Le Fri, 26 Nov 2021 17:54:55 +0000, Vladimir Oltean <vladimir.oltean@xxxxxxx> a écrit : > On Fri, Nov 26, 2021 at 06:27:38PM +0100, Clément Léger wrote: > > IFH preparation can take quite some time on slow processors (up to 5% in > > a iperf3 test for instance). In order to reduce the cost of this > > preparation, pre-compute IFH since most of the parameters are fixed per > > port. Only rew_op and vlan tag will be set when sending if different > > than 0. This allows to remove entirely the calls to packing() with basic > > usage. In the same time, export this function that will be used by FDMA. > > > > Signed-off-by: Clément Léger <clement.leger@xxxxxxxxxxx> > > --- > > If you would move this injection frame header template into struct > ocelot_port_private instead of struct ocelot_port, I would not have > anything against it. Because struct ocelot_port is common with DSA, > whereas struct ocelot_port_private isn't. > > Also, as things stand, all switch drivers call ocelot_init_port, but not > all supported switches have the same IFH format. See seville_xmit() -> > seville_ifh_set_dest(). So even though DSA does not use this for > anything, it wouldn't even contain valid information even if it wanted > to. So maybe you can move this initialization to some place isolated to > vsc7514. Acked, this makes sense, I will do this. -- Clément Léger, Embedded Linux and Kernel engineer at Bootlin https://bootlin.com
