Re: [PATCH v2] wifi: ath11k: fix resource leak on error in ext IRQ setup

On 7/13/2026 6:59 PM, ZhaoJinming wrote:
> In ath11k_ahb_config_irq(), when a CE request_irq() fails, the function
> returns the error immediately without freeing the CE IRQs that were
> successfully registered in previous loop iterations. The probe error
> path does not call ath11k_ahb_free_irq() either, so the previously
> registered CE IRQ handlers remain attached to the interrupt lines and
> are never released.
> 
> In ath11k_ahb_config_ext_irq(), when an external request_irq() fails,
> the error is only logged and the loop continues. The function then
> returns 0 indicating success, leaving the device in a partially
> configured state where some external IRQs are not registered. This
> causes enable_irq()/disable_irq()/free_irq() to be called on
> unregistered IRQs during runtime and remove/shutdown, triggering
> WARN_ON(!desc->action), and missing interrupt handlers lead to data
> loss.
> 
> Additionally, if alloc_netdev_dummy() fails for a later IRQ group, the
> function returns -ENOMEM without freeing the ext IRQs and napi_ndev
> that were successfully set up for earlier groups.
> 
> Fix all three issues: propagate the error up to the caller and unwind
> all successfully registered IRQs and allocated resources on failure.
> 
> Signed-off-by: ZhaoJinming <[email protected]>
> ---
> 
> Changes in v2:
> - Move `irq_grp` declaration from for-loop body to function scope to fix
>   compile error in err_request_irq error path.
> - Set ret = -ENOMEM before goto err_request_irq on alloc_netdev_dummy()
>   failure to avoid returning uninitialized value.
> - When ath11k_ahb_config_ext_irq() fails, unwind the already-registered
>   CE IRQs via a shared err_ce_irq cleanup path to avoid leaking them.
> 
>  drivers/net/wireless/ath/ath11k/ahb.c | 48 ++++++++++++++++++++++++++--
>  1 file changed, 44 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/net/wireless/ath/ath11k/ahb.c b/drivers/net/wireless/ath/ath11k/ahb.c
> index f566d699d074..11bafbd94aff 100644
> --- a/drivers/net/wireless/ath/ath11k/ahb.c
> +++ b/drivers/net/wireless/ath/ath11k/ahb.c
> @@ -524,20 +524,24 @@ static irqreturn_t ath11k_ahb_ext_interrupt_handler(int irq, void *arg)
>  static int ath11k_ahb_config_ext_irq(struct ath11k_base *ab)
>  {
>  	struct ath11k_hw_params *hw = &ab->hw_params;
> +	struct ath11k_ext_irq_grp *irq_grp;
>  	int i, j;
>  	int irq;
>  	int ret;
>  
>  	for (i = 0; i < ATH11K_EXT_IRQ_GRP_NUM_MAX; i++) {
> -		struct ath11k_ext_irq_grp *irq_grp = &ab->ext_irq_grp[i];
> +		irq_grp = &ab->ext_irq_grp[i];
>  		u32 num_irq = 0;

declaration after statement. better to have num_irq declaration at the beginning of the
for loop.

>  
>  		irq_grp->ab = ab;
>  		irq_grp->grp_id = i;
>  
>  		irq_grp->napi_ndev = alloc_netdev_dummy(0);
> -		if (!irq_grp->napi_ndev)
> -			return -ENOMEM;
> +		if (!irq_grp->napi_ndev) {
> +			ret = -ENOMEM;
> +			irq_grp->num_irq = 0;

the ath12k_base is allocated by kzalloc, so the ext_irq_grp array, as a member of
ath12k_base, is already zero'ed, hence this re-initialization is redundant. But I am OK
with it since it improves code readability.

> +			goto err_request_irq;
> +		}
>  
>  		netif_napi_add(irq_grp->napi_ndev, &irq_grp->napi,
>  			       ath11k_ahb_ext_grp_napi_poll);
> @@ -600,11 +604,25 @@ static int ath11k_ahb_config_ext_irq(struct ath11k_base *ab)
>  			if (ret) {
>  				ath11k_err(ab, "failed request_irq for %d\n",
>  					   irq);
> +				irq_grp->num_irq = j;

with this forcing to the number of successful irq request ...

