Received-SPF: pass (google.com: domain of linux-ext4-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Wed, 11 Jan 2023 11:26:01 +0100
From:   Jan Kara <jack@suse.cz>
To:     zhanchengbin <zhanchengbin1@huawei.com>
Cc:     tytso@mit.edu, jack@suse.com, linux-ext4@vger.kernel.org,
        yi.zhang@huawei.com, linfeilong@huawei.com,
        liuzhiqiang26@huawei.com
Subject: Re: [PATCH 2/2] ext4: call ext4_handle_error when read extent failed
 in ext4_ext_insert_extent
Message-ID: <20230111102601.igx6act3cxofml53@quack3>
References: <20230110133407.994711-1-zhanchengbin1@huawei.com>
 <20230110133407.994711-3-zhanchengbin1@huawei.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <20230110133407.994711-3-zhanchengbin1@huawei.com>
Precedence: bulk

On Tue 10-01-23 21:34:07, zhanchengbin wrote:
> In addition to ext4_find_extent reading extent block return
> -EIO, ext4_handle_error will be called when there is a problem that
> may cause file system inconsistency in the ext4_ext_insert_extent
> function.
> So call the ext4_handle_error function when the ext4_find_extent
> read fails，and make the filesystem read-only when mount for
> `errors=remount-ro`, and Check whether the journal is aborted in
> the ext4_split_extent_at function.
> 
> Signed-off-by: zhanchengbin <zhanchengbin1@huawei.com>
> Signed-off-by: Zhang Yi <yi.zhang@huawei.com>

Thanks for the patch! Some comments below.

> diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
> index 3559ea6b0781..3798b2a8e550 100644
> --- a/fs/ext4/extents.c
> +++ b/fs/ext4/extents.c
> @@ -935,6 +935,7 @@ ext4_find_extent(struct inode *inode, ext4_lblk_t block,
>  
>  		bh = read_extent_tree_block(inode, path[ppos].p_idx, --i, flags);
>  		if (IS_ERR(bh)) {
> +			EXT4_ERROR_INODE(inode, "IO error reading extent block");

So this is fine...

>  			ret = PTR_ERR(bh);
>  			goto err;
>  		}
> @@ -3251,7 +3252,7 @@ static int ext4_split_extent_at(handle_t *handle,
>  		ext4_ext_mark_unwritten(ex2);
>  
>  	err = ext4_ext_insert_extent(handle, inode, ppath, &newex, flags);
> -	if (err != -ENOSPC && err != -EDQUOT && err != -ENOMEM)
> +	if (err && is_handle_aborted(handle))
>  		goto out;

But this doesn't look right. Firstly, if err == 0, it will wrongly go to
the zero-out part although we should exit. Secondly, is_handle_aborted() is
not going to work for nojournal mode. So I don't think we can simplify the
condition like this.

								Honza
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR