IB/core: Increase number of char device minors

There is a need to increase number of possible char devices to support
large number of SR-IOV instances. The current limit is in the range of
64-128 devices/ports. Increase it to support up to 1024.

The patch performs the following steps to refactor the code:
1. Removes the split bitmap for fixed and overflow dev numbers.
2. Pre-allocates the non-legacy major number range during driver
   initialization, choosen for simplicity.
3. Add new define (RDMA_MAX_PORTS) that is shared between all drivers.
   This is the maximum total number of ports on all struct ib_devices.
4. Set RDMA_MAX_PORTS to 1024.
Signed-off-by: Huy Nguyen <huyn@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>

drivers/infiniband/core/core_priv.h

@@ -42,6 +42,9 @@
 #include <rdma/ib_mad.h>
 #include "mad_priv.h"
 
+/* Total number of ports combined across all struct ib_devices's */
+#define RDMA_MAX_PORTS 1024
+
 struct pkey_index_qp_list {
 	struct list_head    pkey_index_list;
 	u16                 pkey_index;

drivers/infiniband/core/ucm.c

@@ -53,6 +53,8 @@
 #include <rdma/ib_user_cm.h>
 #include <rdma/ib_marshall.h>
 
+#include "core_priv.h"
+
 MODULE_AUTHOR("Libor Michalek");
 MODULE_DESCRIPTION("InfiniBand userspace Connection Manager access");
 MODULE_LICENSE("Dual BSD/GPL");
@@ -104,10 +106,13 @@ struct ib_ucm_event {
 enum {
 	IB_UCM_MAJOR = 231,
 	IB_UCM_BASE_MINOR = 224,
-	IB_UCM_MAX_DEVICES = 32
+	IB_UCM_MAX_DEVICES = RDMA_MAX_PORTS,
+	IB_UCM_NUM_FIXED_MINOR = 32,
+	IB_UCM_NUM_DYNAMIC_MINOR = IB_UCM_MAX_DEVICES - IB_UCM_NUM_FIXED_MINOR,
 };
 
 #define IB_UCM_BASE_DEV MKDEV(IB_UCM_MAJOR, IB_UCM_BASE_MINOR)
+static dev_t dynamic_ucm_dev;
 
 static void ib_ucm_add_one(struct ib_device *device);
 static void ib_ucm_remove_one(struct ib_device *device, void *client_data);
@@ -1199,7 +1204,6 @@ static int ib_ucm_close(struct inode *inode, struct file *filp)
 	return 0;
 }
 
-static DECLARE_BITMAP(overflow_map, IB_UCM_MAX_DEVICES);
 static void ib_ucm_release_dev(struct device *dev)
 {
 	struct ib_ucm_device *ucm_dev;
@@ -1210,10 +1214,7 @@ static void ib_ucm_release_dev(struct device *dev)
 
 static void ib_ucm_free_dev(struct ib_ucm_device *ucm_dev)
 {
-	if (ucm_dev->devnum < IB_UCM_MAX_DEVICES)
 	clear_bit(ucm_dev->devnum, dev_map);
-	else
-		clear_bit(ucm_dev->devnum - IB_UCM_MAX_DEVICES, overflow_map);
 }
 
 static const struct file_operations ucm_fops = {
@@ -1235,27 +1236,6 @@ static ssize_t show_ibdev(struct device *dev, struct device_attribute *attr,
 }
 static DEVICE_ATTR(ibdev, S_IRUGO, show_ibdev, NULL);
 
