-
Bug
-
Resolution: Fixed
-
P4
-
22
-
b26
| Issue | Fix Version | Assignee | Priority | Status | Resolution | Resolved In Build |
|---|---|---|---|---|---|---|
| JDK-8342500 | 21.0.6-oracle | Thomas Schatzl | P4 | Resolved | Fixed | b03 |
| JDK-8330922 | 21.0.4 | Goetz Lindenmaier | P4 | Resolved | Fixed | b01 |
For debug purposes G1UpdateRemSetTrackingBeforeRebuild::distribute_marked_bytes() accesses the size of the humongous object it is currently processing.
In case of marked_bytes == 0, the object is actually dead, and the code should not access the klass any more.
This currently does not matter because we purge the metaspace for unloaded classes after this code, but still this is ugly.
Makes code that moves the purging before G1UpdateRemSetTrackingBeforeRebuild crash on some platforms (linux-x86).
In case of marked_bytes == 0, the object is actually dead, and the code should not access the klass any more.
This currently does not matter because we purge the metaspace for unloaded classes after this code, but still this is ugly.
Makes code that moves the purging before G1UpdateRemSetTrackingBeforeRebuild crash on some platforms (linux-x86).
- backported by
-
JDK-8330922 G1: G1UpdateRemSetTrackingBeforeRebuild::distribute_marked_bytes accesses partially unloaded klass
-
- Resolved
-
-
-
- Resolved
-
- blocks
-
JDK-8317809 Insertion of free code blobs into code cache can be very slow during class unloading
-
- Resolved
-
- links to
-
Commit
openjdk/jdk21u-dev/bb046b70
-
Commit
openjdk/jdk/21d361e9
-
Review
openjdk/jdk21u-dev/517
-
Review
openjdk/jdk/16766
(2 links to)