linux/fs/btrfs/send.c
<<
>>
Prefs
   1/*
   2 * Copyright (C) 2012 Alexander Block.  All rights reserved.
   3 *
   4 * This program is free software; you can redistribute it and/or
   5 * modify it under the terms of the GNU General Public
   6 * License v2 as published by the Free Software Foundation.
   7 *
   8 * This program is distributed in the hope that it will be useful,
   9 * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  11 * General Public License for more details.
  12 *
  13 * You should have received a copy of the GNU General Public
  14 * License along with this program; if not, write to the
  15 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  16 * Boston, MA 021110-1307, USA.
  17 */
  18
  19#include <linux/bsearch.h>
  20#include <linux/fs.h>
  21#include <linux/file.h>
  22#include <linux/sort.h>
  23#include <linux/mount.h>
  24#include <linux/xattr.h>
  25#include <linux/posix_acl_xattr.h>
  26#include <linux/radix-tree.h>
  27#include <linux/crc32c.h>
  28#include <linux/vmalloc.h>
  29
  30#include "send.h"
  31#include "backref.h"
  32#include "locking.h"
  33#include "disk-io.h"
  34#include "btrfs_inode.h"
  35#include "transaction.h"
  36
  37static int g_verbose = 0;
  38
  39#define verbose_printk(...) if (g_verbose) printk(__VA_ARGS__)
  40
  41/*
  42 * A fs_path is a helper to dynamically build path names with unknown size.
  43 * It reallocates the internal buffer on demand.
  44 * It allows fast adding of path elements on the right side (normal path) and
  45 * fast adding to the left side (reversed path). A reversed path can also be
  46 * unreversed if needed.
  47 */
  48struct fs_path {
  49        union {
  50                struct {
  51                        char *start;
  52                        char *end;
  53                        char *prepared;
  54
  55                        char *buf;
  56                        int buf_len;
  57                        int reversed:1;
  58                        int virtual_mem:1;
  59                        char inline_buf[];
  60                };
  61                char pad[PAGE_SIZE];
  62        };
  63};
  64#define FS_PATH_INLINE_SIZE \
  65        (sizeof(struct fs_path) - offsetof(struct fs_path, inline_buf))
  66
  67
  68/* reused for each extent */
  69struct clone_root {
  70        struct btrfs_root *root;
  71        u64 ino;
  72        u64 offset;
  73
  74        u64 found_refs;
  75};
  76
  77#define SEND_CTX_MAX_NAME_CACHE_SIZE 128
  78#define SEND_CTX_NAME_CACHE_CLEAN_SIZE (SEND_CTX_MAX_NAME_CACHE_SIZE * 2)
  79
  80struct send_ctx {
  81        struct file *send_filp;
  82        loff_t send_off;
  83        char *send_buf;
  84        u32 send_size;
  85        u32 send_max_size;
  86        u64 total_send_size;
  87        u64 cmd_send_size[BTRFS_SEND_C_MAX + 1];
  88
  89        struct vfsmount *mnt;
  90
  91        struct btrfs_root *send_root;
  92        struct btrfs_root *parent_root;
  93        struct clone_root *clone_roots;
  94        int clone_roots_cnt;
  95
  96        /* current state of the compare_tree call */
  97        struct btrfs_path *left_path;
  98        struct btrfs_path *right_path;
  99        struct btrfs_key *cmp_key;
 100
 101        /*
 102         * infos of the currently processed inode. In case of deleted inodes,
 103         * these are the values from the deleted inode.
 104         */
 105        u64 cur_ino;
 106        u64 cur_inode_gen;
 107        int cur_inode_new;
 108        int cur_inode_new_gen;
 109        int cur_inode_deleted;
 110        u64 cur_inode_size;
 111        u64 cur_inode_mode;
 112
 113        u64 send_progress;
 114
 115        struct list_head new_refs;
 116        struct list_head deleted_refs;
 117
 118        struct radix_tree_root name_cache;
 119        struct list_head name_cache_list;
 120        int name_cache_size;
 121
 122        struct file *cur_inode_filp;
 123        char *read_buf;
 124};
 125
 126struct name_cache_entry {
 127        struct list_head list;
 128        /*
 129         * radix_tree has only 32bit entries but we need to handle 64bit inums.
 130         * We use the lower 32bit of the 64bit inum to store it in the tree. If
 131         * more then one inum would fall into the same entry, we use radix_list
 132         * to store the additional entries. radix_list is also used to store
 133         * entries where two entries have the same inum but different
 134         * generations.
 135         */
 136        struct list_head radix_list;
 137        u64 ino;
 138        u64 gen;
 139        u64 parent_ino;
 140        u64 parent_gen;
 141        int ret;
 142        int need_later_update;
 143        int name_len;
 144        char name[];
 145};
 146
 147static void fs_path_reset(struct fs_path *p)
 148{
 149        if (p->reversed) {
 150                p->start = p->buf + p->buf_len - 1;
 151                p->end = p->start;
 152                *p->start = 0;
 153        } else {
 154                p->start = p->buf;
 155                p->end = p->start;
 156                *p->start = 0;
 157        }
 158}
 159
 160static struct fs_path *fs_path_alloc(struct send_ctx *sctx)
 161{
 162        struct fs_path *p;
 163
 164        p = kmalloc(sizeof(*p), GFP_NOFS);
 165        if (!p)
 166                return NULL;
 167        p->reversed = 0;
 168        p->virtual_mem = 0;
 169        p->buf = p->inline_buf;
 170        p->buf_len = FS_PATH_INLINE_SIZE;
 171        fs_path_reset(p);
 172        return p;
 173}
 174
 175static struct fs_path *fs_path_alloc_reversed(struct send_ctx *sctx)
 176{
 177        struct fs_path *p;
 178
 179        p = fs_path_alloc(sctx);
 180        if (!p)
 181                return NULL;
 182        p->reversed = 1;
 183        fs_path_reset(p);
 184        return p;
 185}
 186
 187static void fs_path_free(struct send_ctx *sctx, struct fs_path *p)
 188{
 189        if (!p)
 190                return;
 191        if (p->buf != p->inline_buf) {
 192                if (p->virtual_mem)
 193                        vfree(p->buf);
 194                else
 195                        kfree(p->buf);
 196        }
 197        kfree(p);
 198}
 199
 200static int fs_path_len(struct fs_path *p)
 201{
 202        return p->end - p->start;
 203}
 204
 205static int fs_path_ensure_buf(struct fs_path *p, int len)
 206{
 207        char *tmp_buf;
 208        int path_len;
 209        int old_buf_len;
 210
 211        len++;
 212
 213        if (p->buf_len >= len)
 214                return 0;
 215
 216        path_len = p->end - p->start;
 217        old_buf_len = p->buf_len;
 218        len = PAGE_ALIGN(len);
 219
 220        if (p->buf == p->inline_buf) {
 221                tmp_buf = kmalloc(len, GFP_NOFS);
 222                if (!tmp_buf) {
 223                        tmp_buf = vmalloc(len);
 224                        if (!tmp_buf)
 225                                return -ENOMEM;
 226                        p->virtual_mem = 1;
 227                }
 228                memcpy(tmp_buf, p->buf, p->buf_len);
 229                p->buf = tmp_buf;
 230                p->buf_len = len;
 231        } else {
 232                if (p->virtual_mem) {
 233                        tmp_buf = vmalloc(len);
 234                        if (!tmp_buf)
 235                                return -ENOMEM;
 236                        memcpy(tmp_buf, p->buf, p->buf_len);
 237                        vfree(p->buf);
 238                } else {
 239                        tmp_buf = krealloc(p->buf, len, GFP_NOFS);
 240                        if (!tmp_buf) {
 241                                tmp_buf = vmalloc(len);
 242                                if (!tmp_buf)
 243                                        return -ENOMEM;
 244                                memcpy(tmp_buf, p->buf, p->buf_len);
 245                                kfree(p->buf);
 246                                p->virtual_mem = 1;
 247                        }
 248                }
 249                p->buf = tmp_buf;
 250                p->buf_len = len;
 251        }
 252        if (p->reversed) {
 253                tmp_buf = p->buf + old_buf_len - path_len - 1;
 254                p->end = p->buf + p->buf_len - 1;
 255                p->start = p->end - path_len;
 256                memmove(p->start, tmp_buf, path_len + 1);
 257        } else {
 258                p->start = p->buf;
 259                p->end = p->start + path_len;
 260        }
 261        return 0;
 262}
 263
 264static int fs_path_prepare_for_add(struct fs_path *p, int name_len)
 265{
 266        int ret;
 267        int new_len;
 268
 269        new_len = p->end - p->start + name_len;
 270        if (p->start != p->end)
 271                new_len++;
 272        ret = fs_path_ensure_buf(p, new_len);
 273        if (ret < 0)
 274                goto out;
 275
 276        if (p->reversed) {
 277                if (p->start != p->end)
 278                        *--p->start = '/';
 279                p->start -= name_len;
 280                p->prepared = p->start;
 281        } else {
 282                if (p->start != p->end)
 283                        *p->end++ = '/';
 284                p->prepared = p->end;
 285                p->end += name_len;
 286                *p->end = 0;
 287        }
 288
 289out:
 290        return ret;
 291}
 292
 293static int fs_path_add(struct fs_path *p, const char *name, int name_len)
 294{
 295        int ret;
 296
 297        ret = fs_path_prepare_for_add(p, name_len);
 298        if (ret < 0)
 299                goto out;
 300        memcpy(p->prepared, name, name_len);
 301        p->prepared = NULL;
 302
 303out:
 304        return ret;
 305}
 306
 307static int fs_path_add_path(struct fs_path *p, struct fs_path *p2)
 308{
 309        int ret;
 310
 311        ret = fs_path_prepare_for_add(p, p2->end - p2->start);
 312        if (ret < 0)
 313                goto out;
 314        memcpy(p->prepared, p2->start, p2->end - p2->start);
 315        p->prepared = NULL;
 316
 317out:
 318        return ret;
 319}
 320
 321static int fs_path_add_from_extent_buffer(struct fs_path *p,
 322                                          struct extent_buffer *eb,
 323                                          unsigned long off, int len)
 324{
 325        int ret;
 326
 327        ret = fs_path_prepare_for_add(p, len);
 328        if (ret < 0)
 329                goto out;
 330
 331        read_extent_buffer(eb, p->prepared, off, len);
 332        p->prepared = NULL;
 333
 334out:
 335        return ret;
 336}
 337
 338#if 0
 339static void fs_path_remove(struct fs_path *p)
 340{
 341        BUG_ON(p->reversed);
 342        while (p->start != p->end && *p->end != '/')
 343                p->end--;
 344        *p->end = 0;
 345}
 346#endif
 347
 348static int fs_path_copy(struct fs_path *p, struct fs_path *from)
 349{
 350        int ret;
 351
 352        p->reversed = from->reversed;
 353        fs_path_reset(p);
 354
 355        ret = fs_path_add_path(p, from);
 356
 357        return ret;
 358}
 359
 360
 361static void fs_path_unreverse(struct fs_path *p)
 362{
 363        char *tmp;
 364        int len;
 365
 366        if (!p->reversed)
 367                return;
 368
 369        tmp = p->start;
 370        len = p->end - p->start;
 371        p->start = p->buf;
 372        p->end = p->start + len;
 373        memmove(p->start, tmp, len + 1);
 374        p->reversed = 0;
 375}
 376
 377static struct btrfs_path *alloc_path_for_send(void)
 378{
 379        struct btrfs_path *path;
 380
 381        path = btrfs_alloc_path();
 382        if (!path)
 383                return NULL;
 384        path->search_commit_root = 1;
 385        path->skip_locking = 1;
 386        return path;
 387}
 388
 389int write_buf(struct file *filp, const void *buf, u32 len, loff_t *off)
 390{
 391        int ret;
 392        mm_segment_t old_fs;
 393        u32 pos = 0;
 394
 395        old_fs = get_fs();
 396        set_fs(KERNEL_DS);
 397
 398        while (pos < len) {
 399                ret = vfs_write(filp, (char *)buf + pos, len - pos, off);
 400                /* TODO handle that correctly */
 401                /*if (ret == -ERESTARTSYS) {
 402                        continue;
 403                }*/
 404                if (ret < 0)
 405                        goto out;
 406                if (ret == 0) {
 407                        ret = -EIO;
 408                        goto out;
 409                }
 410                pos += ret;
 411        }
 412
 413        ret = 0;
 414
 415out:
 416        set_fs(old_fs);
 417        return ret;
 418}
 419
 420static int tlv_put(struct send_ctx *sctx, u16 attr, const void *data, int len)
 421{
 422        struct btrfs_tlv_header *hdr;
 423        int total_len = sizeof(*hdr) + len;
 424        int left = sctx->send_max_size - sctx->send_size;
 425
 426        if (unlikely(left < total_len))
 427                return -EOVERFLOW;
 428
 429        hdr = (struct btrfs_tlv_header *) (sctx->send_buf + sctx->send_size);
 430        hdr->tlv_type = cpu_to_le16(attr);
 431        hdr->tlv_len = cpu_to_le16(len);
 432        memcpy(hdr + 1, data, len);
 433        sctx->send_size += total_len;
 434
 435        return 0;
 436}
 437
 438#if 0
 439static int tlv_put_u8(struct send_ctx *sctx, u16 attr, u8 value)
 440{
 441        return tlv_put(sctx, attr, &value, sizeof(value));
 442}
 443
 444static int tlv_put_u16(struct send_ctx *sctx, u16 attr, u16 value)
 445{
 446        __le16 tmp = cpu_to_le16(value);
 447        return tlv_put(sctx, attr, &tmp, sizeof(tmp));
 448}
 449
 450static int tlv_put_u32(struct send_ctx *sctx, u16 attr, u32 value)
 451{
 452        __le32 tmp = cpu_to_le32(value);
 453        return tlv_put(sctx, attr, &tmp, sizeof(tmp));
 454}
 455#endif
 456
 457static int tlv_put_u64(struct send_ctx *sctx, u16 attr, u64 value)
 458{
 459        __le64 tmp = cpu_to_le64(value);
 460        return tlv_put(sctx, attr, &tmp, sizeof(tmp));
 461}
 462
 463static int tlv_put_string(struct send_ctx *sctx, u16 attr,
 464                          const char *str, int len)
 465{
 466        if (len == -1)
 467                len = strlen(str);
 468        return tlv_put(sctx, attr, str, len);
 469}
 470
 471static int tlv_put_uuid(struct send_ctx *sctx, u16 attr,
 472                        const u8 *uuid)
 473{
 474        return tlv_put(sctx, attr, uuid, BTRFS_UUID_SIZE);
 475}
 476
 477#if 0
 478static int tlv_put_timespec(struct send_ctx *sctx, u16 attr,
 479                            struct timespec *ts)
 480{
 481        struct btrfs_timespec bts;
 482        bts.sec = cpu_to_le64(ts->tv_sec);
 483        bts.nsec = cpu_to_le32(ts->tv_nsec);
 484        return tlv_put(sctx, attr, &bts, sizeof(bts));
 485}
 486#endif
 487
 488static int tlv_put_btrfs_timespec(struct send_ctx *sctx, u16 attr,
 489                                  struct extent_buffer *eb,
 490                                  struct btrfs_timespec *ts)
 491{
 492        struct btrfs_timespec bts;
 493        read_extent_buffer(eb, &bts, (unsigned long)ts, sizeof(bts));
 494        return tlv_put(sctx, attr, &bts, sizeof(bts));
 495}
 496
 497
 498#define TLV_PUT(sctx, attrtype, attrlen, data) \
 499        do { \
 500                ret = tlv_put(sctx, attrtype, attrlen, data); \
 501                if (ret < 0) \
 502                        goto tlv_put_failure; \
 503        } while (0)
 504
 505#define TLV_PUT_INT(sctx, attrtype, bits, value) \
 506        do { \
 507                ret = tlv_put_u##bits(sctx, attrtype, value); \
 508                if (ret < 0) \
 509                        goto tlv_put_failure; \
 510        } while (0)
 511
 512#define TLV_PUT_U8(sctx, attrtype, data) TLV_PUT_INT(sctx, attrtype, 8, data)
 513#define TLV_PUT_U16(sctx, attrtype, data) TLV_PUT_INT(sctx, attrtype, 16, data)
 514#define TLV_PUT_U32(sctx, attrtype, data) TLV_PUT_INT(sctx, attrtype, 32, data)
 515#define TLV_PUT_U64(sctx, attrtype, data) TLV_PUT_INT(sctx, attrtype, 64, data)
 516#define TLV_PUT_STRING(sctx, attrtype, str, len) \
 517        do { \
 518                ret = tlv_put_string(sctx, attrtype, str, len); \
 519                if (ret < 0) \
 520                        goto tlv_put_failure; \
 521        } while (0)
 522#define TLV_PUT_PATH(sctx, attrtype, p) \
 523        do { \
 524                ret = tlv_put_string(sctx, attrtype, p->start, \
 525                        p->end - p->start); \
 526                if (ret < 0) \
 527                        goto tlv_put_failure; \
 528        } while(0)
 529#define TLV_PUT_UUID(sctx, attrtype, uuid) \
 530        do { \
 531                ret = tlv_put_uuid(sctx, attrtype, uuid); \
 532                if (ret < 0) \
 533                        goto tlv_put_failure; \
 534        } while (0)
 535#define TLV_PUT_TIMESPEC(sctx, attrtype, ts) \
 536        do { \
 537                ret = tlv_put_timespec(sctx, attrtype, ts); \
 538                if (ret < 0) \
 539                        goto tlv_put_failure; \
 540        } while (0)
 541#define TLV_PUT_BTRFS_TIMESPEC(sctx, attrtype, eb, ts) \
 542        do { \
 543                ret = tlv_put_btrfs_timespec(sctx, attrtype, eb, ts); \
 544                if (ret < 0) \
 545                        goto tlv_put_failure; \
 546        } while (0)
 547
 548static int send_header(struct send_ctx *sctx)
 549{
 550        struct btrfs_stream_header hdr;
 551
 552        strcpy(hdr.magic, BTRFS_SEND_STREAM_MAGIC);
 553        hdr.version = cpu_to_le32(BTRFS_SEND_STREAM_VERSION);
 554
 555        return write_buf(sctx->send_filp, &hdr, sizeof(hdr),
 556                                        &sctx->send_off);
 557}
 558
 559/*
 560 * For each command/item we want to send to userspace, we call this function.
