From: lkp-developer-request(a)eclists.intel.com
[mailto:lkp-developer-request@eclists.intel.com] On Behalf Of kernel test robot
FYI, we noticed the following commit:
commit: c55a9697e10aba9946692c56e9236993c1da023a ("Reimplement IDR
and IDA using the radix tree")
git://git.infradead.org/users/willy/linux-dax.git idr-2016-12-26
caused below changes:
+-------------------------------------------+------------+------------+
| | ef9e389452 | c55a9697e1 |
+-------------------------------------------+------------+------------+
| BUG:KASAN:null-ptr-deref_on_address(null) | 0 | 20 |
| BUG:unable_to_handle_kernel | 0 | 20 |
| Oops | 0 | 20 |
| Kernel_panic-not_syncing:Fatal_exception | 0 | 20 |
+-------------------------------------------+------------+------------+
Thank you! Bug diagnosed, test-case written, test case verified to fail for the c55a9
commit, bug squashed, test case verified to pass, new branch pushed out to
idr-2016-01-03.
(in case anybody cares, the bug was that removing a NULL entry from index 0 of an IDR with
one entry in it caused a NULL pointer dereference when we checked node->shift without
first checking that node wasn't NULL. This was a half-hearted attempt at an
optimization, so I deleted that clause of the if statement. That exposed an unrelated bug
in get_node_tag() which had the sense of the condition wrong).