ARM Sunxi Platform Development
 help / color / mirror / Atom feed
* [PATCH net-next v4 0/5] net: stmmac: dwmac-sun8i: Fix the EPHY and MDIO mux ordering
@ 2026-11-20 10:45 Arun Iyengar via B4 Relay
  2026-11-20 10:45 ` [PATCH net-next v4 1/5] net: stmmac: dwmac-sun8i: Put the EPHY clock and reset after the MAC Arun Iyengar via B4 Relay
                   ` (4 more replies)
  0 siblings, 5 replies; 7+ messages in thread
From: Arun Iyengar via B4 Relay @ 2026-11-20 10:45 UTC (permalink / raw)
  To: Maxime Chevallier, Andrew Lunn, David S. Miller, Eric Dumazet,
	Jakub Kicinski, Paolo Abeni, Maxime Coquelin, Alexandre Torgue,
	Chen-Yu Tsai, Jernej Skrabec, Samuel Holland, Philipp Zabel,
	Russell King
  Cc: netdev, linux-arm-kernel, linux-sunxi, linux-stm32, linux-kernel,
	Arun Iyengar

Chen-Yu Tsai pointed out that the probe error path and remove() put the
EPHY clock and reset before sun8i_dwmac_exit() uses them (patch 1).
Sashiko and Maxime Chevallier pointed out that the EPHY lookup and the
mux registration run after register_netdev(), and Maxime suggested
registering the mux in the MDIO setup (patches 2, 4 and 5). Patch 2
replaces "Defer probe when the EPHY clock is not ready" [1]. Patch 3
balances runtime PM on the stmmac probe error path, which a mux failure
now takes.

Tested on QEMU orangepi-pc (H3, internal PHY) with KASAN; I have no H3,
H5 or V3s board. On net-next, a test-only mdio_mux_init() failure gives
a KASAN use-after-free in clk_disable() from sun8i_dwmac_exit(), and
unbinding with the link up oopses in phy_stop(). With the series both
are clean, a test device tree whose EPHY clock has no provider
(fw_devlink=permissive) defers before the netdev exists, and three
unbind/bind cycles with the link up pass. Without patch 3, after the
failed probe and a rebind the MAC stays runtime active with usage count
1; with it the MAC runtime-suspends (usage 0).

Developed with Claude (Anthropic); I reviewed it.

[1] https://lore.kernel.org/r/20261113-b4-sun8i-emac-ephy-defer-v1-1-12a8591702fe@gmail.com

---
Changes in v4:
- Put the EPHY clock and reset after stmmac_pltfr_remove() (Chen-Yu Tsai)
- Get the EPHY clock and reset before the MAC probe, defer on the clock (Sashiko, Maxime Chevallier)
- Balance runtime PM when the stmmac probe fails (new patch 3)
- Register the mux through new stmmac mdio_init()/mdio_exit() hooks (Maxime Chevallier)
- Now a series, new cover subject
- Link to v3: https://patch.msgid.link/20261115-sun8i-emac-mux-uaf-v3-1-c69d4189539d@gmail.com

Changes in v3:
- Drop the Fixes tag (Andrew Lunn)
- Shorter commit message
- Link to v2: https://patch.msgid.link/20261113-sun8i-emac-mux-uaf-v2-1-6203666b2687@gmail.com

Changes in v2:
- Target net-next instead of net, drop Cc: stable (Andrew Lunn)
- Shorter commit message
- Rebased onto net-next
- Link to v1: https://patch.msgid.link/20261112-sun8i-emac-mux-uaf-v1-1-308ae7e6fcf2@gmail.com

