Implement the segue optimization for LLVM AOT/JIT (#2230)

Segue is an optimization technology which uses x86 segment register to store
the WebAssembly linear memory base address, so as to remove most of the cost
of SFI (Software-based Fault Isolation) base addition and free up a general
purpose register, by this way it may:
- Improve the performance of JIT/AOT
- Reduce the footprint of JIT/AOT, the JIT/AOT code generated is smaller
- Reduce the compilation time of JIT/AOT

This PR uses the x86-64 GS segment register to apply the optimization, currently
it supports linux and linux-sgx platforms on x86-64 target. By default it is disabled,
developer can use the option below to enable it for wamrc and iwasm(with LLVM
JIT enabled):
```bash
wamrc --enable-segue=[<flags>] -o output_file wasm_file
iwasm --enable-segue=[<flags>] wasm_file [args...]
```
`flags` can be:
    i32.load, i64.load, f32.load, f64.load, v128.load,
    i32.store, i64.store, f32.store, f64.store, v128.store
Use comma to separate them, e.g. `--enable-segue=i32.load,i64.store`,
and `--enable-segue` means all flags are added.

Acknowledgement:
Many thanks to Intel Labs, UC San Diego and UT Austin teams for introducing this
technology and the great support and guidance!

Signed-off-by: Wenyong Huang <wenyong.huang@intel.com>
Co-authored-by: Vahldiek-oberwagner, Anjo Lucas <anjo.lucas.vahldiek-oberwagner@intel.com>
This commit is contained in:
Wenyong Huang
2023-05-26 10:13:33 +08:00
committed by GitHub
parent 27239723a9
commit 76be848ec3
42 changed files with 1864 additions and 123 deletions

View File

@ -16,6 +16,7 @@ WAMR project reused some components from other open source project:
- **asmjit**: for the Fast JIT x86-64 codegen implementation
- **zydis**: for the Fast JIT x86-64 codegen implementation
- **NuttX ELF headers**: used in core/iwasm/aot/debug/elf_parser.c
- **Dhrystone**: for the test benchmakr dhrystone
The WAMR fast interpreter is a clean room development. We would acknowledge the inspirations by [WASM3](https://github.com/wasm3/wasm3) open source project for the approach of pre-calculated oprand stack location.
@ -35,6 +36,7 @@ The WAMR fast interpreter is a clean room development. We would acknowledge the
| asmjit | unspecified | unspecified | https://github.com/asmjit/asmjit | |
| zydis | unspecified | e14a07895136182a5b53e181eec3b1c6e0b434de | https://github.com/zyantific/zydis | |
| NuttX ELF headers | 72313301e23f9c2de969fb64b9a0f67bb4c284df | 10.3.0 | https://github.com/apache/incubator-nuttx | |
| Dhrystone | 2.1 | 2.1 | https://fossies.org/linux/privat/old/ | |
## Licenses
@ -81,15 +83,19 @@ The WAMR fast interpreter is a clean room development. We would acknowledge the
[LICENSE](./tests/wamr-test-suites/spec-test-script/LICENSE)
### libuv
[LICENSE](./core/iwasm/libraries/libc-uvwasi/LICENSE_LIBUV)
### uvwasi
[LICENSE](./core/iwasm/libraries/libc-uvwasi/LICENSE_UVWASI)
### asmjit
[LICENSE](./core/iwasm/fast-jit/cg/LICENSE_ASMJIT)
### zydis
[LICENSE](./core/iwasm/fast-jit/cg/LICENSE_ZYDIS)
### NuttX ELF headers
@ -97,3 +103,7 @@ The WAMR fast interpreter is a clean room development. We would acknowledge the
[LICENSE](./core/iwasm/aot/debug/LICENSE_NUTTX)
[NOTICE](./core/iwasm/aot/debug/NOTICE_NUTTX)
### Dhrystone
[LICENSE](./tests/benchmarks/dhrystone/LICENSE)