ipath_qp.c

来自「linux 内核源代码」· C语言 代码 · 共 1,082 行 · 第 1/2 页

/*
 * Copyright (c) 2006, 2007 QLogic Corporation. All rights reserved.
 * Copyright (c) 2005, 2006 PathScale, Inc. 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/err.h>
#include <linux/vmalloc.h>

#include "ipath_verbs.h"
#include "ipath_kernel.h"

#define BITS_PER_PAGE		(PAGE_SIZE*BITS_PER_BYTE)
#define BITS_PER_PAGE_MASK	(BITS_PER_PAGE-1)
#define mk_qpn(qpt, map, off)	(((map) - (qpt)->map) * BITS_PER_PAGE + \
				 (off))
#define find_next_offset(map, off) find_next_zero_bit((map)->page, \
						      BITS_PER_PAGE, off)

/*
 * Convert the AETH credit code into the number of credits.
 */
static u32 credit_table[31] = {
	0,			/* 0 */
	1,			/* 1 */
	2,			/* 2 */
	3,			/* 3 */
	4,			/* 4 */
	6,			/* 5 */
	8,			/* 6 */
	12,			/* 7 */
	16,			/* 8 */
	24,			/* 9 */
	32,			/* A */
	48,			/* B */
	64,			/* C */
	96,			/* D */
	128,			/* E */
	192,			/* F */
	256,			/* 10 */
	384,			/* 11 */
	512,			/* 12 */
	768,			/* 13 */
	1024,			/* 14 */
	1536,			/* 15 */
	2048,			/* 16 */
	3072,			/* 17 */
	4096,			/* 18 */
	6144,			/* 19 */
	8192,			/* 1A */
	12288,			/* 1B */
	16384,			/* 1C */
	24576,			/* 1D */
	32768			/* 1E */
};


static void get_map_page(struct ipath_qp_table *qpt, struct qpn_map *map)
{
	unsigned long page = get_zeroed_page(GFP_KERNEL);
	unsigned long flags;

	/*
	 * Free the page if someone raced with us installing it.
	 */

	spin_lock_irqsave(&qpt->lock, flags);
	if (map->page)
		free_page(page);
	else
		map->page = (void *)page;
	spin_unlock_irqrestore(&qpt->lock, flags);
}


static int alloc_qpn(struct ipath_qp_table *qpt, enum ib_qp_type type)
{
	u32 i, offset, max_scan, qpn;
	struct qpn_map *map;
	u32 ret = -1;

	if (type == IB_QPT_SMI)
		ret = 0;
	else if (type == IB_QPT_GSI)
		ret = 1;

	if (ret != -1) {
		map = &qpt->map[0];
		if (unlikely(!map->page)) {
			get_map_page(qpt, map);
			if (unlikely(!map->page)) {
				ret = -ENOMEM;
				goto bail;
			}
		}
		if (!test_and_set_bit(ret, map->page))
			atomic_dec(&map->n_free);
		else
			ret = -EBUSY;
		goto bail;
	}

	qpn = qpt->last + 1;
	if (qpn >= QPN_MAX)
		qpn = 2;
	offset = qpn & BITS_PER_PAGE_MASK;
	map = &qpt->map[qpn / BITS_PER_PAGE];
	max_scan = qpt->nmaps - !offset;
	for (i = 0;;) {
		if (unlikely(!map->page)) {
			get_map_page(qpt, map);
			if (unlikely(!map->page))
				break;
		}
		if (likely(atomic_read(&map->n_free))) {
			do {
				if (!test_and_set_bit(offset, map->page)) {
					atomic_dec(&map->n_free);
					qpt->last = qpn;
					ret = qpn;
					goto bail;
				}
				offset = find_next_offset(map, offset);
				qpn = mk_qpn(qpt, map, offset);
				/*
				 * This test differs from alloc_pidmap().
				 * If find_next_offset() does find a zero
				 * bit, we don't need to check for QPN
				 * wrapping around past our starting QPN.
				 * We just need to be sure we don't loop
				 * forever.
				 */
			} while (offset < BITS_PER_PAGE && qpn < QPN_MAX);
		}
		/*
		 * In order to keep the number of pages allocated to a
		 * minimum, we scan the all existing pages before increasing
		 * the size of the bitmap table.
		 */
		if (++i > max_scan) {
			if (qpt->nmaps == QPNMAP_ENTRIES)
				break;
			map = &qpt->map[qpt->nmaps++];
			offset = 0;
		} else if (map < &qpt->map[qpt->nmaps]) {
			++map;
			offset = 0;
		} else {
			map = &qpt->map[0];
			offset = 2;
		}
		qpn = mk_qpn(qpt, map, offset);
	}

