/*
* Copyright (C) 2005, 2006
- * Avishay Traeger (avishay@gmail.com) (avishay@il.ibm.com)
- * Copyright (C) 2005, 2006
- * International Business Machines
+ * Avishay Traeger (avishay@gmail.com)
* Copyright (C) 2008, 2009
* Boaz Harrosh <bharrosh@panasas.com>
*
# define EXOFS_DEBUG_OBJ_ISIZE 1
#endif
+#define EXOFS_DBGMSG2(M...) do {} while (0)
+
struct page_collect {
struct exofs_sb_info *sbi;
struct request_queue *req_q;
struct inode *inode)
{
struct exofs_sb_info *sbi = inode->i_sb->s_fs_info;
- struct request_queue *req_q = sbi->s_dev->scsi_device->request_queue;
pcol->sbi = sbi;
- pcol->req_q = req_q;
+ pcol->req_q = osd_request_queue(sbi->s_dev);
pcol->inode = inode;
pcol->expected_pages = expected_pages;
else
page_stat = ret;
- EXOFS_DBGMSG(" readpages_done(0x%lx, 0x%lx) %s\n",
+ EXOFS_DBGMSG2(" readpages_done(0x%lx, 0x%lx) %s\n",
inode->i_ino, page->index,
page_stat ? "bad_bytes" : "good_bytes");
goto err;
}
- osd_req_read(or, &obj, pcol->bio, i_start);
+ osd_req_read(or, &obj, i_start, pcol->bio, pcol->length);
if (is_sync) {
exofs_sync_op(or, pcol->sbi->s_timeout, oi->i_cred);
err:
if (!is_sync)
_unlock_pcol_pages(pcol, ret, READ);
+ else /* Pages unlocked by caller in sync mode only free bio */
+ pcol_free(pcol);
+
kfree(pcol_copy);
if (or)
osd_end_request(or);
if (len != PAGE_CACHE_SIZE)
zero_user(page, len, PAGE_CACHE_SIZE - len);
- EXOFS_DBGMSG(" readpage_strip(0x%lx, 0x%lx) len=0x%zx\n",
+ EXOFS_DBGMSG2(" readpage_strip(0x%lx, 0x%lx) len=0x%zx\n",
inode->i_ino, page->index, len);
ret = pcol_add_page(pcol, page, len);
if (ret) {
- EXOFS_DBGMSG("Failed pcol_add_page pages[i]=%p "
+ EXOFS_DBGMSG2("Failed pcol_add_page pages[i]=%p "
"this_len=0x%zx nr_pages=%u length=0x%lx\n",
page, len, pcol->nr_pages, pcol->length);
update_write_page(page, page_stat);
unlock_page(page);
- EXOFS_DBGMSG(" writepages_done(0x%lx, 0x%lx) status=%d\n",
+ EXOFS_DBGMSG2(" writepages_done(0x%lx, 0x%lx) status=%d\n",
inode->i_ino, page->index, page_stat);
length += bvec->bv_len;
*pcol_copy = *pcol;
- osd_req_write(or, &obj, pcol_copy->bio, i_start);
+ pcol_copy->bio->bi_rw |= (1 << BIO_RW); /* FIXME: bio_set_dir() */
+ osd_req_write(or, &obj, i_start, pcol_copy->bio, pcol_copy->length);
ret = exofs_async_op(or, writepages_done, pcol_copy, oi->i_cred);
if (unlikely(ret)) {
EXOFS_ERR("write_exec: exofs_async_op() Faild\n");
goto fail;
}
- EXOFS_DBGMSG(" writepage_strip(0x%lx, 0x%lx) len=0x%zx\n",
+ EXOFS_DBGMSG2(" writepage_strip(0x%lx, 0x%lx) len=0x%zx\n",
inode->i_ino, page->index, len);
ret = pcol_add_page(pcol, page, len);
error = inode_setattr(inode, iattr);
return error;
}
+
+/*
+ * Read an inode from the OSD, and return it as is. We also return the size
+ * attribute in the 'sanity' argument if we got compiled with debugging turned
+ * on.