---
Arun Iyengar (5):
      net: stmmac: dwmac-sun8i: Put the EPHY clock and reset after the MAC
      net: stmmac: dwmac-sun8i: Get the EPHY clock and reset before the MAC probe
      net: stmmac: Balance runtime PM when the probe fails
      net: stmmac: Add MDIO init and exit platform hooks
      net: stmmac: dwmac-sun8i: Register the MDIO mux in the MDIO setup

 drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c | 82 +++++++++--------------
 drivers/net/ethernet/stmicro/stmmac/stmmac_main.c |  2 +
 drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c | 11 +++
 include/linux/stmmac.h                            |  2 +
 4 files changed, 48 insertions(+), 49 deletions(-)
---
base-commit: 2208ab00f4f213ade91ab775e7f8c4dacdcc856e
change-id: 20261112-sun8i-emac-mux-uaf-1ba885219fee

Best regards,
--  
Arun Iyengar <aiyengar1990@gmail.com>



^ permalink raw reply	[flat|nested] 7+ messages in thread

* [PATCH net-next v4 1/5] net: stmmac: dwmac-sun8i: Put the EPHY clock and reset after the MAC
  2026-11-20 10:45 [PATCH net-next v4 0/5] net: stmmac: dwmac-sun8i: Fix the EPHY and MDIO mux ordering Arun Iyengar via B4 Relay
@ 2026-11-20 10:45 ` Arun Iyengar via B4 Relay
  2026-11-20 10:45 ` [PATCH net-next v4 2/5] net: stmmac: dwmac-sun8i: Get the EPHY clock and reset before the MAC probe Arun Iyengar via B4 Relay
                   ` (3 subsequent siblings)
  4 siblings, 0 replies; 7+ messages in thread
From: Arun Iyengar via B4 Relay @ 2026-11-20 10:45 UTC (permalink / raw)
  To: Maxime Chevallier, Andrew Lunn, David S. Miller, Eric Dumazet,
	Jakub Kicinski, Paolo Abeni, Maxime Coquelin, Alexandre Torgue,
	Chen-Yu Tsai, Jernej Skrabec, Samuel Holland, Philipp Zabel,
	Russell King
  Cc: netdev, linux-arm-kernel, linux-sunxi, linux-stm32, linux-kernel,
	Arun Iyengar

From: Arun Iyengar <aiyengar1990@gmail.com>

sun8i_dwmac_exit() unpowers the internal PHY through ephy_clk and
rst_ephy, but the probe error path and sun8i_dwmac_remove() put both
before stmmac_pltfr_remove() calls it. If mdio_mux_init() fails after
the scan of the internal bus has powered the PHY, this is a
use-after-free.

Put them after stmmac_pltfr_remove(). The explicit unpower in remove()
is then redundant.

Suggested-by: Chen-Yu Tsai <wens@kernel.org>
Signed-off-by: Arun Iyengar <aiyengar1990@gmail.com>
---
 drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c | 15 ++++++---------
 1 file changed, 6 insertions(+), 9 deletions(-)

diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c b/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c
index 2ece7ba9dd73..9fb398f7cf9e 100644
--- a/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c
@@ -1236,7 +1236,7 @@ static int sun8i_dwmac_probe(struct platform_device *pdev)
 		ret = sun8i_dwmac_register_mdio_mux(priv);
 		if (ret) {
 			dev_err(&pdev->dev, "Failed to register mux\n");
-			goto dwmac_mux;
+			goto dwmac_remove;
 		}
 	}
 
@@ -1244,12 +1244,11 @@ static int sun8i_dwmac_probe(struct platform_device *pdev)
 
 	return 0;
 
-dwmac_mux:
-	reset_control_put(gmac->rst_ephy);
-	clk_put(gmac->ephy_clk);
 dwmac_remove:
 	pm_runtime_put_noidle(&pdev->dev);
 	stmmac_pltfr_remove(pdev);
+	reset_control_put(gmac->rst_ephy);
+	clk_put(gmac->ephy_clk);
 dwmac_syscon:
 	sun8i_dwmac_unset_syscon(gmac);
 
@@ -1262,14 +1261,12 @@ static void sun8i_dwmac_remove(struct platform_device *pdev)
 	struct stmmac_priv *priv = netdev_priv(ndev);
 	struct sunxi_priv_data *gmac = priv->plat->bsp_priv;
 
