Btrfs: progress on file_write

Signed-off-by: Chris Mason <chris.mason@oracle.com>

1 files changed, 269 insertions, 15 deletions

diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
index ec689992fdf4..6a56416147e6 100644
--- a/fs/btrfs/super.c
+++ b/fs/btrfs/super.c
@@ -1089,7 +1089,6 @@ static int btrfs_get_block_lock(struct inode *inode, sector_t iblock,
 
         extent_start = btrfs_disk_key_offset(&leaf->items[path->slots[0]].key);
         extent_start = extent_start >> inode->i_blkbits;
-        extent_start += btrfs_file_extent_offset(item);
         extent_end = extent_start + btrfs_file_extent_num_blocks(item);
         if (iblock >= extent_start && iblock < extent_end) {
                 err = 0;
@@ -1103,6 +1102,7 @@ allocate:
                 err = 0;
                 goto out;
         }
+#if 0
         ret = btrfs_alloc_file_extent(trans, root, objectid,
                                       iblock << inode->i_blkbits,
                                       1, extent_end, &blocknr);
@@ -1115,9 +1115,11 @@ allocate:
         map_bh(result, inode->i_sb, blocknr);
 
         btrfs_map_bh_to_logical(root, result, blocknr);
+#endif
 out:
         btrfs_release_path(root, path);
         btrfs_free_path(path);
+printk("mapping iblock %lu to %lu\n", iblock, result->b_blocknr);
         if (trans)
                 btrfs_end_transaction(trans, root);
         return err;
@@ -1273,8 +1275,244 @@ failed:
         return err;
 }
 
