Skip to contentSkip to conversation

Add new PCP and APPTRUST attributes to dcbnl

PATCH Superseded Daniel Machon 2 participants 15 messages Oct 24, 2022 – Oct 27, 2022

Subsystems (3)

v3 of 6 Superseded Latest is v6 →

lore
Share & tools
lore

Daniel Machon

author maintainer
#0 lore raw

v3 cover Add new PCP and APPTRUST attributes to dcbnl

This patch series adds new extension attributes to dcbnl, to support PCP prioritization (and thereby hw offloadable pcp-based queue classification) and per-selector trust and trust order. Additionally, the microchip sparx5 driver has been dcb-enabled to make use of the new attributes to offload PCP, DSCP and Default prio to the switch, and implement trust order of selectors.

For pre-RFC discussion see:
https://lore.kernel.org/netdev/Yv9VO1DYAxNduw6A@DEN-LT-70577/

For RFC series see:
https://lore.kernel.org/netdev/[email protected]/

In summary: there currently exist no convenient way to offload per-port PCP-based queue classification to hardware. The DCB subsystem offers different ways to prioritize through its APP table, but lacks an option for PCP. Similarly, there is no way to indicate the notion of trust for APP table selectors. This patch series addresses both topics.

PCP based queue classification:
- 8021Q standardizes the Priority Code Point table (see 6.9.3 of IEEE Std 802.1Q-2018). This patch series makes it possible, to offload the PCP classification to said table. The new PCP selector is not a standard part of the APP managed object, therefore it is encapsulated in a new non-std extension attribute.

Selector trust:
- ASIC's often has the notion of trust DSCP and trust PCP. The new attribute makes it possible to specify a trust order of app selectors, which drivers can then react on.

DCB-enable sparx5 driver:
- Now supports offloading of DSCP, PCP and default priority. Only one mapping of protocol:priority is allowed. Consecutive mappings of the same protocol to some new priority, will overwrite the previous. This is to keep a consistent view of the app table and the hardware.
- Now supports dscp and pcp trust, by use of the introduced dcbnl_set/getapptrust ops. Sparx5 supports trust orders: [], [dscp], [pcp] and [dscp, pcp]. For now, only DSCP and PCP selectors are supported by the driver, everything else is bounced.

Patch #1 introduces a new PCP selector to the APP object, which makes it possible to encode PCP and DEI in the app triplet and offload it to the PCP table of the ASIC.

Patch #2 Introduces the new extension attributes DCB_ATTR_DCB_APP_TRUST_TABLE and DCB_ATTR_DCB_APP_TRUST. Trusted selectors are passed in the nested DCB_ATTR_DCB_APP_TRUST_TABLE attribute, and assembled into an array of selectors:

  u8 selectors[256];

where lower indexes has higher precedence. In the array, selectors are stored consecutively, starting from index zero. With a maximum number of 256 unique selectors, the list has the same maximum size.

Patch #3 Sets up the dcbnl ops hook, and adds support for offloading pcp app entries, to the PCP table of the switch.

Patch #4 Makes use of the dcbnl_set/getapptrust ops, to set a per-port trust order.

Patch #5 Adds support for offloading dscp app entries to the DSCP table of the switch.

Patch #6 Adds support for offloading default prio app entries to the switch.

================================================================================

RFC v1:
https://lore.kernel.org/netdev/[email protected]/

RFC v1 -> RFC v2:
- Added new nested attribute type DCB_ATTR_DCB_APP_TRUST_TABLE.
- Renamed attributes from DCB_ATTR_IEEE_* to DCB_ATTR_DCB_*.
- Renamed ieee_set/getapptrust to dcbnl_set/getapptrust.
- Added -EOPNOTSUPP if dcbnl_setapptrust is not set.
- Added sanitization of selector array, before passing to driver.

RFC v2 -> (non-RFC) v1
- Added additional check for selector validity.
- Fixed a few style errors.
- using nla_start_nest() instead of nla_start_nest_no_flag().
- Moved DCB_ATTR_DCB_APP_TRUST into new enum.
- Added new DCB_ATTR_DCB_APP extension attribute, for non-std selector values.
- Added support for offloading dscp, pcp and default prio in the sparx5 driver.
- Added support for per-selector trust and trust order in the sparx5 driver.

v1 -> v2
- Fixed compiler and kdoc warning

v2 -> v3
- Moved back to 255 as PCP selector value.
- Fixed return value in dcbnl_app_attr_type_get() to enum.
- Modified in dcbnl_app_attr_type_get() dcbnl_app_attr_type_validate() to return directly.
- Added nselector check in sparx5_dcb_apptrust_validate().
- Added const qualifier to "names" variable in struct sparx5_dcb_apptrust.
- Added new SPARX5_DCB config. Fixes issues reported by kernel test robot.

Daniel Machon (6):
net: dcb: add new pcp selector to app object
net: dcb: add new apptrust attribute
net: microchip: sparx5: add support for offloading pcp table
net: microchip: sparx5: add support for apptrust
net: microchip: sparx5: add support for offloading dscp table
net: microchip: sparx5: add support for offloading default prio

 drivers/net/ethernet/microchip/sparx5/Kconfig |   8 +
 .../net/ethernet/microchip/sparx5/Makefile    |   2 +
 .../ethernet/microchip/sparx5/sparx5_dcb.c    | 293 ++++++++++++++++++
 .../ethernet/microchip/sparx5/sparx5_main.h   |  11 +
 .../microchip/sparx5/sparx5_main_regs.h       | 127 +++++++-
 .../ethernet/microchip/sparx5/sparx5_port.c   |  99 ++++++
 .../ethernet/microchip/sparx5/sparx5_port.h   |  37 +++
 .../ethernet/microchip/sparx5/sparx5_qos.c    |   4 +
 include/net/dcbnl.h                           |   4 +
 include/uapi/linux/dcbnl.h                    |  15 +
 net/dcb/dcbnl.c                               | 113 ++++++-
 11 files changed, 703 insertions(+), 10 deletions(-)
 create mode 100644 drivers/net/ethernet/microchip/sparx5/sparx5_dcb.c

--
2.34.1

author maintainer
#1 lore raw

PATCH v3 1/6 net: dcb: add new pcp selector to app object

Add new PCP selector for the 8021Qaz APP managed object.

As the PCP selector is not part of the 8021Qaz standard, a new non-std extension attribute DCB_ATTR_DCB_APP has been introduced. Also two helper functions to translate between selector and app attribute type has been added. The new selector has been given a value of 255, to minimize the risk of future overlap of std- and non-std attributes.

The new DCB_ATTR_DCB_APP is sent alongside the ieee std attribute in the app table. This means that the dcb_app struct can now both contain std- and non-std app attributes. Currently there is no overlap between the selector values of the two attributes.

The purpose of adding the PCP selector, is to be able to offload PCP-based queue classification to the 8021Q Priority Code Point table, see 6.9.3 of IEEE Std 802.1Q-2018.

PCP and DEI is encoded in the protocol field as 8*dei+pcp, so that a mapping of PCP 2 and DEI 1 to priority 3 is encoded as {255, 10, 3}.

---
 include/uapi/linux/dcbnl.h |  6 ++++++
 net/dcb/dcbnl.c            | 36 ++++++++++++++++++++++++++++++++----
 2 files changed, 38 insertions(+), 4 deletions(-)
include/uapi/linux/dcbnl.h +6 -0
index a791a94013a6..dc7ef96207ca 100644
@@ -218,6 +218,9 @@ struct cee_pfc {
 #define IEEE_8021QAZ_APP_SEL_ANY	4
 #define IEEE_8021QAZ_APP_SEL_DSCP       5
 
+/* Non-std selector values */
+#define DCB_APP_SEL_PCP 255
+
 /* This structure contains the IEEE 802.1Qaz APP managed object. This
  * object is also used for the CEE std as well.
  *
@@ -247,6 +250,8 @@ struct dcb_app {
 	__u16	protocol;
 };
 
+#define IEEE_8021QAZ_APP_SEL_MAX 255
+
 /**
  * struct dcb_peer_app_info - APP feature information sent by the peer
  *
@@ -425,6 +430,7 @@ enum ieee_attrs {
 enum ieee_attrs_app {
 	DCB_ATTR_IEEE_APP_UNSPEC,
 	DCB_ATTR_IEEE_APP,
+	DCB_ATTR_DCB_APP,
 	__DCB_ATTR_IEEE_APP_MAX
 };
 #define DCB_ATTR_IEEE_APP_MAX (__DCB_ATTR_IEEE_APP_MAX - 1)
net/dcb/dcbnl.c +32 -4
index dc4fb699b56c..92c32bc11374 100644
@@ -179,6 +179,33 @@ static const struct nla_policy dcbnl_featcfg_nest[DCB_FEATCFG_ATTR_MAX + 1] = {
 static LIST_HEAD(dcb_app_list);
 static DEFINE_SPINLOCK(dcb_lock);
 
