#include "xattr.h"
#include "compat.h"
#include "tree-log.h"
+#include "ref-cache.h"
+#include "compression.h"
struct btrfs_iget_args {
u64 ino;
};
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)
{
return ret;
}
-static int cow_file_range(struct inode *inode, u64 start, u64 end)
+/*
+ * 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;
u64 alloc_hint = 0;
u64 num_bytes;
+ unsigned long ram_size;
+ u64 orig_start;
+ u64 disk_num_bytes;
u64 cur_alloc_size;
u64 blocksize = root->sectorsize;
- u64 orig_num_bytes;
+ 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);
num_bytes = max(blocksize, num_bytes);
- orig_num_bytes = num_bytes;
+ disk_num_bytes = num_bytes;
+ total_in = 0;
+ ret = 0;
- if (alloc_hint == EXTENT_MAP_INLINE)
- 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));
- BUG_ON(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);
+ btrfs_drop_extent_cache(inode, start, start + num_bytes - 1, 0);
mutex_unlock(&BTRFS_I(inode)->extent_mutex);
- while(num_bytes > 0) {
- cur_alloc_size = min(num_bytes, root->fs_info->max_extent);
+ 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,
- root->sectorsize, 0, alloc_hint,
+ min_bytes, 0, alloc_hint,
(u64)-1, &ins, 1);
if (ret) {
WARN_ON(1);
- goto out;
+ goto free_pages_out_fail;
}
em = alloc_extent_map(GFP_NOFS);
em->start = start;
- em->len = ins.offset;
+
+ 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);
break;
}
btrfs_drop_extent_cache(inode, start,
- start + ins.offset - 1);
+ 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,
- ins.offset, 0);
+ ram_size, cur_alloc_size, 0,
+ will_compress);
BUG_ON(ret);
- if (num_bytes < cur_alloc_size) {
- printk("num_bytes %Lu cur_alloc %Lu\n", num_bytes,
+
+ 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 = 0;
out:
btrfs_end_transaction(trans, root);
+
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;
}
-static int run_delalloc_nocow(struct inode *inode, 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)
{
u64 extent_start;
u64 extent_end;
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;
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, 1);
+ extent_num_bytes,
+ extent_num_bytes, 1, 0);
if (ret) {
err = ret;
goto out;
not_found:
btrfs_end_transaction(trans, root);
btrfs_free_path(path);
- return cow_file_range(inode, start, end);
+ return cow_file_range(inode, locked_page, start, end,
+ page_started);
}
out:
WARN_ON(err);
return err;
}
-static int run_delalloc_range(struct inode *inode, u64 start, u64 end)
+/*
+ * 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, start, end);
+ ret = run_delalloc_nocow(inode, locked_page, start, end,
+ page_started);
else
- ret = cow_file_range(inode, start, end);
+ ret = cow_file_range(inode, locked_page, start, end,
+ page_started);
return ret;
}
+/*
+ * 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)
{
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)
{
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)
+ 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 = bio->bi_sector << 9;
+ u64 logical = (u64)bio->bi_sector << 9;
u64 length = 0;
u64 map_length;
int ret;
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)
+ int mirror_num, unsigned long bio_flags)
{
struct btrfs_root *root = BTRFS_I(inode)->root;
int ret = 0;
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)
+ int mirror_num, unsigned long bio_flags)
{
struct btrfs_root *root = BTRFS_I(inode)->root;
int ret = 0;
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,
- __btrfs_submit_bio_hook);
+ 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)
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;
};
-/* see btrfs_writepage_start_hook for details on why this is required */
void btrfs_writepage_fixup_worker(struct btrfs_work *work)
{
struct btrfs_writepage_fixup *fixup;
* 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 EXTENT_ORDERED bit set
+ * 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.
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;
btrfs_set_file_extent_disk_bytenr(leaf, extent_item,
ordered_extent->start);
btrfs_set_file_extent_disk_num_bytes(leaf, extent_item,
- ordered_extent->len);
+ 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);
+ ordered_extent->len - 1, 0);
mutex_unlock(&BTRFS_I(inode)->extent_mutex);
ins.objectid = ordered_extent->start;
- ins.offset = ordered_extent->len;
+ 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,
- ordered_extent->file_offset, &ins);
+ trans->transid, inode->i_ino, &ins);
BUG_ON(ret);
btrfs_release_path(root, path);
- inode->i_blocks += ordered_extent->len >> 9;
+ 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);
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;
int ret;
int rw;
u64 logical;
+ unsigned long bio_flags = 0;
ret = get_state_private(failure_tree, start, &private);
if (ret) {
}
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 |
rw = READ;
BTRFS_I(inode)->io_tree.ops->submit_bio_hook(inode, rw, bio,
- failrec->last_mirror);
+ 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;
return 0;
}
+/*
+ * 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)
{
int ret = 0, nr_unlink = 0, nr_truncate = 0;
/* don't do orphan cleanup if the fs is readonly. */
- if (root->inode->i_sb->s_flags & MS_RDONLY)
+ if (root->fs_info->sb->s_flags & MS_RDONLY)
return;
path = btrfs_alloc_path();
btrfs_set_key_type(&key, BTRFS_ORPHAN_ITEM_KEY);
key.offset = (u64)-1;
- trans = btrfs_start_transaction(root, 1);
- btrfs_set_trans_block_group(trans, root->inode);
while (1) {
ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
* crossing root thing. we store the inode number in the
* offset of the orphan item.
