Commit Graph

76 Commits

Author SHA1 Message Date
Wenyong Huang
644044522c Merge branch main into dev/gc_refactor 2024-01-12 18:40:47 +08:00
Wenyong Huang
b21f17dd6d
Refine AOT/JIT code call wasm-c-api import process (#2982)
Allow to invoke the quick call entry wasm_runtime_quick_invoke_c_api_import to
call the wasm-c-api import functions to speedup the calling process, which reduces
the data copying.

Use `wamrc --invoke-c-api-import` to generate the optimized AOT code, and set
`jit_options->quick_invoke_c_api_import` true in wasm_engine_new when LLVM JIT
is enabled.
2024-01-10 18:37:02 +08:00
Wenyong Huang
d31455fc4e
Refactor aot stack frame commit (#2976)
- Commit locals, stacks and stack pointer to aot frame only when gc is enabled
- Commit instruction pointer to aot frame when stack frame is enabled
- Refine alloc/free aot frame when gc isn't enabled: use fixed frame size
- Support dump call stack with bytecode offset
2024-01-08 11:18:49 +08:00
Wenyong Huang
cc23a09d5e
Fix test script error and related wamrc and GC issues (#2975)
The wamrc execution errors were skipped in the runtest.py due to `sys.exit(1)`
was caught by upper try-catch sentence and ignored with ret_code 0 returned.

This PR fixed this script issue and fixed the related wamrc/GC issues found.
2024-01-08 09:41:56 +08:00
Wenyong Huang
e6d210a67f Merge branch main into dev/gc_refactor 2024-01-03 12:11:57 +08:00
liang.he
5c3ad0279a
Enable AOT linux perf support (#2930)
And refactor the original perf support
- use WAMR_BUILD_LINUX_PERF as the cmake compilation control
- use WASM_ENABLE_LINUX_PERF as the compiler macro
- use `wamrc --enable-linux-perf` to generate aot file which contains fp operations
- use `iwasm --enable-linux-perf` to create perf map for `perf record`
2024-01-02 15:58:17 +08:00
Daniel Mangum
9779f922b9
Fix boundary check typo in AOT compiler context (#2913)
Fixes typo in docstrings for boundary check in the AOT compiler context.

Signed-off-by: Daniel Mangum <georgedanielmangum@gmail.com>
2023-12-20 09:08:13 +08:00
Wenyong Huang
cf7350f069
Implement aot_alloc_frame/aot_free_frame with LLVM IRs (#2830) 2023-12-12 12:12:02 +08:00
Wenyong Huang
9c71ca6c46 Merge branch main into dev/gc_refactor 2023-11-29 09:23:00 +08:00
Wenyong Huang
238208a6bd
Refine the GC frame ref flag commit (#2822) 2023-11-27 18:17:35 +08:00
liang.he
8aa813f44a
Generate jitdump to support linux perf for LLVM JIT (#2788) 2023-11-27 15:42:00 +08:00
Xu Jun
c4010b0d3a
Fix GC AOT on 32bit target (#2814) 2023-11-23 18:48:19 +08:00
Wenyong Huang
5d9cbfe109
Implement GC AOT object reclaim process (#2762) 2023-11-22 14:39:07 +08:00
Huang Qi
f92630108c
Merge dupplicated compile options for wamrc (#2748) 2023-11-10 19:37:38 +08:00
Wenyong Huang
64e40c13a9
Fix more GC AOT/JIT issues (#2727)
Fix opcode translation for br_on_null/br_on_non_null/br_on_cast/br_on_cast_fail
Fix global data size/offset calculation for 32-bit/64-bit targets
Fix issues in AOT file emitting and AOT loader, refine AOT file format
Fix invalid table element address used for table.get/table.set
Fix invalid struct field offset used for struct.get/struct.set
Fix aot stack frame commit for function call/call_indirect/call_ref
Add GC AOT/JIT to CI test
2023-11-09 08:52:56 +08:00
Wenyong Huang
3a31e517a4
Fix GC AOT issues (#2640) 2023-10-16 13:30:56 +08:00
Wenyong Huang
ae508e25fa Merge branch dev/aot_stack_frame into dev/gc_refactor 2023-10-09 14:21:39 +08:00
Wenyong Huang
85869ed604 Merge branch main into dev/gc_refactor 2023-10-08 10:53:13 +08:00
Wenyong Huang
7a9ed07411 Merge branch main into dev/aot_stack_frame 2023-10-08 10:01:45 +08:00
Alfred E. Neumayer
99b47fd334
Support AOT compiler with LLVM 17 (#2567)
Adapt API usage to new interfaces where applicable, including LLVM function
usage, obsoleted llvm::Optional type and removal of unavailable headers.

Know issues:
- AOT static PGO isn't enabled
- LLVM JIT may run failed due to llvm_orc_registerEHFrameSectionWrapper
  isn't linked into iwasm
2023-09-25 19:00:46 +08:00
Wenyong Huang
33ac031a3b
Enhance LLVM AOT/JIT stack frame dump (#2350)
Implement a full LLVM AOT/JIT stack frame dump:
commit the function arguments, locals, stack operands from LLVM values to the stack frame,
which is required by the GC AOT/JIT feature, and may be required by the AOT debugger,
AOT snapshot and other features.

Refer to:
https://github.com/bytecodealliance/wasm-micro-runtime/issues/2144
https://github.com/bytecodealliance/wasm-micro-runtime/issues/2333
https://github.com/bytecodealliance/wasm-micro-runtime/issues/2506
2023-08-28 09:45:02 +08:00
TianlongLiang
7d86111a53
Refactor the LLVM IR translation of the existing opcodes for GC AOT (#2376)
Refactor the LLVM IR translation of below opcodes:
  CALL_INDIRECT, SELECT_T, TABLE_GET, TABLE_SET,
  REF_NULL, REF_IS_NULL, REF_FUNC,
  TABLE_INIT, TABLE_GROW, TABLE_FILL,
  GET_LOCAL, SET_LOCAL, TEE_LOCAL

And refactor aot_drop_table_seg and add aot_create_func_obj.
2023-08-23 16:04:10 +08:00
Wenyong Huang
803597dc55 Merge branch main into dev/gc_refactor 2023-08-07 09:35:10 +08:00
Huang Qi
aafea39b8c
Add "--enable-builtin-intrinsics=<flags>" option to wamrc (#2341)
Refer to doc/xip.md for details.
2023-07-06 18:20:35 +08:00
Huang Qi
44f4b4f062
Add "--enable-llvm-passes=<passes>" option to wamrc (#2335)
Add "--enable-llvm-passes=<passes>" option to wamrc for customizing LLVM passes
2023-07-04 12:20:52 +08:00
Wenyong Huang
76faf89029 Merge branch main into dev/gc_refactor 2023-06-27 17:03:41 +08:00
YAMAMOTO Takashi
cd7941cc39
AOT/JIT native stack bound check improvement (#2244)
Move the native stack overflow check from the caller to the callee because the
former doesn't work for call_indirect and imported functions.

Make the stack usage estimation more accurate. Instead of making a guess from
the number of wasm locals in the function, use the LLVM's idea of the stack size
of each MachineFunction. The former is inaccurate because a) it doesn't reflect
optimization passes, and b) wasm locals are not the only reason to use stack.

To use the post-compilation stack usage information without requiring 2-pass
compilation or machine-code imm rewriting, introduce a global array to store
stack consumption of each functions:
For JIT, use a custom IRCompiler with an extra pass to fill the array.
For AOT, use `clang -fstack-usage` equivalent because we support external llc.

Re-implement function call stack usage estimation to reflect the real calling
conventions better. (aot_estimate_stack_usage_for_function_call)

Re-implement stack estimation logic (--enable-memory-profiling) based on the new
machinery.

Discussions: #2105.
2023-06-22 07:27:07 +08:00
YAMAMOTO Takashi
92e073b8ce
AOTFuncContext: Remove a stale comment (#2283) 2023-06-09 22:31:08 +08:00
Wenyong Huang
366b3b7683 Merge branch main into dev/gc_refactor 2023-06-05 14:13:45 +08:00
Wenyong Huang
8d88471c46
Implement AOT static PGO (#2243)
LLVM PGO (Profile-Guided Optimization) allows the compiler to better optimize code
for how it actually runs. This PR implements the AOT static PGO, and is tested on
Linux x86-64 and x86-32. The basic steps are:

1. Use `wamrc --enable-llvm-pgo -o <aot_file_of_pgo> <wasm_file>`
   to generate an instrumented aot file.
2. Compile iwasm with `cmake -DWAMR_BUILD_STATIC_PGO=1` and run
      `iwasm --gen-prof-file=<raw_profile_file> <aot_file_of_pgo>`
    to generate the raw profile file.
3. Run `llvm-profdata merge -output=<profile_file> <raw_profile_file>`
    to merge the raw profile file into the profile file.
4. Run `wamrc --use-prof-file=<profile_file> -o <aot_file> <wasm_file>`
    to generate the optimized aot file.
5. Run the optimized aot_file: `iwasm <aot_file>`.

The test scripts are also added for each benchmark, run `test_pgo.sh` under
each benchmark's folder to test the AOT static pgo.
2023-06-05 09:17:39 +08:00
Wenyong Huang
76be848ec3
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>
2023-05-26 10:13:33 +08:00
YAMAMOTO Takashi
670567f8b3
core/iwasm/compilation: constify a bit (#2223)
Just to make the code a bit easier to read.
2023-05-20 11:55:02 +08:00
Huang Qi
035cca062f
Add --enable-gc option to wamrc (#2190)
The runtime instance memory layout changed when GC was enabled. With this patch
the GC is enabled for wamrc, but it keeps the compatibility with iwasm no matter GC
is enabled for it or not.

It may waste some memory for iwasm without GC support since the GC relative fields
for the table instance are always here, let's optimization it after AOT fully supports GC.
2023-05-15 14:32:57 +08:00
Wenyong Huang
8704532675
Merge pull request #2191 from bytecodealliance/main
Merge branch main into dev/gc_refactor
2023-05-09 12:20:20 +08:00
Huang Qi
02f0156475
GC: Change table_elem_type_t to pointer width to simplify the related handlings (#2178) 2023-05-09 11:10:40 +08:00
YAMAMOTO Takashi
2b896c80ef
wamrc: Add --stack-usage option (#2158) 2023-04-28 13:56:44 +08:00
YAMAMOTO Takashi
7d3b2a8773
Make memory profiling show native stack usage (#1917) 2023-02-01 11:52:15 +08:00
Wenyong Huang
7fd37190e8
Add control for the native stack check with hardware trap (#1682)
Add a new options to control the native stack hw bound check feature:
- Besides the original option `cmake -DWAMR_DISABLE_HW_BOUND_CHECK=1/0`,
  add a new option `cmake -DWAMR_DISABLE_STACK_HW_BOUND_CHECK=1/0`
- When the linear memory hw bound check is disabled, the stack hw bound check
   will be disabled automatically, no matter what the input option is
- When the linear memory hw bound check is enabled, the stack hw bound check
  is enabled/disabled according to the value of input option
- Besides the original option `--bounds-checks=1/0`, add a new option
  `--stack-bounds-checks=1/0` for wamrc

Refer to: https://github.com/bytecodealliance/wasm-micro-runtime/issues/1677
2022-11-07 18:26:33 +08:00
Wenyong Huang
5b144c491d
Avoid initialize LLVM repeatedly (#1671)
Currently we initialize and destroy LLVM environment in aot_create_comp_context
and aot_destroy_comp_context, which are called in wasm_module_load/unload,
and the latter may be invoked multiple times, which leads to duplicated LLVM
initialization/destroy and may result in unexpected behaviors.

Move the LLVM init/destroy into runtime init/destroy to resolve the issue.
2022-11-02 16:13:58 +08:00
Wenyong Huang
e87a554616
Refactor LLVM JIT (#1613)
Refactor LLVM JIT for some purposes:
- To simplify the source code of JIT compilation
- To simplify the JIT modes
- To align with LLVM latest changes
- To prepare for the Multi-tier JIT compilation, refer to #1302

The changes mainly include:
- Remove the MCJIT mode, replace it with ORC JIT eager mode
- Remove the LLVM legacy pass manager (only keep the LLVM new pass manager)
- Change the lazy mode's LLVM module/function binding:
  change each function in an individual LLVM module into all functions in a single LLVM module
- Upgraded ORC JIT to ORCv2 JIT to enable lazy compilation

Refer to #1468
2022-10-18 20:17:34 +08:00
Petr Penzin
c075844001
Add our own legacy loop unswitch pass (#1339)
Since legacy binding for loop unswitch pass was removed and we can't get
it back. Implement its equivalent in `aot_llvm_extra.cpp` and use it in
`aot_compiler.c`.

Follow up to #1183.
2022-07-29 10:29:02 +08:00
Wenyong Huang
5e238322c2
Enable aot compiler with llvm-14/15 (#1252)
Enable aot compiler and jit based on llvm-14.0 and llvm-15.0git,
replace LLVMBuildLoad/LLVMBuildInBoundsGEP/LLVMBuildCall with
LLVMBuildLoad2/LLVMBuildInBoundsGEP2/LLVMBuildCall2, and pass
them with related types, so as to meet the requirements of opaque
pointers.

And fix several compilation errors for llvm-14.0/15.0git.

Most spec cases and standalone cases are tested.
2022-06-28 14:53:01 +08:00
Xu Jun
53b775aa4b
Support integrate 3rd-party toolchains into wamrc (#1237)
Support integrating 3rd-party toolchain llc compiler or asm compiler
into wamrc by setting environment variable WAMRC_LLC_COMPILER
or WAMRC_ASM_COMPILER, wamrc will use these tools to generate
object file from LLVM IR firstly, and then refactor the object file into
aot file.
2022-06-20 13:13:41 +08:00
Xu Jun
77595c9560
Support emit specified custom sections into AoT file (#1207)
And add API to get the content of custom section with
section name for both wasm file and aot file.
2022-06-10 21:51:13 +08:00
Wenyong Huang
5631a2aa18
Use LLVM new pass manager for wamrc (#978)
Use LLVM new pass manager for wamrc to replace the legacy pass manger,
so as to gain better performance and reduce the compilation time.
Reference links:
- https://llvm.org/docs/NewPassManager.html
- https://blog.llvm.org/posts/2021-03-26-the-new-pass-manager

And add an option to use the legacy pm mode when building wamrc:
cmake .. -DWAMR_BUILD_LLVM_LEGACY_PM=1

For JIT mode, keep it unchanged as it only runs several function passes and
using new pass manager will increase the compilation time.

And refactor the codes of applying LLVM passes.
2022-01-24 11:10:37 +08:00
Wenyong Huang
7636d86a76
Refactor Orc JIT to enable lazy compilation (#974)
Refactor LLVM Orc JIT to actually enable the lazy compilation and speedup
the launching process:
  https://llvm.org/docs/ORCv2.html#laziness

Main modifications:
- Create LLVM module for each wasm function, wrap it with thread safe module
  so that the modules can be compiled parallelly
- Lookup function from aot module instance's func_ptrs but not directly call the
  function to decouple the module relationship
- Compile the function when it is first called and hasn't been compiled
- Create threads to pre-compile the WASM functions parallelly when loading
- Set Lazy JIT as default, update document and build/test scripts
2022-01-20 18:40:13 +08:00
Wenyong Huang
5be427bfa2
xip: Lookup float constants from table to reduce relocations (#894)
Lookup float/double constants from exec_env->native_symbol table
but not construct them with LLVMBuildConst if XIP mode is enabled,
these constants are introduced by f32/f64.const opcodes and some
float/double conversion opcodes, and make wamrc generate some
relocations in text section of AOT XIP file. This patch eliminates such
relocations when "--enable-indirect-mode" is added to wamrc.
2021-12-16 21:39:23 +08:00
Huang Qi
8d1c56bda4
Implement atomic and memset/memmove/memcpy intrinsic for riscv (#841)
- Lower aotmic instruction to non-atomic form on some platforms
- Lower memset/memmove/memcpy for XIP
- Disable rtti in cmake
2021-11-30 17:10:14 +08:00
Wenyong Huang
a1ad950ae1
Enable LLVM link time optimization (LTO) for AOT (#831)
Enable LLVM link time optimization for AOT and enable it by default,
and provide "wamrc --disable-llvm-lto" option to disable it.
And modify the spec test script, disable it by default when testing
spec cases.

Signed-off-by: Wenyong Huang <wenyong.huang@intel.com>
2021-11-13 16:59:35 +08:00
Wenyong Huang
fb4afc7ca4
Apply clang-format for core/iwasm compilation and libraries (#784)
Apply clang-format for core/iwasm/compilation and core/iwasm/libraries files.
Add wasm-c-api empty_imports sample to workflow test.
And enable triggering workflow when core/config.h changes.
2021-10-13 15:13:00 +08:00