Merge branch 'for-linus' of master.kernel.org:/pub/scm/linux/kernel/git/roland/infiniband

* 'for-linus' of master.kernel.org:/pub/scm/linux/kernel/git/roland/infiniband:
  IB/ipath: kbuild infrastructure
  IB/ipath: infiniband verbs support
  IB/ipath: misc infiniband code, part 2
  IB/ipath: misc infiniband code, part 1
  IB/ipath: infiniband RC protocol support
  IB/ipath: infiniband UC and UD protocol support
  IB/ipath: infiniband header files
  IB/ipath: layering interfaces used by higher-level driver code
  IB/ipath: support for userspace apps using core driver
  IB/ipath: sysfs and ipathfs support for core driver
  IB/ipath: misc driver support code
  IB/ipath: chip initialisation code, and diag support
  IB/ipath: support for PCI Express devices
  IB/ipath: support for HyperTransport devices
  IB/ipath: core driver header files
  IB/ipath: core device driver

MAINTAINERS

@@ -1451,6 +1451,12 @@ P:	Juanjo Ciarlante
 M:	jjciarla@raiz.uncu.edu.ar
 S:	Maintained
 
+IPATH DRIVER:
+P:	Bryan O'Sullivan
+M:	support@pathscale.com
+L:	openib-general@openib.org
+S:	Supported
+
 IPX NETWORK LAYER
 P:	Arnaldo Carvalho de Melo
 M:	acme@conectiva.com.br

drivers/Makefile

@@ -69,6 +69,7 @@ obj-$(CONFIG_CPU_FREQ)		+= cpufreq/
 obj-$(CONFIG_MMC)		+= mmc/
 obj-$(CONFIG_NEW_LEDS)		+= leds/
 obj-$(CONFIG_INFINIBAND)	+= infiniband/
+obj-$(CONFIG_IPATH_CORE)	+= infiniband/
 obj-$(CONFIG_SGI_SN)		+= sn/
 obj-y				+= firmware/
 obj-$(CONFIG_CRYPTO)		+= crypto/

drivers/infiniband/Kconfig

@@ -30,6 +30,7 @@ config INFINIBAND_USER_ACCESS
 	  <http://www.openib.org>.
 
 source "drivers/infiniband/hw/mthca/Kconfig"
+source "drivers/infiniband/hw/ipath/Kconfig"
 
 source "drivers/infiniband/ulp/ipoib/Kconfig"
 

drivers/infiniband/Makefile

 obj-$(CONFIG_INFINIBAND)		+= core/
 obj-$(CONFIG_INFINIBAND_MTHCA)		+= hw/mthca/
+obj-$(CONFIG_IPATH_CORE)		+= hw/ipath/
 obj-$(CONFIG_INFINIBAND_IPOIB)		+= ulp/ipoib/
 obj-$(CONFIG_INFINIBAND_SRP)		+= ulp/srp/

drivers/infiniband/hw/ipath/Kconfig

+config IPATH_CORE
+	tristate "PathScale InfiniPath Driver"
+	depends on 64BIT && PCI_MSI && NET
+	---help---
+	This is a low-level driver for PathScale InfiniPath host channel
+	adapters (HCAs) based on the HT-400 and PE-800 chips.
+
+config INFINIBAND_IPATH
+	tristate "PathScale InfiniPath Verbs Driver"
+	depends on IPATH_CORE && INFINIBAND
+	---help---
+	This is a driver that provides InfiniBand verbs support for
+	PathScale InfiniPath host channel adapters (HCAs).  This
+	allows these devices to be used with both kernel upper level
+	protocols such as IP-over-InfiniBand as well as with userspace
+	applications (in conjunction with InfiniBand userspace access).