*/
- inode = btrfs_iget_locked(root->inode->i_sb,
+ inode = btrfs_iget_locked(root->fs_info->sb,
found_key.offset, root);
if (!inode)
break;
* 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;
}
printk(KERN_INFO "btrfs: truncated %d orphans\n", nr_truncate);
btrfs_free_path(path);
- btrfs_end_transaction(trans, root);
}
+/*
+ * read an inode from the btree into the in-memory inode
+ */
void btrfs_read_locked_inode(struct inode *inode)
{
struct btrfs_path *path;
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_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;
make_bad_inode(inode);
}
+/*
+ * 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,
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_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_I(inode)->block_group->key.objectid);
}
+/*
+ * 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)
}
+/*
+ * 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,
}
/*
+ * 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_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);
+
+ 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);
+
+ 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.
u64 mask = root->sectorsize - 1;
if (root->ref_cows)
- btrfs_drop_extent_cache(inode,
- new_size & (~mask), (u64)-1);
+ btrfs_drop_extent_cache(inode, new_size & (~mask), (u64)-1, 0);
path = btrfs_alloc_path();
path->reada = -1;
BUG_ON(!path);
key.type = (u8)-1;
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) {
item_end +=
btrfs_file_extent_num_bytes(leaf, fi);
} else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
- struct btrfs_item *item = btrfs_item_nr(leaf,
- path->slots[0]);
item_end += btrfs_file_extent_inline_len(leaf,
- item);
+ fi);
}
item_end--;
}
num_dec = (orig_num_bytes -
extent_num_bytes);
if (root->ref_cows && extent_start != 0)
- dec_i_blocks(inode, num_dec);
+ inode_sub_bytes(inode, num_dec);
btrfs_mark_buffer_dirty(leaf);
} else {
extent_num_bytes =
if (extent_start != 0) {
found_extent = 1;
if (root->ref_cows)
- dec_i_blocks(inode, num_dec);
+ 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) {
- if (!del_item) {
+ /*
+ * 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) {
- dec_i_blocks(inode, item_end + 1 -
- found_key.offset - size);
+ inode_sub_bytes(inode, item_end + 1 -
+ new_size);
}
size =
btrfs_file_extent_calc_inline_size(size);
size, 1);
BUG_ON(ret);
} else if (root->ref_cows) {
- dec_i_blocks(inode, item_end + 1 -
- found_key.offset);
+ inode_sub_bytes(inode, item_end + 1 -
+ found_key.offset);
}
}
delete:
ret = btrfs_free_extent(trans, root, extent_start,
extent_num_bytes,
leaf->start, root_owner,
- root_gen, inode->i_ino,
- found_key.offset, 0);
+ root_gen, inode->i_ino, 0);
BUG_ON(ret);
}
next:
err = btrfs_insert_file_extent(trans, root,
inode->i_ino,
hole_start, 0, 0,
- hole_size, 0);
+ hole_size, 0, hole_size,
+ 0, 0, 0);
btrfs_drop_extent_cache(inode, hole_start,
- (u64)-1);
+ (u64)-1, 0);
btrfs_check_file(root, inode);
}
mutex_unlock(&BTRFS_I(inode)->extent_mutex);
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)
{
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;
return ret;
}
+/*
+ * 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)
{
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);
return btrfs_type_by_mode[(inode->i_mode & S_IFMT) >> S_SHIFT];
}
+/*
+ * 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)
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,
start_diff = map_start - em->start;
em->start = map_start;
em->len = map_len;
- if (em->block_start < EXTENT_MAP_LAST_BYTE)
+ 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 pg_offset, u64 start, u64 len,
int create)
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;
+ int compressed;
again:
spin_lock(&em_tree->lock);
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();
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);
em->block_start = EXTENT_MAP_HOLE;
goto insert;
}
- bytenr += btrfs_file_extent_offset(leaf, item);
- em->block_start = bytenr;
em->start = extent_start;
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;
size_t extent_offset;
size_t copy_size;
- size = btrfs_file_extent_inline_len(leaf, btrfs_item_nr(leaf,
