Mercurial > hg > graal-compiler
view src/os_cpu/windows_x86/vm/os_windows_x86.cpp @ 18954:e8a4803f27f0
Simplifications around PEA and using earliest possible schedule for nodes.
author | Thomas Wuerthinger <thomas.wuerthinger@oracle.com> |
---|---|
date | Mon, 26 Jan 2015 23:24:06 +0100 |
parents | 89152779163c |
children | 7848fc12602b |
line wrap: on
line source
/* * Copyright (c) 1999, 2013, 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. * */ // no precompiled headers #include "asm/macroAssembler.hpp" #include "classfile/classLoader.hpp" #include "classfile/systemDictionary.hpp" #include "classfile/vmSymbols.hpp" #include "code/icBuffer.hpp" #include "code/vtableStubs.hpp" #include "decoder_windows.hpp" #include "interpreter/interpreter.hpp" #include "jvm_windows.h" #include "memory/allocation.inline.hpp" #include "mutex_windows.inline.hpp" #include "nativeInst_x86.hpp" #include "os_share_windows.hpp" #include "prims/jniFastGetField.hpp" #include "prims/jvm.h" #include "prims/jvm_misc.hpp" #include "runtime/arguments.hpp" #include "runtime/extendedPC.hpp" #include "runtime/frame.inline.hpp" #include "runtime/interfaceSupport.hpp" #include "runtime/java.hpp" #include "runtime/javaCalls.hpp" #include "runtime/mutexLocker.hpp" #include "runtime/osThread.hpp" #include "runtime/sharedRuntime.hpp" #include "runtime/stubRoutines.hpp" #include "runtime/thread.inline.hpp" #include "runtime/timer.hpp" #include "utilities/events.hpp" #include "utilities/vmError.hpp" # include "unwind_windows_x86.hpp" #undef REG_SP #undef REG_FP #undef REG_PC #ifdef AMD64 #define REG_SP Rsp #define REG_FP Rbp #define REG_PC Rip #else #define REG_SP Esp #define REG_FP Ebp #define REG_PC Eip #endif // AMD64 extern LONG WINAPI topLevelExceptionFilter(_EXCEPTION_POINTERS* ); // Install a win32 structured exception handler around thread. void os::os_exception_wrapper(java_call_t f, JavaValue* value, methodHandle* method, JavaCallArguments* args, Thread* thread) { __try { #ifndef AMD64 // We store the current thread in this wrapperthread location // and determine how far away this address is from the structured // execption pointer that FS:[0] points to. This get_thread // code can then get the thread pointer via FS. // // Warning: This routine must NEVER be inlined since we'd end up with // multiple offsets. // volatile Thread* wrapperthread = thread; if ( ThreadLocalStorage::get_thread_ptr_offset() == 0 ) { int thread_ptr_offset; __asm { lea eax, dword ptr wrapperthread; sub eax, dword ptr FS:[0H]; mov thread_ptr_offset, eax }; ThreadLocalStorage::set_thread_ptr_offset(thread_ptr_offset); } #ifdef ASSERT // Verify that the offset hasn't changed since we initally captured // it. This might happen if we accidentally ended up with an // inlined version of this routine. else { int test_thread_ptr_offset; __asm { lea eax, dword ptr wrapperthread; sub eax, dword ptr FS:[0H]; mov test_thread_ptr_offset, eax }; assert(test_thread_ptr_offset == ThreadLocalStorage::get_thread_ptr_offset(), "thread pointer offset from SEH changed"); } #endif // ASSERT #endif // !AMD64 f(value, method, args, thread); } __except(topLevelExceptionFilter((_EXCEPTION_POINTERS*)_exception_info())) { // Nothing to do. } } #ifdef AMD64 // This is the language specific handler for exceptions // originating from dynamically generated code. // We call the standard structured exception handler // We only expect Continued Execution since we cannot unwind // from generated code. LONG HandleExceptionFromCodeCache( IN PEXCEPTION_RECORD ExceptionRecord, IN ULONG64 EstablisherFrame, IN OUT PCONTEXT ContextRecord, IN OUT PDISPATCHER_CONTEXT DispatcherContext) { EXCEPTION_POINTERS ep; LONG