-	if (gmac->variant->soc_has_internal_phy) {
+	if (gmac->variant->soc_has_internal_phy)
 		mdio_mux_uninit(gmac->mux_handle);
-		sun8i_dwmac_unpower_internal_phy(gmac);
-		reset_control_put(gmac->rst_ephy);
-		clk_put(gmac->ephy_clk);
-	}
 
 	stmmac_pltfr_remove(pdev);
+	reset_control_put(gmac->rst_ephy);
+	clk_put(gmac->ephy_clk);
 	sun8i_dwmac_unset_syscon(gmac);
 }
 

-- 
2.55.0



^ permalink raw reply related	[flat|nested] 7+ messages in thread

* [PATCH net-next v4 2/5] net: stmmac: dwmac-sun8i: Get the EPHY clock and reset before the MAC probe
  2026-11-20 10:45 [PATCH net-next v4 0/5] net: stmmac: dwmac-sun8i: Fix the EPHY and MDIO mux ordering Arun Iyengar via B4 Relay
  2026-11-20 10:45 ` [PATCH net-next v4 1/5] net: stmmac: dwmac-sun8i: Put the EPHY clock and reset after the MAC Arun Iyengar via B4 Relay
@ 2026-11-20 10:45 ` Arun Iyengar via B4 Relay
  2026-11-20 10:45 ` [PATCH net-next v4 3/5] net: stmmac: Balance runtime PM when the probe fails Arun Iyengar via B4 Relay
                   ` (2 subsequent siblings)
  4 siblings, 0 replies; 7+ messages in thread
From: Arun Iyengar via B4 Relay @ 2026-11-20 10:45 UTC (permalink / raw)
  To: Maxime Chevallier, Andrew Lunn, David S. Miller, Eric Dumazet,
	Jakub Kicinski, Paolo Abeni, Maxime Coquelin, Alexandre Torgue,
	Chen-Yu Tsai, Jernej Skrabec, Samuel Holland, Philipp Zabel,
	Russell King
  Cc: netdev, linux-arm-kernel, linux-sunxi, linux-stm32, linux-kernel,
	Arun Iyengar

From: Arun Iyengar <aiyengar1990@gmail.com>

get_ephy_nodes() runs after stmmac_pltfr_probe() has registered the
netdev, so its -EPROBE_DEFER comes too late. An EPHY clock that is not
ready yet does not defer at all: the node is skipped and the probe
fails with -ENODEV.

Look up the clock and reset before stmmac_pltfr_probe(), and defer on
the clock as on the reset.

Signed-off-by: Arun Iyengar <aiyengar1990@gmail.com>
---
 drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c | 34 ++++++++++++++---------
 1 file changed, 21 insertions(+), 13 deletions(-)

diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c b/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c
index 9fb398f7cf9e..4acfda235394 100644
--- a/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c
@@ -775,16 +775,15 @@ static void sun8i_dwmac_flow_ctrl(struct mac_device_info *hw,
 }
 
 /* Search in mdio-mux node for internal PHY node and get its clk/reset */
-static int get_ephy_nodes(struct stmmac_priv *priv)
+static int get_ephy_nodes(struct device *dev, struct sunxi_priv_data *gmac)
 {
-	struct sunxi_priv_data *gmac = priv->plat->bsp_priv;
 	struct device_node *mdio_internal;
 	struct device_node *mdio_mux;
 	int ret;
 
-	mdio_mux = of_get_child_by_name(priv->device->of_node, "mdio-mux");
+	mdio_mux = of_get_child_by_name(dev->of_node, "mdio-mux");
 	if (!mdio_mux) {
-		dev_err(priv->device, "Cannot get mdio-mux node\n");
+		dev_err(dev, "Cannot get mdio-mux node\n");
 		return -ENODEV;
 	}
 
@@ -792,7 +791,7 @@ static int get_ephy_nodes(struct stmmac_priv *priv)
 						"allwinner,sun8i-h3-mdio-internal");
 	of_node_put(mdio_mux);
 	if (!mdio_internal) {
-		dev_err(priv->device, "Cannot get internal_mdio node\n");
+		dev_err(dev, "Cannot get internal_mdio node\n");
 		return -ENODEV;
 	}
 
