[mirror_ubuntu-impish-kernel.git] / mm / memfd.c

/*
 * memfd_create system call and file sealing support
 *
 * Code was originally included in shmem.c, and broken out to facilitate
 * use by hugetlbfs as well as tmpfs.
 *
 * This file is released under the GPL.
 */

#include <linux/fs.h>
#include <linux/vfs.h>
#include <linux/pagemap.h>
#include <linux/file.h>
#include <linux/mm.h>
#include <linux/sched/signal.h>
#include <linux/khugepaged.h>
#include <linux/syscalls.h>
#include <linux/hugetlb.h>
#include <linux/shmem_fs.h>
#include <linux/memfd.h>
#include <uapi/linux/memfd.h>

/*
 * We need a tag: a new tag would expand every xa_node by 8 bytes,
 * so reuse a tag which we firmly believe is never set or cleared on tmpfs
 * or hugetlbfs because they are memory only filesystems.
 */
#define MEMFD_TAG_PINNED        PAGECACHE_TAG_TOWRITE
#define LAST_SCAN               4       /* about 150ms max */

static void memfd_tag_pins(struct xa_state *xas)
{
	struct page *page;
	unsigned int tagged = 0;

	lru_add_drain();

	xas_lock_irq(xas);
	xas_for_each(xas, page, ULONG_MAX) {
		if (xa_is_value(page))
			continue;
		page = find_subpage(page, xas->xa_index);
		if (page_count(page) - page_mapcount(page) > 1)
			xas_set_mark(xas, MEMFD_TAG_PINNED);

		if (++tagged % XA_CHECK_SCHED)
			continue;

		xas_pause(xas);
		xas_unlock_irq(xas);
		cond_resched();
		xas_lock_irq(xas);
	}
	xas_unlock_irq(xas);
}

/*
 * Setting SEAL_WRITE requires us to verify there's no pending writer. However,
 * via get_user_pages(), drivers might have some pending I/O without any active
 * user-space mappings (eg., direct-IO, AIO). Therefore, we look at all pages
 * and see whether it has an elevated ref-count. If so, we tag them and wait for
 * them to be dropped.
 * The caller must guarantee that no new user will acquire writable references
 * to those pages to avoid races.
 */
static int memfd_wait_for_pins(struct address_space *mapping)
{
	XA_STATE(xas, &mapping->i_pages, 0);
	struct page *page;
	int error, scan;

	memfd_tag_pins(&xas);

	error = 0;
	for (scan = 0; scan <= LAST_SCAN; scan++) {
		unsigned int tagged = 0;

		if (!xas_marked(&xas, MEMFD_TAG_PINNED))
			break;

		if (!scan)
			lru_add_drain_all();
		else if (schedule_timeout_killable((HZ << scan) / 200))
			scan = LAST_SCAN;

		xas_set(&xas, 0);
		xas_lock_irq(&xas);
		xas_for_each_marked(&xas, page, ULONG_MAX, MEMFD_TAG_PINNED) {
			bool clear = true;
			if (xa_is_value(page))
				continue;
			page = find_subpage(page, xas.xa_index);
			if (page_count(page) - page_mapcount(page) != 1) {
				/*
				 * On the last scan, we clean up all those tags
				 * we inserted; but make a note that we still
				 * found pages pinned.
				 */
				if (scan == LAST_SCAN)
					error = -EBUSY;
				else
					clear = false;
			}
			if (clear)
				xas_clear_mark(&xas, MEMFD_TAG_PINNED);
			if (++tagged % XA_CHECK_SCHED)
				continue;

			xas_pause(&xas);
			xas_unlock_irq(&xas);
			cond_resched();
			xas_lock_irq(&xas);
		}
		xas_unlock_irq(&xas);
	}

	return error;
}

static unsigned int *memfd_file_seals_ptr(struct file *file)
{
	if (shmem_file(file))
		return &SHMEM_I(file_inode(file))->seals;

#ifdef CONFIG_HUGETLBFS
	if (is_file_hugepages(file))
		return &HUGETLBFS_I(file_inode(file))->seals;
#endif

	return NULL;
}

#define F_ALL_SEALS (F_SEAL_SEAL | \
		     F_SEAL_SHRINK | \
		     F_SEAL_GROW | \
		     F_SEAL_WRITE | \
		     F_SEAL_FUTURE_WRITE)

static int memfd_add_seals(struct file *file, unsigned int seals)
{
	struct inode *inode = file_inode(file);
	unsigned int *file_seals;
	int error;

