Re: [PATCH net-next 2/3] net/mlx5: Allocate cpu_mask on heap to fix frame size warning for large NR_CPUS

[Zhu Yanjun <yanjun.zhu@linux.dev>]

在 2025/7/17 10:03, Tariq Toukan 写道:
> From: Shay Drory <shayd@nvidia.com>
> 
> When NR_CPUS is set to 8192 or higher, the current implementation that
> allocates struct cpu_mask on the stack leads to a compiler warning
> about the frame size[1].
> 
> This patch addresses the issue by moving the allocation of struct
> cpu_mask to the heap.
> 
> [1]
> drivers/net/ethernet/mellanox/mlx5/core/irq_affinity.c: In function ‘irq_pool_request_irq’:
> drivers/net/ethernet/mellanox/mlx5/core/irq_affinity.c:70:1: warning:
> the frame size of 1048 bytes is larger than 1024 bytes
> [-Wframe-larger-than=]
>     70 | }
>        | ^
> drivers/net/ethernet/mellanox/mlx5/core/pci_irq.c: In function ‘mlx5_ctrl_irq_request’:
> drivers/net/ethernet/mellanox/mlx5/core/pci_irq.c:478:1: warning: the
> frame size of 1040 bytes is larger than 1024 bytes [-Wframe-larger-than=]
>    478 | }
>        | ^
> drivers/net/ethernet/mellanox/mlx5/core/pci_irq.c: In function ‘mlx5_irq_request_vector’:
> drivers/net/ethernet/mellanox/mlx5/core/pci_irq.c:597:1: warning: the
> frame size of 1040 bytes is larger than 1024 bytes [-Wframe-larger-than=]
>    597 | }
>        | ^
> 
> drivers/net/ethernet/mellanox/mlx5/core/eq.c: In function ‘comp_irq_request_sf’:
> drivers/net/ethernet/mellanox/mlx5/core/eq.c:925:1: warning: the frame
> size of 1064 bytes is larger than 1024 bytes [-Wframe-larger-than=]
>    925 | }
>        | ^
> 
> drivers/net/ethernet/mellanox/mlx5/core/irq_affinity.c: In function ‘irq_pool_request_irq’:
> drivers/net/ethernet/mellanox/mlx5/core/irq_affinity.c:74:1: warning:
> the frame size of 1048 bytes is larger than 1024 bytes
> [-Wframe-larger-than=]
>     74 | }
>        | ^
> 

https://patchwork.kernel.org/project/linux-rdma/patch/20250711030359.4419-1-yanjun.zhu@linux.dev/

This commit appears to be the same as the one above.
Does the issue addressed in this commit still occur after the previous 
one is applied?

