1 files changed, 338 insertions, 441 deletions

diff --git a/drivers/infiniband/hw/hfi1/user_exp_rcv.c b/drivers/infiniband/hw/hfi1/user_exp_rcv.c
index a8f0aa4722f6..62b4f16dab27 100644
--- a/drivers/infiniband/hw/hfi1/user_exp_rcv.c
+++ b/drivers/infiniband/hw/hfi1/user_exp_rcv.c
@@ -1,212 +1,59 @@
+// SPDX-License-Identifier: GPL-2.0 OR BSD-3-Clause
 /*
- * Copyright(c) 2015-2017 Intel Corporation.
- *
- * This file is provided under a dual BSD/GPLv2 license.  When using or
- * redistributing this file, you may do so under either license.
- *
- * GPL LICENSE SUMMARY
- *
- * This program 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.
- *
- * This program 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.
- *
- * BSD LICENSE
- *
- * 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.
- *  - Neither the name of Intel Corporation nor the names of its
- *    contributors may be used to endorse or promote products derived
- *    from this software without specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
- * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
- * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
- * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
- * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
+ * Copyright(c) 2020 Cornelis Networks, Inc.
+ * Copyright(c) 2015-2018 Intel Corporation.
  */
 #include <asm/page.h>
 #include <linux/string.h>
 
+#include "mmu_rb.h"
 #include "user_exp_rcv.h"
 #include "trace.h"
-#include "mmu_rb.h"
-
-struct tid_group {
-	struct list_head list;
-	u32 base;
-	u8 size;
-	u8 used;
-	u8 map;
-};
-
-struct tid_rb_node {
-	struct mmu_rb_node mmu;
-	unsigned long phys;
-	struct tid_group *grp;
-	u32 rcventry;
-	dma_addr_t dma_addr;
-	bool freed;
-	unsigned npages;
-	struct page *pages[0];
-};
-
-struct tid_pageset {
-	u16 idx;
-	u16 count;
-};
-
-#define EXP_TID_SET_EMPTY(set) (set.count == 0 && list_empty(&set.list))
-
-#define num_user_pages(vaddr, len)				       \
-	(1 + (((((unsigned long)(vaddr) +			       \
-		 (unsigned long)(len) - 1) & PAGE_MASK) -	       \
-	       ((unsigned long)vaddr & PAGE_MASK)) >> PAGE_SHIFT))
 
 static void unlock_exp_tids(struct hfi1_ctxtdata *uctxt,
 			    struct exp_tid_set *set,
 			    struct hfi1_filedata *fd);
-static u32 find_phys_blocks(struct page **pages, unsigned npages,
-			    struct tid_pageset *list);
-static int set_rcvarray_entry(struct hfi1_filedata *fd, unsigned long vaddr,
+static u32 find_phys_blocks(struct tid_user_buf *tidbuf, unsigned int npages);
+static int set_rcvarray_entry(struct hfi1_filedata *fd,
+			      struct tid_user_buf *tbuf,
 			      u32 rcventry, struct tid_group *grp,
-			      struct page **pages, unsigned npages);
-static int tid_rb_insert(void *arg, struct mmu_rb_node *node);
+			      u16 pageidx, unsigned int npages);
 static void cacheless_tid_rb_remove(struct hfi1_filedata *fdata,
 				    struct tid_rb_node *tnode);
-static void tid_rb_remove(void *arg, struct mmu_rb_node *node);
-static int tid_rb_invalidate(void *arg, struct mmu_rb_node *mnode);
-static int program_rcvarray(struct hfi1_filedata *fd, unsigned long vaddr,
-			    struct tid_group *grp, struct tid_pageset *sets,
-			    unsigned start, u16 count, struct page **pages,
-			    u32 *tidlist, unsigned *tididx, unsigned *pmapped);
-static int unprogram_rcvarray(struct hfi1_filedata *fd, u32 tidinfo,
-			      struct tid_group **grp);
+static bool tid_rb_invalidate(struct mmu_interval_notifier *mni,
+			      const struct mmu_notifier_range *range,
+			      unsigned long cur_seq);
+static bool tid_cover_invalidate(struct mmu_interval_notifier *mni,
+			         const struct mmu_notifier_range *range,
+			         unsigned long cur_seq);
+static int program_rcvarray(struct hfi1_filedata *fd, struct tid_user_buf *,
+			    struct tid_group *grp, u16 count,
+			    u32 *tidlist, unsigned int *tididx,
+			    unsigned int *pmapped);
+static int unprogram_rcvarray(struct hfi1_filedata *fd, u32 tidinfo);
+static void __clear_tid_node(struct hfi1_filedata *fd,
+			     struct tid_rb_node *node);
 static void clear_tid_node(struct hfi1_filedata *fd, struct tid_rb_node *node);
 