+static enum ieee_attrs_app dcbnl_app_attr_type_get(u8 selector)
+{
+	switch (selector) {
+	case IEEE_8021QAZ_APP_SEL_ETHERTYPE:
+	case IEEE_8021QAZ_APP_SEL_STREAM:
+	case IEEE_8021QAZ_APP_SEL_DGRAM:
+	case IEEE_8021QAZ_APP_SEL_ANY:
+	case IEEE_8021QAZ_APP_SEL_DSCP:
+		return DCB_ATTR_IEEE_APP;
+	case DCB_APP_SEL_PCP:
+		return DCB_ATTR_DCB_APP;
+	default:
+		return DCB_ATTR_IEEE_APP_UNSPEC;
+	}
+}
+
+static bool dcbnl_app_attr_type_validate(enum ieee_attrs_app type)
+{
+	switch (type) {
+	case DCB_ATTR_IEEE_APP:
+	case DCB_ATTR_DCB_APP:
+		return true;
+	default:
+		return false;
+	}
+}
+
 static struct sk_buff *dcbnl_newmsg(int type, u8 cmd, u32 port, u32 seq,
 				    u32 flags, struct nlmsghdr **nlhp)
 {
@@ -1116,8 +1143,9 @@ static int dcbnl_ieee_fill(struct sk_buff *skb, struct net_device *netdev)
 	spin_lock_bh(&dcb_lock);
 	list_for_each_entry(itr, &dcb_app_list, list) {
 		if (itr->ifindex == netdev->ifindex) {
-			err = nla_put(skb, DCB_ATTR_IEEE_APP, sizeof(itr->app),
-					 &itr->app);
+			enum ieee_attrs_app type =
+				dcbnl_app_attr_type_get(itr->app.selector);
+			err = nla_put(skb, type, sizeof(itr->app), &itr->app);
 			if (err) {
 				spin_unlock_bh(&dcb_lock);
 				return -EMSGSIZE;
@@ -1495,7 +1523,7 @@ static int dcbnl_ieee_set(struct net_device *netdev, struct nlmsghdr *nlh,
 		nla_for_each_nested(attr, ieee[DCB_ATTR_IEEE_APP_TABLE], rem) {
 			struct dcb_app *app_data;
 
-			if (nla_type(attr) != DCB_ATTR_IEEE_APP)
+			if (!dcbnl_app_attr_type_validate(nla_type(attr)))
 				continue;
 
 			if (nla_len(attr) < sizeof(struct dcb_app)) {
@@ -1556,7 +1584,7 @@ static int dcbnl_ieee_del(struct net_device *netdev, struct nlmsghdr *nlh,
 		nla_for_each_nested(attr, ieee[DCB_ATTR_IEEE_APP_TABLE], rem) {
 			struct dcb_app *app_data;
 
-			if (nla_type(attr) != DCB_ATTR_IEEE_APP)
+			if (!dcbnl_app_attr_type_validate(nla_type(attr)))
 				continue;
 			app_data = nla_data(attr);
 			if (ops->ieee_delapp)
Signature · 1 line
2.34.1
author maintainer
#7 lore raw

PATCH v3 2/6 net: dcb: add new apptrust attribute

In reply to Daniel Machon

Add new apptrust extension attributes to the 8021Qaz APP managed object.

Two new attributes, DCB_ATTR_DCB_APP_TRUST_TABLE and DCB_ATTR_DCB_APP_TRUST, has been added. Trusted selectors are passed in the nested attribute DCB_ATTR_DCB_APP_TRUST, in order of precedence.

The new attributes are meant to allow drivers, whose hw supports the notion of trust, to be able to set whether a particular app selector is trusted - and in which order.

---
 include/net/dcbnl.h        |  4 ++
 include/uapi/linux/dcbnl.h | 10 +++++
 net/dcb/dcbnl.c            | 77 ++++++++++++++++++++++++++++++++++++--
 3 files changed, 87 insertions(+), 4 deletions(-)
include/net/dcbnl.h +4 -0
index 2b2d86fb3131..8841ab6c2de7 100644
@@ -109,6 +109,10 @@ struct dcbnl_rtnl_ops {
 	/* buffer settings */
 	int (*dcbnl_getbuffer)(struct net_device *, struct dcbnl_buffer *);
 	int (*dcbnl_setbuffer)(struct net_device *, struct dcbnl_buffer *);
+
+	/* apptrust */
+	int (*dcbnl_setapptrust)(struct net_device *, u8 *, int);
+	int (*dcbnl_getapptrust)(struct net_device *, u8 *, int *);
 };
 
 #endif /* __NET_DCBNL_H__ */
include/uapi/linux/dcbnl.h +10 -0
index dc7ef96207ca..9344e3ba5768 100644
@@ -410,6 +410,7 @@ enum dcbnl_attrs {
  * @DCB_ATTR_IEEE_PEER_ETS: peer ETS configuration - get only
  * @DCB_ATTR_IEEE_PEER_PFC: peer PFC configuration - get only
  * @DCB_ATTR_IEEE_PEER_APP: peer APP tlv - get only
+ * @DCB_ATTR_DCB_APP_TRUST_TABLE: selector trust table
  */
 enum ieee_attrs {
 	DCB_ATTR_IEEE_UNSPEC,
@@ -423,6 +424,7 @@ enum ieee_attrs {
 	DCB_ATTR_IEEE_QCN,
 	DCB_ATTR_IEEE_QCN_STATS,
 	DCB_ATTR_DCB_BUFFER,
+	DCB_ATTR_DCB_APP_TRUST_TABLE,
 	__DCB_ATTR_IEEE_MAX
 };
 #define DCB_ATTR_IEEE_MAX (__DCB_ATTR_IEEE_MAX - 1)
@@ -435,6 +437,14 @@ enum ieee_attrs_app {
 };
 #define DCB_ATTR_IEEE_APP_MAX (__DCB_ATTR_IEEE_APP_MAX - 1)
 
+enum dcbnl_attrs_apptrust {
+	DCB_ATTR_DCB_APP_TRUST_UNSPEC,
+	DCB_ATTR_DCB_APP_TRUST,
+	__DCB_ATTR_DCB_APP_TRUST_MAX
+};
+
+#define DCB_ATTR_DCB_APP_TRUST_MAX (__DCB_ATTR_DCB_APP_TRUST_MAX - 1)
+
 /**
  * enum cee_attrs - CEE DCBX get attributes.
  *
net/dcb/dcbnl.c +73 -4

110 lines not on this page · raw message

Signature · 1 line
2.34.1
author maintainer
#11 lore raw

PATCH v3 3/6 net: microchip: sparx5: add support for offloading pcp table

In reply to Daniel Machon

Add new registers and functions to support offload of pcp app entries.

---
 drivers/net/ethernet/microchip/sparx5/Kconfig |  10 ++
 .../net/ethernet/microchip/sparx5/Makefile    |   2 +
 .../ethernet/microchip/sparx5/sparx5_dcb.c    | 116 ++++++++++++++++
 .../ethernet/microchip/sparx5/sparx5_main.h   |  11 ++
 .../microchip/sparx5/sparx5_main_regs.h       | 127 +++++++++++++++++-
 .../ethernet/microchip/sparx5/sparx5_port.c   |  37 +++++
 .../ethernet/microchip/sparx5/sparx5_port.h   |  17 +++
 .../ethernet/microchip/sparx5/sparx5_qos.c    |   4 +
 8 files changed, 322 insertions(+), 2 deletions(-)
 create mode 100644 drivers/net/ethernet/microchip/sparx5/sparx5_dcb.c
drivers/net/ethernet/microchip/sparx5/Kconfig +10 -0
index cc5e48e1bb4c..19f2443e4407 100644
@@ -11,3 +11,13 @@ config SPARX5_SWITCH
 	select RESET_CONTROLLER
 	help
 	  This driver supports the Sparx5 network switch device.
+
+config SPARX5_DCB
+	bool "Data Center Bridging (DCB) support"
+	depends on SPARX5_SWITCH && DCB
+	default y
+	help
+	  Say Y here if you want to use Data Center Bridging (DCB) in the
+	  driver.
+
+	  If unsure, set to Y.
drivers/net/ethernet/microchip/sparx5/Makefile +2 -0
index d1c6ad966747..fbdea8bebd5d 100644
@@ -9,3 +9,5 @@ sparx5-switch-objs  := sparx5_main.o sparx5_packet.o \
  sparx5_netdev.o sparx5_phylink.o sparx5_port.o sparx5_mactable.o sparx5_vlan.o \
  sparx5_switchdev.o sparx5_calendar.o sparx5_ethtool.o sparx5_fdma.o \
  sparx5_ptp.o sparx5_pgid.o sparx5_tc.o sparx5_qos.o
+
+ sparx5-switch-$(CONFIG_SPARX5_DCB) += sparx5_dcb.o
\ No newline at end of file
drivers/net/ethernet/microchip/sparx5/sparx5_dcb.c new file +116 -0

