EGRESS_COUNT_MODE_FRAMES - Count layer
+			  2 frames (assumed to be 64kb).
+			  MV88E6XXX_PORT_EGRESS_COUNT_MODE_L1 - Count all layer
+			  1 bits
+			  MV88E6XXX_PORT_EGRESS_COUNT_MODE_L2 - Count all layer
+			  2 bits
+			  MV88E6XXX_PORT_EGRESS_COUNT_MODE_L3 - Count all layer
+			  3 bits
+			  Must also specify egress-limit-count.
+- egress-limit-count=<n>: Set port egress rate limiting count. If
+			  egress-limit-mode is FRAMES, this specifies the
+			  maximum number of ethernet frames to allow to egress
+			  from this port per second, otherwise it is number of
+			  bits as counted based on the mode allowed to egress
+			  from this port per second.
+			  The HW has limitations which the driver adheres to:
+			  between 64 Kbps to 1 Mbps in 16 Kbps increments
+			  between 1 Mbps to 100 Mbps in 1Mbps increments
+			  between 100 Mbps to 1 Gbps in 10 Mbps increments.
+			  Other values will be rounded down the previous
+			  increment.
 
 Example:
 
include/dt-bindings/net/dsa-mv88e6xxx.h +5 -0
index 3f62003841ce..33ecd94f5e22 100644
@@ -9,6 +9,11 @@
 #ifndef _DT_BINDINGS_MV88E6XXX_H
 #define _DT_BINDINGS_MV88E6XXX_H
 
+#define MV88E6XXX_PORT_EGRESS_COUNT_MODE_FRAMES	0
+#define MV88E6XXX_PORT_EGRESS_COUNT_MODE_L1	1
+#define MV88E6XXX_PORT_EGRESS_COUNT_MODE_L2	2
+#define MV88E6XXX_PORT_EGRESS_COUNT_MODE_L3	3
+
 #define MV88E6XXX_PORT_SCHED_ROUND_ROBIN	0
 #define MV88E6XXX_PORT_SCHED_STRICT_3		1
 #define MV88E6XXX_PORT_SCHED_STRICT_3_2		2
Signature · 1 line
2.18.0
#26 lore raw
In reply to Robert Beckett

+Ido, Jiri,

Quoted 9 lines · Robert Beckett

This patch-set adds support for some features of the Marvell switch chips that can be used to handle packet storms.

The rationale for this was a setup that requires the ability to receive traffic from one port, while a packet storm is occuring on another port (via an external switch with a deliberate loop). This is needed to ensure vital data delivery from a specific port, while mitigating any loops or DoS that a user may introduce on another port (can't guarantee sensible users).

The use case is reasonable, but the implementation is not really. You are using Device Tree which is meant to describe hardware as a policy holder for setting up queue priorities and likewise for queue scheduling.

The tool that should be used for that purpose is tc and possibly an appropriately offloaded queue scheduler in order to map the desired scheduling class to what the hardware supports.

Jiri, Ido, how do you guys support this with mlxsw?

Quoted 42 lines

[patch 1/7] configures auto negotiation for CPU ports connected with phys to enable pause frame propogation.

[patch 2/7] allows setting of port's default output queue priority for any ingressing packets on that port.

[patch 3/7] dt-bindings for patch 2.

[patch 4/7] allows setting of a port's queue scheduling so that it can prioritise egress of traffic routed from high priority ports.

[patch 5/7] dt-bindings for patch 4.

[patch 6/7] allows ports to rate limit their egress. This can be used to stop the host CPU from becoming swamped by packet delivery and exhasting descriptors.

[patch 7/7] dt-bindings for patch 6.

Robert Beckett (7):
net/dsa: configure autoneg for CPU port
net: dsa: mv88e6xxx: add ability to set default queue priorities per
port
dt-bindings: mv88e6xxx: add ability to set default queue priorities
per port
net: dsa: mv88e6xxx: add ability to set queue scheduling
dt-bindings: mv88e6xxx: add ability to set queue scheduling
net: dsa: mv88e6xxx: add egress rate limiting
dt-bindings: mv88e6xxx: add egress rate limiting

 .../devicetree/bindings/net/dsa/marvell.txt   |  38 +++++
 drivers/net/dsa/mv88e6xxx/chip.c              | 122 ++++++++++++---
 drivers/net/dsa/mv88e6xxx/chip.h              |   5 +-
 drivers/net/dsa/mv88e6xxx/port.c              | 140 +++++++++++++++++-
 drivers/net/dsa/mv88e6xxx/port.h              |  24 ++-
 include/dt-bindings/net/dsa-mv88e6xxx.h       |  22 +++
 net/dsa/port.c                                |  10 ++
 7 files changed, 327 insertions(+), 34 deletions(-)
 create mode 100644 include/dt-bindings/net/dsa-mv88e6xxx.h
Signature · 1 line
Florian
#27 lore raw
Quoted 17 lines · Florian Fainelli

+Ido, Jiri,

Quoted 13 lines · Robert Beckett

This patch-set adds support for some features of the Marvell switch chips that can be used to handle packet storms.

The rationale for this was a setup that requires the ability to receive
traffic from one port, while a packet storm is occuring on another port
(via an external switch with a deliberate loop). This is needed to ensure vital data delivery from a specific port, while mitigating any
loops or DoS that a user may introduce on another port (can't guarantee
sensible users).

Quoted 8 lines

The use case is reasonable, but the implementation is not really. You are using Device Tree which is meant to describe hardware as a policy holder for setting up queue priorities and likewise for queue scheduling.

The tool that should be used for that purpose is tc and possibly an appropriately offloaded queue scheduler in order to map the desired scheduling class to what the hardware supports.

Thanks for the review and tip about tc. Im currently not familiar with that tool. Ill investigate it as an alternative approach.

Quoted 56 lines

Jiri, Ido, how do you guys support this with mlxsw?

Quoted 51 lines

[patch 1/7] configures auto negotiation for CPU ports connected with
phys to enable pause frame propogation.

[patch 2/7] allows setting of port's default output queue priority for
any ingressing packets on that port.

[patch 3/7] dt-bindings for patch 2.

[patch 4/7] allows setting of a port's queue scheduling so that it can
prioritise egress of traffic routed from high priority ports.

[patch 5/7] dt-bindings for patch 4.

[patch 6/7] allows ports to rate limit their egress. This can be used to
stop the host CPU from becoming swamped by packet delivery and exhasting
descriptors.

[patch 7/7] dt-bindings for patch 6.

