diff options
Diffstat (limited to 'fs/btrfs/file.c')
| -rw-r--r-- | fs/btrfs/file.c | 4627 |
1 files changed, 2973 insertions, 1654 deletions
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index a005fe2c072a..7a501e73d880 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -1,526 +1,113 @@ +// SPDX-License-Identifier: GPL-2.0 /* * Copyright (C) 2007 Oracle. All rights reserved. - * - * This program is free software; you can redistribute it and/or - * modify it under the terms of the GNU General Public - * License v2 as published by the Free Software Foundation. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU - * General Public License for more details. - * - * You should have received a copy of the GNU General Public - * License along with this program; if not, write to the - * Free Software Foundation, Inc., 59 Temple Place - Suite 330, - * Boston, MA 021110-1307, USA. */ #include <linux/fs.h> #include <linux/pagemap.h> -#include <linux/highmem.h> #include <linux/time.h> #include <linux/init.h> #include <linux/string.h> #include <linux/backing-dev.h> -#include <linux/mpage.h> -#include <linux/aio.h> #include <linux/falloc.h> -#include <linux/swap.h> #include <linux/writeback.h> -#include <linux/statfs.h> #include <linux/compat.h> #include <linux/slab.h> #include <linux/btrfs.h> +#include <linux/uio.h> +#include <linux/iversion.h> +#include <linux/fsverity.h> #include "ctree.h" +#include "direct-io.h" #include "disk-io.h" #include "transaction.h" #include "btrfs_inode.h" -#include "print-tree.h" #include "tree-log.h" #include "locking.h" -#include "compat.h" -#include "volumes.h" - -static struct kmem_cache *btrfs_inode_defrag_cachep; -/* - * when auto defrag is enabled we - * queue up these defrag structs to remember which - * inodes need defragging passes - */ -struct inode_defrag { - struct rb_node rb_node; - /* objectid */ - u64 ino; - /* - * transid where the defrag was added, we search for - * extents newer than this - */ - u64 transid; - - /* root objectid */ - u64 root; - - /* last offset we were able to defrag */ - u64 last_offset; - - /* if we've wrapped around back to zero once already */ - int cycled; -}; - -static int __compare_inode_defrag(struct inode_defrag *defrag1, - struct inode_defrag *defrag2) -{ - if (defrag1->root > defrag2->root) - return 1; - else if (defrag1->root < defrag2->root) - return -1; - else if (defrag1->ino > defrag2->ino) - return 1; - else if (defrag1->ino < defrag2->ino) - return -1; - else - return 0; -} - -/* pop a record for an inode into the defrag tree. The lock - * must be held already - * - * If you're inserting a record for an older transid than an - * existing record, the transid already in the tree is lowered - * - * If an existing record is found the defrag item you - * pass in is freed - */ -static int __btrfs_add_inode_defrag(struct inode *inode, - struct inode_defrag *defrag) -{ - struct btrfs_root *root = BTRFS_I(inode)->root; - struct inode_defrag *entry; - struct rb_node **p; - struct rb_node *parent = NULL; - int ret; - - p = &root->fs_info->defrag_inodes.rb_node; - while (*p) { - parent = *p; - entry = rb_entry(parent, struct inode_defrag, rb_node); - - ret = __compare_inode_defrag(defrag, entry); - if (ret < 0) - p = &parent->rb_left; - else if (ret > 0) - p = &parent->rb_right; - else { - /* if we're reinserting an entry for - * an old defrag run, make sure to - * lower the transid of our existing record - */ - if (defrag->transid < entry->transid) - entry->transid = defrag->transid; - if (defrag->last_offset > entry->last_offset) - entry->last_offset = defrag->last_offset; - return -EEXIST; - } - } - set_bit(BTRFS_INODE_IN_DEFRAG, &BTRFS_I(inode)->runtime_flags); - rb_link_node(&defrag->rb_node, parent, p); - rb_insert_color(&defrag->rb_node, &root->fs_info->defrag_inodes); - return 0; -} - -static inline int __need_auto_defrag(struct btrfs_root *root) -{ - if (!btrfs_test_opt(root, AUTO_DEFRAG)) - return 0; - - if (btrfs_fs_closing(root->fs_info)) - return 0; - - return 1; -} - -/* - * insert a defrag record for this inode if auto defrag is - * enabled - */ -int btrfs_add_inode_defrag(struct btrfs_trans_handle *trans, - struct inode *inode) -{ - struct btrfs_root *root = BTRFS_I(inode)->root; - struct inode_defrag *defrag; - u64 transid; - int ret; - - if (!__need_auto_defrag(root)) - return 0; - - if (test_bit(BTRFS_INODE_IN_DEFRAG, &BTRFS_I(inode)->runtime_flags)) - return 0; - - if (trans) - transid = trans->transid; - else - transid = BTRFS_I(inode)->root->last_trans; - - defrag = kmem_cache_zalloc(btrfs_inode_defrag_cachep, GFP_NOFS); - if (!defrag) - return -ENOMEM; - - defrag->ino = btrfs_ino(inode); - defrag->transid = transid; - defrag->root = root->root_key.objectid; - - spin_lock(&root->fs_info->defrag_inodes_lock); - if (!test_bit(BTRFS_INODE_IN_DEFRAG, &BTRFS_I(inode)->runtime_flags)) { - /* - * If we set IN_DEFRAG flag and evict the inode from memory, - * and then re-read this inode, this new inode doesn't have - * IN_DEFRAG flag. At the case, we may find the existed defrag. - */ - ret = __btrfs_add_inode_defrag(inode, defrag); - if (ret) - kmem_cache_free(btrfs_inode_defrag_cachep, defrag); - } else { - kmem_cache_free(btrfs_inode_defrag_cachep, defrag); - } - spin_unlock(&root->fs_info->defrag_inodes_lock); - return 0; -} +#include "qgroup.h" +#include "compression.h" +#include "delalloc-space.h" +#include "reflink.h" +#include "subpage.h" +#include "fs.h" +#include "accessors.h" +#include "extent-tree.h" +#include "file-item.h" +#include "ioctl.h" +#include "file.h" +#include "super.h" +#include "print-tree.h" /* - * Requeue the defrag object. If there is a defrag object that points to - * the same inode in the tree, we will merge them together (by - * __btrfs_add_inode_defrag()) and free the one that we want to requeue. + * Unlock folio after btrfs_file_write() is done with it. */ -static void btrfs_requeue_inode_defrag(struct inode *inode, - struct inode_defrag *defrag) +static void btrfs_drop_folio(struct btrfs_fs_info *fs_info, struct folio *folio, + u64 pos, u64 copied) { - struct btrfs_root *root = BTRFS_I(inode)->root; - int ret; - - if (!__need_auto_defrag(root)) - goto out; + u64 block_start = round_down(pos, fs_info->sectorsize); + u64 block_len = round_up(pos + copied, fs_info->sectorsize) - block_start; + ASSERT(block_len <= U32_MAX); /* - * Here we don't check the IN_DEFRAG flag, because we need merge - * them together. + * Folio checked is some magic around finding folios that have been + * modified without going through btrfs_dirty_folio(). Clear it here. + * There should be no need to mark the pages accessed as + * prepare_one_folio() should have marked them accessed in + * prepare_one_folio() via find_or_create_page() */ - spin_lock(&root->fs_info->defrag_inodes_lock); - ret = __btrfs_add_inode_defrag(inode, defrag); - spin_unlock(&root->fs_info->defrag_inodes_lock); - if (ret) - goto out; - return; -out: - kmem_cache_free(btrfs_inode_defrag_cachep, defrag); + btrfs_folio_clamp_clear_checked(fs_info, folio, block_start, block_len); + folio_unlock(folio); + folio_put(folio); } /* - * pick the defragable inode that we want, if it doesn't exist, we will get - * the next one. + * After copy_folio_from_iter_atomic(), update the following things for delalloc: + * - Mark newly dirtied folio as DELALLOC in the io tree. + * Used to advise which range is to be written back. + * - Mark modified folio as Uptodate/Dirty and not needing COW fixup + * - Update inode size for past EOF write */ -static struct inode_defrag * -btrfs_pick_defrag_inode(struct btrfs_fs_info *fs_info, u64 root, u64 ino) +int btrfs_dirty_folio(struct btrfs_inode *inode, struct folio *folio, loff_t pos, + size_t write_bytes, struct extent_state **cached, bool noreserve) { - struct inode_defrag *entry = NULL; - struct inode_defrag tmp; - struct rb_node *p; - struct rb_node *parent = NULL; - int ret; - - tmp.ino = ino; - tmp.root = root; - - spin_lock(&fs_info->defrag_inodes_lock); - p = fs_info->defrag_inodes.rb_node; - while (p) { - parent = p; - entry = rb_entry(parent, struct inode_defrag, rb_node); - - ret = __compare_inode_defrag(&tmp, entry); - if (ret < 0) - p = parent->rb_left; - else if (ret > 0) - p = parent->rb_right; - else - goto out; - } - - if (parent && __compare_inode_defrag(&tmp, entry) > 0) { - parent = rb_next(parent); - if (parent) - entry = rb_entry(parent, struct inode_defrag, rb_node); - else - entry = NULL; - } -out: - if (entry) - rb_erase(parent, &fs_info->defrag_inodes); - spin_unlock(&fs_info->defrag_inodes_lock); - return entry; -} - -void btrfs_cleanup_defrag_inodes(struct btrfs_fs_info *fs_info) -{ - struct inode_defrag *defrag; - struct rb_node *node; - - spin_lock(&fs_info->defrag_inodes_lock); - node = rb_first(&fs_info->defrag_inodes); - while (node) { - rb_erase(node, &fs_info->defrag_inodes); - defrag = rb_entry(node, struct inode_defrag, rb_node); - kmem_cache_free(btrfs_inode_defrag_cachep, defrag); - - if (need_resched()) { - spin_unlock(&fs_info->defrag_inodes_lock); - cond_resched(); - spin_lock(&fs_info->defrag_inodes_lock); - } - - node = rb_first(&fs_info->defrag_inodes); - } - spin_unlock(&fs_info->defrag_inodes_lock); -} - -#define BTRFS_DEFRAG_BATCH 1024 - -static int __btrfs_run_defrag_inode(struct btrfs_fs_info *fs_info, - struct inode_defrag *defrag) -{ - struct btrfs_root *inode_root; - struct inode *inode; - struct btrfs_key key; - struct btrfs_ioctl_defrag_range_args range; - int num_defrag; - int index; - int ret; - - /* get the inode */ - key.objectid = defrag->root; - btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY); - key.offset = (u64)-1; - - index = srcu_read_lock(&fs_info->subvol_srcu); - - inode_root = btrfs_read_fs_root_no_name(fs_info, &key); - if (IS_ERR(inode_root)) { - ret = PTR_ERR(inode_root); - goto cleanup; - } - - key.objectid = defrag->ino; - btrfs_set_key_type(&key, BTRFS_INODE_ITEM_KEY); - key.offset = 0; - inode = btrfs_iget(fs_info->sb, &key, inode_root, NULL); - if (IS_ERR(inode)) { - ret = PTR_ERR(inode); - goto cleanup; - } - srcu_read_unlock(&fs_info->subvol_srcu, index); - - /* do a chunk of defrag */ - clear_bit(BTRFS_INODE_IN_DEFRAG, &BTRFS_I(inode)->runtime_flags); - memset(&range, 0, sizeof(range)); - range.len = (u64)-1; - range.start = defrag->last_offset; - - sb_start_write(fs_info->sb); - num_defrag = btrfs_defrag_file(inode, NULL, &range, defrag->transid, - BTRFS_DEFRAG_BATCH); - sb_end_write(fs_info->sb); - /* - * if we filled the whole defrag batch, there - * must be more work to do. Queue this defrag - * again - */ - if (num_defrag == BTRFS_DEFRAG_BATCH) { - defrag->last_offset = range.start; - btrfs_requeue_inode_defrag(inode, defrag); - } else if (defrag->last_offset && !defrag->cycled) { - /* - * we didn't fill our defrag batch, but - * we didn't start at zero. Make sure we loop - * around to the start of the file. - */ - defrag->last_offset = 0; - defrag->cycled = 1; - btrfs_requeue_inode_defrag(inode, defrag); - } else { - kmem_cache_free(btrfs_inode_defrag_cachep, defrag); - } - - iput(inode); - return 0; -cleanup: - srcu_read_unlock(&fs_info->subvol_srcu, index); - kmem_cache_free(btrfs_inode_defrag_cachep, defrag); - return ret; -} - -/* - * run through the list of inodes in the FS that need - * defragging - */ -int btrfs_run_defrag_inodes(struct btrfs_fs_info *fs_info) -{ - struct inode_defrag *defrag; - u64 first_ino = 0; - u64 root_objectid = 0; - - atomic_inc(&fs_info->defrag_running); - while(1) { - /* Pause the auto defragger. */ - if (test_bit(BTRFS_FS_STATE_REMOUNTING, - &fs_info->fs_state)) - break; + struct btrfs_fs_info *fs_info = inode->root->fs_info; + int ret = 0; + u64 num_bytes; + u64 start_pos; + u64 end_of_last_block; + const u64 end_pos = pos + write_bytes; + loff_t isize = i_size_read(&inode->vfs_inode); + unsigned int extra_bits = 0; - if (!__need_auto_defrag(fs_info->tree_root)) - break; + if (write_bytes == 0) + return 0; - /* find an inode to defrag */ - defrag = btrfs_pick_defrag_inode(fs_info, root_objectid, - first_ino); - if (!defrag) { - if (root_objectid || first_ino) { - root_objectid = 0; - first_ino = 0; - continue; - } else { - break; - } - } + if (noreserve) + extra_bits |= EXTENT_NORESERVE; - first_ino = defrag->ino + 1; - root_objectid = defrag->root; + start_pos = round_down(pos, fs_info->sectorsize); + num_bytes = round_up(end_pos - start_pos, fs_info->sectorsize); + ASSERT(num_bytes <= U32_MAX); + ASSERT(folio_pos(folio) <= pos && folio_next_pos(folio) >= end_pos); - __btrfs_run_defrag_inode(fs_info, defrag); - } - atomic_dec(&fs_info->defrag_running); + end_of_last_block = start_pos + num_bytes - 1; /* - * during unmount, we use the transaction_wait queue to - * wait for the defragger to stop + * The pages may have already been dirty, clear out old accounting so + * we can set things up properly */ - wake_up(&fs_info->transaction_wait); - return 0; -} - -/* simple helper to fault in pages and copy. This should go away - * and be replaced with calls into generic code. - */ -static noinline int btrfs_copy_from_user(loff_t pos, int num_pages, - size_t write_bytes, - struct page **prepared_pages, - struct iov_iter *i) -{ - size_t copied = 0; - size_t total_copied = 0; - int pg = 0; - int offset = pos & (PAGE_CACHE_SIZE - 1); - - while (write_bytes > 0) { - size_t count = min_t(size_t, - PAGE_CACHE_SIZE - offset, write_bytes); - struct page *page = prepared_pages[pg]; - /* - * Copy data from userspace to the current page - * - * Disable pagefault to avoid recursive lock since - * the pages are already locked - */ - pagefault_disable(); - copied = iov_iter_copy_from_user_atomic(page, i, offset, count); - pagefault_enable(); - - /* Flush processor's dcache for this page */ - flush_dcache_page(page); - - /* - * if we get a partial write, we can end up with - * partially up to date pages. These add - * a lot of complexity, so make sure they don't - * happen by forcing this copy to be retried. - * - * The rest of the btrfs_file_write code will fall - * back to page at a time copies after we return 0. - */ - if (!PageUptodate(page) && copied < count) - copied = 0; - - iov_iter_advance(i, copied); - write_bytes -= copied; - total_copied += copied; - - /* Return to btrfs_file_aio_write to fault page */ - if (unlikely(copied == 0)) - break; - - if (unlikely(copied < PAGE_CACHE_SIZE - offset)) { - offset += copied; - } else { - pg++; - offset = 0; - } - } - return total_copied; -} - -/* - * unlocks pages after btrfs_file_write is done with them - */ -static void btrfs_drop_pages(struct page **pages, size_t num_pages) -{ - size_t i; - for (i = 0; i < num_pages; i++) { - /* page checked is some magic around finding pages that - * have been modified without going through btrfs_set_page_dirty - * clear it here - */ - ClearPageChecked(pages[i]); - unlock_page(pages[i]); - mark_page_accessed(pages[i]); - page_cache_release(pages[i]); - } -} - -/* - * after copy_from_user, pages need to be dirtied and we need to make - * sure holes are created between the current EOF and the start of - * any next extents (if required). - * - * this also makes the decision about creating an inline extent vs - * doing real data extents, marking pages dirty and delalloc as required. - */ -int btrfs_dirty_pages(struct btrfs_root *root, struct inode *inode, - struct page **pages, size_t num_pages, - loff_t pos, size_t write_bytes, - struct extent_state **cached) -{ - int err = 0; - int i; - u64 num_bytes; - u64 start_pos; - u64 end_of_last_block; - u64 end_pos = pos + write_bytes; - loff_t isize = i_size_read(inode); + btrfs_clear_extent_bit(&inode->io_tree, start_pos, end_of_last_block, + EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, + cached); - start_pos = pos & ~((u64)root->sectorsize - 1); - num_bytes = ALIGN(write_bytes + pos - start_pos, root->sectorsize); - - end_of_last_block = start_pos + num_bytes - 1; - err = btrfs_set_extent_delalloc(inode, start_pos, end_of_last_block, - cached); - if (err) - return err; + ret = btrfs_set_extent_delalloc(inode, start_pos, end_of_last_block, + extra_bits, cached); + if (ret) + return ret; - for (i = 0; i < num_pages; i++) { - struct page *p = pages[i]; - SetPageUptodate(p); - ClearPageChecked(p); - set_page_dirty(p); - } + btrfs_folio_clamp_set_uptodate(fs_info, folio, start_pos, num_bytes); + btrfs_folio_clamp_clear_checked(fs_info, folio, start_pos, num_bytes); + btrfs_folio_clamp_set_dirty(fs_info, folio, start_pos, num_bytes); /* * we've only changed i_size in ram, and we haven't updated @@ -528,142 +115,11 @@ int btrfs_dirty_pages(struct btrfs_root *root, struct inode *inode, * at this time. */ if (end_pos > isize) - i_size_write(inode, end_pos); + i_size_write(&inode->vfs_inode, end_pos); return 0; } /* - * this drops all the extents in the cache that intersect the range - * [start, end]. Existing extents are split as required. - */ -void btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end, - int skip_pinned) -{ - struct extent_map *em; - struct extent_map *split = NULL; - struct extent_map *split2 = NULL; - struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; - u64 len = end - start + 1; - u64 gen; - int ret; - int testend = 1; - unsigned long flags; - int compressed = 0; - bool modified; - - WARN_ON(end < start); - if (end == (u64)-1) { - len = (u64)-1; - testend = 0; - } - while (1) { - int no_splits = 0; - - modified = false; - if (!split) - split = alloc_extent_map(); - if (!split2) - split2 = alloc_extent_map(); - if (!split || !split2) - no_splits = 1; - - write_lock(&em_tree->lock); - em = lookup_extent_mapping(em_tree, start, len); - if (!em) { - write_unlock(&em_tree->lock); - break; - } - flags = em->flags; - gen = em->generation; - if (skip_pinned && test_bit(EXTENT_FLAG_PINNED, &em->flags)) { - if (testend && em->start + em->len >= start + len) { - free_extent_map(em); - write_unlock(&em_tree->lock); - break; - } - start = em->start + em->len; - if (testend) - len = start + len - (em->start + em->len); - free_extent_map(em); - write_unlock(&em_tree->lock); - continue; - } - compressed = test_bit(EXTENT_FLAG_COMPRESSED, &em->flags); - clear_bit(EXTENT_FLAG_PINNED, &em->flags); - clear_bit(EXTENT_FLAG_LOGGING, &flags); - modified = !list_empty(&em->list); - remove_extent_mapping(em_tree, em); - if (no_splits) - goto next; - - if (em->block_start < EXTENT_MAP_LAST_BYTE && - em->start < start) { - split->start = em->start; - split->len = start - em->start; - split->orig_start = em->orig_start; - split->block_start = em->block_start; - - if (compressed) - split->block_len = em->block_len; - else - split->block_len = split->len; - split->ram_bytes = em->ram_bytes; - split->orig_block_len = max(split->block_len, - em->orig_block_len); - split->generation = gen; - split->bdev = em->bdev; - split->flags = flags; - split->compress_type = em->compress_type; - ret = add_extent_mapping(em_tree, split, modified); - BUG_ON(ret); /* Logic error */ - free_extent_map(split); - split = split2; - split2 = NULL; - } - if (em->block_start < EXTENT_MAP_LAST_BYTE && - testend && em->start + em->len > start + len) { - u64 diff = start + len - em->start; - - split->start = start + len; - split->len = em->start + em->len - (start + len); - split->bdev = em->bdev; - split->flags = flags; - split->compress_type = em->compress_type; - split->generation = gen; - split->orig_block_len = max(em->block_len, - em->orig_block_len); - split->ram_bytes = em->ram_bytes; - - if (compressed) { - split->block_len = em->block_len; - split->block_start = em->block_start; - split->orig_start = em->orig_start; - } else { - split->block_len = split->len; - split->block_start = em->block_start + diff; - split->orig_start = em->orig_start; - } - - ret = add_extent_mapping(em_tree, split, modified); - BUG_ON(ret); /* Logic error */ - free_extent_map(split); - split = NULL; - } -next: - write_unlock(&em_tree->lock); - - /* once for us */ - free_extent_map(em); - /* once for the tree*/ - free_extent_map(em); - } - if (split) - free_extent_map(split); - if (split2) - free_extent_map(split2); -} - -/* * this is very complex, but the basic idea is to drop all extents * in the range start - end. hint_block is filled in with a block number * that would be a good hint to the block allocator for this file. @@ -671,44 +127,67 @@ next: * If an extent intersects the range but is not entirely inside the range * it is either truncated or split. Anything entirely inside the range * is deleted from the tree. + * + * Note: the VFS' inode number of bytes is not updated, it's up to the caller + * to deal with that. We set the field 'bytes_found' of the arguments structure + * with the number of allocated bytes found in the target range, so that the + * caller can update the inode's number of bytes in an atomic way when + * replacing extents in a range to avoid races with stat(2). */ -int __btrfs_drop_extents(struct btrfs_trans_handle *trans, - struct btrfs_root *root, struct inode *inode, - struct btrfs_path *path, u64 start, u64 end, - u64 *drop_end, int drop_cache) +int btrfs_drop_extents(struct btrfs_trans_handle *trans, + struct btrfs_root *root, struct btrfs_inode *inode, + struct btrfs_drop_extents_args *args) { + struct btrfs_fs_info *fs_info = root->fs_info; struct extent_buffer *leaf; struct btrfs_file_extent_item *fi; struct btrfs_key key; struct btrfs_key new_key; u64 ino = btrfs_ino(inode); - u64 search_start = start; + u64 search_start = args->start; u64 disk_bytenr = 0; u64 num_bytes = 0; u64 extent_offset = 0; u64 extent_end = 0; + u64 last_end = args->start; int del_nr = 0; int del_slot = 0; int extent_type; int recow; int ret; int modify_tree = -1; - int update_refs = (root->ref_cows || root == root->fs_info->tree_root); + int update_refs; int found = 0; + struct btrfs_path *path = args->path; - if (drop_cache) - btrfs_drop_extent_cache(inode, start, end - 1, 0); + args->bytes_found = 0; + args->extent_inserted = false; - if (start >= BTRFS_I(inode)->disk_i_size) + /* Must always have a path if ->replace_extent is true */ + ASSERT(!