	ret = -ENOMEM;

bail:
	return ret;
}

static void free_qpn(struct ipath_qp_table *qpt, u32 qpn)
{
	struct qpn_map *map;

	map = qpt->map + qpn / BITS_PER_PAGE;
	if (map->page)
		clear_bit(qpn & BITS_PER_PAGE_MASK, map->page);
	atomic_inc(&map->n_free);
}

/**
 * ipath_alloc_qpn - allocate a QP number
 * @qpt: the QP table
 * @qp: the QP
 * @type: the QP type (IB_QPT_SMI and IB_QPT_GSI are special)
 *
 * Allocate the next available QPN and put the QP into the hash table.
 * The hash table holds a reference to the QP.
 */
static int ipath_alloc_qpn(struct ipath_qp_table *qpt, struct ipath_qp *qp,
			   enum ib_qp_type type)
{
	unsigned long flags;
	int ret;

	ret = alloc_qpn(qpt, type);
	if (ret < 0)
		goto bail;
	qp->ibqp.qp_num = ret;

	/* Add the QP to the hash table. */
	spin_lock_irqsave(&qpt->lock, flags);

	ret %= qpt->max;
	qp->next = qpt->table[ret];
	qpt->table[ret] = qp;
	atomic_inc(&qp->refcount);

	spin_unlock_irqrestore(&qpt->lock, flags);
	ret = 0;

bail:
	return ret;
}

/**
 * ipath_free_qp - remove a QP from the QP table
 * @qpt: the QP table
 * @qp: the QP to remove
 *
 * Remove the QP from the table so it can't be found asynchronously by
 * the receive interrupt routine.
 */
static void ipath_free_qp(struct ipath_qp_table *qpt, struct ipath_qp *qp)
{
	struct ipath_qp *q, **qpp;
	unsigned long flags;
	int fnd = 0;

	spin_lock_irqsave(&qpt->lock, flags);

	/* Remove QP from the hash table. */
	qpp = &qpt->table[qp->ibqp.qp_num % qpt->max];
	for (; (q = *qpp) != NULL; qpp = &q->next) {
		if (q == qp) {
			*qpp = qp->next;
			qp->next = NULL;
			atomic_dec(&qp->refcount);
			fnd = 1;
			break;
		}
	}

	spin_unlock_irqrestore(&qpt->lock, flags);

	if (!fnd)
		return;

	free_qpn(qpt, qp->ibqp.qp_num);

	wait_event(qp->wait, !atomic_read(&qp->refcount));
}

/**
 * ipath_free_all_qps - remove all QPs from the table
 * @qpt: the QP table to empty
 */
void ipath_free_all_qps(struct ipath_qp_table *qpt)
{
	unsigned long flags;
	struct ipath_qp *qp, *nqp;
	u32 n;

	for (n = 0; n < qpt->max; n++) {
		spin_lock_irqsave(&qpt->lock, flags);
		qp = qpt->table[n];
		qpt->table[n] = NULL;
		spin_unlock_irqrestore(&qpt->lock, flags);

		while (qp) {
			nqp = qp->next;
			free_qpn(qpt, qp->ibqp.qp_num);
			if (!atomic_dec_and_test(&qp->refcount) ||
			    !ipath_destroy_qp(&qp->ibqp))
				ipath_dbg("QP memory leak!\n");
			qp = nqp;
		}
	}

	for (n = 0; n < ARRAY_SIZE(qpt->map); n++) {
		if (qpt->map[n].page)
			free_page((unsigned long)qpt->map[n].page);
	}
}

/**
 * ipath_lookup_qpn - return the QP with the given QPN
 * @qpt: the QP table
 * @qpn: the QP number to look up
 *
 * The caller is responsible for decrementing the QP reference count
 * when done.
 */
struct ipath_qp *ipath_lookup_qpn(struct ipath_qp_table *qpt, u32 qpn)
{
	unsigned long flags;
	struct ipath_qp *qp;

	spin_lock_irqsave(&qpt->lock, flags);

	for (qp = qpt->table[qpn % qpt->max]; qp; qp = qp->next) {
		if (qp->ibqp.qp_num == qpn) {
			atomic_inc(&qp->refcount);
			break;
		}
	}

	spin_unlock_irqrestore(&qpt->lock, flags);
	return qp;
}

/**
 * ipath_reset_qp - initialize the QP state to the reset state
 * @qp: the QP to reset
 */
static void ipath_reset_qp(struct ipath_qp *qp)
{
	qp->remote_qpn = 0;
	qp->qkey = 0;
	qp->qp_access_flags = 0;
	qp->s_busy = 0;
	qp->s_flags &= IPATH_S_SIGNAL_REQ_WR;
	qp->s_hdrwords = 0;
	qp->s_wqe = NULL;
	qp->s_psn = 0;
	qp->r_psn = 0;
	qp->r_msn = 0;
	if (qp->ibqp.qp_type == IB_QPT_RC) {
		qp->s_state = IB_OPCODE_RC_SEND_LAST;
		qp->r_state = IB_OPCODE_RC_SEND_LAST;
	} else {
		qp->s_state = IB_OPCODE_UC_SEND_LAST;
		qp->r_state = IB_OPCODE_UC_SEND_LAST;
	}
	qp->s_ack_state = IB_OPCODE_RC_ACKNOWLEDGE;
	qp->r_nak_state = 0;
	qp->r_wrid_valid = 0;
	qp->s_rnr_timeout = 0;
	qp->s_head = 0;
	qp->s_tail = 0;
	qp->s_cur = 0;
	qp->s_last = 0;
	qp->s_ssn = 1;
	qp->s_lsn = 0;
	qp->s_wait_credit = 0;
	memset(qp->s_ack_queue, 0, sizeof(qp->s_ack_queue));
	qp->r_head_ack_queue = 0;
	qp->s_tail_ack_queue = 0;
	qp->s_num_rd_atomic = 0;
	if (qp->r_rq.wq) {
		qp->r_rq.wq->head = 0;
		qp->r_rq.wq->tail = 0;
	}
	qp->r_reuse_sge = 0;
}

/**
 * ipath_error_qp - put a QP into an error state
 * @qp: the QP to put into an error state
 * @err: the receive completion error to signal if a RWQE is active
 *
 * Flushes both send and receive work queues.
 * Returns true if last WQE event should be generated.
 * The QP s_lock should be held and interrupts disabled.
 */

int ipath_error_qp(struct ipath_qp *qp, enum ib_wc_status err)
{
	struct ipath_ibdev *dev = to_idev(qp->ibqp.device);
	struct ib_wc wc;
	int ret = 0;

	ipath_dbg("QP%d/%d in error state\n",
		  qp->ibqp.qp_num, qp->remote_qpn);

	spin_lock(&dev->pending_lock);
	/* XXX What if its already removed by the timeout code? */
	if (!list_empty(&qp->timerwait))
		list_del_init(&qp->timerwait);
	if (!list_empty(&qp->piowait))
		list_del_init(&qp->piowait);
	spin_unlock(&dev->pending_lock);

	wc.vendor_err = 0;
	wc.byte_len = 0;
	wc.imm_data = 0;
	wc.qp = &qp->ibqp;
	wc.src_qp = 0;
	wc.wc_flags = 0;
	wc.pkey_index = 0;
	wc.slid = 0;
	wc.sl = 0;
	wc.dlid_path_bits = 0;
	wc.port_num = 0;
	if (qp->r_wrid_valid) {
		qp->r_wrid_valid = 0;
		wc.wr_id = qp->r_wr_id;
		wc.opcode = IB_WC_RECV;
		wc.status = err;
		ipath_cq_enter(to_icq(qp->ibqp.send_cq), &wc, 1);
	}
	wc.status = IB_WC_WR_FLUSH_ERR;

	while (qp->s_last != qp->s_head) {
		struct ipath_swqe *wqe = get_swqe_ptr(qp, qp->s_last);

