* [PATCH v3 net-next 0/6] tc-gate offload for SJA1105 DSA switch
@ 2020-05-05 19:20 Vladimir Oltean
2020-05-05 19:20 ` [PATCH v3 net-next 1/6] net: dsa: introduce a dsa_port_from_netdev public helper Vladimir Oltean
` (6 more replies)
0 siblings, 7 replies; 9+ messages in thread
From: Vladimir Oltean @ 2020-05-05 19:20 UTC (permalink / raw)
To: netdev; +Cc: andrew, f.fainelli, vivien.didelot, vinicius.gomes, po.liu
From: Vladimir Oltean <vladimir.oltean@nxp.com>
Expose the TTEthernet hardware features of the switch using standard
tc-flower actions: trap, drop, redirect and gate.
v1 was submitted at:
https://patchwork.ozlabs.org/project/netdev/cover/20200503211035.19363-1-olteanv@gmail.com/
v2 was submitted at:
https://patchwork.ozlabs.org/project/netdev/cover/20200503211035.19363-1-olteanv@gmail.com/
Changes in v3:
Made sure there are no compilation warnings when
CONFIG_NET_DSA_SJA1105_TAS or CONFIG_NET_DSA_SJA1105_VL are disabled.
Changes in v2:
Using a newly introduced dsa_port_from_netdev public helper.
Vladimir Oltean (6):
net: dsa: introduce a dsa_port_from_netdev public helper
net: dsa: sja1105: add static tables for virtual links
net: dsa: sja1105: make room for virtual link parsing in flower
offload
net: dsa: sja1105: support flow-based redirection via virtual links
net: dsa: sja1105: implement tc-gate using time-triggered virtual
links
docs: net: dsa: sja1105: document intended usage of virtual links
Documentation/networking/dsa/sja1105.rst | 116 +++
drivers/net/dsa/sja1105/Kconfig | 9 +
drivers/net/dsa/sja1105/Makefile | 4 +
drivers/net/dsa/sja1105/sja1105.h | 59 +-
.../net/dsa/sja1105/sja1105_dynamic_config.c | 51 ++
drivers/net/dsa/sja1105/sja1105_flower.c | 215 ++++-
drivers/net/dsa/sja1105/sja1105_main.c | 13 +-
drivers/net/dsa/sja1105/sja1105_ptp.h | 13 +
drivers/net/dsa/sja1105/sja1105_spi.c | 2 +
.../net/dsa/sja1105/sja1105_static_config.c | 202 +++++
.../net/dsa/sja1105/sja1105_static_config.h | 65 ++
drivers/net/dsa/sja1105/sja1105_tas.c | 127 ++-
drivers/net/dsa/sja1105/sja1105_tas.h | 36 +
drivers/net/dsa/sja1105/sja1105_vl.c | 796 ++++++++++++++++++
drivers/net/dsa/sja1105/sja1105_vl.h | 72 ++
include/net/dsa.h | 1 +
net/dsa/dsa.c | 9 +
17 files changed, 1746 insertions(+), 44 deletions(-)
create mode 100644 drivers/net/dsa/sja1105/sja1105_vl.c
create mode 100644 drivers/net/dsa/sja1105/sja1105_vl.h
--
2.17.1
^ permalink raw reply [flat|nested] 9+ messages in thread
* [PATCH v3 net-next 1/6] net: dsa: introduce a dsa_port_from_netdev public helper
2020-05-05 19:20 [PATCH v3 net-next 0/6] tc-gate offload for SJA1105 DSA switch Vladimir Oltean
@ 2020-05-05 19:20 ` Vladimir Oltean
2020-05-06 15:55 ` Vivien Didelot
2020-05-05 19:20 ` [PATCH v3 net-next 2/6] net: dsa: sja1105: add static tables for virtual links Vladimir Oltean
` (5 subsequent siblings)
6 siblings, 1 reply; 9+ messages in thread
From: Vladimir Oltean @ 2020-05-05 19:20 UTC (permalink / raw)
To: netdev; +Cc: andrew, f.fainelli, vivien.didelot, vinicius.gomes, po.liu
From: Vladimir Oltean <vladimir.oltean@nxp.com>
As its implementation shows, this is synonimous with calling
dsa_slave_dev_check followed by dsa_slave_to_port, so it is quite simple
already and provides functionality which is already there.
However there is now a need for these functions outside dsa_priv.h, for
example in drivers that perform mirroring and redirection through
tc-flower offloads (they are given raw access to the flow_cls_offload
structure), where they need to call this function on act->dev.
But simply exporting dsa_slave_to_port would make it non-inline and
would result in an extra function call in the hotpath, as can be seen
for example in sja1105:
Before:
000006dc <sja1105_xmit>:
{
6dc: e92d4ff0 push {r4, r5, r6, r7, r8, r9, sl, fp, lr}
6e0: e1a04000 mov r4, r0
6e4: e591958c ldr r9, [r1, #1420] ; 0x58c <- Inline dsa_slave_to_port
6e8: e1a05001 mov r5, r1
6ec: e24dd004 sub sp, sp, #4
u16 tx_vid = dsa_8021q_tx_vid(dp->ds, dp->index);
6f0: e1c901d8 ldrd r0, [r9, #24]
6f4: ebfffffe bl 0 <dsa_8021q_tx_vid>
6f4: R_ARM_CALL dsa_8021q_tx_vid
u8 pcp = netdev_txq_to_tc(netdev, queue_mapping);
6f8: e1d416b0 ldrh r1, [r4, #96] ; 0x60
u16 tx_vid = dsa_8021q_tx_vid(dp->ds, dp->index);
6fc: e1a08000 mov r8, r0
After:
000006e4 <sja1105_xmit>:
{
6e4: e92d4ff0 push {r4, r5, r6, r7, r8, r9, sl, fp, lr}
6e8: e1a04000 mov r4, r0
6ec: e24dd004 sub sp, sp, #4
struct dsa_port *dp = dsa_slave_to_port(netdev);
6f0: e1a00001 mov r0, r1
{
6f4: e1a05001 mov r5, r1
struct dsa_port *dp = dsa_slave_to_port(netdev);
6f8: ebfffffe bl 0 <dsa_slave_to_port>
6f8: R_ARM_CALL dsa_slave_to_port
6fc: e1a09000 mov r9, r0
u16 tx_vid = dsa_8021q_tx_vid(dp->ds, dp->index);
700: e1c001d8 ldrd r0, [r0, #24]
704: ebfffffe bl 0 <dsa_8021q_tx_vid>
704: R_ARM_CALL dsa_8021q_tx_vid
Because we want to avoid possible performance regressions, introduce
this new function which is designed to be public.
Suggested-by: Vivien Didelot <vivien.didelot@gmail.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
Changes in v3:
None.
Changes in v2:
Previous patch "net: dsa: export dsa_slave_dev_check and
dsa_slave_to_port" was replaced with this one.
include/net/dsa.h | 1 +
net/dsa/dsa.c | 9 +++++++++
2 files changed, 10 insertions(+)
diff --git a/include/net/dsa.h b/include/net/dsa.h
index fb3f9222f2a1..6dfc8c2f68b8 100644
--- a/include/net/dsa.h
+++ b/include/net/dsa.h
@@ -637,6 +637,7 @@ void dsa_devlink_resource_occ_get_register(struct dsa_switch *ds,
void *occ_get_priv);
void dsa_devlink_resource_occ_get_unregister(struct dsa_switch *ds,
u64 resource_id);
+struct dsa_port *dsa_port_from_netdev(struct net_device *netdev);
struct dsa_devlink_priv {
struct dsa_switch *ds;
diff --git a/net/dsa/dsa.c b/net/dsa/dsa.c
index 0384a911779e..1ce9ba8cf545 100644
--- a/net/dsa/dsa.c
+++ b/net/dsa/dsa.c
@@ -412,6 +412,15 @@ void dsa_devlink_resource_occ_get_unregister(struct dsa_switch *ds,
}
EXPORT_SYMBOL_GPL(dsa_devlink_resource_occ_get_unregister);
+struct dsa_port *dsa_port_from_netdev(struct net_device *netdev)
+{
+ if (!netdev || !dsa_slave_dev_check(netdev))
+ return ERR_PTR(-ENODEV);
+
+ return dsa_slave_to_port(netdev);
+}
+EXPORT_SYMBOL_GPL(dsa_port_from_netdev);
+
static int __init dsa_init_module(void)
{
int rc;
--
2.17.1
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH v3 net-next 2/6] net: dsa: sja1105: add static tables for virtual links
2020-05-05 19:20 [PATCH v3 net-next 0/6] tc-gate offload for SJA1105 DSA switch Vladimir Oltean
2020-05-05 19:20 ` [PATCH v3 net-next 1/6] net: dsa: introduce a dsa_port_from_netdev public helper Vladimir Oltean
@ 2020-05-05 19:20 ` Vladimir Oltean
2020-05-05 19:20 ` [PATCH v3 net-next 3/6] net: dsa: sja1105: make room for virtual link parsing in flower offload Vladimir Oltean
` (4 subsequent siblings)
6 siblings, 0 replies; 9+ messages in thread
From: Vladimir Oltean @ 2020-05-05 19:20 UTC (permalink / raw)
To: netdev; +Cc: andrew, f.fainelli, vivien.didelot, vinicius.gomes, po.liu
From: Vladimir Oltean <vladimir.oltean@nxp.com>
This patch adds the register definitions for the:
- VL Lookup Table
- VL Policing Table
- VL Forwarding Table
- VL Forwarding Parameters Table
These are needed in order to perform TTEthernet operations: QoS
classification, flow-based policing and/or frame redirecting with the
switch.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
Changes from v2:
None.
Changes from v1:
None.
Changes from RFC:
None.
drivers/net/dsa/sja1105/sja1105.h | 2 +
.../net/dsa/sja1105/sja1105_dynamic_config.c | 51 +++++
.../net/dsa/sja1105/sja1105_static_config.c | 202 ++++++++++++++++++
.../net/dsa/sja1105/sja1105_static_config.h | 63 ++++++
4 files changed, 318 insertions(+)
diff --git a/drivers/net/dsa/sja1105/sja1105.h b/drivers/net/dsa/sja1105/sja1105.h
index 2f62942692ec..602aa30c832f 100644
--- a/drivers/net/dsa/sja1105/sja1105.h
+++ b/drivers/net/dsa/sja1105/sja1105.h
@@ -254,6 +254,8 @@ size_t sja1105pqrs_mac_config_entry_packing(void *buf, void *entry_ptr,
enum packing_op op);
size_t sja1105pqrs_avb_params_entry_packing(void *buf, void *entry_ptr,
enum packing_op op);
+size_t sja1105_vl_lookup_entry_packing(void *buf, void *entry_ptr,
+ enum packing_op op);
/* From sja1105_flower.c */
int sja1105_cls_flower_del(struct dsa_switch *ds, int port,
diff --git a/drivers/net/dsa/sja1105/sja1105_dynamic_config.c b/drivers/net/dsa/sja1105/sja1105_dynamic_config.c
index bf9b36ff35bf..bdee01811960 100644
--- a/drivers/net/dsa/sja1105/sja1105_dynamic_config.c
+++ b/drivers/net/dsa/sja1105/sja1105_dynamic_config.c
@@ -97,6 +97,12 @@
#define SJA1105_SIZE_DYN_CMD 4
+#define SJA1105ET_SJA1105_SIZE_VL_LOOKUP_DYN_CMD \
+ SJA1105_SIZE_DYN_CMD
+
+#define SJA1105PQRS_SJA1105_SIZE_VL_LOOKUP_DYN_CMD \
+ (SJA1105_SIZE_DYN_CMD + SJA1105_SIZE_VL_LOOKUP_ENTRY)
+
#define SJA1105ET_SIZE_MAC_CONFIG_DYN_ENTRY \
SJA1105_SIZE_DYN_CMD
@@ -146,6 +152,29 @@ enum sja1105_hostcmd {
SJA1105_HOSTCMD_INVALIDATE = 4,
};
+static void
+sja1105_vl_lookup_cmd_packing(void *buf, struct sja1105_dyn_cmd *cmd,
+ enum packing_op op)
+{
+ const int size = SJA1105_SIZE_DYN_CMD;
+
+ sja1105_packing(buf, &cmd->valid, 31, 31, size, op);
+ sja1105_packing(buf, &cmd->errors, 30, 30, size, op);
+ sja1105_packing(buf, &cmd->rdwrset, 29, 29, size, op);
+ sja1105_packing(buf, &cmd->index, 9, 0, size, op);
+}
+
+static size_t sja1105et_vl_lookup_entry_packing(void *buf, void *entry_ptr,
+ enum packing_op op)
+{
+ struct sja1105_vl_lookup_entry *entry = entry_ptr;
+ const int size = SJA1105ET_SJA1105_SIZE_VL_LOOKUP_DYN_CMD;
+
+ sja1105_packing(buf, &entry->egrmirr, 21, 17, size, op);
+ sja1105_packing(buf, &entry->ingrmirr, 16, 16, size, op);
+ return size;
+}
+
static void
sja1105pqrs_l2_lookup_cmd_packing(void *buf, struct sja1105_dyn_cmd *cmd,
enum packing_op op)
@@ -505,6 +534,16 @@ sja1105pqrs_avb_params_cmd_packing(void *buf, struct sja1105_dyn_cmd *cmd,
struct sja1105_dynamic_table_ops sja1105et_dyn_ops[BLK_IDX_MAX_DYN] = {
[BLK_IDX_SCHEDULE] = {0},
[BLK_IDX_SCHEDULE_ENTRY_POINTS] = {0},
+ [BLK_IDX_VL_LOOKUP] = {
+ .entry_packing = sja1105et_vl_lookup_entry_packing,
+ .cmd_packing = sja1105_vl_lookup_cmd_packing,
+ .access = OP_WRITE,
+ .max_entry_count = SJA1105_MAX_VL_LOOKUP_COUNT,
+ .packed_size = SJA1105ET_SJA1105_SIZE_VL_LOOKUP_DYN_CMD,
+ .addr = 0x35,
+ },
+ [BLK_IDX_VL_POLICING] = {0},
+ [BLK_IDX_VL_FORWARDING] = {0},
[BLK_IDX_L2_LOOKUP] = {
.entry_packing = sja1105et_dyn_l2_lookup_entry_packing,
.cmd_packing = sja1105et_l2_lookup_cmd_packing,
@@ -548,6 +587,7 @@ struct sja1105_dynamic_table_ops sja1105et_dyn_ops[BLK_IDX_MAX_DYN] = {
},
[BLK_IDX_SCHEDULE_PARAMS] = {0},
[BLK_IDX_SCHEDULE_ENTRY_POINTS_PARAMS] = {0},
+ [BLK_IDX_VL_FORWARDING_PARAMS] = {0},
[BLK_IDX_L2_LOOKUP_PARAMS] = {
.entry_packing = sja1105et_l2_lookup_params_entry_packing,
.cmd_packing = sja1105et_l2_lookup_params_cmd_packing,
@@ -573,6 +613,16 @@ struct sja1105_dynamic_table_ops sja1105et_dyn_ops[BLK_IDX_MAX_DYN] = {
struct sja1105_dynamic_table_ops sja1105pqrs_dyn_ops[BLK_IDX_MAX_DYN] = {
[BLK_IDX_SCHEDULE] = {0},
[BLK_IDX_SCHEDULE_ENTRY_POINTS] = {0},
+ [BLK_IDX_VL_LOOKUP] = {
+ .entry_packing = sja1105_vl_lookup_entry_packing,
+ .cmd_packing = sja1105_vl_lookup_cmd_packing,
+ .access = (OP_READ | OP_WRITE),
+ .max_entry_count = SJA1105_MAX_VL_LOOKUP_COUNT,
+ .packed_size = SJA1105PQRS_SJA1105_SIZE_VL_LOOKUP_DYN_CMD,
+ .addr = 0x47,
+ },
+ [BLK_IDX_VL_POLICING] = {0},
+ [BLK_IDX_VL_FORWARDING] = {0},
[BLK_IDX_L2_LOOKUP] = {
.entry_packing = sja1105pqrs_dyn_l2_lookup_entry_packing,
.cmd_packing = sja1105pqrs_l2_lookup_cmd_packing,
@@ -616,6 +666,7 @@ struct sja1105_dynamic_table_ops sja1105pqrs_dyn_ops[BLK_IDX_MAX_DYN] = {
},
[BLK_IDX_SCHEDULE_PARAMS] = {0},
[BLK_IDX_SCHEDULE_ENTRY_POINTS_PARAMS] = {0},
+ [BLK_IDX_VL_FORWARDING_PARAMS] = {0},
[BLK_IDX_L2_LOOKUP_PARAMS] = {
.entry_packing = sja1105et_l2_lookup_params_entry_packing,
.cmd_packing = sja1105et_l2_lookup_params_cmd_packing,
diff --git a/drivers/net/dsa/sja1105/sja1105_static_config.c b/drivers/net/dsa/sja1105/sja1105_static_config.c
index bbfe034910a0..b68c9c92c248 100644
--- a/drivers/net/dsa/sja1105/sja1105_static_config.c
+++ b/drivers/net/dsa/sja1105/sja1105_static_config.c
@@ -432,6 +432,84 @@ static size_t sja1105_schedule_entry_packing(void *buf, void *entry_ptr,
return size;
}
+static size_t
+sja1105_vl_forwarding_params_entry_packing(void *buf, void *entry_ptr,
+ enum packing_op op)
+{
+ struct sja1105_vl_forwarding_params_entry *entry = entry_ptr;
+ const size_t size = SJA1105_SIZE_VL_FORWARDING_PARAMS_ENTRY;
+ int offset, i;
+
+ for (i = 0, offset = 16; i < 8; i++, offset += 10)
+ sja1105_packing(buf, &entry->partspc[i],
+ offset + 9, offset + 0, size, op);
+ sja1105_packing(buf, &entry->debugen, 15, 15, size, op);
