elaborate on why we don't support hardware checksum offloading, and probably

never will
This commit is contained in:
jdolecek 2020-04-23 15:54:36 +00:00
parent 0576a33000
commit 00eb0f15bf
1 changed files with 8 additions and 2 deletions

View File

@ -1,4 +1,4 @@
.\" $NetBSD: sk.4,v 1.23 2019/08/30 09:24:19 wiz Exp $
.\" $NetBSD: sk.4,v 1.24 2020/04/23 15:54:36 jdolecek Exp $
.\"
.\" Copyright (c) 2003, The NetBSD Foundation, Inc.
.\" All rights reserved.
@ -58,7 +58,7 @@
.\"
.\" $FreeBSD: src/share/man/man4/man4.i386/sk.4,v 1.3 1999/08/28 00:20:29 peter Exp $
.\"
.Dd August 24, 2019
.Dd April 23, 2020
.Dt SK 4
.Os
.Sh NAME
@ -285,6 +285,12 @@ Support for the Marvell Yukon-2 was added by
.An Mark Kettenis Aq Mt kettenis@openbsd.org .
.Sh BUGS
Support for checksum offload is unimplemented.
Particularly for Yukon-II hardware, there are multiple different
receive and transmit offload silicon bugs which have to be worked
around in software when using hardware offloading.
For this reason, support for hardware offloading
is not very desirable for therse controllers, and unlikely to be
ever implemented.
.Pp
Performance with at least some Marvell-based adapters is poor,
especially on loaded PCI buses or when the adapters are behind