-static struct mmu_rb_ops tid_rb_ops = {
-	.insert = tid_rb_insert,
-	.remove = tid_rb_remove,
-	.invalidate = tid_rb_invalidate
+static const struct mmu_interval_notifier_ops tid_mn_ops = {
+	.invalidate = tid_rb_invalidate,
+};
+static const struct mmu_interval_notifier_ops tid_cover_ops = {
+	.invalidate = tid_cover_invalidate,
 };
-
-static inline u32 rcventry2tidinfo(u32 rcventry)
-{
-	u32 pair = rcventry & ~0x1;
-
-	return EXP_TID_SET(IDX, pair >> 1) |
-		EXP_TID_SET(CTRL, 1 << (rcventry - pair));
-}
-
-static inline void exp_tid_group_init(struct exp_tid_set *set)
-{
-	INIT_LIST_HEAD(&set->list);
-	set->count = 0;
-}
-
-static inline void tid_group_remove(struct tid_group *grp,
-				    struct exp_tid_set *set)
-{
-	list_del_init(&grp->list);
-	set->count--;
-}
-
-static inline void tid_group_add_tail(struct tid_group *grp,
-				      struct exp_tid_set *set)
-{
-	list_add_tail(&grp->list, &set->list);
-	set->count++;
-}
-
-static inline struct tid_group *tid_group_pop(struct exp_tid_set *set)
-{
-	struct tid_group *grp =
-		list_first_entry(&set->list, struct tid_group, list);
-	list_del_init(&grp->list);
-	set->count--;
-	return grp;
-}
-
-static inline void tid_group_move(struct tid_group *group,
-				  struct exp_tid_set *s1,
-				  struct exp_tid_set *s2)
-{
-	tid_group_remove(group, s1);
-	tid_group_add_tail(group, s2);
-}
-
-int hfi1_user_exp_rcv_grp_init(struct hfi1_filedata *fd)
-{
-	struct hfi1_ctxtdata *uctxt = fd->uctxt;
-	struct hfi1_devdata *dd = fd->dd;
-	u32 tidbase;
-	u32 i;
-	struct tid_group *grp, *gptr;
-
-	exp_tid_group_init(&uctxt->tid_group_list);
-	exp_tid_group_init(&uctxt->tid_used_list);
-	exp_tid_group_init(&uctxt->tid_full_list);
-
-	tidbase = uctxt->expected_base;
-	for (i = 0; i < uctxt->expected_count /
-		     dd->rcv_entries.group_size; i++) {
-		grp = kzalloc(sizeof(*grp), GFP_KERNEL);
-		if (!grp)
-			goto grp_failed;
-
-		grp->size = dd->rcv_entries.group_size;
-		grp->base = tidbase;
-		tid_group_add_tail(grp, &uctxt->tid_group_list);
-		tidbase += dd->rcv_entries.group_size;
-	}
-
-	return 0;
-
-grp_failed:
-	list_for_each_entry_safe(grp, gptr, &uctxt->tid_group_list.list,
-				 list) {
-		list_del_init(&grp->list);
-		kfree(grp);
-	}
-
-	return -ENOMEM;
-}
 
 /*
  * Initialize context and file private data needed for Expected
  * receive caching. This needs to be done after the context has
  * been configured with the eager/expected RcvEntry counts.
  */
-int hfi1_user_exp_rcv_init(struct hfi1_filedata *fd)
+int hfi1_user_exp_rcv_init(struct hfi1_filedata *fd,
+			   struct hfi1_ctxtdata *uctxt)
 {
-	struct hfi1_ctxtdata *uctxt = fd->uctxt;
-	struct hfi1_devdata *dd = uctxt->dd;
 	int ret = 0;
 
-	spin_lock_init(&fd->tid_lock);
-	spin_lock_init(&fd->invalid_lock);
-
 	fd->entry_to_rb = kcalloc(uctxt->expected_count,
-				  sizeof(struct rb_node *),
+				  sizeof(*fd->entry_to_rb),
 				  GFP_KERNEL);
 	if (!fd->entry_to_rb)
 		return -ENOMEM;
@@ -221,20 +68,7 @@ int hfi1_user_exp_rcv_init(struct hfi1_filedata *fd)
 			fd->entry_to_rb = NULL;
 			return -ENOMEM;
 		}
-
-		/*
-		 * Register MMU notifier callbacks. If the registration
-		 * fails, continue without TID caching for this context.
-		 */
-		ret = hfi1_mmu_rb_register(fd, fd->mm, &tid_rb_ops,
-					   dd->pport->hfi1_wq,
-					   &fd->handler);
-		if (ret) {
-			dd_dev_info(dd,
-				    "Failed MMU notifier registration %d\n",
-				    ret);
-			ret = 0;
-		}
+		fd->use_mn = true;
 	}
 
 	/*
@@ -251,7 +85,7 @@ int hfi1_user_exp_rcv_init(struct hfi1_filedata *fd)
 	 * init.
 	 */
 	spin_lock(&fd->tid_lock);
-	if (uctxt->subctxt_cnt && fd->handler) {
+	if (uctxt->subctxt_cnt && fd->use_mn) {
 		u16 remainder;
 
 		fd->tid_limit = uctxt->expected_count / uctxt->subctxt_cnt;
@@ -266,34 +100,16 @@ int hfi1_user_exp_rcv_init(struct hfi1_filedata *fd)
 	return ret;
 }
 
-void hfi1_user_exp_rcv_grp_free(struct hfi1_ctxtdata *uctxt)
-{
-	struct tid_group *grp, *gptr;
-
-	list_for_each_entry_safe(grp, gptr, &uctxt->tid_group_list.list,
-				 list) {
-		list_del_init(&grp->list);
-		kfree(grp);
-	}
-	hfi1_clear_tids(uctxt);
-}
-
 void hfi1_user_exp_rcv_free(struct hfi1_filedata *fd)
 {
 	struct hfi1_ctxtdata *uctxt = fd->uctxt;
 
-	/*
-	 * The notifier would have been removed when the process'es mm
-	 * was freed.
-	 */
-	if (fd->handler) {
-		hfi1_mmu_rb_unregister(fd->handler);
-	} else {
-		if (!EXP_TID_SET_EMPTY(uctxt->tid_full_list))
-			unlock_exp_tids(uctxt, &uctxt->tid_full_list, fd);
-		if (!EXP_TID_SET_EMPTY(uctxt->tid_used_list))
-			unlock_exp_tids(uctxt, &uctxt->tid_used_list, fd);
-	}
+	mutex_lock(&uctxt->exp_mutex);
+	if (!EXP_TID_SET_EMPTY(uctxt->tid_full_list))
+		unlock_exp_tids(uctxt, &uctxt->tid_full_list, fd);
+	if (!EXP_TID_SET_EMPTY(uctxt->tid_used_list))
+		unlock_exp_tids(uctxt, &uctxt->tid_used_list, fd);
+	mutex_unlock(&uctxt->exp_mutex);
 
 	kfree(fd->invalid_tids);
 	fd->invalid_tids = NULL;
@@ -303,20 +119,79 @@ void hfi1_user_exp_rcv_free(struct hfi1_filedata *fd)
 }
 