+ return size;
+}
+
+static size_t sja1105_vl_forwarding_entry_packing(void *buf, void *entry_ptr,
+ enum packing_op op)
+{
+ struct sja1105_vl_forwarding_entry *entry = entry_ptr;
+ const size_t size = SJA1105_SIZE_VL_FORWARDING_ENTRY;
+
+ sja1105_packing(buf, &entry->type, 31, 31, size, op);
+ sja1105_packing(buf, &entry->priority, 30, 28, size, op);
+ sja1105_packing(buf, &entry->partition, 27, 25, size, op);
+ sja1105_packing(buf, &entry->destports, 24, 20, size, op);
+ return size;
+}
+
+size_t sja1105_vl_lookup_entry_packing(void *buf, void *entry_ptr,
+ enum packing_op op)
+{
+ struct sja1105_vl_lookup_entry *entry = entry_ptr;
+ const size_t size = SJA1105_SIZE_VL_LOOKUP_ENTRY;
+
+ if (entry->format == SJA1105_VL_FORMAT_PSFP) {
+ /* Interpreting vllupformat as 0 */
+ sja1105_packing(buf, &entry->destports,
+ 95, 91, size, op);
+ sja1105_packing(buf, &entry->iscritical,
+ 90, 90, size, op);
+ sja1105_packing(buf, &entry->macaddr,
+ 89, 42, size, op);
+ sja1105_packing(buf, &entry->vlanid,
+ 41, 30, size, op);
+ sja1105_packing(buf, &entry->port,
+ 29, 27, size, op);
+ sja1105_packing(buf, &entry->vlanprior,
+ 26, 24, size, op);
+ } else {
+ /* Interpreting vllupformat as 1 */
+ sja1105_packing(buf, &entry->egrmirr,
+ 95, 91, size, op);
+ sja1105_packing(buf, &entry->ingrmirr,
+ 90, 90, size, op);
+ sja1105_packing(buf, &entry->vlid,
+ 57, 42, size, op);
+ sja1105_packing(buf, &entry->port,
+ 29, 27, size, op);
+ }
+ return size;
+}
+
+static size_t sja1105_vl_policing_entry_packing(void *buf, void *entry_ptr,
+ enum packing_op op)
+{
+ struct sja1105_vl_policing_entry *entry = entry_ptr;
+ const size_t size = SJA1105_SIZE_VL_POLICING_ENTRY;
+
+ sja1105_packing(buf, &entry->type, 63, 63, size, op);
+ sja1105_packing(buf, &entry->maxlen, 62, 52, size, op);
+ sja1105_packing(buf, &entry->sharindx, 51, 42, size, op);
+ if (entry->type == 0) {
+ sja1105_packing(buf, &entry->bag, 41, 28, size, op);
+ sja1105_packing(buf, &entry->jitter, 27, 18, size, op);
+ }
+ return size;
+}
+
size_t sja1105_vlan_lookup_entry_packing(void *buf, void *entry_ptr,
enum packing_op op)
{
@@ -510,6 +588,9 @@ static void sja1105_table_write_crc(u8 *table_start, u8 *crc_ptr)
static u64 blk_id_map[BLK_IDX_MAX] = {
[BLK_IDX_SCHEDULE] = BLKID_SCHEDULE,
[BLK_IDX_SCHEDULE_ENTRY_POINTS] = BLKID_SCHEDULE_ENTRY_POINTS,
+ [BLK_IDX_VL_LOOKUP] = BLKID_VL_LOOKUP,
+ [BLK_IDX_VL_POLICING] = BLKID_VL_POLICING,
+ [BLK_IDX_VL_FORWARDING] = BLKID_VL_FORWARDING,
[BLK_IDX_L2_LOOKUP] = BLKID_L2_LOOKUP,
[BLK_IDX_L2_POLICING] = BLKID_L2_POLICING,
[BLK_IDX_VLAN_LOOKUP] = BLKID_VLAN_LOOKUP,
@@ -517,6 +598,7 @@ static u64 blk_id_map[BLK_IDX_MAX] = {
[BLK_IDX_MAC_CONFIG] = BLKID_MAC_CONFIG,
[BLK_IDX_SCHEDULE_PARAMS] = BLKID_SCHEDULE_PARAMS,
[BLK_IDX_SCHEDULE_ENTRY_POINTS_PARAMS] = BLKID_SCHEDULE_ENTRY_POINTS_PARAMS,
+ [BLK_IDX_VL_FORWARDING_PARAMS] = BLKID_VL_FORWARDING_PARAMS,
[BLK_IDX_L2_LOOKUP_PARAMS] = BLKID_L2_LOOKUP_PARAMS,
[BLK_IDX_L2_FORWARDING_PARAMS] = BLKID_L2_FORWARDING_PARAMS,
[BLK_IDX_AVB_PARAMS] = BLKID_AVB_PARAMS,
@@ -533,6 +615,9 @@ const char *sja1105_static_config_error_msg[] = {
"schedule-table present, but one of "
"schedule-entry-points-table, schedule-parameters-table or "
"schedule-entry-points-parameters table is empty",
+ [SJA1105_INCORRECT_VIRTUAL_LINK_CONFIGURATION] =
+ "vl-lookup-table present, but one of vl-policing-table, "
+ "vl-forwarding-table or vl-forwarding-parameters-table is empty",
[SJA1105_MISSING_L2_POLICING_TABLE] =
"l2-policing-table needs to have at least one entry",
[SJA1105_MISSING_L2_FORWARDING_TABLE] =
@@ -560,6 +645,7 @@ static sja1105_config_valid_t
static_config_check_memory_size(const struct sja1105_table *tables)
{
const struct sja1105_l2_forwarding_params_entry *l2_fwd_params;
+ const struct sja1105_vl_forwarding_params_entry *vl_fwd_params;
int i, mem = 0;
l2_fwd_params = tables[BLK_IDX_L2_FORWARDING_PARAMS].entries;
@@ -567,6 +653,12 @@ static_config_check_memory_size(const struct sja1105_table *tables)
for (i = 0; i < 8; i++)
mem += l2_fwd_params->part_spc[i];
+ if (tables[BLK_IDX_VL_FORWARDING_PARAMS].entry_count) {
+ vl_fwd_params = tables[BLK_IDX_VL_FORWARDING_PARAMS].entries;
+ for (i = 0; i < 8; i++)
+ mem += vl_fwd_params->partspc[i];
+ }
+
if (mem > SJA1105_MAX_FRAME_MEMORY)
return SJA1105_OVERCOMMITTED_FRAME_MEMORY;
@@ -594,6 +686,32 @@ sja1105_static_config_check_valid(const struct sja1105_static_config *config)
if (!IS_FULL(BLK_IDX_SCHEDULE_ENTRY_POINTS_PARAMS))
return SJA1105_INCORRECT_TTETHERNET_CONFIGURATION;
}
+ if (tables[BLK_IDX_VL_LOOKUP].entry_count) {
+ struct sja1105_vl_lookup_entry *vl_lookup;
+ bool has_critical_links = false;
+ int i;
+
+ vl_lookup = tables[BLK_IDX_VL_LOOKUP].entries;
+
+ for (i = 0; i < tables[BLK_IDX_VL_LOOKUP].entry_count; i++) {
+ if (vl_lookup[i].iscritical) {
+ has_critical_links = true;
+ break;
+ }
+ }
+
+ if (tables[BLK_IDX_VL_POLICING].entry_count == 0 &&
+ has_critical_links)
+ return SJA1105_INCORRECT_VIRTUAL_LINK_CONFIGURATION;
+
+ if (tables[BLK_IDX_VL_FORWARDING].entry_count == 0 &&
+ has_critical_links)
+ return SJA1105_INCORRECT_VIRTUAL_LINK_CONFIGURATION;
+
+ if (tables[BLK_IDX_VL_FORWARDING_PARAMS].entry_count == 0 &&
+ has_critical_links)
+ return SJA1105_INCORRECT_VIRTUAL_LINK_CONFIGURATION;
+ }
if (tables[BLK_IDX_L2_POLICING].entry_count == 0)
return SJA1105_MISSING_L2_POLICING_TABLE;
@@ -703,6 +821,9 @@ sja1105_static_config_get_length(const struct sja1105_static_config *config)
struct sja1105_table_ops sja1105e_table_ops[BLK_IDX_MAX] = {
[BLK_IDX_SCHEDULE] = {0},
[BLK_IDX_SCHEDULE_ENTRY_POINTS] = {0},
+ [BLK_IDX_VL_LOOKUP] = {0},
+ [BLK_IDX_VL_POLICING] = {0},
+ [BLK_IDX_VL_FORWARDING] = {0},
[BLK_IDX_L2_LOOKUP] = {
.packing = sja1105et_l2_lookup_entry_packing,
.unpacked_entry_size = sizeof(struct sja1105_l2_lookup_entry),
@@ -735,6 +856,7 @@ struct sja1105_table_ops sja1105e_table_ops[BLK_IDX_MAX] = {
},
[BLK_IDX_SCHEDULE_PARAMS] = {0},
[BLK_IDX_SCHEDULE_ENTRY_POINTS_PARAMS] = {0},
+ [BLK_IDX_VL_FORWARDING_PARAMS] = {0},
[BLK_IDX_L2_LOOKUP_PARAMS] = {
.packing = sja1105et_l2_lookup_params_entry_packing,
.unpacked_entry_size = sizeof(struct sja1105_l2_lookup_params_entry),
@@ -781,6 +903,24 @@ struct sja1105_table_ops sja1105t_table_ops[BLK_IDX_MAX] = {
.packed_entry_size = SJA1105_SIZE_SCHEDULE_ENTRY_POINTS_ENTRY,
.max_entry_count = SJA1105_MAX_SCHEDULE_ENTRY_POINTS_COUNT,
},
+ [BLK_IDX_VL_LOOKUP] = {
+ .packing = sja1105_vl_lookup_entry_packing,
+ .unpacked_entry_size = sizeof(struct sja1105_vl_lookup_entry),
+ .packed_entry_size = SJA1105_SIZE_VL_LOOKUP_ENTRY,
+ .max_entry_count = SJA1105_MAX_VL_LOOKUP_COUNT,
+ },
+ [BLK_IDX_VL_POLICING] = {
+ .packing = sja1105_vl_policing_entry_packing,
+ .unpacked_entry_size = sizeof(struct sja1105_vl_policing_entry),
+ .packed_entry_size = SJA1105_SIZE_VL_POLICING_ENTRY,
+ .max_entry_count = SJA1105_MAX_VL_POLICING_COUNT,
+ },
+ [BLK_IDX_VL_FORWARDING] = {
+ .packing = sja1105_vl_forwarding_entry_packing,
+ .unpacked_entry_size = sizeof(struct sja1105_vl_forwarding_entry),
+ .packed_entry_size = SJA1105_SIZE_VL_FORWARDING_ENTRY,
+ .max_entry_count = SJA1105_MAX_VL_FORWARDING_COUNT,
+ },
[BLK_IDX_L2_LOOKUP] = {
.packing = sja1105et_l2_lookup_entry_packing,
.unpacked_entry_size = sizeof(struct sja1105_l2_lookup_entry),
@@ -823,6 +963,12 @@ struct sja1105_table_ops sja1105t_table_ops[BLK_IDX_MAX] = {
.packed_entry_size = SJA1105_SIZE_SCHEDULE_ENTRY_POINTS_PARAMS_ENTRY,
.max_entry_count = SJA1105_MAX_SCHEDULE_ENTRY_POINTS_PARAMS_COUNT,
},
+ [BLK_IDX_VL_FORWARDING_PARAMS] = {
+ .packing = sja1105_vl_forwarding_params_entry_packing,
+ .unpacked_entry_size = sizeof(struct sja1105_vl_forwarding_params_entry),
+ .packed_entry_size = SJA1105_SIZE_VL_FORWARDING_PARAMS_ENTRY,
+ .max_entry_count = SJA1105_MAX_VL_FORWARDING_PARAMS_COUNT,
+ },
[BLK_IDX_L2_LOOKUP_PARAMS] = {
.packing = sja1105et_l2_lookup_params_entry_packing,
.unpacked_entry_size = sizeof(struct sja1105_l2_lookup_params_entry),
@@ -859,6 +1005,9 @@ struct sja1105_table_ops sja1105t_table_ops[BLK_IDX_MAX] = {
struct sja1105_table_ops sja1105p_table_ops[BLK_IDX_MAX] = {
[BLK_IDX_SCHEDULE] = {0},
[BLK_IDX_SCHEDULE_ENTRY_POINTS] = {0},
+ [BLK_IDX_VL_LOOKUP] = {0},
+ [BLK_IDX_VL_POLICING] = {0},
+ [BLK_IDX_VL_FORWARDING] = {0},
[BLK_IDX_L2_LOOKUP] = {
.packing = sja1105pqrs_l2_lookup_entry_packing,
.unpacked_entry_size = sizeof(struct sja1105_l2_lookup_entry),
@@ -891,6 +1040,7 @@ struct sja1105_table_ops sja1105p_table_ops[BLK_IDX_MAX] = {
},
[BLK_IDX_SCHEDULE_PARAMS] = {0},
[BLK_IDX_SCHEDULE_ENTRY_POINTS_PARAMS] = {0},
+ [BLK_IDX_VL_FORWARDING_PARAMS] = {0},
[BLK_IDX_L2_LOOKUP_PARAMS] = {
.packing = sja1105pqrs_l2_lookup_params_entry_packing,
.unpacked_entry_size = sizeof(struct sja1105_l2_lookup_params_entry),
@@ -937,6 +1087,24 @@ struct sja1105_table_ops sja1105q_table_ops[BLK_IDX_MAX] = {
.packed_entry_size = SJA1105_SIZE_SCHEDULE_ENTRY_POINTS_ENTRY,
.max_entry_count = SJA1105_MAX_SCHEDULE_ENTRY_POINTS_COUNT,
},
+ [BLK_IDX_VL_LOOKUP] = {
+ .packing = sja1105_vl_lookup_entry_packing,
+ .unpacked_entry_size = sizeof(struct sja1105_vl_lookup_entry),
+ .packed_entry_size = SJA1105_SIZE_VL_LOOKUP_ENTRY,
+ .max_entry_count = SJA1105_MAX_VL_LOOKUP_COUNT,
+ },
+ [BLK_IDX_VL_POLICING] = {
+ .packing = sja1105_vl_policing_entry_packing,
+ .unpacked_entry_size = sizeof(struct sja1105_vl_policing_entry),
+ .packed_entry_size = SJA1105_SIZE_VL_POLICING_ENTRY,
+ .max_entry_count = SJA1105_MAX_VL_POLICING_COUNT,
+ },
+ [BLK_IDX_VL_FORWARDING] = {
+ .packing = sja1105_vl_forwarding_entry_packing,
+ .unpacked_entry_size = sizeof(struct sja1105_vl_forwarding_entry),
+ .packed_entry_size = SJA1105_SIZE_VL_FORWARDING_ENTRY,
+ .max_entry_count = SJA1105_MAX_VL_FORWARDING_COUNT,
+ },
[BLK_IDX_L2_LOOKUP] = {
.packing = sja1105pqrs_l2_lookup_entry_packing,
.unpacked_entry_size = sizeof(struct sja1105_l2_lookup_entry),
@@ -979,6 +1147,12 @@ struct sja1105_table_ops sja1105q_table_ops[BLK_IDX_MAX] = {
.packed_entry_size = SJA1105_SIZE_SCHEDULE_ENTRY_POINTS_PARAMS_ENTRY,
.max_entry_count = SJA1105_MAX_SCHEDULE_ENTRY_POINTS_PARAMS_COUNT,
},
+ [BLK_IDX_VL_FORWARDING_PARAMS] = {
+ .packing = sja1105_vl_forwarding_params_entry_packing,
+ .unpacked_entry_size = sizeof(struct sja1105_vl_forwarding_params_entry),
+ .packed_entry_size = SJA1105_SIZE_VL_FORWARDING_PARAMS_ENTRY,
+ .max_entry_count = SJA1105_MAX_VL_FORWARDING_PARAMS_COUNT,
+ },
[BLK_IDX_L2_LOOKUP_PARAMS] = {
.packing = sja1105pqrs_l2_lookup_params_entry_packing,
.unpacked_entry_size = sizeof(struct sja1105_l2_lookup_params_entry),
@@ -1015,6 +1189,9 @@ struct sja1105_table_ops sja1105q_table_ops[BLK_IDX_MAX] = {
struct sja1105_table_ops sja1105r_table_ops[BLK_IDX_MAX] = {
[BLK_IDX_SCHEDULE] = {0},
[BLK_IDX_SCHEDULE_ENTRY_POINTS] = {0},
+ [BLK_IDX_VL_LOOKUP] = {0},
+ [BLK_IDX_VL_POLICING] = {0},
+ [BLK_IDX_VL_FORWARDING] = {0},
[BLK_IDX_L2_LOOKUP] = {
.packing = sja1105pqrs_l2_lookup_entry_packing,
.unpacked_entry_size = sizeof(struct sja1105_l2_lookup_entry),
@@ -1047,6 +1224,7 @@ struct sja1105_table_ops sja1105r_table_ops[BLK_IDX_MAX] = {
},
[BLK_IDX_SCHEDULE_PARAMS] = {0},
[BLK_IDX_SCHEDULE_ENTRY_POINTS_PARAMS] = {0},
+ [BLK_IDX_VL_FORWARDING_PARAMS] = {0},
[BLK_IDX_L2_LOOKUP_PARAMS] = {
.packing = sja1105pqrs_l2_lookup_params_entry_packing,
.unpacked_entry_size = sizeof(struct sja1105_l2_lookup_params_entry),
@@ -1093,6 +1271,24 @@ struct sja1105_table_ops sja1105s_table_ops[BLK_IDX_MAX] = {
.packed_entry_size = SJA1105_SIZE_SCHEDULE_ENTRY_POINTS_ENTRY,
.max_entry_count = SJA1105_MAX_SCHEDULE_ENTRY_POINTS_COUNT,
},
+ [BLK_IDX_VL_LOOKUP] = {
+ .packing = sja1105_vl_lookup_entry_packing,
+ .unpacked_entry_size = sizeof(struct sja1105_vl_lookup_entry),
+ .packed_entry_size = SJA1105_SIZE_VL_LOOKUP_ENTRY,
+ .max_entry_count = SJA1105_MAX_VL_LOOKUP_COUNT,
+ },
+ [BLK_IDX_VL_POLICING] = {
+ .packing = sja1105_vl_policing_entry_packing,
+ .unpacked_entry_size = sizeof(struct sja1105_vl_policing_entry),
+ .packed_entry_size = SJA1105_SIZE_VL_POLICING_ENTRY,
+ .max_entry_count = SJA1105_MAX_VL_POLICING_COUNT,
+ },
+ [BLK_IDX_VL_FORWARDING] = {
+ .packing = sja1105_vl_forwarding_entry_packing,
+ .unpacked_entry_size = sizeof(struct sja1105_vl_forwarding_entry),
+ .packed_entry_size = SJA1105_SIZE_VL_FORWARDING_ENTRY,
+ .max_entry_count = SJA1105_MAX_VL_FORWARDING_COUNT,
+ },
[BLK_IDX_L2_LOOKUP] = {
.packing = sja1105pqrs_l2_lookup_entry_packing,
.unpacked_entry_size = sizeof(struct sja1105_l2_lookup_entry),
@@ -1135,6 +1331,12 @@ struct sja1105_table_ops sja1105s_table_ops[BLK_IDX_MAX] = {
.packed_entry_size = SJA1105_SIZE_SCHEDULE_ENTRY_POINTS_PARAMS_ENTRY,
.max_entry_count = SJA1105_MAX_SCHEDULE_ENTRY_POINTS_PARAMS_COUNT,
},
