[PATCH net v2] amd-xgbe: fix sleep while atomic on suspend/resume

[PATCH net v2] amd-xgbe: fix sleep while atomic on suspend/resume

[Raju Rangoju]

The xgbe_powerdown() and xgbe_powerup() functions use spinlocks
(spin_lock_irqsave) while calling functions that may sleep:
- napi_disable() can sleep waiting for NAPI polling to complete
- flush_workqueue() can sleep waiting for pending work items

This causes a "BUG: scheduling while atomic" error during suspend/resume
cycles on systems using the AMD XGBE Ethernet controller.

The spinlock protection in these functions is unnecessary as these
functions are called from suspend/resume paths which are already serialized
by the PM core

Fix this by removing the spinlock from xgbe_powerdown() and xgbe_powerup()

Fixes: c5aa9e3b8156 ("amd-xgbe: Initial AMD 10GbE platform driver")
Signed-off-by: Raju Rangoju <Raju.Rangoju@amd.com>
---
 drivers/net/ethernet/amd/xgbe/xgbe-drv.c | 10 ----------
 1 file changed, 10 deletions(-)

diff --git a/drivers/net/ethernet/amd/xgbe/xgbe-drv.c b/drivers/net/ethernet/amd/xgbe/xgbe-drv.c
index 62bb4b8a68e1..8b79d88480db 100644
--- a/drivers/net/ethernet/amd/xgbe/xgbe-drv.c
+++ b/drivers/net/ethernet/amd/xgbe/xgbe-drv.c
@@ -1120,7 +1120,6 @@ int xgbe_powerdown(struct net_device *netdev, unsigned int caller)
 {
 	struct xgbe_prv_data *pdata = netdev_priv(netdev);
 	struct xgbe_hw_if *hw_if = &pdata->hw_if;
-	unsigned long flags;
 
 	DBGPR("-->xgbe_powerdown\n");
 
@@ -1131,8 +1130,6 @@ int xgbe_powerdown(struct net_device *netdev, unsigned int caller)
 		return -EINVAL;
 	}
 
-	spin_lock_irqsave(&pdata->lock, flags);
-
 	if (caller == XGMAC_DRIVER_CONTEXT)
 		netif_device_detach(netdev);
 
@@ -1148,8 +1145,6 @@ int xgbe_powerdown(struct net_device *netdev, unsigned int caller)
 
 	pdata->power_down = 1;
 
-	spin_unlock_irqrestore(&pdata->lock, flags);
-
 	DBGPR("<--xgbe_powerdown\n");
 
 	return 0;
@@ -1159,7 +1154,6 @@ int xgbe_powerup(struct net_device *netdev, unsigned int caller)
 {
 	struct xgbe_prv_data *pdata = netdev_priv(netdev);
 	struct xgbe_hw_if *hw_if = &pdata->hw_if;
-	unsigned long flags;
 
 	DBGPR("-->xgbe_powerup\n");
 
@@ -1170,8 +1164,6 @@ int xgbe_powerup(struct net_device *netdev, unsigned int caller)
 		return -EINVAL;
 	}
 
-	spin_lock_irqsave(&pdata->lock, flags);
-
 	pdata->power_down = 0;
 
 	xgbe_napi_enable(pdata, 0);
@@ -1186,8 +1178,6 @@ int xgbe_powerup(struct net_device *netdev, unsigned int caller)
 
 	xgbe_start_timers(pdata);
 
-	spin_unlock_irqrestore(&pdata->lock, flags);
-
 	DBGPR("<--xgbe_powerup\n");
 
 	return 0;
-- 
2.34.1

Re: [PATCH net v2] amd-xgbe: fix sleep while atomic on suspend/resume

[Jakub Kicinski]

On Thu, 26 Feb 2026 21:46:40 +0530 Raju Rangoju wrote:
> The xgbe_powerdown() and xgbe_powerup() functions use spinlocks
> (spin_lock_irqsave) while calling functions that may sleep:
> - napi_disable() can sleep waiting for NAPI polling to complete
> - flush_workqueue() can sleep waiting for pending work items
> 
> This causes a "BUG: scheduling while atomic" error during suspend/resume
> cycles on systems using the AMD XGBE Ethernet controller.
> 
> The spinlock protection in these functions is unnecessary as these
> functions are called from suspend/resume paths which are already serialized
> by the PM core
> 
> Fix this by removing the spinlock from xgbe_powerdown() and xgbe_powerup()

I think you're deleting the only code that takes this lock.
Why not delete it completely?
-- 
pw-bot: cr