Re: [PATCH RFC net-next 1/5] net: stmmac: call phylink_start() and phylink_stop() in XDP functions

["Russell King (Oracle)" <linux@armlinux.org.uk>]

On Fri, Feb 28, 2025 at 02:14:29PM +0100, Andrew Lunn wrote:
> On Fri, Feb 28, 2025 at 03:31:22PM +0800, Furong Xu wrote:
> > On Thu, 27 Feb 2025 15:05:02 +0000
> > "Russell King (Oracle)" <rmk+kernel@armlinux.org.uk> wrote:
> > 
> > > Phylink does not permit drivers to mess with the netif carrier, as
> > > this will de-synchronise phylink with the MAC driver. Moreover,
> > > setting and clearing the TE and RE bits via stmmac_mac_set() in this
> > > path is also wrong as the link may not be up.
> > > 
> > > Replace the netif_carrier_on(), netif_carrier_off() and
> > > stmmac_mac_set() calls with the appropriate phylink_start() and
> > > phylink_stop() calls, thereby allowing phylink to manage the netif
> > > carrier and TE/RE bits through the .mac_link_up() and .mac_link_down()
> > > methods.
> > > 
> > > Note that RE should only be set after the DMA is ready to avoid the
> > > receive FIFO between the MAC and DMA blocks overflowing, so
> > > phylink_start() needs to be placed after DMA has been started.
> > > 
> > > Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
>  
> > XDP programs work like a charm both before and after this patch.
> > 
> > Tested-by: Furong Xu <0x1207@gmail.com>
> 
> Thanks for testing this.
> 
> Could you give a little details of how you actually tested it?
> 
> The issues here is, when is the link set admin up, requiring that
> phylink triggers an autoneg etc.
> 
> For plain old TCP/IP, you at some point use:
> 
> ip link set eth42 up
> 
> which will cause the core to call the drivers open() method, which
> then triggers phylnk.
> 
> The carrier manipulation which this code replaces seems to suggest you
> can load an XDP program while the interface is admin down, and that
> action of loading the program will implicitly set the carrier up.

It won't. See
drivers/net/ethernet/stmicro/stmmac/stmmac_xdp.c::stmmac_xdp_set_prog():

        if_running = netif_running(dev);

        need_update = !!priv->xdp_prog != !!prog;

        if (if_running && need_update)
                stmmac_xdp_release(dev);

	...

        if (if_running && need_update)
                stmmac_xdp_open(dev);

stmmac_xdp_open() and stmmac_xdp_release() will only be called if
netif_running() returns true as explained yesterday (but maybe
without enough detail). This tests __LINK_STATE_START in dev->state,
which is set just before calling .ndo_open(), and cleared if it fails
or before calling .ndo_stop().

Thus, netif_running() indicates whether the net core thinks the
device is adminsitratively "up".

Therefore, in the old code, if the programme is changed while the
device is administratively down, netif_running() will return false,
if_running will be false, and neither stmmac_xdp_release() nor
stmmac_xdp_open() will be called.

Hope that addresses your concern.

Thanks.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!