file.c 42.4 KB
Newer Older
1
#include <linux/ceph/ceph_debug.h>
S
Sage Weil 已提交
2

3
#include <linux/module.h>
S
Sage Weil 已提交
4
#include <linux/sched.h>
5
#include <linux/slab.h>
S
Sage Weil 已提交
6
#include <linux/file.h>
S
Sage Weil 已提交
7
#include <linux/mount.h>
S
Sage Weil 已提交
8 9
#include <linux/namei.h>
#include <linux/writeback.h>
L
Li Wang 已提交
10
#include <linux/falloc.h>
S
Sage Weil 已提交
11 12 13

#include "super.h"
#include "mds_client.h"
14
#include "cache.h"
S
Sage Weil 已提交
15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

/*
 * Ceph file operations
 *
 * Implement basic open/close functionality, and implement
 * read/write.
 *
 * We implement three modes of file I/O:
 *  - buffered uses the generic_file_aio_{read,write} helpers
 *
 *  - synchronous is used when there is multi-client read/write
 *    sharing, avoids the page cache, and synchronously waits for an
 *    ack from the OSD.
 *
 *  - direct io takes the variant of the sync path that references
 *    user pages directly.
 *
 * fsync() flushes and waits on dirty pages, but just queues metadata
 * for writeback: since the MDS can recover size and mtime there is no
 * need to wait for MDS acknowledgement.
 */

37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104
/*
 * Calculate the length sum of direct io vectors that can
 * be combined into one page vector.
 */
static size_t dio_get_pagev_size(const struct iov_iter *it)
{
    const struct iovec *iov = it->iov;
    const struct iovec *iovend = iov + it->nr_segs;
    size_t size;

    size = iov->iov_len - it->iov_offset;
    /*
     * An iov can be page vectored when both the current tail
     * and the next base are page aligned.
     */
    while (PAGE_ALIGNED((iov->iov_base + iov->iov_len)) &&
           (++iov < iovend && PAGE_ALIGNED((iov->iov_base)))) {
        size += iov->iov_len;
    }
    dout("dio_get_pagevlen len = %zu\n", size);
    return size;
}

/*
 * Allocate a page vector based on (@it, @nbytes).
 * The return value is the tuple describing a page vector,
 * that is (@pages, @page_align, @num_pages).
 */
static struct page **
dio_get_pages_alloc(const struct iov_iter *it, size_t nbytes,
		    size_t *page_align, int *num_pages)
{
	struct iov_iter tmp_it = *it;
	size_t align;
	struct page **pages;
	int ret = 0, idx, npages;

	align = (unsigned long)(it->iov->iov_base + it->iov_offset) &
		(PAGE_SIZE - 1);
	npages = calc_pages_for(align, nbytes);
	pages = kmalloc(sizeof(*pages) * npages, GFP_KERNEL);
	if (!pages) {
		pages = vmalloc(sizeof(*pages) * npages);
		if (!pages)
			return ERR_PTR(-ENOMEM);
	}

	for (idx = 0; idx < npages; ) {
		size_t start;
		ret = iov_iter_get_pages(&tmp_it, pages + idx, nbytes,
					 npages - idx, &start);
		if (ret < 0)
			goto fail;

		iov_iter_advance(&tmp_it, ret);
		nbytes -= ret;
		idx += (ret + start + PAGE_SIZE - 1) / PAGE_SIZE;
	}

	BUG_ON(nbytes != 0);
	*num_pages = npages;
	*page_align = align;
	dout("dio_get_pages_alloc: got %d pages align %zu\n", npages, align);
	return pages;
fail:
	ceph_put_page_vector(pages, idx, false);
	return ERR_PTR(ret);
}
S
Sage Weil 已提交
105 106 107 108 109 110 111 112

/*
 * Prepare an open request.  Preallocate ceph_cap to avoid an
 * inopportune ENOMEM later.
 */
static struct ceph_mds_request *
prepare_open_request(struct super_block *sb, int flags, int create_mode)
{
113 114
	struct ceph_fs_client *fsc = ceph_sb_to_client(sb);
	struct ceph_mds_client *mdsc = fsc->mdsc;
S
Sage Weil 已提交
115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142
	struct ceph_mds_request *req;
	int want_auth = USE_ANY_MDS;
	int op = (flags & O_CREAT) ? CEPH_MDS_OP_CREATE : CEPH_MDS_OP_OPEN;

	if (flags & (O_WRONLY|O_RDWR|O_CREAT|O_TRUNC))
		want_auth = USE_AUTH_MDS;

	req = ceph_mdsc_create_request(mdsc, op, want_auth);
	if (IS_ERR(req))
		goto out;
	req->r_fmode = ceph_flags_to_mode(flags);
	req->r_args.open.flags = cpu_to_le32(flags);
	req->r_args.open.mode = cpu_to_le32(create_mode);
out:
	return req;
}

/*
 * initialize private struct file data.
 * if we fail, clean up by dropping fmode reference on the ceph_inode
 */
static int ceph_init_file(struct inode *inode, struct file *file, int fmode)
{
	struct ceph_file_info *cf;
	int ret = 0;

	switch (inode->i_mode & S_IFMT) {
	case S_IFREG:
143 144
		ceph_fscache_register_inode_cookie(inode);
		ceph_fscache_file_set_cookie(inode, file);
S
Sage Weil 已提交
145 146 147
	case S_IFDIR:
		dout("init_file %p %p 0%o (regular)\n", inode, file,
		     inode->i_mode);
G
Geliang Tang 已提交
148
		cf = kmem_cache_zalloc(ceph_file_cachep, GFP_KERNEL);
S
Sage Weil 已提交
149 150 151 152 153 154
		if (cf == NULL) {
			ceph_put_fmode(ceph_inode(inode), fmode); /* clean up */
			return -ENOMEM;
		}
		cf->fmode = fmode;
		cf->next_offset = 2;
Y
Yan, Zheng 已提交
155
		cf->readdir_cache_idx = -1;
S
Sage Weil 已提交
156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181
		file->private_data = cf;
		BUG_ON(inode->i_fop->release != ceph_release);
		break;

	case S_IFLNK:
		dout("init_file %p %p 0%o (symlink)\n", inode, file,
		     inode->i_mode);
		ceph_put_fmode(ceph_inode(inode), fmode); /* clean up */
		break;

	default:
		dout("init_file %p %p 0%o (special)\n", inode, file,
		     inode->i_mode);
		/*
		 * we need to drop the open ref now, since we don't
		 * have .release set to ceph_release.
		 */
		ceph_put_fmode(ceph_inode(inode), fmode); /* clean up */
		BUG_ON(inode->i_fop->release == ceph_release);

		/* call the proper open fop */
		ret = inode->i_fop->open(inode, file);
	}
	return ret;
}

182 183 184 185 186 187 188 189 190 191 192 193 194
/*
 * try renew caps after session gets killed.
 */
int ceph_renew_caps(struct inode *inode)
{
	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
	struct ceph_inode_info *ci = ceph_inode(inode);
	struct ceph_mds_request *req;
	int err, flags, wanted;

	spin_lock(&ci->i_ceph_lock);
	wanted = __ceph_caps_file_wanted(ci);
	if (__ceph_is_any_real_caps(ci) &&
195
	    (!(wanted & CEPH_CAP_ANY_WR) || ci->i_auth_cap)) {
196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234
		int issued = __ceph_caps_issued(ci, NULL);
		spin_unlock(&ci->i_ceph_lock);
		dout("renew caps %p want %s issued %s updating mds_wanted\n",
		     inode, ceph_cap_string(wanted), ceph_cap_string(issued));
		ceph_check_caps(ci, 0, NULL);
		return 0;
	}
	spin_unlock(&ci->i_ceph_lock);

	flags = 0;
	if ((wanted & CEPH_CAP_FILE_RD) && (wanted & CEPH_CAP_FILE_WR))
		flags = O_RDWR;
	else if (wanted & CEPH_CAP_FILE_RD)
		flags = O_RDONLY;
	else if (wanted & CEPH_CAP_FILE_WR)
		flags = O_WRONLY;
#ifdef O_LAZY
	if (wanted & CEPH_CAP_FILE_LAZYIO)
		flags |= O_LAZY;
#endif

	req = prepare_open_request(inode->i_sb, flags, 0);
	if (IS_ERR(req)) {
		err = PTR_ERR(req);
		goto out;
	}

	req->r_inode = inode;
	ihold(inode);
	req->r_num_caps = 1;
	req->r_fmode = -1;

	err = ceph_mdsc_do_request(mdsc, NULL, req);
	ceph_mdsc_put_request(req);
out:
	dout("renew caps %p open result=%d\n", inode, err);
	return err < 0 ? err : 0;
}

