From patchwork Sat Jun 29 23:46:48 2013 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Theodore Ts'o X-Patchwork-Id: 255806 Return-Path: X-Original-To: patchwork-incoming@ozlabs.org Delivered-To: patchwork-incoming@ozlabs.org Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by ozlabs.org (Postfix) with ESMTP id EACD42C02C5 for ; Sun, 30 Jun 2013 09:47:00 +1000 (EST) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753435Ab3F2Xq7 (ORCPT ); Sat, 29 Jun 2013 19:46:59 -0400 Received: from li9-11.members.linode.com ([67.18.176.11]:33998 "EHLO imap.thunk.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751431Ab3F2Xq7 (ORCPT ); Sat, 29 Jun 2013 19:46:59 -0400 Received: from root (helo=closure.thunk.org) by imap.thunk.org with local-esmtp (Exim 4.80) (envelope-from ) id 1Ut4wI-0005o7-GN; Sat, 29 Jun 2013 23:52:22 +0000 Received: by closure.thunk.org (Postfix, from userid 15806) id 37E1D5808F8; Sat, 29 Jun 2013 19:46:48 -0400 (EDT) Date: Sat, 29 Jun 2013 19:46:48 -0400 From: Theodore Ts'o To: Younger Liu Cc: Ext4 Developers List , jbacik@fusionio.com, Jan Kara Subject: Re: [PATCH] jbd2: invalidate handle if jbd2_journal_restart() fails Message-ID: <20130629234648.GA4560@thunk.org> References: <20130623173628.GC16620@thunk.org> <1372009468-11651-1-git-send-email-tytso@mit.edu> <51C965F2.3090704@huawei.com> MIME-Version: 1.0 Content-Disposition: inline In-Reply-To: <51C965F2.3090704@huawei.com> User-Agent: Mutt/1.5.21 (2010-09-15) X-SA-Exim-Connect-IP: X-SA-Exim-Mail-From: tytso@thunk.org X-SA-Exim-Scanned: No (on imap.thunk.org); SAEximRunCond expanded to false Sender: linux-ext4-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-ext4@vger.kernel.org On Tue, Jun 25, 2013 at 05:42:10PM +0800, Younger Liu wrote: > If jbd2__journal_restart fails, handle->h_transaction may be NULL. > So we should check handle->h_transaction before > "journal = transaction->t_journal", Right? Good point, yes. Here's a new version of the patch which I have in my tree. - Ted From fd8d369f9ad921eb6dc5c56e87e4c6d6106bad56 Mon Sep 17 00:00:00 2001 From: Theodore Ts'o Date: Sun, 23 Jun 2013 12:59:01 -0400 Subject: [PATCH] jbd2: invalidate handle if jbd2_journal_restart() fails If jbd2_journal_restart() fails the handle will have been disconnected from the current transaction. In this situation, the handle must not be used for for any jbd2 function other than jbd2_journal_stop(). Enforce this with by treating a handle which has a NULL transaction pointer as an aborted handle, and issue a kernel warning if jbd2_journal_extent(), jbd2_journal_get_write_access(), jbd2_journal_dirty_metadata(), etc. is called with an invalid handle. This commit also fixes a bug where jbd2_journal_stop() would trip over a kernel jbd2 assertion check when trying to free an invalid handle. Also move the responsibility of setting current->journal_info to start_this_handle(), simplifying the three users of this function. Signed-off-by: "Theodore Ts'o" Reported-by: Younger Liu Cc: Jan Kara Tested-by: Younger Liu Tested-by: Younger Liu --- fs/jbd2/transaction.c | 74 ++++++++++++++++++++++++++++++--------------------- include/linux/jbd2.h | 2 +- 2 files changed, 44 insertions(+), 32 deletions(-) diff --git a/fs/jbd2/transaction.c b/fs/jbd2/transaction.c index 383b0fb..7aa9a32 100644 --- a/fs/jbd2/transaction.c +++ b/fs/jbd2/transaction.c @@ -368,6 +368,7 @@ repeat: atomic_read(&transaction->t_outstanding_credits), jbd2_log_space_left(journal)); read_unlock(&journal->j_state_lock); + current->journal_info = handle; lock_map_acquire(&handle->h_lockdep_map); jbd2_journal_free_transaction(new_transaction); @@ -442,14 +443,11 @@ handle_t *jbd2__journal_start(journal_t *journal, int nblocks, int rsv_blocks, handle->h_rsv_handle = rsv_handle; } - current->journal_info = handle; - err = start_this_handle(journal, handle, gfp_mask); if (err < 0) { if (handle->h_rsv_handle) jbd2_free_handle(handle->h_rsv_handle); jbd2_free_handle(handle); - current->journal_info = NULL; return ERR_PTR(err); } handle->h_type = type; @@ -511,16 +509,13 @@ int jbd2_journal_start_reserved(handle_t *handle, unsigned int type, } handle->h_journal = NULL; - current->journal_info = handle; /* * GFP_NOFS is here because callers are likely from writeback or * similarly constrained call sites */ ret = start_this_handle(journal, handle, GFP_NOFS); - if (ret < 0) { - current->journal_info = NULL; + if (ret < 0) jbd2_journal_free_reserved(handle); - } handle->h_type = type; handle->h_line_no = line_no; return ret; @@ -550,20 +545,21 @@ EXPORT_SYMBOL(jbd2_journal_start_reserved); int jbd2_journal_extend(handle_t *handle, int nblocks) { transaction_t *transaction = handle->h_transaction; - journal_t *journal = transaction->t_journal; + journal_t *journal; int result; int wanted; - result = -EIO; + WARN_ON(!transaction); if (is_handle_aborted(handle)) - goto out; + return -EROFS; + journal = transaction->t_journal; result = 1; read_lock(&journal->j_state_lock); /* Don't extend a locked-down transaction! */ - if (handle->h_transaction->t_state != T_RUNNING) { + if (transaction->t_state != T_RUNNING) { jbd_debug(3, "denied handle %p %d blocks: " "transaction not running\n", handle, nblocks); goto error_out; @@ -589,7 +585,7 @@ int jbd2_journal_extend(handle_t *handle, int nblocks) } trace_jbd2_handle_extend(journal->j_fs_dev->bd_dev, - handle->h_transaction->t_tid, + transaction->t_tid, handle->h_type, handle->h_line_no, handle->h_buffer_credits, nblocks); @@ -603,7 +599,6 @@ unlock: spin_unlock(&transaction->t_handle_lock); error_out: read_unlock(&journal->j_state_lock); -out: return result; } @@ -626,14 +621,16 @@ out: int jbd2__journal_restart(handle_t *handle, int nblocks, gfp_t gfp_mask) { transaction_t *transaction = handle->h_transaction; - journal_t *journal = transaction->t_journal; + journal_t *journal; tid_t tid; int need_to_start, ret; + WARN_ON(!transaction); /* If we've had an abort of any type, don't even think about * actually doing the restart! */ if (is_handle_aborted(handle)) return 0; + journal = transaction->t_journal; /* * First unlink the handle from its current transaction, and start the @@ -654,6 +651,8 @@ int jbd2__journal_restart(handle_t *handle, int nblocks, gfp_t gfp_mask) wake_up(&journal->j_wait_updates); tid = transaction->t_tid; spin_unlock(&transaction->t_handle_lock); + handle->h_transaction = NULL; + current->journal_info = NULL; jbd_debug(2, "restarting handle %p\n", handle); need_to_start = !tid_geq(journal->j_commit_request, tid); @@ -783,17 +782,16 @@ do_get_write_access(handle_t *handle, struct journal_head *jh, int force_copy) { struct buffer_head *bh; - transaction_t *transaction; + transaction_t *transaction = handle->h_transaction; journal_t *journal; int error; char *frozen_buffer = NULL; int need_copy = 0; unsigned long start_lock, time_lock; + WARN_ON(!transaction); if (is_handle_aborted(handle)) return -EROFS; - - transaction = handle->h_transaction; journal = transaction->t_journal; jbd_debug(5, "journal_head %p, force_copy %d\n", jh, force_copy); @@ -1052,14 +1050,16 @@ int jbd2_journal_get_write_access(handle_t *handle, struct buffer_head *bh) int jbd2_journal_get_create_access(handle_t *handle, struct buffer_head *bh) { transaction_t *transaction = handle->h_transaction; - journal_t *journal = transaction->t_journal; + journal_t *journal; struct journal_head *jh = jbd2_journal_add_journal_head(bh); int err; jbd_debug(5, "journal_head %p\n", jh); + WARN_ON(!transaction); err = -EROFS; if (is_handle_aborted(handle)) goto out; + journal = transaction->t_journal; err = 0; JBUFFER_TRACE(jh, "entry"); @@ -1265,12 +1265,14 @@ void jbd2_buffer_abort_trigger(struct journal_head *jh, int jbd2_journal_dirty_metadata(handle_t *handle, struct buffer_head *bh) { transaction_t *transaction = handle->h_transaction; - journal_t *journal = transaction->t_journal; + journal_t *journal; struct journal_head *jh; int ret = 0; + WARN_ON(!transaction); if (is_handle_aborted(handle)) - goto out; + return -EROFS; + journal = transaction->t_journal; jh = jbd2_journal_grab_journal_head(bh); if (!jh) { ret = -EUCLEAN; @@ -1364,7 +1366,7 @@ int jbd2_journal_dirty_metadata(handle_t *handle, struct buffer_head *bh) JBUFFER_TRACE(jh, "file as BJ_Metadata"); spin_lock(&journal->j_list_lock); - __jbd2_journal_file_buffer(jh, handle->h_transaction, BJ_Metadata); + __jbd2_journal_file_buffer(jh, transaction, BJ_Metadata); spin_unlock(&journal->j_list_lock); out_unlock_bh: jbd_unlock_bh_state(bh); @@ -1395,12 +1397,17 @@ out: int jbd2_journal_forget (handle_t *handle, struct buffer_head *bh) { transaction_t *transaction = handle->h_transaction; - journal_t *journal = transaction->t_journal; + journal_t *journal; struct journal_head *jh; int drop_reserve = 0; int err = 0; int was_modified = 0; + WARN_ON(!transaction); + if (is_handle_aborted(handle)) + return -EROFS; + journal = transaction->t_journal; + BUFFER_TRACE(bh, "entry"); jbd_lock_bh_state(bh); @@ -1427,7 +1434,7 @@ int jbd2_journal_forget (handle_t *handle, struct buffer_head *bh) */ jh->b_modified = 0; - if (jh->b_transaction == handle->h_transaction) { + if (jh->b_transaction == transaction) { J_ASSERT_JH(jh, !jh->b_frozen_data); /* If we are forgetting a buffer which is already part @@ -1522,19 +1529,21 @@ drop: int jbd2_journal_stop(handle_t *handle) { transaction_t *transaction = handle->h_transaction; - journal_t *journal = transaction->t_journal; - int err, wait_for_commit = 0; + journal_t *journal; + int err = 0, wait_for_commit = 0; tid_t tid; pid_t pid; + if (!transaction) + goto free_and_exit; + journal = transaction->t_journal; + J_ASSERT(journal_current_handle() == handle); if (is_handle_aborted(handle)) err = -EIO; - else { + else J_ASSERT(atomic_read(&transaction->t_updates) > 0); - err = 0; - } if (--handle->h_ref > 0) { jbd_debug(4, "h_ref %d -> %d\n", handle->h_ref + 1, @@ -1544,7 +1553,7 @@ int jbd2_journal_stop(handle_t *handle) jbd_debug(4, "Handle %p going down\n", handle); trace_jbd2_handle_stats(journal->j_fs_dev->bd_dev, - handle->h_transaction->t_tid, + transaction->t_tid, handle->h_type, handle->h_line_no, jiffies - handle->h_start_jiffies, handle->h_sync, handle->h_requested_credits, @@ -1657,6 +1666,7 @@ int jbd2_journal_stop(handle_t *handle) if (handle->h_rsv_handle) jbd2_journal_free_reserved(handle->h_rsv_handle); +free_and_exit: jbd2_free_handle(handle); return err; } @@ -2362,10 +2372,12 @@ void jbd2_journal_refile_buffer(journal_t *journal, struct journal_head *jh) int jbd2_journal_file_inode(handle_t *handle, struct jbd2_inode *jinode) { transaction_t *transaction = handle->h_transaction; - journal_t *journal = transaction->t_journal; + journal_t *journal; + WARN_ON(!transaction); if (is_handle_aborted(handle)) - return -EIO; + return -EROFS; + journal = transaction->t_journal; jbd_debug(4, "Adding inode %lu, tid:%d\n", jinode->i_vfs_inode->i_ino, transaction->t_tid); diff --git a/include/linux/jbd2.h b/include/linux/jbd2.h index 0302f3f..d5b50a1 100644 --- a/include/linux/jbd2.h +++ b/include/linux/jbd2.h @@ -1266,7 +1266,7 @@ static inline int is_journal_aborted(journal_t *journal) static inline int is_handle_aborted(handle_t *handle) { - if (handle->h_aborted) + if (handle->h_aborted || !handle->h_transaction) return 1; return is_journal_aborted(handle->h_transaction->t_journal); }