		wc.wr_id = wqe->wr.wr_id;
		wc.opcode = ib_ipath_wc_opcode[wqe->wr.opcode];
		if (++qp->s_last >= qp->s_size)
			qp->s_last = 0;
		ipath_cq_enter(to_icq(qp->ibqp.send_cq), &wc, 1);
	}
	qp->s_cur = qp->s_tail = qp->s_head;
	qp->s_hdrwords = 0;
	qp->s_ack_state = IB_OPCODE_RC_ACKNOWLEDGE;

	if (qp->r_rq.wq) {
		struct ipath_rwq *wq;
		u32 head;
		u32 tail;

		spin_lock(&qp->r_rq.lock);

		/* sanity check pointers before trusting them */
		wq = qp->r_rq.wq;
		head = wq->head;
		if (head >= qp->r_rq.size)
			head = 0;
		tail = wq->tail;
		if (tail >= qp->r_rq.size)
			tail = 0;
		wc.opcode = IB_WC_RECV;
		while (tail != head) {
			wc.wr_id = get_rwqe_ptr(&qp->r_rq, tail)->wr_id;
			if (++tail >= qp->r_rq.size)
				tail = 0;
			ipath_cq_enter(to_icq(qp->ibqp.recv_cq), &wc, 1);
		}
		wq->tail = tail;

		spin_unlock(&qp->r_rq.lock);
	} else if (qp->ibqp.event_handler)
		ret = 1;

	return ret;
}

/**
 * ipath_modify_qp - modify the attributes of a queue pair
 * @ibqp: the queue pair who's attributes we're modifying
 * @attr: the new attributes
 * @attr_mask: the mask of attributes to modify
 * @udata: user data for ipathverbs.so
 *
 * Returns 0 on success, otherwise returns an errno.
 */
int ipath_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
		    int attr_mask, struct ib_udata *udata)
{
	struct ipath_ibdev *dev = to_idev(ibqp->device);
	struct ipath_qp *qp = to_iqp(ibqp);
	enum ib_qp_state cur_state, new_state;
	unsigned long flags;
	int lastwqe = 0;
	int ret;

	spin_lock_irqsave(&qp->s_lock, flags);

	cur_state = attr_mask & IB_QP_CUR_STATE ?
		attr->cur_qp_state : qp->state;
	new_state = attr_mask & IB_QP_STATE ? attr->qp_state : cur_state;

	if (!ib_modify_qp_is_ok(cur_state, new_state, ibqp->qp_type,
				attr_mask))
		goto inval;

	if (attr_mask & IB_QP_AV) {
		if (attr->ah_attr.dlid == 0 ||
		    attr->ah_attr.dlid >= IPATH_MULTICAST_LID_BASE)
			goto inval;

		if ((attr->ah_attr.ah_flags & IB_AH_GRH) &&
		    (attr->ah_attr.grh.sgid_index > 1))
			goto inval;
	}

	if (attr_mask & IB_QP_PKEY_INDEX)
		if (attr->pkey_index >= ipath_get_npkeys(dev->dd))
			goto inval;

	if (attr_mask & IB_QP_MIN_RNR_TIMER)
		if (attr->min_rnr_timer > 31)
			goto inval;

	if (attr_mask & IB_QP_PORT)
		if (attr->port_num == 0 ||
		    attr->port_num > ibqp->device->phys_port_cnt)
			goto inval;

	/*
	 * Note: the chips support a maximum MTU of 4096, but the driver
	 * hasn't implemented this feature yet, so don't allow Path MTU
	 * values greater than 2048.
	 */
	if (attr_mask & IB_QP_PATH_MTU)
		if (attr->path_mtu > IB_MTU_2048)
			goto inval;

	if (attr_mask & IB_QP_PATH_MIG_STATE)
		if (attr->path_mig_state != IB_MIG_MIGRATED &&
		    attr->path_mig_state != IB_MIG_REARM)
			goto inval;

	if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC)
		if (attr->max_dest_rd_atomic > IPATH_MAX_RDMA_ATOMIC)
			goto inval;

	switch (new_state) {
	case IB_QPS_RESET:
		ipath_reset_qp(qp);
		break;

	case IB_QPS_ERR:
		lastwqe = ipath_error_qp(qp, IB_WC_WR_FLUSH_ERR);