[linux-block.git] / io_uring / openclose.c

// SPDX-License-Identifier: GPL-2.0
#include <linux/kernel.h>
#include <linux/errno.h>
#include <linux/fs.h>
#include <linux/file.h>
#include <linux/fdtable.h>
#include <linux/fsnotify.h>
#include <linux/namei.h>
#include <linux/io_uring.h>

#include <uapi/linux/io_uring.h>

#include "../fs/internal.h"

#include "io_uring.h"
#include "rsrc.h"
#include "openclose.h"

struct io_open {
	struct file			*file;
	int				dfd;
	u32				file_slot;
	struct filename			*filename;
	struct open_how			how;
	unsigned long			nofile;
};

struct io_close {
	struct file			*file;
	int				fd;
	u32				file_slot;
};

static bool io_openat_force_async(struct io_open *open)
{
	/*
	 * Don't bother trying for O_TRUNC, O_CREAT, or O_TMPFILE open,
	 * it'll always -EAGAIN
	 */
	return open->how.flags & (O_TRUNC | O_CREAT | O_TMPFILE);
}

static int __io_openat_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
{
	struct io_open *open = io_kiocb_to_cmd(req, struct io_open);
	const char __user *fname;
	int ret;

	if (unlikely(sqe->buf_index))
		return -EINVAL;
	if (unlikely(req->flags & REQ_F_FIXED_FILE))
		return -EBADF;

	/* open.how should be already initialised */
	if (!(open->how.flags & O_PATH) && force_o_largefile())
		open->how.flags |= O_LARGEFILE;

	open->dfd = READ_ONCE(sqe->fd);
	fname = u64_to_user_ptr(READ_ONCE(sqe->addr));
	open->filename = getname(fname);
	if (IS_ERR(open->filename)) {
		ret = PTR_ERR(open->filename);
		open->filename = NULL;
		return ret;
	}

	open->file_slot = READ_ONCE(sqe->file_index);
	if (open->file_slot && (open->how.flags & O_CLOEXEC))
		return -EINVAL;

	open->nofile = rlimit(RLIMIT_NOFILE);
	req->flags |= REQ_F_NEED_CLEANUP;
	if (io_openat_force_async(open))
		req->flags |= REQ_F_FORCE_ASYNC;
	return 0;
}

int io_openat_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
{
	struct io_open *open = io_kiocb_to_cmd(req, struct io_open);
	u64 mode = READ_ONCE(sqe->len);
	u64 flags = READ_ONCE(sqe->open_flags);

	open->how = build_open_how(flags, mode);
	return __io_openat_prep(req, sqe);
}

int io_openat2_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
{
	struct io_open *open = io_kiocb_to_cmd(req, struct io_open);
	struct open_how __user *how;
	size_t len;
	int ret;

	how = u64_to_user_ptr(READ_ONCE(sqe->addr2));
	len = READ_ONCE(sqe->len);
	if (len < OPEN_HOW_SIZE_VER0)
		return -EINVAL;

	ret = copy_struct_from_user(&open->how, sizeof(open->how), how, len);
	if (ret)
		return ret;

	return __io_openat_prep(req, sqe);
}

int io_openat2(struct io_kiocb *req, unsigned int issue_flags)
{
	struct io_open *open = io_kiocb_to_cmd(req, struct io_open);
	struct open_flags op;
	struct file *file;
	bool resolve_nonblock, nonblock_set;
	bool fixed = !!open->file_slot;
	int ret;

	ret = build_open_flags(&open->how, &op);
	if (ret)
		goto err;
	nonblock_set = op.open_flag & O_NONBLOCK;
	resolve_nonblock = open->how.resolve & RESOLVE_CACHED;
	if (issue_flags & IO_URING_F_NONBLOCK) {
		WARN_ON_ONCE(io_openat_force_async(open));
		op.lookup_flags |= LOOKUP_CACHED;
		op.open_flag |= O_NONBLOCK;
	}

	if (!fixed) {
		ret = __get_unused_fd_flags(open->how.flags, open->nofile);
		if (ret < 0)
			goto err;
	}

	file = do_filp_open(open->dfd, open->filename, &op);
	if (IS_ERR(file)) {
		/*
		 * We could hang on to this 'fd' on retrying, but seems like
		 * marginal gain for something that is now known to be a slower
		 * path. So just put it, and we'll get a new one when we retry.
		 */
		if (!fixed)
			put_unused_fd(ret);

		ret = PTR_ERR(file);
		/* only retry if RESOLVE_CACHED wasn't already set by application */
		if (ret == -EAGAIN &&
		    (!resolve_nonblock && (issue_flags & IO_URING_F_NONBLOCK)))
			return -EAGAIN;
		goto err;
	}

