Skip to contentSkip to conversation

net: phylib: fix fixed-speed >= 1G

PATCH Applied Russell King (Oracle) 4 participants 6 messages Aug 8, 2024 – Aug 11, 2024

Subsystems (3)
lore
Share & tools
lore

Russell King (Oracle)

author reviewer
#0 lore raw

v2 cover net: phylib: fix fixed-speed >= 1G

Hi,

This is v2 of the patch (now patches) adding support for ethtool !autoneg while respecting the requirements of IEEE 802.3.

v2 fixes the build errors in the previous patch by first constifying the "advertisement" argument to the linkmode functions that only read from this pointer. It also fixes the incorrectly named linkmode_set function.

 drivers/net/phy/phy_device.c | 34 +++++++++++++++++++++++++---------
 include/linux/mii.h          |  7 ++++---
 2 files changed, 29 insertions(+), 12 deletions(-)
Signature · 2 lines
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!
#1 lore raw

PATCH v2 1/2 net: mii: constify advertising mask

Constify the advertising mask to linkmode functions that only read from the advertising mask.

---
 include/linux/mii.h | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
include/linux/mii.h +4 -3
index d5a959ce4877..b8f26d4513c3 100644
@@ -140,7 +140,7 @@ static inline u32 ethtool_adv_to_mii_adv_t(u32 ethadv)
  * settings to phy autonegotiation advertisements for the
  * MII_ADVERTISE register.
  */
-static inline u32 linkmode_adv_to_mii_adv_t(unsigned long *advertising)
+static inline u32 linkmode_adv_to_mii_adv_t(const unsigned long *advertising)
 {
 	u32 result = 0;
 
@@ -215,7 +215,8 @@ static inline u32 ethtool_adv_to_mii_ctrl1000_t(u32 ethadv)
  * settings to phy autonegotiation advertisements for the
  * MII_CTRL1000 register when in 1000T mode.
  */
-static inline u32 linkmode_adv_to_mii_ctrl1000_t(unsigned long *advertising)
+static inline u32
+linkmode_adv_to_mii_ctrl1000_t(const unsigned long *advertising)
 {
 	u32 result = 0;
 
@@ -453,7 +454,7 @@ static inline void mii_ctrl1000_mod_linkmode_adv_t(unsigned long *advertising,
  * A small helper function that translates linkmode advertising to LVL
  * pause capabilities.
  */
-static inline u32 linkmode_adv_to_lcl_adv_t(unsigned long *advertising)
+static inline u32 linkmode_adv_to_lcl_adv_t(const unsigned long *advertising)
 {
 	u32 lcl_adv = 0;
 
Signature · 1 line
2.30.2
#3 lore raw

PATCH v2 2/2 net: phylib: do not disable autoneg for fixed speeds >= 1G

In reply to Russell King (Oracle)

We have an increasing number of drivers that are forcing auto-negotiation to be enabled for speeds of 1G or faster.

It would appear that auto-negotiation is mandatory for speeds above 100M. In 802.3, Annex 40C's state diagrams seems to imply that mr_autoneg_enable (BMCR AN ENABLE) doesn't affect whether or not the AN state machines work for 1000base-T, and some PHY datasheets (e.g. Marvell Alaska) state that disabling mr_autoneg_enable leaves AN enabled but forced to 1G full duplex.

Other PHY datasheets imply that BMCR AN ENABLE should not be cleared for >= 1G.

Thus, this should be handled in phylib rather than in each driver.

Rather than erroring out, arrange to implement the Marvell Alaska solution but in software for all PHYs: generate an appropriate single-speed advertisement for the requested speed, and keep AN enabled to the PHY driver. However, to avoid userspace API breakage, continue to report to userspace that we have AN disabled.

---
 drivers/net/phy/phy_device.c | 34 +++++++++++++++++++++++++---------
 1 file changed, 25 insertions(+), 9 deletions(-)
drivers/net/phy/phy_device.c +25 -9
index 29a4225cb712..69fe207ac2eb 100644
@@ -2103,22 +2103,20 @@ EXPORT_SYMBOL(phy_reset_after_clk_enable);
 /**
  * genphy_config_advert - sanitize and advertise auto-negotiation parameters
  * @phydev: target phy_device struct
+ * @advert: auto-negotiation parameters to advertise
  *
  * Description: Writes MII_ADVERTISE with the appropriate values,
  *   after sanitizing the values to make sure we only advertise
  *   what is supported.  Returns < 0 on error, 0 if the PHY's advertisement
  *   hasn't changed, and > 0 if it has changed.
  */
-static int genphy_config_advert(struct phy_device *phydev)
+static int genphy_config_advert(struct phy_device *phydev,
+				const unsigned long *advert)
 {
 	int err, bmsr, changed = 0;
 	u32 adv;
 
