-
Type:
Bug
-
Resolution: Fixed
-
Priority:
P3
-
Affects Version/s: 8, 11, 15, 16
-
Component/s: hotspot
-
b21
| Issue | Fix Version | Assignee | Priority | Status | Resolution | Resolved In Build |
|---|---|---|---|---|---|---|
| JDK-8256231 | 15u-cpu | Aleksey Shipilev | P3 | Resolved | Fixed | master |
| JDK-8255537 | 15.0.2 | Aleksey Shipilev | P3 | Resolved | Fixed | b05 |
| JDK-8256035 | 11.0.10 | Aleksey Shipilev | P3 | Resolved | Fixed | b02 |
While doing a change in related area, I noticed there is no IRIW handling block in accessor_entry when reading volatile fields. This probably breaks PPC64 Zero.
There is a block in bytecodeInterpreter.cpp for common code:
if (cache->is_volatile()) {
if (support_IRIW_for_not_multiple_copy_atomic_cpu) {
OrderAccess::fence();
}
AFAIU, this mostly affects 8u that does not have PPC64 port yet.
There is a block in bytecodeInterpreter.cpp for common code:
if (cache->is_volatile()) {
if (support_IRIW_for_not_multiple_copy_atomic_cpu) {
OrderAccess::fence();
}
AFAIU, this mostly affects 8u that does not have PPC64 port yet.
- backported by
-
JDK-8255537 Zero: accessor_entry misses the IRIW case
-
- Resolved
-
-
-
- Resolved
-
-
-
- Resolved
-
