ucma.c

来自「linux 内核源代码」· C语言 代码 · 共 1,148 行 · 第 1/2 页

/*
 * Copyright (c) 2005-2006 Intel Corporation.  All rights reserved.
 *
 * This software is available to you under a choice of one of two
 * licenses.  You may choose to be licensed under the terms of the GNU
 * General Public License (GPL) Version 2, available from the file
 * COPYING in the main directory of this source tree, or the
 * OpenIB.org BSD license below:
 *
 *     Redistribution and use in source and binary forms, with or
 *     without modification, are permitted provided that the following
 *     conditions are met:
 *
 *      - Redistributions of source code must retain the above
 *	copyright notice, this list of conditions and the following
 *	disclaimer.
 *
 *      - Redistributions in binary form must reproduce the above
 *	copyright notice, this list of conditions and the following
 *	disclaimer in the documentation and/or other materials
 *	provided with the distribution.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 */

#include <linux/completion.h>
#include <linux/mutex.h>
#include <linux/poll.h>
#include <linux/idr.h>
#include <linux/in.h>
#include <linux/in6.h>
#include <linux/miscdevice.h>

#include <rdma/rdma_user_cm.h>
#include <rdma/ib_marshall.h>
#include <rdma/rdma_cm.h>

MODULE_AUTHOR("Sean Hefty");
MODULE_DESCRIPTION("RDMA Userspace Connection Manager Access");
MODULE_LICENSE("Dual BSD/GPL");

enum {
	UCMA_MAX_BACKLOG	= 128
};

struct ucma_file {
	struct mutex		mut;
	struct file		*filp;
	struct list_head	ctx_list;
	struct list_head	event_list;
	wait_queue_head_t	poll_wait;
};

struct ucma_context {
	int			id;
	struct completion	comp;
	atomic_t		ref;
	int			events_reported;
	int			backlog;

	struct ucma_file	*file;
	struct rdma_cm_id	*cm_id;
	u64			uid;

	struct list_head	list;
	struct list_head	mc_list;
};

struct ucma_multicast {
	struct ucma_context	*ctx;
	int			id;
	int			events_reported;

	u64			uid;
	struct list_head	list;
	struct sockaddr		addr;
	u8			pad[sizeof(struct sockaddr_in6) -
				    sizeof(struct sockaddr)];
};

struct ucma_event {
	struct ucma_context	*ctx;
	struct ucma_multicast	*mc;
	struct list_head	list;
	struct rdma_cm_id	*cm_id;
	struct rdma_ucm_event_resp resp;
};

static DEFINE_MUTEX(mut);
static DEFINE_IDR(ctx_idr);
static DEFINE_IDR(multicast_idr);

static inline struct ucma_context *_ucma_find_context(int id,
						      struct ucma_file *file)
{
	struct ucma_context *ctx;

	ctx = idr_find(&ctx_idr, id);
	if (!ctx)
		ctx = ERR_PTR(-ENOENT);
	else if (ctx->file != file)
		ctx = ERR_PTR(-EINVAL);
	return ctx;
}

static struct ucma_context *ucma_get_ctx(struct ucma_file *file, int id)
{
	struct ucma_context *ctx;

	mutex_lock(&mut);
	ctx = _ucma_find_context(id, file);
	if (!IS_ERR(ctx))
		atomic_inc(&ctx->ref);
	mutex_unlock(&mut);
	return ctx;
}

static void ucma_put_ctx(struct ucma_context *ctx)
{
	if (atomic_dec_and_test(&ctx->ref))
		complete(&ctx->comp);
}

static struct ucma_context *ucma_alloc_ctx(struct ucma_file *file)
{
	struct ucma_context *ctx;
	int ret;

	ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
	if (!ctx)
		return NULL;

	atomic_set(&ctx->ref, 1);
	init_completion(&ctx->comp);
	INIT_LIST_HEAD(&ctx->mc_list);
	ctx->file = file;

	do {
		ret = idr_pre_get(&ctx_idr, GFP_KERNEL);
		if (!ret)
			goto error;

		mutex_lock(&mut);
		ret = idr_get_new(&ctx_idr, ctx, &ctx->id);
		mutex_unlock(&mut);
	} while (ret == -EAGAIN);

	if (ret)
		goto error;

	list_add_tail(&ctx->list, &file->ctx_list);
	return ctx;

error:
	kfree(ctx);
	return NULL;
}

static struct ucma_multicast* ucma_alloc_multicast(struct ucma_context *ctx)
{
	struct ucma_multicast *mc;
	int ret;

	mc = kzalloc(sizeof(*mc), GFP_KERNEL);
	if (!mc)
		return NULL;

	do {
		ret = idr_pre_get(&multicast_idr, GFP_KERNEL);
		if (!ret)
			goto error;

		mutex_lock(&mut);
		ret = idr_get_new(&multicast_idr, mc, &mc->id);
		mutex_unlock(&mut);
	} while (ret == -EAGAIN);

	if (ret)
		goto error;

	mc->ctx = ctx;
	list_add_tail(&mc->list, &ctx->mc_list);
	return mc;

error:
	kfree(mc);
	return NULL;
}

static void ucma_copy_conn_event(struct rdma_ucm_conn_param *dst,
				 struct rdma_conn_param *src)
{
	if (src->private_data_len)
		memcpy(dst->private_data, src->private_data,
		       src->private_data_len);
	dst->private_data_len = src->private_data_len;
	dst->responder_resources =src->responder_resources;
	dst->initiator_depth = src->initiator_depth;
	dst->flow_control = src->flow_control;
	dst->retry_count = src->retry_count;
	dst->rnr_retry_count = src->rnr_retry_count;
	dst->srq = src->srq;
	dst->qp_num = src->qp_num;
}

static void ucma_copy_ud_event(struct rdma_ucm_ud_param *dst,
			       struct rdma_ud_param *src)
{
	if (src->private_data_len)
		memcpy(dst->private_data, src->private_data,
		       src->private_data_len);
	dst->private_data_len = src->private_data_len;
	ib_copy_ah_attr_to_user(&dst->ah_attr, &src->ah_attr);
	dst->qp_num = src->qp_num;
	dst->qkey = src->qkey;
}

static void ucma_set_event_context(struct ucma_context *ctx,
				   struct rdma_cm_event *event,
				   struct ucma_event *uevent)
{
	uevent->ctx = ctx;
	switch (event->event) {
	case RDMA_CM_EVENT_MULTICAST_JOIN:
	case RDMA_CM_EVENT_MULTICAST_ERROR:
		uevent->mc = (struct ucma_multicast *)
			     event->param.ud.private_data;
		uevent->resp.uid = uevent->mc->uid;
		uevent->resp.id = uevent->mc->id;
		break;
	default:
		uevent->resp.uid = ctx->uid;
		uevent->resp.id = ctx->id;
		break;
	}
}

static int ucma_event_handler(struct rdma_cm_id *cm_id,
			      struct rdma_cm_event *event)
{
	struct ucma_event *uevent;
	struct ucma_context *ctx = cm_id->context;
	int ret = 0;

	uevent = kzalloc(sizeof(*uevent), GFP_KERNEL);
	if (!uevent)
		return event->event == RDMA_CM_EVENT_CONNECT_REQUEST;

	uevent->cm_id = cm_id;
	ucma_set_event_context(ctx, event, uevent);
	uevent->resp.event = event->event;
	uevent->resp.status = event->status;
	if (cm_id->ps == RDMA_PS_UDP || cm_id->ps == RDMA_PS_IPOIB)
		ucma_copy_ud_event(&uevent->resp.param.ud, &event->param.ud);
	else
		ucma_copy_conn_event(&uevent->resp.param.conn,
				     &event->param.conn);

	mutex_lock(&ctx->file->mut);
	if (event->event == RDMA_CM_EVENT_CONNECT_REQUEST) {
		if (!ctx->backlog) {
			ret = -ENOMEM;
			kfree(uevent);
			goto out;
		}
		ctx->backlog--;
	} else if (!ctx->uid) {
		/*
		 * We ignore events for new connections until userspace has set
		 * their context.  This can only happen if an error occurs on a
		 * new connection before the user accepts it.  This is okay,
		 * since the accept will just fail later.
		 */
		kfree(uevent);
		goto out;
	}

	list_add_tail(&uevent->list, &ctx->file->event_list);
	wake_up_interruptible(&ctx->file->poll_wait);
out:
	mutex_unlock(&ctx->file->mut);
	return ret;
}

