-/* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
- * vim:expandtab:shiftwidth=8:tabstop=8:
- *
- * Copyright (C) 2001, 2002 Cluster File Systems, Inc.
- *
- * This file is part of Lustre, http://www.lustre.org.
- *
- * Lustre is free software; you can redistribute it and/or
- * modify it under the terms of version 2 of the GNU General Public
- * License as published by the Free Software Foundation.
- *
- * Lustre is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- * GNU General Public License for more details.
+// SPDX-License-Identifier: GPL-2.0
+
+/* Copyright (c) 2004, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Use is subject to license terms.
*
- * You should have received a copy of the GNU General Public License
- * along with Lustre; if not, write to the Free Software
- * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ * Copyright (c) 2012, 2017, Intel Corporation.
*/
-#ifndef EXPORT_SYMTAB
-# define EXPORT_SYMTAB
-#endif
-#define DEBUG_SUBSYSTEM S_PORTALS
-
-#include <linux/config.h>
-#include <linux/module.h>
-#include <linux/kernel.h>
-#include <linux/mm.h>
-#include <linux/string.h>
-#include <linux/stat.h>
-#include <linux/init.h>
-#include <linux/errno.h>
-#include <linux/smp_lock.h>
-#include <linux/unistd.h>
-
-#include <asm/system.h>
-#include <asm/uaccess.h>
-
-#include <linux/fs.h>
-#include <linux/stat.h>
-#include <asm/uaccess.h>
-#include <asm/segment.h>
+/* This file is part of Lustre, http://www.lustre.org/ */
+
+#define DEBUG_SUBSYSTEM S_LNET
+
#include <linux/miscdevice.h>
+#include <lnet/lib-lnet.h>
+#include <uapi/linux/lnet/lnet-dlc.h>
+#include <uapi/linux/lustre/lustre_ver.h>
+
+static int config_on_load = 0;
+module_param(config_on_load, int, 0444);
+MODULE_PARM_DESC(config_on_load, "configure network at module load");
+
+static DEFINE_MUTEX(lnet_config_mutex);
+
+int lnet_configure(void *arg)
+{
+ /* 'arg' only there so I can be passed to cfs_create_thread() */
+ int rc = 0;
+
+ mutex_lock(&lnet_config_mutex);
+
+ if (!the_lnet.ln_niinit_self) {
+ rc = try_module_get(THIS_MODULE);
+
+ if (rc != 1)
+ goto out;
+
+ rc = LNetNIInit(LNET_PID_LUSTRE);
+ if (rc >= 0) {
+ the_lnet.ln_niinit_self = 1;
+ rc = 0;
+ } else {
+ module_put(THIS_MODULE);
+ }
+ }
+
+out:
+ mutex_unlock(&lnet_config_mutex);
+ return rc;
+}
+
+int lnet_unconfigure(void)
+{
+ int refcount;
+
+ mutex_lock(&lnet_config_mutex);
+
+ if (the_lnet.ln_niinit_self) {
+ the_lnet.ln_niinit_self = 0;
+ LNetNIFini();
+ module_put(THIS_MODULE);
+ }
+
+ mutex_lock(&the_lnet.ln_api_mutex);
+ refcount = the_lnet.ln_refcount;
