On 2025/08/05 7:00, Viacheslav Dubeyko wrote:
>> Please show us your patch that solves your issue.
>
> OK. It will be faster to write my own patch. It works for me.
I haven't heard from you about your own patch.
I guess that your patch will include
diff --git a/fs/hfs/inode.c b/fs/hfs/inode.c
index bf4cb7e78396..8d033ffeb8af 100644
--- a/fs/hfs/inode.c
+++ b/fs/hfs/inode.c
@@ -361,6 +361,10 @@ static int hfs_read_inode(struct inode *inode, void *data)
break;
case HFS_CDR_DIR:
inode->i_ino = be32_to_cpu(rec->dir.DirID);
+ if (inode->i_ino < HFS_FIRSTUSER_CNID && inode->i_ino != HFS_ROOT_CNID) {
+ make_bad_inode(inode);
+ break;
+ }
inode->i_size = be16_to_cpu(rec->dir.Val) + 2;
HFS_I(inode)->fs_blocks = 0;
inode->i_mode = S_IFDIR | (S_IRWXUGO & ~hsb->s_dir_umask);
change, which results in the following.
----------
The root inode's i_ino is 0 or 1 = fail with EINVAL
The root inode's i_ino is 2 = success
The root inode's i_ino is 3 or 4 = fail with ENOTDIR
The root inode's i_ino is 5 to 15 = fail with EINVAL
The root inode's i_ino is 16 and beyond = success
----------
But my patch has extra validation on the root inode's i_ino,
which results in the following.
----------
The root inode's i_ino is 2 = success
The root inode's i_ino is all (i.e. including 16 and beyond) but 2 = fail with EIO
----------
Therefore, while you can propose your patch,
I consider that there is no reason to defer my patch.
