mirror of
https://github.com/bytecodealliance/wasm-micro-runtime.git
synced 2025-11-27 18:11:21 +00:00
* 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.43249378dabecause 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/jnQdJeabxiU1feb00a28c* 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>
1883 lines
52 KiB
C
1883 lines
52 KiB
C
/*
|
|
* Copyright (C) 2021 Intel Corporation. All rights reserved.
|
|
* SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
|
|
*/
|
|
|
|
#ifndef _JIT_IR_H_
|
|
#define _JIT_IR_H_
|
|
|
|
#include "bh_platform.h"
|
|
#include "../interpreter/wasm.h"
|
|
#include "jit_utils.h"
|
|
|
|
#ifdef __cplusplus
|
|
extern "C" {
|
|
#endif
|
|
|
|
/**
|
|
* Register (operand) representation of JIT IR.
|
|
*
|
|
* Encoding: [4-bit: kind, 28-bit register no.]
|
|
*
|
|
* Registers in JIT IR are classified into different kinds according
|
|
* to types of values they can hold. The classification is based on
|
|
* most processors' hardware register classifications, which include
|
|
* various sets of integer, floating point and vector registers with
|
|
* different sizes. These registers can be mapped onto corresponding
|
|
* kinds of hardware registers by register allocator. Instructions
|
|
* can only operate on allowed kinds of registers. For example, an
|
|
* integer instruction cannot operate on floating point or vector
|
|
* registers. Some encodings of these kinds of registers also
|
|
* represent immediate constant values and indexes to constant tables
|
|
* (see below). In that case, those registers are read-only. Writing
|
|
* to them is illegal. Reading from an immediate constant value
|
|
* register always returns the constant value encoded in the register
|
|
* no. Reading from a constant table index register always returns
|
|
* the constant value stored at the encoded index of the constant
|
|
* table of the register's kind. Immediate constant values and values
|
|
* indexed by constant table indexes can only be loaded into the
|
|
* corresponding kinds of registers if they must be loaded into
|
|
* registers. Besides these common kinds of registers, labels of
|
|
* basic blocks are also treated as registers of a special kind, which
|
|
* hold code addresses of basic block labels and are read-only. Each
|
|
* basic block is assigned one unique label register. With this
|
|
* unification, we can use the same set of load instructions to load
|
|
* values either from addresses stored in normal registers or from
|
|
* addresses of labels. Besides these register kinds, the void kind
|
|
* is a special kind of registers to denote some error occurs when a
|
|
* normal register is expected. Or it can be used as result operand
|
|
* of call and invoke instructions to denote no return values. The
|
|
* variable registers are classified into two sets: the hard registers
|
|
* whose register numbers are less than the hard register numbers of
|
|
* their kinds and the virtual registers whose register numbers are
|
|
* greater than or equal to the hard register numbers. Before
|
|
* register allocation is done, hard registers may appear in the IR
|
|
* due to special usages of passes frontend (e.g. fp_reg and exec_env_reg)
|
|
* or lower_cg. In the mean time (including during register
|
|
* allocation), those hard registers are treated same as virtual
|
|
* registers except that they may not be SSA and they can only be
|
|
* allocated to the hard registers of themselves.
|
|
*
|
|
* Classification of registers:
|
|
* + void register (kind == JIT_REG_KIND_VOID, no. must be 0)
|
|
* + label registers (kind == JIT_REG_KIND_L32)
|
|
* + value registers (kind == JIT_REG_KIND_I32/I64/F32/F64/V64/V128/V256)
|
|
* | + constants (_JIT_REG_CONST_VAL_FLAG | _JIT_REG_CONST_IDX_FLAG)
|
|
* | | + constant values (_JIT_REG_CONST_VAL_FLAG)
|
|
* | | + constant indexes (_JIT_REG_CONST_IDX_FLAG)
|
|
* | + variables (!(_JIT_REG_CONST_VAL_FLAG | _JIT_REG_CONST_IDX_FLAG))
|
|
* | | + hard registers (no. < hard register number)
|
|
* | | + virtual registers (no. >= hard register number)
|
|
*/
|
|
typedef uint32 JitReg;
|
|
|
|
/*
|
|
* Mask and shift bits of register kind.
|
|
*/
|
|
#define _JIT_REG_KIND_MASK 0xf0000000
|
|
#define _JIT_REG_KIND_SHIFT 28
|
|
|
|
/*
|
|
* Mask of register no. which must be the least significant bits.
|
|
*/
|
|
#define _JIT_REG_NO_MASK (~_JIT_REG_KIND_MASK)
|
|
|
|
/*
|
|
* Constant value flag (the most significant bit) of register
|
|
* no. field of integer, floating point and vector registers. If this
|
|
* flag is set in the register no., the rest bits of register
|
|
* no. represent a signed (27-bit) integer constant value of the
|
|
* corresponding type of the register and the register is read-only.
|
|
*/
|
|
#define _JIT_REG_CONST_VAL_FLAG ((_JIT_REG_NO_MASK >> 1) + 1)
|
|
|
|
/*
|
|
* Constant index flag of non-constant-value (constant value flag is
|
|
* not set in register no. field) integer, floating point and vector
|
|
* registers. If this flag is set, the rest bits of the register
|
|
* no. represent an index to the constant value table of the
|
|
* corresponding type of the register and the register is read-only.
|
|
*/
|
|
#define _JIT_REG_CONST_IDX_FLAG (_JIT_REG_CONST_VAL_FLAG >> 1)
|
|
|
|
/**
|
|
* Register kinds. Don't change the order of the defined values. The
|
|
* L32 kind must be after all normal kinds (see _const_val and _reg_ann
|
|
* of JitCompContext).
|
|
*/
|
|
typedef enum JitRegKind {
|
|
JIT_REG_KIND_VOID = 0x00, /* void type */
|
|
JIT_REG_KIND_I32 = 0x01, /* 32-bit signed or unsigned integer */
|
|
JIT_REG_KIND_I64 = 0x02, /* 64-bit signed or unsigned integer */
|
|
JIT_REG_KIND_F32 = 0x03, /* 32-bit floating point */
|
|
JIT_REG_KIND_F64 = 0x04, /* 64-bit floating point */
|
|
JIT_REG_KIND_V64 = 0x05, /* 64-bit vector */
|
|
JIT_REG_KIND_V128 = 0x06, /* 128-bit vector */
|
|
JIT_REG_KIND_V256 = 0x07, /* 256-bit vector */
|
|
JIT_REG_KIND_L32 = 0x08, /* 32-bit label address */
|
|
JIT_REG_KIND_NUM /* number of register kinds */
|
|
} JitRegKind;
|
|
|
|
#if UINTPTR_MAX == UINT64_MAX
|
|
#define JIT_REG_KIND_PTR JIT_REG_KIND_I64
|
|
#else
|
|
#define JIT_REG_KIND_PTR JIT_REG_KIND_I32
|
|
#endif
|
|
|
|
/**
|
|
* Construct a new JIT IR register from the kind and no.
|
|
*
|
|
* @param reg_kind register kind
|
|
* @param reg_no register no.
|
|
*
|
|
* @return the new register with the given kind and no.
|
|
*/
|
|
static inline JitReg
|
|
jit_reg_new(unsigned reg_kind, unsigned reg_no)
|
|
{
|
|
return (JitReg)((reg_kind << _JIT_REG_KIND_SHIFT) | reg_no);
|
|
}
|
|
|
|
/**
|
|
* Get the register kind of the given register.
|
|
*
|
|
* @param r a JIT IR register
|
|
*
|
|
* @return the register kind of register r
|
|
*/
|
|
static inline int
|
|
jit_reg_kind(JitReg r)
|
|
{
|
|
return (r & _JIT_REG_KIND_MASK) >> _JIT_REG_KIND_SHIFT;
|
|
}
|
|
|
|
/**
|
|
* Get the register no. of the given JIT IR register.
|
|
*
|
|
* @param r a JIT IR register
|
|
*
|
|
* @return the register no. of register r
|
|
*/
|
|
static inline int
|
|
jit_reg_no(JitReg r)
|
|
{
|
|
return r & _JIT_REG_NO_MASK;
|
|
}
|
|
|
|
/**
|
|
* Check whether the given register is a normal value register.
|
|
*
|
|
* @param r a JIT IR register
|
|
*
|
|
* @return true iff the register is a normal value register
|
|
*/
|
|
static inline bool
|
|
jit_reg_is_value(JitReg r)
|
|
{
|
|
unsigned kind = jit_reg_kind(r);
|
|
return kind > JIT_REG_KIND_VOID && kind < JIT_REG_KIND_L32;
|
|
}
|
|
|
|
/**
|
|
* Check whether the given register is a constant value.
|
|
*
|
|
* @param r a JIT IR register
|
|
*
|
|
* @return true iff register r is a constant value
|
|
*/
|
|
static inline bool
|
|
jit_reg_is_const_val(JitReg r)
|
|
{
|
|
return jit_reg_is_value(r) && (r & _JIT_REG_CONST_VAL_FLAG);
|
|
}
|
|
|
|
/**
|
|
* Check whether the given register is a constant table index.
|
|
*
|
|
* @param r a JIT IR register
|
|
*
|
|
* @return true iff register r is a constant table index
|
|
*/
|
|
static inline bool
|
|
jit_reg_is_const_idx(JitReg r)
|
|
{
|
|
return (jit_reg_is_value(r) && !jit_reg_is_const_val(r)
|
|
&& (r & _JIT_REG_CONST_IDX_FLAG));
|
|
}
|
|
|
|
/**
|
|
* Check whether the given register is a constant.
|
|
*
|
|
* @param r a JIT IR register
|
|
*
|
|
* @return true iff register r is a constant
|
|
*/
|
|
static inline bool
|
|
jit_reg_is_const(JitReg r)
|
|
{
|
|
return (jit_reg_is_value(r)
|
|
&& (r & (_JIT_REG_CONST_VAL_FLAG | _JIT_REG_CONST_IDX_FLAG)));
|
|
}
|
|
|
|
/**
|
|
* Check whether the given register is a normal variable register.
|
|
*
|
|
* @param r a JIT IR register
|
|
*
|
|
* @return true iff the register is a normal variable register
|
|
*/
|
|
static inline bool
|
|
jit_reg_is_variable(JitReg r)
|
|
{
|
|
return (jit_reg_is_value(r)
|
|
&& !(r & (_JIT_REG_CONST_VAL_FLAG | _JIT_REG_CONST_IDX_FLAG)));
|
|
}
|
|
|
|
/**
|
|
* Test whether the register is the given kind.
|
|
*
|
|
* @param KIND register kind name
|
|
* @param R register
|
|
*
|
|
* @return true if the register is the given kind
|
|
*/
|
|
#define jit_reg_is_kind(KIND, R) (jit_reg_kind(R) == JIT_REG_KIND_##KIND)
|
|
|
|
/**
|
|
* Construct a zero IR register with given the kind.
|
|
*
|
|
* @param kind the kind of the value
|
|
*
|
|
* @return a constant register of zero
|
|
*/
|
|
static inline JitReg
|
|
jit_reg_new_zero(unsigned kind)
|
|
{
|
|
bh_assert(kind != JIT_REG_KIND_VOID && kind < JIT_REG_KIND_L32);
|
|
return jit_reg_new(kind, _JIT_REG_CONST_VAL_FLAG);
|
|
}
|
|
|
|
/**
|
|
* Test whether the register is a zero constant value.
|
|
*
|
|
* @param reg an IR register
|
|
*
|
|
* @return true iff the register is a constant zero
|
|
*/
|
|
static inline JitReg
|
|
jit_reg_is_zero(JitReg reg)
|
|
{
|
|
return (jit_reg_is_value(reg)
|
|
&& jit_reg_no(reg) == _JIT_REG_CONST_VAL_FLAG);
|
|
}
|
|
|
|
/**
|
|
* Operand of instructions with fixed-number register operand(s).
|
|
*/
|
|
typedef JitReg JitOpndReg;
|
|
|
|
/**
|
|
* Operand of instructions with variable-number register operand(s).
|
|
*/
|
|
typedef struct JitOpndVReg {
|
|
uint32 _reg_num;
|
|
JitReg _reg[1];
|
|
} JitOpndVReg;
|
|
|
|
/**
|
|
* Operand of lookupswitch instruction.
|
|
*/
|
|
typedef struct JitOpndLookupSwitch {
|
|
/* NOTE: distance between JitReg operands must be the same (see
|
|
jit_insn_opnd_regs). */
|
|
JitReg value; /* the value to be compared */
|
|
uint32 match_pairs_num; /* match pairs number */
|
|
/* NOTE: offset between adjacent targets must be sizeof
|
|
(match_pairs[0]) (see implementation of jit_basic_block_succs),
|
|
so the default_target field must be here. */
|
|
JitReg default_target; /* default target BB */
|
|
struct {
|
|
int32 value; /* match value of the match pair */
|
|
JitReg target; /* target BB of the match pair */
|
|
} match_pairs[1]; /* match pairs of the instruction */
|
|
} JitOpndLookupSwitch;
|
|
|
|
/**
|
|
* Instruction of JIT IR.
|
|
*/
|
|
typedef struct JitInsn {
|
|
/* Pointers to the previous and next instructions. */
|
|
struct JitInsn *prev;
|
|
struct JitInsn *next;
|
|
|
|
/* Opcode of the instruction. */
|
|
uint16 opcode;
|
|
|
|
/* Reserved field that may be used by optimizations locally.
|
|
* bit_0(Least Significant Bit) is atomic flag for load/store */
|
|
uint8 flags_u8;
|
|
|
|
/* The unique ID of the instruction. */
|
|
uint16 uid;
|
|
|
|
/* Operands for different kinds of instructions. */
|
|
union {
|
|
/* For instructions with fixed-number register operand(s). */
|
|
JitOpndReg _opnd_Reg[1];
|
|
|
|
/* For instructions with variable-number register operand(s). */
|
|
JitOpndVReg _opnd_VReg;
|
|
|
|
/* For lookupswitch instruction. */
|
|
JitOpndLookupSwitch _opnd_LookupSwitch;
|
|
} _opnd;
|
|
} JitInsn;
|
|
|
|
/**
|
|
* Opcodes of IR instructions.
|
|
*/
|
|
typedef enum JitOpcode {
|
|
#define INSN(NAME, OPND_KIND, OPND_NUM, FIRST_USE) JIT_OP_##NAME,
|
|
#include "jit_ir.def"
|
|
#undef INSN
|
|
JIT_OP_OPCODE_NUMBER
|
|
} JitOpcode;
|
|
|
|
/*
|
|
* Helper functions for creating new instructions. Don't call them
|
|
* directly. Use jit_insn_new_NAME, such as jit_insn_new_MOV instead.
|
|
*/
|
|
|
|
JitInsn *
|
|
_jit_insn_new_Reg_0(JitOpcode opc);
|
|
JitInsn *
|
|
_jit_insn_new_Reg_1(JitOpcode opc, JitReg r0);
|
|
JitInsn *
|
|
_jit_insn_new_Reg_2(JitOpcode opc, JitReg r0, JitReg r1);
|
|
JitInsn *
|
|
_jit_insn_new_Reg_3(JitOpcode opc, JitReg r0, JitReg r1, JitReg r2);
|
|
JitInsn *
|
|
_jit_insn_new_Reg_4(JitOpcode opc, JitReg r0, JitReg r1, JitReg r2, JitReg r3);
|
|
JitInsn *
|
|
_jit_insn_new_Reg_5(JitOpcode opc, JitReg r0, JitReg r1, JitReg r2, JitReg r3,
|
|
JitReg r4);
|
|
JitInsn *
|
|
_jit_insn_new_VReg_1(JitOpcode opc, JitReg r0, int n);
|
|
JitInsn *
|
|
_jit_insn_new_VReg_2(JitOpcode opc, JitReg r0, JitReg r1, int n);
|
|
JitInsn *
|
|
_jit_insn_new_LookupSwitch_1(JitOpcode opc, JitReg value, uint32 num);
|
|
|
|
/*
|
|
* Instruction creation functions jit_insn_new_NAME, where NAME is the
|
|
* name of the instruction defined in jit_ir.def.
|
|
*/
|
|
#define ARG_DECL_Reg_0
|
|
#define ARG_LIST_Reg_0
|
|
#define ARG_DECL_Reg_1 JitReg r0
|
|
#define ARG_LIST_Reg_1 , r0
|
|
#define ARG_DECL_Reg_2 JitReg r0, JitReg r1
|
|
#define ARG_LIST_Reg_2 , r0, r1
|
|
#define ARG_DECL_Reg_3 JitReg r0, JitReg r1, JitReg r2
|
|
#define ARG_LIST_Reg_3 , r0, r1, r2
|
|
#define ARG_DECL_Reg_4 JitReg r0, JitReg r1, JitReg r2, JitReg r3
|
|
#define ARG_LIST_Reg_4 , r0, r1, r2, r3
|
|
#define ARG_DECL_Reg_5 JitReg r0, JitReg r1, JitReg r2, JitReg r3, JitReg r4
|
|
#define ARG_LIST_Reg_5 , r0, r1, r2, r3, r4
|
|
#define ARG_DECL_VReg_1 JitReg r0, int n
|
|
#define ARG_LIST_VReg_1 , r0, n
|
|
#define ARG_DECL_VReg_2 JitReg r0, JitReg r1, int n
|
|
#define ARG_LIST_VReg_2 , r0, r1, n
|
|
#define ARG_DECL_LookupSwitch_1 JitReg value, uint32 num
|
|
#define ARG_LIST_LookupSwitch_1 , value, num
|
|
#define INSN(NAME, OPND_KIND, OPND_NUM, FIRST_USE) \
|
|
static inline JitInsn *jit_insn_new_##NAME( \
|
|
ARG_DECL_##OPND_KIND##_##OPND_NUM) \
|
|
{ \
|
|
return _jit_insn_new_##OPND_KIND##_##OPND_NUM( \
|
|
JIT_OP_##NAME ARG_LIST_##OPND_KIND##_##OPND_NUM); \
|
|
}
|
|
#include "jit_ir.def"
|
|
#undef INSN
|
|
#undef ARG_DECL_Reg_0
|
|
#undef ARG_LIST_Reg_0
|
|
#undef ARG_DECL_Reg_1
|
|
#undef ARG_LIST_Reg_1
|
|
#undef ARG_DECL_Reg_2
|
|
#undef ARG_LIST_Reg_2
|
|
#undef ARG_DECL_Reg_3
|
|
#undef ARG_LIST_Reg_3
|
|
#undef ARG_DECL_Reg_4
|
|
#undef ARG_LIST_Reg_4
|
|
#undef ARG_DECL_Reg_5
|
|
#undef ARG_LIST_Reg_5
|
|
#undef ARG_DECL_VReg_1
|
|
#undef ARG_LIST_VReg_1
|
|
#undef ARG_DECL_VReg_2
|
|
#undef ARG_LIST_VReg_2
|
|
#undef ARG_DECL_LookupSwitch_1
|
|
#undef ARG_LIST_LookupSwitch_1
|
|
|
|
/**
|
|
* Delete an instruction
|
|
*
|
|
* @param insn an instruction to be deleted
|
|
*/
|
|
static inline void
|
|
jit_insn_delete(JitInsn *insn)
|
|
{
|
|
jit_free(insn);
|
|
}
|
|
|
|
/*
|
|
* Runtime type check functions that check whether accessing the n-th
|
|
* operand is legal. They are only used for in self-verification
|
|
* mode.
