mlx4_core: Allow choosing flow steering mode

Device managed flow steering will be enabled only under administrator
directive provided through setting the existing module parameter
log_num_mgm_entry_size to -1 (if the device actually supports flow
steering).  If flow steering isn't requested or not available, the
driver will use the value of log_num_mgm_entry_size and B0 steering.

Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>

4 files changed, 50 insertions, 16 deletions

diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c
index 680d81026cbd..e1bafffbc3b1 100644
--- a/drivers/net/ethernet/mellanox/mlx4/main.c
+++ b/drivers/net/ethernet/mellanox/mlx4/main.c
@@ -85,15 +85,15 @@ static int probe_vf;
 module_param(probe_vf, int, 0644);
 MODULE_PARM_DESC(probe_vf, "number of vfs to probe by pf driver (num_vfs > 0)");
 
-int mlx4_log_num_mgm_entry_size = 10;
+int mlx4_log_num_mgm_entry_size = MLX4_DEFAULT_MGM_LOG_ENTRY_SIZE;
 module_param_named(log_num_mgm_entry_size,
                         mlx4_log_num_mgm_entry_size, int, 0444);
 MODULE_PARM_DESC(log_num_mgm_entry_size, "log mgm size, that defines the num"
                                          " of qp per mcg, for example:"
-                                         " 10 gives 248.range: 9<="
+                                         " 10 gives 248.range: 7 <="
                                          " log_num_mgm_entry_size <= 12."
-                                         " Not in use with device managed"
-                                         " flow steering");
+                                         " To activate device managed"
+                                         " flow steering when available, set to -1");
 
 static bool enable_64b_cqe_eqe;
 module_param(enable_64b_cqe_eqe, bool, 0444);
@@ -1318,12 +1318,30 @@ static void mlx4_parav_master_pf_caps(struct mlx4_dev *dev)
         }
 }
 
+static int choose_log_fs_mgm_entry_size(int qp_per_entry)
+{
+        int i = MLX4_MIN_MGM_LOG_ENTRY_SIZE;
+
+        for (i = MLX4_MIN_MGM_LOG_ENTRY_SIZE; i <= MLX4_MAX_MGM_LOG_ENTRY_SIZE;
+              i++) {
+                if (qp_per_entry <= 4 * ((1 << i) / 16 - 2))
+                        break;
+        }
+
+        return (i <= MLX4_MAX_MGM_LOG_ENTRY_SIZE) ? i : -1;
+}
+
 static void choose_steering_mode(struct mlx4_dev *dev,
                                  struct mlx4_dev_cap *dev_cap)
 {
-        if (dev_cap->flags2 & MLX4_DEV_CAP_FLAG2_FS_EN &&
+        if (mlx4_log_num_mgm_entry_size == -1 &&
+            dev_cap->flags2 & MLX4_DEV_CAP_FLAG2_FS_EN &&
             (!mlx4_is_mfunc(dev) ||
-             (dev_cap->fs_max_num_qp_per_entry >= (num_vfs + 1)))) {
+             (dev_cap->fs_max_num_qp_per_entry >= (num_vfs + 1))) &&
+            choose_log_fs_mgm_entry_size(dev_cap->fs_max_num_qp_per_entry) >=
+                MLX4_MIN_MGM_LOG_ENTRY_SIZE) {
+                dev->oper_log_mgm_entry_size =
+                        choose_log_fs_mgm_entry_size(dev_cap->fs_max_num_qp_per_entry);
                 dev->caps.steering_mode = MLX4_STEERING_MODE_DEVICE_MANAGED;
                 dev->caps.num_qp_per_mgm = dev_cap->fs_max_num_qp_per_entry;
                 dev->caps.fs_log_max_ucast_qp_range_size =
@@ -1340,10 +1358,17 @@ static void choose_steering_mode(struct mlx4_dev *dev,
                                 mlx4_warn(dev, "Must have both UC_STEER and MC_STEER flags "
                                           "set to use B0 steering. Falling back to A0 steering mode.\n");
                 }
+                dev->oper_log_mgm_entry_size =
+                        mlx4_log_num_mgm_entry_size > 0 ?
+                        mlx4_log_num_mgm_entry_size :
+                        MLX4_DEFAULT_MGM_LOG_ENTRY_SIZE;
                 dev->caps.num_qp_per_mgm = mlx4_get_qp_per_mgm(dev);
         }
-        mlx4_dbg(dev, "Steering mode is: %s\n",
-                 mlx4_steering_mode_str(dev->caps.steering_mode));
+        mlx4_dbg(dev, "Steering mode is: %s, oper_log_mgm_entry_size = %d, "
+                 "modparam log_num_mgm_entry_size = %d\n",
+                 mlx4_steering_mode_str(dev->caps.steering_mode),
+                 dev->oper_log_mgm_entry_size,
+                 mlx4_log_num_mgm_entry_size);
 }
 
