Linux Audio

Check our new training course

Loading...
Note: File does not exist in v3.1.
  1/*
  2 * Copyright(c) 2016 Intel Corporation.
  3 *
  4 * This file is provided under a dual BSD/GPLv2 license.  When using or
  5 * redistributing this file, you may do so under either license.
  6 *
  7 * GPL LICENSE SUMMARY
  8 *
  9 * This program is free software; you can redistribute it and/or modify
 10 * it under the terms of version 2 of the GNU General Public License as
 11 * published by the Free Software Foundation.
 12 *
 13 * This program is distributed in the hope that it will be useful, but
 14 * WITHOUT ANY WARRANTY; without even the implied warranty of
 15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 16 * General Public License for more details.
 17 *
 18 * BSD LICENSE
 19 *
 20 * Redistribution and use in source and binary forms, with or without
 21 * modification, are permitted provided that the following conditions
 22 * are met:
 23 *
 24 *  - Redistributions of source code must retain the above copyright
 25 *    notice, this list of conditions and the following disclaimer.
 26 *  - Redistributions in binary form must reproduce the above copyright
 27 *    notice, this list of conditions and the following disclaimer in
 28 *    the documentation and/or other materials provided with the
 29 *    distribution.
 30 *  - Neither the name of Intel Corporation nor the names of its
 31 *    contributors may be used to endorse or promote products derived
 32 *    from this software without specific prior written permission.
 33 *
 34 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
 35 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
 36 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
 37 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
 38 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
 39 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
 40 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
 41 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
 42 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 43 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 44 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 45 *
 46 */
 47
 48#include <linux/slab.h>
 49#include <linux/vmalloc.h>
 50#include <rdma/ib_umem.h>
 51#include <rdma/rdma_vt.h>
 52#include "vt.h"
 53#include "mr.h"
 54#include "trace.h"
 55
 56/**
 57 * rvt_driver_mr_init - Init MR resources per driver
 58 * @rdi: rvt dev struct
 59 *
 60 * Do any intilization needed when a driver registers with rdmavt.
 61 *
 62 * Return: 0 on success or errno on failure
 63 */
 64int rvt_driver_mr_init(struct rvt_dev_info *rdi)
 65{
 66	unsigned int lkey_table_size = rdi->dparms.lkey_table_size;
 67	unsigned lk_tab_size;
 68	int i;
 69
 70	/*
 71	 * The top hfi1_lkey_table_size bits are used to index the
 72	 * table.  The lower 8 bits can be owned by the user (copied from
 73	 * the LKEY).  The remaining bits act as a generation number or tag.
 74	 */
 75	if (!lkey_table_size)
 76		return -EINVAL;
 77
 78	spin_lock_init(&rdi->lkey_table.lock);
 79
 80	/* ensure generation is at least 4 bits */
 81	if (lkey_table_size > RVT_MAX_LKEY_TABLE_BITS) {
 82		rvt_pr_warn(rdi, "lkey bits %u too large, reduced to %u\n",
 83			    lkey_table_size, RVT_MAX_LKEY_TABLE_BITS);
 84		rdi->dparms.lkey_table_size = RVT_MAX_LKEY_TABLE_BITS;
 85		lkey_table_size = rdi->dparms.lkey_table_size;