 561 */
 562static int begin_cmd(struct send_ctx *sctx, int cmd)
 563{
 564        struct btrfs_cmd_header *hdr;
 565
 566        if (!sctx->send_buf) {
 567                WARN_ON(1);
 568                return -EINVAL;
 569        }
 570
 571        BUG_ON(sctx->send_size);
 572
 573        sctx->send_size += sizeof(*hdr);
 574        hdr = (struct btrfs_cmd_header *)sctx->send_buf;
 575        hdr->cmd = cpu_to_le16(cmd);
 576
 577        return 0;
 578}
 579
 580static int send_cmd(struct send_ctx *sctx)
 581{
 582        int ret;
 583        struct btrfs_cmd_header *hdr;
 584        u32 crc;
 585
 586        hdr = (struct btrfs_cmd_header *)sctx->send_buf;
 587        hdr->len = cpu_to_le32(sctx->send_size - sizeof(*hdr));
 588        hdr->crc = 0;
 589
 590        crc = crc32c(0, (unsigned char *)sctx->send_buf, sctx->send_size);
 591        hdr->crc = cpu_to_le32(crc);
 592
 593        ret = write_buf(sctx->send_filp, sctx->send_buf, sctx->send_size,
 594                                        &sctx->send_off);
 595
 596        sctx->total_send_size += sctx->send_size;
 597        sctx->cmd_send_size[le16_to_cpu(hdr->cmd)] += sctx->send_size;
 598        sctx->send_size = 0;
 599
 600        return ret;
 601}
 602
 603/*
 604 * Sends a move instruction to user space
 605 */
 606static int send_rename(struct send_ctx *sctx,
 607                     struct fs_path *from, struct fs_path *to)
 608{
 609        int ret;
 610
 611verbose_printk("btrfs: send_rename %s -> %s\n", from->start, to->start);
 612
 613        ret = begin_cmd(sctx, BTRFS_SEND_C_RENAME);
 614        if (ret < 0)
 615                goto out;
 616
 617        TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, from);
 618        TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH_TO, to);
 619
 620        ret = send_cmd(sctx);
 621
 622tlv_put_failure:
 623out:
 624        return ret;
 625}
 626
 627/*
 628 * Sends a link instruction to user space
 629 */
 630static int send_link(struct send_ctx *sctx,
 631                     struct fs_path *path, struct fs_path *lnk)
 632{
 633        int ret;
 634
 635verbose_printk("btrfs: send_link %s -> %s\n", path->start, lnk->start);
 636
 637        ret = begin_cmd(sctx, BTRFS_SEND_C_LINK);
 638        if (ret < 0)
 639                goto out;
 640
 641        TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, path);
 642        TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH_LINK, lnk);
 643
 644        ret = send_cmd(sctx);
 645
 646tlv_put_failure:
 647out:
 648        return ret;
 649}
 650
 651/*
 652 * Sends an unlink instruction to user space
 653 */
 654static int send_unlink(struct send_ctx *sctx, struct fs_path *path)
 655{
 656        int ret;
 657
 658verbose_printk("btrfs: send_unlink %s\n", path->start);
 659
 660        ret = begin_cmd(sctx, BTRFS_SEND_C_UNLINK);
 661        if (ret < 0)
 662                goto out;
 663
 664        TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, path);
 665
 666        ret = send_cmd(sctx);
 667
 668tlv_put_failure:
 669out:
 670        return ret;
 671}
 672
 673/*
 674 * Sends a rmdir instruction to user space
 675 */
 676static int send_rmdir(struct send_ctx *sctx, struct fs_path *path)
 677{
 678        int ret;
 679
 680verbose_printk("btrfs: send_rmdir %s\n", path->start);
 681
 682        ret = begin_cmd(sctx, BTRFS_SEND_C_RMDIR);
 683        if (ret < 0)
 684                goto out;
 685
 686        TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, path);
 687
 688        ret = send_cmd(sctx);
 689
 690tlv_put_failure:
 691out:
 692        return ret;
 693}
 694
 695/*
 696 * Helper function to retrieve some fields from an inode item.
 697 */
 698static int get_inode_info(struct btrfs_root *root,
 699                          u64 ino, u64 *size, u64 *gen,
 700                          u64 *mode, u64 *uid, u64 *gid,
 701                          u64 *rdev)
 702{
 703        int ret;
 704        struct btrfs_inode_item *ii;
 705        struct btrfs_key key;
 706        struct btrfs_path *path;
 707
 708        path = alloc_path_for_send();
 709        if (!path)
 710                return -ENOMEM;
 711
 712        key.objectid = ino;
 713        key.type = BTRFS_INODE_ITEM_KEY;
 714        key.offset = 0;
 715        ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
 716        if (ret < 0)
 717                goto out;
 718        if (ret) {
 719                ret = -ENOENT;
 720                goto out;
 721        }
 722
 723        ii = btrfs_item_ptr(path->nodes[0], path->slots[0],
 724                        struct btrfs_inode_item);
 725        if (size)
 726                *size = btrfs_inode_size(path->nodes[0], ii);
 727        if (gen)
 728                *gen = btrfs_inode_generation(path->nodes[0], ii);
 729        if (mode)
 730                *mode = btrfs_inode_mode(path->nodes[0], ii);
 731        if (uid)
 732                *uid = btrfs_inode_uid(path->nodes[0], ii);
 733        if (gid)
 734                *gid = btrfs_inode_gid(path->nodes[0], ii);
 735        if (rdev)
 736                *rdev = btrfs_inode_rdev(path->nodes[0], ii);
 737
 738out:
 739        btrfs_free_path(path);
 740        return ret;
 741}
 742
 743typedef int (*iterate_inode_ref_t)(int num, u64 dir, int index,
 744                                   struct fs_path *p,
 745                                   void *ctx);
 746
 747/*
 748 * Helper function to iterate the entries in ONE btrfs_inode_ref or
 749 * btrfs_inode_extref.
 750 * The iterate callback may return a non zero value to stop iteration. This can
 751 * be a negative value for error codes or 1 to simply stop it.
 752 *
 753 * path must point to the INODE_REF or INODE_EXTREF when called.
 754 */
 755static int iterate_inode_ref(struct send_ctx *sctx,
 756                             struct btrfs_root *root, struct btrfs_path *path,
 757                             struct btrfs_key *found_key, int resolve,
 758                             iterate_inode_ref_t iterate, void *ctx)
 759{
 760        struct extent_buffer *eb = path->nodes[0];
 761        struct btrfs_item *item;
 762        struct btrfs_inode_ref *iref;
 763        struct btrfs_inode_extref *extref;
 764        struct btrfs_path *tmp_path;
 765        struct fs_path *p;
 766        u32 cur = 0;
 767        u32 total;
 768        int slot = path->slots[0];
 769        u32 name_len;
 770        char *start;
 771        int ret = 0;
 772        int num = 0;
 773        int index;
 774        u64 dir;
 775        unsigned long name_off;
 776        unsigned long elem_size;
 777        unsigned long ptr;
 778
 779        p = fs_path_alloc_reversed(sctx);
 780        if (!p)
 781                return -ENOMEM;
 782
 783        tmp_path = alloc_path_for_send();
 784        if (!tmp_path) {
 785                fs_path_free(sctx, p);
 786                return -ENOMEM;
 787        }
 788
 789
 790        if (found_key->type == BTRFS_INODE_REF_KEY) {
 791                ptr = (unsigned long)btrfs_item_ptr(eb, slot,
 792                                                    struct btrfs_inode_ref);
 793                item = btrfs_item_nr(eb, slot);
 794                total = btrfs_item_size(eb, item);
 795                elem_size = sizeof(*iref);
 796        } else {
 797                ptr = btrfs_item_ptr_offset(eb, slot);
 798                total = btrfs_item_size_nr(eb, slot);
 799                elem_size = sizeof(*extref);
 800        }
 801
 802        while (cur < total) {
 803                fs_path_reset(p);
 804
 805                if (found_key->type == BTRFS_INODE_REF_KEY) {
 806                        iref = (struct btrfs_inode_ref *)(ptr + cur);
 807                        name_len = btrfs_inode_ref_name_len(eb, iref);
 808                        name_off = (unsigned long)(iref + 1);
 809                        index = btrfs_inode_ref_index(eb, iref);
 810                        dir = found_key->offset;
 811                } else {
 812                        extref = (struct btrfs_inode_extref *)(ptr + cur);
 813                        name_len = btrfs_inode_extref_name_len(eb, extref);
 814                        name_off = (unsigned long)&extref->name;
 815                        index = btrfs_inode_extref_index(eb, extref);
 816                        dir = btrfs_inode_extref_parent(eb, extref);
 817                }
 818
 819                if (resolve) {
 820                        start = btrfs_ref_to_path(root, tmp_path, name_len,
 821                                                  name_off, eb, dir,
 822                                                  p->buf, p->buf_len);
 823                        if (IS_ERR(start)) {
 824                                ret = PTR_ERR(start);
 825                                goto out;
 826                        }
 827                        if (start < p->buf) {
 828                                /* overflow , try again with larger buffer */
 829                                ret = fs_path_ensure_buf(p,
 830                                                p->buf_len + p->buf - start);
 831                                if (ret < 0)
 832                                        goto out;
 833                                start = btrfs_ref_to_path(root, tmp_path,
 834                                                          name_len, name_off,
 835                                                          eb, dir,
 836                                                          p->buf, p->buf_len);
 837                                if (IS_ERR(start)) {
 838                                        ret = PTR_ERR(start);
 839                                        goto out;
 840                                }
 841                                BUG_ON(start < p->buf);
 842                        }
 843                        p->start = start;
 844                } else {
 845                        ret = fs_path_add_from_extent_buffer(p, eb, name_off,
 846                                                             name_len);
 847                        if (ret < 0)
 848                                goto out;
 849                }
 850
 851                cur += elem_size + name_len;
 852                ret = iterate(num, dir, index, p, ctx);
 853                if (ret)
 854                        goto out;
 855                num++;
 856        }
 857
 858out:
 859        btrfs_free_path(tmp_path);
 860        fs_path_free(sctx, p);
 861        return ret;
 862}
 863
 864typedef int (*iterate_dir_item_t)(int num, struct btrfs_key *di_key,
 865                                  const char *name, int name_len,
 866                                  const char *data, int data_len,
 867                                  u8 type, void *ctx);
 868
 869/*
 870 * Helper function to iterate the entries in ONE btrfs_dir_item.
 871 * The iterate callback may return a non zero value to stop iteration. This can
 872 * be a negative value for error codes or 1 to simply stop it.
 873 *
 874 * path must point to the dir item when called.
 875 */
 876static int iterate_dir_item(struct send_ctx *sctx,
 877                            struct btrfs_root *root, struct btrfs_path *path,
 878                            struct btrfs_key *found_key,
 879                            iterate_dir_item_t iterate, void *ctx)
 880{
 881        int ret = 0;
 882        struct extent_buffer *eb;
 883        struct btrfs_item *item;
 884        struct btrfs_dir_item *di;
 885        struct btrfs_key di_key;
 886        char *buf = NULL;
 887        char *buf2 = NULL;
 888        int buf_len;
 889        int buf_virtual = 0;
 890        u32 name_len;
 891        u32 data_len;
 892        u32 cur;
 893        u32 len;
 894        u32 total;
 895        int slot;
 896        int num;
 897        u8 type;
 898
 899        buf_len = PAGE_SIZE;
 900        buf = kmalloc(buf_len, GFP_NOFS);
 901        if (!buf) {
 902                ret = -ENOMEM;
 903                goto out;
 904        }
 905
 906        eb = path->nodes[0];
 907        slot = path->slots[0];
 908        item = btrfs_item_nr(eb, slot);
 909        di = btrfs_item_ptr(eb, slot, struct btrfs_dir_item);
 910        cur = 0;
 911        len = 0;
 912        total = btrfs_item_size(eb, item);
 913
 914        num = 0;
 915        while (cur < total) {
 916                name_len = btrfs_dir_name_len(eb, di);
 917                data_len = btrfs_dir_data_len(eb, di);
 918                type = btrfs_dir_type(eb, di);
 919                btrfs_dir_item_key_to_cpu(eb, di, &di_key);
 920
 921                if (name_len + data_len > buf_len) {
 922                        buf_len = PAGE_ALIGN(name_len + data_len);
 923                        if (buf_virtual) {
 924                                buf2 = vmalloc(buf_len);
 925                                if (!buf2) {
 926                                        ret = -ENOMEM;
 927                                        goto out;
 928                                }
 929                                vfree(buf);
 930                        } else {
 931                                buf2 = krealloc(buf, buf_len, GFP_NOFS);
 932                                if (!buf2) {
 933                                        buf2 = vmalloc(buf_len);
 934                                        if (!buf2) {
 935                                                ret = -ENOMEM;
 936                                                goto out;
 937                                        }
 938                                        kfree(buf);
 939                                        buf_virtual = 1;
 940                                }
 941                        }
 942
 943                        buf = buf2;
 944                        buf2 = NULL;
 945                }
 946
 947                read_extent_buffer(eb, buf, (unsigned long)(di + 1),
 948                                name_len + data_len);
 949
 950                len = sizeof(*di) + name_len + data_len;
 951                di = (struct btrfs_dir_item *)((char *)di + len);
 952                cur += len;
 953
 954                ret = iterate(num, &di_key, buf, name_len, buf + name_len,
 955                                data_len, type, ctx);
 956                if (ret < 0)
 957                        goto out;
 958                if (ret) {
 959                        ret = 0;
 960                        goto out;
 961                }
 962
 963                num++;
 964        }
 965
 966out:
 967        if (buf_virtual)
 968                vfree(buf);
 969        else
 970                kfree(buf);
 971        return ret;
 972}
 973
 974static int __copy_first_ref(int num, u64 dir, int index,
 975                            struct fs_path *p, void *ctx)
 976{
 977        int ret;
 978        struct fs_path *pt = ctx;
 979
 980        ret = fs_path_copy(pt, p);
 981        if (ret < 0)
 982                return ret;
 983
 984        /* we want the first only */
 985        return 1;
 986}
 987
 988/*
 989 * Retrieve the first path of an inode. If an inode has more then one
 990 * ref/hardlink, this is ignored.
 991 */
 992static int get_inode_path(struct send_ctx *sctx, struct btrfs_root *root,
 993                          u64 ino, struct fs_path *path)
 994{
 995        int ret;
 996        struct btrfs_key key, found_key;
 997        struct btrfs_path *p;
 998
 999        p = alloc_path_for_send();
1000        if (!p)
1001                return -ENOMEM;
1002
1003        fs_path_reset(path);
1004
1005        key.objectid = ino;
1006        key.type = BTRFS_INODE_REF_KEY;
1007        key.offset = 0;
1008
1009        ret = btrfs_search_slot_for_read(root, &key, p, 1, 0);
1010        if (ret < 0)
1011                goto out;
1012        if (ret) {
1013                ret = 1;
1014                goto out;
1015        }
1016        btrfs_item_key_to_cpu(p->nodes[0], &found_key, p->slots[0]);
1017        if (found_key.objectid != ino ||
1018            (found_key.type != BTRFS_INODE_REF_KEY &&
1019             found_key.type != BTRFS_INODE_EXTREF_KEY)) {
1020                ret = -ENOENT;
1021                goto out;
1022        }
1023
1024        ret = iterate_inode_ref(sctx, root, p, &found_key, 1,
1025                        __copy_first_ref, path);
1026        if (ret < 0)
1027                goto out;
1028        ret = 0;
1029
1030out:
1031        btrfs_free_path(p);
1032        return ret;
1033}
1034
1035struct backref_ctx {
1036        struct send_ctx *sctx;
1037
1038        /* number of total found references */
1039        u64 found;
1040
1041        /*
1042         * used for clones found in send_root. clones found behind cur_objectid
1043         * and cur_offset are not considered as allowed clones.
1044         */
1045        u64 cur_objectid;
1046        u64 cur_offset;
1047
1048        /* may be truncated in case it's the last extent in a file */
1049        u64 extent_len;
1050
1051        /* Just to check for bugs in backref resolving */
1052        int found_itself;
1053};
1054
1055static int __clone_root_cmp_bsearch(const void *key, const void *elt)
1056{
1057        u64 root = (u64)(uintptr_t)key;
1058        struct clone_root *cr = (struct clone_root *)elt;
1059
1060        if (root < cr->root->objectid)
1061                return -1;
1062        if (root > cr->root->objectid)
1063                return 1;
1064        return 0;
1065}
1066
1067static int __clone_root_cmp_sort(const void *e1, const void *e2)
1068{
1069        struct clone_root *cr1 = (struct clone_root *)e1;
1070        struct clone_root *cr2 = (struct clone_root *)e2;
1071
1072        if (cr1->root->objectid < cr2->root->objectid)
1073                return -1;
1074        if (cr1->root->objectid > cr2->root->objectid)
1075                return 1;
1076        return 0;
1077}
1078
1079/*
1080 * Called for every backref that is found for the current extent.
1081 * Results are collected in sctx->clone_roots->ino/offset/found_refs
1082 */
1083static int __iterate_backrefs(u64 ino, u64 offset, u64 root, void *ctx_)
1084{
1085        struct backref_ctx *bctx = ctx_;
1086        struct clone_root *found;
1087        int ret;
1088        u64 i_size;
1089
1090        /* First check if the root is in the list of accepted clone sources */
1091        found = bsearch((void *)(uintptr_t)root, bctx->sctx->clone_roots,
1092                        bctx->sctx->clone_roots_cnt,
1093                        sizeof(struct clone_root),
1094                        __clone_root_cmp_bsearch);
1095        if (!found)
1096                return 0;
1097
1098        if (found->root == bctx->sctx->send_root &&
1099            ino == bctx->cur_objectid &&
1100            offset == bctx->cur_offset) {
1101                bctx->found_itself = 1;
1102        }
1103
1104        /*
1105         * There are inodes that have extents that lie behind its i_size. Don't
1106         * accept clones from these extents.
1107         */
1108        ret = get_inode_info(found->root, ino, &i_size, NULL, NULL, NULL, NULL,
1109                        NULL);
1110        if (ret < 0)
1111                return ret;
1112
1113        if (offset + bctx->extent_len > i_size)
1114                return 0;
1115
1116        /*
1117         * Make sure we don't consider clones from send_root that are
1118         * behind the current inode/offset.
1119         */
1120        if (found->root == bctx->sctx->send_root) {
1121                /*
1122                 * TODO for the moment we don't accept clones from the inode
1123                 * that is currently send. We may change this when
1124                 * BTRFS_IOC_CLONE_RANGE supports cloning from and to the same
1125                 * file.
1126                 */
1127                if (ino >= bctx->cur_objectid)
1128                        return 0;
1129#if 0
1130                if (ino > bctx->cur_objectid)
1131                        return 0;
1132                if (offset + bctx->extent_len > bctx->cur_offset)
1133                        return 0;
1134#endif
1135        }
1136
1137        bctx->found++;
1138        found->found_refs++;
1139        if (ino < found->ino) {
1140                found->ino = ino;
1141                found->offset = offset;
1142        } else if (found->ino == ino) {
1143                /*
1144                 * same extent found more then once in the same file.
1145                 */
1146                if (found->offset > offset + bctx->extent_len)
1147                        found->offset = offset;
1148        }
1149
1150        return 0;
1151}
1152
1153/*
1154 * Given an inode, offset and extent item, it finds a good clone for a clone
1155 * instruction. Returns -ENOENT when none could be found. The function makes
1156 * sure that the returned clone is usable at the point where sending is at the
1157 * moment. This means, that no clones are accepted which lie behind the current
1158 * inode+offset.
1159 *
1160 * path must point to the extent item when called.
1161 */
1162static int find_extent_clone(struct send_ctx *sctx,
1163                             struct btrfs_path *path,
1164                             u64 ino, u64 data_offset,
1165                             u64 ino_size,
1166                             struct clone_root **found)
1167{
1168        int ret;
1169        int extent_type;
1170        u64 logical;
1171        u64 disk_byte;
1172        u64 num_bytes;
1173        u64 extent_item_pos;
1174        u64 flags = 0;
1175        struct btrfs_file_extent_item *fi;
1176        struct extent_buffer *eb = path->nodes[0];
1177        struct backref_ctx *backref_ctx = NULL;
1178        struct clone_root *cur_clone_root;
1179        struct btrfs_key found_key;
1180        struct btrfs_path *tmp_path;
1181        int compressed;
1182        u32 i;
1183
1184        tmp_path = alloc_path_for_send();
1185        if (!tmp_path)
1186                return -ENOMEM;
1187
1188        backref_ctx = kmalloc(sizeof(*backref_ctx), GFP_NOFS);
1189        if (!backref_ctx) {
1190                ret = -ENOMEM;
1191                goto out;
1192        }
1193
1194        if (data_offset >= ino_size) {
1195                /*
1196                 * There may be extents that lie behind the file's size.
1197                 * I at least had this in combination with snapshotting while
1198                 * writing large files.
1199                 */
1200                ret = 0;
1201                goto out;
1202        }
1203
1204        fi = btrfs_item_ptr(eb, path->slots[0],
1205                        struct btrfs_file_extent_item);
1206        extent_type = btrfs_file_extent_type(eb, fi);
1207        if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
1208                ret = -ENOENT;
1209                goto out;
1210        }
1211        compressed = btrfs_file_extent_compression(eb, fi);
1212
1213        num_bytes = btrfs_file_extent_num_bytes(eb, fi);
1214        disk_byte = btrfs_file_extent_disk_bytenr(eb, fi);
1215        if (disk_byte == 0) {
1216                ret = -ENOENT;
1217                goto out;
1218        }
1219        logical = disk_byte + btrfs_file_extent_offset(eb, fi);
1220
1221        ret = extent_from_logical(sctx->send_root->fs_info, disk_byte, tmp_path,
1222                                  &found_key, &flags);
1223        btrfs_release_path(tmp_path);
1224
1225        if (ret < 0)
1226                goto out;
1227        if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
1228                ret = -EIO;
1229                goto out;
1230        }
1231
1232        /*
1233         * Setup the clone roots.