@@ -801,8 +800,13 @@ static int get_ephy_nodes(struct stmmac_priv *priv)
 		struct clk *ephy_clk;
 
 		ephy_clk = of_clk_get(iphynode, 0);
-		if (IS_ERR(ephy_clk))
+		if (IS_ERR(ephy_clk)) {
+			if (PTR_ERR(ephy_clk) == -EPROBE_DEFER) {
+				of_node_put(mdio_internal);
+				return -EPROBE_DEFER;
+			}
 			continue;
+		}
 		gmac->rst_ephy = of_reset_control_get_exclusive(iphynode, NULL);
 		if (IS_ERR(gmac->rst_ephy)) {
 			ret = PTR_ERR(gmac->rst_ephy);
@@ -814,7 +818,7 @@ static int get_ephy_nodes(struct stmmac_priv *priv)
 			continue;
 		}
 		gmac->ephy_clk = ephy_clk;
-		dev_info(priv->device, "Found internal PHY node\n");
+		dev_info(dev, "Found internal PHY node\n");
 		of_node_put(mdio_internal);
 		return 0;
 	}
@@ -1209,9 +1213,15 @@ static int sun8i_dwmac_probe(struct platform_device *pdev)
 	plat_dat->tx_fifo_size = 4096;
 	plat_dat->rx_fifo_size = 16384;
 
+	if (gmac->variant->soc_has_internal_phy) {
+		ret = get_ephy_nodes(dev, gmac);
+		if (ret)
+			return ret;
+	}
+
 	ret = sun8i_dwmac_set_syscon(&pdev->dev, plat_dat);
 	if (ret)
-		return ret;
+		goto dwmac_ephy;
 
 	ret = stmmac_pltfr_probe(pdev, plat_dat, &stmmac_res);
 	if (ret)
@@ -1230,9 +1240,6 @@ static int sun8i_dwmac_probe(struct platform_device *pdev)
 	 * so after stmmac_dvr_probe()
 	 */
 	if (gmac->variant->soc_has_internal_phy) {
-		ret = get_ephy_nodes(priv);
-		if (ret)
-			goto dwmac_remove;
 		ret = sun8i_dwmac_register_mdio_mux(priv);
 		if (ret) {
 			dev_err(&pdev->dev, "Failed to register mux\n");
@@ -1247,10 +1254,11 @@ static int sun8i_dwmac_probe(struct platform_device *pdev)
 dwmac_remove:
 	pm_runtime_put_noidle(&pdev->dev);
 	stmmac_pltfr_remove(pdev);
-	reset_control_put(gmac->rst_ephy);
-	clk_put(gmac->ephy_clk);
 dwmac_syscon:
 	sun8i_dwmac_unset_syscon(gmac);
+dwmac_ephy:
+	reset_control_put(gmac->rst_ephy);
+	clk_put(gmac->ephy_clk);
 