 86	}
 87	rdi->lkey_table.max = 1 << lkey_table_size;
 88	rdi->lkey_table.shift = 32 - lkey_table_size;
 89	lk_tab_size = rdi->lkey_table.max * sizeof(*rdi->lkey_table.table);
 90	rdi->lkey_table.table = (struct rvt_mregion __rcu **)
 91			       vmalloc_node(lk_tab_size, rdi->dparms.node);
 92	if (!rdi->lkey_table.table)
 93		return -ENOMEM;
 94
 95	RCU_INIT_POINTER(rdi->dma_mr, NULL);
 96	for (i = 0; i < rdi->lkey_table.max; i++)
 97		RCU_INIT_POINTER(rdi->lkey_table.table[i], NULL);
 98
 99	rdi->dparms.props.max_mr = rdi->lkey_table.max;
100	return 0;
101}
102
103/**
104 * rvt_mr_exit - clean up MR
105 * @rdi: rvt dev structure
106 *
107 * called when drivers have unregistered or perhaps failed to register with us
108 */
109void rvt_mr_exit(struct rvt_dev_info *rdi)
110{
111	if (rdi->dma_mr)
112		rvt_pr_err(rdi, "DMA MR not null!\n");
113
114	vfree(rdi->lkey_table.table);
115}
116
117static void rvt_deinit_mregion(struct rvt_mregion *mr)
118{
119	int i = mr->mapsz;
120
121	mr->mapsz = 0;
122	while (i)
123		kfree(mr->map[--i]);
124	percpu_ref_exit(&mr->refcount);
125}
126
127static void __rvt_mregion_complete(struct percpu_ref *ref)
128{
129	struct rvt_mregion *mr = container_of(ref, struct rvt_mregion,
130					      refcount);
131
132	complete(&mr->comp);
133}
134
135static int rvt_init_mregion(struct rvt_mregion *mr, struct ib_pd *pd,
136			    int count, unsigned int percpu_flags)
137{
138	int m, i = 0;
139	struct rvt_dev_info *dev = ib_to_rvt(pd->device);
140
141	mr->mapsz = 0;
142	m = (count + RVT_SEGSZ - 1) / RVT_SEGSZ;
143	for (; i < m; i++) {
144		mr->map[i] = kzalloc_node(sizeof(*mr->map[0]), GFP_KERNEL,
145					  dev->dparms.node);
146		if (!mr->map[i])
147			goto bail;
148		mr->mapsz++;
149	}
150	init_completion(&mr->comp);
151	/* count returning the ptr to user */
152	if (percpu_ref_init(&mr->refcount, &__rvt_mregion_complete,
153			    percpu_flags, GFP_KERNEL))
154		goto bail;
155
156	atomic_set(&mr->lkey_invalid, 0);
157	mr->pd = pd;
158	mr->max_segs = count;
159	return 0;
160bail:
161	rvt_deinit_mregion(mr);
162	return -ENOMEM;
163}
164
165/**
166 * rvt_alloc_lkey - allocate an lkey
167 * @mr: memory region that this lkey protects
168 * @dma_region: 0->normal key, 1->restricted DMA key
169 *
170 * Returns 0 if successful, otherwise returns -errno.
171 *
172 * Increments mr reference count as required.
173 *
174 * Sets the lkey field mr for non-dma regions.
175 *
176 */
177static int rvt_alloc_lkey(struct rvt_mregion *mr, int dma_region)
178{
179	unsigned long flags;
180	u32 r;
181	u32 n;
182	int ret = 0;
183	struct rvt_dev_info *dev = ib_to_rvt(mr->pd->device);
184	struct rvt_lkey_table *rkt = &dev->lkey_table;
185
186	rvt_get_mr(mr);
187	spin_lock_irqsave(&rkt->lock, flags);
188
189	/* special case for dma_mr lkey == 0 */
190	if (dma_region) {
191		struct rvt_mregion *tmr;
192
193		tmr = rcu_access_pointer(dev->dma_mr);
194		if (!tmr) {
195			mr->lkey_published = 1;
196			/* Insure published written first */
197			rcu_assign_pointer(dev->dma_mr, mr);
198			rvt_get_mr(mr);
199		}
200		goto success;
201	}
202
203	/* Find the next available LKEY */
204	r = rkt->next;
205	n = r;
206	for (;;) {
207		if (!rcu_access_pointer(rkt->table[r]))
208			break;
209		r = (r + 1) & (rkt->max - 1);
210		if (r == n)
211			goto bail;
212	}
213	rkt->next = (r + 1) & (rkt->max - 1);
214	/*
215	 * Make sure lkey is never zero which is reserved to indicate an
216	 * unrestricted LKEY.