(args->replace_extent && !args->path)); + + if (!path) { + path = btrfs_alloc_path(); + if (!path) { + ret = -ENOMEM; + goto out; + } + } + + if (args->drop_cache) + btrfs_drop_extent_map_range(inode, args->start, args->end - 1, false); + + if (data_race(args->start >= inode->disk_i_size) && !args->replace_extent) modify_tree = 0; + update_refs = (btrfs_root_id(root) != BTRFS_TREE_LOG_OBJECTID); while (1) { recow = 0; ret = btrfs_lookup_file_extent(trans, root, path, ino, search_start, modify_tree); if (ret < 0) break; - if (ret > 0 && path->slots[0] > 0 && search_start == start) { + if (ret > 0 && path->slots[0] > 0 && search_start == args->start) { leaf = path->nodes[0]; btrfs_item_key_to_cpu(leaf, &key, path->slots[0] - 1); if (key.objectid == ino && @@ -719,7 +198,11 @@ int __btrfs_drop_extents(struct btrfs_trans_handle *trans, next_slot: leaf = path->nodes[0]; if (path->slots[0] >= btrfs_header_nritems(leaf)) { - BUG_ON(del_nr > 0); + if (WARN_ON(del_nr > 0)) { + btrfs_print_leaf(leaf); + ret = -EINVAL; + break; + } ret = btrfs_next_leaf(root, path); if (ret < 0) break; @@ -732,8 +215,16 @@ next_slot: } btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); - if (key.objectid > ino || - key.type > BTRFS_EXTENT_DATA_KEY || key.offset >= end) + + if (key.objectid > ino) + break; + if (WARN_ON_ONCE(key.objectid < ino) || + key.type < BTRFS_EXTENT_DATA_KEY) { + ASSERT(del_nr == 0); + path->slots[0]++; + goto next_slot; + } + if (key.type > BTRFS_EXTENT_DATA_KEY || key.offset >= args->end) break; fi = btrfs_item_ptr(leaf, path->slots[0], @@ -749,10 +240,24 @@ next_slot: btrfs_file_extent_num_bytes(leaf, fi); } else if (extent_type == BTRFS_FILE_EXTENT_INLINE) { extent_end = key.offset + - btrfs_file_extent_inline_len(leaf, fi); + btrfs_file_extent_ram_bytes(leaf, fi); } else { - WARN_ON(1); - extent_end = search_start; + /* can't happen */ + BUG(); + } + + /* + * Don't skip extent items representing 0 byte lengths. They + * used to be created (bug) if while punching holes we hit + * -ENOSPC condition. So if we find one here, just ensure we + * delete it, otherwise we would insert a new file extent item + * with the same key (offset) as that 0 bytes length file + * extent item in the call to setup_items_for_insert() later + * in this function. + */ + if (extent_end == key.offset && extent_end >= search_start) { + last_end = extent_end; + goto delete_extent_item; } if (extent_end <= search_start) { @@ -761,7 +266,7 @@ next_slot: } found = 1; - search_start = max(key.offset, start); + search_start = max(key.offset, args->start); if (recow || !modify_tree) { modify_tree = -1; btrfs_release_path(path); @@ -772,12 +277,19 @@ next_slot: * | - range to drop - | * | -------- extent -------- | */ - if (start > key.offset && end < extent_end) { - BUG_ON(del_nr > 0); - BUG_ON(extent_type == BTRFS_FILE_EXTENT_INLINE); + if (args->start > key.offset && args->end < extent_end) { + if (WARN_ON(del_nr > 0)) { + btrfs_print_leaf(leaf); + ret = -EINVAL; + break; + } + if (extent_type == BTRFS_FILE_EXTENT_INLINE) { + ret = -EOPNOTSUPP; + break; + } memcpy(&new_key, &key, sizeof(new_key)); - new_key.offset = start; + new_key.offset = args->start; ret = btrfs_duplicate_item(trans, root, path, &new_key); if (ret == -EAGAIN) { @@ -791,45 +303,62 @@ next_slot: fi = btrfs_item_ptr(leaf, path->slots[0] - 1, struct btrfs_file_extent_item); btrfs_set_file_extent_num_bytes(leaf, fi, - start - key.offset); + args->start - key.offset); fi = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_file_extent_item); - extent_offset += start - key.offset; + extent_offset += args->start - key.offset; btrfs_set_file_extent_offset(leaf, fi, extent_offset); btrfs_set_file_extent_num_bytes(leaf, fi, - extent_end - start); - btrfs_mark_buffer_dirty(leaf); + extent_end - args->start); if (update_refs && disk_bytenr > 0) { - ret = btrfs_inc_extent_ref(trans, root, - disk_bytenr, num_bytes, 0, - root->root_key.objectid, - new_key.objectid, - start - extent_offset, 0); - BUG_ON(ret); /* -ENOMEM */ + struct btrfs_ref ref = { + .action = BTRFS_ADD_DELAYED_REF, + .bytenr = disk_bytenr, + .num_bytes = num_bytes, + .parent = 0, + .owning_root = btrfs_root_id(root), + .ref_root = btrfs_root_id(root), + }; + btrfs_init_data_ref(&ref, new_key.objectid, + args->start - extent_offset, + 0, false); + ret = btrfs_inc_extent_ref(trans, &ref); + if (unlikely(ret)) { + btrfs_abort_transaction(trans, ret); + break; + } } - key.offset = start; + key.offset = args->start; } /* + * From here on out we will have actually dropped something, so + * last_end can be updated. + */ + last_end = extent_end; + + /* * | ---- range to drop ----- | * | -------- extent -------- | */ - if (start <= key.offset && end < extent_end) { - BUG_ON(extent_type == BTRFS_FILE_EXTENT_INLINE); + if (args->start <= key.offset && args->end < extent_end) { + if (extent_type == BTRFS_FILE_EXTENT_INLINE) { + ret = -EOPNOTSUPP; + break; + } memcpy(&new_key, &key, sizeof(new_key)); - new_key.offset = end; - btrfs_set_item_key_safe(root, path, &new_key); + new_key.offset = args->end; + btrfs_set_item_key_safe(trans, path, &new_key); - extent_offset += end - key.offset; + extent_offset += args->end - key.offset; btrfs_set_file_extent_offset(leaf, fi, extent_offset); btrfs_set_file_extent_num_bytes(leaf, fi, - extent_end - end); - btrfs_mark_buffer_dirty(leaf); + extent_end - args->end); if (update_refs && disk_bytenr > 0) - inode_sub_bytes(inode, end - key.offset); + args->bytes_found += args->end - key.offset; break; } @@ -838,16 +367,22 @@ next_slot: * | ---- range to drop ----- | * | -------- extent -------- | */ - if (start > key.offset && end >= extent_end) { - BUG_ON(del_nr > 0); - BUG_ON(extent_type == BTRFS_FILE_EXTENT_INLINE); + if (args->start > key.offset && args->end >= extent_end) { + if (WARN_ON(del_nr > 0)) { + btrfs_print_leaf(leaf); + ret = -EINVAL; + break; + } + if (extent_type == BTRFS_FILE_EXTENT_INLINE) { + ret = -EOPNOTSUPP; + break; + } btrfs_set_file_extent_num_bytes(leaf, fi, - start - key.offset); - btrfs_mark_buffer_dirty(leaf); + args->start - key.offset); if (update_refs && disk_bytenr > 0) - inode_sub_bytes(inode, extent_end - start); - if (end == extent_end) + args->bytes_found += extent_end - args->start; + if (args->end == extent_end) break; path->slots[0]++; @@ -858,33 +393,46 @@ next_slot: * | ---- range to drop ----- | * | ------ extent ------ | */ - if (start <= key.offset && end >= extent_end) { + if (args->start <= key.offset && args->end >= extent_end) { +delete_extent_item: if (del_nr == 0) { del_slot = path->slots[0]; del_nr = 1; } else { - BUG_ON(del_slot + del_nr != path->slots[0]); + if (WARN_ON(del_slot + del_nr != path->slots[0])) { + btrfs_print_leaf(leaf); + ret = -EINVAL; + break; + } del_nr++; } if (update_refs && extent_type == BTRFS_FILE_EXTENT_INLINE) { - inode_sub_bytes(inode, - extent_end - key.offset); + args->bytes_found += extent_end - key.offset; extent_end = ALIGN(extent_end, - root->sectorsize); + fs_info->sectorsize); } else if (update_refs && disk_bytenr > 0) { - ret = btrfs_free_extent(trans, root, - disk_bytenr, num_bytes, 0, - root->root_key.objectid, - key.objectid, key.offset - - extent_offset, 0); - BUG_ON(ret); /* -ENOMEM */ - inode_sub_bytes(inode, - extent_end - key.offset); + struct btrfs_ref ref = { + .action = BTRFS_DROP_DELAYED_REF, + .bytenr = disk_bytenr, + .num_bytes = num_bytes, + .parent = 0, + .owning_root = btrfs_root_id(root), + .ref_root = btrfs_root_id(root), + }; + btrfs_init_data_ref(&ref, key.objectid, + key.offset - extent_offset, + 0, false); + ret = btrfs_free_extent(trans, &ref); + if (unlikely(ret)) { + btrfs_abort_transaction(trans, ret); + break; + } + args->bytes_found += extent_end - key.offset; } - if (end == extent_end) + if (args->end == extent_end) break; if (path->slots[0] + 1 < btrfs_header_nritems(leaf)) { @@ -894,8 +442,8 @@ next_slot: ret = btrfs_del_items(trans, root, path, del_slot, del_nr); - if (ret) { - btrfs_abort_transaction(trans, root, ret); + if (unlikely(ret)) { + btrfs_abort_transaction(trans, ret); break; } @@ -906,51 +454,71 @@ next_slot: continue; } - BUG_ON(1); + BUG(); } if (!ret && del_nr > 0) { + /* + * Set path->slots[0] to first slot, so that after the delete + * if items are move off from our leaf to its immediate left or + * right neighbor leafs, we end up with a correct and adjusted + * path->slots[0] for our insertion (if args->replace_extent). + */ + path->slots[0] = del_slot; ret = btrfs_del_items(trans, root, path, del_slot, del_nr); if (ret) - btrfs_abort_transaction(trans, root, ret); + btrfs_abort_transaction(trans, ret); } - if (drop_end) - *drop_end = found ? min(end, extent_end) : end; - btrfs_release_path(path); - return ret; -} + leaf = path->nodes[0]; + /* + * If btrfs_del_items() was called, it might have deleted a leaf, in + * which case it unlocked our path, so check path->locks[0] matches a + * write lock. + */ + if (!ret && args->replace_extent && + path->locks[0] == BTRFS_WRITE_LOCK && + btrfs_leaf_free_space(leaf) >= + sizeof(struct btrfs_item) + args->extent_item_size) { + + key.objectid = ino; + key.type = BTRFS_EXTENT_DATA_KEY; + key.offset = args->start; + if (!del_nr && path->slots[0] < btrfs_header_nritems(leaf)) { + struct btrfs_key slot_key; + + btrfs_item_key_to_cpu(leaf, &slot_key, path->slots[0]); + if (btrfs_comp_cpu_keys(&key, &slot_key) > 0) + path->slots[0]++; + } + btrfs_setup_item_for_insert(trans, root, path, &key, + args->extent_item_size); + args->extent_inserted = true; + } -int btrfs_drop_extents(struct btrfs_trans_handle *trans, - struct btrfs_root *root, struct inode *inode, u64 start, - u64 end, int drop_cache) -{ - struct btrfs_path *path; - int ret; + if (!args->path) + btrfs_free_path(path); + else if (!args->extent_inserted) + btrfs_release_path(path); +out: + args->drop_end = found ? min(args->end, last_end) : args->end; - path = btrfs_alloc_path(); - if (!path) - return -ENOMEM; - ret = __btrfs_drop_extents(trans, root, inode, path, start, end, NULL, - drop_cache); - btrfs_free_path(path); return ret; } -static int extent_mergeable(struct extent_buffer *leaf, int slot, - u64 objectid, u64 bytenr, u64 orig_offset, - u64 *start, u64 *end) +static bool extent_mergeable(struct extent_buffer *leaf, int slot, u64 objectid, + u64 bytenr, u64 orig_offset, u64 *start, u64 *end) { struct btrfs_file_extent_item *fi; struct btrfs_key key; u64 extent_end; if (slot < 0 || slot >= btrfs_header_nritems(leaf)) - return 0; + return false; btrfs_item_key_to_cpu(leaf, &key, slot); if (key.objectid != objectid || key.type != BTRFS_EXTENT_DATA_KEY) - return 0; + return false; fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item); if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_REG || @@ -959,15 +527,15 @@ static int extent_mergeable(struct extent_buffer *leaf, int slot, btrfs_file_extent_compression(leaf, fi) || btrfs_file_extent_encryption(leaf, fi) || btrfs_file_extent_other_encoding(leaf, fi)) - return 0; + return false; extent_end = key.offset + btrfs_file_extent_num_bytes(leaf, fi); if ((*start && *start != key.offset) || (*end && *end != extent_end)) - return 0; + return false; *start = key.offset; *end = extent_end; - return 1; + return true; } /* @@ -978,12 +546,13 @@ static int extent_mergeable(struct extent_buffer *leaf, int slot, * two or three. */ int btrfs_mark_extent_written(struct btrfs_trans_handle *trans, - struct inode *inode, u64 start, u64 end) + struct btrfs_inode *inode, u64 start, u64 end) { - struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_root *root = inode->root; struct extent_buffer *leaf; - struct btrfs_path *path; + BTRFS_PATH_AUTO_FREE(path); struct btrfs_file_extent_item *fi; + struct btrfs_ref ref = { 0 }; struct btrfs_key key; struct btrfs_key new_key; u64 bytenr; @@ -996,7 +565,7 @@ int btrfs_mark_extent_written(struct btrfs_trans_handle *trans, int del_nr = 0; int del_slot = 0; int recow; - int ret; + int ret = 0; u64 ino = btrfs_ino(inode); path = btrfs_alloc_path(); @@ -1017,13 +586,24 @@ again: leaf = path->nodes[0]; btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); - BUG_ON(key.objectid != ino || key.type != BTRFS_EXTENT_DATA_KEY); + if (unlikely(key.objectid != ino || key.type != BTRFS_EXTENT_DATA_KEY)) { + ret = -EINVAL; + btrfs_abort_transaction(trans, ret); + goto out; + } fi = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_file_extent_item); - BUG_ON(btrfs_file_extent_type(leaf, fi) != - BTRFS_FILE_EXTENT_PREALLOC); + if (unlikely(btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_PREALLOC)) { + ret = -EINVAL; + btrfs_abort_transaction(trans, ret); + goto out; + } extent_end = key.offset + btrfs_file_extent_num_bytes(leaf, fi); - BUG_ON(key.offset > start || extent_end < end); + if (unlikely(key.offset > start || extent_end < end)) { + ret = -EINVAL; + btrfs_abort_transaction(trans, ret); + goto out; + } bytenr = btrfs_file_extent_disk_bytenr(leaf, fi); num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi); @@ -1037,7 +617,7 @@ again: ino, bytenr, orig_offset, &other_start, &other_end)) { new_key.offset = end; - btrfs_set_item_key_safe(root, path, &new_key); + btrfs_set_item_key_safe(trans, path, &new_key); fi = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_file_extent_item); btrfs_set_file_extent_generation(leaf, fi, @@ -1052,7 +632,6 @@ again: trans->transid); btrfs_set_file_extent_num_bytes(leaf, fi, end - other_start); - btrfs_mark_buffer_dirty(leaf); goto out; } } @@ -1071,7 +650,7 @@ again: trans->transid); path->slots[0]++; new_key.offset = start; - btrfs_set_item_key_safe(root, path, &new_key); + btrfs_set_item_key_safe(trans, path, &new_key); fi = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_file_extent_item); @@ -1081,7 +660,6 @@ again: other_end - start); btrfs_set_file_extent_offset(leaf, fi, start - orig_offset); - btrfs_mark_buffer_dirty(leaf); goto out; } } @@ -1096,8 +674,8 @@ again: btrfs_release_path(path); goto again; } - if (ret < 0) { - btrfs_abort_transaction(trans, root, ret); + if (unlikely(ret < 0)) { + btrfs_abort_transaction(trans, ret); goto out; } @@ -1115,17 +693,28 @@ again: btrfs_set_file_extent_offset(leaf, fi, split - orig_offset); btrfs_set_file_extent_num_bytes(leaf, fi, extent_end - split); - btrfs_mark_buffer_dirty(leaf); - ret = btrfs_inc_extent_ref(trans, root, bytenr, num_bytes, 0, - root->root_key.objectid, - ino, orig_offset, 0); - BUG_ON(ret); /* -ENOMEM */ + ref.action = BTRFS_ADD_DELAYED_REF; + ref.bytenr = bytenr; + ref.num_bytes = num_bytes; + ref.parent = 0; + ref.owning_root = btrfs_root_id(root); + ref.ref_root = btrfs_root_id(root); + btrfs_init_data_ref(&ref, ino, orig_offset, 0, false); + ret = btrfs_inc_extent_ref(trans, &ref); + if (unlikely(ret)) { + btrfs_abort_transaction(trans, ret); + goto out; + } if (split == start) { key.offset = start; } else { - BUG_ON(start != key.offset); + if (unlikely(start != key.offset)) { + ret = -EINVAL; + btrfs_abort_transaction(trans, ret); + goto out; + } path->slots[0]--; extent_end = end; } @@ -1134,6 +723,14 @@ again: other_start = end; other_end = 0; + + ref.action = BTRFS_DROP_DELAYED_REF; + ref.bytenr = bytenr; + ref.num_bytes = num_bytes; + ref.parent = 0; + ref.owning_root = btrfs_root_id(root); + ref.ref_root = btrfs_root_id(root); + btrfs_init_data_ref(&ref, ino, orig_offset, 0, false); if (extent_mergeable(leaf, path->slots[0] + 1, ino, bytenr, orig_offset, &other_start, &other_end)) { @@ -1144,10 +741,11 @@ again: extent_end = other_end; del_slot = path->slots[0] + 1; del_nr++; - ret = btrfs_free_extent(trans, root, bytenr, num_bytes, - 0, root->root_key.objectid, - ino, orig_offset, 0); - BUG_ON(ret); /* -ENOMEM */ + ret = btrfs_free_extent(trans, &ref); + if (unlikely(ret)) { + btrfs_abort_transaction(trans, ret); + goto out; + } } other_start = 0; other_end = start; @@ -1161,10 +759,11 @@ again: key.offset = other_start; del_slot = path->slots[0]; del_nr++; - ret = btrfs_free_extent(trans, root, bytenr, num_bytes, - 0, root->root_key.objectid, - ino, orig_offset, 0); - BUG_ON(ret); /* -ENOMEM */ + ret = btrfs_free_extent(trans, &ref); + if (unlikely(ret)) { + btrfs_abort_transaction(trans, ret); + goto out; + } } if (del_nr == 0) { fi = btrfs_item_ptr(leaf, path->slots[0], @@ -1172,7 +771,6 @@ again: btrfs_set_file_extent_type(leaf, fi, BTRFS_FILE_EXTENT_REG); btrfs_set_file_extent_generation(leaf, fi, trans->transid); - btrfs_mark_buffer_dirty(leaf); } else { fi = btrfs_item_ptr(leaf, del_slot - 1, struct btrfs_file_extent_item); @@ -1181,577 +779,770 @@ again: btrfs_set_file_extent_generation(leaf, fi, trans->transid); btrfs_set_file_extent_num_bytes(leaf, fi, extent_end - key.offset); - btrfs_mark_buffer_dirty(leaf); ret = btrfs_del_items(trans, root, path, del_slot, del_nr); - if (ret < 0) { - btrfs_abort_transaction(trans, root, ret); + if (unlikely(ret < 0)) { + btrfs_abort_transaction(trans, ret); goto out; } } out: - btrfs_free_path(path); + return ret; +} + +/* + * On error return an unlocked folio and the error value + * On success return a locked folio and 0 + */ +static int prepare_uptodate_folio(struct inode *inode, struct folio *folio, u64 pos, + u64 len) +{ + u64 clamp_start = max_t(u64, pos, folio_pos(folio)); + u64 clamp_end = min_t(u64, pos + len, folio_next_pos(folio)); + const u32 blocksize = inode_to_fs_info(inode)->sectorsize; + int ret = 0; + + if (folio_test_uptodate(folio)) + return 0; + + if (IS_ALIGNED(clamp_start, blocksize) && + IS_ALIGNED(clamp_end, blocksize)) + return 0; + + ret = btrfs_read_folio(NULL, folio); + if (ret) + return ret; + folio_lock(folio); + if (unlikely(!folio_test_uptodate(folio))) { + folio_unlock(folio); + return -EIO; + } + + /* + * Since btrfs_read_folio() will unlock the folio before it returns, + * there is a window where btrfs_release_folio() can be called to + * release