 /*
- * Write an "empty" RcvArray entry.
- * This function exists so the TID registaration code can use it
- * to write to unused/unneeded entries and still take advantage
- * of the WC performance improvements. The HFI will ignore this
- * write to the RcvArray entry.
+ * Release pinned receive buffer pages.
+ *
+ * @mapped: true if the pages have been DMA mapped. false otherwise.
+ * @idx: Index of the first page to unpin.
+ * @npages: No of pages to unpin.
+ *
+ * If the pages have been DMA mapped (indicated by mapped parameter), their
+ * info will be passed via a struct tid_rb_node. If they haven't been mapped,
+ * their info will be passed via a struct tid_user_buf.
  */
-static inline void rcv_array_wc_fill(struct hfi1_devdata *dd, u32 index)
+static void unpin_rcv_pages(struct hfi1_filedata *fd,
+			    struct tid_user_buf *tidbuf,
+			    struct tid_rb_node *node,
+			    unsigned int idx,
+			    unsigned int npages,
+			    bool mapped)
 {
+	struct page **pages;
+	struct hfi1_devdata *dd = fd->uctxt->dd;
+	struct mm_struct *mm;
+
+	if (mapped) {
+		dma_unmap_single(&dd->pcidev->dev, node->dma_addr,
+				 node->npages * PAGE_SIZE, DMA_FROM_DEVICE);
+		pages = &node->pages[idx];
+		mm = mm_from_tid_node(node);
+	} else {
+		pages = &tidbuf->pages[idx];
+		mm = current->mm;
+	}
+	hfi1_release_user_pages(mm, pages, npages, mapped);
+	fd->tid_n_pinned -= npages;
+}
+
+/*
+ * Pin receive buffer pages.
+ */
+static int pin_rcv_pages(struct hfi1_filedata *fd, struct tid_user_buf *tidbuf)
+{
+	int pinned;
+	unsigned int npages = tidbuf->npages;
+	unsigned long vaddr = tidbuf->vaddr;
+	struct page **pages = NULL;
+	struct hfi1_devdata *dd = fd->uctxt->dd;
+
+	if (npages > fd->uctxt->expected_count) {
+		dd_dev_err(dd, "Expected buffer too big\n");
+		return -EINVAL;
+	}
+
+	/* Allocate the array of struct page pointers needed for pinning */
+	pages = kcalloc(npages, sizeof(*pages), GFP_KERNEL);
+	if (!pages)
+		return -ENOMEM;
+
 	/*
-	 * Doing the WC fill writes only makes sense if the device is
-	 * present and the RcvArray has been mapped as WC memory.
+	 * Pin all the pages of the user buffer. If we can't pin all the
+	 * pages, accept the amount pinned so far and program only that.
+	 * User space knows how to deal with partially programmed buffers.
 	 */
-	if ((dd->flags & HFI1_PRESENT) && dd->rcvarray_wc)
-		writeq(0, dd->rcvarray_wc + (index * 8));
+	if (!hfi1_can_pin_pages(dd, current->mm, fd->tid_n_pinned, npages)) {
+		kfree(pages);
+		return -ENOMEM;
+	}
+
+	pinned = hfi1_acquire_user_pages(current->mm, vaddr, npages, true, pages);
+	if (pinned <= 0) {
+		kfree(pages);
+		return pinned;
+	}
+	tidbuf->pages = pages;
+	fd->tid_n_pinned += pinned;
+	return pinned;
 }
 
 /*
@@ -374,83 +249,70 @@ int hfi1_user_exp_rcv_setup(struct hfi1_filedata *fd,
 	int ret = 0, need_group = 0, pinned;
 	struct hfi1_ctxtdata *uctxt = fd->uctxt;
 	struct hfi1_devdata *dd = uctxt->dd;
-	unsigned npages, ngroups, pageidx = 0, pageset_count, npagesets,
+	unsigned int ngroups, pageset_count,
 		tididx = 0, mapped, mapped_pages = 0;
-	unsigned long vaddr = tinfo->vaddr;
-	struct page **pages = NULL;
 	u32 *tidlist = NULL;
-	struct tid_pageset *pagesets = NULL;
+	struct tid_user_buf *tidbuf;
+	unsigned long mmu_seq = 0;
 
-	/* Get the number of pages the user buffer spans */
-	npages = num_user_pages(vaddr, tinfo->length);
-	if (!npages)
+	if (!PAGE_ALIGNED(tinfo->vaddr))
 		return -EINVAL;
-
-	if (npages > uctxt->expected_count) {
-		dd_dev_err(dd, "Expected buffer too big\n");
+	if (tinfo->length == 0)
 		return -EINVAL;
-	}
-
-	/* Verify that access is OK for the user buffer */
-	if (!access_ok(VERIFY_WRITE, (void __user *)vaddr,
-		       npages * PAGE_SIZE)) {
-		dd_dev_err(dd, "Fail vaddr %p, %u pages, !access_ok\n",
-			   (void *)vaddr, npages);
-		return -EFAULT;
-	}
 
-	pagesets = kcalloc(uctxt->expected_count, sizeof(*pagesets),
-			   GFP_KERNEL);
-	if (!pagesets)
+	tidbuf = kzalloc(sizeof(*tidbuf), GFP_KERNEL);
+	if (!tidbuf)
 		return -ENOMEM;
 
-	/* Allocate the array of struct page pointers needed for pinning */
-	pages = kcalloc(npages, sizeof(*pages), GFP_KERNEL);
-	if (!pages) {
+	mutex_init(&tidbuf->cover_mutex);
+	tidbuf->vaddr = tinfo->vaddr;
+	tidbuf->length = tinfo->length;
+	tidbuf->npages = num_user_pages(tidbuf->vaddr, tidbuf->length);
+	tidbuf->psets = kcalloc(uctxt->expected_count, sizeof(*tidbuf->psets),
+				GFP_KERNEL);
+	if (!tidbuf->psets) {
 		ret = -ENOMEM;
-		goto bail;
+		goto fail_release_mem;
 	}
 
