Commit Graph

19 Commits

Author SHA1 Message Date
Wenyong Huang
e8ce4c542e Implement Multi-tier JIT (#1774)
Implement 2-level Multi-tier JIT engine: tier-up from Fast JIT to LLVM JIT to
get quick cold startup by Fast JIT and better performance by gradually
switching to LLVM JIT when the LLVM JIT functions are compiled by the
backend threads.

Refer to:
https://github.com/bytecodealliance/wasm-micro-runtime/issues/1302
2022-12-20 14:37:07 +08:00
Wenyong Huang
97fda83c94 Publish the docker image with tar/zip files (#1808)
And fix issue found in fast jit call indirect.
2022-12-20 14:37:07 +08:00
liang.he
216b2cb540
Fix a typo in workflow (#1760) 2022-11-30 14:00:30 +08:00
Wenyong Huang
87c3195d47
Revert "Implement call Fast JIT function from LLVM JIT jitted code" (#1737)
Reverts bytecodealliance/wasm-micro-runtime#1714, which was merged mistakenly.
2022-11-22 14:04:48 +08:00
Wenyong Huang
440bbea81e
Fix interp/fast-jit float min/max issues (#1733) 2022-11-22 09:14:20 +08:00
Wenyong Huang
89f9d695ed
Fix fast jit issue of translating opcode i32.rem_s/i64.rem_s (#1731) 2022-11-21 16:37:10 +08:00
Wenyong Huang
cf7b01ad82
Implement call Fast JIT function from LLVM JIT jitted code (#1714)
Basically implement the Multi-tier JIT engine.
And update document and wamr-test-suites script.
2022-11-21 10:42:18 +08:00
Wenyong Huang
ef21f0c951
Implement Fast JIT dump call stack and perf profiling (#1633)
Implement dump call stack and perf profiling features for Fast JIT,
and refine some code.
2022-10-27 09:28:32 +08:00
Wenyong Huang
bbfa39c32c
Refine Fast JIT accessing memory/table instance and global data (#1623)
Some offsets can be directly gotten at the compilation stage after the interp/AOT
module instance refactoring PR was merged, so as to reduce some unnecessary
load instructions and improve the Fast JIT performance:
- Access fields of wasm memory instance structure
- Access fields of wasm table instance structure
- Access the global data
2022-10-21 09:32:51 +08:00
Wenyong Huang
1d4cbfceac
Refine Fast JIT call indirect and call native process (#1620)
Translate call_indirect opcode by calling wasm functions with Fast JIT IRs instead of
calling jit_call_indirect runtime API, so as to improve the performance.

Translate call native function process with Fast JIT IRs to validate each pointer argument
and convert it into native address, and then call the native function directly instead
of calling jit_invoke_native runtime API, so as to improve the performance.
2022-10-19 17:11:38 +08:00
Wenyong Huang
a182926a73
Refactor interpreter/AOT module instance layout (#1559)
Refactor the layout of interpreter and AOT module instance:
- Unify the interp/AOT module instance, use the same WASMModuleInstance/
  WASMMemoryInstance/WASMTableInstance data structures for both interpreter
  and AOT
- Make the offset of most fields the same in module instance for both interpreter
  and AOT, append memory instance structure, global data and table instances to
  the end of module instance for interpreter mode (like AOT mode)
- For extra fields in WASM module instance, use WASMModuleInstanceExtra to
  create a field `e` for interpreter
- Change the LLVM JIT module instance creating process, LLVM JIT uses the WASM
  module and module instance same as interpreter/Fast-JIT mode. So that Fast JIT
  and LLVM JIT can access the same data structures, and make it possible to
  implement the Multi-tier JIT (tier-up from Fast JIT to LLVM JIT) in the future
- Unify some APIs: merge some APIs for module instance and memory instance's
  related operations (only implement one copy)

Note that the AOT ABI is same, the AOT file format, AOT relocation types, how AOT
code accesses the AOT module instance and so on are kept unchanged.

Refer to:
https://github.com/bytecodealliance/wasm-micro-runtime/issues/1384
2022-10-18 10:59:28 +08:00
Xu Jun
826cf4f8e1
Fix threads spec test issues (#1586) 2022-10-13 13:53:09 +08:00
Qiang
32d2d16908
Fix Fast JIT issues reported by instrument test (#1488)
- Add checks for `pack_argv`
- Fix the checks in creating cc entry/exit basic blocks
2022-09-22 19:37:59 +08:00
Qiang
da79e3e9b2
Add more checks for Fast JIT gen insn and jit_lock_reg_in_insn (#1449)
Add more checks for Fast JIT generating insn and jit_lock_reg_in_insn
in compile_int_div_no_check and other places to avoid accessing
NULL insn.
2022-09-07 16:52:35 +08:00
Qiang
18774805cc
Add more checks for Fast JIT (#1433)
Add more checks for Fast JIT to fix the issues reported by instrument test:
- add check for the jit_value before pushing it into the stack
- add check at the end of form_and_translate_func
- add checks after each jit pass
2022-09-01 15:31:00 +08:00
Wenyong Huang
903a27aff2
Fix fast jit destroy block issue (#1412)
Only destroy the block if it hasn't been pushed into the block stack,
or it will be destroyed again when destroying the block stack.
2022-08-24 22:00:16 +08:00
Qiang
1dcc59f234
Fix fast jit destroy entry/exit basic block issue in jit_cc_destroy (#1401)
The entry and exit basic blocks might be destroyed before they are created.
Found by instrument test. Add checks to fix the issue.
2022-08-24 17:56:38 +08:00
Wenyong Huang
08eeeb748c
Add more fast jit fixed virtual regs (#1409)
Add fast jit fixed virtual register import_func_ptrs_reg and func_type_indexes_reg
for the preparation of further optimizations.
2022-08-24 14:05:51 +08:00
Wenyong Huang
bf28030993
Import WAMR Fast JIT (#1343)
Import WAMR Fast JIT which is a lightweight JIT with quick startup, small footprint,
relatively good performance (~40% to ~50% of LLVM JIT) and good portability.

Platforms supported: Linux, MacOS and Linux SGX.
Arch supported: x86-64.
2022-08-02 16:03:50 +08:00