Mercurial > hg > icedtea8-forest > hotspot
view src/share/vm/utilities/copy.hpp @ 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 | dee6a1ce4a0c |
children | f79e943d15a7 |
line wrap: on
line source
/* * Copyright (c) 2003, 2018, 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. * */ #ifndef SHARE_VM_UTILITIES_COPY_HPP #define SHARE_VM_UTILITIES_COPY_HPP #include "runtime/stubRoutines.hpp" // Assembly code for platforms that need it. extern "C" { void _Copy_conjoint_words(HeapWord* from, HeapWord* to, size_t count); void _Copy_disjoint_words(HeapWord* from, HeapWord* to, size_t count); void _Copy_conjoint_words_atomic(HeapWord* from, HeapWord* to, size_t count); void _Copy_disjoint_words_atomic(HeapWord* from, HeapWord* to, size_t count); void _Copy_aligned_conjoint_words(HeapWord* from, HeapWord* to, size_t count); void _Copy_aligned_disjoint_words(HeapWord* from, HeapWord* to, size_t count); void _Copy_conjoint_bytes(void* from, void* to, size_t count); void _Copy_conjoint_bytes_atomic (void* from, void* to, size_t count); void _Copy_conjoint_jshorts_atomic(jshort* from, jshort* to, size_t count); void _Copy_conjoint_jints_atomic (jint* from, jint* to, size_t count); void _Copy_conjoint_jlongs_atomic (jlong* from, jlong* to, size_t count); void _Copy_conjoint_oops_atomic (oop* from, oop* to, size_t count); void _Copy_arrayof_conjoint_bytes (HeapWord* from, HeapWord* to, size_t count); void _Copy_arrayof_conjoint_jshorts(HeapWord* from, HeapWord* to, size_t count); void _Copy_arrayof_conjoint_jints (HeapWord* from, HeapWord* to, size_t count); void _Copy_arrayof_conjoint_jlongs (HeapWord* from, HeapWord* to, size_t count); void _Copy_arrayof_conjoint_oops (HeapWord* from, HeapWord* to, size_t count); } class Copy : AllStatic { public: // Block copy methods have four attributes. We don't define all possibilities. // alignment: aligned to BytesPerLong // arrayof: arraycopy operation with both operands aligned on the same // boundary as the first element of an array of the copy unit. // This is currently a HeapWord boundary on all platforms, except // for long and double arrays, which are aligned on an 8-byte // boundary on all platforms. // arraycopy operations are implicitly atomic on each array element. // overlap: disjoint or conjoint. // copy unit: bytes or words (i.e., HeapWords) or oops (i.e., pointers). // atomicity: atomic or non-atomic on the copy unit. // // Names are constructed thusly: // // [ 'aligned_' | 'arrayof_' ] // ('conjoint_' | 'disjoint_') // ('words' | 'bytes' | 'jshorts' | 'jints' | 'jlongs' | 'oops') // [ '_atomic' ] // // Except in the arrayof case, whatever the alignment is, we assume we can copy // whole alignment units. E.g., if BytesPerLong is 2x word alignment, an odd // count may copy an extra word. In the arrayof case, we are allowed to copy // only the number of copy units specified. // // All callees check count for 0. // // HeapWords // Word-aligned words, conjoint, not atomic on each word static void conjoint_words(HeapWord* from, HeapWord* to, size_t count) { assert_params_ok(from, to, LogHeapWordSize); pd_conjoint_words(from, to, count); } // Word-aligned words, disjoint, not atomic on each word static void disjoint_words(HeapWord* from, HeapWord* to, size_t count) { assert_params_ok(from, to, LogHeapWordSize); assert_disjoint(from, to, count); pd_disjoint_words(from, to, count); } // Word-aligned words, disjoint, atomic on each word static void disjoint_words_atomic(HeapWord* from, HeapWord* to, size_t count) { assert_params_ok(from, to, LogHeapWordSize); assert_disjoint(from, to, count); pd_disjoint_words_atomic(from, to, count); } // Object-aligned words, conjoint, not atomic on each word static void aligned_conjoint_words(HeapWord* from, HeapWord* to, size_t count) { assert_params_aligned(from, to); pd_aligned_conjoint_words(from, to, count); } // Object-aligned