* Boston, MA 021110-1307, USA.
*/
+#include <linux/kernel.h>
+#include <linux/bio.h>
#include <linux/buffer_head.h>
+#include <linux/file.h>
#include <linux/fs.h>
#include <linux/pagemap.h>
#include <linux/highmem.h>
#include <linux/compat.h>
#include <linux/bit_spinlock.h>
#include <linux/version.h>
+#include <linux/xattr.h>
+#include <linux/posix_acl.h>
#include "ctree.h"
#include "disk-io.h"
#include "transaction.h"
#include "btrfs_inode.h"
#include "ioctl.h"
#include "print-tree.h"
+#include "volumes.h"
+#include "ordered-data.h"
+#include "xattr.h"
+#include "compat.h"
+#include "tree-log.h"
+#include "ref-cache.h"
+#include "compression.h"
struct btrfs_iget_args {
u64 ino;
static struct address_space_operations btrfs_aops;
static struct address_space_operations btrfs_symlink_aops;
static struct file_operations btrfs_dir_file_operations;
-static struct extent_map_ops btrfs_extent_map_ops;
+static struct extent_io_ops btrfs_extent_io_ops;
static struct kmem_cache *btrfs_inode_cachep;
struct kmem_cache *btrfs_trans_handle_cachep;
[S_IFLNK >> S_SHIFT] = BTRFS_FT_SYMLINK,
};
-static int run_delalloc_range(struct inode *inode, u64 start, u64 end)
+static void btrfs_truncate(struct inode *inode);
+static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end);
+
+/*
+ * a very lame attempt at stopping writes when the FS is 85% full. There
+ * are countless ways this is incorrect, but it is better than nothing.
+ */
+int btrfs_check_free_space(struct btrfs_root *root, u64 num_required,
+ int for_del)
+{
+ u64 total;
+ u64 used;
+ u64 thresh;
+ unsigned long flags;
+ int ret = 0;
+
+ spin_lock_irqsave(&root->fs_info->delalloc_lock, flags);
+ total = btrfs_super_total_bytes(&root->fs_info->super_copy);
+ used = btrfs_super_bytes_used(&root->fs_info->super_copy);
+ if (for_del)
+ thresh = total * 90;
+ else
+ thresh = total * 85;
+
+ do_div(thresh, 100);
+
+ if (used + root->fs_info->delalloc_bytes + num_required > thresh)
+ ret = -ENOSPC;
+ spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags);
+ return ret;
+}
+
+/*
+ * this does all the hard work for inserting an inline extent into
+ * the btree. The caller should have done a btrfs_drop_extents so that
+ * no overlapping inline items exist in the btree
+ */
+static int noinline insert_inline_extent(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root, struct inode *inode,
+ u64 start, size_t size, size_t compressed_size,
+ struct page **compressed_pages)
+{
+ struct btrfs_key key;
+ struct btrfs_path *path;
+ struct extent_buffer *leaf;
+ struct page *page = NULL;
+ char *kaddr;
+ unsigned long ptr;
+ struct btrfs_file_extent_item *ei;
+ int err = 0;
+ int ret;
+ size_t cur_size = size;
+ size_t datasize;
+ unsigned long offset;
+ int use_compress = 0;
+
+ if (compressed_size && compressed_pages) {
+ use_compress = 1;
+ cur_size = compressed_size;
+ }
+
+ path = btrfs_alloc_path(); if (!path)
+ return -ENOMEM;
+
+ btrfs_set_trans_block_group(trans, inode);
+
+ key.objectid = inode->i_ino;
+ key.offset = start;
+ btrfs_set_key_type(&key, BTRFS_EXTENT_DATA_KEY);
+ inode_add_bytes(inode, size);
+ datasize = btrfs_file_extent_calc_inline_size(cur_size);
+
+ inode_add_bytes(inode, size);
+ ret = btrfs_insert_empty_item(trans, root, path, &key,
+ datasize);
+ BUG_ON(ret);
+ if (ret) {
+ err = ret;
+ printk("got bad ret %d\n", ret);
+ goto fail;
+ }
+ leaf = path->nodes[0];
+ ei = btrfs_item_ptr(leaf, path->slots[0],
+ struct btrfs_file_extent_item);
+ btrfs_set_file_extent_generation(leaf, ei, trans->transid);
+ btrfs_set_file_extent_type(leaf, ei, BTRFS_FILE_EXTENT_INLINE);
+ btrfs_set_file_extent_encryption(leaf, ei, 0);
+ btrfs_set_file_extent_other_encoding(leaf, ei, 0);
+ btrfs_set_file_extent_ram_bytes(leaf, ei, size);
+ ptr = btrfs_file_extent_inline_start(ei);
+
+ if (use_compress) {
+ struct page *cpage;
+ int i = 0;
+ while(compressed_size > 0) {
+ cpage = compressed_pages[i];
+ cur_size = min(compressed_size,
+ PAGE_CACHE_SIZE);
+
+ kaddr = kmap(cpage);
+ write_extent_buffer(leaf, kaddr, ptr, cur_size);
+ kunmap(cpage);
+
+ i++;
+ ptr += cur_size;
+ compressed_size -= cur_size;
+ }
+ btrfs_set_file_extent_compression(leaf, ei,
+ BTRFS_COMPRESS_ZLIB);
+ } else {
+ page = find_get_page(inode->i_mapping,
+ start >> PAGE_CACHE_SHIFT);
+ btrfs_set_file_extent_compression(leaf, ei, 0);
+ kaddr = kmap_atomic(page, KM_USER0);
+ offset = start & (PAGE_CACHE_SIZE - 1);
+ write_extent_buffer(leaf, kaddr + offset, ptr, size);
+ kunmap_atomic(kaddr, KM_USER0);
+ page_cache_release(page);
+ }
+ btrfs_mark_buffer_dirty(leaf);
+ btrfs_free_path(path);
+
+ BTRFS_I(inode)->disk_i_size = inode->i_size;
+ btrfs_update_inode(trans, root, inode);
+ return 0;
+fail:
+ btrfs_free_path(path);
+ return err;
+}
+
+
+/*
+ * conditionally insert an inline extent into the file. This
+ * does the checks required to make sure the data is small enough
+ * to fit as an inline extent.
+ */
+static int cow_file_range_inline(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct inode *inode, u64 start, u64 end,
+ size_t compressed_size,
+ struct page **compressed_pages)
+{
+ u64 isize = i_size_read(inode);
+ u64 actual_end = min(end + 1, isize);
+ u64 inline_len = actual_end - start;
+ u64 aligned_end = (end + root->sectorsize - 1) &
+ ~((u64)root->sectorsize - 1);
+ u64 hint_byte;
+ u64 data_len = inline_len;
+ int ret;
+
+ if (compressed_size)
+ data_len = compressed_size;
+
+ if (start > 0 ||
+ data_len >= BTRFS_MAX_INLINE_DATA_SIZE(root) ||
+ (!compressed_size &&
+ (actual_end & (root->sectorsize - 1)) == 0) ||
+ end + 1 < isize ||
+ data_len > root->fs_info->max_inline) {
+ return 1;
+ }
+
+ mutex_lock(&BTRFS_I(inode)->extent_mutex);
+ ret = btrfs_drop_extents(trans, root, inode, start,
+ aligned_end, aligned_end, &hint_byte);
+ BUG_ON(ret);
+
+ if (isize > actual_end)
+ inline_len = min_t(u64, isize, actual_end);
+ ret = insert_inline_extent(trans, root, inode, start,
+ inline_len, compressed_size,
+ compressed_pages);
+ BUG_ON(ret);
+ btrfs_drop_extent_cache(inode, start, aligned_end, 0);
+ mutex_unlock(&BTRFS_I(inode)->extent_mutex);
+ return 0;
+}
+
+/*
+ * when extent_io.c finds a delayed allocation range in the file,
+ * the call backs end up in this code. The basic idea is to
+ * allocate extents on disk for the range, and create ordered data structs
+ * in ram to track those extents.
+ *
+ * locked_page is the page that writepage had locked already. We use
+ * it to make sure we don't do extra locks or unlocks.
+ *
+ * *page_started is set to one if we unlock locked_page and do everything
+ * required to start IO on it. It may be clean and already done with
+ * IO when we return.
+ */
+static int cow_file_range(struct inode *inode, struct page *locked_page,
+ u64 start, u64 end, int *page_started)
{
struct btrfs_root *root = BTRFS_I(inode)->root;
struct btrfs_trans_handle *trans;
- struct btrfs_key ins;
u64 alloc_hint = 0;
u64 num_bytes;
- int ret;
+ unsigned long ram_size;
+ u64 orig_start;
+ u64 disk_num_bytes;
+ u64 cur_alloc_size;
u64 blocksize = root->sectorsize;
-
- mutex_lock(&root->fs_info->fs_mutex);
- trans = btrfs_start_transaction(root, 1);
- btrfs_set_trans_block_group(trans, inode);
+ u64 actual_end;
+ struct btrfs_key ins;
+ struct extent_map *em;
+ struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+ int ret = 0;
+ struct page **pages = NULL;
+ unsigned long nr_pages;
+ unsigned long nr_pages_ret = 0;
+ unsigned long total_compressed = 0;
+ unsigned long total_in = 0;
+ unsigned long max_compressed = 128 * 1024;
+ unsigned long max_uncompressed = 256 * 1024;
+ int i;
+ int will_compress;
+
+ trans = btrfs_join_transaction(root, 1);
BUG_ON(!trans);
+ btrfs_set_trans_block_group(trans, inode);
+ orig_start = start;
+
+ /*
+ * compression made this loop a bit ugly, but the basic idea is to
+ * compress some pages but keep the total size of the compressed
+ * extent relatively small. If compression is off, this goto target
+ * is never used.
+ */
+again:
+ will_compress = 0;
+ nr_pages = (end >> PAGE_CACHE_SHIFT) - (start >> PAGE_CACHE_SHIFT) + 1;
+ nr_pages = min(nr_pages, (128 * 1024UL) / PAGE_CACHE_SIZE);
+
+ actual_end = min_t(u64, i_size_read(inode), end + 1);
+ total_compressed = actual_end - start;
+
+ /* we want to make sure that amount of ram required to uncompress
+ * an extent is reasonable, so we limit the total size in ram
+ * of a compressed extent to 256k
+ */
+ total_compressed = min(total_compressed, max_uncompressed);
num_bytes = (end - start + blocksize) & ~(blocksize - 1);
- ret = btrfs_drop_extents(trans, root, inode,
- start, start + num_bytes, start, &alloc_hint);
+ num_bytes = max(blocksize, num_bytes);
+ disk_num_bytes = num_bytes;
+ total_in = 0;
+ ret = 0;
- ret = btrfs_alloc_extent(trans, root, inode->i_ino, num_bytes, 0,
- alloc_hint, (u64)-1, &ins, 1);
- if (ret) {
- WARN_ON(1);
- goto out;
+ /* we do compression for mount -o compress and when the
+ * inode has not been flagged as nocompress
+ */
+ if (!btrfs_test_flag(inode, NOCOMPRESS) &&
+ btrfs_test_opt(root, COMPRESS)) {
+ WARN_ON(pages);
+ pages = kmalloc(sizeof(struct page *) * nr_pages, GFP_NOFS);
+
+ /* we want to make sure the amount of IO required to satisfy
+ * a random read is reasonably small, so we limit the size
+ * of a compressed extent to 128k
+ */
+ ret = btrfs_zlib_compress_pages(inode->i_mapping, start,
+ total_compressed, pages,
+ nr_pages, &nr_pages_ret,
+ &total_in,
+ &total_compressed,
+ max_compressed);
+
+ if (!ret) {
+ unsigned long offset = total_compressed &
+ (PAGE_CACHE_SIZE - 1);
+ struct page *page = pages[nr_pages_ret - 1];
+ char *kaddr;
+
+ /* zero the tail end of the last page, we might be
+ * sending it down to disk
+ */
+ if (offset) {
+ kaddr = kmap_atomic(page, KM_USER0);
+ memset(kaddr + offset, 0,
+ PAGE_CACHE_SIZE - offset);
+ kunmap_atomic(kaddr, KM_USER0);
+ }
+ will_compress = 1;
+ }
+ }
+ if (start == 0) {
+ /* lets try to make an inline extent */
+ if (ret || total_in < (end - start + 1)) {
+ /* we didn't compress the entire range, try
+ * to make an uncompressed inline extent. This
+ * is almost sure to fail, but maybe inline sizes
+ * will get bigger later
+ */
+ ret = cow_file_range_inline(trans, root, inode,
+ start, end, 0, NULL);
+ } else {
+ ret = cow_file_range_inline(trans, root, inode,
+ start, end,
+ total_compressed, pages);
+ }
+ if (ret == 0) {
+ extent_clear_unlock_delalloc(inode,
+ &BTRFS_I(inode)->io_tree,
+ start, end, NULL,
+ 1, 1, 1);
+ *page_started = 1;
+ ret = 0;
+ goto free_pages_out;
+ }
+ }
+
+ if (will_compress) {
+ /*
+ * we aren't doing an inline extent round the compressed size
+ * up to a block size boundary so the allocator does sane
+ * things
+ */
+ total_compressed = (total_compressed + blocksize - 1) &
+ ~(blocksize - 1);
+
+ /*
+ * one last check to make sure the compression is really a
+ * win, compare the page count read with the blocks on disk
+ */
+ total_in = (total_in + PAGE_CACHE_SIZE - 1) &
+ ~(PAGE_CACHE_SIZE - 1);
+ if (total_compressed >= total_in) {
+ will_compress = 0;
+ } else {
+ disk_num_bytes = total_compressed;
+ num_bytes = total_in;
+ }
+ }
+ if (!will_compress && pages) {
+ /*
+ * the compression code ran but failed to make things smaller,
+ * free any pages it allocated and our page pointer array
+ */
+ for (i = 0; i < nr_pages_ret; i++) {
+ page_cache_release(pages[i]);
+ }
+ kfree(pages);
+ pages = NULL;
+ total_compressed = 0;
+ nr_pages_ret = 0;
+
+ /* flag the file so we don't compress in the future */
+ btrfs_set_flag(inode, NOCOMPRESS);
+ }
+
+ BUG_ON(disk_num_bytes >
+ btrfs_super_total_bytes(&root->fs_info->super_copy));
+
+ mutex_lock(&BTRFS_I(inode)->extent_mutex);
+ btrfs_drop_extent_cache(inode, start, start + num_bytes - 1, 0);
+ mutex_unlock(&BTRFS_I(inode)->extent_mutex);
+
+ while(disk_num_bytes > 0) {
+ unsigned long min_bytes;
+
+ /*
+ * the max size of a compressed extent is pretty small,
+ * make the code a little less complex by forcing
+ * the allocator to find a whole compressed extent at once
+ */
+ if (will_compress)
+ min_bytes = disk_num_bytes;
+ else
+ min_bytes = root->sectorsize;
+
+ cur_alloc_size = min(disk_num_bytes, root->fs_info->max_extent);
+ ret = btrfs_reserve_extent(trans, root, cur_alloc_size,
+ min_bytes, 0, alloc_hint,
+ (u64)-1, &ins, 1);
+ if (ret) {
+ WARN_ON(1);
+ goto free_pages_out_fail;
+ }
+ em = alloc_extent_map(GFP_NOFS);
+ em->start = start;
+
+ if (will_compress) {
+ ram_size = num_bytes;
+ em->len = num_bytes;
+ } else {
+ /* ramsize == disk size */
+ ram_size = ins.offset;
+ em->len = ins.offset;
+ }
+
+ em->block_start = ins.objectid;
+ em->block_len = ins.offset;
+ em->bdev = root->fs_info->fs_devices->latest_bdev;
+
+ mutex_lock(&BTRFS_I(inode)->extent_mutex);
+ set_bit(EXTENT_FLAG_PINNED, &em->flags);
+
+ if (will_compress)
+ set_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
+
+ while(1) {
+ spin_lock(&em_tree->lock);
+ ret = add_extent_mapping(em_tree, em);
+ spin_unlock(&em_tree->lock);
+ if (ret != -EEXIST) {
+ free_extent_map(em);
+ break;
+ }
+ btrfs_drop_extent_cache(inode, start,
+ start + ram_size - 1, 0);
+ }
+ mutex_unlock(&BTRFS_I(inode)->extent_mutex);
+
+ cur_alloc_size = ins.offset;
+ ret = btrfs_add_ordered_extent(inode, start, ins.objectid,
+ ram_size, cur_alloc_size, 0,
+ will_compress);
+ BUG_ON(ret);
+
+ if (disk_num_bytes < cur_alloc_size) {
+ printk("num_bytes %Lu cur_alloc %Lu\n", disk_num_bytes,
+ cur_alloc_size);
+ break;
+ }
+
+ if (will_compress) {
+ /*
+ * we're doing compression, we and we need to
+ * submit the compressed extents down to the device.
+ *
+ * We lock down all the file pages, clearing their
+ * dirty bits and setting them writeback. Everyone
+ * that wants to modify the page will wait on the
+ * ordered extent above.
+ *
+ * The writeback bits on the file pages are
+ * cleared when the compressed pages are on disk
+ */
+ btrfs_end_transaction(trans, root);
+
+ if (start <= page_offset(locked_page) &&
+ page_offset(locked_page) < start + ram_size) {
+ *page_started = 1;
+ }
+
+ extent_clear_unlock_delalloc(inode,
+ &BTRFS_I(inode)->io_tree,
+ start,
+ start + ram_size - 1,
+ NULL, 1, 1, 0);
+
+ ret = btrfs_submit_compressed_write(inode, start,
+ ram_size, ins.objectid,
+ cur_alloc_size, pages,
+ nr_pages_ret);
+
+ BUG_ON(ret);
+ trans = btrfs_join_transaction(root, 1);
+ if (start + ram_size < end) {
+ start += ram_size;
+ alloc_hint = ins.objectid + ins.offset;
+ /* pages will be freed at end_bio time */
+ pages = NULL;
+ goto again;
+ } else {
+ /* we've written everything, time to go */
+ break;
+ }
+ }
+ /* we're not doing compressed IO, don't unlock the first
+ * page (which the caller expects to stay locked), don't
+ * clear any dirty bits and don't set any writeback bits
+ */
+ extent_clear_unlock_delalloc(inode, &BTRFS_I(inode)->io_tree,
+ start, start + ram_size - 1,
+ locked_page, 0, 0, 0);
+ disk_num_bytes -= cur_alloc_size;
+ num_bytes -= cur_alloc_size;
+ alloc_hint = ins.objectid + ins.offset;
+ start += cur_alloc_size;
}
- ret = btrfs_insert_file_extent(trans, root, inode->i_ino,
- start, ins.objectid, ins.offset,
- ins.offset);
+
+ ret = 0;
out:
btrfs_end_transaction(trans, root);
- mutex_unlock(&root->fs_info->fs_mutex);
+
return ret;
+
+free_pages_out_fail:
+ extent_clear_unlock_delalloc(inode, &BTRFS_I(inode)->io_tree,
+ start, end, locked_page, 0, 0, 0);
+free_pages_out:
+ for (i = 0; i < nr_pages_ret; i++)
+ page_cache_release(pages[i]);
+ if (pages)
+ kfree(pages);
+
+ goto out;
}
-int btrfs_writepage_io_hook(struct page *page, u64 start, u64 end)
+/*
+ * when nowcow writeback call back. This checks for snapshots or COW copies
+ * of the extents that exist in the file, and COWs the file as required.