1234         */
1235        for (i = 0; i < sctx->clone_roots_cnt; i++) {
1236                cur_clone_root = sctx->clone_roots + i;
1237                cur_clone_root->ino = (u64)-1;
1238                cur_clone_root->offset = 0;
1239                cur_clone_root->found_refs = 0;
1240        }
1241
1242        backref_ctx->sctx = sctx;
1243        backref_ctx->found = 0;
1244        backref_ctx->cur_objectid = ino;
1245        backref_ctx->cur_offset = data_offset;
1246        backref_ctx->found_itself = 0;
1247        backref_ctx->extent_len = num_bytes;
1248
1249        /*
1250         * The last extent of a file may be too large due to page alignment.
1251         * We need to adjust extent_len in this case so that the checks in
1252         * __iterate_backrefs work.
1253         */
1254        if (data_offset + num_bytes >= ino_size)
1255                backref_ctx->extent_len = ino_size - data_offset;
1256
1257        /*
1258         * Now collect all backrefs.
1259         */
1260        if (compressed == BTRFS_COMPRESS_NONE)
1261                extent_item_pos = logical - found_key.objectid;
1262        else
1263                extent_item_pos = 0;
1264
1265        extent_item_pos = logical - found_key.objectid;
1266        ret = iterate_extent_inodes(sctx->send_root->fs_info,
1267                                        found_key.objectid, extent_item_pos, 1,
1268                                        __iterate_backrefs, backref_ctx);
1269
1270        if (ret < 0)
1271                goto out;
1272
1273        if (!backref_ctx->found_itself) {
1274                /* found a bug in backref code? */
1275                ret = -EIO;
1276                printk(KERN_ERR "btrfs: ERROR did not find backref in "
1277                                "send_root. inode=%llu, offset=%llu, "
1278                                "disk_byte=%llu found extent=%llu\n",
1279                                ino, data_offset, disk_byte, found_key.objectid);
1280                goto out;
1281        }
1282
1283verbose_printk(KERN_DEBUG "btrfs: find_extent_clone: data_offset=%llu, "
1284                "ino=%llu, "
1285                "num_bytes=%llu, logical=%llu\n",
1286                data_offset, ino, num_bytes, logical);
1287
1288        if (!backref_ctx->found)
1289                verbose_printk("btrfs:    no clones found\n");
1290
1291        cur_clone_root = NULL;
1292        for (i = 0; i < sctx->clone_roots_cnt; i++) {
1293                if (sctx->clone_roots[i].found_refs) {
1294                        if (!cur_clone_root)
1295                                cur_clone_root = sctx->clone_roots + i;
1296                        else if (sctx->clone_roots[i].root == sctx->send_root)
1297                                /* prefer clones from send_root over others */
1298                                cur_clone_root = sctx->clone_roots + i;
1299                }
1300
1301        }
1302
1303        if (cur_clone_root) {
1304                *found = cur_clone_root;
1305                ret = 0;
1306        } else {
1307                ret = -ENOENT;
1308        }
1309
1310out:
1311        btrfs_free_path(tmp_path);
1312        kfree(backref_ctx);
1313        return ret;
1314}
1315
1316static int read_symlink(struct send_ctx *sctx,
1317                        struct btrfs_root *root,
1318                        u64 ino,
1319                        struct fs_path *dest)
1320{
1321        int ret;
1322        struct btrfs_path *path;
1323        struct btrfs_key key;
1324        struct btrfs_file_extent_item *ei;
1325        u8 type;
1326        u8 compression;
1327        unsigned long off;
1328        int len;
1329
1330        path = alloc_path_for_send();
1331        if (!path)
1332                return -ENOMEM;
1333
1334        key.objectid = ino;
1335        key.type = BTRFS_EXTENT_DATA_KEY;
1336        key.offset = 0;
1337        ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
1338        if (ret < 0)
1339                goto out;
1340        BUG_ON(ret);
1341
1342        ei = btrfs_item_ptr(path->nodes[0], path->slots[0],
1343                        struct btrfs_file_extent_item);
1344        type = btrfs_file_extent_type(path->nodes[0], ei);
1345        compression = btrfs_file_extent_compression(path->nodes[0], ei);
1346        BUG_ON(type != BTRFS_FILE_EXTENT_INLINE);
1347        BUG_ON(compression);
1348
1349        off = btrfs_file_extent_inline_start(ei);
1350        len = btrfs_file_extent_inline_len(path->nodes[0], ei);
1351
1352        ret = fs_path_add_from_extent_buffer(dest, path->nodes[0], off, len);
1353
1354out:
1355        btrfs_free_path(path);
1356        return ret;
1357}
1358
1359/*
1360 * Helper function to generate a file name that is unique in the root of
1361 * send_root and parent_root. This is used to generate names for orphan inodes.
1362 */
1363static int gen_unique_name(struct send_ctx *sctx,
1364                           u64 ino, u64 gen,
1365                           struct fs_path *dest)
1366{
1367        int ret = 0;
1368        struct btrfs_path *path;
1369        struct btrfs_dir_item *di;
1370        char tmp[64];
1371        int len;
1372        u64 idx = 0;
1373
1374        path = alloc_path_for_send();
1375        if (!path)
1376                return -ENOMEM;
1377
1378        while (1) {
1379                len = snprintf(tmp, sizeof(tmp) - 1, "o%llu-%llu-%llu",
1380                                ino, gen, idx);
1381                if (len >= sizeof(tmp)) {
1382                        /* should really not happen */
1383                        ret = -EOVERFLOW;
1384                        goto out;
1385                }
1386
1387                di = btrfs_lookup_dir_item(NULL, sctx->send_root,
1388                                path, BTRFS_FIRST_FREE_OBJECTID,
1389                                tmp, strlen(tmp), 0);
1390                btrfs_release_path(path);
1391                if (IS_ERR(di)) {
1392                        ret = PTR_ERR(di);
1393                        goto out;
1394                }
1395                if (di) {
1396                        /* not unique, try again */
1397                        idx++;
1398                        continue;
1399                }
1400
1401                if (!sctx->parent_root) {
1402                        /* unique */
1403                        ret = 0;
1404                        break;
1405                }
1406
1407                di = btrfs_lookup_dir_item(NULL, sctx->parent_root,
1408                                path, BTRFS_FIRST_FREE_OBJECTID,
1409                                tmp, strlen(tmp), 0);
1410                btrfs_release_path(path);
1411                if (IS_ERR(di)) {
1412                        ret = PTR_ERR(di);
1413                        goto out;
1414                }
1415                if (di) {
1416                        /* not unique, try again */
1417                        idx++;
1418                        continue;
1419                }
1420                /* unique */
1421                break;
1422        }
1423
1424        ret = fs_path_add(dest, tmp, strlen(tmp));
1425
1426out:
1427        btrfs_free_path(path);
1428        return ret;
1429}
1430
1431enum inode_state {
1432        inode_state_no_change,
1433        inode_state_will_create,
1434        inode_state_did_create,
1435        inode_state_will_delete,
1436        inode_state_did_delete,
1437};
1438
1439static int get_cur_inode_state(struct send_ctx *sctx, u64 ino, u64 gen)
1440{
1441        int ret;
1442        int left_ret;
1443        int right_ret;
1444        u64 left_gen;
1445        u64 right_gen;
1446
1447        ret = get_inode_info(sctx->send_root, ino, NULL, &left_gen, NULL, NULL,
1448                        NULL, NULL);
1449        if (ret < 0 && ret != -ENOENT)
1450                goto out;
1451        left_ret = ret;
1452
1453        if (!sctx->parent_root) {
1454                right_ret = -ENOENT;
1455        } else {
1456                ret = get_inode_info(sctx->parent_root, ino, NULL, &right_gen,
1457                                NULL, NULL, NULL, NULL);
1458                if (ret < 0 && ret != -ENOENT)
1459                        goto out;
1460                right_ret = ret;
1461        }
1462
1463        if (!left_ret && !right_ret) {
1464                if (left_gen == gen && right_gen == gen) {
1465                        ret = inode_state_no_change;
1466                } else if (left_gen == gen) {
1467                        if (ino < sctx->send_progress)
1468                                ret = inode_state_did_create;
1469                        else
1470                                ret = inode_state_will_create;
1471                } else if (right_gen == gen) {
1472                        if (ino < sctx->send_progress)
1473                                ret = inode_state_did_delete;
1474                        else
1475                                ret = inode_state_will_delete;
1476                } else  {
1477                        ret = -ENOENT;
1478                }
1479        } else if (!left_ret) {
1480                if (left_gen == gen) {
1481                        if (ino < sctx->send_progress)
1482                                ret = inode_state_did_create;
1483                        else
1484                                ret = inode_state_will_create;
1485                } else {
1486                        ret = -ENOENT;
1487                }
1488        } else if (!right_ret) {
1489                if (right_gen == gen) {
1490                        if (ino < sctx->send_progress)
1491                                ret = inode_state_did_delete;
1492                        else
1493                                ret = inode_state_will_delete;
1494                } else {
1495                        ret = -ENOENT;
1496                }
1497        } else {
1498                ret = -ENOENT;
1499        }
1500
1501out:
1502        return ret;
1503}
1504
1505static int is_inode_existent(struct send_ctx *sctx, u64 ino, u64 gen)
1506{
1507        int ret;
1508
1509        ret = get_cur_inode_state(sctx, ino, gen);
1510        if (ret < 0)
1511                goto out;
1512
1513        if (ret == inode_state_no_change ||
1514            ret == inode_state_did_create ||
1515            ret == inode_state_will_delete)
1516                ret = 1;
1517        else
1518                ret = 0;
1519
1520out:
1521        return ret;
1522}
1523
1524/*
1525 * Helper function to lookup a dir item in a dir.
1526 */
1527static int lookup_dir_item_inode(struct btrfs_root *root,
1528                                 u64 dir, const char *name, int name_len,
1529                                 u64 *found_inode,
1530                                 u8 *found_type)
1531{
1532        int ret = 0;
1533        struct btrfs_dir_item *di;
1534        struct btrfs_key key;
1535        struct btrfs_path *path;
1536
1537        path = alloc_path_for_send();
1538        if (!path)
1539                return -ENOMEM;
1540
1541        di = btrfs_lookup_dir_item(NULL, root, path,
1542                        dir, name, name_len, 0);
1543        if (!di) {
1544                ret = -ENOENT;
1545                goto out;
1546        }
1547        if (IS_ERR(di)) {
1548                ret = PTR_ERR(di);
1549                goto out;
1550        }
1551        btrfs_dir_item_key_to_cpu(path->nodes[0], di, &key);
1552        *found_inode = key.objectid;
1553        *found_type = btrfs_dir_type(path->nodes[0], di);
1554
1555out:
1556        btrfs_free_path(path);
1557        return ret;
1558}
1559
1560/*
1561 * Looks up the first btrfs_inode_ref of a given ino. It returns the parent dir,
1562 * generation of the parent dir and the name of the dir entry.
1563 */
1564static int get_first_ref(struct send_ctx *sctx,
1565                         struct btrfs_root *root, u64 ino,
1566                         u64 *dir, u64 *dir_gen, struct fs_path *name)
1567{
1568        int ret;
1569        struct btrfs_key key;
1570        struct btrfs_key found_key;
1571        struct btrfs_path *path;
1572        int len;
1573        u64 parent_dir;
1574
1575        path = alloc_path_for_send();
1576        if (!path)
1577                return -ENOMEM;
1578
1579        key.objectid = ino;
1580        key.type = BTRFS_INODE_REF_KEY;
1581        key.offset = 0;
1582
1583        ret = btrfs_search_slot_for_read(root, &key, path, 1, 0);
1584        if (ret < 0)
1585                goto out;
1586        if (!ret)
1587                btrfs_item_key_to_cpu(path->nodes[0], &found_key,
1588                                path->slots[0]);
1589        if (ret || found_key.objectid != ino ||
1590            (found_key.type != BTRFS_INODE_REF_KEY &&
1591             found_key.type != BTRFS_INODE_EXTREF_KEY)) {
1592                ret = -ENOENT;
1593                goto out;
1594        }
1595
1596        if (key.type == BTRFS_INODE_REF_KEY) {
1597                struct btrfs_inode_ref *iref;
1598                iref = btrfs_item_ptr(path->nodes[0], path->slots[0],
1599                                      struct btrfs_inode_ref);
1600                len = btrfs_inode_ref_name_len(path->nodes[0], iref);
1601                ret = fs_path_add_from_extent_buffer(name, path->nodes[0],
1602                                                     (unsigned long)(iref + 1),
1603                                                     len);
1604                parent_dir = found_key.offset;
1605        } else {
1606                struct btrfs_inode_extref *extref;
1607                extref = btrfs_item_ptr(path->nodes[0], path->slots[0],
1608                                        struct btrfs_inode_extref);
1609                len = btrfs_inode_extref_name_len(path->nodes[0], extref);
1610                ret = fs_path_add_from_extent_buffer(name, path->nodes[0],
1611                                        (unsigned long)&extref->name, len);
1612                parent_dir = btrfs_inode_extref_parent(path->nodes[0], extref);
1613        }
1614        if (ret < 0)
1615                goto out;
1616        btrfs_release_path(path);
1617
1618        ret = get_inode_info(root, parent_dir, NULL, dir_gen, NULL, NULL,
1619                        NULL, NULL);
1620        if (ret < 0)
1621                goto out;
1622
1623        *dir = parent_dir;
1624
1625out:
1626        btrfs_free_path(path);
1627        return ret;
1628}
1629
1630static int is_first_ref(struct send_ctx *sctx,
1631                        struct btrfs_root *root,
1632                        u64 ino, u64 dir,
1633                        const char *name, int name_len)
1634{
1635        int ret;
1636        struct fs_path *tmp_name;
1637        u64 tmp_dir;
1638        u64 tmp_dir_gen;
1639
1640        tmp_name = fs_path_alloc(sctx);
1641        if (!tmp_name)
1642                return -ENOMEM;
1643
1644        ret = get_first_ref(sctx, root, ino, &tmp_dir, &tmp_dir_gen, tmp_name);
1645        if (ret < 0)
1646                goto out;
1647
1648        if (dir != tmp_dir || name_len != fs_path_len(tmp_name)) {
1649                ret = 0;
1650                goto out;
1651        }
1652
1653        ret = !memcmp(tmp_name->start, name, name_len);
1654
1655out:
1656        fs_path_free(sctx, tmp_name);
1657        return ret;
1658}
1659
1660/*
1661 * Used by process_recorded_refs to determine if a new ref would overwrite an
1662 * already existing ref. In case it detects an overwrite, it returns the
1663 * inode/gen in who_ino/who_gen.
1664 * When an overwrite is detected, process_recorded_refs does proper orphanizing
1665 * to make sure later references to the overwritten inode are possible.
1666 * Orphanizing is however only required for the first ref of an inode.
1667 * process_recorded_refs does an additional is_first_ref check to see if
1668 * orphanizing is really required.
1669 */
1670static int will_overwrite_ref(struct send_ctx *sctx, u64 dir, u64 dir_gen,
1671                              const char *name, int name_len,
1672                              u64 *who_ino, u64 *who_gen)
1673{
1674        int ret = 0;
1675        u64 other_inode = 0;
1676        u8 other_type = 0;
1677
1678        if (!sctx->parent_root)
1679                goto out;
1680
1681        ret = is_inode_existent(sctx, dir, dir_gen);
1682        if (ret <= 0)
1683                goto out;
1684
1685        ret = lookup_dir_item_inode(sctx->parent_root, dir, name, name_len,
1686                        &other_inode, &other_type);
1687        if (ret < 0 && ret != -ENOENT)
1688                goto out;
1689        if (ret) {
1690                ret = 0;
1691                goto out;
1692        }
1693
1694        /*
1695         * Check if the overwritten ref was already processed. If yes, the ref
1696         * was already unlinked/moved, so we can safely assume that we will not
1697         * overwrite anything at this point in time.
1698         */
1699        if (other_inode > sctx->send_progress) {
1700                ret = get_inode_info(sctx->parent_root, other_inode, NULL,
1701                                who_gen, NULL, NULL, NULL, NULL);
1702                if (ret < 0)
1703                        goto out;
1704
1705                ret = 1;
1706                *who_ino = other_inode;
1707        } else {
1708                ret = 0;
1709        }
1710
1711out:
1712        return ret;
1713}
1714
1715/*
1716 * Checks if the ref was overwritten by an already processed inode. This is
1717 * used by __get_cur_name_and_parent to find out if the ref was orphanized and
1718 * thus the orphan name needs be used.
1719 * process_recorded_refs also uses it to avoid unlinking of refs that were
1720 * overwritten.
1721 */
1722static int did_overwrite_ref(struct send_ctx *sctx,
1723                            u64 dir, u64 dir_gen,
1724                            u64 ino, u64 ino_gen,
1725                            const char *name, int name_len)
1726{
1727        int ret = 0;
1728        u64 gen;
1729        u64 ow_inode;
1730        u8 other_type;
1731
1732        if (!sctx->parent_root)
1733                goto out;
1734
1735        ret = is_inode_existent(sctx, dir, dir_gen);
1736        if (ret <= 0)
1737                goto out;
1738
1739        /* check if the ref was overwritten by another ref */
1740        ret = lookup_dir_item_inode(sctx->send_root, dir, name, name_len,
1741                        &ow_inode, &other_type);
1742        if (ret < 0 && ret != -ENOENT)
1743                goto out;
1744        if (ret) {
1745                /* was never and will never be overwritten */
1746                ret = 0;
1747                goto out;
1748        }
1749
1750        ret = get_inode_info(sctx->send_root, ow_inode, NULL, &gen, NULL, NULL,
1751                        NULL, NULL);
1752        if (ret < 0)
1753                goto out;
1754
1755        if (ow_inode == ino && gen == ino_gen) {
1756                ret = 0;
1757                goto out;
1758        }
1759
1760        /* we know that it is or will be overwritten. check this now */
1761        if (ow_inode < sctx->send_progress)
1762                ret = 1;
1763        else
1764                ret = 0;
1765
1766out:
1767        return ret;
1768}
1769
1770/*
1771 * Same as did_overwrite_ref, but also checks if it is the first ref of an inode
1772 * that got overwritten. This is used by process_recorded_refs to determine
1773 * if it has to use the path as returned by get_cur_path or the orphan name.
1774 */
1775static int did_overwrite_first_ref(struct send_ctx *sctx, u64 ino, u64 gen)
1776{
1777        int ret = 0;
1778        struct fs_path *name = NULL;
1779        u64 dir;
1780        u64 dir_gen;
1781
1782        if (!sctx->parent_root)
1783                goto out;
1784
1785        name = fs_path_alloc(sctx);
1786        if (!name)
1787                return -ENOMEM;
1788
1789        ret = get_first_ref(sctx, sctx->parent_root, ino, &dir, &dir_gen, name);
1790        if (ret < 0)
1791                goto out;
1792
1793        ret = did_overwrite_ref(sctx, dir, dir_gen, ino, gen,
1794                        name->start, fs_path_len(name));
1795
1796out:
1797        fs_path_free(sctx, name);
1798        return ret;
1799}
1800
1801/*
1802 * Insert a name cache entry. On 32bit kernels the radix tree index is 32bit,
1803 * so we need to do some special handling in case we have clashes. This function
1804 * takes care of this with the help of name_cache_entry::radix_list.
1805 * In case of error, nce is kfreed.
