Re: [PATCH] bus/fslmc: fix shadowed variables in queue storage macros

[Stephen Hemminger <stephen@networkplumber.org>]

On Sat,  4 Apr 2026 12:19:11 -0500
Weijun Pan <wpan3636@gmail.com> wrote:

> diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
> index e625a5c035..bbeccd8fc3 100644
> --- a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
> +++ b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
> @@ -204,33 +204,33 @@ struct swp_active_dqs {
>  
>  #define dpaa2_queue_storage_alloc(q, num) \
>  ({ \
> -	int ret = 0, i; \
> +	int qs_ret = 0, qs_idx; \
>  	\
> -	for (i = 0; i < (num); i++) { \
> -		(q)->q_storage[i] = rte_zmalloc(NULL, \
> +	for (qs_idx = 0; qs_idx < (num); qs_idx++) { \
> +		(q)->q_storage[qs_idx] = rte_zmalloc(NULL, \
>  			sizeof(struct queue_storage_info_t), \
>  			RTE_CACHE_LINE_SIZE); \
> -		if (!(q)->q_storage[i]) { \
> -			ret = -ENOBUFS; \
> +		if (!(q)->q_storage[qs_idx]) { \
> +			qs_ret = -ENOBUFS; \
>  			break; \
>  		} \
> -		ret = dpaa2_alloc_dq_storage((q)->q_storage[i]); \
> -		if (ret) \
> +		qs_ret = dpaa2_alloc_dq_storage((q)->q_storage[qs_idx]); \
> +		if (qs_ret) \
>  			break; \
>  	} \
> -	ret; \
> +	qs_ret; \
>  })
>  
>  #define dpaa2_queue_storage_free(q, num) \
>  ({ \
>  	if (q) { \
> -		int i; \
> +		int qs_idx; \
>  		\
> -		for (i = 0; i < (num); i++) { \
> -			if ((q)->q_storage[i]) { \
> -				dpaa2_free_dq_storage((q)->q_storage[i]); \
> -				rte_free((q)->q_storage[i]); \
> -				(q)->q_storage[i] = NULL; \
> +		for (qs_idx = 0; qs_idx < (num); qs_idx++) { \
> +			if ((q)->q_storage[qs_idx]) { \
> +				dpaa2_free_dq_storage((q)->q_storage[qs_idx]); \
> +				rte_free((q)->q_storage[qs_idx]); \
> +				(q)->q_storage[qs_idx] = NULL; \
>  			} \
>  		} \
>  	} \


Why are these not inline functions.
Macros with lower case names are likely place fore confusion like this?