Re: [PATCH] btrfs-progs: print-tree: do sanity checks for dir items

[Qu Wenruo <quwenruo.btrfs@gmx.com>]

在 2024/6/5 01:28, Josef Bacik 写道:
> On Tue, Jun 04, 2024 at 02:19:08PM +0930, Qu Wenruo wrote:
>> There is a bug report that with UBSAN enabled, fuzz/006 test case would
>> crash.
>>
>> It turns out that the image bko-154021-invalid-drop-level.raw has
>> invalid dir items, that the name/data len is beyond the item.
>>
>> And if we try to read beyond the eb boundary, UBSAN got triggered.
>>
>> Normally in kernel tree-checker would reject such metadata in the first
>> place, but in btrfs-progs we can not go that strict or we can not do a
>> lot of repair.
>>
>> So here just enhance print_dir_item() to do extra sanity checks for
>> data/name len before reading the contents.
>>
>> Issue: #805
>> Signed-off-by: Qu Wenruo <wqu@suse.com>
>
> I'd rather not duplicate this check.
>
> Is the print-tree coming from repair?

Nope, it's "inspect tree-stats" subcommand, it's just really printing
the root of each tree.

Remember we still need to call "btrfs ins dump-tree" to analyze fuzzed
images, if the print-tree is as strict as tree-checker, just image how
miserable the life would be.

Thanks,
Qu

>  If that's the case then I'd manually call
> check_leaf to make sure the pointers are all correct before calling print tree,
> otherwise if it's from a different tool we need to make sure the strict checking
> is happening for that tool, we should only be bypassing the strict checking for
> repair.  Thanks,
>
> Josef
>