RFR: 8356941: AbstractMethodError in HotSpot Due to Incorrect Handling of Private Method [v3]
Dan Heidinga
heidinga at openjdk.org
Fri Jul 18 14:17:51 UTC 2025
On Wed, 16 Jul 2025 07:20:16 GMT, David Holmes <dholmes at openjdk.org> wrote:
>> Private methods should never be considered as the implementation of a default method.
>>
>> The first commit adds some additional logging I used to track down what was happening. I like it, but if reviewers think it too much I can drop it.
>>
>> The second commit is the actual fix to exclude private methods, and adds the missing test case to the existing defmeth tests.
>>
>> Testing:
>> - tiers 1-4
>>
>> Thanks
>
> David Holmes has updated the pull request incrementally with one additional commit since the last revision:
>
> Revert "Additional logging to identify the problem"
>
> This reverts commit 60871844d32cbc58ea97b2186a2758e85613afa4.
Current change is also fine. Both suggestions are more for leaving breadcrumbs to help the next person to dig thru this code
src/hotspot/share/classfile/defaultMethods.cpp line 667:
> 665: if (!already_in_vtable_slots(slots, m)) {
> 666: Method* impl = klass->lookup_method(m->name(), m->signature());
> 667: if (impl == nullptr || impl->is_overpass() || impl->is_static() || impl->is_private()) {
Trying to capture your and Coleen's discussion in a comment. Does this correctly cover the gist of the discussion?
Suggestion:
// Unless the klass provides a non-static public or package-private method for this name
// & sig combo, we need to add the EmptyVtableSlot so default method processing finds
// the correct candidate to fill in the slot later.
if (impl == nullptr || impl->is_overpass() || impl->is_static() || impl->is_private()) {
-------------
Marked as reviewed by heidinga (no project role).
PR Review: https://git.openjdk.org/jdk/pull/26302#pullrequestreview-3033634330
PR Review Comment: https://git.openjdk.org/jdk/pull/26302#discussion_r2216102680
More information about the hotspot-runtime-dev
mailing list