-static int prepare_pages(struct btrfs_trans_handle *trans,
-                         struct btrfs_root *root,
+static int drop_csums(struct btrfs_trans_handle *trans,
+                          struct btrfs_root *root,
+                          struct inode *inode,
+                          u64 start, u64 end)
+{
+        struct btrfs_path *path;
+        struct btrfs_leaf *leaf;
+        struct btrfs_key key;
+        int slot;
+        struct btrfs_csum_item *item;
+        char *old_block = NULL;
+        u64 cur = start;
+        u64 found_end;
+        u64 num_csums;
+        u64 item_size;
+        int ret;
+
+        path = btrfs_alloc_path();
+        if (!path)
+                return -ENOMEM;
+        while(cur < end) {
+                item = btrfs_lookup_csum(trans, root, path,
+                                         inode->i_ino, cur, 1);
+                if (IS_ERR(item)) {
+                        cur += root->blocksize;
+                        continue;
+                }
+                leaf = btrfs_buffer_leaf(path->nodes[0]);
+                slot = path->slots[0];
+                btrfs_disk_key_to_cpu(&key, &leaf->items[slot].key);
+                item_size = btrfs_item_size(leaf->items + slot);
+                num_csums = item_size / sizeof(struct btrfs_csum_item);
+                found_end = key.offset + (num_csums << inode->i_blkbits);
+                cur = found_end;
+
+                if (found_end > end) {
+                        char *src;
+                        old_block = kmalloc(root->blocksize, GFP_NOFS);
+                        src = btrfs_item_ptr(leaf, slot, char);
+                        memcpy(old_block, src, item_size);
+                }
+                if (key.offset < start) {
+                        u64 new_size = (start - key.offset) >>
+                                        inode->i_blkbits;
+                        new_size *= sizeof(struct btrfs_csum_item);
+                        ret = btrfs_truncate_item(trans, root, path, new_size);
+                        BUG_ON(ret);
+                } else {
+                        btrfs_del_item(trans, root, path);
+                }
+                btrfs_release_path(root, path);
+                if (found_end > end) {
+                        char *dst;
+                        int i;
+                        int new_size;
+
+                        num_csums = (found_end - end) >> inode->i_blkbits;
+                        new_size = num_csums * sizeof(struct btrfs_csum_item);
+                        key.offset = end;
+                        ret = btrfs_insert_empty_item(trans, root, path,
+                                                      &key, new_size);
+                        BUG_ON(ret);
+                        dst = btrfs_item_ptr(btrfs_buffer_leaf(path->nodes[0]),
+                                             path->slots[0], char);
+                        memcpy(dst, old_block + item_size - new_size,
+                               new_size);
+                        item = (struct btrfs_csum_item *)dst;
+                        for (i = 0; i < num_csums; i++) {
+                                btrfs_set_csum_extent_offset(item, end);
+                                item++;
+                        }
+                        mark_buffer_dirty(path->nodes[0]);
+                        kfree(old_block);
+                        break;
+                }
+        }
+        btrfs_free_path(path);
+        return 0;
+}
+
+static int drop_extents(struct btrfs_trans_handle *trans,
+                          struct btrfs_root *root,
+                          struct inode *inode,
+                          u64 start, u64 end)
+{
+        int ret;
+        struct btrfs_key key;
+        struct btrfs_leaf *leaf;
+        int slot;
+        struct btrfs_file_extent_item *extent;
+        u64 extent_end;
+        int keep;
+        struct btrfs_file_extent_item old;
+        struct btrfs_path *path;
+        u64 search_start = start;
+        int bookend;
+
+        path = btrfs_alloc_path();
+        if (!path)
+                return -ENOMEM;
+search_again:
+printk("drop extent inode %lu start %Lu end %Lu\n", inode->i_ino, start, end);
+        ret = btrfs_lookup_file_extent(trans, root, path, inode->i_ino,
+                                       search_start, -1);
+        if (ret != 0) {
+printk("lookup failed\n");
+                goto out;
+        }
+        while(1) {
+                keep = 0;
+                bookend = 0;
+                leaf = btrfs_buffer_leaf(path->nodes[0]);
+                slot = path->slots[0];
+                btrfs_disk_key_to_cpu(&key, &leaf->items[slot].key);
+
+printk("found key %Lu %Lu %u\n", key.objectid, key.offset, key.flags);
+
+                extent = btrfs_item_ptr(leaf, slot,
+                                        struct btrfs_file_extent_item);
+                extent_end = key.offset +
+                        (btrfs_file_extent_num_blocks(extent) <<
+                         inode->i_blkbits);
+printk("extent end is %Lu\n", extent_end);
+                if (key.offset >= end || key.objectid != inode->i_ino) {
+                        ret = 0;
+                        goto out;
+                }
+                if (btrfs_key_type(&key) != BTRFS_EXTENT_DATA_KEY)
+                        goto next_leaf;
+
+                if (end < extent_end && end >= key.offset) {
+                        memcpy(&old, extent, sizeof(old));
+                        ret = btrfs_inc_extent_ref(trans, root,
+                                   btrfs_file_extent_disk_blocknr(&old),
+                                   btrfs_file_extent_disk_num_blocks(&old));
+                        BUG_ON(ret);
+                        bookend = 1;
+                }
+
+                if (start > key.offset) {
+                        u64 new_num;
+                        /* truncate existing extent */
+                        keep = 1;
+                        WARN_ON(start & (root->blocksize - 1));
+                        new_num = (start - key.offset) >> inode->i_blkbits;
+printk("truncating existing extent, was %Lu ", btrfs_file_extent_num_blocks(extent));
+                        btrfs_set_file_extent_num_blocks(extent, new_num);
+printk("now %Lu\n", btrfs_file_extent_num_blocks(extent));
+
+                        mark_buffer_dirty(path->nodes[0]);
+                }