 	return ret;
 }

-- 
2.55.0



^ permalink raw reply related	[flat|nested] 7+ messages in thread

* [PATCH net-next v4 3/5] net: stmmac: Balance runtime PM when the probe fails
  2026-11-20 10:45 [PATCH net-next v4 0/5] net: stmmac: dwmac-sun8i: Fix the EPHY and MDIO mux ordering Arun Iyengar via B4 Relay
  2026-11-20 10:45 ` [PATCH net-next v4 1/5] net: stmmac: dwmac-sun8i: Put the EPHY clock and reset after the MAC Arun Iyengar via B4 Relay
  2026-11-20 10:45 ` [PATCH net-next v4 2/5] net: stmmac: dwmac-sun8i: Get the EPHY clock and reset before the MAC probe Arun Iyengar via B4 Relay
@ 2026-11-20 10:45 ` Arun Iyengar via B4 Relay
  2026-11-21 10:46   ` sashiko-bot
  2026-11-20 10:45 ` [PATCH net-next v4 4/5] net: stmmac: Add MDIO init and exit platform hooks Arun Iyengar via B4 Relay
  2026-11-20 10:45 ` [PATCH net-next v4 5/5] net: stmmac: dwmac-sun8i: Register the MDIO mux in the MDIO setup Arun Iyengar via B4 Relay
  4 siblings, 1 reply; 7+ messages in thread
From: Arun Iyengar via B4 Relay @ 2026-11-20 10:45 UTC (permalink / raw)
  To: Maxime Chevallier, Andrew Lunn, David S. Miller, Eric Dumazet,
	Jakub Kicinski, Paolo Abeni, Maxime Coquelin, Alexandre Torgue,
	Chen-Yu Tsai, Jernej Skrabec, Samuel Holland, Philipp Zabel,
	Russell King
  Cc: netdev, linux-arm-kernel, linux-sunxi, linux-stm32, linux-kernel,
	Arun Iyengar

From: Arun Iyengar <aiyengar1990@gmail.com>

__stmmac_dvr_probe() enables runtime PM and takes a usage count before
it registers the MDIO bus, but its error path undoes neither. After a
failed or deferred probe the device never runtime-suspends.

Undo both, as stmmac_dvr_remove() does.

Signed-off-by: Arun Iyengar <aiyengar1990@gmail.com>
---
 drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index f256244ce0d1..4ebf4b625374 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -8407,6 +8407,8 @@ static int __stmmac_dvr_probe(struct device *device,
 error_pcs_setup:
 	stmmac_mdio_unregister(ndev);
 error_mdio_register:
+	pm_runtime_disable(device);
+	pm_runtime_put_noidle(device);
 	stmmac_napi_del(ndev);
 error_hw_init:
 	destroy_workqueue(priv->wq);

-- 
2.55.0



^ permalink raw reply related	[flat|nested] 7+ messages in thread

* [PATCH net-next v4 4/5] net: stmmac: Add MDIO init and exit platform hooks
  2026-11-20 10:45 [PATCH net-next v4 0/5] net: stmmac: dwmac-sun8i: Fix the EPHY and MDIO mux ordering Arun Iyengar via B4 Relay
                   ` (2 preceding siblings ...)
  2026-11-20 10:45 ` [PATCH net-next v4 3/5] net: stmmac: Balance runtime PM when the probe fails Arun Iyengar via B4 Relay
@ 2026-11-20 10:45 ` Arun Iyengar via B4 Relay
  2026-11-20 10:45 ` [PATCH net-next v4 5/5] net: stmmac: dwmac-sun8i: Register the MDIO mux in the MDIO setup Arun Iyengar via B4 Relay
  4 siblings, 0 replies; 7+ messages in thread
From: Arun Iyengar via B4 Relay @ 2026-11-20 10:45 UTC (permalink / raw)
  To: Maxime Chevallier, Andrew Lunn, David S. Miller, Eric Dumazet,
	Jakub Kicinski, Paolo Abeni, Maxime Coquelin, Alexandre Torgue,
	Chen-Yu Tsai, Jernej Skrabec, Samuel Holland, Philipp Zabel,
	Russell King
  Cc: netdev, linux-arm-kernel, linux-sunxi, linux-stm32, linux-kernel,
	Arun Iyengar

From: Arun Iyengar <aiyengar1990@gmail.com>

A glue driver that adds devices on top of the stmmac MDIO bus, such as
an MDIO mux, can only do so after stmmac_dvr_probe() has registered the
netdev.

Add mdio_init(), called once the bus is registered, and mdio_exit(),
called before it is unregistered.