217	 */
218	rkt->gen++;
219	/*
220	 * bits are capped to ensure enough bits for generation number
221	 */
222	mr->lkey = (r << (32 - dev->dparms.lkey_table_size)) |
223		((((1 << (24 - dev->dparms.lkey_table_size)) - 1) & rkt->gen)
224		 << 8);
225	if (mr->lkey == 0) {
226		mr->lkey |= 1 << 8;
227		rkt->gen++;
228	}
229	mr->lkey_published = 1;
230	/* Insure published written first */
231	rcu_assign_pointer(rkt->table[r], mr);
232success:
233	spin_unlock_irqrestore(&rkt->lock, flags);
234out:
235	return ret;
236bail:
237	rvt_put_mr(mr);
238	spin_unlock_irqrestore(&rkt->lock, flags);
239	ret = -ENOMEM;
240	goto out;
241}
242
243/**
244 * rvt_free_lkey - free an lkey
245 * @mr: mr to free from tables
246 */
247static void rvt_free_lkey(struct rvt_mregion *mr)
248{
249	unsigned long flags;
250	u32 lkey = mr->lkey;
251	u32 r;
252	struct rvt_dev_info *dev = ib_to_rvt(mr->pd->device);
253	struct rvt_lkey_table *rkt = &dev->lkey_table;
254	int freed = 0;
255
256	spin_lock_irqsave(&rkt->lock, flags);
257	if (!lkey) {
258		if (mr->lkey_published) {
259			mr->lkey_published = 0;
260			/* insure published is written before pointer */
261			rcu_assign_pointer(dev->dma_mr, NULL);
262			rvt_put_mr(mr);
263		}
264	} else {
265		if (!mr->lkey_published)
266			goto out;
267		r = lkey >> (32 - dev->dparms.lkey_table_size);
268		mr->lkey_published = 0;
269		/* insure published is written before pointer */
270		rcu_assign_pointer(rkt->table[r], NULL);
271	}
272	freed++;
273out:
274	spin_unlock_irqrestore(&rkt->lock, flags);
275	if (freed)
276		percpu_ref_kill(&mr->refcount);
277}
278
279static struct rvt_mr *__rvt_alloc_mr(int count, struct ib_pd *pd)
280{
281	struct rvt_mr *mr;
282	int rval = -ENOMEM;
283	int m;
284
285	/* Allocate struct plus pointers to first level page tables. */
286	m = (count + RVT_SEGSZ - 1) / RVT_SEGSZ;
287	mr = kzalloc(struct_size(mr, mr.map, m), GFP_KERNEL);
288	if (!mr)
289		goto bail;
290
291	rval = rvt_init_mregion(&mr->mr, pd, count, 0);
292	if (rval)
293		goto bail;
294	/*
295	 * ib_reg_phys_mr() will initialize mr->ibmr except for
296	 * lkey and rkey.
297	 */
298	rval = rvt_alloc_lkey(&mr->mr, 0);
299	if (rval)
300		goto bail_mregion;
301	mr->ibmr.lkey = mr->mr.lkey;
302	mr->ibmr.rkey = mr->mr.lkey;
303done:
304	return mr;
305
306bail_mregion:
307	rvt_deinit_mregion(&mr->mr);
308bail:
309	kfree(mr);
310	mr = ERR_PTR(rval);
311	goto done;
312}
313
314static void __rvt_free_mr(struct rvt_mr *mr)
315{
316	rvt_free_lkey(&mr->mr);
317	rvt_deinit_mregion(&mr->mr);
318	kfree(mr);
319}
320
321/**
322 * rvt_get_dma_mr - get a DMA memory region
323 * @pd: protection domain for this memory region
324 * @acc: access flags
325 *
326 * Return: the memory region on success, otherwise returns an errno.
327 */
328struct ib_mr *rvt_get_dma_mr(struct ib_pd *pd, int acc)
329{
330	struct rvt_mr *mr;
331	struct ib_mr *ret;
332	int rval;
333
334	if (ibpd_to_rvtpd(pd)->user)
335		return ERR_PTR(-EPERM);
336
337	mr = kzalloc(sizeof(*mr), GFP_KERNEL);
338	if (!mr) {
339		ret = ERR_PTR(-ENOMEM);
340		goto bail;
341	}
342
343	rval = rvt_init_mregion(&mr->mr, pd, 0, 0);
344	if (rval) {
345		ret = ERR_PTR(rval);
346		goto bail;
347	}
348
349	rval = rvt_alloc_lkey(&mr->mr, 1);
350	if (rval) {
351		ret = ERR_PTR(rval);
352		goto bail_mregion;
353	}
354
355	mr->mr.access_flags = acc;
356	ret = &mr->ibmr;
357done:
358	return ret;
359
360bail_mregion:
361	rvt_deinit_mregion(&mr->mr);
362bail:
363	kfree(mr);
364	goto done;
365}
366
367/**
368 * rvt_reg_user_mr - register a userspace memory region
369 * @pd: protection domain for this memory region
370 * @start: starting userspace address
371 * @length: length of region to register
372 * @virt_addr: associated virtual address
373 * @mr_access_flags: access flags for this memory region
374 * @udata: unused by the driver
375 *
376 * Return: the memory region on success, otherwise returns an errno.
