linux-kernel - Re: [RFC PATCH 7/7] exfat: get mutil-clusters in exfat_get

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <9c47afea-a211-4848-bde7-b29f27466e43@samsung.com>
Date: Thu, 4 Dec 2025 21:18:32 +0900
From: Sungjong Seo <sj1557.seo@...sung.com>
To: Chi Zhiling <chizhiling@....com>, linux-fsdevel@...r.kernel.org,
	linux-kernel@...r.kernel.org
Cc: Alexander Viro <viro@...iv.linux.org.uk>, Christian Brauner
	<brauner@...nel.org>, Jan Kara <jack@...e.cz>, Matthew Wilcox
	<willy@...radead.org>, Namjae Jeon <linkinjeon@...nel.org>, Yuezhang Mo
	<yuezhang.mo@...y.com>, Chi Zhiling <chizhiling@...inos.cn>
Subject: Re: [RFC PATCH 7/7] exfat: get mutil-clusters in exfat_get_block



On 25. 11. 28. 15:18, Chi Zhiling wrote:
> On 11/28/25 10:48, Sungjong Seo wrote:
>>
>> Hi, Chi,
>> On 25. 11. 18. 17:22, Chi Zhiling wrote:
>>> From: Chi Zhiling <chizhiling@...inos.cn>
>>>
>>> mpage uses the get_block of the file system to obtain the mapping of a
>>> file or allocate blocks for writes. Currently exfat only supports
>>> obtaining one cluster in each get_block call.
>>>
>>> Since exfat_count_contig_clusters can obtain multiple consecutive clusters,
>>> it can be used to improve exfat_get_block when page size is larger than
>>> cluster size.
>>
>> I think reusing buffer_head is a good approach!
>> However, for obtaining multiple clusters, it would be better to handle
>> them in exfat_map_cluster.
> 
> Hi, Sungjong
> 
> I agree.
> 
> My original plan was to support multiple clusters for exfat_map_cluster and exfat_get_cluster. since the changes required were quite extensive, I put that plan on hold. This would likely involve refactoring exfat_map_clusterand introducing iterators to reduce the number of parameters it needs
> 
> I will take some time to consider the signature of the new exfat_map_clusters. Do you have any thoughts about this?
Apologies, I missed your email.
IMO, we don't need to rush, so I think expanding exfat_map_cluster(s) would be better.

Thanks.
> 
>>
>>>
>>> Signed-off-by: Chi Zhiling <chizhiling@...inos.cn>
>>> ---
>>>   fs/exfat/inode.c | 14 +++++++++++++-
>>>   1 file changed, 13 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/fs/exfat/inode.c b/fs/exfat/inode.c
>>> index f9501c3a3666..256ba2af34eb 100644
>>> --- a/fs/exfat/inode.c
>>> +++ b/fs/exfat/inode.c
>>> @@ -264,13 +264,14 @@ static int exfat_map_cluster(struct inode *inode, unsigned int clu_offset,
>>>   static int exfat_get_block(struct inode *inode, sector_t iblock,
>>>           struct buffer_head *bh_result, int create)
>>>   {
>>> +    struct exfat_chain chain;
>>>       struct exfat_inode_info *ei = EXFAT_I(inode);
>>>       struct super_block *sb = inode->i_sb;
>>>       struct exfat_sb_info *sbi = EXFAT_SB(sb);
>>>       unsigned long max_blocks = bh_result->b_size >> inode->i_blkbits;
>>>       int err = 0;
>>>       unsigned long mapped_blocks = 0;
>>> -    unsigned int cluster, sec_offset;
>>> +    unsigned int cluster, sec_offset, count;
>>>       sector_t last_block;
>>>       sector_t phys = 0;
>>>       sector_t valid_blks;
>>> @@ -301,6 +302,17 @@ static int exfat_get_block(struct inode *inode, sector_t iblock,
>>>         phys = exfat_cluster_to_sector(sbi, cluster) + sec_offset;
>>>       mapped_blocks = sbi->sect_per_clus - sec_offset;
>>> +
>>> +    if (max_blocks > mapped_blocks && !create) {
>>> +        chain.dir = cluster;
>>> +        chain.size = (max_blocks >> sbi->sect_per_clus_bits) + 1;
>>
>> There seems to be an issue where the code sets chain.size to be one greater than the actual cluster count.
>>
>> For example, assuming a 16KiB page, 512B sector, and 4KiB cluster,
>> for a 16KiB file, chain.size becomes 5 instead of 4.
>> Is this the intended behavior?
> 
> This is not the expected behavior. It's a serious bug. Thank you very much for pointing this out.
> 
>>
>>> +        chain.flags = ei->flags;
>>> +
>>> +        err = exfat_count_contig_clusters(sb, &chain, &count);
>>> +        if (err)
>>> +            return err;
>>> +        max_blocks = (count << sbi->sect_per_clus_bits) - sec_offset;
>>
>> You already said mapped_blocks is correct.
>>
>>> +    }
>>>       max_blocks = min(mapped_blocks, max_blocks);
>>>         map_bh(bh_result, sb, phys);
> 
>