Can net interfaces efficiently transmit data in extpg mbufs?

From: Rick Macklem <>
Date: Sun, 22 May 2022 22:33:16 UTC

The current NFS code knows how to optionally put the large RPC
messages (read/readdir replies and write requests) in extpg mbufs
instead of mbuf clusters.

This was done so that the data did not need to be copied when the
KTLS (which requires extpg mbufs) is being used.

However, I am wondering if putting the data in extpg mbufs might
be a performance win for non-TLS use for some net interfaces?

ps: (I do know that some network interfaces cannot handle extpg mbufs
 and, for them, the data must be copied to mbuf clusters. However,
 I am guessing that some network interface drivers can handle data
 to be sent in extpg mbufs.)