From patchwork Mon Nov 5 18:40:14 2012 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Subject: jffs2: Make handling of erase_completion_lock consistent in jffs2_do_reserve_space() Date: Mon, 05 Nov 2012 08:40:14 -0000 From: Alexey Khoroshilov X-Patchwork-Id: 197272 Message-Id: <1352140814-25208-1-git-send-email-khoroshilov@ispras.ru> To: David Woodhouse Cc: ldv-project@linuxtesting.org, Artem Bityutskiy , linux-mtd@lists.infradead.org, linux-kernel@vger.kernel.org, Alexey Khoroshilov Users of jffs2_do_reserve_space() expect they still held erase_completion_lock after call to it. But there is a path where jffs2_do_reserve_space() leaves erase_completion_lock unlocked. The patch fixes it. Found by Linux Driver Verification project (linuxtesting.org). Signed-off-by: Alexey Khoroshilov --- fs/jffs2/nodemgmt.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/fs/jffs2/nodemgmt.c b/fs/jffs2/nodemgmt.c index 0c96eb5..0331072 100644 --- a/fs/jffs2/nodemgmt.c +++ b/fs/jffs2/nodemgmt.c @@ -417,14 +417,16 @@ static int jffs2_do_reserve_space(struct jffs2_sb_info *c, uint32_t minsize, spin_unlock(&c->erase_completion_lock); ret = jffs2_prealloc_raw_node_refs(c, jeb, 1); - if (ret) - return ret; + /* Just lock it again and continue. Nothing much can change because we hold c->alloc_sem anyway. In fact, it's not entirely clear why we hold c->erase_completion_lock in the majority of this function... but that's a question for another (more caffeine-rich) day. */ spin_lock(&c->erase_completion_lock); + if (ret) + return ret; + waste = jeb->free_size; jffs2_link_node_ref(c, jeb, (jeb->offset + c->sector_size - waste) | REF_OBSOLETE,