+ *
+ * If no cow copies or snapshots exist, we write directly to the existing
+ * blocks on disk
+ */
+static int run_delalloc_nocow(struct inode *inode, struct page *locked_page,
+ u64 start, u64 end, int *page_started)
{
- struct inode *inode = page->mapping->host;
+ u64 extent_start;
+ u64 extent_end;
+ u64 bytenr;
+ u64 loops = 0;
+ u64 total_fs_bytes;
struct btrfs_root *root = BTRFS_I(inode)->root;
+ struct btrfs_block_group_cache *block_group;
struct btrfs_trans_handle *trans;
- char *kaddr;
+ struct extent_buffer *leaf;
+ int found_type;
+ struct btrfs_path *path;
+ struct btrfs_file_extent_item *item;
int ret;
- u64 page_start = page->index << PAGE_CACHE_SHIFT;
- size_t offset = start - page_start;
+ int err = 0;
+ struct btrfs_key found_key;
+
+ total_fs_bytes = btrfs_super_total_bytes(&root->fs_info->super_copy);
+ path = btrfs_alloc_path();
+ BUG_ON(!path);
+ trans = btrfs_join_transaction(root, 1);
+ BUG_ON(!trans);
+again:
+ ret = btrfs_lookup_file_extent(NULL, root, path,
+ inode->i_ino, start, 0);
+ if (ret < 0) {
+ err = ret;
+ goto out;
+ }
+
+ if (ret != 0) {
+ if (path->slots[0] == 0)
+ goto not_found;
+ path->slots[0]--;
+ }
+
+ leaf = path->nodes[0];
+ item = btrfs_item_ptr(leaf, path->slots[0],
+ struct btrfs_file_extent_item);
+
+ /* are we inside the extent that was found? */
+ btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
+ found_type = btrfs_key_type(&found_key);
+ if (found_key.objectid != inode->i_ino ||
+ found_type != BTRFS_EXTENT_DATA_KEY)
+ goto not_found;
+
+ found_type = btrfs_file_extent_type(leaf, item);
+ extent_start = found_key.offset;
+ if (found_type == BTRFS_FILE_EXTENT_REG) {
+ u64 extent_num_bytes;
+
+ extent_num_bytes = btrfs_file_extent_num_bytes(leaf, item);
+ extent_end = extent_start + extent_num_bytes;
+ err = 0;
+
+ if (btrfs_file_extent_compression(leaf, item) ||
+ btrfs_file_extent_encryption(leaf,item) ||
+ btrfs_file_extent_other_encoding(leaf, item))
+ goto not_found;
+
+ if (loops && start != extent_start)
+ goto not_found;
+
+ if (start < extent_start || start >= extent_end)
+ goto not_found;
+
+ bytenr = btrfs_file_extent_disk_bytenr(leaf, item);
+ if (bytenr == 0)
+ goto not_found;
+
+ if (btrfs_cross_ref_exists(trans, root, &found_key, bytenr))
+ goto not_found;
+ /*
+ * we may be called by the resizer, make sure we're inside
+ * the limits of the FS
+ */
+ block_group = btrfs_lookup_block_group(root->fs_info,
+ bytenr);
+ if (!block_group || block_group->ro)
+ goto not_found;
+
+ bytenr += btrfs_file_extent_offset(leaf, item);
+ extent_num_bytes = min(end + 1, extent_end) - start;
+ ret = btrfs_add_ordered_extent(inode, start, bytenr,
+ extent_num_bytes,
+ extent_num_bytes, 1, 0);
+ if (ret) {
+ err = ret;
+ goto out;
+ }
+
+ btrfs_release_path(root, path);
+ start = extent_end;
+ if (start <= end) {
+ loops++;
+ goto again;
+ }
+ } else {
+not_found:
+ btrfs_end_transaction(trans, root);
+ btrfs_free_path(path);
+ return cow_file_range(inode, locked_page, start, end,
+ page_started);
+ }
+out:
+ WARN_ON(err);
+ btrfs_end_transaction(trans, root);
+ btrfs_free_path(path);
+ return err;
+}
+
+/*
+ * extent_io.c call back to do delayed allocation processing
+ */
+static int run_delalloc_range(struct inode *inode, struct page *locked_page,
+ u64 start, u64 end, int *page_started)
+{
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ int ret;
+
+ if (btrfs_test_opt(root, NODATACOW) ||
+ btrfs_test_flag(inode, NODATACOW))
+ ret = run_delalloc_nocow(inode, locked_page, start, end,
+ page_started);
+ else
+ ret = cow_file_range(inode, locked_page, start, end,
+ page_started);
- mutex_lock(&root->fs_info->fs_mutex);
- trans = btrfs_start_transaction(root, 1);
- btrfs_set_trans_block_group(trans, inode);
- kaddr = kmap(page);
- btrfs_csum_file_block(trans, root, inode->i_ino,
- start, kaddr + offset, end - start + 1);
- kunmap(page);
- ret = btrfs_end_transaction(trans, root);
- BUG_ON(ret);
- mutex_unlock(&root->fs_info->fs_mutex);
return ret;
}
-int btrfs_readpage_io_hook(struct page *page, u64 start, u64 end)
+/*
+ * extent_io.c set_bit_hook, used to track delayed allocation
+ * bytes in this file, and to maintain the list of inodes that
+ * have pending delalloc work to be done.
+ */
+int btrfs_set_bit_hook(struct inode *inode, u64 start, u64 end,
+ unsigned long old, unsigned long bits)
+{
+ unsigned long flags;
+ if (!(old & EXTENT_DELALLOC) && (bits & EXTENT_DELALLOC)) {
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ spin_lock_irqsave(&root->fs_info->delalloc_lock, flags);
+ BTRFS_I(inode)->delalloc_bytes += end - start + 1;
+ root->fs_info->delalloc_bytes += end - start + 1;
+ if (list_empty(&BTRFS_I(inode)->delalloc_inodes)) {
+ list_add_tail(&BTRFS_I(inode)->delalloc_inodes,
+ &root->fs_info->delalloc_inodes);
+ }
+ spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags);
+ }
+ return 0;
+}
+
+/*
+ * extent_io.c clear_bit_hook, see set_bit_hook for why
+ */
+int btrfs_clear_bit_hook(struct inode *inode, u64 start, u64 end,
+ unsigned long old, unsigned long bits)
{
+ if ((old & EXTENT_DELALLOC) && (bits & EXTENT_DELALLOC)) {
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ unsigned long flags;
+
+ spin_lock_irqsave(&root->fs_info->delalloc_lock, flags);
+ if (end - start + 1 > root->fs_info->delalloc_bytes) {
+ printk("warning: delalloc account %Lu %Lu\n",
+ end - start + 1, root->fs_info->delalloc_bytes);
+ root->fs_info->delalloc_bytes = 0;
+ BTRFS_I(inode)->delalloc_bytes = 0;
+ } else {
+ root->fs_info->delalloc_bytes -= end - start + 1;
+ BTRFS_I(inode)->delalloc_bytes -= end - start + 1;
+ }
+ if (BTRFS_I(inode)->delalloc_bytes == 0 &&
+ !list_empty(&BTRFS_I(inode)->delalloc_inodes)) {
+ list_del_init(&BTRFS_I(inode)->delalloc_inodes);
+ }
+ spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags);
+ }
+ return 0;
+}
+
+/*
+ * extent_io.c merge_bio_hook, this must check the chunk tree to make sure
+ * we don't create bios that span stripes or chunks
+ */
+int btrfs_merge_bio_hook(struct page *page, unsigned long offset,
+ size_t size, struct bio *bio,
+ unsigned long bio_flags)
+{
+ struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
+ struct btrfs_mapping_tree *map_tree;
+ u64 logical = (u64)bio->bi_sector << 9;
+ u64 length = 0;
+ u64 map_length;
+ int ret;
+
+ length = bio->bi_size;
+ map_tree = &root->fs_info->mapping_tree;
+ map_length = length;
+ ret = btrfs_map_block(map_tree, READ, logical,
+ &map_length, NULL, 0);
+
+ if (map_length < length + size) {
+ return 1;
+ }
+ return 0;
+}
+
+/*
+ * in order to insert checksums into the metadata in large chunks,
+ * we wait until bio submission time. All the pages in the bio are
+ * checksummed and sums are attached onto the ordered extent record.
+ *
+ * At IO completion time the cums attached on the ordered extent record
+ * are inserted into the btree
+ */
+int __btrfs_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
+ int mirror_num, unsigned long bio_flags)
+{
+ struct btrfs_root *root = BTRFS_I(inode)->root;
int ret = 0;
+
+ ret = btrfs_csum_one_bio(root, inode, bio);
+ BUG_ON(ret);
+
+ return btrfs_map_bio(root, rw, bio, mirror_num, 1);
+}
+
+/*
+ * extent_io.c submission hook. This does the right thing for csum calculation on write,
+ * or reading the csums from the tree before a read
+ */
+int btrfs_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
+ int mirror_num, unsigned long bio_flags)
+{
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ int ret = 0;
+
+ ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0);
+ BUG_ON(ret);
+
+ if (btrfs_test_opt(root, NODATASUM) ||
+ btrfs_test_flag(inode, NODATASUM)) {
+ goto mapit;
+ }
+
+ if (!(rw & (1 << BIO_RW))) {
+ btrfs_lookup_bio_sums(root, inode, bio);
+
+ if (bio_flags & EXTENT_BIO_COMPRESSED) {
+ return btrfs_submit_compressed_read(inode, bio,
+ mirror_num, bio_flags);
+ }
+
+ goto mapit;
+ }
+ return btrfs_wq_submit_bio(BTRFS_I(inode)->root->fs_info,
+ inode, rw, bio, mirror_num,
+ bio_flags, __btrfs_submit_bio_hook);
+mapit:
+ return btrfs_map_bio(root, rw, bio, mirror_num, 0);
+}
+
+/*
+ * given a list of ordered sums record them in the inode. This happens
+ * at IO completion time based on sums calculated at bio submission time.
+ */
+static noinline int add_pending_csums(struct btrfs_trans_handle *trans,
+ struct inode *inode, u64 file_offset,
+ struct list_head *list)
+{
+ struct list_head *cur;
+ struct btrfs_ordered_sum *sum;
+
+ btrfs_set_trans_block_group(trans, inode);
+ list_for_each(cur, list) {
+ sum = list_entry(cur, struct btrfs_ordered_sum, list);
+ btrfs_csum_file_blocks(trans, BTRFS_I(inode)->root,
+ inode, sum);
+ }
+ return 0;
+}
+
+int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end)
+{
+ return set_extent_delalloc(&BTRFS_I(inode)->io_tree, start, end,
+ GFP_NOFS);
+}
+
+/* see btrfs_writepage_start_hook for details on why this is required */
+struct btrfs_writepage_fixup {
+ struct page *page;
+ struct btrfs_work work;
+};
+
+void btrfs_writepage_fixup_worker(struct btrfs_work *work)
+{
+ struct btrfs_writepage_fixup *fixup;
+ struct btrfs_ordered_extent *ordered;
+ struct page *page;
+ struct inode *inode;
+ u64 page_start;
+ u64 page_end;
+
+ fixup = container_of(work, struct btrfs_writepage_fixup, work);
+ page = fixup->page;
+again:
+ lock_page(page);
+ if (!page->mapping || !PageDirty(page) || !PageChecked(page)) {
+ ClearPageChecked(page);
+ goto out_page;
+ }
+
+ inode = page->mapping->host;
+ page_start = page_offset(page);
+ page_end = page_offset(page) + PAGE_CACHE_SIZE - 1;
+
+ lock_extent(&BTRFS_I(inode)->io_tree, page_start, page_end, GFP_NOFS);
+
+ /* already ordered? We're done */
+ if (test_range_bit(&BTRFS_I(inode)->io_tree, page_start, page_end,
+ EXTENT_ORDERED, 0)) {
+ goto out;
+ }
+
+ ordered = btrfs_lookup_ordered_extent(inode, page_start);
+ if (ordered) {
+ unlock_extent(&BTRFS_I(inode)->io_tree, page_start,
+ page_end, GFP_NOFS);
+ unlock_page(page);
+ btrfs_start_ordered_extent(inode, ordered, 1);
+ goto again;
+ }
+
+ btrfs_set_extent_delalloc(inode, page_start, page_end);
+ ClearPageChecked(page);
+out:
+ unlock_extent(&BTRFS_I(inode)->io_tree, page_start, page_end, GFP_NOFS);
+out_page:
+ unlock_page(page);
+ page_cache_release(page);
+}
+
+/*
+ * There are a few paths in the higher layers of the kernel that directly
+ * set the page dirty bit without asking the filesystem if it is a
+ * good idea. This causes problems because we want to make sure COW
+ * properly happens and the data=ordered rules are followed.
+ *
+ * In our case any range that doesn't have the ORDERED bit set
+ * hasn't been properly setup for IO. We kick off an async process
+ * to fix it up. The async helper will wait for ordered extents, set
+ * the delalloc bit and make it safe to write the page.
+ */
+int btrfs_writepage_start_hook(struct page *page, u64 start, u64 end)
+{
struct inode *inode = page->mapping->host;
+ struct btrfs_writepage_fixup *fixup;
struct btrfs_root *root = BTRFS_I(inode)->root;
- struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
- struct btrfs_csum_item *item;
+ int ret;
+
+ ret = test_range_bit(&BTRFS_I(inode)->io_tree, start, end,
+ EXTENT_ORDERED, 0);
+ if (ret)
+ return 0;
+
+ if (PageChecked(page))
+ return -EAGAIN;
+
+ fixup = kzalloc(sizeof(*fixup), GFP_NOFS);
+ if (!fixup)
+ return -EAGAIN;
+
+ SetPageChecked(page);
+ page_cache_get(page);
+ fixup->work.func = btrfs_writepage_fixup_worker;
+ fixup->page = page;
+ btrfs_queue_worker(&root->fs_info->fixup_workers, &fixup->work);
+ return -EAGAIN;
+}
+
+/* as ordered data IO finishes, this gets called so we can finish
+ * an ordered extent if the range of bytes in the file it covers are
+ * fully written.
+ */
+static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
+{
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ struct btrfs_trans_handle *trans;
+ struct btrfs_ordered_extent *ordered_extent;
+ struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+ struct btrfs_file_extent_item *extent_item;
struct btrfs_path *path = NULL;
- u32 csum;
+ struct extent_buffer *leaf;
+ u64 alloc_hint = 0;
+ struct list_head list;
+ struct btrfs_key ins;
+ int ret;
+
+ ret = btrfs_dec_test_ordered_pending(inode, start, end - start + 1);
+ if (!ret)
+ return 0;
+
+ trans = btrfs_join_transaction(root, 1);
+
+ ordered_extent = btrfs_lookup_ordered_extent(inode, start);
+ BUG_ON(!ordered_extent);
+ if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags))
+ goto nocow;
- mutex_lock(&root->fs_info->fs_mutex);
path = btrfs_alloc_path();
- item = btrfs_lookup_csum(NULL, root, path, inode->i_ino, start, 0);
- if (IS_ERR(item)) {
- ret = PTR_ERR(item);
- /* a csum that isn't present is a preallocated region. */
- if (ret == -ENOENT || ret == -EFBIG)
- ret = 0;
- csum = 0;
- goto out;
+ BUG_ON(!path);
+
+ lock_extent(io_tree, ordered_extent->file_offset,
+ ordered_extent->file_offset + ordered_extent->len - 1,
+ GFP_NOFS);
+
+ INIT_LIST_HEAD(&list);
+
+ mutex_lock(&BTRFS_I(inode)->extent_mutex);
+
+ ret = btrfs_drop_extents(trans, root, inode,
+ ordered_extent->file_offset,
+ ordered_extent->file_offset +
+ ordered_extent->len,
+ ordered_extent->file_offset, &alloc_hint);
+ BUG_ON(ret);
+
+ ins.objectid = inode->i_ino;
+ ins.offset = ordered_extent->file_offset;
+ ins.type = BTRFS_EXTENT_DATA_KEY;
+ ret = btrfs_insert_empty_item(trans, root, path, &ins,
+ sizeof(*extent_item));
+ BUG_ON(ret);
+ leaf = path->nodes[0];
+ extent_item = btrfs_item_ptr(leaf, path->slots[0],
+ struct btrfs_file_extent_item);
+ btrfs_set_file_extent_generation(leaf, extent_item, trans->transid);
+ btrfs_set_file_extent_type(leaf, extent_item, BTRFS_FILE_EXTENT_REG);
+ btrfs_set_file_extent_disk_bytenr(leaf, extent_item,
+ ordered_extent->start);
+ btrfs_set_file_extent_disk_num_bytes(leaf, extent_item,
+ ordered_extent->disk_len);
+ btrfs_set_file_extent_offset(leaf, extent_item, 0);
+
+ if (test_bit(BTRFS_ORDERED_COMPRESSED, &ordered_extent->flags))
+ btrfs_set_file_extent_compression(leaf, extent_item, 1);
+ else
+ btrfs_set_file_extent_compression(leaf, extent_item, 0);
+ btrfs_set_file_extent_encryption(leaf, extent_item, 0);
+ btrfs_set_file_extent_other_encoding(leaf, extent_item, 0);
+
+ /* ram bytes = extent_num_bytes for now */
+ btrfs_set_file_extent_num_bytes(leaf, extent_item,
+ ordered_extent->len);
+ btrfs_set_file_extent_ram_bytes(leaf, extent_item,
+ ordered_extent->len);
+ btrfs_mark_buffer_dirty(leaf);
+
+ btrfs_drop_extent_cache(inode, ordered_extent->file_offset,
+ ordered_extent->file_offset +
+ ordered_extent->len - 1, 0);
+ mutex_unlock(&BTRFS_I(inode)->extent_mutex);
+
+ ins.objectid = ordered_extent->start;
+ ins.offset = ordered_extent->disk_len;
+ ins.type = BTRFS_EXTENT_ITEM_KEY;
+ ret = btrfs_alloc_reserved_extent(trans, root, leaf->start,
+ root->root_key.objectid,
+ trans->transid, inode->i_ino, &ins);
+ BUG_ON(ret);
+ btrfs_release_path(root, path);
+
+ inode_add_bytes(inode, ordered_extent->len);
+ unlock_extent(io_tree, ordered_extent->file_offset,
+ ordered_extent->file_offset + ordered_extent->len - 1,
+ GFP_NOFS);
+nocow:
+ add_pending_csums(trans, inode, ordered_extent->file_offset,
+ &ordered_extent->list);
+
+ mutex_lock(&BTRFS_I(inode)->extent_mutex);
+ btrfs_ordered_update_i_size(inode, ordered_extent);
+ btrfs_update_inode(trans, root, inode);
+ btrfs_remove_ordered_extent(inode, ordered_extent);
+ mutex_unlock(&BTRFS_I(inode)->extent_mutex);
+
+ /* once for us */
+ btrfs_put_ordered_extent(ordered_extent);
+ /* once for the tree */
+ btrfs_put_ordered_extent(ordered_extent);
+
+ btrfs_end_transaction(trans, root);
+ if (path)
+ btrfs_free_path(path);
+ return 0;
+}
+
+int btrfs_writepage_end_io_hook(struct page *page, u64 start, u64 end,
+ struct extent_state *state, int uptodate)
+{
+ return btrfs_finish_ordered_io(page->mapping->host, start, end);
+}
+
+/*
+ * When IO fails, either with EIO or csum verification fails, we
+ * try other mirrors that might have a good copy of the data. This
+ * io_failure_record is used to record state as we go through all the
+ * mirrors. If another mirror has good data, the page is set up to date
+ * and things continue. If a good mirror can't be found, the original
+ * bio end_io callback is called to indicate things have failed.
+ */
+struct io_failure_record {
+ struct page *page;
+ u64 start;
+ u64 len;
+ u64 logical;
+ int last_mirror;
+};
+
+int btrfs_io_failed_hook(struct bio *failed_bio,
+ struct page *page, u64 start, u64 end,
+ struct extent_state *state)
+{
+ struct io_failure_record *failrec = NULL;
+ u64 private;
+ struct extent_map *em;
+ struct inode *inode = page->mapping->host;
+ struct extent_io_tree *failure_tree = &BTRFS_I(inode)->io_failure_tree;
+ struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+ struct bio *bio;
+ int num_copies;
+ int ret;
+ int rw;
+ u64 logical;
+ unsigned long bio_flags = 0;
+
+ ret = get_state_private(failure_tree, start, &private);
+ if (ret) {
+ failrec = kmalloc(sizeof(*failrec), GFP_NOFS);
+ if (!failrec)
+ return -ENOMEM;
+ failrec->start = start;
+ failrec->len = end - start + 1;
+ failrec->last_mirror = 0;
+
+ spin_lock(&em_tree->lock);
+ em = lookup_extent_mapping(em_tree, start, failrec->len);
+ if (em->start > start || em->start + em->len < start) {
+ free_extent_map(em);
+ em = NULL;
+ }
+ spin_unlock(&em_tree->lock);
+
+ if (!em || IS_ERR(em)) {
+ kfree(failrec);
+ return -EIO;
+ }
+ logical = start - em->start;
+ logical = em->block_start + logical;
+ if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags))
+ bio_flags = EXTENT_BIO_COMPRESSED;
+ failrec->logical = logical;
+ free_extent_map(em);
+ set_extent_bits(failure_tree, start, end, EXTENT_LOCKED |
+ EXTENT_DIRTY, GFP_NOFS);
+ set_state_private(failure_tree, start,
+ (u64)(unsigned long)failrec);
+ } else {
+ failrec = (struct io_failure_record *)(unsigned long)private;
+ }
+ num_copies = btrfs_num_copies(
+ &BTRFS_I(inode)->root->fs_info->mapping_tree,
+ failrec->logical, failrec->len);
+ failrec->last_mirror++;
+ if (!state) {
+ spin_lock_irq(&BTRFS_I(inode)->io_tree.lock);
+ state = find_first_extent_bit_state(&BTRFS_I(inode)->io_tree,
+ failrec->start,
+ EXTENT_LOCKED);
+ if (state && state->start != failrec->start)
+ state = NULL;
+ spin_unlock_irq(&BTRFS_I(inode)->io_tree.lock);
+ }
+ if (!state || failrec->last_mirror > num_copies) {
+ set_state_private(failure_tree, failrec->start, 0);
+ clear_extent_bits(failure_tree, failrec->start,
+ failrec->start + failrec->len - 1,
+ EXTENT_LOCKED | EXTENT_DIRTY, GFP_NOFS);
+ kfree(failrec);
+ return -EIO;
+ }
+ bio = bio_alloc(GFP_NOFS, 1);
+ bio->bi_private = state;
+ bio->bi_end_io = failed_bio->bi_end_io;
+ bio->bi_sector = failrec->logical >> 9;
+ bio->bi_bdev = failed_bio->bi_bdev;
+ bio->bi_size = 0;
+ bio_add_page(bio, page, failrec->len, start - page_offset(page));
+ if (failed_bio->bi_rw & (1 << BIO_RW))
+ rw = WRITE;
+ else
+ rw = READ;
+
+ BTRFS_I(inode)->io_tree.ops->submit_bio_hook(inode, rw, bio,
+ failrec->last_mirror,
+ bio_flags);
+ return 0;
+}
+
+/*
+ * each time an IO finishes, we do a fast check in the IO failure tree
+ * to see if we need to process or clean up an io_failure_record
+ */
+int btrfs_clean_io_failures(struct inode *inode, u64 start)
+{
+ u64 private;
+ u64 private_failure;
+ struct io_failure_record *failure;
+ int ret;
+
+ private = 0;
+ if (count_range_bits(&BTRFS_I(inode)->io_failure_tree, &private,
+ (u64)-1, 1, EXTENT_DIRTY)) {
+ ret = get_state_private(&BTRFS_I(inode)->io_failure_tree,
+ start, &private_failure);
+ if (ret == 0) {
+ failure = (struct io_failure_record *)(unsigned long)
+ private_failure;
+ set_state_private(&BTRFS_I(inode)->io_failure_tree,
+ failure->start, 0);
+ clear_extent_bits(&BTRFS_I(inode)->io_failure_tree,
+ failure->start,
+ failure->start + failure->len - 1,
+ EXTENT_DIRTY | EXTENT_LOCKED,
+ GFP_NOFS);
+ kfree(failure);
+ }
}
- read_extent_buffer(path->nodes[0], &csum, (unsigned long)item,
- BTRFS_CRC32_SIZE);
- set_state_private(em_tree, start, csum);
-out:
- if (path)
- btrfs_free_path(path);
- mutex_unlock(&root->fs_info->fs_mutex);
- return ret;
+ return 0;
}
-int btrfs_readpage_end_io_hook(struct page *page, u64 start, u64 end)
+/*
+ * when reads are done, we need to check csums to verify the data is correct
+ * if there's a match, we allow the bio to finish. If not, we go through
+ * the io_failure_record routines to find good copies
+ */
+int btrfs_readpage_end_io_hook(struct page *page, u64 start, u64 end,
+ struct extent_state *state)
{
- size_t offset = start - (page->index << PAGE_CACHE_SHIFT);
+ size_t offset = start - ((u64)page->index << PAGE_CACHE_SHIFT);
struct inode *inode = page->mapping->host;
- struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+ struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
char *kaddr;
- u64 private;
+ u64 private = ~(u32)0;
int ret;
struct btrfs_root *root = BTRFS_I(inode)->root;
u32 csum = ~(u32)0;
+ unsigned long flags;
- ret = get_state_private(em_tree, start, &private);
+ if (btrfs_test_opt(root, NODATASUM) ||
+ btrfs_test_flag(inode, NODATASUM))
+ return 0;
+ if (state && state->start == start) {
+ private = state->private;
+ ret = 0;
+ } else {
+ ret = get_state_private(io_tree, start, &private);
+ }
+ local_irq_save(flags);
kaddr = kmap_atomic(page, KM_IRQ0);
if (ret) {
goto zeroit;
goto zeroit;
}
kunmap_atomic(kaddr, KM_IRQ0);
+ local_irq_restore(flags);
+
+ /* if the io failure tree for this inode is non-empty,
+ * check to see if we've recovered from a failed IO
+ */
+ btrfs_clean_io_failures(inode, start);
return 0;
zeroit:
- printk("btrfs csum failed ino %lu off %llu\n",
- page->mapping->host->i_ino, (unsigned long long)start);
+ printk("btrfs csum failed ino %lu off %llu csum %u private %Lu\n",
+ page->mapping->host->i_ino, (unsigned long long)start, csum,
+ private);
memset(kaddr + offset, 1, end - start + 1);
flush_dcache_page(page);
kunmap_atomic(kaddr, KM_IRQ0);
- return 0;
+ local_irq_restore(flags);
+ if (private == 0)
+ return 0;
+ return -EIO;
+}
+
+/*
+ * This creates an orphan entry for the given inode in case something goes
+ * wrong in the middle of an unlink/truncate.
+ */
+int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode)
+{
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ int ret = 0;
+
+ spin_lock(&root->list_lock);
+
+ /* already on the orphan list, we're good */
+ if (!list_empty(&BTRFS_I(inode)->i_orphan)) {
+ spin_unlock(&root->list_lock);
+ return 0;
+ }
+
+ list_add(&BTRFS_I(inode)->i_orphan, &root->orphan_list);
+
+ spin_unlock(&root->list_lock);
+
+ /*
+ * insert an orphan item to track this unlinked/truncated file
+ */
+ ret = btrfs_insert_orphan_item(trans, root, inode->i_ino);
+
+ return ret;
+}
+
+/*
+ * We have done the truncate/delete so we can go ahead and remove the orphan
+ * item for this particular inode.
+ */
+int btrfs_orphan_del(struct btrfs_trans_handle *trans, struct inode *inode)
+{
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ int ret = 0;
+
+ spin_lock(&root->list_lock);
+
+ if (list_empty(&BTRFS_I(inode)->i_orphan)) {
+ spin_unlock(&root->list_lock);
+ return 0;
+ }
+
+ list_del_init(&BTRFS_I(inode)->i_orphan);
+ if (!trans) {
+ spin_unlock(&root->list_lock);
+ return 0;
+ }
+
+ spin_unlock(&root->list_lock);
+
+ ret = btrfs_del_orphan_item(trans, root, inode->i_ino);
+
+ return ret;
+}
+
+/*
+ * this cleans up any orphans that may be left on the list from the last use
+ * of this root.
+ */
+void btrfs_orphan_cleanup(struct btrfs_root *root)
+{
+ struct btrfs_path *path;
+ struct extent_buffer *leaf;
+ struct btrfs_item *item;
+ struct btrfs_key key, found_key;
+ struct btrfs_trans_handle *trans;
+ struct inode *inode;
+ int ret = 0, nr_unlink = 0, nr_truncate = 0;
+
+ /* don't do orphan cleanup if the fs is readonly. */
+ if (root->fs_info->sb->s_flags & MS_RDONLY)
+ return;
+
+ path = btrfs_alloc_path();
+ if (!path)
+ return;
+ path->reada = -1;
+
+ key.objectid = BTRFS_ORPHAN_OBJECTID;
+ btrfs_set_key_type(&key, BTRFS_ORPHAN_ITEM_KEY);
+ key.offset = (u64)-1;
+
+
+ while (1) {
+ ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+ if (ret < 0) {
+ printk(KERN_ERR "Error searching slot for orphan: %d"
+ "\n", ret);
+ break;
+ }
+
+ /*
+ * if ret == 0 means we found what we were searching for, which
+ * is weird, but possible, so only screw with path if we didnt
+ * find the key and see if we have stuff that matches
+ */
+ if (ret > 0) {
+ if (path->slots[0] == 0)
+ break;
+ path->slots[0]--;
+ }
+
+ /* pull out the item */
+ leaf = path->nodes[0];
+ item = btrfs_item_nr(leaf, path->slots[0]);
+ btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
+
+ /* make sure the item matches what we want */
+ if (found_key.objectid != BTRFS_ORPHAN_OBJECTID)
+ break;
+ if (btrfs_key_type(&found_key) != BTRFS_ORPHAN_ITEM_KEY)
+ break;
+
+ /* release the path since we're done with it */
+ btrfs_release_path(root, path);
+
+ /*
+ * this is where we are basically btrfs_lookup, without the
+ * crossing root thing. we store the inode number in the
+ * offset of the orphan item.
+ */
+ inode = btrfs_iget_locked(root->fs_info->sb,
+ found_key.offset, root);
+ if (!inode)
+ break;
+
+ if (inode->i_state & I_NEW) {
+ BTRFS_I(inode)->root = root;
+
+ /* have to set the location manually */
+ BTRFS_I(inode)->location.objectid = inode->i_ino;
+ BTRFS_I(inode)->location.type = BTRFS_INODE_ITEM_KEY;
+ BTRFS_I(inode)->location.offset = 0;
+
+ btrfs_read_locked_inode(inode);
+ unlock_new_inode(inode);
+ }
+
+ /*
+ * add this inode to the orphan list so btrfs_orphan_del does
+ * the proper thing when we hit it
+ */
+ spin_lock(&root->list_lock);
+ list_add(&BTRFS_I(inode)->i_orphan, &root->orphan_list);
+ spin_unlock(&root->list_lock);
+
+ /*
+ * if this is a bad inode, means we actually succeeded in
+ * removing the inode, but not the orphan record, which means
+ * we need to manually delete the orphan since iput will just
+ * do a destroy_inode
+ */
+ if (is_bad_inode(inode)) {
+ trans = btrfs_start_transaction(root, 1);
+ btrfs_orphan_del(trans, inode);
+ btrfs_end_transaction(trans, root);
+ iput(inode);
+ continue;
+ }
+
+ /* if we have links, this was a truncate, lets do that */
+ if (inode->i_nlink) {
+ nr_truncate++;
+ btrfs_truncate(inode);
+ } else {
+ nr_unlink++;
+ }
+
+ /* this will do delete_inode and everything for us */
+ iput(inode);
+ }
+
+ if (nr_unlink)
+ printk(KERN_INFO "btrfs: unlinked %d orphans\n", nr_unlink);
+ if (nr_truncate)
+ printk(KERN_INFO "btrfs: truncated %d orphans\n", nr_truncate);
+
+ btrfs_free_path(path);
}
+/*
+ * read an inode from the btree into the in-memory inode
+ */
void btrfs_read_locked_inode(struct inode *inode)
{
struct btrfs_path *path;
struct extent_buffer *leaf;
struct btrfs_inode_item *inode_item;
- struct btrfs_inode_timespec *tspec;
+ struct btrfs_timespec *tspec;
struct btrfs_root *root = BTRFS_I(inode)->root;
struct btrfs_key location;
u64 alloc_group_block;
path = btrfs_alloc_path();
BUG_ON(!path);
- mutex_lock(&root->fs_info->fs_mutex);
-
memcpy(&location, &BTRFS_I(inode)->location, sizeof(location));
+
ret = btrfs_lookup_inode(NULL, root, path, &location, 0);
if (ret)
goto make_bad;
inode->i_nlink = btrfs_inode_nlink(leaf, inode_item);
inode->i_uid = btrfs_inode_uid(leaf, inode_item);
inode->i_gid = btrfs_inode_gid(leaf, inode_item);
- inode->i_size = btrfs_inode_size(leaf, inode_item);
+ btrfs_i_size_write(inode, btrfs_inode_size(leaf, inode_item));
tspec = btrfs_inode_atime(inode_item);
inode->i_atime.tv_sec = btrfs_timespec_sec(leaf, tspec);
inode->i_ctime.tv_sec = btrfs_timespec_sec(leaf, tspec);
inode->i_ctime.tv_nsec = btrfs_timespec_nsec(leaf, tspec);
- inode->i_blocks = btrfs_inode_nblocks(leaf, inode_item);
- inode->i_generation = btrfs_inode_generation(leaf, inode_item);
+ inode_set_bytes(inode, btrfs_inode_nbytes(leaf, inode_item));
+ BTRFS_I(inode)->generation = btrfs_inode_generation(leaf, inode_item);
+ inode->i_generation = BTRFS_I(inode)->generation;
inode->i_rdev = 0;
rdev = btrfs_inode_rdev(leaf, inode_item);
+ BTRFS_I(inode)->index_cnt = (u64)-1;
+
alloc_group_block = btrfs_inode_block_group(leaf, inode_item);
BTRFS_I(inode)->block_group = btrfs_lookup_block_group(root->fs_info,
alloc_group_block);
-
+ BTRFS_I(inode)->flags = btrfs_inode_flags(leaf, inode_item);
+ if (!BTRFS_I(inode)->block_group) {
+ BTRFS_I(inode)->block_group = btrfs_find_block_group(root,
+ NULL, 0,
+ BTRFS_BLOCK_GROUP_METADATA, 0);
+ }
btrfs_free_path(path);
inode_item = NULL;
- mutex_unlock(&root->fs_info->fs_mutex);
-
switch (inode->i_mode & S_IFMT) {
case S_IFREG:
inode->i_mapping->a_ops = &btrfs_aops;
- BTRFS_I(inode)->extent_tree.ops = &btrfs_extent_map_ops;
+ inode->i_mapping->backing_dev_info = &root->fs_info->bdi;
+ BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
inode->i_fop = &btrfs_file_operations;
inode->i_op = &btrfs_file_inode_operations;
break;
case S_IFLNK:
inode->i_op = &btrfs_symlink_inode_operations;
inode->i_mapping->a_ops = &btrfs_symlink_aops;
+ inode->i_mapping->backing_dev_info = &root->fs_info->bdi;
break;
default:
init_special_inode(inode, inode->i_mode, rdev);
return;
make_bad:
- btrfs_release_path(root, path);
btrfs_free_path(path);
- mutex_unlock(&root->fs_info->fs_mutex);
make_bad_inode(inode);
}
-static void fill_inode_item(struct extent_buffer *leaf,
+/*
+ * given a leaf and an inode, copy the inode fields into the leaf
+ */
+static void fill_inode_item(struct btrfs_trans_handle *trans,
+ struct extent_buffer *leaf,
struct btrfs_inode_item *item,
struct inode *inode)
{
btrfs_set_inode_uid(leaf, item, inode->i_uid);
btrfs_set_inode_gid(leaf, item, inode->i_gid);
- btrfs_set_inode_size(leaf, item, inode->i_size);
+ btrfs_set_inode_size(leaf, item, BTRFS_I(inode)->disk_i_size);
btrfs_set_inode_mode(leaf, item, inode->i_mode);
btrfs_set_inode_nlink(leaf, item, inode->i_nlink);
btrfs_set_timespec_nsec(leaf, btrfs_inode_ctime(item),
inode->i_ctime.tv_nsec);
- btrfs_set_inode_nblocks(leaf, item, inode->i_blocks);
- btrfs_set_inode_generation(leaf, item, inode->i_generation);
+ btrfs_set_inode_nbytes(leaf, item, inode_get_bytes(inode));
+ btrfs_set_inode_generation(leaf, item, BTRFS_I(inode)->generation);
+ btrfs_set_inode_transid(leaf, item, trans->transid);
btrfs_set_inode_rdev(leaf, item, inode->i_rdev);
+ btrfs_set_inode_flags(leaf, item, BTRFS_I(inode)->flags);
btrfs_set_inode_block_group(leaf, item,
BTRFS_I(inode)->block_group->key.objectid);
}
-int btrfs_update_inode(struct btrfs_trans_handle *trans,
+/*
+ * copy everything in the in-memory inode into the btree.
+ */
+int noinline btrfs_update_inode(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
struct inode *inode)
{
inode_item = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_inode_item);
- fill_inode_item(leaf, inode_item, inode);
+ fill_inode_item(trans, leaf, inode_item, inode);
btrfs_mark_buffer_dirty(leaf);
btrfs_set_inode_last_trans(trans, inode);
ret = 0;
failed:
- btrfs_release_path(root, path);
btrfs_free_path(path);
return ret;
}
-static int btrfs_unlink_trans(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- struct inode *dir,
- struct dentry *dentry)
+/*
+ * unlink helper that gets used here in inode.c and in the tree logging
+ * recovery code. It remove a link in a directory with a given name, and
+ * also drops the back refs in the inode to the directory
+ */
+int btrfs_unlink_inode(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct inode *dir, struct inode *inode,
+ const char *name, int name_len)
{
struct btrfs_path *path;
- const char *name = dentry->d_name.name;
- int name_len = dentry->d_name.len;
int ret = 0;
struct extent_buffer *leaf;
struct btrfs_dir_item *di;
struct btrfs_key key;
+ u64 index;
path = btrfs_alloc_path();
if (!path) {
goto err;
btrfs_release_path(root, path);
+ ret = btrfs_del_inode_ref(trans, root, name, name_len,
+ inode->i_ino,
+ dir->i_ino, &index);
+ if (ret) {
+ printk("failed to delete reference to %.*s, "
+ "inode %lu parent %lu\n", name_len, name,
+ inode->i_ino, dir->i_ino);
+ goto err;
+ }
+
di = btrfs_lookup_dir_index_item(trans, root, path, dir->i_ino,
- key.objectid, name, name_len, -1);
+ index, name, name_len, -1);
if (IS_ERR(di)) {
ret = PTR_ERR(di);
goto err;
goto err;
}
ret = btrfs_delete_one_dir_name(trans, root, path, di);
+ btrfs_release_path(root, path);
+
+ ret = btrfs_del_inode_ref_in_log(trans, root, name, name_len,
+ inode, dir->i_ino);
+ BUG_ON(ret != 0 && ret != -ENOENT);
+ if (ret != -ENOENT)
+ BTRFS_I(dir)->log_dirty_trans = trans->transid;
- dentry->d_inode->i_ctime = dir->i_ctime;
+ ret = btrfs_del_dir_entries_in_log(trans, root, name, name_len,
+ dir, index);
+ BUG_ON(ret);
err:
btrfs_free_path(path);
- if (!ret) {
- dir->i_size -= name_len * 2;
- dir->i_mtime = dir->i_ctime = CURRENT_TIME;
- btrfs_update_inode(trans, root, dir);
- drop_nlink(dentry->d_inode);
- ret = btrfs_update_inode(trans, root, dentry->d_inode);
- dir->i_sb->s_dirt = 1;
- }
+ if (ret)
+ goto out;
+
+ btrfs_i_size_write(dir, dir->i_size - name_len * 2);
+ inode->i_ctime = dir->i_mtime = dir->i_ctime = CURRENT_TIME;
+ btrfs_update_inode(trans, root, dir);
+ btrfs_drop_nlink(inode);
+ ret = btrfs_update_inode(trans, root, inode);
+ dir->i_sb->s_dirt = 1;
+out:
return ret;
}
{
struct btrfs_root *root;
struct btrfs_trans_handle *trans;
+ struct inode *inode = dentry->d_inode;
int ret;
- unsigned long nr;
+ unsigned long nr = 0;
root = BTRFS_I(dir)->root;
- mutex_lock(&root->fs_info->fs_mutex);
+
+ ret = btrfs_check_free_space(root, 1, 1);
+ if (ret)
+ goto fail;
+
trans = btrfs_start_transaction(root, 1);
btrfs_set_trans_block_group(trans, dir);
- ret = btrfs_unlink_trans(trans, root, dir, dentry);
+ ret = btrfs_unlink_inode(trans, root, dir, dentry->d_inode,
+ dentry->d_name.name, dentry->d_name.len);
+
+ if (inode->i_nlink == 0)
+ ret = btrfs_orphan_add(trans, inode);
+
nr = trans->blocks_used;
- btrfs_end_transaction(trans, root);
- mutex_unlock(&root->fs_info->fs_mutex);
+ btrfs_end_transaction_throttle(trans, root);
+fail:
btrfs_btree_balance_dirty(root, nr);
-
return ret;
}
static int btrfs_rmdir(struct inode *dir, struct dentry *dentry)
{
struct inode *inode = dentry->d_inode;
- int err;
+ int err = 0;
int ret;
struct btrfs_root *root = BTRFS_I(dir)->root;
- struct btrfs_path *path;
- struct btrfs_key key;
struct btrfs_trans_handle *trans;
- struct btrfs_key found_key;
- int found_type;
- struct extent_buffer *leaf;
- char *goodnames = "..";
- unsigned long nr;
+ unsigned long nr = 0;
- path = btrfs_alloc_path();
- BUG_ON(!path);
- mutex_lock(&root->fs_info->fs_mutex);
- trans = btrfs_start_transaction(root, 1);
+ if (inode->i_size > BTRFS_EMPTY_DIR_SIZE) {
+ return -ENOTEMPTY;
+ }
+
+ ret = btrfs_check_free_space(root, 1, 1);
+ if (ret)
+ goto fail;
+ trans = btrfs_start_transaction(root, 1);
btrfs_set_trans_block_group(trans, dir);
- key.objectid = inode->i_ino;
- key.offset = (u64)-1;
- key.type = (u8)-1;
- while(1) {
- ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
- if (ret < 0) {
- err = ret;
- goto out;
- }
- BUG_ON(ret == 0);
- if (path->slots[0] == 0) {
- err = -ENOENT;
- goto out;
- }
- path->slots[0]--;
- leaf = path->nodes[0];
- btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
- found_type = btrfs_key_type(&found_key);
- if (found_key.objectid != inode->i_ino) {
- err = -ENOENT;
- goto out;
- }
- if ((found_type != BTRFS_DIR_ITEM_KEY &&
- found_type != BTRFS_DIR_INDEX_KEY) ||
- (!btrfs_match_dir_item_name(root, path, goodnames, 2) &&
- !btrfs_match_dir_item_name(root, path, goodnames, 1))) {
- err = -ENOTEMPTY;
- goto out;
- }
- ret = btrfs_del_item(trans, root, path);
- BUG_ON(ret);
- if (found_type == BTRFS_DIR_ITEM_KEY && found_key.offset == 1)
- break;
- btrfs_release_path(root, path);
- }
- ret = 0;
- btrfs_release_path(root, path);
+ err = btrfs_orphan_add(trans, inode);
+ if (err)
+ goto fail_trans;
/* now the directory is empty */
- err = btrfs_unlink_trans(trans, root, dir, dentry);
+ err = btrfs_unlink_inode(trans, root, dir, dentry->d_inode,
+ dentry->d_name.name, dentry->d_name.len);
if (!err) {
- inode->i_size = 0;
+ btrfs_i_size_write(inode, 0);
}
-out:
- btrfs_release_path(root, path);
- btrfs_free_path(path);
- mutex_unlock(&root->fs_info->fs_mutex);
+
+fail_trans:
nr = trans->blocks_used;
- ret = btrfs_end_transaction(trans, root);
+ ret = btrfs_end_transaction_throttle(trans, root);
+fail:
btrfs_btree_balance_dirty(root, nr);
+
if (ret && !err)
err = ret;
return err;
}
-static int btrfs_free_inode(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- struct inode *inode)
+/*
+ * when truncating bytes in a file, it is possible to avoid reading
+ * the leaves that contain only checksum items. This can be the
+ * majority of the IO required to delete a large file, but it must
+ * be done carefully.
+ *
+ * The keys in the level just above the leaves are checked to make sure
+ * the lowest key in a given leaf is a csum key, and starts at an offset
+ * after the new size.
+ *
+ * Then the key for the next leaf is checked to make sure it also has
+ * a checksum item for the same file. If it does, we know our target leaf
+ * contains only checksum items, and it can be safely freed without reading
+ * it.
+ *
+ * This is just an optimization targeted at large files. It may do
+ * nothing. It will return 0 unless things went badly.
+ */
+static noinline int drop_csum_leaves(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct btrfs_path *path,
+ struct inode *inode, u64 new_size)
{
- struct btrfs_path *path;
+ struct btrfs_key key;
int ret;
+ int nritems;
+ struct btrfs_key found_key;
+ struct btrfs_key other_key;
+ struct btrfs_leaf_ref *ref;
+ u64 leaf_gen;
+ u64 leaf_start;
+
+ path->lowest_level = 1;
+ key.objectid = inode->i_ino;
+ key.type = BTRFS_CSUM_ITEM_KEY;
+ key.offset = new_size;
+again:
+ ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
+ if (ret < 0)
+ goto out;
+
+ if (path->nodes[1] == NULL) {
+ ret = 0;
+ goto out;
+ }
+ ret = 0;
+ btrfs_node_key_to_cpu(path->nodes[1], &found_key, path->slots[1]);
+ nritems = btrfs_header_nritems(path->nodes[1]);
+
+ if (!nritems)
+ goto out;
+
+ if (path->slots[1] >= nritems)
+ goto next_node;
+
+ /* did we find a key greater than anything we want to delete? */
+ if (found_key.objectid > inode->i_ino ||
+ (found_key.objectid == inode->i_ino && found_key.type > key.type))
+ goto out;
+
+ /* we check the next key in the node to make sure the leave contains
+ * only checksum items. This comparison doesn't work if our
+ * leaf is the last one in the node
+ */
+ if (path->slots[1] + 1 >= nritems) {
+next_node:
+ /* search forward from the last key in the node, this
+ * will bring us into the next node in the tree
+ */
+ btrfs_node_key_to_cpu(path->nodes[1], &found_key, nritems - 1);
+
+ /* unlikely, but we inc below, so check to be safe */
+ if (found_key.offset == (u64)-1)
+ goto out;
+
+ /* search_forward needs a path with locks held, do the
+ * search again for the original key. It is possible
+ * this will race with a balance and return a path that
+ * we could modify, but this drop is just an optimization
+ * and is allowed to miss some leaves.
+ */
+ btrfs_release_path(root, path);
+ found_key.offset++;
+
+ /* setup a max key for search_forward */
+ other_key.offset = (u64)-1;
+ other_key.type = key.type;
+ other_key.objectid = key.objectid;
+
+ path->keep_locks = 1;
+ ret = btrfs_search_forward(root, &found_key, &other_key,
+ path, 0, 0);
+ path->keep_locks = 0;
+ if (ret || found_key.objectid != key.objectid ||
+ found_key.type != key.type) {
+ ret = 0;
+ goto out;
+ }
+
+ key.offset = found_key.offset;
+ btrfs_release_path(root, path);
+ cond_resched();
+ goto again;
+ }
+
+ /* we know there's one more slot after us in the tree,
+ * read that key so we can verify it is also a checksum item
+ */
+ btrfs_node_key_to_cpu(path->nodes[1], &other_key, path->slots[1] + 1);
+
+ if (found_key.objectid < inode->i_ino)
+ goto next_key;
+
+ if (found_key.type != key.type || found_key.offset < new_size)
+ goto next_key;
+
+ /*
+ * if the key for the next leaf isn't a csum key from this objectid,
+ * we can't be sure there aren't good items inside this leaf.
+ * Bail out
+ */
+ if (other_key.objectid != inode->i_ino || other_key.type != key.type)
+ goto out;
+
+ leaf_start = btrfs_node_blockptr(path->nodes[1], path->slots[1]);
+ leaf_gen = btrfs_node_ptr_generation(path->nodes[1], path->slots[1]);
+ /*
+ * it is safe to delete this leaf, it contains only
+ * csum items from this inode at an offset >= new_size
+ */
+ ret = btrfs_del_leaf(trans, root, path, leaf_start);
+ BUG_ON(ret);
- clear_inode(inode);
+ if (root->ref_cows && leaf_gen < trans->transid) {
+ ref = btrfs_alloc_leaf_ref(root, 0);
+ if (ref) {
+ ref->root_gen = root->root_key.offset;
+ ref->bytenr = leaf_start;
+ ref->owner = 0;
+ ref->generation = leaf_gen;
+ ref->nritems = 0;
+
+ ret = btrfs_add_leaf_ref(root, ref, 0);
+ WARN_ON(ret);
+ btrfs_free_leaf_ref(root, ref);
+ } else {
+ WARN_ON(1);
+ }
+ }
+next_key:
+ btrfs_release_path(root, path);
- path = btrfs_alloc_path();
- BUG_ON(!path);
- ret = btrfs_lookup_inode(trans, root, path,
- &BTRFS_I(inode)->location, -1);
- if (ret > 0)
- ret = -ENOENT;
- if (!ret)
- ret = btrfs_del_item(trans, root, path);
- btrfs_free_path(path);
+ if (other_key.objectid == inode->i_ino &&
+ other_key.type == key.type && other_key.offset > key.offset) {
+ key.offset = other_key.offset;
+ cond_resched();
+ goto again;
+ }
+ ret = 0;
+out:
+ /* fixup any changes we've made to the path */
+ path->lowest_level = 0;
+ path->keep_locks = 0;
+ btrfs_release_path(root, path);
return ret;
}
/*
* this can truncate away extent items, csum items and directory items.
* It starts at a high offset and removes keys until it can't find
- * any higher than i_size.
+ * any higher than new_size
*
* csum items that cross the new i_size are truncated to the new size
* as well.
+ *
+ * min_type is the minimum key type to truncate down to. If set to 0, this
+ * will kill all the items on this inode, including the INODE_ITEM_KEY.
*/
-static int btrfs_truncate_in_trans(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- struct inode *inode)
+noinline int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans,
+ struct btrfs_root *root,
+ struct inode *inode,
+ u64 new_size, u32 min_type)
{
int ret;
struct btrfs_path *path;
u64 extent_start = 0;
u64 extent_num_bytes = 0;
u64 item_end = 0;
+ u64 root_gen = 0;
+ u64 root_owner = 0;
int found_extent;
int del_item;
+ int pending_del_nr = 0;
+ int pending_del_slot = 0;
+ int extent_type = -1;
+ u64 mask = root->sectorsize - 1;
- btrfs_drop_extent_cache(inode, inode->i_size, (u64)-1);
+ if (root->ref_cows)
+ btrfs_drop_extent_cache(inode, new_size & (~mask), (u64)-1, 0);
path = btrfs_alloc_path();
path->reada = -1;
BUG_ON(!path);
key.offset = (u64)-1;
key.type = (u8)-1;
- while(1) {
- btrfs_init_path(path);
- fi = NULL;
- ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
- if (ret < 0) {
+ btrfs_init_path(path);
+
+ ret = drop_csum_leaves(trans, root, path, inode, new_size);
+ BUG_ON(ret);
+
+search_again:
+ ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
+ if (ret < 0) {
+ goto error;
+ }
+ if (ret > 0) {
+ /* there are no items in the tree for us to truncate, we're
+ * done
+ */
+ if (path->slots[0] == 0) {
+ ret = 0;
goto error;
}
- if (ret > 0) {
- BUG_ON(path->slots[0] == 0);
- path->slots[0]--;
- }
+ path->slots[0]--;
+ }
+
+ while(1) {
+ fi = NULL;
leaf = path->nodes[0];
btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
found_type = btrfs_key_type(&found_key);
if (found_key.objectid != inode->i_ino)
break;
- if (found_type != BTRFS_CSUM_ITEM_KEY &&
- found_type != BTRFS_DIR_ITEM_KEY &&
- found_type != BTRFS_DIR_INDEX_KEY &&
- found_type != BTRFS_EXTENT_DATA_KEY)
+ if (found_type < min_type)
break;
item_end = found_key.offset;
if (found_type == BTRFS_EXTENT_DATA_KEY) {
fi = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_file_extent_item);
- if (btrfs_file_extent_type(leaf, fi) !=
- BTRFS_FILE_EXTENT_INLINE) {
+ extent_type = btrfs_file_extent_type(leaf, fi);
+ if (extent_type != BTRFS_FILE_EXTENT_INLINE) {
item_end +=
btrfs_file_extent_num_bytes(leaf, fi);
+ } else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
+ item_end += btrfs_file_extent_inline_len(leaf,
+ fi);
}
+ item_end--;
}
if (found_type == BTRFS_CSUM_ITEM_KEY) {
ret = btrfs_csum_truncate(trans, root, path,
- inode->i_size);
+ new_size);
BUG_ON(ret);
}
- if (item_end < inode->i_size) {
+ if (item_end < new_size) {
if (found_type == BTRFS_DIR_ITEM_KEY) {
found_type = BTRFS_INODE_ITEM_KEY;
} else if (found_type == BTRFS_EXTENT_ITEM_KEY) {
found_type = BTRFS_CSUM_ITEM_KEY;
+ } else if (found_type == BTRFS_EXTENT_DATA_KEY) {
+ found_type = BTRFS_XATTR_ITEM_KEY;
+ } else if (found_type == BTRFS_XATTR_ITEM_KEY) {
+ found_type = BTRFS_INODE_REF_KEY;
} else if (found_type) {
found_type--;
} else {
break;
}
btrfs_set_key_type(&key, found_type);
- continue;
+ goto next;
}
- if (found_key.offset >= inode->i_size)
+ if (found_key.offset >= new_size)
del_item = 1;
else
del_item = 0;
found_extent = 0;
/* FIXME, shrink the extent if the ref count is only 1 */
- if (found_type == BTRFS_EXTENT_DATA_KEY &&
- btrfs_file_extent_type(leaf, fi) !=
- BTRFS_FILE_EXTENT_INLINE) {
+ if (found_type != BTRFS_EXTENT_DATA_KEY)
+ goto delete;
+
+ if (extent_type != BTRFS_FILE_EXTENT_INLINE) {
u64 num_dec;
extent_start = btrfs_file_extent_disk_bytenr(leaf, fi);
if (!del_item) {
u64 orig_num_bytes =
btrfs_file_extent_num_bytes(leaf, fi);
- extent_num_bytes = inode->i_size -
+ extent_num_bytes = new_size -
found_key.offset + root->sectorsize - 1;
+ extent_num_bytes = extent_num_bytes &
+ ~((u64)root->sectorsize - 1);
btrfs_set_file_extent_num_bytes(leaf, fi,
extent_num_bytes);
num_dec = (orig_num_bytes -
- extent_num_bytes) >> 9;
- if (extent_start != 0) {
- inode->i_blocks -= num_dec;
- }
+ extent_num_bytes);
+ if (root->ref_cows && extent_start != 0)
+ inode_sub_bytes(inode, num_dec);
btrfs_mark_buffer_dirty(leaf);
} else {
extent_num_bytes =
btrfs_file_extent_disk_num_bytes(leaf,
fi);
/* FIXME blocksize != 4096 */
- num_dec = btrfs_file_extent_num_bytes(leaf,
- fi) >> 9;
+ num_dec = btrfs_file_extent_num_bytes(leaf, fi);
if (extent_start != 0) {
found_extent = 1;
- inode->i_blocks -= num_dec;
+ if (root->ref_cows)
+ inode_sub_bytes(inode, num_dec);
}
+ root_gen = btrfs_header_generation(leaf);
+ root_owner = btrfs_header_owner(leaf);
+ }
+ } else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
+ /*
+ * we can't truncate inline items that have had
+ * special encodings
+ */
+ if (!del_item &&
+ btrfs_file_extent_compression(leaf, fi) == 0 &&
+ btrfs_file_extent_encryption(leaf, fi) == 0 &&
+ btrfs_file_extent_other_encoding(leaf, fi) == 0) {
+ u32 size = new_size - found_key.offset;
+
+ if (root->ref_cows) {
+ inode_sub_bytes(inode, item_end + 1 -
+ new_size);
+ }
+ size =
+ btrfs_file_extent_calc_inline_size(size);
+ ret = btrfs_truncate_item(trans, root, path,
+ size, 1);
+ BUG_ON(ret);
+ } else if (root->ref_cows) {
+ inode_sub_bytes(inode, item_end + 1 -
+ found_key.offset);
}
}
+delete:
if (del_item) {
- ret = btrfs_del_item(trans, root, path);
- if (ret)
- goto error;
+ if (!pending_del_nr) {
+ /* no pending yet, add ourselves */
+ pending_del_slot = path->slots[0];
+ pending_del_nr = 1;
+ } else if (pending_del_nr &&
+ path->slots[0] + 1 == pending_del_slot) {
+ /* hop on the pending chunk */
+ pending_del_nr++;
+ pending_del_slot = path->slots[0];
+ } else {
+ printk("bad pending slot %d pending_del_nr %d pending_del_slot %d\n", path->slots[0], pending_del_nr, pending_del_slot);
+ }
} else {
break;
}
- btrfs_release_path(root, path);
if (found_extent) {
ret = btrfs_free_extent(trans, root, extent_start,
- extent_num_bytes, 0);
+ extent_num_bytes,
+ leaf->start, root_owner,
+ root_gen, inode->i_ino, 0);
+ BUG_ON(ret);
+ }
+next:
+ if (path->slots[0] == 0) {
+ if (pending_del_nr)
+ goto del_pending;
+ btrfs_release_path(root, path);
+ goto search_again;
+ }
+
+ path->slots[0]--;
+ if (pending_del_nr &&
+ path->slots[0] + 1 != pending_del_slot) {
+ struct btrfs_key debug;
+del_pending:
+ btrfs_item_key_to_cpu(path->nodes[0], &debug,
+ pending_del_slot);
+ ret = btrfs_del_items(trans, root, path,
+ pending_del_slot,
+ pending_del_nr);
BUG_ON(ret);
+ pending_del_nr = 0;
+ btrfs_release_path(root, path);
+ goto search_again;
}
}
ret = 0;
error:
- btrfs_release_path(root, path);
+ if (pending_del_nr) {
+ ret = btrfs_del_items(trans, root, path, pending_del_slot,
+ pending_del_nr);
+ }
btrfs_free_path(path);
inode->i_sb->s_dirt = 1;
return ret;
}
-static int btrfs_cow_one_page(struct inode *inode, struct page *page,
- size_t zero_start)
-{
- char *kaddr;
- int ret = 0;
- struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
- u64 page_start = page->index << PAGE_CACHE_SHIFT;
- u64 page_end = page_start + PAGE_CACHE_SIZE - 1;
-
- set_page_extent_mapped(page);
-
- lock_extent(em_tree, page_start, page_end, GFP_NOFS);
- set_extent_delalloc(&BTRFS_I(inode)->extent_tree, page_start,
- page_end, GFP_NOFS);
- if (zero_start != PAGE_CACHE_SIZE) {
- kaddr = kmap(page);
- memset(kaddr + zero_start, 0, PAGE_CACHE_SIZE - zero_start);
- flush_dcache_page(page);
- kunmap(page);
- }
- set_page_dirty(page);
- unlock_extent(em_tree, page_start, page_end, GFP_NOFS);
-
- return ret;
-}
-
/*
* taken from block_truncate_page, but does cow as it zeros out
* any bytes left in the last page in the file.
