Message ID | E1tAtcW-002RBS-LB@rmk-PC.armlinux.org.uk (mailing list archive) |
---|---|
State | Accepted |
Commit | 671154f174e0e7f242507cd074497661deb41bfd |
Delegated to: | Netdev Maintainers |
Headers | show |
Series | [net] net: phylink: ensure PHY momentary link-fails are handled | expand |
On Tue, Nov 12, 2024 at 04:20:00PM +0000, Russell King (Oracle) wrote: > Normally, phylib won't notify changes in quick succession. However, as > a result of commit 3e43b903da04 ("net: phy: Immediately call > adjust_link if only tx_lpi_enabled changes") this is no longer true - > it is now possible that phy_link_down() and phy_link_up() will both > complete before phylink's resolver has run, which means it'll miss that > pl->phy_state.link momentarily became false. > > Rename "mac_link_dropped" to be more generic "link_failed" since it will > cover more than the MAC/PCS end of the link failing, and arrange to set > this in phylink_phy_change() if we notice that the PHY reports that the > link is down. > > This will ensure that we capture an EEE reconfiguration event. > > Fixes: 3e43b903da04 ("net: phy: Immediately call adjust_link if only tx_lpi_enabled changes") > Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk> Reviewed-by: Oleksij Rempel <o.rempel@pengutronix.de> Thank you! > --- > drivers/net/phy/phylink.c | 14 ++++++++------ > 1 file changed, 8 insertions(+), 6 deletions(-) > > diff --git a/drivers/net/phy/phylink.c b/drivers/net/phy/phylink.c > index 4309317de3d1..3e9957b6aa14 100644 > --- a/drivers/net/phy/phylink.c > +++ b/drivers/net/phy/phylink.c > @@ -78,7 +78,7 @@ struct phylink { > unsigned int pcs_neg_mode; > unsigned int pcs_state; > > - bool mac_link_dropped; > + bool link_failed; > bool using_mac_select_pcs; > > struct sfp_bus *sfp_bus; > @@ -1475,9 +1475,9 @@ static void phylink_resolve(struct work_struct *w) > cur_link_state = pl->old_link_state; > > if (pl->phylink_disable_state) { > - pl->mac_link_dropped = false; > + pl->link_failed = false; > link_state.link = false; > - } else if (pl->mac_link_dropped) { > + } else if (pl->link_failed) { > link_state.link = false; > retrigger = true; > } else { > @@ -1572,7 +1572,7 @@ static void phylink_resolve(struct work_struct *w) > phylink_link_up(pl, link_state); > } > if (!link_state.link && retrigger) { > - pl->mac_link_dropped = false; > + pl->link_failed = false; > queue_work(system_power_efficient_wq, &pl->resolve); > } > mutex_unlock(&pl->state_mutex); > @@ -1835,6 +1835,8 @@ static void phylink_phy_change(struct phy_device *phydev, bool up) > pl->phy_state.pause |= MLO_PAUSE_RX; > pl->phy_state.interface = phydev->interface; > pl->phy_state.link = up; > + if (!up) > + pl->link_failed = true; > mutex_unlock(&pl->state_mutex); > > phylink_run_resolve(pl); > @@ -2158,7 +2160,7 @@ EXPORT_SYMBOL_GPL(phylink_disconnect_phy); > static void phylink_link_changed(struct phylink *pl, bool up, const char *what) > { > if (!up) > - pl->mac_link_dropped = true; > + pl->link_failed = true; > phylink_run_resolve(pl); > phylink_dbg(pl, "%s link %s\n", what, up ? "up" : "down"); > } > @@ -2792,7 +2794,7 @@ int phylink_ethtool_set_pauseparam(struct phylink *pl, > * link will cycle. > */ > if (manual_changed) { > - pl->mac_link_dropped = true; > + pl->link_failed = true; > phylink_run_resolve(pl); > } > > -- > 2.30.2 > >
Hello: This patch was applied to netdev/net.git (main) by Jakub Kicinski <kuba@kernel.org>: On Tue, 12 Nov 2024 16:20:00 +0000 you wrote: > Normally, phylib won't notify changes in quick succession. However, as > a result of commit 3e43b903da04 ("net: phy: Immediately call > adjust_link if only tx_lpi_enabled changes") this is no longer true - > it is now possible that phy_link_down() and phy_link_up() will both > complete before phylink's resolver has run, which means it'll miss that > pl->phy_state.link momentarily became false. > > [...] Here is the summary with links: - [net] net: phylink: ensure PHY momentary link-fails are handled https://git.kernel.org/netdev/net/c/671154f174e0 You are awesome, thank you!
diff --git a/drivers/net/phy/phylink.c b/drivers/net/phy/phylink.c index 4309317de3d1..3e9957b6aa14 100644 --- a/drivers/net/phy/phylink.c +++ b/drivers/net/phy/phylink.c @@ -78,7 +78,7 @@ struct phylink { unsigned int pcs_neg_mode; unsigned int pcs_state; - bool mac_link_dropped; + bool link_failed; bool using_mac_select_pcs; struct sfp_bus *sfp_bus; @@ -1475,9 +1475,9 @@ static void phylink_resolve(struct work_struct *w) cur_link_state = pl->old_link_state; if (pl->phylink_disable_state) { - pl->mac_link_dropped = false; + pl->link_failed = false; link_state.link = false; - } else if (pl->mac_link_dropped) { + } else if (pl->link_failed) { link_state.link = false; retrigger = true; } else { @@ -1572,7 +1572,7 @@ static void phylink_resolve(struct work_struct *w) phylink_link_up(pl, link_state); } if (!link_state.link && retrigger) { - pl->mac_link_dropped = false; + pl->link_failed = false; queue_work(system_power_efficient_wq, &pl->resolve); } mutex_unlock(&pl->state_mutex); @@ -1835,6 +1835,8 @@ static void phylink_phy_change(struct phy_device *phydev, bool up) pl->phy_state.pause |= MLO_PAUSE_RX; pl->phy_state.interface = phydev->interface; pl->phy_state.link = up; + if (!up) + pl->link_failed = true; mutex_unlock(&pl->state_mutex); phylink_run_resolve(pl); @@ -2158,7 +2160,7 @@ EXPORT_SYMBOL_GPL(phylink_disconnect_phy); static void phylink_link_changed(struct phylink *pl, bool up, const char *what) { if (!up) - pl->mac_link_dropped = true; + pl->link_failed = true; phylink_run_resolve(pl); phylink_dbg(pl, "%s link %s\n", what, up ? "up" : "down"); } @@ -2792,7 +2794,7 @@ int phylink_ethtool_set_pauseparam(struct phylink *pl, * link will cycle. */ if (manual_changed) { - pl->mac_link_dropped = true; + pl->link_failed = true; phylink_run_resolve(pl); }
Normally, phylib won't notify changes in quick succession. However, as a result of commit 3e43b903da04 ("net: phy: Immediately call adjust_link if only tx_lpi_enabled changes") this is no longer true - it is now possible that phy_link_down() and phy_link_up() will both complete before phylink's resolver has run, which means it'll miss that pl->phy_state.link momentarily became false. Rename "mac_link_dropped" to be more generic "link_failed" since it will cover more than the MAC/PCS end of the link failing, and arrange to set this in phylink_phy_change() if we notice that the PHY reports that the link is down. This will ensure that we capture an EEE reconfiguration event. Fixes: 3e43b903da04 ("net: phy: Immediately call adjust_link if only tx_lpi_enabled changes") Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk> --- drivers/net/phy/phylink.c | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-)