|
|
*
|
|
* @param insn any JIT IR instruction
|
|
* @param n index of the operand to access
|
|
*
|
|
* @return true if the access is legal
|
|
*/
|
|
bool
|
|
_jit_insn_check_opnd_access_Reg(const JitInsn *insn, unsigned n);
|
|
bool
|
|
_jit_insn_check_opnd_access_VReg(const JitInsn *insn, unsigned n);
|
|
bool
|
|
_jit_insn_check_opnd_access_LookupSwitch(const JitInsn *insn);
|
|
|
|
/**
|
|
* Get the pointer to the n-th register operand of the given
|
|
* instruction. The instruction format must be Reg.
|
|
*
|
|
* @param insn a Reg format instruction
|
|
* @param n index of the operand to get
|
|
*
|
|
* @return pointer to the n-th operand
|
|
*/
|
|
static inline JitReg *
|
|
jit_insn_opnd(JitInsn *insn, int n)
|
|
{
|
|
bh_assert(_jit_insn_check_opnd_access_Reg(insn, n));
|
|
return &insn->_opnd._opnd_Reg[n];
|
|
}
|
|
|
|
/**
|
|
* Get the pointer to the n-th register operand of the given
|
|
* instruction. The instruction format must be VReg.
|
|
*
|
|
* @param insn a VReg format instruction
|
|
* @param n index of the operand to get
|
|
*
|
|
* @return pointer to the n-th operand
|
|
*/
|
|
static inline JitReg *
|
|
jit_insn_opndv(JitInsn *insn, int n)
|
|
{
|
|
bh_assert(_jit_insn_check_opnd_access_VReg(insn, n));
|
|
return &insn->_opnd._opnd_VReg._reg[n];
|
|
}
|
|
|
|
/**
|
|
* Get the operand number of the given instruction. The instruction
|
|
* format must be VReg.
|
|
*
|
|
* @param insn a VReg format instruction
|
|
*
|
|
* @return operand number of the instruction
|
|
*/
|
|
static inline unsigned
|
|
jit_insn_opndv_num(const JitInsn *insn)
|
|
{
|
|
bh_assert(_jit_insn_check_opnd_access_VReg(insn, 0));
|
|
return insn->_opnd._opnd_VReg._reg_num;
|
|
}
|
|
|
|
/**
|
|
* Get the pointer to the LookupSwitch operand of the given
|
|
* instruction. The instruction format must be LookupSwitch.
|
|
*
|
|
* @param insn a LookupSwitch format instruction
|
|
*
|
|
* @return pointer to the operand
|
|
*/
|
|
static inline JitOpndLookupSwitch *
|
|
jit_insn_opndls(JitInsn *insn)
|
|
{
|
|
bh_assert(_jit_insn_check_opnd_access_LookupSwitch(insn));
|
|
return &insn->_opnd._opnd_LookupSwitch;
|
|
}
|
|
|
|
/**
|
|
* Insert instruction @p insn2 before instruction @p insn1.
|
|
*
|
|
* @param insn1 any instruction
|
|
* @param insn2 any instruction
|
|
*/
|
|
void
|
|
jit_insn_insert_before(JitInsn *insn1, JitInsn *insn2);
|
|
|
|
/**
|
|
* Insert instruction @p insn2 after instruction @p insn1.
|
|
*
|
|
* @param insn1 any instruction
|
|
* @param insn2 any instruction
|
|
*/
|
|
void
|
|
jit_insn_insert_after(JitInsn *insn1, JitInsn *insn2);
|
|
|
|
/**
|
|
* Unlink the instruction @p insn from the containing list.
|
|
*
|
|
* @param insn an instruction
|
|
*/
|
|
void
|
|
jit_insn_unlink(JitInsn *insn);
|
|
|
|
/**
|
|
* Get the hash value of the comparable instruction (pure functions
|
|
* and exception check instructions).
|
|
*
|
|
* @param insn an instruction
|
|
*
|
|
* @return hash value of the instruction
|
|
*/
|
|
unsigned
|
|
jit_insn_hash(JitInsn *insn);
|
|
|
|
/**
|
|
* Compare whether the two comparable instructions are the same.
|
|
*
|
|
* @param insn1 the first instruction
|
|
* @param insn2 the second instruction
|
|
*
|
|
* @return true if the two instructions are the same
|
|
*/
|
|
bool
|
|
jit_insn_equal(JitInsn *insn1, JitInsn *insn2);
|
|
|
|
/**
|
|
* Register vector for accessing predecessors and successors of a
|
|
* basic block.
|
|
*/
|
|
typedef struct JitRegVec {
|
|
JitReg *_base; /* points to the first register */
|
|
int32 _stride; /* stride to the next register */
|
|
uint32 num; /* number of registers */
|
|
} JitRegVec;
|
|
|
|
/**
|
|
* Get the address of the i-th register in the register vector.
|
|
*
|
|
* @param vec a register vector
|
|
* @param i index to the register vector
|
|
*
|
|
* @return the address of the i-th register in the vector
|
|
*/
|
|
static inline JitReg *
|
|
jit_reg_vec_at(const JitRegVec *vec, unsigned i)
|
|
{
|
|
bh_assert(i < vec->num);
|
|
return vec->_base + vec->_stride * i;
|
|
}
|
|
|
|
/**
|
|
* Visit each element in a register vector.
|
|
*
|
|
* @param V (JitRegVec) the register vector
|
|
* @param I (unsigned) index variable in the vector
|
|
* @param R (JitReg *) resiger pointer variable
|
|
*/
|
|
#define JIT_REG_VEC_FOREACH(V, I, R) \
|
|
for ((I) = 0, (R) = (V)._base; (I) < (V).num; (I)++, (R) += (V)._stride)
|
|
|
|
/**
|
|
* Visit each register defined by an instruction.
|
|
*
|
|
* @param V (JitRegVec) register vector of the instruction
|
|
* @param I (unsigned) index variable in the vector
|
|
* @param R (JitReg *) resiger pointer variable
|
|
* @param F index of the first used register
|
|
*/
|
|
#define JIT_REG_VEC_FOREACH_DEF(V, I, R, F) \
|
|
for ((I) = 0, (R) = (V)._base; (I) < (F); (I)++, (R) += (V)._stride)
|
|
|
|
/**
|
|
* Visit each register used by an instruction.
|
|
*
|
|
* @param V (JitRegVec) register vector of the instruction
|
|
* @param I (unsigned) index variable in the vector
|
|
* @param R (JitReg *) resiger pointer variable
|
|
* @param F index of the first used register
|
|
*/
|
|
#define JIT_REG_VEC_FOREACH_USE(V, I, R, F) \
|
|
for ((I) = (F), (R) = (V)._base + (F) * (V)._stride; (I) < (V).num; \
|
|
(I)++, (R) += (V)._stride)
|
|
|
|
/**
|
|
* Get a generic register vector that contains all register operands.
|
|
* The registers defined by the instruction, if any, appear before the
|
|
* registers used by the instruction.
|
|
*
|
|
* @param insn an instruction
|
|
*
|
|
* @return a register vector containing register operands
|
|
*/
|
|
JitRegVec
|
|
jit_insn_opnd_regs(JitInsn *insn);
|
|
|
|
/**
|
|
* Get the index of the first use register in the register vector
|
|
* returned by jit_insn_opnd_regs.
|
|
*
|
|
* @param insn an instruction
|
|
*
|
|
* @return the index of the first use register in the register vector
|
|
*/
|
|
unsigned
|
|
jit_insn_opnd_first_use(JitInsn *insn);
|
|
|
|
/**
|
|
* Basic Block of JIT IR. It is a basic block only if the IR is not in
|
|
* non-BB form. The block is represented by a special phi node, whose
|
|
* result and arguments are label registers. The result label is the
|
|
* containing block's label. The arguments are labels of predecessors
|
|
* of the block. Successor labels are stored in the last instruction,
|
|
* which must be a control flow instruction. Instructions of a block
|
|
* are linked in a circular linked list with the block phi node as the
|
|
* end of the list. The next and prev field of the block phi node
|
|
* point to the first and last instructions of the block.
|
|
*/
|
|
typedef JitInsn JitBasicBlock;
|
|
|
|
/**
|
|
* Create a new basic block instance.
