Merge branch 'net-make-rss-rxnfc-semantics-more-explicit'

Edward Cree says:

====================
net: make RSS+RXNFC semantics more explicit

The original semantics of ntuple filters with FLOW_RSS were not
 fully understood by all drivers, some ignoring the ring_cookie from
 the flow rule.  Require this support to be explicitly declared by
 the driver for filters relying on it to be inserted, and add self-
 test coverage for this functionality.
Also teach ethtool_check_max_channel() about this.
====================

Link: https://patch.msgid.link/cover.1731499021.git.ecree.xilinx@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

6 files changed, 113 insertions, 19 deletions

diff --git a/drivers/net/ethernet/sfc/ef100_ethtool.c b/drivers/net/ethernet/sfc/ef100_ethtool.c
index 5c2551369812..6c3b74000d3b 100644
--- a/drivers/net/ethernet/sfc/ef100_ethtool.c
+++ b/drivers/net/ethernet/sfc/ef100_ethtool.c
@@ -59,6 +59,7 @@ const struct ethtool_ops ef100_ethtool_ops = {
 	.get_rxfh_indir_size	= efx_ethtool_get_rxfh_indir_size,
 	.get_rxfh_key_size	= efx_ethtool_get_rxfh_key_size,
 	.rxfh_per_ctx_key	= true,
+	.cap_rss_rxnfc_adds	= true,
 	.rxfh_priv_size		= sizeof(struct efx_rss_context_priv),
 	.get_rxfh		= efx_ethtool_get_rxfh,
 	.set_rxfh		= efx_ethtool_set_rxfh,
diff --git a/drivers/net/ethernet/sfc/ethtool.c b/drivers/net/ethernet/sfc/ethtool.c
index bb1930818beb..83d715544f7f 100644
--- a/drivers/net/ethernet/sfc/ethtool.c
+++ b/drivers/net/ethernet/sfc/ethtool.c
@@ -263,6 +263,7 @@ const struct ethtool_ops efx_ethtool_ops = {
 	.get_rxfh_indir_size	= efx_ethtool_get_rxfh_indir_size,
 	.get_rxfh_key_size	= efx_ethtool_get_rxfh_key_size,
 	.rxfh_per_ctx_key	= true,
+	.cap_rss_rxnfc_adds	= true,
 	.rxfh_priv_size		= sizeof(struct efx_rss_context_priv),
 	.get_rxfh		= efx_ethtool_get_rxfh,
 	.set_rxfh		= efx_ethtool_set_rxfh,
diff --git a/include/linux/ethtool.h b/include/linux/ethtool.h
index 1199e308c8dd..299280c94d07 100644
--- a/include/linux/ethtool.h
+++ b/include/linux/ethtool.h
@@ -734,6 +734,9 @@ struct kernel_ethtool_ts_info {
  * @rxfh_per_ctx_key: device supports setting different RSS key for each
  *	additional context. Netlink API should report hfunc, key, and input_xfrm
  *	for every context, not just context 0.
+ * @cap_rss_rxnfc_adds: device supports nonzero ring_cookie in filters with
+ *	%FLOW_RSS flag; the queue ID from the filter is added to the value from
+ *	the indirection table to determine the delivery queue.
  * @rxfh_indir_space: max size of RSS indirection tables, if indirection table
  *	size as returned by @get_rxfh_indir_size may change during lifetime
  *	of the device. Leave as 0 if the table size is constant.
@@ -956,6 +959,7 @@ struct ethtool_ops {
 	u32     cap_rss_ctx_supported:1;
 	u32	cap_rss_sym_xor_supported:1;
 	u32	rxfh_per_ctx_key:1;
+	u32	cap_rss_rxnfc_adds:1;
 	u32	rxfh_indir_space;
 	u16	rxfh_key_space;
 	u16	rxfh_priv_size;
diff --git a/net/ethtool/common.c b/net/ethtool/common.c
index 0d62363dbd9d..05ce4f8080b3 100644
--- a/net/ethtool/common.c
+++ b/net/ethtool/common.c
@@ -538,6 +538,20 @@ static int ethtool_get_rxnfc_rule_count(struct net_device *dev)
 	return info.rule_cnt;
 }
 
+/* Max offset for one RSS context */
+static u32 ethtool_get_rss_ctx_max_channel(struct ethtool_rxfh_context *ctx)
+{
+	u32 max_ring = 0;
+	u32 i, *tbl;
+
+	if (WARN_ON_ONCE(!ctx))
+		return 0;
+	tbl = ethtool_rxfh_context_indir(ctx);
+	for (i = 0; i < ctx->indir_size; i++)
+		max_ring = max(max_ring, tbl[i]);
+	return max_ring;
+}
+
 static int ethtool_get_max_rxnfc_channel(struct net_device *dev, u64 *max)
 {
 	const struct ethtool_ops *ops = dev->ethtool_ops;
@@ -574,10 +588,18 @@ static int ethtool_get_max_rxnfc_channel(struct net_device *dev, u64 *max)
 
 		if (rule_info.fs.ring_cookie != RX_CLS_FLOW_DISC &&
 		    rule_info.fs.ring_cookie != RX_CLS_FLOW_WAKE &&
-		    !(rule_info.flow_type & FLOW_RSS) &&
-		    !ethtool_get_flow_spec_ring_vf(rule_info.fs.ring_cookie))
-			max_ring =
-				max_t(u64, max_ring, rule_info.fs.ring_cookie);
+		    !ethtool_get_flow_spec_ring_vf(rule_info.fs.ring_cookie)) {
+			u64 ring = rule_info.fs.ring_cookie;
+
+			if (rule_info.flow_type & FLOW_RSS) {
+				struct ethtool_rxfh_context *ctx;
+
+				ctx = xa_load(&dev->ethtool->rss_ctx,
+					      rule_info.rss_context);
+				ring += ethtool_get_rss_ctx_max_channel(ctx);
+			}
+			max_ring = max_t(u64, max_ring, ring);
+		}
 	}
 
 	kvfree(info);
@@ -589,6 +611,7 @@ err_free_info:
 	return err;
 }
 
+/* Max offset across all of a device's RSS contexts */
 static u32 ethtool_get_max_rss_ctx_channel(struct net_device *dev)
 {
 	struct ethtool_rxfh_context *ctx;
@@ -596,13 +619,8 @@ static u32 ethtool_get_max_rss_ctx_channel(struct net_device *dev)
 	u32 max_ring = 0;
 
 	mutex_lock(&dev->ethtool->rss_lock);
-	xa_for_each(&dev->ethtool->rss_ctx, context, ctx) {
-		u32 i, *tbl;
-
-		tbl = ethtool_rxfh_context_indir(ctx);
-		for (i = 0; i < ctx->indir_size; i++)
-			max_ring = max(max_ring, tbl[i]);
-	}
+	xa_for_each(&dev->ethtool->rss_ctx, context, ctx)
+		max_ring = max(max_ring, ethtool_get_rss_ctx_max_channel(ctx));
 	mutex_unlock(&dev->ethtool->rss_lock);
 
 	return max_ring;
@@ -611,7 +629,7 @@ static u32 ethtool_get_max_rss_ctx_channel(struct net_device *dev)
 static u32 ethtool_get_max_rxfh_channel(struct net_device *dev)
 {
 	struct ethtool_rxfh_param rxfh = {};
-	u32 dev_size, current_max;
+	u32 dev_size, current_max = 0;
 	int ret;
 
 	/* While we do track whether RSS context has an indirection
diff --git a/net/ethtool/ioctl.c b/net/ethtool/ioctl.c
index 7da94e26ced6..d86399bcf223 100644
--- a/net/ethtool/ioctl.c
+++ b/net/ethtool/ioctl.c
@@ -992,6 +992,11 @@ static noinline_for_stack int ethtool_set_rxnfc(struct net_device *dev,
 	if (rc)
 		return rc;
 