-static dev_t overflow_maj;
-static int find_overflow_devnum(void)
-{
-	int ret;
-
-	if (!overflow_maj) {
-		ret = alloc_chrdev_region(&overflow_maj, 0, IB_UCM_MAX_DEVICES,
-					  "infiniband_cm");
-		if (ret) {
-			pr_err("ucm: couldn't register dynamic device number\n");
-			return ret;
-		}
-	}
-
-	ret = find_first_zero_bit(overflow_map, IB_UCM_MAX_DEVICES);
-	if (ret >= IB_UCM_MAX_DEVICES)
-		return -1;
-
-	return ret;
-}
-
 static void ib_ucm_add_one(struct ib_device *device)
 {
 	int devnum;
@@ -1274,19 +1254,14 @@ static void ib_ucm_add_one(struct ib_device *device)
 	ucm_dev->dev.release = ib_ucm_release_dev;
 
 	devnum = find_first_zero_bit(dev_map, IB_UCM_MAX_DEVICES);
-	if (devnum >= IB_UCM_MAX_DEVICES) {
-		devnum = find_overflow_devnum();
-		if (devnum < 0)
+	if (devnum >= IB_UCM_MAX_DEVICES)
 		goto err;
-
-		ucm_dev->devnum = devnum + IB_UCM_MAX_DEVICES;
-		base = devnum + overflow_maj;
-		set_bit(devnum, overflow_map);
-	} else {
 	ucm_dev->devnum = devnum;
-		base = devnum + IB_UCM_BASE_DEV;
 	set_bit(devnum, dev_map);
-	}
+	if (devnum >= IB_UCM_NUM_FIXED_MINOR)
+		base = dynamic_ucm_dev + devnum - IB_UCM_NUM_FIXED_MINOR;
+	else
+		base = IB_UCM_BASE_DEV + devnum;
 
 	cdev_init(&ucm_dev->cdev, &ucm_fops);
 	ucm_dev->cdev.owner = THIS_MODULE;
@@ -1334,13 +1309,20 @@ static int __init ib_ucm_init(void)
 {
 	int ret;
 
-	ret = register_chrdev_region(IB_UCM_BASE_DEV, IB_UCM_MAX_DEVICES,
+	ret = register_chrdev_region(IB_UCM_BASE_DEV, IB_UCM_NUM_FIXED_MINOR,
 				     "infiniband_cm");
 	if (ret) {
 		pr_err("ucm: couldn't register device number\n");
 		goto error1;
 	}
 
+	ret = alloc_chrdev_region(&dynamic_ucm_dev, 0, IB_UCM_NUM_DYNAMIC_MINOR,
+				  "infiniband_cm");
+	if (ret) {
+		pr_err("ucm: couldn't register dynamic device number\n");
+		goto err_alloc;
+	}
+
 	ret = class_create_file(&cm_class, &class_attr_abi_version.attr);
 	if (ret) {
 		pr_err("ucm: couldn't create abi_version attribute\n");
@@ -1357,7 +1339,9 @@ static int __init ib_ucm_init(void)
 error3:
 	class_remove_file(&cm_class, &class_attr_abi_version.attr);
 error2:
-	unregister_chrdev_region(IB_UCM_BASE_DEV, IB_UCM_MAX_DEVICES);
+	unregister_chrdev_region(dynamic_ucm_dev, IB_UCM_NUM_DYNAMIC_MINOR);
+err_alloc:
+	unregister_chrdev_region(IB_UCM_BASE_DEV, IB_UCM_NUM_FIXED_MINOR);
 error1:
 	return ret;
 }
@@ -1366,9 +1350,8 @@ static void __exit ib_ucm_cleanup(void)
 {
 	ib_unregister_client(&ucm_client);
 	class_remove_file(&cm_class, &class_attr_abi_version.attr);
-	unregister_chrdev_region(IB_UCM_BASE_DEV, IB_UCM_MAX_DEVICES);
-	if (overflow_maj)
-		unregister_chrdev_region(overflow_maj, IB_UCM_MAX_DEVICES);
+	unregister_chrdev_region(IB_UCM_BASE_DEV, IB_UCM_NUM_FIXED_MINOR);
+	unregister_chrdev_region(dynamic_ucm_dev, IB_UCM_NUM_DYNAMIC_MINOR);
 	idr_destroy(&ctx_id_table);
 }
 

drivers/infiniband/core/user_mad.c

@@ -55,16 +55,21 @@
 #include <rdma/ib_mad.h>
 #include <rdma/ib_user_mad.h>
 
+#include "core_priv.h"
+
 MODULE_AUTHOR("Roland Dreier");
 MODULE_DESCRIPTION("InfiniBand userspace MAD packet access");
 MODULE_LICENSE("Dual BSD/GPL");
 
