[PATCH net] net: pcs: lynx: fix phylink validation regression for phy-mode = "10g-qxgmii"

[PATCH net] net: pcs: lynx: fix phylink validation regression for phy-mode = "10g-qxgmii"

[Vladimir Oltean]

Added by commit ce312bbc2351 ("net: pcs: lynx: accept phy-mode =
"10g-qxgmii" and use in felix driver"), this mode broke when the
lynx_interfaces[] array was introduced to populate
lynx->pcs.supported_interfaces, because it is absent from there.

mscc_felix 0000:00:00.5: MAC returned PCS which does not support 10g-qxgmii
mscc_felix 0000:00:00.5: failed to validate link configuration for inband

In reality we should be querying the SerDes driver of the attached lane
to restrict which PHY modes are truly supported, but we currently lack
that infrastructure upstream, so just add 10g-qxgmii to the array.

Fixes: b0f88c1b9a53 ("net: pcs: lynx: fill in PCS supported_interfaces")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
 drivers/net/pcs/pcs-lynx.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/net/pcs/pcs-lynx.c b/drivers/net/pcs/pcs-lynx.c
index c0238360eb40..07c570e2d7c1 100644
--- a/drivers/net/pcs/pcs-lynx.c
+++ b/drivers/net/pcs/pcs-lynx.c
@@ -536,6 +536,7 @@ static const phy_interface_t lynx_interfaces[] = {
 	PHY_INTERFACE_MODE_2500BASEX,
 	PHY_INTERFACE_MODE_10GBASER,
 	PHY_INTERFACE_MODE_USXGMII,
+	PHY_INTERFACE_MODE_10G_QXGMII,
 };
 
 void lynx_pcs_set_supported_interfaces(struct phylink_pcs *pcs,
-- 
2.34.1

Re: [PATCH net] net: pcs: lynx: fix phylink validation regression for phy-mode = "10g-qxgmii"

[Russell King (Oracle)]

On Mon, Mar 31, 2025 at 06:39:06PM +0300, Vladimir Oltean wrote:
> Added by commit ce312bbc2351 ("net: pcs: lynx: accept phy-mode =
> "10g-qxgmii" and use in felix driver"), this mode broke when the
> lynx_interfaces[] array was introduced to populate
> lynx->pcs.supported_interfaces, because it is absent from there.

This commit is not in net-next:

$ git log -p ce312bbc2351
fatal: ambiguous argument 'ce312bbc2351': unknown revision or path not in the working tree.

Checking Linus' tree, it's the same.

Are you sure lynx in mainline supports this mode?

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

Re: [PATCH net] net: pcs: lynx: fix phylink validation regression for phy-mode = "10g-qxgmii"

[Vladimir Oltean]

On Mon, Mar 31, 2025 at 04:46:42PM +0100, Russell King (Oracle) wrote:
> On Mon, Mar 31, 2025 at 06:39:06PM +0300, Vladimir Oltean wrote:
> > Added by commit ce312bbc2351 ("net: pcs: lynx: accept phy-mode =
> > "10g-qxgmii" and use in felix driver"), this mode broke when the
> > lynx_interfaces[] array was introduced to populate
> > lynx->pcs.supported_interfaces, because it is absent from there.
> 
> This commit is not in net-next:
> 
> $ git log -p ce312bbc2351
> fatal: ambiguous argument 'ce312bbc2351': unknown revision or path not in the working tree.
> 
> Checking Linus' tree, it's the same.
> 
> Are you sure lynx in mainline supports this mode?
> 
> -- 
> RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
> FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

Oops, you're right, please ignore me. I was working on a rebase and I
didn't even think to check whether the driver support for this new PHY
mode wasn't upstream. Now I'm starting to remember how the QCA8084 also
required it, and Luo Jie upstreamed the core support without users, and
without the Felix driver patch.

pw-bot: cr

Re: [PATCH net] net: pcs: lynx: fix phylink validation regression for phy-mode = "10g-qxgmii"

[Russell King (Oracle)]

On Mon, Mar 31, 2025 at 06:54:36PM +0300, Vladimir Oltean wrote:
> On Mon, Mar 31, 2025 at 04:46:42PM +0100, Russell King (Oracle) wrote:
> > On Mon, Mar 31, 2025 at 06:39:06PM +0300, Vladimir Oltean wrote:
> > > Added by commit ce312bbc2351 ("net: pcs: lynx: accept phy-mode =
> > > "10g-qxgmii" and use in felix driver"), this mode broke when the
> > > lynx_interfaces[] array was introduced to populate
> > > lynx->pcs.supported_interfaces, because it is absent from there.
> > 
> > This commit is not in net-next:
> > 
> > $ git log -p ce312bbc2351
> > fatal: ambiguous argument 'ce312bbc2351': unknown revision or path not in the working tree.
> > 
> > Checking Linus' tree, it's the same.
> > 
> > Are you sure lynx in mainline supports this mode?
> > 
> > -- 
> > RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
> > FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!
> 
> Oops, you're right, please ignore me. I was working on a rebase and I
> didn't even think to check whether the driver support for this new PHY
> mode wasn't upstream. Now I'm starting to remember how the QCA8084 also
> required it, and Luo Jie upstreamed the core support without users, and
> without the Felix driver patch.

Oh no. A feature merged that didn't get any users, which happened six
months ago. :(

Can we please have a user of it in mainline soon?

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!

Re: [PATCH net] net: pcs: lynx: fix phylink validation regression for phy-mode = "10g-qxgmii"

[Vladimir Oltean]

On Mon, Mar 31, 2025 at 05:04:20PM +0100, Russell King (Oracle) wrote:
> Oh no. A feature merged that didn't get any users, which happened six
> months ago. :(
> 
> Can we please have a user of it in mainline soon?

Yes, that looks easily possible. I'll send two patches when net-next
opens.