result; ep.ExceptionRecord = ExceptionRecord; ep.ContextRecord = ContextRecord; result = topLevelExceptionFilter(&ep); // We better only get a CONTINUE_EXECUTION from our handler // since we don't have unwind information registered. guarantee( result == EXCEPTION_CONTINUE_EXECUTION, "Unexpected result from topLevelExceptionFilter"); return(ExceptionContinueExecution); } // Structure containing the Windows Data Structures required // to register our Code Cache exception handler. // We put these in the CodeCache since the API requires // all addresses in these structures are relative to the Code // area registered with RtlAddFunctionTable. typedef struct { char ExceptionHandlerInstr[16]; // jmp HandleExceptionFromCodeCache RUNTIME_FUNCTION rt; UNWIND_INFO_EH_ONLY unw; } DynamicCodeData, *pDynamicCodeData; #endif // AMD64 // // Register our CodeCache area with the OS so it will dispatch exceptions // to our topLevelExceptionFilter when we take an exception in our // dynamically generated code. // // Arguments: low and high are the address of the full reserved // codeCache area // bool os::register_code_area(char *low, char *high) { #ifdef AMD64 ResourceMark rm; pDynamicCodeData pDCD; PRUNTIME_FUNCTION prt; PUNWIND_INFO_EH_ONLY punwind; BufferBlob* blob = BufferBlob::create("CodeCache Exception Handler", sizeof(DynamicCodeData)); CodeBuffer cb(blob); MacroAssembler* masm = new MacroAssembler(&cb); pDCD = (pDynamicCodeData) masm->pc(); masm->jump(ExternalAddress((address)&HandleExceptionFromCodeCache)); masm->flush(); // Create an Unwind Structure specifying no unwind info // other than an Exception Handler punwind = &pDCD->unw; punwind->Version = 1; punwind->Flags = UNW_FLAG_EHANDLER; punwind->SizeOfProlog = 0; punwind->CountOfCodes = 0; punwind->FrameRegister = 0; punwind->FrameOffset = 0; punwind->ExceptionHandler = (char *)(&(pDCD->ExceptionHandlerInstr[0])) - (char*)low; punwind->ExceptionData[0] = 0; // This structure describes the covered dynamic code area. // Addresses are relative to the beginning on the code cache area prt = &pDCD->rt; prt->BeginAddress = 0; prt->EndAddress = (ULONG)(high - low); prt->UnwindData = ((char *)punwind - low); guarantee(RtlAddFunctionTable(prt, 1, (ULONGLONG)low), "Failed to register Dynamic Code Exception Handler with RtlAddFunctionTable"); #endif // AMD64 return true; } void os::initialize_thread(Thread* thr) { // Nothing to do. } // Atomics and Stub Functions typedef jint xchg_func_t (jint, volatile jint*); typedef intptr_t xchg_ptr_func_t (intptr_t, volatile intptr_t*); typedef jint cmpxchg_func_t (jint, volatile jint*, jint); typedef jlong cmpxchg_long_func_t (jlong, volatile jlong*, jlong); typedef jint add_func_t (jint, volatile jint*); typedef intptr_t add_ptr_func_t (intptr_t, volatile intptr_t*); #ifdef AMD64 jint os::atomic_xchg_bootstrap(jint exchange_value, volatile jint* dest) { // try to use the stub: xchg_func_t* func = CAST_TO_FN_PTR(xchg_func_t*, StubRoutines::atomic_xchg_entry()); if (func != NULL) { os::atomic_xchg_func = func; return (*func)(exchange_value, dest); } assert(Threads::number_of_threads() == 0, "for bootstrap only"); jint old_value = *dest; *dest = exchange_value; return old_value; } intptr_t os::atomic_xchg_ptr_bootstrap(intptr_t exchange_value, volatile intptr_t* dest) { // try to use the stub: xchg_ptr_func_t* func = CAST_TO_FN_PTR(xchg_ptr_func_t*, StubRoutines::atomic_xchg_ptr_entry()); if (func != NULL) { os::atomic_xchg_ptr_func = func; return (*func)(exchange_value, dest); } assert(Threads::number_of_threads() == 0, "for bootstrap only"); intptr_t old_value = *dest; *dest = exchange_value; return old_value; } jint os::atomic_cmpxchg_bootstrap(jint exchange_value, volatile