 enum {
-	IB_UMAD_MAX_PORTS  = 64,
+	IB_UMAD_MAX_PORTS  = RDMA_MAX_PORTS,
 	IB_UMAD_MAX_AGENTS = 32,
 
 	IB_UMAD_MAJOR      = 231,
-	IB_UMAD_MINOR_BASE = 0
+	IB_UMAD_MINOR_BASE = 0,
+	IB_UMAD_NUM_FIXED_MINOR = 64,
+	IB_UMAD_NUM_DYNAMIC_MINOR = IB_UMAD_MAX_PORTS - IB_UMAD_NUM_FIXED_MINOR,
+	IB_ISSM_MINOR_BASE        = IB_UMAD_NUM_FIXED_MINOR,
 };
 
 /*
@@ -127,7 +132,11 @@ struct ib_umad_packet {
 
 static struct class *umad_class;
 
-static const dev_t base_dev = MKDEV(IB_UMAD_MAJOR, IB_UMAD_MINOR_BASE);
+static const dev_t base_umad_dev = MKDEV(IB_UMAD_MAJOR, IB_UMAD_MINOR_BASE);
+static const dev_t base_issm_dev = MKDEV(IB_UMAD_MAJOR, IB_UMAD_MINOR_BASE) +
+				   IB_UMAD_NUM_FIXED_MINOR;
+static dev_t dynamic_umad_dev;
+static dev_t dynamic_issm_dev;
 
 static DECLARE_BITMAP(dev_map, IB_UMAD_MAX_PORTS);
 
@@ -1141,49 +1150,25 @@ static DEVICE_ATTR(port, S_IRUGO, show_port, NULL);
 static CLASS_ATTR_STRING(abi_version, S_IRUGO,
 			 __stringify(IB_USER_MAD_ABI_VERSION));
 
-static dev_t overflow_maj;
-static DECLARE_BITMAP(overflow_map, IB_UMAD_MAX_PORTS);
-static int find_overflow_devnum(struct ib_device *device)
-{
-	int ret;
-
-	if (!overflow_maj) {
-		ret = alloc_chrdev_region(&overflow_maj, 0, IB_UMAD_MAX_PORTS * 2,
-					  "infiniband_mad");
-		if (ret) {
-			dev_err(&device->dev,
-				"couldn't register dynamic device number\n");
-			return ret;
-		}
-	}
-
-	ret = find_first_zero_bit(overflow_map, IB_UMAD_MAX_PORTS);
-	if (ret >= IB_UMAD_MAX_PORTS)
-		return -1;
-
-	return ret;
-}
-
 static int ib_umad_init_port(struct ib_device *device, int port_num,
 			     struct ib_umad_device *umad_dev,
 			     struct ib_umad_port *port)
 {
 	int devnum;
-	dev_t base;
+	dev_t base_umad;
+	dev_t base_issm;
 
 	devnum = find_first_zero_bit(dev_map, IB_UMAD_MAX_PORTS);
-	if (devnum >= IB_UMAD_MAX_PORTS) {
-		devnum = find_overflow_devnum(device);
-		if (devnum < 0)
+	if (devnum >= IB_UMAD_MAX_PORTS)
 		return -1;
-
-		port->dev_num = devnum + IB_UMAD_MAX_PORTS;
-		base = devnum + overflow_maj;
-		set_bit(devnum, overflow_map);
-	} else {
 	port->dev_num = devnum;
-		base = devnum + base_dev;
 	set_bit(devnum, dev_map);
+	if (devnum >= IB_UMAD_NUM_FIXED_MINOR) {
+		base_umad = dynamic_umad_dev + devnum - IB_UMAD_NUM_FIXED_MINOR;
+		base_issm = dynamic_issm_dev + devnum - IB_UMAD_NUM_FIXED_MINOR;
+	} else {
+		base_umad = devnum + base_umad_dev;
+		base_issm = devnum + base_issm_dev;
 	}
 
