[PATCH 2/2] UBIFS: add a log overlap assertion

hujianyang hujianyang at huawei.com
Mon Jul 21 18:35:23 PDT 2014


On 2014/7/16 20:42, Artem Bityutskiy wrote:
> From: Artem Bityutskiy <artem.bityutskiy at linux.intel.com>
> 
> Add an assertion which checkes that the head of the log never overlaps with the
> tail of the log.
> 
> Suggested-by: hujianyang <hujianyang at huawei.com>
> Signed-off-by: Artem Bityutskiy <artem.bityutskiy at linux.intel.com>
> ---
>  fs/ubifs/misc.h | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/fs/ubifs/misc.h b/fs/ubifs/misc.h
> index ee7cb5e..81bbf79 100644
> --- a/fs/ubifs/misc.h
> +++ b/fs/ubifs/misc.h
> @@ -297,6 +297,7 @@ static inline int ubifs_next_log_lnum(const struct ubifs_info *c, int lnum)
>  	if (lnum > c->log_last)
>  		lnum = UBIFS_LOG_LNUM;
>  
> +	ubifs_assert(lnum != c->ltail_lnum);
>  	return lnum;
>  }
>  
> 

Hi Artem,

I just come back to work today.

I have to tell you this patch seems wrong. I've tested this patch
and saw lots of 'assert_failed()'. We use ubifs_next_log_lnum()
as an iterator to scan the hole log area and the return value of
this function may equal to c->ltail_lnum.

see:
	fixup_free_space() in fs/ubifs/sb.c line 712
	ubifs_log_post_commit() in fs/ubifs/log.c line 520
and so on.


How about my former patch or a new patch like this?


diff --git a/fs/ubifs/log.c b/fs/ubifs/log.c
index 7e818ec..c14628f 100644
--- a/fs/ubifs/log.c
+++ b/fs/ubifs/log.c
@@ -244,6 +244,7 @@ int ubifs_add_bud_to_log(struct ubifs_info *c, int jhead, int lnum, int offs)

 	if (c->lhead_offs > c->leb_size - c->ref_node_alsz) {
 		c->lhead_lnum = ubifs_next_log_lnum(c, c->lhead_lnum);
+		ubifs_assert(c->lhead_lnum != c->ltail_lnum);
 		c->lhead_offs = 0;
 	}

@@ -408,6 +409,7 @@ int ubifs_log_start_commit(struct ubifs_info *c, int *ltail_lnum)
 	/* Switch to the next log LEB */
 	if (c->lhead_offs) {
 		c->lhead_lnum = ubifs_next_log_lnum(c, c->lhead_lnum);
+		ubifs_assert(c->lhead_lnum != c->ltail_lnum);
 		c->lhead_offs = 0;
 	}

-- 
1.8.1.4








More information about the linux-mtd mailing list