the page. Here we check both inode mapping and page + * private to make sure the page was not released. + * + * The private flag check is essential for subpage as we need to store + * extra bitmap using folio private. + */ + if (folio->mapping != inode->i_mapping || !folio_test_private(folio)) { + folio_unlock(folio); + return -EAGAIN; + } return 0; } +static gfp_t get_prepare_gfp_flags(struct inode *inode, bool nowait) +{ + gfp_t gfp; + + gfp = btrfs_alloc_write_mask(inode->i_mapping); + if (nowait) { + gfp &= ~__GFP_DIRECT_RECLAIM; + gfp |= GFP_NOWAIT; + } + + return gfp; +} + /* - * on error we return an unlocked page and the error value - * on success we return a locked page and 0 + * Get folio into the page cache and lock it. */ -static int prepare_uptodate_page(struct page *page, u64 pos, - bool force_uptodate) +static noinline int prepare_one_folio(struct inode *inode, struct folio **folio_ret, + loff_t pos, size_t write_bytes, + bool nowait) { + const pgoff_t index = pos >> PAGE_SHIFT; + gfp_t mask = get_prepare_gfp_flags(inode, nowait); + fgf_t fgp_flags = (nowait ? FGP_WRITEBEGIN | FGP_NOWAIT : FGP_WRITEBEGIN) | + fgf_set_order(write_bytes); + struct folio *folio; int ret = 0; - if (((pos & (PAGE_CACHE_SIZE - 1)) || force_uptodate) && - !PageUptodate(page)) { - ret = btrfs_readpage(NULL, page); - if (ret) - return ret; - lock_page(page); - if (!PageUptodate(page)) { - unlock_page(page); - return -EIO; +again: + folio = __filemap_get_folio(inode->i_mapping, index, fgp_flags, mask); + if (IS_ERR(folio)) + return PTR_ERR(folio); + + ret = set_folio_extent_mapped(folio); + if (ret < 0) { + folio_unlock(folio); + folio_put(folio); + return ret; + } + ret = prepare_uptodate_folio(inode, folio, pos, write_bytes); + if (ret) { + /* The folio is already unlocked. */ + folio_put(folio); + if (!nowait && ret == -EAGAIN) { + ret = 0; + goto again; } + return ret; } + *folio_ret = folio; return 0; } /* - * this gets pages into the page cache and locks them down, it also properly - * waits for data=ordered extents to finish before allowing the pages to be - * modified. + * Locks the extent and properly waits for data=ordered extents to finish + * before allowing the folios to be modified if need. + * + * Return: + * 1 - the extent is locked + * 0 - the extent is not locked, and everything is OK + * -EAGAIN - need to prepare the folios again */ -static noinline int prepare_pages(struct btrfs_root *root, struct file *file, - struct page **pages, size_t num_pages, - loff_t pos, unsigned long first_index, - size_t write_bytes, bool force_uptodate) +static noinline int +lock_and_cleanup_extent_if_need(struct btrfs_inode *inode, struct folio *folio, + loff_t pos, size_t write_bytes, + u64 *lockstart, u64 *lockend, bool nowait, + struct extent_state **cached_state) { - struct extent_state *cached_state = NULL; - int i; - unsigned long index = pos >> PAGE_CACHE_SHIFT; - struct inode *inode = file_inode(file); - gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping); - int err = 0; - int faili = 0; + struct btrfs_fs_info *fs_info = inode->root->fs_info; u64 start_pos; u64 last_pos; + int ret = 0; - start_pos = pos & ~((u64)root->sectorsize - 1); - last_pos = ((u64)index + num_pages) << PAGE_CACHE_SHIFT; + start_pos = round_down(pos, fs_info->sectorsize); + last_pos = round_up(pos + write_bytes, fs_info->sectorsize) - 1; -again: - for (i = 0; i < num_pages; i++) { - pages[i] = find_or_create_page(inode->i_mapping, index + i, - mask | __GFP_WRITE); - if (!pages[i]) { - faili = i - 1; - err = -ENOMEM; - goto fail; - } - - if (i == 0) - err = prepare_uptodate_page(pages[i], pos, - force_uptodate); - if (i == num_pages - 1) - err = prepare_uptodate_page(pages[i], - pos + write_bytes, false); - if (err) { - page_cache_release(pages[i]); - faili = i - 1; - goto fail; - } - wait_on_page_writeback(pages[i]); - } - err = 0; - if (start_pos < inode->i_size) { + if (start_pos < inode->vfs_inode.i_size) { struct btrfs_ordered_extent *ordered; - lock_extent_bits(&BTRFS_I(inode)->io_tree, - start_pos, last_pos - 1, 0, &cached_state); - ordered = btrfs_lookup_first_ordered_extent(inode, - last_pos - 1); + + if (nowait) { + if (!btrfs_try_lock_extent(&inode->io_tree, start_pos, + last_pos, cached_state)) { + folio_unlock(folio); + folio_put(folio); + return -EAGAIN; + } + } else { + btrfs_lock_extent(&inode->io_tree, start_pos, last_pos, + cached_state); + } + + ordered = btrfs_lookup_ordered_range(inode, start_pos, + last_pos - start_pos + 1); if (ordered && - ordered->file_offset + ordered->len > start_pos && - ordered->file_offset < last_pos) { + ordered->file_offset + ordered->num_bytes > start_pos && + ordered->file_offset <= last_pos) { + btrfs_unlock_extent(&inode->io_tree, start_pos, last_pos, + cached_state); + folio_unlock(folio); + folio_put(folio); + btrfs_start_ordered_extent(ordered); btrfs_put_ordered_extent(ordered); - unlock_extent_cached(&BTRFS_I(inode)->io_tree, - start_pos, last_pos - 1, - &cached_state, GFP_NOFS); - for (i = 0; i < num_pages; i++) { - unlock_page(pages[i]); - page_cache_release(pages[i]); - } - btrfs_wait_ordered_range(inode, start_pos, - last_pos - start_pos); - goto again; + return -EAGAIN; } if (ordered) btrfs_put_ordered_extent(ordered); - clear_extent_bit(&BTRFS_I(inode)->io_tree, start_pos, - last_pos - 1, EXTENT_DIRTY | EXTENT_DELALLOC | - EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, - 0, 0, &cached_state, GFP_NOFS); - unlock_extent_cached(&BTRFS_I(inode)->io_tree, - start_pos, last_pos - 1, &cached_state, - GFP_NOFS); - } - for (i = 0; i < num_pages; i++) { - if (clear_page_dirty_for_io(pages[i])) - account_page_redirty(pages[i]); - set_page_extent_mapped(pages[i]); - WARN_ON(!PageLocked(pages[i])); - } - return 0; -fail: - while (faili >= 0) { - unlock_page(pages[faili]); - page_cache_release(pages[faili]); - faili--; + *lockstart = start_pos; + *lockend = last_pos; + ret = 1; } - return err; + /* + * We should be called after prepare_one_folio() which should have locked + * all pages in the range. + */ + WARN_ON(!folio_test_locked(folio)); + + return ret; } -static noinline int check_can_nocow(struct inode *inode, loff_t pos, - size_t *write_bytes) +/* + * Check if we can do nocow write into the range [@pos, @pos + @write_bytes) + * + * @pos: File offset. + * @write_bytes: The length to write, will be updated to the nocow writeable + * range. + * @nowait: Indicate if we can block or not (non-blocking IO context). + * + * This function will flush ordered extents in the range to ensure proper + * nocow checks. + * + * Return: + * > 0 If we can nocow, and updates @write_bytes. + * 0 If we can't do a nocow write. + * -EAGAIN If we can't do a nocow write because snapshotting of the inode's + * root is in progress or because we are in a non-blocking IO + * context and need to block (@nowait is true). + * < 0 If an error happened. + * + * NOTE: Callers need to call btrfs_check_nocow_unlock() if we return > 0. + */ +int btrfs_check_nocow_lock(struct btrfs_inode *inode, loff_t pos, + size_t *write_bytes, bool nowait) { - struct btrfs_trans_handle *trans; - struct btrfs_root *root = BTRFS_I(inode)->root; - struct btrfs_ordered_extent *ordered; + struct btrfs_fs_info *fs_info = inode->root->fs_info; + struct btrfs_root *root = inode->root; + struct extent_state *cached_state = NULL; u64 lockstart, lockend; - u64 num_bytes; - int ret; + u64 cur_offset; + int ret = 0; - lockstart = round_down(pos, root->sectorsize); - lockend = lockstart + round_up(*write_bytes, root->sectorsize) - 1; + if (!(inode->flags & (BTRFS_INODE_NODATACOW | BTRFS_INODE_PREALLOC))) + return 0; - while (1) { - lock_extent(&BTRFS_I(inode)->io_tree, lockstart, lockend); - ordered = btrfs_lookup_ordered_range(inode, lockstart, - lockend - lockstart + 1); - if (!ordered) { - break; + if (!btrfs_drew_try_write_lock(&root->snapshot_lock)) + return -EAGAIN; + + lockstart = round_down(pos, fs_info->sectorsize); + lockend = round_up(pos + *write_bytes, + fs_info->sectorsize) - 1; + + if (nowait) { + if (!btrfs_try_lock_ordered_range(inode, lockstart, lockend, + &cached_state)) { + btrfs_drew_write_unlock(&root->snapshot_lock); + return -EAGAIN; } - unlock_extent(&BTRFS_I(inode)->io_tree, lockstart, lockend); - btrfs_start_ordered_extent(inode, ordered, 1); - btrfs_put_ordered_extent(ordered); + } else { + btrfs_lock_and_flush_ordered_range(inode, lockstart, lockend, + &cached_state); } - trans = btrfs_join_transaction(root); - if (IS_ERR(trans)) { - unlock_extent(&BTRFS_I(inode)->io_tree, lockstart, lockend); - return PTR_ERR(trans); - } + cur_offset = lockstart; + while (cur_offset < lockend) { + u64 num_bytes = lockend - cur_offset + 1; - num_bytes = lockend - lockstart + 1; - ret = can_nocow_extent(trans, inode, lockstart, &num_bytes, NULL, NULL, - NULL); - btrfs_end_transaction(trans, root); - if (ret <= 0) { - ret = 0; - } else { - clear_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, lockend, - EXTENT_DIRTY | EXTENT_DELALLOC | - EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 0, 0, - NULL, GFP_NOFS); - *write_bytes = min_t(size_t, *write_bytes, num_bytes); + ret = can_nocow_extent(inode, cur_offset, &num_bytes, NULL, nowait); + if (ret <= 0) { + /* + * If cur_offset == lockstart it means we haven't found + * any extent against which we can NOCOW, so unlock the + * snapshot lock. + */ + if (cur_offset == lockstart) + btrfs_drew_write_unlock(&root->snapshot_lock); + break; + } + cur_offset += num_bytes; } - unlock_extent(&BTRFS_I(inode)->io_tree, lockstart, lockend); + btrfs_unlock_extent(&inode->io_tree, lockstart, lockend, &cached_state); + + /* + * cur_offset > lockstart means there's at least a partial range we can + * NOCOW, and that range can cover one or more extents. + */ + if (cur_offset > lockstart) { + *write_bytes = min_t(size_t, *write_bytes, cur_offset - pos); + return 1; + } return ret; } -static noinline ssize_t __btrfs_buffered_write(struct file *file, - struct iov_iter *i, - loff_t pos) +void btrfs_check_nocow_unlock(struct btrfs_inode *inode) { - struct inode *inode = file_inode(file); - struct btrfs_root *root = BTRFS_I(inode)->root; - struct page **pages = NULL; - u64 release_bytes = 0; - unsigned long first_index; - size_t num_written = 0; - int nrptrs; - int ret = 0; - bool only_release_metadata = false; - bool force_page_uptodate = false; - - nrptrs = min((iov_iter_count(i) + PAGE_CACHE_SIZE - 1) / - PAGE_CACHE_SIZE, PAGE_CACHE_SIZE / - (sizeof(struct page *))); - nrptrs = min(nrptrs, current->nr_dirtied_pause - current->nr_dirtied); - nrptrs = max(nrptrs, 8); - pages = kmalloc(nrptrs * sizeof(struct page *), GFP_KERNEL); - if (!pages) - return -ENOMEM; + btrfs_drew_write_unlock(&inode->root->snapshot_lock); +} - first_index = pos >> PAGE_CACHE_SHIFT; +int btrfs_write_check(struct kiocb *iocb, size_t count) +{ + struct file *file = iocb->ki_filp; + struct inode *inode = file_inode(file); + struct btrfs_fs_info *fs_info = inode_to_fs_info(inode); + loff_t pos = iocb->ki_pos; + int ret; + loff_t oldsize; - while (iov_iter_count(i) > 0) { - size_t offset = pos & (PAGE_CACHE_SIZE - 1); - size_t write_bytes = min(iov_iter_count(i), - nrptrs * (size_t)PAGE_CACHE_SIZE - - offset); - size_t num_pages = (write_bytes + offset + - PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT; - size_t reserve_bytes; - size_t dirty_pages; - size_t copied; + /* + * Quickly bail out on NOWAIT writes if we don't have the nodatacow or + * prealloc flags, as without those flags we always have to COW. We will + * later check if we can really COW into the target range (using + * can_nocow_extent() at btrfs_get_blocks_direct_write()). + */ + if ((iocb->ki_flags & IOCB_NOWAIT) && + !(BTRFS_I(inode)->flags & (BTRFS_INODE_NODATACOW | BTRFS_INODE_PREALLOC))) + return -EAGAIN; - WARN_ON(num_pages > nrptrs); + ret = file_remove_privs(file); + if (ret) + return ret; - /* - * Fault pages before locking them in prepare_pages - * to avoid recursive lock - */ - if (unlikely(iov_iter_fault_in_readable(i, write_bytes))) { - ret = -EFAULT; - break; - } + /* + * We reserve space for updating the inode when we reserve space for the + * extent we are going to write, so we will enospc out there. We don't + * need to start yet another transaction to update the inode as we will + * update the inode when we finish writing whatever data we write. + */ + if (!IS_NOCMTIME(inode)) { + inode_set_mtime_to_ts(inode, inode_set_ctime_current(inode)); + inode_inc_iversion(inode); + } - reserve_bytes = num_pages << PAGE_CACHE_SHIFT; - ret = btrfs_check_data_free_space(inode, reserve_bytes); - if (ret == -ENOSPC && - (BTRFS_I(inode)->flags & (BTRFS_INODE_NODATACOW | - BTRFS_INODE_PREALLOC))) { - ret = check_can_nocow(inode, pos, &write_bytes); - if (ret > 0) { - only_release_metadata = true; - /* - * our prealloc extent may be smaller than - * write_bytes, so scale down. - */ - num_pages = (write_bytes + offset + - PAGE_CACHE_SIZE - 1) >> - PAGE_CACHE_SHIFT; - reserve_bytes = num_pages << PAGE_CACHE_SHIFT; - ret = 0; - } else { - ret = -ENOSPC; - } - } + oldsize = i_size_read(inode); + if (pos > oldsize) { + /* Expand hole size to cover write data, preventing empty gap */ + loff_t end_pos = round_up(pos + count, fs_info->sectorsize); + ret = btrfs_cont_expand(BTRFS_I(inode), oldsize, end_pos); if (ret) - break; + return ret; + } - ret = btrfs_delalloc_reserve_metadata(inode, reserve_bytes); - if (ret) { - if (!only_release_metadata) - btrfs_free_reserved_data_space(inode, - reserve_bytes); - break; - } + return 0; +} + +static void release_space(struct btrfs_inode *inode, struct extent_changeset *data_reserved, + u64 start, u64 len, bool only_release_metadata) +{ + if (len == 0) + return; - release_bytes = reserve_bytes; + if (only_release_metadata) { + btrfs_check_nocow_unlock(inode); + btrfs_delalloc_release_metadata(inode, len, true); + } else { + const struct btrfs_fs_info *fs_info = inode->root->fs_info; + + btrfs_delalloc_release_space(inode, data_reserved, + round_down(start, fs_info->sectorsize), + len, true); + } +} + +/* + * Reserve data and metadata space for this buffered write range. + * + * Return >0 for the number of bytes reserved, which is always block aligned. + * Return <0 for error. + */ +static ssize_t reserve_space(struct btrfs_inode *inode, + struct extent_changeset **data_reserved, + u64 start, size_t *len, bool nowait, + bool *only_release_metadata) +{ + const struct btrfs_fs_info *fs_info = inode->root->fs_info; + const unsigned int block_offset = (start & (fs_info->sectorsize - 1)); + size_t reserve_bytes; + int ret; + + ret = btrfs_check_data_free_space(inode, data_reserved, start, *len, nowait); + if (ret < 0) { + int can_nocow; + + if (nowait && (ret == -ENOSPC || ret == -EAGAIN)) + return -EAGAIN; /* - * This is going to setup the pages array with the number of - * pages we want, so we don't really need to worry about the - * contents of pages from loop to loop + * If we don't have to COW at the offset, reserve metadata only. + * write_bytes may get smaller than requested here. */ - ret = prepare_pages(root, file, pages, num_pages, - pos, first_index, write_bytes, - force_page_uptodate); + can_nocow = btrfs_check_nocow_lock(inode, start, len, nowait); + if (can_nocow < 0) + ret = can_nocow; + if (can_nocow > 0) + ret = 0; if (ret) - break; + return ret; + *only_release_metadata = true; + } - copied = btrfs_copy_from_user(pos, num_pages, - write_bytes, pages, i); + reserve_bytes = round_up(*len + block_offset, fs_info->sectorsize); + WARN_ON(reserve_bytes == 0); + ret = btrfs_delalloc_reserve_metadata(inode, reserve_bytes, + reserve_bytes, nowait); + if (ret) { + if (!*only_release_metadata) + btrfs_free_reserved_data_space(inode, *data_reserved, + start, *len); + else + btrfs_check_nocow_unlock(inode); - /* - * if we have trouble faulting in the pages, fall - * back to one page at a time - */ - if (copied < write_bytes) - nrptrs = 1; + if (nowait && ret == -ENOSPC) + ret = -EAGAIN; + return ret; + } + return reserve_bytes; +} - if (copied == 0) { - force_page_uptodate = true; - dirty_pages = 0; - } else { - force_page_uptodate = false; - dirty_pages = (copied + offset + - PAGE_CACHE_SIZE - 1) >> - PAGE_CACHE_SHIFT; - } +/* Shrink the reserved data and metadata space from @reserved_len to @new_len. */ +static void shrink_reserved_space(struct btrfs_inode *inode, + struct extent_changeset *data_reserved, + u64 reserved_start, u64 reserved_len, + u64 new_len, bool only_release_metadata) +{ + const u64 diff = reserved_len - new_len; - /* - * If we had a short copy we need to release the excess delaloc - * bytes we reserved. We need to increment outstanding_extents - * because btrfs_delalloc_release_space will decrement it, but - * we still have an outstanding extent for the chunk we actually - * managed to copy. - */ - if (num_pages > dirty_pages) { - release_bytes = (num_pages - dirty_pages) << - PAGE_CACHE_SHIFT; - if (copied > 0) { - spin_lock(&BTRFS_I(inode)->lock); - BTRFS_I(inode)->outstanding_extents++; - spin_unlock(&BTRFS_I(inode)->lock); - } - if (only_release_metadata) - btrfs_delalloc_release_metadata(inode, - release_bytes); - else - btrfs_delalloc_release_space(inode, - release_bytes); - } + ASSERT(new_len <= reserved_len); + btrfs_delalloc_shrink_extents(inode, reserved_len, new_len); + if (only_release_metadata) + btrfs_delalloc_release_metadata(inode, diff, true); + else + btrfs_delalloc_release_space(inode, data_reserved, + reserved_start + new_len, diff, true); +} - release_bytes = dirty_pages << PAGE_CACHE_SHIFT; - if (copied > 0) { - ret = btrfs_dirty_pages(root, inode, pages, - dirty_pages, pos, copied, - NULL); - if (ret) { - btrfs_drop_pages(pages, num_pages); - break; - } - } +/* Calculate the maximum amount of bytes we can write into one folio. */ +static size_t calc_write_bytes(const struct btrfs_inode *inode, + const struct iov_iter *iter, u64 start) +{ + const size_t max_folio_size = mapping_max_folio_size(inode->vfs_inode.i_mapping); + + return min(max_folio_size - (start & (max_folio_size - 1)), + iov_iter_count(iter)); +} + +/* + * Do the heavy-lifting work to copy one range into one folio of the page cache. + * + * Return > 0 in case we copied all bytes or just some of them. + * Return 0 if no bytes were copied, in which case the caller should retry. + * Return <0 on error. + */ +static int copy_one_range(struct btrfs_inode *inode, struct iov_iter *iter, + struct extent_changeset **data_reserved, u64 start, + bool nowait) +{ + struct btrfs_fs_info *fs_info = inode->root->fs_info; + struct extent_state *cached_state = NULL; + size_t write_bytes = calc_write_bytes(inode, iter, start); + size_t copied; + const u64 reserved_start = round_down(start, fs_info->sectorsize); + u64 reserved_len; + struct folio *folio = NULL; + int extents_locked; + u64 lockstart; + u64 lockend; + bool only_release_metadata = false; + const unsigned int bdp_flags = (nowait ? BDP_ASYNC : 0); + int ret; + + /* + * Fault all pages before locking them in prepare_one_folio() to avoid + * recursive lock. + */ + if (unlikely(fault_in_iov_iter_readable(iter, write_bytes))) + return -EFAULT; + extent_changeset_release(*data_reserved); + ret = reserve_space(inode, data_reserved, start, &write_bytes, nowait, + &only_release_metadata); + if (ret < 0) + return ret; + reserved_len = ret; + /* Write range must be inside the reserved range. */ + ASSERT(reserved_start <= start); + ASSERT(start + write_bytes <= reserved_start + reserved_len); + +again: + ret = balance_dirty_pages_ratelimited_flags(inode->vfs_inode.i_mapping, + bdp_flags); + if (ret) { + btrfs_delalloc_release_extents(inode, reserved_len); + release_space(inode, *data_reserved, reserved_start, reserved_len, + only_release_metadata); + return ret; + } - release_bytes = 0; - btrfs_drop_pages(pages, num_pages); + ret = prepare_one_folio(&inode->vfs_inode, &folio, start, write_bytes, false); + if (ret) { + btrfs_delalloc_release_extents(inode, reserved_len); + release_space(inode, *data_reserved, reserved_start, reserved_len, + only_release_metadata); + return ret; + } - if (only_release_metadata && copied > 0) { - u64 lockstart = round_down(pos, root->sectorsize); - u64 lockend = lockstart + - (dirty_pages << PAGE_CACHE_SHIFT) - 1; + /* + * The reserved range goes beyond the current folio, shrink the reserved + * space to the folio boundary. + */ + if (reserved_start + reserved_len > folio_next_pos(folio)) { + const u64 last_block = folio_next_pos(folio); + + shrink_reserved_space(inode, *data_reserved, reserved_start, + reserved_len, last_block - reserved_start, + only_release_metadata); + write_bytes = last_block - start; + reserved_len = last_block - reserved_start; + } + + extents_locked = lock_and_cleanup_extent_if_need(inode, folio, start, + write_bytes, &lockstart, + &lockend, nowait, + &cached_state); + if (extents_locked < 0) { + if (!nowait && extents_locked == -EAGAIN) + goto again; + + btrfs_delalloc_release_extents(inode, reserved_len); + release_space(inode, *data_reserved, reserved_start, reserved_len, + only_release_metadata); + ret = extents_locked; + return ret; + } - set_extent_bit(&BTRFS_I(inode)->io_tree, lockstart, - lockend, EXTENT_NORESERVE, NULL, - NULL, GFP_NOFS); - only_release_metadata = false; + copied = copy_folio_from_iter_atomic(folio, offset_in_folio(folio, start), + write_bytes, iter); + flush_dcache_folio(folio); + + if (unlikely(copied < write_bytes)) { + u64 last_block; + + /* + * The original write range doesn't need an uptodate folio as + * the range is block aligned. But now a short copy happened. + * We cannot handle it without an uptodate folio. + * + * So just revert the range and we will retry. + */ + if (!folio_test_uptodate(folio)) { + iov_iter_revert(iter, copied); + copied = 0; } - cond_resched(); + /* No copied bytes, unlock, release reserved space and exit. */ + if (copied == 0) { + if (extents_locked) + btrfs_unlock_extent(&inode->io_tree, lockstart, lockend, + &cached_state); + else + btrfs_free_extent_state(cached_state); + btrfs_delalloc_release_extents(inode, reserved_len); + release_space(inode, *data_reserved, reserved_start, reserved_len, + only_release_metadata); + btrfs_drop_folio(fs_info, folio, start, copied); + return 0; + } - balance_dirty_pages_ratelimited(inode->i_mapping); - if (dirty_pages < (root->leafsize >> PAGE_CACHE_SHIFT) + 1) - btrfs_btree_balance_dirty(root); + /* Release the reserved space beyond the last block. */ + last_block = round_up(start + copied, fs_info->sectorsize); - pos += copied; - num_written += copied; + shrink_reserved_space(inode, *data_reserved, reserved_start, + reserved_len, last_block - reserved_start, + only_release_metadata); + reserved_len = last_block - reserved_start; } - kfree(pages); + ret = btrfs_dirty_folio(inode, folio, start, copied, &cached_state, + only_release_metadata); + /* + * If we have not locked the extent range, because the range's start + * offset is >= i_size, we might still have a non-NULL cached extent + * state, acquired while marking the extent range as delalloc through + * btrfs_dirty_page(). Therefore free any possible cached extent state + * to avoid a memory leak. + */ + if (extents_locked) + btrfs_unlock_extent(&inode->io_tree, lockstart, lockend, &cached_state); + else + btrfs_free_extent_state(cached_state); - if (release_bytes) { - if (only_release_metadata) - btrfs_delalloc_release_metadata(inode, release_bytes); - else - btrfs_delalloc_release_space(inode, release_bytes); + btrfs_delalloc_release_extents(inode, reserved_len); + if (ret) { + btrfs_drop_folio(fs_info, folio, start, copied); + release_space(inode, *data_reserved, reserved_start, reserved_len, + only_release_metadata); + return ret; } + if (only_release_metadata) + btrfs_check_nocow_unlock(inode); - return num_written ? num_written : ret; + btrfs_drop_folio(fs_info, folio, start, copied); + return copied; } -static ssize_t __btrfs_direct_write(struct kiocb *iocb, - const struct iovec *iov, - unsigned long nr_segs, loff_t pos, - loff_t *ppos, size_t count, size_t ocount) +ssize_t btrfs_buffered_write(struct kiocb *iocb, struct iov_iter *iter) { struct file *file = iocb->ki_filp; - struct iov_iter i; - ssize_t written; - ssize_t written_buffered; - loff_t endbyte; - int err; - - written = generic_file_direct_write(iocb, iov, &nr_segs, pos, ppos, - count, ocount); - - if (written < 0 || written == count) - return written; - - pos += written; - count -= written; - iov_iter_init(&i, iov, nr_segs, count, written); - written_buffered = __btrfs_buffered_write(file, &i, pos); - if (written_buffered < 0) { - err = written_buffered; - goto out; - } - endbyte = pos + written_buffered - 1; - err = filemap_write_and_wait_range(file->f_mapping, pos, endbyte); - if (err) - goto out; - written += written_buffered; - *ppos = pos + written_buffered; - invalidate_mapping_pages(file->f_mapping, pos >> PAGE_CACHE_SHIFT, - endbyte >> PAGE_CACHE_SHIFT); -out: - return written ? written : err; -} + loff_t pos; + struct inode *inode = file_inode(file); + struct extent_changeset *data_reserved = NULL; + size_t num_written = 0; + ssize_t ret; + loff_t old_isize; + unsigned int ilock_flags = 0; + const bool nowait = (iocb->ki_flags & IOCB_NOWAIT); -static void update_time_for_write(struct inode *inode) -{ - struct timespec now; + if (nowait) + ilock_flags |= BTRFS_ILOCK_TRY; - if (IS_NOCMTIME(inode)) - return; + ret = btrfs_inode_lock(BTRFS_I(inode), ilock_flags); + if (ret < 0) + return ret; + + /* + * We can only trust the isize with inode lock held, or it can race with + * other buffered writes and cause incorrect call of + * pagecache_isize_extended() to overwrite existing data. + */ + old_isize = i_size_read(inode); - now = current_fs_time(inode->i_sb); - if (!timespec_equal(&inode->i_mtime, &now)) - inode->i_mtime = now; + ret = generic_write_checks(iocb, iter); + if (ret <= 0) + goto out; - if (!timespec_equal(&inode->i_ctime, &now)) - inode->i_ctime = now; + ret = btrfs_write_check(iocb, ret); + if (ret < 0) + goto out; - if (IS_I_VERSION(inode)) - inode_inc_iversion(inode); + pos = iocb->ki_pos; + while (iov_iter_count(iter) > 0) { + ret = copy_one_range(BTRFS_I(inode), iter, &data_reserved, pos, nowait); + if (ret < 0) + break; + pos += ret; + num_written += ret; + cond_resched(); + } + + extent_changeset_free(data_reserved); + if (num_written > 0) { + pagecache_isize_extended(inode, old_isize, iocb->ki_pos); + iocb->ki_pos += num_written; + } +out: + btrfs_inode_unlock(BTRFS_I(inode), ilock_flags); + return num_written ? num_written : ret; } -static ssize_t btrfs_file_aio_write(struct kiocb *iocb, - const struct iovec *iov, - unsigned long nr_segs, loff_t pos) +static ssize_t btrfs_encoded_write(struct kiocb *iocb, struct iov_iter *from, + const struct btrfs_ioctl_encoded_io_args *encoded) { struct file *file = iocb->ki_filp; struct inode *inode = file_inode(file); - struct btrfs_root *root = BTRFS_I(inode)->root; - loff_t *ppos = &iocb->ki_pos; - u64 start_pos; - ssize_t num_written = 0; - ssize_t err = 0; - size_t count, ocount; - bool sync = (file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host); + loff_t count; + ssize_t ret; - mutex_lock(&inode->i_mutex); - - err = generic_segment_checks(iov, &nr_segs, &ocount, VERIFY_READ); - if (err) { - mutex_unlock(&inode->i_mutex); - goto out; + btrfs_inode_lock(BTRFS_I(inode), 0); + count = encoded->len; + ret = generic_write_checks_count(iocb, &count); + if (ret == 0 && count != encoded->len) { + /* + * The write got truncated by generic_write_checks_count(). We + * can't do a partial encoded write. + */ + ret = -EFBIG; } - count = ocount; - - current->backing_dev_info = inode->i_mapping->backing_dev_info; - err = generic_write_checks(file, &pos, &count, S_ISBLK(inode->i_mode)); - if (err) { - mutex_unlock(&inode->i_mutex); + if (ret || encoded->len == 0) goto out; - } - if (count == 0) { - mutex_unlock(&inode->i_mutex); + ret = btrfs_write_check(iocb, encoded->len); + if (ret < 0) goto out; - } - err = file_remove_suid(file); - if (err) { - mutex_unlock(&inode->i_mutex); - goto out; - } + ret = btrfs_do_encoded_write(iocb, from, encoded); +out: + btrfs_inode_unlock(BTRFS_I(inode), 0); + return ret; +} +ssize_t btrfs_do_write_iter(struct kiocb *iocb, struct iov_iter *from, + const struct btrfs_ioctl_encoded_io_args *encoded) +{ + struct file *file = iocb->ki_filp; + struct btrfs_inode *inode = BTRFS_I(file_inode(file)); + ssize_t num_written, num_sync; + + if (unlikely(btrfs_is_shutdown(inode->root->fs_info))) + return -EIO; /* - * If BTRFS flips readonly due to some impossible error - * (fs_info->fs_state now has BTRFS_SUPER_FLAG_ERROR), - * although we have opened a file as writable, we have - * to stop this write operation to ensure FS consistency. + * If the fs flips readonly due to some impossible error, although we + * have opened a file as writable, we have to stop this write operation + * to ensure consistency. */ - if (test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state)) { - mutex_unlock(&inode->i_mutex); - err = -EROFS; - goto out; + if (BTRFS_FS_ERROR(inode->root->fs_info)) + return -EROFS; + + if (encoded && (iocb->ki_flags & IOCB_NOWAIT)) + return -EOPNOTSUPP; + + if (encoded) { + num_written = btrfs_encoded_write(iocb, from, encoded); + num_sync = encoded->len; + } else if (iocb->ki_flags & IOCB_DIRECT) { + num_written = btrfs_direct_write(iocb, from); + num_sync = num_written; + } else { + num_written = btrfs_buffered_write(iocb, from); + num_sync = num_written; } - /* - * We reserve space for updating the inode when we reserve space for the - * extent we are going to write, so we will enospc out there. We don't - * need to start yet another transaction to update the inode as we will - * update the inode when we finish writing whatever data we write. - */ - update_time_for_write(inode); + btrfs_set_inode_last_sub_trans(inode); - start_pos = round_down(pos, root->sectorsize); - if (start_pos > i_size_read(inode)) { - err = btrfs_cont_expand(inode, i_size_read(inode), start_pos); - if (err) { - mutex_unlock(&inode->i_mutex); - goto out; - } + if (num_sync > 0) { + num_sync = generic_write_sync(iocb, num_sync); + if (num_sync < 0) + num_written = num_sync; } - if (sync) - atomic_inc(&BTRFS_I(inode)->sync_writers); + return num_written; +} - if (unlikely(file->f_flags & O_DIRECT)) { - num_written = __btrfs_direct_write(iocb, iov, nr_segs, - pos, ppos, count, ocount); - } else { - struct iov_iter i; +static ssize_t btrfs_file_write_iter(struct kiocb *iocb, struct iov_iter *from) +{ + return btrfs_do_write_iter(iocb, from, NULL); +} - iov_iter_init(&i, iov, nr_segs, count, num_written); +int btrfs_release_file(struct inode *inode, struct file *filp) +{ + struct btrfs_file_private *private = filp->private_data; - num_written = __btrfs_buffered_write(file, &i, pos); - if (num_written > 0) - *ppos = pos + num_written; + if (private) { + kfree(private->filldir_buf); + btrfs_free_extent_state(private->llseek_cached_state); + kfree(private); + filp->private_data = NULL; } - mutex_unlock(&inode->i_mutex); + /* + * Set by setattr when we are about to truncate a file from a non-zero + * size to a zero size. This tries to flush down new bytes that may + * have been written if the application were using truncate to replace + * a file in place. + */ + if (test_and_clear_bit(BTRFS_INODE_FLUSH_ON_CLOSE, + &BTRFS_I(inode)->runtime_flags)) + filemap_flush(inode->i_mapping); + return 0; +} + +static int start_ordered_ops(struct btrfs_inode *inode, loff_t start, loff_t end) +{ + int ret; + struct blk_plug plug; /* - * we want to make sure fsync finds this change - * but we haven't joined a transaction running right now. - * - * Later on, someone is sure to update the inode and get the - * real transid recorded. - * - * We set last_trans now to the fs_info generation + 1, - * this will either be one more than the running transaction - * or the generation used for the next transaction if there isn't - * one running right now. - * - * We also have to set last_sub_trans to the current log transid, - * otherwise subsequent syncs to a file that's been synced in this - * transaction will appear to have already occured. + * This is only called in fsync, which would do synchronous writes, so + * a plug can merge adjacent IOs as much as possible. Esp. in case of + * multiple disks using raid profile, a large IO can be split to + * several segments of stripe length (currently 64K). */ - BTRFS_I(inode)->last_trans = root->fs_info->generation + 1; - BTRFS_I(inode)->last_sub_trans = root->log_transid; - if (num_written > 0 || num_written == -EIOCBQUEUED) { - err = generic_write_sync(file, pos, num_written); - if (err < 0 && num_written > 0) - num_written = err; - } + blk_start_plug(&plug); + ret = btrfs_fdatawrite_range(inode, start, end); + blk_finish_plug(&plug); - if (sync) - atomic_dec(&BTRFS_I(inode)->sync_writers); -out: - current->backing_dev_info = NULL; - return num_written ? num_written : err; + return ret; } -int btrfs_release_file(struct inode *inode, struct file *filp) +static inline bool skip_inode_logging(const struct btrfs_log_ctx *ctx) { + struct btrfs_inode *inode = ctx->inode; + struct btrfs_fs_info *fs_info = inode->root->fs_info; + + if (btrfs_inode_in_log(inode, btrfs_get_fs_generation(fs_info)) && + list_empty(&ctx->ordered_extents)) + return true; + /* - * ordered_data_close is set by settattr when we are about to truncate - * a file from a non-zero size to a zero size. This tries to - * flush down new bytes that may have been written if the - * application were using truncate to replace a file in place. + * If we are doing a fast fsync we can not bail out if the inode's + * last_trans is <= then the last committed transaction, because we only + * update the last_trans of the inode during ordered extent completion, + * and for a fast fsync we don't wait for that, we only wait for the + * writeback to complete. */ - if (test_and_clear_bit(BTRFS_INODE_ORDERED_DATA_CLOSE, - &BTRFS_I(inode)->runtime_flags)) { - struct btrfs_trans_handle *trans; - struct btrfs_root *root = BTRFS_I(inode)->root; + if (inode->last_trans <= btrfs_get_last_trans_committed(fs_info) && + (test_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &inode->runtime_flags) || + list_empty(&ctx->ordered_extents))) + return true; - /* - * We need to block on a committing transaction to keep us from - * throwing a ordered operation on to the list and causing - * something like sync to deadlock trying to flush out this - * inode. - */ - trans = btrfs_start_transaction(root, 0); - if (IS_ERR(trans)) - return PTR_ERR(trans); - btrfs_add_ordered_operation(trans, BTRFS_I(inode)->root, inode); - btrfs_end_transaction(trans, root); - if (inode->i_size > BTRFS_ORDERED_OPERATIONS_FLUSH_LIMIT) - filemap_flush(inode->i_mapping); - } - if (filp->private_data) - btrfs_ioctl_trans_end(filp); - return 0; + return false; } /* @@ -1767,91 +1558,192 @@ int btrfs_release_file(struct inode *inode, struct file *filp) */ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) { - struct dentry *dentry = file->f_path.dentry; - struct inode *inode = dentry->d_inode; - struct btrfs_root *root = BTRFS_I(inode)->root; - int ret = 0; + struct dentry *dentry = file_dentry(file); + struct btrfs_inode *inode = BTRFS_I(d_inode(dentry)); + struct btrfs_root *root = inode->root; + struct btrfs_fs_info *fs_info = root->fs_info; struct btrfs_trans_handle *trans; - bool full_sync = 0; + struct btrfs_log_ctx ctx; + int ret = 0, err; + u64 len; + bool full_sync; + bool skip_ilock = false; + + if (current->journal_info == BTRFS_TRANS_DIO_WRITE_STUB) { + skip_ilock = true; + current->journal_info = NULL; + btrfs_assert_inode_locked(inode); + } trace_btrfs_sync_file(file, datasync); + btrfs_init_log_ctx(&ctx, inode); + + /* + * Always set the range to a full range, otherwise we can get into + * several problems, from missing file extent items to represent holes + * when not using the NO_HOLES feature, to log tree corruption due to + * races between hole detection during logging and completion of ordered + * extents outside the range, to missing checksums due to ordered extents + * for which we flushed only a subset of their pages. + */ + start = 0; + end = LLONG_MAX; + len = (u64)LLONG_MAX + 1; + /* * We write the dirty pages in the range and wait until they complete * out of the ->i_mutex. If so, we can flush the dirty pages by * multi-task, and make the performance up. See * btrfs_wait_ordered_range for an explanation of the ASYNC check. */ - atomic_inc(&BTRFS_I(inode)->sync_writers); - ret = filemap_fdatawrite_range(inode->i_mapping, start, end); - if (!ret && test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT, - &BTRFS_I(inode)->runtime_flags)) - ret = filemap_fdatawrite_range(inode->i_mapping, start, end); - atomic_dec(&BTRFS_I(inode)->sync_writers); + ret = start_ordered_ops(inode, start, end); if (ret) - return ret; + goto out; - mutex_lock(&inode->i_mutex); + if (skip_ilock) + down_write(&inode->i_mmap_lock); + else + btrfs_inode_lock(inode, BTRFS_ILOCK_MMAP); - /* - * We flush the dirty pages again to avoid some dirty pages in the - * range being left. - */ - atomic_inc(&root->log_batch); - full_sync = test_bit(BTRFS_INODE_NEEDS_FULL_SYNC, - &BTRFS_I(inode)->runtime_flags); - if (full_sync) - btrfs_wait_ordered_range(inode, start, end - start + 1); atomic_inc(&root->log_batch); /* - * check the transaction that last modified this inode - * and see if its already been committed + * Before we acquired the inode's lock and the mmap lock, someone may + * have dirtied more pages in the target range. We need to make sure + * that writeback for any such pages does not start while we are logging + * the inode, because if it does, any of the following might happen when + * we are not doing a full inode sync: + * + * 1) We log an extent after its writeback finishes but before its + * checksums are added to the csum tree, leading to -EIO errors + * when attempting to read the extent after a log replay. + * + * 2) We can end up logging an extent before its writeback finishes. + * Therefore after the log replay we will have a file extent item + * pointing to an unwritten extent (and no data checksums as well). + * + * So trigger writeback for any eventual new dirty pages and then we + * wait for all ordered extents to complete below. */ - if (!BTRFS_I(inode)->last_trans) { - mutex_unlock(&inode->i_mutex); + ret = start_ordered_ops(inode, start, end); + if (ret) { + if (skip_ilock) + up_write(&inode->i_mmap_lock); + else + btrfs_inode_unlock(inode, BTRFS_ILOCK_MMAP); goto out; } /* - * if the last transaction that changed this file was before - * the current transaction, we can bail out now without any - * syncing + * Always check for the full sync flag while holding the inode's lock, + * to avoid races with other tasks. The flag must be either set all the + * time during logging or always off all the time while logging. + * We check the flag here after starting delalloc above, because when + * running delalloc the full sync flag may be set if we need to drop + * extra extent map ranges due to temporary memory allocation failures. + */ + full_sync = test_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &inode->runtime_flags); + + /* + * We have to do this here to avoid the priority inversion of waiting on + * IO of a lower priority task while holding a transaction open. + * + * For a full fsync we wait for the ordered extents to complete while + * for a fast fsync we wait just for writeback to complete, and then + * attach the ordered extents to the transaction so that a transaction + * commit waits for their completion, to avoid data loss if we fsync, + * the current transaction commits before the ordered extents complete + * and a power failure happens right after that. + * + * For zoned filesystem, if a write IO