Robert Beckett (7):
net/dsa: configure autoneg for CPU port
net: dsa: mv88e6xxx: add ability to set default queue priorities
per
port
dt-bindings: mv88e6xxx: add ability to set default queue
priorities
per port
net: dsa: mv88e6xxx: add ability to set queue scheduling
dt-bindings: mv88e6xxx: add ability to set queue scheduling
net: dsa: mv88e6xxx: add egress rate limiting
dt-bindings: mv88e6xxx: add egress rate limiting

 .../devicetree/bindings/net/dsa/marvell.txt   |  38 +++++
 drivers/net/dsa/mv88e6xxx/chip.c              | 122 ++++++++++++
---
 drivers/net/dsa/mv88e6xxx/chip.h              |   5 +-
 drivers/net/dsa/mv88e6xxx/port.c              | 140
+++++++++++++++++-
 drivers/net/dsa/mv88e6xxx/port.h              |  24 ++-
 include/dt-bindings/net/dsa-mv88e6xxx.h       |  22 +++
 net/dsa/port.c                                |  10 ++
 7 files changed, 327 insertions(+), 34 deletions(-)
 create mode 100644 include/dt-bindings/net/dsa-mv88e6xxx.h
#28 lore raw
In reply to Florian Fainelli
Quoted 17 lines · Florian Fainelli

+Ido, Jiri,

Quoted 9 lines · Robert Beckett

This patch-set adds support for some features of the Marvell switch chips that can be used to handle packet storms.

The rationale for this was a setup that requires the ability to receive traffic from one port, while a packet storm is occuring on another port (via an external switch with a deliberate loop). This is needed to ensure vital data delivery from a specific port, while mitigating any loops or DoS that a user may introduce on another port (can't guarantee sensible users).

The use case is reasonable, but the implementation is not really. You are using Device Tree which is meant to describe hardware as a policy holder for setting up queue priorities and likewise for queue scheduling.

Quoted 5 lines

The tool that should be used for that purpose is tc and possibly an appropriately offloaded queue scheduler in order to map the desired scheduling class to what the hardware supports.

Jiri, Ido, how do you guys support this with mlxsw?

Hi Florian,

Are you referring to policing traffic towards the CPU using a policer on the egress of the CPU port? At least that's what I understand from the description of patch 6 below.

If so, mlxsw sets policers for different traffic types during its initialization sequence. These policers are not exposed to the user nor configurable. While the default settings are good for most users, we do want to allow users to change these and expose current settings.

I agree that tc seems like the right choice, but the question is where are we going to install the filters?

Quoted 47 lines
Quoted 43 lines

[patch 1/7] configures auto negotiation for CPU ports connected with phys to enable pause frame propogation.

[patch 2/7] allows setting of port's default output queue priority for any ingressing packets on that port.

[patch 3/7] dt-bindings for patch 2.

[patch 4/7] allows setting of a port's queue scheduling so that it can prioritise egress of traffic routed from high priority ports.

[patch 5/7] dt-bindings for patch 4.

[patch 6/7] allows ports to rate limit their egress. This can be used to stop the host CPU from becoming swamped by packet delivery and exhasting descriptors.

[patch 7/7] dt-bindings for patch 6.

Robert Beckett (7):
net/dsa: configure autoneg for CPU port
net: dsa: mv88e6xxx: add ability to set default queue priorities per
port
dt-bindings: mv88e6xxx: add ability to set default queue priorities
per port
net: dsa: mv88e6xxx: add ability to set queue scheduling
dt-bindings: mv88e6xxx: add ability to set queue scheduling
net: dsa: mv88e6xxx: add egress rate limiting
dt-bindings: mv88e6xxx: add egress rate limiting

 .../devicetree/bindings/net/dsa/marvell.txt   |  38 +++++
 drivers/net/dsa/mv88e6xxx/chip.c              | 122 ++++++++++++---
 drivers/net/dsa/mv88e6xxx/chip.h              |   5 +-
 drivers/net/dsa/mv88e6xxx/port.c              | 140 +++++++++++++++++-
 drivers/net/dsa/mv88e6xxx/port.h              |  24 ++-
 include/dt-bindings/net/dsa-mv88e6xxx.h       |  22 +++
 net/dsa/port.c                                |  10 ++
 7 files changed, 327 insertions(+), 34 deletions(-)
 create mode 100644 include/dt-bindings/net/dsa-mv88e6xxx.h
Signature · 1 line
Florian
#29 lore raw
Quoted 48 lines · Ido Schimmel
Quoted 26 lines · Florian Fainelli

+Ido, Jiri,

Quoted 4 lines · Robert Beckett

This patch-set adds support for some features of the Marvell switch
chips that can be used to handle packet storms.

Quoted 11 lines

The rationale for this was a setup that requires the ability to receive
traffic from one port, while a packet storm is occuring on another port
(via an external switch with a deliberate loop). This is needed to
ensure vital data delivery from a specific port, while mitigating any
loops or DoS that a user may introduce on another port (can't guarantee
sensible users).

The use case is reasonable, but the implementation is not really. You
are using Device Tree which is meant to describe hardware as a policy
holder for setting up queue priorities and likewise for queue scheduling.

Quoted 5 lines

The tool that should be used for that purpose is tc and possibly an appropriately offloaded queue scheduler in order to map the desired scheduling class to what the hardware supports.

Jiri, Ido, how do you guys support this with mlxsw?

Hi Florian,

Are you referring to policing traffic towards the CPU using a policer on
the egress of the CPU port? At least that's what I understand from the
description of patch 6 below.

If so, mlxsw sets policers for different traffic types during its initialization sequence. These policers are not exposed to the user nor
configurable. While the default settings are good for most users, we do
want to allow users to change these and expose current settings.

Quoted 4 lines

I agree that tc seems like the right choice, but the question is where
are we going to install the filters?

Before I go too far down the rabbit hole of tc traffic shaping, maybe it would be good to explain in more detail the problem I am trying to solve.

We have a setup as follows:

Marvell 88E6240 switch chip, accepting traffic from 4 ports. Port 1 (P1) is critical priority, no dropped packets allowed, all others can be best effort.

CPU port of swtich chip is connected via phy to phy of intel i210 (igb driver).

i210 is connected via pcie switch to imx6.

When too many small packets attempt to be delivered to CPU port (e.g. during broadcast flood) we saw dropped packets.

The packets were being received by i210 in to rx descriptor buffer fine, but the CPU could not keep up with the load. We saw rx_fifo_errors increasing rapidly and ksoftirqd at ~100% CPU.

With this in mind, I am wondering whether any amount of tc traffic shaping would help? Would tc shaping require that the packet reception manages to keep up before it can enact its policies? Does the infrastructure have accelerator offload hooks to be able to apply it via HW? I dont see how it would be able to inspect the packets to apply filtering if they were dropped due to rx descriptor exhaustion. (please bear with me with the basic questions, I am not familiar with this part of the stack).

Assuming that tc is still the way to go, after a brief look in to the man pages and the documentation at largc.org, it seems like it would need to use the ingress qdisc, with some sort of system to segregate and priortise based on ingress port. Is this possible?

