Re: [PATCH] bcache: remove problematic code block from register_bcache()

[Coly Li <i@coly.li>]

On 2017/9/5 下午3:10, Coly Li wrote:
> There is a leak of bdev reference in register_bcache(), and Jan Kara
> posted a patch to fix it. Discussion happened on bcache mailing list,
> Christoph Hellwig pointed out that whole chunk of code where the bdev
> reference leak lied was problematic,
> 
> 	"adding a lookup_bdev and resulting mess just for a
> 	 slightly different error message is just insane"
> 
> This patch jsut removes whole chunk of the problematic code, of cause
> the leak of bdev reference is removed too.
> 
> Signed-off-by: Coly Li <colyli@suse.de>
> Cc: Jan Kara <jack@suse.cz>
> Cc: Christoph Hellwig <hch@infradead.org>
> ---
>  drivers/md/bcache/super.c | 14 +-------------
>  1 file changed, 1 insertion(+), 13 deletions(-)
> 
> diff --git a/drivers/md/bcache/super.c b/drivers/md/bcache/super.c
> index 8352fad765f6..85262f2628fb 100644
> --- a/drivers/md/bcache/super.c
> +++ b/drivers/md/bcache/super.c
> @@ -1955,20 +1955,8 @@ static ssize_t register_bcache(struct kobject *k, struct kobj_attribute *attr,
>  	bdev = blkdev_get_by_path(strim(path),
>  				  FMODE_READ|FMODE_WRITE|FMODE_EXCL,
>  				  sb);
> -	if (IS_ERR(bdev)) {
> -		if (bdev == ERR_PTR(-EBUSY)) {
> -			bdev = lookup_bdev(strim(path));
> -			mutex_lock(&bch_register_lock);
> -			if (!IS_ERR(bdev) && bch_is_open(bdev))
> -				err = "device already registered";
> -			else
> -				err = "device busy";
> -			mutex_unlock(&bch_register_lock);
> -			if (attr == &ksysfs_register_quiet)
> -				goto out;
> -		}
> +	if (IS_ERR(bdev))
>  		goto err;
> -	}
>  
>  	err = "failed to set blocksize";
>  	if (set_blocksize(bdev, 4096))
> 

Hi Jan and Christoph,

There are other three functions only referenced in the above chunk of code,
- bch_is_open()
- bch_is_open_cache()
- bch_is_open_backing()
they should be removed as well.

So you may ignore this email, an update version of this patch will be
posted to linux-bache and linux-block mailing lists for review.

Thanks.

-- 
Coly Li