The branch main has been updated by kp:
URL:
https://cgit.FreeBSD.org/src/commit/?id=fb69ed397e4f455d4c338edcfa210d763bb87189
commit fb69ed397e4f455d4c338edcfa210d763bb87189
Author: Kristof Provost <k...@freebsd.org>
AuthorDate: 2023-08-12 13:56:21 +0000
Commit: Kristof Provost <k...@freebsd.org>
CommitDate: 2023-08-12 13:56:21 +0000
Revert "if_vlan: do not enable LRO for bridge interaces"
This reverts commit 5f11a33ceeb385477cb22d9ad5941061c5a26be9.
As requested by Kevin Bowling. He explains:
> The subtle bug was that vlan_capabilities() in if_vlan was not obeying
> the requested mask from its IFCAP ioctl.
---
sys/net/if_vlan.c | 22 +++++++---------------
1 file changed, 7 insertions(+), 15 deletions(-)
diff --git a/sys/net/if_vlan.c b/sys/net/if_vlan.c
index 92e4e4247e3d..6aa872a19364 100644
--- a/sys/net/if_vlan.c
+++ b/sys/net/if_vlan.c
@@ -2067,22 +2067,14 @@ vlan_capabilities(struct ifvlan *ifv)
}
/*
- * If the parent interface is not a bridge and can do LRO and
- * checksum offloading on VLANs, then guess it may do LRO on VLANs.
- * False positive here cost nothing, while false negative may lead
- * to some confusions. According to Wikipedia:
- *
- * "LRO should not operate on machines acting as routers, as it breaks
- * the end-to-end principle and can significantly impact performance."
- *
- * The same reasoning applies to machines acting as bridges.
+ * If the parent interface can do LRO and checksum offloading on
+ * VLANs, then guess it may do LRO on VLANs. False positive here
+ * cost nothing, while false negative may lead to some confusions.
*/
- if (ifp->if_bridge == NULL) {
- if (p->if_capabilities & IFCAP_VLAN_HWCSUM)
- cap |= p->if_capabilities & IFCAP_LRO;
- if (p->if_capenable & IFCAP_VLAN_HWCSUM)
- ena |= p->if_capenable & IFCAP_LRO;
- }
+ if (p->if_capabilities & IFCAP_VLAN_HWCSUM)
+ cap |= p->if_capabilities & IFCAP_LRO;
+ if (p->if_capenable & IFCAP_VLAN_HWCSUM)
+ ena |= p->if_capenable & IFCAP_LRO;
/*
* If the parent interface can offload TCP connections over VLANs then