+ mutex_unlock(&the_lnet.ln_api_mutex);
+
+ mutex_unlock(&lnet_config_mutex);
+
+ return (refcount == 0) ? 0 : -EBUSY;
+}
+
+static int
+lnet_dyn_configure_net(struct libcfs_ioctl_hdr *hdr)
+{
+ struct lnet_ioctl_config_data *conf =
+ (struct lnet_ioctl_config_data *)hdr;
+ int rc;
-#include <portals/lib-p30.h>
-#include <portals/p30.h>
-#include <portals/nal.h>
-#include <linux/kp30.h>
-#include <linux/kpr.h>
-#include <linux/portals_compat25.h>
+ if (conf->cfg_hdr.ioc_len < sizeof(*conf))
+ return -EINVAL;
-extern void (kping_client)(struct portal_ioctl_data *);
+ mutex_lock(&lnet_config_mutex);
+ if (the_lnet.ln_niinit_self)
+ rc = lnet_dyn_add_net(conf);
+ else
+ rc = -EINVAL;
+ mutex_unlock(&lnet_config_mutex);
-static int kportal_ioctl(struct portal_ioctl_data *data,
- unsigned int cmd, unsigned long arg)
+ return rc;
+}
+
+static int
+lnet_dyn_unconfigure_net(struct libcfs_ioctl_hdr *hdr)
{
- int err;
- char str[PTL_NALFMT_SIZE];
- ENTRY;
-
- switch (cmd) {
- case IOC_PORTAL_PING: {
- void (*ping)(struct portal_ioctl_data *);
-
- CDEBUG(D_IOCTL, "doing %d pings to nid "LPX64" (%s)\n",
- data->ioc_count, data->ioc_nid,
- portals_nid2str(data->ioc_nal, data->ioc_nid, str));
- ping = PORTAL_SYMBOL_GET(kping_client);
- if (!ping)
- CERROR("PORTAL_SYMBOL_GET failed\n");
- else {
- ping(data);
- PORTAL_SYMBOL_PUT(kping_client);
- }
- RETURN(0);
- }
-
- case IOC_PORTAL_GET_NID: {
- ptl_handle_ni_t nih;
- ptl_process_id_t pid;
-
- CDEBUG (D_IOCTL, "Getting nid for nal [%d]\n", data->ioc_nal);
-
- err = PtlNIInit(data->ioc_nal, 0, NULL, NULL, &nih);
- if (!(err == PTL_OK || err == PTL_IFACE_DUP))
- RETURN (-EINVAL);
-
- err = PtlGetId (nih, &pid);
- LASSERT (err == PTL_OK);
-
- PtlNIFini(nih);
-
- data->ioc_nid = pid.nid;
- if (copy_to_user ((char *)arg, data, sizeof (*data)))
- RETURN (-EFAULT);
- RETURN(0);
- }
-
- case IOC_PORTAL_FAIL_NID: {
- ptl_handle_ni_t nih;
-
- CDEBUG (D_IOCTL, "fail nid: [%d] "LPU64" count %d\n",
- data->ioc_nal, data->ioc_nid, data->ioc_count);
-
- err = PtlNIInit(data->ioc_nal, 0, NULL, NULL, &nih);
- if (!(err == PTL_OK || err == PTL_IFACE_DUP))
- return (-EINVAL);
-
- if (err == PTL_OK) {
- /* There's no point in failing an interface that
- * came into existance just for this */
- err = -EINVAL;
- } else {
- err = PtlFailNid (nih, data->ioc_nid, data->ioc_count);
- if (err != PTL_OK)
- err = -EINVAL;
- }
-
- PtlNIFini(nih);
- RETURN (err);
- }
- default:
- RETURN(-EINVAL);
- }
- /* Not Reached */
+ struct lnet_ioctl_config_data *conf =
+ (struct lnet_ioctl_config_data *) hdr;
+ int rc;
+
+ if (conf->cfg_hdr.ioc_len < sizeof(*conf))
