Mercurial > hg > icedtea8-forest > hotspot
view src/cpu/aarch64/vm/methodHandles_aarch64.cpp @ 10905:f57189b7648d
8257192: Integrate AArch64 JIT port into 8u
7009641: Don't fail VM when CodeCache is full
8073108: [AArch64] Use x86 and SPARC CPU instructions for GHASH acceleration
8130309: Need to bailout cleanly if creation of stubs fails when codecache is out of space (AArch64 changes)
8131779: AARCH64: add Montgomery multiply intrinsic
8132875: AArch64: Fix error introduced into AArch64 CodeCache by commit for 8130309
8135018: AARCH64: Missing memory barriers for CMS collector
8145320: Create unsafe_arraycopy and generic_arraycopy for AArch64
8148328: aarch64: redundant lsr instructions in stub code.
8148783: aarch64: SEGV running SpecJBB2013
8148948: aarch64: generate_copy_longs calls align() incorrectly
8149080: AArch64: Recognise disjoint array copy in stub code
8149365: aarch64: memory copy does not prefetch on backwards copy
8149907: aarch64: use load/store pair instructions in call_stub
8150038: aarch64: make use of CBZ and CBNZ when comparing narrow pointer with zero
8150045: arraycopy causes segfaults in SATB during garbage collection
8150082: aarch64: optimise small array copy
8150229: aarch64: pipeline class for several instructions is not set correctly
8150313: aarch64: optimise array copy using SIMD instructions
8150394: aarch64: add support for 8.1 LSE CAS instructions
8150652: Remove unused code in AArch64 back end
8151340: aarch64: prefetch the destination word for write prior to ldxr/stxr loops.
8151502: optimize pd_disjoint_words and pd_conjoint_words
8151775: aarch64: add support for 8.1 LSE atomic operations
8152537: aarch64: Make use of CBZ and CBNZ when comparing unsigned values with zero.
8152840: aarch64: improve _unsafe_arraycopy stub routine
8153172: aarch64: hotspot crashes after the 8.1 LSE patch is merged
8153713: aarch64: improve short array clearing using store pair
8153797: aarch64: Add Arrays.fill stub code
8154413: AArch64: Better byte behaviour
8154537: AArch64: some integer rotate instructions are never emitted
8154739: AArch64: TemplateTable::fast_xaccess loads in wrong mode
8155015: Aarch64: bad assert in spill generation code
8155100: AArch64: Relax alignment requirement for byte_map_base
8155612: Aarch64: vector nodes need to support misaligned offset
8155617: aarch64: ClearArray does not use DC ZVA
8155627: Enable SA on AArch64
8155653: TestVectorUnalignedOffset.java not pushed with 8155612
8156731: aarch64: java/util/Arrays/Correct.java fails due to _generic_arraycopy stub routine
8157841: aarch64: prefetch ignores cache line size
8157906: aarch64: some more integer rotate instructions are never emitted
8158913: aarch64: SEGV running Spark terasort
8159052: aarch64: optimise unaligned copies in pd_disjoint_words and pd_conjoint_words
8159063: aarch64: optimise unaligned array copy long
8160748: [AArch64] Inconsistent types for ideal_reg
8161072: AArch64: jtreg compiler/uncommontrap/TestDeoptOOM failure
8161190: AArch64: Fix overflow in immediate cmp instruction
8164113: AArch64: follow-up the fix for 8161598
8165673: AArch64: Fix JNI floating point argument handling
8167200: AArch64: Broken stack pointer adjustment in interpreter
8167421: AArch64: in one core system, fatal error: Illegal threadstate encountered
8167595: AArch64: SEGV in stub code cipherBlockChaining_decryptAESCrypt
8168699: Validate special case invocations [AArch64 support]
8168888: Port 8160591: Improve internal array handling to AArch64.
8170100: AArch64: Crash in C1-compiled code accessing References
8170188: jtreg test compiler/types/TestMeetIncompatibleInterfaceArrays.java causes JVM crash
8170873: PPC64/aarch64: Poor StrictMath performance due to non-optimized compilation
8171537: aarch64: compiler/c1/Test6849574.java generates guarantee failure in C1
8172881: AArch64: assertion failure: the int pressure is incorrect
8173472: AArch64: C1 comparisons with null only use 32-bit instructions
8176100: [AArch64] [REDO][REDO] G1 Needs pre barrier on dereference of weak JNI handles
8177661: Correct ad rule output register types from iRegX to iRegXNoSp
8179954: AArch64: C1 and C2 volatile accesses are not sequentially consistent
8182581: aarch64: fix for crash caused by earlyret of compiled method
8183925: [AArch64] Decouple crash protection from watcher thread
8186325: AArch64: jtreg test hotspot/test/gc/g1/TestJNIWeakG1/TestJNIWeakG1.java SEGV
8187224: aarch64: some inconsistency between aarch64_ad.m4 and aarch64.ad
8189170: [AArch64] Add option to disable stack overflow checking in primordial thread for use with JNI_CreateJavaJVM
8193133: Assertion failure because 0xDEADDEAD can be in-heap
8195685: AArch64 port of 8174962: Better interface invocations
8195859: AArch64: vtableStubs gtest fails after 8174962
8196136: AArch64: Correct register use in patch for JDK-8194686
8196221: AArch64: Mistake in committed patch for JDK-8195859
8199712: [AArch64] Flight Recorder
8203481: Incorrect constraint for unextended_sp in frame:safe_for_sender
8203699: java/lang/invoke/SpecialInterfaceCall fails with SIGILL on aarch64
8205421: AARCH64: StubCodeMark should be placed after alignment
8206163: AArch64: incorrect code generation for StoreCM
8207345: Trampoline generation code reads from uninitialized memory
8207838: AArch64: Float registers incorrectly restored in JNI call
8209413: AArch64: NPE in clhsdb jstack command
8209414: [AArch64] method handle invocation does not respect JVMTI interp_only mode
8209415: Fix JVMTI test failure HS202
8209420: Track membars for volatile accesses so they can be properly optimized
8209835: Aarch64: elide barriers on all volatile operations
8210425: [AArch64] sharedRuntimeTrig/sharedRuntimeTrans compiled without optimization
8211064: [AArch64] Interpreter and c1 don't correctly handle jboolean results in native calls
8211233: MemBarNode::trailing_membar() and MemBarNode::leading_membar() need to handle dying subgraphs better
8213134: AArch64: vector shift failed with MaxVectorSize=8
8213419: [AArch64] C2 may hang in MulLNode::Ideal()/MulINode::Ideal() with gcc 8.2.1
8214857: "bad trailing membar" assert failure at memnode.cpp:3220
8215951: AArch64: jtreg test vmTestbase/nsk/jvmti/PopFrame/popframe005 segfaults
8215961: jdk/jfr/event/os/TestCPUInformation.java fails on AArch64
8216350: AArch64: monitor unlock fast path not called
8216989: CardTableBarrierSetAssembler::gen_write_ref_array_post_barrier() does not check for zero length on AARCH64
8217368: AArch64: C2 recursive stack locking optimisation not triggered
8218185: aarch64: missing LoadStore barrier in TemplateTable::putfield_or_static
8219011: Implement MacroAssembler::warn method on AArch64
8219635: aarch64: missing LoadStore barrier in TemplateTable::fast_storefield
8221220: AArch64: Add StoreStore membar explicitly for Volatile Writes in TemplateTable
8221658: aarch64: add necessary predicate for ubfx patterns
8224671: AArch64: mauve System.arraycopy test failure
8224828: aarch64: rflags is not correct after safepoint poll
8224851: AArch64: fix warnings and errors with Clang and GCC 8.3
8224880: AArch64: java/javac error with AllocatePrefetchDistance
8228400: Remove built-in AArch64 simulator
8228406: Superfluous change in chaitin.hpp
8228593: Revert explicit JDK 7 support additions
8228716: Revert InstanceKlass::print_on debug additions
8228718: Revert incorrect backport of JDK-8129757 to 8-aarch64
8228725: AArch64: Purge method call format support
8228747: Revert "unused" attribute from test_arraycopy_func
8228767: Revert ResourceMark additions
8228770: Revert development hsdis changes
8229123: Revert build fixes for aarch64/zero
8229124: Revert disassembler.cpp changes
8229145: Revert TemplateTable::bytecode() visibility change
8233839: aarch64: missing memory barrier in NewObjectArrayStub and NewTypeArrayStub
8237512: AArch64: aarch64TestHook leaks a BufferBlob
8246482: Build failures with +JFR -PCH
8247979: aarch64: missing side effect of killing flags for clearArray_reg_reg
8248219: aarch64: missing memory barrier in fast_storefield and fast_accessfield
Reviewed-by: shade, aph
author | andrew |
---|---|
date | Mon, 01 Feb 2021 03:48:36 +0000 |
parents | |
children | f79e943d15a7 |
line wrap: on
line source
/* * Copyright (c) 2013, Red Hat Inc. * Copyright (c) 1997, 2012, Oracle and/or its affiliates. * All rights reserved. * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. * * This code is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License version 2 only, as * published by the Free Software Foundation. * * This code is distributed in the hope that it will be useful, but WITHOUT * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * version 2 for more details (a copy is included in the LICENSE file that * accompanied this code). * * You should have received a copy of the GNU General Public License version * 2 along with this work; if not, write to the Free Software Foundation, * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA. * * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA * or visit www.oracle.com if you need additional information or have any * questions. * */ #include "precompiled.hpp" #include "asm/macroAssembler.hpp" #include "interpreter/interpreter.hpp" #include "interpreter/interpreterRuntime.hpp" #include "memory/allocation.inline.hpp" #include "prims/methodHandles.hpp" #define __ _masm-> #ifdef PRODUCT #define BLOCK_COMMENT(str) /* nothing */ #else #define BLOCK_COMMENT(str) __ block_comment(str) #endif #define BIND(label) bind(label); BLOCK_COMMENT(#label ":") void MethodHandles::load_klass_from_Class(MacroAssembler* _masm, Register klass_reg) { if (VerifyMethodHandles) verify_klass(_masm, klass_reg, SystemDictionary::WK_KLASS_ENUM_NAME(java_lang_Class), "MH argument is a Class"); __ ldr(klass_reg, Address(klass_reg, java_lang_Class::klass_offset_in_bytes())); } #ifdef ASSERT static int check_nonzero(const char* xname, int x) { assert(x != 0, err_msg("%s should be nonzero", xname)); return x; } #define NONZERO(x) check_nonzero(#x, x) #else //ASSERT #define NONZERO(x) (x) #endif //PRODUCT #ifdef ASSERT void MethodHandles::verify_klass(MacroAssembler* _masm, Register obj, SystemDictionary::WKID klass_id, const char* error_message) { Klass** klass_addr = SystemDictionary::well_known_klass_addr(klass_id); KlassHandle klass = SystemDictionary::well_known_klass(klass_id); Register temp = rscratch2; Register temp2 = rscratch1; // used by MacroAssembler::cmpptr Label L_ok, L_bad; BLOCK_COMMENT("verify_klass {"); __ verify_oop(obj); __ cbz(obj, L_bad); __ push(RegSet::of(temp, temp2), sp); __ load_klass(temp, obj); __ cmpptr(temp, ExternalAddress((address) klass_addr)); __ br(Assembler::EQ, L_ok); intptr_t super_check_offset = klass->super_check_offset(); __ ldr(temp, Address(temp, super_check_offset)); __ cmpptr(temp, ExternalAddress((address) klass_addr)); __ br(Assembler::EQ, L_ok); __ pop(RegSet::of(temp, temp2), sp); __ bind(L_bad); __ stop(error_message); __ BIND(L_ok); __ pop(RegSet::of(temp, temp2), sp); BLOCK_COMMENT("} verify_klass"); } void MethodHandles::verify_ref_kind(MacroAssembler* _masm, int ref_kind, Register member_reg, Register temp) { } #endif //ASSERT void MethodHandles::jump_from_method_handle(MacroAssembler* _masm, Register method, Register temp, bool for_compiler_entry) { assert(method == rmethod, "interpreter calling convention"); Label L_no_such_method; __ cbz(rmethod, L_no_such_method); __ verify_method_ptr(method); if (!for_compiler_entry && JvmtiExport::can_post_interpreter_events()) { Label run_compiled_code; // JVMTI events, such as single-stepping, are implemented partly by avoiding running // compiled code in threads for which the event is enabled. Check here for // interp_only_mode if these events CAN be enabled. __ ldrw(rscratch1, Address(rthread, JavaThread::interp_only_mode_offset())); __ cbzw(rscratch1, run_compiled_code); __ ldr(rscratch1, Address(method, Method::interpreter_entry_offset())); __ br(rscratch1); __ BIND(run_compiled_code); } const ByteSize entry_offset = for_compiler_entry ? Method::from_compiled_offset() : Method::from_interpreted_offset(); __ ldr(rscratch1,Address(method, entry_offset)); __ br(rscratch1); __ bind(L_no_such_method); __ far_jump(RuntimeAddress(StubRoutines::throw_AbstractMethodError_entry())); } void MethodHandles::jump_to_lambda_form(MacroAssembler* _masm, Register recv, Register method_temp, Register temp2, bool for_compiler_entry) { BLOCK_COMMENT("jump_to_lambda_form {"); // This is the initial entry point of a lazy method handle. // After type checking, it picks up the invoker from the LambdaForm. assert_different_registers(recv, method_temp, temp2); assert(recv != noreg, "required register"); assert(method_temp == rmethod, "required register for loading method"); //NOT_PRODUCT({ FlagSetting fs(TraceMethodHandles, true); trace_method_handle(_masm, "LZMH"); }); // Load the invoker, as MH -> MH.form -> LF.vmentry __ verify_oop(recv); __ load_heap_oop(method_temp, Address(recv, NONZERO(java_lang_invoke_MethodHandle::form_offset_in_bytes()))); __ verify_oop(method_temp); __ load_heap_oop(method_temp, Address(method_temp, NONZERO(java_lang_invoke_LambdaForm::vmentry_offset_in_bytes()))); __ verify_oop(method_temp); // the following assumes that a Method* is normally compressed in the vmtarget field: __ ldr(method_temp, Address(method_temp, NONZERO(java_lang_invoke_MemberName::vmtarget_offset_in_bytes()))); if (VerifyMethodHandles && !for_compiler_entry) { // make sure recv is already on stack __ ldr(temp2, Address(method_temp, Method::const_offset())); __ load_sized_value(temp2, Address(temp2, ConstMethod::size_of_parameters_offset()), sizeof(u2), /*is_signed*/ false); // assert(sizeof(u2) == sizeof(Method::_size_of_parameters), ""); Label L; __ ldr(rscratch1, __ argument_address(temp2, -1)); __ cmp(recv, rscratch1); __ br(Assembler::EQ, L); __ ldr(r0, __ argument_address(temp2, -1)); __ hlt(0); __ BIND(L); } jump_from_method_handle(_masm, method_temp, temp2, for_compiler_entry); BLOCK_COMMENT("} jump_to_lambda_form"); } // Code generation address MethodHandles::generate_method_handle_interpreter_entry(MacroAssembler* _masm, vmIntrinsics::ID iid) { const bool not_for_compiler_entry = false; // this is the interpreter entry assert(is_signature_polymorphic(iid), "expected invoke iid"); if (iid == vmIntrinsics::_invokeGeneric || iid == vmIntrinsics::_compiledLambdaForm) { // Perhaps surprisingly, the symbolic references visible to Java are not directly used. // They are linked to Java-generated adapters via MethodHandleNatives.linkMethod. // They all allow an appendix argument. __ hlt(0); // empty stubs make SG sick return NULL; } // r13: sender SP (must preserve; see prepare_to_jump_from_interpreted) // rmethod: Method* // r3: argument locator (parameter slot count, added to rsp) // r1: used as temp to hold mh or receiver // r0, r11: garbage temps, blown away Register argp = r3; // argument list ptr, live on error paths Register temp = r0; Register mh = r1; // MH receiver; dies quickly and is recycled // here's where control starts out: __ align(CodeEntryAlignment); address entry_point = __ pc(); if (VerifyMethodHandles) { Label L; BLOCK_COMMENT("verify_intrinsic_id {"); __ ldrb(rscratch1, Address(rmethod, Method::intrinsic_id_offset_in_bytes())); __ cmp(rscratch1, (int) iid); __ br(Assembler::EQ, L); if (iid == vmIntrinsics::_linkToVirtual || iid == vmIntrinsics::_linkToSpecial) { // could do this for all kinds, but would explode assembly code size trace_method_handle(_masm, "bad Method*::intrinsic_id"); } __ hlt(0); __ bind(L); BLOCK_COMMENT("} verify_intrinsic_id"); } // First task: Find