static ssize_t ucma_get_event(struct ucma_file *file, const char __user *inbuf,
			      int in_len, int out_len)
{
	struct ucma_context *ctx;
	struct rdma_ucm_get_event cmd;
	struct ucma_event *uevent;
	int ret = 0;
	DEFINE_WAIT(wait);

	if (out_len < sizeof uevent->resp)
		return -ENOSPC;

	if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
		return -EFAULT;

	mutex_lock(&file->mut);
	while (list_empty(&file->event_list)) {
		mutex_unlock(&file->mut);

		if (file->filp->f_flags & O_NONBLOCK)
			return -EAGAIN;

		if (wait_event_interruptible(file->poll_wait,
					     !list_empty(&file->event_list)))
			return -ERESTARTSYS;

		mutex_lock(&file->mut);
	}

	uevent = list_entry(file->event_list.next, struct ucma_event, list);

	if (uevent->resp.event == RDMA_CM_EVENT_CONNECT_REQUEST) {
		ctx = ucma_alloc_ctx(file);
		if (!ctx) {
			ret = -ENOMEM;
			goto done;
		}
		uevent->ctx->backlog++;
		ctx->cm_id = uevent->cm_id;
		ctx->cm_id->context = ctx;
		uevent->resp.id = ctx->id;
	}

	if (copy_to_user((void __user *)(unsigned long)cmd.response,
			 &uevent->resp, sizeof uevent->resp)) {
		ret = -EFAULT;
		goto done;
	}

	list_del(&uevent->list);
	uevent->ctx->events_reported++;
	if (uevent->mc)
		uevent->mc->events_reported++;
	kfree(uevent);
done:
	mutex_unlock(&file->mut);
	return ret;
}

static ssize_t ucma_create_id(struct ucma_file *file,
				const char __user *inbuf,
				int in_len, int out_len)
{
	struct rdma_ucm_create_id cmd;
	struct rdma_ucm_create_id_resp resp;
	struct ucma_context *ctx;
	int ret;

	if (out_len < sizeof(resp))
		return -ENOSPC;

	if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
		return -EFAULT;

	mutex_lock(&file->mut);
	ctx = ucma_alloc_ctx(file);
	mutex_unlock(&file->mut);
	if (!ctx)
		return -ENOMEM;

	ctx->uid = cmd.uid;
	ctx->cm_id = rdma_create_id(ucma_event_handler, ctx, cmd.ps);
	if (IS_ERR(ctx->cm_id)) {
		ret = PTR_ERR(ctx->cm_id);
		goto err1;
	}

	resp.id = ctx->id;
	if (copy_to_user((void __user *)(unsigned long)cmd.response,
			 &resp, sizeof(resp))) {
		ret = -EFAULT;
		goto err2;
	}
	return 0;

err2:
	rdma_destroy_id(ctx->cm_id);
err1:
	mutex_lock(&mut);
	idr_remove(&ctx_idr, ctx->id);
	mutex_unlock(&mut);
	kfree(ctx);
	return ret;
}

static void ucma_cleanup_multicast(struct ucma_context *ctx)
{
	struct ucma_multicast *mc, *tmp;

	mutex_lock(&mut);
	list_for_each_entry_safe(mc, tmp, &ctx->mc_list, list) {
		list_del(&mc->list);
		idr_remove(&multicast_idr, mc->id);
		kfree(mc);
	}
	mutex_unlock(&mut);
}

static void ucma_cleanup_events(struct ucma_context *ctx)
{
	struct ucma_event *uevent, *tmp;

	list_for_each_entry_safe(uevent, tmp, &ctx->file->event_list, list) {
		if (uevent->ctx != ctx)
			continue;

		list_del(&uevent->list);

		/* clear incoming connections. */
		if (uevent->resp.event == RDMA_CM_EVENT_CONNECT_REQUEST)
			rdma_destroy_id(uevent->cm_id);

		kfree(uevent);
	}
}

static void ucma_cleanup_mc_events(struct ucma_multicast *mc)
{
	struct ucma_event *uevent, *tmp;

	list_for_each_entry_safe(uevent, tmp, &mc->ctx->file->event_list, list) {
		if (uevent->mc != mc)
			continue;

		list_del(&uevent->list);
		kfree(uevent);
	}
}

static int ucma_free_ctx(struct ucma_context *ctx)
{
	int events_reported;

