umem.c 7.1 KB
Newer Older
1 2 3
/*
 * Copyright (c) 2005 Topspin Communications.  All rights reserved.
 * Copyright (c) 2005 Cisco Systems.  All rights reserved.
4
 * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54
 *
 * 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.
 *
 * $Id: uverbs_mem.c 2743 2005-06-28 22:27:59Z roland $
 */

#include <linux/mm.h>
#include <linux/dma-mapping.h>

#include "uverbs.h"

struct ib_umem_account_work {
	struct work_struct work;
	struct mm_struct  *mm;
	unsigned long      diff;
};


static void __ib_umem_release(struct ib_device *dev, struct ib_umem *umem, int dirty)
{
	struct ib_umem_chunk *chunk, *tmp;
	int i;

	list_for_each_entry_safe(chunk, tmp, &umem->chunk_list, list) {
55 56
		ib_dma_unmap_sg(dev, chunk->page_list,
				chunk->nents, DMA_BIDIRECTIONAL);
57 58 59 60 61 62 63 64 65 66
		for (i = 0; i < chunk->nents; ++i) {
			if (umem->writable && dirty)
				set_page_dirty_lock(chunk->page_list[i].page);
			put_page(chunk->page_list[i].page);
		}

		kfree(chunk);
	}
}

67 68 69 70 71 72 73 74 75
/**
 * ib_umem_get - Pin and DMA map userspace memory.
 * @context: userspace context to pin memory for
 * @addr: userspace virtual address to start at
 * @size: length of region to pin
 * @access: IB_ACCESS_xxx flags for memory being pinned
 */
struct ib_umem *ib_umem_get(struct ib_ucontext *context, unsigned long addr,
			    size_t size, int access)
76
{
77
	struct ib_umem *umem;
78 79 80 81 82 83
	struct page **page_list;
	struct ib_umem_chunk *chunk;
	unsigned long locked;
	unsigned long lock_limit;
	unsigned long cur_base;
	unsigned long npages;
84
	int ret;
85 86 87 88
	int off;
	int i;

	if (!can_do_mlock())
89
		return ERR_PTR(-EPERM);
90

91 92 93
	umem = kmalloc(sizeof *umem, GFP_KERNEL);
	if (!umem)
		return ERR_PTR(-ENOMEM);
94

95 96 97 98 99 100 101 102 103 104 105 106
	umem->context   = context;
	umem->length    = size;
	umem->offset    = addr & ~PAGE_MASK;
	umem->page_size = PAGE_SIZE;
	/*
	 * We ask for writable memory if any access flags other than
	 * "remote read" are set.  "Local write" and "remote write"
	 * obviously require write access.  "Remote atomic" can do
	 * things like fetch and add, which will modify memory, and
	 * "MW bind" can change permissions by binding a window.
	 */
	umem->writable  = !!(access & ~IB_ACCESS_REMOTE_READ);
107

108 109 110 111 112 113 114
	INIT_LIST_HEAD(&umem->chunk_list);

	page_list = (struct page **) __get_free_page(GFP_KERNEL);
	if (!page_list) {
		kfree(umem);
		return ERR_PTR(-ENOMEM);
	}
115

116
	npages = PAGE_ALIGN(size + umem->offset) >> PAGE_SHIFT;
117 118 119 120 121 122 123 124 125 126 127

	down_write(&current->mm->mmap_sem);

	locked     = npages + current->mm->locked_vm;
	lock_limit = current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur >> PAGE_SHIFT;

	if ((locked > lock_limit) && !capable(CAP_IPC_LOCK)) {
		ret = -ENOMEM;
		goto out;
	}

128
	cur_base = addr & PAGE_MASK;
129 130 131 132 133

	while (npages) {
		ret = get_user_pages(current, current->mm, cur_base,
				     min_t(int, npages,
					   PAGE_SIZE / sizeof (struct page *)),
134
				     1, !umem->writable, page_list, NULL);
135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159

		if (ret < 0)
			goto out;

		cur_base += ret * PAGE_SIZE;
		npages   -= ret;

		off = 0;

		while (ret) {
			chunk = kmalloc(sizeof *chunk + sizeof (struct scatterlist) *
					min_t(int, ret, IB_UMEM_MAX_PAGE_CHUNK),
					GFP_KERNEL);
			if (!chunk) {
				ret = -ENOMEM;
				goto out;
			}

			chunk->nents = min_t(int, ret, IB_UMEM_MAX_PAGE_CHUNK);
			for (i = 0; i < chunk->nents; ++i) {
				chunk->page_list[i].page   = page_list[i + off];
				chunk->page_list[i].offset = 0;
				chunk->page_list[i].length = PAGE_SIZE;
			}