+ */
+static int exofs_get_inode(struct super_block *sb, struct exofs_i_info *oi,
+ struct exofs_fcb *inode, uint64_t *sanity)
+{
+ struct exofs_sb_info *sbi = sb->s_fs_info;
+ struct osd_request *or;
+ struct osd_attr attr;
+ struct osd_obj_id obj = {sbi->s_pid,
+ oi->vfs_inode.i_ino + EXOFS_OBJ_OFF};
+ int ret;
+
+ exofs_make_credential(oi->i_cred, &obj);
+
+ or = osd_start_request(sbi->s_dev, GFP_KERNEL);
+ if (unlikely(!or)) {
+ EXOFS_ERR("exofs_get_inode: osd_start_request failed.\n");
+ return -ENOMEM;
+ }
+ osd_req_get_attributes(or, &obj);
+
+ /* we need the inode attribute */
+ osd_req_add_get_attr_list(or, &g_attr_inode_data, 1);
+
+#ifdef EXOFS_DEBUG_OBJ_ISIZE
+ /* we get the size attributes to do a sanity check */
+ osd_req_add_get_attr_list(or, &g_attr_logical_length, 1);
+#endif
+
+ ret = exofs_sync_op(or, sbi->s_timeout, oi->i_cred);
+ if (ret)
+ goto out;
+
+ attr = g_attr_inode_data;
+ ret = extract_attr_from_req(or, &attr);
+ if (ret) {
+ EXOFS_ERR("exofs_get_inode: extract_attr_from_req failed\n");
+ goto out;
+ }
+
+ WARN_ON(attr.len != EXOFS_INO_ATTR_SIZE);
+ memcpy(inode, attr.val_ptr, EXOFS_INO_ATTR_SIZE);
+
+#ifdef EXOFS_DEBUG_OBJ_ISIZE
+ attr = g_attr_logical_length;
+ ret = extract_attr_from_req(or, &attr);
+ if (ret) {
+ EXOFS_ERR("ERROR: extract attr from or failed\n");
+ goto out;
+ }
+ *sanity = get_unaligned_be64(attr.val_ptr);
+#endif
+
+out:
+ osd_end_request(or);
+ return ret;
+}
+
+/*
+ * Fill in an inode read from the OSD and set it up for use
+ */
+struct inode *exofs_iget(struct super_block *sb, unsigned long ino)
+{
+ struct exofs_i_info *oi;
+ struct exofs_fcb fcb;
+ struct inode *inode;
+ uint64_t uninitialized_var(sanity);
+ int ret;
+
+ inode = iget_locked(sb, ino);
+ if (!inode)
+ return ERR_PTR(-ENOMEM);
+ if (!(inode->i_state & I_NEW))
+ return inode;
+ oi = exofs_i(inode);
+
+ /* read the inode from the osd */
+ ret = exofs_get_inode(sb, oi, &fcb, &sanity);
+ if (ret)
+ goto bad_inode;
+
+ init_waitqueue_head(&oi->i_wq);
+ set_obj_created(oi);
+
+ /* copy stuff from on-disk struct to in-memory struct */
+ inode->i_mode = le16_to_cpu(fcb.i_mode);
+ inode->i_uid = le32_to_cpu(fcb.i_uid);
+ inode->i_gid = le32_to_cpu(fcb.i_gid);
+ inode->i_nlink = le16_to_cpu(fcb.i_links_count);
+ inode->i_ctime.tv_sec = (signed)le32_to_cpu(fcb.i_ctime);
+ inode->i_atime.tv_sec = (signed)le32_to_cpu(fcb.i_atime);
+ inode->i_mtime.tv_sec = (signed)le32_to_cpu(fcb.i_mtime);
+ inode->i_ctime.tv_nsec =
+ inode->i_atime.tv_nsec = inode->i_mtime.tv_nsec = 0;
+ oi->i_commit_size = le64_to_cpu(fcb.i_size);
+ i_size_write(inode, oi->i_commit_size);
+ inode->i_blkbits = EXOFS_BLKSHIFT;
+ inode->i_generation = le32_to_cpu(fcb.i_generation);
+
+#ifdef EXOFS_DEBUG_OBJ_ISIZE
+ if ((inode->i_size != sanity) &&
+ (!exofs_inode_is_fast_symlink(inode))) {
+ EXOFS_ERR("WARNING: Size of inode=%llu != object=%llu\n",
+ inode->i_size, _LLU(sanity));