drivers/infiniband/hw/ipath/Makefile

+EXTRA_CFLAGS += -DIPATH_IDSTR='"PathScale kernel.org driver"' \
+	-DIPATH_KERN_TYPE=0
+
+obj-$(CONFIG_IPATH_CORE) += ipath_core.o
+obj-$(CONFIG_INFINIBAND_IPATH) += ib_ipath.o
+
+ipath_core-y := \
+	ipath_diag.o \
+	ipath_driver.o \
+	ipath_eeprom.o \
+	ipath_file_ops.o \
+	ipath_fs.o \
+	ipath_ht400.o \
+	ipath_init_chip.o \
+	ipath_intr.o \
+	ipath_layer.o \
+	ipath_pe800.o \
+	ipath_stats.o \
+	ipath_sysfs.o \
+	ipath_user_pages.o
+
+ipath_core-$(CONFIG_X86_64) += ipath_wc_x86_64.o
+
+ib_ipath-y := \
+	ipath_cq.o \
+	ipath_keys.o \
+	ipath_mad.o \
+	ipath_mr.o \
+	ipath_qp.o \
+	ipath_rc.o \
+	ipath_ruc.o \
+	ipath_srq.o \
+	ipath_uc.o \
+	ipath_ud.o \
+	ipath_verbs.o \
+	ipath_verbs_mcast.o