377 */
378struct ib_mr *rvt_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
379			      u64 virt_addr, int mr_access_flags,
380			      struct ib_udata *udata)
381{
382	struct rvt_mr *mr;
383	struct ib_umem *umem;
384	struct sg_page_iter sg_iter;
385	int n, m;
386	struct ib_mr *ret;
387
388	if (length == 0)
389		return ERR_PTR(-EINVAL);
390
391	umem = ib_umem_get(pd->device, start, length, mr_access_flags);
392	if (IS_ERR(umem))
393		return (void *)umem;
394
395	n = ib_umem_num_pages(umem);
396
397	mr = __rvt_alloc_mr(n, pd);
398	if (IS_ERR(mr)) {
399		ret = (struct ib_mr *)mr;
400		goto bail_umem;
401	}
402
403	mr->mr.user_base = start;
404	mr->mr.iova = virt_addr;
405	mr->mr.length = length;
406	mr->mr.offset = ib_umem_offset(umem);
407	mr->mr.access_flags = mr_access_flags;
408	mr->umem = umem;
409
410	mr->mr.page_shift = PAGE_SHIFT;
411	m = 0;
412	n = 0;
413	for_each_sg_page (umem->sg_head.sgl, &sg_iter, umem->nmap, 0) {
414		void *vaddr;
415
416		vaddr = page_address(sg_page_iter_page(&sg_iter));
417		if (!vaddr) {
418			ret = ERR_PTR(-EINVAL);
419			goto bail_inval;
420		}
421		mr->mr.map[m]->segs[n].vaddr = vaddr;
422		mr->mr.map[m]->segs[n].length = PAGE_SIZE;
423		trace_rvt_mr_user_seg(&mr->mr, m, n, vaddr, PAGE_SIZE);
424		if (++n == RVT_SEGSZ) {
425			m++;
426			n = 0;
427		}
428	}
429	return &mr->ibmr;
430
431bail_inval:
432	__rvt_free_mr(mr);
433
434bail_umem:
435	ib_umem_release(umem);
436
437	return ret;
438}
439
440/**
441 * rvt_dereg_clean_qp_cb - callback from iterator
442 * @qp: the qp
443 * @v: the mregion (as u64)
444 *
445 * This routine fields the callback for all QPs and
446 * for QPs in the same PD as the MR will call the
447 * rvt_qp_mr_clean() to potentially cleanup references.
448 */
449static void rvt_dereg_clean_qp_cb(struct rvt_qp *qp, u64 v)
450{
451	struct rvt_mregion *mr = (struct rvt_mregion *)v;
452
453	/* skip PDs that are not ours */
454	if (mr->pd != qp->ibqp.pd)
455		return;
456	rvt_qp_mr_clean(qp, mr->lkey);
457}
458
459/**
460 * rvt_dereg_clean_qps - find QPs for reference cleanup
461 * @mr: the MR that is being deregistered
462 *
463 * This routine iterates RC QPs looking for references
464 * to the lkey noted in mr.
465 */
466static void rvt_dereg_clean_qps(struct rvt_mregion *mr)
467{
468	struct rvt_dev_info *rdi = ib_to_rvt(mr->pd->device);
469
470	rvt_qp_iter(rdi, (u64)mr, rvt_dereg_clean_qp_cb);
471}
472
473/**
474 * rvt_check_refs - check references
475 * @mr: the megion
476 * @t: the caller identification
477 *
478 * This routine checks MRs holding a reference during
479 * when being de-registered.
480 *
481 * If the count is non-zero, the code calls a clean routine then
482 * waits for the timeout for the count to zero.
