Skip to contentSkip to conversation

sfc: fix error unwinds in TC offload

PATCH Applied [email protected] 3 participants 3 messages May 30, 2023 – Jun 1, 2023

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 sfc: fix error unwinds in TC offload

From: Edward Cree <[email protected]>

Failure ladders weren't exactly unwinding what the function had done up
to that point; most seriously, when we encountered an already offloaded rule, the failure path tried to remove the new rule from the hashtable, which would in fact remove the already-present 'old' rule (since it has the same key) from the table, and leak its resources.

Fixes d902e1a737d4 ("sfc: bare bones TC offload on EF100")
Fixes 17654d84b47c ("sfc: add offloading of 'foreign' TC (decap) rules")
---
 drivers/net/ethernet/sfc/tc.c | 27 ++++++++++++---------------
 1 file changed, 12 insertions(+), 15 deletions(-)
drivers/net/ethernet/sfc/tc.c +12 -15
index 0327639a628a..c004443c1d58 100644
@@ -624,13 +624,12 @@ static int efx_tc_flower_replace_foreign(struct efx_nic *efx,
 	if (!found) { /* We don't care. */
 		netif_dbg(efx, drv, efx->net_dev,
 			  "Ignoring foreign filter that doesn't egdev us\n");
-		rc = -EOPNOTSUPP;
-		goto release;
+		return -EOPNOTSUPP;
 	}
 
 	rc = efx_mae_match_check_caps(efx, &match.mask, NULL);
 	if (rc)
-		goto release;
+		return rc;
 
 	if (efx_tc_match_is_encap(&match.mask)) {
 		enum efx_encap_type type;
@@ -639,8 +638,7 @@ static int efx_tc_flower_replace_foreign(struct efx_nic *efx,
 		if (type == EFX_ENCAP_TYPE_NONE) {
 			NL_SET_ERR_MSG_MOD(extack,
 					   "Egress encap match on unsupported tunnel device");
-			rc = -EOPNOTSUPP;
-			goto release;
+			return -EOPNOTSUPP;
 		}
 
 		rc = efx_mae_check_encap_type_supported(efx, type);
@@ -648,25 +646,24 @@ static int efx_tc_flower_replace_foreign(struct efx_nic *efx,
 			NL_SET_ERR_MSG_FMT_MOD(extack,
 					       "Firmware reports no support for %s encap match",
 					       efx_tc_encap_type_name(type));
-			goto release;
+			return rc;
 		}
 
 		rc = efx_tc_flower_record_encap_match(efx, &match, type,
 						      extack);
 		if (rc)
-			goto release;
+			return rc;
 	} else {
 		/* This is not a tunnel decap rule, ignore it */
 		netif_dbg(efx, drv, efx->net_dev,
 			  "Ignoring foreign filter without encap match\n");
-		rc = -EOPNOTSUPP;
-		goto release;
+		return -EOPNOTSUPP;
 	}
 
 	rule = kzalloc(sizeof(*rule), GFP_USER);
 	if (!rule) {
 		rc = -ENOMEM;
-		goto release;
+		goto out_free;
 	}
 	INIT_LIST_HEAD(&rule->acts.list);
 	rule->cookie = tc->cookie;
@@ -678,7 +675,7 @@ static int efx_tc_flower_replace_foreign(struct efx_nic *efx,
 			  "Ignoring already-offloaded rule (cookie %lx)\n",
 			  tc->cookie);
 		rc = -EEXIST;
-		goto release;
+		goto out_free;
 	}
 
 	act = kzalloc(sizeof(*act), GFP_USER);
@@ -843,6 +840,7 @@ static int efx_tc_flower_replace_foreign(struct efx_nic *efx,
 				       efx_tc_match_action_ht_params);
 		efx_tc_free_action_set_list(efx, &rule->acts, false);
 	}
+out_free:
 	kfree(rule);
 	if (match.encap)
 		efx_tc_flower_release_encap_match(efx, match.encap);
@@ -899,8 +897,7 @@ static int efx_tc_flower_replace(struct efx_nic *efx,
 		return rc;
 	if (efx_tc_match_is_encap(&match.mask)) {
 		NL_SET_ERR_MSG_MOD(extack, "Ingress enc_key matches not supported");
-		rc = -EOPNOTSUPP;
-		goto release;
+		return -EOPNOTSUPP;
 	}
 
 	if (tc->common.chain_index) {
@@ -924,9 +921,9 @@ static int efx_tc_flower_replace(struct efx_nic *efx,
 	if (old) {
 		netif_dbg(efx, drv, efx->net_dev,
 			  "Already offloaded rule (cookie %lx)\n", tc->cookie);
-		rc = -EEXIST;
 		NL_SET_ERR_MSG_MOD(extack, "Rule already offloaded");
-		goto release;
+		kfree(rule);
+		return -EEXIST;
 	}
 
 	/* Parse actions */
Reviewed-by on v1 1/1
#1 lore raw
Quoted 8 lines · On Tue, May 30, 2023 at 09:25:27PM +0100, [email protected] wrote:

From: Edward Cree <[email protected]>

Failure ladders weren't exactly unwinding what the function had done up
to that point; most seriously, when we encountered an already offloaded rule, the failure path tried to remove the new rule from the hashtable, which would in fact remove the already-present 'old' rule (since it has the same key) from the table, and leak its resources.

Quoted 6 lines
Fixes d902e1a737d4 ("sfc: bare bones TC offload on EF100")
Fixes 17654d84b47c ("sfc: add offloading of 'foreign' TC (decap) rules")
#2 lore raw
In reply to [email protected]
This patch was applied to netdev/net.git (main) by Jakub Kicinski <[email protected]>:

Hello:

This patch was applied to netdev/net.git (main)
by Jakub Kicinski <[email protected]>:

Quoted 9 lines · On Tue, 30 May 2023 21:25:27 +0100 you wrote:

From: Edward Cree <[email protected]>

Failure ladders weren't exactly unwinding what the function had done up
to that point; most seriously, when we encountered an already offloaded rule, the failure path tried to remove the new rule from the hashtable, which would in fact remove the already-present 'old' rule (since it has the same key) from the table, and leak its resources.

[...]

Here is the summary with links:
- [net] sfc: fix error unwinds in TC offload

    https://git.kernel.org/netdev/net/c/622ab656344a

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html