jint* dest, jint compare_value) { // try to use the stub: cmpxchg_func_t* func = CAST_TO_FN_PTR(cmpxchg_func_t*, StubRoutines::atomic_cmpxchg_entry()); if (func != NULL) { os::atomic_cmpxchg_func = func; return (*func)(exchange_value, dest, compare_value); } assert(Threads::number_of_threads() == 0, "for bootstrap only"); jint old_value = *dest; if (old_value == compare_value) *dest = exchange_value; return old_value; } #endif // AMD64 jlong os::atomic_cmpxchg_long_bootstrap(jlong exchange_value, volatile jlong* dest, jlong compare_value) { // try to use the stub: cmpxchg_long_func_t* func = CAST_TO_FN_PTR(cmpxchg_long_func_t*, StubRoutines::atomic_cmpxchg_long_entry()); if (func != NULL) { os::atomic_cmpxchg_long_func = func; return (*func)(exchange_value, dest, compare_value); } assert(Threads::number_of_threads() == 0, "for bootstrap only"); jlong old_value = *dest; if (old_value == compare_value) *dest = exchange_value; return old_value; } #ifdef AMD64 jint os::atomic_add_bootstrap(jint add_value, volatile jint* dest) { // try to use the stub: add_func_t* func = CAST_TO_FN_PTR(add_func_t*, StubRoutines::atomic_add_entry()); if (func != NULL) { os::atomic_add_func = func; return (*func)(add_value, dest); } assert(Threads::number_of_threads() == 0, "for bootstrap only"); return (*dest) += add_value; } intptr_t os::atomic_add_ptr_bootstrap(intptr_t add_value, volatile intptr_t* dest) { // try to use the stub: add_ptr_func_t* func = CAST_TO_FN_PTR(add_ptr_func_t*, StubRoutines::atomic_add_ptr_entry()); if (func != NULL) { os::atomic_add_ptr_func = func; return (*func)(add_value, dest); } assert(Threads::number_of_threads() == 0, "for bootstrap only"); return (*dest) += add_value; } xchg_func_t* os::atomic_xchg_func = os::atomic_xchg_bootstrap; xchg_ptr_func_t* os::atomic_xchg_ptr_func = os::atomic_xchg_ptr_bootstrap; cmpxchg_func_t* os::atomic_cmpxchg_func = os::atomic_cmpxchg_bootstrap; add_func_t* os::atomic_add_func = os::atomic_add_bootstrap; add_ptr_func_t* os::atomic_add_ptr_func = os::atomic_add_ptr_bootstrap; #endif // AMD64 cmpxchg_long_func_t* os::atomic_cmpxchg_long_func = os::atomic_cmpxchg_long_bootstrap; #ifdef AMD64 /* * Windows/x64 does not use stack frames the way expected by Java: * [1] in most cases, there is no frame pointer. All locals are addressed via RSP * [2] in rare cases, when alloca() is used, a frame pointer is used, but this may * not be RBP. * See http://msdn.microsoft.com/en-us/library/ew5tede7.aspx * * So it's not possible to print the native stack using the * while (...) {... fr = os::get_sender_for_C_frame(&fr); } * loop in vmError.cpp. We need to roll our own loop. */ bool os::platform_print_native_stack(outputStream* st, void* context, char *buf, int buf_size) { CONTEXT ctx; if (context != NULL) { memcpy(&ctx, context, sizeof(ctx)); } else { RtlCaptureContext(&ctx); } st->print_cr("Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code)"); STACKFRAME stk; memset(&stk, 0, sizeof(stk)); stk.AddrStack.Offset = ctx.Rsp; stk.AddrStack.Mode = AddrModeFlat; stk.AddrFrame.Offset = ctx.Rbp; stk.AddrFrame.Mode = AddrModeFlat; stk.AddrPC.Offset = ctx.Rip; stk.AddrPC.Mode = AddrModeFlat; int count = 0; address lastpc = 0; while (count++ < StackPrintLimit) { intptr_t* sp = (intptr_t*)stk.AddrStack.Offset; intptr_t* fp = (intptr_t*)stk.AddrFrame.Offset; // NOT necessarily the same as ctx.Rbp! address pc = (address)stk.AddrPC.Offset; if (pc != NULL && sp != NULL && fp != NULL) { if (count == 2 && lastpc == pc) { // Skip it -- StackWalk64() may return the same PC // (but different SP) on the first try. } else { // Don't try to create a frame(sp, fp, pc) -- on WinX64, stk.AddrFrame // may not contain what Java expects, and may cause the frame() constructor // to crash. Let's just print out the symbolic address. frame::print_C_frame(st, buf, buf_size, pc); st->cr(); } lastpc = pc; } else { break; } PVOID p = WindowsDbgHelp::SymFunctionTableAccess64(GetCurrentProcess(), stk.AddrPC.Offset); if (!p) { // StackWalk64() can't handle this PC. Calling StackWalk64 again may cause crash. break; } BOOL result = WindowsDbgHelp::StackWalk64( IMAGE_FILE_MACHINE_AMD64, // __in DWORD MachineType, GetCurrentProcess(), // __in HANDLE hProcess, GetCurrentThread(), // __in HANDLE hThread, &stk, // __inout LP STACKFRAME64 StackFrame, &ctx, // __inout PVOID ContextRecord, NULL, // __in_opt PREAD_PROCESS_MEMORY_ROUTINE64 ReadMemoryRoutine, WindowsDbgHelp::pfnSymFunctionTableAccess64(), // __in_opt PFUNCTION_TABLE_ACCESS_ROUTINE64 FunctionTableAccessRoutine, WindowsDbgHelp::pfnSymGetModuleBase64(), // __in_opt PGET_MODULE_BASE_ROUTINE64 GetModuleBaseRoutine, NULL); // __in_opt PTRANSLATE_ADDRESS_ROUTINE64 TranslateAddress if (!result) { break; } } if (count > StackPrintLimit) { st->print_cr("...<more frames>..."); } st->cr(); return true; } #endif // AMD64 ExtendedPC os::fetch_frame_from_context(void* ucVoid, intptr_t** ret_sp, intptr_t** ret_fp) { ExtendedPC epc; CONTEXT* uc = (CONTEXT*)ucVoid; if (uc != NULL) { epc = ExtendedPC((address)uc->REG_PC); if (ret_sp) *ret_sp = (intptr_t*)uc->REG_SP; if (ret_fp) *ret_fp = (intptr_t*)uc->REG_FP; } else { // construct empty ExtendedPC for return value checking epc = ExtendedPC(NULL); if (ret_sp) *ret_sp = (intptr_t *)NULL; if (ret_fp) *ret_fp = (intptr_t *)NULL; } return epc; } frame os::fetch_frame_from_context(void* ucVoid) { intptr_t* sp; intptr_t* fp; ExtendedPC epc = fetch_frame_from_context(ucVoid, &sp, &fp); return frame(sp, fp, epc.pc()); } // VC++ does not save frame pointer on stack in optimized build. It // can be turned off by /Oy-. If we really want to walk C frames, // we can use the StackWalk() API. frame os::get_sender_for_C_frame(frame* fr) { return frame(fr->sender_sp(), fr->link(), fr->sender_pc()); } #ifndef AMD64 // Returns an estimate of the current stack pointer. Result must be guaranteed // to point into the calling threads stack, and be no lower than the current // stack pointer. address os::current_stack_pointer() { int dummy; address sp = (address)&dummy; return sp; } #else // Returns the current stack pointer. Accurate value needed for // os::verify_stack_alignment(). address os::current_stack_pointer() { typedef address get_sp_func(); get_sp_func* func = CAST_TO_FN_PTR(get_sp_func*, StubRoutines::x86::get_previous_sp_entry()); return (*func)(); } #endif #ifndef AMD64 intptr_t* _get_previous_fp() { intptr_t **frameptr; __asm { mov frameptr, ebp }; return *frameptr; } #endif // !AMD64 frame os::current_frame() { #ifdef AMD64 // apparently _asm not supported on windows amd64 typedef intptr_t* get_fp_func (); get_fp_func* func = CAST_TO_FN_PTR(get_fp_func*, StubRoutines::x86::get_previous_fp_entry()); if (func == NULL) return frame(); intptr_t* fp = (*func)(); if (fp == NULL) { return frame(); } #else intptr_t* fp = _get_previous_fp(); #endif // AMD64 frame myframe((intptr_t*)os::current_stack_pointer(), (intptr_t*)fp, CAST_FROM_FN_PTR(address, os::current_frame)); if (os::is_first_C_frame(&myframe)) { // stack is not walkable return frame(); } else { return os::get_sender_for_C_frame(&myframe); } } void os::print_context(outputStream *st, void *context) { if (context == NULL) return; CONTEXT* uc = (CONTEXT*)context; st->print_cr("Registers:"); #ifdef AMD64 