483 */
484static int rvt_check_refs(struct rvt_mregion *mr, const char *t)
485{
486	unsigned long timeout;
487	struct rvt_dev_info *rdi = ib_to_rvt(mr->pd->device);
488
489	if (mr->lkey) {
490		/* avoid dma mr */
491		rvt_dereg_clean_qps(mr);
492		/* @mr was indexed on rcu protected @lkey_table */
493		synchronize_rcu();
494	}
495
496	timeout = wait_for_completion_timeout(&mr->comp, 5 * HZ);
497	if (!timeout) {
498		rvt_pr_err(rdi,
499			   "%s timeout mr %p pd %p lkey %x refcount %ld\n",
500			   t, mr, mr->pd, mr->lkey,
501			   atomic_long_read(&mr->refcount.data->count));
502		rvt_get_mr(mr);
503		return -EBUSY;
504	}
505	return 0;
506}
507
508/**
509 * rvt_mr_has_lkey - is MR
510 * @mr: the mregion
511 * @lkey: the lkey
512 */
513bool rvt_mr_has_lkey(struct rvt_mregion *mr, u32 lkey)
514{
515	return mr && lkey == mr->lkey;
516}
517
518/**
519 * rvt_ss_has_lkey - is mr in sge tests
520 * @ss: the sge state
521 * @lkey: the lkey
522 *
523 * This code tests for an MR in the indicated
524 * sge state.
525 */
526bool rvt_ss_has_lkey(struct rvt_sge_state *ss, u32 lkey)
527{
528	int i;
529	bool rval = false;
530
531	if (!ss->num_sge)
532		return rval;
533	/* first one */
534	rval = rvt_mr_has_lkey(ss->sge.mr, lkey);
535	/* any others */
536	for (i = 0; !rval && i < ss->num_sge - 1; i++)
537		rval = rvt_mr_has_lkey(ss->sg_list[i].mr, lkey);
538	return rval;
539}
540
541/**
542 * rvt_dereg_mr - unregister and free a memory region
543 * @ibmr: the memory region to free
544 * @udata: unused by the driver
545 *
546 * Note that this is called to free MRs created by rvt_get_dma_mr()
547 * or rvt_reg_user_mr().
548 *
549 * Returns 0 on success.
550 */
551int rvt_dereg_mr(struct ib_mr *ibmr, struct ib_udata *udata)
552{
553	struct rvt_mr *mr = to_imr(ibmr);
554	int ret;
555
556	rvt_free_lkey(&mr->mr);
557
558	rvt_put_mr(&mr->mr); /* will set completion if last */
559	ret = rvt_check_refs(&mr->mr, __func__);
560	if (ret)
561		goto out;
562	rvt_deinit_mregion(&mr->mr);
563	ib_umem_release(mr->umem);
564	kfree(mr);
565out:
566	return ret;
567}
568
569/**
570 * rvt_alloc_mr - Allocate a memory region usable with the
571 * @pd: protection domain for this memory region
572 * @mr_type: mem region type
573 * @max_num_sg: Max number of segments allowed
574 *
575 * Return: the memory region on success, otherwise return an errno.
576 */
577struct ib_mr *rvt_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type,
578			   u32 max_num_sg)
579{
580	struct rvt_mr *mr;
581
582	if (mr_type != IB_MR_TYPE_MEM_REG)
583		return ERR_PTR(-EINVAL);
584
585	mr = __rvt_alloc_mr(max_num_sg, pd);
586	if (IS_ERR(mr))
587		return (struct ib_mr *)mr;
588
589	return &mr->ibmr;
590}
591
592/**
593 * rvt_set_page - page assignment function called by ib_sg_to_pages
594 * @ibmr: memory region
595 * @addr: dma address of mapped page
596 *
597 * Return: 0 on success
598 */
599static int rvt_set_page(struct ib_mr *ibmr, u64 addr)
600{
601	struct rvt_mr *mr = to_imr(ibmr);
602	u32 ps = 1 << mr->mr.page_shift;
603	u32 mapped_segs = mr->mr.length >> mr->mr.page_shift;
604	int m, n;
605
606	if (unlikely(mapped_segs == mr->mr.max_segs))
607		return -ENOMEM;
608
609	m = mapped_segs / RVT_SEGSZ;
610	n = mapped_segs % RVT_SEGSZ;
611	mr->mr.map[m]->segs[n].vaddr = (void *)addr;
612	mr->mr.map[m]->segs[n].length = ps;
613	mr->mr.length += ps;
614	trace_rvt_mr_page_seg(&mr->mr, m, n, (void *)addr, ps);
615
616	return 0;
617}
618
619/**
620 * rvt_map_mr_sg - map sg list and set it the memory region
621 * @ibmr: memory region
622 * @sg: dma mapped scatterlist
623 * @sg_nents: number of entries in sg
624 * @sg_offset: offset in bytes into sg
625 *
626 * Overwrite rvt_mr length with mr length calculated by ib_sg_to_pages.