Quoted 56 lines
Quoted 56 lines
Quoted 52 lines

[patch 1/7] configures auto negotiation for CPU ports connected with
phys to enable pause frame propogation.

[patch 2/7] allows setting of port's default output queue priority for
any ingressing packets on that port.

[patch 3/7] dt-bindings for patch 2.

[patch 4/7] allows setting of a port's queue scheduling so that it can
prioritise egress of traffic routed from high priority ports.

[patch 5/7] dt-bindings for patch 4.

[patch 6/7] allows ports to rate limit their egress. This can be used to
stop the host CPU from becoming swamped by packet delivery and exhasting
descriptors.

[patch 7/7] dt-bindings for patch 6.

Robert Beckett (7):
net/dsa: configure autoneg for CPU port
net: dsa: mv88e6xxx: add ability to set default queue
priorities per
port
dt-bindings: mv88e6xxx: add ability to set default queue
priorities
per port
net: dsa: mv88e6xxx: add ability to set queue scheduling
dt-bindings: mv88e6xxx: add ability to set queue scheduling
net: dsa: mv88e6xxx: add egress rate limiting
dt-bindings: mv88e6xxx: add egress rate limiting

 .../devicetree/bindings/net/dsa/marvell.txt   |  38 +++++
 drivers/net/dsa/mv88e6xxx/chip.c              | 122
++++++++++++---
 drivers/net/dsa/mv88e6xxx/chip.h              |   5 +-
 drivers/net/dsa/mv88e6xxx/port.c              | 140
+++++++++++++++++-
 drivers/net/dsa/mv88e6xxx/port.h              |  24 ++-
 include/dt-bindings/net/dsa-mv88e6xxx.h       |  22 +++
 net/dsa/port.c                                |  10 ++
 7 files changed, 327 insertions(+), 34 deletions(-)
 create mode 100644 include/dt-bindings/net/dsa-mv88e6xxx.h
Signature · 1 line
Florian
maintainer
#30 lore raw
Quoted 14 lines

We have a setup as follows:

Marvell 88E6240 switch chip, accepting traffic from 4 ports. Port 1 (P1) is critical priority, no dropped packets allowed, all others can be best effort.

CPU port of swtich chip is connected via phy to phy of intel i210 (igb driver).

i210 is connected via pcie switch to imx6.

When too many small packets attempt to be delivered to CPU port (e.g. during broadcast flood) we saw dropped packets.

Quoted 7 lines

The packets were being received by i210 in to rx descriptor buffer fine, but the CPU could not keep up with the load. We saw rx_fifo_errors increasing rapidly and ksoftirqd at ~100% CPU.

With this in mind, I am wondering whether any amount of tc traffic shaping would help?

Hi Robert

The model in linux is that you start with a software TC filter, and then offload it to the hardware. So the user configures TC just as normal, and then that is used to program the hardware to do the same thing as what would happen in software. This is exactly the same as we do with bridging. You create a software bridge and add interfaces to the bridge. This then gets offloaded to the hardware and it does the bridging for you.

So think about how your can model the Marvell switch capabilities using TC, and implement offload support for it.

Andrew

#31 lore raw
Quoted 2 lines · Andrew Lunn

So think about how your can model the Marvell switch capabilities using TC, and implement offload support for it.

+1 :)

#32 lore raw
In reply to Robert Beckett
Quoted 86 lines · Robert Beckett
Quoted 48 lines · Ido Schimmel
Quoted 26 lines · Florian Fainelli

+Ido, Jiri,

Quoted 4 lines · Robert Beckett

This patch-set adds support for some features of the Marvell switch
chips that can be used to handle packet storms.

Quoted 11 lines

The rationale for this was a setup that requires the ability to receive
traffic from one port, while a packet storm is occuring on another port
(via an external switch with a deliberate loop). This is needed to
ensure vital data delivery from a specific port, while mitigating any
loops or DoS that a user may introduce on another port (can't guarantee
sensible users).

The use case is reasonable, but the implementation is not really. You
are using Device Tree which is meant to describe hardware as a policy
holder for setting up queue priorities and likewise for queue scheduling.

Quoted 5 lines

The tool that should be used for that purpose is tc and possibly an appropriately offloaded queue scheduler in order to map the desired scheduling class to what the hardware supports.

Jiri, Ido, how do you guys support this with mlxsw?

Hi Florian,

Are you referring to policing traffic towards the CPU using a policer on
the egress of the CPU port? At least that's what I understand from the
description of patch 6 below.

If so, mlxsw sets policers for different traffic types during its initialization sequence. These policers are not exposed to the user nor
configurable. While the default settings are good for most users, we do
want to allow users to change these and expose current settings.

Quoted 4 lines

I agree that tc seems like the right choice, but the question is where
are we going to install the filters?

Before I go too far down the rabbit hole of tc traffic shaping, maybe it would be good to explain in more detail the problem I am trying to solve.

We have a setup as follows:

Marvell 88E6240 switch chip, accepting traffic from 4 ports. Port 1 (P1) is critical priority, no dropped packets allowed, all others can be best effort.

CPU port of swtich chip is connected via phy to phy of intel i210 (igb driver).

i210 is connected via pcie switch to imx6.

When too many small packets attempt to be delivered to CPU port (e.g. during broadcast flood) we saw dropped packets.

The packets were being received by i210 in to rx descriptor buffer fine, but the CPU could not keep up with the load. We saw rx_fifo_errors increasing rapidly and ksoftirqd at ~100% CPU.

With this in mind, I am wondering whether any amount of tc traffic shaping would help? Would tc shaping require that the packet reception manages to keep up before it can enact its policies? Does the infrastructure have accelerator offload hooks to be able to apply it via HW? I dont see how it would be able to inspect the packets to apply filtering if they were dropped due to rx descriptor exhaustion. (please bear with me with the basic questions, I am not familiar with this part of the stack).

Quoted 4 lines

Assuming that tc is still the way to go, after a brief look in to the man pages and the documentation at largc.org, it seems like it would need to use the ingress qdisc, with some sort of system to segregate and priortise based on ingress port. Is this possible?

Hi Robert,

As I see it, you have two problems here:

1. Classification: Based on ingress port in your case

2. Scheduling: How to schedule between the different transmission queues

Where the port from which the packets should egress is the CPU port, before they cross the PCI towards the imx6.

Both of these issues can be solved by tc. The main problem is that today we do not have a netdev to represent the CPU port and therefore can't use existing infra like tc. I believe we need to create one. Besides scheduling, we can also use it to permit/deny certain traffic from reaching the CPU and perform policing.

Drivers can run the received packets via taps using dev_queue_xmit_nit(), so that users will see all the traffic directed at the host when running tcpdump on this netdev.

Quoted 60 lines
Quoted 59 lines
Quoted 59 lines
Quoted 55 lines

[patch 1/7] configures auto negotiation for CPU ports connected with
phys to enable pause frame propogation.

