The native function might call wasm function exported, in which the memory.grow opcode might be executed, and interpreter should update memory size after that, or load/store opcodes may run failed with "out of bounds memory access" exception thrown.
Update tensorflow sample patch, allow tensorflow wasm app to grow memory so as to run more models. And fix some compile issues of littlevgl zephyr sample for latest zephyr source code.
Signed-off-by: Wenyong Huang <wenyong.huang@intel.com>
Enable to use BH_VPRINTF macro for platform Linux/Windows/Darwin/VxWorks to redirect the stdout output from platform os_printf/os_vprintf, or the wasi output from wasm app to the vprintf like callback function specified by BH_VPRINTF macro of cmake WAMR_BH_VPRINTF variable.
Signed-off-by: Wenyong Huang <wenyong.huang@intel.com>
Enable SIMD by default for wamrc on x86-64 target, for iwasm on platform Linux and Darwin. And update build wamr document, fix app manager compile warning.
Signed-off-by: Wenyong Huang <wenyong.huang@intel.com>
Remove the limit of app file size no larger than 1 MB, fix possible memory leak issues when fail to install app file, change message size of aee_host_msg_callback() from uint16 type to uint32 type to fix possible integer overflow issue, and fix some coding style issues of app manager.
Signed-off-by: Wenyong Huang <wenyong.huang@intel.com>
The total size actually allocated by ealloc_hmu() might be larger than the size required to allocate, we reset the total size after allocating, so that if heap verification feature is enabled, the data to verify can be written to the right place of the suffix verification area. And in gc_realloc_vo_internal(), free the heap lock until the original data is copied to new object to return.
Signed-off-by: Wenyong Huang <wenyong.huang@intel.com>
And add native symbol node to list head when registering native symbol, so as to lookup developer's registered node firstly when linking.
Signed-off-by: Wenyong Huang <wenyong.huang@intel.com>
In some platforms, allocating memory with size 0 may return NULL but not an empty memory block, which causes runtime load, instantiate or execute wasm/aot file failed. We add checks to try to avoid allocating memory in runtime if the size is 0. And in wasm_runtime_malloc/free, output warning if allocate memory with size 0 and free memory with NULL ptr.
Also fix some coding style issues, fix handle riscv32 ilp32d issue, and fix several wasm-c-api issues.
Signed-off-by: Wenyong Huang <wenyong.huang@intel.com>
And remove redundant FAST_INTERP macros in wasm_interp_fast.c, and fix wamrc --help wrong line order issue.
Signed-off-by: Wenyong Huang <wenyong.huang@intel.com>
Re-implement aot call_indirect opcode translation: when calling non-import function, translate it by LLVM call IR to call the function in AOTed code, so as to avoid calling runtime aot_call_indirect API which is much slower. For import function, keep calling aot_call_indirect API due to the possible pointer/string argument conversion.
And add prompt info while app heap is corrupted, change emit_leb to emit_uint32 inter fast-interp to refine footprint.
Signed-off-by: Wenyong Huang <wenyong.huang@intel.com>
Modify the macro control sentences as darwin platform doesn't support mmap with MAP_32BIT flag to put the mapping address into 0-2G range.
Co-authored-by: Wenyong Huang <wenyong.huang@intel.com>
Update the build scripts of sample workloads (meshoptimizer/bwa/wasm-av1) to use the wasi-sdk-12.0 firstly to build the workload and discard clang-11, as wasi-sdk-12 supports wasi, simd and pthread better. And update the related documents.
Also modify wasm mini loader to sync up with the change of wasm normal loader.
Signed-off-by: Wenyong Huang <wenyong.huang@intel.com>
Update wasm app build scripts for wasi-sdk-12.0: add --export=__main_argc_argv, remove --no-threads
Lookup function with name "__main_argc_argv" as main function besides "main"
Change module_malloc to runtime_malloc in wasi native lib
Refine classic interpreter op_block and op_br_table
Refine faster interpreter op_br_table
Signed-off-by: Wenyong Huang <wenyong.huang@intel.com>
Without this change in a corresponding `CMakeLists.txt` it attempts to build `invokeNative` assembly for `X86_64` on Macs with Apple Silicon.
Co-authored-by: Wenyong Huang <wenyong.huang@intel.com>
Fix potential memory leak issue when using llvm::EngineBuilder().selectTarget()
Fix issue of accessing aot_value's fields after it is freed
Fix JIT not print failed to link import warning
Change some error messages: 'fail to' to 'failed to'
Update error message when SIMD isn't enabled
Fix install littlevgl wasm app of wasi version failed
Signed-off-by: Wenyong Huang <wenyong.huang@intel.com>
The EOL of NuttX is configurable (CR/LF/CRLF), the implementation
of getline in NuttX need CR or LF as line delimiter(CRLF not supported).
Then we should use readline for better compatibility.
Signed-off-by: Huang Qi <huangqi3@xiaomi.com>
Co-authored-by: Huang Qi <huangqi3@xiaomi.com>
Motivation:
we found some issues during enable the WAMR on arm devices, such as relocation issues, stack alignment issues.
Proposed change:
We optimized the relocation process for arm platform, for relocation jump table, the SP should be aligned to 16 bytes. And we also make "getentropy" worked for other non-supported platform.
