[PATCH] cxgb4: flower: add support for fragmentation

[PATCH] cxgb4: flower: add support for fragmentation

[Harshita V Rajput]

This patch adds support for matching fragmented packets in tc flower
filters.

Previously, commit 93a8540aac72 ("cxgb4: flower: validate control flags")
added a check using flow_rule_match_has_control_flags() to reject
any rules with control flags, as the driver did not support
fragmentation at that time.

Now, with this patch, support for FLOW_DIS_IS_FRAGMENT is added:
- The driver checks for control flags using
  flow_rule_is_supp_control_flags(), as recommended in
  commit d11e63119432 ("flow_offload: add control flag checking helpers").
- If the fragmentation flag is present, the driver sets `fs->val.frag` and
  `fs->mask.frag` accordingly in the filter specification.

Since fragmentation is now supported, the earlier check that rejected all
control flags (flow_rule_match_has_control_flags()) has been removed.

Signed-off-by: Harshita V Rajput <harshitha.vr@chelsio.com>
Signed-off-by: Potnuri Bharat Teja <bharat@chelsio.com>
---
 .../ethernet/chelsio/cxgb4/cxgb4_tc_flower.c  | 40 +++++++++++--------
 1 file changed, 24 insertions(+), 16 deletions(-)

diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.c
index 0765d000eaef..e2b5554531b5 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_tc_flower.c
@@ -161,20 +161,9 @@ static struct ch_tc_flower_entry *ch_flower_lookup(struct adapter *adap,
 
 static void cxgb4_process_flow_match(struct net_device *dev,
 				     struct flow_rule *rule,
+				     u16 addr_type,
 				     struct ch_filter_specification *fs)
 {
-	u16 addr_type = 0;
-
-	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_CONTROL)) {
-		struct flow_match_control match;
-
-		flow_rule_match_control(rule, &match);
-		addr_type = match.key->addr_type;
-	} else if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_IPV4_ADDRS)) {
-		addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
-	} else if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_IPV6_ADDRS)) {
-		addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS;
-	}
 
 	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_BASIC)) {
 		struct flow_match_basic match;
@@ -327,9 +316,6 @@ static int cxgb4_validate_flow_match(struct netlink_ext_ack *extack,
 		return -EOPNOTSUPP;
 	}
 
-	if (flow_rule_match_has_control_flags(rule, extack))
-		return -EOPNOTSUPP;
-
 	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_BASIC)) {
 		struct flow_match_basic match;
 
@@ -858,6 +844,7 @@ int cxgb4_flow_rule_replace(struct net_device *dev, struct flow_rule *rule,
 {
 	struct adapter *adap = netdev2adap(dev);
 	struct filter_ctx ctx;
+	u16 addr_type = 0;
 	u8 inet_family;
 	int fidx, ret;
 
@@ -867,7 +854,28 @@ int cxgb4_flow_rule_replace(struct net_device *dev, struct flow_rule *rule,
 	if (cxgb4_validate_flow_match(extack, rule))
 		return -EOPNOTSUPP;
 
-	cxgb4_process_flow_match(dev, rule, fs);
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_CONTROL)) {
+		struct flow_match_control match;
+
+		flow_rule_match_control(rule, &match);
+		addr_type = match.key->addr_type;
+
+		if (match.mask->flags & FLOW_DIS_IS_FRAGMENT) {
+			fs->val.frag = match.key->flags & FLOW_DIS_IS_FRAGMENT;
+			fs->mask.frag = true;
+		}
+
+		if (!flow_rule_is_supp_control_flags(FLOW_DIS_IS_FRAGMENT,
+						     match.mask->flags, extack))
+			return -EOPNOTSUPP;
+
+	} else if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_IPV4_ADDRS)) {
+		addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
+	} else if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_IPV6_ADDRS)) {
+		addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS;
+	}
+
+	cxgb4_process_flow_match(dev, rule, addr_type, fs);
 	cxgb4_process_flow_actions(dev, &rule->action, fs);
 
 	fs->hash = is_filter_exact_match(adap, fs);
-- 
2.43.0

Re: [PATCH] cxgb4: flower: add support for fragmentation

[Simon Horman]

On Tue, Oct 28, 2025 at 01:22:55PM +0530, Harshita V Rajput wrote:
> This patch adds support for matching fragmented packets in tc flower
> filters.
> 
> Previously, commit 93a8540aac72 ("cxgb4: flower: validate control flags")
> added a check using flow_rule_match_has_control_flags() to reject
> any rules with control flags, as the driver did not support
> fragmentation at that time.
> 
> Now, with this patch, support for FLOW_DIS_IS_FRAGMENT is added:
> - The driver checks for control flags using
>   flow_rule_is_supp_control_flags(), as recommended in
>   commit d11e63119432 ("flow_offload: add control flag checking helpers").
> - If the fragmentation flag is present, the driver sets `fs->val.frag` and
>   `fs->mask.frag` accordingly in the filter specification.
> 
> Since fragmentation is now supported, the earlier check that rejected all
> control flags (flow_rule_match_has_control_flags()) has been removed.
> 
> Signed-off-by: Harshita V Rajput <harshitha.vr@chelsio.com>
> Signed-off-by: Potnuri Bharat Teja <bharat@chelsio.com>

Thanks for the comprehensive commit message.

Reviewed-by: Simon Horman <horms@kernel.org>

Re: [PATCH] cxgb4: flower: add support for fragmentation

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to netdev/net-next.git (main)
by Jakub Kicinski <kuba@kernel.org>:

On Tue, 28 Oct 2025 13:22:55 +0530 you wrote:
> This patch adds support for matching fragmented packets in tc flower
> filters.
> 
> Previously, commit 93a8540aac72 ("cxgb4: flower: validate control flags")
> added a check using flow_rule_match_has_control_flags() to reject
> any rules with control flags, as the driver did not support
> fragmentation at that time.
> 
> [...]

Here is the summary with links:
  - cxgb4: flower: add support for fragmentation
    https://git.kernel.org/netdev/net-next/c/0d0eb186421d

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html