	/*
	 * SEALING
	 * Sealing allows multiple parties to share a tmpfs or hugetlbfs file
	 * but restrict access to a specific subset of file operations. Seals
	 * can only be added, but never removed. This way, mutually untrusted
	 * parties can share common memory regions with a well-defined policy.
	 * A malicious peer can thus never perform unwanted operations on a
	 * shared object.
	 *
	 * Seals are only supported on special tmpfs or hugetlbfs files and
	 * always affect the whole underlying inode. Once a seal is set, it
	 * may prevent some kinds of access to the file. Currently, the
	 * following seals are defined:
	 *   SEAL_SEAL: Prevent further seals from being set on this file
	 *   SEAL_SHRINK: Prevent the file from shrinking
	 *   SEAL_GROW: Prevent the file from growing
	 *   SEAL_WRITE: Prevent write access to the file
	 *
	 * As we don't require any trust relationship between two parties, we
	 * must prevent seals from being removed. Therefore, sealing a file
	 * only adds a given set of seals to the file, it never touches
	 * existing seals. Furthermore, the "setting seals"-operation can be
	 * sealed itself, which basically prevents any further seal from being
	 * added.
	 *
	 * Semantics of sealing are only defined on volatile files. Only
	 * anonymous tmpfs and hugetlbfs files support sealing. More
	 * importantly, seals are never written to disk. Therefore, there's
	 * no plan to support it on other file types.
	 */

	if (!(file->f_mode & FMODE_WRITE))
		return -EPERM;
	if (seals & ~(unsigned int)F_ALL_SEALS)
		return -EINVAL;

	inode_lock(inode);

	file_seals = memfd_file_seals_ptr(file);
	if (!file_seals) {
		error = -EINVAL;
		goto unlock;
	}

	if (*file_seals & F_SEAL_SEAL) {
		error = -EPERM;
		goto unlock;
	}

	if ((seals & F_SEAL_WRITE) && !(*file_seals & F_SEAL_WRITE)) {
		error = mapping_deny_writable(file->f_mapping);
		if (error)
			goto unlock;

		error = memfd_wait_for_pins(file->f_mapping);
		if (error) {
			mapping_allow_writable(file->f_mapping);
			goto unlock;
		}
	}

	*file_seals |= seals;
	error = 0;

unlock:
	inode_unlock(inode);
	return error;
}

static int memfd_get_seals(struct file *file)
{
	unsigned int *seals = memfd_file_seals_ptr(file);

	return seals ? *seals : -EINVAL;
}

long memfd_fcntl(struct file *file, unsigned int cmd, unsigned long arg)
{
	long error;

	switch (cmd) {
	case F_ADD_SEALS:
		/* disallow upper 32bit */
		if (arg > UINT_MAX)
			return -EINVAL;

		error = memfd_add_seals(file, arg);
		break;
	case F_GET_SEALS:
		error = memfd_get_seals(file);
		break;
	default:
		error = -EINVAL;
		break;
	}

	return error;
}

#define MFD_NAME_PREFIX "memfd:"
#define MFD_NAME_PREFIX_LEN (sizeof(MFD_NAME_PREFIX) - 1)
#define MFD_NAME_MAX_LEN (NAME_MAX - MFD_NAME_PREFIX_LEN)

#define MFD_ALL_FLAGS (MFD_CLOEXEC | MFD_ALLOW_SEALING | MFD_HUGETLB)

SYSCALL_DEFINE2(memfd_create,
		const char __user *, uname,
		unsigned int, flags)
{
	unsigned int *file_seals;
	struct file *file;
	int fd, error;
	char *name;
	long len;

	if (!(flags & MFD_HUGETLB)) {
		if (flags & ~(unsigned int)MFD_ALL_FLAGS)
			return -EINVAL;
	} else {
		/* Allow huge page size encoding in flags. */
		if (flags & ~(unsigned int)(MFD_ALL_FLAGS |
				(MFD_HUGE_MASK << MFD_HUGE_SHIFT)))
			return -EINVAL;
	}

	/* length includes terminating zero */
	len = strnlen_user(uname, MFD_NAME_MAX_LEN + 1);
	if (len <= 0)
		return -EFAULT;
	if (len > MFD_NAME_MAX_LEN + 1)
		return -EINVAL;

	name = kmalloc(len + MFD_NAME_PREFIX_LEN, GFP_KERNEL);
	if (!name)
		return -ENOMEM;

	strcpy(name, MFD_NAME_PREFIX);
	if (copy_from_user(&name[MFD_NAME_PREFIX_LEN], uname, len)) {
		error = -EFAULT;
		goto err_name;
	}

	/* terminating-zero may have changed after strnlen_user() returned */
	if (name[len + MFD_NAME_PREFIX_LEN - 1]) {
		error = -EFAULT;
		goto err_name;
	}

	fd = get_unused_fd_flags((flags & MFD_CLOEXEC) ? O_CLOEXEC : 0);
	if (fd < 0) {
		error = fd;
		goto err_name;
	}

	if (flags & MFD_HUGETLB) {
		struct user_struct *user = NULL;

		file = hugetlb_file_setup(name, 0, VM_NORESERVE, &user,
					HUGETLB_ANONHUGE_INODE,
					(flags >> MFD_HUGE_SHIFT) &
					MFD_HUGE_MASK);
	} else
		file = shmem_file_setup(name, 0, VM_NORESERVE);
	if (IS_ERR(file)) {
		error = PTR_ERR(file);
		goto err_fd;
	}
	file->f_mode |= FMODE_LSEEK | FMODE_PREAD | FMODE_PWRITE;
	file->f_flags |= O_LARGEFILE;

	if (flags & MFD_ALLOW_SEALING) {
		file_seals = memfd_file_seals_ptr(file);
		*file_seals &= ~F_SEAL_SEAL;
	}

	fd_install(fd, file);
	kfree(name);
	return fd;

err_fd:
	put_unused_fd(fd);
err_name:
	kfree(name);
	return error;
}
Commit	Line	Data
5d752600 MK	1	/*
	2	* memfd_create system call and file sealing support
	3	*
	4	* Code was originally included in shmem.c, and broken out to facilitate
	5	* use by hugetlbfs as well as tmpfs.
	6	*
	7	* This file is released under the GPL.
	8	*/
	9
	10	#include <linux/fs.h>
	11	#include <linux/vfs.h>
	12	#include <linux/pagemap.h>
	13	#include <linux/file.h>
	14	#include <linux/mm.h>
	15	#include <linux/sched/signal.h>
	16	#include <linux/khugepaged.h>
	17	#include <linux/syscalls.h>
	18	#include <linux/hugetlb.h>
	19	#include <linux/shmem_fs.h>
	20	#include <linux/memfd.h>
	21	#include <uapi/linux/memfd.h>
	22
	23	/*
2313216f	24	* We need a tag: a new tag would expand every xa_node by 8 bytes,
5d752600 MK	25	* so reuse a tag which we firmly believe is never set or cleared on tmpfs
	26	* or hugetlbfs because they are memory only filesystems.
	27	*/
	28	#define MEMFD_TAG_PINNED PAGECACHE_TAG_TOWRITE
	29	#define LAST_SCAN 4 /* about 150ms max */
	30
ef3038a5	31	static void memfd_tag_pins(struct xa_state *xas)
5d752600	32	{
5d752600	33	struct page *page;
ef3038a5	34	unsigned int tagged = 0;
5d752600 MK	35
5d752600 MK	36	lru_add_drain();
5d752600	37
ef3038a5 MW	38	xas_lock_irq(xas);
	39	xas_for_each(xas, page, ULONG_MAX) {
	40	if (xa_is_value(page))
	41	continue;
4101196b	42	page = find_subpage(page, xas->xa_index);
ef3038a5 MW	43	if (page_count(page) - page_mapcount(page) > 1)
ef3038a5 MW	44	xas_set_mark(xas, MEMFD_TAG_PINNED);
5d752600	45
ef3038a5 MW	46	if (++tagged % XA_CHECK_SCHED)
	47	continue;
	48
	49	xas_pause(xas);
	50	xas_unlock_irq(xas);
	51	cond_resched();
	52	xas_lock_irq(xas);
5d752600	53	}
ef3038a5	54	xas_unlock_irq(xas);
5d752600 MK	55	}
	56
	57	/*
	58	* Setting SEAL_WRITE requires us to verify there's no pending writer. However,
	59	* via get_user_pages(), drivers might have some pending I/O without any active
	60	* user-space mappings (eg., direct-IO, AIO). Therefore, we look at all pages
	61	* and see whether it has an elevated ref-count. If so, we tag them and wait for
	62	* them to be dropped.
	63	* The caller must guarantee that no new user will acquire writable references
	64	* to those pages to avoid races.
	65	*/
	66	static int memfd_wait_for_pins(struct address_space *mapping)
	67	{
2313216f	68	XA_STATE(xas, &mapping->i_pages, 0);
5d752600 MK	69	struct page *page;
	70	int error, scan;
	71
ef3038a5	72	memfd_tag_pins(&xas);
5d752600 MK	73
	74	error = 0;
	75	for (scan = 0; scan <= LAST_SCAN; scan++) {
2313216f MW	76	unsigned int tagged = 0;
	77
