[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20260126171324.1392638-1-horms@kernel.org>
Date: Mon, 26 Jan 2026 17:13:24 +0000
From: Simon Horman <horms@...nel.org>
To: mmyangfl@...il.com
Cc: Simon Horman <horms@...nel.org>,
edumazet@...gle.com,
olteanv@...il.com,
pabeni@...hat.com,
kuba@...nel.org,
netdev@...r.kernel.org,
davem@...emloft.net,
linux-kernel@...r.kernel.org,
linux@...linux.org.uk,
andrew@...n.ch
Subject: Re: [net-next,v3,1/3] net: dsa: tag_yt921x: fix priority support
This is an AI-generated review of your patch. The human sending this
email has considered the AI review valid, or at least plausible.
For local reproduction steps see https://netdev-ai.bots.linux.dev/ai-local.html
---
net: dsa: yt921x: Add DCB/QoS support
This commit adds DCB/QoS support to the yt921x DSA switch driver,
setting up global DSCP/PCP priority mappings and adding related DCB
methods.
> diff --git a/drivers/net/dsa/yt921x.c b/drivers/net/dsa/yt921x.c
> index af8464c42b51..87e33d0df4ac 100644
> --- a/drivers/net/dsa/yt921x.c
> +++ b/drivers/net/dsa/yt921x.c
[ ... ]
> +/* Not "port" - DSCP mapping is global */
> +static int __maybe_unused
> +yt921x_dsa_port_get_dscp_prio(struct dsa_switch *ds, int port, u8 dscp)
> +{
> + struct yt921x_priv *priv = to_yt921x_priv(ds);
> + u32 val;
> + int res;
> +
> + mutex_lock(&priv->reg_lock);
> + res = yt921x_reg_read(priv, YT921X_IPM_DSCPn(dscp), &val);
> + mutex_unlock(&priv->reg_lock);
> +
> + if (res)
> + return res;
> +
> + return FIELD_GET(YT921X_IPM_PRIO_M, val);
> +}
> +
> +static int __maybe_unused
> +yt921x_dsa_port_del_dscp_prio(struct dsa_switch *ds, int port, u8 dscp, u8 prio)
> +{
> + struct yt921x_priv *priv = to_yt921x_priv(ds);
> + int res;
> +
> + mutex_lock(&priv->reg_lock);
> + res = yt921x_reg_write(priv, YT921X_IPM_DSCPn(dscp),
> + YT921X_IPM_PRIO(IEEE8021Q_TT_BK));
> + mutex_unlock(&priv->reg_lock);
> +
> + return res;
> +}
Should yt921x_dsa_port_del_dscp_prio() check the prio parameter before
resetting the DSCP mapping? The ocelot driver (ocelot_port_del_dscp_prio
in drivers/net/ethernet/mscc/ocelot.c) has a comment explaining that
during a "dcb app replace" command, the new entry is added first, then
the old entry is deleted. If the delete operation ignores the prio
parameter, it will incorrectly reset the newly-added mapping instead
of only deleting when the current priority matches the old one.
The ksz driver also checks this condition in ksz_port_del_dscp_prio().
Without this check, the sequence for "dcb app replace ... dscp X prio N"
followed by "dcb app replace ... dscp X prio M" would be:
1. add dscp X -> prio M (correct)
2. del dscp X, prio N (but code ignores prio N and resets to BK)
3. result: dscp X -> BK instead of prio M
Perhaps something like this is needed:
if (yt921x_dsa_port_get_dscp_prio(ds, port, dscp) != prio)
return 0;
Powered by blists - more mailing lists