diff mbox series

[iproute2] iplink: document XDP subcommand to force the XDP mode.

Message ID 20190213144030.15160-1-mcroce@redhat.com
State Accepted
Delegated to: stephen hemminger
Headers show
Series [iproute2] iplink: document XDP subcommand to force the XDP mode. | expand

Commit Message

Matteo Croce Feb. 13, 2019, 2:40 p.m. UTC
When attaching an eBPF program to a device, ip link can force the XDP mode
by using the xdp{generic,drv,offload} keyword instead of just 'xdp'.
Document this behaviour also in the help output.

Signed-off-by: Matteo Croce <mcroce@redhat.com>
Fixes: 14683814 ("bpf: add xdpdrv for requesting XDP driver mode")
Fixes: 1b5e8094 ("bpf: allow requesting XDP HW offload")
---
 ip/iplink.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Comments

Stephen Hemminger Feb. 13, 2019, 10:04 p.m. UTC | #1
On Wed, 13 Feb 2019 15:40:30 +0100
Matteo Croce <mcroce@redhat.com> wrote:

> When attaching an eBPF program to a device, ip link can force the XDP mode
> by using the xdp{generic,drv,offload} keyword instead of just 'xdp'.
> Document this behaviour also in the help output.
> 
> Signed-off-by: Matteo Croce <mcroce@redhat.com>
> Fixes: 14683814 ("bpf: add xdpdrv for requesting XDP driver mode")
> Fixes: 1b5e8094 ("bpf: allow requesting XDP HW offload")

Applied, thanks. 
The man page already has this as well.
Daniel Borkmann Feb. 13, 2019, 10:07 p.m. UTC | #2
On 02/13/2019 03:40 PM, Matteo Croce wrote:
> When attaching an eBPF program to a device, ip link can force the XDP mode
> by using the xdp{generic,drv,offload} keyword instead of just 'xdp'.
> Document this behaviour also in the help output.
> 
> Signed-off-by: Matteo Croce <mcroce@redhat.com>
> Fixes: 14683814 ("bpf: add xdpdrv for requesting XDP driver mode")
> Fixes: 1b5e8094 ("bpf: allow requesting XDP HW offload")

Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Matteo Croce Feb. 14, 2019, 2:01 p.m. UTC | #3
On Wed, Feb 13, 2019 at 11:04 PM Stephen Hemminger
<stephen@networkplumber.org> wrote:
>
> On Wed, 13 Feb 2019 15:40:30 +0100
> Matteo Croce <mcroce@redhat.com> wrote:
>
> > When attaching an eBPF program to a device, ip link can force the XDP mode
> > by using the xdp{generic,drv,offload} keyword instead of just 'xdp'.
> > Document this behaviour also in the help output.
> >
> > Signed-off-by: Matteo Croce <mcroce@redhat.com>
> > Fixes: 14683814 ("bpf: add xdpdrv for requesting XDP driver mode")
> > Fixes: 1b5e8094 ("bpf: allow requesting XDP HW offload")
>
> Applied, thanks.
> The man page already has this as well.
>

Yes, I found it just after I made the patch. However, it could be nice
to have the generic "xdp" and a command like "type" or "mode" to
specify the XDP mode, eg.
ip link set dev eth0 xdp mode [ auto | generic | drv | offload ]
I was trying to add it, but unfortunately it seems that the arguments
aren't parsed in a loop, and are required to be in the exact order.
Would this change make sense?

Regards,

--
Matteo Croce
per aspera ad upstream
Stephen Hemminger Feb. 14, 2019, 4:29 p.m. UTC | #4
On Thu, 14 Feb 2019 15:01:26 +0100
Matteo Croce <mcroce@redhat.com> wrote:

> On Wed, Feb 13, 2019 at 11:04 PM Stephen Hemminger
> <stephen@networkplumber.org> wrote:
> >
> > On Wed, 13 Feb 2019 15:40:30 +0100
> > Matteo Croce <mcroce@redhat.com> wrote:
> >  
> > > When attaching an eBPF program to a device, ip link can force the XDP mode
> > > by using the xdp{generic,drv,offload} keyword instead of just 'xdp'.
> > > Document this behaviour also in the help output.
> > >
> > > Signed-off-by: Matteo Croce <mcroce@redhat.com>
> > > Fixes: 14683814 ("bpf: add xdpdrv for requesting XDP driver mode")
> > > Fixes: 1b5e8094 ("bpf: allow requesting XDP HW offload")  
> >
> > Applied, thanks.
> > The man page already has this as well.
> >  
> 
> Yes, I found it just after I made the patch. However, it could be nice
> to have the generic "xdp" and a command like "type" or "mode" to
> specify the XDP mode, eg.
> ip link set dev eth0 xdp mode [ auto | generic | drv | offload ]
> I was trying to add it, but unfortunately it seems that the arguments
> aren't parsed in a loop, and are required to be in the exact order.
> Would this change make sense?
> 
> Regards,
> 
> --
> Matteo Croce
> per aspera ad upstream

It would have made sense to start with, but the syntax is more or less fixed
by now.
diff mbox series

Patch

diff --git a/ip/iplink.c b/ip/iplink.c
index b5519201..3a0cf459 100644
--- a/ip/iplink.c
+++ b/ip/iplink.c
@@ -98,7 +98,7 @@  void iplink_usage(void)
 		"				   [ trust { on | off} ] ]\n"
 		"				   [ node_guid { eui64 } ]\n"
 		"				   [ port_guid { eui64 } ]\n"
-		"			  [ xdp { off |\n"
+		"			  [ { xdp | xdpgeneric | xdpdrv | xdpoffload } { off |\n"
 		"				  object FILE [ section NAME ] [ verbose ] |\n"
 		"				  pinned FILE } ]\n"
 		"			  [ master DEVICE ][ vrf NAME ]\n"