[patch 2/7] allows setting of port's default output queue priority for
any ingressing packets on that port.

[patch 3/7] dt-bindings for patch 2.

[patch 4/7] allows setting of a port's queue scheduling so that it can
prioritise egress of traffic routed from high priority ports.

[patch 5/7] dt-bindings for patch 4.

[patch 6/7] allows ports to rate limit their egress. This can be used to
stop the host CPU from becoming swamped by packet delivery and exhasting
descriptors.

[patch 7/7] dt-bindings for patch 6.

Robert Beckett (7):
net/dsa: configure autoneg for CPU port
net: dsa: mv88e6xxx: add ability to set default queue
priorities per
port
dt-bindings: mv88e6xxx: add ability to set default queue
priorities
per port
net: dsa: mv88e6xxx: add ability to set queue scheduling
dt-bindings: mv88e6xxx: add ability to set queue scheduling
net: dsa: mv88e6xxx: add egress rate limiting
dt-bindings: mv88e6xxx: add egress rate limiting

 .../devicetree/bindings/net/dsa/marvell.txt   |  38 +++++
 drivers/net/dsa/mv88e6xxx/chip.c              | 122
++++++++++++---
 drivers/net/dsa/mv88e6xxx/chip.h              |   5 +-
 drivers/net/dsa/mv88e6xxx/port.c              | 140
+++++++++++++++++-
 drivers/net/dsa/mv88e6xxx/port.h              |  24 ++-
 include/dt-bindings/net/dsa-mv88e6xxx.h       |  22 +++
 net/dsa/port.c                                |  10 ++
 7 files changed, 327 insertions(+), 34 deletions(-)
 create mode 100644 include/dt-bindings/net/dsa-mv88e6xxx.h
Signature · 1 line
Florian
maintainer
#33 lore raw
Quoted 4 lines

2. Scheduling: How to schedule between the different transmission queues

Where the port from which the packets should egress is the CPU port, before they cross the PCI towards the imx6.

Hi Ido

This is DSA, so the switch is connected via Ethernet to the IMX6, not PCI. Minor detail, but that really is the core of what makes DSA DSA.

Andrew

#34 lore raw
In reply to Ido Schimmel
Quoted 100 lines · Ido Schimmel
Quoted 86 lines · Robert Beckett
Quoted 48 lines · Ido Schimmel
Quoted 26 lines · Florian Fainelli

+Ido, Jiri,

Quoted 4 lines · Robert Beckett

This patch-set adds support for some features of the Marvell switch
chips that can be used to handle packet storms.

Quoted 11 lines

The rationale for this was a setup that requires the ability to receive
traffic from one port, while a packet storm is occuring on another port
(via an external switch with a deliberate loop). This is needed to
ensure vital data delivery from a specific port, while mitigating any
loops or DoS that a user may introduce on another port (can't guarantee
sensible users).

The use case is reasonable, but the implementation is not really. You
are using Device Tree which is meant to describe hardware as a policy
holder for setting up queue priorities and likewise for queue scheduling.

Quoted 5 lines

The tool that should be used for that purpose is tc and possibly an appropriately offloaded queue scheduler in order to map the desired scheduling class to what the hardware supports.

Jiri, Ido, how do you guys support this with mlxsw?

Hi Florian,

Are you referring to policing traffic towards the CPU using a policer on
the egress of the CPU port? At least that's what I understand from the
description of patch 6 below.

If so, mlxsw sets policers for different traffic types during its initialization sequence. These policers are not exposed to the user nor
configurable. While the default settings are good for most users, we do
want to allow users to change these and expose current settings.

Quoted 4 lines

I agree that tc seems like the right choice, but the question is where
are we going to install the filters?

Before I go too far down the rabbit hole of tc traffic shaping, maybe it would be good to explain in more detail the problem I am trying to solve.

We have a setup as follows:

Marvell 88E6240 switch chip, accepting traffic from 4 ports. Port 1 (P1) is critical priority, no dropped packets allowed, all others can be best effort.

CPU port of swtich chip is connected via phy to phy of intel i210 (igb driver).

i210 is connected via pcie switch to imx6.

When too many small packets attempt to be delivered to CPU port (e.g. during broadcast flood) we saw dropped packets.

The packets were being received by i210 in to rx descriptor buffer fine, but the CPU could not keep up with the load. We saw rx_fifo_errors increasing rapidly and ksoftirqd at ~100% CPU.

With this in mind, I am wondering whether any amount of tc traffic shaping would help? Would tc shaping require that the packet reception manages to keep up before it can enact its policies? Does the infrastructure have accelerator offload hooks to be able to apply it via HW? I dont see how it would be able to inspect the packets to apply filtering if they were dropped due to rx descriptor exhaustion. (please bear with me with the basic questions, I am not familiar with this part of the stack).

Quoted 4 lines

Assuming that tc is still the way to go, after a brief look in to the man pages and the documentation at largc.org, it seems like it would need to use the ingress qdisc, with some sort of system to segregate and priortise based on ingress port. Is this possible?

Hi Robert,

As I see it, you have two problems here:

1. Classification: Based on ingress port in your case

2. Scheduling: How to schedule between the different transmission queues

Quoted 8 lines

Where the port from which the packets should egress is the CPU port, before they cross the PCI towards the imx6.

Both of these issues can be solved by tc. The main problem is that today we do not have a netdev to represent the CPU port and therefore can't use existing infra like tc. I believe we need to create one. Besides scheduling, we can also use it to permit/deny certain traffic from reaching the CPU and perform policing.

We do not necessarily have to create a CPU netdev, we can overlay netdev operations onto the DSA master interface (fec in that case), and whenever you configure the DSA master interface, we also call back into the switch side for the CPU port. This is not necessarily the cleanest way to do things, but that is how we support ethtool operations (and some netdev operations incidentally), and it works

Signature · 1 line
Florian
#35 lore raw
Quoted 148 lines · Florian Fainelli
Quoted 138 lines · Ido Schimmel
Quoted 116 lines · Robert Beckett
Quoted 64 lines · Ido Schimmel
Quoted 33 lines · Florian Fainelli

+Ido, Jiri,

Quoted 5 lines · Robert Beckett

This patch-set adds support for some features of the Marvell
switch
chips that can be used to handle packet storms.

Quoted 15 lines

The rationale for this was a setup that requires the ability to
receive
traffic from one port, while a packet storm is occuring on another port
(via an external switch with a deliberate loop). This is needed
to
ensure vital data delivery from a specific port, while mitigating
any
loops or DoS that a user may introduce on another port (can't
guarantee
sensible users).

The use case is reasonable, but the implementation is not really.
You
are using Device Tree which is meant to describe hardware as a
policy
holder for setting up queue priorities and likewise for queue scheduling.