- path->slots[0]));
+ 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->block_start = EXTENT_MAP_INLINE;
- if (!page) {
+ if (!page || create) {
em->start = extent_start;
- em->len = size;
+ em->len = (size + root->sectorsize - 1) &
+ ~((u64)root->sectorsize - 1);
goto out;
}
em->start = extent_start + extent_offset;
em->len = (copy_size + root->sectorsize - 1) &
~((u64)root->sectorsize - 1);
- map = kmap(page);
+ if (compressed)
+ set_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
ptr = btrfs_file_extent_inline_start(item) + extent_offset;
if (create == 0 && !PageUptodate(page)) {
- read_extent_buffer(leaf, map + pg_offset, ptr,
- copy_size);
+ 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) {
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);
}
- kunmap(page);
set_extent_uptodate(io_tree, em->start,
extent_map_end(em) - 1, GFP_NOFS);
goto insert;
return em;
}
-#if 0 /* waiting for O_DIRECT reads */
-static int btrfs_get_block(struct inode *inode, sector_t iblock,
- struct buffer_head *bh_result, int create)
-{
- struct extent_map *em;
- u64 start = (u64)iblock << inode->i_blkbits;
- struct btrfs_multi_bio *multi = NULL;
- struct btrfs_root *root = BTRFS_I(inode)->root;
- u64 len;
- u64 logical;
- u64 map_length;
- int ret = 0;
-
- em = btrfs_get_extent(inode, NULL, 0, start, bh_result->b_size, 0);
-
- if (!em || IS_ERR(em))
- goto out;
-
- if (em->start > start || em->start + em->len <= start) {
- goto out;
- }
-
- if (em->block_start == EXTENT_MAP_INLINE) {
- ret = -EINVAL;
- goto out;
- }
-
- len = em->start + em->len - start;
- len = min_t(u64, len, INT_LIMIT(typeof(bh_result->b_size)));
-
- if (em->block_start == EXTENT_MAP_HOLE ||
- em->block_start == EXTENT_MAP_DELALLOC) {
- bh_result->b_size = len;
- goto out;
- }
-
- logical = start - em->start;
- logical = em->block_start + logical;
-
- map_length = len;
- ret = btrfs_map_block(&root->fs_info->mapping_tree, READ,
- logical, &map_length, &multi, 0);
- BUG_ON(ret);
- bh_result->b_blocknr = multi->stripes[0].physical >> inode->i_blkbits;
- bh_result->b_size = min(map_length, len);
-
- bh_result->b_bdev = multi->stripes[0].dev->bdev;
- set_buffer_mapped(bh_result);
- kfree(multi);
-out:
- free_extent_map(em);
- return ret;
-}
-#endif
-
static ssize_t btrfs_direct_IO(int rw, struct kiocb *iocb,
const struct iovec *iov, loff_t offset,
unsigned long nr_segs)
{
return -EINVAL;
-#if 0
- struct file *file = iocb->ki_filp;
- struct inode *inode = file->f_mapping->host;
-
- if (rw == WRITE)
- return -EINVAL;
-
- return blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
- offset, nr_segs, btrfs_get_block, NULL);
-#endif
}
static sector_t btrfs_bmap(struct address_space *mapping, sector_t iblock)
}
}
-int btrfs_create_subvol_root(struct btrfs_root *new_root,
+/*
+ * 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 inode *inode;
+ int error;
u64 index = 0;
inode = btrfs_new_inode(trans, new_root, NULL, "..", 2, new_dirid,
inode->i_nlink = 1;
btrfs_i_size_write(inode, 0);
- return btrfs_update_inode(trans, new_root, inode);
+ error = btrfs_update_inode(trans, new_root, inode);
+ if (error)
+ return error;
+
+ d_instantiate(dentry, inode);
+ return 0;
}
+/* 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)
btrfs_put_ordered_extent(ordered);
}
}
- btrfs_drop_extent_cache(inode, 0, (u64)-1);
+ btrfs_drop_extent_cache(inode, 0, (u64)-1, 0);
kmem_cache_free(btrfs_inode_cachep, BTRFS_I(inode));
}
struct inode *inode = dentry->d_inode;
generic_fillattr(inode, stat);
stat->blksize = PAGE_CACHE_SIZE;
- stat->blocks = inode->i_blocks + (BTRFS_I(inode)->delalloc_bytes >> 9);
+ stat->blocks = (inode_get_bytes(inode) +
+ BTRFS_I(inode)->delalloc_bytes) >> 9;
return 0;
}
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);
- atomic_inc(&binode->vfs_inode.i_count);
+ inode = igrab(&binode->vfs_inode);
+ if (!inode)
+ list_del_init(&binode->delalloc_inodes);
spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags);
- filemap_write_and_wait(binode->vfs_inode.i_mapping);
- iput(&binode->vfs_inode);
+ 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;
}
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);
git://ftp.safe.ca / safe / jmp / linux-2.6 / blobdiff