Re: [PATCH net-next v2 4/4] net: macb: Add TSN CBS TC offload support

[Charles Perry]

On Thu, Nov 12, 2026 at 03:39:26PM +0530, Vineeth Karumanchi wrote:
> Hi Charles,
> 
> On 11/10/2026 8:29 PM, Charles Perry wrote:
> > Hello Vineeth, 
> > 
> > On Wed, Sep 09, 2026 at 07:50:56PM +0530, Vineeth Karumanchi wrote:
> >> +
> >> +	/* qopt->idleslope is in kbps; convert to the units the hardware
> >> +	 * register expects:
> >> +	 * - High-speed GEM: fraction of port bandwidth, scaled to the full
> >> +	 *   32-bit register range
> >> +	 * - Standard MACB: the register counts bytes/sec in 1G (8-bit GMII)
> >> +	 *   mode and nibbles/sec in 10/100 (4-bit MII) mode, so scale kbps
> >> +	 *   by 1000/8 (125) or 1000/4 (250) respectively
> >> +	 */
> >> +	if (bp->caps & MACB_CAPS_HIGH_SPEED)
> >> +		idleslope = DIV_ROUND_UP_ULL((u64)qopt->idleslope * U32_MAX, speed_kbps);
> >> +	else
> >> +		idleslope = (u32)qopt->idleslope * (kset.base.speed >= 1000 ?
> >> +					       MACB_CBS_PORT_RATE_1G : MACB_CBS_PORT_RATE_10_100M);
> >> +
> > 
> > PIC64-HPSC, which has MACB_CAPS_HIGH_SPEED, doesn't have the first idle
> > slope calculation method which scales to the full 32 bits register range.
> > It uses nibble/sec for 10/100M, byte/sec for 1G, 2.5bytes/sec for 2.5G,
> > 5bytes/sec for 5G and 10bytes/sec for 10G.
> > 
> > Is it possible that you have another design config bit for the "full 32
> > bits register range" method used here?
> > 
> 
> The Cadence GEM revision `r1p14f2` in our implementation uses linear
> scaling across the full 32-bit idleSlope register range, as confirmed by
> Cadence support. However, the register description specifies only the
> second, speed-dependent formula, which does not match our observations.

Ok, on my side, Arya (cced) who did the testing confirmed that the measured
bandwidth is off with the 32-bits method. I think he tested at 1G.

> 
> I couldn’t find a configuration bit that distinguishes between these
> behaviors. Which Cadence IP revision is used in PIC64-HPSC?

r1p14f2a

The revision register reads: 0x220c010e
DCFG1: 08108510
DCFG2: 7af13fff
DCFG3: 24000000
DCFG4: 00000000
DCFG5: 502e7344
DCFG6: 0f9300fe
DCFG7: 00000000
DCFG8: 10100820
DCFG9: 00000000
DCFG10: 24454444
DCFG11: 000000f6
DCFG12: 37770101
DCFG13: 00037640

> 
> I can follow up with Cadence support to determine how software can
> identify which encoding applies.

Cool! Thank you.

> 
> Thanks,
> -- 
> 🙏 Vineeth
> 

-- 
Thanks,
Charles