627 *
628 * Return: number of sg elements mapped to the memory region
629 */
630int rvt_map_mr_sg(struct ib_mr *ibmr, struct scatterlist *sg,
631		  int sg_nents, unsigned int *sg_offset)
632{
633	struct rvt_mr *mr = to_imr(ibmr);
634	int ret;
635
636	mr->mr.length = 0;
637	mr->mr.page_shift = PAGE_SHIFT;
638	ret = ib_sg_to_pages(ibmr, sg, sg_nents, sg_offset, rvt_set_page);
639	mr->mr.user_base = ibmr->iova;
640	mr->mr.iova = ibmr->iova;
641	mr->mr.offset = ibmr->iova - (u64)mr->mr.map[0]->segs[0].vaddr;
642	mr->mr.length = (size_t)ibmr->length;
643	trace_rvt_map_mr_sg(ibmr, sg_nents, sg_offset);
644	return ret;
645}
646
647/**
648 * rvt_fast_reg_mr - fast register physical MR
649 * @qp: the queue pair where the work request comes from
650 * @ibmr: the memory region to be registered
651 * @key: updated key for this memory region
652 * @access: access flags for this memory region
653 *
654 * Returns 0 on success.
655 */
656int rvt_fast_reg_mr(struct rvt_qp *qp, struct ib_mr *ibmr, u32 key,
657		    int access)
658{
659	struct rvt_mr *mr = to_imr(ibmr);
660
661	if (qp->ibqp.pd != mr->mr.pd)
662		return -EACCES;
663
664	/* not applicable to dma MR or user MR */
665	if (!mr->mr.lkey || mr->umem)
666		return -EINVAL;
667
668	if ((key & 0xFFFFFF00) != (mr->mr.lkey & 0xFFFFFF00))
669		return -EINVAL;
670
671	ibmr->lkey = key;
672	ibmr->rkey = key;
673	mr->mr.lkey = key;
674	mr->mr.access_flags = access;
675	mr->mr.iova = ibmr->iova;
676	atomic_set(&mr->mr.lkey_invalid, 0);
677
678	return 0;
679}
680EXPORT_SYMBOL(rvt_fast_reg_mr);
681
682/**
683 * rvt_invalidate_rkey - invalidate an MR rkey
684 * @qp: queue pair associated with the invalidate op
685 * @rkey: rkey to invalidate
686 *
687 * Returns 0 on success.
688 */
689int rvt_invalidate_rkey(struct rvt_qp *qp, u32 rkey)
690{
691	struct rvt_dev_info *dev = ib_to_rvt(qp->ibqp.device);
692	struct rvt_lkey_table *rkt = &dev->lkey_table;
693	struct rvt_mregion *mr;
694
695	if (rkey == 0)
696		return -EINVAL;
697
698	rcu_read_lock();
699	mr = rcu_dereference(
700		rkt->table[(rkey >> (32 - dev->dparms.lkey_table_size))]);
701	if (unlikely(!mr || mr->lkey != rkey || qp->ibqp.pd != mr->pd))
702		goto bail;
703
704	atomic_set(&mr->lkey_invalid, 1);
705	rcu_read_unlock();
706	return 0;
707
708bail:
709	rcu_read_unlock();
710	return -EINVAL;
711}
712EXPORT_SYMBOL(rvt_invalidate_rkey);
713
714/**
715 * rvt_sge_adjacent - is isge compressible
716 * @last_sge: last outgoing SGE written
717 * @sge: SGE to check
718 *
719 * If adjacent will update last_sge to add length.
