Date:   Wed, 15 Mar 2023 01:34:54 +0200
From:   Vladimir Oltean <olteanv@...il.com>
To:     Clément Léger <clement.leger@...tlin.com>
Cc:     Andrew Lunn <andrew@...n.ch>,
        Florian Fainelli <f.fainelli@...il.com>,
        "David S. Miller" <davem@...emloft.net>,
        Eric Dumazet <edumazet@...gle.com>,
        Jakub Kicinski <kuba@...nel.org>,
        Paolo Abeni <pabeni@...hat.com>,
        Thomas Petazzoni <thomas.petazzoni@...tlin.com>,
        Herve Codina <herve.codina@...tlin.com>,
        Miquèl Raynal <miquel.raynal@...tlin.com>,
        Milan Stevanovic <milan.stevanovic@...com>,
        Jimmy Lalande <jimmy.lalande@...com>,
        Pascal Eberhard <pascal.eberhard@...com>,
        Arun Ramadoss <Arun.Ramadoss@...rochip.com>,
        linux-renesas-soc@...r.kernel.org, netdev@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH RESEND net-next v4 3/3] net: dsa: rzn1-a5psw: add vlan
 support

On Tue, Mar 14, 2023 at 05:36:51PM +0100, Clément Léger wrote:
> Add support for vlan operation (add, del, filtering) on the RZN1
> driver. The a5psw switch supports up to 32 VLAN IDs with filtering,
> tagged/untagged VLANs and PVID for each ports.
> 
> Signed-off-by: Clément Léger <clement.leger@...tlin.com>
> ---
>  drivers/net/dsa/rzn1_a5psw.c | 164 +++++++++++++++++++++++++++++++++++
>  drivers/net/dsa/rzn1_a5psw.h |   8 +-
>  2 files changed, 169 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/net/dsa/rzn1_a5psw.c b/drivers/net/dsa/rzn1_a5psw.c
> index 5059b2814cdd..a9a42a8bc7e3 100644
> --- a/drivers/net/dsa/rzn1_a5psw.c
> +++ b/drivers/net/dsa/rzn1_a5psw.c
> @@ -583,6 +583,144 @@ static int a5psw_port_fdb_dump(struct dsa_switch *ds, int port,
>  	return ret;
>  }
>  
> +static int a5psw_port_vlan_filtering(struct dsa_switch *ds, int port,
> +				     bool vlan_filtering,
> +				     struct netlink_ext_ack *extack)
> +{
> +	u32 mask = BIT(port + A5PSW_VLAN_VERI_SHIFT) |
> +		   BIT(port + A5PSW_VLAN_DISC_SHIFT);

I'm curious what the A5PSW_VLAN_VERI_SHIFT and A5PSW_VLAN_DISC_SHIFT
bits do. Also curious in general what does this hardware do w.r.t.
VLANs. There would be several things on the checklist:

- can it drop a VLAN which isn't present in the port membership list?
  I guess this is what A5PSW_VLAN_DISC_SHIFT does.

- can it use VLAN information from the packet (with a fallback on the
  port PVID) to determine where to send, and where *not* to send the
  packet? How does this relate to the flooding registers? Is the flood
  mask restricted by the VLAN mask? Is there a default VLAN installed in
  the hardware tables, which is also the PVID of all ports, and all
  ports are members of it? Could you implement standalone/bridged port
  forwarding isolation based on VLANs, rather than the flimsy and most
  likely buggy implementation done based on flooding domains, from this
  patch set?

- is the FDB looked up per {MAC DA, VLAN ID} or just MAC DA? Looking at
  a5psw_port_fdb_add(), there's absolutely no sign of "vid" being used,
  so I guess it's Shared VLAN Learning. In that case, there's absolutely
  no hope to implement ds->fdb_isolation for this hardware. But at the
  *very* least, please disable address learning on standalone ports,
  *and* implement ds->ops->port_fast_age() so that ports quickly forget
  their learned MAC adddresses after leaving a bridge and become
  standalone again.