Suggested-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Signed-off-by: Arun Iyengar <aiyengar1990@gmail.com>
---
 drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c | 11 +++++++++++
 include/linux/stmmac.h                            |  2 ++
 2 files changed, 13 insertions(+)

diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c
index 7396b68899c6..fb1f7b1a5323 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_mdio.c
@@ -682,8 +682,16 @@ int stmmac_mdio_register(struct net_device *ndev)
 bus_register_done:
 	priv->mii = new_bus;
 
+	if (priv->plat->mdio_init) {
+		err = priv->plat->mdio_init(priv);
+		if (err)
+			goto mdio_init_fail;
+	}
+
 	return 0;
 
+mdio_init_fail:
+	priv->mii = NULL;
 no_phy_found:
 	mdiobus_unregister(new_bus);
 bus_register_fail:
@@ -703,6 +711,9 @@ int stmmac_mdio_unregister(struct net_device *ndev)
 	if (!priv->mii)
 		return 0;
 
+	if (priv->plat->mdio_exit)
+		priv->plat->mdio_exit(priv);
+
 	mdiobus_unregister(priv->mii);
 	priv->mii->priv = NULL;
 	mdiobus_free(priv->mii);
diff --git a/include/linux/stmmac.h b/include/linux/stmmac.h
index 00be2df63d22..e94f9cab973d 100644
--- a/include/linux/stmmac.h
+++ b/include/linux/stmmac.h
@@ -296,6 +296,8 @@ struct plat_stmmacenet_data {
 	int (*crosststamp)(ktime_t *device, struct system_counterval_t *system,
 			   void *ctx);
 	void (*dump_debug_regs)(void *priv);
+	int (*mdio_init)(struct stmmac_priv *priv);
+	void (*mdio_exit)(struct stmmac_priv *priv);
 	int (*pcs_init)(struct stmmac_priv *priv);
 	void (*pcs_exit)(struct stmmac_priv *priv);
 	struct phylink_pcs *(*select_pcs)(struct stmmac_priv *priv,

-- 
2.55.0



^ permalink raw reply related	[flat|nested] 7+ messages in thread

* [PATCH net-next v4 5/5] net: stmmac: dwmac-sun8i: Register the MDIO mux in the MDIO setup
  2026-11-20 10:45 [PATCH net-next v4 0/5] net: stmmac: dwmac-sun8i: Fix the EPHY and MDIO mux ordering Arun Iyengar via B4 Relay
                   ` (3 preceding siblings ...)
  2026-11-20 10:45 ` [PATCH net-next v4 4/5] net: stmmac: Add MDIO init and exit platform hooks Arun Iyengar via B4 Relay
@ 2026-11-20 10:45 ` Arun Iyengar via B4 Relay
  4 siblings, 0 replies; 7+ messages in thread
From: Arun Iyengar via B4 Relay @ 2026-11-20 10:45 UTC (permalink / raw)
  To: Maxime Chevallier, Andrew Lunn, David S. Miller, Eric Dumazet,
	Jakub Kicinski, Paolo Abeni, Maxime Coquelin, Alexandre Torgue,
	Chen-Yu Tsai, Jernej Skrabec, Samuel Holland, Philipp Zabel,
	Russell King
  Cc: netdev, linux-arm-kernel, linux-sunxi, linux-stm32, linux-kernel,
	Arun Iyengar

From: Arun Iyengar <aiyengar1990@gmail.com>

Both PHYs sit behind the mux, but the mux is registered only after
stmmac_pltfr_probe() has registered the netdev, so an early open does
not find the PHY. On remove, the mux goes away before
unregister_netdev(), and unbinding with the link up oopses in
phy_stop().

Register and unregister the mux through the new mdio_init() and
mdio_exit() hooks.

Signed-off-by: Arun Iyengar <aiyengar1990@gmail.com>
---
 drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c | 41 ++++++-----------------
 1 file changed, 10 insertions(+), 31 deletions(-)

diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c b/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c
index 4acfda235394..0ec25ae0b77e 100644
--- a/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c
+++ b/drivers/net/ethernet/stmicro/stmmac/dwmac-sun8i.c
@@ -17,7 +17,6 @@
 #include <linux/of_platform.h>
 #include <linux/phy.h>
 #include <linux/platform_device.h>
-#include <linux/pm_runtime.h>
 #include <linux/regulator/consumer.h>
 #include <linux/regmap.h>
 #include <linux/stmmac.h>
@@ -941,6 +940,13 @@ static int sun8i_dwmac_register_mdio_mux(struct stmmac_priv *priv)
 	return ret;
 }
 