st->print( "RAX=" INTPTR_FORMAT, uc->Rax); st->print(", RBX=" INTPTR_FORMAT, uc->Rbx); st->print(", RCX=" INTPTR_FORMAT, uc->Rcx); st->print(", RDX=" INTPTR_FORMAT, uc->Rdx); st->cr(); st->print( "RSP=" INTPTR_FORMAT, uc->Rsp); st->print(", RBP=" INTPTR_FORMAT, uc->Rbp); st->print(", RSI=" INTPTR_FORMAT, uc->Rsi); st->print(", RDI=" INTPTR_FORMAT, uc->Rdi); st->cr(); st->print( "R8 =" INTPTR_FORMAT, uc->R8); st->print(", R9 =" INTPTR_FORMAT, uc->R9); st->print(", R10=" INTPTR_FORMAT, uc->R10); st->print(", R11=" INTPTR_FORMAT, uc->R11); st->cr(); st->print( "R12=" INTPTR_FORMAT, uc->R12); st->print(", R13=" INTPTR_FORMAT, uc->R13); st->print(", R14=" INTPTR_FORMAT, uc->R14); st->print(", R15=" INTPTR_FORMAT, uc->R15); st->cr(); st->print( "RIP=" INTPTR_FORMAT, uc->Rip); st->print(", EFLAGS=" INTPTR_FORMAT, uc->EFlags); #else st->print( "EAX=" INTPTR_FORMAT, uc->Eax); st->print(", EBX=" INTPTR_FORMAT, uc->Ebx); st->print(", ECX=" INTPTR_FORMAT, uc->Ecx); st->print(", EDX=" INTPTR_FORMAT, uc->Edx); st->cr(); st->print( "ESP=" INTPTR_FORMAT, uc->Esp); st->print(", EBP=" INTPTR_FORMAT, uc->Ebp); st->print(", ESI=" INTPTR_FORMAT, uc->Esi); st->print(", EDI=" INTPTR_FORMAT, uc->Edi); st->cr(); st->print( "EIP=" INTPTR_FORMAT, uc->Eip); st->print(", EFLAGS=" INTPTR_FORMAT, uc->EFlags); #endif // AMD64 st->cr(); st->cr(); intptr_t *sp = (intptr_t *)uc->REG_SP; st->print_cr("Top of Stack: (sp=" PTR_FORMAT ")", sp); print_hex_dump(st, (address)sp, (address)(sp + 32), sizeof(intptr_t)); st->cr(); // Note: it may be unsafe to inspect memory near pc. For example, pc may // point to garbage if entry point in an nmethod is corrupted. Leave // this at the end, and hope for the best. address pc = (address)uc->REG_PC; st->print_cr("Instructions: (pc=" PTR_FORMAT ")", pc); print_hex_dump(st, pc - 32, pc + 32, sizeof(char)); st->cr(); } void os::print_register_info(outputStream *st, void *context) { if (context == NULL) return; CONTEXT* uc = (CONTEXT*)context; st->print_cr("Register to memory mapping:"); st->cr(); // this is only for the "general purpose" registers #ifdef AMD64 st->print("RAX="); print_location(st, uc->Rax); st->print("RBX="); print_location(st, uc->Rbx); st->print("RCX="); print_location(st, uc->Rcx); st->print("RDX="); print_location(st, uc->Rdx); st->print("RSP="); print_location(st, uc->Rsp); st->print("RBP="); print_location(st, uc->Rbp); st->print("RSI="); print_location(st, uc->Rsi); st->print("RDI="); print_location(st, uc->Rdi); st->print("R8 ="); print_location(st, uc->R8); st->print("R9 ="); print_location(st, uc->R9); st->print("R10="); print_location(st, uc->R10); st->print("R11="); print_location(st, uc->R11); st->print("R12="); print_location(st, uc->R12); st->print("R13="); print_location(st, uc->R13); st->print("R14="); print_location(st, uc->R14); st->print("R15="); print_location(st, uc->R15); #else st->print("EAX="); print_location(st, uc->Eax); st->print("EBX="); print_location(st, uc->Ebx); st->print("ECX="); print_location(st, uc->Ecx); st->print("EDX="); print_location(st, uc->Edx); st->print("ESP="); print_location(st, uc->Esp); st->print("EBP="); print_location(st, uc->Ebp); st->print("ESI="); print_location(st, uc->Esi); st->print("EDI="); print_location(st, uc->Edi); #endif st->cr(); } extern "C" int SpinPause () { #ifdef AMD64 return 0 ; #else // pause == rep:nop // On systems that don't support pause a rep:nop // is executed as a nop. The rep: prefix is ignored. _asm { pause ; }; return 1 ; #endif // AMD64 } void os::setup_fpu() { #ifndef AMD64 int fpu_cntrl_word = StubRoutines::fpu_cntrl_wrd_std(); __asm fldcw fpu_cntrl_word; #endif // !AMD64 } #ifndef PRODUCT void os::verify_stack_alignment() { #ifdef AMD64 assert(((intptr_t)os::current_stack_pointer() & (StackAlignmentInBytes-1)) == 0, "incorrect stack alignment"); #endif } #endif