-	/*
-	 * Pin all the pages of the user buffer. If we can't pin all the
-	 * pages, accept the amount pinned so far and program only that.
-	 * User space knows how to deal with partially programmed buffers.
-	 */
-	if (!hfi1_can_pin_pages(dd, fd->mm, fd->tid_n_pinned, npages)) {
-		ret = -ENOMEM;
-		goto bail;
+	if (fd->use_mn) {
+		ret = mmu_interval_notifier_insert(
+			&tidbuf->notifier, current->mm,
+			tidbuf->vaddr, tidbuf->npages * PAGE_SIZE,
+			&tid_cover_ops);
+		if (ret)
+			goto fail_release_mem;
+		mmu_seq = mmu_interval_read_begin(&tidbuf->notifier);
 	}
 
-	pinned = hfi1_acquire_user_pages(fd->mm, vaddr, npages, true, pages);
+	pinned = pin_rcv_pages(fd, tidbuf);
 	if (pinned <= 0) {
-		ret = pinned;
-		goto bail;
+		ret = (pinned < 0) ? pinned : -ENOSPC;
+		goto fail_unpin;
 	}
-	fd->tid_n_pinned += npages;
 
 	/* Find sets of physically contiguous pages */
-	npagesets = find_phys_blocks(pages, pinned, pagesets);
+	tidbuf->n_psets = find_phys_blocks(tidbuf, pinned);
 
-	/*
-	 * We don't need to access this under a lock since tid_used is per
-	 * process and the same process cannot be in hfi1_user_exp_rcv_clear()
-	 * and hfi1_user_exp_rcv_setup() at the same time.
-	 */
+	/* Reserve the number of expected tids to be used. */
 	spin_lock(&fd->tid_lock);
-	if (fd->tid_used + npagesets > fd->tid_limit)
+	if (fd->tid_used + tidbuf->n_psets > fd->tid_limit)
 		pageset_count = fd->tid_limit - fd->tid_used;
 	else
-		pageset_count = npagesets;
+		pageset_count = tidbuf->n_psets;
+	fd->tid_used += pageset_count;
 	spin_unlock(&fd->tid_lock);
 
-	if (!pageset_count)
-		goto bail;
+	if (!pageset_count) {
+		ret = -ENOSPC;
+		goto fail_unreserve;
+	}
 
 	ngroups = pageset_count / dd->rcv_entries.group_size;
 	tidlist = kcalloc(pageset_count, sizeof(*tidlist), GFP_KERNEL);
 	if (!tidlist) {
 		ret = -ENOMEM;
-		goto nomem;
+		goto fail_unreserve;
 	}
 
 	tididx = 0;
@@ -459,7 +321,7 @@ int hfi1_user_exp_rcv_setup(struct hfi1_filedata *fd,
 	 * From this point on, we are going to be using shared (between master
 	 * and subcontexts) context resources. We need to take the lock.
 	 */
-	mutex_lock(&uctxt->exp_lock);
+	mutex_lock(&uctxt->exp_mutex);
 	/*
 	 * The first step is to program the RcvArray entries which are complete
 	 * groups.
@@ -468,9 +330,9 @@ int hfi1_user_exp_rcv_setup(struct hfi1_filedata *fd,
 		struct tid_group *grp =
 			tid_group_pop(&uctxt->tid_group_list);
 
-		ret = program_rcvarray(fd, vaddr, grp, pagesets,
-				       pageidx, dd->rcv_entries.group_size,
-				       pages, tidlist, &tididx, &mapped);
+		ret = program_rcvarray(fd, tidbuf, grp,
+				       dd->rcv_entries.group_size,
+				       tidlist, &tididx, &mapped);
 		/*
 		 * If there was a failure to program the RcvArray
 		 * entries for the entire group, reset the grp fields
@@ -485,11 +347,10 @@ int hfi1_user_exp_rcv_setup(struct hfi1_filedata *fd,
 
 		tid_group_add_tail(grp, &uctxt->tid_full_list);
 		ngroups--;
-		pageidx += ret;
 		mapped_pages += mapped;
 	}
 
-	while (pageidx < pageset_count) {
+	while (tididx < pageset_count) {
 		struct tid_group *grp, *ptr;
 		/*
 		 * If we don't have any partially used tid groups, check
@@ -511,28 +372,26 @@ int hfi1_user_exp_rcv_setup(struct hfi1_filedata *fd,
 		 */
 		list_for_each_entry_safe(grp, ptr, &uctxt->tid_used_list.list,
 					 list) {
-			unsigned use = min_t(unsigned, pageset_count - pageidx,
+			unsigned use = min_t(unsigned, pageset_count - tididx,
 					     grp->size - grp->used);
 