S
Sage Weil 已提交
235 236 237 238 239 240 241 242 243
/*
 * If we already have the requisite capabilities, we can satisfy
 * the open request locally (no need to request new caps from the
 * MDS).  We do, however, need to inform the MDS (asynchronously)
 * if our wanted caps set expands.
 */
int ceph_open(struct inode *inode, struct file *file)
{
	struct ceph_inode_info *ci = ceph_inode(inode);
244 245
	struct ceph_fs_client *fsc = ceph_sb_to_client(inode->i_sb);
	struct ceph_mds_client *mdsc = fsc->mdsc;
S
Sage Weil 已提交
246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271
	struct ceph_mds_request *req;
	struct ceph_file_info *cf = file->private_data;
	int err;
	int flags, fmode, wanted;

	if (cf) {
		dout("open file %p is already opened\n", file);
		return 0;
	}

	/* filter out O_CREAT|O_EXCL; vfs did that already.  yuck. */
	flags = file->f_flags & ~(O_CREAT|O_EXCL);
	if (S_ISDIR(inode->i_mode))
		flags = O_DIRECTORY;  /* mds likes to know */

	dout("open inode %p ino %llx.%llx file %p flags %d (%d)\n", inode,
	     ceph_vinop(inode), file, flags, file->f_flags);
	fmode = ceph_flags_to_mode(flags);
	wanted = ceph_caps_for_mode(fmode);

	/* snapped files are read-only */
	if (ceph_snap(inode) != CEPH_NOSNAP && (file->f_mode & FMODE_WRITE))
		return -EROFS;

	/* trivially open snapdir */
	if (ceph_snap(inode) == CEPH_SNAPDIR) {
272
		spin_lock(&ci->i_ceph_lock);
S
Sage Weil 已提交
273
		__ceph_get_fmode(ci, fmode);
274
		spin_unlock(&ci->i_ceph_lock);
S
Sage Weil 已提交
275 276 277 278
		return ceph_init_file(inode, file, fmode);
	}

	/*
279 280
	 * No need to block if we have caps on the auth MDS (for
	 * write) or any MDS (for read).  Update wanted set
S
Sage Weil 已提交
281 282
	 * asynchronously.
	 */
283
	spin_lock(&ci->i_ceph_lock);
284 285
	if (__ceph_is_any_real_caps(ci) &&
	    (((fmode & CEPH_FILE_MODE_WR) == 0) || ci->i_auth_cap)) {
286
		int mds_wanted = __ceph_caps_mds_wanted(ci, true);
S
Sage Weil 已提交
287 288 289 290 291 292
		int issued = __ceph_caps_issued(ci, NULL);

		dout("open %p fmode %d want %s issued %s using existing\n",
		     inode, fmode, ceph_cap_string(wanted),
		     ceph_cap_string(issued));
		__ceph_get_fmode(ci, fmode);
293
		spin_unlock(&ci->i_ceph_lock);
S
Sage Weil 已提交
294 295 296 297 298 299 300 301 302 303 304

		/* adjust wanted? */
		if ((issued & wanted) != wanted &&
		    (mds_wanted & wanted) != wanted &&
		    ceph_snap(inode) != CEPH_SNAPDIR)
			ceph_check_caps(ci, 0, NULL);

		return ceph_init_file(inode, file, fmode);
	} else if (ceph_snap(inode) != CEPH_NOSNAP &&
		   (ci->i_snap_caps & wanted) == wanted) {
		__ceph_get_fmode(ci, fmode);
305
		spin_unlock(&ci->i_ceph_lock);
S
Sage Weil 已提交
306 307
		return ceph_init_file(inode, file, fmode);
	}
308

309
	spin_unlock(&ci->i_ceph_lock);
S
Sage Weil 已提交
310 311 312 313 314 315 316

	dout("open fmode %d wants %s\n", fmode, ceph_cap_string(wanted));
	req = prepare_open_request(inode->i_sb, flags, 0);
	if (IS_ERR(req)) {
		err = PTR_ERR(req);
		goto out;
	}
317 318
	req->r_inode = inode;
	ihold(inode);
319

S
Sage Weil 已提交
320
	req->r_num_caps = 1;
321
	err = ceph_mdsc_do_request(mdsc, NULL, req);
S
Sage Weil 已提交
322 323 324 325 326 327 328 329 330 331
	if (!err)
		err = ceph_init_file(inode, file, req->r_fmode);
	ceph_mdsc_put_request(req);
	dout("open result=%d on %llx.%llx\n", err, ceph_vinop(inode));
out:
	return err;
}


/*
S
Sage Weil 已提交
332 333
 * Do a lookup + open with a single request.  If we get a non-existent
 * file or symlink, return 1 so the VFS can retry.
S
Sage Weil 已提交
334
 */
S
Sage Weil 已提交
335
int ceph_atomic_open(struct inode *dir, struct dentry *dentry,
A
Al Viro 已提交
336
		     struct file *file, unsigned flags, umode_t mode,
A
Al Viro 已提交
337
		     int *opened)
S
Sage Weil 已提交
338
{
339 340
	struct ceph_fs_client *fsc = ceph_sb_to_client(dir->i_sb);
	struct ceph_mds_client *mdsc = fsc->mdsc;
S
Sage Weil 已提交
341
	struct ceph_mds_request *req;
S
Sage Weil 已提交
342
	struct dentry *dn;
343
	struct ceph_acls_info acls = {};
Y
Yan, Zheng 已提交
344
       int mask;
S
Sage Weil 已提交
345 346
	int err;

A
Al Viro 已提交
347 348
	dout("atomic_open %p dentry %p '%pd' %s flags %d mode 0%o\n",
	     dir, dentry, dentry,
S
Sage Weil 已提交
349 350 351 352 353
	     d_unhashed(dentry) ? "unhashed" : "hashed", flags, mode);

	if (dentry->d_name.len > NAME_MAX)
		return -ENAMETOOLONG;

354 355 356 357 358 359
	if (flags & O_CREAT) {
		err = ceph_pre_init_acls(dir, &mode, &acls);
		if (err < 0)
			return err;
	}

S
Sage Weil 已提交
360 361
	/* do the open */
	req = prepare_open_request(dir->i_sb, flags, mode);
362 363 364 365
	if (IS_ERR(req)) {
		err = PTR_ERR(req);
		goto out_acl;
	}
S
Sage Weil 已提交
366 367 368 369 370
	req->r_dentry = dget(dentry);
	req->r_num_caps = 2;
	if (flags & O_CREAT) {
		req->r_dentry_drop = CEPH_CAP_FILE_SHARED;
		req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
371 372 373 374
		if (acls.pagelist) {
			req->r_pagelist = acls.pagelist;
			acls.pagelist = NULL;
		}
S
Sage Weil 已提交
375
	}
Y
Yan, Zheng 已提交
376 377 378 379 380 381

       mask = CEPH_STAT_CAP_INODE | CEPH_CAP_AUTH_SHARED;
       if (ceph_security_xattr_wanted(dir))
               mask |= CEPH_CAP_XATTR_SHARED;
       req->r_args.open.mask = cpu_to_le32(mask);

382 383
	req->r_parent = dir;
	set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
384 385 386
	err = ceph_mdsc_do_request(mdsc,
				   (flags & (O_CREAT|O_TRUNC)) ? dir : NULL,
				   req);
Y
Yan, Zheng 已提交
387
	err = ceph_handle_snapdir(req, dentry, err);
388
	if (err)
389
		goto out_req;
390

J
Jianpeng Ma 已提交
391
	if ((flags & O_CREAT) && !req->r_reply_info.head->is_dentry)
S
Sage Weil 已提交
392
		err = ceph_handle_notrace_create(dir, dentry);
393

394
	if (d_in_lookup(dentry)) {
S
Sage Weil 已提交
395 396 397 398 399 400 401 402
		dn = ceph_finish_lookup(req, dentry, err);
		if (IS_ERR(dn))
			err = PTR_ERR(dn);
	} else {
		/* we were given a hashed negative dentry */
		dn = NULL;
	}
	if (err)
403
		goto out_req;
404
	if (dn || d_really_is_negative(dentry) || d_is_symlink(dentry)) {
S
Sage Weil 已提交
405 406 407 408 409
		/* make vfs retry on splice, ENOENT, or symlink */
		dout("atomic_open finish_no_open on dn %p\n", dn);
		err = finish_no_open(file, dn);
	} else {
		dout("atomic_open finish_open on dn %p\n", dn);
410
		if (req->r_op == CEPH_MDS_OP_CREATE && req->r_reply_info.has_create_ino) {
411
			ceph_init_inode_acls(d_inode(dentry), &acls);
412 413
			*opened |= FILE_CREATED;
		}
S
Sage Weil 已提交
414 415
		err = finish_open(file, dentry, ceph_open, opened);
	}
416
out_req:
417 418
	if (!req->r_err && req->r_target_inode)
		ceph_put_fmode(ceph_inode(req->r_target_inode), req->r_fmode);
S
Sage Weil 已提交
419
	ceph_mdsc_put_request(req);
420 421
out_acl:
	ceph_release_acls_info(&acls);
S
Sage Weil 已提交
422
	dout("atomic_open result=%d\n", err);
A
Al Viro 已提交
423
	return err;
S
Sage Weil 已提交
424 425 426 427 428 429 430 431 432 433 434 435 436 437
}

int ceph_release(struct inode *inode, struct file *file)
{
	struct ceph_inode_info *ci = ceph_inode(inode);
	struct ceph_file_info *cf = file->private_data;

	dout("release inode %p file %p\n", inode, file);
	ceph_put_fmode(ci, cf->fmode);
	if (cf->last_readdir)
		ceph_mdsc_put_request(cf->last_readdir);
	kfree(cf->last_name);
	kfree(cf->dir_info);
	kmem_cache_free(ceph_file_cachep, cf);
438 439