+static void sun8i_dwmac_unregister_mdio_mux(struct stmmac_priv *priv)
+{
+	struct sunxi_priv_data *gmac = priv->plat->bsp_priv;
+
+	mdio_mux_uninit(gmac->mux_handle);
+}
+
 static int sun8i_dwmac_set_syscon(struct device *dev,
 				  struct plat_stmmacenet_data *plat)
 {
@@ -1133,8 +1139,6 @@ static int sun8i_dwmac_probe(struct platform_device *pdev)
 	struct stmmac_resources stmmac_res;
 	struct sunxi_priv_data *gmac;
 	struct device *dev = &pdev->dev;
-	struct stmmac_priv *priv;
-	struct net_device *ndev;
 	struct regmap *regmap;
 	int ret;
 
@@ -1217,6 +1221,9 @@ static int sun8i_dwmac_probe(struct platform_device *pdev)
 		ret = get_ephy_nodes(dev, gmac);
 		if (ret)
 			return ret;
+
+		plat_dat->mdio_init = sun8i_dwmac_register_mdio_mux;
+		plat_dat->mdio_exit = sun8i_dwmac_unregister_mdio_mux;
 	}
 
 	ret = sun8i_dwmac_set_syscon(&pdev->dev, plat_dat);
@@ -1227,33 +1234,8 @@ static int sun8i_dwmac_probe(struct platform_device *pdev)
 	if (ret)
 		goto dwmac_syscon;
 
-	ndev = dev_get_drvdata(&pdev->dev);
-	priv = netdev_priv(ndev);
-
-	/* the MAC is runtime suspended after stmmac_dvr_probe(), so we
-	 * need to ensure the MAC resume back before other operations such
-	 * as reset.
-	 */
-	pm_runtime_get_sync(&pdev->dev);
-
-	/* The mux must be registered after parent MDIO
-	 * so after stmmac_dvr_probe()
-	 */
-	if (gmac->variant->soc_has_internal_phy) {
-		ret = sun8i_dwmac_register_mdio_mux(priv);
-		if (ret) {
-			dev_err(&pdev->dev, "Failed to register mux\n");
-			goto dwmac_remove;
-		}
-	}
-
-	pm_runtime_put(&pdev->dev);
-
 	return 0;
 
-dwmac_remove:
-	pm_runtime_put_noidle(&pdev->dev);
-	stmmac_pltfr_remove(pdev);
 dwmac_syscon:
 	sun8i_dwmac_unset_syscon(gmac);
 dwmac_ephy:
@@ -1269,9 +1251,6 @@ static void sun8i_dwmac_remove(struct platform_device *pdev)
 	struct stmmac_priv *priv = netdev_priv(ndev);
 	struct sunxi_priv_data *gmac = priv->plat->bsp_priv;
 
-	if (gmac->variant->soc_has_internal_phy)
-		mdio_mux_uninit(gmac->mux_handle);
-
 	stmmac_pltfr_remove(pdev);
 	reset_control_put(gmac->rst_ephy);
 	clk_put(gmac->ephy_clk);

-- 
2.55.0



^ permalink raw reply related	[flat|nested] 7+ messages in thread

* Re: [PATCH net-next v4 3/5] net: stmmac: Balance runtime PM when the probe fails
  2026-11-20 10:45 ` [PATCH net-next v4 3/5] net: stmmac: Balance runtime PM when the probe fails Arun Iyengar via B4 Relay
@ 2026-11-21 10:46   ` sashiko-bot
  0 siblings, 0 replies; 7+ messages in thread
From: sashiko-bot @ 2026-11-21 10:46 UTC (permalink / raw)
  To: Arun Iyengar; +Cc: linux-sunxi