|
|
*
|
|
* @param label the label of the new basic block
|
|
* @param n number of predecessors
|
|
*
|
|
* @return the created new basic block instance
|
|
*/
|
|
JitBasicBlock *
|
|
jit_basic_block_new(JitReg label, int n);
|
|
|
|
/**
|
|
* Delete a basic block instance and all instructions init.
|
|
*
|
|
* @param block the basic block to be deleted
|
|
*/
|
|
void
|
|
jit_basic_block_delete(JitBasicBlock *block);
|
|
|
|
/**
|
|
* Get the label of the basic block.
|
|
*
|
|
* @param block a basic block instance
|
|
*
|
|
* @return the label of the basic block
|
|
*/
|
|
static inline JitReg
|
|
jit_basic_block_label(JitBasicBlock *block)
|
|
{
|
|
return *(jit_insn_opndv(block, 0));
|
|
}
|
|
|
|
/**
|
|
* Get the first instruction of the basic block.
|
|
*
|
|
* @param block a basic block instance
|
|
*
|
|
* @return the first instruction of the basic block
|
|
*/
|
|
static inline JitInsn *
|
|
jit_basic_block_first_insn(JitBasicBlock *block)
|
|
{
|
|
return block->next;
|
|
}
|
|
|
|
/**
|
|
* Get the last instruction of the basic block.
|
|
*
|
|
* @param block a basic block instance
|
|
*
|
|
* @return the last instruction of the basic block
|
|
*/
|
|
static inline JitInsn *
|
|
jit_basic_block_last_insn(JitBasicBlock *block)
|
|
{
|
|
return block->prev;
|
|
}
|
|
|
|
/**
|
|
* Get the end of instruction list of the basic block (which is always
|
|
* the block itself).
|
|
*
|
|
* @param block a basic block instance
|
|
*
|
|
* @return the end of instruction list of the basic block
|
|
*/
|
|
static inline JitInsn *
|
|
jit_basic_block_end_insn(JitBasicBlock *block)
|
|
{
|
|
return block;
|
|
}
|
|
|
|
/**
|
|
* Visit each instruction in the block from the first to the last. In
|
|
* the code block, the instruction pointer @p I must be a valid
|
|
* pointer to an instruction in the block. That means if the
|
|
* instruction may be deleted, @p I must point to the previous or next
|
|
* valid instruction before the next iteration.
|
|
*
|
|
* @param B (JitBasicBlock *) the block
|
|
* @param I (JitInsn *) instruction visited
|
|
*/
|
|
#define JIT_FOREACH_INSN(B, I) \
|
|
for (I = jit_basic_block_first_insn(B); I != jit_basic_block_end_insn(B); \
|
|
I = I->next)
|
|
|
|
/**
|
|
* Visit each instruction in the block from the last to the first. In
|
|
* the code block, the instruction pointer @p I must be a valid
|
|
* pointer to an instruction in the block. That means if the
|
|
* instruction may be deleted, @p I must point to the previous or next
|
|
* valid instruction before the next iteration.
|
|
*
|
|
* @param B (JitBasicBlock *) the block
|
|
* @param I (JitInsn *) instruction visited
|
|
*/
|
|
#define JIT_FOREACH_INSN_REVERSE(B, I) \
|
|
for (I = jit_basic_block_last_insn(B); I != jit_basic_block_end_insn(B); \
|
|
I = I->prev)
|
|
|
|
/**
|
|
* Prepend an instruction in the front of the block. The position is
|
|
* just after the block phi node (the block instance itself).
|
|
*
|
|
* @param block a block
|
|
* @param insn an instruction to be prepended
|
|
*/
|
|
static inline void
|
|
jit_basic_block_prepend_insn(JitBasicBlock *block, JitInsn *insn)
|
|
{
|
|
jit_insn_insert_after(block, insn);
|
|
}
|
|
|
|
/**
|
|
* Append an instruction to the end of the basic block.
|
|
*
|
|
* @param block a basic block
|
|
* @param insn an instruction to be appended
|
|
*/
|
|
static inline void
|
|
jit_basic_block_append_insn(JitBasicBlock *block, JitInsn *insn)
|
|
{
|
|
jit_insn_insert_before(block, insn);
|
|
}
|
|
|
|
/**
|
|
* Get the register vector of predecessors of a basic block.
|
|
*
|
|
* @param block a JIT IR block
|
|
*
|
|
* @return register vector of the predecessors
|
|
*/
|
|
JitRegVec
|
|
jit_basic_block_preds(JitBasicBlock *block);
|
|
|
|
/**
|
|
* Get the register vector of successors of a basic block.
|
|
*
|
|
* @param block a JIT IR basic block
|
|
*
|
|
* @return register vector of the successors
|
|
*/
|
|
JitRegVec
|
|
jit_basic_block_succs(JitBasicBlock *block);
|
|
|
|
/**
|
|
* Hard register information of one kind.
|
|
*/
|
|
typedef struct JitHardRegInfo {
|
|
struct {
|
|
/* Hard register number of this kind. */
|
|
uint32 num;
|
|
|
|
/* Whether each register is fixed. */
|
|
const uint8 *fixed;
|
|
|
|
/* Whether each register is caller-saved in the native ABI. */
|
|
const uint8 *caller_saved_native;
|
|
|
|
/* Whether each register is caller-saved in the JITed ABI. */
|
|
const uint8 *caller_saved_jitted;
|
|
} info[JIT_REG_KIND_L32];
|
|
|
|
/* The indexes of hard registers of frame pointer, exec_env and cmp. */
|
|
uint32 fp_hreg_index;
|
|
uint32 exec_env_hreg_index;
|
|
uint32 cmp_hreg_index;
|
|
} JitHardRegInfo;
|
|
|
|
struct JitBlock;
|
|
struct JitCompContext;
|
|
struct JitValueSlot;
|
|
|
|
/**
|
|
* Value in the WASM operation stack, each stack element
|
|
* is a Jit register
|
|
*/
|
|
typedef struct JitValue {
|
|
struct JitValue *next;
|
|
struct JitValue *prev;
|
|
struct JitValueSlot *value;
|
|
/* VALUE_TYPE_I32/I64/F32/F64/VOID */
|
|
uint8 type;
|
|
} JitValue;
|
|
|
|
/**
|
|
* Value stack, represents stack elements in a WASM block
|
|
*/
|
|
typedef struct JitValueStack {
|
|
JitValue *value_list_head;
|
|
JitValue *value_list_end;
|
|
} JitValueStack;
|
|
|
|
/* Record information of a value slot of local variable or stack
|
|
during translation. */
|
|
typedef struct JitValueSlot {
|
|
/* The virtual register that holds the value of the slot if the
|
|
value of the slot is in register. */
|
|
JitReg reg;
|
|
|
|
/* The dirty bit of the value slot. It's set if the value in
|
|
register is newer than the value in memory. */
|
|
uint32 dirty : 1;
|
|
|
|
/* Whether the new value in register is a reference, which is valid
|
|
only when the dirty bit is set. */
|
|
uint32 ref : 1;
|
|
|
|
/* Committed reference flag. 0: unknown, 1: not-reference, 2:
|
|
reference. */
|
|
uint32 committed_ref : 2;
|
|
} JitValueSlot;
|
|
|
|
typedef struct JitMemRegs {
|
|
/* The following registers should be re-loaded after
|
|
memory.grow, callbc and callnative */
|
|
JitReg memory_inst;
|
|
JitReg cur_page_count;
|
|
JitReg memory_data;
|
|
JitReg memory_data_end;
|
|
JitReg mem_bound_check_1byte;
|
|
JitReg mem_bound_check_2bytes;
|
|
JitReg mem_bound_check_4bytes;
|
|
JitReg mem_bound_check_8bytes;
|
|
JitReg mem_bound_check_16bytes;
|
|
} JitMemRegs;
|
|
|
|
typedef struct JitTableRegs {
|
|
JitReg table_elems;
|
|
/* Should be re-loaded after table.grow,
|
|
callbc and callnative */
|
|
JitReg table_cur_size;
|
|
} JitTableRegs;
|
|
|
|
/* Frame information for translation */
|
|
typedef struct JitFrame {
|
|
/* The current wasm module */
|
|
WASMModule *cur_wasm_module;
|
|
/* The current wasm function */
|
|
WASMFunction *cur_wasm_func;
|
|
/* The current wasm function index */
|
|
uint32 cur_wasm_func_idx;
|
|
/* The current compilation context */
|
|
struct JitCompContext *cc;
|
|
|
|
/* Max local slot number. */
|
|
uint32 max_locals;
|
|
|
|
/* Max operand stack slot number. */
|
|
uint32 max_stacks;
|
|
|
|
/* Instruction pointer */
|
|
uint8 *ip;
|
|
|
|
/* Stack top pointer */
|
|
JitValueSlot *sp;
|
|
|
|
/* Committed instruction pointer */