-			ret = program_rcvarray(fd, vaddr, grp, pagesets,
-					       pageidx, use, pages, tidlist,
+			ret = program_rcvarray(fd, tidbuf, grp,
+					       use, tidlist,
 					       &tididx, &mapped);
 			if (ret < 0) {
 				hfi1_cdbg(TID,
 					  "Failed to program RcvArray entries %d",
 					  ret);
-				ret = -EFAULT;
 				goto unlock;
 			} else if (ret > 0) {
 				if (grp->used == grp->size)
 					tid_group_move(grp,
 						       &uctxt->tid_used_list,
 						       &uctxt->tid_full_list);
-				pageidx += ret;
 				mapped_pages += mapped;
 				need_group = 0;
 				/* Check if we are done so we break out early */
-				if (pageidx >= pageset_count)
+				if (tididx >= pageset_count)
 					break;
 			} else if (WARN_ON(ret == 0)) {
 				/*
@@ -546,46 +405,79 @@ int hfi1_user_exp_rcv_setup(struct hfi1_filedata *fd,
 		}
 	}
 unlock:
-	mutex_unlock(&uctxt->exp_lock);
-nomem:
+	mutex_unlock(&uctxt->exp_mutex);
 	hfi1_cdbg(TID, "total mapped: tidpairs:%u pages:%u (%d)", tididx,
 		  mapped_pages, ret);
-	if (tididx) {
-		spin_lock(&fd->tid_lock);
-		fd->tid_used += tididx;
-		spin_unlock(&fd->tid_lock);
-		tinfo->tidcnt = tididx;
-		tinfo->length = mapped_pages * PAGE_SIZE;
-
-		if (copy_to_user((void __user *)(unsigned long)tinfo->tidlist,
-				 tidlist, sizeof(tidlist[0]) * tididx)) {
-			/*
-			 * On failure to copy to the user level, we need to undo
-			 * everything done so far so we don't leak resources.
-			 */
-			tinfo->tidlist = (unsigned long)&tidlist;
-			hfi1_user_exp_rcv_clear(fd, tinfo);
-			tinfo->tidlist = 0;
-			ret = -EFAULT;
-			goto bail;
+
+	/* fail if nothing was programmed, set error if none provided */
+	if (tididx == 0) {
+		if (ret >= 0)
+			ret = -ENOSPC;
+		goto fail_unreserve;
+	}
+
+	/* adjust reserved tid_used to actual count */
+	spin_lock(&fd->tid_lock);
+	fd->tid_used -= pageset_count - tididx;
+	spin_unlock(&fd->tid_lock);
+
+	/* unpin all pages not covered by a TID */
+	unpin_rcv_pages(fd, tidbuf, NULL, mapped_pages, pinned - mapped_pages,
+			false);
+
+	if (fd->use_mn) {
+		/* check for an invalidate during setup */
+		bool fail = false;
+
+		mutex_lock(&tidbuf->cover_mutex);
+		fail = mmu_interval_read_retry(&tidbuf->notifier, mmu_seq);
+		mutex_unlock(&tidbuf->cover_mutex);
+
+		if (fail) {
+			ret = -EBUSY;
+			goto fail_unprogram;
 		}
 	}
 
-	/*
-	 * If not everything was mapped (due to insufficient RcvArray entries,
-	 * for example), unpin all unmapped pages so we can pin them nex time.
-	 */
-	if (mapped_pages != pinned) {
-		hfi1_release_user_pages(fd->mm, &pages[mapped_pages],
-					pinned - mapped_pages,
-					false);
-		fd->tid_n_pinned -= pinned - mapped_pages;
+	tinfo->tidcnt = tididx;
+	tinfo->length = mapped_pages * PAGE_SIZE;
+
+	if (copy_to_user(u64_to_user_ptr(tinfo->tidlist),
+			 tidlist, sizeof(tidlist[0]) * tididx)) {
+		ret = -EFAULT;
+		goto fail_unprogram;
 	}
-bail:
-	kfree(pagesets);
-	kfree(pages);
+
+	if (fd->use_mn)
+		mmu_interval_notifier_remove(&tidbuf->notifier);
+	kfree(tidbuf->pages);
+	kfree(tidbuf->psets);
+	kfree(tidbuf);
 	kfree(tidlist);
-	return ret > 0 ? 0 : ret;
+	return 0;
+
+fail_unprogram:
+	/* unprogram, unmap, and unpin all allocated TIDs */
+	tinfo->tidlist = (unsigned long)tidlist;
+	hfi1_user_exp_rcv_clear(fd, tinfo);
+	tinfo->tidlist = 0;
+	pinned = 0;		/* nothing left to unpin */
+	pageset_count = 0;	/* nothing left reserved */
+fail_unreserve:
+	spin_lock(&fd->tid_lock);
+	fd->tid_used -= pageset_count;
+	spin_unlock(&fd->tid_lock);
+fail_unpin:
+	if (fd->use_mn)
+		mmu_interval_notifier_remove(&tidbuf->notifier);
+	if (pinned > 0)
+		unpin_rcv_pages(fd, tidbuf, NULL, 0, pinned, false);
+fail_release_mem:
+	kfree(tidbuf->pages);
+	kfree(tidbuf->psets);
+	kfree(tidbuf);
+	kfree(tidlist);
+	return ret;
 }
 
 int hfi1_user_exp_rcv_clear(struct hfi1_filedata *fd,
@@ -599,14 +491,14 @@ int hfi1_user_exp_rcv_clear(struct hfi1_filedata *fd,
 	if (unlikely(tinfo->tidcnt > fd->tid_used))
 		return -EINVAL;
 
-	tidinfo = memdup_user((void __user *)(unsigned long)tinfo->tidlist,
-			      sizeof(tidinfo[0]) * tinfo->tidcnt);
+	tidinfo = memdup_array_user(u64_to_user_ptr(tinfo->tidlist),
+				    tinfo->tidcnt, sizeof(tidinfo[0]));
 	if (IS_ERR(tidinfo))
 		return PTR_ERR(tidinfo);
 
-	mutex_lock(&uctxt->exp_lock);
+	mutex_lock(&uctxt->exp_mutex);
 	for (tididx = 0; tididx < tinfo->tidcnt; tididx++) {
-		ret = unprogram_rcvarray(fd, tidinfo[tididx], NULL);
+		ret = unprogram_rcvarray(fd, tidinfo[tididx]);
 		if (ret) {
 			hfi1_cdbg(TID, "Failed to unprogram rcv array %d",
 				  ret);
@@ -617,7 +509,7 @@ int hfi1_user_exp_rcv_clear(struct hfi1_filedata *fd,
 	fd->tid_used -= tididx;
 	spin_unlock(&fd->tid_lock);
 	tinfo->tidcnt = tididx;
-	mutex_unlock(&uctxt->exp_lock);
+	mutex_unlock(&uctxt->exp_mutex);
 
 	kfree(tidinfo);
 	return ret;
@@ -628,14 +520,10 @@ int hfi1_user_exp_rcv_invalid(struct hfi1_filedata *fd,
 {
 	struct hfi1_ctxtdata *uctxt = fd->uctxt;
 	unsigned long *ev = uctxt->dd->events +
-		(((uctxt->ctxt - uctxt->dd->first_dyn_alloc_ctxt) *
-		  HFI1_MAX_SHARED_CTXTS) + fd->subctxt);
+		(uctxt_offset(uctxt) + fd->subctxt);
 	u32 *array;
 	int ret = 0;
 
