Re: [PATCH] idr: fix invalid ptr dereference on item delete

[Dmitry Vyukov <dvyukov@google.com>]

On Fri, May 11, 2018 at 1:54 AM, Paolo Bonzini <pbonzini@redhat.com> wrote:
> On 10/05/2018 21:16, Roman Kagan wrote:
>> If an IDR contains a single entry at index==0, the underlying radix tree
>> has a single item in its root node, in which case
>> __radix_tree_lookup(index!=0) doesn't set its *@nodep argument (in
>> addition to returning NULL).
>>
>> However, the tree itself is not empty, i.e. the tree root doesn't have
>> IDR_FREE tag.
>>
>> As a result, on an attempt to remove an index!=0 entry from such an IDR,
>> radix_tree_delete_item doesn't return early and calls
>> __radix_tree_delete with invalid parameters which are then dereferenced.
>>
>> Reported-by: syzbot+35666cba7f0a337e2e79@syzkaller.appspotmail.com
>> Signed-off-by: Roman Kagan <rkagan@virtuozzo.com>
>> ---
>>  lib/radix-tree.c | 5 +++--
>>  1 file changed, 3 insertions(+), 2 deletions(-)
>>
>> diff --git a/lib/radix-tree.c b/lib/radix-tree.c
>> index da9e10c827df..10ff1bfae952 100644
>> --- a/lib/radix-tree.c
>> +++ b/lib/radix-tree.c
>> @@ -2040,8 +2040,9 @@ void *radix_tree_delete_item(struct radix_tree_root *root,
>>       void *entry;
>>
>>       entry = __radix_tree_lookup(root, index, &node, &slot);
>> -     if (!entry && (!is_idr(root) || node_tag_get(root, node, IDR_FREE,
>> -                                             get_slot_offset(node, slot))))
>> +     if (!entry && (!is_idr(root) || !node ||
>> +                    node_tag_get(root, node, IDR_FREE,
>> +                                 get_slot_offset(node, slot))))
>>               return NULL;
>>
>>       if (item && entry != item)
>>
>
> I cannot really vouch for the patch, but if it is correct it's
> definitely stuff for stable.  The KVM testcase is only for 4.17-rc but
> this is a really nasty bug in a core data structure.
>
> Cc: stable@vger.kernel.org
>
> Should radix-tree be compilable in userspace, so that we can add unit
> tests for it?...

Good point.

For my education, what/where are the tests that run as user-space code?