{
struct inode *inode = mapping->host;
struct btrfs_root *root = BTRFS_I(inode)->root;
+ struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+ struct btrfs_ordered_extent *ordered;
+ char *kaddr;
u32 blocksize = root->sectorsize;
pgoff_t index = from >> PAGE_CACHE_SHIFT;
unsigned offset = from & (PAGE_CACHE_SIZE-1);
struct page *page;
int ret = 0;
u64 page_start;
+ u64 page_end;
if ((offset & (blocksize - 1)) == 0)
goto out;
- down_read(&root->snap_sem);
ret = -ENOMEM;
+again:
page = grab_cache_page(mapping, index);
if (!page)
goto out;
+
+ page_start = page_offset(page);
+ page_end = page_start + PAGE_CACHE_SIZE - 1;
+
if (!PageUptodate(page)) {
ret = btrfs_readpage(NULL, page);
lock_page(page);
+ if (page->mapping != mapping) {
+ unlock_page(page);
+ page_cache_release(page);
+ goto again;
+ }
if (!PageUptodate(page)) {
ret = -EIO;
- goto out;
+ goto out_unlock;
}
}
- page_start = page->index << PAGE_CACHE_SHIFT;
+ wait_on_page_writeback(page);
+
+ lock_extent(io_tree, page_start, page_end, GFP_NOFS);
+ set_page_extent_mapped(page);
+
+ ordered = btrfs_lookup_ordered_extent(inode, page_start);
+ if (ordered) {
+ unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
+ unlock_page(page);
+ page_cache_release(page);
+ btrfs_start_ordered_extent(inode, ordered, 1);
+ btrfs_put_ordered_extent(ordered);
+ goto again;
+ }
- ret = btrfs_cow_one_page(inode, page, offset);
+ btrfs_set_extent_delalloc(inode, page_start, page_end);
+ ret = 0;
+ if (offset != PAGE_CACHE_SIZE) {
+ kaddr = kmap(page);
+ memset(kaddr + offset, 0, PAGE_CACHE_SIZE - offset);
+ flush_dcache_page(page);
+ kunmap(page);
+ }
+ ClearPageChecked(page);
+ set_page_dirty(page);
+ unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
+out_unlock:
unlock_page(page);
page_cache_release(page);
- up_read(&BTRFS_I(inode)->root->snap_sem);
out:
return ret;
}
attr->ia_valid & ATTR_SIZE && attr->ia_size > inode->i_size) {
struct btrfs_trans_handle *trans;
struct btrfs_root *root = BTRFS_I(inode)->root;
- struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+ struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
u64 mask = root->sectorsize - 1;
- u64 pos = (inode->i_size + mask) & ~mask;
- u64 block_end = attr->ia_size | mask;
+ u64 hole_start = (inode->i_size + mask) & ~mask;
+ u64 block_end = (attr->ia_size + mask) & ~mask;
u64 hole_size;
- u64 alloc_hint;
+ u64 alloc_hint = 0;
- if (attr->ia_size <= pos)
+ if (attr->ia_size <= hole_start)
goto out;
+ err = btrfs_check_free_space(root, 1, 0);
+ if (err)
+ goto fail;
+
btrfs_truncate_page(inode->i_mapping, inode->i_size);
- lock_extent(em_tree, pos, block_end, GFP_NOFS);
- hole_size = (attr->ia_size - pos + mask) & ~mask;
+ hole_size = block_end - hole_start;
+ while(1) {
+ struct btrfs_ordered_extent *ordered;
+ btrfs_wait_ordered_range(inode, hole_start, hole_size);
+
+ lock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS);
+ ordered = btrfs_lookup_ordered_extent(inode, hole_start);
+ if (ordered) {
+ unlock_extent(io_tree, hole_start,
+ block_end - 1, GFP_NOFS);
+ btrfs_put_ordered_extent(ordered);
+ } else {
+ break;
+ }
+ }
- mutex_lock(&root->fs_info->fs_mutex);
trans = btrfs_start_transaction(root, 1);
btrfs_set_trans_block_group(trans, inode);
+ mutex_lock(&BTRFS_I(inode)->extent_mutex);
err = btrfs_drop_extents(trans, root, inode,
- pos, pos + hole_size, pos,
+ hole_start, block_end, hole_start,
&alloc_hint);
- err = btrfs_insert_file_extent(trans, root, inode->i_ino,
- pos, 0, 0, hole_size);
+ if (alloc_hint != EXTENT_MAP_INLINE) {
+ err = btrfs_insert_file_extent(trans, root,
+ inode->i_ino,
+ hole_start, 0, 0,
+ hole_size, 0, hole_size,
+ 0, 0, 0);
+ btrfs_drop_extent_cache(inode, hole_start,
+ (u64)-1, 0);
+ btrfs_check_file(root, inode);
+ }
+ mutex_unlock(&BTRFS_I(inode)->extent_mutex);
btrfs_end_transaction(trans, root);
- mutex_unlock(&root->fs_info->fs_mutex);
- unlock_extent(em_tree, pos, block_end, GFP_NOFS);
+ unlock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS);
if (err)
return err;
}
out:
err = inode_setattr(inode, attr);
+ if (!err && ((attr->ia_valid & ATTR_MODE)))
+ err = btrfs_acl_chmod(inode);
+fail:
return err;
}
+
void btrfs_delete_inode(struct inode *inode)
{
struct btrfs_trans_handle *trans;
truncate_inode_pages(&inode->i_data, 0);
if (is_bad_inode(inode)) {
+ btrfs_orphan_del(NULL, inode);
goto no_delete;
}
+ btrfs_wait_ordered_range(inode, 0, (u64)-1);
- inode->i_size = 0;
- mutex_lock(&root->fs_info->fs_mutex);
+ btrfs_i_size_write(inode, 0);
trans = btrfs_start_transaction(root, 1);
btrfs_set_trans_block_group(trans, inode);
- ret = btrfs_truncate_in_trans(trans, root, inode);
- if (ret)
- goto no_delete_lock;
- ret = btrfs_free_inode(trans, root, inode);
- if (ret)
+ ret = btrfs_truncate_inode_items(trans, root, inode, inode->i_size, 0);
+ if (ret) {
+ btrfs_orphan_del(NULL, inode);
goto no_delete_lock;
+ }
+
+ btrfs_orphan_del(trans, inode);
+
nr = trans->blocks_used;
+ clear_inode(inode);
btrfs_end_transaction(trans, root);
- mutex_unlock(&root->fs_info->fs_mutex);
btrfs_btree_balance_dirty(root, nr);
return;
no_delete_lock:
nr = trans->blocks_used;
btrfs_end_transaction(trans, root);
- mutex_unlock(&root->fs_info->fs_mutex);
btrfs_btree_balance_dirty(root, nr);
no_delete:
clear_inode(inode);
struct btrfs_dir_item *di;
struct btrfs_path *path;
struct btrfs_root *root = BTRFS_I(dir)->root;
- int ret;
+ int ret = 0;
path = btrfs_alloc_path();
BUG_ON(!path);
+
di = btrfs_lookup_dir_item(NULL, root, path, dir->i_ino, name,
namelen, 0);
+ if (IS_ERR(di))
+ ret = PTR_ERR(di);
if (!di || IS_ERR(di)) {
- location->objectid = 0;
- ret = 0;
- goto out;
+ goto out_err;
}
btrfs_dir_item_key_to_cpu(path->nodes[0], di, location);
out:
- btrfs_release_path(root, path);
btrfs_free_path(path);
return ret;
+out_err:
+ location->objectid = 0;
+ goto out;
}
/*
struct btrfs_root **sub_root,
struct dentry *dentry)
{
- struct btrfs_path *path;
struct btrfs_root_item *ri;
if (btrfs_key_type(location) != BTRFS_ROOT_ITEM_KEY)
if (location->objectid == BTRFS_ROOT_TREE_OBJECTID)
return 0;
- path = btrfs_alloc_path();
- BUG_ON(!path);
- mutex_lock(&root->fs_info->fs_mutex);
-
*sub_root = btrfs_read_fs_root(root->fs_info, location,
dentry->d_name.name,
dentry->d_name.len);
btrfs_set_key_type(location, BTRFS_INODE_ITEM_KEY);
location->offset = 0;
- btrfs_free_path(path);
- mutex_unlock(&root->fs_info->fs_mutex);
return 0;
}
+static noinline void init_btrfs_i(struct inode *inode)
+{
+ struct btrfs_inode *bi = BTRFS_I(inode);
+
+ bi->i_acl = NULL;
+ bi->i_default_acl = NULL;
+
+ bi->generation = 0;
+ bi->last_trans = 0;
+ bi->logged_trans = 0;
+ bi->delalloc_bytes = 0;
+ bi->disk_i_size = 0;
+ bi->flags = 0;
+ bi->index_cnt = (u64)-1;
+ bi->log_dirty_trans = 0;
+ extent_map_tree_init(&BTRFS_I(inode)->extent_tree, GFP_NOFS);
+ extent_io_tree_init(&BTRFS_I(inode)->io_tree,
+ inode->i_mapping, GFP_NOFS);
+ extent_io_tree_init(&BTRFS_I(inode)->io_failure_tree,
+ inode->i_mapping, GFP_NOFS);
+ INIT_LIST_HEAD(&BTRFS_I(inode)->delalloc_inodes);
+ btrfs_ordered_inode_tree_init(&BTRFS_I(inode)->ordered_tree);
+ mutex_init(&BTRFS_I(inode)->csum_mutex);
+ mutex_init(&BTRFS_I(inode)->extent_mutex);
+ mutex_init(&BTRFS_I(inode)->log_mutex);
+}
+
static int btrfs_init_locked_inode(struct inode *inode, void *p)
{
struct btrfs_iget_args *args = p;
inode->i_ino = args->ino;
+ init_btrfs_i(inode);
BTRFS_I(inode)->root = args->root;
- extent_map_tree_init(&BTRFS_I(inode)->extent_tree,
- inode->i_mapping, GFP_NOFS);
return 0;
}
args->root == BTRFS_I(inode)->root);
}
+struct inode *btrfs_ilookup(struct super_block *s, u64 objectid,
+ struct btrfs_root *root, int wait)
+{
+ struct inode *inode;
+ struct btrfs_iget_args args;
+ args.ino = objectid;
+ args.root = root;
+
+ if (wait) {
+ inode = ilookup5(s, objectid, btrfs_find_actor,
+ (void *)&args);
+ } else {
+ inode = ilookup5_nowait(s, objectid, btrfs_find_actor,
+ (void *)&args);
+ }
+ return inode;
+}
+
struct inode *btrfs_iget_locked(struct super_block *s, u64 objectid,
struct btrfs_root *root)
{
return inode;
}
+/* Get an inode object given its location and corresponding root.
+ * Returns in *is_new if the inode was read from disk
+ */
+struct inode *btrfs_iget(struct super_block *s, struct btrfs_key *location,
+ struct btrfs_root *root, int *is_new)
+{
+ struct inode *inode;
+
+ inode = btrfs_iget_locked(s, location->objectid, root);
+ if (!inode)
+ return ERR_PTR(-EACCES);
+
+ if (inode->i_state & I_NEW) {
+ BTRFS_I(inode)->root = root;
+ memcpy(&BTRFS_I(inode)->location, location, sizeof(*location));
+ btrfs_read_locked_inode(inode);
+ unlock_new_inode(inode);
+ if (is_new)
+ *is_new = 1;
+ } else {
+ if (is_new)
+ *is_new = 0;
+ }
+
+ return inode;
+}
+
static struct dentry *btrfs_lookup(struct inode *dir, struct dentry *dentry,
struct nameidata *nd)
{
struct btrfs_root *root = bi->root;
struct btrfs_root *sub_root = root;
struct btrfs_key location;
- int ret;
+ int ret, new, do_orphan = 0;
if (dentry->d_name.len > BTRFS_NAME_LEN)
return ERR_PTR(-ENAMETOOLONG);
- mutex_lock(&root->fs_info->fs_mutex);
ret = btrfs_inode_by_name(dir, dentry, &location);
- mutex_unlock(&root->fs_info->fs_mutex);
if (ret < 0)
return ERR_PTR(ret);
return ERR_PTR(ret);
if (ret > 0)
return ERR_PTR(-ENOENT);
- inode = btrfs_iget_locked(dir->i_sb, location.objectid,
- sub_root);
- if (!inode)
- return ERR_PTR(-EACCES);
- if (inode->i_state & I_NEW) {
- /* the inode and parent dir are two different roots */
- if (sub_root != root) {
- igrab(inode);
- sub_root->inode = inode;
- }
- BTRFS_I(inode)->root = sub_root;
- memcpy(&BTRFS_I(inode)->location, &location,
- sizeof(location));
- btrfs_read_locked_inode(inode);
- unlock_new_inode(inode);
+ inode = btrfs_iget(dir->i_sb, &location, sub_root, &new);
+ if (IS_ERR(inode))
+ return ERR_CAST(inode);
+
+ /* the inode and parent dir are two different roots */
+ if (new && root != sub_root) {
+ igrab(inode);
+ sub_root->inode = inode;
+ do_orphan = 1;
}
}
+
+ if (unlikely(do_orphan))
+ btrfs_orphan_cleanup(sub_root);
+
return d_splice_alias(inode, dentry);
}
DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
};
-static int btrfs_readdir(struct file *filp, void *dirent, filldir_t filldir)
+static int btrfs_real_readdir(struct file *filp, void *dirent,
+ filldir_t filldir)
{
- struct inode *inode = filp->f_path.dentry->d_inode;
+ struct inode *inode = filp->f_dentry->d_inode;
struct btrfs_root *root = BTRFS_I(inode)->root;
struct btrfs_item *item;
struct btrfs_dir_item *di;
if (root->fs_info->tree_root == root)
key_type = BTRFS_DIR_ITEM_KEY;
- mutex_lock(&root->fs_info->fs_mutex);
- key.objectid = inode->i_ino;
- btrfs_set_key_type(&key, key_type);
- key.offset = filp->f_pos;
+ /* special case for "." */
+ if (filp->f_pos == 0) {
+ over = filldir(dirent, ".", 1,
+ 1, inode->i_ino,
+ DT_DIR);
+ if (over)
+ return 0;
+ filp->f_pos = 1;
+ }
+ /* special case for .., just use the back ref */
+ if (filp->f_pos == 1) {
+ u64 pino = parent_ino(filp->f_path.dentry);
+ over = filldir(dirent, "..", 2,
+ 2, pino, DT_DIR);
+ if (over)
+ return 0;
+ filp->f_pos = 2;
+ }
path = btrfs_alloc_path();
path->reada = 2;
+
+ btrfs_set_key_type(&key, key_type);
+ key.offset = filp->f_pos;
+ key.objectid = inode->i_ino;
+
ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
if (ret < 0)
goto err;
advance = 0;
- while(1) {
+
+ while (1) {
leaf = path->nodes[0];
nritems = btrfs_header_nritems(leaf);
slot = path->slots[0];
if (advance || slot >= nritems) {
- if (slot >= nritems -1) {
+ if (slot >= nritems - 1) {
ret = btrfs_next_leaf(root, path);
if (ret)
break;
continue;
filp->f_pos = found_key.offset;
- advance = 1;
+
di = btrfs_item_ptr(leaf, slot, struct btrfs_dir_item);
di_cur = 0;
di_total = btrfs_item_size(leaf, item);
- while(di_cur < di_total) {
+
+ while (di_cur < di_total) {
struct btrfs_key location;
name_len = btrfs_dir_name_len(leaf, di);
- if (name_len < 32) {
+ if (name_len <= sizeof(tmp_name)) {
name_ptr = tmp_name;
} else {
name_ptr = kmalloc(name_len, GFP_NOFS);
- BUG_ON(!name_ptr);
+ if (!name_ptr) {
+ ret = -ENOMEM;
+ goto err;
+ }
}
read_extent_buffer(leaf, name_ptr,
(unsigned long)(di + 1), name_len);
d_type = btrfs_filetype_table[btrfs_dir_type(leaf, di)];
btrfs_dir_item_key_to_cpu(leaf, di, &location);
+ over = filldir(dirent, name_ptr, name_len,
+ found_key.offset, location.objectid,
+ d_type);
+
+ if (name_ptr != tmp_name)
+ kfree(name_ptr);
+
+ if (over)
+ goto nopos;
+
+ di_len = btrfs_dir_name_len(leaf, di) +
+ btrfs_dir_data_len(leaf, di) + sizeof(*di);
+ di_cur += di_len;
+ di = (struct btrfs_dir_item *)((char *)di + di_len);
+ }
+ }
+
+ /* Reached end of directory/root. Bump pos past the last item. */
+ if (key_type == BTRFS_DIR_INDEX_KEY)
+ filp->f_pos = INT_LIMIT(typeof(filp->f_pos));
+ else
+ filp->f_pos++;
+nopos:
+ ret = 0;
+err:
+ btrfs_free_path(path);
+ return ret;
+}
+
+int btrfs_write_inode(struct inode *inode, int wait)
+{
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ struct btrfs_trans_handle *trans;
+ int ret = 0;
+
+ if (root->fs_info->closing > 1)
+ return 0;
+
+ if (wait) {
+ trans = btrfs_join_transaction(root, 1);
+ btrfs_set_trans_block_group(trans, inode);
+ ret = btrfs_commit_transaction(trans, root);
+ }
+ return ret;
+}
+
+/*
+ * This is somewhat expensive, updating the tree every time the
+ * inode changes. But, it is most likely to find the inode in cache.
+ * FIXME, needs more benchmarking...there are no reasons other than performance
+ * to keep or drop this code.
+ */
+void btrfs_dirty_inode(struct inode *inode)
+{
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ struct btrfs_trans_handle *trans;
+
+ trans = btrfs_join_transaction(root, 1);
+ btrfs_set_trans_block_group(trans, inode);
+ btrfs_update_inode(trans, root, inode);
+ btrfs_end_transaction(trans, root);
+}
+
+/*
+ * find the highest existing sequence number in a directory
+ * and then set the in-memory index_cnt variable to reflect
+ * free sequence numbers
+ */
+static int btrfs_set_inode_index_count(struct inode *inode)
+{
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ struct btrfs_key key, found_key;
+ struct btrfs_path *path;
+ struct extent_buffer *leaf;
+ int ret;
+
+ key.objectid = inode->i_ino;
+ btrfs_set_key_type(&key, BTRFS_DIR_INDEX_KEY);
+ key.offset = (u64)-1;
+
+ path = btrfs_alloc_path();
+ if (!path)
+ return -ENOMEM;
+
+ ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+ if (ret < 0)
+ goto out;
+ /* FIXME: we should be able to handle this */
+ if (ret == 0)
+ goto out;
+ ret = 0;
- over = filldir(dirent, name_ptr, name_len,
- found_key.offset,
- location.objectid,
- d_type);
+ /*
+ * MAGIC NUMBER EXPLANATION:
+ * since we search a directory based on f_pos we have to start at 2
+ * since '.' and '..' have f_pos of 0 and 1 respectively, so everybody
+ * else has to start at 2
+ */
+ if (path->slots[0] == 0) {
+ BTRFS_I(inode)->index_cnt = 2;
+ goto out;
+ }
- if (name_ptr != tmp_name)
- kfree(name_ptr);
+ path->slots[0]--;
- if (over)
- goto nopos;
- di_len = btrfs_dir_name_len(leaf, di) + sizeof(*di);
- di_cur += di_len;
- di = (struct btrfs_dir_item *)((char *)di + di_len);
- }
+ leaf = path->nodes[0];
+ btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
+
+ if (found_key.objectid != inode->i_ino ||
+ btrfs_key_type(&found_key) != BTRFS_DIR_INDEX_KEY) {
+ BTRFS_I(inode)->index_cnt = 2;
+ goto out;
}
- filp->f_pos++;
-nopos:
- ret = 0;
-err:
- btrfs_release_path(root, path);
+
+ BTRFS_I(inode)->index_cnt = found_key.offset + 1;
+out:
btrfs_free_path(path);
- mutex_unlock(&root->fs_info->fs_mutex);
return ret;
}
-int btrfs_write_inode(struct inode *inode, int wait)
+/*
+ * helper to find a free sequence number in a given directory. This current
+ * code is very simple, later versions will do smarter things in the btree
+ */
+static int btrfs_set_inode_index(struct inode *dir, struct inode *inode,
+ u64 *index)
{
- struct btrfs_root *root = BTRFS_I(inode)->root;
- struct btrfs_trans_handle *trans;
int ret = 0;
- if (wait) {
- mutex_lock(&root->fs_info->fs_mutex);
- trans = btrfs_start_transaction(root, 1);
- btrfs_set_trans_block_group(trans, inode);
- ret = btrfs_commit_transaction(trans, root);
- mutex_unlock(&root->fs_info->fs_mutex);
+ if (BTRFS_I(dir)->index_cnt == (u64)-1) {
+ ret = btrfs_set_inode_index_count(dir);
+ if (ret) {
+ return ret;
+ }
}
- return ret;
-}
-/*
- * This is somewhat expensive, updating the tree every time the
- * inode changes. But, it is most likely to find the inode in cache.
- * FIXME, needs more benchmarking...there are no reasons other than performance
- * to keep or drop this code.
- */
-void btrfs_dirty_inode(struct inode *inode)
-{
- struct btrfs_root *root = BTRFS_I(inode)->root;
- struct btrfs_trans_handle *trans;
+ *index = BTRFS_I(dir)->index_cnt;
+ BTRFS_I(dir)->index_cnt++;
- mutex_lock(&root->fs_info->fs_mutex);
- trans = btrfs_start_transaction(root, 1);
- btrfs_set_trans_block_group(trans, inode);
- btrfs_update_inode(trans, root, inode);
- btrfs_end_transaction(trans, root);
- mutex_unlock(&root->fs_info->fs_mutex);
+ return ret;
}
static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans,
struct btrfs_root *root,
+ struct inode *dir,
+ const char *name, int name_len,
+ u64 ref_objectid,
u64 objectid,
struct btrfs_block_group_cache *group,
- int mode)
+ int mode, u64 *index)
{
struct inode *inode;
struct btrfs_inode_item *inode_item;
+ struct btrfs_block_group_cache *new_inode_group;
struct btrfs_key *location;
struct btrfs_path *path;
+ struct btrfs_inode_ref *ref;
+ struct btrfs_key key[2];
+ u32 sizes[2];
+ unsigned long ptr;
int ret;
int owner;
if (!inode)
return ERR_PTR(-ENOMEM);
- extent_map_tree_init(&BTRFS_I(inode)->extent_tree,
- inode->i_mapping, GFP_NOFS);
+ if (dir) {
+ ret = btrfs_set_inode_index(dir, inode, index);
+ if (ret)
+ return ERR_PTR(ret);
+ }
+ /*
+ * index_cnt is ignored for everything but a dir,
+ * btrfs_get_inode_index_count has an explanation for the magic
+ * number
+ */
+ init_btrfs_i(inode);
+ BTRFS_I(inode)->index_cnt = 2;
BTRFS_I(inode)->root = root;
+ BTRFS_I(inode)->generation = trans->transid;
if (mode & S_IFDIR)
owner = 0;
else
owner = 1;
- group = btrfs_find_block_group(root, group, 0, 0, owner);
- BTRFS_I(inode)->block_group = group;
+ new_inode_group = btrfs_find_block_group(root, group, 0,
+ BTRFS_BLOCK_GROUP_METADATA, owner);
+ if (!new_inode_group) {
+ printk("find_block group failed\n");
+ new_inode_group = group;
+ }
+ BTRFS_I(inode)->block_group = new_inode_group;
- ret = btrfs_insert_empty_inode(trans, root, path, objectid);
- if (ret)
+ key[0].objectid = objectid;
+ btrfs_set_key_type(&key[0], BTRFS_INODE_ITEM_KEY);
+ key[0].offset = 0;
+
+ key[1].objectid = objectid;
+ btrfs_set_key_type(&key[1], BTRFS_INODE_REF_KEY);
+ key[1].offset = ref_objectid;
+
+ sizes[0] = sizeof(struct btrfs_inode_item);
+ sizes[1] = name_len + sizeof(*ref);
+
+ ret = btrfs_insert_empty_items(trans, root, path, key, sizes, 2);
+ if (ret != 0)
goto fail;
+ if (objectid > root->highest_inode)
+ root->highest_inode = objectid;
+
inode->i_uid = current->fsuid;
inode->i_gid = current->fsgid;
inode->i_mode = mode;
inode->i_ino = objectid;
- inode->i_blocks = 0;
+ inode_set_bytes(inode, 0);
inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
inode_item = btrfs_item_ptr(path->nodes[0], path->slots[0],
struct btrfs_inode_item);
- fill_inode_item(path->nodes[0], inode_item, inode);
+ fill_inode_item(trans, path->nodes[0], inode_item, inode);
+
+ ref = btrfs_item_ptr(path->nodes[0], path->slots[0] + 1,
+ struct btrfs_inode_ref);
+ btrfs_set_inode_ref_name_len(path->nodes[0], ref, name_len);
+ btrfs_set_inode_ref_index(path->nodes[0], ref, *index);
+ ptr = (unsigned long)(ref + 1);
+ write_extent_buffer(path->nodes[0], name, ptr, name_len);
+
btrfs_mark_buffer_dirty(path->nodes[0]);
btrfs_free_path(path);
insert_inode_hash(inode);
return inode;
fail:
+ if (dir)
+ BTRFS_I(dir)->index_cnt--;
btrfs_free_path(path);
return ERR_PTR(ret);
}
return btrfs_type_by_mode[(inode->i_mode & S_IFMT) >> S_SHIFT];
}
-static int btrfs_add_link(struct btrfs_trans_handle *trans,
- struct dentry *dentry, struct inode *inode)
+/*
+ * utility function to add 'inode' into 'parent_inode' with
+ * a give name and a given sequence number.