	/* wake up anyone waiting for caps on this inode */
440
	wake_up_all(&ci->i_cap_wq);
S
Sage Weil 已提交
441 442 443
	return 0;
}

Y
Yan, Zheng 已提交
444
enum {
Y
Yan, Zheng 已提交
445 446 447
	HAVE_RETRIED = 1,
	CHECK_EOF =    2,
	READ_INLINE =  3,
Y
Yan, Zheng 已提交
448 449
};

S
Sage Weil 已提交
450 451 452 453 454 455 456 457
/*
 * Read a range of bytes striped over one or more objects.  Iterate over
 * objects we stripe over.  (That's not atomic, but good enough for now.)
 *
 * If we get a short result from the OSD, check against i_size; we need to
 * only return a short read to the caller if we hit EOF.
 */
static int striped_read(struct inode *inode,
458
			u64 pos, u64 len,
459
			struct page **pages, int num_pages,
460
			int page_align, int *checkeof)
S
Sage Weil 已提交
461
{
462
	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
S
Sage Weil 已提交
463
	struct ceph_inode_info *ci = ceph_inode(inode);
464
	u64 this_len;
465
	loff_t i_size;
466 467
	int page_idx;
	int ret, read = 0;
S
Sage Weil 已提交
468 469 470 471 472 473
	bool hit_stripe, was_short;

	/*
	 * we may need to do multiple reads.  not atomic, unfortunately.
	 */
more:
474 475
	this_len = len;
	page_idx = (page_align + read) >> PAGE_SHIFT;
476
	ret = ceph_osdc_readpages(&fsc->client->osdc, ceph_vino(inode),
S
Sage Weil 已提交
477
				  &ci->i_layout, pos, &this_len,
478 479 480
				  ci->i_truncate_seq, ci->i_truncate_size,
				  pages + page_idx, num_pages - page_idx,
				  ((page_align + read) & ~PAGE_MASK));
S
Sage Weil 已提交
481 482
	if (ret == -ENOENT)
		ret = 0;
483
	hit_stripe = this_len < len;
484
	was_short = ret >= 0 && ret < this_len;
485
	dout("striped_read %llu~%llu (read %u) got %d%s%s\n", pos, len, read,
S
Sage Weil 已提交
486 487
	     ret, hit_stripe ? " HITSTRIPE" : "", was_short ? " SHORT" : "");

488
	i_size = i_size_read(inode);
489
	if (ret >= 0) {
490 491
		if (was_short && (pos + ret < i_size)) {
			int zlen = min(this_len - ret, i_size - pos - ret);
492
			int zoff = page_align + read + ret;
493
			dout(" zero gap %llu to %llu\n",
494
			     pos + ret, pos + ret + zlen);
495 496
			ceph_zero_page_vector_range(zoff, zlen, pages);
			ret += zlen;
S
Sage Weil 已提交
497
		}
498

499
		read += ret;
S
Sage Weil 已提交
500
		pos += ret;
501
		len -= ret;
S
Sage Weil 已提交
502

503
		/* hit stripe and need continue*/
504
		if (len && hit_stripe && pos < i_size)
S
Sage Weil 已提交
505 506 507
			goto more;
	}

508
	if (read > 0) {
509
		ret = read;
510
		/* did we bounce off eof? */
511
		if (pos + len > i_size)
Y
Yan, Zheng 已提交
512
			*checkeof = CHECK_EOF;
S
Sage Weil 已提交
513 514 515 516 517 518 519 520 521 522 523 524
	}

	dout("striped_read returns %d\n", ret);
	return ret;
}

/*
 * Completely synchronous read and write methods.  Direct from __user
 * buffer to osd, or directly to user pages (if O_DIRECT).
 *
 * If the read spans object boundary, just do multiple reads.
 */
525 526
static ssize_t ceph_sync_read(struct kiocb *iocb, struct iov_iter *to,
			      int *checkeof)
S
Sage Weil 已提交
527
{
528
	struct file *file = iocb->ki_filp;
A
Al Viro 已提交
529
	struct inode *inode = file_inode(file);
S
Sage Weil 已提交
530
	struct page **pages;
531
	u64 off = iocb->ki_pos;
532 533 534
	int num_pages;
	ssize_t ret;
	size_t len = iov_iter_count(to);
S
Sage Weil 已提交
535

536 537
	dout("sync_read on file %p %llu~%u %s\n", file, off,
	     (unsigned)len,
S
Sage Weil 已提交
538
	     (file->f_flags & O_DIRECT) ? "O_DIRECT" : "");
539 540 541

	if (!len)
		return 0;
542 543 544 545 546 547
	/*
	 * flush any page cache pages in this range.  this
	 * will make concurrent normal and sync io slow,
	 * but it will at least behave sensibly when they are
	 * in sequence.
	 */
548 549
	ret = filemap_write_and_wait_range(inode->i_mapping, off,
						off + len);
550
	if (ret < 0)
551
		return ret;
552

553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592
	if (unlikely(to->type & ITER_PIPE)) {
		size_t page_off;
		ret = iov_iter_get_pages_alloc(to, &pages, len,
					       &page_off);
		if (ret <= 0)
			return -ENOMEM;
		num_pages = DIV_ROUND_UP(ret + page_off, PAGE_SIZE);

		ret = striped_read(inode, off, ret, pages, num_pages,
				   page_off, checkeof);
		if (ret > 0) {
			iov_iter_advance(to, ret);
			off += ret;
		} else {
			iov_iter_advance(to, 0);
		}
		ceph_put_page_vector(pages, num_pages, false);
	} else {
		num_pages = calc_pages_for(off, len);
		pages = ceph_alloc_page_vector(num_pages, GFP_KERNEL);
		if (IS_ERR(pages))
			return PTR_ERR(pages);

		ret = striped_read(inode, off, len, pages, num_pages,
				   (off & ~PAGE_MASK), checkeof);
		if (ret > 0) {
			int l, k = 0;
			size_t left = ret;

			while (left) {
				size_t page_off = off & ~PAGE_MASK;
				size_t copy = min_t(size_t, left,
						    PAGE_SIZE - page_off);
				l = copy_page_to_iter(pages[k++], page_off,
						      copy, to);
				off += l;
				left -= l;
				if (l < copy)
					break;
			}
593
		}
594
		ceph_release_page_vector(pages, num_pages);
595
	}
S
Sage Weil 已提交
596

597 598 599 600
	if (off > iocb->ki_pos) {
		ret = off - iocb->ki_pos;
		iocb->ki_pos = off;
	}
S
Sage Weil 已提交
601

602
	dout("sync_read result %zd\n", ret);
S
Sage Weil 已提交
603 604 605
	return ret;
}

Y
Yan, Zheng 已提交
606 607 608 609 610 611 612 613
struct ceph_aio_request {
	struct kiocb *iocb;
	size_t total_len;
	int write;
	int error;
	struct list_head osd_reqs;
	unsigned num_reqs;
	atomic_t pending_reqs;
614
	struct timespec mtime;
Y
Yan, Zheng 已提交
615 616 617
	struct ceph_cap_flush *prealloc_cf;
};

618 619 620 621 622 623 624
struct ceph_aio_work {
	struct work_struct work;
	struct ceph_osd_request *req;
};

static void ceph_aio_retry_work(struct work_struct *work);

Y
Yan, Zheng 已提交
625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667
static void ceph_aio_complete(struct inode *inode,
			      struct ceph_aio_request *aio_req)
{
	struct ceph_inode_info *ci = ceph_inode(inode);
	int ret;

	if (!atomic_dec_and_test(&aio_req->pending_reqs))
		return;

	ret = aio_req->error;
	if (!ret)
		ret = aio_req->total_len;

	dout("ceph_aio_complete %p rc %d\n", inode, ret);

	if (ret >= 0 && aio_req->write) {
		int dirty;

		loff_t endoff = aio_req->iocb->ki_pos + aio_req->total_len;
		if (endoff > i_size_read(inode)) {
			if (ceph_inode_set_size(inode, endoff))
				ceph_check_caps(ci, CHECK_CAPS_AUTHONLY, NULL);
		}

		spin_lock(&ci->i_ceph_lock);
		ci->i_inline_version = CEPH_INLINE_NONE;
		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR,
					       &aio_req->prealloc_cf);
		spin_unlock(&ci->i_ceph_lock);
		if (dirty)
			__mark_inode_dirty(inode, dirty);