+ [BLK_IDX_VL_FORWARDING_PARAMS] = {
+ .packing = sja1105_vl_forwarding_params_entry_packing,
+ .unpacked_entry_size = sizeof(struct sja1105_vl_forwarding_params_entry),
+ .packed_entry_size = SJA1105_SIZE_VL_FORWARDING_PARAMS_ENTRY,
+ .max_entry_count = SJA1105_MAX_VL_FORWARDING_PARAMS_COUNT,
+ },
[BLK_IDX_L2_LOOKUP_PARAMS] = {
.packing = sja1105pqrs_l2_lookup_params_entry_packing,
.unpacked_entry_size = sizeof(struct sja1105_l2_lookup_params_entry),
diff --git a/drivers/net/dsa/sja1105/sja1105_static_config.h b/drivers/net/dsa/sja1105/sja1105_static_config.h
index 8afafb6aef12..1a8fcbbb57b6 100644
--- a/drivers/net/dsa/sja1105/sja1105_static_config.h
+++ b/drivers/net/dsa/sja1105/sja1105_static_config.h
@@ -13,6 +13,9 @@
#define SJA1105_SIZE_TABLE_HEADER 12
#define SJA1105_SIZE_SCHEDULE_ENTRY 8
#define SJA1105_SIZE_SCHEDULE_ENTRY_POINTS_ENTRY 4
+#define SJA1105_SIZE_VL_LOOKUP_ENTRY 12
+#define SJA1105_SIZE_VL_POLICING_ENTRY 8
+#define SJA1105_SIZE_VL_FORWARDING_ENTRY 4
#define SJA1105_SIZE_L2_POLICING_ENTRY 8
#define SJA1105_SIZE_VLAN_LOOKUP_ENTRY 8
#define SJA1105_SIZE_L2_FORWARDING_ENTRY 8
@@ -20,6 +23,7 @@
#define SJA1105_SIZE_XMII_PARAMS_ENTRY 4
#define SJA1105_SIZE_SCHEDULE_PARAMS_ENTRY 12
#define SJA1105_SIZE_SCHEDULE_ENTRY_POINTS_PARAMS_ENTRY 4
+#define SJA1105_SIZE_VL_FORWARDING_PARAMS_ENTRY 12
#define SJA1105ET_SIZE_L2_LOOKUP_ENTRY 12
#define SJA1105ET_SIZE_MAC_CONFIG_ENTRY 28
#define SJA1105ET_SIZE_L2_LOOKUP_PARAMS_ENTRY 4
@@ -35,6 +39,9 @@
enum {
BLKID_SCHEDULE = 0x00,
BLKID_SCHEDULE_ENTRY_POINTS = 0x01,
+ BLKID_VL_LOOKUP = 0x02,
+ BLKID_VL_POLICING = 0x03,
+ BLKID_VL_FORWARDING = 0x04,
BLKID_L2_LOOKUP = 0x05,
BLKID_L2_POLICING = 0x06,
BLKID_VLAN_LOOKUP = 0x07,
@@ -42,6 +49,7 @@ enum {
BLKID_MAC_CONFIG = 0x09,
BLKID_SCHEDULE_PARAMS = 0x0A,
BLKID_SCHEDULE_ENTRY_POINTS_PARAMS = 0x0B,
+ BLKID_VL_FORWARDING_PARAMS = 0x0C,
BLKID_L2_LOOKUP_PARAMS = 0x0D,
BLKID_L2_FORWARDING_PARAMS = 0x0E,
BLKID_AVB_PARAMS = 0x10,
@@ -52,6 +60,9 @@ enum {
enum sja1105_blk_idx {
BLK_IDX_SCHEDULE = 0,
BLK_IDX_SCHEDULE_ENTRY_POINTS,
+ BLK_IDX_VL_LOOKUP,
+ BLK_IDX_VL_POLICING,
+ BLK_IDX_VL_FORWARDING,
BLK_IDX_L2_LOOKUP,
BLK_IDX_L2_POLICING,
BLK_IDX_VLAN_LOOKUP,
@@ -59,6 +70,7 @@ enum sja1105_blk_idx {
BLK_IDX_MAC_CONFIG,
BLK_IDX_SCHEDULE_PARAMS,
BLK_IDX_SCHEDULE_ENTRY_POINTS_PARAMS,
+ BLK_IDX_VL_FORWARDING_PARAMS,
BLK_IDX_L2_LOOKUP_PARAMS,
BLK_IDX_L2_FORWARDING_PARAMS,
BLK_IDX_AVB_PARAMS,
@@ -73,6 +85,9 @@ enum sja1105_blk_idx {
#define SJA1105_MAX_SCHEDULE_COUNT 1024
#define SJA1105_MAX_SCHEDULE_ENTRY_POINTS_COUNT 2048
+#define SJA1105_MAX_VL_LOOKUP_COUNT 1024
+#define SJA1105_MAX_VL_POLICING_COUNT 1024
+#define SJA1105_MAX_VL_FORWARDING_COUNT 1024
#define SJA1105_MAX_L2_LOOKUP_COUNT 1024
#define SJA1105_MAX_L2_POLICING_COUNT 45
#define SJA1105_MAX_VLAN_LOOKUP_COUNT 4096
@@ -80,6 +95,7 @@ enum sja1105_blk_idx {
#define SJA1105_MAX_MAC_CONFIG_COUNT 5
#define SJA1105_MAX_SCHEDULE_PARAMS_COUNT 1
#define SJA1105_MAX_SCHEDULE_ENTRY_POINTS_PARAMS_COUNT 1
+#define SJA1105_MAX_VL_FORWARDING_PARAMS_COUNT 1
#define SJA1105_MAX_L2_LOOKUP_PARAMS_COUNT 1
#define SJA1105_MAX_L2_FORWARDING_PARAMS_COUNT 1
#define SJA1105_MAX_GENERAL_PARAMS_COUNT 1
@@ -262,6 +278,52 @@ struct sja1105_xmii_params_entry {
u64 xmii_mode[5];
};
+enum {
+ SJA1105_VL_FORMAT_PSFP = 0,
+ SJA1105_VL_FORMAT_ARINC664 = 1,
+};
+
+struct sja1105_vl_lookup_entry {
+ u64 format;
+ u64 port;
+ union {
+ /* SJA1105_VL_FORMAT_PSFP */
+ struct {
+ u64 destports;
+ u64 iscritical;
+ u64 macaddr;
+ u64 vlanid;
+ u64 vlanprior;
+ };
+ /* SJA1105_VL_FORMAT_ARINC664 */
+ struct {
+ u64 egrmirr;
+ u64 ingrmirr;
+ u64 vlid;
+ };
+ };
+};
+
+struct sja1105_vl_policing_entry {
+ u64 type;
+ u64 maxlen;
+ u64 sharindx;
+ u64 bag;
+ u64 jitter;
+};
+
+struct sja1105_vl_forwarding_entry {
+ u64 type;
+ u64 priority;
+ u64 partition;
+ u64 destports;
+};
+
+struct sja1105_vl_forwarding_params_entry {
+ u64 partspc[8];
+ u64 debugen;
+};
+
struct sja1105_table_header {
u64 block_id;
u64 len;
@@ -303,6 +365,7 @@ typedef enum {
SJA1105_CONFIG_OK = 0,
SJA1105_TTETHERNET_NOT_SUPPORTED,
SJA1105_INCORRECT_TTETHERNET_CONFIGURATION,
+ SJA1105_INCORRECT_VIRTUAL_LINK_CONFIGURATION,
SJA1105_MISSING_L2_POLICING_TABLE,
SJA1105_MISSING_L2_FORWARDING_TABLE,
SJA1105_MISSING_L2_FORWARDING_PARAMS_TABLE,
--
2.17.1
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH v3 net-next 3/6] net: dsa: sja1105: make room for virtual link parsing in flower offload
2020-05-05 19:20 [PATCH v3 net-next 0/6] tc-gate offload for SJA1105 DSA switch Vladimir Oltean
2020-05-05 19:20 ` [PATCH v3 net-next 1/6] net: dsa: introduce a dsa_port_from_netdev public helper Vladimir Oltean
2020-05-05 19:20 ` [PATCH v3 net-next 2/6] net: dsa: sja1105: add static tables for virtual links Vladimir Oltean
@ 2020-05-05 19:20 ` Vladimir Oltean
2020-05-05 19:20 ` [PATCH v3 net-next 4/6] net: dsa: sja1105: support flow-based redirection via virtual links Vladimir Oltean
` (3 subsequent siblings)
6 siblings, 0 replies; 9+ messages in thread
From: Vladimir Oltean @ 2020-05-05 19:20 UTC (permalink / raw)
To: netdev; +Cc: andrew, f.fainelli, vivien.didelot, vinicius.gomes, po.liu
From: Vladimir Oltean <vladimir.oltean@nxp.com>
Virtual links are a sja1105 hardware concept of executing various flow
actions based on a key extracted from the frame's DMAC, VID and PCP.
Currently the tc-flower offload code supports only parsing the DMAC if
that is the broadcast MAC address, and the VLAN PCP. Extract the key
parsing logic from the L2 policers functionality and move it into its
own function, after adding extra logic for matching on any DMAC and VID.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
Changes from v2:
None.
Changes from v1:
None.
Changes from RFC:
None.
drivers/net/dsa/sja1105/sja1105.h | 28 +++++-
drivers/net/dsa/sja1105/sja1105_flower.c | 111 +++++++++++++++++------
2 files changed, 112 insertions(+), 27 deletions(-)
diff --git a/drivers/net/dsa/sja1105/sja1105.h b/drivers/net/dsa/sja1105/sja1105.h
index 602aa30c832f..95633ad9bfb7 100644
--- a/drivers/net/dsa/sja1105/sja1105.h
+++ b/drivers/net/dsa/sja1105/sja1105.h
@@ -97,6 +97,32 @@ struct sja1105_info {
const char *name;
};
+enum sja1105_key_type {
+ SJA1105_KEY_BCAST,
+ SJA1105_KEY_TC,
+ SJA1105_KEY_VLAN_UNAWARE_VL,
+ SJA1105_KEY_VLAN_AWARE_VL,
+};
+
+struct sja1105_key {
+ enum sja1105_key_type type;
+
+ union {
+ /* SJA1105_KEY_TC */
+ struct {
+ int pcp;
+ } tc;
+
+ /* SJA1105_KEY_VLAN_UNAWARE_VL */
+ /* SJA1105_KEY_VLAN_AWARE_VL */
+ struct {
+ u64 dmac;
+ u16 vid;
+ u16 pcp;
+ } vl;
+ };
+};
+
enum sja1105_rule_type {
SJA1105_RULE_BCAST_POLICER,
SJA1105_RULE_TC_POLICER,
@@ -106,6 +132,7 @@ struct sja1105_rule {
struct list_head list;
unsigned long cookie;
unsigned long port_mask;
+ struct sja1105_key key;
enum sja1105_rule_type type;
union {
@@ -117,7 +144,6 @@ struct sja1105_rule {
/* SJA1105_RULE_TC_POLICER */
struct {
int sharindx;
- int tc;
} tc_pol;
};
};
diff --git a/drivers/net/dsa/sja1105/sja1105_flower.c b/drivers/net/dsa/sja1105/sja1105_flower.c
index 5288a722e625..3246d5a49436 100644
--- a/drivers/net/dsa/sja1105/sja1105_flower.c
+++ b/drivers/net/dsa/sja1105/sja1105_flower.c
@@ -46,6 +46,7 @@ static int sja1105_setup_bcast_policer(struct sja1105_private *priv,
rule->cookie = cookie;
rule->type = SJA1105_RULE_BCAST_POLICER;
rule->bcast_pol.sharindx = sja1105_find_free_l2_policer(priv);
+ rule->key.type = SJA1105_KEY_BCAST;
new_rule = true;
}
@@ -117,7 +118,8 @@ static int sja1105_setup_tc_policer(struct sja1105_private *priv,
rule->cookie = cookie;
rule->type = SJA1105_RULE_TC_POLICER;
rule->tc_pol.sharindx = sja1105_find_free_l2_policer(priv);
- rule->tc_pol.tc = tc;
+ rule->key.type = SJA1105_KEY_TC;
+ rule->key.tc.pcp = tc;
new_rule = true;
}
@@ -169,14 +171,37 @@ static int sja1105_setup_tc_policer(struct sja1105_private *priv,
return rc;
}
-static int sja1105_flower_parse_policer(struct sja1105_private *priv, int port,
- struct netlink_ext_ack *extack,
- struct flow_cls_offload *cls,
- u64 rate_bytes_per_sec,
- s64 burst)
+static int sja1105_flower_policer(struct sja1105_private *priv, int port,
+ struct netlink_ext_ack *extack,
+ unsigned long cookie, struct sja1105_key *key,
+ u64 rate_bytes_per_sec,
+ s64 burst)
+{
+ switch (key->type) {
+ case SJA1105_KEY_BCAST:
+ return sja1105_setup_bcast_policer(priv, extack, cookie, port,
+ rate_bytes_per_sec, burst);
+ case SJA1105_KEY_TC:
+ return sja1105_setup_tc_policer(priv, extack, cookie, port,
+ key->tc.pcp, rate_bytes_per_sec,
+ burst);
+ default:
+ NL_SET_ERR_MSG_MOD(extack, "Unknown keys for policing");
+ return -EOPNOTSUPP;
+ }
+}
+
+static int sja1105_flower_parse_key(struct sja1105_private *priv,
+ struct netlink_ext_ack *extack,
+ struct flow_cls_offload *cls,
+ struct sja1105_key *key)
{
struct flow_rule *rule = flow_cls_offload_flow_rule(cls);
struct flow_dissector *dissector = rule->match.dissector;
+ bool is_bcast_dmac = false;
+ u64 dmac = U64_MAX;
+ u16 vid = U16_MAX;
+ u16 pcp = U16_MAX;
if (dissector->used_keys &
~(BIT(FLOW_DISSECTOR_KEY_BASIC) |
@@ -213,16 +238,14 @@ static int sja1105_flower_parse_policer(struct sja1105_private *priv, int port,
return -EOPNOTSUPP;
}
- if (!ether_addr_equal_masked(match.key->dst, bcast,
- match.mask->dst)) {
+ if (!ether_addr_equal(match.mask->dst, bcast)) {
NL_SET_ERR_MSG_MOD(extack,
- "Only matching on broadcast DMAC is supported");
+ "Masked matching on MAC not supported");
return -EOPNOTSUPP;
}
- return sja1105_setup_bcast_policer(priv, extack, cls->cookie,
- port, rate_bytes_per_sec,
- burst);
+ dmac = ether_addr_to_u64(match.key->dst);
+ is_bcast_dmac = ether_addr_equal(match.key->dst, bcast);
}
if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_VLAN)) {
@@ -230,22 +253,46 @@ static int sja1105_flower_parse_policer(struct sja1105_private *priv, int port,
flow_rule_match_vlan(rule, &match);
- if (match.key->vlan_id & match.mask->vlan_id) {
+ if (match.mask->vlan_id &&
+ match.mask->vlan_id != VLAN_VID_MASK) {
NL_SET_ERR_MSG_MOD(extack,
- "Matching on VID is not supported");
+ "Masked matching on VID is not supported");
return -EOPNOTSUPP;
}
- if (match.mask->vlan_priority != 0x7) {
+ if (match.mask->vlan_priority &&
+ match.mask->vlan_priority != 0x7) {
NL_SET_ERR_MSG_MOD(extack,
"Masked matching on PCP is not supported");
return -EOPNOTSUPP;
}
- return sja1105_setup_tc_policer(priv, extack, cls->cookie, port,
- match.key->vlan_priority,
- rate_bytes_per_sec,
- burst);
+ if (match.mask->vlan_id)
+ vid = match.key->vlan_id;
+ if (match.mask->vlan_priority)
+ pcp = match.key->vlan_priority;
+ }
+
+ if (is_bcast_dmac && vid == U16_MAX && pcp == U16_MAX) {
+ key->type = SJA1105_KEY_BCAST;
+ return 0;
+ }
+ if (dmac == U64_MAX && vid == U16_MAX && pcp != U16_MAX) {
+ key->type = SJA1105_KEY_TC;
+ key->tc.pcp = pcp;
+ return 0;
+ }
+ if (dmac != U64_MAX && vid != U16_MAX && pcp != U16_MAX) {
+ key->type = SJA1105_KEY_VLAN_AWARE_VL;
+ key->vl.dmac = dmac;
+ key->vl.vid = vid;
+ key->vl.pcp = pcp;
+ return 0;
+ }
+ if (dmac != U64_MAX) {
+ key->type = SJA1105_KEY_VLAN_UNAWARE_VL;
+ key->vl.dmac = dmac;
+ return 0;
}
NL_SET_ERR_MSG_MOD(extack, "Not matching on any known key");
@@ -259,22 +306,34 @@ int sja1105_cls_flower_add(struct dsa_switch *ds, int port,
struct netlink_ext_ack *extack = cls->common.extack;
struct sja1105_private *priv = ds->priv;
const struct flow_action_entry *act;
- int rc = -EOPNOTSUPP, i;
+ unsigned long cookie = cls->cookie;
+ struct sja1105_key key;
+ int rc, i;
+
+ rc = sja1105_flower_parse_key(priv, extack, cls, &key);
+ if (rc)
+ return rc;
+
+ rc = -EOPNOTSUPP;
flow_action_for_each(i, act, &rule->action) {
switch (act->id) {
case FLOW_ACTION_POLICE:
- rc = sja1105_flower_parse_policer(priv, port, extack, cls,
- act->police.rate_bytes_ps,
- act->police.burst);
+ rc = sja1105_flower_policer(priv, port,
+ extack, cookie, &key,
+ act->police.rate_bytes_ps,
+ act->police.burst);
+ if (rc)
+ goto out;
break;
default:
NL_SET_ERR_MSG_MOD(extack,
"Action not supported");
- break;
+ rc = -EOPNOTSUPP;
+ goto out;
}
}
-
+out:
return rc;
}
@@ -297,7 +356,7 @@ int sja1105_cls_flower_del(struct dsa_switch *ds, int port,
old_sharindx = policing[bcast].sharindx;
policing[bcast].sharindx = port;
} else if (rule->type == SJA1105_RULE_TC_POLICER) {
- int index = (port * SJA1105_NUM_TC) + rule->tc_pol.tc;
+ int index = (port * SJA1105_NUM_TC) + rule->key.tc.pcp;
old_sharindx = policing[index].sharindx;
policing[index].sharindx = port;
--
2.17.1
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH v3 net-next 4/6] net: dsa: sja1105: support flow-based redirection via virtual links
2020-05-05 19:20 [PATCH v3 net-next 0/6] tc-gate offload for SJA1105 DSA switch Vladimir Oltean
` (2 preceding siblings ...)
