[PATCH v2] erofs: fix file-backed mounts over FUSE

Gao Xiang posted 1 patch 1 week, 1 day ago
There is a newer version of this series
fs/erofs/data.c     | 8 +++++---
fs/erofs/internal.h | 1 +
2 files changed, 6 insertions(+), 3 deletions(-)
[PATCH v2] erofs: fix file-backed mounts over FUSE
Posted by Gao Xiang 1 week, 1 day ago
syzbot reported a null-ptr-deref in fuse_read_args_fill:
 fuse_read_folio+0xb0/0x100 fs/fuse/file.c:905
 filemap_read_folio+0xc6/0x2a0 mm/filemap.c:2367
 do_read_cache_folio+0x263/0x5c0 mm/filemap.c:3825
 read_mapping_folio include/linux/pagemap.h:1011 [inline]
 erofs_bread+0x34d/0x7e0 fs/erofs/data.c:41
 erofs_read_superblock fs/erofs/super.c:281 [inline]
 erofs_fc_fill_super+0x2b9/0x2500 fs/erofs/super.c:625

Unlike most filesystems, some network filesystems and FUSE need
unavoidable valid `file` pointers for their read I/Os [1].
Anyway, those use cases need to be supported too.

[1] https://docs.kernel.org/filesystems/vfs.html
Reported-by: syzbot+0b1279812c46e48bb0c1@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/r/6727bbdf.050a0220.3c8d68.0a7e.GAE@google.com
Fixes: fb176750266a ("erofs: add file-backed mount support")
Tested-by: syzbot+0b1279812c46e48bb0c1@syzkaller.appspotmail.com
Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>
---
v1: https://lore.kernel.org/r/20241114051957.419551-1-hsiangkao@linux.alibaba.com
changes since v1 (suggested by Al):
 - rename `filp` to `file`;
 - don't use union for `mapping` and `file`.

 fs/erofs/data.c     | 8 +++++---
 fs/erofs/internal.h | 1 +
 2 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/fs/erofs/data.c b/fs/erofs/data.c
index 6355866220ff..f741c3847ff2 100644
--- a/fs/erofs/data.c
+++ b/fs/erofs/data.c
@@ -38,7 +38,9 @@ void *erofs_bread(struct erofs_buf *buf, erofs_off_t offset,
 	}
 	if (!folio || !folio_contains(folio, index)) {
 		erofs_put_metabuf(buf);
-		folio = read_mapping_folio(buf->mapping, index, NULL);
+		folio = buf->file ? read_mapping_folio(buf->file->f_mapping,
+					index, buf->file) :
+			read_mapping_folio(buf->mapping, index, NULL);
 		if (IS_ERR(folio))
 			return folio;
 	}
@@ -61,8 +63,8 @@ void erofs_init_metabuf(struct erofs_buf *buf, struct super_block *sb)
 {
 	struct erofs_sb_info *sbi = EROFS_SB(sb);
 
-	if (erofs_is_fileio_mode(sbi))
-		buf->mapping = file_inode(sbi->fdev)->i_mapping;
+	if (erofs_is_fileio_mode(sbi))	/* some fs like FUSE needs it */
+		buf->file = sbi->fdev;
 	else if (erofs_is_fscache_mode(sb))
 		buf->mapping = sbi->s_fscache->inode->i_mapping;
 	else
diff --git a/fs/erofs/internal.h b/fs/erofs/internal.h
index 5459d0b39415..9844ee8a07e5 100644
--- a/fs/erofs/internal.h
+++ b/fs/erofs/internal.h
@@ -209,6 +209,7 @@ enum erofs_kmap_type {
 
 struct erofs_buf {
 	struct address_space *mapping;
+	struct file *file;
 	struct page *page;
 	void *base;
 	enum erofs_kmap_type kmap_type;
-- 
2.43.5
Re: [PATCH v2] erofs: fix file-backed mounts over FUSE
Posted by Al Viro 1 week, 1 day ago
On Thu, Nov 14, 2024 at 05:01:09PM +0800, Gao Xiang wrote:
> --- a/fs/erofs/data.c
> +++ b/fs/erofs/data.c
> @@ -38,7 +38,9 @@ void *erofs_bread(struct erofs_buf *buf, erofs_off_t offset,
>  	}
>  	if (!folio || !folio_contains(folio, index)) {
>  		erofs_put_metabuf(buf);
> -		folio = read_mapping_folio(buf->mapping, index, NULL);
> +		folio = buf->file ? read_mapping_folio(buf->file->f_mapping,
> +					index, buf->file) :
> +			read_mapping_folio(buf->mapping, index, NULL);
>  		if (IS_ERR(folio))
>  			return folio;
>  	}
> @@ -61,8 +63,8 @@ void erofs_init_metabuf(struct erofs_buf *buf, struct super_block *sb)
>  {
>  	struct erofs_sb_info *sbi = EROFS_SB(sb);
>  
> -	if (erofs_is_fileio_mode(sbi))
> -		buf->mapping = file_inode(sbi->fdev)->i_mapping;
> +	if (erofs_is_fileio_mode(sbi))	/* some fs like FUSE needs it */
> +		buf->file = sbi->fdev;

Would be easier to set *both* ->mapping and ->file at that point, so that
erofs_bread() would just have read_mapping_folio(buf->mapping, index, buf->file),
unconditionally...
Re: [PATCH v2] erofs: fix file-backed mounts over FUSE
Posted by Gao Xiang 1 week, 1 day ago

On 2024/11/15 00:18, Al Viro wrote:
> On Thu, Nov 14, 2024 at 05:01:09PM +0800, Gao Xiang wrote:
>> --- a/fs/erofs/data.c
>> +++ b/fs/erofs/data.c
>> @@ -38,7 +38,9 @@ void *erofs_bread(struct erofs_buf *buf, erofs_off_t offset,
>>   	}
>>   	if (!folio || !folio_contains(folio, index)) {
>>   		erofs_put_metabuf(buf);
>> -		folio = read_mapping_folio(buf->mapping, index, NULL);
>> +		folio = buf->file ? read_mapping_folio(buf->file->f_mapping,
>> +					index, buf->file) :
>> +			read_mapping_folio(buf->mapping, index, NULL);
>>   		if (IS_ERR(folio))
>>   			return folio;
>>   	}
>> @@ -61,8 +63,8 @@ void erofs_init_metabuf(struct erofs_buf *buf, struct super_block *sb)
>>   {
>>   	struct erofs_sb_info *sbi = EROFS_SB(sb);
>>   
>> -	if (erofs_is_fileio_mode(sbi))
>> -		buf->mapping = file_inode(sbi->fdev)->i_mapping;
>> +	if (erofs_is_fileio_mode(sbi))	/* some fs like FUSE needs it */
>> +		buf->file = sbi->fdev;
> 
> Would be easier to set *both* ->mapping and ->file at that point, so that
> erofs_bread() would just have read_mapping_folio(buf->mapping, index, buf->file),
> unconditionally...

Ok, will do the next version.

Thanks,
Gao Xiang