Re: [PATCH] bcachefs: Fix invalid shift in validate_sb_layout()

[Gianfranco Trad <gianf.trad@gmail.com>]

On 24/10/24 13:24, Kent Overstreet wrote:
> On Thu, Oct 24, 2024 at 12:48:30PM +0200, Gianfranco Trad wrote:
>> On 23/10/24 23:30, Gianfranco Trad wrote:
>>> Add check on layout->sb_max_size_bits against BCH_SB_LAYOUT_SIZE_BITS_MAX
>>> to prevent UBSAN shift-out-of-bounds in validate_sb_layout().
>>>
>>> Reported-by: syzbot+089fad5a3a5e77825426@syzkaller.appspotmail.com
>>> Closes: https://syzkaller.appspot.com/bug?extid=089fad5a3a5e77825426
>>> Fixes: 03ef80b469d5 ("bcachefs: Ignore unknown mount options")
>>> Tested-by: syzbot+089fad5a3a5e77825426@syzkaller.appspotmail.com
>>> Signed-off-by: Gianfranco Trad <gianf.trad@gmail.com>
>>> ---
>>>    fs/bcachefs/errcode.h  | 1 +
>>>    fs/bcachefs/super-io.c | 5 +++++
>>>    2 files changed, 6 insertions(+)
>>>
>>> diff --git a/fs/bcachefs/errcode.h b/fs/bcachefs/errcode.h
>>> index 649263516ab1..b6cbd716000b 100644
>>> --- a/fs/bcachefs/errcode.h
>>> +++ b/fs/bcachefs/errcode.h
>>> @@ -222,6 +222,7 @@
>>>    	x(BCH_ERR_invalid_sb_layout,	invalid_sb_layout_type)			\
>>>    	x(BCH_ERR_invalid_sb_layout,	invalid_sb_layout_nr_superblocks)	\
>>>    	x(BCH_ERR_invalid_sb_layout,	invalid_sb_layout_superblocks_overlap)	\
>>> +	x(BCH_ERR_invalid_sb_layout,    invalid_sb_layout_sb_max_size_bits)     \
>>>    	x(BCH_ERR_invalid_sb,		invalid_sb_members_missing)		\
>>>    	x(BCH_ERR_invalid_sb,		invalid_sb_members)			\
>>>    	x(BCH_ERR_invalid_sb,		invalid_sb_disk_groups)			\
>>> diff --git a/fs/bcachefs/super-io.c b/fs/bcachefs/super-io.c
>>> index ce7410d72089..44d0ac9b00dd 100644
>>> --- a/fs/bcachefs/super-io.c
>>> +++ b/fs/bcachefs/super-io.c
>>> @@ -287,6 +287,11 @@ static int validate_sb_layout(struct bch_sb_layout *layout, struct printbuf *out
>>>    		return -BCH_ERR_invalid_sb_layout_nr_superblocks;
>>>    	}
>>> +	if (layout->sb_max_size_bits > BCH_SB_LAYOUT_SIZE_BITS_MAX) {
>>> +		prt_printf(out, "Invalid superblock layout: max_size_bits too high");
>>> +		return -BCH_ERR_invalid_sb_layout_sb_max_size_bits;
>>> +	}
>>> +
>>>    	max_sectors = 1 << layout->sb_max_size_bits;
>>>    	prev_offset = le64_to_cpu(layout->sb_offset[0]);
>>
>> Wondering if this other patch might be considered more correct to prevent
>> shift oob, given also [0]:
> 
> Your first patch looks better, we want to know if we're feeding it
> garbage
> 

Understood. Thanks Kent.
Have a nice day,
--Gian
>>
>> diff --git a/fs/bcachefs/super-io.c b/fs/bcachefs/super-io.c
>> index ce7410d72089..428172897501 100644
>> --- a/fs/bcachefs/super-io.c
>> +++ b/fs/bcachefs/super-io.c
>> @@ -287,7 +287,7 @@ static int validate_sb_layout(struct bch_sb_layout
>> *layout, struct printbuf *out
>>   		return -BCH_ERR_invalid_sb_layout_nr_superblocks;
>>   	}
>>
>> -	max_sectors = 1 << layout->sb_max_size_bits;
>> +	max_sectors = 1 << min(BCH_SB_LAYOUT_SIZE_BITS_MAX,
>> layout->sb_max_size_bits);
>>
>>   	prev_offset = le64_to_cpu(layout->sb_offset[0]);
>>
>> Also this patch was already tested by syzbot [1]
>>
>> [0] 71dac2482ad3c8d4a8b8998a96751f009bad895f ("bcachefs:
>> BCH_SB_LAYOUT_SIZE_BITS_MAX")
>> [1] https://syzkaller.appspot.com/x/log.txt?x=1640b640580000
>>
>> Thanks for your time,
>>
>> -- Gian
>>