+                if (!keep) {
+                        u64 disk_blocknr;
+                        u64 disk_num_blocks;
+printk("del old\n");
+                        disk_blocknr = btrfs_file_extent_disk_blocknr(extent);
+                        disk_num_blocks =
+                                btrfs_file_extent_disk_num_blocks(extent);
+                        search_start = key.offset +
+                                (btrfs_file_extent_num_blocks(extent) <<
+                                inode->i_blkbits);
+                        ret = btrfs_del_item(trans, root, path);
+                        BUG_ON(ret);
+                        btrfs_release_path(root, path);
+
+                        ret = btrfs_free_extent(trans, root, disk_blocknr,
+                                                disk_num_blocks, 0);
+
+                        BUG_ON(ret);
+                        if (!bookend && search_start >= end) {
+                                ret = 0;
+                                goto out;
+                        }
+                        if (!bookend)
+                                goto search_again;
+                }
+                if (bookend) {
+                        /* create bookend */
+                        struct btrfs_key ins;
+printk("bookend! extent end %Lu\n", extent_end);
+                        ins.objectid = inode->i_ino;
+                        ins.offset = end;
+                        ins.flags = 0;
+                        btrfs_set_key_type(&ins, BTRFS_EXTENT_DATA_KEY);
+
+                        btrfs_release_path(root, path);
+                        ret = drop_csums(trans, root, inode, start, end);
+                        BUG_ON(ret);
+                        ret = btrfs_insert_empty_item(trans, root, path, &ins,
+                                                      sizeof(*extent));
+                        BUG_ON(ret);
+                        extent = btrfs_item_ptr(
+                                    btrfs_buffer_leaf(path->nodes[0]),
+                                    path->slots[0],
+                                    struct btrfs_file_extent_item);
+                        btrfs_set_file_extent_disk_blocknr(extent,
+                                    btrfs_file_extent_disk_blocknr(&old));
+                        btrfs_set_file_extent_disk_num_blocks(extent,
+                                    btrfs_file_extent_disk_num_blocks(&old));
+
+                        btrfs_set_file_extent_offset(extent,
+                                    btrfs_file_extent_offset(&old) +
+                                    ((end - key.offset) >> inode->i_blkbits));
+                        WARN_ON(btrfs_file_extent_num_blocks(&old) <
+                                (end - key.offset) >> inode->i_blkbits);
+                        btrfs_set_file_extent_num_blocks(extent,
+                                    btrfs_file_extent_num_blocks(&old) -
+                                    ((end - key.offset) >> inode->i_blkbits));
+
+                        btrfs_set_file_extent_generation(extent,
+                                    btrfs_file_extent_generation(&old));
+printk("new bookend at offset %Lu, file_extent_offset %Lu, file_extent_num_blocks %Lu\n", end, btrfs_file_extent_offset(extent), btrfs_file_extent_num_blocks(extent));
+                        btrfs_mark_buffer_dirty(path->nodes[0]);
+                        ret = 0;
+                        goto out_nocsum;
+                }
+next_leaf:
+                if (slot >= btrfs_header_nritems(&leaf->header) - 1) {
+                        ret = btrfs_next_leaf(root, path);
+                        if (ret) {
+                                ret = 0;
+                                goto out;
+                        }
+                } else {
+                        path->slots[0]++;
+                }
+        }
+
+out:
+        ret = drop_csums(trans, root, inode, start, end);
+        BUG_ON(ret);
+
+out_nocsum:
+        btrfs_free_path(path);
+        return ret;
+}
+
+static int prepare_pages(struct btrfs_root *root,
                          struct file *file,
                          struct page **pages,
                          size_t num_pages,
@@ -1289,7 +1527,6 @@ static int prepare_pages(struct btrfs_trans_handle *trans,
         struct inode *inode = file->f_path.dentry->d_inode;
         int offset;
         int err = 0;
-        int ret;
         int this_write;
         struct buffer_head *bh;
         struct buffer_head *head;
@@ -1305,18 +1542,21 @@ static int prepare_pages(struct btrfs_trans_handle *trans,
                 }
                 offset = pos & (PAGE_CACHE_SIZE -1);
                 this_write = min(PAGE_CACHE_SIZE - offset, write_bytes);
-                if (!PageUptodate(pages[i]) &&
-                   (pages[i]->index == first_index ||
-                    pages[i]->index == last_index) && pos < isize) {
+#if 0
+                if ((pages[i]->index == first_index ||
+                    pages[i]->index == last_index) && pos < isize &&
+                    !PageUptodate(pages[i])) {
                         ret = mpage_readpage(pages[i], btrfs_get_block);
                         BUG_ON(ret);
                         lock_page(pages[i]);
                 }
+#endif
                 create_empty_buffers(pages[i], root->fs_info->sb->s_blocksize,
                                      (1 << BH_Uptodate));
                 head = page_buffers(pages[i]);
                 bh = head;
                 do {
+printk("mapping page %lu to block %Lu\n", pages[i]->index, alloc_extent_start);
                         err = btrfs_map_bh_to_logical(root, bh,
                                                       alloc_extent_start);
                         BUG_ON(err);
@@ -1351,7 +1591,7 @@ static ssize_t btrfs_file_write(struct file *file, const char __user *buf,
         int ret = 0;
         struct inode *inode = file->f_path.dentry->d_inode;
         struct btrfs_root *root = BTRFS_I(inode)->root;
-        struct page *pages[1];
+        struct page *pages[8];
         unsigned long first_index;
         unsigned long last_index;
         u64 start_pos;
@@ -1359,6 +1599,7 @@ static ssize_t btrfs_file_write(struct file *file, const char __user *buf,
         u64 alloc_extent_start;
         u64 orig_extent_start;
         struct btrfs_trans_handle *trans;
+        struct btrfs_key ins;
 