> +				goto err_request_irq;
>  			}
>  		}
>  	}
>  
>  	return 0;
> +
> +err_request_irq:
> +	for ( ; i >= 0; i--) {

then with the index starting from the failed loop ...

> +		irq_grp = &ab->ext_irq_grp[i];
> +		for (j = irq_grp->num_irq - 1; j >= 0; j--)
> +			free_irq(ab->irq_num[irq_grp->irqs[j]], irq_grp);
> +		if (irq_grp->napi_ndev) {

and with the guard ...

> +			netif_napi_del(&irq_grp->napi);
> +			free_netdev(irq_grp->napi_ndev);
> +		}
> +	}

at the cost of above three, the centralized err handling is working. Yeah this simplifies
the code, but makes it hard to understand.

what about:

1. refactor the existing ath11k_ahb_free_ext_irq():

ath11k_ahb_free_ext_irq_grp()
{
	for (j = 0; j < irq_grp->num_irq; j++)
		free_irq(ab->irq_num[irq_grp->irqs[j]], irq_grp);

	netif_napi_del(&irq_grp->napi);
	free_netdev(irq_grp->napi_ndev);
}

then we can have

ath11k_ahb_free_ext_irq()
{
	for (i = 0; i < ATH11K_EXT_IRQ_GRP_NUM_MAX; i++)
		ath11k_ahb_free_ext_irq_grp(irq_grp)
}

and the complete err handling of ext irq can be

ath11k_ahb_config_ext_irq()
{
	for (i = 0; i < ATH11K_EXT_IRQ_GRP_NUM_MAX; i++) {

		irq_grp->napi_ndev = alloc_netdev_dummy(0);
		if (!irq_grp->napi_ndev) {
			ret = -ENOMEM;
			irq_grp->num_irq = 0;
			goto err_request_irq;
		}

		...

		for (j = 0; j < num_irq; j++) {
			ret = request_irq();
			if (ret) {
				irq_grp->num_irq = j;
				ath11k_ahb_free_ext_irq_grp(irq_grp);
				goto err_request_irq;
			}
		}

		irq_grp->num_irq = num_irq;
	}

err_request_irq:
	for ( i--; i >= 0; i--) {
		irq_grp = &ab->ext_irq_grp[i];
		ath11k_ahb_free_ext_irq_grp(irq_grp);
	}

	return ret;
}
	
> +	return ret;
>  }
>  
>  static int ath11k_ahb_config_irq(struct ath11k_base *ab)
> @@ -629,15 +647,30 @@ static int ath11k_ahb_config_irq(struct ath11k_base *ab)
>  		ret = request_irq(irq, ath11k_ahb_ce_interrupt_handler,
>  				  IRQF_TRIGGER_RISING, irq_name[irq_idx],
>  				  ce_pipe);
> -		if (ret)
> -			return ret;
> +		if (ret) {
> +			ath11k_err(ab, "failed request_irq for %d\n", irq);
> +			goto err_ce_irq;
> +		}
>  
>  		ab->irq_num[irq_idx] = irq;
>  	}
>  
>  	/* Configure external interrupts */
>  	ret = ath11k_ahb_config_ext_irq(ab);
> +	if (ret) {
> +		ath11k_err(ab, "failed to configure ext irq: %d\n", ret);
> +		goto err_ce_irq;
> +	}
> +
> +	return 0;
>  
> +err_ce_irq:
> +	for (i--; i >= 0; i--) {
> +		if (ath11k_ce_get_attr_flags(ab, i) & CE_ATTR_DIS_INTR)
> +			continue;
> +		free_irq(ab->irq_num[ATH11K_IRQ_CE0_OFFSET + i],
> +			 &ab->ce.ce_pipe[i]);
> +	}

this is also a duplication of the CE part of ath11k_ahb_free_irq(), consider refactoring
the exsiting helper and reuse it.

>  	return ret;
>  }
>