drivers/infiniband/hw/ipath/ipath_cq.c

+/*
+ * 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"
+
+/**
+ * ipath_cq_enter - add a new entry to the completion queue
+ * @cq: completion queue
+ * @entry: work completion entry to add
+ * @sig: true if @entry is a solicitated entry
+ *
+ * This may be called with one of the qp->s_lock or qp->r_rq.lock held.
+ */
+void ipath_cq_enter(struct ipath_cq *cq, struct ib_wc *entry, int solicited)
+{
+	unsigned long flags;
+	u32 next;
+
+	spin_lock_irqsave(&cq->lock, flags);
+
+	if (cq->head == cq->ibcq.cqe)
+		next = 0;
+	else
+		next = cq->head + 1;
+	if (unlikely(next == cq->tail)) {
+		spin_unlock_irqrestore(&cq->lock, flags);
+		if (cq->ibcq.event_handler) {
+			struct ib_event ev;
+
+			ev.device = cq->ibcq.device;
+			ev.element.cq = &cq->ibcq;
+			ev.event = IB_EVENT_CQ_ERR;
+			cq->ibcq.event_handler(&ev, cq->ibcq.cq_context);
+		}
+		return;
+	}
+	cq->queue[cq->head] = *entry;
+	cq->head = next;
+
+	if (cq->notify == IB_CQ_NEXT_COMP ||
+	    (cq->notify == IB_CQ_SOLICITED && solicited)) {
+		cq->notify = IB_CQ_NONE;
+		cq->triggered++;
+		/*
+		 * This will cause send_complete() to be called in
+		 * another thread.
+		 */
+		tasklet_hi_schedule(&cq->comptask);
+	}
+
+	spin_unlock_irqrestore(&cq->lock, flags);
+
+	if (entry->status != IB_WC_SUCCESS)
+		to_idev(cq->ibcq.device)->n_wqe_errs++;
+}
+
+/**
+ * ipath_poll_cq - poll for work completion entries
+ * @ibcq: the completion queue to poll
+ * @num_entries: the maximum number of entries to return
+ * @entry: pointer to array where work completions are placed
+ *
+ * Returns the number of completion entries polled.
+ *
+ * This may be called from interrupt context.  Also called by ib_poll_cq()
+ * in the generic verbs code.
+ */
+int ipath_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *entry)
+{
+	struct ipath_cq *cq = to_icq(ibcq);
+	unsigned long flags;
+	int npolled;
+
+	spin_lock_irqsave(&cq->lock, flags);
+
+	for (npolled = 0; npolled < num_entries; ++npolled, ++entry) {
+		if (cq->tail == cq->head)
+			break;
+		*entry = cq->queue[cq->tail];
+		if (cq->tail == cq->ibcq.cqe)
+			cq->tail = 0;
+		else
+			cq->tail++;
+	}
+
+	spin_unlock_irqrestore(&cq->lock, flags);
+
+	return npolled;
+}
+
+static void send_complete(unsigned long data)
+{
+	struct ipath_cq *cq = (struct ipath_cq *)data;
+
+	/*
+	 * The completion handler will most likely rearm the notification
+	 * and poll for all pending entries.  If a new completion entry
+	 * is added while we are in this routine, tasklet_hi_schedule()
+	 * won't call us again until we return so we check triggered to
+	 * see if we need to call the handler again.
+	 */
+	for (;;) {
+		u8 triggered = cq->triggered;
+
+		cq->ibcq.comp_handler(&cq->ibcq, cq->ibcq.cq_context);
+
+		if (cq->triggered == triggered)
+			return;
+	}
+}
+
+/**
+ * ipath_create_cq - create a completion queue
+ * @ibdev: the device this completion queue is attached to
+ * @entries: the minimum size of the completion queue
+ * @context: unused by the InfiniPath driver
+ * @udata: unused by the InfiniPath driver
+ *
+ * Returns a pointer to the completion queue or negative errno values