Quoted 7 lines

The tool that should be used for that purpose is tc and possibly an
appropriately offloaded queue scheduler in order to map the desired
scheduling class to what the hardware supports.

Jiri, Ido, how do you guys support this with mlxsw?

Hi Florian,

Are you referring to policing traffic towards the CPU using a policer
on
the egress of the CPU port? At least that's what I understand from
the
description of patch 6 below.

If so, mlxsw sets policers for different traffic types during its
initialization sequence. These policers are not exposed to the user
nor
configurable. While the default settings are good for most users, we
do
want to allow users to change these and expose current settings.

Quoted 5 lines

I agree that tc seems like the right choice, but the question is
where
are we going to install the filters?

Before I go too far down the rabbit hole of tc traffic shaping, maybe
it would be good to explain in more detail the problem I am trying to
solve.

We have a setup as follows:

Marvell 88E6240 switch chip, accepting traffic from 4 ports. Port 1
(P1) is critical priority, no dropped packets allowed, all others can
be best effort.

CPU port of swtich chip is connected via phy to phy of intel i210 (igb
driver).

i210 is connected via pcie switch to imx6.

When too many small packets attempt to be delivered to CPU port (e.g.
during broadcast flood) we saw dropped packets.

The packets were being received by i210 in to rx descriptor buffer
fine, but the CPU could not keep up with the load. We saw rx_fifo_errors increasing rapidly and ksoftirqd at ~100% CPU.

With this in mind, I am wondering whether any amount of tc traffic
shaping would help? Would tc shaping require that the packet reception
manages to keep up before it can enact its policies? Does the infrastructure have accelerator offload hooks to be able to apply it
via HW? I dont see how it would be able to inspect the packets to apply
filtering if they were dropped due to rx descriptor exhaustion. (please
bear with me with the basic questions, I am not familiar with this part
of the stack).

Quoted 7 lines

Assuming that tc is still the way to go, after a brief look in to the
man pages and the documentation at largc.org, it seems like it would
need to use the ingress qdisc, with some sort of system to segregate
and priortise based on ingress port. Is this possible?

Hi Robert,

As I see it, you have two problems here:

1. Classification: Based on ingress port in your case

2. Scheduling: How to schedule between the different transmission queues

Where the port from which the packets should egress is the CPU port,
before they cross the PCI towards the imx6.

Quoted 8 lines

Both of these issues can be solved by tc. The main problem is that today
we do not have a netdev to represent the CPU port and therefore can't
use existing infra like tc. I believe we need to create one. Besides
scheduling, we can also use it to permit/deny certain traffic from reaching the CPU and perform policing.

Quoted 9 lines

We do not necessarily have to create a CPU netdev, we can overlay netdev
operations onto the DSA master interface (fec in that case), and whenever you configure the DSA master interface, we also call back into
the switch side for the CPU port. This is not necessarily the cleanest
way to do things, but that is how we support ethtool operations (and some netdev operations incidentally), and it works

After reading up on tc, I am not sure how this would work given the semantics of the tool currently.

My initial thought was to model the switch's 4 output queues using an mqprio qdisc for the CPU port, and then use either iptables's classify module on the input ports to set which queue it egresses from on the CPU port, or use vlan tagging with id 0 and priority set. (with the many detail of how to implement them still left to discover).

However, it looks like the mqprio qdisc could only be used for egress, so without a netdev representing the CPU port, I dont know how it could be used.

Another thing I thought of using was just to use iptable's TOS module to set the minimal delay bit and rely on default behaviours, but Ive yet to find anything in the Marvell manual that indicates it could set that bit on all frames entering a port.

Another option might be to use vlans with their priority bits being used to steer to output queues, but I really dont want to introduce more virtual interfaces in to the setup, and I cant see how to configure an enforce default vlan tag with id 0 and priority bits set via linux userland tools.

It does look like tc would be quite nice for configuring the egress rate limiting assuming we a netdev to target with the rate controls of the qdisc.

So far, this seems like I am trying to shoe horn this stuff in to tc. It seems like tc is meant to configure how the ip stack configures flow within the stack, whereas in a switch chip, the packets go nowhere near the CPUs kernel ip stack. I cant help thinking that it would be good have a specific utility for configuring switches that operates on the port level for manage flow within the chip, or maybe simple sysfs attributes to set the ports priority.

#36 lore raw
Quoted 168 lines · Robert Beckett
Quoted 148 lines · Florian Fainelli
Quoted 138 lines · Ido Schimmel
Quoted 116 lines · Robert Beckett
Quoted 64 lines · Ido Schimmel
Quoted 33 lines · Florian Fainelli

+Ido, Jiri,

Quoted 5 lines · Robert Beckett

This patch-set adds support for some features of the Marvell
switch
chips that can be used to handle packet storms.

Quoted 15 lines

The rationale for this was a setup that requires the ability to
receive
traffic from one port, while a packet storm is occuring on another port
(via an external switch with a deliberate loop). This is needed
to
ensure vital data delivery from a specific port, while mitigating
any
loops or DoS that a user may introduce on another port (can't
guarantee
sensible users).

The use case is reasonable, but the implementation is not really.
You
are using Device Tree which is meant to describe hardware as a
policy
holder for setting up queue priorities and likewise for queue scheduling.

Quoted 7 lines

The tool that should be used for that purpose is tc and possibly an
appropriately offloaded queue scheduler in order to map the desired
scheduling class to what the hardware supports.

Jiri, Ido, how do you guys support this with mlxsw?

Hi Florian,

Are you referring to policing traffic towards the CPU using a policer
on
the egress of the CPU port? At least that's what I understand from
the
description of patch 6 below.

If so, mlxsw sets policers for different traffic types during its
initialization sequence. These policers are not exposed to the user
nor
configurable. While the default settings are good for most users, we
do
want to allow users to change these and expose current settings.

Quoted 5 lines

I agree that tc seems like the right choice, but the question is
where
are we going to install the filters?

Before I go too far down the rabbit hole of tc traffic shaping, maybe
it would be good to explain in more detail the problem I am trying to
solve.

We have a setup as follows:

Marvell 88E6240 switch chip, accepting traffic from 4 ports. Port 1
(P1) is critical priority, no dropped packets allowed, all others can
be best effort.

CPU port of swtich chip is connected via phy to phy of intel i210 (igb
driver).

i210 is connected via pcie switch to imx6.

When too many small packets attempt to be delivered to CPU port (e.g.
during broadcast flood) we saw dropped packets.

The packets were being received by i210 in to rx descriptor buffer
fine, but the CPU could not keep up with the load. We saw rx_fifo_errors increasing rapidly and ksoftirqd at ~100% CPU.

With this in mind, I am wondering whether any amount of tc traffic
shaping would help? Would tc shaping require that the packet reception
manages to keep up before it can enact its policies? Does the infrastructure have accelerator offload hooks to be able to apply it
via HW? I dont see how it would be able to inspect the packets to apply
filtering if they were dropped due to rx descriptor exhaustion. (please
bear with me with the basic questions, I am not familiar with this part
of the stack).

Quoted 7 lines

Assuming that tc is still the way to go, after a brief look in to the
man pages and the documentation at largc.org, it seems like it would
need to use the ingress qdisc, with some sort of system to segregate
and priortise based on ingress port. Is this possible?

Hi Robert,

As I see it, you have two problems here:

1. Classification: Based on ingress port in your case

2. Scheduling: How to schedule between the different transmission queues

Where the port from which the packets should egress is the CPU port,
before they cross the PCI towards the imx6.

Quoted 8 lines

Both of these issues can be solved by tc. The main problem is that today
we do not have a netdev to represent the CPU port and therefore can't
use existing infra like tc. I believe we need to create one. Besides
scheduling, we can also use it to permit/deny certain traffic from reaching the CPU and perform policing.

Quoted 9 lines

We do not necessarily have to create a CPU netdev, we can overlay netdev
operations onto the DSA master interface (fec in that case), and whenever you configure the DSA master interface, we also call back into
the switch side for the CPU port. This is not necessarily the cleanest
way to do things, but that is how we support ethtool operations (and some netdev operations incidentally), and it works

After reading up on tc, I am not sure how this would work given the semantics of the tool currently.

My initial thought was to model the switch's 4 output queues using an mqprio qdisc for the CPU port, and then use either iptables's classify module on the input ports to set which queue it egresses from on the CPU port, or use vlan tagging with id 0 and priority set. (with the many detail of how to implement them still left to discover).

Quoted 3 lines

However, it looks like the mqprio qdisc could only be used for egress, so without a netdev representing the CPU port, I dont know how it could be used.

If you are looking at mapping your DSA master/CPU port egress queues to actual switch egress queues, you can look at what bcm_sf2.c and bcmsysport.c do and read the commit messages that introduced the mapping functionality for background on why this was done. In a nutshell, the hardware has the ability to back pressure the Ethernet MAC behind the CPU port in order to automatically rate limit the egress out of the switch. So for instance, if your CPU tries to send 1Gb/sec of traffic to a port that is linked to a link partner at 100Mbits/sec, there is out of band information between the switch and the Ethernet DMA of the CPU port to pace the TX completion interrupt rate to match 100Mbits/sec.

This is going to be different for you here obviously because the hardware has not been specifically designed for that, so you do need to rely on more standard constructs, like actual egress QoS on both ends.

Quoted 18 lines

Another thing I thought of using was just to use iptable's TOS module to set the minimal delay bit and rely on default behaviours, but Ive yet to find anything in the Marvell manual that indicates it could set that bit on all frames entering a port.

Another option might be to use vlans with their priority bits being used to steer to output queues, but I really dont want to introduce more virtual interfaces in to the setup, and I cant see how to configure an enforce default vlan tag with id 0 and priority bits set via linux userland tools.

It does look like tc would be quite nice for configuring the egress rate limiting assuming we a netdev to target with the rate controls of the qdisc.

Quoted 7 lines

So far, this seems like I am trying to shoe horn this stuff in to tc. It seems like tc is meant to configure how the ip stack configures flow within the stack, whereas in a switch chip, the packets go nowhere near the CPUs kernel ip stack. I cant help thinking that it would be good have a specific utility for configuring switches that operates on the port level for manage flow within the chip, or maybe simple sysfs attributes to set the ports priority.

I am not looking at tc the same way you are doing, tc is just the tool to configure all QoS/ingress/egress related operations on a network device. Whether that network device can offload some of the TC operations or not is where things get interesting.

TC has ingress filtering support, which is what you could use for offloading broadcast storms, I would imagine that the following should be possible to be offloaded (this is not a working command but you get the idea):

tc qdisc add dev sw0p0 handle ffff: ingress
tc filter add dev sw0p0 parent ffff: protocol ip prio 1 u32 match ether src 0xfffffffffffff police rate 100k burst 10k skip_sw

something along those lines is how I would implement ingress rate limiting leveraging what the switch could do. This might mean adding support for offloading specific TC filters, Jiri and Ido can certainly suggest a cleverer way of achieving that same functionality.

Signature · 1 line
Florian
#37 lore raw
Quoted 279 lines · Florian Fainelli
Quoted 196 lines · Robert Beckett
Quoted 171 lines · Florian Fainelli
Quoted 159 lines · Ido Schimmel
Quoted 134 lines · Robert Beckett
Quoted 73 lines · Ido Schimmel
Quoted 37 lines · Florian Fainelli

+Ido, Jiri,

Quoted 14 lines · Robert Beckett

This patch-set adds support for some features of the Marvell
switch
chips that can be used to handle packet storms.

The rationale for this was a setup that requires the ability to
receive
traffic from one port, while a packet storm is occuring on
another port
(via an external switch with a deliberate loop). This is
needed

Quoted 8 lines

to
ensure vital data delivery from a specific port, while mitigating
any
loops or DoS that a user may introduce on another port (can't
guarantee
sensible users).

The use case is reasonable, but the implementation is not really.
You
are using Device Tree which is meant to describe hardware as
a
policy
holder for setting up queue priorities and likewise for queue
scheduling.

Quoted 8 lines

The tool that should be used for that purpose is tc and possibly an
appropriately offloaded queue scheduler in order to map the
desired
scheduling class to what the hardware supports.

Jiri, Ido, how do you guys support this with mlxsw?

Hi Florian,

Are you referring to policing traffic towards the CPU using a
policer
on
the egress of the CPU port? At least that's what I understand
from
the
description of patch 6 below.

If so, mlxsw sets policers for different traffic types during
its
initialization sequence. These policers are not exposed to the
user
nor
configurable. While the default settings are good for most users, we
do
want to allow users to change these and expose current settings.

Quoted 6 lines

I agree that tc seems like the right choice, but the question
is
where
are we going to install the filters?

Before I go too far down the rabbit hole of tc traffic shaping,
maybe
it would be good to explain in more detail the problem I am trying to
solve.

We have a setup as follows:

Marvell 88E6240 switch chip, accepting traffic from 4 ports. Port
1
(P1) is critical priority, no dropped packets allowed, all others
can
be best effort.

CPU port of swtich chip is connected via phy to phy of intel i210
(igb
driver).

i210 is connected via pcie switch to imx6.

When too many small packets attempt to be delivered to CPU port
(e.g.
during broadcast flood) we saw dropped packets.

The packets were being received by i210 in to rx descriptor buffer
fine, but the CPU could not keep up with the load. We saw rx_fifo_errors increasing rapidly and ksoftirqd at ~100% CPU.

