net/mlx5_core: Add base sriov support

This patch adds SRIOV base support for mlx5 supported devices. The same
driver is used for both PFs and VFs; VFs are identified by the driver
through the flag MLX5_PCI_DEV_IS_VF added to the pci table entries.
Virtual functions are created as usual through writing a value to the
sriov_numvs sysfs file of the PF device. Upon instantiating VFs, they will
all be probed by the driver on the hypervisor. One can gracefully unbind
them through /sys/bus/pci/drivers/mlx5_core/unbind.

mlx5_wait_for_vf_pages() was added to ensure that when a VF dies without
executing proper teardown, the hypervisor driver waits till all of the
pages that were allocated at the hypervisor to maintain its operation
are returned.

In order for the VF to be operational, the PF needs to call enable_hca
for it. This can be done before the VFs are created through a call to
pci_enable_sriov.

If the there are VFs assigned to a VMs when the driver of the PF is
unloaded, all the VF will experience system error and PF driver unloads
cleanly; in this case pci_disable_sriov is not called and the devices
will show when running lspci. Once the PF driver is reloaded, it will
sync its data structures which maintain state on its VFs.

Signed-off-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Author: Eli Cohen

drivers/net/ethernet/mellanox/mlx5/core/Makefile

@@ -2,7 +2,7 @@ obj-$(CONFIG_MLX5_CORE)		+= mlx5_core.o
 
 mlx5_core-y :=	main.o cmd.o debugfs.o fw.o eq.o uar.o pagealloc.o \
 		health.o mcg.o cq.o srq.o alloc.o qp.o port.o mr.o pd.o   \
-		mad.o transobj.o vport.o
+		mad.o transobj.o vport.o sriov.o
 mlx5_core-$(CONFIG_MLX5_CORE_EN) += wq.o flow_table.o \
 		en_main.o en_flow_table.o en_ethtool.o en_tx.o en_rx.o \
 		en_txrx.o

drivers/net/ethernet/mellanox/mlx5/core/main.c

@@ -454,6 +454,9 @@ static int set_hca_ctrl(struct mlx5_core_dev *dev)
 	struct mlx5_reg_host_endianess he_out;
 	int err;
 
+	if (!mlx5_core_is_pf(dev))
+		return 0;
+
 	memset(&he_in, 0, sizeof(he_in));
 	he_in.he = MLX5_SET_HOST_ENDIANNESS;
 	err = mlx5_core_access_reg(dev, &he_in,  sizeof(he_in),
@@ -1049,6 +1052,12 @@ static int mlx5_load_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv)
 	mlx5_init_srq_table(dev);
 	mlx5_init_mr_table(dev);
 
+	err = mlx5_sriov_init(dev);
+	if (err) {
+		dev_err(&pdev->dev, "sriov init failed %d\n", err);
+		goto err_sriov;
+	}
+
 	err = mlx5_register_device(dev);
 	if (err) {
 		dev_err(&pdev->dev, "mlx5_register_device failed %d\n", err);
@@ -1065,6 +1074,10 @@ static int mlx5_load_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv)
 
 	return 0;
 
+err_sriov:
+	if (mlx5_sriov_cleanup(dev))
+		dev_err(&dev->pdev->dev, "sriov cleanup failed\n");
+
 err_reg_dev:
 	mlx5_cleanup_mr_table(dev);
 	mlx5_cleanup_srq_table(dev);
@@ -1120,6 +1133,13 @@ static int mlx5_unload_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv)
 {
 	int err = 0;
 
+	err = mlx5_sriov_cleanup(dev);
+	if (err) {
+		dev_warn(&dev->pdev->dev, "%s: sriov cleanup failed - abort\n",
+			 __func__);
+		return err;
+	}
+
 	mutex_lock(&dev->intf_state_mutex);
 	if (dev->interface_state == MLX5_INTERFACE_STATE_DOWN) {
 		dev_warn(&dev->pdev->dev, "%s: interface is down, NOP\n",
@@ -1192,6 +1212,7 @@ static int init_one(struct pci_dev *pdev,
 		return -ENOMEM;
 	}
 	priv = &dev->priv;
+	priv->pci_dev_data = id->driver_data;
 
 	pci_set_drvdata(pdev, dev);
 
@@ -1362,12 +1383,12 @@ static const struct pci_error_handlers mlx5_err_handler = {
 };
 
 static const struct pci_device_id mlx5_core_pci_table[] = {
-	{ PCI_VDEVICE(MELLANOX, 0x1011) }, /* Connect-IB */
-	{ PCI_VDEVICE(MELLANOX, 0x1012) }, /* Connect-IB VF */
-	{ PCI_VDEVICE(MELLANOX, 0x1013) }, /* ConnectX-4 */
-	{ PCI_VDEVICE(MELLANOX, 0x1014) }, /* ConnectX-4 VF */
-	{ PCI_VDEVICE(MELLANOX, 0x1015) }, /* ConnectX-4LX */
-	{ PCI_VDEVICE(MELLANOX, 0x1016) }, /* ConnectX-4LX VF */
+	{ PCI_VDEVICE(MELLANOX, 0x1011) },			/* Connect-IB */
+	{ PCI_VDEVICE(MELLANOX, 0x1012), MLX5_PCI_DEV_IS_VF},	/* Connect-IB VF */
+	{ PCI_VDEVICE(MELLANOX, 0x1013) },			/* ConnectX-4 */
+	{ PCI_VDEVICE(MELLANOX, 0x1014), MLX5_PCI_DEV_IS_VF},	/* ConnectX-4 VF */
+	{ PCI_VDEVICE(MELLANOX, 0x1015) },			/* ConnectX-4LX */
+	{ PCI_VDEVICE(MELLANOX, 0x1016), MLX5_PCI_DEV_IS_VF},	/* ConnectX-4LX VF */
 	{ 0, }
 };
 
