Skip to contentSkip to conversation

Fix ntuple rules targeting default RSS

PATCH Superseded Gal Pressman 3 participants 10 messages Jun 9, 2025 – Jun 11, 2025

Subsystems (4)

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

v2 cover Fix ntuple rules targeting default RSS

This series addresses a regression in ethtool flow steering where rules targeting the default RSS context (context 0) were incorrectly rejected.

The default RSS context always exists but is not stored in the rss_ctx xarray like additional contexts. The current validation logic was checking for the existence of context 0 in this array, causing valid flow steering rules to be rejected.

This prevented configurations such as:
- High priority rules directing specific traffic to the default context
- Low priority catch-all rules directing remaining traffic to additional contexts

Patch 1 fixes the validation logic to skip the existence check for context 0.

Patch 2 adds a selftest that verifies this behavior.

Changelog -
v1->v2: https://lore.kernel.org/all/[email protected]/
* Reworded commit message.
* Added a selftest.

Gal Pressman (2):
net: ethtool: Don't check if RSS context exists in case of context 0
selftests: drv-net: rss_ctx: Add test for ntuple rules targeting
default RSS context

 net/ethtool/ioctl.c                           |  3 +-
 .../selftests/drivers/net/hw/rss_ctx.py       | 59 ++++++++++++++++++-
 2 files changed, 60 insertions(+), 2 deletions(-)
Signature · 1 line
2.40.1
#1 lore raw

PATCH v2 1/2 net: ethtool: Don't check if RSS context exists in case of context 0

Context 0 (default context) always exists, there is no need to check whether it exists or not when adding a flow steering rule.

The existing check fails when creating a flow steering rule for context 0 as it is not stored in the rss_ctx xarray.

For example:

$ ethtool --config-ntuple eth2 flow-type tcp4 dst-ip 194.237.147.23 dst-port 19983 context 0 loc 618
rmgr: Cannot insert RX class rule: Invalid argument
Cannot insert classification rule

An example usecase for this could be:
- A high-priority rule (loc 0) directing specific port traffic to context 0.
- A low-priority rule (loc 1) directing all other TCP traffic to context 1.

Fixes de7f7582dff2 ("net: ethtool: prevent flow steering to RSS contexts which don't exist")
Cc Jakub Kicinski <[email protected]>
---
 net/ethtool/ioctl.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
net/ethtool/ioctl.c +2 -1
index 39ec920f5de7..71c828d0bf31 100644
@@ -1083,7 +1083,8 @@ static noinline_for_stack int ethtool_set_rxnfc(struct net_device *dev,
 		    ethtool_get_flow_spec_ring(info.fs.ring_cookie))
 			return -EINVAL;
 
-		if (!xa_load(&dev->ethtool->rss_ctx, info.rss_context))
+		if (info.rss_context &&
+		    !xa_load(&dev->ethtool->rss_ctx, info.rss_context))
 			return -EINVAL;
 	}
 
Signature · 1 line
2.40.1
Reviewed-by on v2 1/2
#2 lore raw

Re: [PATCH net v2 1/2] net: ethtool: Don't check if RSS context exists in case of context 0

Quoted 17 lines · Gal Pressman

Context 0 (default context) always exists, there is no need to check whether it exists or not when adding a flow steering rule.

The existing check fails when creating a flow steering rule for context 0 as it is not stored in the rss_ctx xarray.

For example:

$ ethtool --config-ntuple eth2 flow-type tcp4 dst-ip 194.237.147.23 dst-port 19983 context 0 loc 618
rmgr: Cannot insert RX class rule: Invalid argument
Cannot insert classification rule

An example usecase for this could be:
- A high-priority rule (loc 0) directing specific port traffic to context 0.
- A low-priority rule (loc 1) directing all other TCP traffic to context 1.

Quoted 6 lines
Fixes de7f7582dff2 ("net: ethtool: prevent flow steering to RSS contexts which don't exist")
Cc Jakub Kicinski <[email protected]>

Thanks,
Sundeep

Quoted 21 lines
---
 net/ethtool/ioctl.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
net/ethtool/ioctl.c +2 -1
index 39ec920f5de7..71c828d0bf31 100644
@@ -1083,7 +1083,8 @@ static noinline_for_stack int ethtool_set_rxnfc(struct net_device *dev,
 		    ethtool_get_flow_spec_ring(info.fs.ring_cookie))
 			return -EINVAL;
 
-		if (!xa_load(&dev->ethtool->rss_ctx, info.rss_context))
+		if (info.rss_context &&
+		    !xa_load(&dev->ethtool->rss_ctx, info.rss_context))
 			return -EINVAL;
 	}
 
Signature · 1 line
2.40.1
maintainer
#3 lore raw

Re: [PATCH net v2 1/2] net: ethtool: Don't check if RSS context exists in case of context 0

In reply to Gal Pressman
Quoted 5 lines · On Mon, 9 Jun 2025 15:02:49 +0300 Gal Pressman wrote:

Did you receive a user report for this change in behavior?

#4 lore raw

Re: [PATCH net v2 1/2] net: ethtool: Don't check if RSS context exists in case of context 0

Quoted 8 lines · Jakub Kicinski
Quoted 5 lines · On Mon, 9 Jun 2025 15:02:49 +0300 Gal Pressman wrote:

Did you receive a user report for this change in behavior?

No.

maintainer
#5 lore raw

Re: [PATCH net v2 1/2] net: ethtool: Don't check if RSS context exists in case of context 0

Quoted 11 lines · On Tue, 10 Jun 2025 08:58:52 +0300 Gal Pressman wrote:
Quoted 8 lines · Jakub Kicinski
Quoted 5 lines · On Mon, 9 Jun 2025 15:02:49 +0300 Gal Pressman wrote:

Did you receive a user report for this change in behavior?

No.

Please make this clear in the commit message.

#6 lore raw

PATCH v2 2/2 selftests: drv-net: rss_ctx: Add test for ntuple rules targeting default RSS context

In reply to Gal Pressman

Add test_rss_default_context_rule() to verify that ntuple rules can correctly direct traffic to the default RSS context (context 0).

The test creates two ntuple rules with explicit location priorities:
- A high-priority rule (loc 0) directing specific port traffic to context 0.
- A low-priority rule (loc 1) directing all other TCP traffic to context 1.

This validates that:
1. Rules targeting the default context function properly.
2. Traffic steering works as expected when mixing default and additional RSS contexts.

The test was written by AI, and reviewed by humans.

---
 .../selftests/drivers/net/hw/rss_ctx.py       | 59 ++++++++++++++++++-
 1 file changed, 58 insertions(+), 1 deletion(-)
tools/testing/selftests/drivers/net/hw/rss_ctx.py +58 -1
index ca60ae325c22..6aef8f643ca4 100755
@@ -747,6 +747,62 @@ def test_rss_ntuple_addition(cfg):
                                                           'noise' : (0,) })
 
