wasm-micro-runtime/core/iwasm/compilation/aot_emit_function.c
liang.he 3ce549948d
Merge main to dev zephyr (#4384)
* Optimize memory initialization handling in AOT loader (#3983)

Save memory if the file buffer is always exist before exit.

Signed-off-by: Huang Qi <huangqi3@xiaomi.com>

* Break aot_create_comp_data into small functions

Signed-off-by: Huang Qi <huangqi3@xiaomi.com>

* Handle a new scenario where an item is both exported and imported. (#3984)

* Error message improvement (#4000)

Improve error message in the scenario where the runtime was built
with ref types disabled but the module uses reference types feature.

* Ensure __heap_base and __data_end global indices are validated against import count (#3996)

* Fix table index calculations in wasm_loader and wasm_mini_loader (#4004)

* Add an example of how to embed WAMR in Zephyr user mode  (#3998)

* [fuzzing] Use software bound-check during fuzzing (#4003)

* Update CMakeLists.txt of fuzzing

- enable software bound-check
- enable wasi
- disable libc builtin and multiple modules

* Fix off-by-one error in result offset calculation for function calls

* Check whether related table has funcref elem in opcode call_indirect (#3999)

* check whether table has funcref elem in call_indirect
* check whether table has funcref elem in call_indirect when gc is enabled

* Improve stack consistency by ensuring sufficient space for dummy offsets (#4011)

One more corner case: if the `frame_offset` increases and becomes equal to
the `frame_offset_boundary` after the last assignment within the for loop.

* Add documentation regarding security issues and the status of Wasm proposals (#3972)

Add documentation regarding security issues and the status of Wasm proposals.

* Enable shrunk memory by default and add related configurations (#4008)

- Enable shrunk memory by default and add related configurations
- Improve error messages for memory access alignment checks
- Add documentation for WAMR shrunk memory build option
- Update NuttX workflow to disable shrunk memory build option

* build(deps): Bump actions/upload-artifact from 4.5.0 to 4.6.0 (#4021)

Bumps [actions/upload-artifact](https://github.com/actions/upload-artifact) from 4.5.0 to 4.6.0.
- [Release notes](https://github.com/actions/upload-artifact/releases)
- [Commits](https://github.com/actions/upload-artifact/compare/v4.5.0...v4.6.0)

---
updated-dependencies:
- dependency-name: actions/upload-artifact
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* build(deps): Bump github/codeql-action from 3.28.0 to 3.28.1 (#4020)

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.28.0 to 3.28.1.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Commits](https://github.com/github/codeql-action/compare/v3.28.0...v3.28.1)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Refine read leb int wasm loader of fast interpreter (#4017)

* .github: Add shared lib builds (#3975)

So far, no workflows would attempt to build the shared version of the
iwasm library (namely, vmlib).

Note that, as opposed to GC_EH_BUILD_OPTIONS and DEFAULT_BUILD_OPTIONS,
the actual default options defined by the build system are assumed, for
the sake of simplicity and avoiding repeated code.

* fixes for compiling on windows (#4026)

* Refine getting const offsets in wasm loader of fast-interp (#4012)

- Refine const offsets in loader for fast-interp
- handle const cell num overflow
- Use const array, remove list

* Synchronize the GC spec tests to the commit from December 9. 2024. (#4022)

- Synchronize the GC spec tests to the commit from December 9. 2024.
- Revise the error messages to be consistent with the spec test cases.
- bypass gc spec test on the nuttx platform as a workaround

* Fix wasm loader check data segment count (#4039)

correctly report error when datacount section has non-zero data segment count while the data section is not present

* Update Rust target from 'wasm32-wasi' to 'wasm32-wasip1' in CI (#4050)

- update Rust target from 'wasm32-wasi' to 'wasm32-wasip1' in ci

* build(deps): Bump github/codeql-action from 3.28.1 to 3.28.5

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.28.1 to 3.28.5.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Commits](https://github.com/github/codeql-action/compare/v3.28.1...v3.28.5)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>

* build(deps): Bump github/codeql-action from 3.28.5 to 3.28.8

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.28.5 to 3.28.8.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Commits](https://github.com/github/codeql-action/compare/v3.28.5...v3.28.8)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>

* Use wasm32-wasip1 instead of wasm32-wasi target for rust code (#4057)

Rust compiler previously deprecated, and now removed the wasm32-wasi target and replaced it with wasm32-wasip1. This
change updates all the occurrences of wasm32-wasi in the context of Rust compilation.

covers the wasi-nn/test.

* add a validator for aot module (#3995)

- Add AOT module validation to ensure memory constraints are met
- Enable AOT validator in build configuration and update related source files

* Show wasm proposals status during compilation and execution (#3989)

- add default build configuration options and enhance message output for WAMR features
- Add Wasm proposal status printing functionality

* initial

* Add versioning support and update CMake configuration

* Add versioning information for libraries and executables across multiple platforms

* Refactor versioning documentation and adopt semantic versioning guidelines

* Remove deprecated version.h file and update versioning documentation

* Add version.h and update versioning documentation for embedded platforms

* Add workflow to confirm version.h is in sync and integrate it into Android compilation workflow

* Cleanup check_version_h workflow by removing unnecessary outputs and permissions

* Update memory allocation functions to use allocator user data (#4043)

* [fuzzing] execute every exported function (#3959)

- Enhance wasm mutator fuzz tests by adding export function execution and random value generation
- Use --fuel to limit  loop size
- Use predefined values and enhance argument logging in execution

* In wasm32, fix potential conversion overflow when enlarging 65536 pages (#4064)

fix enlarge 65536 pages conversion overflow in wasm32

* fix(aot): ensure value_cmp does not exceed br_count in branch table compilation (#4065)

* build(deps): Bump github/codeql-action from 3.28.8 to 3.28.9 (#4074)

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.28.8 to 3.28.9.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Commits](https://github.com/github/codeql-action/compare/v3.28.8...v3.28.9)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Unit test:type matching issue and code redundancy (#4079)

* Add a conditional check for the macro __STDC_VERSION__ (#4080)

* build_llvm.py: Allow to build xtensa target on non-xtensa host

Signed-off-by: Huang Qi <huangqi3@xiaomi.com>

* [gc] Subtyping fix (#4075)

* fix(build_llvm.py): clean up whitespace and formatting in build script

Signed-off-by: Huang Qi <huangqi3@xiaomi.com>

* feat: add support for EXTERNREF value type and enable AOT validator in fuzz tests (#4083)

* fix(unit-test): libc_builtin_test issues (#4073)

- uninitialized buffer pointers (crashes)
- match integer constant size with printf specifier

Signed-off-by: Peter Tatrai <peter.tatrai.ext@siemens.com>

* fix(build_llvm_libraries.yml): Correct script path for build_llvm.py

Signed-off-by: Huang Qi <huangqi3@xiaomi.com>

* fix(aot_emit_aot_file): prevent buffer emission for zero byte_count (#4095)

if using a debug building of wamrc to run spec test. there will be:
core/iwasm/compilation/aot_emit_aot_file.c:1794:13: runtime error: null pointer passed as argument 2, which is declared to never be null

* Cmake improvements (#4076)

- Utilizes the standard CMake variable BUILD_SHARED_LIBS to simplify the CMake configuration.
- Allows the use of a single library definition for both static and
  shared library cases, improving maintainability and readability of the CMake configuration.
- Install vmlib public header files
- Installs the public header files for the vmlib target to the include/iwasm directory.
- Install cmake package
- Adds the necessary CMake configuration files (iwasmConfig.cmake and iwasmConfigVersion.cmake).
- Configures the installation of these files to the appropriate directory (lib/cmake/iwasm).
- Ensures compatibility with the same major version.
- Improve windows product-mini CMakeLists.txt
- Fix missing symbols when linking windows product-mini with shared vmlib
- Improve Darwin product-mini CMakeLists.txt

---------

Signed-off-by: Peter Tatrai <peter.tatrai.ext@siemens.com>

* fix: when load aot init expr,no type_idx set. (#4094)

Fix an assertion from *gc_object.c line 91*  `bh_assert(rtt_type->type_flag == WASM_TYPE_STRUCT;`

* prevent data overflow on 32 bit platform for memory.grow

* cr suggestions

* cr suggestions

* format

* cr suggestions

* feat: use C linkage in aot_comp_option.h for C++ embeding (#4106)

Co-authored-by: xiangjia.xj <xiangjia.xj@alibaba-inc.com>

* build(deps): Bump actions/upload-artifact from 4.6.0 to 4.6.1

Bumps [actions/upload-artifact](https://github.com/actions/upload-artifact) from 4.6.0 to 4.6.1.
- [Release notes](https://github.com/actions/upload-artifact/releases)
- [Commits](https://github.com/actions/upload-artifact/compare/v4.6.0...v4.6.1)

---
updated-dependencies:
- dependency-name: actions/upload-artifact
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>

* build(deps): Bump github/codeql-action from 3.28.9 to 3.28.10

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.28.9 to 3.28.10.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Commits](https://github.com/github/codeql-action/compare/v3.28.9...v3.28.10)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>

* Apply suggestions from code review

remove confusing comments.

* build(deps): Bump ossf/scorecard-action from 2.4.0 to 2.4.1

Bumps [ossf/scorecard-action](https://github.com/ossf/scorecard-action) from 2.4.0 to 2.4.1.
- [Release notes](https://github.com/ossf/scorecard-action/releases)
- [Changelog](https://github.com/ossf/scorecard-action/blob/main/RELEASE.md)
- [Commits](62b2cac7ed...f49aabe0b5)

---
updated-dependencies:
- dependency-name: ossf/scorecard-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>

* fix: add dispose of the debug information builder when destroying compilation context (#4105)

Co-authored-by: xiangjia.xj <xiangjia.xj@alibaba-inc.com>

* wasm_loader allocates more spaces for elements (#4099)

- allocate memory for array initialization based on length
- update reference type mapping for struct initialization

* log warning instaed of assertion (#4119)

* fix: fix load aarch64 aot failed (#4114)

Co-authored-by: xiangjia.xj <xiangjia.xj@alibaba-inc.com>

* fix: correct typos and improve comments across multiple files by codespell (#4116)

Signed-off-by: Huang Qi <huangqi3@xiaomi.com>

* avoid Windows perform newline translation (#4128)

* Iterate callstack API

* wamr bool type

* clang-format

* meaning of the return bool type in the callback

* keep devs notes out of public API

* format

* support standard frames as well

* format

* Calculate func_index instead of adding an extra field to wasm frame

* ignore frames with no function

* update typo in the comment

* update signature

* add correct frame size for aot standard frames

* standard frame is not supported when GC is enabled

* Copy read only API behind a flag instead of using user defined callback

* Cleaning up

* remove unnecessary includes

* formatting

* define if not defined

* formatting

* address comments

* formatting

* remove spare diff line

* address comments

* clang format

* spare line

* spare lines

* last fixes

* identation

* fix bug for return value when skip_n is passed

* build(deps): Bump github/codeql-action from 3.28.10 to 3.28.11

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.28.10 to 3.28.11.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Commits](https://github.com/github/codeql-action/compare/v3.28.10...v3.28.11)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>

* Expose WAMR_BUILD_GC_HEAP_SIZE_DEFAULT as a CMake option

This is wired through to the GC_HEAP_SIZE_DEFAULT constant.

Also honor this value when configuring the engine with the
wasm_c_api.

* Address code review feedback

* Move the default heap size initialization

* Restore the doc heading.

* Fix iwasm build error when WAMR_BUILD_WASI_NN enabled

A recent change on ./product-mini/platforms/linux/CMakeLists.txt renamed
libiwasm to vmlib, but wasi-nn.cmake still wants to link libiwasm.so.
Replace libiwasm with vmlib in wasi-nn.cmake to resolve iwasm build error
when WAMR_BUILD_WASI_NN enabled.

* include bh_platform.h (#4135)

This should guarantee that the various macros required by
wasm_proposal.c are defined even if the build system does not supply
them to the compiler command.

* Merge dev/simd for fast-interp (#4131)

* Implement the first few SIMD opcodes for fast interpreter (v128.const, v128.any_true) (#3818)

Tested on the following code:
```
(module
  (import "wasi_snapshot_preview1" "proc_exit" (func $proc_exit (param i32)))
  (memory (export "memory") 1)

  ;; WASI entry point
  (func $main (export "_start")
    v128.const i8x16 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
    v128.any_true
    if
      unreachable
    end
    
    v128.const i8x16 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15     
    v128.any_true
    i32.const 0
    i32.eq
    if
      unreachable
    end

    i32.const 0
    call $proc_exit
  )
)
```

* implement POP_V128()

This is to simplify the simd implementation for fast interpreter

* Add all SIMD operations into wasm_interp_fast switch

* Add V128 comparison operations

Tested using
```
(module
  (import "wasi_snapshot_preview1" "proc_exit" (func $proc_exit (param i32)))

  (memory (export "memory") 1)

  (func $assert_true (param v128)
    local.get 0
    v128.any_true
    i32.eqz
    if
      unreachable
    end
  )

  (func $main (export "_start")
    ;; Test v128.not
    v128.const i8x16 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
    v128.not
    v128.const i8x16 255 255 255 255 255 255 255 255 255 255 255 255 255 255 255 255
    i8x16.eq
    call $assert_true

    ;; Test v128.and
    v128.const i8x16 255 255 255 255 0 0 0 0 255 255 255 255 0 0 0 0
    v128.const i8x16 255 255 0 0 255 255 0 0 255 255 0 0 255 255 0 0
    v128.and
    v128.const i8x16 255 255 0 0 0 0 0 0 255 255 0 0 0 0 0 0
    i8x16.eq
    call $assert_true

    ;; Test v128.andnot
    v128.const i8x16 255 255 255 255 0 0 0 0 255 255 255 255 0 0 0 0
    v128.const i8x16 255 255 0 0 255 255 0 0 255 255 0 0 255 255 0 0
    v128.andnot
    v128.const i8x16 0 0 255 255 0 0 0 0 0 0 255 255 0 0 0 0
    i8x16.eq
    call $assert_true

    ;; Test v128.or
    v128.const i8x16 255 255 0 0 0 0 255 255 255 255 0 0 0 0 255 0
    v128.const i8x16 0 0 255 255 255 255 0 0 0 0 255 255 255 255 0 0
    v128.or
    v128.const i8x16 255 255 255 255 255 255 255 255 255 255 255 255 255 255 255 0
    i8x16.eq
    call $assert_true

    ;; Test v128.xor
    v128.const i8x16 255 255 0 0 255 255 0 0 255 255 0 0 255 255 0 0
    v128.const i8x16 255 255 255 255 0 0 0 0 255 255 255 255 0 0 0 0
    v128.xor
    v128.const i8x16 0 0 255 255 255 255 0 0 0 0 255 255 255 255 0 0
    i8x16.eq
    call $assert_true

    i32.const 0
    call $proc_exit
  )
)
```

* Add first NEON SIMD opcode implementations to fast interpreter (#3859)

Add some implementations of SIMD opcodes using NEON instructions.
Tested using:
```wast
(module
  (import "wasi_snapshot_preview1" "proc_exit" (func $proc_exit (param i32)))
  (memory (export "memory") 1)

  (func $assert_true (param v128)
    local.get 0
    v128.any_true 
    i32.eqz
    if
      unreachable
    end
  )
  (func $main (export "_start")
    i32.const 0
    i32.const 32
    memory.grow
    drop

    i32.const 0
    v128.const i8x16 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
    v128.store

    i32.const 0
    v128.load

    v128.const i8x16 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
    i8x16.eq
    call $assert_true

    i32.const 16
    v128.const i8x16 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
    v128.store

    i32.const 16
    v128.load
    v128.const i8x16 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
    i8x16.eq
    call $assert_true

    i32.const 0
    v128.load
    v128.const i8x16 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
    i8x16.eq
    call $assert_true
    drop

    i32.const 0
    i32.const 1
    memory.grow
    drop

    i32.const 0
    i64.const 0x7F80FF017E02FE80
    i64.store

    i32.const 0
    v128.load8x8_s

    v128.const i16x8 127 -128 -1 1 126 2 -2 -128

    i16x8.eq
    call $assert_true

    i32.const 0
    i64.const 0x80FE027E01FF807F
    i64.store

    i32.const 0
    v128.load8x8_u

    v128.const i16x8 128 254 2 126 1 255 128 127

    i16x8.eq
    call $assert_true

    i32.const 0
    i64.const 0x8000FFFE7FFF0001
    i64.store

    i32.const 0
    v128.load16x4_s

    v128.const i32x4 -32768 -2 32767 1

    i32x4.eq
    call $assert_true

    i32.const 0
    i64.const 0x8000FFFE7FFF0001 
    i64.store

    i32.const 0
    v128.load16x4_u

    v128.const i32x4 32768 65534 32767 1   

    i32x4.eq
    call $assert_true

    i32.const 0
    i64.const 0x8000000000000001
    i64.store

    i32.const 0
    v128.load32x2_s

    v128.const i64x2 -2147483648 1 

    i64x2.eq
    call $assert_true

    i32.const 0
    i64.const 0x8000000000000001
    i64.store

    i32.const 0
    v128.load32x2_u

    v128.const i64x2 2147483648 1

    i64x2.eq
    call $assert_true

    call $proc_exit
  )
)
```

* Emit imm for lane extract and replace (#3906)

* Fix replacement value not being correct (#3919)

* Implement load lanes opcodes for wasm (#3942)

* Implement final SIMD opcodes: store lane (#4001)

* Fix load/store (#4054)

* Correctly use unsigned functions  (#4055)

* implement local and function calls for v128 in the fast interpreter

* Fix splat opcodes, add V128 handling in preserve_referenced_local and reserve_block_ret

* Fix incorrect memory overflow values + SIMD ifdefs

* Fix load/load_splat macros

* correct endif wasm loader

* Update core/iwasm/interpreter/wasm_opcode.h

* Fix spec tests when WASM_CPU_SUPPORTS_UNALIGNED_ADDR_ACCESS is 0

* Resolve merge conflicts arising from main -> dev/simd_for_interp and implement fast interpreter const offset loader support for V128

* Enable SIMDe tests on CI

* Document WAMR_BUILD_LIB_SIMDE

---------

Co-authored-by: James Marsh <mrshnja@amazon.co.uk>
Co-authored-by: jammar1 <108334558+jammar1@users.noreply.github.com>
Co-authored-by: Maks Litskevich <makslit@amazon.com>
Co-authored-by: Marcin Kolny <marcin.kolny@gmail.com>
Co-authored-by: Wenyong Huang <wenyong.huang@intel.com>

* Fix build issues when compiling WAMRC as a cross-compiler (#4112)

* Use CMAKE_INSTALL_BINDIR for wamrc installation
* Fix wamrc build failure for 32bit non-x86 targets
* Handle PIC flags by cmake in wamrc
* Use dummy AOT reloc functions when building wamrc

AOT reloc functions are used only when loading AOT WebAssembly modules
on target, not during AOT compilation. Original code led to build issues
when building wamrc as cross-compiler, using arm header on x86 build.

* Add option to turn off SIMD support in wamrc

* fix(runtest.py): A workaround to bypass errors that occur when deleting temporary files (#4093)

- Replace sys.exit with exceptions for better error handling in test assertions
- Update exception handling in compile_wast_to_wasm to catch all exceptions
- Improve error messages and logging
- Use `--ignore-whitespace` option for git apply in spec_test function
- Use raw string notation for regex patterns.  *The "SyntaxWarning: invalid escape sequence" in Python The warning has been upgraded to SyntaxWarning since Python 3.12, and it is expected to become a SyntaxError in future versions.*
- Add early return for non-loadable AOT compilation to prevent unnecessary assertions
- Redirect stderr to stdout in test_case for unified output
- Update `create_tmpfiles()`  to improve clarity and handling of temporary files

* build(deps): Bump esbuild, @vitejs/plugin-react and vite (#4149)

Bumps [esbuild](https://github.com/evanw/esbuild) to 0.25.1 and updates ancestor dependencies [esbuild](https://github.com/evanw/esbuild), [@vitejs/plugin-react](https://github.com/vitejs/vite-plugin-react/tree/HEAD/packages/plugin-react) and [vite](https://github.com/vitejs/vite/tree/HEAD/packages/vite). These dependencies need to be updated together.


