[PATCH v4] net: mvneta: re-enable percpu interrupt on resume

[PATCH v4] net: mvneta: re-enable percpu interrupt on resume

[Yun Zhou]

On Marvell MPIC platforms (Armada 370/XP/38x), mvneta uses a percpu
IRQ disable/enable scheme for NAPI: the ISR (mvneta_percpu_isr) calls
disable_percpu_irq() to mask the MPIC per-CPU interrupt and schedules
NAPI poll, which calls enable_percpu_irq() on completion to unmask.

If suspend occurs while NAPI poll is pending (between
disable_percpu_irq in the ISR and enable_percpu_irq in poll
completion), the interrupt is never re-enabled:

  1. mvneta_percpu_isr: disable_percpu_irq() + napi_schedule()
     => MPIC masked, percpu_enabled cpumask bit cleared
  2. NAPI poll does not complete before suspend proceeds
     (on PREEMPT_RT this is highly likely since softirqs run in
     ksoftirqd which gets frozen; on non-RT it can happen when
     softirq processing is deferred to ksoftirqd)
  3. mvneta_stop_dev => napi_disable(): cancels the pending poll
     without executing the completion path
  4. suspend_device_irqs => IRQCHIP_MASK_ON_SUSPEND: masks MPIC
     (already masked, but records IRQS_SUSPENDED)
  5. Resume: mpic_resume checks irq_percpu_is_enabled() => false
     (bit was cleared in step 1) => skips unmask
  6. mvneta_start_dev only restores device-level INTR_NEW_MASK,
     does not touch the MPIC per-CPU mask

Result: MPIC per-CPU interrupt stays masked permanently. The NIC
generates interrupts (INTR_NEW_CAUSE != 0) but the CPU never
receives them, causing complete loss of network connectivity.

Fix by calling on_each_cpu(mvneta_percpu_enable) in the resume path
to unconditionally unmask the MPIC per-CPU interrupt regardless of
pre-suspend state.

Fixes: 12bb03b436da ("net: mvneta: Handle per-cpu interrupts")
Signed-off-by: Yun Zhou <yun.zhou@windriver.com>
---
v4:
  - Rewrite commit message with accurate root cause analysis.

v3:
  - Dropped the free_irq/request_irq approach (incorrect root cause).
  - Instead, call on_each_cpu(mvneta_percpu_enable) in the resume path
    to ensure the MPIC percpu IRQ is unmasked, matching mvneta_open().
  - Updated commit message with correct root cause analysis.

v2:
  - Move request_irq before cpuhp registration in resume (matching
    mvneta_open ordering) so that failure does not leave cpuhp
    callbacks registered on a non-functional device.
  - On request_irq failure, call netif_device_detach() to prevent
    further traffic on the dead interface.

 drivers/net/ethernet/marvell/mvneta.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/net/ethernet/marvell/mvneta.c b/drivers/net/ethernet/marvell/mvneta.c
index 488f2663ad2c..543e566425c1 100644
--- a/drivers/net/ethernet/marvell/mvneta.c
+++ b/drivers/net/ethernet/marvell/mvneta.c
@@ -5918,6 +5918,9 @@ static int mvneta_resume(struct device *device)
 	rtnl_unlock();
 	mvneta_set_rx_mode(dev);
 
+	if (!pp->neta_armada3700)
+		on_each_cpu(mvneta_percpu_enable, pp, true);
+
 	return 0;
 }
 #endif
-- 
2.43.0

Re: [PATCH v4] net: mvneta: re-enable percpu interrupt on resume

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:

Pre-existing issues:
- [High] Unsafe call to mvneta_set_rx_mode() without netif_addr_lock_bh() in PM resume.
- [High] Race condition in mvneta_open() and mvneta_resume() with CPU hotplug causing double napi_enable().
--

commit bf4b34826006aca50b2726e755de72b6f2a7c330
Author: Yun Zhou <yun.zhou@windriver.com>

net: mvneta: re-enable percpu interrupt on resume

On Marvell MPIC platforms (Armada 370/XP/38x), mvneta uses a percpu IRQ 
disable/enable scheme for NAPI. This commit fixes an issue where the MPIC 
per-CPU interrupt stays masked permanently after suspend by unmasking it 
unconditionally during the resume path.