1806 */
1807static int name_cache_insert(struct send_ctx *sctx,
1808                             struct name_cache_entry *nce)
1809{
1810        int ret = 0;
1811        struct list_head *nce_head;
1812
1813        nce_head = radix_tree_lookup(&sctx->name_cache,
1814                        (unsigned long)nce->ino);
1815        if (!nce_head) {
1816                nce_head = kmalloc(sizeof(*nce_head), GFP_NOFS);
1817                if (!nce_head)
1818                        return -ENOMEM;
1819                INIT_LIST_HEAD(nce_head);
1820
1821                ret = radix_tree_insert(&sctx->name_cache, nce->ino, nce_head);
1822                if (ret < 0) {
1823                        kfree(nce_head);
1824                        kfree(nce);
1825                        return ret;
1826                }
1827        }
1828        list_add_tail(&nce->radix_list, nce_head);
1829        list_add_tail(&nce->list, &sctx->name_cache_list);
1830        sctx->name_cache_size++;
1831
1832        return ret;
1833}
1834
1835static void name_cache_delete(struct send_ctx *sctx,
1836                              struct name_cache_entry *nce)
1837{
1838        struct list_head *nce_head;
1839
1840        nce_head = radix_tree_lookup(&sctx->name_cache,
1841                        (unsigned long)nce->ino);
1842        BUG_ON(!nce_head);
1843
1844        list_del(&nce->radix_list);
1845        list_del(&nce->list);
1846        sctx->name_cache_size--;
1847
1848        if (list_empty(nce_head)) {
1849                radix_tree_delete(&sctx->name_cache, (unsigned long)nce->ino);
1850                kfree(nce_head);
1851        }
1852}
1853
1854static struct name_cache_entry *name_cache_search(struct send_ctx *sctx,
1855                                                    u64 ino, u64 gen)
1856{
1857        struct list_head *nce_head;
1858        struct name_cache_entry *cur;
1859
1860        nce_head = radix_tree_lookup(&sctx->name_cache, (unsigned long)ino);
1861        if (!nce_head)
1862                return NULL;
1863
1864        list_for_each_entry(cur, nce_head, radix_list) {
1865                if (cur->ino == ino && cur->gen == gen)
1866                        return cur;
1867        }
1868        return NULL;
1869}
1870
1871/*
1872 * Removes the entry from the list and adds it back to the end. This marks the
1873 * entry as recently used so that name_cache_clean_unused does not remove it.
1874 */
1875static void name_cache_used(struct send_ctx *sctx, struct name_cache_entry *nce)
1876{
1877        list_del(&nce->list);
1878        list_add_tail(&nce->list, &sctx->name_cache_list);
1879}
1880
1881/*
1882 * Remove some entries from the beginning of name_cache_list.
1883 */
1884static void name_cache_clean_unused(struct send_ctx *sctx)
1885{
1886        struct name_cache_entry *nce;
1887
1888        if (sctx->name_cache_size < SEND_CTX_NAME_CACHE_CLEAN_SIZE)
1889                return;
1890
1891        while (sctx->name_cache_size > SEND_CTX_MAX_NAME_CACHE_SIZE) {
1892                nce = list_entry(sctx->name_cache_list.next,
1893                                struct name_cache_entry, list);
1894                name_cache_delete(sctx, nce);
1895                kfree(nce);
1896        }
1897}
1898
1899static void name_cache_free(struct send_ctx *sctx)
1900{
1901        struct name_cache_entry *nce;
1902
1903        while (!list_empty(&sctx->name_cache_list)) {
1904                nce = list_entry(sctx->name_cache_list.next,
1905                                struct name_cache_entry, list);
1906                name_cache_delete(sctx, nce);
1907                kfree(nce);
1908        }
1909}
1910
1911/*
1912 * Used by get_cur_path for each ref up to the root.
1913 * Returns 0 if it succeeded.
1914 * Returns 1 if the inode is not existent or got overwritten. In that case, the
1915 * name is an orphan name. This instructs get_cur_path to stop iterating. If 1
1916 * is returned, parent_ino/parent_gen are not guaranteed to be valid.
1917 * Returns <0 in case of error.
1918 */
1919static int __get_cur_name_and_parent(struct send_ctx *sctx,
1920                                     u64 ino, u64 gen,
1921                                     u64 *parent_ino,
1922                                     u64 *parent_gen,
1923                                     struct fs_path *dest)
1924{
1925        int ret;
1926        int nce_ret;
1927        struct btrfs_path *path = NULL;
1928        struct name_cache_entry *nce = NULL;
1929
1930        /*
1931         * First check if we already did a call to this function with the same
1932         * ino/gen. If yes, check if the cache entry is still up-to-date. If yes
1933         * return the cached result.
1934         */
1935        nce = name_cache_search(sctx, ino, gen);
1936        if (nce) {
1937                if (ino < sctx->send_progress && nce->need_later_update) {
1938                        name_cache_delete(sctx, nce);
1939                        kfree(nce);
1940                        nce = NULL;
1941                } else {
1942                        name_cache_used(sctx, nce);
1943                        *parent_ino = nce->parent_ino;
1944                        *parent_gen = nce->parent_gen;
1945                        ret = fs_path_add(dest, nce->name, nce->name_len);
1946                        if (ret < 0)
1947                                goto out;
1948                        ret = nce->ret;
1949                        goto out;
1950                }
1951        }
1952
1953        path = alloc_path_for_send();
1954        if (!path)
1955                return -ENOMEM;
1956
1957        /*
1958         * If the inode is not existent yet, add the orphan name and return 1.
1959         * This should only happen for the parent dir that we determine in
1960         * __record_new_ref
1961         */
1962        ret = is_inode_existent(sctx, ino, gen);
1963        if (ret < 0)
1964                goto out;
1965
1966        if (!ret) {
1967                ret = gen_unique_name(sctx, ino, gen, dest);
1968                if (ret < 0)
1969                        goto out;
1970                ret = 1;
1971                goto out_cache;
1972        }
1973
1974        /*
1975         * Depending on whether the inode was already processed or not, use
1976         * send_root or parent_root for ref lookup.
1977         */
1978        if (ino < sctx->send_progress)
1979                ret = get_first_ref(sctx, sctx->send_root, ino,
1980                                parent_ino, parent_gen, dest);
1981        else
1982                ret = get_first_ref(sctx, sctx->parent_root, ino,
1983                                parent_ino, parent_gen, dest);
1984        if (ret < 0)
1985                goto out;
1986
1987        /*
1988         * Check if the ref was overwritten by an inode's ref that was processed
1989         * earlier. If yes, treat as orphan and return 1.
1990         */
1991        ret = did_overwrite_ref(sctx, *parent_ino, *parent_gen, ino, gen,
1992                        dest->start, dest->end - dest->start);
1993        if (ret < 0)
1994                goto out;
1995        if (ret) {
1996                fs_path_reset(dest);
1997                ret = gen_unique_name(sctx, ino, gen, dest);
1998                if (ret < 0)
1999                        goto out;
2000                ret = 1;
2001        }
2002
2003out_cache:
2004        /*
2005         * Store the result of the lookup in the name cache.
2006         */
2007        nce = kmalloc(sizeof(*nce) + fs_path_len(dest) + 1, GFP_NOFS);
2008        if (!nce) {
2009                ret = -ENOMEM;
2010                goto out;
2011        }
2012
2013        nce->ino = ino;
2014        nce->gen = gen;
2015        nce->parent_ino = *parent_ino;
2016        nce->parent_gen = *parent_gen;
2017        nce->name_len = fs_path_len(dest);
2018        nce->ret = ret;
2019        strcpy(nce->name, dest->start);
2020
2021        if (ino < sctx->send_progress)
2022                nce->need_later_update = 0;
2023        else
2024                nce->need_later_update = 1;
2025
2026        nce_ret = name_cache_insert(sctx, nce);
2027        if (nce_ret < 0)
2028                ret = nce_ret;
2029        name_cache_clean_unused(sctx);
2030
2031out:
2032        btrfs_free_path(path);
2033        return ret;
2034}
2035
2036/*
2037 * Magic happens here. This function returns the first ref to an inode as it
2038 * would look like while receiving the stream at this point in time.
2039 * We walk the path up to the root. For every inode in between, we check if it
2040 * was already processed/sent. If yes, we continue with the parent as found
2041 * in send_root. If not, we continue with the parent as found in parent_root.
2042 * If we encounter an inode that was deleted at this point in time, we use the
2043 * inodes "orphan" name instead of the real name and stop. Same with new inodes
2044 * that were not created yet and overwritten inodes/refs.
2045 *
2046 * When do we have have orphan inodes:
2047 * 1. When an inode is freshly created and thus no valid refs are available yet
2048 * 2. When a directory lost all it's refs (deleted) but still has dir items
2049 *    inside which were not processed yet (pending for move/delete). If anyone
2050 *    tried to get the path to the dir items, it would get a path inside that
2051 *    orphan directory.
2052 * 3. When an inode is moved around or gets new links, it may overwrite the ref
2053 *    of an unprocessed inode. If in that case the first ref would be
2054 *    overwritten, the overwritten inode gets "orphanized". Later when we
2055 *    process this overwritten inode, it is restored at a new place by moving
2056 *    the orphan inode.
2057 *
2058 * sctx->send_progress tells this function at which point in time receiving
2059 * would be.
2060 */
2061static int get_cur_path(struct send_ctx *sctx, u64 ino, u64 gen,
2062                        struct fs_path *dest)
2063{
2064        int ret = 0;
2065        struct fs_path *name = NULL;
2066        u64 parent_inode = 0;
2067        u64 parent_gen = 0;
2068        int stop = 0;
2069
2070        name = fs_path_alloc(sctx);
2071        if (!name) {
2072                ret = -ENOMEM;
2073                goto out;
2074        }
2075
2076        dest->reversed = 1;
2077        fs_path_reset(dest);
2078
2079        while (!stop && ino != BTRFS_FIRST_FREE_OBJECTID) {
2080                fs_path_reset(name);
2081
2082                ret = __get_cur_name_and_parent(sctx, ino, gen,
2083                                &parent_inode, &parent_gen, name);
2084                if (ret < 0)
2085                        goto out;
2086                if (ret)
2087                        stop = 1;
2088
2089                ret = fs_path_add_path(dest, name);
2090                if (ret < 0)
2091                        goto out;
2092
2093                ino = parent_inode;
2094                gen = parent_gen;
2095        }
2096
2097out:
2098        fs_path_free(sctx, name);
2099        if (!ret)
2100                fs_path_unreverse(dest);
2101        return ret;
2102}
2103
2104/*
2105 * Called for regular files when sending extents data. Opens a struct file
2106 * to read from the file.
2107 */
2108static int open_cur_inode_file(struct send_ctx *sctx)
2109{
2110        int ret = 0;
2111        struct btrfs_key key;
2112        struct path path;
2113        struct inode *inode;
2114        struct dentry *dentry;
2115        struct file *filp;
2116        int new = 0;
2117
2118        if (sctx->cur_inode_filp)
2119                goto out;
2120
2121        key.objectid = sctx->cur_ino;
2122        key.type = BTRFS_INODE_ITEM_KEY;
2123        key.offset = 0;
2124
2125        inode = btrfs_iget(sctx->send_root->fs_info->sb, &key, sctx->send_root,
2126                        &new);
2127        if (IS_ERR(inode)) {
2128                ret = PTR_ERR(inode);
2129                goto out;
2130        }
2131
2132        dentry = d_obtain_alias(inode);
2133        inode = NULL;
2134        if (IS_ERR(dentry)) {
2135                ret = PTR_ERR(dentry);
2136                goto out;
2137        }
2138
2139        path.mnt = sctx->mnt;
2140        path.dentry = dentry;
2141        filp = dentry_open(&path, O_RDONLY | O_LARGEFILE, current_cred());
2142        dput(dentry);
2143        dentry = NULL;
2144        if (IS_ERR(filp)) {
2145                ret = PTR_ERR(filp);
2146                goto out;
2147        }
2148        sctx->cur_inode_filp = filp;
2149
2150out:
2151        /*
2152         * no xxxput required here as every vfs op
2153         * does it by itself on failure
2154         */
2155        return ret;
2156}
2157
2158/*
2159 * Closes the struct file that was created in open_cur_inode_file
2160 */
2161static int close_cur_inode_file(struct send_ctx *sctx)
2162{
2163        int ret = 0;
2164
2165        if (!sctx->cur_inode_filp)
2166                goto out;
2167
2168        ret = filp_close(sctx->cur_inode_filp, NULL);
2169        sctx->cur_inode_filp = NULL;
2170
2171out:
2172        return ret;
2173}
2174
2175/*
2176 * Sends a BTRFS_SEND_C_SUBVOL command/item to userspace
2177 */
2178static int send_subvol_begin(struct send_ctx *sctx)
2179{
2180        int ret;
2181        struct btrfs_root *send_root = sctx->send_root;
2182        struct btrfs_root *parent_root = sctx->parent_root;
2183        struct btrfs_path *path;
2184        struct btrfs_key key;
2185        struct btrfs_root_ref *ref;
2186        struct extent_buffer *leaf;
2187        char *name = NULL;
2188        int namelen;
2189
2190        path = alloc_path_for_send();
2191        if (!path)
2192                return -ENOMEM;
2193
2194        name = kmalloc(BTRFS_PATH_NAME_MAX, GFP_NOFS);
2195        if (!name) {
2196                btrfs_free_path(path);
2197                return -ENOMEM;
2198        }
2199
2200        key.objectid = send_root->objectid;
2201        key.type = BTRFS_ROOT_BACKREF_KEY;
2202        key.offset = 0;
2203
2204        ret = btrfs_search_slot_for_read(send_root->fs_info->tree_root,
2205                                &key, path, 1, 0);
2206        if (ret < 0)
2207                goto out;
2208        if (ret) {
2209                ret = -ENOENT;
2210                goto out;
2211        }
2212
2213        leaf = path->nodes[0];
2214        btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
2215        if (key.type != BTRFS_ROOT_BACKREF_KEY ||
2216            key.objectid != send_root->objectid) {
2217                ret = -ENOENT;
2218                goto out;
2219        }
2220        ref = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_root_ref);
2221        namelen = btrfs_root_ref_name_len(leaf, ref);
2222        read_extent_buffer(leaf, name, (unsigned long)(ref + 1), namelen);
2223        btrfs_release_path(path);
2224
2225        if (parent_root) {
2226                ret = begin_cmd(sctx, BTRFS_SEND_C_SNAPSHOT);
2227                if (ret < 0)
2228                        goto out;
2229        } else {
2230                ret = begin_cmd(sctx, BTRFS_SEND_C_SUBVOL);
2231                if (ret < 0)
2232                        goto out;
2233        }
2234
2235        TLV_PUT_STRING(sctx, BTRFS_SEND_A_PATH, name, namelen);
2236        TLV_PUT_UUID(sctx, BTRFS_SEND_A_UUID,
2237                        sctx->send_root->root_item.uuid);
2238        TLV_PUT_U64(sctx, BTRFS_SEND_A_CTRANSID,
2239                        sctx->send_root->root_item.ctransid);
2240        if (parent_root) {
2241                TLV_PUT_UUID(sctx, BTRFS_SEND_A_CLONE_UUID,
2242                                sctx->parent_root->root_item.uuid);
2243                TLV_PUT_U64(sctx, BTRFS_SEND_A_CLONE_CTRANSID,
2244                                sctx->parent_root->root_item.ctransid);
2245        }
2246
2247        ret = send_cmd(sctx);
2248
2249tlv_put_failure:
2250out:
2251        btrfs_free_path(path);
2252        kfree(name);
2253        return ret;
2254}
2255
2256static int send_truncate(struct send_ctx *sctx, u64 ino, u64 gen, u64 size)
2257{
2258        int ret = 0;
2259        struct fs_path *p;
2260
2261verbose_printk("btrfs: send_truncate %llu size=%llu\n", ino, size);
2262
2263        p = fs_path_alloc(sctx);
2264        if (!p)
2265                return -ENOMEM;
2266
2267        ret = begin_cmd(sctx, BTRFS_SEND_C_TRUNCATE);
2268        if (ret < 0)
2269                goto out;
2270
2271        ret = get_cur_path(sctx, ino, gen, p);
2272        if (ret < 0)
2273                goto out;
2274        TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, p);
2275        TLV_PUT_U64(sctx, BTRFS_SEND_A_SIZE, size);
2276
2277        ret = send_cmd(sctx);
2278
2279tlv_put_failure:
2280out:
2281        fs_path_free(sctx, p);
2282        return ret;
2283}
2284
2285static int send_chmod(struct send_ctx *sctx, u64 ino, u64 gen, u64 mode)
2286{
2287        int ret = 0;
2288        struct fs_path *p;
2289
2290verbose_printk("btrfs: send_chmod %llu mode=%llu\n", ino, mode);
2291
2292        p = fs_path_alloc(sctx);
2293        if (!p)
2294                return -ENOMEM;
2295
2296        ret = begin_cmd(sctx, BTRFS_SEND_C_CHMOD);
2297        if (ret < 0)
2298                goto out;
2299
2300        ret = get_cur_path(sctx, ino, gen, p);
2301        if (ret < 0)
2302                goto out;
2303        TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, p);
2304        TLV_PUT_U64(sctx, BTRFS_SEND_A_MODE, mode & 07777);
2305
2306        ret = send_cmd(sctx);
2307
2308tlv_put_failure:
2309out:
2310        fs_path_free(sctx, p);
2311        return ret;
2312}
2313
2314static int send_chown(struct send_ctx *sctx, u64 ino, u64 gen, u64 uid, u64 gid)
2315{
2316        int ret = 0;
2317        struct fs_path *p;
2318
2319verbose_printk("btrfs: send_chown %llu uid=%llu, gid=%llu\n", ino, uid, gid);
2320
2321        p = fs_path_alloc(sctx);
2322        if (!p)
2323                return -ENOMEM;
2324
2325        ret = begin_cmd(sctx, BTRFS_SEND_C_CHOWN);
2326        if (ret < 0)
2327                goto out;
2328
2329        ret = get_cur_path(sctx, ino, gen, p);
2330        if (ret < 0)
2331                goto out;
2332        TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, p);
2333        TLV_PUT_U64(sctx, BTRFS_SEND_A_UID, uid);
2334        TLV_PUT_U64(sctx, BTRFS_SEND_A_GID, gid);
2335
2336        ret = send_cmd(sctx);
2337
2338tlv_put_failure:
2339out:
2340        fs_path_free(sctx, p);
2341        return ret;
2342}
2343
2344static int send_utimes(struct send_ctx *sctx, u64 ino, u64 gen)
2345{
2346        int ret = 0;
2347        struct fs_path *p = NULL;
2348        struct btrfs_inode_item *ii;
2349        struct btrfs_path *path = NULL;
2350        struct extent_buffer *eb;
2351        struct btrfs_key key;
2352        int slot;
2353
2354verbose_printk("btrfs: send_utimes %llu\n", ino);
2355
2356        p = fs_path_alloc(sctx);
2357        if (!p)
2358                return -ENOMEM;
2359
2360        path = alloc_path_for_send();
2361        if (!path) {
2362                ret = -ENOMEM;
2363                goto out;
2364        }
2365
2366        key.objectid = ino;
2367        key.type = BTRFS_INODE_ITEM_KEY;
2368        key.offset = 0;
2369        ret = btrfs_search_slot(NULL, sctx->send_root, &key, path, 0, 0);
2370        if (ret < 0)
2371                goto out;
2372
2373        eb = path->nodes[0];
2374        slot = path->slots[0];
2375        ii = btrfs_item_ptr(eb, slot, struct btrfs_inode_item);
2376
2377        ret = begin_cmd(sctx, BTRFS_SEND_C_UTIMES);
2378        if (ret < 0)
2379                goto out;
2380
2381        ret = get_cur_path(sctx, ino, gen, p);
2382        if (ret < 0)
2383                goto out;
2384        TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, p);
2385        TLV_PUT_BTRFS_TIMESPEC(sctx, BTRFS_SEND_A_ATIME, eb,
2386                        btrfs_inode_atime(ii));
2387        TLV_PUT_BTRFS_TIMESPEC(sctx, BTRFS_SEND_A_MTIME, eb,
2388                        btrfs_inode_mtime(ii));
2389        TLV_PUT_BTRFS_TIMESPEC(sctx, BTRFS_SEND_A_CTIME, eb,
2390                        btrfs_inode_ctime(ii));
2391        /* TODO Add otime support when the otime patches get into upstream */
2392
2393        ret = send_cmd(sctx);
2394
2395tlv_put_failure:
2396out:
2397        fs_path_free(sctx, p);
2398        btrfs_free_path(path);
2399        return ret;
2400}
2401
2402/*
2403 * Sends a BTRFS_SEND_C_MKXXX or SYMLINK command to user space. We don't have
2404 * a valid path yet because we did not process the refs yet. So, the inode
2405 * is created as orphan.
