Re: [PATCH v2 1/2] dt-bindings: ufs: Document static TX Equalization settings properties

[Can Guo]

On 5/27/2026 5:24 PM, Can Guo wrote:
> On 5/27/2026 5:15 PM, Manivannan Sadhasivam wrote:
> > On Wed, May 27, 2026 at 04:51:04PM +0800, Can Guo wrote:
> > > Hi Mani,
> > >
> > > On 5/23/2026 10:14 PM, Manivannan Sadhasivam wrote:
> > > > On Sat, May 23, 2026 at 06:47:10AM -0700, Can Guo wrote:
> > > > > UFS v5.0/UFSHCI v5.0 add HS-G6 support (46.6 Gbps/lane) via UniPro 
> > > > > v3.0
> > > > > and M-PHY v6.0. In these specs, TX Equalization is defined for all 
> > > > > High
> > > > > Speed Gears (not only HS-G6) to compensate channel loss and 
> > > > > improve signal
> > > > > integrity at high speed operation.
> > > > >
> > > > > For HS-G6, M-PHY uses PAM4 1b1b line coding, Pre-Coding may also be
> > > > > required depending on channel characteristics.
> > > > >
> > > > > Add vendor-neutral DT patternProperties:
> > > > > txeq-settings-g[1-6]
> > > > >
> > > > > Each property is a uint32 array of per-lane tuples:
> > > > > (PreShoot, DeEmphasis, PrecodeEn)
> > > > I don't think combining all EQ settings (PreShoot, DeEmphasis, 
> > > > PrecodeEn) in a
> > > > single property as opaque tuples is the right approach. These are 
> > > > three
> > > > semantically distinct parameters with independent value ranges. So 
> > > > packing
> > > > them into a uint32 array makes validation impossible in the schema.
> > > >
> > > > AFACIS, PrecodeEn is applicable only to HS-G6 (PAM4), but the proposed
> > > > patternProperties forces it into G1-G5 tuples as well, which is 
> > > > semantically
> > > > wrong.
> > > Point taken for the PrecodeEn.
> > > > PCIe binding defines one property per data rate for EQ presets:
> > > > https://github.com/devicetree-org/dt-schema/blob/main/dtschema/schemas/pci/pci-bus-common.yaml#L193 
> > > >
> > > >
> > > > Similarly, UFS should define one property per gear per (like, 
> > > > txeq-preshoot-g6,
> > > > txeq-deemphasis-g6, txeq-precode-enable-g6,...) rather than 
> > > > clubbing everything
> > > > into opaque tuples.
> > > Thanks for the suggestion. I will go with below approach:
> > >
> > > txeq-preshoot-g6 = <Host Lane 0 PreShoot, Device Lane 0 PreShoot, 
> > > Host Lane
> > > 1 PreShoot, Device Lane 1 PreShoot>;
> > > txeq-deemphasis-g6 = <Host Lane 0 DeEmphasis, Device Lane 0 
> > > DeEmphasis, Host
> > > Lane 1 DeEmphasis, Device Lane 1 DeEmphasis>;
> > > txeq-precode-en-g6 = <Host Lane 0 PrecodeEn, Device Lane 0 
> > > PrecodeEn, Host
> > > Lane 1 PrecodeEn, Device Lane 1 PrecodeEn>;
> > How about encoding Host and Device values in a single tuple. Like,
> >
> >     txeq-preshoot-g6 = <Lane_0 Host_PreShoot Device_PreShoot>, <Lane 
> > 1...>,
> Man, I don't like that way to be honest, it took me a few minutes to 
> understand it on the PCIe side...
>
> Let's go with my approach please...
I misunderstood your point. I thought you meant the 0x5555 vs 0x55 
stuffs in PCIe side.
I am OK to go with single tuple approach as you suggested, which is more 
readable.

Thanks,
Can Guo.
> Thanks,
> Can Guo.
> > - Mani