2020-05-05 19:20 ` [PATCH v3 net-next 3/6] net: dsa: sja1105: make room for virtual link parsing in flower offload Vladimir Oltean
@ 2020-05-05 19:20 ` Vladimir Oltean
2020-05-05 19:20 ` [PATCH v3 net-next 5/6] net: dsa: sja1105: implement tc-gate using time-triggered " Vladimir Oltean
` (2 subsequent siblings)
6 siblings, 0 replies; 9+ messages in thread
From: Vladimir Oltean @ 2020-05-05 19:20 UTC (permalink / raw)
To: netdev; +Cc: andrew, f.fainelli, vivien.didelot, vinicius.gomes, po.liu
From: Vladimir Oltean <vladimir.oltean@nxp.com>
Implement tc-flower offloads for redirect, trap and drop using
non-critical virtual links.
Commands which were tested to work are:
# Send frames received on swp2 with a DA of 42:be:24:9b:76:20 to the
# CPU and to swp3. This type of key (DA only) when the port's VLAN
# awareness state is off.
tc qdisc add dev swp2 clsact
tc filter add dev swp2 ingress flower skip_sw dst_mac 42:be:24:9b:76:20 \
action mirred egress redirect dev swp3 \
action trap
# Drop frames received on swp2 with a DA of 42:be:24:9b:76:20, a VID
# of 100 and a PCP of 0.
tc filter add dev swp2 ingress protocol 802.1Q flower skip_sw \
dst_mac 42:be:24:9b:76:20 vlan_id 100 vlan_prio 0 action drop
Under the hood, all rules match on DMAC, VID and PCP, but when VLAN
filtering is disabled, those are set internally by the driver to the
port-based defaults. Because we would be put in an awkward situation if
the user were to change the VLAN filtering state while there are active
rules (packets would no longer match on the specified keys), we simply
deny changing vlan_filtering unless the list of flows offloaded via
virtual links is empty. Then the user can re-add new rules.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
Changes from v2:
None.
Changes from v1:
Using the new dsa_port_from_netdev helper.
Changes from RFC:
None.
drivers/net/dsa/sja1105/Kconfig | 9 +
drivers/net/dsa/sja1105/Makefile | 4 +
drivers/net/dsa/sja1105/sja1105.h | 18 ++
drivers/net/dsa/sja1105/sja1105_flower.c | 57 ++++-
drivers/net/dsa/sja1105/sja1105_main.c | 12 +-
drivers/net/dsa/sja1105/sja1105_vl.c | 302 +++++++++++++++++++++++
drivers/net/dsa/sja1105/sja1105_vl.h | 41 +++
7 files changed, 437 insertions(+), 6 deletions(-)
create mode 100644 drivers/net/dsa/sja1105/sja1105_vl.c
create mode 100644 drivers/net/dsa/sja1105/sja1105_vl.h
diff --git a/drivers/net/dsa/sja1105/Kconfig b/drivers/net/dsa/sja1105/Kconfig
index 0fe1ae173aa1..bc59c3fbab50 100644
--- a/drivers/net/dsa/sja1105/Kconfig
+++ b/drivers/net/dsa/sja1105/Kconfig
@@ -33,3 +33,12 @@ config NET_DSA_SJA1105_TAS
This enables support for the TTEthernet-based egress scheduling
engine in the SJA1105 DSA driver, which is controlled using a
hardware offload of the tc-tqprio qdisc.
+
+config NET_DSA_SJA1105_VL
+ bool "Support for Virtual Links on NXP SJA1105"
+ depends on NET_DSA_SJA1105_TAS
+ help
+ This enables support for flow classification using capable devices
+ (SJA1105T, SJA1105Q, SJA1105S). The following actions are supported:
+ - redirect, trap, drop
+ - time-based ingress policing, via the tc-gate action
diff --git a/drivers/net/dsa/sja1105/Makefile b/drivers/net/dsa/sja1105/Makefile
index 8943d8d66f2b..c88e56a29db8 100644
--- a/drivers/net/dsa/sja1105/Makefile
+++ b/drivers/net/dsa/sja1105/Makefile
@@ -17,3 +17,7 @@ endif
ifdef CONFIG_NET_DSA_SJA1105_TAS
sja1105-objs += sja1105_tas.o
endif
+
+ifdef CONFIG_NET_DSA_SJA1105_VL
+sja1105-objs += sja1105_vl.o
+endif
diff --git a/drivers/net/dsa/sja1105/sja1105.h b/drivers/net/dsa/sja1105/sja1105.h
index 95633ad9bfb7..1756000f6936 100644
--- a/drivers/net/dsa/sja1105/sja1105.h
+++ b/drivers/net/dsa/sja1105/sja1105.h
@@ -126,6 +126,13 @@ struct sja1105_key {
enum sja1105_rule_type {
SJA1105_RULE_BCAST_POLICER,
SJA1105_RULE_TC_POLICER,
+ SJA1105_RULE_VL,
+};
+
+enum sja1105_vl_type {
+ SJA1105_VL_NONCRITICAL,
+ SJA1105_VL_RATE_CONSTRAINED,
+ SJA1105_VL_TIME_TRIGGERED,
};
struct sja1105_rule {
@@ -135,6 +142,7 @@ struct sja1105_rule {
struct sja1105_key key;
enum sja1105_rule_type type;
+ /* Action */
union {
/* SJA1105_RULE_BCAST_POLICER */
struct {
@@ -145,12 +153,19 @@ struct sja1105_rule {
struct {
int sharindx;
} tc_pol;
+
+ /* SJA1105_RULE_VL */
+ struct {
+ unsigned long destports;
+ enum sja1105_vl_type type;
+ } vl;
};
};
struct sja1105_flow_block {
struct list_head rules;
bool l2_policer_used[SJA1105_NUM_L2_POLICERS];
+ int num_virtual_links;
};
struct sja1105_private {
@@ -187,6 +202,7 @@ enum sja1105_reset_reason {
SJA1105_AGEING_TIME,
SJA1105_SCHEDULING,
SJA1105_BEST_EFFORT_POLICING,
+ SJA1105_VIRTUAL_LINKS,
};
int sja1105_static_config_reload(struct sja1105_private *priv,
@@ -290,5 +306,7 @@ int sja1105_cls_flower_add(struct dsa_switch *ds, int port,
struct flow_cls_offload *cls, bool ingress);
void sja1105_flower_setup(struct dsa_switch *ds);
void sja1105_flower_teardown(struct dsa_switch *ds);
+struct sja1105_rule *sja1105_rule_find(struct sja1105_private *priv,
+ unsigned long cookie);
#endif
diff --git a/drivers/net/dsa/sja1105/sja1105_flower.c b/drivers/net/dsa/sja1105/sja1105_flower.c
index 3246d5a49436..5f08eed0b1fc 100644
--- a/drivers/net/dsa/sja1105/sja1105_flower.c
+++ b/drivers/net/dsa/sja1105/sja1105_flower.c
@@ -2,9 +2,10 @@
/* Copyright 2020, NXP Semiconductors
*/
#include "sja1105.h"
+#include "sja1105_vl.h"
-static struct sja1105_rule *sja1105_rule_find(struct sja1105_private *priv,
- unsigned long cookie)
+struct sja1105_rule *sja1105_rule_find(struct sja1105_private *priv,
+ unsigned long cookie)
{
struct sja1105_rule *rule;
@@ -173,7 +174,8 @@ static int sja1105_setup_tc_policer(struct sja1105_private *priv,
static int sja1105_flower_policer(struct sja1105_private *priv, int port,
struct netlink_ext_ack *extack,
- unsigned long cookie, struct sja1105_key *key,
+ unsigned long cookie,
+ struct sja1105_key *key,
u64 rate_bytes_per_sec,
s64 burst)
{
@@ -308,6 +310,7 @@ int sja1105_cls_flower_add(struct dsa_switch *ds, int port,
const struct flow_action_entry *act;
unsigned long cookie = cls->cookie;
struct sja1105_key key;
+ bool vl_rule = false;
int rc, i;
rc = sja1105_flower_parse_key(priv, extack, cls, &key);
@@ -319,13 +322,50 @@ int sja1105_cls_flower_add(struct dsa_switch *ds, int port,
flow_action_for_each(i, act, &rule->action) {
switch (act->id) {
case FLOW_ACTION_POLICE:
- rc = sja1105_flower_policer(priv, port,
- extack, cookie, &key,
+ rc = sja1105_flower_policer(priv, port, extack, cookie,
+ &key,
act->police.rate_bytes_ps,
act->police.burst);
if (rc)
goto out;
break;
+ case FLOW_ACTION_TRAP: {
+ int cpu = dsa_upstream_port(ds, port);
+
+ vl_rule = true;
+
+ rc = sja1105_vl_redirect(priv, port, extack, cookie,
+ &key, BIT(cpu), true);
+ if (rc)
+ goto out;
+ break;
+ }
+ case FLOW_ACTION_REDIRECT: {
+ struct dsa_port *to_dp;
+
+ to_dp = dsa_port_from_netdev(act->dev);
+ if (IS_ERR(to_dp)) {
+ NL_SET_ERR_MSG_MOD(extack,
+ "Destination not a switch port");
+ return -EOPNOTSUPP;
+ }
+
+ vl_rule = true;
+
+ rc = sja1105_vl_redirect(priv, port, extack, cookie,
+ &key, BIT(to_dp->index), true);
+ if (rc)
+ goto out;
+ break;
+ }
+ case FLOW_ACTION_DROP:
+ vl_rule = true;
+
+ rc = sja1105_vl_redirect(priv, port, extack, cookie,
+ &key, 0, false);
+ if (rc)
+ goto out;
+ break;
default:
NL_SET_ERR_MSG_MOD(extack,
"Action not supported");
@@ -333,6 +373,10 @@ int sja1105_cls_flower_add(struct dsa_switch *ds, int port,
goto out;
}
}
+
+ if (vl_rule && !rc)
+ rc = sja1105_static_config_reload(priv, SJA1105_VIRTUAL_LINKS);
+
out:
return rc;
}
@@ -348,6 +392,9 @@ int sja1105_cls_flower_del(struct dsa_switch *ds, int port,
if (!rule)
return 0;
+ if (rule->type == SJA1105_RULE_VL)
+ return sja1105_vl_delete(priv, port, rule, cls->common.extack);
+
policing = priv->static_config.tables[BLK_IDX_L2_POLICING].entries;
if (rule->type == SJA1105_RULE_BCAST_POLICER) {
diff --git a/drivers/net/dsa/sja1105/sja1105_main.c b/drivers/net/dsa/sja1105/sja1105_main.c
index 472f4eb20c49..8bb104ee73d5 100644
--- a/drivers/net/dsa/sja1105/sja1105_main.c
+++ b/drivers/net/dsa/sja1105/sja1105_main.c
@@ -445,7 +445,7 @@ static int sja1105_init_general_params(struct sja1105_private *priv)
*/
.casc_port = SJA1105_NUM_PORTS,
/* No TTEthernet */
- .vllupformat = 0,
+ .vllupformat = SJA1105_VL_FORMAT_PSFP,
.vlmarker = 0,
.vlmask = 0,
/* Only update correctionField for 1-step PTP (L2 transport) */
@@ -1589,6 +1589,7 @@ static const char * const sja1105_reset_reasons[] = {
[SJA1105_AGEING_TIME] = "Ageing time",
[SJA1105_SCHEDULING] = "Time-aware scheduling",
[SJA1105_BEST_EFFORT_POLICING] = "Best-effort policing",
+ [SJA1105_VIRTUAL_LINKS] = "Virtual links",
};
/* For situations where we need to change a setting at runtime that is only
@@ -1831,9 +1832,18 @@ static int sja1105_vlan_filtering(struct dsa_switch *ds, int port, bool enabled)
struct sja1105_general_params_entry *general_params;
struct sja1105_private *priv = ds->priv;
struct sja1105_table *table;
+ struct sja1105_rule *rule;
u16 tpid, tpid2;
int rc;
+ list_for_each_entry(rule, &priv->flow_block.rules, list) {
+ if (rule->type == SJA1105_RULE_VL) {
+ dev_err(ds->dev,
+ "Cannot change VLAN filtering state while VL rules are active\n");
+ return -EBUSY;
+ }
+ }
+
if (enabled) {
/* Enable VLAN filtering. */
tpid = ETH_P_8021Q;
diff --git a/drivers/net/dsa/sja1105/sja1105_vl.c b/drivers/net/dsa/sja1105/sja1105_vl.c
new file mode 100644
index 000000000000..c226779b8275
--- /dev/null
+++ b/drivers/net/dsa/sja1105/sja1105_vl.c
@@ -0,0 +1,302 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright 2020, NXP Semiconductors
+ */
+#include <linux/dsa/8021q.h>
+#include "sja1105.h"
+
+/* The switch flow classification core implements TTEthernet, which 'thinks' in
+ * terms of Virtual Links (VL), a concept borrowed from ARINC 664 part 7.
+ * However it also has one other operating mode (VLLUPFORMAT=0) where it acts
+ * somewhat closer to a pre-standard implementation of IEEE 802.1Qci
+ * (Per-Stream Filtering and Policing), which is what the driver is going to be
+ * implementing.
+ *
+ * VL Lookup
+ * Key = {DMAC && VLANID +---------+ Key = { (DMAC[47:16] & VLMASK ==
+ * && VLAN PCP | | VLMARKER)
+ * && INGRESS PORT} +---------+ (both fixed)
+ * (exact match, | && DMAC[15:0] == VLID
+ * all specified in rule) | (specified in rule)
+ * v && INGRESS PORT }
+ * ------------
+ * 0 (PSFP) / \ 1 (ARINC664)
+ * +-----------/ VLLUPFORMAT \----------+
+ * | \ (fixed) / |
+ * | \ / |
+ * 0 (forwarding) v ------------ |
+ * ------------ |
+ * / \ 1 (QoS classification) |
+ * +---/ ISCRITICAL \-----------+ |
+ * | \ (per rule) / | |
+ * | \ / VLID taken from VLID taken from
+ * v ------------ index of rule contents of rule
+ * select that matched that matched
+ * DESTPORTS | |
+ * | +---------+--------+
+ * | |
+ * | v
+ * | VL Forwarding
+ * | (indexed by VLID)
+ * | +---------+
+ * | +--------------| |
+ * | | select TYPE +---------+
+ * | v
+ * | 0 (rate ------------ 1 (time
+ * | constrained) / \ triggered)
+ * | +------/ TYPE \------------+
+ * | | \ (per VLID) / |
+ * | v \ / v
+ * | VL Policing ------------ VL Policing
+ * | (indexed by VLID) (indexed by VLID)
+ * | +---------+ +---------+
+ * | | TYPE=0 | | TYPE=1 |
+ * | +---------+ +---------+
+ * | select SHARINDX select SHARINDX to
+ * | to rate-limit re-enter VL Forwarding
+ * | groups of VL's with new VLID for egress
+ * | to same quota |
+ * | | |
+ * | select MAXLEN -> exceed => drop select MAXLEN -> exceed => drop
+ * | | |
+ * | v v
+ * | VL Forwarding VL Forwarding
+ * | (indexed by SHARINDX) (indexed by SHARINDX)
+ * | +---------+ +---------+
+ * | | TYPE=0 | | TYPE=1 |
+ * | +---------+ +---------+
+ * | select PRIORITY, select PRIORITY,
+ * | PARTITION, DESTPORTS PARTITION, DESTPORTS
+ * | | |
+ * | v v
+ * | VL Policing VL Policing
+ * | (indexed by SHARINDX) (indexed by SHARINDX)
+ * | +---------+ +---------+
+ * | | TYPE=0 | | TYPE=1 |
+ * | +---------+ +---------+
+ * | | |
+ * | v |
+ * | select BAG, -> exceed => drop |
+ * | JITTER v
+ * | | ----------------------------------------------
+ * | | / Reception Window is open for this VL \
+ * | | / (the Schedule Table executes an entry i \
+ * | | / M <= i < N, for which these conditions hold): \ no
+ * | | +----/ \-+
+ * | | |yes \ WINST[M] == 1 && WINSTINDEX[M] == VLID / |
+ * | | | \ WINEND[N] == 1 && WINSTINDEX[N] == VLID / |
+ * | | | \ / |
+ * | | | \ (the VL window has opened and not yet closed)/ |
+ * | | | ---------------------------------------------- |
+ * | | v v
+ * | | dispatch to DESTPORTS when the Schedule Table drop
+ * | | executes an entry i with TXEN == 1 && VLINDEX == i
+ * v v
+ * dispatch immediately to DESTPORTS
+ *
+ * The per-port classification key is always composed of {DMAC, VID, PCP} and
+ * is non-maskable. This 'looks like' the NULL stream identification function
+ * from IEEE 802.1CB clause 6, except for the extra VLAN PCP. When the switch
+ * ports operate as VLAN-unaware, we do allow the user to not specify the VLAN
+ * ID and PCP, and then the port-based defaults will be used.