 	port->ib_dev   = device;
@@ -1196,7 +1181,7 @@ static int ib_umad_init_port(struct ib_device *device, int port_num,
 	port->cdev.owner = THIS_MODULE;
 	cdev_set_parent(&port->cdev, &umad_dev->kobj);
 	kobject_set_name(&port->cdev.kobj, "umad%d", port->dev_num);
-	if (cdev_add(&port->cdev, base, 1))
+	if (cdev_add(&port->cdev, base_umad, 1))
 		goto err_cdev;
 
 	port->dev = device_create(umad_class, device->dev.parent,
@@ -1210,12 +1195,11 @@ static int ib_umad_init_port(struct ib_device *device, int port_num,
 	if (device_create_file(port->dev, &dev_attr_port))
 		goto err_dev;
 
-	base += IB_UMAD_MAX_PORTS;
 	cdev_init(&port->sm_cdev, &umad_sm_fops);
 	port->sm_cdev.owner = THIS_MODULE;
 	cdev_set_parent(&port->sm_cdev, &umad_dev->kobj);
 	kobject_set_name(&port->sm_cdev.kobj, "issm%d", port->dev_num);
-	if (cdev_add(&port->sm_cdev, base, 1))
+	if (cdev_add(&port->sm_cdev, base_issm, 1))
 		goto err_sm_cdev;
 
 	port->sm_dev = device_create(umad_class, device->dev.parent,
@@ -1242,10 +1226,7 @@ static int ib_umad_init_port(struct ib_device *device, int port_num,
 
 err_cdev:
 	cdev_del(&port->cdev);
-	if (port->dev_num < IB_UMAD_MAX_PORTS)
 	clear_bit(devnum, dev_map);
-	else
-		clear_bit(devnum, overflow_map);
 
 	return -1;
 }
@@ -1279,11 +1260,7 @@ static void ib_umad_kill_port(struct ib_umad_port *port)
 	}
 
 	mutex_unlock(&port->file_mutex);
-
-	if (port->dev_num < IB_UMAD_MAX_PORTS)
 	clear_bit(port->dev_num, dev_map);
-	else
-		clear_bit(port->dev_num - IB_UMAD_MAX_PORTS, overflow_map);
 }
 
 static void ib_umad_add_one(struct ib_device *device)
@@ -1359,13 +1336,23 @@ static int __init ib_umad_init(void)
 {
 	int ret;
 
-	ret = register_chrdev_region(base_dev, IB_UMAD_MAX_PORTS * 2,
+	ret = register_chrdev_region(base_umad_dev,
+				     IB_UMAD_NUM_FIXED_MINOR * 2,
 				     "infiniband_mad");
 	if (ret) {
 		pr_err("couldn't register device number\n");
 		goto out;
 	}
 
+	ret = alloc_chrdev_region(&dynamic_umad_dev, 0,
+				  IB_UMAD_NUM_DYNAMIC_MINOR * 2,
+				  "infiniband_mad");
+	if (ret) {
+		pr_err("couldn't register dynamic device number\n");
+		goto out_alloc;
+	}
+	dynamic_issm_dev = dynamic_umad_dev + IB_UMAD_NUM_DYNAMIC_MINOR;
+
 	umad_class = class_create(THIS_MODULE, "infiniband_mad");
 	if (IS_ERR(umad_class)) {
 		ret = PTR_ERR(umad_class);
@@ -1393,7 +1380,12 @@ static int __init ib_umad_init(void)
 	class_destroy(umad_class);
 
 out_chrdev:
-	unregister_chrdev_region(base_dev, IB_UMAD_MAX_PORTS * 2);
+	unregister_chrdev_region(dynamic_umad_dev,
+				 IB_UMAD_NUM_DYNAMIC_MINOR * 2);
+
+out_alloc:
+	unregister_chrdev_region(base_umad_dev,
+				 IB_UMAD_NUM_FIXED_MINOR * 2);
 
