On Tue, Dec 03, 2024 at 02:45:36PM +0000, Rufus Segar wrote:
From 2f97d889cecfa2ba06652469765f31c70a68e867 Mon Sep 17 00:00:00 2001 From: Rufus Segar rufuss@carallon.com Date: Tue, 3 Dec 2024 14:35:10 +0000 Subject: [PATCH] Revert "net: phy: marvell 88e151x: Fix handling of bare RGMII interface type"
This reverts commit 431be621c6cbc72efd1d45fa36686a682cbb470a.
This change is erroneous, the RX/TX delay should only be added when in RGMII-ID mode. RGMII mode should have no RX/TX delay.
drivers/net/phy/marvell.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/drivers/net/phy/marvell.c b/drivers/net/phy/marvell.c index b0a0b7fcb3..08608a99b9 100644 --- a/drivers/net/phy/marvell.c +++ b/drivers/net/phy/marvell.c @@ -461,9 +461,7 @@ static int m88e151x_config(struct phy_device *phydev)
reg = phy_read(phydev, MDIO_DEVAD_NONE, MIIM_88E151x_PHY_MSCR); reg &= ~MIIM_88E151x_RGMII_RXTX_DELAY;
if (phydev->interface == PHY_INTERFACE_MODE_RGMII ||phydev->interface == PHY_INTERFACE_MODE_RGMII_ID)
if (phydev->interface == PHY_INTERFACE_MODE_RGMII_ID) reg |= MIIM_88E151x_RGMII_RXTX_DELAY;
Please provide more details and ideally links to relevant documentation as to why a 6 year old commit that itself says it's fixing a use case is wrong, thanks.