Thanks,
Yanjun.Zhu

> Signed-off-by: Shay Drory <shayd@nvidia.com>
> Reported-by: Arnd Bergmann <arnd@kernel.org>
> Closes: https://lore.kernel.org/all/20250620111010.3364606-1-arnd@kernel.org
> Reviewed-by: Maher Sanalla <msanalla@nvidia.com>
> Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
> Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
> Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
> ---
>   drivers/net/ethernet/mellanox/mlx5/core/eq.c  | 19 +++++++---
>   .../mellanox/mlx5/core/irq_affinity.c         | 21 ++++++++---
>   .../net/ethernet/mellanox/mlx5/core/pci_irq.c | 37 +++++++++++++------
>   3 files changed, 53 insertions(+), 24 deletions(-)
> 
> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eq.c b/drivers/net/ethernet/mellanox/mlx5/core/eq.c
> index 66dce17219a6..779efc186255 100644
> --- a/drivers/net/ethernet/mellanox/mlx5/core/eq.c
> +++ b/drivers/net/ethernet/mellanox/mlx5/core/eq.c
> @@ -876,19 +876,25 @@ static int comp_irq_request_sf(struct mlx5_core_dev *dev, u16 vecidx)
>   {
>   	struct mlx5_irq_pool *pool = mlx5_irq_table_get_comp_irq_pool(dev);
>   	struct mlx5_eq_table *table = dev->priv.eq_table;
> -	struct irq_affinity_desc af_desc = {};
> +	struct irq_affinity_desc *af_desc;
>   	struct mlx5_irq *irq;
>   
>   	/* In case SF irq pool does not exist, fallback to the PF irqs*/
>   	if (!mlx5_irq_pool_is_sf_pool(pool))
>   		return comp_irq_request_pci(dev, vecidx);
>   
> -	af_desc.is_managed = false;
> -	cpumask_copy(&af_desc.mask, cpu_online_mask);
> -	cpumask_andnot(&af_desc.mask, &af_desc.mask, &table->used_cpus);
> -	irq = mlx5_irq_affinity_request(dev, pool, &af_desc);
> -	if (IS_ERR(irq))
> +	af_desc = kzalloc(sizeof(*af_desc), GFP_KERNEL);
> +	if (!af_desc)
> +		return -ENOMEM;
> +
> +	af_desc->is_managed = false;
> +	cpumask_copy(&af_desc->mask, cpu_online_mask);
> +	cpumask_andnot(&af_desc->mask, &af_desc->mask, &table->used_cpus);
> +	irq = mlx5_irq_affinity_request(dev, pool, af_desc);
> +	if (IS_ERR(irq)) {
> +		kfree(af_desc);
>   		return PTR_ERR(irq);
> +	}
>   
>   	cpumask_or(&table->used_cpus, &table->used_cpus, mlx5_irq_get_affinity_mask(irq));
>   	mlx5_core_dbg(pool->dev, "IRQ %u mapped to cpu %*pbl, %u EQs on this irq\n",
> @@ -896,6 +902,7 @@ static int comp_irq_request_sf(struct mlx5_core_dev *dev, u16 vecidx)
>   		      cpumask_pr_args(mlx5_irq_get_affinity_mask(irq)),
>   		      mlx5_irq_read_locked(irq) / MLX5_EQ_REFS_PER_IRQ);
>   
> +	kfree(af_desc);
>   	return xa_err(xa_store(&table->comp_irqs, vecidx, irq, GFP_KERNEL));
>   }
>   
> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/irq_affinity.c b/drivers/net/ethernet/mellanox/mlx5/core/irq_affinity.c
> index 2691d88cdee1..d0a845579d33 100644
> --- a/drivers/net/ethernet/mellanox/mlx5/core/irq_affinity.c
> +++ b/drivers/net/ethernet/mellanox/mlx5/core/irq_affinity.c
> @@ -47,29 +47,38 @@ static int cpu_get_least_loaded(struct mlx5_irq_pool *pool,
>   static struct mlx5_irq *
>   irq_pool_request_irq(struct mlx5_irq_pool *pool, struct irq_affinity_desc *af_desc)
>   {
> -	struct irq_affinity_desc auto_desc = {};
> +	struct irq_affinity_desc *auto_desc;
>   	struct mlx5_irq *irq;
>   	u32 irq_index;
>   	int err;
>   
> +	auto_desc = kzalloc(sizeof(*auto_desc), GFP_KERNEL);
> +	if (!auto_desc)
> +		return ERR_PTR(-ENOMEM);
> +
>   	err = xa_alloc(&pool->irqs, &irq_index, NULL, pool->xa_num_irqs, GFP_KERNEL);
> -	if (err)
> -		return ERR_PTR(err);
> +	if (err) {
> +		irq = ERR_PTR(err);
