Re: [PATCH net-next v4 5/6] net: phy: mediatek: add calibration logic for AN7581

From
Christian Marangi <ansuelsmth@gmail.com>
Date
2026-11-04 18:24:35
Also in
linux-arm-kernel, linux-mediatek, lkml
Reply to
Andrew Lunn, 2026-11-04
On Wed, Nov 04, 2026 at 04:47:09AM +0100, Andrew Lunn wrote:
quoted
+static int airoha_cal_cycle(struct phy_device *phydev, int devad,
+			    u32 regnum, u16 mask, u16 cal_val)
+{
+	struct airoha_socphy_shared *shared = phy_package_get_priv(phydev);
+	struct phy_device *phydev_p0;
+	int ret;
+
+	phydev_p0 = shared->phydev_p0;
+
+	phy_modify_mmd(phydev, devad, regnum, mask, cal_val);
+
+	ret = mtk_cal_cycle_wait(phydev_p0);
+	phydev_dbg(phydev, "cal_val: 0x%x, ret: %d\n", cal_val, ret);
What is the locking scheme here?

When using PHY packages, the typical usage is:

	phy_lock_mdio_bus(phydev);

and then use

	val = __phy_package_read(phydev, X, Y);

All the read/write helpers are __ indicating they assume the bus is
locked.

Here you are doing something similar, accessing another PHY in the
same package, but without locking. Is that safe?
Hi Andrew, thanks a lot for the review tag.

It's safe in the measure that the worst is that link goes down during the
calibration... I will rework the function to assign the base address to the
default address for port 0 (it's hardcoded by the SoC) and use the lock and
read/write helper instead.

The main problem of using __phy_package_read is that I have to duplicate
the mtk_cal_cycle_wait function in the airoha-ge-soc.c as it would use
different helper... is it ok?

-- 
	Ansuel
Thread (17 messages) 17 messages, 3 authors, last reply 10d ago
Keyboard shortcuts
hback out one level
jnext message in thread / next series-diff position
kprevious message in thread / previous series-diff position
ldrill in
Escclose help / fold thread tree
?toggle this help