@@ -1378,7 +1399,8 @@ static struct pci_driver mlx5_core_driver = {
 	.id_table       = mlx5_core_pci_table,
 	.probe          = init_one,
 	.remove         = remove_one,
-	.err_handler	= &mlx5_err_handler
+	.err_handler	= &mlx5_err_handler,
+	.sriov_configure   = mlx5_core_sriov_configure,
 };
 
 static int __init init(void)

drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h

@@ -90,8 +90,10 @@ void mlx5_core_event(struct mlx5_core_dev *dev, enum mlx5_dev_event event,
 		     unsigned long param);
 void mlx5_enter_error_state(struct mlx5_core_dev *dev);
 void mlx5_disable_device(struct mlx5_core_dev *dev);
+int mlx5_core_sriov_configure(struct pci_dev *dev, int num_vfs);
 int mlx5_core_enable_hca(struct mlx5_core_dev *dev, u16 func_id);
 int mlx5_core_disable_hca(struct mlx5_core_dev *dev, u16 func_id);
+int mlx5_wait_for_vf_pages(struct mlx5_core_dev *dev);
 
 void mlx5e_init(void);
 void mlx5e_cleanup(void);

drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c

@@ -33,6 +33,7 @@
 #include <linux/highmem.h>
 #include <linux/kernel.h>
 #include <linux/module.h>
+#include <linux/delay.h>
 #include <linux/mlx5/driver.h>
 #include <linux/mlx5/cmd.h>
 #include "mlx5_core.h"
@@ -95,6 +96,7 @@ struct mlx5_manage_pages_outbox {
 
 enum {
 	MAX_RECLAIM_TIME_MSECS	= 5000,
+	MAX_RECLAIM_VFS_PAGES_TIME_MSECS = 2 * 1000 * 60,
 };
 
 enum {
@@ -352,6 +354,10 @@ static int give_pages(struct mlx5_core_dev *dev, u16 func_id, int npages,
 		goto out_4k;
 	}
 
+	dev->priv.fw_pages += npages;
+	if (func_id)
+		dev->priv.vfs_pages += npages;
+
 	mlx5_core_dbg(dev, "err %d\n", err);
 
 	kvfree(in);
@@ -405,13 +411,22 @@ static int reclaim_pages(struct mlx5_core_dev *dev, u32 func_id, int npages,
 	}
 
 	num_claimed = be32_to_cpu(out->num_entries);
+	if (num_claimed > npages) {
+		mlx5_core_warn(dev, "fw returned %d, driver asked %d => corruption\n",
+			       num_claimed, npages);
+		err = -EINVAL;
+		goto out_free;
+	}
 	if (nclaimed)
 		*nclaimed = num_claimed;
 
 	for (i = 0; i < num_claimed; i++) {
 		addr = be64_to_cpu(out->pas[i]);
 		free_4k(dev, addr);
 	}
+	dev->priv.fw_pages -= num_claimed;
+	if (func_id)
+		dev->priv.vfs_pages -= num_claimed;
 
 out_free:
 	kvfree(out);
@@ -548,3 +563,26 @@ void mlx5_pagealloc_stop(struct mlx5_core_dev *dev)
 {
 	destroy_workqueue(dev->priv.pg_wq);
 }
+
+int mlx5_wait_for_vf_pages(struct mlx5_core_dev *dev)
+{
+	unsigned long end = jiffies + msecs_to_jiffies(MAX_RECLAIM_VFS_PAGES_TIME_MSECS);
+	int prev_vfs_pages = dev->priv.vfs_pages;
+
+	mlx5_core_dbg(dev, "Waiting for %d pages from %s\n", prev_vfs_pages,
+		      dev->priv.name);
+	while (dev->priv.vfs_pages) {
+		if (time_after(jiffies, end)) {
+			mlx5_core_warn(dev, "aborting while there are %d pending pages\n", dev->priv.vfs_pages);
+			return -ETIMEDOUT;
+		}
+		if (dev->priv.vfs_pages < prev_vfs_pages) {
+			end = jiffies + msecs_to_jiffies(MAX_RECLAIM_VFS_PAGES_TIME_MSECS);
+			prev_vfs_pages = dev->priv.vfs_pages;
+		}
+		msleep(50);
+	}
+
+	mlx5_core_dbg(dev, "All pages received from %s\n", dev->priv.name);
+	return 0;
+}