net: phy: honor eee_disabled_modes in phy_support_eee()

[PATCH] net: phy: honor eee_disabled_modes in phy_support_eee()

Posted by Nicolai Buchwitz 1 week, 2 days ago

phy_support_eee() copies supported_eee into advertising_eee
unconditionally, overwriting any filtering applied during phy_probe()
based on DT eee-broken-* properties or driver-populated
eee_disabled_modes. MAC drivers that call phy_support_eee() after
probe (e.g. bcmgenet, fec, lan743x, lan78xx, r8169) then cause the PHY
to advertise EEE for modes the user marked as broken.

The symptom is that ethtool --show-eee on the local interface reports
"not supported" (supported & ~eee_disabled_modes is empty) while the
link partner sees EEE negotiated and active.

phy_probe() already filters advertising_eee via eee_disabled_modes
after calling of_set_phy_eee_broken(). Apply the same mask in
phy_support_eee() so the filtering survives the copy.

Fixes: 49168d1980e2 ("net: phy: Add phy_support_eee() indicating MAC support EEE")
Signed-off-by: Nicolai Buchwitz <nb@tipi-net.de>
---
 drivers/net/phy/phy_device.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/net/phy/phy_device.c b/drivers/net/phy/phy_device.c
index c2cdf1ae354249c033445f2d942e07220205b41c..83b074bc4a8f716b3e10c8ff57aff7c7abba0236 100644
--- a/drivers/net/phy/phy_device.c
+++ b/drivers/net/phy/phy_device.c
@@ -2903,7 +2903,8 @@ EXPORT_SYMBOL_GPL(phy_advertise_eee_all);
  */
 void phy_support_eee(struct phy_device *phydev)
 {
-	linkmode_copy(phydev->advertising_eee, phydev->supported_eee);
+	linkmode_andnot(phydev->advertising_eee, phydev->supported_eee,
+			phydev->eee_disabled_modes);
 	phydev->eee_cfg.tx_lpi_enabled = true;
 	phydev->eee_cfg.eee_enabled = true;
 

---
base-commit: aaec7096f9961eb223b5b149abe9495525c205d9
change-id: 20260516-devel-phy-support-eee-fix-d9cf040a6557

Best regards,
-- 
Nicolai Buchwitz <nb@tipi-net.de>

Re: [PATCH] net: phy: honor eee_disabled_modes in phy_support_eee()

Posted by Andrew Lunn 1 week, 1 day ago

On Sat, May 16, 2026 at 06:51:02AM +0200, Nicolai Buchwitz wrote:
> phy_support_eee() copies supported_eee into advertising_eee
> unconditionally, overwriting any filtering applied during phy_probe()
> based on DT eee-broken-* properties or driver-populated
> eee_disabled_modes. MAC drivers that call phy_support_eee() after
> probe (e.g. bcmgenet, fec, lan743x, lan78xx, r8169) then cause the PHY
> to advertise EEE for modes the user marked as broken.
> 
> The symptom is that ethtool --show-eee on the local interface reports
> "not supported" (supported & ~eee_disabled_modes is empty) while the
> link partner sees EEE negotiated and active.
> 
> phy_probe() already filters advertising_eee via eee_disabled_modes
> after calling of_set_phy_eee_broken(). Apply the same mask in
> phy_support_eee() so the filtering survives the copy.

I don't think this not the full fix.

of_set_phy_eee_broken() sets phydev->eee_disabled_modes.

genphy_c45_ethtool_get_eee() masks supported_eee with this when
reporting to user space what is supported.

genphy_c45_ethtool_set_eee() will also mask what user space passes in,
if it is not empty. However, if nothing is passed in,
phy_advertise_eee_all() is used, and that does not make use of
phydev->eee_disabled_modes, which i think it should.

So while your fix seems correct, i don't think on its own, it is
sufficient. Whenever phydev->supported_eee is used, we need to review
the code and see if it should be masked by phydev->eee_disabled_modes.

    Andrew

---
pw-bot: cr

Re: [PATCH] net: phy: honor eee_disabled_modes in phy_support_eee()

Posted by Nicolai Buchwitz 1 week ago

Hi Andrew

On May 16, 2026 4:40:25 PM GMT+02:00, Andrew Lunn <andrew@lunn.ch> wrote:
>On Sat, May 16, 2026 at 06:51:02AM +0200, Nicolai Buchwitz wrote:
>> phy_support_eee() copies supported_eee into advertising_eee
>> unconditionally, overwriting any filtering applied during phy_probe()
>> based on DT eee-broken-* properties or driver-populated
>> eee_disabled_modes. MAC drivers that call phy_support_eee() after
>> probe (e.g. bcmgenet, fec, lan743x, lan78xx, r8169) then cause the PHY
>> to advertise EEE for modes the user marked as broken.
>> 
>> The symptom is that ethtool --show-eee on the local interface reports
>> "not supported" (supported & ~eee_disabled_modes is empty) while the
>> link partner sees EEE negotiated and active.
>> 
>> phy_probe() already filters advertising_eee via eee_disabled_modes
>> after calling of_set_phy_eee_broken(). Apply the same mask in
>> phy_support_eee() so the filtering survives the copy.
>
>I don't think this not the full fix.

You're right, this only fixes the initial behavior on boot.

>
>of_set_phy_eee_broken() sets phydev->eee_disabled_modes.
>
>genphy_c45_ethtool_get_eee() masks supported_eee with this when
>reporting to user space what is supported.
>
>genphy_c45_ethtool_set_eee() will also mask what user space passes in,
>if it is not empty. However, if nothing is passed in,
>phy_advertise_eee_all() is used, and that does not make use of
>phydev->eee_disabled_modes, which i think it should.

Agree, will fix this in v2.

>
>So while your fix seems correct, i don't think on its own, it is
>sufficient. Whenever phydev->supported_eee is used, we need to review
>the code and see if it should be masked by phydev->eee_disabled_modes.

I will have a look and try to catch all scenarios.

>
>    Andrew
>
>---
>pw-bot: cr

Thanks
Nicolai

Re: [PATCH] net: phy: honor eee_disabled_modes in phy_support_eee()

Posted by Nicolai Buchwitz 1 week, 2 days ago

On 16.5.2026 06:51, Nicolai Buchwitz wrote:
> phy_support_eee() copies supported_eee into advertising_eee
> unconditionally, overwriting any filtering applied during phy_probe()
> based on DT eee-broken-* properties or driver-populated
> eee_disabled_modes. MAC drivers that call phy_support_eee() after
> probe (e.g. bcmgenet, fec, lan743x, lan78xx, r8169) then cause the PHY
> to advertise EEE for modes the user marked as broken.
> 
> The symptom is that ethtool --show-eee on the local interface reports
> "not supported" (supported & ~eee_disabled_modes is empty) while the
> link partner sees EEE negotiated and active.
> 
> phy_probe() already filters advertising_eee via eee_disabled_modes
> after calling of_set_phy_eee_broken(). Apply the same mask in
> phy_support_eee() so the filtering survives the copy.
> 
> Fixes: 49168d1980e2 ("net: phy: Add phy_support_eee() indicating MAC 
> support EEE")
> Signed-off-by: Nicolai Buchwitz <nb@tipi-net.de>
> ---

> [...]


Sorry for the noise, this patch is missing the net prefix in the 
subject. Will resend as v2 with the correct prefix tomorrow. Still 
learning b4...

Have a nice weekend,
Nicolai