+ * if 'add_backref' is true, also insert a backref from the
+ * inode to the parent directory.
+ */
+int btrfs_add_link(struct btrfs_trans_handle *trans,
+ struct inode *parent_inode, struct inode *inode,
+ const char *name, int name_len, int add_backref, u64 index)
{
int ret;
struct btrfs_key key;
- struct btrfs_root *root = BTRFS_I(dentry->d_parent->d_inode)->root;
- struct inode *parent_inode;
+ struct btrfs_root *root = BTRFS_I(parent_inode)->root;
key.objectid = inode->i_ino;
btrfs_set_key_type(&key, BTRFS_INODE_ITEM_KEY);
key.offset = 0;
- ret = btrfs_insert_dir_item(trans, root,
- dentry->d_name.name, dentry->d_name.len,
- dentry->d_parent->d_inode->i_ino,
- &key, btrfs_inode_type(inode));
+ ret = btrfs_insert_dir_item(trans, root, name, name_len,
+ parent_inode->i_ino,
+ &key, btrfs_inode_type(inode),
+ index);
if (ret == 0) {
- parent_inode = dentry->d_parent->d_inode;
- parent_inode->i_size += dentry->d_name.len * 2;
+ if (add_backref) {
+ ret = btrfs_insert_inode_ref(trans, root,
+ name, name_len,
+ inode->i_ino,
+ parent_inode->i_ino,
+ index);
+ }
+ btrfs_i_size_write(parent_inode, parent_inode->i_size +
+ name_len * 2);
parent_inode->i_mtime = parent_inode->i_ctime = CURRENT_TIME;
- ret = btrfs_update_inode(trans, root,
- dentry->d_parent->d_inode);
+ ret = btrfs_update_inode(trans, root, parent_inode);
}
return ret;
}
static int btrfs_add_nondir(struct btrfs_trans_handle *trans,
- struct dentry *dentry, struct inode *inode)
+ struct dentry *dentry, struct inode *inode,
+ int backref, u64 index)
{
- int err = btrfs_add_link(trans, dentry, inode);
+ int err = btrfs_add_link(trans, dentry->d_parent->d_inode,
+ inode, dentry->d_name.name,
+ dentry->d_name.len, backref, index);
if (!err) {
d_instantiate(dentry, inode);
return 0;
{
struct btrfs_trans_handle *trans;
struct btrfs_root *root = BTRFS_I(dir)->root;
- struct inode *inode;
+ struct inode *inode = NULL;
int err;
int drop_inode = 0;
u64 objectid;
- unsigned long nr;
+ unsigned long nr = 0;
+ u64 index = 0;
if (!new_valid_dev(rdev))
return -EINVAL;
- mutex_lock(&root->fs_info->fs_mutex);
+ err = btrfs_check_free_space(root, 1, 0);
+ if (err)
+ goto fail;
+
trans = btrfs_start_transaction(root, 1);
btrfs_set_trans_block_group(trans, dir);
goto out_unlock;
}
- inode = btrfs_new_inode(trans, root, objectid,
- BTRFS_I(dir)->block_group, mode);
+ inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
+ dentry->d_name.len,
+ dentry->d_parent->d_inode->i_ino, objectid,
+ BTRFS_I(dir)->block_group, mode, &index);
err = PTR_ERR(inode);
if (IS_ERR(inode))
goto out_unlock;
+ err = btrfs_init_acl(inode, dir);
+ if (err) {
+ drop_inode = 1;
+ goto out_unlock;
+ }
+
btrfs_set_trans_block_group(trans, inode);
- err = btrfs_add_nondir(trans, dentry, inode);
+ err = btrfs_add_nondir(trans, dentry, inode, 0, index);
if (err)
drop_inode = 1;
else {
btrfs_update_inode_block_group(trans, dir);
out_unlock:
nr = trans->blocks_used;
- btrfs_end_transaction(trans, root);
- mutex_unlock(&root->fs_info->fs_mutex);
-
+ btrfs_end_transaction_throttle(trans, root);
+fail:
if (drop_inode) {
inode_dec_link_count(inode);
iput(inode);
{
struct btrfs_trans_handle *trans;
struct btrfs_root *root = BTRFS_I(dir)->root;
- struct inode *inode;
+ struct inode *inode = NULL;
int err;
int drop_inode = 0;
- unsigned long nr;
+ unsigned long nr = 0;
u64 objectid;
+ u64 index = 0;
- mutex_lock(&root->fs_info->fs_mutex);
+ err = btrfs_check_free_space(root, 1, 0);
+ if (err)
+ goto fail;
trans = btrfs_start_transaction(root, 1);
btrfs_set_trans_block_group(trans, dir);
goto out_unlock;
}
- inode = btrfs_new_inode(trans, root, objectid,
- BTRFS_I(dir)->block_group, mode);
+ inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
+ dentry->d_name.len,
+ dentry->d_parent->d_inode->i_ino,
+ objectid, BTRFS_I(dir)->block_group, mode,
+ &index);
err = PTR_ERR(inode);
if (IS_ERR(inode))
goto out_unlock;
+ err = btrfs_init_acl(inode, dir);
+ if (err) {
+ drop_inode = 1;
+ goto out_unlock;
+ }
+
btrfs_set_trans_block_group(trans, inode);
- err = btrfs_add_nondir(trans, dentry, inode);
+ err = btrfs_add_nondir(trans, dentry, inode, 0, index);
if (err)
drop_inode = 1;
else {
inode->i_mapping->a_ops = &btrfs_aops;
+ inode->i_mapping->backing_dev_info = &root->fs_info->bdi;
inode->i_fop = &btrfs_file_operations;
inode->i_op = &btrfs_file_inode_operations;
- extent_map_tree_init(&BTRFS_I(inode)->extent_tree,
- inode->i_mapping, GFP_NOFS);
- BTRFS_I(inode)->extent_tree.ops = &btrfs_extent_map_ops;
+ BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
}
dir->i_sb->s_dirt = 1;
btrfs_update_inode_block_group(trans, inode);
btrfs_update_inode_block_group(trans, dir);
out_unlock:
nr = trans->blocks_used;
- btrfs_end_transaction(trans, root);
- mutex_unlock(&root->fs_info->fs_mutex);
-
+ btrfs_end_transaction_throttle(trans, root);
+fail:
if (drop_inode) {
inode_dec_link_count(inode);
iput(inode);
struct btrfs_trans_handle *trans;
struct btrfs_root *root = BTRFS_I(dir)->root;
struct inode *inode = old_dentry->d_inode;
- unsigned long nr;
+ u64 index;
+ unsigned long nr = 0;
int err;
int drop_inode = 0;
if (inode->i_nlink == 0)
return -ENOENT;
- inc_nlink(inode);
- mutex_lock(&root->fs_info->fs_mutex);
+ btrfs_inc_nlink(inode);
+ err = btrfs_check_free_space(root, 1, 0);
+ if (err)
+ goto fail;
+ err = btrfs_set_inode_index(dir, inode, &index);
+ if (err)
+ goto fail;
+
trans = btrfs_start_transaction(root, 1);
btrfs_set_trans_block_group(trans, dir);
atomic_inc(&inode->i_count);
- err = btrfs_add_nondir(trans, dentry, inode);
+
+ err = btrfs_add_nondir(trans, dentry, inode, 1, index);
if (err)
drop_inode = 1;
drop_inode = 1;
nr = trans->blocks_used;
- btrfs_end_transaction(trans, root);
- mutex_unlock(&root->fs_info->fs_mutex);
-
+ btrfs_end_transaction_throttle(trans, root);
+fail:
if (drop_inode) {
inode_dec_link_count(inode);
iput(inode);
return err;
}
-static int btrfs_make_empty_dir(struct btrfs_trans_handle *trans,
- struct btrfs_root *root,
- u64 objectid, u64 dirid)
-{
- int ret;
- char buf[2];
- struct btrfs_key key;
-
- buf[0] = '.';
- buf[1] = '.';
-
- key.objectid = objectid;
- key.offset = 0;
- btrfs_set_key_type(&key, BTRFS_INODE_ITEM_KEY);
-
- ret = btrfs_insert_dir_item(trans, root, buf, 1, objectid,
- &key, BTRFS_FT_DIR);
- if (ret)
- goto error;
-
- key.objectid = dirid;
- ret = btrfs_insert_dir_item(trans, root, buf, 2, objectid,
- &key, BTRFS_FT_DIR);
- if (ret)
- goto error;
-error:
- return ret;
-}
-
static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode)
{
- struct inode *inode;
+ struct inode *inode = NULL;
struct btrfs_trans_handle *trans;
struct btrfs_root *root = BTRFS_I(dir)->root;
int err = 0;
int drop_on_err = 0;
- u64 objectid;
+ u64 objectid = 0;
+ u64 index = 0;
unsigned long nr = 1;
- mutex_lock(&root->fs_info->fs_mutex);
+ err = btrfs_check_free_space(root, 1, 0);
+ if (err)
+ goto out_unlock;
+
trans = btrfs_start_transaction(root, 1);
btrfs_set_trans_block_group(trans, dir);
goto out_unlock;
}
- inode = btrfs_new_inode(trans, root, objectid,
- BTRFS_I(dir)->block_group, S_IFDIR | mode);
+ inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
+ dentry->d_name.len,
+ dentry->d_parent->d_inode->i_ino, objectid,
+ BTRFS_I(dir)->block_group, S_IFDIR | mode,
+ &index);
if (IS_ERR(inode)) {
err = PTR_ERR(inode);
goto out_fail;
}
drop_on_err = 1;
- inode->i_op = &btrfs_dir_inode_operations;
- inode->i_fop = &btrfs_dir_file_operations;
- btrfs_set_trans_block_group(trans, inode);
- err = btrfs_make_empty_dir(trans, root, inode->i_ino, dir->i_ino);
+ err = btrfs_init_acl(inode, dir);
if (err)
goto out_fail;
- inode->i_size = 6;
+ inode->i_op = &btrfs_dir_inode_operations;
+ inode->i_fop = &btrfs_dir_file_operations;
+ btrfs_set_trans_block_group(trans, inode);
+
+ btrfs_i_size_write(inode, 0);
err = btrfs_update_inode(trans, root, inode);
if (err)
goto out_fail;
- err = btrfs_add_link(trans, dentry, inode);
+ err = btrfs_add_link(trans, dentry->d_parent->d_inode,
+ inode, dentry->d_name.name,
+ dentry->d_name.len, 0, index);
if (err)
goto out_fail;
out_fail:
nr = trans->blocks_used;
- btrfs_end_transaction(trans, root);
+ btrfs_end_transaction_throttle(trans, root);
out_unlock:
- mutex_unlock(&root->fs_info->fs_mutex);
if (drop_on_err)
iput(inode);
btrfs_btree_balance_dirty(root, nr);
return err;
}
+/* helper for btfs_get_extent. Given an existing extent in the tree,
+ * and an extent that you want to insert, deal with overlap and insert
+ * the new extent into the tree.
+ */
+static int merge_extent_mapping(struct extent_map_tree *em_tree,
+ struct extent_map *existing,
+ struct extent_map *em,
+ u64 map_start, u64 map_len)
+{
+ u64 start_diff;
+
+ BUG_ON(map_start < em->start || map_start >= extent_map_end(em));
+ start_diff = map_start - em->start;
+ em->start = map_start;
+ em->len = map_len;
+ if (em->block_start < EXTENT_MAP_LAST_BYTE &&
+ !test_bit(EXTENT_FLAG_COMPRESSED, &em->flags)) {
+ em->block_start += start_diff;
+ em->block_len -= start_diff;
+ }
+ return add_extent_mapping(em_tree, em);
+}
+
+static noinline int uncompress_inline(struct btrfs_path *path,
+ struct inode *inode, struct page *page,
+ size_t pg_offset, u64 extent_offset,
+ struct btrfs_file_extent_item *item)
+{
+ int ret;
+ struct extent_buffer *leaf = path->nodes[0];
+ char *tmp;
+ size_t max_size;
+ unsigned long inline_size;
+ unsigned long ptr;
+
+ WARN_ON(pg_offset != 0);
+ max_size = btrfs_file_extent_ram_bytes(leaf, item);
+ inline_size = btrfs_file_extent_inline_item_len(leaf,
+ btrfs_item_nr(leaf, path->slots[0]));
+ tmp = kmalloc(inline_size, GFP_NOFS);
+ ptr = btrfs_file_extent_inline_start(item);
+
+ read_extent_buffer(leaf, tmp, ptr, inline_size);
+
+ max_size = min(PAGE_CACHE_SIZE, max_size);
+ ret = btrfs_zlib_decompress(tmp, page, extent_offset,
+ inline_size, max_size);
+ if (ret) {
+ char *kaddr = kmap_atomic(page, KM_USER0);
+ unsigned long copy_size = min_t(u64,
+ PAGE_CACHE_SIZE - pg_offset,
+ max_size - extent_offset);
+ memset(kaddr + pg_offset, 0, copy_size);
+ kunmap_atomic(kaddr, KM_USER0);
+ }
+ kfree(tmp);
+ return 0;
+}
+
+/*
+ * a bit scary, this does extent mapping from logical file offset to the disk.
+ * the ugly parts come from merging extents from the disk with the
+ * in-ram representation. This gets more complex because of the data=ordered code,
+ * where the in-ram extents might be locked pending data=ordered completion.
+ *
+ * This also copies inline extents directly into the page.