	if ((issue_flags & IO_URING_F_NONBLOCK) && !nonblock_set)
		file->f_flags &= ~O_NONBLOCK;

	if (!fixed)
		fd_install(ret, file);
	else
		ret = io_fixed_fd_install(req, issue_flags, file,
						open->file_slot);
err:
	putname(open->filename);
	req->flags &= ~REQ_F_NEED_CLEANUP;
	if (ret < 0)
		req_set_fail(req);
	io_req_set_res(req, ret, 0);
	return IOU_OK;
}

int io_openat(struct io_kiocb *req, unsigned int issue_flags)
{
	return io_openat2(req, issue_flags);
}

void io_open_cleanup(struct io_kiocb *req)
{
	struct io_open *open = io_kiocb_to_cmd(req, struct io_open);

	if (open->filename)
		putname(open->filename);
}

int __io_close_fixed(struct io_ring_ctx *ctx, unsigned int issue_flags,
		     unsigned int offset)
{
	int ret;

	io_ring_submit_lock(ctx, issue_flags);
	ret = io_fixed_fd_remove(ctx, offset);
	io_ring_submit_unlock(ctx, issue_flags);

	return ret;
}

static inline int io_close_fixed(struct io_kiocb *req, unsigned int issue_flags)
{
	struct io_close *close = io_kiocb_to_cmd(req, struct io_close);

	return __io_close_fixed(req->ctx, issue_flags, close->file_slot - 1);
}

int io_close_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
{
	struct io_close *close = io_kiocb_to_cmd(req, struct io_close);

	if (sqe->off || sqe->addr || sqe->len || sqe->rw_flags || sqe->buf_index)
		return -EINVAL;
	if (req->flags & REQ_F_FIXED_FILE)
		return -EBADF;

	close->fd = READ_ONCE(sqe->fd);
	close->file_slot = READ_ONCE(sqe->file_index);
	if (close->file_slot && close->fd)
		return -EINVAL;

	return 0;
}

int io_close(struct io_kiocb *req, unsigned int issue_flags)
{
	struct files_struct *files = current->files;
	struct io_close *close = io_kiocb_to_cmd(req, struct io_close);
	struct fdtable *fdt;
	struct file *file;
	int ret = -EBADF;

	if (close->file_slot) {
		ret = io_close_fixed(req, issue_flags);
		goto err;
	}

	spin_lock(&files->file_lock);
	fdt = files_fdtable(files);
	if (close->fd >= fdt->max_fds) {
		spin_unlock(&files->file_lock);
		goto err;
	}
	file = rcu_dereference_protected(fdt->fd[close->fd],
			lockdep_is_held(&files->file_lock));
	if (!file || io_is_uring_fops(file)) {
		spin_unlock(&files->file_lock);
		goto err;
	}

	/* if the file has a flush method, be safe and punt to async */
	if (file->f_op->flush && (issue_flags & IO_URING_F_NONBLOCK)) {
		spin_unlock(&files->file_lock);
		return -EAGAIN;
	}

	file = __close_fd_get_file(close->fd);
	spin_unlock(&files->file_lock);
	if (!file)
		goto err;