> diff --git a/drivers/net/ethernet/marvell/mvneta.c b/drivers/net/ethernet/marvell/mvneta.c
> index 0c061fb0ed072..744d6585a949d 100644
> --- a/drivers/net/ethernet/marvell/mvneta.c
> +++ b/drivers/net/ethernet/marvell/mvneta.c
> @@ -5900,6 +5900,9 @@ static int mvneta_resume(struct device *device)
>  	rtnl_unlock();
>  	mvneta_set_rx_mode(dev);

[Severity: High]
This isn't a bug introduced by this patch, but is it safe to call 
mvneta_set_rx_mode() here without holding netif_addr_lock_bh()?

The mvneta_set_rx_mode() function iterates over the net_device's multicast 
list:

drivers/net/ethernet/marvell/mvneta.c:mvneta_set_rx_mode() {
	...
			if (!netdev_mc_empty(dev)) {
				netdev_for_each_mc_addr(ha, dev) {
					mvneta_mcast_addr_set(pp, ha->addr,
							      pp->rxq_def);
				}
			}
	...
}

Since rtnl_unlock() is called immediately before this, it looks like this 
unprotected iteration could race with concurrent softirqs handling IGMP/MLD 
packets. Could this corrupt the list and lead to use-after-free crashes or 
infinite loops?

>  
> +	if (!pp->neta_armada3700)
> +		on_each_cpu(mvneta_percpu_enable, pp, true);
> +

[Severity: High]
This is a pre-existing issue, but does the resume path have a race condition 
with CPU hotplug?

In functions like mvneta_open() and mvneta_resume(), the device is added to 
the CPU hotplug state via cpuhp_state_add_instance_nocalls() without 
preventing concurrent hotplug events (e.g., via cpus_read_lock()).

If a CPU hotplug event triggers concurrently, mvneta_cpu_online() could call 
napi_enable_locked() for the new CPU:

drivers/net/ethernet/marvell/mvneta.c:mvneta_cpu_online() {
	...
	napi_enable_locked(&port->napi);
	...
}

If mvneta_start_dev() then runs and blindly iterates over all online CPUs 
calling napi_enable(), could this result in a double napi_enable() on the 
same NAPI instance and trigger a BUG_ON() panic in the core NAPI code?

>  	return 0;
>  }
> #endif

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260622074350.1666290-1-yun.zhou@windriver.com?part=1

Re: [PATCH v4] net: mvneta: re-enable percpu interrupt on resume

[Sebastian Andrzej Siewior]

On 2026-06-22 15:43:50 [+0800], Yun Zhou wrote:
> On Marvell MPIC platforms (Armada 370/XP/38x), mvneta uses a percpu
…
> Fix by calling on_each_cpu(mvneta_percpu_enable) in the resume path
> to unconditionally unmask the MPIC per-CPU interrupt regardless of
> pre-suspend state.
> 
> Fixes: 12bb03b436da ("net: mvneta: Handle per-cpu interrupts")
> Signed-off-by: Yun Zhou <yun.zhou@windriver.com>

Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

Sebastian

Re: [PATCH v4] net: mvneta: re-enable percpu interrupt on resume

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to netdev/net.git (main)
by Jakub Kicinski <kuba@kernel.org>:

On Mon, 22 Jun 2026 15:43:50 +0800 you wrote:
> On Marvell MPIC platforms (Armada 370/XP/38x), mvneta uses a percpu
> IRQ disable/enable scheme for NAPI: the ISR (mvneta_percpu_isr) calls
> disable_percpu_irq() to mask the MPIC per-CPU interrupt and schedules
> NAPI poll, which calls enable_percpu_irq() on completion to unmask.
> 
> If suspend occurs while NAPI poll is pending (between
> disable_percpu_irq in the ISR and enable_percpu_irq in poll
> completion), the interrupt is never re-enabled:
> 
> [...]

Here is the summary with links:
  - [v4] net: mvneta: re-enable percpu interrupt on resume
    https://git.kernel.org/netdev/net/c/fd398d648098

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html