Updates `esbuild` from 0.14.54 to 0.25.1
- [Release notes](https://github.com/evanw/esbuild/releases)
- [Changelog](https://github.com/evanw/esbuild/blob/main/CHANGELOG-2022.md)
- [Commits](https://github.com/evanw/esbuild/compare/v0.14.54...v0.25.1)

Updates `@vitejs/plugin-react` from 2.0.1 to 4.3.4
- [Release notes](https://github.com/vitejs/vite-plugin-react/releases)
- [Changelog](https://github.com/vitejs/vite-plugin-react/blob/main/packages/plugin-react/CHANGELOG.md)
- [Commits](https://github.com/vitejs/vite-plugin-react/commits/v4.3.4/packages/plugin-react)

Updates `vite` from 3.0.9 to 6.2.2
- [Release notes](https://github.com/vitejs/vite/releases)
- [Changelog](https://github.com/vitejs/vite/blob/main/packages/vite/CHANGELOG.md)
- [Commits](https://github.com/vitejs/vite/commits/v6.2.2/packages/vite)

---
updated-dependencies:
- dependency-name: esbuild
  dependency-type: indirect
- dependency-name: "@vitejs/plugin-react"
  dependency-type: direct:development
- dependency-name: vite
  dependency-type: direct:development
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Update NuttX and NuttX Apps references to releases/12.9 in workflow f… (#4148)

* Update NuttX and NuttX Apps references to releases/12.9 in workflow files
* Remove Kconfig modification step for NuttX in spec test workflow

* platform/nuttx: Flush icache/dcache properly (#4147)

Enhance the os_dcache_flush and os_icache_flush functions to ensure
proper cache invalidation, improving memory management efficiency.
* Added cache invalidation for data cache
* Implemented cache invalidation for instruction cache

Signed-off-by: Huang Qi <huangqi3@xiaomi.com>

* build(deps): Bump github/codeql-action from 3.28.11 to 3.28.12 (#4160)

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.28.11 to 3.28.12.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Commits](https://github.com/github/codeql-action/compare/v3.28.11...v3.28.12)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* build(deps): Bump actions/upload-artifact from 4.6.1 to 4.6.2 (#4159)

Bumps [actions/upload-artifact](https://github.com/actions/upload-artifact) from 4.6.1 to 4.6.2.
- [Release notes](https://github.com/actions/upload-artifact/releases)
- [Commits](https://github.com/actions/upload-artifact/compare/v4.6.1...v4.6.2)

---
updated-dependencies:
- dependency-name: actions/upload-artifact
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* test: temporarily skip 'skip-stack-guard-page' test case until issue is resolved

* nuttx: remove the up_x API for kernel build (#4154)

Signed-off-by: buxiasen <buxiasen@xiaomi.com>
Co-authored-by: buxiasen <buxiasen@xiaomi.com>

* docs: Update build instructions suggestions for using Valgrind (#4164)

* build(deps): Bump github/codeql-action from 3.28.12 to 3.28.13 (#4170)

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.28.12 to 3.28.13.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Commits](https://github.com/github/codeql-action/compare/v3.28.12...v3.28.13)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* dwarf_extractor.cpp: use macro control to be compatible with lower version toolchain (#4169)

* Update cmake min to 3.14 (#4175)

3.14 is used and tested by linux mini-product

to fix

```
CMake Error at CMakeLists.txt:4 (cmake_minimum_required):
  Compatibility with CMake < 3.5 has been removed from CMake.

  Update the VERSION argument <min> value.  Or, use the <min>...<max> syntax
  to tell CMake that the project requires at least <min> but has been updated
  to work with policies introduced by <max> or earlier.

  Or, add -DCMAKE_POLICY_VERSION_MINIMUM=3.5 to try configuring anyway.
```

* fix format specifier warning on 32bit builds (#4177)

* Remove indirect-load for constants on Xtensa Target to improve performance (#4162)

* Remove indirect-load for constants on Xtensa Target to improve performance
* Remove const intrinsics flags for xtensa instead of adding too much #ifdef
* Add AOT_INTRINSIC_FLAG_F32_CONST for xtensa frontend, because espressif xtensa llvm backend does not support float-point immediate for now

---------
Co-authored-by: zhanheng1 <Zhanheng.Qin@sony.com>

* cmake: Enhance target selection for ARM architectures with FPU (#4185)

Improve the target selection logic for ARM architectures in the NuttX platform configuration.
* Added support for FPU detection in THUMB and ARM targets
* Ensured correct target is set based on architecture and configuration options

Signed-off-by: Huang Qi <huangqi3@xiaomi.com>

* build(deps): Bump github/codeql-action from 3.28.13 to 3.28.14 (#4184)

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.28.13 to 3.28.14.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Commits](https://github.com/github/codeql-action/compare/v3.28.13...v3.28.14)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-version: 3.28.14
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* aot: add new u64 intrinsics (#4168)

* Refactor Dockerfile and update .dockerignore for wasi-nn tests; adjust map-dir parameters in smoke test script (#4158)

* Add import memory/table flag assert check for miniloader (#4179)

* Fix few integer overflowing (#4161)

- fix(interpreter): correct offset calculations in wasm_loader_get_const_offset function
- fix(mem-alloc): update offset calculation in gc_migrate for memory migration
- add pointer-overflow sanitizer

* prevent frame_offset underflow in wasm_loader (#4165)

* improve variable naming and code clarity in SIMD operations (#4157)

Fix compilation warning about shadow, like

```sh
 declaration of ‘val’ shadows a previous local [-Wshadow]
```

* fix: Remove unused variables in SIMD_v128_const case (#4197)

Fix compiler warnings about unused variables `high` and `low` in the `SIMD_v128_const` case. These variables are only needed inside the `WASM_ENABLE_FAST_INTERP != 0` conditional block, but were incorrectly declared outside of it, causing unused variable warnings.

Signed-off-by: Huang Qi <huangqi3@xiaomi.com>

* build(deps): Bump github/codeql-action from 3.28.14 to 3.28.15 (#4198)

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.28.14 to 3.28.15.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Commits](https://github.com/github/codeql-action/compare/v3.28.14...v3.28.15)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-version: 3.28.15
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* fix false native stack overflow detections with HW_BOUND_CHECK (#4196)

In call_wasm_with_hw_bound_check/call_native_with_hw_bound_check,
ensure to set up the stack boundary (wasm_exec_env_set_thread_info)
before checking the overflow.

It seems that the problem was introduced by:
https://github.com/bytecodealliance/wasm-micro-runtime/pull/2940

* Keep fix the CMake compatibility issue (#4180)

```
CMake Error at CMakeLists.txt:4 (cmake_minimum_required):
  Compatibility with CMake < 3.5 has been removed from CMake.

  Update the VERSION argument <min> value.  Or, use the <min>...<max> syntax
  to tell CMake that the project requires at least <min> but has been updated
  to work with policies introduced by <max> or earlier.

  Or, add -DCMAKE_POLICY_VERSION_MINIMUM=3.5 to try configuring anyway.
```

* Fix the error of AOT mode on the "i386-windows-msvc" platform (#4183)

* Fix errors on the "i386-windows-msvc" platform
* Refactor symbol name handling for AOT COFF32 binary format
* Fix preprocessor directive placement for Windows compatibility in aot_reloc_x86_32.c

---------

Co-authored-by: liang.he@intel.com <liang.he@intel.com>

* debug-engine: fix a few type mismatches (#4189)

- use strict prototypes complained by GCC `-Wstrict-prototypes`
- use `int*` instead of `int32*`

Note: on some targets, int32_t is a long.
for example, GCC shipped with the recent ESP-IDF has such a
configuration.

- https://github.com/apache/nuttx/issues/15755#issuecomment-2635652808
- https://github.com/apache/nuttx/pull/16022
- https://docs.espressif.com/projects/esp-idf/en/stable/esp32/migration-guides/release-5.x/5.0/gcc.html#espressif-toolchain-changes

* Replace CMAKE_CURRENT_FUNCTION_LIST_DIR (#4200)

`CMAKE_CURRENT_FUNCTION_LIST_DIR` is added in version 3.17 and currently
most of `cmake_minimum_required()` with 3.14.

Refer to https://cmake.org/cmake/help/latest/variable/CMAKE_CURRENT_FUNCTION_LIST_DIR.html

* Raise CI runner to ubuntu 22.04 (#4191)

update workflows and scripts for Ubuntu 22.04 compatibility. It includes
  - install Intel SGX SDK 2.25
  - use a reusable action to install sgx required
  - keep improve error handling in AOT compilation process in runtest.py

add a workaround to fix receiving a shutdown signal problem. Refers to https://github.com/actions/runner-images/issues/6680 and https://github.com/actions/runner-images/discussions/7188

* Remove the dlen to optimize it. (#4193)

There are two reasons for this optimization:
- The value of dlen can equal 0x1_0000_0000, even in wasm32 mode, because it is derived from (4G-0). This results in a truncation when it is passed to b_memmove_s(). Consequently, s1max becomes 0 and n is greater than s1max. To correct this, a longer type is required.
- The dlen is only used to check if there is enough space in b_memmove_s(). However, from a different angle, after confirming that both src+len and dst+len are within the memory range, we can be assured and there is no need for this explicit check.

* Add missing casts and improve error handling in performance map functions (#4202)

Wrong type of arguments to formatting function.

* Raise wasi-sdk to 25 and wabt to 1.0.37 (#4187)

 Raise wasi-sdk to 25 and wabt to 1.0.37. It includes
  - Refactor CI workflow to install WASI-SDK and WABT from a composite action
  - Use ExternalProject to bring wasm-apps for few samples. file/ wasi-threads/
  - Refactor sample build and test steps in SGX compilation workflow for improved clarity and efficiency (workaround)

Add CMake support for EMSCRIPTEN and WAMRC, update module paths

* fix potential memory leak (#4205)

* Add missing V128 handling in WASM_OP_BR, reported in #4173

* Update unit test cases (#4214)

* Update gc unit test cases
* Update aot stack frame unit test cases

* fix print_help when libc wasi is enabled (#4218)

* LLVM: don't verify instcombine fixpoint (#4219)

LLVM 18 and later, instcombine perfoms only one iteration.
it performs extra "verify fixpoint" operation when instcombine
is specified in certain ways, including how we do so here.
a problem is that the verification raises a fatal error when it
finds we didn't reach a fixpoint:

    LLVM ERROR: Instruction Combining did not reach a fixpoint
    after 1 iterations

while it should be rare, it's quite normal not to reach a fixpoint.
this commit fixes the issue by simply disabing the verification.

cf. 41895843b5

* LLVMCreateTargetMachineWithOpts: disable large data (#4220)

for x86-64, llvm 17 and later sometimes uses "l" prefix
for data sections.
cf. 43249378da

because our aot file emitter/loader doesn't support such
sections, it ends up with load-time errors solving symbols like ".lrodata".

this commit fixes it by avoid placing data in the large data sections.

references:
https://groups.google.com/g/x86-64-abi/c/jnQdJeabxiU
1feb00a28c

* wamrc: add --disable-llvm-jump-tables option (#4224)

while ideally a user should not need to care this kind of
optimization details, in reality i guess it's sometimes useful.
both of clang and GCC expose a similar option.  (-fno-jump-tables)

* feat(fuzz): add a new fuzzing target about aot compiler (#4121)

support llvm-jit running mode as another fuzzing target

* bypass vptr santizier (#4231)

LLVM, by default, disables the use of C++'s built-in Run-Time Type Information.
This decision is primarily driven by concerns about code size and efficiency.

But '-fsanitize=vptr' not allowed with '-fno-rtti'.

* use a selected llvm libs list to replace the full list (#4232)

* set default value of `WAMR_BUILD_REF_TYPES` to 1 in standalone cases (#4227)

- set default value of WAMR_BUILD_REF_TYPES to 1 in CMakeLists.txt

* teach aot emitter/loader about .srodata and .srodata.cst* sections (#4240)

LLVM 19 and later started to use srodata ("small read only data")
sections for RISCV.  cf. https://github.com/llvm/llvm-project/pull/82214
this commit makes our aot emitter/loader deal with those sections.

an alternative would be to disable small data sections completely by
setting the "SmallDataLimit" module attribute to zero. however, i feel
this commit is more straightforward and consisitent as we are already
dealing with sdata sections.

* run_clang_format_diff: mention homebrew for clang-format installation (#4237)

* platform/nuttx: Fix dcache operation in os_dcache_flush (#4225)

Replace up_invalidate_dcache_all() with up_flush_dcache_all() in
os_dcache_flush() to properly flush the data cache instead of just
invalidating it. This ensures that any modified data in the cache
is written back to memory before execution.

Signed-off-by: Huang Qi <huangqi3@xiaomi.com>

* Use --target to pass a triple in wamrc (#4199)

Provide a triple string in the format of <arch>-<vendor>-<os>-<abi>
via --target.

* fix return types of our 64-bit clz/ctz/popcount intrinsics (#4238)

the corresponding LLVM intrinsics' return types are same as
their first argument. eg. i64 for llvm.cttz.i64.
cf. https://llvm.org/docs/LangRef.html#llvm-cttz-intrinsic

this commit changes the return types of our versions of the
intrinsics to match llvm versions as our aot compiler,
specifically __call_llvm_intrinsic, assumes.

strictly speaking, this is a potential AOT ABI change.
however, I suppose it isn't a problem for many of 64-bit ABIs
out there, where (lower half of) a 64-bit register is used to
return a 32-bit value anyway.  (for such ABIs, this commit
would fix the upper 32-bit value of the register.)

* build(deps): Bump github/codeql-action from 3.28.15 to 3.28.17 (#4243)

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.28.15 to 3.28.17.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Commits](https://github.com/github/codeql-action/compare/v3.28.15...v3.28.17)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-version: 3.28.17
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* samples/wasm-c-api: skip aot compilation unless necessary (#4239)

* riscv: avoid llvm.cttz.i32/i64 for xip (#4248)

LLVM 16 and later expands cttz intrinsic to a table lookup,
which involves some relocations. (unless ZBB is available,
in which case the native instructions are preferred over
the table-based lowering.)

cf. https://reviews.llvm.org/D128911

* Add overflow check for preserved local offset in preserve_referenced_local (#4211)

* aot_resolve_object_relocation_group: adapt to LLVM 16 (#4250)

cf. https://reviews.llvm.org/D123264

* samples/wasm-c-api: remove unused valgrind detection (#4249)

- it's unused
- valgrind is basically a linux-only software.
  it isn't a good idea to make it a hard requirement.
  if we want to use valgrind, it's better to introduce
  a separate option to control it.

* More detail to python setup, and fixed small typo (#4247)

* initialize WASI stdio handles to invalid for better error handling (#4092)

* initialize WASI stdio handles to invalid for better error handling
* implement os_invalid_raw_handle function for consistent invalid handle representation

* Modifying build flags to ensure libiwasm.so is built (#4255)

* JIT: don't join worker threads twice (#4252)

in case of WASM_ENABLE_LAZY_JIT==0, compile_jit_functions should
have already joined these threads. joining them again here is
an undefined behavior.

* aot_resolve_object_relocation_group: adapt to LLVM 19 (#4254)

cf.
https://github.com/llvm/llvm-project/pull/95031
https://github.com/llvm/llvm-project/pull/89693

* Stop pretending to support extended-const proposal (#4258)

As far as I know, we don't implement the proposal at all.

```
spacetanuki% wasm2wat --enable-all data.28.wasm
(module
  (memory (;0;) 1)
  (data (;0;) (i32.const 42
    i32.const 0
    i32.sub) ""))
spacetanuki% toywasm --load data.28.wasm
spacetanuki% ~/git/wasm-micro-runtime/product-mini/platforms/darwin/b.classic/iwasm data.28.wasm
WASM module load failed: illegal opcode or constant expression required or type mismatch
spacetanuki%
```

data.28.wasm in the above example is a binary version of:
8d4f6aa2b0/test/core/data.wast (L184-L187)

* Improve readlinkat_dup() to handle symlink size correctly (#4229)

* In readlinkat_dup(), use fstatat() to estimate size first.
* Reduce additional space in samples/file

* build-scripts/build_llvm.py: bump to llvm 18 (#4259)

* build-scripts/build_llvm.py: bump to llvm 18

cf. https://github.com/bytecodealliance/wasm-micro-runtime/issues/4210

why not 20?
because, as of writing this, 19 is the latest released version for
the xtensa fork of llvm: https://github.com/espressif/llvm-project

why not 19?
because of a bug in the xtensa fork of llvm:
https://github.com/espressif/llvm-project/issues/112

while we can use different versions for different targets,
it's nicer to use the same version everywhere when possible.

* spec-test-script/runtest.py: --size-level=0 for x86-64

with the recent version of LLVM, wamrc --size-level=1 often
generates R_X86_64_32S relocations which fail on load with
the infamous error:

"relocation truncated to fit R_X86_64_32S failed"

it seems that these relocations are often for jump tables.

this commit workarounds it with --size-level=0.

an alternative is to disable jump tables. (although it seems that
jump tables are not the only source of these relocations.)

cf. https://github.com/bytecodealliance/wasm-micro-runtime/issues/3035

it might be better to do this in wamrc itself. however, currently
target info is not available there in case of native compilation.
related: https://github.com/bytecodealliance/wasm-micro-runtime/issues/3356

* wamr-compiler: size_level=0 for sgx mode

cf. https://github.com/bytecodealliance/wasm-micro-runtime/issues/3035

* fix: improve error handling of snprintf() in send_thread_stop_status() (#4234)

Prevent `MAX_PACKET_SIZE - len` from overflowing.

* Don't call os_thread_get_stack_boundary unless we actually use it (#4264)

Previously, if the user sets their own stack boundary, we still compute
the thread stack boundary (which is expensive), then immediately discard
the result. This change makes the expensive call only if we need it for
sure.

* CI: make macos' build_samples_wasm_c_api similar to ubuntu (#4253)

* avoid access null pointer (#4262)

* disable compiler to prevent get_current_target() crash (#4251)

* product-mini/platforms/windows: set C++17 explicitly (#4269)

The recent LLVM uses std::optional, which is C++17.

* fix buf checking in load_table_section (#4276)

Signed-off-by: Su Yihan <yihan.su@intel.com>

* Refactor fast-interpreter SIMD compilation flags (#4261)

- enable SIMD flag by default unless hardware limitation
- use SIMDE flag to control fast-interpreter behavior

* Bypass wamr_ide-related components from the release process. (#4268)

Mostly because of some observations:
- There is no actual usage reported.
- Both ide-ext and ide-docker-image have not been maintained for quite a while.
- At the very least, there is no need to recompile it every time when there are no modifications.

* Set CMAKE_OSX_SYSROOT when building lldb (#4274)

CMake 4 no longer sets the CMAKE_OSX_SYSROOT variable by default, causing the
lldb build to fail after all GitHub-hosted runners have been upgraded to
CMake 4.

As a workaround, the variable is set using CMake command line options. There
is a PR to fix this issue in the llvm-project:
https://github.com/llvm/llvm-project/pull/138020. We might want to remove
this workaround after that PR has been merged.

* Check for WASM_ENABLE_SIMDE in a couple more places (#4266)

For WAMR users who don't use cmake, it's possible that WASM_ENABLE_SIMD
is set when WASM_ENABLE_SIMDE isn't. This was causing build failures.

* Add error handling for sgx ci (#4222)

> Process completed with exit code 143.

It will attempt to run spec test scripts three times if they end with code 143.

It is a known issue with GitHub-hosted runners. Usually, increasing the swap
file can help avoid it. However, sometimes error 143 still occurs. To prevent
confusion, let's capture error 143 and allow the CI to pass.

* Add select 128 (#4236)

Add select 128

* Merge commit from fork

* Update version to 2.3.0 (#4171)

- Update version to 2.3.0
- Update RELEASE_NOTES.md. Remove commits that forget to squash when PRs were merged, and some updates on commit messages

---------

Co-authored-by: James Marsh <mrshnja@amazon.co.uk>
Co-authored-by: liang.he@intel.com <liang.he@intel.com>
Co-authored-by: TianlongLiang <111852609+TianlongLiang@users.noreply.github.com>

* Fix SIMD load lane to avoid incompatible pointer types (#4278)

* Fixed unit tests on X86_32 (#4279)

* fix unit tests on x86_32
* enbale wasm-c-api unit test on X86_32
* enable aot-stack-frame unit test on X86_32
* add ci: unit tests on X86_32

* feat(yml): Add ESP32-P4 and ESP32-C5 support (#4270)

- Add ESP32-P4 and ESP32-C5 support
- Support for compiler options of different floating-point types in various RISC-V chips

* build(deps): Bump github/codeql-action from 3.28.17 to 3.28.18 (#4285)

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.28.17 to 3.28.18.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Commits](https://github.com/github/codeql-action/compare/v3.28.17...v3.28.18)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-version: 3.28.18
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Improve Embedding WAMR guideline (#4263) (#4284)

* Fix CMakeList example by adding -lm
* Add bh_read_file inclusion to CMakeList
* replace non-existing read_binary_to_buffer() to existing bh_read_file_to_buffer()
* add #include initialization

Signed-off-by: Krisztian Szilvasi <34309983+kr-t@users.noreply.github.com>

* add a sample to use cmake package (#4291)

- add a sample to use cmake package

* feat: Add instruction metering for interpreter (#4122)

- add instruction metering support with execution limit
- initialize instruction execution limit in exec_env
- docs: add instruction metering section to build_wamr documentation

* Fix Compiler Error C2491 (#4286)

> Data, static data members, and functions can be declared as `dllimports` but not defined as `dllimports`.

https://learn.microsoft.com/en-us/cpp/error-messages/compiler-errors-1/compiler-error-c2491?view=msvc-170

* Revert the location to install public headers (#4295)

This partly reverts "Cmake improvements".
(https://github.com/bytecodealliance/wasm-micro-runtime/pull/4076)

Recently we changed the location to install public headers.
For example,
Old: include/wasm_export.h
New: include/iwasm/wasm_export.h

For cmake-based user applications using find_package(iwasm),
the cmake package, namely target_include_directories(INSTALL_INTERFACE),
is expected to add necessary compiler options like -isystem automatically.
(See samples/printversion for an example of such user applications.)

However, in reality, not every user application uses cmake.
This commit reverts the location to install public headers
for now, to avoid breakage for non-cmake user applications.

In case we want to re-apply the location change in future,
we should better communicate to the users. (eg. document
migration proceduces in release notes.)

Fixes: https://github.com/bytecodealliance/wasm-micro-runtime/issues/4290

References:
https://cmake.org/cmake/help/latest/prop_tgt/INTERFACE_INCLUDE_DIRECTORIES.html

* Enhance type checking for function types in loader and improve error handling (#4294)

Especially when GC is enabled, a valid item of `module->types` needs additional
checks before casting to WASMFuncType.

Also, avoid overflowing if reftype_map_count is 0.

Additionally, correctly set IN_OSS_FUZZ based on CFLAGS_ENV for sanitizer
configuration. Update ASan and UBSan messages for clarity in non-oss-fuzz
environments.

* Dockerfile.vx-delegate build error fix  (#4273)

- specify tensorflow version & bugfix

* Enable runtime API exposure for MSVC builds (#4287)

* updating WASI stdio handle initialization and build options for UVWASI (#4260)

* Bump version to 2.3.1 and update release notes (#4303)

* Fix a linking error caused by commit #3580d1 (#4311)

> **Fix a release-blocking issue**

---

Like:
```
vmlib.lib(blocking_op.obj) : error LNK2019: unresolved external symbol
__imp_wasm_runtime_begin_blocking_op referenced in function
blocking_op_close
[D:\a\wasm-micro-runtime\wasm-micro-runtime\wamr-compiler\build\wamrc.vcxproj]
vmlib.lib(blocking_op.obj) : error LNK2019: unresolved external symbol
__imp_wasm_runtime_end_blocking_op referenced in function
blocking_op_close
[D:\a\wasm-micro-runtime\wasm-micro-runtime\wamr-compiler\build\wamrc.vcxproj]
```

* add load_by_name in wasi-nn (#4298)

* build(deps): Bump ossf/scorecard-action from 2.4.1 to 2.4.2 (#4315)

Bumps [ossf/scorecard-action](https://github.com/ossf/scorecard-action) from 2.4.1 to 2.4.2.
- [Release notes](https://github.com/ossf/scorecard-action/releases)
- [Changelog](https://github.com/ossf/scorecard-action/blob/main/RELEASE.md)
- [Commits](f49aabe0b5...05b42c6244)

---
updated-dependencies:
- dependency-name: ossf/scorecard-action
  dependency-version: 2.4.2
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Bump uvwasi to latest commit #392e1f1 (#4312)

* wasi_nn_openvino.c: fix a few printf formats (#4310)

* wasi-nn: remove "backends" argument from detect_and_load_backend() (#4309)

it seems meaningless and quite confusing to access a table with
two aliases ("lookup" and "backends") within a function.

no functional changes are intended.

* fix wasi-nn abi definitions (#4307)

sync with a more appropriate version of the definitions.

as we use the "wasi_ephemeral_nn", which is p1-based, it seems
more appropriate to use definitions from witx, not wit.

it's a bit unfortunate p2-based wasi-nn made gratuitous changes
like this from p1.

note: this is an ABI change.

* wasi-nn: fix shared library filenames for macOS (#4306)

tested with openvino

* wasi_nn_openvino.c: make this buildable (#4305)

* handle nullable heap reference types in import section (#4302)

* wasi-nn: protect the backend lookup table with a lock (#4319)

this would avoid potential issues when multiple instances happen to
make an attempt to load a backend at the same time.

Fixes: https://github.com/bytecodealliance/wasm-micro-runtime/issues/4314

* wasi_nn.h: add import_name attribute (#4328)

this would fix undefined symbol errors by making it clear
these functions are imported.

references:
e2c698c7e8/llvm/lib/MC/WasmObjectWriter.cpp (L1798-L1799)
e2c698c7e8/llvm/lib/Object/WasmObjectFile.cpp (L749-L752)
e2c698c7e8/lld/wasm/Symbols.cpp (L203)
e2c698c7e8/lld/wasm/Relocations.cpp (L36-L40)

* wasi-nn: remove unused wasi_nn_dump_tensor_dimension prototype (#4325)

* Add wamrc compilation into Windows CI workflow (#4327)

+formatting

* Update binary compression steps to follow symlinks for actual files (#4321)

By default, zip follows symbolic links and includes the actual files
or directories they point to in the archive.

* Update Dockerfile for Zephyr SDK and Zephyr-project versioning (#4335)

Use a minimum manifest to reduce time consumption

* Collective fix: fix some typos (#4337)

* wasi-nn: move some host-only things out of wasi_nn_types.h (#4334)

cf. https://github.com/bytecodealliance/wasm-micro-runtime/issues/4324

* wasi-nn: fix the size of tensor->type (#4333)

* this enum is (@witx tag u8) in witx
* it seems that some wasm modules actually use non-zero padding
  and cause errors
* it's a bad practice to use C enum for ABI description anyway

* remove temporary wasi-libc build steps from CI workflows (#4343)

Ref:
https://github.com/bytecodealliance/wasm-micro-runtime/pull/2465

* wasi_nn.h: make this compatible with wasi_ephemeral_nn (#4330)

- wasi_nn.h: make this compatible with wasi_ephemeral_nn
cf. https://github.com/bytecodealliance/wasm-micro-runtime/issues/4323

- fix WASM_ENABLE_WASI_EPHEMERAL_NN build
this structure is used by host logic as well.
ideally definitions for wasm and host should be separated.
until it happens, check __wasm__ to avoid the breakage.

* wasi-nn: do not assign wasi_nn_ctx->backend multiple times (#4329)

* build(deps): Bump github/codeql-action from 3.28.18 to 3.28.19 (#4346)

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.28.18 to 3.28.19.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Commits](https://github.com/github/codeql-action/compare/v3.28.18...v3.28.19)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-version: 3.28.19
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* wasi_socket_ext.c: avoid tls to make this library-friendly (#4338)

* Enable aot memory64 sw bounds checks by default (#4350)

- enable aot memory64 sw bounds checks by default

* build(deps): Bump requests from 2.32.3 to 2.32.4 in /build-scripts (#4349)

Bumps [requests](https://github.com/psf/requests) from 2.32.3 to 2.32.4.
- [Release notes](https://github.com/psf/requests/releases)
- [Changelog](https://github.com/psf/requests/blob/main/HISTORY.md)
- [Commits](https://github.com/psf/requests/compare/v2.32.3...v2.32.4)

---
updated-dependencies:
- dependency-name: requests
  dependency-version: 2.32.4
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* wasi_nn_types.h: remove a seemingly stale comment (#4348)

* add heap-type check for GC when ref.null (#4300)

- According to [Link 1](https://webassembly.github.io/gc/core/valid/instructions.html#xref-syntax-instructions-syntax-instr-ref-mathsf-ref-null-mathit-ht), we must ensure that the heap type is valid when ref.null.
- According to [Link 2](https://webassembly.github.io/gc/core/valid/types.html#heap-types), a heap type is considered valid if it is either a concrete heap type or an abstract heap type.

However, in this function, the check for abstract heap types (absheaptype) was clearly missing, so this condition needs to be added explicitly in the if statement.

- When GC is disabled, no change is needed.
- When GC is enabled, heap types in WAMR are LEB-encoded values ([Link 3](https://webassembly.github.io/gc/core/appendix/index-types.html)). Therefore, we must use read_leb_int32 to parse the heap type correctly. And we can compute the original type1 using type1 = (uint8)((int32)0x80 + heap_type);.

* wamr-wasi-extensions: add a cmake package to provide our wasi extension (#4344)

* wasi_ephemeral_nn.h: add a convenience wrapper header
* wamr-wasi-extensions: add a cmake package to provide our wasi extension

the sample app was tested with:
* wasmtime
* iwasm with https://github.com/bytecodealliance/wasm-micro-runtime/pull/4308

currently only contains wasi-nn.
maybe it makes sense to add lib-socket things as well.

cf. https://github.com/bytecodealliance/wasm-micro-runtime/issues/4288

* wasi_nn_openvino.c: remove the tensor layout adjustment logic (#4308)

the logic in question seems like an attempt to work around
some application bugs.
my wild guess is that it was for classification-example.
cf. https://github.com/bytecodealliance/wasmtime/issues/10867

* Update type validation in load_table_import() and load_table() (#4296)

Prevent from value type.

https://webassembly.github.io/spec/core/valid/types.html#table-types
https://webassembly.github.io/gc/core/syntax/types.html#reference-types

* Follow #4268 to deprecate wamr_ide-related components (#4341)

refer to: Bypass wamr_ide-related components from the release process. (#4268)

* clean up incompatible running mode checks in test script and ci (#4342)

Rearrange the content of do_execute_in_running_mode() in alphabetical
order. 

Add an incompatible check for x86_32. Now, all belows will be bypassed:
- jit, fast-jit, multi-tier-jit
- memory64
- multi-memory
- simd

* Update WABT downloads URL (#4357)

Plus, skip unsupported running mode instead quit during wamr compiler
test

* Modify AOT static PGO to conform to llvm-18 and add a CI job to test static PGO on the coremark benchmark (#4345)

* static PGO compatible with llvm18 and add CI job to test static PGO on coremark benchmark
* update comments and warning info, bitmaps section in llvm profdata shouldn't be used in PGO

* Collective fix for typos and minor bugs (#4369)

* wasi-nn: fix backend leak on multiple loads (#4366)

cf. https://github.com/bytecodealliance/wasm-micro-runtime/issues/4340

* build(deps): Bump github/codeql-action from 3.28.19 to 3.29.0 (#4371)

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.28.19 to 3.29.0.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Commits](https://github.com/github/codeql-action/compare/v3.28.19...v3.29.0)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-version: 3.29.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* add validation for array type in load_init_expr(GC only) (#4370)

* wasi_nn_openvino.c: remove broken xml check (#4365)

`xml.buf[xml.size]` check is broken because it accesses past
the end of the buffer.

anyway, openvino doesn't seem to care the NUL termination.

* wamr-wasi-extensions: add lib-socket things (#4360)

* improve installation steps for wasi-sdk and wabt on Windows (#4359)

* wasi_ephemeral_nn.h: prefix identfiers to avoid too generic names (#4358)

* wasi_nn_openvino.c: add a missing buffer overflow check in get_output (#4353)

cf. https://github.com/bytecodealliance/wasm-micro-runtime/issues/4351

* send an empty/error reply from server (#4362)

Signed-off-by: Su Yihan <yihan.su@intel.com>

* wasi_nn_openvino.c: remove pre/postprocessing and layout assumptions (#4361)

as wasi-nn doesn't have these concepts, the best we can do without
risking breaking certain applications here is to pass through tensors
as they are.

this matches wasmtime's behavior.

tested with:

* wasmtime classification-example
  (with this change, this example fails on tensor size mismatch
  instead of implicitly resizing it.)

* license-plate-recognition-barrier-0007, a converted version
  with non-fp32 output. [1]
  (with this change, this model outputs integers as expected.)

[1] cd7ebe313b/models/public/license-plate-recognition-barrier-0007

* add nn-cli example (#4373)

an example application with flexible cli options which
aims to allow us to perform any wasi-nn operations.

eg.
```
--load-graph=file=fixture/model.xml,file=fixture/model.bin,id=graph
--init-execution-context=graph-id=graph,id=ctx
--set-input=file=fixture/tensor.bgr,context-id=ctx,dim=1,dim=3,dim=224,dim=224
--compute=context-id=ctx
--get-output=context-id=ctx,file=output.bin
```

* wasi-nn: apply the shared library hack to darwin as well (#4374)

copied from the linux version.

i'm a bit skeptical with this workaround though.
it might be simpler to prohibit the use of wamr api in these
shared libraries. after all, what these libraries do is nothing
specific to wasm.

* wasi-nn: don't try to deinit uninitialized backend (#4375)

cf. https://github.com/bytecodealliance/wasm-micro-runtime/issues/4339

* core/iwasm/libraries/wasi-nn/test/build.sh: add a tip for intel mac (#4389)

i keep forgetting this and had to re-investigate it at least twice.
hopefully this can be helpful for others too.

* wasi_nn_tensorflowlite.cpp: reject non-fp32 input earlier (#4388)

this backend assumes fp32 here and there.
it's safer to reject unexpected inputs explicitly.

* Fix several issues related to night-run CI and test scripts. (#4385)

- remove duplicated options
- fix test script
- change ci to use binary

* core/iwasm/libraries/wasi-nn/test: use the correct version of keras (#4383)

---------

Signed-off-by: Huang Qi <huangqi3@xiaomi.com>
Signed-off-by: dependabot[bot] <support@github.com>
Signed-off-by: Peter Tatrai <peter.tatrai.ext@siemens.com>
Signed-off-by: buxiasen <buxiasen@xiaomi.com>
Signed-off-by: Su Yihan <yihan.su@intel.com>
Signed-off-by: Krisztian Szilvasi <34309983+kr-t@users.noreply.github.com>
Co-authored-by: Huang Qi <huangqi3@xiaomi.com>
Co-authored-by: Marcin Kolny <mkolny@amazon.com>
Co-authored-by: TianlongLiang <111852609+TianlongLiang@users.noreply.github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Wenyong Huang <wenyong.huang@intel.com>
Co-authored-by: Xavier Del Campo <90845888+midokura-xavi92@users.noreply.github.com>
Co-authored-by: Tomáš Malý <malytomas@users.noreply.github.com>
Co-authored-by: Viacheslav Palchikov <palchikov@gmail.com>
Co-authored-by: yangkun27 <yangkun27@xiaomi.com>
Co-authored-by: Jérôme Vouillon <jerome.vouillon@gmail.com>
Co-authored-by: peter-tatrai <peter.tatrai.ext@siemens.com>
Co-authored-by: eric <chenliuyang_1989@163.com>
Co-authored-by: TL <tianlong.liang@intel.com>
Co-authored-by: jia xiang <58927968+Jiax-cn@users.noreply.github.com>
Co-authored-by: xiangjia.xj <xiangjia.xj@alibaba-inc.com>
Co-authored-by: Georgii Rylov <godjan@amazon.co.uk>
Co-authored-by: Jesse Wilson <jwilson@squareup.com>
Co-authored-by: Zhen Kong <zhkon@microsoft.com>
Co-authored-by: James Ring <sjr@jdns.org>
Co-authored-by: James Marsh <mrshnja@amazon.co.uk>
Co-authored-by: jammar1 <108334558+jammar1@users.noreply.github.com>
Co-authored-by: Maks Litskevich <makslit@amazon.com>
Co-authored-by: Marcin Kolny <marcin.kolny@gmail.com>
Co-authored-by: buxiasen <buxiasen@xiaomi.com>
Co-authored-by: dongsheng28849455 <68947925+dongsheng28849455@users.noreply.github.com>
Co-authored-by: zhanheng1 <Zhanheng.Qin@sony.com>
Co-authored-by: Raul Hernandez <raul.hernandez@spaceface.dev>
Co-authored-by: YAMAMOTO Takashi <yamamoto@midokura.com>
Co-authored-by: a seven <w4454962@users.noreply.github.com>
Co-authored-by: Zhenwei Jin <109658203+kylo5aby@users.noreply.github.com>
Co-authored-by: Liu Jia <jia3.liu@intel.com>
Co-authored-by: Chris Woods <6069113+woodsmc@users.noreply.github.com>
Co-authored-by: Su Yihan <yihan.su@intel.com>
Co-authored-by: ChenWen <63690793+cwespressif@users.noreply.github.com>
Co-authored-by: Krisztian <34309983+kr-t@users.noreply.github.com>
Co-authored-by: Alix ANNERAUD <alix@anneraud.fr>
Co-authored-by: hongxia <103626902+HongxiaWangSSSS@users.noreply.github.com>
2025-06-20 09:52:46 +08:00

3256 lines
122 KiB
C

/*
* Copyright (C) 2019 Intel Corporation. All rights reserved.
* SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
*/
#include "aot_emit_function.h"
#include "aot_emit_exception.h"
#include "aot_emit_control.h"
#include "aot_emit_table.h"
#include "aot_stack_frame_comp.h"
#include "../aot/aot_runtime.h"
#if WASM_ENABLE_GC != 0
#include "aot_emit_gc.h"
#endif
#define ADD_BASIC_BLOCK(block, name) \
do { \
if (!(block = LLVMAppendBasicBlockInContext(comp_ctx->context, \
func_ctx->func, name))) { \
aot_set_last_error("llvm add basic block failed."); \
goto fail; \
} \
} while (0)
static bool
is_win_platform(AOTCompContext *comp_ctx)
{
char *triple = LLVMGetTargetMachineTriple(comp_ctx->target_machine);
bool ret;
bh_assert(triple);
ret = (strstr(triple, "win32") || strstr(triple, "win")) ? true : false;
LLVMDisposeMessage(triple);
return ret;
}
static bool
create_func_return_block(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx)
{
LLVMBasicBlockRef block_curr = LLVMGetInsertBlock(comp_ctx->builder);
AOTFuncType *aot_func_type = func_ctx->aot_func->func_type;
/* Create function return block if it isn't created */
if (!func_ctx->func_return_block) {
if (!(func_ctx->func_return_block = LLVMAppendBasicBlockInContext(
comp_ctx->context, func_ctx->func, "func_ret"))) {
aot_set_last_error("llvm add basic block failed.");
return false;
}
/* Create return IR */
LLVMPositionBuilderAtEnd(comp_ctx->builder,
func_ctx->func_return_block);
if (!comp_ctx->enable_bound_check) {
if (!aot_emit_exception(comp_ctx, func_ctx, EXCE_ALREADY_THROWN,
false, NULL, NULL)) {
return false;
}
}
else if (!aot_build_zero_function_ret(comp_ctx, func_ctx,
aot_func_type)) {
return false;
}
}
LLVMPositionBuilderAtEnd(comp_ctx->builder, block_curr);
return true;
}
/* Check whether there was exception thrown, if yes, return directly */
static bool
check_exception_thrown(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx)
{
LLVMBasicBlockRef block_curr, check_exce_succ;
LLVMValueRef value, cmp;
/* Create function return block if it isn't created */
if (!create_func_return_block(comp_ctx, func_ctx))
return false;
/* Load the first byte of aot_module_inst->cur_exception, and check
whether it is '\0'. If yes, no exception was thrown. */
if (!(value = LLVMBuildLoad2(comp_ctx->builder, INT8_TYPE,
func_ctx->cur_exception, "exce_value"))
|| !(cmp = LLVMBuildICmp(comp_ctx->builder, LLVMIntEQ, value, I8_ZERO,
"cmp"))) {
aot_set_last_error("llvm build icmp failed.");
return false;
}
/* Add check exception success block */
if (!(check_exce_succ = LLVMAppendBasicBlockInContext(
comp_ctx->context, func_ctx->func, "check_exce_succ"))) {
aot_set_last_error("llvm add basic block failed.");
return false;
}
block_curr = LLVMGetInsertBlock(comp_ctx->builder);
LLVMMoveBasicBlockAfter(check_exce_succ, block_curr);
LLVMPositionBuilderAtEnd(comp_ctx->builder, block_curr);
/* Create condition br */
if (!LLVMBuildCondBr(comp_ctx->builder, cmp, check_exce_succ,
func_ctx->func_return_block)) {
aot_set_last_error("llvm build cond br failed.");
return false;
}
LLVMPositionBuilderAtEnd(comp_ctx->builder, check_exce_succ);
return true;
}
/* Check whether there was exception thrown, if yes, return directly */
static bool
check_call_return(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
LLVMValueRef res)
{
LLVMBasicBlockRef block_curr, check_call_succ;
LLVMValueRef cmp;
/* Create function return block if it isn't created */
if (!create_func_return_block(comp_ctx, func_ctx))
return false;
if (!(cmp = LLVMBuildICmp(comp_ctx->builder, LLVMIntNE, res, I8_ZERO,
"cmp"))) {
aot_set_last_error("llvm build icmp failed.");
return false;
}
/* Add check exception success block */
if (!(check_call_succ = LLVMAppendBasicBlockInContext(
comp_ctx->context, func_ctx->func, "check_call_succ"))) {
aot_set_last_error("llvm add basic block failed.");
return false;
}
block_curr = LLVMGetInsertBlock(comp_ctx->builder);
LLVMMoveBasicBlockAfter(check_call_succ, block_curr);
LLVMPositionBuilderAtEnd(comp_ctx->builder, block_curr);
/* Create condition br */
if (!LLVMBuildCondBr(comp_ctx->builder, cmp, check_call_succ,
func_ctx->func_return_block)) {
aot_set_last_error("llvm build cond br failed.");
return false;
}
LLVMPositionBuilderAtEnd(comp_ctx->builder, check_call_succ);
return true;
}
static bool
call_aot_invoke_native_func(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
LLVMValueRef func_idx, AOTFuncType *aot_func_type,
LLVMTypeRef *param_types,
LLVMValueRef *param_values, uint32 param_count,
uint32 param_cell_num, LLVMTypeRef ret_type,
uint8 wasm_ret_type, LLVMValueRef *p_value_ret,
LLVMValueRef *p_res)
{
LLVMTypeRef func_type, func_ptr_type, func_param_types[4];
LLVMTypeRef ret_ptr_type, elem_ptr_type;
LLVMValueRef func, elem_idx, elem_ptr;
LLVMValueRef func_param_values[4], value_ret = NULL, res;
char buf[32], *func_name = "aot_invoke_native";
uint32 i, cell_num = 0;
/* prepare function type of aot_invoke_native */
func_param_types[0] = comp_ctx->exec_env_type; /* exec_env */
func_param_types[1] = I32_TYPE; /* func_idx */
func_param_types[2] = I32_TYPE; /* argc */
func_param_types[3] = INT32_PTR_TYPE; /* argv */
if (!(func_type =
LLVMFunctionType(INT8_TYPE, func_param_types, 4, false))) {
aot_set_last_error("llvm add function type failed.");
return false;
}
/* prepare function pointer */
if (comp_ctx->is_jit_mode) {
if (!(func_ptr_type = LLVMPointerType(func_type, 0))) {
aot_set_last_error("create LLVM function type failed.");
return false;
}
/* JIT mode, call the function directly */
if (!(func = I64_CONST((uint64)(uintptr_t)llvm_jit_invoke_native))
|| !(func = LLVMConstIntToPtr(func, func_ptr_type))) {
aot_set_last_error("create LLVM value failed.");
return false;
}
}
else if (comp_ctx->is_indirect_mode) {
int32 func_index;
if (!(func_ptr_type = LLVMPointerType(func_type, 0))) {
aot_set_last_error("create LLVM function type failed.");
return false;
}
func_index = aot_get_native_symbol_index(comp_ctx, func_name);
if (func_index < 0) {
return false;
}
if (!(func = aot_get_func_from_table(comp_ctx, func_ctx->native_symbol,
func_ptr_type, func_index))) {
return false;
}
}
else {
if (!(func = LLVMGetNamedFunction(func_ctx->module, func_name))
&& !(func =
LLVMAddFunction(func_ctx->module, func_name, func_type))) {
aot_set_last_error("add LLVM function failed.");
return false;
}
}
if (param_cell_num > 64) {
aot_set_last_error("prepare native arguments failed: "
"maximum 64 parameter cell number supported.");
return false;
}
/* prepare frame_lp */
for (i = 0; i < param_count; i++) {
if (!(elem_idx = I32_CONST(cell_num))
|| !(elem_ptr_type = LLVMPointerType(param_types[i], 0))) {
aot_set_last_error("llvm add const or pointer type failed.");
return false;
}
snprintf(buf, sizeof(buf), "%s%d", "elem", i);
if (!(elem_ptr =
LLVMBuildInBoundsGEP2(comp_ctx->builder, I32_TYPE,
func_ctx->argv_buf, &elem_idx, 1, buf))
|| !(elem_ptr = LLVMBuildBitCast(comp_ctx->builder, elem_ptr,
elem_ptr_type, buf))) {
aot_set_last_error("llvm build bit cast failed.");
return false;
}
if (!(res = LLVMBuildStore(comp_ctx->builder, param_values[i],
elem_ptr))) {
aot_set_last_error("llvm build store failed.");
return false;
}
LLVMSetAlignment(res, 1);
cell_num += wasm_value_type_cell_num_internal(aot_func_type->types[i],
comp_ctx->pointer_size);
}
func_param_values[0] = func_ctx->exec_env;
func_param_values[1] = func_idx;
func_param_values[2] = I32_CONST(param_cell_num);
func_param_values[3] = func_ctx->argv_buf;
if (!func_param_values[2]) {
aot_set_last_error("llvm create const failed.");
return false;
}
/* call aot_invoke_native() function */
if (!(res = LLVMBuildCall2(comp_ctx->builder, func_type, func,
func_param_values, 4, "res"))) {
aot_set_last_error("llvm build call failed.");
return false;
}
/* get function return value */
if (wasm_ret_type != VALUE_TYPE_VOID) {
if (!(ret_ptr_type = LLVMPointerType(ret_type, 0))) {
aot_set_last_error("llvm add pointer type failed.");
return false;
}
if (!(value_ret =
LLVMBuildBitCast(comp_ctx->builder, func_ctx->argv_buf,
ret_ptr_type, "argv_ret"))) {
aot_set_last_error("llvm build bit cast failed.");
return false;
}
if (!(*p_value_ret = LLVMBuildLoad2(comp_ctx->builder, ret_type,
value_ret, "value_ret"))) {
aot_set_last_error("llvm build load failed.");
return false;
}
}
*p_res = res;
return true;
}
static bool
call_aot_invoke_c_api_native(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
uint32 import_func_idx, AOTFuncType *aot_func_type,
LLVMValueRef *params)
{
LLVMTypeRef int8_ptr_type, param_types[6], ret_type;
LLVMTypeRef value_ptr_type = NULL, value_type = NULL;
LLVMTypeRef func_type, func_ptr_type;
LLVMValueRef param_values[6], res, func, value = NULL, offset;
LLVMValueRef c_api_func_imports, c_api_func_import;
LLVMValueRef c_api_params, c_api_results, value_ret;
LLVMValueRef c_api_param_kind, c_api_param_value;
LLVMValueRef c_api_result_value;
uint32 offset_c_api_func_imports, i;
uint32 offset_param_kind, offset_param_value;
char buf[16];
/* `int8 **` type */
int8_ptr_type = LLVMPointerType(INT8_PTR_TYPE, 0);
if (!int8_ptr_type) {
aot_set_last_error("create llvm pointer type failed");
return false;
}
param_types[0] = INT8_PTR_TYPE; /* module_inst */
param_types[1] = INT8_PTR_TYPE; /* CApiFuncImport *c_api_import */
param_types[2] = INT8_PTR_TYPE; /* wasm_val_t *params */
param_types[3] = I32_TYPE; /* uint32 param_count */
param_types[4] = INT8_PTR_TYPE; /* wasm_val_t *results */
param_types[5] = I32_TYPE; /* uint32 result_count */
ret_type = INT8_TYPE;
GET_AOT_FUNCTION(wasm_runtime_quick_invoke_c_api_native, 6);
param_values[0] = func_ctx->aot_inst;
/* Get module_inst->c_api_func_imports, jit mode WASMModuleInstance is the
* same layout with AOTModuleInstance */
offset_c_api_func_imports = offsetof(AOTModuleInstance, c_api_func_imports);
offset = I32_CONST(offset_c_api_func_imports);
CHECK_LLVM_CONST(offset);
c_api_func_imports =
LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE, func_ctx->aot_inst,
&offset, 1, "c_api_func_imports_addr");
c_api_func_imports =
LLVMBuildBitCast(comp_ctx->builder, c_api_func_imports, int8_ptr_type,
"c_api_func_imports_ptr");
c_api_func_imports =
LLVMBuildLoad2(comp_ctx->builder, INT8_PTR_TYPE, c_api_func_imports,
"c_api_func_imports");
/* Get &c_api_func_imports[func_idx], note size of CApiFuncImport
is pointer_size * 3 */
offset = I32_CONST((comp_ctx->pointer_size * 3) * import_func_idx);
CHECK_LLVM_CONST(offset);
c_api_func_import =
LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE, c_api_func_imports,
&offset, 1, "c_api_func_import");
param_values[1] = c_api_func_import;
param_values[2] = c_api_params = func_ctx->argv_buf;
param_values[3] = I32_CONST(aot_func_type->param_count);
CHECK_LLVM_CONST(param_values[3]);
/* Ensure sizeof(wasm_val_t) is 16 bytes */
offset = I32_CONST(sizeof(wasm_val_t) * aot_func_type->param_count);
c_api_results =
LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE, func_ctx->argv_buf,
&offset, 1, "results");
param_values[4] = c_api_results;
param_values[5] = I32_CONST(aot_func_type->result_count);
CHECK_LLVM_CONST(param_values[5]);
/* Set each c api param */
for (i = 0; i < aot_func_type->param_count; i++) {
/* Ensure sizeof(wasm_val_t) is 16 bytes */
offset_param_kind = sizeof(wasm_val_t) * i;
offset = I32_CONST(offset_param_kind);
CHECK_LLVM_CONST(offset);
c_api_param_kind =
LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE, c_api_params,
&offset, 1, "c_api_param_kind_addr");
c_api_param_kind =
LLVMBuildBitCast(comp_ctx->builder, c_api_param_kind, INT8_PTR_TYPE,
"c_api_param_kind_ptr");
switch (aot_func_type->types[i]) {
case VALUE_TYPE_I32:
value = I8_CONST(WASM_I32);
break;
case VALUE_TYPE_F32:
value = I8_CONST(WASM_F32);
break;
case VALUE_TYPE_I64:
value = I8_CONST(WASM_I64);
break;
case VALUE_TYPE_F64:
value = I8_CONST(WASM_F64);
break;
default:
bh_assert(0);
break;
}
CHECK_LLVM_CONST(value);
LLVMBuildStore(comp_ctx->builder, value, c_api_param_kind);
/* Ensure offsetof(wasm_val_t, of) is 8 bytes */
offset_param_value = offset_param_kind + offsetof(wasm_val_t, of);
offset = I32_CONST(offset_param_value);
CHECK_LLVM_CONST(offset);
c_api_param_value =
LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE, c_api_params,
&offset, 1, "c_api_param_value_addr");
switch (aot_func_type->types[i]) {
case VALUE_TYPE_I32:
value_ptr_type = INT32_PTR_TYPE;
break;
case VALUE_TYPE_F32:
value_ptr_type = F32_PTR_TYPE;
break;
case VALUE_TYPE_I64:
value_ptr_type = INT64_PTR_TYPE;
break;
case VALUE_TYPE_F64:
value_ptr_type = F64_PTR_TYPE;
break;
default:
bh_assert(0);
break;
}
c_api_param_value =
LLVMBuildBitCast(comp_ctx->builder, c_api_param_value,
value_ptr_type, "c_api_param_value_ptr");
LLVMBuildStore(comp_ctx->builder, params[i], c_api_param_value);
}
/* Call the function */
if (!(res = LLVMBuildCall2(comp_ctx->builder, func_type, func, param_values,
6, "call"))) {
aot_set_last_error("LLVM build call failed.");
goto fail;
}
/* Check whether exception was thrown when executing the function */
if (comp_ctx->enable_bound_check
&& !check_call_return(comp_ctx, func_ctx, res)) {
goto fail;
}
for (i = 0; i < aot_func_type->result_count; i++) {
/* Ensure sizeof(wasm_val_t) is 16 bytes and
offsetof(wasm_val_t, of) is 8 bytes */
uint32 offset_result_value =
sizeof(wasm_val_t) * i + offsetof(wasm_val_t, of);
offset = I32_CONST(offset_result_value);
CHECK_LLVM_CONST(offset);
c_api_result_value =
LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE, c_api_results,
&offset, 1, "c_api_result_value_addr");
switch (aot_func_type->types[aot_func_type->param_count + i]) {
case VALUE_TYPE_I32:
value_type = I32_TYPE;
value_ptr_type = INT32_PTR_TYPE;
break;
case VALUE_TYPE_F32:
value_type = F32_TYPE;
value_ptr_type = F32_PTR_TYPE;
break;
case VALUE_TYPE_I64:
value_type = I64_TYPE;
value_ptr_type = INT64_PTR_TYPE;
break;
case VALUE_TYPE_F64:
value_type = F64_TYPE;
value_ptr_type = F64_PTR_TYPE;
break;
default:
bh_assert(0);
break;
}
c_api_result_value =
LLVMBuildBitCast(comp_ctx->builder, c_api_result_value,
value_ptr_type, "c_api_result_value_ptr");
snprintf(buf, sizeof(buf), "%s%u", "ret", i);
value_ret = LLVMBuildLoad2(comp_ctx->builder, value_type,
c_api_result_value, buf);
PUSH(value_ret, aot_func_type->types[aot_func_type->param_count + i]);
}
return true;
fail:
return false;
}
#if WASM_ENABLE_AOT_STACK_FRAME != 0
static bool
call_aot_alloc_frame_func(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
LLVMValueRef func_idx)
{
LLVMValueRef param_values[2], ret_value, value, func;
LLVMTypeRef param_types[2], ret_type, func_type, func_ptr_type;
LLVMBasicBlockRef block_curr = LLVMGetInsertBlock(comp_ctx->builder);
LLVMBasicBlockRef frame_alloc_fail, frame_alloc_success;
AOTFuncType *aot_func_type = func_ctx->aot_func->func_type;
param_types[0] = comp_ctx->exec_env_type;
param_types[1] = I32_TYPE;
ret_type = INT8_TYPE;
#if WASM_ENABLE_JIT != 0
if (comp_ctx->is_jit_mode)
GET_AOT_FUNCTION(llvm_jit_alloc_frame, 2);
else
#endif
GET_AOT_FUNCTION(aot_alloc_frame, 2);
param_values[0] = func_ctx->exec_env;
param_values[1] = func_idx;
if (!(ret_value =
LLVMBuildCall2(comp_ctx->builder, func_type, func, param_values,
2, "call_aot_alloc_frame"))) {
aot_set_last_error("llvm build call failed.");
return false;
}
if (!(ret_value = LLVMBuildICmp(comp_ctx->builder, LLVMIntUGT, ret_value,
I8_ZERO, "frame_alloc_ret"))) {
aot_set_last_error("llvm build icmp failed.");
return false;
}
ADD_BASIC_BLOCK(frame_alloc_fail, "frame_alloc_fail");
ADD_BASIC_BLOCK(frame_alloc_success, "frame_alloc_success");
LLVMMoveBasicBlockAfter(frame_alloc_fail, block_curr);
LLVMMoveBasicBlockAfter(frame_alloc_success, block_curr);
if (!LLVMBuildCondBr(comp_ctx->builder, ret_value, frame_alloc_success,
frame_alloc_fail)) {
aot_set_last_error("llvm build cond br failed.");
return false;
}
/* If frame alloc failed, return this function
so the runtime can catch the exception */
LLVMPositionBuilderAtEnd(comp_ctx->builder, frame_alloc_fail);
if (!aot_build_zero_function_ret(comp_ctx, func_ctx, aot_func_type)) {
return false;
}
LLVMPositionBuilderAtEnd(comp_ctx->builder, frame_alloc_success);
return true;
fail:
return false;
}
static bool
alloc_frame_for_aot_func(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
uint32 func_idx)
{
LLVMValueRef wasm_stack_top_bound = func_ctx->wasm_stack_top_bound;
LLVMValueRef wasm_stack_top_ptr = func_ctx->wasm_stack_top_ptr,
wasm_stack_top;
LLVMValueRef wasm_stack_top_max, offset, cmp;
LLVMValueRef cur_frame, new_frame, prev_frame_ptr;
LLVMValueRef cur_frame_ptr = func_ctx->cur_frame_ptr;
LLVMValueRef func_idx_ptr, func_idx_val, func_inst_ptr, func_inst;
LLVMTypeRef int8_ptr_type;
LLVMBasicBlockRef check_wasm_stack_succ;
uint32 import_func_count = comp_ctx->comp_data->import_func_count;
uint32 param_cell_num = 0, local_cell_num = 0, i;
uint32 max_local_cell_num, max_stack_cell_num;
uint32 all_cell_num, frame_size, frame_size_with_outs_area;
uint32 aot_frame_ptr_num = offsetof(AOTFrame, lp) / sizeof(uintptr_t);
AOTImportFunc *import_funcs = comp_ctx->comp_data->import_funcs;
AOTFuncType *aot_func_type;
AOTFunc *aot_func = NULL;
/* `int8 **` type */
int8_ptr_type = LLVMPointerType(INT8_PTR_TYPE, 0);
if (!int8_ptr_type) {
aot_set_last_error("create llvm pointer type failed");
return false;
}
/* Get param_cell_num, local_cell_num and max_stack_cell_num */
if (func_idx < import_func_count) {
aot_func_type = import_funcs[func_idx].func_type;
for (i = 0; i < aot_func_type->param_count; i++)
param_cell_num += wasm_value_type_cell_num_internal(
aot_func_type->types[i], comp_ctx->pointer_size);
max_local_cell_num = param_cell_num > 2 ? param_cell_num : 2;
max_stack_cell_num = 0;
}
else {
aot_func = comp_ctx->comp_data->funcs[func_idx - import_func_count];
param_cell_num = aot_func->param_cell_num;
local_cell_num = aot_func->local_cell_num;
max_local_cell_num = param_cell_num + local_cell_num;
max_stack_cell_num = aot_func->max_stack_cell_num;
}
all_cell_num = max_local_cell_num + max_stack_cell_num;
/* Get size of the frame to allocate and get size with outs_area to
check whether wasm operand stack is overflow */
if (!comp_ctx->is_jit_mode) {
/* Refer to aot_alloc_frame */
if (!comp_ctx->enable_gc) {
frame_size = frame_size_with_outs_area =
comp_ctx->pointer_size * aot_frame_ptr_num;
}
else {
frame_size = comp_ctx->pointer_size * aot_frame_ptr_num
+ align_uint(all_cell_num * 5, 4);
frame_size_with_outs_area =
frame_size + comp_ctx->pointer_size * aot_frame_ptr_num
+ max_stack_cell_num * 4;
}
}
else {
/* Refer to wasm_interp_interp_frame_size */
if (!comp_ctx->enable_gc) {
frame_size = frame_size_with_outs_area =
offsetof(WASMInterpFrame, lp);
}
else {
frame_size =
offsetof(WASMInterpFrame, lp) + align_uint(all_cell_num * 5, 4);
frame_size_with_outs_area = frame_size
+ offsetof(WASMInterpFrame, lp)
+ max_stack_cell_num * 4;
}
}
cur_frame = func_ctx->cur_frame;
if (!comp_ctx->enable_gc) {
offset = I32_CONST(frame_size);
CHECK_LLVM_CONST(offset);
if (!(wasm_stack_top =
LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE, cur_frame,
&offset, 1, "wasm_stack_top"))) {
aot_set_last_error("llvm build in bounds gep failed");
return false;
}
offset = I32_CONST(frame_size * 2);
CHECK_LLVM_CONST(offset);
if (!(wasm_stack_top_max =
LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE, cur_frame,
&offset, 1, "wasm_stack_top_max"))) {
aot_set_last_error("llvm build in bounds gep failed");
return false;
}
}
else {
/* Get exec_env->wasm_stack.top */
if (!(wasm_stack_top =
LLVMBuildLoad2(comp_ctx->builder, INT8_PTR_TYPE,
wasm_stack_top_ptr, "wasm_stack_top"))) {
aot_set_last_error("load wasm_stack.top failed");
return false;
}
/* Check whether wasm operand stack is overflow */
offset = I32_CONST(frame_size_with_outs_area);
CHECK_LLVM_CONST(offset);
if (!(wasm_stack_top_max = LLVMBuildInBoundsGEP2(
comp_ctx->builder, INT8_TYPE, wasm_stack_top, &offset, 1,
"wasm_stack_top_max"))) {
aot_set_last_error("llvm build in bounds gep failed");
return false;
}
}
new_frame = wasm_stack_top;
if (comp_ctx->call_stack_features.bounds_checks) {
if (!(check_wasm_stack_succ = LLVMAppendBasicBlockInContext(
comp_ctx->context, func_ctx->func,
"check_wasm_stack_succ"))) {
aot_set_last_error("llvm add basic block failed.");
return false;
}
LLVMMoveBasicBlockAfter(check_wasm_stack_succ,
LLVMGetInsertBlock(comp_ctx->builder));
if (!(cmp = LLVMBuildICmp(comp_ctx->builder, LLVMIntUGT,
wasm_stack_top_max, wasm_stack_top_bound,
"cmp"))) {
aot_set_last_error("llvm build icmp failed");
return false;
}
if (!(aot_emit_exception(comp_ctx, func_ctx,
EXCE_OPERAND_STACK_OVERFLOW, true, cmp,
check_wasm_stack_succ))) {
return false;
}
}
#if WASM_ENABLE_GC != 0
if (comp_ctx->enable_gc) {
LLVMValueRef wasm_stack_top_new, frame_ref, frame_ref_ptr;
uint32 j, k;
/* exec_env->wasm_stack.top += frame_size */
offset = I32_CONST(frame_size);
CHECK_LLVM_CONST(offset);
if (!(wasm_stack_top_new = LLVMBuildInBoundsGEP2(
comp_ctx->builder, INT8_TYPE, wasm_stack_top, &offset, 1,
"wasm_stack_top_new"))) {
aot_set_last_error("llvm build in bounds gep failed");
return false;
}
if (!LLVMBuildStore(comp_ctx->builder, wasm_stack_top_new,
wasm_stack_top_ptr)) {
aot_set_last_error("llvm build store failed");
return false;
}
if (func_idx < import_func_count) {
LLVMValueRef frame_sp, frame_sp_ptr;
/* Only need to initialize new_frame->sp when it's import function
otherwise they will be committed in AOT code if needed */
/* new_frame->sp = new_frame->lp + max_local_cell_num */
if (!comp_ctx->is_jit_mode)
offset = I32_CONST(comp_ctx->pointer_size * 5);
else
offset = I32_CONST(offsetof(WASMInterpFrame, sp));
CHECK_LLVM_CONST(offset);
if (!(frame_sp_ptr = LLVMBuildInBoundsGEP2(
comp_ctx->builder, INT8_TYPE, new_frame, &offset, 1,
"frame_sp_addr"))
|| !(frame_sp_ptr =
LLVMBuildBitCast(comp_ctx->builder, frame_sp_ptr,
int8_ptr_type, "frame_sp_ptr"))) {
aot_set_last_error("llvm get frame_sp_ptr failed");
return false;
}
if (!comp_ctx->is_jit_mode)
offset = I32_CONST(comp_ctx->pointer_size * aot_frame_ptr_num
+ max_local_cell_num * sizeof(uint32));
else
offset = I32_CONST(offsetof(WASMInterpFrame, lp)
+ max_local_cell_num * sizeof(uint32));
CHECK_LLVM_CONST(offset);
if (!(frame_sp = LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE,
new_frame, &offset, 1,
"frame_sp"))) {
aot_set_last_error("llvm build in bounds gep failed");
return false;
}
if (!LLVMBuildStore(comp_ctx->builder, frame_sp, frame_sp_ptr)) {
aot_set_last_error("llvm build store failed");
return false;
}
}
if (!comp_ctx->is_jit_mode) {
/* new_frame->frame_ref = new_frame->lp + max_local_cell_num
+ max_stack_cell_num */
offset = I32_CONST(comp_ctx->pointer_size * 6);
CHECK_LLVM_CONST(offset);
if (!(frame_ref_ptr = LLVMBuildInBoundsGEP2(
comp_ctx->builder, INT8_TYPE, new_frame, &offset, 1,
"frame_ref_addr"))
|| !(frame_ref_ptr =
LLVMBuildBitCast(comp_ctx->builder, frame_ref_ptr,
int8_ptr_type, "frame_ref_ptr"))) {
aot_set_last_error("llvm get frame_ref_ptr failed");
return false;
}
offset = I32_CONST(comp_ctx->pointer_size * aot_frame_ptr_num
+ (max_local_cell_num + max_stack_cell_num)
* sizeof(uint32));
CHECK_LLVM_CONST(offset);
if (!(frame_ref = LLVMBuildInBoundsGEP2(comp_ctx->builder,
INT8_TYPE, new_frame,
&offset, 1, "frame_ref"))) {
aot_set_last_error("llvm build in bounds gep failed");
return false;
}
if (!LLVMBuildStore(comp_ctx->builder, frame_ref, frame_ref_ptr)) {
aot_set_last_error("llvm build store failed");
return false;
}
}
else {
/* Get frame_ref in WASMInterpFrame */
offset = I32_CONST(offsetof(WASMInterpFrame, lp)
+ (max_local_cell_num + max_stack_cell_num)
* sizeof(uint32));
CHECK_LLVM_CONST(offset);
if (!(frame_ref = LLVMBuildInBoundsGEP2(comp_ctx->builder,
INT8_TYPE, new_frame,
&offset, 1, "frame_ref"))) {
aot_set_last_error("llvm build in bounds gep failed");
return false;
}
}
/* Initialize frame ref flags for import function only in JIT mode */
if (func_idx < import_func_count && comp_ctx->is_jit_mode) {
aot_func_type = import_funcs[func_idx].func_type;
for (i = 0, j = 0; i < aot_func_type->param_count; i++) {
if (aot_is_type_gc_reftype(aot_func_type->types[i])
&& !wasm_is_reftype_i31ref(aot_func_type->types[i])) {
for (k = 0; k < comp_ctx->pointer_size / sizeof(uint32);
k++) {
/* frame_ref[j++] = 1 */
offset = I32_CONST(j);
CHECK_LLVM_CONST(offset);
frame_ref_ptr = LLVMBuildInBoundsGEP2(
comp_ctx->builder, INT8_TYPE, frame_ref, &offset, 1,
"frame_ref_ptr");
if (!LLVMBuildStore(comp_ctx->builder, I8_ONE,
frame_ref_ptr)) {
aot_set_last_error("llvm build store failed");
return false;
}
j++;
}
}
else {
uint32 value_type_cell_num =
wasm_value_type_cell_num_internal(
aot_func_type->types[i], comp_ctx->pointer_size);
for (k = 0; k < value_type_cell_num; k++) {
/* frame_ref[j++] = 0 */
offset = I32_CONST(j);
CHECK_LLVM_CONST(offset);
frame_ref_ptr = LLVMBuildInBoundsGEP2(
comp_ctx->builder, INT8_TYPE, frame_ref, &offset, 1,
"frame_ref_ptr");
if (!LLVMBuildStore(comp_ctx->builder, I8_ZERO,
frame_ref_ptr)) {
aot_set_last_error("llvm build store failed");
return false;
}
j++;
}
}
}
for (; j < 2; j++) {
/* frame_ref[j++] = 0 */
offset = I32_CONST(j);
CHECK_LLVM_CONST(offset);
frame_ref_ptr = LLVMBuildInBoundsGEP2(
comp_ctx->builder, INT8_TYPE, frame_ref, &offset, 1,
"frame_ref_ptr");
if (!LLVMBuildStore(comp_ctx->builder, I8_ZERO,
frame_ref_ptr)) {
aot_set_last_error("llvm build store failed");
return false;
}
}
}
}
#endif /* end of WASM_ENABLE_GC != 0 */
/* new_frame->prev_frame = cur_frame */
if (!(prev_frame_ptr = LLVMBuildBitCast(comp_ctx->builder, new_frame,
int8_ptr_type, "prev_frame_ptr"))) {
aot_set_last_error("llvm build bitcast failed");
return false;
}
if (!LLVMBuildStore(comp_ctx->builder, cur_frame, prev_frame_ptr)) {
aot_set_last_error("llvm build store failed");
return false;
}
if (!comp_ctx->is_jit_mode) {
if (comp_ctx->call_stack_features.func_idx) {
/* aot mode: new_frame->func_idx = func_idx */
func_idx_val = comp_ctx->pointer_size == sizeof(uint64)
? I64_CONST(func_idx)
: I32_CONST(func_idx);
offset = I32_CONST(comp_ctx->pointer_size);
CHECK_LLVM_CONST(func_idx_val);
CHECK_LLVM_CONST(offset);
if (!(func_idx_ptr = LLVMBuildInBoundsGEP2(
comp_ctx->builder, INT8_TYPE, new_frame, &offset, 1,
"func_idx_addr"))
|| !(func_idx_ptr =
LLVMBuildBitCast(comp_ctx->builder, func_idx_ptr,
INTPTR_T_PTR_TYPE, "func_idx_ptr"))) {
aot_set_last_error("llvm get func_idx_ptr failed");
return false;
}
if (!LLVMBuildStore(comp_ctx->builder, func_idx_val,
func_idx_ptr)) {
aot_set_last_error("llvm build store failed");
return false;
}
}
}
else {
/* jit mode: frame->function = module_inst->e->functions + func_index */
LLVMValueRef functions;
uint32 offset_functions =
get_module_inst_extra_offset(comp_ctx)
+ offsetof(WASMModuleInstanceExtra, functions);
offset = I32_CONST(offset_functions);
CHECK_LLVM_CONST(offset);
if (!(functions = LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE,
func_ctx->aot_inst, &offset, 1,
"functions_addr"))) {
aot_set_last_error("llvm build inbounds gep failed");
return false;
}
if (!(functions = LLVMBuildBitCast(comp_ctx->builder, functions,
int8_ptr_type, "functions_ptr"))) {
aot_set_last_error("llvm build bitcast failed");
return false;
}
if (!(functions = LLVMBuildLoad2(comp_ctx->builder, INT8_PTR_TYPE,
functions, "functions"))) {
aot_set_last_error("llvm build load failed");
return false;
}
offset = I32_CONST(sizeof(WASMFunctionInstance) * func_idx);
CHECK_LLVM_CONST(offset);
if (!(func_inst =
LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE, functions,
&offset, 1, "func_inst"))) {
aot_set_last_error("llvm build inbounds gep failed");
return false;
}
offset = I32_CONST(offsetof(WASMInterpFrame, function));
CHECK_LLVM_CONST(offset);
if (!(func_inst_ptr =
LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE, new_frame,
&offset, 1, "func_inst_addr"))
|| !(func_inst_ptr =
LLVMBuildBitCast(comp_ctx->builder, func_inst_ptr,
int8_ptr_type, "func_inst_ptr"))) {
aot_set_last_error("llvm get func_inst_ptr failed");
return false;
}
if (!LLVMBuildStore(comp_ctx->builder, func_inst, func_inst_ptr)) {
aot_set_last_error("llvm build store failed");
return false;
}
}
/* No need to initialize new_frame->sp and new_frame->ip_offset
since they will be committed in AOT/JIT code if needed */
/* exec_env->cur_frame = new_frame */
if (!LLVMBuildStore(comp_ctx->builder, new_frame, cur_frame_ptr)) {
aot_set_last_error("llvm build store failed");
return false;
}
if (comp_ctx->enable_perf_profiling || comp_ctx->enable_memory_profiling) {
LLVMTypeRef param_types[2], func_type, func_ptr_type;
LLVMValueRef param_values[2], func = NULL, res;
char *func_name = "aot_frame_update_profile_info";
/* Call aot_frame_update_profile_info for AOT or
llvm_jit_frame_update_profile_info for JIT */
param_types[0] = comp_ctx->exec_env_type;
param_types[1] = INT8_TYPE;
if (!(func_type = LLVMFunctionType(VOID_TYPE, param_types, 2, false))) {
aot_set_last_error("llvm add function type failed.");
return false;
}
if (comp_ctx->is_jit_mode) {
if (!(func_ptr_type = LLVMPointerType(func_type, 0))) {
aot_set_last_error("create LLVM function type failed.");
return false;
}
#if WASM_ENABLE_JIT != 0 \
&& (WASM_ENABLE_PERF_PROFILING != 0 || WASM_ENABLE_MEMORY_PROFILING != 0)
/* JIT mode, call the function directly */
if (!(func = I64_CONST(
(uint64)(uintptr_t)llvm_jit_frame_update_profile_info))
|| !(func = LLVMConstIntToPtr(func, func_ptr_type))) {
aot_set_last_error("create LLVM value failed.");
return false;
}
#endif
}
else if (comp_ctx->is_indirect_mode) {
int32 func_index;
if (!(func_ptr_type = LLVMPointerType(func_type, 0))) {
aot_set_last_error("create LLVM function type failed.");
return false;
}
func_index = aot_get_native_symbol_index(comp_ctx, func_name);
if (func_index < 0) {
return false;
}
if (!(func =
aot_get_func_from_table(comp_ctx, func_ctx->native_symbol,
func_ptr_type, func_index))) {
return false;
}
}
else {
if (!(func = LLVMGetNamedFunction(func_ctx->module, func_name))
&& !(func = LLVMAddFunction(func_ctx->module, func_name,
func_type))) {
aot_set_last_error("add LLVM function failed.");
return false;
}
}
param_values[0] = func_ctx->exec_env;
param_values[1] = I8_ONE;
if (!(res = LLVMBuildCall2(comp_ctx->builder, func_type, func,
param_values, 2, ""))) {
aot_set_last_error("llvm build call failed.");
return false;
}
}
return true;
fail:
return false;
}
static bool
free_frame_for_aot_func(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx)
{
LLVMValueRef cur_frame_ptr = func_ctx->cur_frame_ptr, cur_frame;
LLVMValueRef wasm_stack_top_ptr = func_ctx->wasm_stack_top_ptr;
LLVMTypeRef int8_ptr_type;
/* `int8 **` type */
int8_ptr_type = LLVMPointerType(INT8_PTR_TYPE, 0);
if (!int8_ptr_type) {
aot_set_last_error("create llvm pointer type failed");
return false;
}
if (comp_ctx->enable_perf_profiling) {
LLVMTypeRef param_types[2], func_type, func_ptr_type;
LLVMValueRef param_values[2], func = NULL, res;
char *func_name = "aot_frame_update_profile_info";
/* call aot_frame_update_profile_info for AOT or
llvm_jit_frame_update_profile_info for JIT */
param_types[0] = comp_ctx->exec_env_type;
param_types[1] = INT8_TYPE;
if (!(func_type = LLVMFunctionType(VOID_TYPE, param_types, 2, false))) {
aot_set_last_error("llvm add function type failed.");
return false;
}
if (comp_ctx->is_jit_mode) {
if (!(func_ptr_type = LLVMPointerType(func_type, 0))) {
aot_set_last_error("create LLVM function type failed.");
return false;
}
#if WASM_ENABLE_JIT != 0 \
&& (WASM_ENABLE_PERF_PROFILING != 0 || WASM_ENABLE_MEMORY_PROFILING != 0)
/* JIT mode, call the function directly */
if (!(func = I64_CONST(
(uint64)(uintptr_t)llvm_jit_frame_update_profile_info))
|| !(func = LLVMConstIntToPtr(func, func_ptr_type))) {
aot_set_last_error("create LLVM value failed.");
return false;
}
#endif
}
else if (comp_ctx->is_indirect_mode) {
int32 func_index;
if (!(func_ptr_type = LLVMPointerType(func_type, 0))) {
aot_set_last_error("create LLVM function type failed.");
return false;
}
func_index = aot_get_native_symbol_index(comp_ctx, func_name);
if (func_index < 0) {
return false;
}
if (!(func =
aot_get_func_from_table(comp_ctx, func_ctx->native_symbol,
func_ptr_type, func_index))) {
return false;
}
}
else {
if (!(func = LLVMGetNamedFunction(func_ctx->module, func_name))
&& !(func = LLVMAddFunction(func_ctx->module, func_name,
func_type))) {
aot_set_last_error("add LLVM function failed.");
return false;
}
}
param_values[0] = func_ctx->exec_env;
param_values[1] = I8_ZERO;
if (!(res = LLVMBuildCall2(comp_ctx->builder, func_type, func,
param_values, 2, ""))) {
aot_set_last_error("llvm build call failed.");
return false;
}
}
if (comp_ctx->enable_gc) {
/* cur_frame = exec_env->cur_frame */
if (!(cur_frame = LLVMBuildLoad2(comp_ctx->builder, INT8_PTR_TYPE,
cur_frame_ptr, "cur_frame"))) {
aot_set_last_error("llvm build load failed");
return false;
}
/* exec_env->wasm_stack.top = cur_frame */
if (!LLVMBuildStore(comp_ctx->builder, cur_frame, wasm_stack_top_ptr)) {
aot_set_last_error("llvm build store failed");
return false;
}
}
/* exec_env->cur_frame = prev_frame */
if (!LLVMBuildStore(comp_ctx->builder, func_ctx->cur_frame,
cur_frame_ptr)) {
aot_set_last_error("llvm build store failed");
return false;
}
return true;
}
#endif /* end of WASM_ENABLE_AOT_STACK_FRAME != 0 */
/**
* Check whether the app address and its buffer are inside the linear memory,
* if no, throw exception
*/
static bool
check_app_addr_and_convert(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
bool is_str_arg, LLVMValueRef app_addr,
LLVMValueRef buf_size,
LLVMValueRef *p_native_addr_converted)
{
LLVMTypeRef func_type, func_ptr_type, func_param_types[5];
LLVMValueRef func, func_param_values[5], res, native_addr_ptr;
char *func_name = "aot_check_app_addr_and_convert";
/* prepare function type of aot_check_app_addr_and_convert */
func_param_types[0] = comp_ctx->aot_inst_type; /* module_inst */
func_param_types[1] = INT8_TYPE; /* is_str_arg */
func_param_types[2] = I64_TYPE; /* app_offset */
func_param_types[3] = I64_TYPE; /* buf_size */
func_param_types[4] =
comp_ctx->basic_types.int8_pptr_type; /* p_native_addr */
if (!(func_type =
LLVMFunctionType(INT8_TYPE, func_param_types, 5, false))) {
aot_set_last_error("llvm add function type failed.");
return false;
}
/* prepare function pointer */
if (comp_ctx->is_jit_mode) {
if (!(func_ptr_type = LLVMPointerType(func_type, 0))) {
aot_set_last_error("create LLVM function type failed.");
return false;
}
/* JIT mode, call the function directly */
if (!(func =
I64_CONST((uint64)(uintptr_t)jit_check_app_addr_and_convert))
|| !(func = LLVMConstIntToPtr(func, func_ptr_type))) {
aot_set_last_error("create LLVM value failed.");
return false;
}
}
else if (comp_ctx->is_indirect_mode) {
int32 func_index;
if (!(func_ptr_type = LLVMPointerType(func_type, 0))) {
aot_set_last_error("create LLVM function type failed.");
return false;
}
func_index = aot_get_native_symbol_index(comp_ctx, func_name);
if (func_index < 0) {
return false;
}
if (!(func = aot_get_func_from_table(comp_ctx, func_ctx->native_symbol,
func_ptr_type, func_index))) {
return false;
}
}
else {
if (!(func = LLVMGetNamedFunction(func_ctx->module, func_name))
&& !(func =
LLVMAddFunction(func_ctx->module, func_name, func_type))) {
aot_set_last_error("add LLVM function failed.");
return false;
}
}
if (!(native_addr_ptr = LLVMBuildBitCast(
comp_ctx->builder, func_ctx->argv_buf,
comp_ctx->basic_types.int8_pptr_type, "p_native_addr"))) {
aot_set_last_error("llvm build bit cast failed.");
return false;
}
func_param_values[0] = func_ctx->aot_inst;
func_param_values[1] = I8_CONST(is_str_arg);
func_param_values[2] = app_addr;
func_param_values[3] = buf_size;
func_param_values[4] = native_addr_ptr;
if (!func_param_values[1]) {
aot_set_last_error("llvm create const failed.");
return false;
}
/* call aot_check_app_addr_and_convert() function */
if (!(res = LLVMBuildCall2(comp_ctx->builder, func_type, func,
func_param_values, 5, "res"))) {
aot_set_last_error("llvm build call failed.");
return false;
}
/* Check whether exception was thrown when executing the function */
if ((comp_ctx->enable_bound_check || is_win_platform(comp_ctx))
&& !check_call_return(comp_ctx, func_ctx, res)) {
return false;
}
if (!(*p_native_addr_converted =
LLVMBuildLoad2(comp_ctx->builder, OPQ_PTR_TYPE, native_addr_ptr,
"native_addr"))) {
aot_set_last_error("llvm build load failed.");
return false;
}
return true;
}
static void
aot_estimate_and_record_stack_usage_for_function_call(
const AOTCompContext *comp_ctx, AOTFuncContext *caller_func_ctx,
const AOTFuncType *callee_func_type)
{
unsigned int size;
if (!(comp_ctx->enable_stack_bound_check
|| comp_ctx->enable_stack_estimation)) {
return;
}
size =
aot_estimate_stack_usage_for_function_call(comp_ctx, callee_func_type);
/*
* only record the max value, assuming that LLVM emits machine code
* which rewinds the stack before making the next call in the
* function.
*/
if (caller_func_ctx->stack_consumption_for_func_call < size) {
caller_func_ctx->stack_consumption_for_func_call = size;
}
}
static bool
commit_params_to_frame_of_import_func(AOTCompContext *comp_ctx,
AOTFuncContext *func_ctx,
AOTFuncType *func_type,
const LLVMValueRef *param_values)
{
uint32 i, n;
if (!comp_ctx->call_stack_features.values) {
return true;
}
for (i = 0, n = 0; i < func_type->param_count; i++, n++) {
switch (func_type->types[i]) {
case VALUE_TYPE_I32:
if (!aot_frame_store_value(
comp_ctx, param_values[i], VALUE_TYPE_I32,
func_ctx->cur_frame,
offset_of_local_in_outs_area(comp_ctx, n)))
return false;
break;
case VALUE_TYPE_I64:
if (!aot_frame_store_value(
comp_ctx, param_values[i], VALUE_TYPE_I64,
func_ctx->cur_frame,
offset_of_local_in_outs_area(comp_ctx, n)))
return false;
n++;
break;
case VALUE_TYPE_F32:
if (!aot_frame_store_value(
comp_ctx, param_values[i], VALUE_TYPE_F32,
func_ctx->cur_frame,
offset_of_local_in_outs_area(comp_ctx, n)))
return false;
break;
case VALUE_TYPE_F64:
if (!aot_frame_store_value(
comp_ctx, param_values[i], VALUE_TYPE_F64,
func_ctx->cur_frame,
offset_of_local_in_outs_area(comp_ctx, n)))
return false;
n++;
break;
case VALUE_TYPE_FUNCREF:
case VALUE_TYPE_EXTERNREF:
if (comp_ctx->enable_ref_types) {
if (!aot_frame_store_value(
comp_ctx, param_values[i], VALUE_TYPE_I32,
func_ctx->cur_frame,
offset_of_local_in_outs_area(comp_ctx, n)))
return false;
}
#if WASM_ENABLE_GC != 0
else if (comp_ctx->enable_gc) {
if (!aot_frame_store_value(
comp_ctx, param_values[i], VALUE_TYPE_GC_REF,
func_ctx->cur_frame,
offset_of_local_in_outs_area(comp_ctx, n)))
return false;
if (comp_ctx->pointer_size == sizeof(uint64))
n++;
}
#endif
else {
bh_assert(0);
}
break;
#if WASM_ENABLE_GC != 0
case REF_TYPE_NULLFUNCREF:
case REF_TYPE_NULLEXTERNREF:
case REF_TYPE_NULLREF:
/* case REF_TYPE_FUNCREF: */
/* case REF_TYPE_EXTERNREF: */
case REF_TYPE_ANYREF:
case REF_TYPE_EQREF:
case REF_TYPE_HT_NULLABLE:
case REF_TYPE_HT_NON_NULLABLE:
case REF_TYPE_I31REF:
case REF_TYPE_STRUCTREF:
case REF_TYPE_ARRAYREF:
case VALUE_TYPE_GC_REF:
#if WASM_ENABLE_STRINGREF != 0
case REF_TYPE_STRINGREF:
case REF_TYPE_STRINGVIEWWTF8:
case REF_TYPE_STRINGVIEWWTF16:
case REF_TYPE_STRINGVIEWITER:
#endif
if (!aot_frame_store_value(
comp_ctx, param_values[i], VALUE_TYPE_GC_REF,
func_ctx->cur_frame,
offset_of_local_in_outs_area(comp_ctx, n)))
return false;
if (comp_ctx->pointer_size == sizeof(uint64))
n++;
break;
#endif
default:
bh_assert(0);
break;
}
}
return true;
}
bool
aot_compile_op_call(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
uint32 func_idx, bool tail_call)
{
uint32 import_func_count = comp_ctx->comp_data->import_func_count;
AOTImportFunc *import_funcs = comp_ctx->comp_data->import_funcs;
uint32 func_count = comp_ctx->func_ctx_count, param_cell_num = 0;
uint32 ext_ret_cell_num = 0, cell_num = 0;
AOTFuncContext **func_ctxes = comp_ctx->func_ctxes;
AOTFuncType *func_type;
LLVMTypeRef *param_types = NULL, ret_type;
LLVMTypeRef ext_ret_ptr_type;
LLVMValueRef *param_values = NULL, value_ret = NULL, func;
LLVMValueRef import_func_idx, res;
LLVMValueRef ext_ret, ext_ret_ptr, ext_ret_idx;
#if WASM_ENABLE_AOT_STACK_FRAME != 0
LLVMValueRef func_idx_ref;
#endif
int32 i, j = 0, param_count, result_count, ext_ret_count;
uint64 total_size;
uint8 wasm_ret_type;
uint8 *ext_ret_types = NULL;
const char *signature = NULL;
bool ret = false;
char buf[32];
bool quick_invoke_c_api_import = false;
/* Check function index */
if (func_idx >= import_func_count + func_count) {
aot_set_last_error("Function index out of range.");
return false;
}
/* Get function type */
if (func_idx < import_func_count) {
func_type = import_funcs[func_idx].func_type;
signature = import_funcs[func_idx].signature;
}
else {
func_type =
func_ctxes[func_idx - import_func_count]->aot_func->func_type;
}
aot_estimate_and_record_stack_usage_for_function_call(comp_ctx, func_ctx,
func_type);
/* Commit stack operands, sp and ip */
if (comp_ctx->aot_frame) {
if (comp_ctx->enable_gc && !aot_gen_commit_values(comp_ctx->aot_frame))
return false;
/* Commit sp if gc is enabled and commit ip for func call */
if (!aot_gen_commit_sp_ip(comp_ctx->aot_frame, comp_ctx->enable_gc,
true))
return false;
}
/* Insert suspend check point */
if (comp_ctx->enable_thread_mgr) {
if (!check_suspend_flags(comp_ctx, func_ctx, true))
return false;
}
#if WASM_ENABLE_AOT_STACK_FRAME != 0
if (comp_ctx->aux_stack_frame_type) {
if (func_idx < import_func_count
&& comp_ctx->call_stack_features.frame_per_function) {
INT_CONST(func_idx_ref, func_idx, I32_TYPE, true);
if (!aot_alloc_frame_per_function_frame_for_aot_func(
comp_ctx, func_ctx, func_idx_ref)) {
return false;
}
}
else if (!comp_ctx->call_stack_features.frame_per_function) {
if (comp_ctx->aux_stack_frame_type
!= AOT_STACK_FRAME_TYPE_STANDARD) {
aot_set_last_error("unsupported mode");
return false;
}
if (!alloc_frame_for_aot_func(comp_ctx, func_ctx, func_idx)) {
return false;
}
}
}
#endif
/* Get param cell number */
param_cell_num = func_type->param_cell_num;
/* Allocate memory for parameters.
* Parameters layout:
* - exec env
* - wasm function's parameters
* - extra results'(except the first one) addresses
*/
param_count = (int32)func_type->param_count;
result_count = (int32)func_type->result_count;
ext_ret_count = result_count > 1 ? result_count - 1 : 0;
total_size =
sizeof(LLVMValueRef) * (uint64)(param_count + 1 + ext_ret_count);
if (total_size >= UINT32_MAX
|| !(param_values = wasm_runtime_malloc((uint32)total_size))) {
aot_set_last_error("allocate memory failed.");
return false;
}
/* First parameter is exec env */
param_values[j++] = func_ctx->exec_env;
/* Pop parameters from stack */
for (i = param_count - 1; i >= 0; i--)
POP(param_values[i + j], func_type->types[i]);
/* Set parameters for multiple return values, the first return value
is returned by function return value, and the other return values
are returned by function parameters with pointer types */
if (ext_ret_count > 0) {
ext_ret_types = func_type->types + param_count + 1;
ext_ret_cell_num = wasm_get_cell_num(ext_ret_types, ext_ret_count);
if (ext_ret_cell_num > 64) {
aot_set_last_error("prepare extra results's return "
"address arguments failed: "
"maximum 64 parameter cell number supported.");
goto fail;
}
for (i = 0; i < ext_ret_count; i++) {
if (!(ext_ret_idx = I32_CONST(cell_num))
|| !(ext_ret_ptr_type =
LLVMPointerType(TO_LLVM_TYPE(ext_ret_types[i]), 0))) {
aot_set_last_error("llvm add const or pointer type failed.");
goto fail;
}
snprintf(buf, sizeof(buf), "ext_ret%d_ptr", i);
if (!(ext_ret_ptr = LLVMBuildInBoundsGEP2(
comp_ctx->builder, I32_TYPE, func_ctx->argv_buf,
&ext_ret_idx, 1, buf))) {
aot_set_last_error("llvm build GEP failed.");
goto fail;
}
snprintf(buf, sizeof(buf), "ext_ret%d_ptr_cast", i);
if (!(ext_ret_ptr = LLVMBuildBitCast(comp_ctx->builder, ext_ret_ptr,
ext_ret_ptr_type, buf))) {
aot_set_last_error("llvm build bit cast failed.");
goto fail;
}
param_values[param_count + 1 + i] = ext_ret_ptr;
cell_num += wasm_value_type_cell_num_internal(
ext_ret_types[i], comp_ctx->pointer_size);
}
}
if (func_idx < import_func_count) {
if (comp_ctx->aux_stack_frame_type == AOT_STACK_FRAME_TYPE_STANDARD
&& !commit_params_to_frame_of_import_func(
comp_ctx, func_ctx, func_type, param_values + 1)) {
goto fail;
}
if (!(import_func_idx = I32_CONST(func_idx))) {
aot_set_last_error("llvm build inbounds gep failed.");
goto fail;
}
/* Initialize parameter types of the LLVM function */
total_size = sizeof(LLVMTypeRef) * (uint64)(param_count + 1);
if (total_size >= UINT32_MAX
|| !(param_types = wasm_runtime_malloc((uint32)total_size))) {
aot_set_last_error("allocate memory failed.");
goto fail;
}
j = 0;
param_types[j++] = comp_ctx->exec_env_type;
for (i = 0; i < param_count; i++, j++) {
param_types[j] = TO_LLVM_TYPE(func_type->types[i]);
/* If the signature can be gotten, e.g. the signature of the builtin
native libraries, just check the app offset and buf size, and
then convert app offset to native addr and call the native func
directly, no need to call aot_invoke_native to call it */
if (signature) {
LLVMValueRef native_addr, native_addr_size;
if (signature[i + 1] == '*' || signature[i + 1] == '$') {
param_types[j] = INT8_PTR_TYPE;
}
if (signature[i + 1] == '*') {
if (signature[i + 2] == '~')
native_addr_size = param_values[i + 2];
else
native_addr_size = I64_CONST(1);
if (!(native_addr_size = LLVMBuildZExtOrBitCast(
comp_ctx->builder, native_addr_size, I64_TYPE,
"native_addr_size_i64"))) {
aot_set_last_error("llvm build zextOrBitCast failed.");
goto fail;
}
if (!(param_values[j] = LLVMBuildZExtOrBitCast(
comp_ctx->builder, param_values[j], I64_TYPE,
"native_addr_i64"))) {
aot_set_last_error("llvm build zextOrBitCast failed.");
goto fail;
}
if (!check_app_addr_and_convert(
comp_ctx, func_ctx, false, param_values[j],
native_addr_size, &native_addr)) {
goto fail;
}
param_values[j] = native_addr;
}
else if (signature[i + 1] == '$') {
native_addr_size = I64_ZERO;
if (!(param_values[j] = LLVMBuildZExtOrBitCast(
comp_ctx->builder, param_values[j], I64_TYPE,
"native_addr_i64"))) {
aot_set_last_error("llvm build zextOrBitCast failed.");
goto fail;
}
if (!check_app_addr_and_convert(
comp_ctx, func_ctx, true, param_values[j],
native_addr_size, &native_addr)) {
goto fail;
}
param_values[j] = native_addr;
}
}
}
if (func_type->result_count) {
wasm_ret_type = func_type->types[func_type->param_count];
ret_type = TO_LLVM_TYPE(wasm_ret_type);
}
else {
wasm_ret_type = VALUE_TYPE_VOID;
ret_type = VOID_TYPE;
}
if (!signature) {
if (comp_ctx->quick_invoke_c_api_import) {
uint32 buf_size_needed =
sizeof(wasm_val_t) * (param_count + result_count);
/* length of exec_env->argv_buf is 64 */
if (buf_size_needed < sizeof(uint32) * 64) {
for (i = 0; i < param_count + result_count; i++) {
/* Only support i32/i64/f32/f64 now */
if (!(func_type->types[i] == VALUE_TYPE_I32
|| func_type->types[i] == VALUE_TYPE_I64
|| func_type->types[i] == VALUE_TYPE_F32
|| func_type->types[i] == VALUE_TYPE_F64))
break;
}
if (i == param_count + result_count)
quick_invoke_c_api_import = true;
}
}
if (quick_invoke_c_api_import) {
if (!call_aot_invoke_c_api_native(comp_ctx, func_ctx, func_idx,
func_type, param_values + 1))
goto fail;
}
else {
/* call aot_invoke_native() */
if (!call_aot_invoke_native_func(
comp_ctx, func_ctx, import_func_idx, func_type,
param_types + 1, param_values + 1, param_count,
param_cell_num, ret_type, wasm_ret_type, &value_ret,
&res))
goto fail;
/* Check whether there was exception thrown when executing
the function */
if ((comp_ctx->enable_bound_check || is_win_platform(comp_ctx))
&& !check_call_return(comp_ctx, func_ctx, res))
goto fail;
}
}
else { /* call native func directly */
LLVMTypeRef native_func_type, func_ptr_type;
LLVMValueRef func_ptr;
if (!(native_func_type = LLVMFunctionType(
ret_type, param_types, param_count + 1, false))) {
aot_set_last_error("llvm add function type failed.");
goto fail;
}
if (!(func_ptr_type = LLVMPointerType(native_func_type, 0))) {
aot_set_last_error("create LLVM function type failed.");
goto fail;
}
/* Load function pointer */
if (!(func_ptr = LLVMBuildInBoundsGEP2(
comp_ctx->builder, OPQ_PTR_TYPE, func_ctx->func_ptrs,
&import_func_idx, 1, "native_func_ptr_tmp"))) {
aot_set_last_error("llvm build inbounds gep failed.");
goto fail;
}
if (!(func_ptr = LLVMBuildLoad2(comp_ctx->builder, OPQ_PTR_TYPE,
func_ptr, "native_func_ptr"))) {
aot_set_last_error("llvm build load failed.");
goto fail;
}
if (!(func = LLVMBuildBitCast(comp_ctx->builder, func_ptr,
func_ptr_type, "native_func"))) {
aot_set_last_error("llvm bit cast failed.");
goto fail;
}
/* Call the function */
if (!(value_ret = LLVMBuildCall2(
comp_ctx->builder, native_func_type, func, param_values,
(uint32)param_count + 1 + ext_ret_count,
(func_type->result_count > 0 ? "call" : "")))) {
aot_set_last_error("LLVM build call failed.");
goto fail;
}
/* Check whether there was exception thrown when executing
the function */
if (!check_exception_thrown(comp_ctx, func_ctx)) {
goto fail;
}
}
}
else {
#if LLVM_VERSION_MAJOR >= 14
LLVMTypeRef llvm_func_type;
#endif
if (comp_ctx->is_indirect_mode) {
LLVMTypeRef func_ptr_type;
if (!(func_ptr_type = LLVMPointerType(
func_ctxes[func_idx - import_func_count]->func_type,
0))) {
aot_set_last_error("construct func ptr type failed.");
goto fail;
}
if (!(func = aot_get_func_from_table(comp_ctx, func_ctx->func_ptrs,
func_ptr_type, func_idx))) {
goto fail;
}
}
else {
if (func_ctxes[func_idx - import_func_count] == func_ctx) {
/* recursive call */
func = func_ctx->precheck_func;
}
else {
if (!comp_ctx->is_jit_mode) {
func =
func_ctxes[func_idx - import_func_count]->precheck_func;
}
else {
#if !(WASM_ENABLE_FAST_JIT != 0 && WASM_ENABLE_LAZY_JIT != 0)
func =
func_ctxes[func_idx - import_func_count]->precheck_func;
#else
/* JIT tier-up, load func ptr from func_ptrs[func_idx] */
LLVMValueRef func_ptr, func_idx_const;
LLVMTypeRef func_ptr_type;
if (!(func_idx_const = I32_CONST(func_idx))) {
aot_set_last_error("llvm build const failed.");
goto fail;
}
if (!(func_ptr = LLVMBuildInBoundsGEP2(
comp_ctx->builder, OPQ_PTR_TYPE,
func_ctx->func_ptrs, &func_idx_const, 1,
"func_ptr_tmp"))) {
aot_set_last_error("llvm build inbounds gep failed.");
goto fail;
}
if (!(func_ptr =
LLVMBuildLoad2(comp_ctx->builder, OPQ_PTR_TYPE,
func_ptr, "func_ptr"))) {
aot_set_last_error("llvm build load failed.");
goto fail;
}
if (!(func_ptr_type = LLVMPointerType(
func_ctxes[func_idx - import_func_count]
->func_type,
0))) {
aot_set_last_error("construct func ptr type failed.");
goto fail;
}
if (!(func = LLVMBuildBitCast(comp_ctx->builder, func_ptr,
func_ptr_type,
"indirect_func"))) {
aot_set_last_error("llvm build bit cast failed.");
goto fail;
}
#endif /* end of !(WASM_ENABLE_FAST_JIT != 0 && WASM_ENABLE_LAZY_JIT != 0) */
}
}
}
#if LLVM_VERSION_MAJOR >= 14
llvm_func_type = func_ctxes[func_idx - import_func_count]->func_type;
#endif
/* Call the function */
if (!(value_ret = LLVMBuildCall2(
comp_ctx->builder, llvm_func_type, func, param_values,
(uint32)param_count + 1 + ext_ret_count,
(func_type->result_count > 0 ? "call" : "")))) {
aot_set_last_error("LLVM build call failed.");
goto fail;
}
if (tail_call)
LLVMSetTailCall(value_ret, true);
/* Check whether there was exception thrown when executing
the function */
if (!tail_call
&& (comp_ctx->enable_bound_check || is_win_platform(comp_ctx))
&& !check_exception_thrown(comp_ctx, func_ctx))
goto fail;
}
if (func_type->result_count > 0 && !quick_invoke_c_api_import) {
/* Push the first result to stack */
PUSH(value_ret, func_type->types[func_type->param_count]);
/* Load extra result from its address and push to stack */
for (i = 0; i < ext_ret_count; i++) {
snprintf(buf, sizeof(buf), "func%d_ext_ret%d", func_idx, i);
if (!(ext_ret = LLVMBuildLoad2(
comp_ctx->builder, TO_LLVM_TYPE(ext_ret_types[i]),
param_values[1 + param_count + i], buf))) {
aot_set_last_error("llvm build load failed.");
goto fail;
}
LLVMSetAlignment(ext_ret, 4);
PUSH(ext_ret, ext_ret_types[i]);
}
}
#if WASM_ENABLE_AOT_STACK_FRAME != 0
if (comp_ctx->aux_stack_frame_type) {
if (func_idx < import_func_count
&& comp_ctx->call_stack_features.frame_per_function) {
if (!aot_free_frame_per_function_frame_for_aot_func(comp_ctx,
func_ctx)) {
goto fail;
}
}
else if (!comp_ctx->call_stack_features.frame_per_function) {
if (comp_ctx->aux_stack_frame_type
!= AOT_STACK_FRAME_TYPE_STANDARD) {
aot_set_last_error("unsupported mode");
}
if (!free_frame_for_aot_func(comp_ctx, func_ctx)) {
goto fail;
}
}
}
#endif
/* Insert suspend check point */
if (comp_ctx->enable_thread_mgr) {
if (!check_suspend_flags(comp_ctx, func_ctx, false))
goto fail;
}
ret = true;
fail:
if (param_types)
wasm_runtime_free(param_types);
if (param_values)
wasm_runtime_free(param_values);
return ret;
}
#if WASM_ENABLE_GC != 0
static LLVMValueRef
call_aot_func_type_is_super_of_func(AOTCompContext *comp_ctx,
AOTFuncContext *func_ctx,
LLVMValueRef type_idx1,
LLVMValueRef type_idx2)
{
LLVMValueRef param_values[3], ret_value, value, func;
LLVMTypeRef param_types[3], ret_type, func_type, func_ptr_type;
param_types[0] = comp_ctx->aot_inst_type;
param_types[1] = I32_TYPE;
param_types[2] = I32_TYPE;
ret_type = INT8_TYPE;
#if WASM_ENABLE_JIT != 0
if (comp_ctx->is_jit_mode)
GET_AOT_FUNCTION(llvm_jit_func_type_is_super_of, 3);
else
#endif
GET_AOT_FUNCTION(aot_func_type_is_super_of, 3);
param_values[0] = func_ctx->aot_inst;
param_values[1] = type_idx1;
param_values[2] = type_idx2;
if (!(ret_value =
LLVMBuildCall2(comp_ctx->builder, func_type, func, param_values,
3, "call_aot_func_type_is_super_of"))) {
aot_set_last_error("llvm build call failed.");
return NULL;
}
if (!(ret_value = LLVMBuildICmp(comp_ctx->builder, LLVMIntEQ, ret_value,
I8_ZERO, "check_fail"))) {
aot_set_last_error("llvm build icmp failed.");
return NULL;
}
return ret_value;
fail:
return NULL;
}
#endif
static bool
call_aot_call_indirect_func(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
AOTFuncType *aot_func_type,
LLVMValueRef func_type_idx, LLVMValueRef table_idx,
LLVMValueRef table_elem_idx,
LLVMTypeRef *param_types,
LLVMValueRef *param_values, uint32 param_count,
uint32 param_cell_num, uint32 result_count,
uint8 *wasm_ret_types, LLVMValueRef *value_rets,
LLVMValueRef *p_res)
{
LLVMTypeRef func_type, func_ptr_type, func_param_types[6];
LLVMTypeRef ret_type, ret_ptr_type, elem_ptr_type;
LLVMValueRef func, ret_idx, ret_ptr, elem_idx, elem_ptr;
LLVMValueRef func_param_values[6], res = NULL;
char buf[32], *func_name = "aot_call_indirect";
uint32 i, cell_num = 0, ret_cell_num, argv_cell_num;
/* prepare function type of aot_call_indirect */
func_param_types[0] = comp_ctx->exec_env_type; /* exec_env */
func_param_types[1] = I32_TYPE; /* table_idx */
func_param_types[2] = I32_TYPE; /* table_elem_idx */
func_param_types[3] = I32_TYPE; /* argc */
func_param_types[4] = INT32_PTR_TYPE; /* argv */
if (!(func_type =
LLVMFunctionType(INT8_TYPE, func_param_types, 5, false))) {
aot_set_last_error("llvm add function type failed.");
return false;
}
/* prepare function pointer */
if (comp_ctx->is_jit_mode) {
if (!(func_ptr_type = LLVMPointerType(func_type, 0))) {
aot_set_last_error("create LLVM function type failed.");
return false;
}
/* JIT mode, call the function directly */
if (!(func = I64_CONST((uint64)(uintptr_t)llvm_jit_call_indirect))
|| !(func = LLVMConstIntToPtr(func, func_ptr_type))) {
aot_set_last_error("create LLVM value failed.");
return false;
}
}
else if (comp_ctx->is_indirect_mode) {
int32 func_index;
if (!(func_ptr_type = LLVMPointerType(func_type, 0))) {
aot_set_last_error("create LLVM function type failed.");
return false;
}
func_index = aot_get_native_symbol_index(comp_ctx, func_name);
if (func_index < 0) {
return false;
}
if (!(func = aot_get_func_from_table(comp_ctx, func_ctx->native_symbol,
func_ptr_type, func_index))) {
return false;
}
}
else {
if (!(func = LLVMGetNamedFunction(func_ctx->module, func_name))
&& !(func =
LLVMAddFunction(func_ctx->module, func_name, func_type))) {
aot_set_last_error("add LLVM function failed.");
return false;
}
}
ret_cell_num = wasm_get_cell_num(wasm_ret_types, result_count);
argv_cell_num =
param_cell_num > ret_cell_num ? param_cell_num : ret_cell_num;
if (argv_cell_num > 64) {
aot_set_last_error("prepare native arguments failed: "
"maximum 64 parameter cell number supported.");
return false;
}
/* prepare frame_lp */
for (i = 0; i < param_count; i++) {
if (!(elem_idx = I32_CONST(cell_num))
|| !(elem_ptr_type = LLVMPointerType(param_types[i], 0))) {
aot_set_last_error("llvm add const or pointer type failed.");
return false;
}
snprintf(buf, sizeof(buf), "%s%d", "elem", i);
if (!(elem_ptr =
LLVMBuildInBoundsGEP2(comp_ctx->builder, I32_TYPE,
func_ctx->argv_buf, &elem_idx, 1, buf))
|| !(elem_ptr = LLVMBuildBitCast(comp_ctx->builder, elem_ptr,
elem_ptr_type, buf))) {
aot_set_last_error("llvm build bit cast failed.");
return false;
}
if (!(res = LLVMBuildStore(comp_ctx->builder, param_values[i],
elem_ptr))) {
aot_set_last_error("llvm build store failed.");
return false;
}
LLVMSetAlignment(res, 1);
cell_num += wasm_value_type_cell_num_internal(aot_func_type->types[i],
comp_ctx->pointer_size);
}
func_param_values[0] = func_ctx->exec_env;
func_param_values[1] = table_idx;
func_param_values[2] = table_elem_idx;
func_param_values[3] = I32_CONST(param_cell_num);
func_param_values[4] = func_ctx->argv_buf;
if (!func_param_values[3]) {
aot_set_last_error("llvm create const failed.");
return false;
}
/* call aot_call_indirect() function */
if (!(res = LLVMBuildCall2(comp_ctx->builder, func_type, func,
func_param_values, 5, "res"))) {
aot_set_last_error("llvm build call failed.");
return false;
}
/* get function result values */
cell_num = 0;
for (i = 0; i < result_count; i++) {
ret_type = TO_LLVM_TYPE(wasm_ret_types[i]);
if (!(ret_idx = I32_CONST(cell_num))
|| !(ret_ptr_type = LLVMPointerType(ret_type, 0))) {
aot_set_last_error("llvm add const or pointer type failed.");
return false;
}
snprintf(buf, sizeof(buf), "argv_ret%d", i);
if (!(ret_ptr =
LLVMBuildInBoundsGEP2(comp_ctx->builder, I32_TYPE,
func_ctx->argv_buf, &ret_idx, 1, buf))
|| !(ret_ptr = LLVMBuildBitCast(comp_ctx->builder, ret_ptr,
ret_ptr_type, buf))) {
aot_set_last_error("llvm build GEP or bit cast failed.");
return false;
}
snprintf(buf, sizeof(buf), "ret%d", i);
if (!(value_rets[i] =
LLVMBuildLoad2(comp_ctx->builder, ret_type, ret_ptr, buf))) {
aot_set_last_error("llvm build load failed.");
return false;
}
LLVMSetAlignment(value_rets[i], 4);
cell_num += wasm_value_type_cell_num_internal(wasm_ret_types[i],
comp_ctx->pointer_size);
}
*p_res = res;
return true;
}
bool
aot_compile_op_call_indirect(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
uint32 type_idx, uint32 tbl_idx)
{
AOTFuncType *func_type;
LLVMValueRef tbl_idx_value, elem_idx, func_idx;
LLVMValueRef table_elem_base, table_elem_addr, table_elem;
LLVMValueRef ftype_idx_ptr, ftype_idx, ftype_idx_const;
LLVMValueRef cmp_func_obj, cmp_elem_idx, cmp_func_idx, cmp_ftype_idx;
LLVMValueRef func, func_ptr, table_size_const;
LLVMValueRef ext_ret_offset, ext_ret_ptr, ext_ret, res;
LLVMValueRef *param_values = NULL, *value_rets = NULL;
LLVMValueRef *result_phis = NULL, value_ret, import_func_count;
#if WASM_ENABLE_MEMORY64 != 0
LLVMValueRef u32_max, u32_cmp_result = NULL;
#endif
LLVMTypeRef *param_types = NULL, ret_type;
LLVMTypeRef llvm_func_type, llvm_func_ptr_type;
LLVMTypeRef ext_ret_ptr_type;
LLVMBasicBlockRef check_func_obj_succ, check_elem_idx_succ,
check_ftype_idx_succ;
LLVMBasicBlockRef check_func_idx_succ, block_return, block_curr;
LLVMBasicBlockRef block_call_import, block_call_non_import;
LLVMValueRef offset;
uint32 total_param_count, func_param_count, func_result_count;
uint32 ext_cell_num, param_cell_num, i, j;
uint8 wasm_ret_type, *wasm_ret_types;
uint64 total_size;
char buf[32];
bool ret = false;
/* Check function type index */
if (type_idx >= comp_ctx->comp_data->type_count) {
aot_set_last_error("function type index out of range");
return false;
}
if (!comp_ctx->enable_gc) {
/* Find the equivalent function type whose type index is the smallest:
the callee function's type index is also converted to the smallest
one in wasm loader, so we can just check whether the two type indexes
are equal (the type index of call_indirect opcode and callee func),
we don't need to check whether the whole function types are equal,
including param types and result types. */
type_idx = wasm_get_smallest_type_idx(
(WASMTypePtr *)comp_ctx->comp_data->types,
comp_ctx->comp_data->type_count, type_idx);
}
else {
/* Call aot_func_type_is_super_of to check whether the func type
provided in the bytecode is a super type of the func type of
the function to call */
}
ftype_idx_const = I32_CONST(type_idx);
CHECK_LLVM_CONST(ftype_idx_const);
func_type = (AOTFuncType *)comp_ctx->comp_data->types[type_idx];
aot_estimate_and_record_stack_usage_for_function_call(comp_ctx, func_ctx,
func_type);
/* Commit stack operands, sp and ip */
if (comp_ctx->aot_frame) {
if (comp_ctx->enable_gc && !aot_gen_commit_values(comp_ctx->aot_frame))
return false;
/* Commit sp if gc is enabled and always commit ip for call_indirect */
if (!aot_gen_commit_sp_ip(comp_ctx->aot_frame, comp_ctx->enable_gc,
true))
return false;
}
/* Insert suspend check point */
if (comp_ctx->enable_thread_mgr) {
if (!check_suspend_flags(comp_ctx, func_ctx, true))
return false;
}
func_param_count = func_type->param_count;
func_result_count = func_type->result_count;
POP_TBL_ELEM_IDX(elem_idx);
/* get the cur size of the table instance */
if (!(offset = I32_CONST(get_tbl_inst_offset(comp_ctx, func_ctx, tbl_idx)
+ offsetof(AOTTableInstance, cur_size)))) {
HANDLE_FAILURE("LLVMConstInt");
goto fail;
}
if (!(table_size_const = LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE,
func_ctx->aot_inst, &offset,
1, "cur_size_i8p"))) {
HANDLE_FAILURE("LLVMBuildGEP");
goto fail;
}
if (!(table_size_const =
LLVMBuildBitCast(comp_ctx->builder, table_size_const,
INT32_PTR_TYPE, "cur_size_i32p"))) {
HANDLE_FAILURE("LLVMBuildBitCast");
goto fail;
}
if (!(table_size_const = LLVMBuildLoad2(comp_ctx->builder, I32_TYPE,
table_size_const, "cur_size"))) {
HANDLE_FAILURE("LLVMBuildLoad");
goto fail;
}
#if WASM_ENABLE_MEMORY64 != 0
/* Check if elem index >= UINT32_MAX */
if (IS_TABLE64(tbl_idx)) {
if (!(u32_max = I64_CONST(UINT32_MAX))) {
aot_set_last_error("llvm build const failed");
goto fail;
}
if (!(u32_cmp_result =
LLVMBuildICmp(comp_ctx->builder, LLVMIntUGE, elem_idx,
u32_max, "cmp_elem_idx_u32_max"))) {
aot_set_last_error("llvm build icmp failed.");
goto fail;
}
if (!(elem_idx = LLVMBuildTrunc(comp_ctx->builder, elem_idx, I32_TYPE,
"elem_idx_i32"))) {
aot_set_last_error("llvm build trunc failed.");
goto fail;
}
}
#endif
/* Check if (uint32)elem index >= table size */
if (!(cmp_elem_idx = LLVMBuildICmp(comp_ctx->builder, LLVMIntUGE, elem_idx,
table_size_const, "cmp_elem_idx"))) {
aot_set_last_error("llvm build icmp failed.");
goto fail;
}
#if WASM_ENABLE_MEMORY64 != 0
if (IS_TABLE64(tbl_idx)) {
if (!(cmp_elem_idx =
LLVMBuildOr(comp_ctx->builder, cmp_elem_idx, u32_cmp_result,
"larger_than_u32_max_or_cur_size"))) {
aot_set_last_error("llvm build or failed.");
goto fail;
}
}
#endif
/* Throw exception if elem index >= table size or elem index >= UINT32_MAX
*/
if (!(check_elem_idx_succ = LLVMAppendBasicBlockInContext(
comp_ctx->context, func_ctx->func, "check_elem_idx_succ"))) {
aot_set_last_error("llvm add basic block failed.");
goto fail;
}
LLVMMoveBasicBlockAfter(check_elem_idx_succ,
LLVMGetInsertBlock(comp_ctx->builder));
if (!(aot_emit_exception(comp_ctx, func_ctx, EXCE_UNDEFINED_ELEMENT, true,
cmp_elem_idx, check_elem_idx_succ)))
goto fail;
/* load data as i32* */
if (!(offset = I32_CONST(get_tbl_inst_offset(comp_ctx, func_ctx, tbl_idx)
+ offsetof(AOTTableInstance, elems)))) {
HANDLE_FAILURE("LLVMConstInt");
goto fail;
}
if (!(table_elem_base = LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE,
func_ctx->aot_inst, &offset,
1, "table_elem_base_i8p"))) {
aot_set_last_error("llvm build add failed.");
goto fail;
}
/* Load function index */
if (comp_ctx->enable_gc) {
/* table elem is func_obj when gc is enabled */
if (!(table_elem_base =
LLVMBuildBitCast(comp_ctx->builder, table_elem_base,
GC_REF_PTR_TYPE, "table_elem_base"))) {
HANDLE_FAILURE("LLVMBuildBitCast");
goto fail;
}
if (!(table_elem_addr = LLVMBuildInBoundsGEP2(
comp_ctx->builder, GC_REF_TYPE, table_elem_base, &elem_idx, 1,
"table_elem_addr"))) {
HANDLE_FAILURE("LLVMBuildNUWAdd");
goto fail;
}
if (!(table_elem = LLVMBuildLoad2(comp_ctx->builder, GC_REF_TYPE,
table_elem_addr, "table_elem"))) {
aot_set_last_error("llvm build load failed.");
goto fail;
}
/* Check if func object is NULL */
if (!(cmp_func_obj = LLVMBuildIsNull(comp_ctx->builder, table_elem,
"cmp_func_obj"))) {
aot_set_last_error("llvm build isnull failed.");
goto fail;
}
/* Throw exception if func object is NULL */
if (!(check_func_obj_succ = LLVMAppendBasicBlockInContext(
comp_ctx->context, func_ctx->func, "check_func_obj_succ"))) {
aot_set_last_error("llvm add basic block failed.");
goto fail;
}
LLVMMoveBasicBlockAfter(check_func_obj_succ,
LLVMGetInsertBlock(comp_ctx->builder));
if (!(aot_emit_exception(comp_ctx, func_ctx, EXCE_UNINITIALIZED_ELEMENT,
true, cmp_func_obj, check_func_obj_succ)))
goto fail;
/* Get the func idx bound of the WASMFuncObject, the offset may be
* different in 32-bit runtime and 64-bit runtime since WASMObjectHeader
* is uintptr_t. Use comp_ctx->pointer_size as the
* offsetof(WASMFuncObject, func_idx_bound)
*/
if (!(offset = I32_CONST(comp_ctx->pointer_size))) {
HANDLE_FAILURE("LLVMConstInt");
goto fail;
}
if (!(func_idx = LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE,
table_elem, &offset, 1,
"func_idx_bound_i8p"))) {
HANDLE_FAILURE("LLVMBuildGEP");
goto fail;
}
if (!(func_idx =
LLVMBuildBitCast(comp_ctx->builder, func_idx, INT32_PTR_TYPE,
"func_idx_bound_i32p"))) {
HANDLE_FAILURE("LLVMBuildBitCast");
goto fail;
}
if (!(func_idx = LLVMBuildLoad2(comp_ctx->builder, I32_TYPE, func_idx,
"func_idx_bound"))) {
HANDLE_FAILURE("LLVMBuildLoad");
goto fail;
}
}
else {
if (!(table_elem_base =
LLVMBuildBitCast(comp_ctx->builder, table_elem_base,
INTPTR_T_PTR_TYPE, "table_elem_base"))) {
HANDLE_FAILURE("LLVMBuildBitCast");
goto fail;
}
if (!(table_elem_addr = LLVMBuildInBoundsGEP2(
comp_ctx->builder, INTPTR_T_TYPE, table_elem_base, &elem_idx,
1, "table_elem_addr"))) {
HANDLE_FAILURE("LLVMBuildNUWAdd");
goto fail;
}
if (!(func_idx = LLVMBuildLoad2(comp_ctx->builder, INTPTR_T_TYPE,
table_elem_addr, "func_idx"))) {
aot_set_last_error("llvm build load failed.");
goto fail;
}
if (!(func_idx = LLVMBuildIntCast2(comp_ctx->builder, func_idx,
I32_TYPE, true, "func_idx_i32"))) {
aot_set_last_error("llvm build int cast failed.");
goto fail;
}
/* Check if func_idx == -1 */
if (!(cmp_func_idx =
LLVMBuildICmp(comp_ctx->builder, LLVMIntEQ, func_idx,
I32_NEG_ONE, "cmp_func_idx"))) {
aot_set_last_error("llvm build icmp failed.");
goto fail;
}
/* Throw exception if func_idx == -1 */
if (!(check_func_idx_succ = LLVMAppendBasicBlockInContext(
comp_ctx->context, func_ctx->func, "check_func_idx_succ"))) {
aot_set_last_error("llvm add basic block failed.");
goto fail;
}
LLVMMoveBasicBlockAfter(check_func_idx_succ,
LLVMGetInsertBlock(comp_ctx->builder));
if (!(aot_emit_exception(comp_ctx, func_ctx, EXCE_UNINITIALIZED_ELEMENT,
true, cmp_func_idx, check_func_idx_succ)))
goto fail;
}
/* Load function type index */
if (!(ftype_idx_ptr = LLVMBuildInBoundsGEP2(
comp_ctx->builder, I32_TYPE, func_ctx->func_type_indexes,
&func_idx, 1, "ftype_idx_ptr"))) {
aot_set_last_error("llvm build inbounds gep failed.");
goto fail;
}
if (!(ftype_idx = LLVMBuildLoad2(comp_ctx->builder, I32_TYPE, ftype_idx_ptr,
"ftype_idx"))) {
aot_set_last_error("llvm build load failed.");
goto fail;
}
#if WASM_ENABLE_GC != 0
if (comp_ctx->enable_gc) {
if (!(cmp_ftype_idx = call_aot_func_type_is_super_of_func(
comp_ctx, func_ctx, ftype_idx_const, ftype_idx))) {
goto fail;
}
}
else
#endif
{
/* Check if function type index not equal */
if (!(cmp_ftype_idx =
LLVMBuildICmp(comp_ctx->builder, LLVMIntNE, ftype_idx,
ftype_idx_const, "cmp_ftype_idx"))) {
aot_set_last_error("llvm build icmp failed.");
goto fail;
}
}
/* Throw exception if ftype_idx != ftype_idx_const */
if (!(check_ftype_idx_succ = LLVMAppendBasicBlockInContext(
comp_ctx->context, func_ctx->func, "check_ftype_idx_succ"))) {
aot_set_last_error("llvm add basic block failed.");
goto fail;
}
LLVMMoveBasicBlockAfter(check_ftype_idx_succ,
LLVMGetInsertBlock(comp_ctx->builder));
if (!(aot_emit_exception(comp_ctx, func_ctx,
EXCE_INVALID_FUNCTION_TYPE_INDEX, true,
cmp_ftype_idx, check_ftype_idx_succ)))
goto fail;
/* Initialize parameter types of the LLVM function */
total_param_count = 1 + func_param_count;
/* Extra function results' addresses (except the first one) are
appended to aot function parameters. */
if (func_result_count > 1)
total_param_count += func_result_count - 1;
total_size = sizeof(LLVMTypeRef) * (uint64)total_param_count;
if (total_size >= UINT32_MAX
|| !(param_types = wasm_runtime_malloc((uint32)total_size))) {
aot_set_last_error("allocate memory failed.");
goto fail;
}
/* Prepare param types */
j = 0;
param_types[j++] = comp_ctx->exec_env_type;
for (i = 0; i < func_param_count; i++)
param_types[j++] = TO_LLVM_TYPE(func_type->types[i]);
for (i = 1; i < func_result_count; i++, j++) {
param_types[j] = TO_LLVM_TYPE(func_type->types[func_param_count + i]);
if (!(param_types[j] = LLVMPointerType(param_types[j], 0))) {
aot_set_last_error("llvm get pointer type failed.");
goto fail;
}
}
/* Resolve return type of the LLVM function */
if (func_result_count) {
wasm_ret_type = func_type->types[func_param_count];
ret_type = TO_LLVM_TYPE(wasm_ret_type);
}
else {
wasm_ret_type = VALUE_TYPE_VOID;
ret_type = VOID_TYPE;
}
/* Allocate memory for parameters */
total_size = sizeof(LLVMValueRef) * (uint64)total_param_count;
if (total_size >= UINT32_MAX
|| !(param_values = wasm_runtime_malloc((uint32)total_size))) {
aot_set_last_error("allocate memory failed.");
goto fail;
}
/* First parameter is exec env */
j = 0;
param_values[j++] = func_ctx->exec_env;
/* Pop parameters from stack */
for (i = func_param_count - 1; (int32)i >= 0; i--)
POP(param_values[i + j], func_type->types[i]);
/* Prepare extra parameters */
ext_cell_num = 0;
for (i = 1; i < func_result_count; i++) {
ext_ret_offset = I32_CONST(ext_cell_num);
CHECK_LLVM_CONST(ext_ret_offset);
snprintf(buf, sizeof(buf), "ext_ret%d_ptr", i - 1);
if (!(ext_ret_ptr = LLVMBuildInBoundsGEP2(comp_ctx->builder, I32_TYPE,
func_ctx->argv_buf,
&ext_ret_offset, 1, buf))) {
aot_set_last_error("llvm build GEP failed.");
goto fail;
}
ext_ret_ptr_type = param_types[func_param_count + i];
snprintf(buf, sizeof(buf), "ext_ret%d_ptr_cast", i - 1);
if (!(ext_ret_ptr = LLVMBuildBitCast(comp_ctx->builder, ext_ret_ptr,
ext_ret_ptr_type, buf))) {
aot_set_last_error("llvm build bit cast failed.");
goto fail;
}
param_values[func_param_count + i] = ext_ret_ptr;
ext_cell_num += wasm_value_type_cell_num_internal(
func_type->types[func_param_count + i], comp_ctx->pointer_size);
}
if (ext_cell_num > 64) {
aot_set_last_error("prepare call-indirect arguments failed: "
"maximum 64 extra cell number supported.");
goto fail;
}
if (comp_ctx->aux_stack_frame_type
&& !comp_ctx->call_stack_features.frame_per_function) {
#if WASM_ENABLE_AOT_STACK_FRAME != 0
/* TODO: use current frame instead of allocating new frame
for WASM_OP_RETURN_CALL_INDIRECT */
if (!call_aot_alloc_frame_func(comp_ctx, func_ctx, func_idx))
goto fail;
#endif
}
/* Add basic blocks */
block_call_import = LLVMAppendBasicBlockInContext(
comp_ctx->context, func_ctx->func, "call_import");
block_call_non_import = LLVMAppendBasicBlockInContext(
comp_ctx->context, func_ctx->func, "call_non_import");
block_return = LLVMAppendBasicBlockInContext(comp_ctx->context,
func_ctx->func, "func_return");
if (!block_call_import || !block_call_non_import || !block_return) {
aot_set_last_error("llvm add basic block failed.");
goto fail;
}
LLVMMoveBasicBlockAfter(block_call_import,
LLVMGetInsertBlock(comp_ctx->builder));
LLVMMoveBasicBlockAfter(block_call_non_import, block_call_import);
LLVMMoveBasicBlockAfter(block_return, block_call_non_import);
import_func_count = I32_CONST(comp_ctx->comp_data->import_func_count);
CHECK_LLVM_CONST(import_func_count);
/* Check if func_idx < import_func_count */
if (!(cmp_func_idx = LLVMBuildICmp(comp_ctx->builder, LLVMIntULT, func_idx,
import_func_count, "cmp_func_idx"))) {
aot_set_last_error("llvm build icmp failed.");
goto fail;
}
/* If func_idx < import_func_count, jump to call import block,
else jump to call non-import block */
if (!LLVMBuildCondBr(comp_ctx->builder, cmp_func_idx, block_call_import,
block_call_non_import)) {
aot_set_last_error("llvm build cond br failed.");
goto fail;
}
/* Add result phis for return block */
LLVMPositionBuilderAtEnd(comp_ctx->builder, block_return);
if (func_result_count > 0) {
total_size = sizeof(LLVMValueRef) * (uint64)func_result_count;
if (total_size >= UINT32_MAX
|| !(result_phis = wasm_runtime_malloc((uint32)total_size))) {
aot_set_last_error("allocate memory failed.");
goto fail;
}
memset(result_phis, 0, (uint32)total_size);
for (i = 0; i < func_result_count; i++) {
LLVMTypeRef tmp_type =
TO_LLVM_TYPE(func_type->types[func_param_count + i]);
if (!(result_phis[i] =
LLVMBuildPhi(comp_ctx->builder, tmp_type, "phi"))) {
aot_set_last_error("llvm build phi failed.");
goto fail;
}
}
}
/* Translate call import block */
LLVMPositionBuilderAtEnd(comp_ctx->builder, block_call_import);
if (comp_ctx->aot_frame && comp_ctx->call_stack_features.frame_per_function
&& !aot_alloc_frame_per_function_frame_for_aot_func(comp_ctx, func_ctx,
func_idx)) {
goto fail;
}
if (comp_ctx->aux_stack_frame_type == AOT_STACK_FRAME_TYPE_STANDARD
&& !commit_params_to_frame_of_import_func(comp_ctx, func_ctx, func_type,
param_values + 1)) {
goto fail;
}
/* Allocate memory for result values */
if (func_result_count > 0) {
total_size = sizeof(LLVMValueRef) * (uint64)func_result_count;
if (total_size >= UINT32_MAX
|| !(value_rets = wasm_runtime_malloc((uint32)total_size))) {
aot_set_last_error("allocate memory failed.");
goto fail;
}
memset(value_rets, 0, (uint32)total_size);
}
param_cell_num = func_type->param_cell_num;
wasm_ret_types = func_type->types + func_type->param_count;
tbl_idx_value = I32_CONST(tbl_idx);
if (!tbl_idx_value) {
aot_set_last_error("llvm create const failed.");
goto fail;
}
if (!call_aot_call_indirect_func(
comp_ctx, func_ctx, func_type, ftype_idx, tbl_idx_value, elem_idx,
param_types + 1, param_values + 1, func_param_count, param_cell_num,
func_result_count, wasm_ret_types, value_rets, &res))
goto fail;
/* Check whether exception was thrown when executing the function */
if ((comp_ctx->enable_bound_check || is_win_platform(comp_ctx))
&& !check_call_return(comp_ctx, func_ctx, res))
goto fail;
if (comp_ctx->aot_frame && comp_ctx->call_stack_features.frame_per_function
&& !aot_free_frame_per_function_frame_for_aot_func(comp_ctx,
func_ctx)) {
goto fail;
}
block_curr = LLVMGetInsertBlock(comp_ctx->builder);
for (i = 0; i < func_result_count; i++) {
LLVMAddIncoming(result_phis[i], &value_rets[i], &block_curr, 1);
}
if (!LLVMBuildBr(comp_ctx->builder, block_return)) {
aot_set_last_error("llvm build br failed.");
goto fail;
}
/* Translate call non-import block */
LLVMPositionBuilderAtEnd(comp_ctx->builder, block_call_non_import);
/* Load function pointer */
if (!(func_ptr = LLVMBuildInBoundsGEP2(comp_ctx->builder, OPQ_PTR_TYPE,
func_ctx->func_ptrs, &func_idx, 1,
"func_ptr_tmp"))) {
aot_set_last_error("llvm build inbounds gep failed.");
goto fail;
}
if (!(func_ptr = LLVMBuildLoad2(comp_ctx->builder, OPQ_PTR_TYPE, func_ptr,
"func_ptr"))) {
aot_set_last_error("llvm build load failed.");
goto fail;
}
if (!(llvm_func_type =
LLVMFunctionType(ret_type, param_types, total_param_count, false))
|| !(llvm_func_ptr_type = LLVMPointerType(llvm_func_type, 0))) {
aot_set_last_error("llvm add function type failed.");
goto fail;
}
if (!(func = LLVMBuildBitCast(comp_ctx->builder, func_ptr,
llvm_func_ptr_type, "indirect_func"))) {
aot_set_last_error("llvm build bit cast failed.");
goto fail;
}
if (!(value_ret = LLVMBuildCall2(comp_ctx->builder, llvm_func_type, func,
param_values, total_param_count,
func_result_count > 0 ? "ret" : ""))) {
aot_set_last_error("llvm build call failed.");
goto fail;
}
/* Check whether exception was thrown when executing the function */
if ((comp_ctx->enable_bound_check || is_win_platform(comp_ctx))
&& !check_exception_thrown(comp_ctx, func_ctx))
goto fail;
if (func_result_count > 0) {
block_curr = LLVMGetInsertBlock(comp_ctx->builder);
/* Push the first result to stack */
LLVMAddIncoming(result_phis[0], &value_ret, &block_curr, 1);
/* Load extra result from its address and push to stack */
for (i = 1; i < func_result_count; i++) {
ret_type = TO_LLVM_TYPE(func_type->types[func_param_count + i]);
snprintf(buf, sizeof(buf), "ext_ret%d", i - 1);
if (!(ext_ret = LLVMBuildLoad2(comp_ctx->builder, ret_type,
param_values[func_param_count + i],
buf))) {
aot_set_last_error("llvm build load failed.");
goto fail;
}
LLVMSetAlignment(ext_ret, 4);
LLVMAddIncoming(result_phis[i], &ext_ret, &block_curr, 1);
}
}
if (!LLVMBuildBr(comp_ctx->builder, block_return)) {
aot_set_last_error("llvm build br failed.");
goto fail;
}
/* Translate function return block */
LLVMPositionBuilderAtEnd(comp_ctx->builder, block_return);
for (i = 0; i < func_result_count; i++) {
PUSH(result_phis[i], func_type->types[func_param_count + i]);
}
if (comp_ctx->aux_stack_frame_type
&& !comp_ctx->call_stack_features.frame_per_function) {
#if WASM_ENABLE_AOT_STACK_FRAME != 0
if (!free_frame_for_aot_func(comp_ctx, func_ctx))
goto fail;
#endif
}
/* Insert suspend check point */
if (comp_ctx->enable_thread_mgr) {
if (!check_suspend_flags(comp_ctx, func_ctx, false))
goto fail;
}
ret = true;
fail:
if (param_values)
wasm_runtime_free(param_values);
if (param_types)
wasm_runtime_free(param_types);
if (value_rets)
wasm_runtime_free(value_rets);
if (result_phis)
wasm_runtime_free(result_phis);
return ret;
}
bool
aot_compile_op_ref_null(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx)
{
if (comp_ctx->enable_gc)
PUSH_GC_REF(GC_REF_NULL);
else
PUSH_I32(REF_NULL);
return true;
fail:
return false;
}
bool
aot_compile_op_ref_is_null(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx)
{
LLVMValueRef lhs = NULL, res;
if (comp_ctx->enable_gc) {
POP_GC_REF(lhs);
if (!(res = LLVMBuildIsNull(comp_ctx->builder, lhs, "lhs is null"))) {
HANDLE_FAILURE("LLVMBuildIsNull");
goto fail;
}
}
else {
POP_I32(lhs);
if (!(res = LLVMBuildICmp(comp_ctx->builder, LLVMIntEQ, lhs, REF_NULL,
"cmp_w_null"))) {
HANDLE_FAILURE("LLVMBuildICmp");
goto fail;
}
}
if (!(res = LLVMBuildZExt(comp_ctx->builder, res, I32_TYPE, "r_i"))) {
HANDLE_FAILURE("LLVMBuildZExt");
goto fail;
}
PUSH_I32(res);
return true;
fail:
return false;
}
bool
aot_compile_op_ref_func(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
uint32 func_idx)
{
LLVMValueRef ref_idx;
#if WASM_ENABLE_GC != 0
LLVMValueRef gc_obj;
#endif
if (!(ref_idx = I32_CONST(func_idx))) {
HANDLE_FAILURE("LLVMConstInt");
goto fail;
}
#if WASM_ENABLE_GC != 0
if (comp_ctx->enable_gc) {
if (!aot_gen_commit_values(comp_ctx->aot_frame))
return false;
/* Commit sp and ip if gc is enabled */
if (!aot_gen_commit_sp_ip(comp_ctx->aot_frame, true, true))
return false;
if (!aot_call_aot_create_func_obj(comp_ctx, func_ctx, ref_idx,
&gc_obj)) {
goto fail;
}
PUSH_GC_REF(gc_obj);
}
else
#endif
{
PUSH_I32(ref_idx);
}
return true;
fail:
return false;
}
#if WASM_ENABLE_GC != 0
bool
aot_compile_op_call_ref(AOTCompContext *comp_ctx, AOTFuncContext *func_ctx,
uint32 type_idx, bool tail_call)
{
AOTFuncType *func_type;
LLVMValueRef func_obj, func_idx;
LLVMValueRef cmp_func_obj, cmp_func_idx;
LLVMValueRef func, func_ptr;
LLVMValueRef ext_ret_offset, ext_ret_ptr, ext_ret, res;
LLVMValueRef *param_values = NULL;
LLVMValueRef *result_phis = NULL, value_ret, import_func_count;
LLVMTypeRef *param_types = NULL, ret_type;
LLVMTypeRef llvm_func_type, llvm_func_ptr_type;
LLVMTypeRef ext_ret_ptr_type;
LLVMBasicBlockRef check_func_obj_succ, block_return, block_curr;
LLVMBasicBlockRef block_call_import, block_call_non_import;
LLVMValueRef offset;
uint32 total_param_count, func_param_count, func_result_count;
uint32 ext_cell_num, param_cell_num, i, j;
uint8 wasm_ret_type;
uint64 total_size;
char buf[32];
bool ret = false;
/* Check function type index */
bh_assert(type_idx < comp_ctx->comp_data->type_count);
func_type = (AOTFuncType *)comp_ctx->comp_data->types[type_idx];
aot_estimate_and_record_stack_usage_for_function_call(comp_ctx, func_ctx,
func_type);
func_param_count = func_type->param_count;
func_result_count = func_type->result_count;
param_cell_num = func_type->param_cell_num;
/* Commit stack operands, sp and ip to aot frame */
if (comp_ctx->aot_frame) {
/* Note that GC is enabled, no need to check it again */
if (!aot_gen_commit_values(comp_ctx->aot_frame))
return false;
/* Commit sp if gc is enabled and always commit ip for call_ref */
if (!aot_gen_commit_sp_ip(comp_ctx->aot_frame, true, true))
return false;
}
/* Insert suspend check point */
if (comp_ctx->enable_thread_mgr) {
if (!check_suspend_flags(comp_ctx, func_ctx, true))
return false;
}
POP_GC_REF(func_obj);
/* Check if func object is NULL */
if (!(cmp_func_obj =
LLVMBuildIsNull(comp_ctx->builder, func_obj, "cmp_func_obj"))) {
aot_set_last_error("llvm build isnull failed.");
goto fail;
}
/* Throw exception if func object is NULL */
if (!(check_func_obj_succ = LLVMAppendBasicBlockInContext(
comp_ctx->context, func_ctx->func, "check_func_obj_succ"))) {
aot_set_last_error("llvm add basic block failed.");
goto fail;
}
LLVMMoveBasicBlockAfter(check_func_obj_succ,
LLVMGetInsertBlock(comp_ctx->builder));
if (!(aot_emit_exception(comp_ctx, func_ctx, EXCE_NULL_FUNC_OBJ, true,
cmp_func_obj, check_func_obj_succ)))
goto fail;
/* Get the func idx bound of the WASMFuncObject, the offset may be
* different in 32-bit runtime and 64-bit runtime since WASMObjectHeader
* is uintptr_t. Use comp_ctx->pointer_size as the
* offsetof(WASMFuncObject, func_idx_bound) */
if (!(offset = I32_CONST(comp_ctx->pointer_size))) {
HANDLE_FAILURE("LLVMConstInt");
goto fail;
}
if (!(func_idx =
LLVMBuildInBoundsGEP2(comp_ctx->builder, INT8_TYPE, func_obj,
&offset, 1, "func_idx_bound_i8p"))) {
HANDLE_FAILURE("LLVMBuildGEP");
goto fail;
}
if (!(func_idx = LLVMBuildBitCast(comp_ctx->builder, func_idx,
INT32_PTR_TYPE, "func_idx_bound_i32p"))) {
HANDLE_FAILURE("LLVMBuildBitCast");
goto fail;
}
if (!(func_idx = LLVMBuildLoad2(comp_ctx->builder, I32_TYPE, func_idx,
"func_idx_bound"))) {
HANDLE_FAILURE("LLVMBuildLoad");
goto fail;
}
/* Initialize parameter types of the LLVM function */
total_param_count = 1 + func_param_count;
/* Extra function results' addresses (except the first one) are
appended to aot function parameters. */
if (func_result_count > 1)
total_param_count += func_result_count - 1;
total_size = sizeof(LLVMTypeRef) * (uint64)total_param_count;
if (total_size >= UINT32_MAX
|| !(param_types = wasm_runtime_malloc((uint32)total_size))) {
aot_set_last_error("allocate memory failed.");
goto fail;
}
/* Prepare param types */
j = 0;
param_types[j++] = comp_ctx->exec_env_type;
for (i = 0; i < func_param_count; i++)
param_types[j++] = TO_LLVM_TYPE(func_type->types[i]);
for (i = 1; i < func_result_count; i++, j++) {
param_types[j] = TO_LLVM_TYPE(func_type->types[func_param_count + i]);
if (!(param_types[j] = LLVMPointerType(param_types[j], 0))) {
aot_set_last_error("llvm get pointer type failed.");
goto fail;
}
}
/* Resolve return type of the LLVM function */
if (func_result_count) {
wasm_ret_type = func_type->types[func_param_count];
ret_type = TO_LLVM_TYPE(wasm_ret_type);
}
else {
wasm_ret_type = VALUE_TYPE_VOID;
ret_type = VOID_TYPE;
}
/* Allocate memory for parameters */
total_size = sizeof(LLVMValueRef) * (uint64)total_param_count;
if (total_size >= UINT32_MAX
|| !(param_values = wasm_runtime_malloc((uint32)total_size))) {
aot_set_last_error("allocate memory failed.");
goto fail;
}
/* First parameter is exec env */
j = 0;
param_values[j++] = func_ctx->exec_env;
/* Pop parameters from stack */
for (i = func_param_count - 1; (int32)i >= 0; i--)
POP(param_values[i + j], func_type->types[i]);
/* Prepare extra parameters */
ext_cell_num = 0;
for (i = 1; i < func_result_count; i++) {
ext_ret_offset = I32_CONST(ext_cell_num);
CHECK_LLVM_CONST(ext_ret_offset);
snprintf(buf, sizeof(buf), "ext_ret%d_ptr", i - 1);
if (!(ext_ret_ptr = LLVMBuildInBoundsGEP2(comp_ctx->builder, I32_TYPE,
func_ctx->argv_buf,
&ext_ret_offset, 1, buf))) {
aot_set_last_error("llvm build GEP failed.");
goto fail;
}
ext_ret_ptr_type = param_types[func_param_count + i];
snprintf(buf, sizeof(buf), "ext_ret%d_ptr_cast", i - 1);
if (!(ext_ret_ptr = LLVMBuildBitCast(comp_ctx->builder, ext_ret_ptr,
ext_ret_ptr_type, buf))) {
aot_set_last_error("llvm build bit cast failed.");
goto fail;
}
param_values[func_param_count + i] = ext_ret_ptr;
ext_cell_num += wasm_value_type_cell_num_internal(
func_type->types[func_param_count + i], comp_ctx->pointer_size);
}
if (ext_cell_num > 64) {
aot_set_last_error("prepare call-indirect arguments failed: "
"maximum 64 extra cell number supported.");
goto fail;
}
if (comp_ctx->aux_stack_frame_type
&& !comp_ctx->call_stack_features.frame_per_function) {
#if WASM_ENABLE_AOT_STACK_FRAME != 0
/* TODO: use current frame instead of allocating new frame
for WASM_OP_RETURN_CALL_REF */
if (!call_aot_alloc_frame_func(comp_ctx, func_ctx, func_idx))
goto fail;
#endif
}
/* Add basic blocks */
block_call_import = LLVMAppendBasicBlockInContext(
comp_ctx->context, func_ctx->func, "call_import");
block_call_non_import = LLVMAppendBasicBlockInContext(
comp_ctx->context, func_ctx->func, "call_non_import");
block_return = LLVMAppendBasicBlockInContext(comp_ctx->context,
func_ctx->func, "func_return");
if (!block_call_import || !block_call_non_import || !block_return) {
aot_set_last_error("llvm add basic block failed.");
goto fail;
}
LLVMMoveBasicBlockAfter(block_call_import,
LLVMGetInsertBlock(comp_ctx->builder));
LLVMMoveBasicBlockAfter(block_call_non_import, block_call_import);
LLVMMoveBasicBlockAfter(block_return, block_call_non_import);
import_func_count = I32_CONST(comp_ctx->comp_data->import_func_count);
CHECK_LLVM_CONST(import_func_count);
/* Check if func_idx < import_func_count */
if (!(cmp_func_idx = LLVMBuildICmp(comp_ctx->builder, LLVMIntULT, func_idx,
import_func_count, "cmp_func_idx"))) {
aot_set_last_error("llvm build icmp failed.");
goto fail;
}
/* If func_idx < import_func_count, jump to call import block,
else jump to call non-import block */
if (!LLVMBuildCondBr(comp_ctx->builder, cmp_func_idx, block_call_import,
block_call_non_import)) {
aot_set_last_error("llvm build cond br failed.");
goto fail;
}
/* Add result phis for return block */
LLVMPositionBuilderAtEnd(comp_ctx->builder, block_return);
if (func_result_count > 0) {
total_size = sizeof(LLVMValueRef) * (uint64)func_result_count;
if (total_size >= UINT32_MAX
|| !(result_phis = wasm_runtime_malloc((uint32)total_size))) {
aot_set_last_error("allocate memory failed.");
goto fail;
}
memset(result_phis, 0, (uint32)total_size);
for (i = 0; i < func_result_count; i++) {
LLVMTypeRef tmp_type =
TO_LLVM_TYPE(func_type->types[func_param_count + i]);
if (!(result_phis[i] =
LLVMBuildPhi(comp_ctx->builder, tmp_type, "phi"))) {
aot_set_last_error("llvm build phi failed.");
goto fail;
}
}
}
/* Translate call import block */
LLVMPositionBuilderAtEnd(comp_ctx->builder, block_call_import);
if (comp_ctx->aux_stack_frame_type == AOT_STACK_FRAME_TYPE_STANDARD
&& !commit_params_to_frame_of_import_func(comp_ctx, func_ctx, func_type,
param_values + 1)) {
goto fail;
}
/* Similar to opcode call_indirect, but for opcode ref.func needs to call
* aot_invoke_native_func instead */
if (!call_aot_invoke_native_func(comp_ctx, func_ctx, func_idx, func_type,
param_types + 1, param_values + 1,
func_param_count, param_cell_num, ret_type,
wasm_ret_type, &value_ret, &res))
goto fail;
/* Check whether exception was thrown when executing the function */
if (comp_ctx->enable_bound_check
&& !check_call_return(comp_ctx, func_ctx, res))
goto fail;
block_curr = LLVMGetInsertBlock(comp_ctx->builder);
/* Get function return values, for aot_invoke_native_func, the extra ret
* values are put into param's array */
if (func_result_count > 0) {
/* Push the first result to stack */
LLVMAddIncoming(result_phis[0], &value_ret, &block_curr, 1);
/* Load extra result from its address and push to stack */
for (i = 1; i < func_result_count; i++) {
ret_type = TO_LLVM_TYPE(func_type->types[func_param_count + i]);
snprintf(buf, sizeof(buf), "ext_ret%d", i - 1);
if (!(ext_ret = LLVMBuildLoad2(comp_ctx->builder, ret_type,
param_values[func_param_count + i],
buf))) {
aot_set_last_error("llvm build load failed.");
goto fail;
}
LLVMSetAlignment(ext_ret, 4);
LLVMAddIncoming(result_phis[i], &ext_ret, &block_curr, 1);
}
}
if (!LLVMBuildBr(comp_ctx->builder, block_return)) {
aot_set_last_error("llvm build br failed.");
goto fail;
}
/* Translate call non-import block */
LLVMPositionBuilderAtEnd(comp_ctx->builder, block_call_non_import);
/* Load function pointer */
if (!(func_ptr = LLVMBuildInBoundsGEP2(comp_ctx->builder, OPQ_PTR_TYPE,
func_ctx->func_ptrs, &func_idx, 1,
"func_ptr_tmp"))) {
aot_set_last_error("llvm build inbounds gep failed.");
goto fail;
}
if (!(func_ptr = LLVMBuildLoad2(comp_ctx->builder, OPQ_PTR_TYPE, func_ptr,
"func_ptr"))) {
aot_set_last_error("llvm build load failed.");
goto fail;
}
if (!(llvm_func_type =
LLVMFunctionType(ret_type, param_types, total_param_count, false))
|| !(llvm_func_ptr_type = LLVMPointerType(llvm_func_type, 0))) {
aot_set_last_error("llvm add function type failed.");
goto fail;
}
if (!(func = LLVMBuildBitCast(comp_ctx->builder, func_ptr,
llvm_func_ptr_type, "indirect_func"))) {
aot_set_last_error("llvm build bit cast failed.");
goto fail;
}
if (!(value_ret = LLVMBuildCall2(comp_ctx->builder, llvm_func_type, func,
param_values, total_param_count,
func_result_count > 0 ? "ret" : ""))) {
aot_set_last_error("llvm build call failed.");
goto fail;
}
/* Set calling convention for the call with the func's calling
convention */
LLVMSetInstructionCallConv(value_ret, LLVMGetFunctionCallConv(func));
if (tail_call)
LLVMSetTailCall(value_ret, true);
/* Check whether exception was thrown when executing the function */
if (!tail_call
&& (comp_ctx->enable_bound_check || is_win_platform(comp_ctx))
&& !check_exception_thrown(comp_ctx, func_ctx))
goto fail;
if (func_result_count > 0) {
block_curr = LLVMGetInsertBlock(comp_ctx->builder);
/* Push the first result to stack */
LLVMAddIncoming(result_phis[0], &value_ret, &block_curr, 1);
/* Load extra result from its address and push to stack */
for (i = 1; i < func_result_count; i++) {
ret_type = TO_LLVM_TYPE(func_type->types[func_param_count + i]);
snprintf(buf, sizeof(buf), "ext_ret%d", i - 1);
if (!(ext_ret = LLVMBuildLoad2(comp_ctx->builder, ret_type,
param_values[func_param_count + i],
buf))) {
aot_set_last_error("llvm build load failed.");
goto fail;
}
LLVMSetAlignment(ext_ret, 4);
LLVMAddIncoming(result_phis[i], &ext_ret, &block_curr, 1);
}
}
if (!LLVMBuildBr(comp_ctx->builder, block_return)) {
aot_set_last_error("llvm build br failed.");
goto fail;
}
/* Translate function return block */
LLVMPositionBuilderAtEnd(comp_ctx->builder, block_return);
for (i = 0; i < func_result_count; i++) {
PUSH(result_phis[i], func_type->types[func_param_count + i]);
}
if (comp_ctx->aux_stack_frame_type
&& !comp_ctx->call_stack_features.frame_per_function) {
#if WASM_ENABLE_AOT_STACK_FRAME != 0
if (!free_frame_for_aot_func(comp_ctx, func_ctx))
goto fail;
#endif
}
/* Insert suspend check point */
if (comp_ctx->enable_thread_mgr) {
if (!check_suspend_flags(comp_ctx, func_ctx, false))
goto fail;
}
ret = true;
fail:
if (param_values)
wasm_runtime_free(param_values);
if (param_types)
wasm_runtime_free(param_types);
if (result_phis)
wasm_runtime_free(result_phis);
return ret;
}
#endif /* end of WASM_ENABLE_GC != 0 */