2406 */
2407static int send_create_inode(struct send_ctx *sctx, u64 ino)
2408{
2409        int ret = 0;
2410        struct fs_path *p;
2411        int cmd;
2412        u64 gen;
2413        u64 mode;
2414        u64 rdev;
2415
2416verbose_printk("btrfs: send_create_inode %llu\n", ino);
2417
2418        p = fs_path_alloc(sctx);
2419        if (!p)
2420                return -ENOMEM;
2421
2422        ret = get_inode_info(sctx->send_root, ino, NULL, &gen, &mode, NULL,
2423                        NULL, &rdev);
2424        if (ret < 0)
2425                goto out;
2426
2427        if (S_ISREG(mode)) {
2428                cmd = BTRFS_SEND_C_MKFILE;
2429        } else if (S_ISDIR(mode)) {
2430                cmd = BTRFS_SEND_C_MKDIR;
2431        } else if (S_ISLNK(mode)) {
2432                cmd = BTRFS_SEND_C_SYMLINK;
2433        } else if (S_ISCHR(mode) || S_ISBLK(mode)) {
2434                cmd = BTRFS_SEND_C_MKNOD;
2435        } else if (S_ISFIFO(mode)) {
2436                cmd = BTRFS_SEND_C_MKFIFO;
2437        } else if (S_ISSOCK(mode)) {
2438                cmd = BTRFS_SEND_C_MKSOCK;
2439        } else {
2440                printk(KERN_WARNING "btrfs: unexpected inode type %o",
2441                                (int)(mode & S_IFMT));
2442                ret = -ENOTSUPP;
2443                goto out;
2444        }
2445
2446        ret = begin_cmd(sctx, cmd);
2447        if (ret < 0)
2448                goto out;
2449
2450        ret = gen_unique_name(sctx, ino, gen, p);
2451        if (ret < 0)
2452                goto out;
2453
2454        TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, p);
2455        TLV_PUT_U64(sctx, BTRFS_SEND_A_INO, ino);
2456
2457        if (S_ISLNK(mode)) {
2458                fs_path_reset(p);
2459                ret = read_symlink(sctx, sctx->send_root, ino, p);
2460                if (ret < 0)
2461                        goto out;
2462                TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH_LINK, p);
2463        } else if (S_ISCHR(mode) || S_ISBLK(mode) ||
2464                   S_ISFIFO(mode) || S_ISSOCK(mode)) {
2465                TLV_PUT_U64(sctx, BTRFS_SEND_A_RDEV, new_encode_dev(rdev));
2466                TLV_PUT_U64(sctx, BTRFS_SEND_A_MODE, mode);
2467        }
2468
2469        ret = send_cmd(sctx);
2470        if (ret < 0)
2471                goto out;
2472
2473
2474tlv_put_failure:
2475out:
2476        fs_path_free(sctx, p);
2477        return ret;
2478}
2479
2480/*
2481 * We need some special handling for inodes that get processed before the parent
2482 * directory got created. See process_recorded_refs for details.
2483 * This function does the check if we already created the dir out of order.
2484 */
2485static int did_create_dir(struct send_ctx *sctx, u64 dir)
2486{
2487        int ret = 0;
2488        struct btrfs_path *path = NULL;
2489        struct btrfs_key key;
2490        struct btrfs_key found_key;
2491        struct btrfs_key di_key;
2492        struct extent_buffer *eb;
2493        struct btrfs_dir_item *di;
2494        int slot;
2495
2496        path = alloc_path_for_send();
2497        if (!path) {
2498                ret = -ENOMEM;
2499                goto out;
2500        }
2501
2502        key.objectid = dir;
2503        key.type = BTRFS_DIR_INDEX_KEY;
2504        key.offset = 0;
2505        while (1) {
2506                ret = btrfs_search_slot_for_read(sctx->send_root, &key, path,
2507                                1, 0);
2508                if (ret < 0)
2509                        goto out;
2510                if (!ret) {
2511                        eb = path->nodes[0];
2512                        slot = path->slots[0];
2513                        btrfs_item_key_to_cpu(eb, &found_key, slot);
2514                }
2515                if (ret || found_key.objectid != key.objectid ||
2516                    found_key.type != key.type) {
2517                        ret = 0;
2518                        goto out;
2519                }
2520
2521                di = btrfs_item_ptr(eb, slot, struct btrfs_dir_item);
2522                btrfs_dir_item_key_to_cpu(eb, di, &di_key);
2523
2524                if (di_key.objectid < sctx->send_progress) {
2525                        ret = 1;
2526                        goto out;
2527                }
2528
2529                key.offset = found_key.offset + 1;
2530                btrfs_release_path(path);
2531        }
2532
2533out:
2534        btrfs_free_path(path);
2535        return ret;
2536}
2537
2538/*
2539 * Only creates the inode if it is:
2540 * 1. Not a directory
2541 * 2. Or a directory which was not created already due to out of order
2542 *    directories. See did_create_dir and process_recorded_refs for details.
2543 */
2544static int send_create_inode_if_needed(struct send_ctx *sctx)
2545{
2546        int ret;
2547
2548        if (S_ISDIR(sctx->cur_inode_mode)) {
2549                ret = did_create_dir(sctx, sctx->cur_ino);
2550                if (ret < 0)
2551                        goto out;
2552                if (ret) {
2553                        ret = 0;
2554                        goto out;
2555                }
2556        }
2557
2558        ret = send_create_inode(sctx, sctx->cur_ino);
2559        if (ret < 0)
2560                goto out;
2561
2562out:
2563        return ret;
2564}
2565
2566struct recorded_ref {
2567        struct list_head list;
2568        char *dir_path;
2569        char *name;
2570        struct fs_path *full_path;
2571        u64 dir;
2572        u64 dir_gen;
2573        int dir_path_len;
2574        int name_len;
2575};
2576
2577/*
2578 * We need to process new refs before deleted refs, but compare_tree gives us
2579 * everything mixed. So we first record all refs and later process them.
2580 * This function is a helper to record one ref.
2581 */
2582static int record_ref(struct list_head *head, u64 dir,
2583                      u64 dir_gen, struct fs_path *path)
2584{
2585        struct recorded_ref *ref;
2586        char *tmp;
2587
2588        ref = kmalloc(sizeof(*ref), GFP_NOFS);
2589        if (!ref)
2590                return -ENOMEM;
2591
2592        ref->dir = dir;
2593        ref->dir_gen = dir_gen;
2594        ref->full_path = path;
2595
2596        tmp = strrchr(ref->full_path->start, '/');
2597        if (!tmp) {
2598                ref->name_len = ref->full_path->end - ref->full_path->start;
2599                ref->name = ref->full_path->start;
2600                ref->dir_path_len = 0;
2601                ref->dir_path = ref->full_path->start;
2602        } else {
2603                tmp++;
2604                ref->name_len = ref->full_path->end - tmp;
2605                ref->name = tmp;
2606                ref->dir_path = ref->full_path->start;
2607                ref->dir_path_len = ref->full_path->end -
2608                                ref->full_path->start - 1 - ref->name_len;
2609        }
2610
2611        list_add_tail(&ref->list, head);
2612        return 0;
2613}
2614
2615static void __free_recorded_refs(struct send_ctx *sctx, struct list_head *head)
2616{
2617        struct recorded_ref *cur;
2618
2619        while (!list_empty(head)) {
2620                cur = list_entry(head->next, struct recorded_ref, list);
2621                fs_path_free(sctx, cur->full_path);
2622                list_del(&cur->list);
2623                kfree(cur);
2624        }
2625}
2626
2627static void free_recorded_refs(struct send_ctx *sctx)
2628{
2629        __free_recorded_refs(sctx, &sctx->new_refs);
2630        __free_recorded_refs(sctx, &sctx->deleted_refs);
2631}
2632
2633/*
2634 * Renames/moves a file/dir to its orphan name. Used when the first
2635 * ref of an unprocessed inode gets overwritten and for all non empty
2636 * directories.
2637 */
2638static int orphanize_inode(struct send_ctx *sctx, u64 ino, u64 gen,
2639                          struct fs_path *path)
2640{
2641        int ret;
2642        struct fs_path *orphan;
2643
2644        orphan = fs_path_alloc(sctx);
2645        if (!orphan)
2646                return -ENOMEM;
2647
2648        ret = gen_unique_name(sctx, ino, gen, orphan);
2649        if (ret < 0)
2650                goto out;
2651
2652        ret = send_rename(sctx, path, orphan);
2653
2654out:
2655        fs_path_free(sctx, orphan);
2656        return ret;
2657}
2658
2659/*
2660 * Returns 1 if a directory can be removed at this point in time.
2661 * We check this by iterating all dir items and checking if the inode behind
2662 * the dir item was already processed.
2663 */
2664static int can_rmdir(struct send_ctx *sctx, u64 dir, u64 send_progress)
2665{
2666        int ret = 0;
2667        struct btrfs_root *root = sctx->parent_root;
2668        struct btrfs_path *path;
2669        struct btrfs_key key;
2670        struct btrfs_key found_key;
2671        struct btrfs_key loc;
2672        struct btrfs_dir_item *di;
2673
2674        /*
2675         * Don't try to rmdir the top/root subvolume dir.
2676         */
2677        if (dir == BTRFS_FIRST_FREE_OBJECTID)
2678                return 0;
2679
2680        path = alloc_path_for_send();
2681        if (!path)
2682                return -ENOMEM;
2683
2684        key.objectid = dir;
2685        key.type = BTRFS_DIR_INDEX_KEY;
2686        key.offset = 0;
2687
2688        while (1) {
2689                ret = btrfs_search_slot_for_read(root, &key, path, 1, 0);
2690                if (ret < 0)
2691                        goto out;
2692                if (!ret) {
2693                        btrfs_item_key_to_cpu(path->nodes[0], &found_key,
2694                                        path->slots[0]);
2695                }
2696                if (ret || found_key.objectid != key.objectid ||
2697                    found_key.type != key.type) {
2698                        break;
2699                }
2700
2701                di = btrfs_item_ptr(path->nodes[0], path->slots[0],
2702                                struct btrfs_dir_item);
2703                btrfs_dir_item_key_to_cpu(path->nodes[0], di, &loc);
2704
2705                if (loc.objectid > send_progress) {
2706                        ret = 0;
2707                        goto out;
2708                }
2709
2710                btrfs_release_path(path);
2711                key.offset = found_key.offset + 1;
2712        }
2713
2714        ret = 1;
2715
2716out:
2717        btrfs_free_path(path);
2718        return ret;
2719}
2720
2721/*
2722 * This does all the move/link/unlink/rmdir magic.
2723 */
2724static int process_recorded_refs(struct send_ctx *sctx)
2725{
2726        int ret = 0;
2727        struct recorded_ref *cur;
2728        struct recorded_ref *cur2;
2729        struct ulist *check_dirs = NULL;
2730        struct ulist_iterator uit;
2731        struct ulist_node *un;
2732        struct fs_path *valid_path = NULL;
2733        u64 ow_inode = 0;
2734        u64 ow_gen;
2735        int did_overwrite = 0;
2736        int is_orphan = 0;
2737
2738verbose_printk("btrfs: process_recorded_refs %llu\n", sctx->cur_ino);
2739
2740        /*
2741         * This should never happen as the root dir always has the same ref
2742         * which is always '..'
2743         */
2744        BUG_ON(sctx->cur_ino <= BTRFS_FIRST_FREE_OBJECTID);
2745
2746        valid_path = fs_path_alloc(sctx);
2747        if (!valid_path) {
2748                ret = -ENOMEM;
2749                goto out;
2750        }
2751
2752        check_dirs = ulist_alloc(GFP_NOFS);
2753        if (!check_dirs) {
2754                ret = -ENOMEM;
2755                goto out;
2756        }
2757
2758        /*
2759         * First, check if the first ref of the current inode was overwritten
2760         * before. If yes, we know that the current inode was already orphanized
2761         * and thus use the orphan name. If not, we can use get_cur_path to
2762         * get the path of the first ref as it would like while receiving at
2763         * this point in time.
2764         * New inodes are always orphan at the beginning, so force to use the
2765         * orphan name in this case.
2766         * The first ref is stored in valid_path and will be updated if it
2767         * gets moved around.
2768         */
2769        if (!sctx->cur_inode_new) {
2770                ret = did_overwrite_first_ref(sctx, sctx->cur_ino,
2771                                sctx->cur_inode_gen);
2772                if (ret < 0)
2773                        goto out;
2774                if (ret)
2775                        did_overwrite = 1;
2776        }
2777        if (sctx->cur_inode_new || did_overwrite) {
2778                ret = gen_unique_name(sctx, sctx->cur_ino,
2779                                sctx->cur_inode_gen, valid_path);
2780                if (ret < 0)
2781                        goto out;
2782                is_orphan = 1;
2783        } else {
2784                ret = get_cur_path(sctx, sctx->cur_ino, sctx->cur_inode_gen,
2785                                valid_path);
2786                if (ret < 0)
2787                        goto out;
2788        }
2789
2790        list_for_each_entry(cur, &sctx->new_refs, list) {
2791                /*
2792                 * We may have refs where the parent directory does not exist
2793                 * yet. This happens if the parent directories inum is higher
2794                 * the the current inum. To handle this case, we create the
2795                 * parent directory out of order. But we need to check if this
2796                 * did already happen before due to other refs in the same dir.
2797                 */
2798                ret = get_cur_inode_state(sctx, cur->dir, cur->dir_gen);
2799                if (ret < 0)
2800                        goto out;
2801                if (ret == inode_state_will_create) {
2802                        ret = 0;
2803                        /*
2804                         * First check if any of the current inodes refs did
2805                         * already create the dir.
2806                         */
2807                        list_for_each_entry(cur2, &sctx->new_refs, list) {
2808                                if (cur == cur2)
2809                                        break;
2810                                if (cur2->dir == cur->dir) {
2811                                        ret = 1;
2812                                        break;
2813                                }
2814                        }
2815
2816                        /*
2817                         * If that did not happen, check if a previous inode
2818                         * did already create the dir.
2819                         */
2820                        if (!ret)
2821                                ret = did_create_dir(sctx, cur->dir);
2822                        if (ret < 0)
2823                                goto out;
2824                        if (!ret) {
2825                                ret = send_create_inode(sctx, cur->dir);
2826                                if (ret < 0)
2827                                        goto out;
2828                        }
2829                }
2830
2831                /*
2832                 * Check if this new ref would overwrite the first ref of
2833                 * another unprocessed inode. If yes, orphanize the
2834                 * overwritten inode. If we find an overwritten ref that is
2835                 * not the first ref, simply unlink it.
2836                 */
2837                ret = will_overwrite_ref(sctx, cur->dir, cur->dir_gen,
2838                                cur->name, cur->name_len,
2839                                &ow_inode, &ow_gen);
2840                if (ret < 0)
2841                        goto out;
2842                if (ret) {
2843                        ret = is_first_ref(sctx, sctx->parent_root,
2844                                        ow_inode, cur->dir, cur->name,
2845                                        cur->name_len);
2846                        if (ret < 0)
2847                                goto out;
2848                        if (ret) {
2849                                ret = orphanize_inode(sctx, ow_inode, ow_gen,
2850                                                cur->full_path);
2851                                if (ret < 0)
2852                                        goto out;
2853                        } else {
2854                                ret = send_unlink(sctx, cur->full_path);
2855                                if (ret < 0)
2856                                        goto out;
2857                        }
2858                }
2859
2860                /*
2861                 * link/move the ref to the new place. If we have an orphan
2862                 * inode, move it and update valid_path. If not, link or move
2863                 * it depending on the inode mode.
2864                 */
2865                if (is_orphan) {
2866                        ret = send_rename(sctx, valid_path, cur->full_path);
2867                        if (ret < 0)
2868                                goto out;
2869                        is_orphan = 0;
2870                        ret = fs_path_copy(valid_path, cur->full_path);
2871                        if (ret < 0)
2872                                goto out;
2873                } else {
2874                        if (S_ISDIR(sctx->cur_inode_mode)) {
2875                                /*
2876                                 * Dirs can't be linked, so move it. For moved
2877                                 * dirs, we always have one new and one deleted
2878                                 * ref. The deleted ref is ignored later.
2879                                 */
2880                                ret = send_rename(sctx, valid_path,
2881                                                cur->full_path);
2882                                if (ret < 0)
2883                                        goto out;
2884                                ret = fs_path_copy(valid_path, cur->full_path);
2885                                if (ret < 0)
2886                                        goto out;
2887                        } else {
2888                                ret = send_link(sctx, cur->full_path,
2889                                                valid_path);
2890                                if (ret < 0)
2891                                        goto out;
2892                        }
2893                }
2894                ret = ulist_add(check_dirs, cur->dir, cur->dir_gen,
2895                                GFP_NOFS);
2896                if (ret < 0)
2897                        goto out;
2898        }
2899
2900        if (S_ISDIR(sctx->cur_inode_mode) && sctx->cur_inode_deleted) {
2901                /*
2902                 * Check if we can already rmdir the directory. If not,
2903                 * orphanize it. For every dir item inside that gets deleted
2904                 * later, we do this check again and rmdir it then if possible.
2905                 * See the use of check_dirs for more details.
2906                 */
2907                ret = can_rmdir(sctx, sctx->cur_ino, sctx->cur_ino);
2908                if (ret < 0)
2909                        goto out;
2910                if (ret) {
2911                        ret = send_rmdir(sctx, valid_path);
2912                        if (ret < 0)
2913                                goto out;
2914                } else if (!is_orphan) {
2915                        ret = orphanize_inode(sctx, sctx->cur_ino,
2916                                        sctx->cur_inode_gen, valid_path);
2917                        if (ret < 0)
2918                                goto out;
2919                        is_orphan = 1;
2920                }
2921
2922                list_for_each_entry(cur, &sctx->deleted_refs, list) {
2923                        ret = ulist_add(check_dirs, cur->dir, cur->dir_gen,
2924                                        GFP_NOFS);
2925                        if (ret < 0)
2926                                goto out;
2927                }
2928        } else if (S_ISDIR(sctx->cur_inode_mode) &&
2929                   !list_empty(&sctx->deleted_refs)) {
2930                /*
2931                 * We have a moved dir. Add the old parent to check_dirs
2932                 */
2933                cur = list_entry(sctx->deleted_refs.next, struct recorded_ref,
2934                                list);
2935                ret = ulist_add(check_dirs, cur->dir, cur->dir_gen,
2936                                GFP_NOFS);
2937                if (ret < 0)
2938                        goto out;
2939        } else if (!S_ISDIR(sctx->cur_inode_mode)) {
2940                /*
2941                 * We have a non dir inode. Go through all deleted refs and
2942                 * unlink them if they were not already overwritten by other
2943                 * inodes.
2944                 */
2945                list_for_each_entry(cur, &sctx->deleted_refs, list) {
2946                        ret = did_overwrite_ref(sctx, cur->dir, cur->dir_gen,
2947                                        sctx->cur_ino, sctx->cur_inode_gen,
2948                                        cur->name, cur->name_len);
2949                        if (ret < 0)
2950                                goto out;
2951                        if (!ret) {
2952                                ret = send_unlink(sctx, cur->full_path);
2953                                if (ret < 0)
2954                                        goto out;
2955                        }
2956                        ret = ulist_add(check_dirs, cur->dir, cur->dir_gen,
2957                                        GFP_NOFS);
2958                        if (ret < 0)
2959                                goto out;
2960                }
2961
2962                /*
2963                 * If the inode is still orphan, unlink the orphan. This may
2964                 * happen when a previous inode did overwrite the first ref
2965                 * of this inode and no new refs were added for the current
2966                 * inode. Unlinking does not mean that the inode is deleted in
2967                 * all cases. There may still be links to this inode in other
2968                 * places.
