From: Andrew Lunn <andrew@lunn.ch>
To: David Yang <mmyangfl@gmail.com>
Cc: netdev@vger.kernel.org, Vladimir Oltean <olteanv@gmail.com>,
"David S. Miller" <davem@davemloft.net>,
Eric Dumazet <edumazet@google.com>,
Jakub Kicinski <kuba@kernel.org>, Paolo Abeni <pabeni@redhat.com>,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH net-next 2/3] net: dsa: yt921x: Add port police support
Date: Wed, 25 Feb 2026 16:06:24 +0100 [thread overview]
Message-ID: <a11993f5-144b-4c3e-863d-2e3104d41bf1@lunn.ch> (raw)
In-Reply-To: <CAAXyoMO7YS28fwAkf93qN8Bc45qE2M6N9OzzKW+wSTJ_ujeXnQ@mail.gmail.com>
On Wed, Feb 25, 2026 at 10:47:18PM +0800, David Yang wrote:
> On Wed, Feb 25, 2026 at 10:17 PM Andrew Lunn <andrew@lunn.ch> wrote:
> ...
> > > +/* v * 2^e */
> > > +#define ldexpi(v, e) ({ \
> > > + __auto_type _v = (v); \
> > > + __auto_type _e = (e); \
> > > + _e >= 0 ? _v << _e : _v >> -_e; \
> > > +})
> > > +
> > > +/* slot (ns) * rate (/s) / 10^9 (ns/s) = 2^C * token * 4^unit */
> > > +#define rate2token(rate, unit, slot_ns, C) \
> > > + ((u32)(ldexpi((slot_ns) * (rate), \
> > > + -(2 * (unit) + (C) + YT921X_TOKEN_RATE_C)) / 1000000000))
> > > +#define token2rate(token, unit, slot_ns, C) \
> > > + (ldexpi(1000000000 * (u64)(token), \
> > > + 2 * (unit) + (C) + YT921X_TOKEN_RATE_C) / (slot_ns))
> > > +
> > > +/* burst = 2^C * token * 4^unit */
> > > +#define burst2token(burst, unit, C) ((u32)ldexpi((burst), -(2 * (unit) + (C))))
> > > +#define token2burst(token, unit, C) ldexpi((u64)(token), 2 * (unit) + (C))
> >
> > Please avoid macros like this. Write functions. The compiler will
> > inline them if that makes sense. But you gain type checking, etc.
> >
>
> ldexpi is generic over integer types and we do type conversions in the
> formulas below. So I think it cannot be expressed as a function.
Writing functions will then help makes the types clearer. Why are type
conversions needed? Making them explicit would make the code better.
> > > +static struct yt921x_meter
> > > +yt921x_meter_tfm(struct yt921x_priv *priv, int port, unsigned int slot_ns,
> > > + u64 rate, u64 burst, unsigned int flags,
> > > + u32 cir_max, u32 cbs_max, int unit_max)
> > > +{
> > > + const int C = flags & YT921X_METER_PKT_MODE ? YT921X_TOKEN_PKT_C :
> > > + YT921X_TOKEN_BYTE_C;
> > > + struct device *dev = to_device(priv);
> > > + struct yt921x_meter meter;
> > > + bool distorted;
> > > + u64 burst_est;
> > > + u64 burst_max;
> > > + u64 rate_max;
> > > +
> > > + meter.unit = unit_max;
> > > + rate_max = token2rate(cir_max, meter.unit, slot_ns, C);
> > > + burst_max = token2burst(cbs_max, meter.unit, C);
> > > +
> > > + /* Clamp rate and burst */
> > > + if (rate > rate_max || burst > burst_max) {
> > > + dev_warn(dev,
> > > + "rate %llu, burst %llu too high, max is %llu, %llu, clamping...\n",
> > > + rate, burst, rate_max, burst_max);
> >
> > I'm not sure clamping is the correct thing to do here. -ERANGE would
> > make it clearer that what the user requests is not possible.
> >
>
> The maxima are way larger than the physical limitations of GbE, so
> they are also acceptable. Still it would be quite unusual for users to
> supply such values.
If they are unusual, then maybe ERANGE makes more sense, the user has
probably misunderstood the API, which can be quite easy with tc :-(
> > > + } else if (meter.cir > cir_max) {
> > > + WARN_ON(1);
> > > + meter.cir = cir_max;
> >
> > What does this WARN_ON() indicate? If this fires, you want somebody
> > who is debugging it to be able to understand it.
> >
>
> It means logical error during unit selection. Surely internal error
> and the debugger must be familiar with the implementation details, I
> think.
} else if (WARN_ON(meter.cir > cir_max)) {
meter.cir = cir_max;
would be more normal.
Andrew
next prev parent reply other threads:[~2026-02-25 15:06 UTC|newest]
Thread overview: 10+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-02-25 9:08 [PATCH net-next 0/3] net: dsa: yt921x: Add port police/tbf support David Yang
2026-02-25 9:08 ` [PATCH net-next 1/3] net: dsa: yt921x: Refactor long register helpers David Yang
2026-02-25 13:52 ` Andrew Lunn
2026-02-25 14:12 ` David Yang
2026-02-25 14:21 ` Andrew Lunn
2026-02-25 9:08 ` [PATCH net-next 2/3] net: dsa: yt921x: Add port police support David Yang
2026-02-25 14:17 ` Andrew Lunn
2026-02-25 14:47 ` David Yang
2026-02-25 15:06 ` Andrew Lunn [this message]
2026-02-25 9:08 ` [PATCH net-next 3/3] net: dsa: yt921x: Add port qdisc tbf support David Yang
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=a11993f5-144b-4c3e-863d-2e3104d41bf1@lunn.ch \
--to=andrew@lunn.ch \
--cc=davem@davemloft.net \
--cc=edumazet@google.com \
--cc=kuba@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mmyangfl@gmail.com \
--cc=netdev@vger.kernel.org \
--cc=olteanv@gmail.com \
--cc=pabeni@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®