Date: Sun, 22 May 2022 22:33:16 +0000 From: Rick Macklem <rmacklem@uoguelph.ca> To: freebsd-net <freebsd-net@freebsd.org> Subject: Can net interfaces efficiently transmit data in extpg mbufs? Message-ID: <YQBPR0101MB974247DA6698DEE079B920DADDD59@YQBPR0101MB9742.CANPRD01.PROD.OUTLOOK.COM>
next in thread | raw e-mail | index | archive | help
Hi,=0A= =0A= The current NFS code knows how to optionally put the large RPC=0A= messages (read/readdir replies and write requests) in extpg mbufs=0A= instead of mbuf clusters.=0A= =0A= This was done so that the data did not need to be copied when the=0A= KTLS (which requires extpg mbufs) is being used.=0A= =0A= However, I am wondering if putting the data in extpg mbufs might=0A= be a performance win for non-TLS use for some net interfaces?=0A= =0A= rick=0A= ps: (I do know that some network interfaces cannot handle extpg mbufs=0A= and, for them, the data must be copied to mbuf clusters. However,=0A= I am guessing that some network interface drivers can handle data=0A= to be sent in extpg mbufs.)=0A= =0A=
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?YQBPR0101MB974247DA6698DEE079B920DADDD59>