+ }
+#endif
+
+ oi->i_dir_start_lookup = 0;
+
+ if ((inode->i_nlink == 0) && (inode->i_mode == 0)) {
+ ret = -ESTALE;
+ goto bad_inode;
+ }
+
+ if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode)) {
+ if (fcb.i_data[0])
+ inode->i_rdev =
+ old_decode_dev(le32_to_cpu(fcb.i_data[0]));
+ else
+ inode->i_rdev =
+ new_decode_dev(le32_to_cpu(fcb.i_data[1]));
+ } else {
+ memcpy(oi->i_data, fcb.i_data, sizeof(fcb.i_data));
+ }
+
+ if (S_ISREG(inode->i_mode)) {
+ inode->i_op = &exofs_file_inode_operations;
+ inode->i_fop = &exofs_file_operations;
+ inode->i_mapping->a_ops = &exofs_aops;
+ } else if (S_ISDIR(inode->i_mode)) {
+ inode->i_op = &exofs_dir_inode_operations;
+ inode->i_fop = &exofs_dir_operations;
+ inode->i_mapping->a_ops = &exofs_aops;
+ } else if (S_ISLNK(inode->i_mode)) {
+ if (exofs_inode_is_fast_symlink(inode))
+ inode->i_op = &exofs_fast_symlink_inode_operations;
+ else {
+ inode->i_op = &exofs_symlink_inode_operations;
+ inode->i_mapping->a_ops = &exofs_aops;
+ }
+ } else {
+ inode->i_op = &exofs_special_inode_operations;
+ if (fcb.i_data[0])
+ init_special_inode(inode, inode->i_mode,
+ old_decode_dev(le32_to_cpu(fcb.i_data[0])));
+ else
+ init_special_inode(inode, inode->i_mode,
+ new_decode_dev(le32_to_cpu(fcb.i_data[1])));
+ }
+
+ unlock_new_inode(inode);
+ return inode;
+
+bad_inode:
+ iget_failed(inode);
+ return ERR_PTR(ret);
+}
+
+int __exofs_wait_obj_created(struct exofs_i_info *oi)
+{
+ if (!obj_created(oi)) {
+ BUG_ON(!obj_2bcreated(oi));
+ wait_event(oi->i_wq, obj_created(oi));
+ }
+ return unlikely(is_bad_inode(&oi->vfs_inode)) ? -EIO : 0;
+}
+/*
+ * Callback function from exofs_new_inode(). The important thing is that we
+ * set the obj_created flag so that other methods know that the object exists on
+ * the OSD.
+ */
+static void create_done(struct osd_request *or, void *p)
+{
+ struct inode *inode = p;
+ struct exofs_i_info *oi = exofs_i(inode);
+ struct exofs_sb_info *sbi = inode->i_sb->s_fs_info;
+ int ret;
+
+ ret = exofs_check_ok(or);
+ osd_end_request(or);
+ atomic_dec(&sbi->s_curr_pending);
+
+ if (unlikely(ret)) {
+ EXOFS_ERR("object=0x%llx creation faild in pid=0x%llx",
+ _LLU(sbi->s_pid), _LLU(inode->i_ino + EXOFS_OBJ_OFF));
+ make_bad_inode(inode);
+ } else
+ set_obj_created(oi);
+
+ atomic_dec(&inode->i_count);
+ wake_up(&oi->i_wq);
+}
+
+/*
+ * Set up a new inode and create an object for it on the OSD
+ */
+struct inode *exofs_new_inode(struct inode *dir, int mode)
+{
+ struct super_block *sb;
+ struct inode *inode;
+ struct exofs_i_info *oi;
+ struct exofs_sb_info *sbi;
+ struct osd_request *or;
+ struct osd_obj_id obj;
+ int ret;
+
+ sb = dir->i_sb;
+ inode = new_inode(sb);
+ if (!inode)
+ return ERR_PTR(-ENOMEM);
+
+ oi = exofs_i(inode);
+
+ init_waitqueue_head(&oi->i_wq);
+ set_obj_2bcreated(oi);
+
+ sbi = sb->s_fs_info;
+
+ sb->s_dirt = 1;
+ inode->i_uid = current->cred->fsuid;