+ */
struct extent_map *btrfs_get_extent(struct inode *inode, struct page *page,
- size_t page_offset, u64 start, u64 end,
+ size_t pg_offset, u64 start, u64 len,
int create)
{
int ret;
u64 extent_end = 0;
u64 objectid = inode->i_ino;
u32 found_type;
- int failed_insert = 0;
- struct btrfs_path *path;
+ struct btrfs_path *path = NULL;
struct btrfs_root *root = BTRFS_I(inode)->root;
struct btrfs_file_extent_item *item;
struct extent_buffer *leaf;
struct btrfs_key found_key;
struct extent_map *em = NULL;
struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
+ struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
struct btrfs_trans_handle *trans = NULL;
-
- path = btrfs_alloc_path();
- BUG_ON(!path);
- mutex_lock(&root->fs_info->fs_mutex);
+ int compressed;
again:
- em = lookup_extent_mapping(em_tree, start, end);
+ spin_lock(&em_tree->lock);
+ em = lookup_extent_mapping(em_tree, start, len);
+ if (em)
+ em->bdev = root->fs_info->fs_devices->latest_bdev;
+ spin_unlock(&em_tree->lock);
+
if (em) {
- goto out;
+ if (em->start > start || em->start + em->len <= start)
+ free_extent_map(em);
+ else if (em->block_start == EXTENT_MAP_INLINE && page)
+ free_extent_map(em);
+ else
+ goto out;
}
+ em = alloc_extent_map(GFP_NOFS);
if (!em) {
- em = alloc_extent_map(GFP_NOFS);
- if (!em) {
- err = -ENOMEM;
- goto out;
- }
- em->start = EXTENT_MAP_HOLE;
- em->end = EXTENT_MAP_HOLE;
+ err = -ENOMEM;
+ goto out;
}
- em->bdev = inode->i_sb->s_bdev;
- ret = btrfs_lookup_file_extent(NULL, root, path,
- objectid, start, 0);
+ em->bdev = root->fs_info->fs_devices->latest_bdev;
+ em->start = EXTENT_MAP_HOLE;
+ em->len = (u64)-1;
+ em->block_len = (u64)-1;
+
+ if (!path) {
+ path = btrfs_alloc_path();
+ BUG_ON(!path);
+ }
+
+ ret = btrfs_lookup_file_extent(trans, root, path,
+ objectid, start, trans != NULL);
if (ret < 0) {
err = ret;
goto out;
found_type = btrfs_file_extent_type(leaf, item);
extent_start = found_key.offset;
+ compressed = btrfs_file_extent_compression(leaf, item);
if (found_type == BTRFS_FILE_EXTENT_REG) {
extent_end = extent_start +
btrfs_file_extent_num_bytes(leaf, item);
if (start < extent_start || start >= extent_end) {
em->start = start;
if (start < extent_start) {
- if (end < extent_start)
+ if (start + len <= extent_start)
goto not_found;
- em->end = extent_end - 1;
+ em->len = extent_end - extent_start;
} else {
- em->end = end;
+ em->len = len;
}
goto not_found_em;
}
bytenr = btrfs_file_extent_disk_bytenr(leaf, item);
if (bytenr == 0) {
em->start = extent_start;
- em->end = extent_end - 1;
+ em->len = extent_end - extent_start;
em->block_start = EXTENT_MAP_HOLE;
- em->block_end = EXTENT_MAP_HOLE;
goto insert;
}
- bytenr += btrfs_file_extent_offset(leaf, item);
- em->block_start = bytenr;
- em->block_end = em->block_start +
- btrfs_file_extent_num_bytes(leaf, item) - 1;
em->start = extent_start;
- em->end = extent_end - 1;
+ em->len = extent_end - extent_start;
+ if (compressed) {
+ set_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
+ em->block_start = bytenr;
+ em->block_len = btrfs_file_extent_disk_num_bytes(leaf,
+ item);
+ } else {
+ bytenr += btrfs_file_extent_offset(leaf, item);
+ em->block_start = bytenr;
+ em->block_len = em->len;
+ }
goto insert;
} else if (found_type == BTRFS_FILE_EXTENT_INLINE) {
+ u64 page_start;
unsigned long ptr;
char *map;
size_t size;
size_t extent_offset;
size_t copy_size;
- size = btrfs_file_extent_inline_len(leaf, btrfs_item_nr(leaf,
- path->slots[0]));
-
- extent_end = (extent_start + size) |
- ((u64)root->sectorsize - 1);
+ size = btrfs_file_extent_inline_len(leaf, item);
+ extent_end = (extent_start + size + root->sectorsize - 1) &
+ ~((u64)root->sectorsize - 1);
if (start < extent_start || start >= extent_end) {
em->start = start;
if (start < extent_start) {
- if (end < extent_start)
+ if (start + len <= extent_start)
goto not_found;
- em->end = extent_end;
+ em->len = extent_end - extent_start;
} else {
- em->end = end;
+ em->len = len;
}
goto not_found_em;
}
-
- extent_offset = (page->index << PAGE_CACHE_SHIFT) -
- extent_start;
- ptr = btrfs_file_extent_inline_start(item) + extent_offset;
- map = kmap(page);
- copy_size = min(PAGE_CACHE_SIZE - page_offset,
- size - extent_offset);
-
em->block_start = EXTENT_MAP_INLINE;
- em->block_end = EXTENT_MAP_INLINE;
- em->start = extent_start + extent_offset;
- em->end = (em->start + copy_size -1) |
- ((u64)root->sectorsize -1);
- if (!page) {
- goto insert;
+ if (!page || create) {
+ em->start = extent_start;
+ em->len = (size + root->sectorsize - 1) &
+ ~((u64)root->sectorsize - 1);
+ goto out;
}
- read_extent_buffer(leaf, map + page_offset, ptr, copy_size);
- /*
- memset(map + page_offset + copy_size, 0,
- PAGE_CACHE_SIZE - copy_size - page_offset);
- */
- flush_dcache_page(page);
- kunmap(page);
- set_extent_uptodate(em_tree, em->start, em->end, GFP_NOFS);
+ page_start = page_offset(page) + pg_offset;
+ extent_offset = page_start - extent_start;
+ copy_size = min_t(u64, PAGE_CACHE_SIZE - pg_offset,
+ size - extent_offset);
+ em->start = extent_start + extent_offset;
+ em->len = (copy_size + root->sectorsize - 1) &
+ ~((u64)root->sectorsize - 1);
+ if (compressed)
+ set_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
+ ptr = btrfs_file_extent_inline_start(item) + extent_offset;
+ if (create == 0 && !PageUptodate(page)) {
+ if (btrfs_file_extent_compression(leaf, item) ==
+ BTRFS_COMPRESS_ZLIB) {
+ ret = uncompress_inline(path, inode, page,
+ pg_offset,
+ extent_offset, item);
+ BUG_ON(ret);
+ } else {
+ map = kmap(page);
+ read_extent_buffer(leaf, map + pg_offset, ptr,
+ copy_size);
+ kunmap(page);
+ }
+ flush_dcache_page(page);
+ } else if (create && PageUptodate(page)) {
+ if (!trans) {
+ kunmap(page);
+ free_extent_map(em);
+ em = NULL;
+ btrfs_release_path(root, path);
+ trans = btrfs_join_transaction(root, 1);
+ goto again;
+ }
+ map = kmap(page);
+ write_extent_buffer(leaf, map + pg_offset, ptr,
+ copy_size);
+ kunmap(page);
+ btrfs_mark_buffer_dirty(leaf);
+ }
+ set_extent_uptodate(io_tree, em->start,
+ extent_map_end(em) - 1, GFP_NOFS);
goto insert;
} else {
printk("unkknown found_type %d\n", found_type);
}
not_found:
em->start = start;
- em->end = end;
+ em->len = len;
not_found_em:
em->block_start = EXTENT_MAP_HOLE;
- em->block_end = EXTENT_MAP_HOLE;
insert:
btrfs_release_path(root, path);
- if (em->start > start || em->end < start) {
- printk("bad extent! em: [%Lu %Lu] passed [%Lu %Lu]\n", em->start, em->end, start, end);
+ if (em->start > start || extent_map_end(em) <= start) {
+ printk("bad extent! em: [%Lu %Lu] passed [%Lu %Lu]\n", em->start, em->len, start, len);
err = -EIO;
goto out;
}
+
+ err = 0;
+ spin_lock(&em_tree->lock);
ret = add_extent_mapping(em_tree, em);
+ /* it is possible that someone inserted the extent into the tree
+ * while we had the lock dropped. It is also possible that
+ * an overlapping map exists in the tree
+ */
if (ret == -EEXIST) {
- free_extent_map(em);
- em = NULL;
- failed_insert++;
- if (failed_insert > 5) {
- printk("failing to insert %Lu %Lu\n", start, end);
- err = -EIO;
- goto out;
+ struct extent_map *existing;
+
+ ret = 0;
+
+ existing = lookup_extent_mapping(em_tree, start, len);
+ if (existing && (existing->start > start ||
+ existing->start + existing->len <= start)) {
+ free_extent_map(existing);
+ existing = NULL;
+ }
+ if (!existing) {
+ existing = lookup_extent_mapping(em_tree, em->start,
+ em->len);
+ if (existing) {
+ err = merge_extent_mapping(em_tree, existing,
+ em, start,
+ root->sectorsize);
+ free_extent_map(existing);
+ if (err) {
+ free_extent_map(em);
+ em = NULL;
+ }
+ } else {
+ err = -EIO;
+ printk("failing to insert %Lu %Lu\n",
+ start, len);
+ free_extent_map(em);
+ em = NULL;
+ }
+ } else {
+ free_extent_map(em);
+ em = existing;
+ err = 0;
}
- goto again;
}
- err = 0;
+ spin_unlock(&em_tree->lock);
out:
- btrfs_free_path(path);
+ if (path)
+ btrfs_free_path(path);
if (trans) {
ret = btrfs_end_transaction(trans, root);
- if (!err)
+ if (!err) {
err = ret;
+ }
}
- mutex_unlock(&root->fs_info->fs_mutex);
if (err) {
free_extent_map(em);
WARN_ON(1);
return em;
}
-static sector_t btrfs_bmap(struct address_space *mapping, sector_t iblock)
+static ssize_t btrfs_direct_IO(int rw, struct kiocb *iocb,
+ const struct iovec *iov, loff_t offset,
+ unsigned long nr_segs)
{
- return extent_bmap(mapping, iblock, btrfs_get_extent);
+ return -EINVAL;
}
-static int btrfs_prepare_write(struct file *file, struct page *page,
- unsigned from, unsigned to)
+static sector_t btrfs_bmap(struct address_space *mapping, sector_t iblock)
{
- return extent_prepare_write(&BTRFS_I(page->mapping->host)->extent_tree,
- page->mapping->host, page, from, to,
- btrfs_get_extent);
+ return extent_bmap(mapping, iblock, btrfs_get_extent);
}
int btrfs_readpage(struct file *file, struct page *page)
{
- struct extent_map_tree *tree;
- tree = &BTRFS_I(page->mapping->host)->extent_tree;
+ struct extent_io_tree *tree;
+ tree = &BTRFS_I(page->mapping->host)->io_tree;
return extent_read_full_page(tree, page, btrfs_get_extent);
}
+
static int btrfs_writepage(struct page *page, struct writeback_control *wbc)
{
- struct extent_map_tree *tree;
+ struct extent_io_tree *tree;
if (current->flags & PF_MEMALLOC) {
unlock_page(page);
return 0;
}
- tree = &BTRFS_I(page->mapping->host)->extent_tree;
+ tree = &BTRFS_I(page->mapping->host)->io_tree;
return extent_write_full_page(tree, page, btrfs_get_extent, wbc);
}
-static int btrfs_releasepage(struct page *page, gfp_t unused_gfp_flags)
+int btrfs_writepages(struct address_space *mapping,
+ struct writeback_control *wbc)
+{
+ struct extent_io_tree *tree;
+ tree = &BTRFS_I(mapping->host)->io_tree;
+ return extent_writepages(tree, mapping, btrfs_get_extent, wbc);
+}
+
+static int
+btrfs_readpages(struct file *file, struct address_space *mapping,
+ struct list_head *pages, unsigned nr_pages)
{
- struct extent_map_tree *tree;
+ struct extent_io_tree *tree;
+ tree = &BTRFS_I(mapping->host)->io_tree;
+ return extent_readpages(tree, mapping, pages, nr_pages,
+ btrfs_get_extent);
+}
+static int __btrfs_releasepage(struct page *page, gfp_t gfp_flags)
+{
+ struct extent_io_tree *tree;
+ struct extent_map_tree *map;
int ret;
- tree = &BTRFS_I(page->mapping->host)->extent_tree;
- ret = try_release_extent_mapping(tree, page);
+ tree = &BTRFS_I(page->mapping->host)->io_tree;
+ map = &BTRFS_I(page->mapping->host)->extent_tree;
+ ret = try_release_extent_mapping(map, tree, page, gfp_flags);
if (ret == 1) {
ClearPagePrivate(page);
set_page_private(page, 0);
return ret;
}
+static int btrfs_releasepage(struct page *page, gfp_t gfp_flags)
+{
+ if (PageWriteback(page) || PageDirty(page))
+ return 0;
+ return __btrfs_releasepage(page, gfp_flags);
+}
+
static void btrfs_invalidatepage(struct page *page, unsigned long offset)
{
- struct extent_map_tree *tree;
+ struct extent_io_tree *tree;
+ struct btrfs_ordered_extent *ordered;
+ u64 page_start = page_offset(page);
+ u64 page_end = page_start + PAGE_CACHE_SIZE - 1;
+
+ wait_on_page_writeback(page);
+ tree = &BTRFS_I(page->mapping->host)->io_tree;
+ if (offset) {
+ btrfs_releasepage(page, GFP_NOFS);
+ return;
+ }
- tree = &BTRFS_I(page->mapping->host)->extent_tree;
- extent_invalidatepage(tree, page, offset);
- btrfs_releasepage(page, GFP_NOFS);
+ lock_extent(tree, page_start, page_end, GFP_NOFS);
+ ordered = btrfs_lookup_ordered_extent(page->mapping->host,
+ page_offset(page));
+ if (ordered) {
+ /*
+ * IO on this page will never be started, so we need
+ * to account for any ordered extents now
+ */
+ clear_extent_bit(tree, page_start, page_end,
+ EXTENT_DIRTY | EXTENT_DELALLOC |
+ EXTENT_LOCKED, 1, 0, GFP_NOFS);
+ btrfs_finish_ordered_io(page->mapping->host,
+ page_start, page_end);
+ btrfs_put_ordered_extent(ordered);
+ lock_extent(tree, page_start, page_end, GFP_NOFS);
+ }
+ clear_extent_bit(tree, page_start, page_end,
+ EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC |
+ EXTENT_ORDERED,
+ 1, 1, GFP_NOFS);
+ __btrfs_releasepage(page, GFP_NOFS);
+
+ ClearPageChecked(page);
+ if (PagePrivate(page)) {
+ ClearPagePrivate(page);
+ set_page_private(page, 0);
+ page_cache_release(page);
+ }
}
/*
*/
int btrfs_page_mkwrite(struct vm_area_struct *vma, struct page *page)
{
- struct inode *inode = vma->vm_file->f_path.dentry->d_inode;
- unsigned long end;
+ struct inode *inode = fdentry(vma->vm_file)->d_inode;
+ struct btrfs_root *root = BTRFS_I(inode)->root;
+ struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+ struct btrfs_ordered_extent *ordered;
+ char *kaddr;
+ unsigned long zero_start;
loff_t size;
- int ret = -EINVAL;
+ int ret;
u64 page_start;
+ u64 page_end;
+
+ ret = btrfs_check_free_space(root, PAGE_CACHE_SIZE, 0);
+ if (ret)
+ goto out;
- down_read(&BTRFS_I(inode)->root->snap_sem);
+ ret = -EINVAL;
+again:
lock_page(page);
- wait_on_page_writeback(page);
size = i_size_read(inode);
- page_start = page->index << PAGE_CACHE_SHIFT;
+ page_start = page_offset(page);
+ page_end = page_start + PAGE_CACHE_SIZE - 1;
if ((page->mapping != inode->i_mapping) ||
- (page_start > size)) {
+ (page_start >= size)) {
/* page got truncated out from underneath us */
goto out_unlock;
}
+ wait_on_page_writeback(page);
+
+ lock_extent(io_tree, page_start, page_end, GFP_NOFS);
+ set_page_extent_mapped(page);
+
+ /*
+ * 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_extent(inode, page_start);
+ if (ordered) {
+ unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
+ unlock_page(page);
+ btrfs_start_ordered_extent(inode, ordered, 1);
+ btrfs_put_ordered_extent(ordered);
+ goto again;
+ }
+
+ btrfs_set_extent_delalloc(inode, page_start, page_end);
+ ret = 0;
/* page is wholly or partially inside EOF */
if (page_start + PAGE_CACHE_SIZE > size)
- end = size & ~PAGE_CACHE_MASK;
+ zero_start = size & ~PAGE_CACHE_MASK;
else
- end = PAGE_CACHE_SIZE;
+ zero_start = PAGE_CACHE_SIZE;
- ret = btrfs_cow_one_page(inode, page, end);
+ if (zero_start != PAGE_CACHE_SIZE) {
+ kaddr = kmap(page);
+ memset(kaddr + zero_start, 0, PAGE_CACHE_SIZE - zero_start);
+ flush_dcache_page(page);
+ kunmap(page);
+ }
+ ClearPageChecked(page);
+ set_page_dirty(page);
+ unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
out_unlock:
- up_read(&BTRFS_I(inode)->root->snap_sem);
unlock_page(page);
+out:
return ret;
}
int ret;
struct btrfs_trans_handle *trans;
unsigned long nr;
+ u64 mask = root->sectorsize - 1;
if (!S_ISREG(inode->i_mode))
return;
return;
btrfs_truncate_page(inode->i_mapping, inode->i_size);
+ btrfs_wait_ordered_range(inode, inode->i_size & (~mask), (u64)-1);
- mutex_lock(&root->fs_info->fs_mutex);
trans = btrfs_start_transaction(root, 1);
btrfs_set_trans_block_group(trans, inode);
+ btrfs_i_size_write(inode, inode->i_size);
+ ret = btrfs_orphan_add(trans, inode);
+ if (ret)
+ goto out;
/* FIXME, add redo link to tree so we don't leak on crash */
- ret = btrfs_truncate_in_trans(trans, root, inode);
+ ret = btrfs_truncate_inode_items(trans, root, inode, inode->i_size,
+ BTRFS_EXTENT_DATA_KEY);
btrfs_update_inode(trans, root, inode);
- nr = trans->blocks_used;
- ret = btrfs_end_transaction(trans, root);
+ ret = btrfs_orphan_del(trans, inode);
+ BUG_ON(ret);
+
+out:
+ nr = trans->blocks_used;
+ ret = btrfs_end_transaction_throttle(trans, root);
BUG_ON(ret);
- mutex_unlock(&root->fs_info->fs_mutex);
btrfs_btree_balance_dirty(root, nr);
}
-int btrfs_commit_write(struct file *file, struct page *page,
- unsigned from, unsigned to)
+/*
+ * Invalidate a single dcache entry at the root of the filesystem.
+ * Needed after creation of snapshot or subvolume.
+ */
+void btrfs_invalidate_dcache_root(struct btrfs_root *root, char *name,
+ int namelen)
{
- return extent_commit_write(&BTRFS_I(page->mapping->host)->extent_tree,
- page->mapping->host, page, from, to);
+ struct dentry *alias, *entry;
+ struct qstr qstr;
+
+ alias = d_find_alias(root->fs_info->sb->s_root->d_inode);
+ if (alias) {
+ qstr.name = name;
+ qstr.len = namelen;
+ /* change me if btrfs ever gets a d_hash operation */
+ qstr.hash = full_name_hash(qstr.name, qstr.len);
+ entry = d_lookup(alias, &qstr);
+ dput(alias);
+ if (entry) {
+ d_invalidate(entry);
+ dput(entry);
+ }
+ }
}
-static int create_subvol(struct btrfs_root *root, char *name, int namelen)
+/*
+ * create a new subvolume directory/inode (helper for the ioctl).
+ */
+int btrfs_create_subvol_root(struct btrfs_root *new_root, struct dentry *dentry,
+ struct btrfs_trans_handle *trans, u64 new_dirid,
+ struct btrfs_block_group_cache *block_group)
{
- struct btrfs_trans_handle *trans;
- struct btrfs_key key;
- struct btrfs_root_item root_item;
- struct btrfs_inode_item *inode_item;
- struct extent_buffer *leaf;
- struct btrfs_root *new_root;
struct inode *inode;
- struct inode *dir;
- int ret;
- int err;
- u64 objectid;
- u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID;
- unsigned long nr = 1;
-
- mutex_lock(&root->fs_info->fs_mutex);
- trans = btrfs_start_transaction(root, 1);
- BUG_ON(!trans);
-
- leaf = btrfs_alloc_free_block(trans, root, root->leafsize, 0, 0);
- if (IS_ERR(leaf))
- return PTR_ERR(leaf);
-
- btrfs_set_header_nritems(leaf, 0);
- btrfs_set_header_level(leaf, 0);
- btrfs_set_header_bytenr(leaf, leaf->start);
- btrfs_set_header_generation(leaf, trans->transid);
- btrfs_set_header_owner(leaf, root->root_key.objectid);
- write_extent_buffer(leaf, root->fs_info->fsid,
- (unsigned long)btrfs_header_fsid(leaf),
- BTRFS_FSID_SIZE);
- btrfs_mark_buffer_dirty(leaf);
-
- inode_item = &root_item.inode;
- memset(inode_item, 0, sizeof(*inode_item));
- inode_item->generation = cpu_to_le64(1);
- inode_item->size = cpu_to_le64(3);
- inode_item->nlink = cpu_to_le32(1);
- inode_item->nblocks = cpu_to_le64(1);
- inode_item->mode = cpu_to_le32(S_IFDIR | 0755);
-
- btrfs_set_root_bytenr(&root_item, leaf->start);
- btrfs_set_root_level(&root_item, 0);
- btrfs_set_root_refs(&root_item, 1);
- btrfs_set_root_used(&root_item, 0);
-
- memset(&root_item.drop_progress, 0, sizeof(root_item.drop_progress));
- root_item.drop_level = 0;
-
- free_extent_buffer(leaf);
- leaf = NULL;
-
- ret = btrfs_find_free_objectid(trans, root->fs_info->tree_root,
- 0, &objectid);
- if (ret)
- goto fail;
-
- btrfs_set_root_dirid(&root_item, new_dirid);
-
- key.objectid = objectid;
- key.offset = 1;
- btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY);
- ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key,
- &root_item);
- if (ret)
- goto fail;
-
- /*
- * insert the directory item
- */
- key.offset = (u64)-1;
- dir = root->fs_info->sb->s_root->d_inode;
- ret = btrfs_insert_dir_item(trans, root->fs_info->tree_root,
- name, namelen, dir->i_ino, &key,
- BTRFS_FT_DIR);
- if (ret)
- goto fail;
-
- ret = btrfs_commit_transaction(trans, root);
- if (ret)
- goto fail_commit;
-
- new_root = btrfs_read_fs_root(root->fs_info, &key, name, namelen);
- BUG_ON(!new_root);
-
- trans = btrfs_start_transaction(new_root, 1);
- BUG_ON(!trans);
+ int error;
+ u64 index = 0;
- inode = btrfs_new_inode(trans, new_root, new_dirid,