uses a ZONE_APPEND command, the + * logical address recorded in the ordered extent may change. We need + * to wait for the IO to stabilize the logical address. */ - smp_mb(); - if (btrfs_inode_in_log(inode, root->fs_info->generation) || - BTRFS_I(inode)->last_trans <= - root->fs_info->last_trans_committed) { - BTRFS_I(inode)->last_trans = 0; + if (full_sync || btrfs_is_zoned(fs_info)) { + ret = btrfs_wait_ordered_range(inode, start, len); + clear_bit(BTRFS_INODE_COW_WRITE_ERROR, &inode->runtime_flags); + } else { + /* + * Get our ordered extents as soon as possible to avoid doing + * checksum lookups in the csum tree, and use instead the + * checksums attached to the ordered extents. + */ + btrfs_get_ordered_extents_for_logging(inode, &ctx.ordered_extents); + ret = filemap_fdatawait_range(inode->vfs_inode.i_mapping, start, end); + if (ret) + goto out_release_extents; + + /* + * Check and clear the BTRFS_INODE_COW_WRITE_ERROR now after + * starting and waiting for writeback, because for buffered IO + * it may have been set during the end IO callback + * (end_bbio_data_write() -> btrfs_finish_ordered_extent()) in + * case an error happened and we need to wait for ordered + * extents to complete so that any extent maps that point to + * unwritten locations are dropped and we don't log them. + */ + if (test_and_clear_bit(BTRFS_INODE_COW_WRITE_ERROR, &inode->runtime_flags)) + ret = btrfs_wait_ordered_range(inode, start, len); + } + + if (ret) + goto out_release_extents; + atomic_inc(&root->log_batch); + + if (skip_inode_logging(&ctx)) { /* - * We'v had everything committed since the last time we were + * We've had everything committed since the last time we were * modified so clear this flag in case it was set for whatever * reason, it's no longer relevant. */ - clear_bit(BTRFS_INODE_NEEDS_FULL_SYNC, - &BTRFS_I(inode)->runtime_flags); - mutex_unlock(&inode->i_mutex); - goto out; + clear_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &inode->runtime_flags); + /* + * An ordered extent might have started before and completed + * already with io errors, in which case the inode was not + * updated and we end up here. So check the inode's mapping + * for any errors that might have happened since we last + * checked called fsync. + */ + ret = filemap_check_wb_err(inode->vfs_inode.i_mapping, file->f_wb_err); + goto out_release_extents; } + btrfs_init_log_ctx_scratch_eb(&ctx); + /* - * ok we haven't committed the transaction yet, lets do a commit + * We use start here because we will need to wait on the IO to complete + * in btrfs_sync_log, which could require joining a transaction (for + * example checking cross references in the nocow path). If we use join + * here we could get into a situation where we're waiting on IO to + * happen that is blocked on a transaction trying to commit. With start + * we inc the extwriter counter, so we wait for all extwriters to exit + * before we start blocking joiners. This comment is to keep somebody + * from thinking they are super smart and changing this to + * btrfs_join_transaction *cough*Josef*cough*. */ - if (file->private_data) - btrfs_ioctl_trans_end(file); - trans = btrfs_start_transaction(root, 0); if (IS_ERR(trans)) { ret = PTR_ERR(trans); - mutex_unlock(&inode->i_mutex); - goto out; + goto out_release_extents; } + trans->in_fsync = true; - ret = btrfs_log_dentry_safe(trans, root, dentry); + ret = btrfs_log_dentry_safe(trans, dentry, &ctx); + /* + * Scratch eb no longer needed, release before syncing log or commit + * transaction, to avoid holding unnecessary memory during such long + * operations. + */ + if (ctx.scratch_eb) { + free_extent_buffer(ctx.scratch_eb); + ctx.scratch_eb = NULL; + } + btrfs_release_log_ctx_extents(&ctx); if (ret < 0) { - mutex_unlock(&inode->i_mutex); - goto out; + /* Fallthrough and commit/free transaction. */ + ret = BTRFS_LOG_FORCE_COMMIT; } /* we've logged all the items and now have a consistent @@ -1864,108 +1756,366 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync) * file again, but that will end up using the synchronization * inside btrfs_sync_log to keep things safe. */ - mutex_unlock(&inode->i_mutex); + if (skip_ilock) + up_write(&inode->i_mmap_lock); + else + btrfs_inode_unlock(inode, BTRFS_ILOCK_MMAP); + + if (ret == BTRFS_NO_LOG_SYNC) { + ret = btrfs_end_transaction(trans); + goto out; + } + + /* We successfully logged the inode, attempt to sync the log. */ + if (!ret) { + ret = btrfs_sync_log(trans, root, &ctx); + if (!ret) { + ret = btrfs_end_transaction(trans); + goto out; + } + } + + /* + * At this point we need to commit the transaction because we had + * btrfs_need_log_full_commit() or some other error. + * + * If we didn't do a full sync we have to stop the trans handle, wait on + * the ordered extents, start it again and commit the transaction. If + * we attempt to wait on the ordered extents here we could deadlock with + * something like fallocate() that is holding the extent lock trying to + * start a transaction while some other thread is trying to commit the + * transaction while we (fsync) are currently holding the transaction + * open. + */ + if (!full_sync) { + ret = btrfs_end_transaction(trans); + if (ret) + goto out; + ret = btrfs_wait_ordered_range(inode, start, len); + if (ret) + goto out; + + /* + * This is safe to use here because we're only interested in + * making sure the transaction that had the ordered extents is + * committed. We aren't waiting on anything past this point, + * we're purely getting the transaction and committing it. + */ + trans = btrfs_attach_transaction_barrier(root); + if (IS_ERR(trans)) { + ret = PTR_ERR(trans); - if (ret != BTRFS_NO_LOG_SYNC) { - if (ret > 0) { /* - * If we didn't already wait for ordered extents we need - * to do that now. + * We committed the transaction and there's no currently + * running transaction, this means everything we care + * about made it to disk and we are done. */ - if (!full_sync) - btrfs_wait_ordered_range(inode, start, - end - start + 1); - ret = btrfs_commit_transaction(trans, root); - } else { - ret = btrfs_sync_log(trans, root); - if (ret == 0) { - ret = btrfs_end_transaction(trans, root); - } else { - if (!full_sync) - btrfs_wait_ordered_range(inode, start, - end - - start + 1); - ret = btrfs_commit_transaction(trans, root); - } + if (ret == -ENOENT) + ret = 0; + goto out; } - } else { - ret = btrfs_end_transaction(trans, root); } + + ret = btrfs_commit_transaction(trans); out: + free_extent_buffer(ctx.scratch_eb); + ASSERT(list_empty(&ctx.list)); + ASSERT(list_empty(&ctx.conflict_inodes)); + err = file_check_and_advance_wb_err(file); + if (!ret) + ret = err; return ret > 0 ? -EIO : ret; + +out_release_extents: + btrfs_release_log_ctx_extents(&ctx); + if (skip_ilock) + up_write(&inode->i_mmap_lock); + else + btrfs_inode_unlock(inode, BTRFS_ILOCK_MMAP); + goto out; +} + +/* + * btrfs_page_mkwrite() is not allowed to change the file size as it gets + * called from a page fault handler when a page is first dirtied. Hence we must + * be careful to check for EOF conditions here. We set the page up correctly + * for a written page which means we get ENOSPC checking when writing into + * holes and correct delalloc and unwritten extent mapping on filesystems that + * support these features. + * + * We are not allowed to take the i_mutex here so we have to play games to + * protect against truncate races as the page could now be beyond EOF. Because + * truncate_setsize() writes the inode size before removing pages, once we have + * the page lock we can determine safely if the page is beyond EOF. If it is not + * beyond EOF, then the page is guaranteed safe against truncation until we + * unlock the page. + */ +static vm_fault_t btrfs_page_mkwrite(struct vm_fault *vmf) +{ + struct page *page = vmf->page; + struct folio *folio = page_folio(page); + struct btrfs_inode *inode = BTRFS_I(file_inode(vmf->vma->vm_file)); + struct btrfs_fs_info *fs_info = inode->root->fs_info; + struct extent_io_tree *io_tree = &inode->io_tree; + struct btrfs_ordered_extent *ordered; + struct extent_state *cached_state = NULL; + struct extent_changeset *data_reserved = NULL; + unsigned long zero_start; + loff_t size; + size_t fsize = folio_size(folio); + int ret; + bool only_release_metadata = false; + u64 reserved_space; + u64 page_start; + u64 page_end; + u64 end; + + reserved_space = fsize; + + sb_start_pagefault(inode->vfs_inode.i_sb); + page_start = folio_pos(folio); + page_end = page_start + folio_size(folio) - 1; + end = page_end; + + /* + * Reserving delalloc space after obtaining the page lock can lead to + * deadlock. For example, if a dirty page is locked by this function + * and the call to btrfs_delalloc_reserve_space() ends up triggering + * dirty page write out, then the btrfs_writepages() function could + * end up waiting indefinitely to get a lock on the page currently + * being processed by btrfs_page_mkwrite() function. + */ + ret = btrfs_check_data_free_space(inode, &data_reserved, page_start, + reserved_space, false); + if (ret < 0) { + size_t write_bytes = reserved_space; + + if (btrfs_check_nocow_lock(inode, page_start, &write_bytes, false) <= 0) + goto out_noreserve; + + only_release_metadata = true; + + /* + * Can't write the whole range, there may be shared extents or + * holes in the range, bail out with @only_release_metadata set + * to true so that we unlock the nocow lock before returning the + * error. + */ + if (write_bytes < reserved_space) + goto out_noreserve; + } + ret = btrfs_delalloc_reserve_metadata(inode, reserved_space, + reserved_space, false); + if (ret < 0) { + if (!only_release_metadata) + btrfs_free_reserved_data_space(inode, data_reserved, + page_start, reserved_space); + goto out_noreserve; + } + + ret = file_update_time(vmf->vma->vm_file); + if (ret < 0) + goto out; +again: + down_read(&inode->i_mmap_lock); + folio_lock(folio); + size = i_size_read(&inode->vfs_inode); + + if ((folio->mapping != inode->vfs_inode.i_mapping) || + (page_start >= size)) { + /* Page got truncated out from underneath us. */ + goto out_unlock; + } + folio_wait_writeback(folio); + + btrfs_lock_extent(io_tree, page_start, page_end, &cached_state); + ret = set_folio_extent_mapped(folio); + if (ret < 0) { + btrfs_unlock_extent(io_tree, page_start, page_end, &cached_state); + goto out_unlock; + } + + /* + * We can't set the delalloc bits if there are pending ordered + * extents. Drop our locks and wait for them to finish. + */ + ordered = btrfs_lookup_ordered_range(inode, page_start, fsize); + if (ordered) { + btrfs_unlock_extent(io_tree, page_start, page_end, &cached_state); + folio_unlock(folio); + up_read(&inode->i_mmap_lock); + btrfs_start_ordered_extent(ordered); + btrfs_put_ordered_extent(ordered); + goto again; + } + + if (folio_contains(folio, (size - 1) >> PAGE_SHIFT)) { + reserved_space = round_up(size - page_start, fs_info->sectorsize); + if (reserved_space < fsize) { + const u64 to_free = fsize - reserved_space; + + end = page_start + reserved_space - 1; + if (only_release_metadata) + btrfs_delalloc_release_metadata(inode, to_free, true); + else + btrfs_delalloc_release_space(inode, data_reserved, + end + 1, to_free, true); + } + } + + /* + * page_mkwrite gets called when the page is firstly dirtied after it's + * faulted in, but write(2) could also dirty a page and set delalloc + * bits, thus in this case for space account reason, we still need to + * clear any delalloc bits within this page range since we have to + * reserve data&meta space before lock_page() (see above comments). + */ + btrfs_clear_extent_bit(io_tree, page_start, end, + EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING | + EXTENT_DEFRAG, &cached_state); + + ret = btrfs_set_extent_delalloc(inode, page_start, end, 0, &cached_state); + if (ret < 0) { + btrfs_unlock_extent(io_tree, page_start, page_end, &cached_state); + goto out_unlock; + } + + /* Page is wholly or partially inside EOF. */ + if (page_start + folio_size(folio) > size) + zero_start = offset_in_folio(folio, size); + else + zero_start = fsize; + + if (zero_start != fsize) + folio_zero_range(folio, zero_start, folio_size(folio) - zero_start); + + btrfs_folio_clear_checked(fs_info, folio, page_start, fsize); + btrfs_folio_set_dirty(fs_info, folio, page_start, end + 1 - page_start); + btrfs_folio_set_uptodate(fs_info, folio, page_start, end + 1 - page_start); + + btrfs_set_inode_last_sub_trans(inode); + + if (only_release_metadata) + btrfs_set_extent_bit(io_tree, page_start, end, EXTENT_NORESERVE, + &cached_state); + + btrfs_unlock_extent(io_tree, page_start, page_end, &cached_state); + up_read(&inode->i_mmap_lock); + + btrfs_delalloc_release_extents(inode, fsize); + if (only_release_metadata) + btrfs_check_nocow_unlock(inode); + sb_end_pagefault(inode->vfs_inode.i_sb); + extent_changeset_free(data_reserved); + return VM_FAULT_LOCKED; + +out_unlock: + folio_unlock(folio); + up_read(&inode->i_mmap_lock); +out: + btrfs_delalloc_release_extents(inode, fsize); + if (only_release_metadata) + btrfs_delalloc_release_metadata(inode, reserved_space, true); + else + btrfs_delalloc_release_space(inode, data_reserved, page_start, + reserved_space, true); + extent_changeset_free(data_reserved); +out_noreserve: + if (only_release_metadata) + btrfs_check_nocow_unlock(inode); + + sb_end_pagefault(inode->vfs_inode.i_sb); + + if (ret < 0) + return vmf_error(ret); + + /* Make the VM retry the fault. */ + return VM_FAULT_NOPAGE; } static const struct vm_operations_struct btrfs_file_vm_ops = { .fault = filemap_fault, + .map_pages = filemap_map_pages, .page_mkwrite = btrfs_page_mkwrite, - .remap_pages = generic_file_remap_pages, }; -static int btrfs_file_mmap(struct file *filp, struct vm_area_struct *vma) +static int btrfs_file_mmap_prepare(struct vm_area_desc *desc) { + struct file *filp = desc->file; struct address_space *mapping = filp->f_mapping; - if (!mapping->a_ops->readpage) + if (unlikely(btrfs_is_shutdown(inode_to_fs_info(file_inode(filp))))) + return -EIO; + if (!mapping->a_ops->read_folio) return -ENOEXEC; file_accessed(filp); - vma->vm_ops = &btrfs_file_vm_ops; + desc->vm_ops = &btrfs_file_vm_ops; return 0; } -static int hole_mergeable(struct inode *inode, struct extent_buffer *leaf, - int slot, u64 start, u64 end) +static bool hole_mergeable(struct btrfs_inode *inode, struct extent_buffer *leaf, + int slot, u64 start, u64 end) { struct btrfs_file_extent_item *fi; struct btrfs_key key; if (slot < 0 || slot >= btrfs_header_nritems(leaf)) - return 0; + return false; btrfs_item_key_to_cpu(leaf, &key, slot); if (key.objectid != btrfs_ino(inode) || key.type != BTRFS_EXTENT_DATA_KEY) - return 0; + return false; fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item); if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_REG) - return 0; + return false; if (btrfs_file_extent_disk_bytenr(leaf, fi)) - return 0; + return false; if (key.offset == end) - return 1; + return true; if (key.offset + btrfs_file_extent_num_bytes(leaf, fi) == start) - return 1; - return 0; + return true; + return false; } -static int fill_holes(struct btrfs_trans_handle *trans, struct inode *inode, - struct btrfs_path *path, u64 offset, u64 end) +static int fill_holes(struct btrfs_trans_handle *trans, + struct btrfs_inode *inode, + struct btrfs_path *path, u64 offset, u64 end) { - struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_fs_info *fs_info = trans->fs_info; + struct btrfs_root *root = inode->root; struct extent_buffer *leaf; struct btrfs_file_extent_item *fi; struct extent_map *hole_em; - struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; struct btrfs_key key; int ret; + if (btrfs_fs_incompat(fs_info, NO_HOLES)) + goto out; + key.objectid = btrfs_ino(inode); key.type = BTRFS_EXTENT_DATA_KEY; key.offset = offset; - ret = btrfs_search_slot(trans, root, &key, path, 0, 1); - if (ret < 0) + if (ret <= 0) { + /* + * We should have dropped this offset, so if we find it then + * something has gone horribly wrong. + */ + if (ret == 0) + ret = -EINVAL; return ret; - BUG_ON(!ret); + } leaf = path->nodes[0]; - if (hole_mergeable(inode, leaf, path->slots[0]-1, offset, end)) { + if (hole_mergeable(inode, leaf, path->slots[0] - 1, offset, end)) { u64 num_bytes; path->slots[0]--; @@ -1976,16 +2126,15 @@ static int fill_holes(struct btrfs_trans_handle *trans, struct inode *inode, btrfs_set_file_extent_num_bytes(leaf, fi, num_bytes); btrfs_set_file_extent_ram_bytes(leaf, fi, num_bytes); btrfs_set_file_extent_offset(leaf, fi, 0); - btrfs_mark_buffer_dirty(leaf); + btrfs_set_file_extent_generation(leaf, fi, trans->transid); goto out; } - if (hole_mergeable(inode, leaf, path->slots[0]+1, offset, end)) { + if (hole_mergeable(inode, leaf, path->slots[0], offset, end)) { u64 num_bytes; - path->slots[0]++; key.offset = offset; - btrfs_set_item_key_safe(root, path, &key); + btrfs_set_item_key_safe(trans, path, &key); fi = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_file_extent_item); num_bytes = btrfs_file_extent_num_bytes(leaf, fi) + end - @@ -1993,246 +2142,954 @@ static int fill_holes(struct btrfs_trans_handle *trans, struct inode *inode, btrfs_set_file_extent_num_bytes(leaf, fi, num_bytes); btrfs_set_file_extent_ram_bytes(leaf, fi, num_bytes); btrfs_set_file_extent_offset(leaf, fi, 0); - btrfs_mark_buffer_dirty(leaf); + btrfs_set_file_extent_generation(leaf, fi, trans->transid); goto out; } btrfs_release_path(path); - ret = btrfs_insert_file_extent(trans, root, btrfs_ino(inode), offset, - 0, 0, end - offset, 0, end - offset, - 0, 0, 0); + ret = btrfs_insert_hole_extent(trans, root, btrfs_ino(inode), offset, + end - offset); if (ret) return ret; out: btrfs_release_path(path); - hole_em = alloc_extent_map(); + hole_em = btrfs_alloc_extent_map(); if (!hole_em) { - btrfs_drop_extent_cache(inode, offset, end - 1, 0); - set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, - &BTRFS_I(inode)->runtime_flags); + btrfs_drop_extent_map_range(inode, offset, end - 1, false); + btrfs_set_inode_full_sync(inode); } else { hole_em->start = offset; hole_em->len = end - offset; hole_em->ram_bytes = hole_em->len; - hole_em->orig_start = offset; - hole_em->block_start = EXTENT_MAP_HOLE; - hole_em->block_len = 0; - hole_em->orig_block_len = 0; - hole_em->bdev = root->fs_info->fs_devices->latest_bdev; - hole_em->compress_type = BTRFS_COMPRESS_NONE; + hole_em->disk_bytenr = EXTENT_MAP_HOLE; + hole_em->disk_num_bytes = 0; hole_em->generation = trans->transid; - do { - btrfs_drop_extent_cache(inode, offset, end - 1, 0); - write_lock(&em_tree->lock); - ret = add_extent_mapping(em_tree, hole_em, 1); - write_unlock(&em_tree->lock); - } while (ret == -EEXIST); - free_extent_map(hole_em); + ret = btrfs_replace_extent_map_range(inode, hole_em, true); + btrfs_free_extent_map(hole_em); if (ret) - set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, - &BTRFS_I(inode)->runtime_flags); + btrfs_set_inode_full_sync(inode); } return 0; } -static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len) +/* + * Find a hole extent on given inode and change start/len to the end of hole + * extent.