+ if (dir->i_mode & S_ISGID) {
+ inode->i_gid = dir->i_gid;
+ if (S_ISDIR(mode))
+ mode |= S_ISGID;
+ } else {
+ inode->i_gid = current->cred->fsgid;
+ }
+ inode->i_mode = mode;
+
+ inode->i_ino = sbi->s_nextid++;
+ inode->i_blkbits = EXOFS_BLKSHIFT;
+ inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
+ oi->i_commit_size = inode->i_size = 0;
+ spin_lock(&sbi->s_next_gen_lock);
+ inode->i_generation = sbi->s_next_generation++;
+ spin_unlock(&sbi->s_next_gen_lock);
+ insert_inode_hash(inode);
+
+ mark_inode_dirty(inode);
+
+ obj.partition = sbi->s_pid;
+ obj.id = inode->i_ino + EXOFS_OBJ_OFF;
+ exofs_make_credential(oi->i_cred, &obj);
+
+ or = osd_start_request(sbi->s_dev, GFP_KERNEL);
+ if (unlikely(!or)) {
+ EXOFS_ERR("exofs_new_inode: osd_start_request failed\n");
+ return ERR_PTR(-ENOMEM);
+ }
+
+ osd_req_create_object(or, &obj);
+
+ /* increment the refcount so that the inode will still be around when we
+ * reach the callback
+ */
+ atomic_inc(&inode->i_count);
+
+ ret = exofs_async_op(or, create_done, inode, oi->i_cred);
+ if (ret) {
+ atomic_dec(&inode->i_count);
+ osd_end_request(or);
+ return ERR_PTR(-EIO);
+ }
+ atomic_inc(&sbi->s_curr_pending);
+
+ return inode;
+}
+
+/*
+ * struct to pass two arguments to update_inode's callback
+ */
+struct updatei_args {
+ struct exofs_sb_info *sbi;
+ struct exofs_fcb fcb;
+};
+
+/*
+ * Callback function from exofs_update_inode().
+ */
+static void updatei_done(struct osd_request *or, void *p)
+{
+ struct updatei_args *args = p;
+
+ osd_end_request(or);
+
+ atomic_dec(&args->sbi->s_curr_pending);
+
+ kfree(args);
+}
+
+/*
+ * Write the inode to the OSD. Just fill up the struct, and set the attribute
+ * synchronously or asynchronously depending on the do_sync flag.
+ */
+static int exofs_update_inode(struct inode *inode, int do_sync)
+{
+ struct exofs_i_info *oi = exofs_i(inode);
+ struct super_block *sb = inode->i_sb;
+ struct exofs_sb_info *sbi = sb->s_fs_info;
+ struct osd_obj_id obj = {sbi->s_pid, inode->i_ino + EXOFS_OBJ_OFF};
+ struct osd_request *or;
+ struct osd_attr attr;
+ struct exofs_fcb *fcb;
+ struct updatei_args *args;
+ int ret;
+
+ args = kzalloc(sizeof(*args), GFP_KERNEL);
+ if (!args)
+ return -ENOMEM;
+
+ fcb = &args->fcb;
+
+ fcb->i_mode = cpu_to_le16(inode->i_mode);
+ fcb->i_uid = cpu_to_le32(inode->i_uid);
+ fcb->i_gid = cpu_to_le32(inode->i_gid);
+ fcb->i_links_count = cpu_to_le16(inode->i_nlink);
+ fcb->i_ctime = cpu_to_le32(inode->i_ctime.tv_sec);
+ fcb->i_atime = cpu_to_le32(inode->i_atime.tv_sec);
+ fcb->i_mtime = cpu_to_le32(inode->i_mtime.tv_sec);
+ oi->i_commit_size = i_size_read(inode);
+ fcb->i_size = cpu_to_le64(oi->i_commit_size);
+ fcb->i_generation = cpu_to_le32(inode->i_generation);
+
+ if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode)) {
+ if (old_valid_dev(inode->i_rdev)) {
+ fcb->i_data[0] =
+ cpu_to_le32(old_encode_dev(inode->i_rdev));
