On Wed, Aug 5, 2015 at 5:38 PM, Jesse Brandeburg <jesse.brandeb...@intel.com> wrote: > On Wed, 5 Aug 2015 17:13:21 -0700 > Tom Herbert <t...@herbertland.com> wrote: > >> On Wed, Aug 5, 2015 at 4:52 PM, Jeff Kirsher >> <jeffrey.t.kirs...@intel.com> wrote: >> > From: Anjali Singhai Jain <anjali.sing...@intel.com> >> > if (vsi->back->flags & I40E_FLAG_WB_ON_ITR_CAPABLE) >> > tx_ring->flags = I40E_TXR_FLAGS_WB_ON_ITR; >> > + if (vsi->back->flags & I40E_FLAG_OUTER_UDP_CSUM_CAPABLE) >> > + tx_ring->flags |= I40E_TXR_FLAGS_OUTER_UDP_CSUM; >> >> Just curious... is there a difference between enabling the outer UDP >> checksum (of a tunnel) and just enabling checksum offload for UDP >> packets? > > Yes, the hardware knows the difference (or we actually tell it > the difference) between a UDP packet and a tunnel inside a UDP > packet.
There should be no difference between handling the UDP checksum for tunneling versus a non-tunnel packet with a UDP checksum. While VXLAN (RFC7348) allows for non-zero checksums to be not be verified, this fundamentally violates UDP specification (RFC1122). Neither is there any API that allows a driver to indicate to the stack that a non-zero UDP checksum is accepted but not verified, using checksum unnecessary for that would also be incorrect. Not to belabor the point, but if devices provide checksum-complete this complexity in drivers and devices needing to parse specific UDP encapsulations goes away. Tom -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
