From: Daniel Jurgens <danielj@mellanox.com>
Date: Thu, 5 Dec 2019 16:58:10 -0600
Subject: net/mlx5: Don't request more than supported EQs
Patch-mainline: v5.12-rc7
Git-commit: a7b76002ae78cd230ee652ccdfedf21aa94fcecc
References: git-fixes

Calculating the number of compeltion EQs based on the number of
available IRQ vectors doesn't work now that all async EQs share one IRQ.
Thus the max number of EQs can be exceeded on systems with more than
approximately 256 CPUs. Take this into account when calculating the
number of available completion EQs.

Fixes: 81bfa206032a ("net/mlx5: Use a single IRQ for all async EQs")
Signed-off-by: Daniel Jurgens <danielj@mellanox.com>
Reviewed-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Acked-by: Thomas Bogendoerfer <tbogendoerfer@suse.de>
---
 drivers/net/ethernet/mellanox/mlx5/core/eq.c |   13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)

--- a/drivers/net/ethernet/mellanox/mlx5/core/eq.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eq.c
@@ -886,13 +886,24 @@ void mlx5_core_eq_free_irqs(struct mlx5_
 	mutex_unlock(&table->lock);
 }
 
+#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
+#define MLX5_MAX_ASYNC_EQS 4
+#else
+#define MLX5_MAX_ASYNC_EQS 3
+#endif
+
 int mlx5_eq_table_create(struct mlx5_core_dev *dev)
 {
 	struct mlx5_eq_table *eq_table = dev->priv.eq_table;
+	int num_eqs = MLX5_CAP_GEN(dev, max_num_eqs) ?
+		      MLX5_CAP_GEN(dev, max_num_eqs) :
+		      1 << MLX5_CAP_GEN(dev, log_max_eq);
 	int err;
 
 	eq_table->num_comp_eqs =
-		mlx5_irq_get_num_comp(eq_table->irq_table);
+		min_t(int,
+		      mlx5_irq_get_num_comp(eq_table->irq_table),
+		      num_eqs - MLX5_MAX_ASYNC_EQS);
 
 	err = create_async_eqs(dev);
 	if (err) {