 out:
 	return ret;
@@ -1403,9 +1395,10 @@ static void __exit ib_umad_cleanup(void)
 {
 	ib_unregister_client(&umad_client);
 	class_destroy(umad_class);
-	unregister_chrdev_region(base_dev, IB_UMAD_MAX_PORTS * 2);
-	if (overflow_maj)
-		unregister_chrdev_region(overflow_maj, IB_UMAD_MAX_PORTS * 2);
+	unregister_chrdev_region(base_umad_dev,
+				 IB_UMAD_NUM_FIXED_MINOR * 2);
+	unregister_chrdev_region(dynamic_umad_dev,
+				 IB_UMAD_NUM_DYNAMIC_MINOR * 2);
 }
 
 module_init(ib_umad_init);

drivers/infiniband/core/uverbs_main.c

@@ -62,11 +62,14 @@ MODULE_LICENSE("Dual BSD/GPL");
 enum {
 	IB_UVERBS_MAJOR       = 231,
 	IB_UVERBS_BASE_MINOR  = 192,
-	IB_UVERBS_MAX_DEVICES = 32
+	IB_UVERBS_MAX_DEVICES = RDMA_MAX_PORTS,
+	IB_UVERBS_NUM_FIXED_MINOR = 32,
+	IB_UVERBS_NUM_DYNAMIC_MINOR = IB_UVERBS_MAX_DEVICES - IB_UVERBS_NUM_FIXED_MINOR,
 };
 
 #define IB_UVERBS_BASE_DEV	MKDEV(IB_UVERBS_MAJOR, IB_UVERBS_BASE_MINOR)
 
+static dev_t dynamic_uverbs_dev;
 static struct class *uverbs_class;
 
 static DECLARE_BITMAP(dev_map, IB_UVERBS_MAX_DEVICES);
@@ -1004,34 +1007,6 @@ static DEVICE_ATTR(abi_version, S_IRUGO, show_dev_abi_version, NULL);
 static CLASS_ATTR_STRING(abi_version, S_IRUGO,
 			 __stringify(IB_USER_VERBS_ABI_VERSION));
 
-static dev_t overflow_maj;
-static DECLARE_BITMAP(overflow_map, IB_UVERBS_MAX_DEVICES);
-
-/*
- * If we have more than IB_UVERBS_MAX_DEVICES, dynamically overflow by
- * requesting a new major number and doubling the number of max devices we
- * support. It's stupid, but simple.
- */
-static int find_overflow_devnum(void)
-{
-	int ret;
-
-	if (!overflow_maj) {
-		ret = alloc_chrdev_region(&overflow_maj, 0, IB_UVERBS_MAX_DEVICES,
-					  "infiniband_verbs");
-		if (ret) {
-			pr_err("user_verbs: couldn't register dynamic device number\n");
-			return ret;
-		}
-	}
-
-	ret = find_first_zero_bit(overflow_map, IB_UVERBS_MAX_DEVICES);
-	if (ret >= IB_UVERBS_MAX_DEVICES)
-		return -1;
-
-	return ret;
-}
-
 static void ib_uverbs_add_one(struct ib_device *device)
 {
 	int devnum;
@@ -1062,19 +1037,14 @@ static void ib_uverbs_add_one(struct ib_device *device)
 	INIT_LIST_HEAD(&uverbs_dev->uverbs_events_file_list);
 
 	devnum = find_first_zero_bit(dev_map, IB_UVERBS_MAX_DEVICES);
-	if (devnum >= IB_UVERBS_MAX_DEVICES) {
-		devnum = find_overflow_devnum();
-		if (devnum < 0)
+	if (devnum >= IB_UVERBS_MAX_DEVICES)
 		goto err;
-
-		uverbs_dev->devnum = devnum + IB_UVERBS_MAX_DEVICES;
-		base = devnum + overflow_maj;
-		set_bit(devnum, overflow_map);
-	} else {
 	uverbs_dev->devnum = devnum;
-		base = devnum + IB_UVERBS_BASE_DEV;
 	set_bit(devnum, dev_map);
-	}
+	if (devnum >= IB_UVERBS_NUM_FIXED_MINOR)
+		base = dynamic_uverbs_dev + devnum - IB_UVERBS_NUM_FIXED_MINOR;
+	else
+		base = IB_UVERBS_BASE_DEV + devnum;
 
