RE: [PATCH net-next] net: phy: micrel: Adding SQI support for lan8814 phy

[Divya.Koppera]

Hi Andrew,

> -----Original Message-----
> From: Andrew Lunn <andrew@xxxxxxx>
> Sent: Saturday, August 27, 2022 1:13 AM
> To: Divya Koppera - I30481 <Divya.Koppera@xxxxxxxxxxxxx>
> Cc: hkallweit1@xxxxxxxxx; linux@xxxxxxxxxxxxxxx; davem@xxxxxxxxxxxxx;
> edumazet@xxxxxxxxxx; kuba@xxxxxxxxxx; pabeni@xxxxxxxxxx;
> netdev@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; UNGLinuxDriver
> <UNGLinuxDriver@xxxxxxxxxxxxx>
> Subject: Re: [PATCH net-next] net: phy: micrel: Adding SQI support for
> lan8814 phy
> 
> EXTERNAL EMAIL: Do not click links or open attachments unless you know the
> content is safe
> 
> > > I just took a quick look at the datasheet. It says:
> > >
> >
> > I'm not sure the datasheet you looked into is the right one. Could you
> please crosscheck if its lan8814 or lan8841.
> > Lan8814 is quad port phy where register access are of extended page.
> Lan8841 is 1 port phy where register access are mmd access.
> >
> > > All registers references in this section are in MMD Device Address 1
> > >
> > > So you should be using phy_read_mmd(phydev,
> MDIO_MMD_PMAPMD,
> > > xxx) to read/write these registers. The datasheet i have however is
> > > missing the register map, so i've no idea if it is still 0xe6.
> 
> https://ww1.microchip.com/downloads/aemDocuments/documents/UNG/
> ProductDocuments/DataSheets/DS-LAN8814-00003592C.pdf
> 
> 5.13.4 OPEN ALLIANCE TC1/TC12 DCQ SIGNAL QUALITY INDEX
> 
> Note: All registers references in this section are in MMD Device Address 1.
> 

Ah.. I too looked into it. Its a mistake. I suggested for correction of document internally. Also requested team to add 
SQI register set in register document that is published.

> This section defines the implementation of section 6.1.2 of the TC1 and TC12
> specifications. This mode builds upon the OPEN Alliance
> TC1/TC12 DCQ Mean Square Error method by mapping the MSE value onto a
> simple quality index. This mode is enabled by setting the sqi_enable bit, in
> the DCQ Configuration register.
> 
> The MSE value is compared to the thresholds set in the DCQ SQI Table
> Registers to provide an SQI value between 0 (worst value) and 7 (best
> value) as follows:
> 
> In order to capture the SQI value, the DCQ Read Capture bit in the DCQ
> Configuration register needs to be written as a high with the desired cable
> pair specified in the DCQ Channel Number field of the same register. The
> DCQ Read Capture bit will immediately self-clear and the result will be
> available in the DCQ SQI register.  In addition to the current SQI, the worst
> case (lowest) SQI since the last read is available in the SQI Worst Case field.
> The correlation between the SQI values stored in the DCQ SQI register and an
> according Signal to Noise Ratio (SNR) based on Additive White Gaussian
> (AWG) noise (bandwidth of 80 MHz @ 100 Mbps / 550 MHz @ 1000 Mbps) is
> shown in Table 5-5. The bit error rates to be expected in the case of white
> noise as interference signal is shown in the table as well for information
> purposes.
> 
> I had a quick look at OPEN ALLIANCE specification. It seems to specify how
> each of these registers should look. It just failed to specify where in the
> address map they are. So if you look at drivers implementing SQI, you see
> most poke around in MDIO_MMD_VEND1.  I wounder if we can actually
> share the implementation between drivers, those that follow the standard,
> with some paramatirisation where the registers are.
> 

I don't think it will work, each phy may have different register set and may or may not supports SQI.
Also it contains drivers of too old and that may not support SQI.

Apart from this lan8814 is quad port phy and register set is completely different from other drivers.

>           Andrew

Thanks,
Divya