+ fcb->i_data[1] = 0;
+ } else {
+ fcb->i_data[0] = 0;
+ fcb->i_data[1] =
+ cpu_to_le32(new_encode_dev(inode->i_rdev));
+ fcb->i_data[2] = 0;
+ }
+ } else
+ memcpy(fcb->i_data, oi->i_data, sizeof(fcb->i_data));
+
+ or = osd_start_request(sbi->s_dev, GFP_KERNEL);
+ if (unlikely(!or)) {
+ EXOFS_ERR("exofs_update_inode: osd_start_request failed.\n");
+ ret = -ENOMEM;
+ goto free_args;
+ }
+
+ osd_req_set_attributes(or, &obj);
+
+ attr = g_attr_inode_data;
+ attr.val_ptr = fcb;
+ osd_req_add_set_attr_list(or, &attr, 1);
+
+ if (!obj_created(oi)) {
+ EXOFS_DBGMSG("!obj_created\n");
+ BUG_ON(!obj_2bcreated(oi));
+ wait_event(oi->i_wq, obj_created(oi));
+ EXOFS_DBGMSG("wait_event done\n");
+ }
+
+ if (do_sync) {
+ ret = exofs_sync_op(or, sbi->s_timeout, oi->i_cred);
+ osd_end_request(or);
+ goto free_args;
+ } else {
+ args->sbi = sbi;
+
+ ret = exofs_async_op(or, updatei_done, args, oi->i_cred);
+ if (ret) {
+ osd_end_request(or);
+ goto free_args;
+ }
+ atomic_inc(&sbi->s_curr_pending);
+ goto out; /* deallocation in updatei_done */
+ }
+
+free_args:
+ kfree(args);
+out:
+ EXOFS_DBGMSG("ret=>%d\n", ret);
+ return ret;
+}
+
+int exofs_write_inode(struct inode *inode, int wait)
+{
+ return exofs_update_inode(inode, wait);
+}
+
+/*
+ * Callback function from exofs_delete_inode() - don't have much cleaning up to
+ * do.
+ */
+static void delete_done(struct osd_request *or, void *p)
+{
+ struct exofs_sb_info *sbi;
+ osd_end_request(or);
+ sbi = p;
+ atomic_dec(&sbi->s_curr_pending);
+}
+
+/*
+ * Called when the refcount of an inode reaches zero. We remove the object
+ * from the OSD here. We make sure the object was created before we try and
+ * delete it.
+ */
+void exofs_delete_inode(struct inode *inode)
+{
+ struct exofs_i_info *oi = exofs_i(inode);
+ struct super_block *sb = inode->i_sb;
+ struct exofs_sb_info *sbi = sb->s_fs_info;
+ struct osd_obj_id obj = {sbi->s_pid, inode->i_ino + EXOFS_OBJ_OFF};
+ struct osd_request *or;
+ int ret;
+
+ truncate_inode_pages(&inode->i_data, 0);
+
+ if (is_bad_inode(inode))
+ goto no_delete;
+
+ mark_inode_dirty(inode);
+ exofs_update_inode(inode, inode_needs_sync(inode));
+
+ inode->i_size = 0;
+ if (inode->i_blocks)
+ exofs_truncate(inode);
+
+ clear_inode(inode);
+
+ or = osd_start_request(sbi->s_dev, GFP_KERNEL);
+ if (unlikely(!or)) {
+ EXOFS_ERR("exofs_delete_inode: osd_start_request failed\n");
+ return;
+ }
+
+ osd_req_remove_object(or, &obj);
+
+ /* if we are deleting an obj that hasn't been created yet, wait */
+ if (!obj_created(oi)) {
+ BUG_ON(!obj_2bcreated(oi));
+ wait_event(oi->i_wq, obj_created(oi));
+ }
+
+ ret = exofs_async_op(or, delete_done, sbi, oi->i_cred);
+ if (ret) {
+ EXOFS_ERR(
+ "ERROR: @exofs_delete_inode exofs_async_op failed\n");
+ osd_end_request(or);
+ return;
+ }
+ atomic_inc(&sbi->s_curr_pending);
+
+ return;
+
+no_delete:
+ clear_inode(inode);
+}
projects / linux-2.6.git / blobdiff