2969                 */
2970                if (is_orphan) {
2971                        ret = send_unlink(sctx, valid_path);
2972                        if (ret < 0)
2973                                goto out;
2974                }
2975        }
2976
2977        /*
2978         * We did collect all parent dirs where cur_inode was once located. We
2979         * now go through all these dirs and check if they are pending for
2980         * deletion and if it's finally possible to perform the rmdir now.
2981         * We also update the inode stats of the parent dirs here.
2982         */
2983        ULIST_ITER_INIT(&uit);
2984        while ((un = ulist_next(check_dirs, &uit))) {
2985                /*
2986                 * In case we had refs into dirs that were not processed yet,
2987                 * we don't need to do the utime and rmdir logic for these dirs.
2988                 * The dir will be processed later.
2989                 */
2990                if (un->val > sctx->cur_ino)
2991                        continue;
2992
2993                ret = get_cur_inode_state(sctx, un->val, un->aux);
2994                if (ret < 0)
2995                        goto out;
2996
2997                if (ret == inode_state_did_create ||
2998                    ret == inode_state_no_change) {
2999                        /* TODO delayed utimes */
3000                        ret = send_utimes(sctx, un->val, un->aux);
3001                        if (ret < 0)
3002                                goto out;
3003                } else if (ret == inode_state_did_delete) {
3004                        ret = can_rmdir(sctx, un->val, sctx->cur_ino);
3005                        if (ret < 0)
3006                                goto out;
3007                        if (ret) {
3008                                ret = get_cur_path(sctx, un->val, un->aux,
3009                                                valid_path);
3010                                if (ret < 0)
3011                                        goto out;
3012                                ret = send_rmdir(sctx, valid_path);
3013                                if (ret < 0)
3014                                        goto out;
3015                        }
3016                }
3017        }
3018
3019        ret = 0;
3020
3021out:
3022        free_recorded_refs(sctx);
3023        ulist_free(check_dirs);
3024        fs_path_free(sctx, valid_path);
3025        return ret;
3026}
3027
3028static int __record_new_ref(int num, u64 dir, int index,
3029                            struct fs_path *name,
3030                            void *ctx)
3031{
3032        int ret = 0;
3033        struct send_ctx *sctx = ctx;
3034        struct fs_path *p;
3035        u64 gen;
3036
3037        p = fs_path_alloc(sctx);
3038        if (!p)
3039                return -ENOMEM;
3040
3041        ret = get_inode_info(sctx->send_root, dir, NULL, &gen, NULL, NULL,
3042                        NULL, NULL);
3043        if (ret < 0)
3044                goto out;
3045
3046        ret = get_cur_path(sctx, dir, gen, p);
3047        if (ret < 0)
3048                goto out;
3049        ret = fs_path_add_path(p, name);
3050        if (ret < 0)
3051                goto out;
3052
3053        ret = record_ref(&sctx->new_refs, dir, gen, p);
3054
3055out:
3056        if (ret)
3057                fs_path_free(sctx, p);
3058        return ret;
3059}
3060
3061static int __record_deleted_ref(int num, u64 dir, int index,
3062                                struct fs_path *name,
3063                                void *ctx)
3064{
3065        int ret = 0;
3066        struct send_ctx *sctx = ctx;
3067        struct fs_path *p;
3068        u64 gen;
3069
3070        p = fs_path_alloc(sctx);
3071        if (!p)
3072                return -ENOMEM;
3073
3074        ret = get_inode_info(sctx->parent_root, dir, NULL, &gen, NULL, NULL,
3075                        NULL, NULL);
3076        if (ret < 0)
3077                goto out;
3078
3079        ret = get_cur_path(sctx, dir, gen, p);
3080        if (ret < 0)
3081                goto out;
3082        ret = fs_path_add_path(p, name);
3083        if (ret < 0)
3084                goto out;
3085
3086        ret = record_ref(&sctx->deleted_refs, dir, gen, p);
3087
3088out:
3089        if (ret)
3090                fs_path_free(sctx, p);
3091        return ret;
3092}
3093
3094static int record_new_ref(struct send_ctx *sctx)
3095{
3096        int ret;
3097
3098        ret = iterate_inode_ref(sctx, sctx->send_root, sctx->left_path,
3099                        sctx->cmp_key, 0, __record_new_ref, sctx);
3100        if (ret < 0)
3101                goto out;
3102        ret = 0;
3103
3104out:
3105        return ret;
3106}
3107
3108static int record_deleted_ref(struct send_ctx *sctx)
3109{
3110        int ret;
3111
3112        ret = iterate_inode_ref(sctx, sctx->parent_root, sctx->right_path,
3113                        sctx->cmp_key, 0, __record_deleted_ref, sctx);
3114        if (ret < 0)
3115                goto out;
3116        ret = 0;
3117
3118out:
3119        return ret;
3120}
3121
3122struct find_ref_ctx {
3123        u64 dir;
3124        struct fs_path *name;
3125        int found_idx;
3126};
3127
3128static int __find_iref(int num, u64 dir, int index,
3129                       struct fs_path *name,
3130                       void *ctx_)
3131{
3132        struct find_ref_ctx *ctx = ctx_;
3133
3134        if (dir == ctx->dir && fs_path_len(name) == fs_path_len(ctx->name) &&
3135            strncmp(name->start, ctx->name->start, fs_path_len(name)) == 0) {
3136                ctx->found_idx = num;
3137                return 1;
3138        }
3139        return 0;
3140}
3141
3142static int find_iref(struct send_ctx *sctx,
3143                     struct btrfs_root *root,
3144                     struct btrfs_path *path,
3145                     struct btrfs_key *key,
3146                     u64 dir, struct fs_path *name)
3147{
3148        int ret;
3149        struct find_ref_ctx ctx;
3150
3151        ctx.dir = dir;
3152        ctx.name = name;
3153        ctx.found_idx = -1;
3154
3155        ret = iterate_inode_ref(sctx, root, path, key, 0, __find_iref, &ctx);
3156        if (ret < 0)
3157                return ret;
3158
3159        if (ctx.found_idx == -1)
3160                return -ENOENT;
3161
3162        return ctx.found_idx;
3163}
3164
3165static int __record_changed_new_ref(int num, u64 dir, int index,
3166                                    struct fs_path *name,
3167                                    void *ctx)
3168{
3169        int ret;
3170        struct send_ctx *sctx = ctx;
3171
3172        ret = find_iref(sctx, sctx->parent_root, sctx->right_path,
3173                        sctx->cmp_key, dir, name);
3174        if (ret == -ENOENT)
3175                ret = __record_new_ref(num, dir, index, name, sctx);
3176        else if (ret > 0)
3177                ret = 0;
3178
3179        return ret;
3180}
3181
3182static int __record_changed_deleted_ref(int num, u64 dir, int index,
3183                                        struct fs_path *name,
3184                                        void *ctx)
3185{
3186        int ret;
3187        struct send_ctx *sctx = ctx;
3188
3189        ret = find_iref(sctx, sctx->send_root, sctx->left_path, sctx->cmp_key,
3190                        dir, name);
3191        if (ret == -ENOENT)
3192                ret = __record_deleted_ref(num, dir, index, name, sctx);
3193        else if (ret > 0)
3194                ret = 0;
3195
3196        return ret;
3197}
3198
3199static int record_changed_ref(struct send_ctx *sctx)
3200{
3201        int ret = 0;
3202
3203        ret = iterate_inode_ref(sctx, sctx->send_root, sctx->left_path,
3204                        sctx->cmp_key, 0, __record_changed_new_ref, sctx);
3205        if (ret < 0)
3206                goto out;
3207        ret = iterate_inode_ref(sctx, sctx->parent_root, sctx->right_path,
3208                        sctx->cmp_key, 0, __record_changed_deleted_ref, sctx);
3209        if (ret < 0)
3210                goto out;
3211        ret = 0;
3212
3213out:
3214        return ret;
3215}
3216
3217/*
3218 * Record and process all refs at once. Needed when an inode changes the
3219 * generation number, which means that it was deleted and recreated.
3220 */
3221static int process_all_refs(struct send_ctx *sctx,
3222                            enum btrfs_compare_tree_result cmd)
3223{
3224        int ret;
3225        struct btrfs_root *root;
3226        struct btrfs_path *path;
3227        struct btrfs_key key;
3228        struct btrfs_key found_key;
3229        struct extent_buffer *eb;
3230        int slot;
3231        iterate_inode_ref_t cb;
3232
3233        path = alloc_path_for_send();
3234        if (!path)
3235                return -ENOMEM;
3236
3237        if (cmd == BTRFS_COMPARE_TREE_NEW) {
3238                root = sctx->send_root;
3239                cb = __record_new_ref;
3240        } else if (cmd == BTRFS_COMPARE_TREE_DELETED) {
3241                root = sctx->parent_root;
3242                cb = __record_deleted_ref;
3243        } else {
3244                BUG();
3245        }
3246
3247        key.objectid = sctx->cmp_key->objectid;
3248        key.type = BTRFS_INODE_REF_KEY;
3249        key.offset = 0;
3250        while (1) {
3251                ret = btrfs_search_slot_for_read(root, &key, path, 1, 0);
3252                if (ret < 0)
3253                        goto out;
3254                if (ret)
3255                        break;
3256
3257                eb = path->nodes[0];
3258                slot = path->slots[0];
3259                btrfs_item_key_to_cpu(eb, &found_key, slot);
3260
3261                if (found_key.objectid != key.objectid ||
3262                    (found_key.type != BTRFS_INODE_REF_KEY &&
3263                     found_key.type != BTRFS_INODE_EXTREF_KEY))
3264                        break;
3265
3266                ret = iterate_inode_ref(sctx, root, path, &found_key, 0, cb,
3267                                sctx);
3268                btrfs_release_path(path);
3269                if (ret < 0)
3270                        goto out;
3271
3272                key.offset = found_key.offset + 1;
3273        }
3274        btrfs_release_path(path);
3275
3276        ret = process_recorded_refs(sctx);
3277
3278out:
3279        btrfs_free_path(path);
3280        return ret;
3281}
3282
3283static int send_set_xattr(struct send_ctx *sctx,
3284                          struct fs_path *path,
3285                          const char *name, int name_len,
3286                          const char *data, int data_len)
3287{
3288        int ret = 0;
3289
3290        ret = begin_cmd(sctx, BTRFS_SEND_C_SET_XATTR);
3291        if (ret < 0)
3292                goto out;
3293
3294        TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, path);
3295        TLV_PUT_STRING(sctx, BTRFS_SEND_A_XATTR_NAME, name, name_len);
3296        TLV_PUT(sctx, BTRFS_SEND_A_XATTR_DATA, data, data_len);
3297
3298        ret = send_cmd(sctx);
3299
3300tlv_put_failure:
3301out:
3302        return ret;
3303}
3304
3305static int send_remove_xattr(struct send_ctx *sctx,
3306                          struct fs_path *path,
3307                          const char *name, int name_len)
3308{
3309        int ret = 0;
3310
3311        ret = begin_cmd(sctx, BTRFS_SEND_C_REMOVE_XATTR);
3312        if (ret < 0)
3313                goto out;
3314
3315        TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, path);
3316        TLV_PUT_STRING(sctx, BTRFS_SEND_A_XATTR_NAME, name, name_len);
3317
3318        ret = send_cmd(sctx);
3319
3320tlv_put_failure:
3321out:
3322        return ret;
3323}
3324
3325static int __process_new_xattr(int num, struct btrfs_key *di_key,
3326                               const char *name, int name_len,
3327                               const char *data, int data_len,
3328                               u8 type, void *ctx)
3329{
3330        int ret;
3331        struct send_ctx *sctx = ctx;
3332        struct fs_path *p;
3333        posix_acl_xattr_header dummy_acl;
3334
3335        p = fs_path_alloc(sctx);
3336        if (!p)
3337                return -ENOMEM;
3338
3339        /*
3340         * This hack is needed because empty acl's are stored as zero byte
3341         * data in xattrs. Problem with that is, that receiving these zero byte
3342         * acl's will fail later. To fix this, we send a dummy acl list that
3343         * only contains the version number and no entries.
3344         */
3345        if (!strncmp(name, XATTR_NAME_POSIX_ACL_ACCESS, name_len) ||
3346            !strncmp(name, XATTR_NAME_POSIX_ACL_DEFAULT, name_len)) {
3347                if (data_len == 0) {
3348                        dummy_acl.a_version =
3349                                        cpu_to_le32(POSIX_ACL_XATTR_VERSION);
3350                        data = (char *)&dummy_acl;
3351                        data_len = sizeof(dummy_acl);
3352                }
3353        }
3354
3355        ret = get_cur_path(sctx, sctx->cur_ino, sctx->cur_inode_gen, p);
3356        if (ret < 0)
3357                goto out;
3358
3359        ret = send_set_xattr(sctx, p, name, name_len, data, data_len);
3360
3361out:
3362        fs_path_free(sctx, p);
3363        return ret;
3364}
3365
3366static int __process_deleted_xattr(int num, struct btrfs_key *di_key,
3367                                   const char *name, int name_len,
3368                                   const char *data, int data_len,
3369                                   u8 type, void *ctx)
3370{
3371        int ret;
3372        struct send_ctx *sctx = ctx;
3373        struct fs_path *p;
3374
3375        p = fs_path_alloc(sctx);
3376        if (!p)
3377                return -ENOMEM;
3378
3379        ret = get_cur_path(sctx, sctx->cur_ino, sctx->cur_inode_gen, p);
3380        if (ret < 0)
3381                goto out;
3382
3383        ret = send_remove_xattr(sctx, p, name, name_len);
3384
3385out:
3386        fs_path_free(sctx, p);
3387        return ret;
3388}
3389
3390static int process_new_xattr(struct send_ctx *sctx)
3391{
3392        int ret = 0;
3393
3394        ret = iterate_dir_item(sctx, sctx->send_root, sctx->left_path,
3395                        sctx->cmp_key, __process_new_xattr, sctx);
3396
3397        return ret;
3398}
3399
3400static int process_deleted_xattr(struct send_ctx *sctx)
3401{
3402        int ret;
3403
3404        ret = iterate_dir_item(sctx, sctx->parent_root, sctx->right_path,
3405                        sctx->cmp_key, __process_deleted_xattr, sctx);
3406
3407        return ret;
3408}
3409
3410struct find_xattr_ctx {
3411        const char *name;
3412        int name_len;
3413        int found_idx;
3414        char *found_data;
3415        int found_data_len;
3416};
3417
3418static int __find_xattr(int num, struct btrfs_key *di_key,
3419                        const char *name, int name_len,
3420                        const char *data, int data_len,
3421                        u8 type, void *vctx)
3422{
3423        struct find_xattr_ctx *ctx = vctx;
3424
3425        if (name_len == ctx->name_len &&
3426            strncmp(name, ctx->name, name_len) == 0) {
3427                ctx->found_idx = num;
3428                ctx->found_data_len = data_len;
3429                ctx->found_data = kmalloc(data_len, GFP_NOFS);
3430                if (!ctx->found_data)
3431                        return -ENOMEM;
3432                memcpy(ctx->found_data, data, data_len);
3433                return 1;
3434        }
3435        return 0;
3436}
3437
3438static int find_xattr(struct send_ctx *sctx,
3439                      struct btrfs_root *root,
3440                      struct btrfs_path *path,
3441                      struct btrfs_key *key,
3442                      const char *name, int name_len,
3443                      char **data, int *data_len)
3444{
3445        int ret;
3446        struct find_xattr_ctx ctx;
3447
3448        ctx.name = name;
3449        ctx.name_len = name_len;
3450        ctx.found_idx = -1;
3451        ctx.found_data = NULL;
3452        ctx.found_data_len = 0;
3453
3454        ret = iterate_dir_item(sctx, root, path, key, __find_xattr, &ctx);
3455        if (ret < 0)
3456                return ret;
3457
3458        if (ctx.found_idx == -1)
3459                return -ENOENT;
3460        if (data) {
3461                *data = ctx.found_data;
3462                *data_len = ctx.found_data_len;
3463        } else {
3464                kfree(ctx.found_data);
3465        }
3466        return ctx.found_idx;
3467}
3468
3469
3470static int __process_changed_new_xattr(int num, struct btrfs_key *di_key,
3471                                       const char *name, int name_len,
3472                                       const char *data, int data_len,
3473                                       u8 type, void *ctx)
3474{
3475        int ret;
3476        struct send_ctx *sctx = ctx;
3477        char *found_data = NULL;
3478        int found_data_len  = 0;
3479        struct fs_path *p = NULL;
3480
3481        ret = find_xattr(sctx, sctx->parent_root, sctx->right_path,
3482                        sctx->cmp_key, name, name_len, &found_data,
3483                        &found_data_len);
3484        if (ret == -ENOENT) {
3485                ret = __process_new_xattr(num, di_key, name, name_len, data,
3486                                data_len, type, ctx);
3487        } else if (ret >= 0) {
3488                if (data_len != found_data_len ||
3489                    memcmp(data, found_data, data_len)) {
3490                        ret = __process_new_xattr(num, di_key, name, name_len,
3491                                        data, data_len, type, ctx);
3492                } else {
3493                        ret = 0;
3494                }
3495        }
3496
3497        kfree(found_data);
3498        fs_path_free(sctx, p);
3499        return ret;
3500}
3501
3502static int __process_changed_deleted_xattr(int num, struct btrfs_key *di_key,
3503                                           const char *name, int name_len,
3504                                           const char *data, int data_len,
3505                                           u8 type, void *ctx)
3506{
3507        int ret;
3508        struct send_ctx *sctx = ctx;
3509
3510        ret = find_xattr(sctx, sctx->send_root, sctx->left_path, sctx->cmp_key,
3511                        name, name_len, NULL, NULL);
3512        if (ret == -ENOENT)
3513                ret = __process_deleted_xattr(num, di_key, name, name_len, data,
3514                                data_len, type, ctx);
3515        else if (ret >= 0)
3516                ret = 0;
3517
3518        return ret;
3519}
3520
3521static int process_changed_xattr(struct send_ctx *sctx)
3522{
3523        int ret = 0;
3524
3525        ret = iterate_dir_item(sctx, sctx->send_root, sctx->left_path,
3526                        sctx->cmp_key, __process_changed_new_xattr, sctx);
3527        if (ret < 0)
3528                goto out;
3529        ret = iterate_dir_item(sctx, sctx->parent_root, sctx->right_path,
3530                        sctx->cmp_key, __process_changed_deleted_xattr, sctx);
3531
3532out:
3533        return ret;
3534}
3535
3536static int process_all_new_xattrs(struct send_ctx *sctx)
3537{
3538        int ret;
3539        struct btrfs_root *root;
3540        struct btrfs_path *path;
3541        struct btrfs_key key;
3542        struct btrfs_key found_key;
3543        struct extent_buffer *eb;
3544        int slot;
3545
3546        path = alloc_path_for_send();
3547        if (!path)
3548                return -ENOMEM;
3549
3550        root = sctx->send_root;
3551
3552        key.objectid = sctx->cmp_key->objectid;
3553        key.type = BTRFS_XATTR_ITEM_KEY;
3554        key.offset = 0;
3555        while (1) {
3556                ret = btrfs_search_slot_for_read(root, &key, path, 1, 0);
3557                if (ret < 0)
3558                        goto out;
3559                if (ret) {
3560                        ret = 0;
3561                        goto out;
3562                }
3563
3564                eb = path->nodes[0];
3565                slot = path->slots[0];
3566                btrfs_item_key_to_cpu(eb, &found_key, slot);
3567
3568                if (found_key.objectid != key.objectid ||
3569                    found_key.type != key.type) {
3570                        ret = 0;
3571                        goto out;
3572                }
3573
3574                ret = iterate_dir_item(sctx, root, path, &found_key,
3575                                __process_new_xattr, sctx);
3576                if (ret < 0)
3577                        goto out;
3578
3579                btrfs_release_path(path);
3580                key.offset = found_key.offset + 1;
3581        }
3582
3583out:
3584        btrfs_free_path(path);
3585        return ret;
3586}
3587
3588/*
3589 * Read some bytes from the current inode/file and send a write command to
3590 * user space.
3591 */
3592static int send_write(struct send_ctx *sctx, u64 offset, u32 len)
3593{
3594        int ret = 0;
3595        struct fs_path *p;
3596        loff_t pos = offset;
3597        int num_read = 0;
3598        mm_segment_t old_fs;
3599
3600        p = fs_path_alloc(sctx);
3601        if (!p)
3602                return -ENOMEM;
3603
3604        /*
3605         * vfs normally only accepts user space buffers for security reasons.
3606         * we only read from the file and also only provide the read_buf buffer
3607         * to vfs. As this buffer does not come from a user space call, it's
3608         * ok to temporary allow kernel space buffers.
3609         */
3610        old_fs = get_fs();
3611        set_fs(KERNEL_DS);
3612
3613verbose_printk("btrfs: send_write offset=%llu, len=%d\n", offset, len);
3614
3615        ret = open_cur_inode_file(sctx);
3616        if (ret < 0)
3617                goto out;
3618
3619        ret = vfs_read(sctx->cur_inode_filp, sctx->read_buf, len, &pos);
3620        if (ret < 0)
3621                goto out;
3622        num_read = ret;
3623        if (!num_read)
3624                goto out;
3625
3626        ret = begin_cmd(sctx, BTRFS_SEND_C_WRITE);
3627        if (ret < 0)
3628                goto out;
3629
3630        ret = get_cur_path(sctx, sctx->cur_ino, sctx->cur_inode_gen, p);
3631        if (ret < 0)
3632                goto out;
3633
3634        TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, p);
3635        TLV_PUT_U64(sctx, BTRFS_SEND_A_FILE_OFFSET, offset);
3636        TLV_PUT(sctx, BTRFS_SEND_A_DATA, sctx->read_buf, num_read);
3637
3638        ret = send_cmd(sctx);
3639
3640tlv_put_failure:
3641out:
3642        fs_path_free(sctx, p);
3643        set_fs(old_fs);
3644        if (ret < 0)
3645                return ret;
3646        return num_read;
3647}
3648
3649/*
3650 * Send a clone command to user space.
3651 */
3652static int send_clone(struct send_ctx *sctx,
3653                      u64 offset, u32 len,
3654                      struct clone_root *clone_root)
3655{
3656        int ret = 0;
3657        struct fs_path *p;
3658        u64 gen;
3659
3660verbose_printk("btrfs: send_clone offset=%llu, len=%d, clone_root=%llu, "
3661               "clone_inode=%llu, clone_offset=%llu\n", offset, len,
3662                clone_root->root->objectid, clone_root->ino,
3663                clone_root->offset);
3664
3665        p = fs_path_alloc(sctx);
3666        if (!p)
3667                return -ENOMEM;
3668
3669        ret = begin_cmd(sctx, BTRFS_SEND_C_CLONE);
3670        if (ret < 0)
3671                goto out;
3672
3673        ret = get_cur_path(sctx, sctx->cur_ino, sctx->cur_inode_gen, p);
3674        if (ret < 0)
3675                goto out;
3676
3677        TLV_PUT_U64(sctx, BTRFS_SEND_A_FILE_OFFSET, offset);
3678        TLV_PUT_U64(sctx, BTRFS_SEND_A_CLONE_LEN, len);
3679        TLV_PUT_PATH(sctx, BTRFS_SEND_A_PATH, p);
3680
3681        if (clone_root->root == sctx->send_root) {
3682                ret = get_inode_info(sctx->send_root, clone_root->ino, NULL,
3683                                &gen, NULL, NULL, NULL, NULL);
3684                if (ret < 0)
3685                        goto out;
3686                ret = get_cur_path(sctx, clone_root->ino, gen, p);
3687        } else {
3688                ret = get_inode_path(sctx, clone_root->root,
3689                                clone_root->ino, p);
3690        }
3691        if (ret < 0)
3692                goto out;
3693
3694        TLV_PUT_UUID(sctx, BTRFS_SEND_A_CLONE_UUID,
3695                        clone_root->root->root_item.uuid);
3696        TLV_PUT_U64(sctx, BTRFS_SEND_A_CLONE_CTRANSID,
3697                        clone_root->root->root_item.ctransid);
3698        TLV_PUT_PATH(sctx, BTRFS_SEND_A_CLONE_PATH, p);
3699        TLV_PUT_U64(sctx, BTRFS_SEND_A_CLONE_OFFSET,
3700                        clone_root->offset);
3701
3702        ret = send_cmd(sctx);
3703
3704tlv_put_failure:
3705out:
3706        fs_path_free(sctx, p);
3707        return ret;
3708}
3709
3710static int send_write_or_clone(struct send_ctx *sctx,
3711                               struct btrfs_path *path,
3712                               struct btrfs_key *key,
3713                               struct clone_root *clone_root)
3714{
3715        int ret = 0;
3716        struct btrfs_file_extent_item *ei;
3717        u64 offset = key->offset;
3718        u64 pos = 0;
3719        u64 len;
3720        u32 l;
3721        u8 type;
3722
3723        ei = btrfs_item_ptr(path->nodes[0], path->slots[0],
3724                        struct btrfs_file_extent_item);
3725        type = btrfs_file_extent_type(path->nodes[0], ei);
3726        if (type == BTRFS_FILE_EXTENT_INLINE) {
3727                len = btrfs_file_extent_inline_len(path->nodes[0], ei);
3728                /*
3729                 * it is possible the inline item won't cover the whole page,
3730                 * but there may be items after this page.  Make
3731                 * sure to send the whole thing
3732                 */
3733                len = PAGE_CACHE_ALIGN(len);
3734        } else {
3735                len = btrfs_file_extent_num_bytes(path->nodes[0], ei);
3736        }
3737
3738        if (offset + len > sctx->cur_inode_size)
3739                len = sctx->cur_inode_size - offset;
3740        if (len == 0) {
3741                ret = 0;
3742                goto out;
3743        }
3744
3745        if (!clone_root) {
3746                while (pos < len) {
3747                        l = len - pos;
3748                        if (l > BTRFS_SEND_READ_SIZE)
3749                                l = BTRFS_SEND_READ_SIZE;
3750                        ret = send_write(sctx, pos + offset, l);
3751                        if (ret < 0)
3752                                goto out;
3753                        if (!ret)
3754                                break;
3755                        pos += ret;
3756                }
3757                ret = 0;
3758        } else {
3759                ret = send_clone(sctx, offset, len, clone_root);
3760        }
3761
3762out:
3763        return ret;
3764}
3765
3766static int is_extent_unchanged(struct send_ctx *sctx,
3767                               struct btrfs_path *left_path,
3768                               struct btrfs_key *ekey)
3769{
3770        int ret = 0;
3771        struct btrfs_key key;
3772        struct btrfs_path *path = NULL;
3773        struct extent_buffer *eb;
3774        int slot;
3775        struct btrfs_key found_key;
3776        struct btrfs_file_extent_item *ei;
3777        u64 left_disknr;
3778        u64 right_disknr;
3779        u64 left_offset;
3780        u64 right_offset;
3781        u64 left_offset_fixed;
3782        u64 left_len;
3783        u64 right_len;
3784        u64 left_gen;
3785        u64 right_gen;
3786        u8 left_type;
3787        u8 right_type;
3788
3789        path = alloc_path_for_send();
3790        if (!path)
3791                return -ENOMEM;
3792
3793        eb = left_path->nodes[0];
3794        slot = left_path->slots[0];
3795        ei = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
3796        left_type = btrfs_file_extent_type(eb, ei);
3797
3798        if (left_type != BTRFS_FILE_EXTENT_REG) {
3799                ret = 0;
3800                goto out;
3801        }
3802        left_disknr = btrfs_file_extent_disk_bytenr(eb, ei);
3803        left_len = btrfs_file_extent_num_bytes(eb, ei);
3804        left_offset = btrfs_file_extent_offset(eb, ei);
3805        left_gen = btrfs_file_extent_generation(eb, ei);
3806
3807        /*
3808         * Following comments will refer to these graphics. L is the left
3809         * extents which we are checking at the moment. 1-8 are the right
3810         * extents that we iterate.
3811         *
3812         *       |-----L-----|
3813         * |-1-|-2a-|-3-|-4-|-5-|-6-|
3814         *
3815         *       |-----L-----|
3816         * |--1--|-2b-|...(same as above)
3817         *
3818         * Alternative situation. Happens on files where extents got split.
3819         *       |-----L-----|
3820         * |-----------7-----------|-6-|
3821         *
3822         * Alternative situation. Happens on files which got larger.
3823         *       |-----L-----|
3824         * |-8-|
3825         * Nothing follows after 8.
3826         */
3827
3828        key.objectid = ekey->objectid;
3829        key.type = BTRFS_EXTENT_DATA_KEY;
3830        key.offset = ekey->offset;
3831        ret = btrfs_search_slot_for_read(sctx->parent_root, &key, path, 0, 0);
3832        if (ret < 0)
3833                goto out;
3834        if (ret) {
3835                ret = 0;
3836                goto out;
3837        }
3838
3839        /*
3840         * Handle special case where the right side has no extents at all.
3841         */
3842        eb = path->nodes[0];
3843        slot = path->slots[0];
3844        btrfs_item_key_to_cpu(eb, &found_key, slot);
3845        if (found_key.objectid != key.objectid ||
3846            found_key.type != key.type) {
3847                ret = 0;
3848                goto out;
3849        }
3850
3851        /*
3852         * We're now on 2a, 2b or 7.
3853         */
3854        key = found_key;
3855        while (key.offset < ekey->offset + left_len) {
3856                ei = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
3857                right_type = btrfs_file_extent_type(eb, ei);
3858                right_disknr = btrfs_file_extent_disk_bytenr(eb, ei);
3859                right_len = btrfs_file_extent_num_bytes(eb, ei);
3860                right_offset = btrfs_file_extent_offset(eb, ei);
3861                right_gen = btrfs_file_extent_generation(eb, ei);
3862
3863                if (right_type != BTRFS_FILE_EXTENT_REG) {
3864                        ret = 0;
3865                        goto out;
3866                }
3867
3868                /*
3869                 * Are we at extent 8? If yes, we know the extent is changed.
3870                 * This may only happen on the first iteration.
3871                 */
3872                if (found_key.offset + right_len <= ekey->offset) {
3873                        ret = 0;
3874                        goto out;
3875                }
3876
3877                left_offset_fixed = left_offset;
3878                if (key.offset < ekey->offset) {
3879                        /* Fix the right offset for 2a and 7. */
3880                        right_offset += ekey->offset - key.offset;
3881                } else {
3882                        /* Fix the left offset for all behind 2a and 2b */
3883                        left_offset_fixed += key.offset - ekey->offset;
3884                }
3885
3886                /*
3887                 * Check if we have the same extent.
3888                 */
3889                if (left_disknr != right_disknr ||
3890                    left_offset_fixed != right_offset ||
3891                    left_gen != right_gen) {
3892                        ret = 0;
3893                        goto out;
3894                }
3895
3896                /*
3897                 * Go to the next extent.
3898                 */
3899                ret = btrfs_next_item(sctx->parent_root, path);
3900                if (ret < 0)
3901                        goto out;
3902                if (!ret) {
3903                        eb = path->nodes[0];
3904                        slot = path->slots[0];
3905                        btrfs_item_key_to_cpu(eb, &found_key, slot);
3906                }
3907                if (ret || found_key.objectid != key.objectid ||
3908                    found_key.type != key.type) {
3909                        key.offset += right_len;
3910                        break;
3911                } else {
3912                        if (found_key.offset != key.offset + right_len) {
3913                                /* Should really not happen */
3914                                ret = -EIO;
3915                                goto out;
3916                        }
3917                }
3918                key = found_key;
3919        }
3920
3921        /*
3922         * We're now behind the left extent (treat as unchanged) or at the end
3923         * of the right side (treat as changed).
3924         */
3925        if (key.offset >= ekey->offset + left_len)
3926                ret = 1;
3927        else
3928                ret = 0;
3929
3930
3931out:
3932        btrfs_free_path(path);
3933        return ret;
3934}
3935
3936static int process_extent(struct send_ctx *sctx,
3937                          struct btrfs_path *path,
3938                          struct btrfs_key *key)
3939{
3940        int ret = 0;
3941        struct clone_root *found_clone = NULL;
3942
3943        if (S_ISLNK(sctx->cur_inode_mode))
3944                return 0;
3945
3946        if (sctx->parent_root && !sctx->cur_inode_new) {
3947                ret = is_extent_unchanged(sctx, path, key);
3948                if (ret < 0)
3949                        goto out;
3950                if (ret) {
3951                        ret = 0;
3952                        goto out;
3953                }
3954        }
3955
3956        ret = find_extent_clone(sctx, path, key->objectid, key->offset,
3957                        sctx->cur_inode_size, &found_clone);
3958        if (ret != -ENOENT && ret < 0)
3959                goto out;
3960
3961        ret = send_write_or_clone(sctx, path, key, found_clone);
3962
3963out:
3964        return ret;
3965}
3966
3967static int process_all_extents(struct send_ctx *sctx)
3968{
3969        int ret;
3970        struct btrfs_root *root;
3971        struct btrfs_path *path;
3972        struct btrfs_key key;
3973        struct btrfs_key found_key;
3974        struct extent_buffer *eb;
3975        int slot;
3976
3977        root = sctx->send_root;
3978        path = alloc_path_for_send();
3979        if (!path)
3980                return -ENOMEM;
3981
3982        key.objectid = sctx->cmp_key->objectid;
3983        key.type = BTRFS_EXTENT_DATA_KEY;
3984        key.offset = 0;
3985        while (1) {
3986                ret = btrfs_search_slot_for_read(root, &key, path, 1, 0);
3987                if (ret < 0)
3988                        goto out;
3989                if (ret) {
3990                        ret = 0;
3991                        goto out;
3992                }
3993
3994                eb = path->nodes[0];
3995                slot = path->slots[0];
3996                btrfs_item_key_to_cpu(eb, &found_key, slot);
3997
3998                if (found_key.objectid != key.objectid ||
3999                    found_key.type != key.type) {
4000                        ret = 0;
4001                        goto out;
4002                }
4003
4004                ret = process_extent(sctx, path, &found_key);
4005                if (ret < 0)
4006                        goto out;
4007
4008                btrfs_release_path(path);
4009                key.offset = found_key.offset + 1;
4010        }
4011
4012out:
4013        btrfs_free_path(path);
4014        return ret;
4015}
4016
4017static int process_recorded_refs_if_needed(struct send_ctx *sctx, int at_end)
4018{
4019        int ret = 0;
4020
4021        if (sctx->cur_ino == 0)
4022                goto out;
4023        if (!at_end && sctx->cur_ino == sctx->cmp_key->objectid &&
4024            sctx->cmp_key->type <= BTRFS_INODE_EXTREF_KEY)
4025                goto out;
4026        if (list_empty(&sctx->new_refs) && list_empty(&sctx->deleted_refs))
4027                goto out;
4028
4029        ret = process_recorded_refs(sctx);
4030        if (ret < 0)
4031                goto out;
4032
4033        /*
4034         * We have processed the refs and thus need to advance send_progress.
4035         * Now, calls to get_cur_xxx will take the updated refs of the current
4036         * inode into account.
4037         */
4038        sctx->send_progress = sctx->cur_ino + 1;
4039
4040out:
4041        return ret;
4042}
4043
4044static int finish_inode_if_needed(struct send_ctx *sctx, int at_end)
4045{
4046        int ret = 0;
4047        u64 left_mode;
4048        u64 left_uid;
4049        u64 left_gid;
4050        u64 right_mode;
4051        u64 right_uid;
4052        u64 right_gid;
4053        int need_chmod = 0;
4054        int need_chown = 0;
4055
4056        ret = process_recorded_refs_if_needed(sctx, at_end);
4057        if (ret < 0)
4058                goto out;
4059
4060        if (sctx->cur_ino == 0 || sctx->cur_inode_deleted)
4061                goto out;
4062        if (!at_end && sctx->cmp_key->objectid == sctx->cur_ino)
4063                goto out;
4064
4065        ret = get_inode_info(sctx->send_root, sctx->cur_ino, NULL, NULL,
4066                        &left_mode, &left_uid, &left_gid, NULL);
4067        if (ret < 0)
4068                goto out;
4069
4070        if (!sctx->parent_root || sctx->cur_inode_new) {
4071                need_chown = 1;
4072                if (!S_ISLNK(sctx->cur_inode_mode))
4073                        need_chmod = 1;
4074        } else {
4075                ret = get_inode_info(sctx->parent_root, sctx->cur_ino,
4076                                NULL, NULL, &right_mode, &right_uid,
4077                                &right_gid, NULL);
4078                if (ret < 0)
4079                        goto out;
4080
4081                if (left_uid != right_uid || left_gid != right_gid)
4082                        need_chown = 1;
4083                if (!S_ISLNK(sctx->cur_inode_mode) && left_mode != right_mode)
4084                        need_chmod = 1;
4085        }
4086
4087        if (S_ISREG(sctx->cur_inode_mode)) {
4088                ret = send_truncate(sctx, sctx->cur_ino, sctx->cur_inode_gen,
4089                                sctx->cur_inode_size);
4090                if (ret < 0)
4091                        goto out;
4092        }
4093
4094        if (need_chown) {
4095                ret = send_chown(sctx, sctx->cur_ino, sctx->cur_inode_gen,
4096                                left_uid, left_gid);
4097                if (ret < 0)
4098                        goto out;
4099        }
4100        if (need_chmod) {
4101                ret = send_chmod(sctx, sctx->cur_ino, sctx->cur_inode_gen,
4102                                left_mode);
4103                if (ret < 0)
4104                        goto out;
4105        }
4106
4107        /*
4108         * Need to send that every time, no matter if it actually changed
4109         * between the two trees as we have done changes to the inode before.
4110         */
4111        ret = send_utimes(sctx, sctx->cur_ino, sctx->cur_inode_gen);
4112        if (ret < 0)
4113                goto out;
4114
4115out:
4116        return ret;
4117}
4118
4119static int changed_inode(struct send_ctx *sctx,
4120                         enum btrfs_compare_tree_result result)
4121{
4122        int ret = 0;
4123        struct btrfs_key *key = sctx->cmp_key;
4124        struct btrfs_inode_item *left_ii = NULL;
4125        struct btrfs_inode_item *right_ii = NULL;
4126        u64 left_gen = 0;
4127        u64 right_gen = 0;
4128
4129        ret = close_cur_inode_file(sctx);
4130        if (ret < 0)
4131                goto out;
4132
4133        sctx->cur_ino = key->objectid;
4134        sctx->cur_inode_new_gen = 0;
4135
4136        /*
4137         * Set send_progress to current inode. This will tell all get_cur_xxx
4138         * functions that the current inode's refs are not updated yet. Later,
4139         * when process_recorded_refs is finished, it is set to cur_ino + 1.
4140         */
4141        sctx->send_progress = sctx->cur_ino;
4142
4143        if (result == BTRFS_COMPARE_TREE_NEW ||
4144            result == BTRFS_COMPARE_TREE_CHANGED) {
4145                left_ii = btrfs_item_ptr(sctx->left_path->nodes[0],
4146                                sctx->left_path->slots[0],
4147                                struct btrfs_inode_item);
4148                left_gen = btrfs_inode_generation(sctx->left_path->nodes[0],
4149                                left_ii);
4150        } else {
4151                right_ii = btrfs_item_ptr(sctx->right_path->nodes[0],
4152                                sctx->right_path->slots[0],
4153                                struct btrfs_inode_item);
4154                right_gen = btrfs_inode_generation(sctx->right_path->nodes[0],
4155                                right_ii);
4156        }
4157        if (result == BTRFS_COMPARE_TREE_CHANGED) {
4158                right_ii = btrfs_item_ptr(sctx->right_path->nodes[0],
4159                                sctx->right_path->slots[0],
4160                                struct btrfs_inode_item);
4161
4162                right_gen = btrfs_inode_generation(sctx->right_path->nodes[0],
4163                                right_ii);
4164
4165                /*
4166                 * The cur_ino = root dir case is special here. We can't treat
4167                 * the inode as deleted+reused because it would generate a
4168                 * stream that tries to delete/mkdir the root dir.
4169                 */
4170                if (left_gen != right_gen &&
4171                    sctx->cur_ino != BTRFS_FIRST_FREE_OBJECTID)
4172                        sctx->cur_inode_new_gen = 1;
4173        }
4174
4175        if (result == BTRFS_COMPARE_TREE_NEW) {
4176                sctx->cur_inode_gen = left_gen;
4177                sctx->cur_inode_new = 1;
4178                sctx->cur_inode_deleted = 0;
4179                sctx->cur_inode_size = btrfs_inode_size(
4180                                sctx->left_path->nodes[0], left_ii);
4181                sctx->cur_inode_mode = btrfs_inode_mode(
4182                                sctx->left_path->nodes[0], left_ii);
4183                if (sctx->cur_ino != BTRFS_FIRST_FREE_OBJECTID)
4184                        ret = send_create_inode_if_needed(sctx);
4185        } else if (result == BTRFS_COMPARE_TREE_DELETED) {
4186                sctx->cur_inode_gen = right_gen;
4187                sctx->cur_inode_new = 0;
4188                sctx->cur_inode_deleted = 1;
4189                sctx->cur_inode_size = btrfs_inode_size(
4190                                sctx->right_path->nodes[0], right_ii);
4191                sctx->cur_inode_mode = btrfs_inode_mode(
4192                                sctx->right_path->nodes[0], right_ii);
4193        } else if (result == BTRFS_COMPARE_TREE_CHANGED) {
4194                /*
4195                 * We need to do some special handling in case the inode was
4196                 * reported as changed with a changed generation number. This
4197                 * means that the original inode was deleted and new inode
4198                 * reused the same inum. So we have to treat the old inode as
4199                 * deleted and the new one as new.
4200                 */
4201                if (sctx->cur_inode_new_gen) {
4202                        /*
4203                         * First, process the inode as if it was deleted.
4204                         */
4205                        sctx->cur_inode_gen = right_gen;
4206                        sctx->cur_inode_new = 0;
4207                        sctx->cur_inode_deleted = 1;
4208                        sctx->cur_inode_size = btrfs_inode_size(
4209                                        sctx->right_path->nodes[0], right_ii);
4210                        sctx->cur_inode_mode = btrfs_inode_mode(
4211                                        sctx->right_path->nodes[0], right_ii);
4212                        ret = process_all_refs(sctx,
4213                                        BTRFS_COMPARE_TREE_DELETED);
4214                        if (ret < 0)
4215                                goto out;
4216
4217                        /*
4218                         * Now process the inode as if it was new.
4219                         */
4220                        sctx->cur_inode_gen = left_gen;
4221                        sctx->cur_inode_new = 1;
4222                        sctx->cur_inode_deleted = 0;
4223                        sctx->cur_inode_size = btrfs_inode_size(
4224                                        sctx->left_path->nodes[0], left_ii);
4225                        sctx->cur_inode_mode = btrfs_inode_mode(
4226                                        sctx->left_path->nodes[0], left_ii);
4227                        ret = send_create_inode_if_needed(sctx);
4228                        if (ret < 0)
4229                                goto out;
4230
4231                        ret = process_all_refs(sctx, BTRFS_COMPARE_TREE_NEW);
4232                        if (ret < 0)
4233                                goto out;
4234                        /*
4235                         * Advance send_progress now as we did not get into
4236                         * process_recorded_refs_if_needed in the new_gen case.
4237                         */
4238                        sctx->send_progress = sctx->cur_ino + 1;
4239
4240                        /*
4241                         * Now process all extents and xattrs of the inode as if
4242                         * they were all new.
4243                         */
4244                        ret = process_all_extents(sctx);
4245                        if (ret < 0)
4246                                goto out;
4247                        ret = process_all_new_xattrs(sctx);
4248                        if (ret < 0)
4249                                goto out;
4250                } else {
4251                        sctx->cur_inode_gen = left_gen;
4252                        sctx->cur_inode_new = 0;
4253                        sctx->cur_inode_new_gen = 0;
4254                        sctx->cur_inode_deleted = 0;
4255                        sctx->cur_inode_size = btrfs_inode_size(
4256                                        sctx->left_path->nodes[0], left_ii);
4257                        sctx->cur_inode_mode = btrfs_inode_mode(
4258                                        sctx->left_path->nodes[0], left_ii);
4259                }
4260        }
4261
4262out:
4263        return ret;
4264}
4265
4266/*
4267 * We have to process new refs before deleted refs, but compare_trees gives us
4268 * the new and deleted refs mixed. To fix this, we record the new/deleted refs
4269 * first and later process them in process_recorded_refs.
4270 * For the cur_inode_new_gen case, we skip recording completely because
4271 * changed_inode did already initiate processing of refs. The reason for this is
4272 * that in this case, compare_tree actually compares the refs of 2 different
4273 * inodes. To fix this, process_all_refs is used in changed_inode to handle all
4274 * refs of the right tree as deleted and all refs of the left tree as new.
4275 */
4276static int changed_ref(struct send_ctx *sctx,
4277                       enum btrfs_compare_tree_result result)
4278{
4279        int ret = 0;
4280
4281        BUG_ON(sctx->cur_ino != sctx->cmp_key->objectid);
4282
4283        if (!sctx->cur_inode_new_gen &&
4284            sctx->cur_ino != BTRFS_FIRST_FREE_OBJECTID) {
4285                if (result == BTRFS_COMPARE_TREE_NEW)
4286                        ret = record_new_ref(sctx);
4287                else if (result == BTRFS_COMPARE_TREE_DELETED)
4288                        ret = record_deleted_ref(sctx);
4289                else if (result == BTRFS_COMPARE_TREE_CHANGED)
4290                        ret = record_changed_ref(sctx);
4291        }
4292
4293        return ret;
4294}
4295
4296/*
4297 * Process new/deleted/changed xattrs. We skip processing in the
4298 * cur_inode_new_gen case because changed_inode did already initiate processing
4299 * of xattrs. The reason is the same as in changed_ref
4300 */
4301static int changed_xattr(struct send_ctx *sctx,
4302                         enum btrfs_compare_tree_result result)
4303{
4304        int ret = 0;
4305
4306        BUG_ON(sctx->cur_ino != sctx->cmp_key->objectid);
4307
4308        if (!sctx->cur_inode_new_gen && !sctx->cur_inode_deleted) {
4309                if (result == BTRFS_COMPARE_TREE_NEW)
4310                        ret = process_new_xattr(sctx);
4311                else if (result == BTRFS_COMPARE_TREE_DELETED)
4312                        ret = process_deleted_xattr(sctx);
4313                else if (result == BTRFS_COMPARE_TREE_CHANGED)
4314                        ret = process_changed_xattr(sctx);
4315        }
4316
4317        return ret;
4318}
4319
4320/*
4321 * Process new/deleted/changed extents. We skip processing in the
4322 * cur_inode_new_gen case because changed_inode did already initiate processing
4323 * of extents. The reason is the same as in changed_ref
4324 */
4325static int changed_extent(struct send_ctx *sctx,
4326                          enum btrfs_compare_tree_result result)
4327{
4328        int ret = 0;
4329
4330        BUG_ON(sctx->cur_ino != sctx->cmp_key->objectid);
4331
4332        if (!sctx->cur_inode_new_gen && !sctx->cur_inode_deleted) {
4333                if (result != BTRFS_COMPARE_TREE_DELETED)
4334                        ret = process_extent(sctx, sctx->left_path,
4335                                        sctx->cmp_key);
4336        }
4337
4338        return ret;
4339}
4340
4341/*
4342 * Updates compare related fields in sctx and simply forwards to the actual
4343 * changed_xxx functions.
4344 */
4345static int changed_cb(struct btrfs_root *left_root,
4346                      struct btrfs_root *right_root,
4347                      struct btrfs_path *left_path,
4348                      struct btrfs_path *right_path,
4349                      struct btrfs_key *key,
4350                      enum btrfs_compare_tree_result result,
4351                      void *ctx)
4352{
4353        int ret = 0;
4354        struct send_ctx *sctx = ctx;
4355
4356        sctx->left_path = left_path;
4357        sctx->right_path = right_path;
4358        sctx->cmp_key = key;
4359
4360        ret = finish_inode_if_needed(sctx, 0);
4361        if (ret < 0)
4362                goto out;
4363
4364        /* Ignore non-FS objects */
4365        if (key->objectid == BTRFS_FREE_INO_OBJECTID ||
4366            key->objectid == BTRFS_FREE_SPACE_OBJECTID)
4367                goto out;
4368
4369        if (key->type == BTRFS_INODE_ITEM_KEY)
4370                ret = changed_inode(sctx, result);
4371        else if (key->type == BTRFS_INODE_REF_KEY ||
4372                 key->type == BTRFS_INODE_EXTREF_KEY)
4373                ret = changed_ref(sctx, result);
4374        else if (key->type == BTRFS_XATTR_ITEM_KEY)
4375                ret = changed_xattr(sctx, result);
4376        else if (key->type == BTRFS_EXTENT_DATA_KEY)
4377                ret = changed_extent(sctx, result);
4378
4379out:
4380        return ret;
4381}
4382
4383static int full_send_tree(struct send_ctx *sctx)
4384{
4385        int ret;
4386        struct btrfs_trans_handle *trans = NULL;
4387        struct btrfs_root *send_root = sctx->send_root;
4388        struct btrfs_key key;
4389        struct btrfs_key found_key;
4390        struct btrfs_path *path;
4391        struct extent_buffer *eb;
4392        int slot;
4393        u64 start_ctransid;
4394        u64 ctransid;
4395
4396        path = alloc_path_for_send();
4397        if (!path)
4398                return -ENOMEM;
4399
4400        spin_lock(&send_root->root_times_lock);
4401        start_ctransid = btrfs_root_ctransid(&send_root->root_item);
4402        spin_unlock(&send_root->root_times_lock);
4403
4404        key.objectid = BTRFS_FIRST_FREE_OBJECTID;
4405        key.type = BTRFS_INODE_ITEM_KEY;
4406        key.offset = 0;
4407
4408join_trans:
4409        /*
4410         * We need to make sure the transaction does not get committed
4411         * while we do anything on commit roots. Join a transaction to prevent
4412         * this.
4413         */
4414        trans = btrfs_join_transaction(send_root);
4415        if (IS_ERR(trans)) {
4416                ret = PTR_ERR(trans);
4417                trans = NULL;
4418                goto out;
4419        }
4420
4421        /*
4422         * Make sure the tree has not changed after re-joining. We detect this
4423         * by comparing start_ctransid and ctransid. They should always match.
4424         */
4425        spin_lock(&send_root->root_times_lock);
4426        ctransid = btrfs_root_ctransid(&send_root->root_item);
4427        spin_unlock(&send_root->root_times_lock);
4428
4429        if (ctransid != start_ctransid) {
4430                WARN(1, KERN_WARNING "btrfs: the root that you're trying to "
4431                                     "send was modified in between. This is "
4432                                     "probably a bug.\n");
4433                ret = -EIO;
4434                goto out;
4435        }
4436
4437        ret = btrfs_search_slot_for_read(send_root, &key, path, 1, 0);
4438        if (ret < 0)
4439                goto out;
4440        if (ret)
4441                goto out_finish;
4442
4443        while (1) {
4444                /*
4445                 * When someone want to commit while we iterate, end the
4446                 * joined transaction and rejoin.
4447                 */
4448                if (btrfs_should_end_transaction(trans, send_root)) {
4449                        ret = btrfs_end_transaction(trans, send_root);
4450                        trans = NULL;
4451                        if (ret < 0)
4452                                goto out;
4453                        btrfs_release_path(path);
4454                        goto join_trans;
4455                }
4456
4457                eb = path->nodes[0];
4458                slot = path->slots[0];
4459                btrfs_item_key_to_cpu(eb, &found_key, slot);
4460
4461                ret = changed_cb(send_root, NULL, path, NULL,
4462                                &found_key, BTRFS_COMPARE_TREE_NEW, sctx);
4463                if (ret < 0)
4464                        goto out;
4465
4466                key.objectid = found_key.objectid;
4467                key.type = found_key.type;
4468                key.offset = found_key.offset + 1;
4469
4470                ret = btrfs_next_item(send_root, path);
4471                if (ret < 0)
4472                        goto out;
4473                if (ret) {
4474                        ret  = 0;
4475                        break;
4476                }
4477        }
4478
4479out_finish:
4480        ret = finish_inode_if_needed(sctx, 1);
4481
4482out:
4483        btrfs_free_path(path);
4484        if (trans) {
4485                if (!ret)
4486                        ret = btrfs_end_transaction(trans, send_root);
4487                else
4488                        btrfs_end_transaction(trans, send_root);
4489        }
4490        return ret;
4491}
4492
4493static int send_subvol(struct send_ctx *sctx)
4494{
4495        int ret;
4496
4497        ret = send_header(sctx);
4498        if (ret < 0)
4499                goto out;
4500
4501        ret = send_subvol_begin(sctx);
4502        if (ret < 0)
4503                goto out;
4504
4505        if (sctx->parent_root) {
4506                ret = btrfs_compare_trees(sctx->send_root, sctx->parent_root,
4507                                changed_cb, sctx);
4508                if (ret < 0)
4509                        goto out;
4510                ret = finish_inode_if_needed(sctx, 1);
4511                if (ret < 0)
4512                        goto out;
4513        } else {
4514                ret = full_send_tree(sctx);
4515                if (ret < 0)
4516                        goto out;
4517        }
4518
4519out:
4520        if (!ret)
4521                ret = close_cur_inode_file(sctx);
4522        else
4523                close_cur_inode_file(sctx);
4524
4525        free_recorded_refs(sctx);
4526        return ret;
4527}
4528
4529long btrfs_ioctl_send(struct file *mnt_file, void __user *arg_)
4530{
4531        int ret = 0;
4532        struct btrfs_root *send_root;
4533        struct btrfs_root *clone_root;
4534        struct btrfs_fs_info *fs_info;
4535        struct btrfs_ioctl_send_args *arg = NULL;
4536        struct btrfs_key key;
4537        struct file *filp = NULL;
4538        struct send_ctx *sctx = NULL;
4539        u32 i;
4540        u64 *clone_sources_tmp = NULL;
4541
4542        if (!capable(CAP_SYS_ADMIN))
4543                return -EPERM;
4544
4545        send_root = BTRFS_I(fdentry(mnt_file)->d_inode)->root;
4546        fs_info = send_root->fs_info;
4547
4548        arg = memdup_user(arg_, sizeof(*arg));
4549        if (IS_ERR(arg)) {
4550                ret = PTR_ERR(arg);
4551                arg = NULL;
4552                goto out;
4553        }
4554
4555        if (!access_ok(VERIFY_READ, arg->clone_sources,
4556                        sizeof(*arg->clone_sources *
4557                        arg->clone_sources_count))) {
4558                ret = -EFAULT;
4559                goto out;
4560        }
4561
4562        sctx = kzalloc(sizeof(struct send_ctx), GFP_NOFS);
4563        if (!sctx) {
4564                ret = -ENOMEM;
4565                goto out;
4566        }
4567
4568        INIT_LIST_HEAD(&sctx->new_refs);
4569        INIT_LIST_HEAD(&sctx->deleted_refs);
4570        INIT_RADIX_TREE(&sctx->name_cache, GFP_NOFS);
4571        INIT_LIST_HEAD(&sctx->name_cache_list);
4572
4573        sctx->send_filp = fget(arg->send_fd);
4574        if (IS_ERR(sctx->send_filp)) {
4575                ret = PTR_ERR(sctx->send_filp);
4576                goto out;
4577        }
4578
4579        sctx->mnt = mnt_file->f_path.mnt;
4580
4581        sctx->send_root = send_root;
4582        sctx->clone_roots_cnt = arg->clone_sources_count;
4583
4584        sctx->send_max_size = BTRFS_SEND_BUF_SIZE;
4585        sctx->send_buf = vmalloc(sctx->send_max_size);
4586        if (!sctx->send_buf) {
4587                ret = -ENOMEM;
4588                goto out;
4589        }
4590
4591        sctx->read_buf = vmalloc(BTRFS_SEND_READ_SIZE);
4592        if (!sctx->read_buf) {
4593                ret = -ENOMEM;
4594                goto out;
4595        }
4596
4597        sctx->clone_roots = vzalloc(sizeof(struct clone_root) *
4598                        (arg->clone_sources_count + 1));
4599        if (!sctx->clone_roots) {
4600                ret = -ENOMEM;
4601                goto out;
4602        }
4603
4604        if (arg->clone_sources_count) {
4605                clone_sources_tmp = vmalloc(arg->clone_sources_count *
4606                                sizeof(*arg->clone_sources));
4607                if (!clone_sources_tmp) {
4608                        ret = -ENOMEM;
4609                        goto out;
4610                }
4611
4612                ret = copy_from_user(clone_sources_tmp, arg->clone_sources,
4613                                arg->clone_sources_count *
4614                                sizeof(*arg->clone_sources));
4615                if (ret) {
4616                        ret = -EFAULT;
4617                        goto out;
4618                }
4619
4620                for (i = 0; i < arg->clone_sources_count; i++) {
4621                        key.objectid = clone_sources_tmp[i];
4622                        key.type = BTRFS_ROOT_ITEM_KEY;
4623                        key.offset = (u64)-1;
4624                        clone_root = btrfs_read_fs_root_no_name(fs_info, &key);
4625                        if (!clone_root) {
4626                                ret = -EINVAL;
4627                                goto out;
4628                        }
4629                        if (IS_ERR(clone_root)) {
4630                                ret = PTR_ERR(clone_root);
4631                                goto out;
4632                        }
4633                        sctx->clone_roots[i].root = clone_root;
4634                }
4635                vfree(clone_sources_tmp);
4636                clone_sources_tmp = NULL;
4637        }
4638
4639        if (arg->parent_root) {
4640                key.objectid = arg->parent_root;
4641                key.type = BTRFS_ROOT_ITEM_KEY;
4642                key.offset = (u64)-1;
4643                sctx->parent_root = btrfs_read_fs_root_no_name(fs_info, &key);
4644                if (!sctx->parent_root) {
4645                        ret = -EINVAL;
4646                        goto out;
4647                }
4648        }
4649
4650        /*
4651         * Clones from send_root are allowed, but only if the clone source
4652         * is behind the current send position. This is checked while searching
4653         * for possible clone sources.
4654         */
4655        sctx->clone_roots[sctx->clone_roots_cnt++].root = sctx->send_root;
4656
4657        /* We do a bsearch later */
4658        sort(sctx->clone_roots, sctx->clone_roots_cnt,
4659                        sizeof(*sctx->clone_roots), __clone_root_cmp_sort,
4660                        NULL);
4661
4662        ret = send_subvol(sctx);
4663        if (ret < 0)
4664                goto out;
4665
4666        ret = begin_cmd(sctx, BTRFS_SEND_C_END);
4667        if (ret < 0)
4668                goto out;
4669        ret = send_cmd(sctx);
4670        if (ret < 0)
4671                goto out;
4672
4673out:
4674        if (filp)
4675                fput(filp);
4676        kfree(arg);
4677        vfree(clone_sources_tmp);
4678
4679        if (sctx) {
4680                if (sctx->send_filp)
4681                        fput(sctx->send_filp);
4682
4683                vfree(sctx->clone_roots);
4684                vfree(sctx->send_buf);
4685                vfree(sctx->read_buf);
4686
4687                name_cache_free(sctx);
4688
4689                kfree(sctx);
4690        }
4691
4692        return ret;
4693}
4694
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.