	}

	ceph_put_cap_refs(ci, (aio_req->write ? CEPH_CAP_FILE_WR :
						CEPH_CAP_FILE_RD));

	aio_req->iocb->ki_complete(aio_req->iocb, ret, 0);

	ceph_free_cap_flush(aio_req->prealloc_cf);
	kfree(aio_req);
}

668
static void ceph_aio_complete_req(struct ceph_osd_request *req)
Y
Yan, Zheng 已提交
669 670 671 672 673 674 675 676 677 678 679 680
{
	int rc = req->r_result;
	struct inode *inode = req->r_inode;
	struct ceph_aio_request *aio_req = req->r_priv;
	struct ceph_osd_data *osd_data = osd_req_op_extent_osd_data(req, 0);
	int num_pages = calc_pages_for((u64)osd_data->alignment,
				       osd_data->length);

	dout("ceph_aio_complete_req %p rc %d bytes %llu\n",
	     inode, rc, osd_data->length);

	if (rc == -EOLDSNAPC) {
681 682 683 684 685 686 687 688 689 690 691 692 693
		struct ceph_aio_work *aio_work;
		BUG_ON(!aio_req->write);

		aio_work = kmalloc(sizeof(*aio_work), GFP_NOFS);
		if (aio_work) {
			INIT_WORK(&aio_work->work, ceph_aio_retry_work);
			aio_work->req = req;
			queue_work(ceph_inode_to_client(inode)->wb_wq,
				   &aio_work->work);
			return;
		}
		rc = -ENOMEM;
	} else if (!aio_req->write) {
Y
Yan, Zheng 已提交
694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718
		if (rc == -ENOENT)
			rc = 0;
		if (rc >= 0 && osd_data->length > rc) {
			int zoff = osd_data->alignment + rc;
			int zlen = osd_data->length - rc;
			/*
			 * If read is satisfied by single OSD request,
			 * it can pass EOF. Otherwise read is within
			 * i_size.
			 */
			if (aio_req->num_reqs == 1) {
				loff_t i_size = i_size_read(inode);
				loff_t endoff = aio_req->iocb->ki_pos + rc;
				if (endoff < i_size)
					zlen = min_t(size_t, zlen,
						     i_size - endoff);
				aio_req->total_len = rc + zlen;
			}

			if (zlen > 0)
				ceph_zero_page_vector_range(zoff, zlen,
							    osd_data->pages);
		}
	}

719
	ceph_put_page_vector(osd_data->pages, num_pages, !aio_req->write);
Y
Yan, Zheng 已提交
720 721 722 723 724 725 726 727 728
	ceph_osdc_put_request(req);

	if (rc < 0)
		cmpxchg(&aio_req->error, 0, rc);

	ceph_aio_complete(inode, aio_req);
	return;
}

729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755
static void ceph_aio_retry_work(struct work_struct *work)
{
	struct ceph_aio_work *aio_work =
		container_of(work, struct ceph_aio_work, work);
	struct ceph_osd_request *orig_req = aio_work->req;
	struct ceph_aio_request *aio_req = orig_req->r_priv;
	struct inode *inode = orig_req->r_inode;
	struct ceph_inode_info *ci = ceph_inode(inode);
	struct ceph_snap_context *snapc;
	struct ceph_osd_request *req;
	int ret;

	spin_lock(&ci->i_ceph_lock);
	if (__ceph_have_pending_cap_snap(ci)) {
		struct ceph_cap_snap *capsnap =
			list_last_entry(&ci->i_cap_snaps,
					struct ceph_cap_snap,
					ci_item);
		snapc = ceph_get_snap_context(capsnap->context);
	} else {
		BUG_ON(!ci->i_head_snapc);
		snapc = ceph_get_snap_context(ci->i_head_snapc);
	}
	spin_unlock(&ci->i_ceph_lock);

	req = ceph_osdc_alloc_request(orig_req->r_osdc, snapc, 2,
			false, GFP_NOFS);
756 757
	if (!req) {
		ret = -ENOMEM;
758 759 760 761
		req = orig_req;
		goto out;
	}

762
	req->r_flags = CEPH_OSD_FLAG_ORDERSNAP | CEPH_OSD_FLAG_WRITE;
763
	ceph_oloc_copy(&req->r_base_oloc, &orig_req->r_base_oloc);
764
	ceph_oid_copy(&req->r_base_oid, &orig_req->r_base_oid);
765

766 767 768 769 770 771
	ret = ceph_osdc_alloc_messages(req, GFP_NOFS);
	if (ret) {
		ceph_osdc_put_request(req);
		req = orig_req;
		goto out;
	}
772 773 774 775

	req->r_ops[0] = orig_req->r_ops[0];
	osd_req_op_init(req, 1, CEPH_OSD_OP_STARTSYNC, 0);

776 777
	req->r_mtime = aio_req->mtime;
	req->r_data_offset = req->r_ops[0].extent.offset;
778 779 780 781 782 783 784 785 786 787 788

	ceph_osdc_put_request(orig_req);

	req->r_callback = ceph_aio_complete_req;
	req->r_inode = inode;
	req->r_priv = aio_req;

	ret = ceph_osdc_start_request(req->r_osdc, req, false);
out:
	if (ret < 0) {
		req->r_result = ret;
789
		ceph_aio_complete_req(req);
790 791
	}

792
	ceph_put_snap_context(snapc);
793 794 795
	kfree(aio_work);
}

796
static ssize_t
Y
Yan, Zheng 已提交
797 798 799
ceph_direct_read_write(struct kiocb *iocb, struct iov_iter *iter,
		       struct ceph_snap_context *snapc,
		       struct ceph_cap_flush **pcf)
S
Sage Weil 已提交
800
{
801
	struct file *file = iocb->ki_filp;
A
Al Viro 已提交
802
	struct inode *inode = file_inode(file);
S
Sage Weil 已提交
803
	struct ceph_inode_info *ci = ceph_inode(inode);
804
	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
805
	struct ceph_vino vino;
S
Sage Weil 已提交
806 807
	struct ceph_osd_request *req;
	struct page **pages;
Y
Yan, Zheng 已提交
808 809
	struct ceph_aio_request *aio_req = NULL;
	int num_pages = 0;
S
Sage Weil 已提交
810 811
	int flags;
	int ret;
812
	struct timespec mtime = current_time(inode);
Y
Yan, Zheng 已提交
813 814 815
	size_t count = iov_iter_count(iter);
	loff_t pos = iocb->ki_pos;
	bool write = iov_iter_rw(iter) == WRITE;
S
Sage Weil 已提交
816

Y
Yan, Zheng 已提交
817
	if (write && ceph_snap(file_inode(file)) != CEPH_NOSNAP)
S
Sage Weil 已提交
818 819
		return -EROFS;

Y
Yan, Zheng 已提交
820 821
	dout("sync_direct_read_write (%s) on file %p %lld~%u\n",
	     (write ? "write" : "read"), file, pos, (unsigned)count);
S
Sage Weil 已提交
822

823
	ret = filemap_write_and_wait_range(inode->i_mapping, pos, pos + count);
824 825 826
	if (ret < 0)
		return ret;

Y
Yan, Zheng 已提交
827
	if (write) {
828
		int ret2 = invalidate_inode_pages2_range(inode->i_mapping,
829 830
					pos >> PAGE_SHIFT,
					(pos + count) >> PAGE_SHIFT);
831
		if (ret2 < 0)
832
			dout("invalidate_inode_pages2_range returned %d\n", ret2);
833

834
		flags = CEPH_OSD_FLAG_ORDERSNAP | CEPH_OSD_FLAG_WRITE;
Y
Yan, Zheng 已提交
835 836 837
	} else {
		flags = CEPH_OSD_FLAG_READ;
	}
S
Sage Weil 已提交
838

Y
Yan, Zheng 已提交
839 840 841 842
	while (iov_iter_count(iter) > 0) {
		u64 size = dio_get_pagev_size(iter);
		size_t start = 0;
		ssize_t len;
843 844 845

		vino = ceph_vino(inode);
		req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
Y
Yan, Zheng 已提交
846 847 848 849 850 851
					    vino, pos, &size, 0,
					    /*include a 'startsync' command*/
					    write ? 2 : 1,
					    write ? CEPH_OSD_OP_WRITE :
						    CEPH_OSD_OP_READ,
					    flags, snapc,
852 853 854 855 856
					    ci->i_truncate_seq,
					    ci->i_truncate_size,
					    false);
		if (IS_ERR(req)) {
			ret = PTR_ERR(req);
857
			break;
858
		}
S
Sage Weil 已提交
859

