-
-
[结贴] 关于asmjit 代码重定位问题
-
发表于: 2018-10-11 23:12 4505
-
新版本的重定位区说明
// -------------------------------------------------------------------------- // [Relocations] // -------------------------------------------------------------------------- //! Create a new relocation entry of type `type` and size `size`. //! //! Additional fields can be set after the relocation entry was created. ASMJIT_API Error newRelocEntry(RelocEntry** dst, uint32_t type, uint32_t size) noexcept; //! Get if the code contains relocations. ASMJIT_INLINE bool hasRelocations() const noexcept { return !_relocations.isEmpty(); } //! Get array of `RelocEntry*` records. ASMJIT_INLINE const ZoneVector<RelocEntry*>& getRelocEntries() const noexcept { return _relocations; } ASMJIT_INLINE RelocEntry* getRelocEntry(uint32_t id) const noexcept { return _relocations[id]; } //! Relocate the code to `baseAddress` and copy it to `dst`. //! //! \param dst Contains the location where the relocated code should be //! copied. The pointer can be address returned by virtual memory allocator //! or any other address that has sufficient space. //! //! \param baseAddress Base address used for relocation. `JitRuntime` always //! sets the `baseAddress` to be the same as `dst`. //! //! \return The number bytes actually used. If the code emitter reserved //! space for possible trampolines, but didn't use it, the number of bytes //! used can actually be less than the expected worst case. Virtual memory //! allocator can shrink the memory it allocated initially. //! //! A given buffer will be overwritten, to get the number of bytes required, //! use `getCodeSize()`. ASMJIT_API size_t relocate(void* dst, uint64_t baseAddress = Globals::kNoBaseAddress) const noexcept;
这个问题主要是关于 X86Assembler jmp() 立即数的使用问题
在之前的测试中,jmp 一个标签 (
Label
) 一个寄存器(
X86Gp
) 都是正常没有问题,但是jmp 一个立即数就没有值
测试代码:
using namespace asmjit::x86; // Easier access to x86/x64 registers. CodeHolder code; // Create a CodeHolder. code.init(CodeInfo(ArchInfo::kTypeX86));// Initialize it for a 32-bit X86 target. // Generate a 32-bit function that sums 4 floats and looks like: // void func(float* dst, const float* a, const float* b) X86Assembler a(&code); // Create and attach X86Assembler to `code`. a.mov(eax, dword_ptr(esp, 4)); // Load the destination pointer. a.mov(ecx, dword_ptr(esp, 8)); // Load the first source pointer. a.mov(edx, dword_ptr(esp, 12)); // Load the second source pointer. a.movups(xmm0, ptr(ecx)); // Load 4 floats from [ecx] to XMM0. a.movups(xmm1, ptr(edx)); // Load 4 floats from [edx] to XMM1. a.addps(xmm0, xmm1); // Add 4 floats in XMM1 to XMM0. a.movups(ptr(eax), xmm0); // Store the result to [eax]. unsigned long addr = 0x401000; a.jmp(addr); a.ret(); // Return from function. // Now we have two options if we want to do something with the code hold // by CodeHolder. In order to use it we must first sync X86Assembler with // the CodeHolder as it doesn't do it for every instruction it generates for // performance reasons. The options are: // // 1. Detach X86Assembler from CodeHolder (will automatically sync). // 2. Sync explicitly, allows to use X86Assembler again if needed. // // NOTE: AsmJit always syncs internally when CodeHolder needs to access these // buffers and knows that there is an Assembler attached, so you have to sync // explicitly only if you bypass CodeHolder and intend to do something on your // own. code.sync(); // So let's sync, it's easy. // We have no Runtime this time, it's on us what we do with the code. // CodeHolder stores code in SectionEntry, which embeds CodeSection // and CodeBuffer structures. We are interested in section's CodeBuffer only. // // NOTE: The first section is always '.text', so it's safe to just use 0 index. CodeBuffer& buf = code.getSectionEntry(0)->getBuffer(); // Print the machine-code generated or do something more interesting with it? for (size_t i = 0; i < buf._length; i++) printf("%02X", buf.getData()[i]);
输出结果:
8B4424048B4C24088B54240C0F10010F100A0F58C10F1100E900000000C3
//重定位后 VMemMgr vm; void* p = vm.alloc(code.getCodeSize()); // Allocate a virtual memory (executable). if (!p) return 0; // Handle a possible out-of-memory case. size_t realSize = code.relocate(p); // Relocate & store the output in 'p'.
输出结果:
8b4424048b4c24088b54240c0f10010f100a0f58c10f1100e9e30f3100c3
问题已经解决,和老版本一样用,昨晚搞太晚迷糊了,两个参数填反了
最后于 2019-3-23 14:24
被Wszzy编辑
,原因:
赞赏
看原图
赞赏
雪币:
留言: