Re: [PATCH net-next] net: dsa: xrs700x: allow HSR/PRP supervision dupes for node_table

From: George McCollister
Date: Fri Jun 04 2021 - 15:52:03 EST


On Fri, Jun 4, 2021 at 1:55 PM Vladimir Oltean <olteanv@xxxxxxxxx> wrote:
>
> On Fri, Jun 04, 2021 at 11:29:22AM -0500, George McCollister wrote:
> > Add an inbound policy filter which matches the HSR/PRP supervision
> > MAC range and forwards to the CPU port without discarding duplicates.
> > This is required to correctly populate time_in[A] and time_in[B] in the
> > HSR/PRP node_table. Leave the policy disabled by default and
> > enable/disable it when joining/leaving hsr.
> >
> > Signed-off-by: George McCollister <george.mccollister@xxxxxxxxx>
> > ---
>
> What happens if duplicates are discarded for supervision frames and
> time_in[A] or time_in[B] is not updated in the node table?

It'll indicate an error condition that doesn't exist though everything
will work just fine.
It'll call hsr_nl_ringerror() and show only time_in[A] or time_in[B]
updating in /sys/kernel/debug/hsr/hsr0/node_table.

IEC 62439-3 specifies an SNMP MIB that contains the node_table. I've
implemented Net-SNMP mibgroup code to provide the node table data for
remote monitoring.

Basically if time_in[A] and time_in[B] aren't both updating it means:
For HSR there is a break in the ring.
For PRP: A device has a cable connecting to a switch unplugged or a failed port.

>
> > drivers/net/dsa/xrs700x/xrs700x.c | 67 +++++++++++++++++++++++++++++++++++++++
> > 1 file changed, 67 insertions(+)
> >
> > diff --git a/drivers/net/dsa/xrs700x/xrs700x.c b/drivers/net/dsa/xrs700x/xrs700x.c
> > index fde6e99274b6..a79066174a77 100644
> > --- a/drivers/net/dsa/xrs700x/xrs700x.c
> > +++ b/drivers/net/dsa/xrs700x/xrs700x.c
> > @@ -79,6 +79,9 @@ static const struct xrs700x_mib xrs700x_mibs[] = {
> > XRS700X_MIB(XRS_EARLY_DROP_L, "early_drop", tx_dropped),
> > };
> >
> > +static const u8 eth_hsrsup_addr[ETH_ALEN] = {
> > + 0x01, 0x15, 0x4e, 0x00, 0x01, 0x00};
> > +
>
> What if the user sets a different last address byte for supervision frames?

Below I only actually use 40 bits in the policy. See comment in the code.

>
> > static void xrs700x_get_strings(struct dsa_switch *ds, int port,
> > u32 stringset, u8 *data)
> > {
> > @@ -329,6 +332,50 @@ static int xrs700x_port_add_bpdu_ipf(struct dsa_switch *ds, int port)
> > return 0;
> > }
> >
> > +/* Add an inbound policy filter which matches the HSR/PRP supervision MAC
> > + * range and forwards to the CPU port without discarding duplicates.
> > + * This is required to correctly populate the HSR/PRP node_table.
> > + * Leave the policy disabled, it will be enabled as needed.
> > + */
> > +static int xrs700x_port_add_hsrsup_ipf(struct dsa_switch *ds, int port)
> > +{
> > + struct xrs700x *priv = ds->priv;
> > + unsigned int val = 0;
> > + int i = 0;
> > + int ret;
> > +
> > + /* Compare 40 bits of the destination MAC address. */
> > + ret = regmap_write(priv->regmap, XRS_ETH_ADDR_CFG(port, 1), 40 << 2);
> > + if (ret)
> > + return ret;
> > +
> > + /* match HSR/PRP supervision destination 01:15:4e:00:01:XX */
> > + for (i = 0; i < sizeof(eth_hsrsup_addr); i += 2) {
> > + ret = regmap_write(priv->regmap, XRS_ETH_ADDR_0(port, 1) + i,
> > + eth_hsrsup_addr[i] |
> > + (eth_hsrsup_addr[i + 1] << 8));
> > + if (ret)
> > + return ret;
> > + }
> > +
> > + /* Mirror HSR/PRP supervision to CPU port */
> > + for (i = 0; i < ds->num_ports; i++) {
> > + if (dsa_is_cpu_port(ds, i))
> > + val |= BIT(i);
> > + }
> > +
> > + ret = regmap_write(priv->regmap, XRS_ETH_ADDR_FWD_MIRROR(port, 1), val);
> > + if (ret)
> > + return ret;
> > +
> > + /* Allow must be set prevent duplicate discard */
> > + ret = regmap_write(priv->regmap, XRS_ETH_ADDR_FWD_ALLOW(port, 1), val);
> > + if (ret)
> > + return ret;
> > +
> > + return 0;
> > +}
> > +
> > static int xrs700x_port_setup(struct dsa_switch *ds, int port)
> > {
> > bool cpu_port = dsa_is_cpu_port(ds, port);
> > @@ -358,6 +405,10 @@ static int xrs700x_port_setup(struct dsa_switch *ds, int port)
> > ret = xrs700x_port_add_bpdu_ipf(ds, port);
> > if (ret)
> > return ret;
> > +
> > + ret = xrs700x_port_add_hsrsup_ipf(ds, port);
> > + if (ret)
> > + return ret;
> > }
> >
> > return 0;
> > @@ -565,6 +616,14 @@ static int xrs700x_hsr_join(struct dsa_switch *ds, int port,
> > XRS_PORT_FORWARDING);
> > regmap_fields_write(priv->ps_forward, port, XRS_PORT_FORWARDING);
> >
> > + /* Enable inbound policy added by xrs700x_port_add_hsrsup_ipf()
> > + * which allows HSR/PRP supervision forwarding to the CPU port without
> > + * discarding duplicates.
> > + */
> > + regmap_update_bits(priv->regmap,
> > + XRS_ETH_ADDR_CFG(partner->index, 1), 1, 1);
> > + regmap_update_bits(priv->regmap, XRS_ETH_ADDR_CFG(port, 1), 1, 1);
> > +
> > hsr_pair[0] = port;
> > hsr_pair[1] = partner->index;
> > for (i = 0; i < ARRAY_SIZE(hsr_pair); i++) {
> > @@ -611,6 +670,14 @@ static int xrs700x_hsr_leave(struct dsa_switch *ds, int port,
> > XRS_PORT_FORWARDING);
> > regmap_fields_write(priv->ps_forward, port, XRS_PORT_FORWARDING);
> >
> > + /* Disable inbound policy added by xrs700x_port_add_hsrsup_ipf()
> > + * which allows HSR/PRP supervision forwarding to the CPU port without
> > + * discarding duplicates.
> > + */
> > + regmap_update_bits(priv->regmap,
> > + XRS_ETH_ADDR_CFG(partner->index, 1), 1, 0);
> > + regmap_update_bits(priv->regmap, XRS_ETH_ADDR_CFG(port, 1), 1, 0);
> > +
> > hsr_pair[0] = port;
> > hsr_pair[1] = partner->index;
> > for (i = 0; i < ARRAY_SIZE(hsr_pair); i++) {
> > --
> > 2.11.0
> >