/*

 * CDDL HEADER START

 *

 * The contents of this file are subject to the terms of the

 * Common Development and Distribution License (the "License").

 * You may not use this file except in compliance with the License.

 *

 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE

 * or http://www.opensolaris.org/os/licensing.

 * See the License for the specific language governing permissions

 * and limitations under the License.

 *

 * When distributing Covered Code, include this CDDL HEADER in each

 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.

 * If applicable, add the following below this CDDL HEADER, with the

 * fields enclosed by brackets "[]" replaced with your own identifying

 * information: Portions Copyright [yyyy] [name of copyright owner]

 *

 * CDDL HEADER END

 */

/*

 * Copyright (c) 2010, Oracle and/or its affiliates. All rights reserved.

 */

/*

 * Copyright (c) 2006 Oracle.  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 <sys/types.h>

#include <sys/stat.h>

#include <sys/conf.h>

#include <sys/ddi.h>

#include <sys/sunddi.h>

#include <sys/modctl.h>

#include <sys/rds.h>

#include <sys/stropts.h>

#include <sys/socket.h>

#include <sys/socketvar.h>

#include <sys/sockio.h>

#include <sys/sysmacros.h>

#include <inet/ip.h>

#include <net/if_types.h>

#include <sys/ib/clients/rdsv3/rdsv3.h>

#include <sys/ib/clients/rdsv3/rdma.h>

#include <sys/ib/clients/rdsv3/rdma_transport.h>

#include <sys/ib/clients/rdsv3/rdsv3_debug.h>

extern void rdsv3_remove_bound(struct rdsv3_sock *rds);

extern int rdsv3_verify_bind_address(ipaddr_t addr);

extern ddi_taskq_t	*rdsv3_taskq;

extern struct rdma_cm_id *rdsv3_rdma_listen_id;

/* this is just used for stats gathering :/ */

kmutex_t rdsv3_sock_lock;

static unsigned long rdsv3_sock_count;

list_t rdsv3_sock_list;

rdsv3_wait_queue_t rdsv3_poll_waitq;

/*

 * This is called as the final descriptor referencing this socket is closed.

 * We have to unbind the socket so that another socket can be bound to the

 * address it was using.

 *

 * We have to be careful about racing with the incoming path.  sock_orphan()

 * sets SOCK_DEAD and we use that as an indicator to the rx path that new

 * messages shouldn't be queued.

 */

/* ARGSUSED */

static int

rdsv3_release(sock_lower_handle_t proto_handle, int flgs, cred_t *cr)

{

	struct rsock *sk = (struct rsock *)proto_handle;

	struct rdsv3_sock *rs;

	if (sk == NULL)

		goto out;

	rs = rdsv3_sk_to_rs(sk);

	RDSV3_DPRINTF4("rdsv3_release", "Enter(rs: %p, sk: %p)", rs, sk);

	rdsv3_sk_sock_orphan(sk);

	rdsv3_cong_remove_socket(rs);

	rdsv3_remove_bound(rs);

	/*

	 * Note - rdsv3_clear_recv_queue grabs rs_recv_lock, so

	 * that ensures the recv path has completed messing

	 * with the socket.

	 */

	rdsv3_clear_recv_queue(rs);

	rdsv3_send_drop_to(rs, NULL);

	rdsv3_rdma_drop_keys(rs);

	(void) rdsv3_notify_queue_get(rs, NULL);

	mutex_enter(&rdsv3_sock_lock);

	list_remove_node(&rs->rs_item);

	rdsv3_sock_count--;

	mutex_exit(&rdsv3_sock_lock);

	while (sk->sk_refcount > 1) {

		/* wait for 1 sec and try again */

		delay(drv_usectohz(1000000));

	}

	/* this will free the rs and sk */

	rdsv3_sk_sock_put(sk);

	RDSV3_DPRINTF4("rdsv3_release", "Return (rds: %p)", rs);

out:

	return (0);

}

void

__rdsv3_wake_sk_sleep(struct rsock *sk)

{

	/* wakup anyone waiting in recvmsg */

	if (!rdsv3_sk_sock_flag(sk, SOCK_DEAD) && sk->sk_sleep)

		rdsv3_wake_up(sk->sk_sleep);

}

/*

 * Careful not to race with rdsv3_release -> sock_orphan which clears sk_sleep.

 * _bh() isn't OK here, we're called from interrupt handlers.  It's probably OK

 * to wake the waitqueue after sk_sleep is clear as we hold a sock ref, but

 * this seems more conservative.

 * NB - normally, one would use sk_callback_lock for this, but we can

 * get here from interrupts, whereas the network code grabs sk_callback_lock

 * with _lock_bh only - so relying on sk_callback_lock introduces livelocks.

 */

void

rdsv3_wake_sk_sleep(struct rdsv3_sock *rs)

{

	RDSV3_DPRINTF4("rdsv3_wake_sk_sleep", "Enter(rs: %p)", rs);

	rw_enter(&rs->rs_recv_lock, RW_READER);

	__rdsv3_wake_sk_sleep(rdsv3_rs_to_sk(rs));

	rw_exit(&rs->rs_recv_lock);

}

/*ARGSUSED*/

static int

rdsv3_getname(sock_lower_handle_t proto_handle, struct sockaddr *addr,

    socklen_t *addr_len, cred_t *cr)

{

	struct rsock *sk = (struct rsock *)proto_handle;

	struct sockaddr_in *sin = (struct sockaddr_in *)addr;

	struct rdsv3_sock *rs = rdsv3_sk_to_rs(sk);

	RDSV3_DPRINTF4("rdsv3_getname", "Enter(rs: %p, port: %d)", rs,

	    rs->rs_bound_port);

	sin->sin_port = rs->rs_bound_port;

	sin->sin_addr.s_addr = rs->rs_bound_addr;

	sin->sin_family = AF_INET_OFFLOAD;

	*addr_len = sizeof (*sin);

	return (0);

}

/*

 * RDS' poll is without a doubt the least intuitive part of the interface,

 * as POLLIN and POLLOUT do not behave entirely as you would expect from

 * a network protocol.

 *

 * POLLIN is asserted if

 *  -	there is data on the receive queue.

 *  -	to signal that a previously congested destination may have become

 *	uncongested

 *  -	A notification has been queued to the socket (this can be a congestion

 *	update, or a RDMA completion).

 *

 * POLLOUT is asserted if there is room on the send queue. This does not mean

 * however, that the next sendmsg() call will succeed. If the application tries

 * to send to a congested destination, the system call may still fail (and

 * return ENOBUFS).

 */

/* ARGSUSED */

static short

rdsv3_poll(sock_lower_handle_t proto_handle, short events, int anyyet,

    cred_t *cr)

{

	struct rsock	*sk = (struct rsock *)proto_handle;

	struct rdsv3_sock	*rs = rdsv3_sk_to_rs(sk);

	unsigned short mask = 0;

#if 0

	RDSV3_DPRINTF4("rdsv3_poll", "enter(%p %x %d)", rs, events, anyyet);

#endif

	/*

	 * If rs_seen_congestion is on, wait until it's off.

	 * This is implemented for the following OFED code.

	 * 	if (rs->rs_seen_congestion)

	 *		poll_wait(file, &rds_poll_waitq, wait);

	 */

	mutex_enter(&rdsv3_poll_waitq.waitq_mutex);

	while (rs->rs_seen_congestion) {

		cv_wait(&rdsv3_poll_waitq.waitq_cv,

		    &rdsv3_poll_waitq.waitq_mutex);

	}

	mutex_exit(&rdsv3_poll_waitq.waitq_mutex);

	rw_enter(&rs->rs_recv_lock, RW_READER);

	if (!rs->rs_cong_monitor) {

		/*

		 * When a congestion map was updated, we signal POLLIN for

		 * "historical" reasons. Applications can also poll for

		 * WRBAND instead.

		 */

		if (rdsv3_cong_updated_since(&rs->rs_cong_track))

			mask |= (POLLIN | POLLRDNORM | POLLWRBAND);

	} else {

		mutex_enter(&rs->rs_lock);

		if (rs->rs_cong_notify)

			mask |= (POLLIN | POLLRDNORM);

		mutex_exit(&rs->rs_lock);

	}

	if (!list_is_empty(&rs->rs_recv_queue) ||

	    !list_is_empty(&rs->rs_notify_queue))

		mask |= (POLLIN | POLLRDNORM);

	if (rs->rs_snd_bytes < rdsv3_sk_sndbuf(rs))

		mask |= (POLLOUT | POLLWRNORM);

	rw_exit(&rs->rs_recv_lock);

	/* clear state any time we wake a seen-congested socket */

	if (mask) {

		mutex_enter(&rdsv3_poll_waitq.waitq_mutex);

		rs->rs_seen_congestion = 0;

		mutex_exit(&rdsv3_poll_waitq.waitq_mutex);

	}

#if 0

	RDSV3_DPRINTF4("rdsv3_poll", "return(%p %x)", rs, mask);

#endif

	return (mask);

}

/* ARGSUSED */

static int

rdsv3_ioctl(sock_lower_handle_t proto_handle, int cmd, intptr_t arg,

    int mode, int32_t *rvalp, cred_t *cr)

{

	ksocket_t	so4;

	struct lifconf	lifc;

	struct lifreq	lifr, *lifrp;

	struct ifconf	ifc;

	struct ifreq	ifr;

	int		rval = 0, rc, len;

	int		numifs;

	int		bufsize;

	void		*buf;

	RDSV3_DPRINTF4("rdsv3_ioctl", "enter: cmd: %d", cmd);

	/* Only ipv4 for now */

	rval = ksocket_socket(&so4, PF_INET, SOCK_DGRAM, 0, KSOCKET_NOSLEEP,

	    CRED());

	if (rval != 0) {

		RDSV3_DPRINTF2("rdsv3_ioctl", "ksocket_socket returned %d",

		    rval);

		return (rval);

	}

	switch (cmd) {

	case SIOCGLIFNUM :

	case SIOCGIFNUM :

		rval = rdsv3_do_ip_ioctl(so4, &buf, &bufsize, &numifs);

		if (rval != 0) break;

		if (cmd == SIOCGLIFNUM) {

		} else {

			len = 0;

			for (lifrp = (struct lifreq *)buf, rc = 0; rc < numifs;

			    rc++, lifrp++) {

				if (strlen(lifrp->lifr_name) <= IFNAMSIZ) {

					len++;

				}

			}

			(void) ddi_copyout(&len, (void *)arg,

			    sizeof (int), 0);

		}

		kmem_free(buf, bufsize);

		break;

	case SIOCGLIFCONF :

		if (ddi_copyin((void *)arg, &lifc, sizeof (struct lifconf), 0)

		    != 0) {

			RDSV3_DPRINTF2("rdsv3_ioctl", "ddi_copyin failed lifc");

			rval = EFAULT;

			break;

		}

		rval = rdsv3_do_ip_ioctl(so4, &buf, &bufsize, &numifs);

		if (rval != 0) {

			RDSV3_DPRINTF2("rdsv3_ioctl",

			    "rdsv3_do_ip_ioctl failed: %d", rval);

			break;

		}

		if ((lifc.lifc_len > 0) && (numifs > 0)) {

			if (ddi_copyout(buf, (void *)lifc.lifc_req,

			    (lifc.lifc_len < bufsize) ? lifc.lifc_len :

			    bufsize, 0) != 0) {

				RDSV3_DPRINTF2("rdsv3_ioctl",

				    "copyout of records failed");

				rval = EFAULT;

			}

		}

		lifc.lifc_len = bufsize;

		if (ddi_copyout(&lifc, (void *)arg, sizeof (struct lifconf),

		    0) != 0) {

			RDSV3_DPRINTF2("rdsv3_ioctl",

			    "copyout of lifconf failed");

			rval = EFAULT;

		}

		kmem_free(buf, bufsize);

		break;

	case SIOCGIFCONF :

	case O_SIOCGIFCONF :

		if (ddi_copyin((void *)arg, &ifc, sizeof (struct ifconf), 0)

		    != 0) {

			RDSV3_DPRINTF2("rdsv3_ioctl", "ddi_copyin failed ifc");

			rval = EFAULT;

			break;

		}

		RDSV3_DPRINTF2("rdsv3_ioctl",

		    "O_SIOCGIFCONF: ifc_len: %d, req: %p",

		    ifc.ifc_len, ifc.ifc_req);

		rval = rdsv3_do_ip_ioctl_old(so4, &buf, &bufsize, &numifs);

		if (rval != 0) {

			RDSV3_DPRINTF2("rdsv3_ioctl",

			    "rdsv3_do_ip_ioctl_old failed: %d", rval);

			break;

		}

		if ((ifc.ifc_len > 0) && (numifs > 0)) {

			if (ddi_copyout(buf, (void *)ifc.ifc_req,

			    (ifc.ifc_len < bufsize) ? ifc.ifc_len :

			    bufsize, 0) != 0) {

				RDSV3_DPRINTF2("rdsv3_ioctl",

				    "copyout of records failed");

				rval = EFAULT;