-	if (!fd->invalid_tids)
-		return -EINVAL;
-
 	/*
 	 * copy_to_user() can sleep, which will leave the invalid_lock
 	 * locked and cause the MMU notifier to be blocked on the lock
@@ -674,11 +562,12 @@ int hfi1_user_exp_rcv_invalid(struct hfi1_filedata *fd,
 	return ret;
 }
 
-static u32 find_phys_blocks(struct page **pages, unsigned npages,
-			    struct tid_pageset *list)
+static u32 find_phys_blocks(struct tid_user_buf *tidbuf, unsigned int npages)
 {
 	unsigned pagecount, pageidx, setcount = 0, i;
 	unsigned long pfn, this_pfn;
+	struct page **pages = tidbuf->pages;
+	struct tid_pageset *list = tidbuf->psets;
 
 	if (!npages)
 		return 0;
@@ -741,13 +630,12 @@ static u32 find_phys_blocks(struct page **pages, unsigned npages,
 /**
  * program_rcvarray() - program an RcvArray group with receive buffers
  * @fd: filedata pointer
- * @vaddr: starting user virtual address
+ * @tbuf: pointer to struct tid_user_buf that has the user buffer starting
+ *	  virtual address, buffer length, page pointers, pagesets (array of
+ *	  struct tid_pageset holding information on physically contiguous
+ *	  chunks from the user buffer), and other fields.
  * @grp: RcvArray group
- * @sets: array of struct tid_pageset holding information on physically
- *        contiguous chunks from the user buffer
- * @start: starting index into sets array
  * @count: number of struct tid_pageset's to program
- * @pages: an array of struct page * for the user buffer
  * @tidlist: the array of u32 elements when the information about the
  *           programmed RcvArray entries is to be encoded.
  * @tididx: starting offset into tidlist
@@ -765,15 +653,15 @@ static u32 find_phys_blocks(struct page **pages, unsigned npages,
  * -ENOMEM or -EFAULT on error from set_rcvarray_entry(), or
  * number of RcvArray entries programmed.
  */
-static int program_rcvarray(struct hfi1_filedata *fd, unsigned long vaddr,
-			    struct tid_group *grp,
-			    struct tid_pageset *sets,
-			    unsigned start, u16 count, struct page **pages,
-			    u32 *tidlist, unsigned *tididx, unsigned *pmapped)
+static int program_rcvarray(struct hfi1_filedata *fd, struct tid_user_buf *tbuf,
+			    struct tid_group *grp, u16 count,
+			    u32 *tidlist, unsigned int *tididx,
+			    unsigned int *pmapped)
 {
 	struct hfi1_ctxtdata *uctxt = fd->uctxt;
 	struct hfi1_devdata *dd = uctxt->dd;
 	u16 idx;
+	unsigned int start = *tididx;
 	u32 tidinfo = 0, rcventry, useidx = 0;
 	int mapped = 0;
 
@@ -808,18 +696,17 @@ static int program_rcvarray(struct hfi1_filedata *fd, unsigned long vaddr,
 		}
 
 		rcventry = grp->base + useidx;
-		npages = sets[setidx].count;
-		pageidx = sets[setidx].idx;
+		npages = tbuf->psets[setidx].count;
+		pageidx = tbuf->psets[setidx].idx;
 
-		ret = set_rcvarray_entry(fd, vaddr + (pageidx * PAGE_SIZE),
-					 rcventry, grp, pages + pageidx,
+		ret = set_rcvarray_entry(fd, tbuf,
+					 rcventry, grp, pageidx,
 					 npages);
 		if (ret)
 			return ret;
 		mapped += npages;
 
-		tidinfo = rcventry2tidinfo(rcventry - uctxt->expected_base) |
-			EXP_TID_SET(LEN, npages);
+		tidinfo = create_tid(rcventry - uctxt->expected_base, npages);
 		tidlist[(*tididx)++] = tidinfo;
 		grp->used++;
 		grp->map |= 1 << useidx++;
@@ -833,28 +720,28 @@ static int program_rcvarray(struct hfi1_filedata *fd, unsigned long vaddr,
 	return idx;
 }
 
-static int set_rcvarray_entry(struct hfi1_filedata *fd, unsigned long vaddr,
+static int set_rcvarray_entry(struct hfi1_filedata *fd,
+			      struct tid_user_buf *tbuf,
 			      u32 rcventry, struct tid_group *grp,
-			      struct page **pages, unsigned npages)
+			      u16 pageidx, unsigned int npages)
 {
 	int ret;
 	struct hfi1_ctxtdata *uctxt = fd->uctxt;
 	struct tid_rb_node *node;
 	struct hfi1_devdata *dd = uctxt->dd;
 	dma_addr_t phys;
+	struct page **pages = tbuf->pages + pageidx;
 
 	/*
 	 * Allocate the node first so we can handle a potential
 	 * failure before we've programmed anything.
 	 */
-	node = kzalloc(sizeof(*node) + (sizeof(struct page *) * npages),
-		       GFP_KERNEL);
+	node = kzalloc(struct_size(node, pages, npages), GFP_KERNEL);
 	if (!node)
 		return -ENOMEM;
 