+ *
+ * In TTEthernet, routing is something that needs to be done manually for each
+ * Virtual Link. So the flow action must always include one of:
+ * a. 'redirect', 'trap' or 'drop': select the egress port list
+ * Additionally, the following actions may be applied on a Virtual Link,
+ * turning it into 'critical' traffic:
+ * b. 'police': turn it into a rate-constrained VL, with bandwidth limitation
+ * given by the maximum frame length, bandwidth allocation gap (BAG) and
+ * maximum jitter.
+ * c. 'gate': turn it into a time-triggered VL, which can be only be received
+ * and forwarded according to a given schedule.
+ */
+
+static bool sja1105_vl_key_lower(struct sja1105_vl_lookup_entry *a,
+ struct sja1105_vl_lookup_entry *b)
+{
+ if (a->macaddr < b->macaddr)
+ return true;
+ if (a->macaddr > b->macaddr)
+ return false;
+ if (a->vlanid < b->vlanid)
+ return true;
+ if (a->vlanid > b->vlanid)
+ return false;
+ if (a->port < b->port)
+ return true;
+ if (a->port > b->port)
+ return false;
+ if (a->vlanprior < b->vlanprior)
+ return true;
+ if (a->vlanprior > b->vlanprior)
+ return false;
+ /* Keys are equal */
+ return false;
+}
+
+static int sja1105_init_virtual_links(struct sja1105_private *priv,
+ struct netlink_ext_ack *extack)
+{
+ struct sja1105_vl_lookup_entry *vl_lookup;
+ struct sja1105_table *table;
+ struct sja1105_rule *rule;
+ int num_virtual_links = 0;
+ int i, j, k;
+
+ /* Figure out the dimensioning of the problem */
+ list_for_each_entry(rule, &priv->flow_block.rules, list) {
+ if (rule->type != SJA1105_RULE_VL)
+ continue;
+ /* Each VL lookup entry matches on a single ingress port */
+ num_virtual_links += hweight_long(rule->port_mask);
+ }
+
+ if (num_virtual_links > SJA1105_MAX_VL_LOOKUP_COUNT) {
+ NL_SET_ERR_MSG_MOD(extack, "Not enough VL entries available");
+ return -ENOSPC;
+ }
+
+ /* Discard previous VL Lookup Table */
+ table = &priv->static_config.tables[BLK_IDX_VL_LOOKUP];
+ if (table->entry_count) {
+ kfree(table->entries);
+ table->entry_count = 0;
+ }
+
+ /* Nothing to do */
+ if (!num_virtual_links)
+ return 0;
+
+ /* Pre-allocate space in the static config tables */
+
+ /* VL Lookup Table */
+ table = &priv->static_config.tables[BLK_IDX_VL_LOOKUP];
+ table->entries = kcalloc(num_virtual_links,
+ table->ops->unpacked_entry_size,
+ GFP_KERNEL);
+ if (!table->entries)
+ return -ENOMEM;
+ table->entry_count = num_virtual_links;
+ vl_lookup = table->entries;
+
+ k = 0;
+
+ list_for_each_entry(rule, &priv->flow_block.rules, list) {
+ unsigned long port;
+
+ if (rule->type != SJA1105_RULE_VL)
+ continue;
+
+ for_each_set_bit(port, &rule->port_mask, SJA1105_NUM_PORTS) {
+ vl_lookup[k].format = SJA1105_VL_FORMAT_PSFP;
+ vl_lookup[k].port = port;
+ vl_lookup[k].macaddr = rule->key.vl.dmac;
+ if (rule->key.type == SJA1105_KEY_VLAN_AWARE_VL) {
+ vl_lookup[k].vlanid = rule->key.vl.vid;
+ vl_lookup[k].vlanprior = rule->key.vl.pcp;
+ } else {
+ u16 vid = dsa_8021q_rx_vid(priv->ds, port);
+
+ vl_lookup[k].vlanid = vid;
+ vl_lookup[k].vlanprior = 0;
+ }
+ /* For critical VLs, the DESTPORTS mask is taken from
+ * the VL Forwarding Table, so no point in putting it
+ * in the VL Lookup Table
+ */
+ if (rule->vl.type == SJA1105_VL_NONCRITICAL)
+ vl_lookup[k].destports = rule->vl.destports;
+ else
+ vl_lookup[k].iscritical = true;
+ k++;
+ }
+ }
+
+ /* UM10944.pdf chapter 4.2.3 VL Lookup table:
+ * "the entries in the VL Lookup table must be sorted in ascending
+ * order (i.e. the smallest value must be loaded first) according to
+ * the following sort order: MACADDR, VLANID, PORT, VLANPRIOR."
+ */
+ for (i = 0; i < num_virtual_links; i++) {
+ struct sja1105_vl_lookup_entry *a = &vl_lookup[i];
+
+ for (j = i + 1; j < num_virtual_links; j++) {
+ struct sja1105_vl_lookup_entry *b = &vl_lookup[j];
+
+ if (sja1105_vl_key_lower(b, a)) {
+ struct sja1105_vl_lookup_entry tmp = *a;
+
+ *a = *b;
+ *b = tmp;
+ }
+ }
+ }
+
+ return 0;
+}
+
+int sja1105_vl_redirect(struct sja1105_private *priv, int port,
+ struct netlink_ext_ack *extack, unsigned long cookie,
+ struct sja1105_key *key, unsigned long destports,
+ bool append)
+{
+ struct sja1105_rule *rule = sja1105_rule_find(priv, cookie);
+ int rc;
+
+ if (dsa_port_is_vlan_filtering(dsa_to_port(priv->ds, port)) &&
+ key->type != SJA1105_KEY_VLAN_AWARE_VL) {
+ NL_SET_ERR_MSG_MOD(extack,
+ "Can only redirect based on {DMAC, VID, PCP}");
+ return -EOPNOTSUPP;
+ } else if (key->type != SJA1105_KEY_VLAN_UNAWARE_VL) {
+ NL_SET_ERR_MSG_MOD(extack,
+ "Can only redirect based on DMAC");
+ return -EOPNOTSUPP;
+ }
+
+ if (!rule) {
+ rule = kzalloc(sizeof(*rule), GFP_KERNEL);
+ if (!rule)
+ return -ENOMEM;
+
+ rule->cookie = cookie;
+ rule->type = SJA1105_RULE_VL;
+ rule->key = *key;
+ list_add(&rule->list, &priv->flow_block.rules);
+ }
+
+ rule->port_mask |= BIT(port);
+ if (append)
+ rule->vl.destports |= destports;
+ else
+ rule->vl.destports = destports;
+
+ rc = sja1105_init_virtual_links(priv, extack);
+ if (rc) {
+ rule->port_mask &= ~BIT(port);
+ if (!rule->port_mask) {
+ list_del(&rule->list);
+ kfree(rule);
+ }
+ }
+
+ return rc;
+}
+
+int sja1105_vl_delete(struct sja1105_private *priv, int port,
+ struct sja1105_rule *rule, struct netlink_ext_ack *extack)
+{
+ int rc;
+
+ rule->port_mask &= ~BIT(port);
+ if (!rule->port_mask) {
+ list_del(&rule->list);
+ kfree(rule);
+ }
+
+ rc = sja1105_init_virtual_links(priv, extack);
+ if (rc)
+ return rc;
+
+ return sja1105_static_config_reload(priv, SJA1105_VIRTUAL_LINKS);
+}
diff --git a/drivers/net/dsa/sja1105/sja1105_vl.h b/drivers/net/dsa/sja1105/sja1105_vl.h
new file mode 100644
index 000000000000..08ee5557b463
--- /dev/null
+++ b/drivers/net/dsa/sja1105/sja1105_vl.h
@@ -0,0 +1,41 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/* Copyright 2020, NXP Semiconductors
+ */
+#ifndef _SJA1105_VL_H
+#define _SJA1105_VL_H
+
+#if IS_ENABLED(CONFIG_NET_DSA_SJA1105_VL)
+
+int sja1105_vl_redirect(struct sja1105_private *priv, int port,
+ struct netlink_ext_ack *extack, unsigned long cookie,
+ struct sja1105_key *key, unsigned long destports,
+ bool append);
+
+int sja1105_vl_delete(struct sja1105_private *priv, int port,
+ struct sja1105_rule *rule,
+ struct netlink_ext_ack *extack);
+
+#else
+
+static inline int sja1105_vl_redirect(struct sja1105_private *priv, int port,
+ struct netlink_ext_ack *extack,
+ unsigned long cookie,
+ struct sja1105_key *key,
+ unsigned long destports,
+ bool append)
+{
+ NL_SET_ERR_MSG_MOD(extack, "Virtual Links not compiled in");
+ return -EOPNOTSUPP;
+}
+
+static inline int sja1105_vl_delete(struct sja1105_private *priv,
+ int port, struct sja1105_rule *rule,
+ struct netlink_ext_ack *extack)
+{
+ NL_SET_ERR_MSG_MOD(extack, "Virtual Links not compiled in");
+ return -EOPNOTSUPP;
+}
+
+#endif /* IS_ENABLED(CONFIG_NET_DSA_SJA1105_VL) */
+
+#endif /* _SJA1105_VL_H */
--
2.17.1
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH v3 net-next 5/6] net: dsa: sja1105: implement tc-gate using time-triggered virtual links
2020-05-05 19:20 [PATCH v3 net-next 0/6] tc-gate offload for SJA1105 DSA switch Vladimir Oltean
` (3 preceding siblings ...)
2020-05-05 19:20 ` [PATCH v3 net-next 4/6] net: dsa: sja1105: support flow-based redirection via virtual links Vladimir Oltean
@ 2020-05-05 19:20 ` Vladimir Oltean
2020-05-05 19:20 ` [PATCH v3 net-next 6/6] docs: net: dsa: sja1105: document intended usage of " Vladimir Oltean
2020-05-08 0:32 ` [PATCH v3 net-next 0/6] tc-gate offload for SJA1105 DSA switch David Miller
6 siblings, 0 replies; 9+ messages in thread
From: Vladimir Oltean @ 2020-05-05 19:20 UTC (permalink / raw)
To: netdev; +Cc: andrew, f.fainelli, vivien.didelot, vinicius.gomes, po.liu
From: Vladimir Oltean <vladimir.oltean@nxp.com>
Restrict the TTEthernet hardware support on this switch to operate as
closely as possible to IEEE 802.1Qci as possible. This means that it can
perform PTP-time-based ingress admission control on streams identified
by {DMAC, VID, PCP}, which is useful when trying to ensure the
determinism of traffic scheduled via IEEE 802.1Qbv.
The oddity comes from the fact that in hardware (and in TTEthernet at
large), virtual links always need a full-blown action, including not
only the type of policing, but also the list of destination ports. So in
practice, a single tc-gate action will result in all packets getting
dropped. Additional actions (either "trap" or "redirect") need to be
specified in the same filter rule such that the conforming packets are
actually forwarded somewhere.
Apart from the VL Lookup, Policing and Forwarding tables which need to
be programmed for each flow (virtual link), the Schedule engine also
needs to be told to open/close the admission gates for each individual
virtual link. A fairly accurate (and detailed) description of how that
works is already present in sja1105_tas.c, since it is already used to
trigger the egress gates for the tc-taprio offload (IEEE 802.1Qbv). Key
point here, we remember that the schedule engine supports 8
"subschedules" (execution threads that iterate through the global
schedule in parallel, and that no 2 hardware threads must execute a
schedule entry at the same time). For tc-taprio, each egress port used
one of these 8 subschedules, leaving a total of 4 subschedules unused.
In principle we could have allocated 1 subschedule for the tc-gate
offload of each ingress port, but actually the schedules of all virtual
links installed on each ingress port would have needed to be merged
together, before they could have been programmed to hardware. So
simplify our life and just merge the entire tc-gate configuration, for
all virtual links on all ingress ports, into a single subschedule. Be
sure to check that against the usual hardware scheduling conflicts, and
program it to hardware alongside any tc-taprio subschedule that may be
present.
The following scenarios were tested:
1. Quantitative testing:
tc qdisc add dev swp2 clsact
tc filter add dev swp2 ingress flower skip_sw \
dst_mac 42:be:24:9b:76:20 \
action gate index 1 base-time 0 \
sched-entry OPEN 1200 -1 -1 \
sched-entry CLOSE 1200 -1 -1 \
action trap
ping 192.168.1.2 -f
PING 192.168.1.2 (192.168.1.2) 56(84) bytes of data.
.............................
--- 192.168.1.2 ping statistics ---
948 packets transmitted, 467 received, 50.7384% packet loss, time 9671ms
2. Qualitative testing (with a phase-aligned schedule - the clocks are
synchronized by ptp4l, not shown here):
Receiver (sja1105):
tc qdisc add dev swp2 clsact
now=$(phc_ctl /dev/ptp1 get | awk '/clock time is/ {print $5}') && \
sec=$(echo $now | awk -F. '{print $1}') && \
base_time="$(((sec + 2) * 1000000000))" && \
echo "base time ${base_time}"
tc filter add dev swp2 ingress flower skip_sw \
dst_mac 42:be:24:9b:76:20 \
action gate base-time ${base_time} \
sched-entry OPEN 60000 -1 -1 \
sched-entry CLOSE 40000 -1 -1 \
action trap
Sender (enetc):
now=$(phc_ctl /dev/ptp0 get | awk '/clock time is/ {print $5}') && \
sec=$(echo $now | awk -F. '{print $1}') && \
base_time="$(((sec + 2) * 1000000000))" && \
echo "base time ${base_time}"
tc qdisc add dev eno0 parent root taprio \
num_tc 8 \
map 0 1 2 3 4 5 6 7 \
queues 1@0 1@1 1@2 1@3 1@4 1@5 1@6 1@7 \
base-time ${base_time} \
sched-entry S 01 50000 \
sched-entry S 00 50000 \
flags 2
ping -A 192.168.1.1
PING 192.168.1.1 (192.168.1.1): 56 data bytes
...
^C
--- 192.168.1.1 ping statistics ---
1425 packets transmitted, 1424 packets received, 0% packet loss
round-trip min/avg/max = 0.322/0.361/0.990 ms
And just for comparison, with the tc-taprio schedule deleted:
ping -A 192.168.1.1
PING 192.168.1.1 (192.168.1.1): 56 data bytes
...
^C
--- 192.168.1.1 ping statistics ---
33 packets transmitted, 19 packets received, 42% packet loss
round-trip min/avg/max = 0.336/0.464/0.597 ms
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
Changes from v2:
Provide a shim implementation of sja1105_init_scheduling.
Changes from v1:
None.
Changes from RFC:
* Deny gating actions in absence of any routing action. Without this
check, the default behavior would be to drop everything, which is
probably not what the user expects.
* Deny offloading IntervalOctetMax, which I initially mistook for meaning
per-flow MTU. It actually means a sort of flow metering within each
time slot. Not supported.
drivers/net/dsa/sja1105/sja1105.h | 13 +-
drivers/net/dsa/sja1105/sja1105_flower.c | 57 +-
drivers/net/dsa/sja1105/sja1105_main.c | 1 +
drivers/net/dsa/sja1105/sja1105_ptp.h | 13 +
drivers/net/dsa/sja1105/sja1105_spi.c | 2 +
.../net/dsa/sja1105/sja1105_static_config.h | 2 +
drivers/net/dsa/sja1105/sja1105_tas.c | 127 ++++-
drivers/net/dsa/sja1105/sja1105_tas.h | 36 ++
drivers/net/dsa/sja1105/sja1105_vl.c | 494 ++++++++++++++++++
drivers/net/dsa/sja1105/sja1105_vl.h | 31 ++
10 files changed, 759 insertions(+), 17 deletions(-)
diff --git a/drivers/net/dsa/sja1105/sja1105.h b/drivers/net/dsa/sja1105/sja1105.h
index 1756000f6936..8df2a5c53b02 100644
--- a/drivers/net/dsa/sja1105/sja1105.h
+++ b/drivers/net/dsa/sja1105/sja1105.h
@@ -36,6 +36,7 @@ struct sja1105_regs {
u64 status;
u64 port_control;
u64 rgu;
+ u64 vl_status;
u64 config;
u64 sgmii;
u64 rmii_pll1;
@@ -156,8 +157,16 @@ struct sja1105_rule {
/* SJA1105_RULE_VL */
struct {
- unsigned long destports;
enum sja1105_vl_type type;
+ unsigned long destports;
+ int sharindx;
+ int maxlen;
+ int ipv;
+ u64 base_time;
+ u64 cycle_time;
+ int num_entries;
+ struct action_gate_entry *entries;
+ struct flow_stats stats;
} vl;
};
};
@@ -304,6 +313,8 @@ int sja1105_cls_flower_del(struct dsa_switch *ds, int port,
struct flow_cls_offload *cls, bool ingress);
int sja1105_cls_flower_add(struct dsa_switch *ds, int port,
struct flow_cls_offload *cls, bool ingress);
+int sja1105_cls_flower_stats(struct dsa_switch *ds, int port,
+ struct flow_cls_offload *cls, bool ingress);
void sja1105_flower_setup(struct dsa_switch *ds);
void sja1105_flower_teardown(struct dsa_switch *ds);
struct sja1105_rule *sja1105_rule_find(struct sja1105_private *priv,
diff --git a/drivers/net/dsa/sja1105/sja1105_flower.c b/drivers/net/dsa/sja1105/sja1105_flower.c
index 5f08eed0b1fc..9ee8968610cd 100644
--- a/drivers/net/dsa/sja1105/sja1105_flower.c
+++ b/drivers/net/dsa/sja1105/sja1105_flower.c
@@ -309,7 +309,9 @@ int sja1105_cls_flower_add(struct dsa_switch *ds, int port,
struct sja1105_private *priv = ds->priv;
const struct flow_action_entry *act;
unsigned long cookie = cls->cookie;
+ bool routing_rule = false;
struct sja1105_key key;
+ bool gate_rule = false;
bool vl_rule = false;
int rc, i;
@@ -332,6 +334,7 @@ int sja1105_cls_flower_add(struct dsa_switch *ds, int port,
case FLOW_ACTION_TRAP: {
int cpu = dsa_upstream_port(ds, port);
+ routing_rule = true;
vl_rule = true;
rc = sja1105_vl_redirect(priv, port, extack, cookie,
@@ -350,6 +353,7 @@ int sja1105_cls_flower_add(struct dsa_switch *ds, int port,
return -EOPNOTSUPP;
}
+ routing_rule = true;
vl_rule = true;
rc = sja1105_vl_redirect(priv, port, extack, cookie,
@@ -366,6 +370,21 @@ int sja1105_cls_flower_add(struct dsa_switch *ds, int port,
if (rc)
goto out;
break;
+ case FLOW_ACTION_GATE:
+ gate_rule = true;
+ vl_rule = true;
+
+ rc = sja1105_vl_gate(priv, port, extack, cookie,
+ &key, act->gate.index,
+ act->gate.prio,
+ act->gate.basetime,
+ act->gate.cycletime,
+ act->gate.cycletimeext,
+ act->gate.num_entries,
+ act->gate.entries);
+ if (rc)
+ goto out;
+ break;
default:
NL_SET_ERR_MSG_MOD(extack,
"Action not supported");
@@ -374,8 +393,23 @@ int sja1105_cls_flower_add(struct dsa_switch *ds, int port,
}
}
- if (vl_rule && !rc)
+ if (vl_rule && !rc) {
+ /* Delay scheduling configuration until DESTPORTS has been
+ * populated by all other actions.