Y
Yan, Zheng 已提交
860 861
		len = size;
		pages = dio_get_pages_alloc(iter, len, &start, &num_pages);
862
		if (IS_ERR(pages)) {
863
			ceph_osdc_put_request(req);
864
			ret = PTR_ERR(pages);
865
			break;
S
Sage Weil 已提交
866 867 868
		}

		/*
Y
Yan, Zheng 已提交
869 870
		 * To simplify error handling, allow AIO when IO within i_size
		 * or IO can be satisfied by single OSD request.
S
Sage Weil 已提交
871
		 */
Y
Yan, Zheng 已提交
872 873 874 875 876 877 878 879
		if (pos == iocb->ki_pos && !is_sync_kiocb(iocb) &&
		    (len == count || pos + count <= i_size_read(inode))) {
			aio_req = kzalloc(sizeof(*aio_req), GFP_KERNEL);
			if (aio_req) {
				aio_req->iocb = iocb;
				aio_req->write = write;
				INIT_LIST_HEAD(&aio_req->osd_reqs);
				if (write) {
880
					aio_req->mtime = mtime;
Y
Yan, Zheng 已提交
881 882 883 884 885 886 887 888 889 890 891 892
					swap(aio_req->prealloc_cf, *pcf);
				}
			}
			/* ignore error */
		}

		if (write) {
			/*
			 * throw out any page cache pages in this range. this
			 * may block.
			 */
			truncate_inode_pages_range(inode->i_mapping, pos,
893
					(pos+len) | (PAGE_SIZE - 1));
Y
Yan, Zheng 已提交
894 895

			osd_req_op_init(req, 1, CEPH_OSD_OP_STARTSYNC, 0);
896
			req->r_mtime = mtime;
Y
Yan, Zheng 已提交
897 898 899 900
		}

		osd_req_op_extent_osd_data_pages(req, 0, pages, len, start,
						 false, false);
901

Y
Yan, Zheng 已提交
902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917
		if (aio_req) {
			aio_req->total_len += len;
			aio_req->num_reqs++;
			atomic_inc(&aio_req->pending_reqs);

			req->r_callback = ceph_aio_complete_req;
			req->r_inode = inode;
			req->r_priv = aio_req;
			list_add_tail(&req->r_unsafe_item, &aio_req->osd_reqs);

			pos += len;
			iov_iter_advance(iter, len);
			continue;
		}

		ret = ceph_osdc_start_request(req->r_osdc, req, false);
918 919 920
		if (!ret)
			ret = ceph_osdc_wait_request(&fsc->client->osdc, req);

Y
Yan, Zheng 已提交
921 922 923 924 925 926 927 928 929 930 931 932 933 934 935
		size = i_size_read(inode);
		if (!write) {
			if (ret == -ENOENT)
				ret = 0;
			if (ret >= 0 && ret < len && pos + ret < size) {
				int zlen = min_t(size_t, len - ret,
						 size - pos - ret);
				ceph_zero_page_vector_range(start + ret, zlen,
							    pages);
				ret += zlen;
			}
			if (ret >= 0)
				len = ret;
		}

936
		ceph_put_page_vector(pages, num_pages, !write);
937 938

		ceph_osdc_put_request(req);
Y
Yan, Zheng 已提交
939
		if (ret < 0)
940
			break;
941

Y
Yan, Zheng 已提交
942 943 944 945
		pos += len;
		iov_iter_advance(iter, len);

		if (!write && pos >= size)
946
			break;
947

Y
Yan, Zheng 已提交
948 949
		if (write && pos > size) {
			if (ceph_inode_set_size(inode, pos))
950 951 952 953
				ceph_check_caps(ceph_inode(inode),
						CHECK_CAPS_AUTHONLY,
						NULL);
		}
954 955
	}

Y
Yan, Zheng 已提交
956
	if (aio_req) {
957 958
		LIST_HEAD(osd_reqs);

Y
Yan, Zheng 已提交
959 960 961 962 963 964 965 966
		if (aio_req->num_reqs == 0) {
			kfree(aio_req);
			return ret;
		}

		ceph_get_cap_refs(ci, write ? CEPH_CAP_FILE_WR :
					      CEPH_CAP_FILE_RD);

967 968 969
		list_splice(&aio_req->osd_reqs, &osd_reqs);
		while (!list_empty(&osd_reqs)) {
			req = list_first_entry(&osd_reqs,
Y
Yan, Zheng 已提交
970 971 972 973 974 975 976 977
					       struct ceph_osd_request,
					       r_unsafe_item);
			list_del_init(&req->r_unsafe_item);
			if (ret >= 0)
				ret = ceph_osdc_start_request(req->r_osdc,
							      req, false);
			if (ret < 0) {
				req->r_result = ret;
978
				ceph_aio_complete_req(req);
Y
Yan, Zheng 已提交
979 980 981 982 983 984 985
			}
		}
		return -EIOCBQUEUED;
	}

	if (ret != -EOLDSNAPC && pos > iocb->ki_pos) {
		ret = pos - iocb->ki_pos;
986 987 988 989 990 991 992 993 994 995 996 997
		iocb->ki_pos = pos;
	}
	return ret;
}

/*
 * Synchronous write, straight from __user pointer or user pages.
 *
 * If write spans object boundary, just do multiple writes.  (For a
 * correct atomic write, we should e.g. take write locks on all
 * objects, rollback on failure, etc.)
 */
Y
Yan, Zheng 已提交
998
static ssize_t
999 1000
ceph_sync_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos,
		struct ceph_snap_context *snapc)
1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014
{
	struct file *file = iocb->ki_filp;
	struct inode *inode = file_inode(file);
	struct ceph_inode_info *ci = ceph_inode(inode);
	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
	struct ceph_vino vino;
	struct ceph_osd_request *req;
	struct page **pages;
	u64 len;
	int num_pages;
	int written = 0;
	int flags;
	int check_caps = 0;
	int ret;
1015
	struct timespec mtime = current_time(inode);
A
Al Viro 已提交
1016
	size_t count = iov_iter_count(from);
1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027

	if (ceph_snap(file_inode(file)) != CEPH_NOSNAP)
		return -EROFS;

	dout("sync_write on file %p %lld~%u\n", file, pos, (unsigned)count);

	ret = filemap_write_and_wait_range(inode->i_mapping, pos, pos + count);
	if (ret < 0)
		return ret;

	ret = invalidate_inode_pages2_range(inode->i_mapping,
1028 1029
					    pos >> PAGE_SHIFT,
					    (pos + count) >> PAGE_SHIFT);
1030 1031 1032
	if (ret < 0)
		dout("invalidate_inode_pages2_range returned %d\n", ret);

1033
	flags = CEPH_OSD_FLAG_ORDERSNAP | CEPH_OSD_FLAG_WRITE;
1034

A
Al Viro 已提交
1035
	while ((len = iov_iter_count(from)) > 0) {
1036 1037 1038 1039 1040
		size_t left;
		int n;

		vino = ceph_vino(inode);
		req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
1041
					    vino, pos, &len, 0, 1,
1042 1043 1044 1045 1046 1047
					    CEPH_OSD_OP_WRITE, flags, snapc,
					    ci->i_truncate_seq,
					    ci->i_truncate_size,
					    false);
		if (IS_ERR(req)) {
			ret = PTR_ERR(req);
1048
			break;
1049 1050 1051 1052 1053 1054
		}

		/*
		 * write from beginning of first page,
		 * regardless of io alignment
		 */
1055
		num_pages = (len + PAGE_SIZE - 1) >> PAGE_SHIFT;
1056

1057
		pages = ceph_alloc_page_vector(num_pages, GFP_KERNEL);
S
Sage Weil 已提交
1058 1059 1060 1061
		if (IS_ERR(pages)) {
			ret = PTR_ERR(pages);
			goto out;
		}
1062 1063 1064

		left = len;
		for (n = 0; n < num_pages; n++) {
1065
			size_t plen = min_t(size_t, left, PAGE_SIZE);
A
Al Viro 已提交
1066
			ret = copy_page_from_iter(pages[n], 0, plen, from);
1067 1068 1069 1070 1071 1072 1073
			if (ret != plen) {
				ret = -EFAULT;
				break;
			}
			left -= ret;
		}

S
Sage Weil 已提交
1074 1075 1076 1077 1078
		if (ret < 0) {
			ceph_release_page_vector(pages, num_pages);
			goto out;
		}

1079
		req->r_inode = inode;
S
Sage Weil 已提交
1080

1081 1082
		osd_req_op_extent_osd_data_pages(req, 0, pages, len, 0,
						false, true);
1083