	/* No new events will be generated after destroying the id. */
	rdma_destroy_id(ctx->cm_id);

	ucma_cleanup_multicast(ctx);

	/* Cleanup events not yet reported to the user. */
	mutex_lock(&ctx->file->mut);
	ucma_cleanup_events(ctx);
	list_del(&ctx->list);
	mutex_unlock(&ctx->file->mut);

	events_reported = ctx->events_reported;
	kfree(ctx);
	return events_reported;
}

static ssize_t ucma_destroy_id(struct ucma_file *file, const char __user *inbuf,
			       int in_len, int out_len)
{
	struct rdma_ucm_destroy_id cmd;
	struct rdma_ucm_destroy_id_resp resp;
	struct ucma_context *ctx;
	int ret = 0;

	if (out_len < sizeof(resp))
		return -ENOSPC;

	if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
		return -EFAULT;

	mutex_lock(&mut);
	ctx = _ucma_find_context(cmd.id, file);
	if (!IS_ERR(ctx))
		idr_remove(&ctx_idr, ctx->id);
	mutex_unlock(&mut);

	if (IS_ERR(ctx))
		return PTR_ERR(ctx);

	ucma_put_ctx(ctx);
	wait_for_completion(&ctx->comp);
	resp.events_reported = ucma_free_ctx(ctx);

	if (copy_to_user((void __user *)(unsigned long)cmd.response,
			 &resp, sizeof(resp)))
		ret = -EFAULT;

	return ret;
}

static ssize_t ucma_bind_addr(struct ucma_file *file, const char __user *inbuf,
			      int in_len, int out_len)
{
	struct rdma_ucm_bind_addr cmd;
	struct ucma_context *ctx;
	int ret;

	if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
		return -EFAULT;

	ctx = ucma_get_ctx(file, cmd.id);
	if (IS_ERR(ctx))
		return PTR_ERR(ctx);

	ret = rdma_bind_addr(ctx->cm_id, (struct sockaddr *) &cmd.addr);
	ucma_put_ctx(ctx);
	return ret;
}

static ssize_t ucma_resolve_addr(struct ucma_file *file,
				 const char __user *inbuf,
				 int in_len, int out_len)
{
	struct rdma_ucm_resolve_addr cmd;
	struct ucma_context *ctx;
	int ret;

	if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
		return -EFAULT;

	ctx = ucma_get_ctx(file, cmd.id);
	if (IS_ERR(ctx))
		return PTR_ERR(ctx);

	ret = rdma_resolve_addr(ctx->cm_id, (struct sockaddr *) &cmd.src_addr,
				(struct sockaddr *) &cmd.dst_addr,
				cmd.timeout_ms);
	ucma_put_ctx(ctx);
	return ret;
}

static ssize_t ucma_resolve_route(struct ucma_file *file,
				  const char __user *inbuf,
				  int in_len, int out_len)
{
	struct rdma_ucm_resolve_route cmd;
	struct ucma_context *ctx;
	int ret;

	if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
		return -EFAULT;

	ctx = ucma_get_ctx(file, cmd.id);
	if (IS_ERR(ctx))
		return PTR_ERR(ctx);

	ret = rdma_resolve_route(ctx->cm_id, cmd.timeout_ms);
	ucma_put_ctx(ctx);
	return ret;
}

static void ucma_copy_ib_route(struct rdma_ucm_query_route_resp *resp,
			       struct rdma_route *route)
{
	struct rdma_dev_addr *dev_addr;

	resp->num_paths = route->num_paths;
	switch (route->num_paths) {
	case 0:
		dev_addr = &route->addr.dev_addr;
		ib_addr_get_dgid(dev_addr,
				 (union ib_gid *) &resp->ib_route[0].dgid);
		ib_addr_get_sgid(dev_addr,
				 (union ib_gid *) &resp->ib_route[0].sgid);
		resp->ib_route[0].pkey = cpu_to_be16(ib_addr_get_pkey(dev_addr));
		break;
	case 2:
		ib_copy_path_rec_to_user(&resp->ib_route[1],
					 &route->path_rec[1]);
		/* fall through */