+ return -EINVAL;
+
+ mutex_lock(&lnet_config_mutex);
+ if (the_lnet.ln_niinit_self)
+ rc = lnet_dyn_del_net(conf->cfg_net);
+ else
+ rc = -EINVAL;
+ mutex_unlock(&lnet_config_mutex);
+
+ return rc;
}
-DECLARE_IOCTL_HANDLER(kportal_ioctl_handler, kportal_ioctl);
+static int
+lnet_dyn_configure_ni(struct libcfs_ioctl_hdr *hdr)
+{
+ struct lnet_ioctl_config_ni *conf =
+ (struct lnet_ioctl_config_ni *)hdr;
+ int rc = -EINVAL;
+
+ if (conf->lic_cfg_hdr.ioc_len < sizeof(*conf))
+ return rc;
-static int init_kportals_module(void)
+ mutex_lock(&lnet_config_mutex);
+ if (the_lnet.ln_niinit_self) {
+ struct lnet_ioctl_config_lnd_tunables *tun = NULL;
+ struct lnet_nid nid;
+ u32 net_id;
+
+ /* get the tunables if they are available */
+ if (conf->lic_cfg_hdr.ioc_len >=
+ sizeof(*conf) + sizeof(*tun))
+ tun = (struct lnet_ioctl_config_lnd_tunables *) conf->lic_bulk;
+
+ lnet_nid4_to_nid(conf->lic_nid, &nid);
+ net_id = LNET_NID_NET(&nid);
+ rc = lnet_dyn_add_ni(conf, net_id, &LNET_ANY_NID, tun);
+ }
+ mutex_unlock(&lnet_config_mutex);
+
+ return rc;
+}
+
+static int
+lnet_dyn_unconfigure_ni(struct libcfs_ioctl_hdr *hdr)
{
- int rc;
- ENTRY;
+ struct lnet_ioctl_config_ni *conf =
+ (struct lnet_ioctl_config_ni *) hdr;
+ struct lnet_nid nid;
+ int rc = -EINVAL;
- rc = PtlInit(NULL);
- if (rc) {
- CERROR("PtlInit: error %d\n", rc);
- RETURN(rc);
- }
+ if (conf->lic_cfg_hdr.ioc_len < sizeof(*conf) ||
+ !the_lnet.ln_niinit_self)
+ return rc;
- rc = libcfs_register_ioctl(&kportal_ioctl_handler);
- LASSERT (rc == 0);
+ lnet_nid4_to_nid(conf->lic_nid, &nid);
+ mutex_lock(&lnet_config_mutex);
+ if (the_lnet.ln_niinit_self)
+ rc = lnet_dyn_del_ni(&nid);
+ else
+ rc = -EINVAL;
+ mutex_unlock(&lnet_config_mutex);
- RETURN(rc);
+ return rc;
}
-static void exit_kportals_module(void)
+static int
+lnet_ioctl(unsigned int cmd, struct libcfs_ioctl_hdr *hdr)
{
- int rc;
+ int rc;
+
+ switch (cmd) {
+ case IOC_LIBCFS_CONFIGURE: {
+ struct libcfs_ioctl_data *data =
+ (struct libcfs_ioctl_data *)hdr;
+
+ if (data->ioc_hdr.ioc_len < sizeof(*data)) {
+ rc = -EINVAL;
+ } else {
+ the_lnet.ln_nis_from_mod_params = data->ioc_flags;
+ rc = lnet_configure(NULL);
+ }
+ break;
+ }
- rc = libcfs_deregister_ioctl(&kportal_ioctl_handler);
- LASSERT (rc == 0);
+ case IOC_LIBCFS_UNCONFIGURE:
+ rc = lnet_unconfigure();
+ break;
- PtlFini();
+ case IOC_LIBCFS_ADD_NET:
+ rc = lnet_dyn_configure_net(hdr);
+ break;
+
+ case IOC_LIBCFS_DEL_NET:
+ rc = lnet_dyn_unconfigure_net(hdr);
+ break;
+
+ case IOC_LIBCFS_ADD_LOCAL_NI:
+ rc = lnet_dyn_configure_ni(hdr);
+ break;
+
+ case IOC_LIBCFS_DEL_LOCAL_NI:
+ rc = lnet_dyn_unconfigure_ni(hdr);
+ break;
+
+ default:
+ /* Passing LNET_PID_ANY only gives me a ref if the net is up
+ * already; I'll need it to ensure the net can't go down while
+ * I'm called into it */
+ rc = LNetNIInit(LNET_PID_ANY);
+ if (rc >= 0) {
+ rc = LNetCtl(cmd, hdr);
+ LNetNIFini();
+ }
+ break;
+ }
+ return rc;
+}
+BLOCKING_NOTIFIER_HEAD(lnet_ioctl_list);
+EXPORT_SYMBOL(lnet_ioctl_list);
+
+static inline size_t lnet_ioctl_packlen(struct libcfs_ioctl_data *data)
+{
+ size_t len = sizeof(*data);
+
+ len += (data->ioc_inllen1 + 7) & ~7;
+ len += (data->ioc_inllen2 + 7) & ~7;
+ return len;
}
-EXPORT_SYMBOL(ptl_register_nal);
-EXPORT_SYMBOL(ptl_unregister_nal);
-
-EXPORT_SYMBOL(ptl_err_str);
-EXPORT_SYMBOL(lib_dispatch);
-EXPORT_SYMBOL(PtlMEAttach);
-EXPORT_SYMBOL(PtlMEInsert);
-EXPORT_SYMBOL(PtlMEUnlink);
-EXPORT_SYMBOL(PtlEQAlloc);
-EXPORT_SYMBOL(PtlMDAttach);
-EXPORT_SYMBOL(PtlMDUnlink);
-EXPORT_SYMBOL(PtlNIInit);
-EXPORT_SYMBOL(PtlNIFini);
-EXPORT_SYMBOL(PtlInit);
-EXPORT_SYMBOL(PtlFini);
-EXPORT_SYMBOL(PtlSnprintHandle);
-EXPORT_SYMBOL(PtlPut);
-EXPORT_SYMBOL(PtlGet);
-EXPORT_SYMBOL(PtlEQWait);
-EXPORT_SYMBOL(PtlEQFree);
-EXPORT_SYMBOL(PtlEQGet);
-EXPORT_SYMBOL(PtlGetId);
-EXPORT_SYMBOL(PtlMDBind);
-EXPORT_SYMBOL(lib_iov_nob);
-EXPORT_SYMBOL(lib_copy_iov2buf);
-EXPORT_SYMBOL(lib_copy_buf2iov);
-EXPORT_SYMBOL(lib_extract_iov);
-EXPORT_SYMBOL(lib_kiov_nob);
-EXPORT_SYMBOL(lib_copy_kiov2buf);
-EXPORT_SYMBOL(lib_copy_buf2kiov);
-EXPORT_SYMBOL(lib_extract_kiov);
-EXPORT_SYMBOL(lib_finalize);
-EXPORT_SYMBOL(lib_parse);
-EXPORT_SYMBOL(lib_create_reply_msg);
-EXPORT_SYMBOL(lib_init);
-EXPORT_SYMBOL(lib_fini);
-EXPORT_SYMBOL(dispatch_name);
-
-MODULE_AUTHOR("Peter J. Braam <braam@clusterfs.com>");
-MODULE_DESCRIPTION("Portals v3.1");
+static bool lnet_ioctl_is_invalid(struct libcfs_ioctl_data *data)
+{
+ const int maxlen = 1 << 30;
+
+ if (data->ioc_hdr.ioc_len > maxlen)
+ return true;
+
+ if (data->ioc_inllen1 > maxlen)
+ return true;
+
+ if (data->ioc_inllen2 > maxlen)
+ return true;
+
+ if (data->ioc_inlbuf1 && !data->ioc_inllen1)
+ return true;
+
+ if (data->ioc_inlbuf2 && !data->ioc_inllen2)
+ return true;
+
+ if (data->ioc_pbuf1 && !data->ioc_plen1)
+ return true;
+
+ if (data->ioc_pbuf2 && !data->ioc_plen2)
+ return true;
+
+ if (data->ioc_plen1 && !data->ioc_pbuf1)
+ return true;
+
+ if (data->ioc_plen2 && !data->ioc_pbuf2)
+ return true;
+
+ if (lnet_ioctl_packlen(data) != data->ioc_hdr.ioc_len)
+ return true;
+
+ if (data->ioc_inllen1 &&
+ data->ioc_bulk[((data->ioc_inllen1 + 7) & ~7) +
+ data->ioc_inllen2 - 1] != '\0')