         if (file->f_flags & O_DIRECT)
                 return -EINVAL;
@@ -1390,16 +1631,24 @@ static ssize_t btrfs_file_write(struct file *file, const char __user *buf,
         trans = btrfs_start_transaction(root, 1);
         if (!trans) {
                 err = -ENOMEM;
+                mutex_unlock(&root->fs_info->fs_mutex);
                 goto out_unlock;
         }
-        ret = btrfs_alloc_file_extent(trans, root, inode->i_ino,
-                                      start_pos, num_blocks, 1,
-                                      &alloc_extent_start);
-        BUG_ON(ret);
-
+        if (start_pos < inode->i_size) {
+                ret = drop_extents(trans, root, inode,
+                                   start_pos,
+                                   (pos + count + root->blocksize -1) &
+                                   ~(root->blocksize - 1));
+        }
         orig_extent_start = start_pos;
-        ret = btrfs_end_transaction(trans, root);
+        ret = btrfs_alloc_extent(trans, root, num_blocks, 1,
+                                 (u64)-1, &ins);
+        BUG_ON(ret);
+        ret = btrfs_insert_file_extent(trans, root, inode->i_ino,
+                                       start_pos, ins.objectid, ins.offset);
         BUG_ON(ret);
+        alloc_extent_start = ins.objectid;
+        ret = btrfs_end_transaction(trans, root);
         mutex_unlock(&root->fs_info->fs_mutex);
 
         while(count > 0) {
@@ -1407,16 +1656,21 @@ static ssize_t btrfs_file_write(struct file *file, const char __user *buf,
                 size_t write_bytes = min(count, PAGE_CACHE_SIZE - offset);
                 size_t num_pages = (write_bytes + PAGE_CACHE_SIZE - 1) >>
                                         PAGE_CACHE_SHIFT;
-                ret = prepare_pages(NULL, root, file, pages, num_pages,
+printk("num_pages is %lu\n", num_pages);
+
+                memset(pages, 0, sizeof(pages));
+                ret = prepare_pages(root, file, pages, num_pages,
                                     pos, first_index, last_index,
                                     write_bytes, alloc_extent_start);
                 BUG_ON(ret);
+
                 /* FIXME blocks != pagesize */
                 alloc_extent_start += num_pages;
                 ret = btrfs_copy_from_user(pos, num_pages,
                                            write_bytes, pages, buf);
                 BUG_ON(ret);
 
+printk("2num_pages is %lu\n", num_pages);
                 ret = dirty_and_release_pages(NULL, root, file, pages,
                                               num_pages, orig_extent_start,
                                               pos, write_bytes);