Re: [PATCH net-next 2/3] dt-bindings: net: cdns,macb: Add wol-arp-packet property
From: Andrew Lunn <andrew@lunn.ch>
Date: 2024-02-01 13:12:29
Also in:
linux-devicetree, lkml
On Thu, Feb 01, 2024 at 12:11:15PM +0530, Vineeth Karumanchi wrote:
Hi Andrew, Krzysztof, On 31/01/24 6:48 pm, Andrew Lunn wrote:quoted
On Wed, Jan 31, 2024 at 01:09:19PM +0530, Vineeth Karumanchi wrote:quoted
Hi Andrew, On 31/01/24 6:56 am, Andrew Lunn wrote:quoted
On Tue, Jan 30, 2024 at 04:18:44PM +0530, Vineeth Karumanchi wrote:quoted
"wol-arp-packet" property enables WOL with ARP packet. It is an extension to "magic-packet for WOL.It not clear why this is needed. Is this not a standard feature of the IP? Is there no hardware bit indicating the capability?WOL via both ARP and Magic packet is supported by the IP version on ZU+ and Versal. However, user can choose which type of packet to recognize as a WOL event - magic packet or ARP.ethtool says: wol p|u|m|b|a|g|s|f|d... Sets Wake-on-LAN options. Not all devices support this. The argument to this option is a string of characters specifying which options to enable. p Wake on PHY activity u Wake on unicast messages m Wake on multicast messages b Wake on broadcast messages a Wake on ARP g Wake on MagicPacket™ s Enable SecureOn™ password for MagicPacket™ f Wake on filter(s) d Disable (wake on nothing). This option clears all previous options. So why do we need a DT property?The earlier implementation of WOL (magic-packet) was using DT property. We added one more packet type using DT property to be in-line with the earlier implementation.
I can understand that. It also suggests we did a bad job reviewing that patch, and should of rejected it. But it was added a long time ago, and we were less strict back then.
However, I echo with you that this feature should be in driver (CAPS). We will re-work the implementation with the below flow: - Add MACB_CAPS_WOL capability to the supported platforms - Advertise supported WOL packet types based on the CAPS in ethtool. - Users can set packet type using ethtool.
Yes, this sounds good. Maybe add to that, mark magic-packet deprecated, and a comment that ethtool should be used instead. Thanks Andrew