(hole/vacuum extent whose em->start <= start && + * em->start + em->len > start) + * When a hole extent is found, return 1 and modify start/len. + */ +static int find_first_non_hole(struct btrfs_inode *inode, u64 *start, u64 *len) { - struct btrfs_root *root = BTRFS_I(inode)->root; - struct extent_state *cached_state = NULL; - struct btrfs_path *path; - struct btrfs_block_rsv *rsv; - struct btrfs_trans_handle *trans; - u64 lockstart = round_up(offset, BTRFS_I(inode)->root->sectorsize); - u64 lockend = round_down(offset + len, - BTRFS_I(inode)->root->sectorsize) - 1; - u64 cur_offset = lockstart; - u64 min_size = btrfs_calc_trunc_metadata_size(root, 1); - u64 drop_end; + struct btrfs_fs_info *fs_info = inode->root->fs_info; + struct extent_map *em; int ret = 0; - int err = 0; - bool same_page = ((offset >> PAGE_CACHE_SHIFT) == - ((offset + len - 1) >> PAGE_CACHE_SHIFT)); - btrfs_wait_ordered_range(inode, offset, len); + em = btrfs_get_extent(inode, NULL, + round_down(*start, fs_info->sectorsize), + round_up(*len, fs_info->sectorsize)); + if (IS_ERR(em)) + return PTR_ERR(em); - mutex_lock(&inode->i_mutex); - /* - * We needn't truncate any page which is beyond the end of the file - * because we are sure there is no data there. - */ + /* Hole or vacuum extent(only exists in no-hole mode) */ + if (em->disk_bytenr == EXTENT_MAP_HOLE) { + ret = 1; + *len = em->start + em->len > *start + *len ? + 0 : *start + *len - em->start - em->len; + *start = em->start + em->len; + } + btrfs_free_extent_map(em); + return ret; +} + +/* + * Check if there is no folio in the range. + * + * We cannot utilize filemap_range_has_page() in a filemap with large folios + * as we can hit the following false positive: + * + * start end + * | | + * |//|//|//|//| | | | | | | | |//|//| + * \ / \ / + * Folio A Folio B + * + * That large folio A and B cover the start and end indexes. + * In that case filemap_range_has_page() will always return true, but the above + * case is fine for btrfs_punch_hole_lock_range() usage. + * + * So here we only ensure that no other folios is in the range, excluding the + * head/tail large folio. + */ +static bool check_range_has_page(struct inode *inode, u64 start, u64 end) +{ + struct folio_batch fbatch; + bool ret = false; /* - * Only do this if we are in the same page and we aren't doing the - * entire page. + * For subpage case, if the range is not at page boundary, we could + * have pages at the leading/tailing part of the range. + * This could lead to dead loop since filemap_range_has_page() + * will always return true. + * So here we need to do extra page alignment for + * filemap_range_has_page(). + * + * And do not decrease page_lockend right now, as it can be 0. */ - if (same_page && len < PAGE_CACHE_SIZE) { - if (offset < round_up(inode->i_size, PAGE_CACHE_SIZE)) - ret = btrfs_truncate_page(inode, offset, len, 0); - mutex_unlock(&inode->i_mutex); - return ret; + const u64 page_lockstart = round_up(start, PAGE_SIZE); + const u64 page_lockend = round_down(end + 1, PAGE_SIZE); + const pgoff_t start_index = page_lockstart >> PAGE_SHIFT; + const pgoff_t end_index = (page_lockend - 1) >> PAGE_SHIFT; + pgoff_t tmp = start_index; + int found_folios; + + /* The same page or adjacent pages. */ + if (page_lockend <= page_lockstart) + return false; + + folio_batch_init(&fbatch); + found_folios = filemap_get_folios(inode->i_mapping, &tmp, end_index, &fbatch); + for (int i = 0; i < found_folios; i++) { + struct folio *folio = fbatch.folios[i]; + + /* A large folio begins before the start. Not a target. */ + if (folio->index < start_index) + continue; + /* A large folio extends beyond the end. Not a target. */ + if (folio_next_index(folio) > end_index) + continue; + /* A folio doesn't cover the head/tail index. Found a target. */ + ret = true; + break; } + folio_batch_release(&fbatch); + return ret; +} - /* zero back part of the first page */ - if (offset < round_up(inode->i_size, PAGE_CACHE_SIZE)) { - ret = btrfs_truncate_page(inode, offset, 0, 0); - if (ret) { - mutex_unlock(&inode->i_mutex); - return ret; - } - } +static void btrfs_punch_hole_lock_range(struct inode *inode, + const u64 lockstart, const u64 lockend, + struct extent_state **cached_state) +{ + while (1) { + truncate_pagecache_range(inode, lockstart, lockend); - /* zero the front end of the last page */ - if (offset + len < round_up(inode->i_size, PAGE_CACHE_SIZE)) { - ret = btrfs_truncate_page(inode, offset + len, 0, 1); - if (ret) { - mutex_unlock(&inode->i_mutex); - return ret; - } - } + btrfs_lock_extent(&BTRFS_I(inode)->io_tree, lockstart, lockend, + cached_state); + /* + * We can't have ordered extents in the range, nor dirty/writeback + * pages, because we have locked the inode's VFS lock in exclusive + * mode, we have locked the inode's i_mmap_lock in exclusive mode, + * we have flushed all delalloc in the range and we have waited + * for any ordered extents in the range to complete. + * We can race with anyone reading pages from this range, so after + * locking the range check if we have pages in the range, and if + * we do, unlock the range and retry. + */ + if (!check_range_has_page(inode, lockstart, lockend)) + break; - if (lockend < lockstart) { - mutex_unlock(&inode->i_mutex); - return 0; + btrfs_unlock_extent(&BTRFS_I(inode)->io_tree, lockstart, lockend, + cached_state); } - while (1) { - struct btrfs_ordered_extent *ordered; + btrfs_assert_inode_range_clean(BTRFS_I(inode), lockstart, lockend); +} - truncate_pagecache_range(inode, lockstart, lockend); +static int btrfs_insert_replace_extent(struct btrfs_trans_handle *trans, + struct btrfs_inode *inode, + struct btrfs_path *path, + struct btrfs_replace_extent_info *extent_info, + const u64 replace_len, + const u64 bytes_to_drop) +{ + struct btrfs_fs_info *fs_info = trans->fs_info; + struct btrfs_root *root = inode->root; + struct btrfs_file_extent_item *extent; + struct extent_buffer *leaf; + struct btrfs_key key; + int slot; + int ret; - lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend, - 0, &cached_state); - ordered = btrfs_lookup_first_ordered_extent(inode, lockend); + if (replace_len == 0) + return 0; - /* - * We need to make sure we have no ordered extents in this range - * and nobody raced in and read a page in this range, if we did - * we need to try again. - */ - if ((!ordered || - (ordered->file_offset + ordered->len < lockstart || - ordered->file_offset > lockend)) && - !test_range_bit(&BTRFS_I(inode)->io_tree, lockstart, - lockend, EXTENT_UPTODATE, 0, - cached_state)) { - if (ordered) - btrfs_put_ordered_extent(ordered); - break; - } - if (ordered) - btrfs_put_ordered_extent(ordered); - unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, - lockend, &cached_state, GFP_NOFS); - btrfs_wait_ordered_range(inode, lockstart, - lockend - lockstart + 1); + if (extent_info->disk_offset == 0 && + btrfs_fs_incompat(fs_info, NO_HOLES)) { + btrfs_update_inode_bytes(inode, 0, bytes_to_drop); + return 0; } - path = btrfs_alloc_path(); - if (!path) { - ret = -ENOMEM; - goto out; + key.objectid = btrfs_ino(inode); + key.type = BTRFS_EXTENT_DATA_KEY; + key.offset = extent_info->file_offset; + ret = btrfs_insert_empty_item(trans, root, path, &key, + sizeof(struct btrfs_file_extent_item)); + if (ret) + return ret; + leaf = path->nodes[0]; + slot = path->slots[0]; + write_extent_buffer(leaf, extent_info->extent_buf, + btrfs_item_ptr_offset(leaf, slot), + sizeof(struct btrfs_file_extent_item)); + extent = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item); + ASSERT(btrfs_file_extent_type(leaf, extent) != BTRFS_FILE_EXTENT_INLINE); + btrfs_set_file_extent_offset(leaf, extent, extent_info->data_offset); + btrfs_set_file_extent_num_bytes(leaf, extent, replace_len); + if (extent_info->is_new_extent) + btrfs_set_file_extent_generation(leaf, extent, trans->transid); + btrfs_release_path(path); + + ret = btrfs_inode_set_file_extent_range(inode, extent_info->file_offset, + replace_len); + if (ret) + return ret; + + /* If it's a hole, nothing more needs to be done. */ + if (extent_info->disk_offset == 0) { + btrfs_update_inode_bytes(inode, 0, bytes_to_drop); + return 0; } - rsv = btrfs_alloc_block_rsv(root, BTRFS_BLOCK_RSV_TEMP); - if (!rsv) { - ret = -ENOMEM; - goto out_free; + btrfs_update_inode_bytes(inode, replace_len, bytes_to_drop); + + if (extent_info->is_new_extent && extent_info->insertions == 0) { + key.objectid = extent_info->disk_offset; + key.type = BTRFS_EXTENT_ITEM_KEY; + key.offset = extent_info->disk_len; + ret = btrfs_alloc_reserved_file_extent(trans, root, + btrfs_ino(inode), + extent_info->file_offset, + extent_info->qgroup_reserved, + &key); + } else { + struct btrfs_ref ref = { + .action = BTRFS_ADD_DELAYED_REF, + .bytenr = extent_info->disk_offset, + .num_bytes = extent_info->disk_len, + .owning_root = btrfs_root_id(root), + .ref_root = btrfs_root_id(root), + }; + u64 ref_offset; + + ref_offset = extent_info->file_offset - extent_info->data_offset; + btrfs_init_data_ref(&ref, btrfs_ino(inode), ref_offset, 0, false); + ret = btrfs_inc_extent_ref(trans, &ref); } - rsv->size = btrfs_calc_trunc_metadata_size(root, 1); - rsv->failfast = 1; + + extent_info->insertions++; + + return ret; +} + +/* + * The respective range must have been previously locked, as well as the inode. + * The end offset is inclusive (last byte of the range). + * @extent_info is NULL for fallocate's hole punching and non-NULL when replacing + * the file range with an extent. + * When not punching a hole, we don't want to end up in a state where we dropped + * extents without inserting a new one, so we must abort the transaction to avoid + * a corruption. + */ +int btrfs_replace_file_extents(struct btrfs_inode *inode, + struct btrfs_path *path, const u64 start, + const u64 end, + struct btrfs_replace_extent_info *extent_info, + struct btrfs_trans_handle **trans_out) +{ + struct btrfs_drop_extents_args drop_args = { 0 }; + struct btrfs_root *root = inode->root; + struct btrfs_fs_info *fs_info = root->fs_info; + u64 min_size = btrfs_calc_insert_metadata_size(fs_info, 1); + u64 ino_size = round_up(inode->vfs_inode.i_size, fs_info->sectorsize); + struct btrfs_trans_handle *trans = NULL; + struct btrfs_block_rsv rsv; + unsigned int rsv_count; + u64 cur_offset; + u64 len = end - start; + int ret = 0; + + if (end <= start) + return -EINVAL; + + btrfs_init_metadata_block_rsv(fs_info, &rsv, BTRFS_BLOCK_RSV_TEMP); + rsv.size = btrfs_calc_insert_metadata_size(fs_info, 1); + rsv.failfast = true; /* * 1 - update the inode * 1 - removing the extents in the range - * 1 - adding the hole extent + * 1 - adding the hole extent if no_holes isn't set or if we are + * replacing the range with a new extent */ - trans = btrfs_start_transaction(root, 3); + if (!btrfs_fs_incompat(fs_info, NO_HOLES) || extent_info) + rsv_count = 3; + else + rsv_count = 2; + + trans = btrfs_start_transaction(root, rsv_count); if (IS_ERR(trans)) { - err = PTR_ERR(trans); - goto out_free; + ret = PTR_ERR(trans); + trans = NULL; + goto out_release; } - ret = btrfs_block_rsv_migrate(&root->fs_info->trans_block_rsv, rsv, - min_size); - BUG_ON(ret); - trans->block_rsv = rsv; - - while (cur_offset < lockend) { - ret = __btrfs_drop_extents(trans, root, inode, path, - cur_offset, lockend + 1, - &drop_end, 1); - if (ret != -ENOSPC) + ret = btrfs_block_rsv_migrate(&fs_info->trans_block_rsv, &rsv, + min_size, false); + if (WARN_ON(ret)) + goto out_trans; + trans->block_rsv = &rsv; + + cur_offset = start; + drop_args.path = path; + drop_args.end = end + 1; + drop_args.drop_cache = true; + while (cur_offset < end) { + drop_args.start = cur_offset; + ret = btrfs_drop_extents(trans, root, inode, &drop_args); + /* If we are punching a hole decrement the inode's byte count */ + if (!extent_info) + btrfs_update_inode_bytes(inode, 0, + drop_args.bytes_found); + if (ret != -ENOSPC) { + /* + * The only time we don't want to abort is if we are + * attempting to clone a partial inline extent, in which + * case we'll get EOPNOTSUPP. However if we aren't + * clone we need to abort no matter what, because if we + * got EOPNOTSUPP via prealloc then we messed up and + * need to abort. + */ + if (unlikely(ret && + (ret != -EOPNOTSUPP || + (extent_info && extent_info->is_new_extent)))) + btrfs_abort_transaction(trans, ret); break; + } - trans->block_rsv = &root->fs_info->trans_block_rsv; + trans->block_rsv = &fs_info->trans_block_rsv; - ret = fill_holes(trans, inode, path, cur_offset, drop_end); - if (ret) { - err = ret; - break; + if (!extent_info && cur_offset < drop_args.drop_end && + cur_offset < ino_size) { + ret = fill_holes(trans, inode, path, cur_offset, + drop_args.drop_end); + if (unlikely(ret)) { + /* + * If we failed then we didn't insert our hole + * entries for the area we dropped, so now the + * fs is corrupted, so we must abort the + * transaction. + */ + btrfs_abort_transaction(trans, ret); + break; + } + } else if (!extent_info && cur_offset < drop_args.drop_end) { + /* + * We are past the i_size here, but since we didn't + * insert holes we need to clear the mapped area so we + * know to not set disk_i_size in this area until a new + * file extent is inserted here. + */ + ret = btrfs_inode_clear_file_extent_range(inode, + cur_offset, + drop_args.drop_end - cur_offset); + if (unlikely(ret)) { + /* + * We couldn't clear our area, so we could + * presumably adjust up and corrupt the fs, so + * we need to abort. + */ + btrfs_abort_transaction(trans, ret); + break; + } } - cur_offset = drop_end; + if (extent_info && + drop_args.drop_end > extent_info->file_offset) { + u64 replace_len = drop_args.drop_end - + extent_info->file_offset; - ret = btrfs_update_inode(trans, root, inode); - if (ret) { - err = ret; - break; + ret = btrfs_insert_replace_extent(trans, inode, path, + extent_info, replace_len, + drop_args.bytes_found); + if (unlikely(ret)) { + btrfs_abort_transaction(trans, ret); + break; + } + extent_info->data_len -= replace_len; + extent_info->data_offset += replace_len; + extent_info->file_offset += replace_len; } - btrfs_end_transaction(trans, root); - btrfs_btree_balance_dirty(root); + /* + * We are releasing our handle on the transaction, balance the + * dirty pages of the btree inode and flush delayed items, and + * then get a new transaction handle, which may now point to a + * new transaction in case someone else may have committed the + * transaction we used to replace/drop file extent items. So + * bump the inode's iversion and update mtime and ctime except + * if we are called from a dedupe context. This is because a + * power failure/crash may happen after the transaction is + * committed and before we finish replacing/dropping all the + * file extent items we need. + */ + inode_inc_iversion(&inode->vfs_inode); - trans = btrfs_start_transaction(root, 3); + if (!extent_info || extent_info->update_times) + inode_set_mtime_to_ts(&inode->vfs_inode, + inode_set_ctime_current(&inode->vfs_inode)); + + ret = btrfs_update_inode(trans, inode); + if (ret) + break; + + btrfs_end_transaction(trans); + btrfs_btree_balance_dirty(fs_info); + + trans = btrfs_start_transaction(root, rsv_count); if (IS_ERR(trans)) { ret = PTR_ERR(trans); trans = NULL; break; } - ret = btrfs_block_rsv_migrate(&root->fs_info->trans_block_rsv, - rsv, min_size); - BUG_ON(ret); /* shouldn't happen */ - trans->block_rsv = rsv; - } + ret = btrfs_block_rsv_migrate(&fs_info->trans_block_rsv, + &rsv, min_size, false); + if (WARN_ON(ret)) + break; + trans->block_rsv = &rsv; - if (ret) { - err = ret; - goto out_trans; + cur_offset = drop_args.drop_end; + len = end - cur_offset; + if (!extent_info && len) { + ret = find_first_non_hole(inode, &cur_offset, &len); + if (unlikely(ret < 0)) + break; + if (ret && !len) { + ret = 0; + break; + } + } } - trans->block_rsv = &root->fs_info->trans_block_rsv; - ret = fill_holes(trans, inode, path, cur_offset, drop_end); - if (ret) { - err = ret; + /* + * If we were cloning, force the next fsync to be a full one since we + * we replaced (or just dropped in the case of cloning holes when + * NO_HOLES is enabled) file extent items and did not setup new extent + * maps for the replacement extents (or holes). + */ + if (extent_info && !extent_info->is_new_extent) + btrfs_set_inode_full_sync(inode); + + if (ret) goto out_trans; + + trans->block_rsv = &fs_info->trans_block_rsv; + /* + * If we are using the NO_HOLES feature we might have had already an + * hole that overlaps a part of the region [lockstart, lockend] and + * ends at (or beyond) lockend. Since we have no file extent items to + * represent holes, drop_end can be less than lockend and so we must + * make sure we have an extent map representing the existing hole (the + * call to __btrfs_drop_extents() might have dropped the existing extent + * map representing the existing hole), otherwise the fast fsync path + * will not record the existence of the hole region + * [existing_hole_start, lockend]. + */ + if (drop_args.drop_end <= end) + drop_args.drop_end = end + 1; + /* + * Don't insert file hole extent item if it's for a range beyond eof + * (because it's useless) or if it represents a 0 bytes range (when + * cur_offset == drop_end). + */ + if (!extent_info && cur_offset < ino_size && + cur_offset < drop_args.drop_end) { + ret = fill_holes(trans, inode, path, cur_offset, + drop_args.drop_end); + if (unlikely(ret)) { + /* Same comment as above. */ + btrfs_abort_transaction(trans, ret); + goto out_trans; + } + } else if (!extent_info && cur_offset < drop_args.drop_end) { + /* See the comment in the loop above for the reasoning here. */ + ret = btrfs_inode_clear_file_extent_range(inode, cur_offset, + drop_args.drop_end - cur_offset); + if (unlikely(ret)) { + btrfs_abort_transaction(trans, ret); + goto out_trans; + } + + } + if (extent_info) { + ret = btrfs_insert_replace_extent(trans, inode, path, + extent_info, extent_info->data_len, + drop_args.bytes_found); + if (unlikely(ret)) { + btrfs_abort_transaction(trans, ret); + goto out_trans; + } } out_trans: if (!trans) - goto out_free; + goto out_release; - inode_inc_iversion(inode); - inode->i_mtime = inode->i_ctime = CURRENT_TIME; + trans->block_rsv = &fs_info->trans_block_rsv; + if (ret) + btrfs_end_transaction(trans); + else + *trans_out = trans; +out_release: + btrfs_block_rsv_release(fs_info, &rsv, (u64)-1, NULL); + return ret; +} + +static int btrfs_punch_hole(struct file *file, loff_t offset, loff_t len) +{ + struct inode *inode = file_inode(file); + struct btrfs_fs_info *fs_info = inode_to_fs_info(inode); + struct btrfs_root *root = BTRFS_I(inode)->root; + struct extent_state *cached_state = NULL; + struct btrfs_path *path; + struct btrfs_trans_handle *trans = NULL; + u64 lockstart; + u64 lockend; + u64 tail_start; + u64 tail_len; + const u64 orig_start = offset; + const u64 orig_end = offset + len - 1; + int ret = 0; + bool same_block; + u64 ino_size; + bool truncated_block = false; + bool updated_inode = false; + + btrfs_inode_lock(BTRFS_I(inode), BTRFS_ILOCK_MMAP); + + ret = btrfs_wait_ordered_range(BTRFS_I(inode), offset, len); + if (ret) + goto out_only_mutex; + + ino_size = round_up(inode->i_size, fs_info->sectorsize); + ret = find_first_non_hole(BTRFS_I(inode), &offset, &len); + if (ret < 0) + goto out_only_mutex; + if (ret && !len) { + /* Already in a large hole */ + ret = 0; + goto out_only_mutex; + } + + ret = file_modified(file); + if (ret) + goto out_only_mutex; + + lockstart = round_up(offset, fs_info->sectorsize); + lockend = round_down(offset + len, fs_info->sectorsize) - 1; + same_block = (BTRFS_BYTES_TO_BLKS(fs_info, offset)) + == (BTRFS_BYTES_TO_BLKS(fs_info, offset + len - 1)); + /* + * Only do this if we are in the same block and we aren't doing the + * entire block. + */ + if (same_block && len < fs_info->sectorsize) { + if (offset < ino_size) { + truncated_block = true; + ret = btrfs_truncate_block(BTRFS_I(inode), offset + len - 1, + orig_start, orig_end); + } else { + ret = 0; + } + goto out_only_mutex; + } + + /* zero back part of the first block */ + if (offset < ino_size) { + truncated_block = true; + ret = btrfs_truncate_block(BTRFS_I(inode), offset, orig_start, orig_end); + if (ret) { + btrfs_inode_unlock(BTRFS_I(inode), BTRFS_ILOCK_MMAP); + return ret; + } + } + + /* Check the aligned pages after the first unaligned page, + * if offset != orig_start, which means the first unaligned page + * including several following pages are already in holes, + * the extra check can be skipped */ + if (offset == orig_start) { + /* after truncate page, check hole again */ + len = offset + len - lockstart; + offset = lockstart; + ret = find_first_non_hole(BTRFS_I(inode), &offset, &len); + if (ret < 0) + goto out_only_mutex; + if (ret && !len) { + ret = 0; + goto out_only_mutex; + } + lockstart = offset; + } + + /* Check the