words, disjoint, not atomic on each word static void aligned_disjoint_words(HeapWord* from, HeapWord* to, size_t count) { assert_params_aligned(from, to); assert_disjoint(from, to, count); pd_aligned_disjoint_words(from, to, count); } // bytes, jshorts, jints, jlongs, oops // bytes, conjoint, not atomic on each byte (not that it matters) static void conjoint_jbytes(void* from, void* to, size_t count) { pd_conjoint_bytes(from, to, count); } // bytes, conjoint, atomic on each byte (not that it matters) static void conjoint_jbytes_atomic(void* from, void* to, size_t count) { pd_conjoint_bytes(from, to, count); } // jshorts, conjoint, atomic on each jshort static void conjoint_jshorts_atomic(jshort* from, jshort* to, size_t count) { assert_params_ok(from, to, LogBytesPerShort); pd_conjoint_jshorts_atomic(from, to, count); } // jints, conjoint, atomic on each jint static void conjoint_jints_atomic(jint* from, jint* to, size_t count) { assert_params_ok(from, to, LogBytesPerInt); pd_conjoint_jints_atomic(from, to, count); } // jlongs, conjoint, atomic on each jlong static void conjoint_jlongs_atomic(jlong* from, jlong* to, size_t count) { assert_params_ok(from, to, LogBytesPerLong); pd_conjoint_jlongs_atomic(from, to, count); } // oops, conjoint, atomic on each oop static void conjoint_oops_atomic(oop* from, oop* to, size_t count) { assert_params_ok(from, to, LogBytesPerHeapOop); pd_conjoint_oops_atomic(from, to, count); } // overloaded for UseCompressedOops static void conjoint_oops_atomic(narrowOop* from, narrowOop* to, size_t count) { assert(sizeof(narrowOop) == sizeof(jint), "this cast is wrong"); assert_params_ok(from, to, LogBytesPerInt); pd_conjoint_jints_atomic((jint*)from, (jint*)to, count); } // Copy a span of memory. If the span is an integral number of aligned // longs, words, or ints, copy those units atomically. // The largest atomic transfer unit is 8 bytes, or the largest power // of two which divides all of from, to, and size, whichever is smaller. static void conjoint_memory_atomic(void* from, void* to, size_t size); // bytes, conjoint array, atomic on each byte (not that it matters) static void arrayof_conjoint_jbytes(HeapWord* from, HeapWord* to, size_t count) { pd_arrayof_conjoint_bytes(from, to, count); } // jshorts, conjoint array, atomic on each jshort static void arrayof_conjoint_jshorts(HeapWord* from, HeapWord* to, size_t count) { assert_params_ok(from, to, LogBytesPerShort); pd_arrayof_conjoint_jshorts(from, to, count); } // jints, conjoint array, atomic on each jint static void arrayof_conjoint_jints(HeapWord* from, HeapWord* to, size_t count) { assert_params_ok(from, to, LogBytesPerInt); pd_arrayof_conjoint_jints(from, to, count); } // jlongs, conjoint array, atomic on each jlong static void arrayof_conjoint_jlongs(HeapWord* from, HeapWord* to, size_t count) { assert_params_ok(from, to, LogBytesPerLong); pd_arrayof_conjoint_jlongs(from, to, count); } // oops, conjoint array, atomic on each oop static void arrayof_conjoint_oops(HeapWord* from, HeapWord* to, size_t count) { assert_params_ok(from, to, LogBytesPerHeapOop); pd_arrayof_conjoint_oops(from, to, count); } // Known overlap methods // Copy word-aligned words from higher to lower addresses, not atomic on each word inline static void conjoint_words_to_lower(HeapWord* from, HeapWord* to, size_t byte_count) { // byte_count is in bytes to check its alignment assert_params_ok(from, to, LogHeapWordSize); assert_byte_count_ok(byte_count, HeapWordSize); size_t count = (size_t)round_to(byte_count, HeapWordSize) >> LogHeapWordSize; assert(to <= from || from + count <= to, "do not overwrite source data"); while (count-- > 0) { *to++ = *from++; } } // Copy word-aligned words from lower to higher addresses, not atomic on each word inline static void