Re: [PATCH v2] hpfs: obsolete check=none mount option

[Tetsuo Handa]

On 2026/06/01 19:10, Mikulas Patocka wrote:
> Hi
> 
> Linus didn't want the patch that disables the check=none mode, so I don't 
> know what else to do.
> 

Hi Mikulas,

Thank you for your reply.

I think there might be a slight misunderstanding regarding Linus's feedback.
My understanding is that what Linus did not like is that your approach

-	{"none",	0},
+	{"none",	1},

will not break existing userspace programs regarding parsing of mount options.
But your approach will break existing userspace programs regarding reporting of
mount options (the

	if (!sbi->sb_chk)
		seq_printf(seq, ",check=none");

logic in hpfs_show_options()).

My proposed patch aligns perfectly with both of Linus's requirements:

1. It preserves the "check=none" mount option (for both parsing and reporting),
   meaning we DO NOT break any existing user space configurations or scripts.
2. It always enforces the absolute minimum buffer boundary and pointer validation
   inside the code (such as in hpfs_get_ea and hpfs_bplus_lookup). This change
   should fundamentally prevent the critical KASAN UAF crashes reported by syzbot,
   regardless of what mount option is specified.

What do you think?