116 lines not on this page · raw message

drivers/net/ethernet/microchip/sparx5/sparx5_main.h +11 -0
index 7a83222caa73..b42c625795ae 100644
@@ -357,6 +357,16 @@ int sparx5_config_dsm_calendar(struct sparx5 *sparx5);
 void sparx5_get_stats64(struct net_device *ndev, struct rtnl_link_stats64 *stats);
 int sparx_stats_init(struct sparx5 *sparx5);
 
+/* sparx5_dcb.c */
+#ifdef CONFIG_SPARX5_DCB
+int sparx5_dcb_init(struct sparx5 *sparx5);
+#else
+static inline int sparx5_dcb_init(struct sparx5 *sparx5)
+{
+	return 0;
+}
+#endif
+
 /* sparx5_netdev.c */
 void sparx5_set_port_ifh_timestamp(void *ifh_hdr, u64 timestamp);
 void sparx5_set_port_ifh_rew_op(void *ifh_hdr, u32 rew_op);
@@ -418,6 +428,7 @@ static inline bool sparx5_is_baser(phy_interface_t interface)
 extern const struct phylink_mac_ops sparx5_phylink_mac_ops;
 extern const struct phylink_pcs_ops sparx5_phylink_pcs_ops;
 extern const struct ethtool_ops sparx5_ethtool_ops;
+extern const struct dcbnl_rtnl_ops sparx5_dcbnl_ops;
 
 /* Calculate raw offset */
 static inline __pure int spx5_offset(int id, int tinst, int tcnt,
drivers/net/ethernet/microchip/sparx5/sparx5_main_regs.h +125 -2

145 lines not on this page · raw message

drivers/net/ethernet/microchip/sparx5/sparx5_port.c +37 -0
index 32709d21ab2f..9ffaaf34d196 100644
@@ -6,6 +6,7 @@
 
 #include <linux/module.h>
 #include <linux/phy/phy.h>
+#include <net/dcbnl.h>
 
 #include "sparx5_main_regs.h"
 #include "sparx5_main.h"
@@ -1144,3 +1145,39 @@ void sparx5_port_enable(struct sparx5_port *port, bool enable)
 		 sparx5,
 		 QFWD_SWITCH_PORT_MODE(port->portno));
 }
+
+int sparx5_port_qos_set(struct sparx5_port *port,
+			struct sparx5_port_qos *qos)
+{
+	sparx5_port_qos_pcp_set(port, &qos->pcp);
+
+	return 0;
+}
+
+int sparx5_port_qos_pcp_set(const struct sparx5_port *port,
+			    struct sparx5_port_qos_pcp *qos)
+{
+	struct sparx5 *sparx5 = port->sparx5;
+	u8 *pcp_itr = qos->map.map;
+	u8 pcp, dp;
+	int i;
+
+	/* Enable/disable pcp and dp for qos classification. */
+	spx5_rmw(ANA_CL_QOS_CFG_PCP_DEI_QOS_ENA_SET(1) |
+		 ANA_CL_QOS_CFG_PCP_DEI_DP_ENA_SET(1),
+		 ANA_CL_QOS_CFG_PCP_DEI_QOS_ENA | ANA_CL_QOS_CFG_PCP_DEI_DP_ENA,
+		 sparx5, ANA_CL_QOS_CFG(port->portno));
+
+	/* Map each pcp and dei value to priority and dp */
+	for (i = 0; i < ARRAY_SIZE(qos->map.map); i++) {
+		pcp = *(pcp_itr + i);
+		dp = (i <= 7) ? 0 : 1;
+		spx5_rmw(ANA_CL_PCP_DEI_MAP_CFG_PCP_DEI_QOS_VAL_SET(pcp) |
+			 ANA_CL_PCP_DEI_MAP_CFG_PCP_DEI_DP_VAL_SET(dp),
+			 ANA_CL_PCP_DEI_MAP_CFG_PCP_DEI_QOS_VAL |
+			 ANA_CL_PCP_DEI_MAP_CFG_PCP_DEI_DP_VAL, sparx5,
+			 ANA_CL_PCP_DEI_MAP_CFG(port->portno, i));
+	}
+
+	return 0;
+}
drivers/net/ethernet/microchip/sparx5/sparx5_port.h +17 -0
index 2f8043eac71b..9c5fb6b651db 100644
@@ -91,4 +91,21 @@ int sparx5_get_port_status(struct sparx5 *sparx5,
 void sparx5_port_enable(struct sparx5_port *port, bool enable);
 int sparx5_port_fwd_urg(struct sparx5 *sparx5, u32 speed);
 
+struct sparx5_port_qos_pcp_map {
+	u8 map[16];
+};
+
+struct sparx5_port_qos_pcp {
+	struct sparx5_port_qos_pcp_map map;
+};
+
+struct sparx5_port_qos {
+	struct sparx5_port_qos_pcp pcp;
+};
+
+int sparx5_port_qos_set(struct sparx5_port *port, struct sparx5_port_qos *qos);
+
+int sparx5_port_qos_pcp_set(const struct sparx5_port *port,
+			    struct sparx5_port_qos_pcp *qos);
+
 #endif	/* __SPARX5_PORT_H__ */
drivers/net/ethernet/microchip/sparx5/sparx5_qos.c +4 -0
index 1e79d0ef0cb8..379e540e5e6a 100644
@@ -389,6 +389,10 @@ int sparx5_qos_init(struct sparx5 *sparx5)
 	if (ret < 0)
 		return ret;
 
+	ret = sparx5_dcb_init(sparx5);
+	if (ret < 0)
+		return ret;
+
 	return 0;
 }
 
Signature · 1 line
2.34.1
author maintainer
#12 lore raw

PATCH v3 4/6 net: microchip: sparx5: add support for apptrust

In reply to Daniel Machon

Make use of set/getapptrust() to implement per-selector trust and trust order.

---
 .../ethernet/microchip/sparx5/sparx5_dcb.c    | 105 ++++++++++++++++++
 .../ethernet/microchip/sparx5/sparx5_port.c   |   4 +-
 .../ethernet/microchip/sparx5/sparx5_port.h   |   2 +
 3 files changed, 109 insertions(+), 2 deletions(-)
drivers/net/ethernet/microchip/sparx5/sparx5_dcb.c +105 -0
index 2a6e875a5860..1fa150d46977 100644
@@ -8,6 +8,22 @@
 
 #include "sparx5_port.h"
 
+static const struct sparx5_dcb_apptrust {
+	u8 selectors[256];
+	int nselectors;
+	const char *names;
+} *apptrust[SPX5_PORTS];
+
+/* Sparx5 supported apptrust configurations */
+static const struct sparx5_dcb_apptrust apptrust_conf[4] = {
+	/* Empty *must* be first */
+	{ { 0                         }, 0, "empty"    },
+	{ { IEEE_8021QAZ_APP_SEL_DSCP }, 1, "dscp"     },
+	{ { DCB_APP_SEL_PCP           }, 1, "pcp"      },
+	{ { IEEE_8021QAZ_APP_SEL_DSCP,
+	    DCB_APP_SEL_PCP           }, 2, "dscp pcp" },
+};
+
 /* Validate app entry.
  *
  * Check for valid selectors and valid protocol and priority ranges.
@@ -37,12 +53,62 @@ static int sparx5_dcb_app_validate(struct net_device *dev,
 	return err;
 }
 
+/* Validate apptrust configuration.
+ *
+ * Return index of supported apptrust configuration if valid, otherwise return
+ * error.
+ */
+static int sparx5_dcb_apptrust_validate(struct net_device *dev, u8 *selectors,
+					int nselectors, int *err)
+{
+	bool match;
+	int i, ii;
+
+	for (i = 0; i < ARRAY_SIZE(apptrust_conf); i++) {
+		if (apptrust_conf[i].nselectors != nselectors)
+			continue;
+		match = true;
+		for (ii = 0; ii < nselectors; ii++) {
+			if (apptrust_conf[i].selectors[ii] !=
+			    *(selectors + ii)) {
+				match = false;
+				break;
+			}
+		}
+		if (match)
+			break;
+	}
+
+	/* Requested trust configuration is not supported */
+	if (!match) {
+		netdev_err(dev, "Valid apptrust configurations are:\n");
+		for (i = 0; i < ARRAY_SIZE(apptrust_conf); i++)
+			pr_info("order: %s\n", apptrust_conf[i].names);
+		*err = -EOPNOTSUPP;
+	}
+
+	return i;
+}
+
+static bool sparx5_dcb_apptrust_contains(int portno, u8 selector)
+{
+	const struct sparx5_dcb_apptrust *conf = apptrust[portno];
+	int i;
+
+	for (i = 0; i < conf->nselectors; i++)
+		if (conf->selectors[i] == selector)
+			return true;
+
+	return false;
+}
+
 static int sparx5_dcb_app_update(struct net_device *dev)
 {
 	struct dcb_app app_itr = { .selector = DCB_APP_SEL_PCP };
 	struct sparx5_port *port = netdev_priv(dev);
 	struct sparx5_port_qos_pcp_map *pcp_map;
 	struct sparx5_port_qos qos = {0};
+	int portno = port->portno;
 	int i;
 
 	pcp_map = &qos.pcp.map;
@@ -53,6 +119,12 @@ static int sparx5_dcb_app_update(struct net_device *dev)
 		pcp_map->map[i] = dcb_getapp(dev, &app_itr);
 	}
 
+	/* Enable use of pcp for queue classification ? */
+	if (sparx5_dcb_apptrust_contains(portno, DCB_APP_SEL_PCP)) {
+		qos.pcp.qos_enable = true;
+		qos.pcp.dp_enable = qos.pcp.qos_enable;
+	}
+
 	return sparx5_port_qos_set(port, &qos);
 }
 
@@ -95,9 +167,40 @@ static int sparx5_dcb_ieee_delapp(struct net_device *dev, struct dcb_app *app)
 	return sparx5_dcb_app_update(dev);
 }
 
+static int sparx5_dcb_setapptrust(struct net_device *dev, u8 *selectors,
+				  int nselectors)
+{
+	struct sparx5_port *port = netdev_priv(dev);
+	int err = 0, idx;
+
+	idx = sparx5_dcb_apptrust_validate(dev, selectors, nselectors, &err);
+	if (err < 0)
+		return err;
+
+	apptrust[port->portno] = &apptrust_conf[idx];
+
+	return sparx5_dcb_app_update(dev);
+}
+
+static int sparx5_dcb_getapptrust(struct net_device *dev, u8 *selectors,
+				  int *nselectors)
+{
+	struct sparx5_port *port = netdev_priv(dev);
+	const struct sparx5_dcb_apptrust *trust;
+
+	trust = apptrust[port->portno];
+
+	memcpy(selectors, trust->selectors, trust->nselectors);
+	*nselectors = trust->nselectors;
+
+	return 0;
+}
+
 const struct dcbnl_rtnl_ops sparx5_dcbnl_ops = {
 	.ieee_setapp = sparx5_dcb_ieee_setapp,
 	.ieee_delapp = sparx5_dcb_ieee_delapp,
+	.dcbnl_setapptrust = sparx5_dcb_setapptrust,
+	.dcbnl_getapptrust = sparx5_dcb_getapptrust,
 };
 
 int sparx5_dcb_init(struct sparx5 *sparx5)
@@ -110,6 +213,8 @@ int sparx5_dcb_init(struct sparx5 *sparx5)
 		if (!port)
 			continue;
 		port->ndev->dcbnl_ops = &sparx5_dcbnl_ops;
+		/* Initialize [dscp, pcp] default trust */
+		apptrust[port->portno] = &apptrust_conf[3];
 	}
 
 	return 0;
drivers/net/ethernet/microchip/sparx5/sparx5_port.c +2 -2
index 9ffaaf34d196..99e86e87aa16 100644
@@ -1163,8 +1163,8 @@ int sparx5_port_qos_pcp_set(const struct sparx5_port *port,
 	int i;
 
 	/* Enable/disable pcp and dp for qos classification. */
-	spx5_rmw(ANA_CL_QOS_CFG_PCP_DEI_QOS_ENA_SET(1) |
-		 ANA_CL_QOS_CFG_PCP_DEI_DP_ENA_SET(1),
+	spx5_rmw(ANA_CL_QOS_CFG_PCP_DEI_QOS_ENA_SET(qos->qos_enable) |
+		 ANA_CL_QOS_CFG_PCP_DEI_DP_ENA_SET(qos->dp_enable),
 		 ANA_CL_QOS_CFG_PCP_DEI_QOS_ENA | ANA_CL_QOS_CFG_PCP_DEI_DP_ENA,
 		 sparx5, ANA_CL_QOS_CFG(port->portno));
 
drivers/net/ethernet/microchip/sparx5/sparx5_port.h +2 -0
index 9c5fb6b651db..fae9f5464548 100644
@@ -97,6 +97,8 @@ struct sparx5_port_qos_pcp_map {
 
 struct sparx5_port_qos_pcp {
 	struct sparx5_port_qos_pcp_map map;
+	bool qos_enable;
+	bool dp_enable;
 };
 
 struct sparx5_port_qos {
Signature · 1 line
2.34.1
author maintainer
#13 lore raw

PATCH v3 5/6 net: microchip: sparx5: add support for offloading dscp table

In reply to Daniel Machon

Add support for offloading dscp app entries. Dscp values are global for all ports on the sparx5 switch. Therefore, we replicate each dscp app entry per-port.

---
 .../ethernet/microchip/sparx5/sparx5_dcb.c    | 66 ++++++++++++++++++-
 .../ethernet/microchip/sparx5/sparx5_port.c   | 39 +++++++++++
 .../ethernet/microchip/sparx5/sparx5_port.h   | 13 ++++
 3 files changed, 115 insertions(+), 3 deletions(-)
drivers/net/ethernet/microchip/sparx5/sparx5_dcb.c +63 -3
index 1fa150d46977..283d5f338e0e 100644
@@ -34,6 +34,13 @@ static int sparx5_dcb_app_validate(struct net_device *dev,
 	int err = 0;
 
 	switch (app->selector) {
+	/* Dscp checks */
+	case IEEE_8021QAZ_APP_SEL_DSCP:
+		if (app->protocol > 63)
+			err = -EINVAL;
+		else if (app->priority >= SPX5_PRIOS)
+			err = -ERANGE;
+		break;
 	/* Pcp checks */
 	case DCB_APP_SEL_PCP:
 		if (app->protocol > 15)
@@ -104,17 +111,27 @@ static bool sparx5_dcb_apptrust_contains(int portno, u8 selector)
 
 static int sparx5_dcb_app_update(struct net_device *dev)
 {
-	struct dcb_app app_itr = { .selector = DCB_APP_SEL_PCP };
 	struct sparx5_port *port = netdev_priv(dev);
+	struct sparx5_port_qos_dscp_map *dscp_map;
 	struct sparx5_port_qos_pcp_map *pcp_map;
 	struct sparx5_port_qos qos = {0};
+	struct dcb_app app_itr = {0};
 	int portno = port->portno;
 	int i;
 
+	dscp_map = &qos.dscp.map;
 	pcp_map = &qos.pcp.map;
 
+	/* Get dscp ingress mapping */
+	for (i = 0; i < ARRAY_SIZE(dscp_map->map); i++) {
+		app_itr.selector = IEEE_8021QAZ_APP_SEL_DSCP;
+		app_itr.protocol = i;
+		dscp_map->map[i] = dcb_getapp(dev, &app_itr);
+	}
+
 	/* Get pcp ingress mapping */
 	for (i = 0; i < ARRAY_SIZE(pcp_map->map); i++) {
+		app_itr.selector = DCB_APP_SEL_PCP;
 		app_itr.protocol = i;
 		pcp_map->map[i] = dcb_getapp(dev, &app_itr);
 	}
@@ -125,9 +142,44 @@ static int sparx5_dcb_app_update(struct net_device *dev)
 		qos.pcp.dp_enable = qos.pcp.qos_enable;
 	}
 
+	/* Enable use of dscp for queue classification ? */
+	if (sparx5_dcb_apptrust_contains(portno, IEEE_8021QAZ_APP_SEL_DSCP)) {
+		qos.dscp.qos_enable = true;
+		qos.dscp.dp_enable = qos.dscp.qos_enable;
+	}
+
 	return sparx5_port_qos_set(port, &qos);
 }
 
+/* Set or delete dscp app entry.
+ *
+ * Dscp mapping is global for all ports, so set and delete app entries are
+ * replicated for each port.
+ */
+static int sparx5_dcb_ieee_dscp_setdel_app(struct net_device *dev,
+					   struct dcb_app *app, bool del)
+{
+	struct sparx5_port *port = netdev_priv(dev);
+	struct dcb_app apps[SPX5_PORTS];
+	struct sparx5_port *port_itr;
+	int err, i;
+
+	for (i = 0; i < SPX5_PORTS; i++) {
+		port_itr = port->sparx5->ports[i];
+		if (!port_itr)
+			continue;
+		memcpy(&apps[i], app, sizeof(struct dcb_app));
+		if (del)
+			err = dcb_ieee_delapp(port_itr->ndev, &apps[i]);
+		else
+			err = dcb_ieee_setapp(port_itr->ndev, &apps[i]);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
 static int sparx5_dcb_ieee_setapp(struct net_device *dev, struct dcb_app *app)
 {
 	struct dcb_app app_itr;
@@ -146,7 +198,11 @@ static int sparx5_dcb_ieee_setapp(struct net_device *dev, struct dcb_app *app)
 		dcb_ieee_delapp(dev, &app_itr);
 	}
 
-	err = dcb_ieee_setapp(dev, app);
+	if (app->selector == IEEE_8021QAZ_APP_SEL_DSCP)
+		err = sparx5_dcb_ieee_dscp_setdel_app(dev, app, false);
+	else
+		err = dcb_ieee_setapp(dev, app);
+
 	if (err)
 		goto out;
 
@@ -160,7 +216,11 @@ static int sparx5_dcb_ieee_delapp(struct net_device *dev, struct dcb_app *app)
 {
 	int err;
 
-	err = dcb_ieee_delapp(dev, app);
+	if (app->selector == IEEE_8021QAZ_APP_SEL_DSCP)
+		err = sparx5_dcb_ieee_dscp_setdel_app(dev, app, true);
+	else
+		err = dcb_ieee_delapp(dev, app);
+
 	if (err < 0)
 		return err;
 
drivers/net/ethernet/microchip/sparx5/sparx5_port.c +39 -0
index 99e86e87aa16..fb5e321c4896 100644
@@ -1149,6 +1149,7 @@ void sparx5_port_enable(struct sparx5_port *port, bool enable)
 int sparx5_port_qos_set(struct sparx5_port *port,
 			struct sparx5_port_qos *qos)
 {
+	sparx5_port_qos_dscp_set(port, &qos->dscp);
 	sparx5_port_qos_pcp_set(port, &qos->pcp);
 
 	return 0;
@@ -1181,3 +1182,41 @@ int sparx5_port_qos_pcp_set(const struct sparx5_port *port,
 
 	return 0;
 }
+
+int sparx5_port_qos_dscp_set(const struct sparx5_port *port,
+			     struct sparx5_port_qos_dscp *qos)
+{
+	struct sparx5 *sparx5 = port->sparx5;
+	u8 *dscp = qos->map.map;
+	int i;
+
+	/* Enable/disable dscp and dp for qos classification.
+	 * Disable rewrite of dscp values for now.
+	 */
+	spx5_rmw(ANA_CL_QOS_CFG_DSCP_QOS_ENA_SET(qos->qos_enable) |
+		 ANA_CL_QOS_CFG_DSCP_DP_ENA_SET(qos->dp_enable) |
+		 ANA_CL_QOS_CFG_DSCP_KEEP_ENA_SET(1),
+		 ANA_CL_QOS_CFG_DSCP_QOS_ENA | ANA_CL_QOS_CFG_DSCP_DP_ENA |
+		 ANA_CL_QOS_CFG_DSCP_KEEP_ENA, sparx5,
+		 ANA_CL_QOS_CFG(port->portno));
+
+	/* Map each dscp value to priority and dp */
+	for (i = 0; i < ARRAY_SIZE(qos->map.map); i++) {
+		spx5_rmw(ANA_CL_DSCP_CFG_DSCP_QOS_VAL_SET(*(dscp + i)) |
+			 ANA_CL_DSCP_CFG_DSCP_DP_VAL_SET(0),
+			 ANA_CL_DSCP_CFG_DSCP_QOS_VAL |
+			 ANA_CL_DSCP_CFG_DSCP_DP_VAL, sparx5,
+			 ANA_CL_DSCP_CFG(i));
+	}
+
+	/* Set per-dscp trust */
+	for (i = 0; i <  ARRAY_SIZE(qos->map.map); i++) {
+		if (qos->qos_enable) {
+			spx5_rmw(ANA_CL_DSCP_CFG_DSCP_TRUST_ENA_SET(1),
+				 ANA_CL_DSCP_CFG_DSCP_TRUST_ENA, sparx5,
+				 ANA_CL_DSCP_CFG(i));
+		}
+	}
+
+	return 0;
+}
drivers/net/ethernet/microchip/sparx5/sparx5_port.h +13 -0
index fae9f5464548..a0cd53fa3ad0 100644
@@ -95,14 +95,25 @@ struct sparx5_port_qos_pcp_map {
 	u8 map[16];
 };
 
+struct sparx5_port_qos_dscp_map {
+	u8 map[64];
+};
+
 struct sparx5_port_qos_pcp {
 	struct sparx5_port_qos_pcp_map map;
 	bool qos_enable;
 	bool dp_enable;
 };
 
+struct sparx5_port_qos_dscp {
+	struct sparx5_port_qos_dscp_map map;
+	bool qos_enable;
+	bool dp_enable;
+};
+
 struct sparx5_port_qos {
 	struct sparx5_port_qos_pcp pcp;
+	struct sparx5_port_qos_dscp dscp;
 };
 
 int sparx5_port_qos_set(struct sparx5_port *port, struct sparx5_port_qos *qos);
@@ -110,4 +121,6 @@ int sparx5_port_qos_set(struct sparx5_port *port, struct sparx5_port_qos *qos);
 int sparx5_port_qos_pcp_set(const struct sparx5_port *port,
 			    struct sparx5_port_qos_pcp *qos);
 
+int sparx5_port_qos_dscp_set(const struct sparx5_port *port,
+			     struct sparx5_port_qos_dscp *qos);
 #endif	/* __SPARX5_PORT_H__ */
Signature · 1 line
2.34.1
author maintainer
#14 lore raw

PATCH v3 6/6 net: microchip: sparx5: add support for offloading default prio

In reply to Daniel Machon

Add support for offloading default prio {ETHERTYPE, 0, prio}.

---
 .../ethernet/microchip/sparx5/sparx5_dcb.c    | 12 ++++++++++
 .../ethernet/microchip/sparx5/sparx5_port.c   | 23 +++++++++++++++++++
 .../ethernet/microchip/sparx5/sparx5_port.h   |  5 ++++
 3 files changed, 40 insertions(+)
drivers/net/ethernet/microchip/sparx5/sparx5_dcb.c +12 -0
index 283d5f338e0e..df2374e64b72 100644
@@ -34,6 +34,13 @@ static int sparx5_dcb_app_validate(struct net_device *dev,
 	int err = 0;
 
 	switch (app->selector) {
+	/* Default priority checks */
+	case IEEE_8021QAZ_APP_SEL_ETHERTYPE:
+		if (app->protocol != 0)
+			err = -EINVAL;
+		else if (app->priority >= SPX5_PRIOS)
+			err = -ERANGE;
+		break;
 	/* Dscp checks */
 	case IEEE_8021QAZ_APP_SEL_DSCP:
 		if (app->protocol > 63)
@@ -122,6 +129,11 @@ static int sparx5_dcb_app_update(struct net_device *dev)
 	dscp_map = &qos.dscp.map;
 	pcp_map = &qos.pcp.map;
 
+	/* Get default prio. */
+	qos.default_prio = dcb_ieee_getapp_default_prio_mask(dev);
+	if (qos.default_prio)
+		qos.default_prio = fls(qos.default_prio) - 1;
+
 	/* Get dscp ingress mapping */
 	for (i = 0; i < ARRAY_SIZE(dscp_map->map); i++) {
 		app_itr.selector = IEEE_8021QAZ_APP_SEL_DSCP;
drivers/net/ethernet/microchip/sparx5/sparx5_port.c +23 -0
index fb5e321c4896..73ebe76d7e50 100644
@@ -1151,6 +1151,7 @@ int sparx5_port_qos_set(struct sparx5_port *port,
 {
 	sparx5_port_qos_dscp_set(port, &qos->dscp);
 	sparx5_port_qos_pcp_set(port, &qos->pcp);
+	sparx5_port_qos_default_set(port, qos);
 
 	return 0;
 }
@@ -1220,3 +1221,25 @@ int sparx5_port_qos_dscp_set(const struct sparx5_port *port,
 
 	return 0;
 }
+
+int sparx5_port_qos_default_set(const struct sparx5_port *port,
+				const struct sparx5_port_qos *qos)
+{
+	struct sparx5 *sparx5 = port->sparx5;
+
+	/* Set default prio and dp level */
+	spx5_rmw(ANA_CL_QOS_CFG_DEFAULT_QOS_VAL_SET(qos->default_prio) |
+		 ANA_CL_QOS_CFG_DEFAULT_DP_VAL_SET(0),
+		 ANA_CL_QOS_CFG_DEFAULT_QOS_VAL |
+		 ANA_CL_QOS_CFG_DEFAULT_DP_VAL,
+		 sparx5, ANA_CL_QOS_CFG(port->portno));
+
+	/* Set default pcp and dei for untagged frames */
+	spx5_rmw(ANA_CL_VLAN_CTRL_PORT_PCP_SET(0) |
+		 ANA_CL_VLAN_CTRL_PORT_DEI_SET(0),
+		 ANA_CL_VLAN_CTRL_PORT_PCP |
+		 ANA_CL_VLAN_CTRL_PORT_DEI,
+		 sparx5, ANA_CL_VLAN_CTRL(port->portno));
+
+	return 0;
+}
drivers/net/ethernet/microchip/sparx5/sparx5_port.h +5 -0
index a0cd53fa3ad0..588eecff59f8 100644
@@ -114,6 +114,7 @@ struct sparx5_port_qos_dscp {
 struct sparx5_port_qos {
 	struct sparx5_port_qos_pcp pcp;
 	struct sparx5_port_qos_dscp dscp;
+	u8 default_prio;
 };
 
 int sparx5_port_qos_set(struct sparx5_port *port, struct sparx5_port_qos *qos);
@@ -123,4 +124,8 @@ int sparx5_port_qos_pcp_set(const struct sparx5_port *port,
 
 int sparx5_port_qos_dscp_set(const struct sparx5_port *port,
 			     struct sparx5_port_qos_dscp *qos);
+
+int sparx5_port_qos_default_set(const struct sparx5_port *port,
+				const struct sparx5_port_qos *qos);
+
 #endif	/* __SPARX5_PORT_H__ */
Signature · 1 line
2.34.1
#2 lore raw

Re: [net-next v3 1/6] net: dcb: add new pcp selector to app object

In reply to Daniel Machon
Quoted 47 lines · Daniel Machon

Add new PCP selector for the 8021Qaz APP managed object.

As the PCP selector is not part of the 8021Qaz standard, a new non-std extension attribute DCB_ATTR_DCB_APP has been introduced. Also two helper functions to translate between selector and app attribute type has been added. The new selector has been given a value of 255, to minimize the risk of future overlap of std- and non-std attributes.

The new DCB_ATTR_DCB_APP is sent alongside the ieee std attribute in the app table. This means that the dcb_app struct can now both contain std- and non-std app attributes. Currently there is no overlap between the selector values of the two attributes.

The purpose of adding the PCP selector, is to be able to offload PCP-based queue classification to the 8021Q Priority Code Point table, see 6.9.3 of IEEE Std 802.1Q-2018.

PCP and DEI is encoded in the protocol field as 8*dei+pcp, so that a mapping of PCP 2 and DEI 1 to priority 3 is encoded as {255, 10, 3}.

static struct sk_buff *dcbnl_newmsg(int type, u8 cmd, u32 port, u32 seq,

net/dcb/dcbnl.c +4 -3
 				    u32 flags, struct nlmsghdr **nlhp)
 {
@@ -1116,8 +1143,9 @@ static int dcbnl_ieee_fill(struct sk_buff *skb, struct net_device *netdev)
 	spin_lock_bh(&dcb_lock);
 	list_for_each_entry(itr, &dcb_app_list, list) {
 		if (itr->ifindex == netdev->ifindex) {
-			err = nla_put(skb, DCB_ATTR_IEEE_APP, sizeof(itr->app),
-					 &itr->app);
+			enum ieee_attrs_app type =
+				dcbnl_app_attr_type_get(itr->app.selector);
+			err = nla_put(skb, type, sizeof(itr->app), &itr->app);
 			if (err) {
 				spin_unlock_bh(&dcb_lock);
 				return -EMSGSIZE;
@@ -1495,7 +1523,7 @@ static int dcbnl_ieee_set(struct net_device *netdev, struct nlmsghdr *nlh,
 		nla_for_each_nested(attr, ieee[DCB_ATTR_IEEE_APP_TABLE], rem) {
 			struct dcb_app *app_data;
 
-			if (nla_type(attr) != DCB_ATTR_IEEE_APP)
+			if (!dcbnl_app_attr_type_validate(nla_type(attr)))
 				continue;
 
 			if (nla_len(attr) < sizeof(struct dcb_app)) {

net/dcb/dcbnl.c static int dcbnl_ieee_del(struct net_device *netdev, struct nlmsghdr *nlh, · line 1590

Quoted 8 lines
net/dcb/dcbnl.c +1 -1
@@ -1556,7 +1584,7 @@ static int dcbnl_ieee_del(struct net_device *netdev, struct nlmsghdr *nlh,
 		nla_for_each_nested(attr, ieee[DCB_ATTR_IEEE_APP_TABLE], rem) {
 			struct dcb_app *app_data;
 
-			if (nla_type(attr) != DCB_ATTR_IEEE_APP)
+			if (!dcbnl_app_attr_type_validate(nla_type(attr)))
 				continue;
 			app_data = nla_data(attr);
 			if (ops->ieee_delapp)

I'm missing a validation that DCB_APP_SEL_PCP is always sent in DCB_ATTR_DCB_APP encapsulation. Wouldn't the current code permit sending it in the IEEE encap? This should be forbidden.

And vice versa: I'm not sure we want to permit sending the standard attributes in the DCB encap.

#8 lore raw

Re: [net-next v3 2/6] net: dcb: add new apptrust attribute

In reply to Daniel Machon
Quoted 99 lines · Daniel Machon

Add new apptrust extension attributes to the 8021Qaz APP managed object.

Two new attributes, DCB_ATTR_DCB_APP_TRUST_TABLE and DCB_ATTR_DCB_APP_TRUST, has been added. Trusted selectors are passed in the nested attribute DCB_ATTR_DCB_APP_TRUST, in order of precedence.

The new attributes are meant to allow drivers, whose hw supports the notion of trust, to be able to set whether a particular app selector is trusted - and in which order.

---
 include/net/dcbnl.h        |  4 ++
 include/uapi/linux/dcbnl.h | 10 +++++
 net/dcb/dcbnl.c            | 77 ++++++++++++++++++++++++++++++++++++--
 3 files changed, 87 insertions(+), 4 deletions(-)
include/net/dcbnl.h +4 -0
index 2b2d86fb3131..8841ab6c2de7 100644
@@ -109,6 +109,10 @@ struct dcbnl_rtnl_ops {
 	/* buffer settings */
 	int (*dcbnl_getbuffer)(struct net_device *, struct dcbnl_buffer *);
 	int (*dcbnl_setbuffer)(struct net_device *, struct dcbnl_buffer *);
+
+	/* apptrust */
+	int (*dcbnl_setapptrust)(struct net_device *, u8 *, int);
+	int (*dcbnl_getapptrust)(struct net_device *, u8 *, int *);
 };
 
 #endif /* __NET_DCBNL_H__ */
include/uapi/linux/dcbnl.h +10 -0
index dc7ef96207ca..9344e3ba5768 100644
@@ -410,6 +410,7 @@ enum dcbnl_attrs {
  * @DCB_ATTR_IEEE_PEER_ETS: peer ETS configuration - get only
  * @DCB_ATTR_IEEE_PEER_PFC: peer PFC configuration - get only
  * @DCB_ATTR_IEEE_PEER_APP: peer APP tlv - get only
+ * @DCB_ATTR_DCB_APP_TRUST_TABLE: selector trust table
  */
 enum ieee_attrs {
 	DCB_ATTR_IEEE_UNSPEC,
@@ -423,6 +424,7 @@ enum ieee_attrs {
 	DCB_ATTR_IEEE_QCN,
 	DCB_ATTR_IEEE_QCN_STATS,
 	DCB_ATTR_DCB_BUFFER,
+	DCB_ATTR_DCB_APP_TRUST_TABLE,
 	__DCB_ATTR_IEEE_MAX
 };
 #define DCB_ATTR_IEEE_MAX (__DCB_ATTR_IEEE_MAX - 1)
@@ -435,6 +437,14 @@ enum ieee_attrs_app {
 };
 #define DCB_ATTR_IEEE_APP_MAX (__DCB_ATTR_IEEE_APP_MAX - 1)
 
+enum dcbnl_attrs_apptrust {
+	DCB_ATTR_DCB_APP_TRUST_UNSPEC,
+	DCB_ATTR_DCB_APP_TRUST,
+	__DCB_ATTR_DCB_APP_TRUST_MAX
+};
+
+#define DCB_ATTR_DCB_APP_TRUST_MAX (__DCB_ATTR_DCB_APP_TRUST_MAX - 1)
+
 /**
  * enum cee_attrs - CEE DCBX get attributes.
  *
net/dcb/dcbnl.c +5 -3
index 92c32bc11374..01310edf6d1b 100644
@@ -166,6 +166,7 @@ static const struct nla_policy dcbnl_ieee_policy[DCB_ATTR_IEEE_MAX + 1] = {
 	[DCB_ATTR_IEEE_QCN]         = {.len = sizeof(struct ieee_qcn)},
 	[DCB_ATTR_IEEE_QCN_STATS]   = {.len = sizeof(struct ieee_qcn_stats)},
 	[DCB_ATTR_DCB_BUFFER]       = {.len = sizeof(struct dcbnl_buffer)},
+	[DCB_ATTR_DCB_APP_TRUST_TABLE] = {.type = NLA_NESTED},
 };
 
 /* DCB number of traffic classes nested attributes. */
@@ -1057,11 +1058,11 @@ static int dcbnl_build_peer_app(struct net_device *netdev, struct sk_buff* skb,
 /* Handle IEEE 802.1Qaz/802.1Qau/802.1Qbb GET commands. */
 static int dcbnl_ieee_fill(struct sk_buff *skb, struct net_device *netdev)
 {
-	struct nlattr *ieee, *app;
-	struct dcb_app_type *itr;
 	const struct dcbnl_rtnl_ops *ops = netdev->dcbnl_ops;
+	struct nlattr *ieee, *app, *apptrust;
+	struct dcb_app_type *itr;
+	int err, i;
 	int dcbx;
-	int err;
 
 	if (nla_put_string(skb, DCB_ATTR_IFNAME, netdev->name))
 		return -EMSGSIZE;
@@ -1161,6 +1162,24 @@ static int dcbnl_ieee_fill(struct sk_buff *skb, struct net_device *netdev)
 	spin_unlock_bh(&dcb_lock);
 	nla_nest_end(skb, app);
 
+	if (ops->dcbnl_getapptrust) {

net/dcb/dcbnl.c static int dcbnl_ieee_fill(struct sk_buff *skb, struct net_device *netdev) · line 1175

Quoted 10 lines
net/dcb/dcbnl.c +10 -0
+		u8 selectors[IEEE_8021QAZ_APP_SEL_MAX + 1] = {0};
+		int nselectors;
+
+		apptrust = nla_nest_start(skb, DCB_ATTR_DCB_APP_TRUST_TABLE);
+		if (!app)
+			return -EMSGSIZE;
+
+		err = ops->dcbnl_getapptrust(netdev, selectors, &nselectors);
+		if (err)
+			return -EMSGSIZE;

This should return the error coming from the driver instead of -EMSGSIZE.

Also, it should cancel the nest before returning.

Quoted 11 lines
net/dcb/dcbnl.c +7 -0
+
+		for (i = 0; i < nselectors; i++)
+			nla_put_u8(skb, DCB_ATTR_DCB_APP_TRUST, selectors[i]);
+
+		nla_nest_end(skb, apptrust);
+	}
+
 	/* get peer info if available */
 	if (ops->ieee_peer_getets) {
 		struct ieee_ets ets;

net/dcb/dcbnl.c static int dcbnl_ieee_set(struct net_device *netdev, struct nlmsghdr *nlh, · line 1458

Quoted 6 lines
net/dcb/dcbnl.c +1 -1
@@ -1454,8 +1473,8 @@ static int dcbnl_ieee_set(struct net_device *netdev, struct nlmsghdr *nlh,
 {
 	const struct dcbnl_rtnl_ops *ops = netdev->dcbnl_ops;
 	struct nlattr *ieee[DCB_ATTR_IEEE_MAX + 1];
+	int err, i;
 	int prio;
-	int err;

I don't really mind, but I have to wonder why this new variable is not with the rest of them in the trust table scope.

Quoted 60 lines
net/dcb/dcbnl.c +50 -0
 	if (!ops)
 		return -EOPNOTSUPP;
@@ -1541,6 +1560,56 @@ static int dcbnl_ieee_set(struct net_device *netdev, struct nlmsghdr *nlh,
 		}
 	}
 
+	if (ieee[DCB_ATTR_DCB_APP_TRUST_TABLE]) {
+		u8 selectors[IEEE_8021QAZ_APP_SEL_MAX + 1] = {0};
+		struct nlattr *attr;
+		int nselectors = 0;
+		u8 selector;
+		int rem;
+
+		if (!ops->dcbnl_setapptrust) {
+			err = -EOPNOTSUPP;
+			goto err;
+		}
+
+		nla_for_each_nested(attr, ieee[DCB_ATTR_DCB_APP_TRUST_TABLE],
+				    rem) {
+			if (nla_type(attr) != DCB_ATTR_DCB_APP_TRUST ||
+			    nla_len(attr) != 1 ||
+			    nselectors >= sizeof(selectors)) {
+				err = -EINVAL;
+				goto err;
+			}
+
+			selector = nla_get_u8(attr);
+			switch (selector) {
+			case IEEE_8021QAZ_APP_SEL_ETHERTYPE:
+			case IEEE_8021QAZ_APP_SEL_STREAM:
+			case IEEE_8021QAZ_APP_SEL_DGRAM:
+			case IEEE_8021QAZ_APP_SEL_ANY:
+			case IEEE_8021QAZ_APP_SEL_DSCP:
+			case DCB_APP_SEL_PCP:
+				break;
+			default:
+				err = -EINVAL;
+				goto err;
+			}
+			/* Duplicate selector ? */
+			for (i = 0; i < nselectors; i++) {
+				if (selectors[i] == selector) {
+					err = -EINVAL;
+					goto err;
+				}
+			}
+
+			selectors[nselectors++] = selector;
+		}
+
+		err = ops->dcbnl_setapptrust(netdev, selectors, nselectors);
+		if (err)
+			goto err;
+	}
+
 err:
 	err = nla_put_u8(skb, DCB_ATTR_IEEE, err);
 	dcbnl_ieee_notify(netdev, RTM_SETDCB, DCB_CMD_IEEE_SET, seq, 0);
author maintainer
#3 lore raw

Re: [net-next v3 1/6] net: dcb: add new pcp selector to app object

In reply to Petr Machata
Quoted 50 lines
Quoted 50 lines

Add new PCP selector for the 8021Qaz APP managed object.

As the PCP selector is not part of the 8021Qaz standard, a new non-std extension attribute DCB_ATTR_DCB_APP has been introduced. Also two helper functions to translate between selector and app attribute type has been added. The new selector has been given a value of 255, to minimize the risk of future overlap of std- and non-std attributes.

The new DCB_ATTR_DCB_APP is sent alongside the ieee std attribute in the app table. This means that the dcb_app struct can now both contain std- and non-std app attributes. Currently there is no overlap between the selector values of the two attributes.

The purpose of adding the PCP selector, is to be able to offload PCP-based queue classification to the 8021Q Priority Code Point table, see 6.9.3 of IEEE Std 802.1Q-2018.

PCP and DEI is encoded in the protocol field as 8*dei+pcp, so that a mapping of PCP 2 and DEI 1 to priority 3 is encoded as {255, 10, 3}.

static struct sk_buff *dcbnl_newmsg(int type, u8 cmd, u32 port, u32 seq,
u32 flags, struct nlmsghdr **nlhp)
{

net/dcb/dcbnl.c +4 -3
@@ -1116,8 +1143,9 @@ static int dcbnl_ieee_fill(struct sk_buff *skb, struct net_device *netdev)
      spin_lock_bh(&dcb_lock);
      list_for_each_entry(itr, &dcb_app_list, list) {
              if (itr->ifindex == netdev->ifindex) {
-                     err = nla_put(skb, DCB_ATTR_IEEE_APP, sizeof(itr->app),
-                                      &itr->app);
+                     enum ieee_attrs_app type =
+                             dcbnl_app_attr_type_get(itr->app.selector);
+                     err = nla_put(skb, type, sizeof(itr->app), &itr->app);
                      if (err) {
                              spin_unlock_bh(&dcb_lock);
                              return -EMSGSIZE;
@@ -1495,7 +1523,7 @@ static int dcbnl_ieee_set(struct net_device *netdev, struct nlmsghdr *nlh,
              nla_for_each_nested(attr, ieee[DCB_ATTR_IEEE_APP_TABLE], rem) {
                      struct dcb_app *app_data;
 
-                     if (nla_type(attr) != DCB_ATTR_IEEE_APP)
+                     if (!dcbnl_app_attr_type_validate(nla_type(attr)))
                              continue;
 
                      if (nla_len(attr) < sizeof(struct dcb_app)) {
@@ -1556,7 +1584,7 @@ static int dcbnl_ieee_del(struct net_device *netdev, struct nlmsghdr *nlh,
              nla_for_each_nested(attr, ieee[DCB_ATTR_IEEE_APP_TABLE], rem) {
                      struct dcb_app *app_data;
 
Quoted 9 lines

net/dcb/dcbnl.c static int dcbnl_ieee_del(struct net_device *netdev, struct nlmsghdr *nlh, · line 1590

Quoted 5 lines
net/dcb/dcbnl.c +1 -1
-                     if (nla_type(attr) != DCB_ATTR_IEEE_APP)
+                     if (!dcbnl_app_attr_type_validate(nla_type(attr)))
                              continue;
                      app_data = nla_data(attr);
                      if (ops->ieee_delapp)

I'm missing a validation that DCB_APP_SEL_PCP is always sent in DCB_ATTR_DCB_APP encapsulation. Wouldn't the current code permit sending it in the IEEE encap? This should be forbidden.