 	rcu_assign_pointer(uverbs_dev->ib_dev, device);
 	uverbs_dev->num_comp_vectors = device->num_comp_vectors;
@@ -1119,10 +1089,7 @@ static void ib_uverbs_add_one(struct ib_device *device)
 
 err_cdev:
 	cdev_del(&uverbs_dev->cdev);
-	if (uverbs_dev->devnum < IB_UVERBS_MAX_DEVICES)
 	clear_bit(devnum, dev_map);
-	else
-		clear_bit(devnum, overflow_map);
 
 err:
 	if (atomic_dec_and_test(&uverbs_dev->refcount))
@@ -1214,11 +1181,7 @@ static void ib_uverbs_remove_one(struct ib_device *device, void *client_data)
 	dev_set_drvdata(uverbs_dev->dev, NULL);
 	device_destroy(uverbs_class, uverbs_dev->cdev.dev);
 	cdev_del(&uverbs_dev->cdev);
-
-	if (uverbs_dev->devnum < IB_UVERBS_MAX_DEVICES)
 	clear_bit(uverbs_dev->devnum, dev_map);
-	else
-		clear_bit(uverbs_dev->devnum - IB_UVERBS_MAX_DEVICES, overflow_map);
 
 	if (device->disassociate_ucontext) {
 		/* We disassociate HW resources and immediately return.
@@ -1260,13 +1223,22 @@ static int __init ib_uverbs_init(void)
 {
 	int ret;
 
-	ret = register_chrdev_region(IB_UVERBS_BASE_DEV, IB_UVERBS_MAX_DEVICES,
+	ret = register_chrdev_region(IB_UVERBS_BASE_DEV,
+				     IB_UVERBS_NUM_FIXED_MINOR,
 				     "infiniband_verbs");
 	if (ret) {
 		pr_err("user_verbs: couldn't register device number\n");
 		goto out;
 	}
 
+	ret = alloc_chrdev_region(&dynamic_uverbs_dev, 0,
+				  IB_UVERBS_NUM_DYNAMIC_MINOR,
+				  "infiniband_verbs");
+	if (ret) {
+		pr_err("couldn't register dynamic device number\n");
+		goto out_alloc;
+	}
+
 	uverbs_class = class_create(THIS_MODULE, "infiniband_verbs");
 	if (IS_ERR(uverbs_class)) {
 		ret = PTR_ERR(uverbs_class);
@@ -1294,7 +1266,12 @@ static int __init ib_uverbs_init(void)
 	class_destroy(uverbs_class);
 
 out_chrdev:
-	unregister_chrdev_region(IB_UVERBS_BASE_DEV, IB_UVERBS_MAX_DEVICES);
+	unregister_chrdev_region(dynamic_uverbs_dev,
+				 IB_UVERBS_NUM_DYNAMIC_MINOR);
+
+out_alloc:
+	unregister_chrdev_region(IB_UVERBS_BASE_DEV,
+				 IB_UVERBS_NUM_FIXED_MINOR);
 
 out:
 	return ret;
@@ -1304,9 +1281,10 @@ static void __exit ib_uverbs_cleanup(void)
 {
 	ib_unregister_client(&uverbs_client);
 	class_destroy(uverbs_class);
-	unregister_chrdev_region(IB_UVERBS_BASE_DEV, IB_UVERBS_MAX_DEVICES);
-	if (overflow_maj)
-		unregister_chrdev_region(overflow_maj, IB_UVERBS_MAX_DEVICES);
+	unregister_chrdev_region(IB_UVERBS_BASE_DEV,
+				 IB_UVERBS_NUM_FIXED_MINOR);
+	unregister_chrdev_region(dynamic_uverbs_dev,
+				 IB_UVERBS_NUM_DYNAMIC_MINOR);
 }
 
 module_init(ib_uverbs_init);