	78	if (!xas_marked(&xas, MEMFD_TAG_PINNED))
5d752600 MK	79	break;
	80
	81	if (!scan)
	82	lru_add_drain_all();
	83	else if (schedule_timeout_killable((HZ << scan) / 200))
	84	scan = LAST_SCAN;
	85
2313216f MW	86	xas_set(&xas, 0);
	87	xas_lock_irq(&xas);
	88	xas_for_each_marked(&xas, page, ULONG_MAX, MEMFD_TAG_PINNED) {
	89	bool clear = true;
	90	if (xa_is_value(page))
	91	continue;
4101196b	92	page = find_subpage(page, xas.xa_index);
2313216f	93	if (page_count(page) - page_mapcount(page) != 1) {
5d752600 MK	94	/*
	95	* On the last scan, we clean up all those tags
	96	* we inserted; but make a note that we still
	97	* found pages pinned.
	98	*/
2313216f MW	99	if (scan == LAST_SCAN)
	100	error = -EBUSY;
	101	else
	102	clear = false;
5d752600	103	}
2313216f MW	104	if (clear)
	105	xas_clear_mark(&xas, MEMFD_TAG_PINNED);
	106	if (++tagged % XA_CHECK_SCHED)
	107	continue;
5d752600	108
2313216f MW	109	xas_pause(&xas);
	110	xas_unlock_irq(&xas);
	111	cond_resched();
	112	xas_lock_irq(&xas);
5d752600	113	}
2313216f	114	xas_unlock_irq(&xas);
5d752600 MK	115	}
	116
	117	return error;
	118	}
	119
	120	static unsigned int memfd_file_seals_ptr(struct file file)
	121	{
	122	if (shmem_file(file))
	123	return &SHMEM_I(file_inode(file))->seals;
	124
	125	#ifdef CONFIG_HUGETLBFS
	126	if (is_file_hugepages(file))
	127	return &HUGETLBFS_I(file_inode(file))->seals;
	128	#endif
	129
	130	return NULL;
	131	}
	132
	133	#define F_ALL_SEALS (F_SEAL_SEAL \| \
	134	F_SEAL_SHRINK \| \
	135	F_SEAL_GROW \| \
ab3948f5 JFG	136	F_SEAL_WRITE \| \
ab3948f5 JFG	137	F_SEAL_FUTURE_WRITE)
5d752600 MK	138
	139	static int memfd_add_seals(struct file *file, unsigned int seals)
	140	{
	141	struct inode *inode = file_inode(file);
	142	unsigned int *file_seals;
	143	int error;
	144
	145	/*
	146	* SEALING
	147	* Sealing allows multiple parties to share a tmpfs or hugetlbfs file
	148	* but restrict access to a specific subset of file operations. Seals
	149	* can only be added, but never removed. This way, mutually untrusted
	150	* parties can share common memory regions with a well-defined policy.
	151	* A malicious peer can thus never perform unwanted operations on a
	152	* shared object.
	153	*
	154	* Seals are only supported on special tmpfs or hugetlbfs files and
	155	* always affect the whole underlying inode. Once a seal is set, it
	156	* may prevent some kinds of access to the file. Currently, the
	157	* following seals are defined:
	158	* SEAL_SEAL: Prevent further seals from being set on this file
	159	* SEAL_SHRINK: Prevent the file from shrinking
	160	* SEAL_GROW: Prevent the file from growing
	161	* SEAL_WRITE: Prevent write access to the file
	162	*
	163	* As we don't require any trust relationship between two parties, we
	164	* must prevent seals from being removed. Therefore, sealing a file
	165	* only adds a given set of seals to the file, it never touches
	166	* existing seals. Furthermore, the "setting seals"-operation can be
	167	* sealed itself, which basically prevents any further seal from being
	168	* added.
	169	*
	170	* Semantics of sealing are only defined on volatile files. Only
	171	* anonymous tmpfs and hugetlbfs files support sealing. More
	172	* importantly, seals are never written to disk. Therefore, there's
	173	* no plan to support it on other file types.
	174	*/
	175
	176	if (!(file->f_mode & FMODE_WRITE))
	177	return -EPERM;
	178	if (seals & ~(unsigned int)F_ALL_SEALS)
	179	return -EINVAL;
	180
	181	inode_lock(inode);
	182