-	phys = pci_map_single(dd->pcidev,
-			      __va(page_to_phys(pages[0])),
-			      npages * PAGE_SIZE, PCI_DMA_FROMDEVICE);
+	phys = dma_map_single(&dd->pcidev->dev, __va(page_to_phys(pages[0])),
+			      npages * PAGE_SIZE, DMA_FROM_DEVICE);
 	if (dma_mapping_error(&dd->pcidev->dev, phys)) {
 		dd_dev_err(dd, "Failed to DMA map Exp Rcv pages 0x%llx\n",
 			   phys);
@@ -862,90 +749,100 @@ static int set_rcvarray_entry(struct hfi1_filedata *fd, unsigned long vaddr,
 		return -EFAULT;
 	}
 
-	node->mmu.addr = vaddr;
-	node->mmu.len = npages * PAGE_SIZE;
+	node->fdata = fd;
+	mutex_init(&node->invalidate_mutex);
 	node->phys = page_to_phys(pages[0]);
 	node->npages = npages;
 	node->rcventry = rcventry;
 	node->dma_addr = phys;
 	node->grp = grp;
 	node->freed = false;
-	memcpy(node->pages, pages, sizeof(struct page *) * npages);
+	memcpy(node->pages, pages, flex_array_size(node, pages, npages));
 
-	if (!fd->handler)
-		ret = tid_rb_insert(fd, &node->mmu);
-	else
-		ret = hfi1_mmu_rb_insert(fd->handler, &node->mmu);
-
-	if (ret) {
-		hfi1_cdbg(TID, "Failed to insert RB node %u 0x%lx, 0x%lx %d",
-			  node->rcventry, node->mmu.addr, node->phys, ret);
-		pci_unmap_single(dd->pcidev, phys, npages * PAGE_SIZE,
-				 PCI_DMA_FROMDEVICE);
-		kfree(node);
-		return -EFAULT;
+	if (fd->use_mn) {
+		ret = mmu_interval_notifier_insert(
+			&node->notifier, current->mm,
+			tbuf->vaddr + (pageidx * PAGE_SIZE), npages * PAGE_SIZE,
+			&tid_mn_ops);
+		if (ret)
+			goto out_unmap;
 	}
+	fd->entry_to_rb[node->rcventry - uctxt->expected_base] = node;
+
 	hfi1_put_tid(dd, rcventry, PT_EXPECTED, phys, ilog2(npages) + 1);
 	trace_hfi1_exp_tid_reg(uctxt->ctxt, fd->subctxt, rcventry, npages,
-			       node->mmu.addr, node->phys, phys);
+			       node->notifier.interval_tree.start, node->phys,
+			       phys);
 	return 0;
+
+out_unmap:
+	hfi1_cdbg(TID, "Failed to insert RB node %u 0x%lx, 0x%lx %d",
+		  node->rcventry, node->notifier.interval_tree.start,
+		  node->phys, ret);
+	dma_unmap_single(&dd->pcidev->dev, phys, npages * PAGE_SIZE,
+			 DMA_FROM_DEVICE);
+	kfree(node);
+	return -EFAULT;
 }
 
-static int unprogram_rcvarray(struct hfi1_filedata *fd, u32 tidinfo,
-			      struct tid_group **grp)
+static int unprogram_rcvarray(struct hfi1_filedata *fd, u32 tidinfo)
 {
 	struct hfi1_ctxtdata *uctxt = fd->uctxt;
 	struct hfi1_devdata *dd = uctxt->dd;
 	struct tid_rb_node *node;
-	u8 tidctrl = EXP_TID_GET(tidinfo, CTRL);
+	u32 tidctrl = EXP_TID_GET(tidinfo, CTRL);
 	u32 tididx = EXP_TID_GET(tidinfo, IDX) << 1, rcventry;
 
-	if (tididx >= uctxt->expected_count) {
-		dd_dev_err(dd, "Invalid RcvArray entry (%u) index for ctxt %u\n",
-			   tididx, uctxt->ctxt);
-		return -EINVAL;
-	}
-
-	if (tidctrl == 0x3)
+	if (tidctrl == 0x3 || tidctrl == 0x0)
 		return -EINVAL;
 
 	rcventry = tididx + (tidctrl - 1);
 
+	if (rcventry >= uctxt->expected_count) {
+		dd_dev_err(dd, "Invalid RcvArray entry (%u) index for ctxt %u\n",
+			   rcventry, uctxt->ctxt);
+		return -EINVAL;
+	}
+
 	node = fd->entry_to_rb[rcventry];
 	if (!node || node->rcventry != (uctxt->expected_base + rcventry))
 		return -EBADF;
 
-	if (grp)
-		*grp = node->grp;
-
-	if (!fd->handler)
-		cacheless_tid_rb_remove(fd, node);
-	else
-		hfi1_mmu_rb_remove(fd->handler, &node->mmu);
+	if (fd->use_mn)
+		mmu_interval_notifier_remove(&node->notifier);
+	cacheless_tid_rb_remove(fd, node);
 
 	return 0;
 }
 
-static void clear_tid_node(struct hfi1_filedata *fd, struct tid_rb_node *node)
+static void __clear_tid_node(struct hfi1_filedata *fd, struct tid_rb_node *node)
 {
 	struct hfi1_ctxtdata *uctxt = fd->uctxt;
 	struct hfi1_devdata *dd = uctxt->dd;
 
+	mutex_lock(&node->invalidate_mutex);
+	if (node->freed)
+		goto done;
+	node->freed = true;
+
 	trace_hfi1_exp_tid_unreg(uctxt->ctxt, fd->subctxt, node->rcventry,
-				 node->npages, node->mmu.addr, node->phys,
+				 node->npages,
+				 node->notifier.interval_tree.start, node->phys,
 				 node->dma_addr);
 