1084
		req->r_mtime = mtime;
1085 1086 1087
		ret = ceph_osdc_start_request(&fsc->client->osdc, req, false);
		if (!ret)
			ret = ceph_osdc_wait_request(&fsc->client->osdc, req);
S
Sage Weil 已提交
1088 1089

out:
1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104
		ceph_osdc_put_request(req);
		if (ret == 0) {
			pos += len;
			written += len;

			if (pos > i_size_read(inode)) {
				check_caps = ceph_inode_set_size(inode, pos);
				if (check_caps)
					ceph_check_caps(ceph_inode(inode),
							CHECK_CAPS_AUTHONLY,
							NULL);
			}
		} else
			break;
	}
S
Sage Weil 已提交
1105

1106
	if (ret != -EOLDSNAPC && written > 0) {
S
Sage Weil 已提交
1107
		ret = written;
1108
		iocb->ki_pos = pos;
S
Sage Weil 已提交
1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119
	}
	return ret;
}

/*
 * Wrap generic_file_aio_read with checks for cap bits on the inode.
 * Atomically grab references, so that those bits are not released
 * back to the MDS mid-read.
 *
 * Hmm, the sync read case isn't actually async... should it be?
 */
A
Al Viro 已提交
1120
static ssize_t ceph_read_iter(struct kiocb *iocb, struct iov_iter *to)
S
Sage Weil 已提交
1121 1122
{
	struct file *filp = iocb->ki_filp;
1123
	struct ceph_file_info *fi = filp->private_data;
C
Christoph Hellwig 已提交
1124
	size_t len = iov_iter_count(to);
A
Al Viro 已提交
1125
	struct inode *inode = file_inode(filp);
S
Sage Weil 已提交
1126
	struct ceph_inode_info *ci = ceph_inode(inode);
1127
	struct page *pinned_page = NULL;
S
Sage Weil 已提交
1128
	ssize_t ret;
1129
	int want, got = 0;
Y
Yan, Zheng 已提交
1130
	int retry_op = 0, read = 0;
S
Sage Weil 已提交
1131

1132
again:
1133 1134 1135
	dout("aio_read %p %llx.%llx %llu~%u trying to get caps on %p\n",
	     inode, ceph_vinop(inode), iocb->ki_pos, (unsigned)len, inode);

1136 1137 1138 1139
	if (fi->fmode & CEPH_FILE_MODE_LAZY)
		want = CEPH_CAP_FILE_CACHE | CEPH_CAP_FILE_LAZYIO;
	else
		want = CEPH_CAP_FILE_CACHE;
1140
	ret = ceph_get_caps(ci, CEPH_CAP_FILE_RD, want, -1, &got, &pinned_page);
S
Sage Weil 已提交
1141
	if (ret < 0)
1142
		return ret;
S
Sage Weil 已提交
1143

1144
	if ((got & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) == 0 ||
1145
	    (iocb->ki_flags & IOCB_DIRECT) ||
1146 1147 1148 1149 1150 1151
	    (fi->flags & CEPH_F_SYNC)) {

		dout("aio_sync_read %p %llx.%llx %llu~%u got cap refs on %s\n",
		     inode, ceph_vinop(inode), iocb->ki_pos, (unsigned)len,
		     ceph_cap_string(got));

Y
Yan, Zheng 已提交
1152
		if (ci->i_inline_version == CEPH_INLINE_NONE) {
Y
Yan, Zheng 已提交
1153 1154 1155 1156 1157 1158 1159 1160
			if (!retry_op && (iocb->ki_flags & IOCB_DIRECT)) {
				ret = ceph_direct_read_write(iocb, to,
							     NULL, NULL);
				if (ret >= 0 && ret < len)
					retry_op = CHECK_EOF;
			} else {
				ret = ceph_sync_read(iocb, to, &retry_op);
			}
Y
Yan, Zheng 已提交
1161 1162 1163
		} else {
			retry_op = READ_INLINE;
		}
1164 1165
	} else {
		dout("aio_read %p %llx.%llx %llu~%u got cap refs on %s\n",
A
Al Viro 已提交
1166
		     inode, ceph_vinop(inode), iocb->ki_pos, (unsigned)len,
1167
		     ceph_cap_string(got));
1168
		current->journal_info = filp;
A
Al Viro 已提交
1169
		ret = generic_file_read_iter(iocb, to);
1170
		current->journal_info = NULL;
1171
	}
S
Sage Weil 已提交
1172 1173
	dout("aio_read %p %llx.%llx dropping cap refs on %s = %d\n",
	     inode, ceph_vinop(inode), ceph_cap_string(got), (int)ret);
1174
	if (pinned_page) {
1175
		put_page(pinned_page);
1176 1177
		pinned_page = NULL;
	}
S
Sage Weil 已提交
1178
	ceph_put_cap_refs(ci, got);
Y
Yan, Zheng 已提交
1179
	if (retry_op > HAVE_RETRIED && ret >= 0) {
Y
Yan, Zheng 已提交
1180 1181 1182 1183
		int statret;
		struct page *page = NULL;
		loff_t i_size;
		if (retry_op == READ_INLINE) {
1184
			page = __page_cache_alloc(GFP_KERNEL);
Y
Yan, Zheng 已提交
1185 1186 1187
			if (!page)
				return -ENOMEM;
		}
1188

Y
Yan, Zheng 已提交
1189 1190 1191
		statret = __ceph_do_getattr(inode, page,
					    CEPH_STAT_CAP_INLINE_DATA, !!page);
		if (statret < 0) {
1192 1193
			if (page)
				__free_page(page);
Y
Yan, Zheng 已提交
1194 1195 1196 1197 1198 1199
			if (statret == -ENODATA) {
				BUG_ON(retry_op != READ_INLINE);
				goto again;
			}
			return statret;
		}
1200

Y
Yan, Zheng 已提交
1201 1202
		i_size = i_size_read(inode);
		if (retry_op == READ_INLINE) {
1203 1204
			BUG_ON(ret > 0 || read > 0);
			if (iocb->ki_pos < i_size &&
1205
			    iocb->ki_pos < PAGE_SIZE) {
Y
Yan, Zheng 已提交
1206 1207
				loff_t end = min_t(loff_t, i_size,
						   iocb->ki_pos + len);
1208
				end = min_t(loff_t, end, PAGE_SIZE);
Y
Yan, Zheng 已提交
1209 1210 1211 1212 1213 1214
				if (statret < end)
					zero_user_segment(page, statret, end);
				ret = copy_page_to_iter(page,
						iocb->ki_pos & ~PAGE_MASK,
						end - iocb->ki_pos, to);
				iocb->ki_pos += ret;
1215 1216 1217 1218 1219 1220 1221 1222
				read += ret;
			}
			if (iocb->ki_pos < i_size && read < len) {
				size_t zlen = min_t(size_t, len - read,
						    i_size - iocb->ki_pos);
				ret = iov_iter_zero(zlen, to);
				iocb->ki_pos += ret;
				read += ret;
Y
Yan, Zheng 已提交
1223 1224
			}
			__free_pages(page, 0);
1225
			return read;
Y
Yan, Zheng 已提交
1226
		}
1227 1228

		/* hit EOF or hole? */
Y
Yan, Zheng 已提交
1229
		if (retry_op == CHECK_EOF && iocb->ki_pos < i_size &&
1230
		    ret < len) {
1231
			dout("sync_read hit hole, ppos %lld < size %lld"
1232
			     ", reading more\n", iocb->ki_pos, i_size);
1233

1234 1235
			read += ret;
			len -= ret;
Y
Yan, Zheng 已提交
1236
			retry_op = HAVE_RETRIED;
1237 1238 1239
			goto again;
		}
	}
1240

1241 1242 1243
	if (ret >= 0)
		ret += read;

S
Sage Weil 已提交
1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256
	return ret;
}

/*
 * Take cap references to avoid releasing caps to MDS mid-write.
 *
 * If we are synchronous, and write with an old snap context, the OSD
 * may return EOLDSNAPC.  In that case, retry the write.. _after_
 * dropping our cap refs and allowing the pending snap to logically
 * complete _before_ this write occurs.
 *
 * If we are near ENOSPC, write synchronously.
 */
A
Al Viro 已提交
1257
static ssize_t ceph_write_iter(struct kiocb *iocb, struct iov_iter *from)
S
Sage Weil 已提交
1258 1259
{
	struct file *file = iocb->ki_filp;
1260
	struct ceph_file_info *fi = file->private_data;
A
Al Viro 已提交
1261
	struct inode *inode = file_inode(file);
S
Sage Weil 已提交
1262
	struct ceph_inode_info *ci = ceph_inode(inode);
1263 1264
	struct ceph_osd_client *osdc =
		&ceph_sb_to_client(inode->i_sb)->client->osdc;
1265
	struct ceph_cap_flush *prealloc_cf;
1266
	ssize_t count, written = 0;
1267
	int err, want, got;
1268
	loff_t pos;
S
Sage Weil 已提交
1269 1270 1271 1272

	if (ceph_snap(inode) != CEPH_NOSNAP)
		return -EROFS;

