JFFS2: Return values of jffs2_block_check_erase error paths

JFFS2: Return values of jffs2_block_check_erase error paths

[Anders Grafström]

It looks the error paths in jffs2_block_check_erase() have wrong return values.
A block that failed to be erased never gets marked as bad.

Any comments on this patch?

Anders


diff --git a/fs/jffs2/erase.c b/fs/jffs2/erase.c
index a1db918..23e9dd6 100644
--- a/fs/jffs2/erase.c
+++ b/fs/jffs2/erase.c
@@ -350,9 +350,11 @@ static int jffs2_block_check_erase(struct jffs2_sb_info *c, struct jffs2_erasebl
 			   break;
 		} while(--retlen);
 		c->mtd->unpoint(c->mtd, ebuf, jeb->offset, c->sector_size);
-		if (retlen)
+		if (retlen) {
 			printk(KERN_WARNING "Newly-erased block contained word 0x%lx at offset 0x%08tx\n",
 			       *wordebuf, jeb->offset + c->sector_size-retlen*sizeof(*wordebuf));
+			return -EIO;
+		}
 		return 0;
 	}
  do_flash_read:
@@ -373,10 +375,12 @@ static int jffs2_block_check_erase(struct jffs2_sb_info *c, struct jffs2_erasebl
 		ret = c->mtd->read(c->mtd, ofs, readlen, &retlen, ebuf);
 		if (ret) {
 			printk(KERN_WARNING "Read of newly-erased block at 0x%08x failed: %d. Putting on bad_list\n", ofs, ret);
+			ret = -EIO;
 			goto fail;
 		}
 		if (retlen != readlen) {
 			printk(KERN_WARNING "Short read from newly-erased block at 0x%08x. Wanted %d, got %zd\n", ofs, readlen, retlen);
+			ret = -EIO;
 			goto fail;
 		}
 		for (i=0; i<readlen; i += sizeof(unsigned long)) {
@@ -385,6 +389,7 @@ static int jffs2_block_check_erase(struct jffs2_sb_info *c, struct jffs2_erasebl
 			if (*datum + 1) {
 				*bad_offset += i;
 				printk(KERN_WARNING "Newly-erased block contained word 0x%lx at offset 0x%08x\n", *datum, *bad_offset);
+				ret = -EIO;
 				goto fail;
 			}
 		}

Re: JFFS2: Return values of jffs2_block_check_erase error paths

[Anders Grafström]

Anders Grafström wrote:
> It looks like the error paths in jffs2_block_check_erase() have wrong return values.
> A block that failed to be erased never gets marked as bad.
> 
> Any comments on this patch?
> 
> Anders
> 
> 
> diff --git a/fs/jffs2/erase.c b/fs/jffs2/erase.c
> index a1db918..23e9dd6 100644
> --- a/fs/jffs2/erase.c
> +++ b/fs/jffs2/erase.c
> @@ -350,9 +350,11 @@ static int jffs2_block_check_erase(struct jffs2_sb_info *c, struct jffs2_erasebl
>  			   break;
>  		} while(--retlen);
>  		c->mtd->unpoint(c->mtd, ebuf, jeb->offset, c->sector_size);
> -		if (retlen)
> +		if (retlen) {
>  			printk(KERN_WARNING "Newly-erased block contained word 0x%lx at offset 0x%08tx\n",
>  			       *wordebuf, jeb->offset + c->sector_size-retlen*sizeof(*wordebuf));
> +			return -EIO;
> +		}
>  		return 0;
>  	}
>   do_flash_read:
> @@ -373,10 +375,12 @@ static int jffs2_block_check_erase(struct jffs2_sb_info *c, struct jffs2_erasebl
>  		ret = c->mtd->read(c->mtd, ofs, readlen, &retlen, ebuf);
>  		if (ret) {
>  			printk(KERN_WARNING "Read of newly-erased block at 0x%08x failed: %d. Putting on bad_list\n", ofs, ret);
> +			ret = -EIO;
>  			goto fail;
>  		}
>  		if (retlen != readlen) {
>  			printk(KERN_WARNING "Short read from newly-erased block at 0x%08x. Wanted %d, got %zd\n", ofs, readlen, retlen);
> +			ret = -EIO;
>  			goto fail;
>  		}
>  		for (i=0; i<readlen; i += sizeof(unsigned long)) {
> @@ -385,6 +389,7 @@ static int jffs2_block_check_erase(struct jffs2_sb_info *c, struct jffs2_erasebl
>  			if (*datum + 1) {
>  				*bad_offset += i;
>  				printk(KERN_WARNING "Newly-erased block contained word 0x%lx at offset 0x%08x\n", *datum, *bad_offset);
> +				ret = -EIO;
>  				goto fail;
>  			}
>  		}


Nobody looked at this?
This issue caused a corrupt file system.

/Anders

RE: JFFS2: Return values of jffs2_block_check_erase error paths

[Joakim Tjernlund]

> -----Original Message-----
> From: linux-mtd-bounces@lists.infradead.org [mailto:linux-mtd-bounces@lists.infradead.org] On Behalf
> Of Anders Grafström
> Sent: den 10 april 2008 22:08
> To: linux-mtd@lists.infradead.org
> Subject: Re: JFFS2: Return values of jffs2_block_check_erase error paths
> 
> Anders Grafström wrote:
> > It looks like the error paths in jffs2_block_check_erase() have wrong return values.
> > A block that failed to be erased never gets marked as bad.
> >
> > Any comments on this patch?
> >
> > Anders
> >
> >
> > diff --git a/fs/jffs2/erase.c b/fs/jffs2/erase.c
> > index a1db918..23e9dd6 100644
> > --- a/fs/jffs2/erase.c


[SNIP]

> 
> 
> Nobody looked at this?
> This issue caused a corrupt file system.
> 
> /Anders

Looks good to me, but I think you should add your own signoff, write a
proper commit message and send it to Linus/Andrew as David appears busy.

Signed-off-by: Joakim Tjernlund <Joakim.Tjenrlund@transmode.se>