	183	file_seals = memfd_file_seals_ptr(file);
	184	if (!file_seals) {
	185	error = -EINVAL;
	186	goto unlock;
	187	}
	188
	189	if (*file_seals & F_SEAL_SEAL) {
	190	error = -EPERM;
	191	goto unlock;
	192	}
	193
	194	if ((seals & F_SEAL_WRITE) && !(*file_seals & F_SEAL_WRITE)) {
	195	error = mapping_deny_writable(file->f_mapping);
	196	if (error)
	197	goto unlock;
	198
	199	error = memfd_wait_for_pins(file->f_mapping);
	200	if (error) {
	201	mapping_allow_writable(file->f_mapping);
202	goto unlock;
203	}
204	}
205
206	*file_seals \|= seals;
207	error = 0;
208
209	unlock:
210	inode_unlock(inode);
211	return error;
212	}
213
214	static int memfd_get_seals(struct file *file)
215	{
216	unsigned int *seals = memfd_file_seals_ptr(file);
217
218	return seals ? *seals : -EINVAL;
219	}
220
221	long memfd_fcntl(struct file *file, unsigned int cmd, unsigned long arg)
222	{
223	long error;
224
225	switch (cmd) {
226	case F_ADD_SEALS:
227	/* disallow upper 32bit */
228	if (arg > UINT_MAX)
229	return -EINVAL;
230
231	error = memfd_add_seals(file, arg);
232	break;
233	case F_GET_SEALS:
234	error = memfd_get_seals(file);
235	break;
236	default:
237	error = -EINVAL;
238	break;
239	}
240
241	return error;
242	}
243
244	#define MFD_NAME_PREFIX "memfd:"
245	#define MFD_NAME_PREFIX_LEN (sizeof(MFD_NAME_PREFIX) - 1)
246	#define MFD_NAME_MAX_LEN (NAME_MAX - MFD_NAME_PREFIX_LEN)
247
248	#define MFD_ALL_FLAGS (MFD_CLOEXEC \| MFD_ALLOW_SEALING \| MFD_HUGETLB)
249
250	SYSCALL_DEFINE2(memfd_create,
251	const char __user *, uname,
252	unsigned int, flags)
253	{
254	unsigned int *file_seals;
255	struct file *file;
256	int fd, error;
257	char *name;
258	long len;
259
260	if (!(flags & MFD_HUGETLB)) {
261	if (flags & ~(unsigned int)MFD_ALL_FLAGS)
262	return -EINVAL;
263	} else {
264	/* Allow huge page size encoding in flags. */
265	if (flags & ~(unsigned int)(MFD_ALL_FLAGS \|
266	(MFD_HUGE_MASK << MFD_HUGE_SHIFT)))
267	return -EINVAL;
268	}
269
270	/* length includes terminating zero */
271	len = strnlen_user(uname, MFD_NAME_MAX_LEN + 1);
272	if (len <= 0)
273	return -EFAULT;
274	if (len > MFD_NAME_MAX_LEN + 1)
275	return -EINVAL;
276
277	name = kmalloc(len + MFD_NAME_PREFIX_LEN, GFP_KERNEL);
278	if (!name)
279	return -ENOMEM;
280
281	strcpy(name, MFD_NAME_PREFIX);
282	if (copy_from_user(&name[MFD_NAME_PREFIX_LEN], uname, len)) {
283	error = -EFAULT;
284	goto err_name;
285	}
286
287	/* terminating-zero may have changed after strnlen_user() returned */
288	if (name[len + MFD_NAME_PREFIX_LEN - 1]) {
289	error = -EFAULT;
290	goto err_name;
291	}
292
293	fd = get_unused_fd_flags((flags & MFD_CLOEXEC) ? O_CLOEXEC : 0);
294	if (fd < 0) {
295	error = fd;
296	goto err_name;
297	}
298
299	if (flags & MFD_HUGETLB) {
300	struct user_struct *user = NULL;
301
302	file = hugetlb_file_setup(name, 0, VM_NORESERVE, &user,
303	HUGETLB_ANONHUGE_INODE,
304	(flags >> MFD_HUGE_SHIFT) &
305	MFD_HUGE_MASK);
306	} else
307	file = shmem_file_setup(name, 0, VM_NORESERVE);
308	if (IS_ERR(file)) {
309	error = PTR_ERR(file);
310	goto err_fd;
311	}
312	file->f_mode \|= FMODE_LSEEK \| FMODE_PREAD \| FMODE_PWRITE;
c9c554f2	313	file->f_flags \|= O_LARGEFILE;
5d752600 MK	314
	315	if (flags & MFD_ALLOW_SEALING) {
	316	file_seals = memfd_file_seals_ptr(file);
	317	*file_seals &= ~F_SEAL_SEAL;
	318	}
	319
	320	fd_install(fd, file);
	321	kfree(name);
	322	return fd;
	323
	324	err_fd:
	325	put_unused_fd(fd);
	326	err_name:
	327	kfree(name);
	328	return error;
	329	}