- BTRFS_I(dir)->block_group, S_IFDIR | 0700);
+ inode = btrfs_new_inode(trans, new_root, NULL, "..", 2, new_dirid,
+ new_dirid, block_group, S_IFDIR | 0700, &index);
if (IS_ERR(inode))
- goto fail;
+ return PTR_ERR(inode);
inode->i_op = &btrfs_dir_inode_operations;
inode->i_fop = &btrfs_dir_file_operations;
new_root->inode = inode;
- ret = btrfs_make_empty_dir(trans, new_root, new_dirid, new_dirid);
- if (ret)
- goto fail;
-
inode->i_nlink = 1;
- inode->i_size = 6;
- ret = btrfs_update_inode(trans, new_root, inode);
- if (ret)
- goto fail;
-fail:
- nr = trans->blocks_used;
- err = btrfs_commit_transaction(trans, root);
- if (err && !ret)
- ret = err;
-fail_commit:
- mutex_unlock(&root->fs_info->fs_mutex);
- btrfs_btree_balance_dirty(root, nr);
- return ret;
-}
-
-static int create_snapshot(struct btrfs_root *root, char *name, int namelen)
-{
- struct btrfs_trans_handle *trans;
- struct btrfs_key key;
- struct btrfs_root_item new_root_item;
- struct extent_buffer *tmp;
- int ret;
- int err;
- u64 objectid;
- unsigned long nr;
-
- if (!root->ref_cows)
- return -EINVAL;
-
- down_write(&root->snap_sem);
- freeze_bdev(root->fs_info->sb->s_bdev);
- thaw_bdev(root->fs_info->sb->s_bdev, root->fs_info->sb);
-
- mutex_lock(&root->fs_info->fs_mutex);
- trans = btrfs_start_transaction(root, 1);
- BUG_ON(!trans);
-
- ret = btrfs_update_inode(trans, root, root->inode);
- if (ret)
- goto fail;
-
- ret = btrfs_find_free_objectid(trans, root->fs_info->tree_root,
- 0, &objectid);
- if (ret)
- goto fail;
-
- memcpy(&new_root_item, &root->root_item,
- sizeof(new_root_item));
-
- key.objectid = objectid;
- key.offset = 1;
- btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY);
-
- btrfs_cow_block(trans, root, root->node, NULL, 0, &tmp);
- btrfs_set_root_bytenr(&new_root_item, root->node->start);
- btrfs_set_root_level(&new_root_item, btrfs_header_level(root->node));
-
- ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key,
- &new_root_item);
- if (ret)
- goto fail;
-
- /*
- * insert the directory item
- */
- key.offset = (u64)-1;
- ret = btrfs_insert_dir_item(trans, root->fs_info->tree_root,
- name, namelen,
- root->fs_info->sb->s_root->d_inode->i_ino,
- &key, BTRFS_FT_DIR);
-
- if (ret)
- goto fail;
-
- ret = btrfs_inc_root_ref(trans, root);
- if (ret)
- goto fail;
-fail:
- nr = trans->blocks_used;
- err = btrfs_commit_transaction(trans, root);
+ btrfs_i_size_write(inode, 0);
- if (err && !ret)
- ret = err;
+ error = btrfs_update_inode(trans, new_root, inode);
+ if (error)
+ return error;
- mutex_unlock(&root->fs_info->fs_mutex);
- up_write(&root->snap_sem);
- btrfs_btree_balance_dirty(root, nr);
- return ret;
+ d_instantiate(dentry, inode);
+ return 0;
}
-static unsigned long force_ra(struct address_space *mapping,
+/* helper function for file defrag and space balancing. This
+ * forces readahead on a given range of bytes in an inode
+ */
+unsigned long btrfs_force_ra(struct address_space *mapping,
struct file_ra_state *ra, struct file *file,
pgoff_t offset, pgoff_t last_index)
{
- pgoff_t req_size;
+ pgoff_t req_size = last_index - offset + 1;
-#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,23)
- req_size = last_index - offset + 1;
- offset = page_cache_readahead(mapping, ra, file, offset, req_size);
- return offset;
-#else
- req_size = min(last_index - offset + 1, (pgoff_t)128);
page_cache_sync_readahead(mapping, ra, file, offset, req_size);
return offset + req_size;
-#endif
-}
-
-int btrfs_defrag_file(struct file *file) {
- struct inode *inode = file->f_path.dentry->d_inode;
- struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
- struct page *page;
- unsigned long last_index;
- unsigned long ra_index = 0;
- u64 page_start;
- u64 page_end;
- unsigned long i;
-
- mutex_lock(&inode->i_mutex);
- last_index = inode->i_size >> PAGE_CACHE_SHIFT;
- for (i = 0; i <= last_index; i++) {
- if (i == ra_index) {
- ra_index = force_ra(inode->i_mapping, &file->f_ra,
- file, ra_index, last_index);
- }
- page = grab_cache_page(inode->i_mapping, i);
- if (!page)
- goto out_unlock;
- if (!PageUptodate(page)) {
- btrfs_readpage(NULL, page);
- lock_page(page);
- if (!PageUptodate(page)) {
- unlock_page(page);
- page_cache_release(page);
- goto out_unlock;
- }
- }
- page_start = page->index << PAGE_CACHE_SHIFT;
- page_end = page_start + PAGE_CACHE_SIZE - 1;
-
- lock_extent(em_tree, page_start, page_end, GFP_NOFS);
- set_extent_delalloc(em_tree, page_start,
- page_end, GFP_NOFS);
- unlock_extent(em_tree, page_start, page_end, GFP_NOFS);
- set_page_dirty(page);
- unlock_page(page);
- page_cache_release(page);
- balance_dirty_pages_ratelimited_nr(inode->i_mapping, 1);
- }
-
-out_unlock:
- mutex_unlock(&inode->i_mutex);
- return 0;
-}
-
-static int btrfs_ioctl_snap_create(struct btrfs_root *root, void __user *arg)
-{
- struct btrfs_ioctl_vol_args vol_args;
- struct btrfs_dir_item *di;
- struct btrfs_path *path;
- int namelen;
- u64 root_dirid;
-
- if (copy_from_user(&vol_args, arg, sizeof(vol_args)))
- return -EFAULT;
-
- namelen = strlen(vol_args.name);
- if (namelen > BTRFS_VOL_NAME_MAX)
- return -EINVAL;
- if (strchr(vol_args.name, '/'))
- return -EINVAL;
-
- path = btrfs_alloc_path();
- if (!path)
- return -ENOMEM;
-
- root_dirid = root->fs_info->sb->s_root->d_inode->i_ino,
- mutex_lock(&root->fs_info->fs_mutex);
- di = btrfs_lookup_dir_item(NULL, root->fs_info->tree_root,
- path, root_dirid,
- vol_args.name, namelen, 0);
- mutex_unlock(&root->fs_info->fs_mutex);
- btrfs_free_path(path);
- if (di && !IS_ERR(di))
- return -EEXIST;
- if (IS_ERR(di))
- return PTR_ERR(di);
-
- if (root == root->fs_info->tree_root)
- return create_subvol(root, vol_args.name, namelen);
- return create_snapshot(root, vol_args.name, namelen);
-}
-
-static int btrfs_ioctl_defrag(struct file *file)
-{
- struct inode *inode = file->f_path.dentry->d_inode;
- struct btrfs_root *root = BTRFS_I(inode)->root;
-
- switch (inode->i_mode & S_IFMT) {
- case S_IFDIR:
- mutex_lock(&root->fs_info->fs_mutex);
- btrfs_defrag_root(root, 0);
- btrfs_defrag_root(root->fs_info->extent_root, 0);
- mutex_unlock(&root->fs_info->fs_mutex);
- break;
- case S_IFREG:
- btrfs_defrag_file(file);
- break;
- }
-
- return 0;
-}
-
-long btrfs_ioctl(struct file *file, unsigned int
- cmd, unsigned long arg)
-{
- struct btrfs_root *root = BTRFS_I(file->f_path.dentry->d_inode)->root;
-
- switch (cmd) {
- case BTRFS_IOC_SNAP_CREATE:
- return btrfs_ioctl_snap_create(root, (void __user *)arg);
- case BTRFS_IOC_DEFRAG:
- return btrfs_ioctl_defrag(file);
- }
-
- return -ENOTTY;
}
-/*
- * Called inside transaction, so use GFP_NOFS
- */
struct inode *btrfs_alloc_inode(struct super_block *sb)
{
struct btrfs_inode *ei;
if (!ei)
return NULL;
ei->last_trans = 0;
+ ei->logged_trans = 0;
+ btrfs_ordered_inode_tree_init(&ei->ordered_tree);
+ ei->i_acl = BTRFS_ACL_NOT_CACHED;
+ ei->i_default_acl = BTRFS_ACL_NOT_CACHED;
+ INIT_LIST_HEAD(&ei->i_orphan);
return &ei->vfs_inode;
}
void btrfs_destroy_inode(struct inode *inode)
{
+ struct btrfs_ordered_extent *ordered;
WARN_ON(!list_empty(&inode->i_dentry));
WARN_ON(inode->i_data.nrpages);
+ if (BTRFS_I(inode)->i_acl &&
+ BTRFS_I(inode)->i_acl != BTRFS_ACL_NOT_CACHED)
+ posix_acl_release(BTRFS_I(inode)->i_acl);
+ if (BTRFS_I(inode)->i_default_acl &&
+ BTRFS_I(inode)->i_default_acl != BTRFS_ACL_NOT_CACHED)
+ posix_acl_release(BTRFS_I(inode)->i_default_acl);
+
+ spin_lock(&BTRFS_I(inode)->root->list_lock);
+ if (!list_empty(&BTRFS_I(inode)->i_orphan)) {
+ printk(KERN_ERR "BTRFS: inode %lu: inode still on the orphan"
+ " list\n", inode->i_ino);
+ dump_stack();
+ }
+ spin_unlock(&BTRFS_I(inode)->root->list_lock);
+
+ while(1) {
+ ordered = btrfs_lookup_first_ordered_extent(inode, (u64)-1);
+ if (!ordered)
+ break;
+ else {
+ printk("found ordered extent %Lu %Lu\n",
+ ordered->file_offset, ordered->len);
+ btrfs_remove_ordered_extent(inode, ordered);
+ btrfs_put_ordered_extent(ordered);
+ btrfs_put_ordered_extent(ordered);
+ }
+ }
+ btrfs_drop_extent_cache(inode, 0, (u64)-1, 0);
kmem_cache_free(btrfs_inode_cachep, BTRFS_I(inode));
}
-static void init_once(void * foo, struct kmem_cache * cachep,
- unsigned long flags)
+static void init_once(void *foo)
{
struct btrfs_inode *ei = (struct btrfs_inode *) foo;
struct kmem_cache *btrfs_cache_create(const char *name, size_t size,
unsigned long extra_flags,
- void (*ctor)(void *, struct kmem_cache *,
- unsigned long))
+ void (*ctor)(void *))
{
return kmem_cache_create(name, size, 0, (SLAB_RECLAIM_ACCOUNT |
- SLAB_MEM_SPREAD | extra_flags), ctor
-#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,23)
- ,NULL
-#endif
- );
+ SLAB_MEM_SPREAD | extra_flags), ctor);
}
int btrfs_init_cachep(void)
{
struct inode *inode = dentry->d_inode;
generic_fillattr(inode, stat);
- stat->blksize = 256 * 1024;
+ stat->blksize = PAGE_CACHE_SIZE;
+ stat->blocks = (inode_get_bytes(inode) +
+ BTRFS_I(inode)->delalloc_bytes) >> 9;
return 0;
}
struct inode *new_inode = new_dentry->d_inode;
struct inode *old_inode = old_dentry->d_inode;
struct timespec ctime = CURRENT_TIME;
- struct btrfs_path *path;
- struct btrfs_dir_item *di;
+ u64 index = 0;
int ret;
if (S_ISDIR(old_inode->i_mode) && new_inode &&
return -ENOTEMPTY;
}
- mutex_lock(&root->fs_info->fs_mutex);
+ ret = btrfs_check_free_space(root, 1, 0);
+ if (ret)
+ goto out_unlock;
+
trans = btrfs_start_transaction(root, 1);
btrfs_set_trans_block_group(trans, new_dir);
- path = btrfs_alloc_path();
- if (!path) {
- ret = -ENOMEM;
- goto out_fail;
- }
- old_dentry->d_inode->i_nlink++;
+ btrfs_inc_nlink(old_dentry->d_inode);
old_dir->i_ctime = old_dir->i_mtime = ctime;
new_dir->i_ctime = new_dir->i_mtime = ctime;
old_inode->i_ctime = ctime;
- if (S_ISDIR(old_inode->i_mode) && old_dir != new_dir) {
- struct btrfs_key *location = &BTRFS_I(new_dir)->location;
- struct btrfs_key old_parent_key;
- di = btrfs_lookup_dir_item(trans, root, path, old_inode->i_ino,
- "..", 2, -1);
- if (IS_ERR(di)) {
- ret = PTR_ERR(di);
- goto out_fail;
- }
- if (!di) {
- ret = -ENOENT;
- goto out_fail;
- }
- btrfs_dir_item_key_to_cpu(path->nodes[0], di, &old_parent_key);
- ret = btrfs_del_item(trans, root, path);
- if (ret) {
- goto out_fail;
- }
- btrfs_release_path(root, path);
-
- di = btrfs_lookup_dir_index_item(trans, root, path,
- old_inode->i_ino,
- old_parent_key.objectid,
- "..", 2, -1);
- if (IS_ERR(di)) {
- ret = PTR_ERR(di);
- goto out_fail;
- }
- if (!di) {
- ret = -ENOENT;
- goto out_fail;
- }
- ret = btrfs_del_item(trans, root, path);
- if (ret) {
- goto out_fail;
- }
- btrfs_release_path(root, path);
-
- ret = btrfs_insert_dir_item(trans, root, "..", 2,
- old_inode->i_ino, location,
- BTRFS_FT_DIR);
- if (ret)
- goto out_fail;
- }
-
-
- ret = btrfs_unlink_trans(trans, root, old_dir, old_dentry);
+ ret = btrfs_unlink_inode(trans, root, old_dir, old_dentry->d_inode,
+ old_dentry->d_name.name,
+ old_dentry->d_name.len);
if (ret)
goto out_fail;
if (new_inode) {
new_inode->i_ctime = CURRENT_TIME;
- ret = btrfs_unlink_trans(trans, root, new_dir, new_dentry);
+ ret = btrfs_unlink_inode(trans, root, new_dir,
+ new_dentry->d_inode,
+ new_dentry->d_name.name,
+ new_dentry->d_name.len);
if (ret)
goto out_fail;
+ if (new_inode->i_nlink == 0) {
+ ret = btrfs_orphan_add(trans, new_dentry->d_inode);
+ if (ret)
+ goto out_fail;
+ }
+
}
- ret = btrfs_add_link(trans, new_dentry, old_inode);
+ ret = btrfs_set_inode_index(new_dir, old_inode, &index);
+ if (ret)
+ goto out_fail;
+
+ ret = btrfs_add_link(trans, new_dentry->d_parent->d_inode,
+ old_inode, new_dentry->d_name.name,
+ new_dentry->d_name.len, 1, index);
if (ret)
goto out_fail;
out_fail:
- btrfs_free_path(path);
- btrfs_end_transaction(trans, root);
- mutex_unlock(&root->fs_info->fs_mutex);
+ btrfs_end_transaction_throttle(trans, root);
+out_unlock:
return ret;
}
+/*
+ * some fairly slow code that needs optimization. This walks the list
+ * of all the inodes with pending delalloc and forces them to disk.
+ */
+int btrfs_start_delalloc_inodes(struct btrfs_root *root)
+{
+ struct list_head *head = &root->fs_info->delalloc_inodes;
+ struct btrfs_inode *binode;
+ struct inode *inode;
+ unsigned long flags;
+
+ spin_lock_irqsave(&root->fs_info->delalloc_lock, flags);
+ while(!list_empty(head)) {
+ binode = list_entry(head->next, struct btrfs_inode,
+ delalloc_inodes);
+ inode = igrab(&binode->vfs_inode);
+ if (!inode)
+ list_del_init(&binode->delalloc_inodes);
+ spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags);
+ if (inode) {
+ filemap_flush(inode->i_mapping);
+ iput(inode);
+ }
+ cond_resched();
+ spin_lock_irqsave(&root->fs_info->delalloc_lock, flags);
+ }
+ spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags);
+
+ /* the filemap_flush will queue IO into the worker threads, but
+ * we have to make sure the IO is actually started and that
+ * ordered extents get created before we return
+ */
+ atomic_inc(&root->fs_info->async_submit_draining);
+ while(atomic_read(&root->fs_info->nr_async_submits)) {
+ wait_event(root->fs_info->async_submit_wait,
+ (atomic_read(&root->fs_info->nr_async_submits) == 0));
+ }
+ atomic_dec(&root->fs_info->async_submit_draining);
+ return 0;
+}
+
static int btrfs_symlink(struct inode *dir, struct dentry *dentry,
const char *symname)
{
struct btrfs_root *root = BTRFS_I(dir)->root;
struct btrfs_path *path;
struct btrfs_key key;
- struct inode *inode;
+ struct inode *inode = NULL;
int err;
int drop_inode = 0;
u64 objectid;
+ u64 index = 0 ;
int name_len;
int datasize;
unsigned long ptr;
struct btrfs_file_extent_item *ei;
struct extent_buffer *leaf;
- unsigned long nr;
+ unsigned long nr = 0;
name_len = strlen(symname) + 1;
if (name_len > BTRFS_MAX_INLINE_DATA_SIZE(root))
return -ENAMETOOLONG;
- mutex_lock(&root->fs_info->fs_mutex);
+
+ err = btrfs_check_free_space(root, 1, 0);
+ if (err)
+ goto out_fail;
+
trans = btrfs_start_transaction(root, 1);
btrfs_set_trans_block_group(trans, dir);
goto out_unlock;
}
- inode = btrfs_new_inode(trans, root, objectid,
- BTRFS_I(dir)->block_group, S_IFLNK|S_IRWXUGO);
+ inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
+ dentry->d_name.len,
+ dentry->d_parent->d_inode->i_ino, objectid,
+ BTRFS_I(dir)->block_group, S_IFLNK|S_IRWXUGO,
+ &index);
err = PTR_ERR(inode);
if (IS_ERR(inode))
goto out_unlock;
+ err = btrfs_init_acl(inode, dir);
+ if (err) {
+ drop_inode = 1;
+ goto out_unlock;
+ }
+
btrfs_set_trans_block_group(trans, inode);
- err = btrfs_add_nondir(trans, dentry, inode);
+ err = btrfs_add_nondir(trans, dentry, inode, 0, index);
if (err)
drop_inode = 1;
else {
inode->i_mapping->a_ops = &btrfs_aops;
+ inode->i_mapping->backing_dev_info = &root->fs_info->bdi;
inode->i_fop = &btrfs_file_operations;
inode->i_op = &btrfs_file_inode_operations;
- extent_map_tree_init(&BTRFS_I(inode)->extent_tree,
- inode->i_mapping, GFP_NOFS);
- BTRFS_I(inode)->extent_tree.ops = &btrfs_extent_map_ops;
+ BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
}
dir->i_sb->s_dirt = 1;
btrfs_update_inode_block_group(trans, inode);
btrfs_set_file_extent_generation(leaf, ei, trans->transid);
btrfs_set_file_extent_type(leaf, ei,
BTRFS_FILE_EXTENT_INLINE);
+ btrfs_set_file_extent_encryption(leaf, ei, 0);
+ btrfs_set_file_extent_compression(leaf, ei, 0);
+ btrfs_set_file_extent_other_encoding(leaf, ei, 0);
+ btrfs_set_file_extent_ram_bytes(leaf, ei, name_len);
+
ptr = btrfs_file_extent_inline_start(ei);
write_extent_buffer(leaf, symname, ptr, name_len);
btrfs_mark_buffer_dirty(leaf);
inode->i_op = &btrfs_symlink_inode_operations;
inode->i_mapping->a_ops = &btrfs_symlink_aops;
- inode->i_size = name_len - 1;
+ inode->i_mapping->backing_dev_info = &root->fs_info->bdi;
+ btrfs_i_size_write(inode, name_len - 1);
err = btrfs_update_inode(trans, root, inode);
if (err)
drop_inode = 1;
out_unlock:
nr = trans->blocks_used;
- btrfs_end_transaction(trans, root);
- mutex_unlock(&root->fs_info->fs_mutex);
+ btrfs_end_transaction_throttle(trans, root);
+out_fail:
if (drop_inode) {
inode_dec_link_count(inode);
iput(inode);
return err;
}
+static int btrfs_set_page_dirty(struct page *page)
+{
+ return __set_page_dirty_nobuffers(page);
+}
+
+static int btrfs_permission(struct inode *inode, int mask)
+{
+ if (btrfs_test_flag(inode, READONLY) && (mask & MAY_WRITE))
+ return -EACCES;
+ return generic_permission(inode, mask, btrfs_check_acl);
+}
+
static struct inode_operations btrfs_dir_inode_operations = {
.lookup = btrfs_lookup,
.create = btrfs_create,
.symlink = btrfs_symlink,
.setattr = btrfs_setattr,
.mknod = btrfs_mknod,
+ .setxattr = btrfs_setxattr,
+ .getxattr = btrfs_getxattr,
+ .listxattr = btrfs_listxattr,
+ .removexattr = btrfs_removexattr,
+ .permission = btrfs_permission,
};
-
static struct inode_operations btrfs_dir_ro_inode_operations = {
.lookup = btrfs_lookup,
+ .permission = btrfs_permission,
};
-
static struct file_operations btrfs_dir_file_operations = {
.llseek = generic_file_llseek,
.read = generic_read_dir,
- .readdir = btrfs_readdir,
+ .readdir = btrfs_real_readdir,
.unlocked_ioctl = btrfs_ioctl,
#ifdef CONFIG_COMPAT
.compat_ioctl = btrfs_ioctl,
#endif
+ .release = btrfs_release_file,
+ .fsync = btrfs_sync_file,
};
-static struct extent_map_ops btrfs_extent_map_ops = {
+static struct extent_io_ops btrfs_extent_io_ops = {
.fill_delalloc = run_delalloc_range,
- .writepage_io_hook = btrfs_writepage_io_hook,
- .readpage_io_hook = btrfs_readpage_io_hook,
+ .submit_bio_hook = btrfs_submit_bio_hook,
+ .merge_bio_hook = btrfs_merge_bio_hook,
.readpage_end_io_hook = btrfs_readpage_end_io_hook,
+ .writepage_end_io_hook = btrfs_writepage_end_io_hook,
+ .writepage_start_hook = btrfs_writepage_start_hook,
+ .readpage_io_failed_hook = btrfs_io_failed_hook,
+ .set_bit_hook = btrfs_set_bit_hook,
+ .clear_bit_hook = btrfs_clear_bit_hook,
};
static struct address_space_operations btrfs_aops = {
.readpage = btrfs_readpage,
.writepage = btrfs_writepage,
+ .writepages = btrfs_writepages,
+ .readpages = btrfs_readpages,
.sync_page = block_sync_page,
- .prepare_write = btrfs_prepare_write,
- .commit_write = btrfs_commit_write,
.bmap = btrfs_bmap,
+ .direct_IO = btrfs_direct_IO,
.invalidatepage = btrfs_invalidatepage,
.releasepage = btrfs_releasepage,
- .set_page_dirty = __set_page_dirty_nobuffers,
+ .set_page_dirty = btrfs_set_page_dirty,
};
static struct address_space_operations btrfs_symlink_aops = {
.truncate = btrfs_truncate,
.getattr = btrfs_getattr,
.setattr = btrfs_setattr,
+ .setxattr = btrfs_setxattr,
+ .getxattr = btrfs_getxattr,
+ .listxattr = btrfs_listxattr,
+ .removexattr = btrfs_removexattr,
+ .permission = btrfs_permission,
};
-
static struct inode_operations btrfs_special_inode_operations = {
.getattr = btrfs_getattr,
.setattr = btrfs_setattr,
+ .permission = btrfs_permission,
+ .setxattr = btrfs_setxattr,
+ .getxattr = btrfs_getxattr,
+ .listxattr = btrfs_listxattr,
+ .removexattr = btrfs_removexattr,
};
-
static struct inode_operations btrfs_symlink_inode_operations = {
.readlink = generic_readlink,
.follow_link = page_follow_link_light,
.put_link = page_put_link,
+ .permission = btrfs_permission,
};