|
|
uint8 *committed_ip;
|
|
|
|
/* Committed stack top pointer */
|
|
JitValueSlot *committed_sp;
|
|
|
|
/* WASM module instance */
|
|
JitReg module_inst_reg;
|
|
/* WASM module */
|
|
JitReg module_reg;
|
|
/* module_inst->import_func_ptrs */
|
|
JitReg import_func_ptrs_reg;
|
|
/* module_inst->fast_jit_func_ptrs */
|
|
JitReg fast_jit_func_ptrs_reg;
|
|
/* module_inst->func_type_indexes */
|
|
JitReg func_type_indexes_reg;
|
|
/* Boundary of auxiliary stack */
|
|
JitReg aux_stack_bound_reg;
|
|
/* Bottom of auxiliary stack */
|
|
JitReg aux_stack_bottom_reg;
|
|
/* Data of memory instances */
|
|
JitMemRegs *memory_regs;
|
|
/* Data of table instances */
|
|
JitTableRegs *table_regs;
|
|
|
|
/* Local variables */
|
|
JitValueSlot lp[1];
|
|
} JitFrame;
|
|
|
|
typedef struct JitIncomingInsn {
|
|
struct JitIncomingInsn *next;
|
|
JitInsn *insn;
|
|
uint32 opnd_idx;
|
|
} JitIncomingInsn, *JitIncomingInsnList;
|
|
|
|
typedef struct JitBlock {
|
|
struct JitBlock *next;
|
|
struct JitBlock *prev;
|
|
|
|
/* The current Jit Block */
|
|
struct JitCompContext *cc;
|
|
|
|
/* LABEL_TYPE_BLOCK/LOOP/IF/FUNCTION */
|
|
uint32 label_type;
|
|
|
|
/* code of else opcode of this block, if it is a IF block */
|
|
uint8 *wasm_code_else;
|
|
/* code of end opcode of this block */
|
|
uint8 *wasm_code_end;
|
|
|
|
/* JIT label points to code begin */
|
|
JitBasicBlock *basic_block_entry;
|
|
/* JIT label points to code else */
|
|
JitBasicBlock *basic_block_else;
|
|
/* JIT label points to code end */
|
|
JitBasicBlock *basic_block_end;
|
|
|
|
/* Incoming INSN for basic_block_else */
|
|
JitInsn *incoming_insn_for_else_bb;
|
|
/* Incoming INSNs for basic_block_end */
|
|
JitIncomingInsnList incoming_insns_for_end_bb;
|
|
|
|
/* WASM operation stack */
|
|
JitValueStack value_stack;
|
|
|
|
/* Param count/types/PHIs of this block */
|
|
uint32 param_count;
|
|
uint8 *param_types;
|
|
|
|
/* Result count/types/PHIs of this block */
|
|
uint32 result_count;
|
|
uint8 *result_types;
|
|
|
|
/* The begin frame stack pointer of this block */
|
|
JitValueSlot *frame_sp_begin;
|
|
} JitBlock;
|
|
|
|
/**
|
|
* Block stack, represents WASM block stack elements
|
|
*/
|
|
typedef struct JitBlockStack {
|
|
JitBlock *block_list_head;
|
|
JitBlock *block_list_end;
|
|
} JitBlockStack;
|
|
|
|
/**
|
|
* The JIT compilation context for one compilation process of a
|
|
* compilation unit.
|
|
*/
|
|
typedef struct JitCompContext {
|
|
/* Hard register information of each kind. */
|
|
const JitHardRegInfo *hreg_info;
|
|
|
|
/* No. of the pass to be applied. */
|
|
uint8 cur_pass_no;
|
|
|
|
/* The current wasm module */
|
|
WASMModule *cur_wasm_module;
|
|
/* The current wasm function */
|
|
WASMFunction *cur_wasm_func;
|
|
/* The current wasm function index */
|
|
uint32 cur_wasm_func_idx;
|
|
/* The block stack */
|
|
JitBlockStack block_stack;
|
|
|
|
bool mem_space_unchanged;
|
|
|
|
/* Entry and exit labels of the compilation unit, whose numbers must
|
|
be 0 and 1 respectively (see JIT_FOREACH_BLOCK). */
|
|
JitReg entry_label;
|
|
JitReg exit_label;
|
|
JitBasicBlock **exce_basic_blocks;
|
|
JitIncomingInsnList *incoming_insns_for_exec_bbs;
|
|
|
|
/* The current basic block to generate instructions */
|
|
JitBasicBlock *cur_basic_block;
|
|
|
|
/* Registers of frame pointer, exec_env and CMP result. */
|
|
JitReg fp_reg;
|
|
JitReg exec_env_reg;
|
|
JitReg cmp_reg;
|
|
|
|
/* WASM module instance */
|
|
JitReg module_inst_reg;
|
|
/* WASM module */
|
|
JitReg module_reg;
|
|
/* module_inst->import_func_ptrs */
|
|
JitReg import_func_ptrs_reg;
|
|
/* module_inst->fast_jit_func_ptrs */
|
|
JitReg fast_jit_func_ptrs_reg;
|
|
/* module_inst->func_type_indexes */
|
|
JitReg func_type_indexes_reg;
|
|
/* Boundary of auxiliary stack */
|
|
JitReg aux_stack_bound_reg;
|
|
/* Bottom of auxiliary stack */
|
|
JitReg aux_stack_bottom_reg;
|
|
/* Data of memory instances */
|
|
JitMemRegs *memory_regs;
|
|
/* Data of table instances */
|
|
JitTableRegs *table_regs;
|
|
|
|
/* Current frame information for translation */
|
|
JitFrame *jit_frame;
|
|
|
|
/* The total frame size of current function */
|
|
uint32 total_frame_size;
|
|
|
|
/* The spill cache offset to the interp frame */
|
|
uint32 spill_cache_offset;
|
|
/* The spill cache size */
|
|
uint32 spill_cache_size;
|
|
|
|
/* The offset of jitted_return_address in the frame, which is set by
|
|
the pass frontend and used by the pass codegen. */
|
|
uint32 jitted_return_address_offset;
|
|
|
|
/* Begin and end addresses of the jitted code produced by the pass
|
|
codegen and consumed by the region registration after codegen and
|
|
the pass dump. */
|
|
void *jitted_addr_begin;
|
|
void *jitted_addr_end;
|
|
|
|
char last_error[128];
|
|
|
|
/* Below fields are all private. Don't access them directly. */
|
|
|
|
/* Reference count of the compilation context. */
|
|
uint16 _reference_count;
|
|
|
|
/* Constant values. */
|
|
struct {
|
|
/* Number of constant values of each kind. */
|
|
uint32 _num[JIT_REG_KIND_L32];
|
|
|
|
/* Capacity of register annotations of each kind. */
|
|
uint32 _capacity[JIT_REG_KIND_L32];
|
|
|
|
/* Constant values of each kind. */
|
|
uint8 *_value[JIT_REG_KIND_L32];
|
|
|
|
/* Next element on the list of values with the same hash code. */
|
|
JitReg *_next[JIT_REG_KIND_L32];
|
|
|
|
/* Size of the hash table. */
|
|
uint32 _hash_table_size;
|
|
|
|
/* Map values to JIT register. */
|
|
JitReg *_hash_table;
|
|
} _const_val;
|
|
|
|
/* Annotations of labels, registers and instructions. */
|
|
struct {
|
|
/* Number of all ever created labels. */
|
|
uint32 _label_num;
|
|
|
|
/* Capacity of label annotations. */
|
|
uint32 _label_capacity;
|
|
|
|
/* Number of all ever created instructions. */
|
|
uint32 _insn_num;
|
|
|
|
/* Capacity of instruction annotations. */
|
|
uint32 _insn_capacity;
|
|
|
|
/* Number of ever created registers of each kind. */
|
|
uint32 _reg_num[JIT_REG_KIND_L32];
|
|
|
|
/* Capacity of register annotations of each kind. */
|
|
uint32 _reg_capacity[JIT_REG_KIND_L32];
|
|
|
|
/* Storage of annotations. */
|
|
#define ANN_LABEL(TYPE, NAME) TYPE *_label_##NAME;
|
|
#define ANN_INSN(TYPE, NAME) TYPE *_insn_##NAME;
|
|
#define ANN_REG(TYPE, NAME) TYPE *_reg_##NAME[JIT_REG_KIND_L32];
|
|
#include "jit_ir.def"
|
|
#undef ANN_LABEL
|
|
#undef ANN_INSN
|
|
#undef ANN_REG
|
|
|
|
/* Flags of annotations. */
|
|
#define ANN_LABEL(TYPE, NAME) uint32 _label_##NAME##_enabled : 1;
|
|
#define ANN_INSN(TYPE, NAME) uint32 _insn_##NAME##_enabled : 1;
|
|
#define ANN_REG(TYPE, NAME) uint32 _reg_##NAME##_enabled : 1;
|
|
#include "jit_ir.def"
|
|
#undef ANN_LABEL
|
|
#undef ANN_INSN
|
|
#undef ANN_REG
|
|
} _ann;
|
|
|
|
/* Instruction hash table. */
|
|
struct {
|
|
/* Size of the hash table. */
|
|
uint32 _size;
|
|
|
|
/* The hash table. */
|
|
JitInsn **_table;
|
|
} _insn_hash_table;
|
|
|
|
/* indicate if the last comparison is about floating-point numbers or not
|
|
*/
|
|
bool last_cmp_on_fp;
|
|
} JitCompContext;
|
|
|
|
/*
|
|
* Annotation accessing functions jit_annl_NAME, jit_anni_NAME and
|
|
* jit_annr_NAME.