Thank you for your contribution! Sashiko AI review found 2 potential
issues to consider:
- [Medium] net: stmmac: unbalanced pm_runtime_disable() in
  __stmmac_dvr_probe()
- [Low] net: stmmac: missing Fixes trailer for probe runtime PM leak
--

commit 150d8ab7f1c2f574c3442bec9dafca1bba988cdd
Author: Arun Iyengar <aiyengar1990@gmail.com>

> net: stmmac: Balance runtime PM when the probe fails
>
> __stmmac_dvr_probe() enables runtime PM and takes a usage count before
> it registers the MDIO bus, but its error path undoes neither. After a
> failed or deferred probe the device never runtime-suspends.
>
> Undo both, as stmmac_dvr_remove() does.
>
> Signed-off-by: Arun Iyengar <aiyengar1990@gmail.com>

[Severity: Low]
Should this bug fix include a Fixes tag for the commit that added runtime
PM handling to probe without cleaning it up on the error path?

  Fixes: 5ec55823438e ("net: stmmac: add clocks management for gmac driver")

> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> index f256244ce0d11..4ebf4b6253743 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> @@ -8407,6 +8407,8 @@ static int __stmmac_dvr_probe(struct device *device,
>  error_pcs_setup:
>  	stmmac_mdio_unregister(ndev);
>  error_mdio_register:
> +	pm_runtime_disable(device);
> +	pm_runtime_put_noidle(device);
>  	stmmac_napi_del(ndev);
>  error_hw_init:
>  	destroy_workqueue(priv->wq);

[Severity: Medium]
Can calling pm_runtime_disable(device) unconditionally here leave runtime
PM unbalanced on buses that enable runtime PM before probing?

Earlier in __stmmac_dvr_probe(), pm_runtime_enable() is only called when
runtime PM is not already enabled:

drivers/net/ethernet/stmicro/stmmac/stmmac_main.c:__stmmac_dvr_probe() {
    ...
	pm_runtime_get_noresume(device);
	pm_runtime_set_active(device);
	if (!pm_runtime_enabled(device))
		pm_runtime_enable(device);
    ...
}

For PCI stmmac devices, pci_bus_add_device() enables runtime PM before the
driver probes, so __stmmac_dvr_probe() skips pm_runtime_enable().  If
stmmac_mdio_register() or a later step fails or defers, this error path
still calls pm_runtime_disable(device), incrementing disable_depth from 0
to 1.  When probe unwind runs pm_runtime_reinit(), it sees runtime PM
disabled and forces the device to RPM_SUSPENDED while still in D0,
decrementing the parent bridge's active child count and causing a
subsequent re-probe's pm_runtime_get_sync() in local_pci_probe() to fail
with -EACCES.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261120-sun8i-emac-mux-uaf-v4-0-57eba8915f25@gmail.com?part=3

^ permalink raw reply	[flat|nested] 7+ messages in thread

end of thread, other threads:[~2026-11-21 10:46 UTC | newest]

Thread overview: 7+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-11-20 10:45 [PATCH net-next v4 0/5] net: stmmac: dwmac-sun8i: Fix the EPHY and MDIO mux ordering Arun Iyengar via B4 Relay
2026-11-20 10:45 ` [PATCH net-next v4 1/5] net: stmmac: dwmac-sun8i: Put the EPHY clock and reset after the MAC Arun Iyengar via B4 Relay
2026-11-20 10:45 ` [PATCH net-next v4 2/5] net: stmmac: dwmac-sun8i: Get the EPHY clock and reset before the MAC probe Arun Iyengar via B4 Relay
2026-11-20 10:45 ` [PATCH net-next v4 3/5] net: stmmac: Balance runtime PM when the probe fails Arun Iyengar via B4 Relay
2026-11-21 10:46   ` sashiko-bot
2026-11-20 10:45 ` [PATCH net-next v4 4/5] net: stmmac: Add MDIO init and exit platform hooks Arun Iyengar via B4 Relay
2026-11-20 10:45 ` [PATCH net-next v4 5/5] net: stmmac: dwmac-sun8i: Register the MDIO mux in the MDIO setup Arun Iyengar via B4 Relay

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox