Refactor AArch64 instruction

[0xb-s]

This PR refactor the AArch64 instruction emitters by replacing manual [bool; 32] construction with u32encoding.

Changes

• Rewrote emit_ret, emit_ret_x30, emit_br, emit_movk, and emit_movz to use binary field.
• Kept all original function names and output types.
• All instruction encodings match official Armv8-A specifications.
• Used helper functions.

[0xb-s]

@microsoft-github-policy-service agree

[mazong1123]

Thanks @0xb-s for helping this! Could you help to fix the CI failure?

[0xb-s]

Thanks @0xb-s for helping this! Could you help to fix the CI failure?

Done. Need CI approval

[mazong1123]

Thanks! I should find a way to auto approve the CI.

[0xb-s]

Thanks! I should find a way to auto approve the CI.

ci fail on aarch64-unknown-linux-gnu with SIGILL. We're manually encoding valid AArch64 instructions, but QEMU in CI likely doesn't support executing them directly. Any suggestion on how to handle this or maybe omdify the CI setup?

[mazong1123]

Thanks! I should find a way to auto approve the CI.

ci fail on aarch64-unknown-linux-gnu with SIGILL. We're manually encoding valid AArch64 instructions, but QEMU in CI likely doesn't support executing them directly. Any suggestion on how to handle this or maybe omdify the CI setup?

This sounds like a test failure and issue on the machine code written to the memory. Is it passing on your local Linux arm64 environment?

Caused by:
  process didn't exit successfully: `/linux-runner aarch64 /target/aarch64-unknown-linux-gnu/debug/deps/async-d3ca3d9cbabe31eb --nocapture` (signal: 4, SIGILL: illegal instruction)
test test_complex_struct_async_func_with_param_should_success ... 

[0xb-s]

This sounds like a test failure and issue on the machine code written to the memory. Is it passing on your local Linux arm64 environment?

Yes, it's passing on my local Linux arm64 environment.