Right. Current impl. does not check that the non-std selectors received, are sent with a DCB_ATTR_DCB_APP type. We could introduce a new check dcbnl_app_attr_selector_validate() that checks combination of type and selector, after the type and nla_len(attr) has been checked, so that:

validate type -> validate nla_len(attr) -> validate selector

Quoted 2 lines

And vice versa: I'm not sure we want to permit sending the standard attributes in the DCB encap.

dcbnl_app_attr_type_get() in dcbnl_ieee_fill() takes care of this. IEEE are always sent in DCB_ATTR_IEEE and non-std are sent in DCB_ATTR_DCB.

/ Daniel

author maintainer
#9 lore raw

Re: [net-next v3 2/6] net: dcb: add new apptrust attribute

In reply to Petr Machata
Quoted 6 lines
Quoted 6 lines
net/dcb/dcbnl.c +6 -0
+     if (ops->dcbnl_getapptrust) {
+             u8 selectors[IEEE_8021QAZ_APP_SEL_MAX + 1] = {0};
+             int nselectors;
+
+             apptrust = nla_nest_start(skb, DCB_ATTR_DCB_APP_TRUST_TABLE);
+             if (!app)
Quoted 8 lines

net/dcb/dcbnl.c static int dcbnl_ieee_fill(struct sk_buff *skb, struct net_device *netdev) · line 1175

Quoted 5 lines
net/dcb/dcbnl.c +5 -0
+                     return -EMSGSIZE;
+
+             err = ops->dcbnl_getapptrust(netdev, selectors, &nselectors);
+             if (err)
+                     return -EMSGSIZE;

This should return the error coming from the driver instead of -EMSGSIZE.

Hmm. The question is whether we should return at all if dcbnl_getapptrust() fails, or just continue to fill whatever other ieee information is available? Seems to be like the rest of the code in ieee_fill() just ignore that error.

Quoted 2 lines

Also, it should cancel the nest before returning.

Ack.

#4 lore raw

Re: [net-next v3 1/6] net: dcb: add new pcp selector to app object

In reply to [email protected]
Quoted 10 lines · <[email protected]> writes:
Quoted 3 lines

I'm missing a validation that DCB_APP_SEL_PCP is always sent in DCB_ATTR_DCB_APP encapsulation. Wouldn't the current code permit sending it in the IEEE encap? This should be forbidden.

Right. Current impl. does not check that the non-std selectors received, are sent with a DCB_ATTR_DCB_APP type. We could introduce a new check dcbnl_app_attr_selector_validate() that checks combination of type and selector, after the type and nla_len(attr) has been checked, so that:

validate type -> validate nla_len(attr) -> validate selector

This needs to be validated, otherwise there's no point in having a dedicated attribute for the non-standard stuff.

Quoted 5 lines
Quoted 2 lines

And vice versa: I'm not sure we want to permit sending the standard attributes in the DCB encap.

dcbnl_app_attr_type_get() in dcbnl_ieee_fill() takes care of this. IEEE are always sent in DCB_ATTR_IEEE and non-std are sent in DCB_ATTR_DCB.

By "sending" I meant userspace sending this to the kernel. So bounce extended opcodes that are wrapped in IEEE and bounce IEEE opcodes wrapped in DCB as well.

#10 lore raw

Re: [net-next v3 2/6] net: dcb: add new apptrust attribute

In reply to [email protected]
Quoted 15 lines · <[email protected]> writes:
Quoted 6 lines
Quoted 6 lines
net/dcb/dcbnl.c +6 -0
+     if (ops->dcbnl_getapptrust) {
+             u8 selectors[IEEE_8021QAZ_APP_SEL_MAX + 1] = {0};
+             int nselectors;
+
+             apptrust = nla_nest_start(skb, DCB_ATTR_DCB_APP_TRUST_TABLE);
+             if (!app)
Quoted 8 lines

net/dcb/dcbnl.c static int dcbnl_ieee_fill(struct sk_buff *skb, struct net_device *netdev) · line 1175

Quoted 5 lines
net/dcb/dcbnl.c +5 -0
+                     return -EMSGSIZE;
+
+             err = ops->dcbnl_getapptrust(netdev, selectors, &nselectors);
+             if (err)
+                     return -EMSGSIZE;

This should return the error coming from the driver instead of -EMSGSIZE.

Quoted 3 lines

Hmm. The question is whether we should return at all if dcbnl_getapptrust() fails, or just continue to fill whatever other ieee information is available? Seems to be like the rest of the code in ieee_fill() just ignore that error.

I see. I guess the reasoning is that one broken callback shouldn't lead to dropping the whole query. So yeah, let's do what the rest of the code does. But it seems rather cavalier to disregard error codes like this.

author maintainer
#5 lore raw

Re: [net-next v3 1/6] net: dcb: add new pcp selector to app object

In reply to Petr Machata
Quoted 9 lines
Quoted 9 lines
Quoted 3 lines

I'm missing a validation that DCB_APP_SEL_PCP is always sent in DCB_ATTR_DCB_APP encapsulation. Wouldn't the current code permit sending it in the IEEE encap? This should be forbidden.

Right. Current impl. does not check that the non-std selectors received, are sent with a DCB_ATTR_DCB_APP type. We could introduce a new check dcbnl_app_attr_selector_validate() that checks combination of type and selector, after the type and nla_len(attr) has been checked, so that:

Quoted 4 lines
Quoted 1 line

validate type -> validate nla_len(attr) -> validate selector

This needs to be validated, otherwise there's no point in having a dedicated attribute for the non-standard stuff.

Agree.

Quoted 10 lines
Quoted 6 lines
Quoted 3 lines

And vice versa: I'm not sure we want to permit sending the standard attributes in the DCB encap.

dcbnl_app_attr_type_get() in dcbnl_ieee_fill() takes care of this. IEEE are always sent in DCB_ATTR_IEEE and non-std are sent in DCB_ATTR_DCB.

By "sending" I meant userspace sending this to the kernel. So bounce extended opcodes that are wrapped in IEEE and bounce IEEE opcodes wrapped in DCB as well.

Right. Then we only need to decide what to do with any opcode in-between (not defined in uapi, neither ieee or extension opcode, 7-254). If they are sent in DCB_ATTR_DCB they should be bounced, because we agreed that we can interpret data in the new attr), _but_ if they are sent in DCB_ATTR_IEEE I guess we should accept them, to not break userspace that is already sending them.

Here is what that could look like:

	/* Make sure any non-std selectors is always encapsulated in the non-std
	* DCB_ATTR_DCB_APP attribute.
	*/
	static bool dcbnl_app_attr_selector_validate(enum ieee_attrs_app type, u32 selector)
	{
		switch (selector) {
		case DCB_APP_SEL_PCP:
			/* Non-std selector in non-std attr? */
			if (type == DCB_ATTR_DCB_APP)
				return true;
		default:
			/* Std selector in std attr? */
			if (type == DCB_ATTR_IEEE_APP)
				return true;
		}

		return false;
	}

/ Daniel

#6 lore raw

Re: [net-next v3 1/6] net: dcb: add new pcp selector to app object

Quoted 10 lines · <[email protected]> writes:
Quoted 9 lines
Quoted 5 lines
Quoted 2 lines

And vice versa: I'm not sure we want to permit sending the standard attributes in the DCB encap.

dcbnl_app_attr_type_get() in dcbnl_ieee_fill() takes care of this. IEEE are always sent in DCB_ATTR_IEEE and non-std are sent in DCB_ATTR_DCB.

By "sending" I meant userspace sending this to the kernel. So bounce extended opcodes that are wrapped in IEEE and bounce IEEE opcodes wrapped in DCB as well.

Quoted 6 lines

Right. Then we only need to decide what to do with any opcode in-between (not defined in uapi, neither ieee or extension opcode, 7-254). If they are sent in DCB_ATTR_DCB they should be bounced, because we agreed that we can interpret data in the new attr), _but_ if they are sent in DCB_ATTR_IEEE I guess we should accept them, to not break userspace that is already sending them.

I see, it's not currently validating at all. It just relies on the driver to do the validation, but e.g. bnxt_dcbnl_ieee_setapp(), just lets nonsense right through.

OK, but this interface is built on standards. The selector has a well-defined, IEEE-backed meaning with enumerators published in the UAPI headers. As before, even though this constitutes API breakage, IMHO if anyone relies on shoving random garbage through this interface, it's on them...

I think it's kosher to start bouncing undefined selectors.