Re: [PATCH net v2] net: phy: aquantia: Introduce custom get_features

From: Abhishek Chauhan (ABC)
Date: Tue Sep 24 2024 - 20:50:28 EST




On 9/24/2024 8:19 AM, Abhishek Chauhan (ABC) wrote:
>
>
> On 9/24/2024 5:04 AM, Andrew Lunn wrote:
>>> +static void aqr_supported_speed(struct phy_device *phydev, u32 max_speed)
>>> +{
>>> + __ETHTOOL_DECLARE_LINK_MODE_MASK(supported) = { 0, };
>>> +
>>> + linkmode_set_bit(ETHTOOL_LINK_MODE_Autoneg_BIT, supported);
>>> + linkmode_set_bit(ETHTOOL_LINK_MODE_Pause_BIT, supported);
>>> + linkmode_set_bit(ETHTOOL_LINK_MODE_Asym_Pause_BIT, supported);
>>> + linkmode_set_bit(ETHTOOL_LINK_MODE_FIBRE_BIT, supported);
>>> + linkmode_set_bit(ETHTOOL_LINK_MODE_TP_BIT, supported);
>>> + linkmode_set_bit(ETHTOOL_LINK_MODE_10baseT_Half_BIT, supported);
>>> + linkmode_set_bit(ETHTOOL_LINK_MODE_10baseT_Full_BIT, supported);
>>> + linkmode_set_bit(ETHTOOL_LINK_MODE_100baseT_Half_BIT, supported);
>>> + linkmode_set_bit(ETHTOOL_LINK_MODE_100baseT_Full_BIT, supported);
>>> + linkmode_set_bit(ETHTOOL_LINK_MODE_1000baseT_Half_BIT, supported);
>>> + linkmode_set_bit(ETHTOOL_LINK_MODE_1000baseT_Full_BIT, supported);
>>> +
>>> + if (max_speed == SPEED_2500) {
>>> + linkmode_set_bit(ETHTOOL_LINK_MODE_2500baseX_Full_BIT, supported);
>>> + linkmode_set_bit(ETHTOOL_LINK_MODE_2500baseT_Full_BIT, supported);
>>> + } else if (max_speed == SPEED_5000) {
>>> + linkmode_set_bit(ETHTOOL_LINK_MODE_2500baseX_Full_BIT, supported);
>>> + linkmode_set_bit(ETHTOOL_LINK_MODE_2500baseT_Full_BIT, supported);
>>> + linkmode_set_bit(ETHTOOL_LINK_MODE_5000baseT_Full_BIT, supported);
>>> + }
>>> +
>>> + linkmode_copy(phydev->supported, supported);
>>> +}
>>
>> So you have got lots of comments....
>>
>> Please split this into two patches. One patch for the PHY you are
>> interested in, and a second patch to remove phy_set_max_speed() and
>> fix up that PHY.
>>
> Noted!
>
>> Also, i would prefer you do the normal feature discovery, calling
>> genphy_read_abilities() and/or genphy_c45_pma_read_abilities() and
>> then fixup the results by removing the modes which should not be
>> there.
>>
> Sounds good!
>> Take a look at bcm84881_get_features() as an example.
>>
> Thanks Andrew!

Andrew, (Let me know if this is okay with you)
On doing the normal feature discovery today I observed that AQR115C
i want to document this misbehavior here for future reference

//Print added by me in the AQR driver
[ 5.583440] AQR supported mask=00,00000000,00018000,000e102c

key points :-

AQR115c supports 10Mbps(F/H) but feature discovery says no
AQR115C supports 1Gbps(F/H) but feature discovery says no
AQR115C supports 2500BaseX but feature discovery says no
AQR115C supports Autoneg but feature discovery says Hell no!
AQR115C does not support 10GBaseT/KX4/KR but feature discovery says yes
AQR115c does not support 5GbaseT but feature discovery says yes

I have got 2 different FW from Marvell but none seem to help.
I also got the confirmation from Marvell folks that Autoneg is supported but
the feature discovery says otherwise.

Here is the thing what i am going to do for now. (Let me know if this is okay with you)
1. Raise FIXUP for AQR115c patch
- remove all the features which are not support
- Add supported features which we really requires such as Autoneg/phy_gbit_features/2500BaseX/BaseT(Clearly see them supported in the NDA internal docs)
Reference for other folks for information which is public :
https://www.marvell.com/content/dam/marvell/en/public-collateral/transceivers/marvell-phys-transceivers-aqrate-gen4-product-brief.pdf


2. Raise FIXUP patch by removing phy_set_max_speed and call
- Generic function which sets speeds up to 2.5Gbps for AQCS109 (2.5Gbps max speed)
- Generic function which sets speeds up to 5 Gbps for AQR111/B0/114c (5Gbps max speed)

Both points 1 and 2 will be a patch series with cover letter.



>> Andrew