 static int mlx4_init_hca(struct mlx4_dev *dev)
@@ -2479,6 +2504,17 @@ static int __init mlx4_verify_params(void)
                 port_type_array[0] = true;
         }
 
+        if (mlx4_log_num_mgm_entry_size != -1 &&
+            (mlx4_log_num_mgm_entry_size < MLX4_MIN_MGM_LOG_ENTRY_SIZE ||
+             mlx4_log_num_mgm_entry_size > MLX4_MAX_MGM_LOG_ENTRY_SIZE)) {
+                pr_warning("mlx4_core: mlx4_log_num_mgm_entry_size (%d) not "
+                           "in legal range (-1 or %d..%d)\n",
+                           mlx4_log_num_mgm_entry_size,
+                           MLX4_MIN_MGM_LOG_ENTRY_SIZE,
+                           MLX4_MAX_MGM_LOG_ENTRY_SIZE);
+                return -1;
+        }
+
         return 0;
 }
 
diff --git a/drivers/net/ethernet/mellanox/mlx4/mcg.c b/drivers/net/ethernet/mellanox/mlx4/mcg.c
index e151c21baf2b..1ee4db3c6400 100644
--- a/drivers/net/ethernet/mellanox/mlx4/mcg.c
+++ b/drivers/net/ethernet/mellanox/mlx4/mcg.c
@@ -54,12 +54,7 @@ struct mlx4_mgm {
 
 int mlx4_get_mgm_entry_size(struct mlx4_dev *dev)
 {
-        if (dev->caps.steering_mode ==
-            MLX4_STEERING_MODE_DEVICE_MANAGED)
-                return 1 << MLX4_FS_MGM_LOG_ENTRY_SIZE;
-        else
-                return min((1 << mlx4_log_num_mgm_entry_size),
-                           MLX4_MAX_MGM_ENTRY_SIZE);
+        return 1 << dev->oper_log_mgm_entry_size;
 }
 
 int mlx4_get_qp_per_mgm(struct mlx4_dev *dev)
diff --git a/drivers/net/ethernet/mellanox/mlx4/mlx4.h b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
index 1cf42036d7bb..116c5c29d2d1 100644
--- a/drivers/net/ethernet/mellanox/mlx4/mlx4.h
+++ b/drivers/net/ethernet/mellanox/mlx4/mlx4.h
@@ -94,8 +94,10 @@ enum {
 };
 
 enum {
-        MLX4_MAX_MGM_ENTRY_SIZE = 0x1000,
-        MLX4_MAX_QP_PER_MGM     = 4 * (MLX4_MAX_MGM_ENTRY_SIZE / 16 - 2),
+        MLX4_DEFAULT_MGM_LOG_ENTRY_SIZE = 10,
+        MLX4_MIN_MGM_LOG_ENTRY_SIZE = 7,
+        MLX4_MAX_MGM_LOG_ENTRY_SIZE = 12,
+        MLX4_MAX_QP_PER_MGM = 4 * ((1 << MLX4_MAX_MGM_LOG_ENTRY_SIZE) / 16 - 2),
         MLX4_MTT_ENTRY_PER_SEG  = 8,
 };
 
diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
index 21821da2abfd..20ea939c22a6 100644
--- a/include/linux/mlx4/device.h
+++ b/include/linux/mlx4/device.h
@@ -625,6 +625,7 @@ struct mlx4_dev {
         u8                      rev_id;
         char                    board_id[MLX4_BOARD_ID_LEN];
         int                     num_vfs;
+        int                     oper_log_mgm_entry_size;
         u64                     regid_promisc_array[MLX4_MAX_PORTS + 1];
         u64                     regid_allmulti_array[MLX4_MAX_PORTS + 1];
 };