|
|
*/
|
|
#define ANN_LABEL(TYPE, NAME) \
|
|
static inline TYPE *jit_annl_##NAME(JitCompContext *cc, JitReg label) \
|
|
{ \
|
|
unsigned idx = jit_reg_no(label); \
|
|
bh_assert(jit_reg_kind(label) == JIT_REG_KIND_L32); \
|
|
bh_assert(idx < cc->_ann._label_num); \
|
|
bh_assert(cc->_ann._label_##NAME##_enabled); \
|
|
return &cc->_ann._label_##NAME[idx]; \
|
|
}
|
|
#define ANN_INSN(TYPE, NAME) \
|
|
static inline TYPE *jit_anni_##NAME(JitCompContext *cc, JitInsn *insn) \
|
|
{ \
|
|
unsigned uid = insn->uid; \
|
|
bh_assert(uid < cc->_ann._insn_num); \
|
|
bh_assert(cc->_ann._insn_##NAME##_enabled); \
|
|
return &cc->_ann._insn_##NAME[uid]; \
|
|
}
|
|
#define ANN_REG(TYPE, NAME) \
|
|
static inline TYPE *jit_annr_##NAME(JitCompContext *cc, JitReg reg) \
|
|
{ \
|
|
unsigned kind = jit_reg_kind(reg); \
|
|
unsigned no = jit_reg_no(reg); \
|
|
bh_assert(kind < JIT_REG_KIND_L32); \
|
|
bh_assert(no < cc->_ann._reg_num[kind]); \
|
|
bh_assert(cc->_ann._reg_##NAME##_enabled); \
|
|
return &cc->_ann._reg_##NAME[kind][no]; \
|
|
}
|
|
#include "jit_ir.def"
|
|
#undef ANN_LABEL
|
|
#undef ANN_INSN
|
|
#undef ANN_REG
|
|
|
|
/*
|
|
* Annotation enabling functions jit_annl_enable_NAME,
|
|
* jit_anni_enable_NAME and jit_annr_enable_NAME, which allocate
|
|
* sufficient memory for the annotations.
|
|
*/
|
|
#define ANN_LABEL(TYPE, NAME) bool jit_annl_enable_##NAME(JitCompContext *cc);
|
|
#define ANN_INSN(TYPE, NAME) bool jit_anni_enable_##NAME(JitCompContext *cc);
|
|
#define ANN_REG(TYPE, NAME) bool jit_annr_enable_##NAME(JitCompContext *cc);
|
|
#include "jit_ir.def"
|
|
#undef ANN_LABEL
|
|
#undef ANN_INSN
|
|
#undef ANN_REG
|
|
|
|
/*
|
|
* Annotation disabling functions jit_annl_disable_NAME,
|
|
* jit_anni_disable_NAME and jit_annr_disable_NAME, which release
|
|
* memory of the annotations. Before calling these functions,
|
|
* resources owned by the annotations must be explicitly released.
|
|
*/
|
|
#define ANN_LABEL(TYPE, NAME) void jit_annl_disable_##NAME(JitCompContext *cc);
|
|
#define ANN_INSN(TYPE, NAME) void jit_anni_disable_##NAME(JitCompContext *cc);
|
|
#define ANN_REG(TYPE, NAME) void jit_annr_disable_##NAME(JitCompContext *cc);
|
|
#include "jit_ir.def"
|
|
#undef ANN_LABEL
|
|
#undef ANN_INSN
|
|
#undef ANN_REG
|
|
|
|
/*
|
|
* Functions jit_annl_is_enabled_NAME, jit_anni_is_enabled_NAME and
|
|
* jit_annr_is_enabled_NAME for checking whether an annotation is
|
|
* enabled.
|
|
*/
|
|
#define ANN_LABEL(TYPE, NAME) \
|
|
static inline bool jit_annl_is_enabled_##NAME(JitCompContext *cc) \
|
|
{ \
|
|
return !!cc->_ann._label_##NAME##_enabled; \
|
|
}
|
|
#define ANN_INSN(TYPE, NAME) \
|
|
static inline bool jit_anni_is_enabled_##NAME(JitCompContext *cc) \
|
|
{ \
|
|
return !!cc->_ann._insn_##NAME##_enabled; \
|
|
}
|
|
#define ANN_REG(TYPE, NAME) \
|
|
static inline bool jit_annr_is_enabled_##NAME(JitCompContext *cc) \
|
|
{ \
|
|
return !!cc->_ann._reg_##NAME##_enabled; \
|
|
}
|
|
#include "jit_ir.def"
|
|
#undef ANN_LABEL
|
|
#undef ANN_INSN
|
|
#undef ANN_REG
|
|
|
|
/**
|
|
* Initialize a compilation context.
|
|
*
|
|
* @param cc the compilation context
|
|
* @param htab_size the initial hash table size of constant pool
|
|
*
|
|
* @return cc if succeeds, NULL otherwise
|
|
*/
|
|
JitCompContext *
|
|
jit_cc_init(JitCompContext *cc, unsigned htab_size);
|
|
|
|
/**
|
|
* Release all resources of a compilation context, which doesn't
|
|
* include the compilation context itself.
|
|
*
|
|
* @param cc the compilation context
|
|
*/
|
|
void
|
|
jit_cc_destroy(JitCompContext *cc);
|
|
|
|
/**
|
|
* Increase the reference count of the compilation context.
|
|
*
|
|
* @param cc the compilation context
|
|
*/
|
|
static inline void
|
|
jit_cc_inc_ref(JitCompContext *cc)
|
|
{
|
|
cc->_reference_count++;
|
|
}
|
|
|
|
/**
|
|
* Decrease the reference_count and destroy and free the compilation
|
|
* context if the reference_count is decreased to zero.
|
|
*
|
|
* @param cc the compilation context
|
|
*/
|
|
void
|
|
jit_cc_delete(JitCompContext *cc);
|
|
|
|
char *
|
|
jit_get_last_error(JitCompContext *cc);
|
|
|
|
void
|
|
jit_set_last_error(JitCompContext *cc, const char *error);
|
|
|
|
void
|
|
jit_set_last_error_v(JitCompContext *cc, const char *format, ...);
|
|
|
|
/**
|
|
* Create a I32 constant value with relocatable into the compilation
|
|
* context. A constant value that has relocation info cannot be
|
|
* constant-folded as normal constants because its value depends on
|
|
* runtime context and may be different in different executions.
|
|
*
|
|
* @param cc compilation context
|
|
* @param val a I32 value
|
|
* @param rel relocation information
|
|
*
|
|
* @return a constant register containing the value
|
|
*/
|
|
JitReg
|
|
jit_cc_new_const_I32_rel(JitCompContext *cc, int32 val, uint32 rel);
|
|
|
|
/**
|
|
* Create a I32 constant value without relocation info (0) into the
|
|
* compilation context.
|
|
*
|
|
* @param cc compilation context
|
|
* @param val a I32 value
|
|
*
|
|
* @return a constant register containing the value
|
|
*/
|
|
static inline JitReg
|
|
jit_cc_new_const_I32(JitCompContext *cc, int32 val)
|
|
{
|
|
return jit_cc_new_const_I32_rel(cc, val, 0);
|
|
}
|
|
|
|
/**
|
|
* Create a I64 constant value into the compilation context.
|
|
*
|
|
* @param cc compilation context
|
|
* @param val a I64 value
|
|
*
|
|
* @return a constant register containing the value
|
|
*/
|
|
JitReg
|
|
jit_cc_new_const_I64(JitCompContext *cc, int64 val);
|
|
|
|
#if UINTPTR_MAX == UINT64_MAX
|
|
#define jit_cc_new_const_PTR jit_cc_new_const_I64
|
|
#else
|
|
#define jit_cc_new_const_PTR jit_cc_new_const_I32
|
|
#endif
|
|
|
|
/**
|
|
* Create a F32 constant value into the compilation context.
|
|
*
|
|
* @param cc compilation context
|
|
* @param val a F32 value
|
|
*
|
|
* @return a constant register containing the value
|
|
*/
|
|
JitReg
|
|
jit_cc_new_const_F32(JitCompContext *cc, float val);
|
|
|
|
/**
|
|
* Create a F64 constant value into the compilation context.
|
|
*
|
|
* @param cc compilation context
|
|
* @param val a F64 value
|
|
*
|
|
* @return a constant register containing the value
|
|
*/
|
|
JitReg
|
|
jit_cc_new_const_F64(JitCompContext *cc, double val);
|
|
|
|
/**
|
|
* Get the relocation info of a I32 constant register.
|
|
*
|
|
* @param cc compilation context
|
|
* @param reg constant register
|
|
*
|
|
* @return the relocation info of the constant
|
|
*/
|
|
uint32
|
|
jit_cc_get_const_I32_rel(JitCompContext *cc, JitReg reg);
|
|
|
|
/**
|
|
* Get the constant value of a I32 constant register.
|
|
*
|
|
* @param cc compilation context
|
|
* @param reg constant register
|
|
*
|
|
* @return the constant value
|
|
*/
|
|
int32
|
|
jit_cc_get_const_I32(JitCompContext *cc, JitReg reg);
|
|
|
|
/**
|
|
* Get the constant value of a I64 constant register.