tail unaligned part is in a hole */ + tail_start = lockend + 1; + tail_len = offset + len - tail_start; + if (tail_len) { + ret = find_first_non_hole(BTRFS_I(inode), &tail_start, &tail_len); + if (unlikely(ret < 0)) + goto out_only_mutex; + if (!ret) { + /* zero the front end of the last page */ + if (tail_start + tail_len < ino_size) { + truncated_block = true; + ret = btrfs_truncate_block(BTRFS_I(inode), + tail_start + tail_len - 1, + orig_start, orig_end); + if (ret) + goto out_only_mutex; + } + } + } + + if (lockend < lockstart) { + ret = 0; + goto out_only_mutex; + } - trans->block_rsv = &root->fs_info->trans_block_rsv; - ret = btrfs_update_inode(trans, root, inode); - btrfs_end_transaction(trans, root); - btrfs_btree_balance_dirty(root); -out_free: + btrfs_punch_hole_lock_range(inode, lockstart, lockend, &cached_state); + + path = btrfs_alloc_path(); + if (!path) { + ret = -ENOMEM; + goto out; + } + + ret = btrfs_replace_file_extents(BTRFS_I(inode), path, lockstart, + lockend, NULL, &trans); btrfs_free_path(path); - btrfs_free_block_rsv(root, rsv); + if (ret) + goto out; + + ASSERT(trans != NULL); + inode_inc_iversion(inode); + inode_set_mtime_to_ts(inode, inode_set_ctime_current(inode)); + ret = btrfs_update_inode(trans, BTRFS_I(inode)); + updated_inode = true; + btrfs_end_transaction(trans); + btrfs_btree_balance_dirty(fs_info); out: - unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend, - &cached_state, GFP_NOFS); - mutex_unlock(&inode->i_mutex); - if (ret && !err) - err = ret; - return err; + btrfs_unlock_extent(&BTRFS_I(inode)->io_tree, lockstart, lockend, + &cached_state); +out_only_mutex: + if (!updated_inode && truncated_block && !ret) { + /* + * If we only end up zeroing part of a page, we still need to + * update the inode item, so that all the time fields are + * updated as well as the necessary btrfs inode in memory fields + * for detecting, at fsync time, if the inode isn't yet in the + * log tree or it's there but not up to date. + */ + struct timespec64 now = inode_set_ctime_current(inode); + + inode_inc_iversion(inode); + inode_set_mtime_to_ts(inode, now); + trans = btrfs_start_transaction(root, 1); + if (IS_ERR(trans)) { + ret = PTR_ERR(trans); + } else { + int ret2; + + ret = btrfs_update_inode(trans, BTRFS_I(inode)); + ret2 = btrfs_end_transaction(trans); + if (!ret) + ret = ret2; + } + } + btrfs_inode_unlock(BTRFS_I(inode), BTRFS_ILOCK_MMAP); + return ret; +} + +/* Helper structure to record which range is already reserved */ +struct falloc_range { + struct list_head list; + u64 start; + u64 len; +}; + +/* + * Helper function to add falloc range + * + * Caller should have locked the larger range of extent containing + * [start, len) + */ +static int add_falloc_range(struct list_head *head, u64 start, u64 len) +{ + struct falloc_range *range = NULL; + + if (!list_empty(head)) { + /* + * As fallocate iterates by bytenr order, we only need to check + * the last range. + */ + range = list_last_entry(head, struct falloc_range, list); + if (range->start + range->len == start) { + range->len += len; + return 0; + } + } + + range = kmalloc(sizeof(*range), GFP_KERNEL); + if (!range) + return -ENOMEM; + range->start = start; + range->len = len; + list_add_tail(&range->list, head); + return 0; +} + +static int btrfs_fallocate_update_isize(struct inode *inode, + const u64 end, + const int mode) +{ + struct btrfs_trans_handle *trans; + struct btrfs_root *root = BTRFS_I(inode)->root; + u64 range_start; + u64 range_end; + int ret; + int ret2; + + if (mode & FALLOC_FL_KEEP_SIZE || end <= i_size_read(inode)) + return 0; + + range_start = round_down(i_size_read(inode), root->fs_info->sectorsize); + range_end = round_up(end, root->fs_info->sectorsize); + + ret = btrfs_inode_set_file_extent_range(BTRFS_I(inode), range_start, + range_end - range_start); + if (ret) + return ret; + + trans = btrfs_start_transaction(root, 1); + if (IS_ERR(trans)) + return PTR_ERR(trans); + + inode_set_ctime_current(inode); + i_size_write(inode, end); + btrfs_inode_safe_disk_i_size_write(BTRFS_I(inode), 0); + ret = btrfs_update_inode(trans, BTRFS_I(inode)); + ret2 = btrfs_end_transaction(trans); + + return ret ? ret : ret2; +} + +enum { + RANGE_BOUNDARY_WRITTEN_EXTENT, + RANGE_BOUNDARY_PREALLOC_EXTENT, + RANGE_BOUNDARY_HOLE, +}; + +static int btrfs_zero_range_check_range_boundary(struct btrfs_inode *inode, + u64 offset) +{ + const u64 sectorsize = inode->root->fs_info->sectorsize; + struct extent_map *em; + int ret; + + offset = round_down(offset, sectorsize); + em = btrfs_get_extent(inode, NULL, offset, sectorsize); + if (IS_ERR(em)) + return PTR_ERR(em); + + if (em->disk_bytenr == EXTENT_MAP_HOLE) + ret = RANGE_BOUNDARY_HOLE; + else if (em->flags & EXTENT_FLAG_PREALLOC) + ret = RANGE_BOUNDARY_PREALLOC_EXTENT; + else + ret = RANGE_BOUNDARY_WRITTEN_EXTENT; + + btrfs_free_extent_map(em); + return ret; +} + +static int btrfs_zero_range(struct inode *inode, + loff_t offset, + loff_t len, + const int mode) +{ + struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info; + struct extent_map *em; + struct extent_changeset *data_reserved = NULL; + int ret; + u64 alloc_hint = 0; + const u64 sectorsize = fs_info->sectorsize; + const u64 orig_start = offset; + const u64 orig_end = offset + len - 1; + u64 alloc_start = round_down(offset, sectorsize); + u64 alloc_end = round_up(offset + len, sectorsize); + u64 bytes_to_reserve = 0; + bool space_reserved = false; + + em = btrfs_get_extent(BTRFS_I(inode), NULL, alloc_start, + alloc_end - alloc_start); + if (IS_ERR(em)) { + ret = PTR_ERR(em); + goto out; + } + + /* + * Avoid hole punching and extent allocation for some cases. More cases + * could be considered, but these are unlikely common and we keep things + * as simple as possible for now. Also, intentionally, if the target + * range contains one or more prealloc extents together with regular + * extents and holes, we drop all the existing extents and allocate a + * new prealloc extent, so that we get a larger contiguous disk extent. + */ + if (em->start <= alloc_start && (em->flags & EXTENT_FLAG_PREALLOC)) { + const u64 em_end = em->start + em->len; + + if (em_end >= offset + len) { + /* + * The whole range is already a prealloc extent, + * do nothing except updating the inode's i_size if + * needed. + */ + btrfs_free_extent_map(em); + ret = btrfs_fallocate_update_isize(inode, offset + len, + mode); + goto out; + } + /* + * Part of the range is already a prealloc extent, so operate + * only on the remaining part of the range. + */ + alloc_start = em_end; + ASSERT(IS_ALIGNED(alloc_start, sectorsize)); + len = offset + len - alloc_start; + offset = alloc_start; + alloc_hint = btrfs_extent_map_block_start(em) + em->len; + } + btrfs_free_extent_map(em); + + if (BTRFS_BYTES_TO_BLKS(fs_info, offset) == + BTRFS_BYTES_TO_BLKS(fs_info, offset + len - 1)) { + em = btrfs_get_extent(BTRFS_I(inode), NULL, alloc_start, sectorsize); + if (IS_ERR(em)) { + ret = PTR_ERR(em); + goto out; + } + + if (em->flags & EXTENT_FLAG_PREALLOC) { + btrfs_free_extent_map(em); + ret = btrfs_fallocate_update_isize(inode, offset + len, + mode); + goto out; + } + if (len < sectorsize && em->disk_bytenr != EXTENT_MAP_HOLE) { + btrfs_free_extent_map(em); + ret = btrfs_truncate_block(BTRFS_I(inode), offset + len - 1, + orig_start, orig_end); + if (!ret) + ret = btrfs_fallocate_update_isize(inode, + offset + len, + mode); + return ret; + } + btrfs_free_extent_map(em); + alloc_start = round_down(offset, sectorsize); + alloc_end = alloc_start + sectorsize; + goto reserve_space; + } + + alloc_start = round_up(offset, sectorsize); + alloc_end = round_down(offset + len, sectorsize); + + /* + * For unaligned ranges, check the pages at the boundaries, they might + * map to an extent, in which case we need to partially zero them, or + * they might map to a hole, in which case we need our allocation range + * to cover them. + */ + if (!IS_ALIGNED(offset, sectorsize)) { + ret = btrfs_zero_range_check_range_boundary(BTRFS_I(inode), + offset); + if (ret < 0) + goto out; + if (ret == RANGE_BOUNDARY_HOLE) { + alloc_start = round_down(offset, sectorsize); + ret = 0; + } else if (ret == RANGE_BOUNDARY_WRITTEN_EXTENT) { + ret = btrfs_truncate_block(BTRFS_I(inode), offset, + orig_start, orig_end); + if (ret) + goto out; + } else { + ret = 0; + } + } + + if (!IS_ALIGNED(offset + len, sectorsize)) { + ret = btrfs_zero_range_check_range_boundary(BTRFS_I(inode), + offset + len); + if (ret < 0) + goto out; + if (ret == RANGE_BOUNDARY_HOLE) { + alloc_end = round_up(offset + len, sectorsize); + ret = 0; + } else if (ret == RANGE_BOUNDARY_WRITTEN_EXTENT) { + ret = btrfs_truncate_block(BTRFS_I(inode), offset + len - 1, + orig_start, orig_end); + if (ret) + goto out; + } else { + ret = 0; + } + } + +reserve_space: + if (alloc_start < alloc_end) { + struct extent_state *cached_state = NULL; + const u64 lockstart = alloc_start; + const u64 lockend = alloc_end - 1; + + bytes_to_reserve = alloc_end - alloc_start; + ret = btrfs_alloc_data_chunk_ondemand(BTRFS_I(inode), + bytes_to_reserve); + if (ret < 0) + goto out; + space_reserved = true; + btrfs_punch_hole_lock_range(inode, lockstart, lockend, + &cached_state); + ret = btrfs_qgroup_reserve_data(BTRFS_I(inode), &data_reserved, + alloc_start, bytes_to_reserve); + if (ret) { + btrfs_unlock_extent(&BTRFS_I(inode)->io_tree, lockstart, + lockend, &cached_state); + goto out; + } + ret = btrfs_prealloc_file_range(inode, mode, alloc_start, + alloc_end - alloc_start, + fs_info->sectorsize, + offset + len, &alloc_hint); + btrfs_unlock_extent(&BTRFS_I(inode)->io_tree, lockstart, lockend, + &cached_state); + /* btrfs_prealloc_file_range releases reserved space on error */ + if (ret) { + space_reserved = false; + goto out; + } + } + ret = btrfs_fallocate_update_isize(inode, offset + len, mode); + out: + if (ret && space_reserved) + btrfs_free_reserved_data_space(BTRFS_I(inode), data_reserved, + alloc_start, bytes_to_reserve); + extent_changeset_free(data_reserved); + + return ret; } static long btrfs_fallocate(struct file *file, int mode, @@ -2240,337 +3097,799 @@ static long btrfs_fallocate(struct file *file, int mode, { struct inode *inode = file_inode(file); struct extent_state *cached_state = NULL; - struct btrfs_root *root = BTRFS_I(inode)->root; + struct extent_changeset *data_reserved = NULL; + struct falloc_range *range; + struct falloc_range *tmp; + LIST_HEAD(reserve_list); u64 cur_offset; u64 last_byte; u64 alloc_start; u64 alloc_end; u64 alloc_hint = 0; u64 locked_end; + u64 actual_end = 0; + u64 data_space_needed = 0; + u64 data_space_reserved = 0; + u64 qgroup_reserved = 0; struct extent_map *em; - int blocksize = BTRFS_I(inode)->root->sectorsize; + int blocksize = BTRFS_I(inode)->root->fs_info->sectorsize; int ret; + if (unlikely(btrfs_is_shutdown(inode_to_fs_info(inode)))) + return -EIO; + + /* Do not allow fallocate in ZONED mode */ + if (btrfs_is_zoned(inode_to_fs_info(inode))) + return -EOPNOTSUPP; + alloc_start = round_down(offset, blocksize); alloc_end = round_up(offset + len, blocksize); + cur_offset = alloc_start; /* Make sure we aren't being give some crap mode */ - if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE)) + if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE | + FALLOC_FL_ZERO_RANGE)) return -EOPNOTSUPP; if (mode & FALLOC_FL_PUNCH_HOLE) - return btrfs_punch_hole(inode, offset, len); + return btrfs_punch_hole(file, offset, len); - /* - * Make sure we have enough space before we do the - * allocation. - */ - ret = btrfs_check_data_free_space(inode, alloc_end - alloc_start); - if (ret) - return ret; - if (root->fs_info->quota_enabled) { - ret = btrfs_qgroup_reserve(root, alloc_end - alloc_start); + btrfs_inode_lock(BTRFS_I(inode), BTRFS_ILOCK_MMAP); + + if (!(mode & FALLOC_FL_KEEP_SIZE) && offset + len > inode->i_size) { + ret = inode_newsize_ok(inode, offset + len); if (ret) - goto out_reserve_fail; + goto out; } - mutex_lock(&inode->i_mutex); - ret = inode_newsize_ok(inode, alloc_end); + ret = file_modified(file); if (ret) goto out; + /* + * TODO: Move these two operations after we have checked + * accurate reserved space, or fallocate can still fail but + * with page truncated or size expanded. + * + * But that's a minor problem and won't do much harm BTW. + */ if (alloc_start > inode->i_size) { - ret = btrfs_cont_expand(inode, i_size_read(inode), + ret = btrfs_cont_expand(BTRFS_I(inode), i_size_read(inode), alloc_start); if (ret) goto out; - } else { + } else if (offset + len > inode->i_size) { /* * If we are fallocating from the end of the file onward we - * need to zero out the end of the page if i_size lands in the - * middle of a page. + * need to zero out the end of the block if i_size lands in the + * middle of a block. */ - ret = btrfs_truncate_page(inode, inode->i_size, 0, 0); + ret = btrfs_truncate_block(BTRFS_I(inode), inode->i_size, + inode->i_size, (u64)-1); if (ret) goto out; } /* - * wait for ordered IO before we have any locks. We'll loop again - * below with the locks held. + * We have locked the inode at the VFS level (in exclusive mode) and we + * have locked the i_mmap_lock lock (in exclusive mode). Now before + * locking the file range, flush all dealloc in the range and wait for + * all ordered extents in the range to complete. After this we can lock + * the file range and, due to the previous locking we did, we know there + * can't be more delalloc or ordered extents in the range. */ - btrfs_wait_ordered_range(inode, alloc_start, alloc_end - alloc_start); - - locked_end = alloc_end - 1; - while (1) { - struct btrfs_ordered_extent *ordered; + ret = btrfs_wait_ordered_range(BTRFS_I(inode), alloc_start, + alloc_end - alloc_start); + if (ret) + goto out; - /* the extent lock is ordered inside the running - * transaction - */ - lock_extent_bits(&BTRFS_I(inode)->io_tree, alloc_start, - locked_end, 0, &cached_state); - ordered = btrfs_lookup_first_ordered_extent(inode, - alloc_end - 1); - if (ordered && - ordered->file_offset + ordered->len > alloc_start && - ordered->file_offset < alloc_end) { - btrfs_put_ordered_extent(ordered); - unlock_extent_cached(&BTRFS_I(inode)->io_tree, - alloc_start, locked_end, - &cached_state, GFP_NOFS); - /* - * we can't wait on the range with the transaction - * running or with the extent lock held - */ - btrfs_wait_ordered_range(inode, alloc_start, - alloc_end - alloc_start); - } else { - if (ordered) - btrfs_put_ordered_extent(ordered); - break; - } + if (mode & FALLOC_FL_ZERO_RANGE) { + ret = btrfs_zero_range(inode, offset, len, mode); + btrfs_inode_unlock(BTRFS_I(inode), BTRFS_ILOCK_MMAP); + return ret; } - cur_offset = alloc_start; - while (1) { - u64 actual_end; + locked_end = alloc_end - 1; + btrfs_lock_extent(&BTRFS_I(inode)->io_tree, alloc_start, locked_end, + &cached_state); - em = btrfs_get_extent(inode, NULL, 0, cur_offset, - alloc_end - cur_offset, 0); - if (IS_ERR_OR_NULL(em)) { - if (!em) - ret = -ENOMEM; - else - ret = PTR_ERR(em); + btrfs_assert_inode_range_clean(BTRFS_I(inode), alloc_start, locked_end); + + /* First, check if we exceed the qgroup limit */ + while (cur_offset < alloc_end) { + em = btrfs_get_extent(BTRFS_I(inode), NULL, cur_offset, + alloc_end - cur_offset); + if (IS_ERR(em)) { + ret = PTR_ERR(em); break; } - last_byte = min(extent_map_end(em), alloc_end); - actual_end = min_t(u64, extent_map_end(em), offset + len); + last_byte = min(btrfs_extent_map_end(em), alloc_end); + actual_end = min_t(u64, btrfs_extent_map_end(em), offset + len); last_byte = ALIGN(last_byte, blocksize); - - if (em->block_start == EXTENT_MAP_HOLE || + if (em->disk_bytenr == EXTENT_MAP_HOLE || (cur_offset >= inode->i_size && - !test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) { - ret = btrfs_prealloc_file_range(inode, mode, cur_offset, - last_byte - cur_offset, - 1 << inode->i_blkbits, - offset + len, - &alloc_hint); + !(em->flags & EXTENT_FLAG_PREALLOC))) { + const u64 range_len = last_byte - cur_offset; + ret = add_falloc_range(&reserve_list, cur_offset, range_len); + if (ret < 0) { + btrfs_free_extent_map(em); + break; + } + ret = btrfs_qgroup_reserve_data(BTRFS_I(inode), + &data_reserved, cur_offset, range_len); if (ret < 0) { - free_extent_map(em); + btrfs_free_extent_map(em); break; } - } else if (actual_end > inode->i_size && - !(mode & FALLOC_FL_KEEP_SIZE)) { + qgroup_reserved += range_len; + data_space_needed += range_len; + } + btrfs_free_extent_map(em); + cur_offset = last_byte; + } + + if (!ret && data_space_needed > 0) { + /* + * We are safe to reserve space here as we can't have delalloc + * in the range, see above. + */ + ret = btrfs_alloc_data_chunk_ondemand(BTRFS_I(inode), + data_space_needed); + if (!ret) + data_space_reserved = data_space_needed; + } + + /* + * If ret is still 0, means we're OK to fallocate. + * Or just cleanup the list and exit. + */ + list_for_each_entry_safe(range, tmp, &reserve_list, list) { + if (!ret) { + ret = btrfs_prealloc_file_range(inode, mode, + range->start, + range->len, blocksize, + offset + len, &alloc_hint); /* - * We didn't need to allocate any more space, but we - * still extended the size of the file so we need to - * update i_size. + * btrfs_prealloc_file_range() releases space even + * if it returns an error. */ - inode->i_ctime = CURRENT_TIME; - i_size_write(inode, actual_end); - btrfs_ordered_update_i_size(inode, actual_end, NULL); + data_space_reserved -= range->len; + qgroup_reserved -= range->len; + } else if (data_space_reserved > 0) { + btrfs_free_reserved_data_space(BTRFS_I(inode), + data_reserved, range->start, + range->len); + data_space_reserved -= range->len; + qgroup_reserved -= range->len; + } else if (qgroup_reserved > 0) { + btrfs_qgroup_free_data(BTRFS_I(inode), data_reserved, + range->start, range->len, NULL); + qgroup_reserved -= range->len; } - free_extent_map(em); + list_del(&range->list); + kfree(range); + } + if (ret < 0) + goto out_unlock; - cur_offset = last_byte; - if (cur_offset >= alloc_end) { - ret = 0; + /* + * We didn't need to allocate any more space, but we still extended the + * size of the file so we need to update i_size and the inode item. + */ + ret = btrfs_fallocate_update_isize(inode, actual_end, mode); +out_unlock: + btrfs_unlock_extent(&BTRFS_I(inode)->io_tree, alloc_start, locked_end, + &cached_state); +out: + btrfs_inode_unlock(BTRFS_I(inode), BTRFS_ILOCK_MMAP); + extent_changeset_free(data_reserved); + return ret; +} + +/* + * Helper for btrfs_find_delalloc_in_range(). Find a subrange in a given range + * that has unflushed and/or flushing delalloc. There might be other adjacent + * subranges after the one it found, so btrfs_find_delalloc_in_range() keeps + * looping while it gets adjacent subranges, and merging them together. + */ +static bool find_delalloc_subrange(struct btrfs_inode *inode, u64 start, u64 end, + struct extent_state **cached_state, + bool *search_io_tree, + u64 *delalloc_start_ret, u64 *delalloc_end_ret) +{ + u64 len = end + 1 - start; + u64 delalloc_len = 0; + struct btrfs_ordered_extent *oe; + u64 oe_start; + u64 oe_end; + + /* + * Search the io tree first for EXTENT_DELALLOC. If we find any, it + * means we have delalloc (dirty pages) for which writeback has not + * started yet. + */ + if (*search_io_tree) { + spin_lock(&inode->lock); + if (inode->delalloc_bytes > 0) { + spin_unlock(&inode->lock); + *delalloc_start_ret = start; + delalloc_len = btrfs_count_range_bits(&inode->io_tree, + delalloc_start_ret, end, + len, EXTENT_DELALLOC, 1, + cached_state); + } else { + spin_unlock(&inode->lock); + } + } + + if (delalloc_len > 0) { + /* + * If delalloc was found then *delalloc_start_ret has a sector size + * aligned value (rounded down). + */ + *delalloc_end_ret = *delalloc_start_ret + delalloc_len - 1; + + if (*delalloc_start_ret == start) { + /* Delalloc for the whole range, nothing more to do. */ + if (*delalloc_end_ret == end) + return true; + /* Else trim our search range for ordered