+	/* Nonzero ring with RSS only makes sense if NIC adds them together */
+	if (info.flow_type & FLOW_RSS && !ops->cap_rss_rxnfc_adds &&
+	    ethtool_get_flow_spec_ring(info.fs.ring_cookie))
+		return -EINVAL;
+
 	if (ops->get_rxfh) {
 		struct ethtool_rxfh_param rxfh = {};
 
diff --git a/tools/testing/selftests/drivers/net/hw/rss_ctx.py b/tools/testing/selftests/drivers/net/hw/rss_ctx.py
index 29995586993c..0b49ce7ae678 100755
--- a/tools/testing/selftests/drivers/net/hw/rss_ctx.py
+++ b/tools/testing/selftests/drivers/net/hw/rss_ctx.py
@@ -215,7 +215,7 @@ def test_rss_queue_reconfigure(cfg, main_ctx=True):
         defer(ethtool, f"-X {cfg.ifname} default")
     else:
         other_key = 'noise'
-        flow = f"flow-type tcp{cfg.addr_ipver} dst-port {port} context {ctx_id}"
+        flow = f"flow-type tcp{cfg.addr_ipver} dst-ip {cfg.addr} dst-port {port} context {ctx_id}"
         ntuple = ethtool_create(cfg, "-N", flow)
         defer(ethtool, f"-N {cfg.ifname} delete {ntuple}")
 
@@ -238,6 +238,32 @@ def test_rss_queue_reconfigure(cfg, main_ctx=True):
     else:
         raise Exception(f"Driver didn't prevent us from deactivating a used queue (context {ctx_id})")
 
+    if not main_ctx:
+        ethtool(f"-L {cfg.ifname} combined 4")
+        flow = f"flow-type tcp{cfg.addr_ipver} dst-ip {cfg.addr} dst-port {port} context {ctx_id} action 1"
+        try:
+            # this targets queue 4, which doesn't exist
+            ntuple2 = ethtool_create(cfg, "-N", flow)
+        except CmdExitFailure:
+            pass
+        else:
+            raise Exception(f"Driver didn't prevent us from targeting a nonexistent queue (context {ctx_id})")
+        # change the table to target queues 0 and 2
+        ethtool(f"-X {cfg.ifname} {ctx_ref} weight 1 0 1 0")
+        # ntuple rule therefore targets queues 1 and 3
+        ntuple2 = ethtool_create(cfg, "-N", flow)
+        # should replace existing filter
+        ksft_eq(ntuple, ntuple2)
+        _send_traffic_check(cfg, port, ctx_ref, { 'target': (1, 3),
+                                                  'noise' : (0, 2) })
+        # Setting queue count to 3 should fail, queue 3 is used
+        try:
+            ethtool(f"-L {cfg.ifname} combined 3")
+        except CmdExitFailure:
+            pass
+        else:
+            raise Exception(f"Driver didn't prevent us from deactivating a used queue (context {ctx_id})")
+
 
 def test_rss_resize(cfg):
     """Test resizing of the RSS table.
@@ -429,7 +455,7 @@ def test_rss_context(cfg, ctx_cnt=1, create_with_cfg=None):
         ksft_eq(max(data['rss-indirection-table']), 2 + i * 2 + 1, "Unexpected context cfg: " + str(data))
 
         ports.append(rand_port())
-        flow = f"flow-type tcp{cfg.addr_ipver} dst-port {ports[i]} context {ctx_id}"
+        flow = f"flow-type tcp{cfg.addr_ipver} dst-ip {cfg.addr} dst-port {ports[i]} context {ctx_id}"
         ntuple = ethtool_create(cfg, "-N", flow)
         defer(ethtool, f"-N {cfg.ifname} delete {ntuple}")
 
@@ -516,7 +542,7 @@ def test_rss_context_out_of_order(cfg, ctx_cnt=4):
         ctx.append(defer(ethtool, f"-X {cfg.ifname} context {ctx_id} delete"))
 
         ports.append(rand_port())
-        flow = f"flow-type tcp{cfg.addr_ipver} dst-port {ports[i]} context {ctx_id}"
+        flow = f"flow-type tcp{cfg.addr_ipver} dst-ip {cfg.addr} dst-port {ports[i]} context {ctx_id}"
         ntuple_id = ethtool_create(cfg, "-N", flow)
         ntuple.append(defer(ethtool, f"-N {cfg.ifname} delete {ntuple_id}"))
 