720 *
721 * Return: true if isge is adjacent to last sge
722 */
723static inline bool rvt_sge_adjacent(struct rvt_sge *last_sge,
724				    struct ib_sge *sge)
725{
726	if (last_sge && sge->lkey == last_sge->mr->lkey &&
727	    ((uint64_t)(last_sge->vaddr + last_sge->length) == sge->addr)) {
728		if (sge->lkey) {
729			if (unlikely((sge->addr - last_sge->mr->user_base +
730			      sge->length > last_sge->mr->length)))
731				return false; /* overrun, caller will catch */
732		} else {
733			last_sge->length += sge->length;
734		}
735		last_sge->sge_length += sge->length;
736		trace_rvt_sge_adjacent(last_sge, sge);
737		return true;
738	}
739	return false;
740}
741
742/**
743 * rvt_lkey_ok - check IB SGE for validity and initialize
744 * @rkt: table containing lkey to check SGE against
745 * @pd: protection domain
746 * @isge: outgoing internal SGE
747 * @last_sge: last outgoing SGE written
748 * @sge: SGE to check
749 * @acc: access flags
750 *
751 * Check the IB SGE for validity and initialize our internal version
752 * of it.
753 *
754 * Increments the reference count when a new sge is stored.
755 *
756 * Return: 0 if compressed, 1 if added , otherwise returns -errno.
757 */
758int rvt_lkey_ok(struct rvt_lkey_table *rkt, struct rvt_pd *pd,
759		struct rvt_sge *isge, struct rvt_sge *last_sge,
760		struct ib_sge *sge, int acc)
761{
762	struct rvt_mregion *mr;
763	unsigned n, m;
764	size_t off;
765
766	/*
767	 * We use LKEY == zero for kernel virtual addresses
768	 * (see rvt_get_dma_mr()).
769	 */
770	if (sge->lkey == 0) {
771		struct rvt_dev_info *dev = ib_to_rvt(pd->ibpd.device);
772
773		if (pd->user)
774			return -EINVAL;
775		if (rvt_sge_adjacent(last_sge, sge))
776			return 0;
777		rcu_read_lock();
778		mr = rcu_dereference(dev->dma_mr);
779		if (!mr)
780			goto bail;
781		rvt_get_mr(mr);
782		rcu_read_unlock();
783
784		isge->mr = mr;
785		isge->vaddr = (void *)sge->addr;
786		isge->length = sge->length;
787		isge->sge_length = sge->length;
788		isge->m = 0;
789		isge->n = 0;
790		goto ok;
791	}
792	if (rvt_sge_adjacent(last_sge, sge))
793		return 0;
794	rcu_read_lock();
795	mr = rcu_dereference(rkt->table[sge->lkey >> rkt->shift]);
796	if (!mr)
797		goto bail;
798	rvt_get_mr(mr);
799	if (!READ_ONCE(mr->lkey_published))
800		goto bail_unref;
801
802	if (unlikely(atomic_read(&mr->lkey_invalid) ||
803		     mr->lkey != sge->lkey || mr->pd != &pd->ibpd))
804		goto bail_unref;
805
806	off = sge->addr - mr->user_base;
807	if (unlikely(sge->addr < mr->user_base ||
808		     off + sge->length > mr->length ||
809		     (mr->access_flags & acc) != acc))
810		goto bail_unref;
811	rcu_read_unlock();
812
813	off += mr->offset;
814	if (mr->page_shift) {
815		/*
816		 * page sizes are uniform power of 2 so no loop is necessary
817		 * entries_spanned_by_off is the number of times the loop below
818		 * would have executed.