-	hfi1_put_tid(dd, node->rcventry, PT_INVALID, 0, 0);
-	/*
-	 * Make sure device has seen the write before we unpin the
-	 * pages.
-	 */
-	flush_wc();
+	/* Make sure device has seen the write before pages are unpinned */
+	hfi1_put_tid(dd, node->rcventry, PT_INVALID_FLUSH, 0, 0);
+
+	unpin_rcv_pages(fd, NULL, node, 0, node->npages, true);
+done:
+	mutex_unlock(&node->invalidate_mutex);
+}
+
+static void clear_tid_node(struct hfi1_filedata *fd, struct tid_rb_node *node)
+{
+	struct hfi1_ctxtdata *uctxt = fd->uctxt;
 
-	pci_unmap_single(dd->pcidev, node->dma_addr, node->mmu.len,
-			 PCI_DMA_FROMDEVICE);
-	hfi1_release_user_pages(fd->mm, node->pages, node->npages, true);
-	fd->tid_n_pinned -= node->npages;
+	__clear_tid_node(fd, node);
 
 	node->grp->used--;
 	node->grp->map &= ~(1 << (node->rcventry - node->grp->base));
@@ -983,39 +880,43 @@ static void unlock_exp_tids(struct hfi1_ctxtdata *uctxt,
 				if (!node || node->rcventry != rcventry)
 					continue;
 
+				if (fd->use_mn)
+					mmu_interval_notifier_remove(
+						&node->notifier);
 				cacheless_tid_rb_remove(fd, node);
 			}
 		}
 	}
 }
 
-/*
- * Always return 0 from this function.  A non-zero return indicates that the
- * remove operation will be called and that memory should be unpinned.
- * However, the driver cannot unpin out from under PSM.  Instead, retain the
- * memory (by returning 0) and inform PSM that the memory is going away.  PSM
- * will call back later when it has removed the memory from its list.
- */
-static int tid_rb_invalidate(void *arg, struct mmu_rb_node *mnode)
+static bool tid_rb_invalidate(struct mmu_interval_notifier *mni,
+			      const struct mmu_notifier_range *range,
+			      unsigned long cur_seq)
 {
-	struct hfi1_filedata *fdata = arg;
-	struct hfi1_ctxtdata *uctxt = fdata->uctxt;
 	struct tid_rb_node *node =
-		container_of(mnode, struct tid_rb_node, mmu);
+		container_of(mni, struct tid_rb_node, notifier);
+	struct hfi1_filedata *fdata = node->fdata;
+	struct hfi1_ctxtdata *uctxt = fdata->uctxt;
 
 	if (node->freed)
-		return 0;
+		return true;
 
-	trace_hfi1_exp_tid_inval(uctxt->ctxt, fdata->subctxt, node->mmu.addr,
+	/* take action only if unmapping */
+	if (range->event != MMU_NOTIFY_UNMAP)
+		return true;
+
+	trace_hfi1_exp_tid_inval(uctxt->ctxt, fdata->subctxt,
+				 node->notifier.interval_tree.start,
 				 node->rcventry, node->npages, node->dma_addr);
-	node->freed = true;
+
+	/* clear the hardware rcvarray entry */
+	__clear_tid_node(fdata, node);
 
 	spin_lock(&fdata->invalid_lock);
 	if (fdata->invalid_tid_idx < uctxt->expected_count) {
 		fdata->invalid_tids[fdata->invalid_tid_idx] =
-			rcventry2tidinfo(node->rcventry - uctxt->expected_base);
-		fdata->invalid_tids[fdata->invalid_tid_idx] |=
-			EXP_TID_SET(LEN, node->npages);
+			create_tid(node->rcventry - uctxt->expected_base,
+				   node->npages);
 		if (!fdata->invalid_tid_idx) {
 			unsigned long *ev;
 
@@ -1029,25 +930,30 @@ static int tid_rb_invalidate(void *arg, struct mmu_rb_node *mnode)
 			 * process in question.
 			 */
 			ev = uctxt->dd->events +
-			  (((uctxt->ctxt - uctxt->dd->first_dyn_alloc_ctxt) *
-			    HFI1_MAX_SHARED_CTXTS) + fdata->subctxt);
+				(uctxt_offset(uctxt) + fdata->subctxt);
 			set_bit(_HFI1_EVENT_TID_MMU_NOTIFY_BIT, ev);
 		}
 		fdata->invalid_tid_idx++;
 	}
 	spin_unlock(&fdata->invalid_lock);
-	return 0;
+	return true;
 }
 
-static int tid_rb_insert(void *arg, struct mmu_rb_node *node)
+static bool tid_cover_invalidate(struct mmu_interval_notifier *mni,
+			         const struct mmu_notifier_range *range,
+			         unsigned long cur_seq)
 {
-	struct hfi1_filedata *fdata = arg;
-	struct tid_rb_node *tnode =
-		container_of(node, struct tid_rb_node, mmu);
-	u32 base = fdata->uctxt->expected_base;
+	struct tid_user_buf *tidbuf =
+		container_of(mni, struct tid_user_buf, notifier);
+
+	/* take action only if unmapping */
+	if (range->event == MMU_NOTIFY_UNMAP) {
+		mutex_lock(&tidbuf->cover_mutex);
+		mmu_interval_set_seq(mni, cur_seq);
+		mutex_unlock(&tidbuf->cover_mutex);
+	}
 
-	fdata->entry_to_rb[tnode->rcventry - base] = tnode;
-	return 0;
+	return true;
 }
 
 static void cacheless_tid_rb_remove(struct hfi1_filedata *fdata,
@@ -1058,12 +964,3 @@ static void cacheless_tid_rb_remove(struct hfi1_filedata *fdata,
 	fdata->entry_to_rb[tnode->rcventry - base] = NULL;
 	clear_tid_node(fdata, tnode);
 }
-
-static void tid_rb_remove(void *arg, struct mmu_rb_node *node)
-{
-	struct hfi1_filedata *fdata = arg;
-	struct tid_rb_node *tnode =
-		container_of(node, struct tid_rb_node, mmu);
-
-	cacheless_tid_rb_remove(fdata, tnode);
-}