Thread (24 messages) 24 messages, 4 authors, 2025-11-13

Re: [PATCH v5 8/8] net: stmmac: qcom-ethqos: add support for sa8255p

From: Simon Horman <horms@kernel.org>
Date: 2025-11-11 11:48:09
Also in: imx, linux-amlogic, linux-arm-msm, linux-devicetree, linux-mips, linux-renesas-soc, linux-riscv, linux-rockchip, linux-sunxi, lkml

On Fri, Nov 07, 2025 at 11:29:58AM +0100, Bartosz Golaszewski wrote:
From: Bartosz Golaszewski <redacted>
...
+static int qcom_ethqos_pd_init(struct platform_device *pdev, void *priv)
+{
+	struct qcom_ethqos *ethqos = priv;
+	int ret;
+
+	/*
+	 * Enable functional clock to prevent DMA reset after timeout due
+	 * to no PHY clock being enabled after the hardware block has been
+	 * power cycled. The actual configuration will be adjusted once
+	 * ethqos_fix_mac_speed() is called.
+	 */
+	ethqos_set_func_clk_en(ethqos);
+
+	ret = qcom_ethqos_domain_on(ethqos, ETHQOS_PD_CORE);
+	if (ret)
+		return ret;
+
+	ret = qcom_ethqos_domain_on(ethqos, ETHQOS_PD_MDIO);
+	if (ret) {
+		qcom_ethqos_domain_off(ethqos, ETHQOS_PD_CORE);
+		return ret;
+	}
+
+	return 0;
+}
+
+static void qcom_ethqos_pd_exit(struct platform_device *pdev, void *data)
+{
+	struct qcom_ethqos *ethqos = data;
+
+	qcom_ethqos_domain_off(ethqos, ETHQOS_PD_MDIO);
+	qcom_ethqos_domain_off(ethqos, ETHQOS_PD_CORE);
+}
+
 static void ethqos_ptp_clk_freq_config(struct stmmac_priv *priv)
 {
 	struct plat_stmmacenet_data *plat_dat = priv->plat;
...
quoted hunk ↗ jump to hunk
@@ -852,28 +993,63 @@ static int qcom_ethqos_probe(struct platform_device *pdev)
 	ethqos->rgmii_config_loopback_en = drv_data->rgmii_config_loopback_en;
 	ethqos->has_emac_ge_3 = drv_data->has_emac_ge_3;
 	ethqos->needs_sgmii_loopback = drv_data->needs_sgmii_loopback;
-
-	ethqos->pm.link_clk = devm_clk_get(dev, clk_name);
-	if (IS_ERR(ethqos->pm.link_clk))
-		return dev_err_probe(dev, PTR_ERR(ethqos->pm.link_clk),
-				     "Failed to get link_clk\n");
-
-	ret = ethqos_clks_config(ethqos, true);
-	if (ret)
-		return ret;
-
-	ret = devm_add_action_or_reset(dev, ethqos_clks_disable, ethqos);
-	if (ret)
-		return ret;
-
-	ethqos->pm.serdes_phy = devm_phy_optional_get(dev, "serdes");
-	if (IS_ERR(ethqos->pm.serdes_phy))
-		return dev_err_probe(dev, PTR_ERR(ethqos->pm.serdes_phy),
-				     "Failed to get serdes phy\n");
-
-	ethqos->set_serdes_speed = ethqos_set_serdes_speed_phy;
 	ethqos->serdes_speed = SPEED_1000;
-	ethqos_update_link_clk(ethqos, SPEED_1000);
+
+	if (pm_data && pm_data->use_domains) {
+		ethqos->set_serdes_speed = ethqos_set_serdes_speed_pd;
+
+		ret = devm_pm_domain_attach_list(dev, &pm_data->pd,
+						 &ethqos->pd.pd_list);
+		if (ret < 0)
+			return dev_err_probe(dev, ret, "Failed to attach power domains\n");
+
+		plat_dat->clks_config = ethqos_pd_clks_config;
+		plat_dat->serdes_powerup = qcom_ethqos_pd_serdes_powerup;
+		plat_dat->serdes_powerdown = qcom_ethqos_pd_serdes_powerdown;
+		plat_dat->exit = qcom_ethqos_pd_exit;
Hi Bartosz,

It seems that the intention of this is to ensure
that domains turned on by qcom_ethqos_pd_init()
are turned off again on exit or clean-up in error paths.
+		plat_dat->init = qcom_ethqos_pd_init;
+		plat_dat->clk_ptp_rate = pm_data->clk_ptp_rate;
+
+		ret = qcom_ethqos_pd_init(pdev, ethqos);
+		if (ret)
+			return ret;
And here those domains are turned on.
+
+		ret = qcom_ethqos_domain_on(ethqos, ETHQOS_PD_SERDES);
+		if (ret)
But it seems that if we reach this error path then the cleanup is not
performed. This is because plat_dat and thus it's exit callback are
registered until the call to devm_stmmac_pltfr_probe() towards the end of
this function.

Sorry if I'm on the wrong track here. I did dig into it.
But this was flagged by Claude Code running
https://github.com/masoncl/review-prompts/
+			return dev_err_probe(dev, ret,
+					     "Failed to enable the serdes power domain\n");
...
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help