+ */
+ if (gate_rule) {
+ if (!routing_rule) {
+ NL_SET_ERR_MSG_MOD(extack,
+ "Can only offload gate action together with redirect or trap");
+ return -EOPNOTSUPP;
+ }
+ rc = sja1105_init_scheduling(priv);
+ if (rc)
+ goto out;
+ }
+
rc = sja1105_static_config_reload(priv, SJA1105_VIRTUAL_LINKS);
+ }
out:
return rc;
@@ -421,6 +455,27 @@ int sja1105_cls_flower_del(struct dsa_switch *ds, int port,
return sja1105_static_config_reload(priv, SJA1105_BEST_EFFORT_POLICING);
}
+int sja1105_cls_flower_stats(struct dsa_switch *ds, int port,
+ struct flow_cls_offload *cls, bool ingress)
+{
+ struct sja1105_private *priv = ds->priv;
+ struct sja1105_rule *rule = sja1105_rule_find(priv, cls->cookie);
+ int rc;
+
+ if (!rule)
+ return 0;
+
+ if (rule->type != SJA1105_RULE_VL)
+ return 0;
+
+ rc = sja1105_vl_stats(priv, port, rule, &cls->stats,
+ cls->common.extack);
+ if (rc)
+ return rc;
+
+ return 0;
+}
+
void sja1105_flower_setup(struct dsa_switch *ds)
{
struct sja1105_private *priv = ds->priv;
diff --git a/drivers/net/dsa/sja1105/sja1105_main.c b/drivers/net/dsa/sja1105/sja1105_main.c
index 8bb104ee73d5..666e54565df0 100644
--- a/drivers/net/dsa/sja1105/sja1105_main.c
+++ b/drivers/net/dsa/sja1105/sja1105_main.c
@@ -2369,6 +2369,7 @@ static const struct dsa_switch_ops sja1105_switch_ops = {
.port_policer_del = sja1105_port_policer_del,
.cls_flower_add = sja1105_cls_flower_add,
.cls_flower_del = sja1105_cls_flower_del,
+ .cls_flower_stats = sja1105_cls_flower_stats,
};
static int sja1105_check_device_id(struct sja1105_private *priv)
diff --git a/drivers/net/dsa/sja1105/sja1105_ptp.h b/drivers/net/dsa/sja1105/sja1105_ptp.h
index 43480b24f1f0..6408d1158f2d 100644
--- a/drivers/net/dsa/sja1105/sja1105_ptp.h
+++ b/drivers/net/dsa/sja1105/sja1105_ptp.h
@@ -48,6 +48,19 @@ static inline s64 future_base_time(s64 base_time, s64 cycle_time, s64 now)
return base_time + n * cycle_time;
}
+/* This is not a preprocessor macro because the "ns" argument may or may not be
+ * s64 at caller side. This ensures it is properly type-cast before div_s64.
+ */
+static inline s64 ns_to_sja1105_delta(s64 ns)
+{
+ return div_s64(ns, 200);
+}
+
+static inline s64 sja1105_delta_to_ns(s64 delta)
+{
+ return delta * 200;
+}
+
struct sja1105_ptp_cmd {
u64 startptpcp; /* start toggling PTP_CLK pin */
u64 stopptpcp; /* stop toggling PTP_CLK pin */
diff --git a/drivers/net/dsa/sja1105/sja1105_spi.c b/drivers/net/dsa/sja1105/sja1105_spi.c
index 43f14a5c2718..0be75c49e6c3 100644
--- a/drivers/net/dsa/sja1105/sja1105_spi.c
+++ b/drivers/net/dsa/sja1105/sja1105_spi.c
@@ -439,6 +439,7 @@ static struct sja1105_regs sja1105et_regs = {
.prod_id = 0x100BC3,
.status = 0x1,
.port_control = 0x11,
+ .vl_status = 0x10000,
.config = 0x020000,
.rgu = 0x100440,
/* UM10944.pdf, Table 86, ACU Register overview */
@@ -472,6 +473,7 @@ static struct sja1105_regs sja1105pqrs_regs = {
.prod_id = 0x100BC3,
.status = 0x1,
.port_control = 0x12,
+ .vl_status = 0x10000,
.config = 0x020000,
.rgu = 0x100440,
/* UM10944.pdf, Table 86, ACU Register overview */
diff --git a/drivers/net/dsa/sja1105/sja1105_static_config.h b/drivers/net/dsa/sja1105/sja1105_static_config.h
index 1a8fcbbb57b6..b569e3de3590 100644
--- a/drivers/net/dsa/sja1105/sja1105_static_config.h
+++ b/drivers/net/dsa/sja1105/sja1105_static_config.h
@@ -302,6 +302,8 @@ struct sja1105_vl_lookup_entry {
u64 vlid;
};
};
+ /* Not part of hardware structure */
+ unsigned long flow_cookie;
};
struct sja1105_vl_policing_entry {
diff --git a/drivers/net/dsa/sja1105/sja1105_tas.c b/drivers/net/dsa/sja1105/sja1105_tas.c
index 77e547b4cd89..3aa1a8b5f766 100644
--- a/drivers/net/dsa/sja1105/sja1105_tas.c
+++ b/drivers/net/dsa/sja1105/sja1105_tas.c
@@ -7,7 +7,6 @@
#define SJA1105_TAS_CLKSRC_STANDALONE 1
#define SJA1105_TAS_CLKSRC_AS6802 2
#define SJA1105_TAS_CLKSRC_PTP 3
-#define SJA1105_TAS_MAX_DELTA BIT(19)
#define SJA1105_GATE_MASK GENMASK_ULL(SJA1105_NUM_TC - 1, 0)
#define work_to_sja1105_tas(d) \
@@ -15,22 +14,10 @@
#define tas_to_sja1105(d) \
container_of((d), struct sja1105_private, tas_data)
-/* This is not a preprocessor macro because the "ns" argument may or may not be
- * s64 at caller side. This ensures it is properly type-cast before div_s64.
- */
-static s64 ns_to_sja1105_delta(s64 ns)
-{
- return div_s64(ns, 200);
-}
-
-static s64 sja1105_delta_to_ns(s64 delta)
-{
- return delta * 200;
-}
-
static int sja1105_tas_set_runtime_params(struct sja1105_private *priv)
{
struct sja1105_tas_data *tas_data = &priv->tas_data;
+ struct sja1105_gating_config *gating_cfg = &tas_data->gating_cfg;
struct dsa_switch *ds = priv->ds;
s64 earliest_base_time = S64_MAX;
s64 latest_base_time = 0;
@@ -59,6 +46,19 @@ static int sja1105_tas_set_runtime_params(struct sja1105_private *priv)
}
}
+ if (!list_empty(&gating_cfg->entries)) {
+ tas_data->enabled = true;
+
+ if (max_cycle_time < gating_cfg->cycle_time)
+ max_cycle_time = gating_cfg->cycle_time;
+ if (latest_base_time < gating_cfg->base_time)
+ latest_base_time = gating_cfg->base_time;
+ if (earliest_base_time > gating_cfg->base_time) {
+ earliest_base_time = gating_cfg->base_time;
+ its_cycle_time = gating_cfg->cycle_time;
+ }
+ }
+
if (!tas_data->enabled)
return 0;
@@ -155,13 +155,14 @@ static int sja1105_tas_set_runtime_params(struct sja1105_private *priv)
* their "subschedule end index" (subscheind) equal to the last valid
* subschedule's end index (in this case 5).
*/
-static int sja1105_init_scheduling(struct sja1105_private *priv)
+int sja1105_init_scheduling(struct sja1105_private *priv)
{
struct sja1105_schedule_entry_points_entry *schedule_entry_points;
struct sja1105_schedule_entry_points_params_entry
*schedule_entry_points_params;
struct sja1105_schedule_params_entry *schedule_params;
struct sja1105_tas_data *tas_data = &priv->tas_data;
+ struct sja1105_gating_config *gating_cfg = &tas_data->gating_cfg;
struct sja1105_schedule_entry *schedule;
struct sja1105_table *table;
int schedule_start_idx;
@@ -213,6 +214,11 @@ static int sja1105_init_scheduling(struct sja1105_private *priv)
}
}
+ if (!list_empty(&gating_cfg->entries)) {
+ num_entries += gating_cfg->num_entries;
+ num_cycles++;
+ }
+
/* Nothing to do */
if (!num_cycles)
return 0;
@@ -312,6 +318,42 @@ static int sja1105_init_scheduling(struct sja1105_private *priv)
cycle++;
}
+ if (!list_empty(&gating_cfg->entries)) {
+ struct sja1105_gate_entry *e;
+
+ /* Relative base time */
+ s64 rbt;
+
+ schedule_start_idx = k;
+ schedule_end_idx = k + gating_cfg->num_entries - 1;
+ rbt = future_base_time(gating_cfg->base_time,
+ gating_cfg->cycle_time,
+ tas_data->earliest_base_time);
+ rbt -= tas_data->earliest_base_time;
+ entry_point_delta = ns_to_sja1105_delta(rbt) + 1;
+
+ schedule_entry_points[cycle].subschindx = cycle;
+ schedule_entry_points[cycle].delta = entry_point_delta;
+ schedule_entry_points[cycle].address = schedule_start_idx;
+
+ for (i = cycle; i < 8; i++)
+ schedule_params->subscheind[i] = schedule_end_idx;
+
+ list_for_each_entry(e, &gating_cfg->entries, list) {
+ schedule[k].delta = ns_to_sja1105_delta(e->interval);
+ schedule[k].destports = e->rule->vl.destports;
+ schedule[k].setvalid = true;
+ schedule[k].txen = true;
+ schedule[k].vlindex = e->rule->vl.sharindx;
+ schedule[k].winstindex = e->rule->vl.sharindx;
+ if (e->gate_state) /* Gate open */
+ schedule[k].winst = true;
+ else /* Gate closed */
+ schedule[k].winend = true;
+ k++;
+ }
+ }
+
return 0;
}
@@ -415,6 +457,54 @@ sja1105_tas_check_conflicts(struct sja1105_private *priv, int port,
return false;
}
+/* Check the tc-taprio configuration on @port for conflicts with the tc-gate
+ * global subschedule. If @port is -1, check it against all ports.
+ * To reuse the sja1105_tas_check_conflicts logic without refactoring it,
+ * convert the gating configuration to a dummy tc-taprio offload structure.
+ */
+bool sja1105_gating_check_conflicts(struct sja1105_private *priv, int port,
+ struct netlink_ext_ack *extack)
+{
+ struct sja1105_gating_config *gating_cfg = &priv->tas_data.gating_cfg;
+ size_t num_entries = gating_cfg->num_entries;
+ struct tc_taprio_qopt_offload *dummy;
+ struct sja1105_gate_entry *e;
+ bool conflict;
+ int i = 0;
+
+ if (list_empty(&gating_cfg->entries))
+ return false;
+
+ dummy = kzalloc(sizeof(struct tc_taprio_sched_entry) * num_entries +
+ sizeof(struct tc_taprio_qopt_offload), GFP_KERNEL);
+ if (!dummy) {
+ NL_SET_ERR_MSG_MOD(extack, "Failed to allocate memory");
+ return true;
+ }
+
+ dummy->num_entries = num_entries;
+ dummy->base_time = gating_cfg->base_time;
+ dummy->cycle_time = gating_cfg->cycle_time;
+
+ list_for_each_entry(e, &gating_cfg->entries, list)
+ dummy->entries[i++].interval = e->interval;
+
+ if (port != -1) {
+ conflict = sja1105_tas_check_conflicts(priv, port, dummy);
+ } else {
+ for (port = 0; port < SJA1105_NUM_PORTS; port++) {
+ conflict = sja1105_tas_check_conflicts(priv, port,
+ dummy);
+ if (conflict)
+ break;
+ }
+ }
+
+ kfree(dummy);
+
+ return conflict;
+}
+
int sja1105_setup_tc_taprio(struct dsa_switch *ds, int port,
struct tc_taprio_qopt_offload *admin)
{
@@ -473,6 +563,11 @@ int sja1105_setup_tc_taprio(struct dsa_switch *ds, int port,
return -ERANGE;
}
+ if (sja1105_gating_check_conflicts(priv, port, NULL)) {
+ dev_err(ds->dev, "Conflict with tc-gate schedule\n");
+ return -ERANGE;
+ }
+
tas_data->offload[port] = taprio_offload_get(admin);
rc = sja1105_init_scheduling(priv);
@@ -779,6 +874,8 @@ void sja1105_tas_setup(struct dsa_switch *ds)
INIT_WORK(&tas_data->tas_work, sja1105_tas_state_machine);
tas_data->state = SJA1105_TAS_STATE_DISABLED;
tas_data->last_op = SJA1105_PTP_NONE;
+
+ INIT_LIST_HEAD(&tas_data->gating_cfg.entries);
}
void sja1105_tas_teardown(struct dsa_switch *ds)
diff --git a/drivers/net/dsa/sja1105/sja1105_tas.h b/drivers/net/dsa/sja1105/sja1105_tas.h
index b226c3dfd5b1..0c173ff51751 100644
--- a/drivers/net/dsa/sja1105/sja1105_tas.h
+++ b/drivers/net/dsa/sja1105/sja1105_tas.h
@@ -6,6 +6,10 @@
#include <net/pkt_sched.h>
+#define SJA1105_TAS_MAX_DELTA BIT(18)
+
+struct sja1105_private;
+
#if IS_ENABLED(CONFIG_NET_DSA_SJA1105_TAS)
enum sja1105_tas_state {
@@ -20,8 +24,23 @@ enum sja1105_ptp_op {
SJA1105_PTP_ADJUSTFREQ,
};
+struct sja1105_gate_entry {
+ struct list_head list;
+ struct sja1105_rule *rule;
+ s64 interval;
+ u8 gate_state;
+};
+
+struct sja1105_gating_config {
+ u64 cycle_time;
+ s64 base_time;
+ int num_entries;
+ struct list_head entries;
+};
+
struct sja1105_tas_data {
struct tc_taprio_qopt_offload *offload[SJA1105_NUM_PORTS];
+ struct sja1105_gating_config gating_cfg;
enum sja1105_tas_state state;
enum sja1105_ptp_op last_op;
struct work_struct tas_work;
@@ -42,6 +61,11 @@ void sja1105_tas_clockstep(struct dsa_switch *ds);
void sja1105_tas_adjfreq(struct dsa_switch *ds);
+bool sja1105_gating_check_conflicts(struct sja1105_private *priv, int port,
+ struct netlink_ext_ack *extack);
+
+int sja1105_init_scheduling(struct sja1105_private *priv);
+
#else
/* C doesn't allow empty structures, bah! */
@@ -63,6 +87,18 @@ static inline void sja1105_tas_clockstep(struct dsa_switch *ds) { }
static inline void sja1105_tas_adjfreq(struct dsa_switch *ds) { }
+static inline bool
+sja1105_gating_check_conflicts(struct dsa_switch *ds, int port,
+ struct netlink_ext_ack *extack)
+{
+ return true;
+}
+
+static inline int sja1105_init_scheduling(struct sja1105_private *priv)
+{
+ return 0;
+}
+
#endif /* IS_ENABLED(CONFIG_NET_DSA_SJA1105_TAS) */
#endif /* _SJA1105_TAS_H */
diff --git a/drivers/net/dsa/sja1105/sja1105_vl.c b/drivers/net/dsa/sja1105/sja1105_vl.c
index c226779b8275..b52f1af6e7e7 100644
--- a/drivers/net/dsa/sja1105/sja1105_vl.c
+++ b/drivers/net/dsa/sja1105/sja1105_vl.c
@@ -1,9 +1,13 @@
// SPDX-License-Identifier: GPL-2.0
/* Copyright 2020, NXP Semiconductors
*/
+#include <net/tc_act/tc_gate.h>
#include <linux/dsa/8021q.h>
#include "sja1105.h"
+#define SJA1105_VL_FRAME_MEMORY 100
+#define SJA1105_SIZE_VL_STATUS 8
+
/* The switch flow classification core implements TTEthernet, which 'thinks' in
* terms of Virtual Links (VL), a concept borrowed from ARINC 664 part 7.