> +		goto out;
> +	}
>   	if (pool->irqs_per_cpu) {
>   		if (cpumask_weight(&af_desc->mask) > 1)
>   			/* if req_mask contain more then one CPU, set the least loadad CPU
>   			 * of req_mask
>   			 */
>   			cpumask_set_cpu(cpu_get_least_loaded(pool, &af_desc->mask),
> -					&auto_desc.mask);
> +					&auto_desc->mask);
>   		else
>   			cpu_get(pool, cpumask_first(&af_desc->mask));
>   	}
>   	irq = mlx5_irq_alloc(pool, irq_index,
> -			     cpumask_empty(&auto_desc.mask) ? af_desc : &auto_desc,
> -			     NULL);
> +			     cpumask_empty(&auto_desc->mask) ?
> +			     af_desc : auto_desc, NULL);
>   	if (IS_ERR(irq))
>   		xa_erase(&pool->irqs, irq_index);
> +
> +out:
> +	kfree(auto_desc);
>   	return irq;
>   }
>   
> diff --git a/drivers/net/ethernet/mellanox/mlx5/core/pci_irq.c b/drivers/net/ethernet/mellanox/mlx5/core/pci_irq.c
> index 40024cfa3099..ac00aa29e61a 100644
> --- a/drivers/net/ethernet/mellanox/mlx5/core/pci_irq.c
> +++ b/drivers/net/ethernet/mellanox/mlx5/core/pci_irq.c
> @@ -470,26 +470,32 @@ void mlx5_ctrl_irq_release(struct mlx5_core_dev *dev, struct mlx5_irq *ctrl_irq)
>   struct mlx5_irq *mlx5_ctrl_irq_request(struct mlx5_core_dev *dev)
>   {
>   	struct mlx5_irq_pool *pool = ctrl_irq_pool_get(dev);
> -	struct irq_affinity_desc af_desc;
> +	struct irq_affinity_desc *af_desc;
>   	struct mlx5_irq *irq;
>   
> -	cpumask_copy(&af_desc.mask, cpu_online_mask);
> -	af_desc.is_managed = false;
> +	af_desc = kzalloc(sizeof(*af_desc), GFP_KERNEL);
> +	if (!af_desc)
> +		return ERR_PTR(-ENOMEM);
> +
> +	cpumask_copy(&af_desc->mask, cpu_online_mask);
> +	af_desc->is_managed = false;
>   	if (!mlx5_irq_pool_is_sf_pool(pool)) {
>   		/* In case we are allocating a control IRQ from a pci device's pool.
>   		 * This can happen also for a SF if the SFs pool is empty.
>   		 */
>   		if (!pool->xa_num_irqs.max) {
> -			cpumask_clear(&af_desc.mask);
> +			cpumask_clear(&af_desc->mask);
>   			/* In case we only have a single IRQ for PF/VF */
> -			cpumask_set_cpu(cpumask_first(cpu_online_mask), &af_desc.mask);
> +			cpumask_set_cpu(cpumask_first(cpu_online_mask),
> +					&af_desc->mask);
>   		}
>   		/* Allocate the IRQ in index 0. The vector was already allocated */
> -		irq = irq_pool_request_vector(pool, 0, &af_desc, NULL);
> +		irq = irq_pool_request_vector(pool, 0, af_desc, NULL);
>   	} else {
> -		irq = mlx5_irq_affinity_request(dev, pool, &af_desc);
> +		irq = mlx5_irq_affinity_request(dev, pool, af_desc);
>   	}
>   
> +	kfree(af_desc);
>   	return irq;
>   }
>   
> @@ -548,16 +554,23 @@ struct mlx5_irq *mlx5_irq_request_vector(struct mlx5_core_dev *dev, u16 cpu,
>   {
>   	struct mlx5_irq_table *table = mlx5_irq_table_get(dev);
>   	struct mlx5_irq_pool *pool = table->pcif_pool;
> -	struct irq_affinity_desc af_desc;
>   	int offset = MLX5_IRQ_VEC_COMP_BASE;
> +	struct irq_affinity_desc *af_desc;
> +	struct mlx5_irq *irq;
> +
> +	af_desc = kzalloc(sizeof(*af_desc), GFP_KERNEL);
> +	if (!af_desc)
> +		return ERR_PTR(-ENOMEM);
>   
>   	if (!pool->xa_num_irqs.max)
>   		offset = 0;
>   
> -	af_desc.is_managed = false;
> -	cpumask_clear(&af_desc.mask);
> -	cpumask_set_cpu(cpu, &af_desc.mask);
> -	return mlx5_irq_request(dev, vecidx + offset, &af_desc, rmap);
> +	af_desc->is_managed = false;
> +	cpumask_clear(&af_desc->mask);
> +	cpumask_set_cpu(cpu, &af_desc->mask);
> +	irq = mlx5_irq_request(dev, vecidx + offset, af_desc, rmap);
> +	kfree(af_desc);
> +	return irq;
>   }
>   
>   static struct mlx5_irq_pool *