Re: [PATCH RFC 25/33] wifi: ath12k: Fix locking in error paths

[Jeff Johnson]

On 2/6/2025 9:51 AM, Bart Van Assche wrote:
> If ag->mutex has been locked, unlock it before returning. If it has not
> been locked, do not unlock it before returning. These bugs have been
> detected by the Clang thread-safety analyzer.
> 
> Cc: Karthikeyan Periyasamy <quic_periyasa@xxxxxxxxxxx>
> Cc: Jeff Johnson <jeff.johnson@xxxxxxxxxxxxxxxx>
> Fixes: ee146e11b4d9 ("wifi: ath12k: refactor core start based on hardware group")
> Signed-off-by: Bart Van Assche <bvanassche@xxxxxxx>
> ---
>  drivers/net/wireless/ath/ath12k/core.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/wireless/ath/ath12k/core.c b/drivers/net/wireless/ath/ath12k/core.c
> index 0606116d6b9c..212cd935e60a 100644
> --- a/drivers/net/wireless/ath/ath12k/core.c
> +++ b/drivers/net/wireless/ath/ath12k/core.c
> @@ -1122,16 +1122,18 @@ int ath12k_core_qmi_firmware_ready(struct ath12k_base *ab)
>  		ath12k_core_stop(ab);
>  		mutex_unlock(&ab->core_lock);
>  	}
> +	mutex_unlock(&ag->mutex);
>  	goto exit;
>  
>  err_dp_free:
>  	ath12k_dp_free(ab);
>  	mutex_unlock(&ab->core_lock);
> +	mutex_unlock(&ag->mutex);
> +
>  err_firmware_stop:
>  	ath12k_qmi_firmware_stop(ab);
>  
>  exit:
> -	mutex_unlock(&ag->mutex);
>  	return ret;
>  }
>  

This looks like a legitimate issue.

Can you submit this as a single patch and cc the ath12k and linux-wireless
lists (as I've done in this reply)