* However it also has one other operating mode (VLLUPFORMAT=0) where it acts
@@ -137,18 +141,33 @@ static bool sja1105_vl_key_lower(struct sja1105_vl_lookup_entry *a,
static int sja1105_init_virtual_links(struct sja1105_private *priv,
struct netlink_ext_ack *extack)
{
+ struct sja1105_l2_forwarding_params_entry *l2_fwd_params;
+ struct sja1105_vl_forwarding_params_entry *vl_fwd_params;
+ struct sja1105_vl_policing_entry *vl_policing;
+ struct sja1105_vl_forwarding_entry *vl_fwd;
struct sja1105_vl_lookup_entry *vl_lookup;
+ bool have_critical_virtual_links = false;
struct sja1105_table *table;
struct sja1105_rule *rule;
int num_virtual_links = 0;
+ int max_sharindx = 0;
int i, j, k;
+ table = &priv->static_config.tables[BLK_IDX_L2_FORWARDING_PARAMS];
+ l2_fwd_params = table->entries;
+ l2_fwd_params->part_spc[0] = SJA1105_MAX_FRAME_MEMORY;
+
/* Figure out the dimensioning of the problem */
list_for_each_entry(rule, &priv->flow_block.rules, list) {
if (rule->type != SJA1105_RULE_VL)
continue;
/* Each VL lookup entry matches on a single ingress port */
num_virtual_links += hweight_long(rule->port_mask);
+
+ if (rule->vl.type != SJA1105_VL_NONCRITICAL)
+ have_critical_virtual_links = true;
+ if (max_sharindx < rule->vl.sharindx)
+ max_sharindx = rule->vl.sharindx;
}
if (num_virtual_links > SJA1105_MAX_VL_LOOKUP_COUNT) {
@@ -156,6 +175,13 @@ static int sja1105_init_virtual_links(struct sja1105_private *priv,
return -ENOSPC;
}
+ if (max_sharindx + 1 > SJA1105_MAX_VL_LOOKUP_COUNT) {
+ NL_SET_ERR_MSG_MOD(extack, "Policer index out of range");
+ return -ENOSPC;
+ }
+
+ max_sharindx = max_t(int, num_virtual_links, max_sharindx) + 1;
+
/* Discard previous VL Lookup Table */
table = &priv->static_config.tables[BLK_IDX_VL_LOOKUP];
if (table->entry_count) {
@@ -163,6 +189,27 @@ static int sja1105_init_virtual_links(struct sja1105_private *priv,
table->entry_count = 0;
}
+ /* Discard previous VL Policing Table */
+ table = &priv->static_config.tables[BLK_IDX_VL_POLICING];
+ if (table->entry_count) {
+ kfree(table->entries);
+ table->entry_count = 0;
+ }
+
+ /* Discard previous VL Forwarding Table */
+ table = &priv->static_config.tables[BLK_IDX_VL_FORWARDING];
+ if (table->entry_count) {
+ kfree(table->entries);
+ table->entry_count = 0;
+ }
+
+ /* Discard previous VL Forwarding Parameters Table */
+ table = &priv->static_config.tables[BLK_IDX_VL_FORWARDING_PARAMS];
+ if (table->entry_count) {
+ kfree(table->entries);
+ table->entry_count = 0;
+ }
+
/* Nothing to do */
if (!num_virtual_links)
return 0;
@@ -208,6 +255,7 @@ static int sja1105_init_virtual_links(struct sja1105_private *priv,
vl_lookup[k].destports = rule->vl.destports;
else
vl_lookup[k].iscritical = true;
+ vl_lookup[k].flow_cookie = rule->cookie;
k++;
}
}
@@ -232,6 +280,68 @@ static int sja1105_init_virtual_links(struct sja1105_private *priv,
}
}
+ if (!have_critical_virtual_links)
+ return 0;
+
+ /* VL Policing Table */
+ table = &priv->static_config.tables[BLK_IDX_VL_POLICING];
+ table->entries = kcalloc(max_sharindx, table->ops->unpacked_entry_size,
+ GFP_KERNEL);
+ if (!table->entries)
+ return -ENOMEM;
+ table->entry_count = max_sharindx;
+ vl_policing = table->entries;
+
+ /* VL Forwarding Table */
+ table = &priv->static_config.tables[BLK_IDX_VL_FORWARDING];
+ table->entries = kcalloc(max_sharindx, table->ops->unpacked_entry_size,
+ GFP_KERNEL);
+ if (!table->entries)
+ return -ENOMEM;
+ table->entry_count = max_sharindx;
+ vl_fwd = table->entries;
+
+ /* VL Forwarding Parameters Table */
+ table = &priv->static_config.tables[BLK_IDX_VL_FORWARDING_PARAMS];
+ table->entries = kcalloc(1, table->ops->unpacked_entry_size,
+ GFP_KERNEL);
+ if (!table->entries)
+ return -ENOMEM;
+ table->entry_count = 1;
+ vl_fwd_params = table->entries;
+
+ /* Reserve some frame buffer memory for the critical-traffic virtual
+ * links (this needs to be done). At the moment, hardcode the value
+ * at 100 blocks of 128 bytes of memory each. This leaves 829 blocks
+ * remaining for best-effort traffic. TODO: figure out a more flexible
+ * way to perform the frame buffer partitioning.
+ */
+ l2_fwd_params->part_spc[0] = SJA1105_MAX_FRAME_MEMORY -
+ SJA1105_VL_FRAME_MEMORY;
+ vl_fwd_params->partspc[0] = SJA1105_VL_FRAME_MEMORY;
+
+ for (i = 0; i < num_virtual_links; i++) {
+ unsigned long cookie = vl_lookup[i].flow_cookie;
+ struct sja1105_rule *rule = sja1105_rule_find(priv, cookie);
+
+ if (rule->vl.type == SJA1105_VL_NONCRITICAL)
+ continue;
+ if (rule->vl.type == SJA1105_VL_TIME_TRIGGERED) {
+ int sharindx = rule->vl.sharindx;
+
+ vl_policing[i].type = 1;
+ vl_policing[i].sharindx = sharindx;
+ vl_policing[i].maxlen = rule->vl.maxlen;
+ vl_policing[sharindx].type = 1;
+
+ vl_fwd[i].type = 1;
+ vl_fwd[sharindx].type = 1;
+ vl_fwd[sharindx].priority = rule->vl.ipv;
+ vl_fwd[sharindx].partition = 0;
+ vl_fwd[sharindx].destports = rule->vl.destports;
+ }
+ }
+
return 0;
}
@@ -300,3 +410,387 @@ int sja1105_vl_delete(struct sja1105_private *priv, int port,
return sja1105_static_config_reload(priv, SJA1105_VIRTUAL_LINKS);
}
+
+/* Insert into the global gate list, sorted by gate action time. */
+static int sja1105_insert_gate_entry(struct sja1105_gating_config *gating_cfg,
+ struct sja1105_rule *rule,
+ u8 gate_state, s64 entry_time,
+ struct netlink_ext_ack *extack)
+{
+ struct sja1105_gate_entry *e;
+ int rc;
+
+ e = kzalloc(sizeof(*e), GFP_KERNEL);
+ if (!e)
+ return -ENOMEM;
+
+ e->rule = rule;
+ e->gate_state = gate_state;
+ e->interval = entry_time;
+
+ if (list_empty(&gating_cfg->entries)) {
+ list_add(&e->list, &gating_cfg->entries);
+ } else {
+ struct sja1105_gate_entry *p;
+
+ list_for_each_entry(p, &gating_cfg->entries, list) {
+ if (p->interval == e->interval) {
+ NL_SET_ERR_MSG_MOD(extack,
+ "Gate conflict");
+ rc = -EBUSY;
+ goto err;
+ }
+
+ if (e->interval < p->interval)
+ break;
+ }
+ list_add(&e->list, p->list.prev);
+ }
+
+ gating_cfg->num_entries++;
+
+ return 0;
+err:
+ kfree(e);
+ return rc;
+}
+
+/* The gate entries contain absolute times in their e->interval field. Convert
+ * that to proper intervals (i.e. "0, 5, 10, 15" to "5, 5, 5, 5").
+ */
+static void
+sja1105_gating_cfg_time_to_interval(struct sja1105_gating_config *gating_cfg,
+ u64 cycle_time)
+{
+ struct sja1105_gate_entry *last_e;
+ struct sja1105_gate_entry *e;
+ struct list_head *prev;
+ u32 prev_time = 0;
+
+ list_for_each_entry(e, &gating_cfg->entries, list) {
+ struct sja1105_gate_entry *p;
+
+ prev = e->list.prev;
+
+ if (prev == &gating_cfg->entries)
+ continue;
+
+ p = list_entry(prev, struct sja1105_gate_entry, list);
+ prev_time = e->interval;
+ p->interval = e->interval - p->interval;
+ }
+ last_e = list_last_entry(&gating_cfg->entries,
+ struct sja1105_gate_entry, list);
+ if (last_e->list.prev != &gating_cfg->entries)
+ last_e->interval = cycle_time - last_e->interval;
+}
+
+static void sja1105_free_gating_config(struct sja1105_gating_config *gating_cfg)
+{
+ struct sja1105_gate_entry *e, *n;
+
+ list_for_each_entry_safe(e, n, &gating_cfg->entries, list) {
+ list_del(&e->list);
+ kfree(e);
+ }
+}
+
+static int sja1105_compose_gating_subschedule(struct sja1105_private *priv,
+ struct netlink_ext_ack *extack)
+{
+ struct sja1105_gating_config *gating_cfg = &priv->tas_data.gating_cfg;
+ struct sja1105_rule *rule;
+ s64 max_cycle_time = 0;
+ s64 its_base_time = 0;
+ int i, rc = 0;
+
+ list_for_each_entry(rule, &priv->flow_block.rules, list) {
+ if (rule->type != SJA1105_RULE_VL)
+ continue;
+ if (rule->vl.type != SJA1105_VL_TIME_TRIGGERED)
+ continue;
+
+ if (max_cycle_time < rule->vl.cycle_time) {
+ max_cycle_time = rule->vl.cycle_time;
+ its_base_time = rule->vl.base_time;
+ }
+ }
+
+ if (!max_cycle_time)
+ return 0;
+
+ dev_dbg(priv->ds->dev, "max_cycle_time %lld its_base_time %lld\n",
+ max_cycle_time, its_base_time);
+
+ sja1105_free_gating_config(gating_cfg);
+
+ gating_cfg->base_time = its_base_time;
+ gating_cfg->cycle_time = max_cycle_time;
+ gating_cfg->num_entries = 0;
+
+ list_for_each_entry(rule, &priv->flow_block.rules, list) {
+ s64 time;
+ s64 rbt;
+
+ if (rule->type != SJA1105_RULE_VL)
+ continue;
+ if (rule->vl.type != SJA1105_VL_TIME_TRIGGERED)
+ continue;
+
+ /* Calculate the difference between this gating schedule's
+ * base time, and the base time of the gating schedule with the
+ * longest cycle time. We call it the relative base time (rbt).
+ */
+ rbt = future_base_time(rule->vl.base_time, rule->vl.cycle_time,
+ its_base_time);
+ rbt -= its_base_time;
+
+ time = rbt;
+
+ for (i = 0; i < rule->vl.num_entries; i++) {
+ u8 gate_state = rule->vl.entries[i].gate_state;
+ s64 entry_time = time;
+
+ while (entry_time < max_cycle_time) {
+ rc = sja1105_insert_gate_entry(gating_cfg, rule,
+ gate_state,
+ entry_time,
+ extack);
+ if (rc)
+ goto err;
+
+ entry_time += rule->vl.cycle_time;
+ }
+ time += rule->vl.entries[i].interval;
+ }
+ }
+
+ sja1105_gating_cfg_time_to_interval(gating_cfg, max_cycle_time);
+
+ return 0;
+err:
+ sja1105_free_gating_config(gating_cfg);
+ return rc;
+}
+
+int sja1105_vl_gate(struct sja1105_private *priv, int port,
+ struct netlink_ext_ack *extack, unsigned long cookie,
+ struct sja1105_key *key, u32 index, s32 prio,
+ u64 base_time, u64 cycle_time, u64 cycle_time_ext,
+ u32 num_entries, struct action_gate_entry *entries)
+{
+ struct sja1105_rule *rule = sja1105_rule_find(priv, cookie);
+ int ipv = -1;
+ int i, rc;
+ s32 rem;
+
+ if (cycle_time_ext) {
+ NL_SET_ERR_MSG_MOD(extack,
+ "Cycle time extension not supported");
+ return -EOPNOTSUPP;
+ }
+
+ div_s64_rem(base_time, sja1105_delta_to_ns(1), &rem);
+ if (rem) {
+ NL_SET_ERR_MSG_MOD(extack,
+ "Base time must be multiple of 200 ns");
+ return -ERANGE;
+ }
+
+ div_s64_rem(cycle_time, sja1105_delta_to_ns(1), &rem);
+ if (rem) {
+ NL_SET_ERR_MSG_MOD(extack,
+ "Cycle time must be multiple of 200 ns");
+ return -ERANGE;
+ }
+
+ if (dsa_port_is_vlan_filtering(dsa_to_port(priv->ds, port)) &&
+ key->type != SJA1105_KEY_VLAN_AWARE_VL) {
+ NL_SET_ERR_MSG_MOD(extack,
+ "Can only gate based on {DMAC, VID, PCP}");
+ return -EOPNOTSUPP;
+ } else if (key->type != SJA1105_KEY_VLAN_UNAWARE_VL) {
+ NL_SET_ERR_MSG_MOD(extack,
+ "Can only gate based on DMAC");
+ return -EOPNOTSUPP;
+ }
+
+ if (!rule) {
+ rule = kzalloc(sizeof(*rule), GFP_KERNEL);
+ if (!rule)
+ return -ENOMEM;
+
+ list_add(&rule->list, &priv->flow_block.rules);
+ rule->cookie = cookie;
+ rule->type = SJA1105_RULE_VL;
+ rule->key = *key;
+ rule->vl.type = SJA1105_VL_TIME_TRIGGERED;
+ rule->vl.sharindx = index;
+ rule->vl.base_time = base_time;
+ rule->vl.cycle_time = cycle_time;
+ rule->vl.num_entries = num_entries;
+ rule->vl.entries = kcalloc(num_entries,
+ sizeof(struct action_gate_entry),
+ GFP_KERNEL);
+ if (!rule->vl.entries) {
+ rc = -ENOMEM;
+ goto out;
+ }
+
+ for (i = 0; i < num_entries; i++) {
+ div_s64_rem(entries[i].interval,
+ sja1105_delta_to_ns(1), &rem);
+ if (rem) {
+ NL_SET_ERR_MSG_MOD(extack,
+ "Interval must be multiple of 200 ns");
+ rc = -ERANGE;
+ goto out;
+ }
+
+ if (!entries[i].interval) {
+ NL_SET_ERR_MSG_MOD(extack,
+ "Interval cannot be zero");
+ rc = -ERANGE;
+ goto out;
+ }
+
+ if (ns_to_sja1105_delta(entries[i].interval) >
+ SJA1105_TAS_MAX_DELTA) {
+ NL_SET_ERR_MSG_MOD(extack,
+ "Maximum interval is 52 ms");
+ rc = -ERANGE;
+ goto out;
+ }
+
+ if (entries[i].maxoctets != -1) {
+ NL_SET_ERR_MSG_MOD(extack,
+ "Cannot offload IntervalOctetMax");
+ rc = -EOPNOTSUPP;
+ goto out;
+ }
+
+ if (ipv == -1) {
+ ipv = entries[i].ipv;
+ } else if (ipv != entries[i].ipv) {
+ NL_SET_ERR_MSG_MOD(extack,
+ "Only support a single IPV per VL");
+ rc = -EOPNOTSUPP;
+ goto out;
+ }
+
+ rule->vl.entries[i] = entries[i];
+ }
+
+ if (ipv == -1) {
+ if (key->type == SJA1105_KEY_VLAN_AWARE_VL)
+ ipv = key->vl.pcp;
+ else
+ ipv = 0;
+ }
+
+ /* TODO: support per-flow MTU */
+ rule->vl.maxlen = VLAN_ETH_FRAME_LEN + ETH_FCS_LEN;
+ rule->vl.ipv = ipv;
+ }
+
+ rule->port_mask |= BIT(port);
+
+ rc = sja1105_compose_gating_subschedule(priv, extack);
+ if (rc)
+ goto out;
+
+ rc = sja1105_init_virtual_links(priv, extack);
+ if (rc)
+ goto out;
+
+ if (sja1105_gating_check_conflicts(priv, -1, extack)) {
+ NL_SET_ERR_MSG_MOD(extack, "Conflict with tc-taprio schedule");
+ rc = -ERANGE;
+ goto out;
+ }
+
+out:
+ if (rc) {
+ rule->port_mask &= ~BIT(port);
+ if (!rule->port_mask) {
+ list_del(&rule->list);
+ kfree(rule->vl.entries);
+ kfree(rule);
+ }
+ }
+
+ return rc;
+}
+
+static int sja1105_find_vlid(struct sja1105_private *priv, int port,
+ struct sja1105_key *key)
+{
+ struct sja1105_vl_lookup_entry *vl_lookup;
+ struct sja1105_table *table;
+ int i;
+
+ if (WARN_ON(key->type != SJA1105_KEY_VLAN_AWARE_VL &&
+ key->type != SJA1105_KEY_VLAN_UNAWARE_VL))
+ return -1;
+
+ table = &priv->static_config.tables[BLK_IDX_VL_LOOKUP];
+ vl_lookup = table->entries;
+
+ for (i = 0; i < table->entry_count; i++) {
+ if (key->type == SJA1105_KEY_VLAN_AWARE_VL) {
+ if (vl_lookup[i].port == port &&
+ vl_lookup[i].macaddr == key->vl.dmac &&
+ vl_lookup[i].vlanid == key->vl.vid &&
+ vl_lookup[i].vlanprior == key->vl.pcp)
+ return i;
+ } else {
+ if (vl_lookup[i].port == port &&
+ vl_lookup[i].macaddr == key->vl.dmac)
+ return i;
+ }
+ }
+
+ return -1;
+}
+
+int sja1105_vl_stats(struct sja1105_private *priv, int port,
+ struct sja1105_rule *rule, struct flow_stats *stats,
+ struct netlink_ext_ack *extack)
+{
+ const struct sja1105_regs *regs = priv->info->regs;
+ u8 buf[SJA1105_SIZE_VL_STATUS] = {0};
+ u64 unreleased;
+ u64 timingerr;
+ u64 lengtherr;
+ int vlid, rc;
+ u64 pkts;
+
+ if (rule->vl.type != SJA1105_VL_TIME_TRIGGERED)
+ return 0;
+
+ vlid = sja1105_find_vlid(priv, port, &rule->key);
+ if (vlid < 0)
+ return 0;
+
+ rc = sja1105_xfer_buf(priv, SPI_READ, regs->vl_status + 2 * vlid, buf,
+ SJA1105_SIZE_VL_STATUS);
+ if (rc) {
+ NL_SET_ERR_MSG_MOD(extack, "SPI access failed");
+ return rc;
+ }
+
+ sja1105_unpack(buf, &timingerr, 31, 16, SJA1105_SIZE_VL_STATUS);
+ sja1105_unpack(buf, &unreleased, 15, 0, SJA1105_SIZE_VL_STATUS);
+ sja1105_unpack(buf, &lengtherr, 47, 32, SJA1105_SIZE_VL_STATUS);
+
+ pkts = timingerr + unreleased + lengtherr;
+
+ flow_stats_update(stats, 0, pkts - rule->vl.stats.pkts,
+ jiffies - rule->vl.stats.lastused,
+ FLOW_ACTION_HW_STATS_IMMEDIATE);
+
+ rule->vl.stats.pkts = pkts;
+ rule->vl.stats.lastused = jiffies;
+
+ return 0;
+}
diff --git a/drivers/net/dsa/sja1105/sja1105_vl.h b/drivers/net/dsa/sja1105/sja1105_vl.h
index 08ee5557b463..323fa0535af7 100644
--- a/drivers/net/dsa/sja1105/sja1105_vl.h
+++ b/drivers/net/dsa/sja1105/sja1105_vl.h
@@ -15,6 +15,16 @@ int sja1105_vl_delete(struct sja1105_private *priv, int port,
struct sja1105_rule *rule,
struct netlink_ext_ack *extack);
+int sja1105_vl_gate(struct sja1105_private *priv, int port,
+ struct netlink_ext_ack *extack, unsigned long cookie,
+ struct sja1105_key *key, u32 index, s32 prio,
+ u64 base_time, u64 cycle_time, u64 cycle_time_ext,
+ u32 num_entries, struct action_gate_entry *entries);
+
+int sja1105_vl_stats(struct sja1105_private *priv, int port,
+ struct sja1105_rule *rule, struct flow_stats *stats,
+ struct netlink_ext_ack *extack);
+
#else
static inline int sja1105_vl_redirect(struct sja1105_private *priv, int port,
@@ -36,6 +46,27 @@ static inline int sja1105_vl_delete(struct sja1105_private *priv,
return -EOPNOTSUPP;
}
+static inline int sja1105_vl_gate(struct sja1105_private *priv, int port,
+ struct netlink_ext_ack *extack,
+ unsigned long cookie,
+ struct sja1105_key *key, u32 index, s32 prio,
+ u64 base_time, u64 cycle_time,
+ u64 cycle_time_ext, u32 num_entries,
+ struct action_gate_entry *entries)
+{
+ NL_SET_ERR_MSG_MOD(extack, "Virtual Links not compiled in");
+ return -EOPNOTSUPP;
+}
+
+static inline int sja1105_vl_stats(struct sja1105_private *priv, int port,
+ struct sja1105_rule *rule,
+ struct flow_stats *stats,
+ struct netlink_ext_ack *extack)
+{
+ NL_SET_ERR_MSG_MOD(extack, "Virtual Links not compiled in");
+ return -EOPNOTSUPP;
+}
+
#endif /* IS_ENABLED(CONFIG_NET_DSA_SJA1105_VL) */
#endif /* _SJA1105_VL_H */
--
2.17.1
^ permalink raw reply related [flat|nested] 9+ messages in thread
* [PATCH v3 net-next 6/6] docs: net: dsa: sja1105: document intended usage of virtual links
2020-05-05 19:20 [PATCH v3 net-next 0/6] tc-gate offload for SJA1105 DSA switch Vladimir Oltean
` (4 preceding siblings ...)
2020-05-05 19:20 ` [PATCH v3 net-next 5/6] net: dsa: sja1105: implement tc-gate using time-triggered " Vladimir Oltean
@ 2020-05-05 19:20 ` Vladimir Oltean
2020-05-08 0:32 ` [PATCH v3 net-next 0/6] tc-gate offload for SJA1105 DSA switch David Miller
6 siblings, 0 replies; 9+ messages in thread
From: Vladimir Oltean @ 2020-05-05 19:20 UTC (permalink / raw)
To: netdev; +Cc: andrew, f.fainelli, vivien.didelot, vinicius.gomes, po.liu
From: Vladimir Oltean <vladimir.oltean@nxp.com>
Add some verbiage describing how the hardware features of the switch are
exposed to users through tc-flower.
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
Changes from v2:
None.
Changes from v1:
None.
Changes from RFC:
Patch is new.
Documentation/networking/dsa/sja1105.rst | 116 +++++++++++++++++++++++
1 file changed, 116 insertions(+)
diff --git a/Documentation/networking/dsa/sja1105.rst b/Documentation/networking/dsa/sja1105.rst
index 64553d8d91cb..34581629dd3f 100644
--- a/Documentation/networking/dsa/sja1105.rst
+++ b/Documentation/networking/dsa/sja1105.rst
@@ -230,6 +230,122 @@ simultaneously on two ports. The driver checks the consistency of the schedules
against this restriction and errors out when appropriate. Schedule analysis is
needed to avoid this, which is outside the scope of the document.
+Routing actions (redirect, trap, drop)
+--------------------------------------
+
+The switch is able to offload flow-based redirection of packets to a set of
+destination ports specified by the user. Internally, this is implemented by
+making use of Virtual Links, a TTEthernet concept.
+
+The driver supports 2 types of keys for Virtual Links:
+
+- VLAN-aware virtual links: these match on destination MAC address, VLAN ID and
+ VLAN PCP.
+- VLAN-unaware virtual links: these match on destination MAC address only.
+
+The VLAN awareness state of the bridge (vlan_filtering) cannot be changed while
+there are virtual link rules installed.
+
+Composing multiple actions inside the same rule is supported. When only routing
+actions are requested, the driver creates a "non-critical" virtual link. When
+the action list also contains tc-gate (more details below), the virtual link
+becomes "time-critical" (draws frame buffers from a reserved memory partition,
+etc).
+
+The 3 routing actions that are supported are "trap", "drop" and "redirect".
+
+Example 1: send frames received on swp2 with a DA of 42:be:24:9b:76:20 to the
+CPU and to swp3. This type of key (DA only) when the port's VLAN awareness
+state is off::
+
+ tc qdisc add dev swp2 clsact
+ tc filter add dev swp2 ingress flower skip_sw dst_mac 42:be:24:9b:76:20 \
+ action mirred egress redirect dev swp3 \
+ action trap
+
+Example 2: drop frames received on swp2 with a DA of 42:be:24:9b:76:20, a VID
+of 100 and a PCP of 0::
+
+ tc filter add dev swp2 ingress protocol 802.1Q flower skip_sw \
+ dst_mac 42:be:24:9b:76:20 vlan_id 100 vlan_prio 0 action drop
+
+Time-based ingress policing
+---------------------------
+
+The TTEthernet hardware abilities of the switch can be constrained to act
+similarly to the Per-Stream Filtering and Policing (PSFP) clause specified in
+IEEE 802.1Q-2018 (formerly 802.1Qci). This means it can be used to perform
+tight timing-based admission control for up to 1024 flows (identified by a
+tuple composed of destination MAC address, VLAN ID and VLAN PCP). Packets which
+are received outside their expected reception window are dropped.
+
+This capability can be managed through the offload of the tc-gate action. As
+routing actions are intrinsic to virtual links in TTEthernet (which performs
+explicit routing of time-critical traffic and does not leave that in the hands
+of the FDB, flooding etc), the tc-gate action may never appear alone when
+asking sja1105 to offload it. One (or more) redirect or trap actions must also
+follow along.
+
+Example: create a tc-taprio schedule that is phase-aligned with a tc-gate
+schedule (the clocks must be synchronized by a 1588 application stack, which is
+outside the scope of this document). No packet delivered by the sender will be
+dropped. Note that the reception window is larger than the transmission window
+(and much more so, in this example) to compensate for the packet propagation
+delay of the link (which can be determined by the 1588 application stack).
+
+Receiver (sja1105)::
+
+ tc qdisc add dev swp2 clsact
+ now=$(phc_ctl /dev/ptp1 get | awk '/clock time is/ {print $5}') && \
+ sec=$(echo $now | awk -F. '{print $1}') && \
+ base_time="$(((sec + 2) * 1000000000))" && \
+ echo "base time ${base_time}"
+ tc filter add dev swp2 ingress flower skip_sw \
+ dst_mac 42:be:24:9b:76:20 \
+ action gate base-time ${base_time} \
+ sched-entry OPEN 60000 -1 -1 \
+ sched-entry CLOSE 40000 -1 -1 \
+ action trap
+
+Sender::
+
+ now=$(phc_ctl /dev/ptp0 get | awk '/clock time is/ {print $5}') && \
+ sec=$(echo $now | awk -F. '{print $1}') && \
+ base_time="$(((sec + 2) * 1000000000))" && \
+ echo "base time ${base_time}"
+ tc qdisc add dev eno0 parent root taprio \
+ num_tc 8 \
+ map 0 1 2 3 4 5 6 7 \
+ queues 1@0 1@1 1@2 1@3 1@4 1@5 1@6 1@7 \
+ base-time ${base_time} \
+ sched-entry S 01 50000 \
+ sched-entry S 00 50000 \
+ flags 2
+
+The engine used to schedule the ingress gate operations is the same that the
+one used for the tc-taprio offload. Therefore, the restrictions regarding the
+fact that no two gate actions (either tc-gate or tc-taprio gates) may fire at
+the same time (during the same 200 ns slot) still apply.
+
+To come in handy, it is possible to share time-triggered virtual links across
+more than 1 ingress port, via flow blocks. In this case, the restriction of
+firing at the same time does not apply because there is a single schedule in
+the system, that of the shared virtual link::
+
+ tc qdisc add dev swp2 ingress_block 1 clsact
+ tc qdisc add dev swp3 ingress_block 1 clsact
+ tc filter add block 1 flower skip_sw dst_mac 42:be:24:9b:76:20 \
+ action gate index 2 \
+ base-time 0 \
+ sched-entry OPEN 50000000 -1 -1 \
+ sched-entry CLOSE 50000000 -1 -1 \
+ action trap
+
+Hardware statistics for each flow are also available ("pkts" counts the number
+of dropped frames, which is a sum of frames dropped due to timing violations,
+lack of destination ports and MTU enforcement checks). Byte-level counters are
+not available.
+
Device Tree bindings and board design
=====================================
--
2.17.1
^ permalink raw reply related [flat|nested] 9+ messages in thread
* Re: [PATCH v3 net-next 1/6] net: dsa: introduce a dsa_port_from_netdev public helper
2020-05-05 19:20 ` [PATCH v3 net-next 1/6] net: dsa: introduce a dsa_port_from_netdev public helper Vladimir Oltean
@ 2020-05-06 15:55 ` Vivien Didelot
0 siblings, 0 replies; 9+ messages in thread
From: Vivien Didelot @ 2020-05-06 15:55 UTC (permalink / raw)
To: Vladimir Oltean; +Cc: netdev, andrew, f.fainelli, vinicius.gomes, po.liu
On Tue, 5 May 2020 22:20:52 +0300, Vladimir Oltean <olteanv@gmail.com> wrote:
> From: Vladimir Oltean <vladimir.oltean@nxp.com>
>
> As its implementation shows, this is synonimous with calling
> dsa_slave_dev_check followed by dsa_slave_to_port, so it is quite simple
> already and provides functionality which is already there.
>
> However there is now a need for these functions outside dsa_priv.h, for
> example in drivers that perform mirroring and redirection through
> tc-flower offloads (they are given raw access to the flow_cls_offload
> structure), where they need to call this function on act->dev.
>
> But simply exporting dsa_slave_to_port would make it non-inline and
> would result in an extra function call in the hotpath, as can be seen
> for example in sja1105:
>
> Before:
>
> 000006dc <sja1105_xmit>:
> {
> 6dc: e92d4ff0 push {r4, r5, r6, r7, r8, r9, sl, fp, lr}
> 6e0: e1a04000 mov r4, r0
> 6e4: e591958c ldr r9, [r1, #1420] ; 0x58c <- Inline dsa_slave_to_port
> 6e8: e1a05001 mov r5, r1
> 6ec: e24dd004 sub sp, sp, #4
> u16 tx_vid = dsa_8021q_tx_vid(dp->ds, dp->index);
> 6f0: e1c901d8 ldrd r0, [r9, #24]
> 6f4: ebfffffe bl 0 <dsa_8021q_tx_vid>
> 6f4: R_ARM_CALL dsa_8021q_tx_vid
> u8 pcp = netdev_txq_to_tc(netdev, queue_mapping);
> 6f8: e1d416b0 ldrh r1, [r4, #96] ; 0x60
> u16 tx_vid = dsa_8021q_tx_vid(dp->ds, dp->index);
> 6fc: e1a08000 mov r8, r0
>
> After:
>
> 000006e4 <sja1105_xmit>:
> {
> 6e4: e92d4ff0 push {r4, r5, r6, r7, r8, r9, sl, fp, lr}
> 6e8: e1a04000 mov r4, r0
> 6ec: e24dd004 sub sp, sp, #4
> struct dsa_port *dp = dsa_slave_to_port(netdev);
> 6f0: e1a00001 mov r0, r1
> {
> 6f4: e1a05001 mov r5, r1
> struct dsa_port *dp = dsa_slave_to_port(netdev);
> 6f8: ebfffffe bl 0 <dsa_slave_to_port>
> 6f8: R_ARM_CALL dsa_slave_to_port
> 6fc: e1a09000 mov r9, r0
> u16 tx_vid = dsa_8021q_tx_vid(dp->ds, dp->index);
> 700: e1c001d8 ldrd r0, [r0, #24]
> 704: ebfffffe bl 0 <dsa_8021q_tx_vid>
> 704: R_ARM_CALL dsa_8021q_tx_vid
>
> Because we want to avoid possible performance regressions, introduce
> this new function which is designed to be public.
>
> Suggested-by: Vivien Didelot <vivien.didelot@gmail.com>
> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Vivien Didelot <vivien.didelot@gmail.com>
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH v3 net-next 0/6] tc-gate offload for SJA1105 DSA switch
2020-05-05 19:20 [PATCH v3 net-next 0/6] tc-gate offload for SJA1105 DSA switch Vladimir Oltean
` (5 preceding siblings ...)
2020-05-05 19:20 ` [PATCH v3 net-next 6/6] docs: net: dsa: sja1105: document intended usage of " Vladimir Oltean
@ 2020-05-08 0:32 ` David Miller
6 siblings, 0 replies; 9+ messages in thread
From: David Miller @ 2020-05-08 0:32 UTC (permalink / raw)
To: olteanv
Cc: netdev, andrew, f.fainelli, vivien.didelot, vinicius.gomes, po.liu
From: Vladimir Oltean <olteanv@gmail.com>
Date: Tue, 5 May 2020 22:20:51 +0300
> From: Vladimir Oltean <vladimir.oltean@nxp.com>
>
> Expose the TTEthernet hardware features of the switch using standard
> tc-flower actions: trap, drop, redirect and gate.
>
> v1 was submitted at:
> https://patchwork.ozlabs.org/project/netdev/cover/20200503211035.19363-1-olteanv@gmail.com/
>
> v2 was submitted at:
> https://patchwork.ozlabs.org/project/netdev/cover/20200503211035.19363-1-olteanv@gmail.com/
>
> Changes in v3:
> Made sure there are no compilation warnings when
> CONFIG_NET_DSA_SJA1105_TAS or CONFIG_NET_DSA_SJA1105_VL are disabled.
>
> Changes in v2:
> Using a newly introduced dsa_port_from_netdev public helper.
Series applied, thanks.
^ permalink raw reply [flat|nested] 9+ messages in thread
end of thread, other threads:[~2020-05-08 0:32 UTC | newest]
Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-05-05 19:20 [PATCH v3 net-next 0/6] tc-gate offload for SJA1105 DSA switch Vladimir Oltean
2020-05-05 19:20 ` [PATCH v3 net-next 1/6] net: dsa: introduce a dsa_port_from_netdev public helper Vladimir Oltean
2020-05-06 15:55 ` Vivien Didelot
2020-05-05 19:20 ` [PATCH v3 net-next 2/6] net: dsa: sja1105: add static tables for virtual links Vladimir Oltean
2020-05-05 19:20 ` [PATCH v3 net-next 3/6] net: dsa: sja1105: make room for virtual link parsing in flower offload Vladimir Oltean
2020-05-05 19:20 ` [PATCH v3 net-next 4/6] net: dsa: sja1105: support flow-based redirection via virtual links Vladimir Oltean
2020-05-05 19:20 ` [PATCH v3 net-next 5/6] net: dsa: sja1105: implement tc-gate using time-triggered " Vladimir Oltean
2020-05-05 19:20 ` [PATCH v3 net-next 6/6] docs: net: dsa: sja1105: document intended usage of " Vladimir Oltean
2020-05-08 0:32 ` [PATCH v3 net-next 0/6] tc-gate offload for SJA1105 DSA switch David Miller
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).