+ * for failure.
+ *
+ * Called by ib_create_cq() in the generic verbs code.
+ */
+struct ib_cq *ipath_create_cq(struct ib_device *ibdev, int entries,
+			      struct ib_ucontext *context,
+			      struct ib_udata *udata)
+{
+	struct ipath_cq *cq;
+	struct ib_wc *wc;
+	struct ib_cq *ret;
+
+	/*
+	 * Need to use vmalloc() if we want to support large #s of
+	 * entries.
+	 */
+	cq = kmalloc(sizeof(*cq), GFP_KERNEL);
+	if (!cq) {
+		ret = ERR_PTR(-ENOMEM);
+		goto bail;
+	}
+
+	/*
+	 * Need to use vmalloc() if we want to support large #s of entries.
+	 */
+	wc = vmalloc(sizeof(*wc) * (entries + 1));
+	if (!wc) {
+		kfree(cq);
+		ret = ERR_PTR(-ENOMEM);
+		goto bail;
+	}
+	/*
+	 * ib_create_cq() will initialize cq->ibcq except for cq->ibcq.cqe.
+	 * The number of entries should be >= the number requested or return
+	 * an error.
+	 */
+	cq->ibcq.cqe = entries;
+	cq->notify = IB_CQ_NONE;
+	cq->triggered = 0;
+	spin_lock_init(&cq->lock);
+	tasklet_init(&cq->comptask, send_complete, (unsigned long)cq);
+	cq->head = 0;
+	cq->tail = 0;
+	cq->queue = wc;
+
+	ret = &cq->ibcq;
+
+bail:
+	return ret;
+}
+
+/**
+ * ipath_destroy_cq - destroy a completion queue
+ * @ibcq: the completion queue to destroy.
+ *
+ * Returns 0 for success.
+ *
+ * Called by ib_destroy_cq() in the generic verbs code.
+ */
+int ipath_destroy_cq(struct ib_cq *ibcq)
+{
+	struct ipath_cq *cq = to_icq(ibcq);
+
+	tasklet_kill(&cq->comptask);
+	vfree(cq->queue);
+	kfree(cq);
+
+	return 0;
+}
+
+/**
+ * ipath_req_notify_cq - change the notification type for a completion queue
+ * @ibcq: the completion queue
+ * @notify: the type of notification to request
+ *
+ * Returns 0 for success.
+ *
+ * This may be called from interrupt context.  Also called by
+ * ib_req_notify_cq() in the generic verbs code.
+ */
+int ipath_req_notify_cq(struct ib_cq *ibcq, enum ib_cq_notify notify)
+{
+	struct ipath_cq *cq = to_icq(ibcq);
+	unsigned long flags;
+
+	spin_lock_irqsave(&cq->lock, flags);
+	/*
+	 * Don't change IB_CQ_NEXT_COMP to IB_CQ_SOLICITED but allow
+	 * any other transitions.
+	 */
+	if (cq->notify != IB_CQ_NEXT_COMP)
+		cq->notify = notify;
+	spin_unlock_irqrestore(&cq->lock, flags);
+	return 0;
+}
+
+int ipath_resize_cq(struct ib_cq *ibcq, int cqe, struct ib_udata *udata)
+{
+	struct ipath_cq *cq = to_icq(ibcq);
+	struct ib_wc *wc, *old_wc;
+	u32 n;
+	int ret;
+
+	/*
+	 * Need to use vmalloc() if we want to support large #s of entries.
+	 */
+	wc = vmalloc(sizeof(*wc) * (cqe + 1));
+	if (!wc) {
+		ret = -ENOMEM;
+		goto bail;
+	}
+
+	spin_lock_irq(&cq->lock);
+	if (cq->head < cq->tail)
+		n = cq->ibcq.cqe + 1 + cq->head - cq->tail;
+	else
+		n = cq->head - cq->tail;
+	if (unlikely((u32)cqe < n)) {
+		spin_unlock_irq(&cq->lock);
+		vfree(wc);
+		ret = -EOVERFLOW;
+		goto bail;
+	}
+	for (n = 0; cq->tail != cq->head; n++) {
+		wc[n] = cq->queue[cq->tail];
+		if (cq->tail == cq->ibcq.cqe)
+			cq->tail = 0;
+		else
+			cq->tail++;
+	}
+	cq->ibcq.cqe = cqe;
+	cq->head = n;
+	cq->tail = 0;
+	old_wc = cq->queue;
+	cq->queue = wc;
+	spin_unlock_irq(&cq->lock);
+
+	vfree(old_wc);
+
+	ret = 0;
+
+bail:
+	return ret;
+}

