-
Bug
-
Resolution: Fixed
-
P3
-
11, 12, 13
-
b28
-
ppc
| Issue | Fix Version | Assignee | Priority | Status | Resolution | Resolved In Build |
|---|---|---|---|---|---|---|
| JDK-8217023 | 13 | Martin Doerr | P3 | Resolved | Fixed | b04 |
| JDK-8217080 | 12.0.1 | Martin Doerr | P3 | Resolved | Fixed | b03 |
| JDK-8217239 | 11.0.3 | Martin Doerr | P3 | Resolved | Fixed | master |
The template interpreter on PPC64 resizes frames which are not on top in order to avoid wasting stack space. On PPC64, SP must always point to the backlink (pointer to the caller frame). This needs to be maintained.
At some places, we restore the previous state of the stack frame by setting the stack pointer back to its previous value relying on not having overwritten the old backlink. However, this is not reliably ensured at some of these places.
In addition, relying on not having overwritten the old backlink is very error-prone so we should better avoid it.
At some places, we restore the previous state of the stack frame by setting the stack pointer back to its previous value relying on not having overwritten the old backlink. However, this is not reliably ensured at some of these places.
In addition, relying on not having overwritten the old backlink is very error-prone so we should better avoid it.
- backported by
-
JDK-8217023 [PPC64] Possibly unreliable stack frame resizing in template interpreter
-
- Resolved
-
-
-
- Resolved
-
-
-
- Resolved
-
- blocks
-
JDK-8216060 [PPC64] Vector CRC implementation should be used by interpreter and be faster for short arrays
-
- Resolved
-