- 
    Bug 
- 
    Resolution: Unresolved
- 
     P4 P4
- 
    15
- 
        x86
                    Explicitly disabling AVX512VL leads to the following assertion failures:
# Internal Error (open/src/hotspot/cpu/x86/macroAssembler_x86.cpp:3306), pid=25774, tid=25788
# assert(((dst->encoding() < 16) || VM_Version::supports_avx512vl())) failed: XMM register should be 0-15
V [libjvm.so+0x107695c] MacroAssembler::vmovdqu(XMMRegisterImpl*, AddressLiteral, RegisterImpl*)+0x20c
V [libjvm.so+0x36fd7a] vshift64B_avxNode::emit(CodeBuffer&, PhaseRegAlloc*) const+0x41a
# Internal Error (open/src/hotspot/cpu/x86/assembler_x86.cpp:7910), pid=26235, tid=26249
# assert(((!attributes->uses_vl()) || (attributes->get_vector_len() == AVX_512bit) || (!_legacy_mode_vl) || (attributes->is_legacy_mode()))) failed: XMM register should be 0-15
V [libjvm.so+0x4fd964] Assembler::vex_prefix(Address, int, int, Assembler::VexSimdPrefix, Assembler::VexOpcode, InstructionAttr*)+0x214
V [libjvm.so+0x4ff184] Assembler::vpmullq(XMMRegisterImpl*, XMMRegisterImpl*, Address, int)+0x104
V [libjvm.so+0x365536] vmulL_mem_0Node::emit(CodeBuffer&, PhaseRegAlloc*) const+0x1d6
# Internal Error (open/src/hotspot/cpu/x86/macroAssembler_x86.cpp:3894), pid=26287, tid=26301
# assert(((dst->encoding() < 16 && shift->encoding() < 16 && nds->encoding() < 16) || VM_Version::supports_avx512vlbw())) failed: XMM register should be 0-15
V [libjvm.so+0x107b38a] MacroAssembler::vpsllw(XMMRegisterImpl*, XMMRegisterImpl*, XMMRegisterImpl*, int)+0x7a
V [libjvm.so+0x372c2a] vshiftSNode::emit(CodeBuffer&, PhaseRegAlloc*) const+0x14a
# Internal Error (open/src/hotspot/cpu/x86/macroAssembler_x86.cpp:3803), pid=26980, tid=26994
# assert(((dst->encoding() < 16 && src->encoding() < 16) || VM_Version::supports_avx512vlbw())) failed: XMM register should be 0-15
V [libjvm.so+0x1079d0d] MacroAssembler::vpbroadcastw(XMMRegisterImpl*, XMMRegisterImpl*, int)+0x6d
V [libjvm.so+0x342213] ReplS_immNode::emit(CodeBuffer&, PhaseRegAlloc*) const+0x373
As of now, the configuration is artificial since there's no such hardware exist.
# Internal Error (open/src/hotspot/cpu/x86/macroAssembler_x86.cpp:3306), pid=25774, tid=25788
# assert(((dst->encoding() < 16) || VM_Version::supports_avx512vl())) failed: XMM register should be 0-15
V [libjvm.so+0x107695c] MacroAssembler::vmovdqu(XMMRegisterImpl*, AddressLiteral, RegisterImpl*)+0x20c
V [libjvm.so+0x36fd7a] vshift64B_avxNode::emit(CodeBuffer&, PhaseRegAlloc*) const+0x41a
# Internal Error (open/src/hotspot/cpu/x86/assembler_x86.cpp:7910), pid=26235, tid=26249
# assert(((!attributes->uses_vl()) || (attributes->get_vector_len() == AVX_512bit) || (!_legacy_mode_vl) || (attributes->is_legacy_mode()))) failed: XMM register should be 0-15
V [libjvm.so+0x4fd964] Assembler::vex_prefix(Address, int, int, Assembler::VexSimdPrefix, Assembler::VexOpcode, InstructionAttr*)+0x214
V [libjvm.so+0x4ff184] Assembler::vpmullq(XMMRegisterImpl*, XMMRegisterImpl*, Address, int)+0x104
V [libjvm.so+0x365536] vmulL_mem_0Node::emit(CodeBuffer&, PhaseRegAlloc*) const+0x1d6
# Internal Error (open/src/hotspot/cpu/x86/macroAssembler_x86.cpp:3894), pid=26287, tid=26301
# assert(((dst->encoding() < 16 && shift->encoding() < 16 && nds->encoding() < 16) || VM_Version::supports_avx512vlbw())) failed: XMM register should be 0-15
V [libjvm.so+0x107b38a] MacroAssembler::vpsllw(XMMRegisterImpl*, XMMRegisterImpl*, XMMRegisterImpl*, int)+0x7a
V [libjvm.so+0x372c2a] vshiftSNode::emit(CodeBuffer&, PhaseRegAlloc*) const+0x14a
# Internal Error (open/src/hotspot/cpu/x86/macroAssembler_x86.cpp:3803), pid=26980, tid=26994
# assert(((dst->encoding() < 16 && src->encoding() < 16) || VM_Version::supports_avx512vlbw())) failed: XMM register should be 0-15
V [libjvm.so+0x1079d0d] MacroAssembler::vpbroadcastw(XMMRegisterImpl*, XMMRegisterImpl*, int)+0x6d
V [libjvm.so+0x342213] ReplS_immNode::emit(CodeBuffer&, PhaseRegAlloc*) const+0x373
As of now, the configuration is artificial since there's no such hardware exist.
- relates to
- 
                    JDK-8309660 C2: failed: XMM register should be 0-15 (UseKNLSetting and ConvF2HF) -           
- Resolved
 
-