1273 1274 1275 1276
	prealloc_cf = ceph_alloc_cap_flush();
	if (!prealloc_cf)
		return -ENOMEM;

A
Al Viro 已提交
1277
	inode_lock(inode);
1278 1279

	/* We can write back this queue in page reclaim */
1280
	current->backing_dev_info = inode_to_bdi(inode);
1281

1282 1283 1284 1285 1286 1287
	if (iocb->ki_flags & IOCB_APPEND) {
		err = ceph_do_getattr(inode, CEPH_STAT_CAP_SIZE, false);
		if (err < 0)
			goto out;
	}

1288 1289
	err = generic_write_checks(iocb, from);
	if (err <= 0)
1290 1291
		goto out;

1292 1293
	pos = iocb->ki_pos;
	count = iov_iter_count(from);
1294
	err = file_remove_privs(file);
1295 1296 1297 1298 1299 1300 1301
	if (err)
		goto out;

	err = file_update_time(file);
	if (err)
		goto out;

1302 1303 1304 1305 1306 1307
	if (ci->i_inline_version != CEPH_INLINE_NONE) {
		err = ceph_uninline_data(file, NULL);
		if (err < 0)
			goto out;
	}

S
Sage Weil 已提交
1308
retry_snap:
1309
	if (ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL)) {
1310
		err = -ENOSPC;
Y
Yan, Zheng 已提交
1311 1312
		goto out;
	}
1313

1314
	dout("aio_write %p %llx.%llx %llu~%zd getting caps. i_size %llu\n",
1315
	     inode, ceph_vinop(inode), pos, count, i_size_read(inode));
S
Sage Weil 已提交
1316 1317 1318 1319
	if (fi->fmode & CEPH_FILE_MODE_LAZY)
		want = CEPH_CAP_FILE_BUFFER | CEPH_CAP_FILE_LAZYIO;
	else
		want = CEPH_CAP_FILE_BUFFER;
1320
	got = 0;
1321 1322
	err = ceph_get_caps(ci, CEPH_CAP_FILE_WR, want, pos + count,
			    &got, NULL);
1323
	if (err < 0)
1324
		goto out;
S
Sage Weil 已提交
1325

1326
	dout("aio_write %p %llx.%llx %llu~%zd got cap refs on %s\n",
1327
	     inode, ceph_vinop(inode), pos, count, ceph_cap_string(got));
S
Sage Weil 已提交
1328 1329

	if ((got & (CEPH_CAP_FILE_BUFFER|CEPH_CAP_FILE_LAZYIO)) == 0 ||
1330
	    (iocb->ki_flags & IOCB_DIRECT) || (fi->flags & CEPH_F_SYNC)) {
1331
		struct ceph_snap_context *snapc;
A
Al Viro 已提交
1332
		struct iov_iter data;
A
Al Viro 已提交
1333
		inode_unlock(inode);
1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347

		spin_lock(&ci->i_ceph_lock);
		if (__ceph_have_pending_cap_snap(ci)) {
			struct ceph_cap_snap *capsnap =
					list_last_entry(&ci->i_cap_snaps,
							struct ceph_cap_snap,
							ci_item);
			snapc = ceph_get_snap_context(capsnap->context);
		} else {
			BUG_ON(!ci->i_head_snapc);
			snapc = ceph_get_snap_context(ci->i_head_snapc);
		}
		spin_unlock(&ci->i_ceph_lock);

A
Al Viro 已提交
1348 1349
		/* we might need to revert back to that point */
		data = *from;
1350
		if (iocb->ki_flags & IOCB_DIRECT)
Y
Yan, Zheng 已提交
1351 1352
			written = ceph_direct_read_write(iocb, &data, snapc,
							 &prealloc_cf);
1353
		else
1354
			written = ceph_sync_write(iocb, &data, pos, snapc);
1355 1356 1357 1358
		if (written == -EOLDSNAPC) {
			dout("aio_write %p %llx.%llx %llu~%u"
				"got EOLDSNAPC, retrying\n",
				inode, ceph_vinop(inode),
A
Al Viro 已提交
1359
				pos, (unsigned)count);
A
Al Viro 已提交
1360
			inode_lock(inode);
1361 1362
			goto retry_snap;
		}
A
Al Viro 已提交
1363 1364
		if (written > 0)
			iov_iter_advance(from, written);
1365
		ceph_put_snap_context(snapc);
S
Sage Weil 已提交
1366
	} else {
Y
Yan, Zheng 已提交
1367 1368 1369 1370 1371 1372 1373
		/*
		 * No need to acquire the i_truncate_mutex. Because
		 * the MDS revokes Fwb caps before sending truncate
		 * message to us. We can't get Fwb cap while there
		 * are pending vmtruncate. So write and vmtruncate
		 * can not run at the same time
		 */
A
Al Viro 已提交
1374
		written = generic_perform_write(file, from, pos);
1375 1376
		if (likely(written >= 0))
			iocb->ki_pos = pos + written;
A
Al Viro 已提交
1377
		inode_unlock(inode);
S
Sage Weil 已提交
1378
	}
1379

1380
	if (written >= 0) {
1381
		int dirty;
1382
		spin_lock(&ci->i_ceph_lock);
1383
		ci->i_inline_version = CEPH_INLINE_NONE;
1384 1385
		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR,
					       &prealloc_cf);
1386
		spin_unlock(&ci->i_ceph_lock);
1387 1388
		if (dirty)
			__mark_inode_dirty(inode, dirty);
S
Sage Weil 已提交
1389
	}
S
Sage Weil 已提交
1390

S
Sage Weil 已提交
1391
	dout("aio_write %p %llx.%llx %llu~%u  dropping cap refs on %s\n",
A
Al Viro 已提交
1392
	     inode, ceph_vinop(inode), pos, (unsigned)count,
S
Sage Weil 已提交
1393
	     ceph_cap_string(got));
S
Sage Weil 已提交
1394
	ceph_put_cap_refs(ci, got);
S
Sage Weil 已提交
1395

C
Christoph Hellwig 已提交
1396
	if (written >= 0) {
1397
		if (ceph_osdmap_flag(osdc, CEPH_OSDMAP_NEARFULL))
C
Christoph Hellwig 已提交
1398 1399 1400
			iocb->ki_flags |= IOCB_DSYNC;

		written = generic_write_sync(iocb, written);
Y
Yan, Zheng 已提交
1401
	}
1402

1403 1404
	goto out_unlocked;

1405
out:
A
Al Viro 已提交
1406
	inode_unlock(inode);
1407
out_unlocked:
1408
	ceph_free_cap_flush(prealloc_cf);
1409 1410
	current->backing_dev_info = NULL;
	return written ? written : err;
S
Sage Weil 已提交
1411 1412 1413 1414 1415
}

/*
 * llseek.  be sure to verify file size on SEEK_END.
 */
1416
static loff_t ceph_llseek(struct file *file, loff_t offset, int whence)
S
Sage Weil 已提交
1417 1418
{
	struct inode *inode = file->f_mapping->host;
1419
	loff_t i_size;
1420
	loff_t ret;
S
Sage Weil 已提交
1421

A
Al Viro 已提交
1422
	inode_lock(inode);
1423

1424
	if (whence == SEEK_END || whence == SEEK_DATA || whence == SEEK_HOLE) {
1425
		ret = ceph_do_getattr(inode, CEPH_STAT_CAP_SIZE, false);
1426
		if (ret < 0)
S
Sage Weil 已提交
1427
			goto out;
1428 1429
	}

1430
	i_size = i_size_read(inode);
1431
	switch (whence) {
1432
	case SEEK_END:
1433
		offset += i_size;
S
Sage Weil 已提交
1434 1435 1436 1437 1438 1439 1440 1441 1442
		break;
	case SEEK_CUR:
		/*
		 * Here we special-case the lseek(fd, 0, SEEK_CUR)
		 * position-querying operation.  Avoid rewriting the "same"
		 * f_pos value back to the file because a concurrent read(),
		 * write() or lseek() might have altered it
		 */
		if (offset == 0) {
1443
			ret = file->f_pos;
S
Sage Weil 已提交
1444 1445 1446 1447
			goto out;
		}
		offset += file->f_pos;
		break;
1448
	case SEEK_DATA:
1449
		if (offset >= i_size) {
1450 1451 1452 1453 1454
			ret = -ENXIO;
			goto out;
		}
		break;
	case SEEK_HOLE:
1455
		if (offset >= i_size) {
1456 1457 1458
			ret = -ENXIO;
			goto out;
		}
1459
		offset = i_size;
1460
		break;
S
Sage Weil 已提交
1461 1462
	}

1463
	ret = vfs_setpos(file, offset, inode->i_sb->s_maxbytes);
S
Sage Weil 已提交
1464 1465

out:
A
Al Viro 已提交
1466
	inode_unlock(inode);
1467
	return ret;
S
Sage Weil 已提交
1468 1469
}