+ return true;
+
+ return false;
+}
+
+static int lnet_ioctl_data_adjust(struct libcfs_ioctl_data *data)
+{
+ ENTRY;
+
+ if (lnet_ioctl_is_invalid(data)) {
+ CERROR("lnet ioctl: parameter not correctly formatted\n");
+ RETURN(-EINVAL);
+ }
+
+ if (data->ioc_inllen1 != 0)
+ data->ioc_inlbuf1 = &data->ioc_bulk[0];
+
+ if (data->ioc_inllen2 != 0)
+ data->ioc_inlbuf2 = (&data->ioc_bulk[0] +
+ round_up(data->ioc_inllen1, 8));
+
+ RETURN(0);
+}
+
+static int lnet_ioctl_getdata(struct libcfs_ioctl_hdr **hdr_pp,
+ struct libcfs_ioctl_hdr __user *uhdr)
+{
+ struct libcfs_ioctl_hdr hdr;
+ int err;
+
+ ENTRY;
+ if (copy_from_user(&hdr, uhdr, sizeof(hdr)))
+ RETURN(-EFAULT);
+
+ if (hdr.ioc_version != LNET_IOCTL_VERSION &&
+ hdr.ioc_version != LNET_IOCTL_VERSION2) {
+ CERROR("lnet ioctl: version mismatch expected %#x, got %#x\n",
+ LNET_IOCTL_VERSION, hdr.ioc_version);
+ RETURN(-EINVAL);
+ }
+
+ if (hdr.ioc_len < sizeof(struct libcfs_ioctl_hdr)) {
+ CERROR("lnet ioctl: user buffer too small for ioctl\n");
+ RETURN(-EINVAL);
+ }
+
+ if (hdr.ioc_len > LIBCFS_IOC_DATA_MAX) {
+ CERROR("lnet ioctl: user buffer is too large %d/%d\n",
+ hdr.ioc_len, LIBCFS_IOC_DATA_MAX);
+ RETURN(-EINVAL);
+ }
+
+ LIBCFS_ALLOC(*hdr_pp, hdr.ioc_len);
+ if (*hdr_pp == NULL)
+ RETURN(-ENOMEM);
+
+ if (copy_from_user(*hdr_pp, uhdr, hdr.ioc_len))
+ GOTO(free, err = -EFAULT);
+
+ if ((*hdr_pp)->ioc_version != hdr.ioc_version ||
+ (*hdr_pp)->ioc_len != hdr.ioc_len) {
+ GOTO(free, err = -EINVAL);
+ }
+
+ RETURN(0);
+
+free:
+ LIBCFS_FREE(*hdr_pp, hdr.ioc_len);
+ RETURN(err);
+}
+
+static long
+lnet_psdev_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
+{
+ struct libcfs_ioctl_data *data = NULL;
+ struct libcfs_ioctl_hdr *hdr;
+ int err;
+ void __user *uparam = (void __user *)arg;
+
+ ENTRY;
+ if (!capable(CAP_SYS_ADMIN))
+ return -EACCES;
+
+ if (_IOC_TYPE(cmd) != IOC_LIBCFS_TYPE ||
+ _IOC_NR(cmd) < IOC_LIBCFS_MIN_NR ||
+ _IOC_NR(cmd) > IOC_LIBCFS_MAX_NR) {
+ CDEBUG(D_IOCTL, "invalid ioctl ( type %d, nr %d, size %d )\n",
+ _IOC_TYPE(cmd), _IOC_NR(cmd), _IOC_SIZE(cmd));
+ return -EINVAL;
+ }
+
+ /* 'cmd' and permissions get checked in our arch-specific caller */
+ err = lnet_ioctl_getdata(&hdr, uparam);
+ if (err != 0) {
+ CDEBUG_LIMIT(D_ERROR,
+ "lnet ioctl: data header error %d\n", err);
+ RETURN(err);
+ }
+
+ if (hdr->ioc_version == LNET_IOCTL_VERSION) {
+ /* The lnet_ioctl_data_adjust() function performs adjustment
+ * operations on the libcfs_ioctl_data structure to make
+ * it usable by the code. This doesn't need to be called
+ * for new data structures added.