160
			chunk->nmap = ib_dma_map_sg(context->device,
161 162 163
						    &chunk->page_list[0],
						    chunk->nents,
						    DMA_BIDIRECTIONAL);
164 165 166 167 168 169 170 171 172 173 174
			if (chunk->nmap <= 0) {
				for (i = 0; i < chunk->nents; ++i)
					put_page(chunk->page_list[i].page);
				kfree(chunk);

				ret = -ENOMEM;
				goto out;
			}

			ret -= chunk->nents;
			off += chunk->nents;
175
			list_add_tail(&chunk->list, &umem->chunk_list);
176 177 178 179 180 181
		}

		ret = 0;
	}

out:
182 183 184 185
	if (ret < 0) {
		__ib_umem_release(context->device, umem, 0);
		kfree(umem);
	} else
186 187 188 189 190
		current->mm->locked_vm = locked;

	up_write(&current->mm->mmap_sem);
	free_page((unsigned long) page_list);

191
	return ret < 0 ? ERR_PTR(ret) : umem;
192
}
193
EXPORT_SYMBOL(ib_umem_get);
194

D
David Howells 已提交
195
static void ib_umem_account(struct work_struct *_work)
196
{
D
David Howells 已提交
197 198
	struct ib_umem_account_work *work =
		container_of(_work, struct ib_umem_account_work, work);
199 200 201 202 203 204 205 206

	down_write(&work->mm->mmap_sem);
	work->mm->locked_vm -= work->diff;
	up_write(&work->mm->mmap_sem);
	mmput(work->mm);
	kfree(work);
}

207 208 209 210 211
/**
 * ib_umem_release - release memory pinned with ib_umem_get
 * @umem: umem struct to release
 */
void ib_umem_release(struct ib_umem *umem)
212 213
{
	struct ib_umem_account_work *work;
214
	struct ib_ucontext *context = umem->context;
215
	struct mm_struct *mm;
216
	unsigned long diff;
217

218
	__ib_umem_release(umem->context->device, umem, 1);
219 220 221 222 223

	mm = get_task_mm(current);
	if (!mm)
		return;

224 225 226
	diff = PAGE_ALIGN(umem->length + umem->offset) >> PAGE_SHIFT;
	kfree(umem);

227 228 229 230 231
	/*
	 * We may be called with the mm's mmap_sem already held.  This
	 * can happen when a userspace munmap() is the call that drops
	 * the last reference to our file and calls our release
	 * method.  If there are memory regions to destroy, we'll end
232 233
	 * up here and not be able to take the mmap_sem.  In that case
	 * we defer the vm_locked accounting to the system workqueue.
234
	 */
235 236 237 238 239 240
	if (context->closing && !down_write_trylock(&mm->mmap_sem)) {
		work = kmalloc(sizeof *work, GFP_KERNEL);
		if (!work) {
			mmput(mm);
			return;
		}
241

242 243 244 245 246
		INIT_WORK(&work->work, ib_umem_account);
		work->mm   = mm;
		work->diff = diff;

		schedule_work(&work->work);
247
		return;
248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264
	} else
		down_write(&mm->mmap_sem);

	current->mm->locked_vm -= diff;
	up_write(&mm->mmap_sem);
	mmput(mm);
}
EXPORT_SYMBOL(ib_umem_release);

int ib_umem_page_count(struct ib_umem *umem)
{
	struct ib_umem_chunk *chunk;
	int shift;
	int i;
	int n;

	shift = ilog2(umem->page_size);
265

266 267 268 269
	n = 0;
	list_for_each_entry(chunk, &umem->chunk_list, list)
		for (i = 0; i < chunk->nmap; ++i)
			n += sg_dma_len(&chunk->page_list[i]) >> shift;
270

271
	return n;
272
}
273
EXPORT_SYMBOL(ib_umem_page_count);