-
Bug
-
Resolution: Fixed
-
P2
-
11, 11.0.2, 12, 13
Issue | Fix Version | Assignee | Priority | Status | Resolution | Resolved In Build |
---|---|---|---|---|---|---|
JDK-8216344 | 13 | Sandhya Viswanathan | P2 | Resolved | Fixed | b03 |
JDK-8216296 | 12.0.1 | Sandhya Viswanathan | P2 | Resolved | Fixed | b02 |
JDK-8226559 | 11.0.5-oracle | Sandhya Viswanathan | P2 | Resolved | Fixed | b02 |
JDK-8216290 | 11.0.3 | Sandhya Viswanathan | P2 | Resolved | Fixed | master |
We noticed that the register to register moves in x86.ad file attempts to generate emovdqu when UseAVX==2.
The instruction emovdquq is only supported on platforms where UseAVX > 2 (AVX 512).
The following rules in x86.ad file need to be corrected:
MoveVecX2Leg
MoveLeg2VecX
MoveVecY2Leg
MoveLeg2VecY
The above move rules when activated through register allocator and could result in illegal instruction exception.
Also there is similar usage of (UseAVX < 2) in c1_LIRAssembler which could be written as (UseAVX <=2) for better code sequence.
The instruction emovdquq is only supported on platforms where UseAVX > 2 (AVX 512).
The following rules in x86.ad file need to be corrected:
MoveVecX2Leg
MoveLeg2VecX
MoveVecY2Leg
MoveLeg2VecY
The above move rules when activated through register allocator and could result in illegal instruction exception.
Also there is similar usage of (UseAVX < 2) in c1_LIRAssembler which could be written as (UseAVX <=2) for better code sequence.
- backported by
-
JDK-8216290 Register to register spill may use AVX 512 move instruction on unsupported platform.
-
- Resolved
-
-
JDK-8216296 Register to register spill may use AVX 512 move instruction on unsupported platform.
-
- Resolved
-
-
JDK-8216344 Register to register spill may use AVX 512 move instruction on unsupported platform.
-
- Resolved
-
-
JDK-8226559 Register to register spill may use AVX 512 move instruction on unsupported platform.
-
- Resolved
-
- relates to
-
JDK-8210764 Update avx512 implementation
-
- Resolved
-