extents. */ + start = *delalloc_end_ret + 1; + len = end + 1 - start; + } + } else { + /* No delalloc, future calls don't need to search again. */ + *search_io_tree = false; + } + + /* + * Now also check if there's any ordered extent in the range. + * We do this because: + * + * 1) When delalloc is flushed, the file range is locked, we clear the + * EXTENT_DELALLOC bit from the io tree and create an extent map and + * an ordered extent for the write. So we might just have been called + * after delalloc is flushed and before the ordered extent completes + * and inserts the new file extent item in the subvolume's btree; + * + * 2) We may have an ordered extent created by flushing delalloc for a + * subrange that starts before the subrange we found marked with + * EXTENT_DELALLOC in the io tree. + * + * We could also use the extent map tree to find such delalloc that is + * being flushed, but using the ordered extents tree is more efficient + * because it's usually much smaller as ordered extents are removed from + * the tree once they complete. With the extent maps, we may have them + * in the extent map tree for a very long time, and they were either + * created by previous writes or loaded by read operations. + */ + oe = btrfs_lookup_first_ordered_range(inode, start, len); + if (!oe) + return (delalloc_len > 0); + + /* The ordered extent may span beyond our search range. */ + oe_start = max(oe->file_offset, start); + oe_end = min(oe->file_offset + oe->num_bytes - 1, end); + + btrfs_put_ordered_extent(oe); + + /* Don't have unflushed delalloc, return the ordered extent range. */ + if (delalloc_len == 0) { + *delalloc_start_ret = oe_start; + *delalloc_end_ret = oe_end; + return true; + } + + /* + * We have both unflushed delalloc (io_tree) and an ordered extent. + * If the ranges are adjacent returned a combined range, otherwise + * return the leftmost range. + */ + if (oe_start < *delalloc_start_ret) { + if (oe_end < *delalloc_start_ret) + *delalloc_end_ret = oe_end; + *delalloc_start_ret = oe_start; + } else if (*delalloc_end_ret + 1 == oe_start) { + *delalloc_end_ret = oe_end; + } + + return true; +} + +/* + * Check if there's delalloc in a given range. + * + * @inode: The inode. + * @start: The start offset of the range. It does not need to be + * sector size aligned. + * @end: The end offset (inclusive value) of the search range. + * It does not need to be sector size aligned. + * @cached_state: Extent state record used for speeding up delalloc + * searches in the inode's io_tree. Can be NULL. + * @delalloc_start_ret: Output argument, set to the start offset of the + * subrange found with delalloc (may not be sector size + * aligned). + * @delalloc_end_ret: Output argument, set to he end offset (inclusive value) + * of the subrange found with delalloc. + * + * Returns true if a subrange with delalloc is found within the given range, and + * if so it sets @delalloc_start_ret and @delalloc_end_ret with the start and + * end offsets of the subrange. + */ +bool btrfs_find_delalloc_in_range(struct btrfs_inode *inode, u64 start, u64 end, + struct extent_state **cached_state, + u64 *delalloc_start_ret, u64 *delalloc_end_ret) +{ + u64 cur_offset = round_down(start, inode->root->fs_info->sectorsize); + u64 prev_delalloc_end = 0; + bool search_io_tree = true; + bool ret = false; + + while (cur_offset <= end) { + u64 delalloc_start; + u64 delalloc_end; + bool delalloc; + + delalloc = find_delalloc_subrange(inode, cur_offset, end, + cached_state, &search_io_tree, + &delalloc_start, + &delalloc_end); + if (!delalloc) + break; + + if (prev_delalloc_end == 0) { + /* First subrange found. */ + *delalloc_start_ret = max(delalloc_start, start); + *delalloc_end_ret = delalloc_end; + ret = true; + } else if (delalloc_start == prev_delalloc_end + 1) { + /* Subrange adjacent to the previous one, merge them. */ + *delalloc_end_ret = delalloc_end; + } else { + /* Subrange not adjacent to the previous one, exit. */ break; } + + prev_delalloc_end = delalloc_end; + cur_offset = delalloc_end + 1; + cond_resched(); } - unlock_extent_cached(&BTRFS_I(inode)->io_tree, alloc_start, locked_end, - &cached_state, GFP_NOFS); -out: - mutex_unlock(&inode->i_mutex); - if (root->fs_info->quota_enabled) - btrfs_qgroup_free(root, alloc_end - alloc_start); -out_reserve_fail: - /* Let go of our reservation. */ - btrfs_free_reserved_data_space(inode, alloc_end - alloc_start); + return ret; } -static int find_desired_extent(struct inode *inode, loff_t *offset, int whence) +/* + * Check if there's a hole or delalloc range in a range representing a hole (or + * prealloc extent) found in the inode's subvolume btree. + * + * @inode: The inode. + * @whence: Seek mode (SEEK_DATA or SEEK_HOLE). + * @start: Start offset of the hole region. It does not need to be sector + * size aligned. + * @end: End offset (inclusive value) of the hole region. It does not + * need to be sector size aligned. + * @start_ret: Return parameter, used to set the start of the subrange in the + * hole that matches the search criteria (seek mode), if such + * subrange is found (return value of the function is true). + * The value returned here may not be sector size aligned. + * + * Returns true if a subrange matching the given seek mode is found, and if one + * is found, it updates @start_ret with the start of the subrange. + */ +static bool find_desired_extent_in_hole(struct btrfs_inode *inode, int whence, + struct extent_state **cached_state, + u64 start, u64 end, u64 *start_ret) { - struct btrfs_root *root = BTRFS_I(inode)->root; - struct extent_map *em; - struct extent_state *cached_state = NULL; - u64 lockstart = *offset; - u64 lockend = i_size_read(inode); - u64 start = *offset; - u64 orig_start = *offset; - u64 len = i_size_read(inode); - u64 last_end = 0; - int ret = 0; + u64 delalloc_start; + u64 delalloc_end; + bool delalloc; - lockend = max_t(u64, root->sectorsize, lockend); - if (lockend <= lockstart) - lockend = lockstart + root->sectorsize; + delalloc = btrfs_find_delalloc_in_range(inode, start, end, cached_state, + &delalloc_start, &delalloc_end); + if (delalloc && whence == SEEK_DATA) { + *start_ret = delalloc_start; + return true; + } - lockend--; - len = lockend - lockstart + 1; + if (delalloc && whence == SEEK_HOLE) { + /* + * We found delalloc but it starts after out start offset. So we + * have a hole between our start offset and the delalloc start. + */ + if (start < delalloc_start) { + *start_ret = start; + return true; + } + /* + * Delalloc range starts at our start offset. + * If the delalloc range's length is smaller than our range, + * then it means we have a hole that starts where the delalloc + * subrange ends. + */ + if (delalloc_end < end) { + *start_ret = delalloc_end + 1; + return true; + } - len = max_t(u64, len, root->sectorsize); - if (inode->i_size == 0) - return -ENXIO; + /* There's delalloc for the whole range. */ + return false; + } + + if (!delalloc && whence == SEEK_HOLE) { + *start_ret = start; + return true; + } - lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend, 0, - &cached_state); + /* + * No delalloc in the range and we are seeking for data. The caller has + * to iterate to the next extent item in the subvolume btree. + */ + return false; +} + +static loff_t find_desired_extent(struct file *file, loff_t offset, int whence) +{ + struct btrfs_inode *inode = BTRFS_I(file->f_mapping->host); + struct btrfs_file_private *private; + struct btrfs_fs_info *fs_info = inode->root->fs_info; + struct extent_state *cached_state = NULL; + struct extent_state **delalloc_cached_state; + const loff_t i_size = i_size_read(&inode->vfs_inode); + const u64 ino = btrfs_ino(inode); + struct btrfs_root *root = inode->root; + struct btrfs_path *path; + struct btrfs_key key; + u64 last_extent_end; + u64 lockstart; + u64 lockend; + u64 start; + int ret; + bool found = false; + + if (i_size == 0 || offset >= i_size) + return -ENXIO; /* - * Delalloc is such a pain. If we have a hole and we have pending - * delalloc for a portion of the hole we will get back a hole that - * exists for the entire range since it hasn't been actually written - * yet. So to take care of this case we need to look for an extent just - * before the position we want in case there is outstanding delalloc - * going on here. + * Quick path. If the inode has no prealloc extents and its number of + * bytes used matches its i_size, then it can not have holes. */ - if (whence == SEEK_HOLE && start != 0) { - if (start <= root->sectorsize) - em = btrfs_get_extent_fiemap(inode, NULL, 0, 0, - root->sectorsize, 0); - else - em = btrfs_get_extent_fiemap(inode, NULL, 0, - start - root->sectorsize, - root->sectorsize, 0); - if (IS_ERR(em)) { - ret = PTR_ERR(em); - goto out; + if (whence == SEEK_HOLE && + !(inode->flags & BTRFS_INODE_PREALLOC) && + inode_get_bytes(&inode->vfs_inode) == i_size) + return i_size; + + spin_lock(&inode->lock); + private = file->private_data; + spin_unlock(&inode->lock); + + if (private && private->owner_task != current) { + /* + * Not allocated by us, don't use it as its cached state is used + * by the task that allocated it and we don't want neither to + * mess with it nor get incorrect results because it reflects an + * invalid state for the current task. + */ + private = NULL; + } else if (!private) { + private = kzalloc(sizeof(*private), GFP_KERNEL); + /* + * No worries if memory allocation failed. + * The private structure is used only for speeding up multiple + * lseek SEEK_HOLE/DATA calls to a file when there's delalloc, + * so everything will still be correct. + */ + if (private) { + bool free = false; + + private->owner_task = current; + + spin_lock(&inode->lock); + if (file->private_data) + free = true; + else + file->private_data = private; + spin_unlock(&inode->lock); + + if (free) { + kfree(private); + private = NULL; + } } - last_end = em->start + em->len; - if (em->block_start == EXTENT_MAP_DELALLOC) - last_end = min_t(u64, last_end, inode->i_size); - free_extent_map(em); } - while (1) { - em = btrfs_get_extent_fiemap(inode, NULL, 0, start, len, 0); - if (IS_ERR(em)) { - ret = PTR_ERR(em); + if (private) + delalloc_cached_state = &private->llseek_cached_state; + else + delalloc_cached_state = NULL; + + /* + * offset can be negative, in this case we start finding DATA/HOLE from + * the very start of the file. + */ + start = max_t(loff_t, 0, offset); + + lockstart = round_down(start, fs_info->sectorsize); + lockend = round_up(i_size, fs_info->sectorsize); + if (lockend <= lockstart) + lockend = lockstart + fs_info->sectorsize; + lockend--; + + path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + path->reada = READA_FORWARD; + + key.objectid = ino; + key.type = BTRFS_EXTENT_DATA_KEY; + key.offset = start; + + last_extent_end = lockstart; + + btrfs_lock_extent(&inode->io_tree, lockstart, lockend, &cached_state); + + ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); + if (ret < 0) { + goto out; + } else if (ret > 0 && path->slots[0] > 0) { + btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0] - 1); + if (key.objectid == ino && key.type == BTRFS_EXTENT_DATA_KEY) + path->slots[0]--; + } + + while (start < i_size) { + struct extent_buffer *leaf = path->nodes[0]; + struct btrfs_file_extent_item *extent; + u64 extent_end; + u8 type; + + if (path->slots[0] >= btrfs_header_nritems(leaf)) { + ret = btrfs_next_leaf(root, path); + if (ret < 0) + goto out; + else if (ret > 0) + break; + + leaf = path->nodes[0]; + } + + btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); + if (key.objectid != ino || key.type != BTRFS_EXTENT_DATA_KEY) break; + + extent_end = btrfs_file_extent_end(path); + + /* + * In the first iteration we may have a slot that points to an + * extent that ends before our start offset, so skip it. + */ + if (extent_end <= start) { + path->slots[0]++; + continue; } - if (em->block_start == EXTENT_MAP_HOLE) { - if (test_bit(EXTENT_FLAG_VACANCY, &em->flags)) { - if (last_end <= orig_start) { - free_extent_map(em); - ret = -ENXIO; - break; - } + /* We have an implicit hole, NO_HOLES feature is likely set. */ + if (last_extent_end < key.offset) { + u64 search_start = last_extent_end; + u64 found_start; + + /* + * First iteration, @start matches @offset and it's + * within the hole. + */ + if (start == offset) + search_start = offset; + + found = find_desired_extent_in_hole(inode, whence, + delalloc_cached_state, + search_start, + key.offset - 1, + &found_start); + if (found) { + start = found_start; + break; } + /* + * Didn't find data or a hole (due to delalloc) in the + * implicit hole range, so need to analyze the extent. + */ + } + + extent = btrfs_item_ptr(leaf, path->slots[0], + struct btrfs_file_extent_item); + type = btrfs_file_extent_type(leaf, extent); + + /* + * Can't access the extent's disk_bytenr field if this is an + * inline extent, since at that offset, it's where the extent + * data starts. + */ + if (type == BTRFS_FILE_EXTENT_PREALLOC || + (type == BTRFS_FILE_EXTENT_REG && + btrfs_file_extent_disk_bytenr(leaf, extent) == 0)) { + /* + * Explicit hole or prealloc extent, search for delalloc. + * A prealloc extent is treated like a hole. + */ + u64 search_start = key.offset; + u64 found_start; - if (whence == SEEK_HOLE) { - *offset = start; - free_extent_map(em); + /* + * First iteration, @start matches @offset and it's + * within the hole. + */ + if (start == offset) + search_start = offset; + + found = find_desired_extent_in_hole(inode, whence, + delalloc_cached_state, + search_start, + extent_end - 1, + &found_start); + if (found) { + start = found_start; break; } + /* + * Didn't find data or a hole (due to delalloc) in the + * implicit hole range, so need to analyze the next + * extent item. + */ } else { + /* + * Found a regular or inline extent. + * If we are seeking for data, adjust the start offset + * and stop, we're done. + */ if (whence == SEEK_DATA) { - if (em->block_start == EXTENT_MAP_DELALLOC) { - if (start >= inode->i_size) { - free_extent_map(em); - ret = -ENXIO; - break; - } - } - - if (!test_bit(EXTENT_FLAG_PREALLOC, - &em->flags)) { - *offset = start; - free_extent_map(em); - break; - } + start = max_t(u64, key.offset, offset); + found = true; + break; } + /* + * Else, we are seeking for a hole, check the next file + * extent item. + */ } - start = em->start + em->len; - last_end = em->start + em->len; - - if (em->block_start == EXTENT_MAP_DELALLOC) - last_end = min_t(u64, last_end, inode->i_size); - - if (test_bit(EXTENT_FLAG_VACANCY, &em->flags)) { - free_extent_map(em); - ret = -ENXIO; - break; + start = extent_end; + last_extent_end = extent_end; + path->slots[0]++; + if (fatal_signal_pending(current)) { + ret = -EINTR; + goto out; } - free_extent_map(em); cond_resched(); } - if (!ret) - *offset = min(*offset, inode->i_size); + + /* We have an implicit hole from the last extent found up to i_size. */ + if (!found && start < i_size) { + found = find_desired_extent_in_hole(inode, whence, + delalloc_cached_state, start, + i_size - 1, &start); + if (!found) + start = i_size; + } + out: - unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend, - &cached_state, GFP_NOFS); - return ret; + btrfs_unlock_extent(&inode->io_tree, lockstart, lockend, &cached_state); + btrfs_free_path(path); + + if (ret < 0) + return ret; + + if (whence == SEEK_DATA && start >= i_size) + return -ENXIO; + + return min_t(loff_t, start, i_size); } static loff_t btrfs_file_llseek(struct file *file, loff_t offset, int whence) { struct inode *inode = file->f_mapping->host; - int ret; - mutex_lock(&inode->i_mutex); switch (whence) { - case SEEK_END: - case SEEK_CUR: - offset = generic_file_llseek(file, offset, whence); - goto out; + default: + return generic_file_llseek(file, offset, whence); case SEEK_DATA: case SEEK_HOLE: - if (offset >= i_size_read(inode)) { - mutex_unlock(&inode->i_mutex); - return -ENXIO; - } + btrfs_inode_lock(BTRFS_I(inode), BTRFS_ILOCK_SHARED); + offset = find_desired_extent(file, offset, whence); + btrfs_inode_unlock(BTRFS_I(inode), BTRFS_ILOCK_SHARED); + break; + } - ret = find_desired_extent(inode, &offset, whence); - if (ret) { - mutex_unlock(&inode->i_mutex); + if (offset < 0) + return offset; + + return vfs_setpos(file, offset, inode->i_sb->s_maxbytes); +} + +static int btrfs_file_open(struct inode *inode, struct file *filp) +{ + int ret; + + if (unlikely(btrfs_is_shutdown(inode_to_fs_info(inode)))) + return -EIO; + + filp->f_mode |= FMODE_NOWAIT | FMODE_CAN_ODIRECT; + + ret = fsverity_file_open(inode, filp); + if (ret) + return ret; + return generic_file_open(inode, filp); +} + +static ssize_t btrfs_file_read_iter(struct kiocb *iocb, struct iov_iter *to) +{ + ssize_t ret = 0; + + if (unlikely(btrfs_is_shutdown(inode_to_fs_info(file_inode(iocb->ki_filp))))) + return -EIO; + + if (iocb->ki_flags & IOCB_DIRECT) { + ret = btrfs_direct_read(iocb, to); + if (ret < 0 || !iov_iter_count(to) || + iocb->ki_pos >= i_size_read(file_inode(iocb->ki_filp))) return ret; - } } - offset = vfs_setpos(file, offset, inode->i_sb->s_maxbytes); -out: - mutex_unlock(&inode->i_mutex); - return offset; + return filemap_read(iocb, to, ret); +} + +static ssize_t btrfs_file_splice_read(struct file *in, loff_t *ppos, + struct pipe_inode_info *pipe, + size_t len, unsigned int flags) +{ + if (unlikely(btrfs_is_shutdown(inode_to_fs_info(file_inode(in))))) + return -EIO; + + return filemap_splice_read(in, ppos, pipe, len, flags); } const struct file_operations btrfs_file_operations = { .llseek = btrfs_file_llseek, - .read = do_sync_read, - .write = do_sync_write, - .aio_read = generic_file_aio_read, - .splice_read = generic_file_splice_read, - .aio_write = btrfs_file_aio_write, - .mmap = btrfs_file_mmap, - .open = generic_file_open, + .read_iter = btrfs_file_read_iter, + .splice_read = btrfs_file_splice_read, + .write_iter = btrfs_file_write_iter, + .splice_write = iter_file_splice_write, + .mmap_prepare = btrfs_file_mmap_prepare, + .open = btrfs_file_open, .release = btrfs_release_file, + .get_unmapped_area = thp_get_unmapped_area, .fsync = btrfs_sync_file, .fallocate = btrfs_fallocate, .unlocked_ioctl = btrfs_ioctl, #ifdef CONFIG_COMPAT - .compat_ioctl = btrfs_ioctl, + .compat_ioctl = btrfs_compat_ioctl, #endif + .remap_file_range = btrfs_remap_file_range, + .uring_cmd = btrfs_uring_cmd, + .fop_flags = FOP_BUFFER_RASYNC | FOP_BUFFER_WASYNC, }; -void btrfs_auto_defrag_exit(void) +int btrfs_fdatawrite_range(struct btrfs_inode *inode, loff_t start, loff_t end) { - if (btrfs_inode_defrag_cachep) - kmem_cache_destroy(btrfs_inode_defrag_cachep); -} + struct address_space *mapping = inode->vfs_inode.i_mapping; + int ret; -int btrfs_auto_defrag_init(void) -{ - btrfs_inode_defrag_cachep = kmem_cache_create("btrfs_inode_defrag", - sizeof(struct inode_defrag), 0, - SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD, - NULL); - if (!btrfs_inode_defrag_cachep) - return -ENOMEM; + /* + * So with compression we will find and lock a dirty page and clear the + * first one as dirty, setup an async extent, and immediately return + * with the entire range locked but with nobody actually marked with + * writeback. So we can't just filemap_write_and_wait_range() and + * expect it to work since it will just kick off a thread to do the + * actual work. So we need to call filemap_fdatawrite_range _again_ + * since it will wait on the page lock, which won't be unlocked until + * after the pages have been marked as writeback and so we're good to go + * from there. We have to do this otherwise we'll miss the ordered + * extents and that results in badness. Please Josef, do not think you + * know better and pull this out at some point in the future, it is + * right and you are wrong. + */ + ret = filemap_fdatawrite_range(mapping, start, end); + if (!ret && test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT, &inode->runtime_flags)) + ret = filemap_fdatawrite_range(mapping, start, end); - return 0; + return ret; } |