On Wed, Feb 24, 2021 at 7:12 AM DENG Qingfang <dqf...@gmail.com> wrote:
> Use port isolation registers to configure bridge offloading.
> Remove the VLAN init, as we have proper CPU tag and bridge offloading
> support now.
>
> Signed-off-by: DENG Qingfang <dqf...@gmail.com>
> ---
> This is not tested, as I don't have a RTL8366RB board. And I think there
> is potential race condition in port_bridge_{join,leave}.
Compilation failed for me like this:
../drivers/net/dsa/rtl8366rb.c:1573:23: error: initialization of 'void
(*)(struct dsa_switch *, int, struct net_device *)' from incompatible
pointer type 'int (*)(struct dsa_switch *, int, struct net_device *)'
[-Werror=incompatible-pointer-types]
1573 | .port_bridge_leave = rtl8366rb_port_bridge_leave,
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~
../drivers/net/dsa/rtl8366rb.c:1573:23: note: (near initialization for
'rtl8366rb_switch_ops.port_bridge_leave')
I fixed it like this:
diff --git a/drivers/net/dsa/rtl8366rb.c b/drivers/net/dsa/rtl8366rb.c
index e7abf846350d..0719fadadc3d 100644
--- a/drivers/net/dsa/rtl8366rb.c
+++ b/drivers/net/dsa/rtl8366rb.c
@@ -1161,7 +1161,7 @@ rtl8366rb_port_bridge_join(struct dsa_switch
*ds, int port,
0, port_bitmap << 1);
}
-static int
+static void
rtl8366rb_port_bridge_leave(struct dsa_switch *ds, int port,
struct net_device *bridge)
{
@@ -1176,14 +1176,17 @@ rtl8366rb_port_bridge_leave(struct dsa_switch
*ds, int port,
continue;
ret = regmap_update_bits(smi->map, RTL8366RB_PORT_ISO(i),
BIT(port + 1), 0);
- if (ret)
- return ret;
+ if (ret) {
+ dev_err(smi->dev, "failed to leave port %d
from bridge\n",
+ port);
+ return;
+ }
port_bitmap |= BIT(i);
}
- return regmap_update_bits(smi->map, RTL8366RB_PORT_ISO(port),
- port_bitmap << 1, 0);
+ regmap_update_bits(smi->map, RTL8366RB_PORT_ISO(port),
+ port_bitmap << 1, 0);
}
After this it works like a charm.
> - ret = rtl8366_init_vlan(smi);
> - if (ret)
> - return ret;
I suppose we can delete that confused VLAN set-up after this.
> - ds->configure_vlan_while_not_filtering = false;
This is default true not IIUC so we should be good!
With my minor changes:
Tested-by: Linus Walleij <linus.wall...@linaro.org>
Yours,
Linus Walleij
