Re: [PATCH net v2 11/12] net: au1000: keep the MAC stopped until the PHY link is ready

From: Florian Fainelli

Date: Tue Nov 17 2026 - 14:18:29 EST


On 11/16/26 05:50, netdev-bot+sashiko@xxxxxxxxxx wrote:
Thank you for your contribution! Sashiko AI review found more severe
issue(s) in other patches of this series. This mail is only an FYI, in case
you want to address the nit picks below when you respin.

Low priority / nit picks for your consideration:

- [Low] The commit message and comments in this patch don't match the
code.

--

net: au1000: keep the MAC stopped until the PHY link is ready

This patch stops the MAC when the PHY reports link-down. On open, it now
calls phy_resume() before au1000_init(). It also keeps MAC RX/TX disabled
in au1000_init() until au1000_adjust_link() has seen a link-up.

diff --git a/drivers/net/ethernet/amd/au1000_eth.c b/drivers/net/ethernet/amd/au1000_eth.c
index 8b7a88ceb3c5e..d60430a3795ed 100644
--- a/drivers/net/ethernet/amd/au1000_eth.c
+++ b/drivers/net/ethernet/amd/au1000_eth.c

[ ... ]

@@ -691,8 +692,8 @@ static u32 au1000_rx_mode(struct net_device *dev, u32 reg);
* When the device powers up, the clocks are disabled and the
* mac is in reset state. When the interface is closed, we
* do the same -- reset the device and disable the clocks to
- * conserve power. Thus, whenever au1000_init() is called,
- * the device should already be in reset state.
+ * conserve power. The device must be stopped before au1000_init() is
+ * called, but may be out of reset for MDIO accesses while resuming the PHY.
*/
static int au1000_init(struct net_device *dev)
{

[ ... ]

@@ -928,11 +934,25 @@ static int au1000_open(struct net_device *dev)
return retval;
}

[ ... ]

+ /* Restore the MII clocks without starting the PHY state machine. */
+ if (dev->phydev) {
+ retval = phy_resume(dev->phydev);
+ if (retval)
+ goto err_reset;
+ }
+
retval = au1000_init(dev);
if (retval) {
netdev_err(dev, "error in au1000_init\n");
- free_irq(dev->irq, dev);
- return retval;
+ if (dev->phydev)
+ phy_suspend(dev->phydev);
+ goto err_reset;
}

[Severity: Low]

Can this error branch in au1000_open() ever run? au1000_init() has only
one return statement:

drivers/net/ethernet/amd/au1000_eth.c:au1000_init() {
...
spin_unlock_irqrestore(&aup->lock, flags);
netif_addr_unlock_bh(dev);
return 0;
}

au1000_enable_mac() returns void, so nothing in au1000_init() can fail.
That makes the new phy_suspend() rollback look like dead code, even
though the commit message describes it as:

if initialization fails, suspend the PHY again before cleanup.

The commit message also says:

Initialize the MAC before phy_start() so the PHY worker cannot access
MDIO or enable traffic during setup.

Didn't au1000_open() already call au1000_init() before phy_start() before
this patch? The diff shows that order as unchanged context.

The actual changes seem to be the new phy_resume() call before
au1000_init(), and making RX/TX depend on aup->old_link. Could the commit
message describe those changes instead?

Separately, the au1000_init() header comment now says the device "may be
out of reset for MDIO accesses while resuming the PHY". The comment in
au1000_probe() was not changed and still says:

/*
* The boot code uses the ethernet controller, so reset it to start
* fresh. au1000_init() expects that the device is in reset state.
*/

Should these two comments be reconciled? They now give conflicting
preconditions for au1000_init(). Neither is strictly accurate, because
au1000_init() always forces a reset itself through
au1000_enable_mac(dev, 1).

Agreed on all three points. In the respin I have:

• Removed the unreachable initialization-error branch, including the new phy_suspend() rollback, since au1000_init() always returns zero.

• Updated the commit message to describe the new synchronous phy_resume() before MAC initialization, while preserving the existing initialization-before-phy_start() ordering.

• Reconciled the comments to explain that au1000_init() forces a reset itself and that preceding MDIO accesses may already have enabled the MAC.

pw-bot: cr
--
Florian