+ */
+ data = container_of(hdr, struct libcfs_ioctl_data, ioc_hdr);
+ err = lnet_ioctl_data_adjust(data);
+ if (err != 0)
+ GOTO(out, err);
+ }
+
+ CDEBUG(D_IOCTL, "lnet ioctl cmd %u\n", cmd);
+
+#if LUSTRE_VERSION_CODE < OBD_OCD_VERSION(2, 18, 53, 0)
+ err = libcfs_ioctl(cmd, data);
+ if (err == -EINVAL)
+#endif
+ err = lnet_ioctl(cmd, hdr);
+ if (err == -EINVAL) {
+ err = blocking_notifier_call_chain(&lnet_ioctl_list,
+ cmd, hdr);
+ if (!(err & NOTIFY_STOP_MASK))
+ /* No-one claimed the ioctl */
+ err = -EINVAL;
+ else
+ err = notifier_to_errno(err);
+ }
+ if (copy_to_user(uparam, hdr, hdr->ioc_len) && !err)
+ err = -EFAULT;
+out:
+ LIBCFS_FREE(hdr, hdr->ioc_len);
+ RETURN(err);
+}
+
+static const struct file_operations lnet_fops = {
+ .owner = THIS_MODULE,
+ .unlocked_ioctl = lnet_psdev_ioctl,
+};
+
+static struct miscdevice lnet_dev = {
+ .minor = MISC_DYNAMIC_MINOR,
+ .name = "lnet",
+ .fops = &lnet_fops,
+};
+
+static int __init lnet_init(void)
+{
+ int rc;
+
+ ENTRY;
+ rc = libcfs_setup();
+ if (rc)
+ return rc;
+
+ rc = cfs_cpu_init();
+ if (rc < 0) {
+ CERROR("cfs_cpu_init: rc = %d\n", rc);
+ RETURN(rc);
+ }
+
+ rc = lnet_lib_init();
+ if (rc != 0) {
+ CERROR("lnet_lib_init: rc = %d\n", rc);
+ cfs_cpu_fini();
+ RETURN(rc);
+ }
+
+ rc = misc_register(&lnet_dev);
+ if (rc) {
+ CERROR("misc_register: rc = %d\n", rc);
+ cfs_cpu_fini();
+ RETURN(rc);
+ }
+
+ if (live_router_check_interval != INT_MIN ||
+ dead_router_check_interval != INT_MIN)
+ LCONSOLE_WARN("live_router_check_interval and dead_router_check_interval have been deprecated. Use alive_router_check_interval instead. Ignoring these deprecated parameters.\n");
+
+ if (config_on_load) {
+ /* Have to schedule a separate thread to avoid deadlocking
+ * in modload */
+ (void)kthread_run(lnet_configure, NULL, "lnet_initd");
+ }
+
+ RETURN(0);
+}
+
+static void __exit lnet_exit(void)
+{
+ misc_deregister(&lnet_dev);
+
+ lnet_router_exit();
+ lnet_lib_exit();
+ cfs_cpu_fini();
+}
+
+MODULE_AUTHOR("OpenSFS, Inc. <http://www.lustre.org/>");
+MODULE_DESCRIPTION("Lustre Networking layer");
+MODULE_VERSION(LNET_VERSION);
MODULE_LICENSE("GPL");
-module_init(init_kportals_module);
-module_exit(exit_kportals_module);
+
+module_init(lnet_init);
+module_exit(lnet_exit);