	/* No ->flush() or already async, safely close from here */
	ret = filp_close(file, current->files);
err:
	if (ret < 0)
		req_set_fail(req);
	io_req_set_res(req, ret, 0);
	return IOU_OK;
}
Commit	Line	Data
cd40cae2 JA	1	// SPDX-License-Identifier: GPL-2.0
	2	#include <linux/kernel.h>
	3	#include <linux/errno.h>
	4	#include <linux/fs.h>
	5	#include <linux/file.h>
	6	#include <linux/fdtable.h>
	7	#include <linux/fsnotify.h>
	8	#include <linux/namei.h>
	9	#include <linux/io_uring.h>
	10
	11	#include <uapi/linux/io_uring.h>
	12
	13	#include "../fs/internal.h"
	14
cd40cae2	15	#include "io_uring.h"
73572984	16	#include "rsrc.h"
cd40cae2 JA	17	#include "openclose.h"
	18
	19	struct io_open {
	20	struct file *file;
	21	int dfd;
	22	u32 file_slot;
	23	struct filename *filename;
	24	struct open_how how;
	25	unsigned long nofile;
	26	};
	27
	28	struct io_close {
	29	struct file *file;
	30	int fd;
	31	u32 file_slot;
	32	};
	33
0ffae640 DY	34	static bool io_openat_force_async(struct io_open *open)
	35	{
	36	/*
	37	* Don't bother trying for O_TRUNC, O_CREAT, or O_TMPFILE open,
	38	* it'll always -EAGAIN
	39	*/
	40	return open->how.flags & (O_TRUNC \| O_CREAT \| O_TMPFILE);
	41	}
	42
cd40cae2 JA	43	static int __io_openat_prep(struct io_kiocb req, const struct io_uring_sqe sqe)
cd40cae2 JA	44	{
f2ccb5ae	45	struct io_open *open = io_kiocb_to_cmd(req, struct io_open);
cd40cae2 JA	46	const char __user *fname;
	47	int ret;
	48
	49	if (unlikely(sqe->buf_index))
	50	return -EINVAL;
	51	if (unlikely(req->flags & REQ_F_FIXED_FILE))
	52	return -EBADF;
	53
	54	/* open.how should be already initialised */
	55	if (!(open->how.flags & O_PATH) && force_o_largefile())
	56	open->how.flags \|= O_LARGEFILE;
	57
	58	open->dfd = READ_ONCE(sqe->fd);
	59	fname = u64_to_user_ptr(READ_ONCE(sqe->addr));
	60	open->filename = getname(fname);
	61	if (IS_ERR(open->filename)) {
	62	ret = PTR_ERR(open->filename);
	63	open->filename = NULL;
	64	return ret;
	65	}
	66
	67	open->file_slot = READ_ONCE(sqe->file_index);
	68	if (open->file_slot && (open->how.flags & O_CLOEXEC))
	69	return -EINVAL;
	70
	71	open->nofile = rlimit(RLIMIT_NOFILE);
	72	req->flags \|= REQ_F_NEED_CLEANUP;
0ffae640 DY	73	if (io_openat_force_async(open))
0ffae640 DY	74	req->flags \|= REQ_F_FORCE_ASYNC;
cd40cae2 JA	75	return 0;
	76	}
	77
	78	int io_openat_prep(struct io_kiocb req, const struct io_uring_sqe sqe)
	79	{
f2ccb5ae	80	struct io_open *open = io_kiocb_to_cmd(req, struct io_open);
cd40cae2 JA	81	u64 mode = READ_ONCE(sqe->len);
	82	u64 flags = READ_ONCE(sqe->open_flags);
	83
	84	open->how = build_open_how(flags, mode);
	85	return __io_openat_prep(req, sqe);
	86	}
	87
	88	int io_openat2_prep(struct io_kiocb req, const struct io_uring_sqe sqe)
	89	{
f2ccb5ae	90	struct io_open *open = io_kiocb_to_cmd(req, struct io_open);
cd40cae2 JA	91	struct open_how __user *how;
	92	size_t len;
	93	int ret;
	94
	95	how = u64_to_user_ptr(READ_ONCE(sqe->addr2));
	96	len = READ_ONCE(sqe->len);
	97	if (len < OPEN_HOW_SIZE_VER0)
	98	return -EINVAL;
	99
	100	ret = copy_struct_from_user(&open->how, sizeof(open->how), how, len);
	101	if (ret)
	102	return ret;
	103
	104	return __io_openat_prep(req, sqe);
	105	}
	106
	107	int io_openat2(struct io_kiocb *req, unsigned int issue_flags)
	108	{
f2ccb5ae	109	struct io_open *open = io_kiocb_to_cmd(req, struct io_open);
cd40cae2 JA	110	struct open_flags op;
	111	struct file *file;
	112	bool resolve_nonblock, nonblock_set;
	113	bool fixed = !!open->file_slot;
	114	int ret;
	115
	116	ret = build_open_flags(&open->how, &op);
	117	if (ret)
	118	goto err;
	119	nonblock_set = op.open_flag & O_NONBLOCK;
	120	resolve_nonblock = open->how.resolve & RESOLVE_CACHED;
	121	if (issue_flags & IO_URING_F_NONBLOCK) {
0ffae640	122	WARN_ON_ONCE(io_openat_force_async(open));
cd40cae2 JA	123	op.lookup_flags \|= LOOKUP_CACHED;
	124	op.open_flag \|= O_NONBLOCK;
	125	}
	126
	127	if (!fixed) {
	128	ret = __get_unused_fd_flags(open->how.flags, open->nofile);