			}

		}

		ifc.ifc_len = bufsize;

		if (ddi_copyout(&ifc, (void *)arg, sizeof (struct ifconf),

		    0) != 0) {

			RDSV3_DPRINTF2("rdsv3_ioctl",

			    "copyout of ifconf failed");

			rval = EFAULT;

		}

		kmem_free(buf, bufsize);

		break;

	case SIOCGLIFFLAGS :

	case SIOCSLIFFLAGS :

	case SIOCGLIFMTU :

	case SIOCGLIFNETMASK :

	case SIOCGLIFINDEX :

		if (ddi_copyin((void *)arg, &lifr, sizeof (struct lifreq), 0)

		    != 0) {

			RDSV3_DPRINTF2("rdsv3_ioctl", "ddi_copyin failed lifr");

			rval = EFAULT;

			break;

		}

		rc = ksocket_ioctl(so4, cmd, (intptr_t)&lifr, &rval, CRED());

		if (rc != 0) {

			RDSV3_DPRINTF2("rdsv3_ioctl",

			    "ksocket_ioctl failed: %d, name: %s cmd: 0x%x",

			    rc, lifr.lifr_name, cmd);

			break;

		}

		(void) ddi_copyout(&lifr, (void *)arg,

		    sizeof (struct lifreq), 0);

		break;

	case SIOCGIFFLAGS :

	case SIOCSIFFLAGS :

	case SIOCGIFMTU :

	case SIOCGIFNETMASK :

	case SIOCGIFINDEX :

		if (ddi_copyin((void *)arg, &ifr, sizeof (struct ifreq), 0)

		    != 0) {

			RDSV3_DPRINTF2("rdsv3_ioctl", "ddi_copyin failed ifr");

			rval = EFAULT;

			break;

		}

		RDSV3_DPRINTF2("rdsv3_ioctl", "1. name: %s", ifr.ifr_name);

		rc = ksocket_ioctl(so4, cmd, (intptr_t)&ifr, &rval, CRED());

		if (rc != 0) {

			RDSV3_DPRINTF2("rdsv3_ioctl",

			    "ksocket_ioctl failed: %d, name: %s cmd: 0x%x",

			    rc, ifr.ifr_name, cmd);

			break;

		}

		RDSV3_DPRINTF2("rdsv3_ioctl", "2. name: %s", ifr.ifr_name);

		(void) ddi_copyout(&ifr, (void *)arg,

		    sizeof (struct ifreq), 0);

		break;

	default:

		cmn_err(CE_CONT, "unsupported IOCTL cmd: %d \n", cmd);

		rval = EOPNOTSUPP;

	}

	(void) ksocket_close(so4, CRED());

	RDSV3_DPRINTF4("rdsv3_ioctl", "return: %d cmd: %d", rval, cmd);

	*rvalp = rval;

	return (rval);

}

static int

rdsv3_cancel_sent_to(struct rdsv3_sock *rs, char *optval, int len)

{

	struct sockaddr_in sin;

	/* racing with another thread binding seems ok here */

	if (rs->rs_bound_addr == 0)

		return (-ENOTCONN); /* XXX not a great errno */

	if (len < sizeof (struct sockaddr_in))

		return (-EINVAL);

	if (ddi_copyin((void *)optval, &sin, sizeof (struct sockaddr_in),

	    0) != 0) {

		RDSV3_DPRINTF2("rdsv3_cancel_sent_to", "ddi_copyin failed sin");

		return (-EFAULT);

	}

	rdsv3_send_drop_to(rs, &sin);

	return (0);

}

static int

rdsv3_set_bool_option(unsigned char *optvar, char *optval, int optlen)

{

	int value = *optval;

	if (optlen < sizeof (int))

		return (-EINVAL);

	*optvar = !!value;

	return (0);

}

static int

rdsv3_cong_monitor(struct rdsv3_sock *rs, char *optval, int optlen)

{

	int ret;

	ret = rdsv3_set_bool_option(&rs->rs_cong_monitor, optval, optlen);

	if (ret == 0) {

		if (rs->rs_cong_monitor) {

			rdsv3_cong_add_socket(rs);

		} else {

			rdsv3_cong_remove_socket(rs);

			rs->rs_cong_mask = 0;

			rs->rs_cong_notify = 0;

		}

	}

	return (ret);

}

/*ARGSUSED*/