drivers/infiniband/hw/ipath/ipath_debug.h

+/*
+ * Copyright (c) 2003, 2004, 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.
+ */
+
+#ifndef _IPATH_DEBUG_H
+#define _IPATH_DEBUG_H
+
+#ifndef _IPATH_DEBUGGING	/* debugging enabled or not */
+#define _IPATH_DEBUGGING 1
+#endif
+
+#if _IPATH_DEBUGGING
+
+/*
+ * Mask values for debugging.  The scheme allows us to compile out any
+ * of the debug tracing stuff, and if compiled in, to enable or disable
+ * dynamically.  This can be set at modprobe time also:
+ *      modprobe infinipath.ko infinipath_debug=7
+ */
+
+#define __IPATH_INFO        0x1	/* generic low verbosity stuff */
+#define __IPATH_DBG         0x2	/* generic debug */
+#define __IPATH_TRSAMPLE    0x8	/* generate trace buffer sample entries */
+/* leave some low verbosity spots open */
+#define __IPATH_VERBDBG     0x40	/* very verbose debug */
+#define __IPATH_PKTDBG      0x80	/* print packet data */
+/* print process startup (init)/exit messages */
+#define __IPATH_PROCDBG     0x100
+/* print mmap/nopage stuff, not using VDBG any more */
+#define __IPATH_MMDBG       0x200
+#define __IPATH_USER_SEND   0x1000	/* use user mode send */
+#define __IPATH_KERNEL_SEND 0x2000	/* use kernel mode send */
+#define __IPATH_EPKTDBG     0x4000	/* print ethernet packet data */
+#define __IPATH_SMADBG      0x8000	/* sma packet debug */
+#define __IPATH_IPATHDBG    0x10000	/* Ethernet (IPATH) general debug on */
+#define __IPATH_IPATHWARN   0x20000	/* Ethernet (IPATH) warnings on */
+#define __IPATH_IPATHERR    0x40000	/* Ethernet (IPATH) errors on */
+#define __IPATH_IPATHPD     0x80000	/* Ethernet (IPATH) packet dump on */
+#define __IPATH_IPATHTABLE  0x100000	/* Ethernet (IPATH) table dump on */
+
+#else				/* _IPATH_DEBUGGING */
+
+/*
+ * define all of these even with debugging off, for the few places that do
+ * if(infinipath_debug & _IPATH_xyzzy), but in a way that will make the
+ * compiler eliminate the code
+ */
+
+#define __IPATH_INFO      0x0	/* generic low verbosity stuff */
+#define __IPATH_DBG       0x0	/* generic debug */
+#define __IPATH_TRSAMPLE  0x0	/* generate trace buffer sample entries */
+#define __IPATH_VERBDBG   0x0	/* very verbose debug */
+#define __IPATH_PKTDBG    0x0	/* print packet data */
+#define __IPATH_PROCDBG   0x0	/* print process startup (init)/exit messages */
+/* print mmap/nopage stuff, not using VDBG any more */
+#define __IPATH_MMDBG     0x0
+#define __IPATH_EPKTDBG   0x0	/* print ethernet packet data */
+#define __IPATH_SMADBG    0x0   /* print process startup (init)/exit messages */#define __IPATH_IPATHDBG  0x0	/* Ethernet (IPATH) table dump on */
+#define __IPATH_IPATHWARN 0x0	/* Ethernet (IPATH) warnings on   */
+#define __IPATH_IPATHERR  0x0	/* Ethernet (IPATH) errors on   */
+#define __IPATH_IPATHPD   0x0	/* Ethernet (IPATH) packet dump on   */
+#define __IPATH_IPATHTABLE 0x0	/* Ethernet (IPATH) packet dump on   */
+
+#endif				/* _IPATH_DEBUGGING */
+
+#define __IPATH_VERBOSEDBG __IPATH_VERBDBG
+
+#endif				/* _IPATH_DEBUG_H */