With this in mind, I am wondering whether any amount of tc traffic
shaping would help? Would tc shaping require that the packet reception
manages to keep up before it can enact its policies? Does the infrastructure have accelerator offload hooks to be able to apply
it
via HW? I dont see how it would be able to inspect the packets to
apply
filtering if they were dropped due to rx descriptor exhaustion.
(please
bear with me with the basic questions, I am not familiar with this part
of the stack).

Quoted 9 lines

Assuming that tc is still the way to go, after a brief look in to
the
man pages and the documentation at largc.org, it seems like it
would
need to use the ingress qdisc, with some sort of system to segregate
and priortise based on ingress port. Is this possible?

Hi Robert,

As I see it, you have two problems here:

1. Classification: Based on ingress port in your case

2. Scheduling: How to schedule between the different transmission
queues

Where the port from which the packets should egress is the CPU port,
before they cross the PCI towards the imx6.

Quoted 10 lines

Both of these issues can be solved by tc. The main problem is that
today
we do not have a netdev to represent the CPU port and therefore can't
use existing infra like tc. I believe we need to create one. Besides
scheduling, we can also use it to permit/deny certain traffic from
reaching the CPU and perform policing.

Quoted 11 lines

We do not necessarily have to create a CPU netdev, we can overlay netdev
operations onto the DSA master interface (fec in that case), and whenever you configure the DSA master interface, we also call back
into
the switch side for the CPU port. This is not necessarily the cleanest
way to do things, but that is how we support ethtool operations (and
some netdev operations incidentally), and it works

After reading up on tc, I am not sure how this would work given the semantics of the tool currently.

My initial thought was to model the switch's 4 output queues using an
mqprio qdisc for the CPU port, and then use either iptables's classify
module on the input ports to set which queue it egresses from on the
CPU port, or use vlan tagging with id 0 and priority set. (with the many detail of how to implement them still left to discover).

Quoted 5 lines

However, it looks like the mqprio qdisc could only be used for egress,
so without a netdev representing the CPU port, I dont know how it could
be used.

If you are looking at mapping your DSA master/CPU port egress queues to
actual switch egress queues, you can look at what bcm_sf2.c and bcmsysport.c do and read the commit messages that introduced the mapping
functionality for background on why this was done. In a nutshell, the hardware has the ability to back pressure the Ethernet MAC behind the CPU port in order to automatically rate limit the egress out of the switch. So for instance, if your CPU tries to send 1Gb/sec of traffic to
a port that is linked to a link partner at 100Mbits/sec, there is out of
band information between the switch and the Ethernet DMA of the CPU port
to pace the TX completion interrupt rate to match 100Mbits/sec.

This is going to be different for you here obviously because the hardware has not been specifically designed for that, so you do need to
rely on more standard constructs, like actual egress QoS on both ends.

Quoted 23 lines

Another thing I thought of using was just to use iptable's TOS module
to set the minimal delay bit and rely on default behaviours, but Ive
yet to find anything in the Marvell manual that indicates it could set
that bit on all frames entering a port.

Another option might be to use vlans with their priority bits being used to steer to output queues, but I really dont want to introduce more virtual interfaces in to the setup, and I cant see how to configure an enforce default vlan tag with id 0 and priority bits set
via linux userland tools.

It does look like tc would be quite nice for configuring the egress rate limiting assuming we a netdev to target with the rate controls of
the qdisc.

Quoted 12 lines

So far, this seems like I am trying to shoe horn this stuff in to tc.
It seems like tc is meant to configure how the ip stack configures flow within the stack, whereas in a switch chip, the packets go nowhere
near the CPUs kernel ip stack. I cant help thinking that it would be
good have a specific utility for configuring switches that operates on
the port level for manage flow within the chip, or maybe simple sysfs
attributes to set the ports priority.

I am not looking at tc the same way you are doing, tc is just the tool
to configure all QoS/ingress/egress related operations on a network device. Whether that network device can offload some of the TC operations or not is where things get interesting.

TC has ingress filtering support, which is what you could use for offloading broadcast storms, I would imagine that the following should
be possible to be offloaded (this is not a working command but you get
the idea):

tc qdisc add dev sw0p0 handle ffff: ingress
tc filter add dev sw0p0 parent ffff: protocol ip prio 1 u32 match ether
src 0xfffffffffffff police rate 100k burst 10k skip_sw

Quoted 5 lines

something along those lines is how I would implement ingress rate limiting leveraging what the switch could do. This might mean adding support for offloading specific TC filters, Jiri and Ido can certainly
suggest a cleverer way of achieving that same functionality.

Thanks for your thoughts on this, its been very helpful in leanring the stack and coming up with ideas for a better design.

I wrote up a set of high level options for discussions internally, and would appreciate any feedback you had on them:

To get this upstreamed, I think we will need something like the following high level design:

1. Handle egress rate limiting
1.1. Add frames per second as a rate metric usable throughout
tc and associated kernel interfaces.
1.2. Handle any changes required to make a command like this
work:

tc qdisc add dev enp4s0 handle ffff: ingress
tc filter add dev enp4s0 parent ffff: protocol ip prio 1 u32 match ether src 0xfffffffffffffffff police rate 50kfrm burst 10kfrm

This should mostly already work as a valid command, maybe some
changes for handling the new frame based rates.

1.3. Add tc bindings in dsa driver that hook in to the parent
netdev's tc bindings (similar to how the ethtool bindings hook in to the parent's ethtool bindings) to setup the HW egress rate limiting as done in the existing patch.

2. Add ability to set output queue scheduling algorithm Currently no netdev is created for the CPU port, so it can't be targeted by tc or any of the other userland utilities. We need to be able to set settings for that port. Currently I can think of 2 options:

2.1. Add a netdev to represent the CPU port. This will likely
face objections from some people upstream, though it has already been suggested as a way to handle this by others upstream.
This would likely require a lot of effort and learning to figure out how to do this in a way that doesn't start to break key assumptions with the rest of dsa (like the CPU port not having its own IP address). If this were achieved, we could then do one of the following:
2.1.1. Use mqprio disc to model the 4 output queues with a new
parameter to select the scheduling mode.
2.1.2. Add an ethtool priv settings capability (similar to priv
flags) that configures the scheduling mode. This would be my preferred method as it allows port prioritization irrespective of linux's qdisc priorities, which seems to model the HW better.

2.2. Add tc bindings similar to 1.3 above, which allow us to
define a new ingress qdisc parameter for scheduling mode, which the CPU port code can see due to hooking in to the parent device's tc bindings. This might be the simplest approach to implement, but feels a bit hinky w.r.t the semantics of ingress qdisc as we are actually specifying the scheduling of output queues for its link partner.