819		*/
820		size_t entries_spanned_by_off;
821
822		entries_spanned_by_off = off >> mr->page_shift;
823		off -= (entries_spanned_by_off << mr->page_shift);
824		m = entries_spanned_by_off / RVT_SEGSZ;
825		n = entries_spanned_by_off % RVT_SEGSZ;
826	} else {
827		m = 0;
828		n = 0;
829		while (off >= mr->map[m]->segs[n].length) {
830			off -= mr->map[m]->segs[n].length;
831			n++;
832			if (n >= RVT_SEGSZ) {
833				m++;
834				n = 0;
835			}
836		}
837	}
838	isge->mr = mr;
839	isge->vaddr = mr->map[m]->segs[n].vaddr + off;
840	isge->length = mr->map[m]->segs[n].length - off;
841	isge->sge_length = sge->length;
842	isge->m = m;
843	isge->n = n;
844ok:
845	trace_rvt_sge_new(isge, sge);
846	return 1;
847bail_unref:
848	rvt_put_mr(mr);
849bail:
850	rcu_read_unlock();
851	return -EINVAL;
852}
853EXPORT_SYMBOL(rvt_lkey_ok);
854
855/**
856 * rvt_rkey_ok - check the IB virtual address, length, and RKEY
857 * @qp: qp for validation
858 * @sge: SGE state
859 * @len: length of data
860 * @vaddr: virtual address to place data
861 * @rkey: rkey to check
862 * @acc: access flags
863 *
864 * Return: 1 if successful, otherwise 0.
865 *
866 * increments the reference count upon success
867 */
868int rvt_rkey_ok(struct rvt_qp *qp, struct rvt_sge *sge,
869		u32 len, u64 vaddr, u32 rkey, int acc)
870{
871	struct rvt_dev_info *dev = ib_to_rvt(qp->ibqp.device);
872	struct rvt_lkey_table *rkt = &dev->lkey_table;
873	struct rvt_mregion *mr;
874	unsigned n, m;
875	size_t off;
876
877	/*
878	 * We use RKEY == zero for kernel virtual addresses
879	 * (see rvt_get_dma_mr()).
880	 */
881	rcu_read_lock();
882	if (rkey == 0) {
883		struct rvt_pd *pd = ibpd_to_rvtpd(qp->ibqp.pd);
884		struct rvt_dev_info *rdi = ib_to_rvt(pd->ibpd.device);
885
886		if (pd->user)
887			goto bail;
888		mr = rcu_dereference(rdi->dma_mr);
889		if (!mr)
890			goto bail;
891		rvt_get_mr(mr);
892		rcu_read_unlock();
893
894		sge->mr = mr;
895		sge->vaddr = (void *)vaddr;
896		sge->length = len;
897		sge->sge_length = len;
898		sge->m = 0;
899		sge->n = 0;
900		goto ok;
901	}
902
903	mr = rcu_dereference(rkt->table[rkey >> rkt->shift]);
904	if (!mr)
905		goto bail;
906	rvt_get_mr(mr);
907	/* insure mr read is before test */
908	if (!READ_ONCE(mr->lkey_published))
909		goto bail_unref;
910	if (unlikely(atomic_read(&mr->lkey_invalid) ||
911		     mr->lkey != rkey || qp->ibqp.pd != mr->pd))
912		goto bail_unref;
913
914	off = vaddr - mr->iova;
915	if (unlikely(vaddr < mr->iova || off + len > mr->length ||
916		     (mr->access_flags & acc) == 0))
917		goto bail_unref;
918	rcu_read_unlock();
919
920	off += mr->offset;
921	if (mr->page_shift) {
922		/*
923		 * page sizes are uniform power of 2 so no loop is necessary
924		 * entries_spanned_by_off is the number of times the loop below
925		 * would have executed.
926		*/
927		size_t entries_spanned_by_off;
928
929		entries_spanned_by_off = off >> mr->page_shift;
930		off -= (entries_spanned_by_off << mr->page_shift);
931		m = entries_spanned_by_off / RVT_SEGSZ;
932		n = entries_spanned_by_off % RVT_SEGSZ;
933	} else {
934		m = 0;
935		n = 0;
936		while (off >= mr->map[m]->segs[n].length) {
937			off -= mr->map[m]->segs[n].length;
938			n++;
939			if (n >= RVT_SEGSZ) {
940				m++;
941				n = 0;
942			}
943		}
944	}
945	sge->mr = mr;
946	sge->vaddr = mr->map[m]->segs[n].vaddr + off;
947	sge->length = mr->map[m]->segs[n].length - off;
948	sge->sge_length = len;
949	sge->m = m;
950	sge->n = n;
951ok:
952	return 1;
953bail_unref:
954	rvt_put_mr(mr);
955bail:
956	rcu_read_unlock();
957	return 0;
958}
959EXPORT_SYMBOL(rvt_rkey_ok);