-	/* Only allow advertising what this PHY supports */
-	linkmode_and(phydev->advertising, phydev->advertising,
-		     phydev->supported);
-
-	adv = linkmode_adv_to_mii_adv_t(phydev->advertising);
+	adv = linkmode_adv_to_mii_adv_t(advert);
 
 	/* Setup standard advertisement */
 	err = phy_modify_changed(phydev, MII_ADVERTISE,
@@ -2141,7 +2139,7 @@ static int genphy_config_advert(struct phy_device *phydev)
 	if (!(bmsr & BMSR_ESTATEN))
 		return changed;
 
-	adv = linkmode_adv_to_mii_ctrl1000_t(phydev->advertising);
+	adv = linkmode_adv_to_mii_ctrl1000_t(advert);
 
 	err = phy_modify_changed(phydev, MII_CTRL1000,
 				 ADVERTISE_1000FULL | ADVERTISE_1000HALF,
@@ -2365,6 +2363,9 @@ EXPORT_SYMBOL(genphy_check_and_restart_aneg);
  */
 int __genphy_config_aneg(struct phy_device *phydev, bool changed)
 {
+	__ETHTOOL_DECLARE_LINK_MODE_MASK(fixed_advert);
+	const struct phy_setting *set;
+	unsigned long *advert;
 	int err;
 
 	err = genphy_c45_an_config_eee_aneg(phydev);
@@ -2379,10 +2380,25 @@ int __genphy_config_aneg(struct phy_device *phydev, bool changed)
 	else if (err)
 		changed = true;
 
-	if (AUTONEG_ENABLE != phydev->autoneg)
+	if (phydev->autoneg == AUTONEG_ENABLE) {
+		/* Only allow advertising what this PHY supports */
+		linkmode_and(phydev->advertising, phydev->advertising,
+			     phydev->supported);
+		advert = phydev->advertising;
+	} else if (phydev->speed < SPEED_1000) {
 		return genphy_setup_forced(phydev);
+	} else {
+		linkmode_zero(fixed_advert);
+
+		set = phy_lookup_setting(phydev->speed, phydev->duplex,
+					 phydev->supported, true);
+		if (set)
+			linkmode_set_bit(set->bit, fixed_advert);
+
+		advert = fixed_advert;
+	}
 
-	err = genphy_config_advert(phydev);
+	err = genphy_config_advert(phydev, advert);
 	if (err < 0) /* error */
 		return err;
 	else if (err)
Signature · 1 line
2.30.2
#5 lore raw
In reply to Russell King (Oracle)
This series was applied to netdev/net-next.git (main) by David S. Miller <[email protected]>:

Hello:

This series was applied to netdev/net-next.git (main) by David S. Miller <[email protected]>:

Quoted 11 lines · On Thu, 8 Aug 2024 12:40:36 +0100 you wrote:

Hi,

This is v2 of the patch (now patches) adding support for ethtool !autoneg while respecting the requirements of IEEE 802.3.

v2 fixes the build errors in the previous patch by first constifying the "advertisement" argument to the linkmode functions that only read from this pointer. It also fixes the incorrectly named linkmode_set function.

[...]

Here is the summary with links:
- [net-next,v2,1/2] net: mii: constify advertising mask
https://git.kernel.org/netdev/net-next/c/aa9fbc5dd9da
- [net-next,v2,2/2] net: phylib: do not disable autoneg for fixed speeds >= 1G

    https://git.kernel.org/netdev/net-next/c/6ff3cddc365b

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html
maintainer Reviewed-by on v2 2/2
#4 lore raw

Re: [PATCH net-next v2 2/2] net: phylib: do not disable autoneg for fixed speeds >= 1G

In reply to Russell King (Oracle)
Quoted 15 lines · Russell King (Oracle)

We have an increasing number of drivers that are forcing auto-negotiation to be enabled for speeds of 1G or faster.

It would appear that auto-negotiation is mandatory for speeds above 100M. In 802.3, Annex 40C's state diagrams seems to imply that mr_autoneg_enable (BMCR AN ENABLE) doesn't affect whether or not the AN state machines work for 1000base-T, and some PHY datasheets (e.g. Marvell Alaska) state that disabling mr_autoneg_enable leaves AN enabled but forced to 1G full duplex.

Other PHY datasheets imply that BMCR AN ENABLE should not be cleared for >= 1G.

Thus, this should be handled in phylib rather than in each driver.

Quoted 7 lines

Rather than erroring out, arrange to implement the Marvell Alaska solution but in software for all PHYs: generate an appropriate single-speed advertisement for the requested speed, and keep AN enabled to the PHY driver. However, to avoid userspace API breakage, continue to report to userspace that we have AN disabled.

Andrew