Subject: Re: [PATCH 7/7] ext4: improve inode table blocks counting in
 ext4_num_overhead_clusters
To:     Dan Carpenter <error27@gmail.com>, oe-kbuild@lists.linux.dev,
        tytso@mit.edu, adilger.kernel@dilger.ca
Cc:     lkp@intel.com, oe-kbuild-all@lists.linux.dev,
        linux-ext4@vger.kernel.org, linux-kernel@vger.kernel.org
References: <202302222219.u328sqfs-lkp@intel.com>
From:   Kemeng Shi <shikemeng@huaweicloud.com>
Message-ID: <1802222f-9c10-9ea3-93fb-6fce086e201a@huaweicloud.com>
Date:   Thu, 23 Feb 2023 09:31:54 +0800
User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:60.0) Gecko/20100101
 Thunderbird/60.5.0
MIME-Version: 1.0
In-Reply-To: <202302222219.u328sqfs-lkp@intel.com>
Content-Type: text/plain; charset=gbk
Content-Transfer-Encoding: 7bit
Precedence: bulk


on 2/22/2023 11:13 PM, Dan Carpenter wrote:
> Hi Kemeng,
> 
> https://git-scm.com/docs/git-format-patch#_base_tree_information]
> 
> url:    https://github.com/intel-lab-lkp/linux/commits/Kemeng-Shi/ext4-properly-handle-error-of-ext4_init_block_bitmap-in-ext4_read_block_bitmap_nowait/20230221-115830
> base:   https://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4.git dev
> patch link:    https://lore.kernel.org/r/20230221115919.1918161-8-shikemeng%40huaweicloud.com
> patch subject: [PATCH 7/7] ext4: improve inode table blocks counting in ext4_num_overhead_clusters
> config: riscv-randconfig-m031-20230219 (https://download.01.org/0day-ci/archive/20230222/202302222219.u328sqfs-lkp@intel.com/config)
> compiler: riscv32-linux-gcc (GCC) 12.1.0
> 
> If you fix the issue, kindly add following tag where applicable
> | Reported-by: kernel test robot <lkp@intel.com>
> | Reported-by: Dan Carpenter <error27@gmail.com>
> | Link: https://lore.kernel.org/r/202302222219.u328sqfs-lkp@intel.com/
> 
> New smatch warnings:
> fs/ext4/balloc.c:153 ext4_num_overhead_clusters() error: uninitialized symbol 'block_cluster'.
> 
> vim +/block_cluster +153 fs/ext4/balloc.c
[...]
> d5b8f31007a937 Theodore Ts'o     2011-09-09  128  	/*
> 2b59a2fd93873a Kemeng Shi        2023-02-21  129  	 * For the allocation bitmaps, we first need to check to see
> 2b59a2fd93873a Kemeng Shi        2023-02-21  130  	 * if the block is in the block group.  If it is, then check
> 2b59a2fd93873a Kemeng Shi        2023-02-21  131  	 * to see if the cluster is already accounted for in the clusters
> 2b59a2fd93873a Kemeng Shi        2023-02-21  132  	 * used for the base metadata cluster and inode tables cluster.
> d5b8f31007a937 Theodore Ts'o     2011-09-09  133  	 * Normally all of these blocks are contiguous, so the special
> d5b8f31007a937 Theodore Ts'o     2011-09-09  134  	 * case handling shouldn't be necessary except for *very*
> d5b8f31007a937 Theodore Ts'o     2011-09-09  135  	 * unusual file system layouts.
> d5b8f31007a937 Theodore Ts'o     2011-09-09  136  	 */
> d5b8f31007a937 Theodore Ts'o     2011-09-09  137  	if (ext4_block_in_group(sb, ext4_block_bitmap(sb, gdp), block_group)) {
> b0dd6b70f0fda1 Theodore Ts'o     2012-06-07  138  		block_cluster = EXT4_B2C(sbi,
> b0dd6b70f0fda1 Theodore Ts'o     2012-06-07  139  					 ext4_block_bitmap(sb, gdp) - start);
> 2b59a2fd93873a Kemeng Shi        2023-02-21  140  		if (block_cluster >= base_clusters &&
> 2b59a2fd93873a Kemeng Shi        2023-02-21  141  		    (block_cluster < itbl_cluster_start ||
> 2b59a2fd93873a Kemeng Shi        2023-02-21  142  		    block_cluster > itbl_cluster_end))
> d5b8f31007a937 Theodore Ts'o     2011-09-09  143  			num_clusters++;
> d5b8f31007a937 Theodore Ts'o     2011-09-09  144  	}
> d5b8f31007a937 Theodore Ts'o     2011-09-09  145  
> d5b8f31007a937 Theodore Ts'o     2011-09-09  146  	if (ext4_block_in_group(sb, ext4_inode_bitmap(sb, gdp), block_group)) {
> 
> These two conditions are exactly the same so I don't see why they can't
> be combined into one condition.  I have read the comment, but I guess I
> don't understand ext4 well enough to really understand it.
These two conditions check two kinds of bitmap block: *block* bitmap block
and *inode* bitmap block. For case that block bitmap in the block group
while inode bitmap in a different group, there is a risk to access
uninitialized  block_cluster.
I will fix this in next version, Thanks!

-- 
Best wishes
Kemeng Shi