Differences

This shows you the differences between two versions of the page.

--- mywiki:linux:checksum [2016/04/06 17:48] – [Linux Checksum] shaoguoh
+++ mywiki:linux:checksum [2019/09/15 18:55] (current) – external edit 127.0.0.1
@@ Line 5: / Line 5: @@
 | ip_fast_csum | |
 | csum_fold | |
-|
+| skb_checksum | |
+| csum_tcpudp_magic | |
+| nf_ip_checksum |
+| csum_tcpudp_nofold |
+| csum_replace4 |
+| csum_replace2 |
+| nf_nat_ipv4_csum_recalc |
+| csum_tcpudp_magic |
+skb->csum_start = skb_transport_header(skb) - skb->head;
+skb->csum_offset = offsetof(struct tcphdr, check);
+**TSO** is a hardware feature whereas **GSO** is a software feature.
+In the sense, TSO needs "the device" to break the frame into MTU sized segments.
+BUT, GSO is a strategy followed by kernel to avoid processing smaller packets throughout the stack to improve the porformance
+====== Linux Checksum offloading ======
+  - For bridging and non-udp/tcp packet, no so called offloading support in tx path
+  - For Fragmented packets, no checksum offloading support in tx path
+  - For UDP/TCP routing packet (incremental), no checksum offloading support in tx path. at the same time, no ip_summed flag change
+  - Only Linux local output UDP/TCP’s checksum can be offloaded in tx path
+  - For L2TP related tunnels, only inner UDP/TCP can be offloaded, outer UDP/TCP is done by linux itself in tx path.
+Below variables in skb is very important for checksum offloading:
+  - ip_summed (CHECKSUM_PARTIAL)
+  - encapsulation
+  - inner_network_header/network_header
+  - inner_transport_header/transport_header
 ====== Linux Checksum ======