- if the port PVID is indeed used to filter the flooding mask of
  untagged packets, then I speculate that when A5PSW_VLAN_VERI_SHIFT
  is set, the hardware searches for a VLAN tag in the packet, whereas if
  it's unset, all packets will be forwarded according just to the port
  PVID (A5PSW_SYSTEM_TAGINFO). That would be absolutely magnificent if
  true, but it also means that you need to be *a lot* more careful when
  programming this register. See the "Address databases" section from
  Documentation/networking/dsa/dsa.rst for an explanation of the
  asynchronous nature of .port_vlan_add() relative to .port_vlan_filtering().
  Also see the call paths of sja1105_commit_pvid() and mv88e6xxx_port_commit_pvid()
  for an example of how this should be managed correctly, and how the
  bridge PVID should be committed to hardware only when the port is
  currently VLAN-aware.

> +	u32 val = vlan_filtering ? mask : 0;
> +	struct a5psw *a5psw = ds->priv;
> +
> +	a5psw_reg_rmw(a5psw, A5PSW_VLAN_VERIFY, mask, val);
> +
> +	return 0;
> +}
> +
> +static int a5psw_port_vlan_del(struct dsa_switch *ds, int port,
> +			       const struct switchdev_obj_port_vlan *vlan)
> +{
> +	struct a5psw *a5psw = ds->priv;
> +	u16 vid = vlan->vid;
> +	int vlan_res_id;
> +
> +	dev_dbg(a5psw->dev, "Removing VLAN %d on port %d\n", vid, port);
> +
> +	vlan_res_id = a5psw_find_vlan_entry(a5psw, vid);
> +	if (vlan_res_id < 0)
> +		return -EINVAL;
> +
> +	a5psw_port_vlan_cfg(a5psw, vlan_res_id, port, false);
> +	a5psw_port_vlan_tagged_cfg(a5psw, vlan_res_id, port, false);
> +
> +	/* Disable PVID if the vid is matching the port one */

What does it mean to disable PVID?

> +	if (vid == a5psw_reg_readl(a5psw, A5PSW_SYSTEM_TAGINFO(port)))
> +		a5psw_reg_rmw(a5psw, A5PSW_VLAN_IN_MODE_ENA, BIT(port), 0);
> +
> +	return 0;
> +}
> +
>  static u64 a5psw_read_stat(struct a5psw *a5psw, u32 offset, int port)
>  {
>  	u32 reg_lo, reg_hi;
> @@ -700,6 +838,27 @@ static void a5psw_get_eth_ctrl_stats(struct dsa_switch *ds, int port,
>  	ctrl_stats->MACControlFramesReceived = stat;
>  }
>  
> +static void a5psw_vlan_setup(struct a5psw *a5psw, int port)
> +{
> +	u32 reg;
> +
> +	/* Enable TAG always mode for the port, this is actually controlled
> +	 * by VLAN_IN_MODE_ENA field which will be used for PVID insertion
> +	 */

What does the "tag always" mode do, and what are the alternatives?

> +	reg = A5PSW_VLAN_IN_MODE_TAG_ALWAYS;
> +	reg <<= A5PSW_VLAN_IN_MODE_PORT_SHIFT(port);
> +	a5psw_reg_rmw(a5psw, A5PSW_VLAN_IN_MODE, A5PSW_VLAN_IN_MODE_PORT(port),
> +		      reg);
> +
> +	/* Set transparent mode for output frame manipulation, this will depend
> +	 * on the VLAN_RES configuration mode
> +	 */

What does the "transparent" output mode do, and how does it compare to
the "dis", "strip" and "tag through" alternatives?

> +	reg = A5PSW_VLAN_OUT_MODE_TRANSPARENT;
> +	reg <<= A5PSW_VLAN_OUT_MODE_PORT_SHIFT(port);
> +	a5psw_reg_rmw(a5psw, A5PSW_VLAN_OUT_MODE,
> +		      A5PSW_VLAN_OUT_MODE_PORT(port), reg);
> +}

Sorry for asking all these questions, but VLAN configuration on a switch
such as to bring it in line with the bridge driver expectations is a
rather tricky thing, so I'd like to have as clear of a mental model of
this hardware as possible, if public documentation isn't available.

Powered by blists - more mailing lists