3. Add ability to set default queue priority of incoming packets on a port.
This could be done as a new parameter for tc's ingress qdisc. I would suggest that it should specify an 802.1p priority number (e.g. "hwprio 3" would specify all traffic ingressing should be considered critical) as this neatly lines up with the 8 priority levels used in mqprio and could be extended further to allow/disallow priority setting per frame from 802.1q tags (e.g. "hwprio 3 notag" or something similar). This might balloon in required effort, particularly if we have to handle the none HW offload path, requiring us to figure out and implement a priority tagging within the kernel's buffers. This would likely be able to reuse a lot of the infrastructure in place for 802.1q tagging that currently exists within the kernel, though Ive not looked in to those code paths to estimate difficulty.

#38 lore raw
In reply to Robert Beckett

Hi Robert,

Quoted 41 lines · Robert Beckett

This patch-set adds support for some features of the Marvell switch chips that can be used to handle packet storms.

The rationale for this was a setup that requires the ability to receive traffic from one port, while a packet storm is occuring on another port (via an external switch with a deliberate loop). This is needed to ensure vital data delivery from a specific port, while mitigating any loops or DoS that a user may introduce on another port (can't guarantee sensible users).

[patch 1/7] configures auto negotiation for CPU ports connected with phys to enable pause frame propogation.

[patch 2/7] allows setting of port's default output queue priority for any ingressing packets on that port.

[patch 3/7] dt-bindings for patch 2.

[patch 4/7] allows setting of a port's queue scheduling so that it can prioritise egress of traffic routed from high priority ports.

[patch 5/7] dt-bindings for patch 4.

[patch 6/7] allows ports to rate limit their egress. This can be used to stop the host CPU from becoming swamped by packet delivery and exhasting descriptors.

[patch 7/7] dt-bindings for patch 6.

Robert Beckett (7):
net/dsa: configure autoneg for CPU port
net: dsa: mv88e6xxx: add ability to set default queue priorities per
port
dt-bindings: mv88e6xxx: add ability to set default queue priorities
per port
net: dsa: mv88e6xxx: add ability to set queue scheduling
dt-bindings: mv88e6xxx: add ability to set queue scheduling
net: dsa: mv88e6xxx: add egress rate limiting
dt-bindings: mv88e6xxx: add egress rate limiting

Quoted 9 lines
 .../devicetree/bindings/net/dsa/marvell.txt   |  38 +++++
 drivers/net/dsa/mv88e6xxx/chip.c              | 122 ++++++++++++---
 drivers/net/dsa/mv88e6xxx/chip.h              |   5 +-
 drivers/net/dsa/mv88e6xxx/port.c              | 140 +++++++++++++++++-
 drivers/net/dsa/mv88e6xxx/port.h              |  24 ++-
 include/dt-bindings/net/dsa-mv88e6xxx.h       |  22 +++
 net/dsa/port.c                                |  10 ++
 7 files changed, 327 insertions(+), 34 deletions(-)
 create mode 100644 include/dt-bindings/net/dsa-mv88e6xxx.h

Feature series targeting netdev must be prefixed "PATCH net-next". As this approach was a PoC, sending it as "RFC net-next" would be even more appropriate.

Thank you,

Vivien

#39 lore raw
Quoted 68 lines · Vivien Didelot

Hi Robert,

Quoted 52 lines · Robert Beckett

This patch-set adds support for some features of the Marvell switch chips that can be used to handle packet storms.

The rationale for this was a setup that requires the ability to receive
traffic from one port, while a packet storm is occuring on another port
(via an external switch with a deliberate loop). This is needed to ensure vital data delivery from a specific port, while mitigating any
loops or DoS that a user may introduce on another port (can't guarantee
sensible users).

[patch 1/7] configures auto negotiation for CPU ports connected with
phys to enable pause frame propogation.

[patch 2/7] allows setting of port's default output queue priority for
any ingressing packets on that port.

[patch 3/7] dt-bindings for patch 2.

[patch 4/7] allows setting of a port's queue scheduling so that it can
prioritise egress of traffic routed from high priority ports.

[patch 5/7] dt-bindings for patch 4.

[patch 6/7] allows ports to rate limit their egress. This can be used to
stop the host CPU from becoming swamped by packet delivery and exhasting
descriptors.

[patch 7/7] dt-bindings for patch 6.

Robert Beckett (7):
net/dsa: configure autoneg for CPU port
net: dsa: mv88e6xxx: add ability to set default queue priorities
per
port
dt-bindings: mv88e6xxx: add ability to set default queue
priorities
per port
net: dsa: mv88e6xxx: add ability to set queue scheduling
dt-bindings: mv88e6xxx: add ability to set queue scheduling
net: dsa: mv88e6xxx: add egress rate limiting
dt-bindings: mv88e6xxx: add egress rate limiting

Quoted 11 lines
 .../devicetree/bindings/net/dsa/marvell.txt   |  38 +++++
 drivers/net/dsa/mv88e6xxx/chip.c              | 122 ++++++++++++
---
 drivers/net/dsa/mv88e6xxx/chip.h              |   5 +-
 drivers/net/dsa/mv88e6xxx/port.c              | 140
+++++++++++++++++-
 drivers/net/dsa/mv88e6xxx/port.h              |  24 ++-
 include/dt-bindings/net/dsa-mv88e6xxx.h       |  22 +++
 net/dsa/port.c                                |  10 ++
 7 files changed, 327 insertions(+), 34 deletions(-)
 create mode 100644 include/dt-bindings/net/dsa-mv88e6xxx.h
Quoted 1 line

Feature series targeting netdev must be prefixed "PATCH net-next". As

Thanks for the info. Out of curiosity, where should I have gleaned this info from? This is my first contribution to netdev, so I wasnt familiar with the etiquette.

Quoted 8 lines

this approach was a PoC, sending it as "RFC net-next" would be even more
appropriate.

Thank you,

Vivien

#40 lore raw

Hi Robert,

Quoted 9 lines · Robert Beckett
Quoted 1 line

Feature series targeting netdev must be prefixed "PATCH net-next". As

Thanks for the info. Out of curiosity, where should I have gleaned this info from? This is my first contribution to netdev, so I wasnt familiar with the etiquette.

Quoted 3 lines

this approach was a PoC, sending it as "RFC net-next" would be even more
appropriate.

Netdev being a huge subsystem has specific rules for subject prefix or merge window, which are described in Documentation/networking/netdev-FAQ.rst

Thank you,

Vivien

maintainer
#41 lore raw
In reply to Robert Beckett
Quoted 5 lines
Quoted 1 line

Feature series targeting netdev must be prefixed "PATCH net-next". As

Thanks for the info. Out of curiosity, where should I have gleaned this info from? This is my first contribution to netdev, so I wasnt familiar with the etiquette.

It is also a good idea to 'lurk' in a mailing list for a while, reading emails flying around, getting to know how things work. This subject of "PATCH net-next" comes up maybe once a week. The idea off offloads gets discussed once every couple of weeks etc.

Andrew