@@ -569,7 +595,7 @@ def test_rss_context_overlap(cfg, other_ctx=0):
 
     port = rand_port()
     if other_ctx:
-        flow = f"flow-type tcp{cfg.addr_ipver} dst-port {port} context {other_ctx}"
+        flow = f"flow-type tcp{cfg.addr_ipver} dst-ip {cfg.addr} dst-port {port} context {other_ctx}"
         ntuple_id = ethtool_create(cfg, "-N", flow)
         ntuple = defer(ethtool, f"-N {cfg.ifname} delete {ntuple_id}")
 
@@ -587,7 +613,7 @@ def test_rss_context_overlap(cfg, other_ctx=0):
     # Now create a rule for context 1 and make sure traffic goes to a subset
     if other_ctx:
         ntuple.exec()
-    flow = f"flow-type tcp{cfg.addr_ipver} dst-port {port} context {ctx_id}"
+    flow = f"flow-type tcp{cfg.addr_ipver} dst-ip {cfg.addr} dst-port {port} context {ctx_id}"
     ntuple_id = ethtool_create(cfg, "-N", flow)
     defer(ethtool, f"-N {cfg.ifname} delete {ntuple_id}")
 
@@ -620,7 +646,7 @@ def test_delete_rss_context_busy(cfg):
 
     # utilize context from ntuple filter
     port = rand_port()
-    flow = f"flow-type tcp{cfg.addr_ipver} dst-port {port} context {ctx_id}"
+    flow = f"flow-type tcp{cfg.addr_ipver} dst-ip {cfg.addr} dst-port {port} context {ctx_id}"
     ntuple_id = ethtool_create(cfg, "-N", flow)
     defer(ethtool, f"-N {cfg.ifname} delete {ntuple_id}")
 
@@ -633,6 +659,45 @@ def test_delete_rss_context_busy(cfg):
         pass
 
 
+def test_rss_ntuple_addition(cfg):
+    """
+    Test that the queue offset (ring_cookie) of an ntuple rule is added
+    to the queue number read from the indirection table.
+    """
+
+    require_ntuple(cfg)
+
+    queue_cnt = len(_get_rx_cnts(cfg))
+    if queue_cnt < 4:
+        try:
+            ksft_pr(f"Increasing queue count {queue_cnt} -> 4")
+            ethtool(f"-L {cfg.ifname} combined 4")
+            defer(ethtool, f"-L {cfg.ifname} combined {queue_cnt}")
+        except:
+            raise KsftSkipEx("Not enough queues for the test")
+
+    # Use queue 0 for normal traffic
+    ethtool(f"-X {cfg.ifname} equal 1")
+    defer(ethtool, f"-X {cfg.ifname} default")
+
+    # create additional rss context
+    ctx_id = ethtool_create(cfg, "-X", "context new equal 2")
+    defer(ethtool, f"-X {cfg.ifname} context {ctx_id} delete")
+
+    # utilize context from ntuple filter
+    port = rand_port()
+    flow = f"flow-type tcp{cfg.addr_ipver} dst-ip {cfg.addr} dst-port {port} context {ctx_id} action 2"
+    try:
+        ntuple_id = ethtool_create(cfg, "-N", flow)
+    except CmdExitFailure:
+        raise KsftSkipEx("Ntuple filter with RSS and nonzero action not supported")
+    defer(ethtool, f"-N {cfg.ifname} delete {ntuple_id}")
+
+    _send_traffic_check(cfg, port, f"context {ctx_id}", { 'target': (2, 3),
+                                                          'empty' : (1,),
+                                                          'noise' : (0,) })
+
+
 def main() -> None:
     with NetDrvEpEnv(__file__, nsim_test=False) as cfg:
         cfg.ethnl = EthtoolFamily()
@@ -644,7 +709,7 @@ def main() -> None:
                   test_rss_context_dump, test_rss_context_queue_reconfigure,
                   test_rss_context_overlap, test_rss_context_overlap2,
                   test_rss_context_out_of_order, test_rss_context4_create_with_cfg,
-                  test_delete_rss_context_busy],
+                  test_delete_rss_context_busy, test_rss_ntuple_addition],
                  args=(cfg, ))
     ksft_exit()