	129	if (ret < 0)
	130	goto err;
	131	}
	132
	133	file = do_filp_open(open->dfd, open->filename, &op);
	134	if (IS_ERR(file)) {
	135	/*
	136	* We could hang on to this 'fd' on retrying, but seems like
	137	* marginal gain for something that is now known to be a slower
	138	* path. So just put it, and we'll get a new one when we retry.
	139	*/
	140	if (!fixed)
	141	put_unused_fd(ret);
	142
	143	ret = PTR_ERR(file);
	144	/* only retry if RESOLVE_CACHED wasn't already set by application */
	145	if (ret == -EAGAIN &&
	146	(!resolve_nonblock && (issue_flags & IO_URING_F_NONBLOCK)))
	147	return -EAGAIN;
	148	goto err;
	149	}
	150
	151	if ((issue_flags & IO_URING_F_NONBLOCK) && !nonblock_set)
	152	file->f_flags &= ~O_NONBLOCK;
cd40cae2 JA	153
	154	if (!fixed)
	155	fd_install(ret, file);
	156	else
	157	ret = io_fixed_fd_install(req, issue_flags, file,
	158	open->file_slot);
	159	err:
	160	putname(open->filename);
	161	req->flags &= ~REQ_F_NEED_CLEANUP;
	162	if (ret < 0)
	163	req_set_fail(req);
	164	io_req_set_res(req, ret, 0);
	165	return IOU_OK;
	166	}
	167
	168	int io_openat(struct io_kiocb *req, unsigned int issue_flags)
	169	{
	170	return io_openat2(req, issue_flags);
	171	}
	172
	173	void io_open_cleanup(struct io_kiocb *req)
	174	{
f2ccb5ae	175	struct io_open *open = io_kiocb_to_cmd(req, struct io_open);
cd40cae2 JA	176
	177	if (open->filename)
	178	putname(open->filename);
	179	}
	180
f110ed84	181	int __io_close_fixed(struct io_ring_ctx *ctx, unsigned int issue_flags,
cd40cae2 JA	182	unsigned int offset)
cd40cae2 JA	183	{
cd40cae2 JA	184	int ret;
	185
	186	io_ring_submit_lock(ctx, issue_flags);
f110ed84	187	ret = io_fixed_fd_remove(ctx, offset);
cd40cae2	188	io_ring_submit_unlock(ctx, issue_flags);
f110ed84	189
cd40cae2 JA	190	return ret;
	191	}
	192
	193	static inline int io_close_fixed(struct io_kiocb *req, unsigned int issue_flags)
	194	{
f2ccb5ae	195	struct io_close *close = io_kiocb_to_cmd(req, struct io_close);
cd40cae2	196
f110ed84	197	return __io_close_fixed(req->ctx, issue_flags, close->file_slot - 1);
cd40cae2 JA	198	}
	199
	200	int io_close_prep(struct io_kiocb req, const struct io_uring_sqe sqe)
	201	{
f2ccb5ae	202	struct io_close *close = io_kiocb_to_cmd(req, struct io_close);
cd40cae2 JA	203
	204	if (sqe->off \|\| sqe->addr \|\| sqe->len \|\| sqe->rw_flags \|\| sqe->buf_index)
	205	return -EINVAL;
	206	if (req->flags & REQ_F_FIXED_FILE)
	207	return -EBADF;
	208
	209	close->fd = READ_ONCE(sqe->fd);
	210	close->file_slot = READ_ONCE(sqe->file_index);
	211	if (close->file_slot && close->fd)
	212	return -EINVAL;
	213
	214	return 0;
	215	}
	216
	217	int io_close(struct io_kiocb *req, unsigned int issue_flags)
	218	{
	219	struct files_struct *files = current->files;
f2ccb5ae	220	struct io_close *close = io_kiocb_to_cmd(req, struct io_close);
cd40cae2 JA	221	struct fdtable *fdt;
	222	struct file *file;
	223	int ret = -EBADF;
	224
	225	if (close->file_slot) {
	226	ret = io_close_fixed(req, issue_flags);
	227	goto err;
	228	}
	229
	230	spin_lock(&files->file_lock);
	231	fdt = files_fdtable(files);
	232	if (close->fd >= fdt->max_fds) {
	233	spin_unlock(&files->file_lock);
	234	goto err;
	235	}
	236	file = rcu_dereference_protected(fdt->fd[close->fd],
	237	lockdep_is_held(&files->file_lock));
	238	if (!file \|\| io_is_uring_fops(file)) {
	239	spin_unlock(&files->file_lock);
	240	goto err;
	241	}
	242
	243	/* if the file has a flush method, be safe and punt to async */
	244	if (file->f_op->flush && (issue_flags & IO_URING_F_NONBLOCK)) {
	245	spin_unlock(&files->file_lock);
	246	return -EAGAIN;
	247	}
	248
	249	file = __close_fd_get_file(close->fd);
	250	spin_unlock(&files->file_lock);
	251	if (!file)
	252	goto err;
	253
	254	/* No ->flush() or already async, safely close from here */
	255	ret = filp_close(file, current->files);
	256	err:
	257	if (ret < 0)
	258	req_set_fail(req);
	259	io_req_set_res(req, ret, 0);
	260	return IOU_OK;
	261	}