out how big the argument list is. Address r3_first_arg_addr; int ref_kind = signature_polymorphic_intrinsic_ref_kind(iid); assert(ref_kind != 0 || iid == vmIntrinsics::_invokeBasic, "must be _invokeBasic or a linkTo intrinsic"); if (ref_kind == 0 || MethodHandles::ref_kind_has_receiver(ref_kind)) { __ ldr(argp, Address(rmethod, Method::const_offset())); __ load_sized_value(argp, Address(argp, ConstMethod::size_of_parameters_offset()), sizeof(u2), /*is_signed*/ false); // assert(sizeof(u2) == sizeof(Method::_size_of_parameters), ""); r3_first_arg_addr = __ argument_address(argp, -1); } else { DEBUG_ONLY(argp = noreg); } if (!is_signature_polymorphic_static(iid)) { __ ldr(mh, r3_first_arg_addr); DEBUG_ONLY(argp = noreg); } // r3_first_arg_addr is live! trace_method_handle_interpreter_entry(_masm, iid); if (iid == vmIntrinsics::_invokeBasic) { generate_method_handle_dispatch(_masm, iid, mh, noreg, not_for_compiler_entry); } else { // Adjust argument list by popping the trailing MemberName argument. Register recv = noreg; if (MethodHandles::ref_kind_has_receiver(ref_kind)) { // Load the receiver (not the MH; the actual MemberName's receiver) up from the interpreter stack. __ ldr(recv = r2, r3_first_arg_addr); } DEBUG_ONLY(argp = noreg); Register rmember = rmethod; // MemberName ptr; incoming method ptr is dead now __ pop(rmember); // extract last argument generate_method_handle_dispatch(_masm, iid, recv, rmember, not_for_compiler_entry); } return entry_point; } void MethodHandles::generate_method_handle_dispatch(MacroAssembler* _masm, vmIntrinsics::ID iid, Register receiver_reg, Register member_reg, bool for_compiler_entry) { assert(is_signature_polymorphic(iid), "expected invoke iid"); // temps used in this code are not used in *either* compiled or interpreted calling sequences Register temp1 = r10; Register temp2 = r11; Register temp3 = r14; // r13 is live by this point: it contains the sender SP if (for_compiler_entry) { assert(receiver_reg == (iid == vmIntrinsics::_linkToStatic ? noreg : j_rarg0), "only valid assignment"); assert_different_registers(temp1, j_rarg0, j_rarg1, j_rarg2, j_rarg3, j_rarg4, j_rarg5, j_rarg6, j_rarg7); assert_different_registers(temp2, j_rarg0, j_rarg1, j_rarg2, j_rarg3, j_rarg4, j_rarg5, j_rarg6, j_rarg7); assert_different_registers(temp3, j_rarg0, j_rarg1, j_rarg2, j_rarg3, j_rarg4, j_rarg5, j_rarg6, j_rarg7); } assert_different_registers(temp1, temp2, temp3, receiver_reg); assert_different_registers(temp1, temp2, temp3, member_reg); if (iid == vmIntrinsics::_invokeBasic) { // indirect through MH.form.vmentry.vmtarget jump_to_lambda_form(_masm, receiver_reg, rmethod, temp1, for_compiler_entry); } else { // The method is a member invoker used by direct method handles. if (VerifyMethodHandles) { // make sure the trailing argument really is a MemberName (caller responsibility) verify_klass(_masm, member_reg, SystemDictionary::WK_KLASS_ENUM_NAME(java_lang_invoke_MemberName), "MemberName required for invokeVirtual etc."); } Address member_clazz( member_reg, NONZERO(java_lang_invoke_MemberName::clazz_offset_in_bytes())); Address member_vmindex( member_reg, NONZERO(java_lang_invoke_MemberName::vmindex_offset_in_bytes())); Address member_vmtarget( member_reg, NONZERO(java_lang_invoke_MemberName::vmtarget_offset_in_bytes())); Register temp1_recv_klass = temp1; if (iid != vmIntrinsics::_linkToStatic) { __ verify_oop(receiver_reg); if (iid == vmIntrinsics::_linkToSpecial) { // Don't actually load the klass; just null-check the receiver. __ null_check(receiver_reg); } else { // load receiver klass itself __ null_check(receiver_reg, oopDesc::klass_offset_in_bytes()); __ load_klass(temp1_recv_klass, receiver_reg); __ verify_klass_ptr(temp1_recv_klass); } BLOCK_COMMENT("check_receiver {"); // The receiver for the MemberName must be in receiver_reg. // Check the receiver against the MemberName.clazz if (VerifyMethodHandles && iid == vmIntrinsics::_linkToSpecial) { // Did not load it above... __ load_klass(temp1_recv_klass, receiver_reg); __ verify_klass_ptr(temp1_recv_klass); } if (VerifyMethodHandles && iid != vmIntrinsics::_linkToInterface) { Label L_ok; Register temp2_defc = temp2; __ load_heap_oop(temp2_defc, member_clazz); load_klass_from_Class(_masm, temp2_defc); __ verify_klass_ptr(temp2_defc); __ check_klass_subtype(temp1_recv_klass, temp2_defc, temp3, L_ok); // If we get here, the type check failed! __ hlt(0); // __ STOP("receiver class disagrees with MemberName.clazz"); __ bind(L_ok); } BLOCK_COMMENT("} check_receiver"); } if (iid == vmIntrinsics::_linkToSpecial || iid == vmIntrinsics::_linkToStatic) { DEBUG_ONLY(temp1_recv_klass = noreg); // these guys didn't load the recv_klass } // Live registers at this point: // member_reg - MemberName that was the trailing argument // temp1_recv_klass - klass of stacked receiver, if needed // r13 - interpreter linkage (if interpreted) ??? FIXME // r1 ... r0 - compiler arguments (if compiled) Label L_incompatible_class_change_error; switch (iid) { case vmIntrinsics::_linkToSpecial: if (VerifyMethodHandles) { verify_ref_kind(_masm, JVM_REF_invokeSpecial, member_reg, temp3); } __ ldr(rmethod, member_vmtarget); break; case vmIntrinsics::_linkToStatic: if (VerifyMethodHandles) { verify_ref_kind(_masm, JVM_REF_invokeStatic, member_reg, temp3); } __ ldr(rmethod, member_vmtarget); break; case vmIntrinsics::_linkToVirtual: { // same as TemplateTable::invokevirtual, // minus the CP setup and profiling: if (VerifyMethodHandles) { verify_ref_kind(_masm, JVM_REF_invokeVirtual, member_reg, temp3); } // pick out the vtable index from the MemberName, and then we can discard it: Register temp2_index = temp2; __ ldr(temp2_index, member_vmindex); if (VerifyMethodHandles) { Label L_index_ok; __ cmpw(temp2_index, 0U); __ br(Assembler::GE, L_index_ok); __ hlt(0); __ BIND(L_index_ok); } // Note: The verifier invariants allow us to ignore MemberName.clazz and vmtarget // at this point. And VerifyMethodHandles has already checked clazz, if needed. // get target Method* & entry point __ lookup_virtual_method(temp1_recv_klass, temp2_index, rmethod); break; } case vmIntrinsics::_linkToInterface: { // same as TemplateTable::invokeinterface // (minus the CP setup and profiling, with different argument motion) if (VerifyMethodHandles) { verify_ref_kind(_masm, JVM_REF_invokeInterface, member_reg, temp3); } Register temp3_intf = temp3; __ load_heap_oop(temp3_intf, member_clazz); load_klass_from_Class(_masm, temp3_intf); __ verify_klass_ptr(temp3_intf); Register rindex = rmethod; __ ldr(rindex, member_vmindex); if (VerifyMethodHandles) { Label L; __ cmpw(rindex, 0U); __ br(Assembler::GE, L); __ hlt(0); __ bind(L); } // given intf, index, and recv klass, dispatch to the implementation method __ lookup_interface_method(temp1_recv_klass, temp3_intf, // note: next two args must be the same: rindex, rmethod, temp2, L_incompatible_class_change_error); break; } default: fatal(err_msg_res("unexpected intrinsic %d: %s", iid, vmIntrinsics::name_at(iid))); break; } // live at this point: rmethod, r13 (if interpreted) // After figuring out which concrete method to call, jump into it. // Note that this works in the interpreter with no data motion. // But the compiled version will require that r2_recv be shifted out. __ verify_method_ptr(rmethod); jump_from_method_handle(_masm, rmethod, temp1, for_compiler_entry); if (iid == vmIntrinsics::_linkToInterface) { __ bind(L_incompatible_class_change_error); __ far_jump(RuntimeAddress(StubRoutines::throw_IncompatibleClassChangeError_entry())); } } } #ifndef PRODUCT void trace_method_handle_stub(const char* adaptername, oop mh, intptr_t* saved_regs, intptr_t* entry_sp) { } // The stub wraps the arguments in a struct on the stack to avoid // dealing with the different calling conventions for passing 6 // arguments. struct MethodHandleStubArguments { const char* adaptername; oopDesc* mh; intptr_t* saved_regs; intptr_t* entry_sp; }; void trace_method_handle_stub_wrapper(MethodHandleStubArguments* args) { } void MethodHandles::trace_method_handle(MacroAssembler* _masm, const char* adaptername) { } #endif //PRODUCT