drivers/infiniband/hw/ipath/ipath_keys.c

+/*
+ * 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 <asm/io.h>
+
+#include "ipath_verbs.h"
+
+/**
+ * ipath_alloc_lkey - allocate an lkey
+ * @rkt: lkey table in which to allocate the lkey
+ * @mr: memory region that this lkey protects
+ *
+ * Returns 1 if successful, otherwise returns 0.
+ */
+
+int ipath_alloc_lkey(struct ipath_lkey_table *rkt, struct ipath_mregion *mr)
+{
+	unsigned long flags;
+	u32 r;
+	u32 n;
+	int ret;
+
+	spin_lock_irqsave(&rkt->lock, flags);
+
+	/* Find the next available LKEY */
+	r = n = rkt->next;
+	for (;;) {
+		if (rkt->table[r] == NULL)
+			break;
+		r = (r + 1) & (rkt->max - 1);
+		if (r == n) {
+			spin_unlock_irqrestore(&rkt->lock, flags);
+			_VERBS_INFO("LKEY table full\n");
+			ret = 0;
+			goto bail;
+		}
+	}
+	rkt->next = (r + 1) & (rkt->max - 1);
+	/*
+	 * Make sure lkey is never zero which is reserved to indicate an
+	 * unrestricted LKEY.
+	 */
+	rkt->gen++;
+	mr->lkey = (r << (32 - ib_ipath_lkey_table_size)) |
+		((((1 << (24 - ib_ipath_lkey_table_size)) - 1) & rkt->gen)
+		 << 8);
+	if (mr->lkey == 0) {
+		mr->lkey |= 1 << 8;
+		rkt->gen++;
+	}
+	rkt->table[r] = mr;
+	spin_unlock_irqrestore(&rkt->lock, flags);
+
+	ret = 1;
+
+bail:
+	return ret;
+}
+
+/**
+ * ipath_free_lkey - free an lkey
+ * @rkt: table from which to free the lkey
+ * @lkey: lkey id to free
+ */
+void ipath_free_lkey(struct ipath_lkey_table *rkt, u32 lkey)
+{
+	unsigned long flags;
+	u32 r;
+
+	if (lkey == 0)
+		return;
+	r = lkey >> (32 - ib_ipath_lkey_table_size);
+	spin_lock_irqsave(&rkt->lock, flags);
+	rkt->table[r] = NULL;
+	spin_unlock_irqrestore(&rkt->lock, flags);
+}
+
+/**
+ * ipath_lkey_ok - check IB SGE for validity and initialize
+ * @rkt: table containing lkey to check SGE against
+ * @isge: outgoing internal SGE
+ * @sge: SGE to check
+ * @acc: access flags
+ *
+ * Return 1 if valid and successful, otherwise returns 0.
+ *
+ * Check the IB SGE for validity and initialize our internal version
+ * of it.
+ */
+int ipath_lkey_ok(struct ipath_lkey_table *rkt, struct ipath_sge *isge,
+		  struct ib_sge *sge, int acc)
+{
+	struct ipath_mregion *mr;
+	size_t off;
+	int ret;
+
+	/*
+	 * We use LKEY == zero to mean a physical kmalloc() address.
+	 * This is a bit of a hack since we rely on dma_map_single()
+	 * being reversible by calling bus_to_virt().
+	 */
+	if (sge->lkey == 0) {
+		isge->mr = NULL;
+		isge->vaddr = bus_to_virt(sge->addr);
+		isge->length = sge->length;
+		isge->sge_length = sge->length;
+		ret = 1;
+		goto bail;
+	}
+	spin_lock(&rkt->lock);
+	mr = rkt->table[(sge->lkey >> (32 - ib_ipath_lkey_table_size))];
+	spin_unlock(&rkt->lock);
+	if (unlikely(mr == NULL || mr->lkey != sge->lkey)) {
+		ret = 0;
+		goto bail;
+	}
+
+	off = sge->addr - mr->user_base;
+	if (unlikely(sge->addr < mr->user_base ||
+		     off + sge->length > mr->length ||
+		     (mr->access_flags & acc) != acc)) {
+		ret = 0;
+		goto bail;
+	}
+
+	off += mr->offset;
+	isge->mr = mr;
+	isge->m = 0;
+	isge->n = 0;
+	while (off >= mr->map[isge->m]->segs[isge->n].length) {
+		off -= mr->map[isge->m]->segs[isge->n].length;
+		isge->n++;
+		if (isge->n >= IPATH_SEGSZ) {
+			isge->m++;
+			isge->n = 0;
+		}
+	}
+	isge->vaddr = mr->map[isge->m]->segs[isge->n].vaddr + off;
+	isge->length = mr->map[isge->m]->segs[isge->n].length - off;
+	isge->sge_length = sge->length;
+
+	ret = 1;
+
+bail:
+	return ret;
+}
+
+/**
+ * ipath_rkey_ok - check the IB virtual address, length, and RKEY
+ * @dev: infiniband device
+ * @ss: SGE state
+ * @len: length of data
+ * @vaddr: virtual address to place data
+ * @rkey: rkey to check
+ * @acc: access flags
+ *
+ * Return 1 if successful, otherwise 0.
+ *
+ * The QP r_rq.lock should be held.
+ */
+int ipath_rkey_ok(struct ipath_ibdev *dev, struct ipath_sge_state *ss,
+		  u32 len, u64 vaddr, u32 rkey, int acc)
+{
+	struct ipath_lkey_table *rkt = &dev->lk_table;
+	struct ipath_sge *sge = &ss->sge;
+	struct ipath_mregion *mr;
+	size_t off;
+	int ret;
+
+	spin_lock(&rkt->lock);
+	mr = rkt->table[(rkey >> (32 - ib_ipath_lkey_table_size))];
+	spin_unlock(&rkt->lock);
+	if (unlikely(mr == NULL || mr->lkey != rkey)) {
+		ret = 0;
+		goto bail;
+	}
+
+	off = vaddr - mr->iova;
+	if (unlikely(vaddr < mr->iova || off + len > mr->length ||
+		     (mr->access_flags & acc) == 0)) {
+		ret = 0;
+		goto bail;
+	}
+
+	off += mr->offset;
+	sge->mr = mr;
+	sge->m = 0;
+	sge->n = 0;
+	while (off >= mr->map[sge->m]->segs[sge->n].length) {
+		off -= mr->map[sge->m]->segs[sge->n].length;
+		sge->n++;
+		if (sge->n >= IPATH_SEGSZ) {
+			sge->m++;
+			sge->n = 0;
+		}
+	}
+	sge->vaddr = mr->map[sge->m]->segs[sge->n].vaddr + off;
+	sge->length = mr->map[sge->m]->segs[sge->n].length - off;
+	sge->sge_length = len;
+	ss->sg_list = NULL;
+	ss->num_sge = 1;
+
+	ret = 1;
+
+bail:
+	return ret;
+}