L
Li Wang 已提交
1470 1471 1472 1473
static inline void ceph_zero_partial_page(
	struct inode *inode, loff_t offset, unsigned size)
{
	struct page *page;
1474
	pgoff_t index = offset >> PAGE_SHIFT;
L
Li Wang 已提交
1475 1476 1477 1478

	page = find_lock_page(inode->i_mapping, index);
	if (page) {
		wait_on_page_writeback(page);
1479
		zero_user(page, offset & (PAGE_SIZE - 1), size);
L
Li Wang 已提交
1480
		unlock_page(page);
1481
		put_page(page);
L
Li Wang 已提交
1482 1483 1484 1485 1486 1487
	}
}

static void ceph_zero_pagecache_range(struct inode *inode, loff_t offset,
				      loff_t length)
{
1488
	loff_t nearly = round_up(offset, PAGE_SIZE);
L
Li Wang 已提交
1489 1490 1491 1492 1493 1494 1495 1496
	if (offset < nearly) {
		loff_t size = nearly - offset;
		if (length < size)
			size = length;
		ceph_zero_partial_page(inode, offset, size);
		offset += size;
		length -= size;
	}
1497 1498
	if (length >= PAGE_SIZE) {
		loff_t size = round_down(length, PAGE_SIZE);
L
Li Wang 已提交
1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526
		truncate_pagecache_range(inode, offset, offset + size - 1);
		offset += size;
		length -= size;
	}
	if (length)
		ceph_zero_partial_page(inode, offset, length);
}

static int ceph_zero_partial_object(struct inode *inode,
				    loff_t offset, loff_t *length)
{
	struct ceph_inode_info *ci = ceph_inode(inode);
	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
	struct ceph_osd_request *req;
	int ret = 0;
	loff_t zero = 0;
	int op;

	if (!length) {
		op = offset ? CEPH_OSD_OP_DELETE : CEPH_OSD_OP_TRUNCATE;
		length = &zero;
	} else {
		op = CEPH_OSD_OP_ZERO;
	}

	req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
					ceph_vino(inode),
					offset, length,
1527
					0, 1, op,
1528
					CEPH_OSD_FLAG_WRITE,
L
Li Wang 已提交
1529 1530 1531 1532 1533 1534
					NULL, 0, 0, false);
	if (IS_ERR(req)) {
		ret = PTR_ERR(req);
		goto out;
	}

1535
	req->r_mtime = inode->i_mtime;
L
Li Wang 已提交
1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551
	ret = ceph_osdc_start_request(&fsc->client->osdc, req, false);
	if (!ret) {
		ret = ceph_osdc_wait_request(&fsc->client->osdc, req);
		if (ret == -ENOENT)
			ret = 0;
	}
	ceph_osdc_put_request(req);

out:
	return ret;
}

static int ceph_zero_objects(struct inode *inode, loff_t offset, loff_t length)
{
	int ret = 0;
	struct ceph_inode_info *ci = ceph_inode(inode);
1552 1553 1554
	s32 stripe_unit = ci->i_layout.stripe_unit;
	s32 stripe_count = ci->i_layout.stripe_count;
	s32 object_size = ci->i_layout.object_size;
S
Sage Weil 已提交
1555 1556 1557 1558 1559 1560 1561
	u64 object_set_size = object_size * stripe_count;
	u64 nearly, t;

	/* round offset up to next period boundary */
	nearly = offset + object_set_size - 1;
	t = nearly;
	nearly -= do_div(t, object_set_size);
L
Li Wang 已提交
1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597

	while (length && offset < nearly) {
		loff_t size = length;
		ret = ceph_zero_partial_object(inode, offset, &size);
		if (ret < 0)
			return ret;
		offset += size;
		length -= size;
	}
	while (length >= object_set_size) {
		int i;
		loff_t pos = offset;
		for (i = 0; i < stripe_count; ++i) {
			ret = ceph_zero_partial_object(inode, pos, NULL);
			if (ret < 0)
				return ret;
			pos += stripe_unit;
		}
		offset += object_set_size;
		length -= object_set_size;
	}
	while (length) {
		loff_t size = length;
		ret = ceph_zero_partial_object(inode, offset, &size);
		if (ret < 0)
			return ret;
		offset += size;
		length -= size;
	}
	return ret;
}

static long ceph_fallocate(struct file *file, int mode,
				loff_t offset, loff_t length)
{
	struct ceph_file_info *fi = file->private_data;
L
Libo Chen 已提交
1598
	struct inode *inode = file_inode(file);
L
Li Wang 已提交
1599 1600 1601
	struct ceph_inode_info *ci = ceph_inode(inode);
	struct ceph_osd_client *osdc =
		&ceph_inode_to_client(inode)->client->osdc;
1602
	struct ceph_cap_flush *prealloc_cf;
L
Li Wang 已提交
1603 1604 1605 1606 1607 1608
	int want, got = 0;
	int dirty;
	int ret = 0;
	loff_t endoff = 0;
	loff_t size;

1609 1610 1611
	if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE))
		return -EOPNOTSUPP;

L
Li Wang 已提交
1612 1613 1614
	if (!S_ISREG(inode->i_mode))
		return -EOPNOTSUPP;

1615 1616 1617 1618
	prealloc_cf = ceph_alloc_cap_flush();
	if (!prealloc_cf)
		return -ENOMEM;

A
Al Viro 已提交
1619
	inode_lock(inode);
L
Li Wang 已提交
1620 1621 1622 1623 1624 1625

	if (ceph_snap(inode) != CEPH_NOSNAP) {
		ret = -EROFS;
		goto unlock;
	}

1626 1627
	if (ceph_osdmap_flag(osdc, CEPH_OSDMAP_FULL) &&
	    !(mode & FALLOC_FL_PUNCH_HOLE)) {
L
Li Wang 已提交
1628 1629 1630 1631
		ret = -ENOSPC;
		goto unlock;
	}

1632 1633 1634 1635 1636 1637
	if (ci->i_inline_version != CEPH_INLINE_NONE) {
		ret = ceph_uninline_data(file, NULL);
		if (ret < 0)
			goto unlock;
	}

L
Li Wang 已提交
1638 1639 1640 1641 1642 1643 1644 1645 1646
	size = i_size_read(inode);
	if (!(mode & FALLOC_FL_KEEP_SIZE))
		endoff = offset + length;

	if (fi->fmode & CEPH_FILE_MODE_LAZY)
		want = CEPH_CAP_FILE_BUFFER | CEPH_CAP_FILE_LAZYIO;
	else
		want = CEPH_CAP_FILE_BUFFER;

1647
	ret = ceph_get_caps(ci, CEPH_CAP_FILE_WR, want, endoff, &got, NULL);
L
Li Wang 已提交
1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663
	if (ret < 0)
		goto unlock;

	if (mode & FALLOC_FL_PUNCH_HOLE) {
		if (offset < size)
			ceph_zero_pagecache_range(inode, offset, length);
		ret = ceph_zero_objects(inode, offset, length);
	} else if (endoff > size) {
		truncate_pagecache_range(inode, size, -1);
		if (ceph_inode_set_size(inode, endoff))
			ceph_check_caps(ceph_inode(inode),
				CHECK_CAPS_AUTHONLY, NULL);
	}

	if (!ret) {
		spin_lock(&ci->i_ceph_lock);
1664
		ci->i_inline_version = CEPH_INLINE_NONE;
1665 1666
		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR,
					       &prealloc_cf);
L
Li Wang 已提交
1667 1668 1669 1670 1671 1672 1673
		spin_unlock(&ci->i_ceph_lock);
		if (dirty)
			__mark_inode_dirty(inode, dirty);
	}

	ceph_put_cap_refs(ci, got);
unlock:
A
Al Viro 已提交
1674
	inode_unlock(inode);
1675
	ceph_free_cap_flush(prealloc_cf);
L
Li Wang 已提交
1676 1677 1678
	return ret;
}

S
Sage Weil 已提交
1679 1680 1681 1682
const struct file_operations ceph_file_fops = {
	.open = ceph_open,
	.release = ceph_release,
	.llseek = ceph_llseek,
A
Al Viro 已提交
1683
	.read_iter = ceph_read_iter,
A
Al Viro 已提交
1684
	.write_iter = ceph_write_iter,
S
Sage Weil 已提交
1685 1686
	.mmap = ceph_mmap,
	.fsync = ceph_fsync,
G
Greg Farnum 已提交
1687 1688
	.lock = ceph_lock,
	.flock = ceph_flock,
1689
	.splice_read = generic_file_splice_read,
1690
	.splice_write = iter_file_splice_write,
S
Sage Weil 已提交
1691 1692
	.unlocked_ioctl = ceph_ioctl,
	.compat_ioctl	= ceph_ioctl,
L
Li Wang 已提交
1693
	.fallocate	= ceph_fallocate,
S
Sage Weil 已提交
1694 1695
};