conjoint_words_to_higher(HeapWord* from, HeapWord* to, size_t byte_count) { // byte_count is in bytes to check its alignment assert_params_ok(from, to, LogHeapWordSize); assert_byte_count_ok(byte_count, HeapWordSize); size_t count = (size_t)round_to(byte_count, HeapWordSize) >> LogHeapWordSize; assert(from <= to || to + count <= from, "do not overwrite source data"); from += count - 1; to += count - 1; while (count-- > 0) { *to-- = *from--; } } /** * Copy and *unconditionally* byte swap elements * * @param src address of source * @param dst address of destination * @param byte_count number of bytes to copy * @param elem_size size of the elements to copy-swap */ static void conjoint_swap(address src, address dst, size_t byte_count, size_t elem_size); // Fill methods // Fill word-aligned words, not atomic on each word // set_words static void fill_to_words(HeapWord* to, size_t count, juint value = 0) { assert_params_ok(to, LogHeapWordSize); pd_fill_to_words(to, count, value); } static void fill_to_aligned_words(HeapWord* to, size_t count, juint value = 0) { assert_params_aligned(to); pd_fill_to_aligned_words(to, count, value); } // Fill bytes static void fill_to_bytes(void* to, size_t count, jubyte value = 0) { pd_fill_to_bytes(to, count, value); } // Fill a span of memory. If the span is an integral number of aligned // longs, words, or ints, store to those units atomically. // The largest atomic transfer unit is 8 bytes, or the largest power // of two which divides both to and size, whichever is smaller. static void fill_to_memory_atomic(void* to, size_t size, jubyte value = 0); // Zero-fill methods // Zero word-aligned words, not atomic on each word static void zero_to_words(HeapWord* to, size_t count) { assert_params_ok(to, LogHeapWordSize); pd_zero_to_words(to, count); } // Zero bytes static void zero_to_bytes(void* to, size_t count) { pd_zero_to_bytes(to, count); } private: static bool params_disjoint(HeapWord* from, HeapWord* to, size_t count) { if (from < to) { return pointer_delta(to, from) >= count; } return pointer_delta(from, to) >= count; } // These methods raise a fatal if they detect a problem. static void assert_disjoint(HeapWord* from, HeapWord* to, size_t count) { #ifdef ASSERT if (!params_disjoint(from, to, count)) basic_fatal("source and dest overlap"); #endif } static void assert_params_ok(void* from, void* to, intptr_t log_align) { #ifdef ASSERT if (mask_bits((uintptr_t)from, right_n_bits(log_align)) != 0) basic_fatal("not aligned"); if (mask_bits((uintptr_t)to, right_n_bits(log_align)) != 0) basic_fatal("not aligned"); #endif } static void assert_params_ok(HeapWord* to, intptr_t log_align) { #ifdef ASSERT if (mask_bits((uintptr_t)to, right_n_bits(log_align)) != 0) basic_fatal("not word aligned"); #endif } static void assert_params_aligned(HeapWord* from, HeapWord* to) { #ifdef ASSERT if (mask_bits((uintptr_t)from, BytesPerLong-1) != 0) basic_fatal("not long aligned"); if (mask_bits((uintptr_t)to, BytesPerLong-1) != 0) basic_fatal("not long aligned"); #endif } static void assert_params_aligned(HeapWord* to) { #ifdef ASSERT if (mask_bits((uintptr_t)to, BytesPerLong-1) != 0) basic_fatal("not long aligned"); #endif } static void assert_byte_count_ok(size_t byte_count, size_t unit_size) { #ifdef ASSERT if ((size_t)round_to(byte_count, unit_size) != byte_count) { basic_fatal("byte count must be aligned"); } #endif } // Platform dependent implementations of the above methods. #ifdef TARGET_ARCH_x86 # include "copy_x86.hpp" #endif #ifdef TARGET_ARCH_aarch64 # include "copy_aarch64.hpp" #endif #ifdef TARGET_ARCH_sparc # include "copy_sparc.hpp" #endif #ifdef TARGET_ARCH_zero # include "copy_zero.hpp" #endif #ifdef TARGET_ARCH_arm # include "copy_arm.hpp" #endif #ifdef TARGET_ARCH_ppc # include "copy_ppc.hpp" #endif }; #endif // SHARE_VM_UTILITIES_COPY_HPP