geode/loader/dobby/source/TrampolineBridge/Trampoline/arm64/trampoline-arm64.cc

#include "platform_macro.h"
#if defined(TARGET_ARCH_ARM64)

#include "dobby_internal.h"

#include "core/modules/assembler/assembler-arm64.h"
#include "core/modules/codegen/codegen-arm64.h"

#include "InstructionRelocation/arm64/ARM64InstructionRelocation.h"

#include "MemoryAllocator/NearMemoryArena.h"
#include "InterceptRouting/RoutingPlugin/RoutingPlugin.h"

using namespace zz::arm64;

CodeBufferBase *GenerateNormalTrampolineBuffer(addr_t from, addr_t to) {
  TurboAssembler turbo_assembler_((void *)from);
#define _ turbo_assembler_.

  uint64_t distance = llabs((int64_t)(from - to));
  uint64_t adrp_range = ((uint64_t)1 << (2 + 19 + 12 - 1));
  if (distance < adrp_range) {
    // adrp, add, br
    _ AdrpAdd(TMP_REG_0, from, to);
    _ br(TMP_REG_0);
    DLOG(0, "Trampoline use [Adrp, Add, Br] combine");
  } else {
    // ldr, br, branch-address
    CodeGen codegen(&turbo_assembler_);
    codegen.LiteralLdrBranch((uint64_t)to);
    DLOG(0, "Trampoline use [Ldr, Br, Label] combine");
  }

  CodeBufferBase *result = NULL;
  result = turbo_assembler_.GetCodeBuffer()->Copy();
  return result;
}

#define ARM64_B_XXX_RANGE ((1 << 25) << 2) // signed

// If BranchType is B_Branch and the branch_range of `B` is not enough, build the transfer to forward the b branch, if
static AssemblyCodeChunk *GenerateFastForwardTrampoline(addr_t source_address, addr_t target_address) {
  AssemblyCodeChunk *cchunk = NULL;

  TurboAssembler turbo_assembler_(0);

  // Use adrp + add branch
  cchunk = NearMemoryArena::AllocateCodeChunk((addr_t)source_address, ARM64_B_XXX_RANGE, 3 * 4);
  if (cchunk == nullptr) {
    ERROR_LOG("Can't found near code chunk");
    return NULL;
  }

  // Use adrp + add branch
  uint64_t distance = llabs((int64_t)((addr_t)cchunk->address - target_address));
  uint64_t adrp_range = ((uint64_t)1 << (2 + 19 + 12 - 1));
  if (distance < adrp_range) { // Use adrp + add branch == (3 * 4) trampoline size
    _ AdrpAdd(TMP_REG_0, (addr_t)cchunk->address, target_address);
    _ br(TMP_REG_0);
    DLOG(0, "Forward Trampoline use [Adrp, Add, Br] combine");
  } else {
    delete cchunk;
    cchunk = NULL;
  }

  // Use absolute branch
  if (cchunk == NULL) {
#if 0
    // Use literal ldr == (4 * 4) trampoline size
  CodeGen codegen(&turbo_assembler_);
  // forward trampoline => target address
  codegen.LiteralLdrBranch(target_address);
  DLOG(0, "Forward Trampoline use [Ldr, Br, Label] combine");
#else
    // Use mov + br == (4 * 5) trampoline size
#define _ turbo_assembler_.
    _ Mov(TMP_REG_0, target_address);
    _ br(TMP_REG_0);
    DLOG(0, "Forward Trampoline use [Mov, Br] combine");
#endif

    size_t tramp_size = turbo_assembler_.GetCodeBuffer()->getSize();
    cchunk = NearMemoryArena::AllocateCodeChunk((addr_t)source_address, ARM64_B_XXX_RANGE, tramp_size);
    if (cchunk == nullptr) {
      ERROR_LOG("Can't found near code chunk");
      return NULL;
    }
  }

  turbo_assembler_.SetRealizedAddress(cchunk->address);

  AssemblyCodeChunk *result = NULL;
  result = AssemblyCodeBuilder::FinalizeFromTurboAssembler(&turbo_assembler_);

  { // release
    delete cchunk;
  }
  return result;
}

CodeBufferBase *GenerateNearTrampolineBuffer(InterceptRouting *routing, addr_t src, addr_t dst) {
  CodeBufferBase *result = NULL;

  TurboAssembler turbo_assembler_((void *)src);
#define _ turbo_assembler_.

  // branch to trampoline_target directly
  if (llabs((long long)dst - (long long)src) < ARM64_B_XXX_RANGE) {
    _ b(dst - src);
  } else {
    AssemblyCodeChunk *fast_forward_trampoline = NULL;
    fast_forward_trampoline = GenerateFastForwardTrampoline(src, dst);
    if (!fast_forward_trampoline)
      return NULL;
    // trampoline => fast_forward_trampoline
    addr_t fast_forward_trampoline_addr = fast_forward_trampoline->raw_instruction_start();
    _ b(fast_forward_trampoline_addr - src);
  }

  // free the original trampoline
  result = turbo_assembler_.GetCodeBuffer()->Copy();
  return result;
}

#endif