diff mbox series

[net-next,1/5] net: dpaa: avoid on-stack arrays of NR_CPUS elements

Message ID 20240710230025.46487-2-vladimir.oltean@nxp.com (mailing list archive)
State New, archived
Headers show
Series Eliminate CONFIG_NR_CPUS dependency in dpaa-eth and enable COMPILE_TEST in fsl_qbman | expand

Commit Message

Vladimir Oltean July 10, 2024, 11 p.m. UTC 
The dpaa-eth driver is written for PowerPC and Arm SoCs which have 1-24
CPUs. It depends on CONFIG_NR_CPUS having a reasonably small value in
Kconfig. Otherwise, there are 2 functions which allocate on-stack arrays
of NR_CPUS elements, and these can quickly explode in size, leading to
warnings such as:

  drivers/net/ethernet/freescale/dpaa/dpaa_eth.c:3280:12: warning:
  stack frame size (16664) exceeds limit (2048) in 'dpaa_eth_probe' [-Wframe-larger-than]

The problem is twofold:
- Reducing the array size to the boot-time num_possible_cpus() (rather
  than the compile-time NR_CPUS) creates a variable-length array,
  which should be avoided in the Linux kernel.
- Using NR_CPUS as an array size makes the driver blow up in stack
  consumption with generic, as opposed to hand-crafted, .config files.

A simple solution is to use dynamic allocation for num_possible_cpus()
elements (aka a small number determined at runtime).

Link: https://lore.kernel.org/all/202406261920.l5pzM1rj-lkp@intel.com/
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
---
 .../net/ethernet/freescale/dpaa/dpaa_eth.c    | 20 ++++++++++++++-----
 .../ethernet/freescale/dpaa/dpaa_ethtool.c    | 10 +++++++++-
 2 files changed, 24 insertions(+), 6 deletions(-)

Comments

Breno Leitao July 11, 2024, 11:16 a.m. UTC | #1 
Hello Vladimir,

On Thu, Jul 11, 2024 at 02:00:21AM +0300, Vladimir Oltean wrote:
> The dpaa-eth driver is written for PowerPC and Arm SoCs which have 1-24
> CPUs. It depends on CONFIG_NR_CPUS having a reasonably small value in
> Kconfig. Otherwise, there are 2 functions which allocate on-stack arrays
> of NR_CPUS elements, and these can quickly explode in size, leading to
> warnings such as:
> 
>   drivers/net/ethernet/freescale/dpaa/dpaa_eth.c:3280:12: warning:
>   stack frame size (16664) exceeds limit (2048) in 'dpaa_eth_probe' [-Wframe-larger-than]
> 
> The problem is twofold:
> - Reducing the array size to the boot-time num_possible_cpus() (rather
>   than the compile-time NR_CPUS) creates a variable-length array,
>   which should be avoided in the Linux kernel.
> - Using NR_CPUS as an array size makes the driver blow up in stack
>   consumption with generic, as opposed to hand-crafted, .config files.
> 
> A simple solution is to use dynamic allocation for num_possible_cpus()
> elements (aka a small number determined at runtime).
> 
> Link: https://lore.kernel.org/all/202406261920.l5pzM1rj-lkp@intel.com/
> Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Breno Leitao <leitao@debian.org>

Thanks for working on it.

--breno
diff mbox series

Patch

diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
index ddeb0a5f2317..c856b556929d 100644
--- a/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_eth.c
@@ -931,14 +931,18 @@  static inline void dpaa_setup_egress(const struct dpaa_priv *priv,
 	}
 }
 
-static void dpaa_fq_setup(struct dpaa_priv *priv,
-			  const struct dpaa_fq_cbs *fq_cbs,
-			  struct fman_port *tx_port)
+static int dpaa_fq_setup(struct dpaa_priv *priv,
+			 const struct dpaa_fq_cbs *fq_cbs,
+			 struct fman_port *tx_port)
 {
 	int egress_cnt = 0, conf_cnt = 0, num_portals = 0, portal_cnt = 0, cpu;
 	const cpumask_t *affine_cpus = qman_affine_cpus();
-	u16 channels[NR_CPUS];
 	struct dpaa_fq *fq;
+	u16 *channels;
+
+	channels = kcalloc(num_possible_cpus(), sizeof(u16), GFP_KERNEL);
+	if (!channels)
+		return -ENOMEM;
 
 	for_each_cpu_and(cpu, affine_cpus, cpu_online_mask)
 		channels[num_portals++] = qman_affine_channel(cpu);
@@ -997,6 +1001,10 @@  static void dpaa_fq_setup(struct dpaa_priv *priv,
 				break;
 		}
 	}
+
+	kfree(channels);
+
+	return 0;
 }
 
 static inline int dpaa_tx_fq_to_id(const struct dpaa_priv *priv,
@@ -3416,7 +3424,9 @@  static int dpaa_eth_probe(struct platform_device *pdev)
 	 */
 	dpaa_eth_add_channel(priv->channel, &pdev->dev);
 
-	dpaa_fq_setup(priv, &dpaa_fq_cbs, priv->mac_dev->port[TX]);
+	err = dpaa_fq_setup(priv, &dpaa_fq_cbs, priv->mac_dev->port[TX]);
+	if (err)
+		goto free_dpaa_bps;
 
 	/* Create a congestion group for this netdev, with
 	 * dynamically-allocated CGR ID.
diff --git a/drivers/net/ethernet/freescale/dpaa/dpaa_ethtool.c b/drivers/net/ethernet/freescale/dpaa/dpaa_ethtool.c
index 5bd0b36d1feb..3f8cd4a7d845 100644
--- a/drivers/net/ethernet/freescale/dpaa/dpaa_ethtool.c
+++ b/drivers/net/ethernet/freescale/dpaa/dpaa_ethtool.c
@@ -457,12 +457,16 @@  static int dpaa_set_coalesce(struct net_device *dev,
 			     struct netlink_ext_ack *extack)
 {
 	const cpumask_t *cpus = qman_affine_cpus();
-	bool needs_revert[NR_CPUS] = {false};
 	struct qman_portal *portal;
 	u32 period, prev_period;
 	u8 thresh, prev_thresh;
+	bool *needs_revert;
 	int cpu, res;
 
+	needs_revert = kcalloc(num_possible_cpus(), sizeof(bool), GFP_KERNEL);
+	if (!needs_revert)
+		return -ENOMEM;
+
 	period = c->rx_coalesce_usecs;
 	thresh = c->rx_max_coalesced_frames;
 
@@ -485,6 +489,8 @@  static int dpaa_set_coalesce(struct net_device *dev,
 		needs_revert[cpu] = true;
 	}
 
+	kfree(needs_revert);
+
 	return 0;
 
 revert_values:
@@ -498,6 +504,8 @@  static int dpaa_set_coalesce(struct net_device *dev,
 		qman_dqrr_set_ithresh(portal, prev_thresh);
 	}
 
+	kfree(needs_revert);
+
 	return res;
 }