|
|
*
|
|
* @param cc compilation context
|
|
* @param reg constant register
|
|
*
|
|
* @return the constant value
|
|
*/
|
|
int64
|
|
jit_cc_get_const_I64(JitCompContext *cc, JitReg reg);
|
|
|
|
/**
|
|
* Get the constant value of a F32 constant register.
|
|
*
|
|
* @param cc compilation context
|
|
* @param reg constant register
|
|
*
|
|
* @return the constant value
|
|
*/
|
|
float
|
|
jit_cc_get_const_F32(JitCompContext *cc, JitReg reg);
|
|
|
|
/**
|
|
* Get the constant value of a F64 constant register.
|
|
*
|
|
* @param cc compilation context
|
|
* @param reg constant register
|
|
*
|
|
* @return the constant value
|
|
*/
|
|
double
|
|
jit_cc_get_const_F64(JitCompContext *cc, JitReg reg);
|
|
|
|
/**
|
|
* Get the number of total created labels.
|
|
*
|
|
* @param cc the compilation context
|
|
*
|
|
* @return the number of total created labels
|
|
*/
|
|
static inline unsigned
|
|
jit_cc_label_num(JitCompContext *cc)
|
|
{
|
|
return cc->_ann._label_num;
|
|
}
|
|
|
|
/**
|
|
* Get the number of total created instructions.
|
|
*
|
|
* @param cc the compilation context
|
|
*
|
|
* @return the number of total created instructions
|
|
*/
|
|
static inline unsigned
|
|
jit_cc_insn_num(JitCompContext *cc)
|
|
{
|
|
return cc->_ann._insn_num;
|
|
}
|
|
|
|
/**
|
|
* Get the number of total created registers.
|
|
*
|
|
* @param cc the compilation context
|
|
* @param kind the register kind
|
|
*
|
|
* @return the number of total created registers
|
|
*/
|
|
static inline unsigned
|
|
jit_cc_reg_num(JitCompContext *cc, unsigned kind)
|
|
{
|
|
bh_assert(kind < JIT_REG_KIND_L32);
|
|
return cc->_ann._reg_num[kind];
|
|
}
|
|
|
|
/**
|
|
* Create a new label in the compilation context.
|
|
*
|
|
* @param cc the compilation context
|
|
*
|
|
* @return a new label in the compilation context
|
|
*/
|
|
JitReg
|
|
jit_cc_new_label(JitCompContext *cc);
|
|
|
|
/**
|
|
* Create a new block with a new label in the compilation context.
|
|
*
|
|
* @param cc the compilation context
|
|
* @param n number of predecessors
|
|
*
|
|
* @return a new block with a new label in the compilation context
|
|
*/
|
|
JitBasicBlock *
|
|
jit_cc_new_basic_block(JitCompContext *cc, int n);
|
|
|
|
/**
|
|
* Resize the predecessor number of a block.
|
|
*
|
|
* @param cc the containing compilation context
|
|
* @param block block to be resized
|
|
* @param n new number of predecessors
|
|
*
|
|
* @return the new block if succeeds, NULL otherwise
|
|
*/
|
|
JitBasicBlock *
|
|
jit_cc_resize_basic_block(JitCompContext *cc, JitBasicBlock *block, int n);
|
|
|
|
/**
|
|
* Initialize the instruction hash table to the given size and enable
|
|
* the instruction's _hash_link annotation.
|
|
*
|
|
* @param cc the containing compilation context
|
|
* @param n size of the hash table
|
|
*
|
|
* @return true if succeeds, false otherwise
|
|
*/
|
|
bool
|
|
jit_cc_enable_insn_hash(JitCompContext *cc, unsigned n);
|
|
|
|
/**
|
|
* Destroy the instruction hash table and disable the instruction's
|
|
* _hash_link annotation.
|
|
*
|
|
* @param cc the containing compilation context
|
|
*/
|
|
void
|
|
jit_cc_disable_insn_hash(JitCompContext *cc);
|
|
|
|
/**
|
|
* Reset the hash table entries.
|
|
*
|
|
* @param cc the containing compilation context
|
|
*/
|
|
void
|
|
jit_cc_reset_insn_hash(JitCompContext *cc);
|
|
|
|
/**
|
|
* Allocate a new instruction ID in the compilation context and set it
|
|
* to the given instruction.
|
|
*
|
|
* @param cc the compilation context
|
|
* @param insn IR instruction
|
|
*
|
|
* @return the insn with uid being set
|
|
*/
|
|
JitInsn *
|
|
jit_cc_set_insn_uid(JitCompContext *cc, JitInsn *insn);
|
|
|
|
/*
|
|
* Similar to jit_cc_set_insn_uid except that if setting uid failed,
|
|
* delete the insn. Only used by jit_cc_new_insn
|
|
*/
|
|
JitInsn *
|
|
_jit_cc_set_insn_uid_for_new_insn(JitCompContext *cc, JitInsn *insn);
|
|
|
|
/**
|
|
* Create a new instruction in the compilation context.
|
|
*
|
|
* @param cc the compilationo context
|
|
* @param NAME instruction name
|
|
*
|
|
* @return a new instruction in the compilation context
|
|
*/
|
|
#define jit_cc_new_insn(cc, NAME, ...) \
|
|
_jit_cc_set_insn_uid_for_new_insn(cc, jit_insn_new_##NAME(__VA_ARGS__))
|
|
|
|
/*
|
|
* Helper function for jit_cc_new_insn_norm.
|
|
*/
|
|
JitInsn *
|
|
_jit_cc_new_insn_norm(JitCompContext *cc, JitReg *result, JitInsn *insn);
|
|
|
|
/**
|
|
* Create a new instruction in the compilation context and normalize
|
|
* the instruction (constant folding and simplification etc.). If the
|
|
* instruction hashing is enabled (anni__hash_link is enabled), try to
|
|
* find the existing equivalent insruction first before adding a new
|
|
* one to the compilation contest.
|
|
*
|
|
* @param cc the compilationo context
|
|
* @param result returned result of the instruction. If the value is
|
|
* non-zero, it is the result of the constant-folding or an existing
|
|
* equivalent instruction, in which case no instruction is added into
|
|
* the compilation context. Otherwise, a new normalized instruction
|
|
* has been added into the compilation context.
|
|
* @param NAME instruction name
|
|
*
|
|
* @return a new or existing instruction in the compilation context
|
|
*/
|
|
#define jit_cc_new_insn_norm(cc, result, NAME, ...) \
|
|
_jit_cc_new_insn_norm(cc, result, jit_insn_new_##NAME(__VA_ARGS__))
|
|
|
|
/**
|
|
* Helper function for GEN_INSN
|
|
*
|
|
* @param cc compilation context
|
|
* @param block the current block
|
|
* @param insn the new instruction
|
|
*
|
|
* @return the new instruction if inserted, NULL otherwise
|
|
*/
|
|
static inline JitInsn *
|
|
_gen_insn(JitCompContext *cc, JitInsn *insn)
|
|
{
|
|
if (insn)
|
|
jit_basic_block_append_insn(cc->cur_basic_block, insn);
|
|
else
|
|
jit_set_last_error(cc, "generate insn failed");
|
|
|
|
return insn;
|
|
}
|
|
|
|
/**
|
|
* Generate and append an instruction to the current block.
|
|
*/
|
|
#define GEN_INSN(...) _gen_insn(cc, jit_cc_new_insn(cc, __VA_ARGS__))
|
|
|
|
/**
|
|
* Create a constant register without relocation info.
|
|
*
|
|
* @param Type type of the register
|
|
* @param val the constant value
|
|
*
|
|
* @return the constant register if succeeds, 0 otherwise
|
|
*/
|
|
#define NEW_CONST(Type, val) jit_cc_new_const_##Type(cc, val)
|
|
|
|
/**
|
|
* Create a new virtual register in the compilation context.
|
|
*
|
|
* @param cc the compilation context
|
|
* @param kind kind of the register
|
|
*
|
|
* @return a new label in the compilation context
|
|
*/
|
|
JitReg
|
|
jit_cc_new_reg(JitCompContext *cc, unsigned kind);
|
|
|
|
/*
|
|
* Create virtual registers with specific types in the compilation
|
|
* context. They are more convenient than the above one.
|
|
*/
|
|
|
|
static inline JitReg
|
|
jit_cc_new_reg_I32(JitCompContext *cc)
|
|
{
|
|
return jit_cc_new_reg(cc, JIT_REG_KIND_I32);
|
|
}
|
|
|
|
static inline JitReg
|
|
jit_cc_new_reg_I64(JitCompContext *cc)
|
|
{
|
|
return jit_cc_new_reg(cc, JIT_REG_KIND_I64);
|
|
}
|
|
|
|
#if UINTPTR_MAX == UINT64_MAX
|
|
#define jit_cc_new_reg_ptr jit_cc_new_reg_I64
|
|
#else
|
|
#define jit_cc_new_reg_ptr jit_cc_new_reg_I32
|
|
#endif
|
|
|
|
static inline JitReg
|
|
jit_cc_new_reg_F32(JitCompContext *cc)
|
|
{
|
|
return jit_cc_new_reg(cc, JIT_REG_KIND_F32);
|
|
}
|
|
|
|
static inline JitReg
|
|
jit_cc_new_reg_F64(JitCompContext *cc)
|
|
{
|
|
return jit_cc_new_reg(cc, JIT_REG_KIND_F64);
|
|
}
|
|
|
|
static inline JitReg
|
|
jit_cc_new_reg_V64(JitCompContext *cc)
|
|
{
|
|
return jit_cc_new_reg(cc, JIT_REG_KIND_V64);
|
|
}
|
|
|
|
static inline JitReg
|
|
jit_cc_new_reg_V128(JitCompContext *cc)
|
|
{
|
|
return jit_cc_new_reg(cc, JIT_REG_KIND_V128);
|
|
}
|
|
|
|
static inline JitReg
|
|
jit_cc_new_reg_V256(JitCompContext *cc)
|
|
{
|
|
return jit_cc_new_reg(cc, JIT_REG_KIND_V256);
|
|
}
|
|
|
|
/**
|
|
* Get the hard register numbe of the given kind
|
|
*
|
|
* @param cc the compilation context
|
|
* @param kind the register kind
|
|
*
|
|
* @return number of hard registers of the given kind
|
|
*/
|
|
static inline unsigned
|
|
jit_cc_hreg_num(JitCompContext *cc, unsigned kind)
|
|
{
|
|
bh_assert(kind < JIT_REG_KIND_L32);
|
|
return cc->hreg_info->info[kind].num;
|
|
}
|
|
|
|
/**
|
|
* Check whether a given register is a hard register.
|
|
*
|
|
* @param cc the compilation context
|
|
* @param reg the register which must be a variable
|
|
*
|
|
* @return true if the register is a hard register
|
|
*/
|
|
static inline bool
|
|
jit_cc_is_hreg(JitCompContext *cc, JitReg reg)
|
|
{
|
|
unsigned kind = jit_reg_kind(reg);
|
|
unsigned no = jit_reg_no(reg);
|
|
bh_assert(jit_reg_is_variable(reg));
|
|
bh_assert(kind < JIT_REG_KIND_L32);
|
|
return no < cc->hreg_info->info[kind].num;
|
|
}
|
|
|
|
/**
|
|
* Check whether the given hard register is fixed.
|
|
*
|
|
* @param cc the compilation context
|
|
* @param reg the hard register
|
|
*
|
|
* @return true if the hard register is fixed
|
|
*/
|
|
static inline bool
|
|
jit_cc_is_hreg_fixed(JitCompContext *cc, JitReg reg)
|
|
{
|
|
unsigned kind = jit_reg_kind(reg);
|
|
unsigned no = jit_reg_no(reg);
|
|
bh_assert(jit_cc_is_hreg(cc, reg));
|
|
bh_assert(kind < JIT_REG_KIND_L32);
|
|
return !!cc->hreg_info->info[kind].fixed[no];
|
|
}
|
|
|
|
/**
|
|
* Check whether the given hard register is caller-saved-native.
|
|
*
|
|
* @param cc the compilation context
|
|
* @param reg the hard register
|
|
*
|
|
* @return true if the hard register is caller-saved-native
|
|
*/
|
|
static inline bool
|
|
jit_cc_is_hreg_caller_saved_native(JitCompContext *cc, JitReg reg)
|
|
{
|
|
unsigned kind = jit_reg_kind(reg);
|
|
unsigned no = jit_reg_no(reg);
|
|
bh_assert(jit_cc_is_hreg(cc, reg));
|
|
bh_assert(kind < JIT_REG_KIND_L32);
|
|
return !!cc->hreg_info->info[kind].caller_saved_native[no];
|
|
}
|
|
|
|
/**
|
|
* Check whether the given hard register is caller-saved-jitted.
|
|
*
|
|
* @param cc the compilation context
|
|
* @param reg the hard register
|
|
*
|
|
* @return true if the hard register is caller-saved-jitted
|
|
*/
|
|
static inline bool
|
|
jit_cc_is_hreg_caller_saved_jitted(JitCompContext *cc, JitReg reg)
|
|
{
|
|
unsigned kind = jit_reg_kind(reg);
|
|
unsigned no = jit_reg_no(reg);
|
|
bh_assert(jit_cc_is_hreg(cc, reg));
|
|
bh_assert(kind < JIT_REG_KIND_L32);
|
|
return !!cc->hreg_info->info[kind].caller_saved_jitted[no];
|
|
}
|
|
|
|
/**
|
|
* Return the entry block of the compilation context.
|
|
*
|
|
* @param cc the compilation context
|
|
*
|
|
* @return the entry block of the compilation context
|
|
*/
|
|
static inline JitBasicBlock *
|
|
jit_cc_entry_basic_block(JitCompContext *cc)
|
|
{
|
|
return *(jit_annl_basic_block(cc, cc->entry_label));
|
|
}
|
|
|
|
/**
|
|
* Return the exit block of the compilation context.
|
|
*
|
|
* @param cc the compilation context
|
|
*
|
|
* @return the exit block of the compilation context
|
|
*/
|
|
static inline JitBasicBlock *
|
|
jit_cc_exit_basic_block(JitCompContext *cc)
|
|
{
|
|
return *(jit_annl_basic_block(cc, cc->exit_label));
|
|
}
|
|
|
|
void
|
|
jit_value_stack_push(JitValueStack *stack, JitValue *value);
|
|
|
|
JitValue *
|
|
jit_value_stack_pop(JitValueStack *stack);
|
|
|
|
void
|
|
jit_value_stack_destroy(JitValueStack *stack);
|
|
|
|
JitBlock *
|
|
jit_block_stack_top(JitBlockStack *stack);
|
|
|
|
void
|
|
jit_block_stack_push(JitBlockStack *stack, JitBlock *block);
|
|
|
|
JitBlock *
|
|
jit_block_stack_pop(JitBlockStack *stack);
|
|
|
|
void
|
|
jit_block_stack_destroy(JitBlockStack *stack);
|
|
|
|
bool
|
|
jit_block_add_incoming_insn(JitBlock *block, JitInsn *insn, uint32 opnd_idx);
|
|
|
|
void
|
|
jit_block_destroy(JitBlock *block);
|
|
|
|
bool
|
|
jit_cc_push_value(JitCompContext *cc, uint8 type, JitReg value);
|
|
|
|
bool
|
|
jit_cc_pop_value(JitCompContext *cc, uint8 type, JitReg *p_value);
|
|
|
|
bool
|
|
jit_lock_reg_in_insn(JitCompContext *cc, JitInsn *the_insn, JitReg reg_to_lock);
|
|
|
|
/**
|
|
* Update the control flow graph after successors of blocks are
|
|
* changed so that the predecessor vector of each block represents the
|
|
* updated status. The predecessors may not be required by all
|
|
* passes, so we don't need to keep them always being updated.
|
|
*
|
|
* @param cc the compilation context
|
|
*
|
|
* @return true if succeeds, false otherwise
|
|
*/
|
|
bool
|
|
jit_cc_update_cfg(JitCompContext *cc);
|
|
|
|
/**
|
|
* Visit each normal block (which is not entry nor exit block) in a
|
|
* compilation context. New blocks can be added in the loop body, but
|
|
* they won't be visited. Blocks can also be removed safely (by
|
|
* setting the label's block annotation to NULL) in the loop body.
|
|
*
|
|
* @param CC (JitCompContext *) the compilation context
|
|
* @param I (unsigned) index variable of the block (label no)
|
|
* @param E (unsigned) end index variable of block (last index + 1)
|
|
* @param B (JitBasicBlock *) block pointer variable
|
|
*/
|
|
#define JIT_FOREACH_BLOCK(CC, I, E, B) \
|
|
for ((I) = 2, (E) = (CC)->_ann._label_num; (I) < (E); (I)++) \
|
|
if (((B) = (CC)->_ann._label_basic_block[(I)]))
|
|
|
|
/**
|
|
* The version that includes entry and exit block.
|
|
*/
|
|
#define JIT_FOREACH_BLOCK_ENTRY_EXIT(CC, I, E, B) \
|
|
for ((I) = 0, (E) = (CC)->_ann._label_num; (I) < (E); (I)++) \
|
|
if (((B) = (CC)->_ann._label_basic_block[(I)]))
|
|
|
|
/**
|
|
* Visit each normal block (which is not entry nor exit block) in a
|
|
* compilation context in reverse order. New blocks can be added in
|
|
* the loop body, but they won't be visited. Blocks can also be
|
|
* removed safely (by setting the label's block annotation to NULL) in
|
|
* the loop body.
|
|
*
|
|
* @param CC (JitCompContext *) the compilation context
|
|
* @param I (unsigned) index of the block (label no)
|
|
* @param B (JitBasicBlock *) block pointer
|
|
*/
|
|
#define JIT_FOREACH_BLOCK_REVERSE(CC, I, B) \
|
|
for ((I) = (CC)->_ann._label_num; (I) > 2; (I)--) \
|
|
if (((B) = (CC)->_ann._label_basic_block[(I)-1]))
|
|
|
|
/**
|
|
* The version that includes entry and exit block.
|
|
*/
|
|
#define JIT_FOREACH_BLOCK_REVERSE_ENTRY_EXIT(CC, I, B) \
|
|
for ((I) = (CC)->_ann._label_num; (I) > 0; (I)--) \
|
|
if (((B) = (CC)->_ann._label_basic_block[(I)-1]))
|
|
|
|
#ifdef __cplusplus
|
|
}
|
|
#endif
|
|
|
|
#endif /* end of _JIT_IR_H_ */
|