 
+def test_rss_default_context_rule(cfg):
+    """
+    Allocate a port, direct this port to context 0, then create a new RSS
+    context and steer all TCP traffic to it (context 1).  Verify that:
+      * Traffic to the specific port continues to use queues of the main
+        context (0/1).
+      * Traffic to any other TCP port is redirected to the new context
+        (queues 2/3).
+    """
+
+    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 queues 0 and 1 for the main context
+    ethtool(f"-X {cfg.ifname} equal 2")
+    defer(ethtool, f"-X {cfg.ifname} default")
+
+    # Create a new RSS context that uses queues 2 and 3
+    ctx_id = ethtool_create(cfg, "-X", "context new start 2 equal 2")
+    defer(ethtool, f"-X {cfg.ifname} context {ctx_id} delete")
+
+    # Generic low-priority rule: redirect all TCP traffic to the new context.
+    # Give it an explicit higher location number (lower priority).
+    flow_generic = f"flow-type tcp{cfg.addr_ipver} dst-ip {cfg.addr} context {ctx_id} loc 1"
+    ethtool(f"-N {cfg.ifname} {flow_generic}")
+    defer(ethtool, f"-N {cfg.ifname} delete 1")
+
+    # Specific high-priority rule for a random port that should stay on context 0.
+    # Assign loc 0 so it is evaluated before the generic rule.
+    port_main = rand_port()
+    flow_main = f"flow-type tcp{cfg.addr_ipver} dst-ip {cfg.addr} dst-port {port_main} context 0 loc 0"
+    ethtool(f"-N {cfg.ifname} {flow_main}")
+    defer(ethtool, f"-N {cfg.ifname} delete 0")
+
+    _ntuple_rule_check(cfg, 1, ctx_id)
+
+    # Verify that traffic matching the specific rule still goes to queues 0/1
+    _send_traffic_check(cfg, port_main, "context 0",
+                        { 'target': (0, 1),
+                          'empty' : (2, 3) })
+
+    # And that traffic for any other port is steered to the new context
+    port_other = rand_port()
+    _send_traffic_check(cfg, port_other, f"context {ctx_id}",
+                        { 'target': (2, 3),
+                          'noise' : (0, 1) })
+
+
 def main() -> None:
     with NetDrvEpEnv(__file__, nsim_test=False) as cfg:
         cfg.context_cnt = None
@@ -760,7 +816,8 @@ def main() -> None:
                   test_rss_context_overlap, test_rss_context_overlap2,
                   test_rss_context_out_of_order, test_rss_context4_create_with_cfg,
                   test_flow_add_context_missing,
-                  test_delete_rss_context_busy, test_rss_ntuple_addition],
+                  test_delete_rss_context_busy, test_rss_ntuple_addition,
+                  test_rss_default_context_rule],
                  args=(cfg, ))
     ksft_exit()
 
Signature · 1 line
2.40.1
maintainer
#7 lore raw

Re: [PATCH net v2 2/2] selftests: drv-net: rss_ctx: Add test for ntuple rules targeting default RSS context

Quoted 1 line · On Mon, 9 Jun 2025 15:02:50 +0300 Gal Pressman wrote:

The test was written by AI, and reviewed by humans.

Please use pylint and avoid adding new warnings:

tools/testing/selftests/drivers/net/hw/rss_ctx.py:788:0: C0301: Line too long (103/100) (line-too-long)
tools/testing/selftests/drivers/net/hw/rss_ctx.py:769:12: W0707: Consider explicitly re-raising using 'except Exception as exc' and 'raise KsftSkipEx('Not enough queues for the test') from exc' (raise-missing-from)

Signature · 1 line
pw-bot: cr
#8 lore raw

Re: [PATCH net v2 2/2] selftests: drv-net: rss_ctx: Add test for ntuple rules targeting default RSS context

Quoted 4 lines · Jakub Kicinski
Quoted 1 line · On Mon, 9 Jun 2025 15:02:50 +0300 Gal Pressman wrote:

The test was written by AI, and reviewed by humans.

Please use pylint and avoid adding new warnings:

Ack.

Quoted 2 lines

tools/testing/selftests/drivers/net/hw/rss_ctx.py:788:0: C0301: Line too long (103/100) (line-too-long)

Are you sure you want to blindly break this line?
It's a single string.

Quoted 1 line

tools/testing/selftests/drivers/net/hw/rss_ctx.py:769:12: W0707: Consider explicitly re-raising using 'except Exception as exc' and 'raise KsftSkipEx('Not enough queues for the test') from exc' (raise-missing-from)

Cool, didn't know python can do that.

maintainer
#9 lore raw

Re: [PATCH net v2 2/2] selftests: drv-net: rss_ctx: Add test for ntuple rules targeting default RSS context

Quoted 4 lines · On Wed, 11 Jun 2025 08:06:38 +0300 Gal Pressman wrote:
Quoted 1 line

tools/testing/selftests/drivers/net/hw/rss_ctx.py:788:0: C0301: Line too long (103/100) (line-too-long)

Are you sure you want to blindly break this line?
It's a single string.

Yeah, if it's a string leave it be. This one is just a check, the other one as a warning. Unhelpfully I don't see an option in pylint to add the bad lines to the report so its a bit hard to tell at a glance. The long strings and the lack of doc string in the main function are the two checks that we are definitely okay to ignore.