Follow-up on #2907. The log level is needed in the host embedder to
better integrate with the embedder's logger.
Allow the developer to customize his bh_log callback with
`cmake -DWAMR_BH_LOG=<log_callback>`,
and update sample/basic to show the usage.
Possible alternatives:
* Make wasm_cluster_destroy_spawned_exec_env take two exec_env.
One for wasm execution and another to specify the target to destroy.
* Make execute functions to switch exec_env as briefly discussed in
https://github.com/bytecodealliance/wasm-micro-runtime/pull/2047
When WAMR is embedded to other application, the lifecycle of the socket
might conflict with other usecases. E.g. if WAMR is deinitialized before any
other use of sockets, the application goes into the invalid state. The new
flag allows host application to take control over the socket initialization.
Check whether the arguments are NULL before calling bh_hash_map_find,
or lots of "HashMap find elem failed: map or key is NULL" warnings may
be dumped. Reported in #3053.
Though SIMD isn't supported by interpreter, when JIT is enabled,
developer may run `iwasm --interp <wasm_file>` to trigger the SIMD
opcode in interpreter, which isn't handled before this PR.
It seems that some users want to wrap rather large chunk of code
with wasm_runtime_begin_blocking_op/wasm_runtime_end_blocking_op.
If the wrapped code happens to have a call to
e.g. wasm_runtime_spawn_exec_env, WASM_SUSPEND_FLAG_BLOCKING is
inherited to the child exec_env and it may cause unexpected behaviors.
- Enable quick aot entry when hw bound check is disabled
- Remove unnecessary ret_type argument in the quick aot entries
- Declare detailed prototype of aot function to call in each quick aot entry
Since there is no so rich api in freertos like embedded system, simply set
CONFIG_HAS_CAP_ENTER to 1 to support posix file api for freertos.
Test file api in wasm app pass.
Enhance the statistic of wasm function execution time, or the performance
profiling feature:
- Add os_time_thread_cputime_us() to get the cputime of a thread,
and use it to calculate the execution time of a wasm function
- Support the statistic of the children execution time of a function,
and dump it in wasm_runtime_dump_perf_profiling
- Expose two APIs:
wasm_runtime_sum_wasm_exec_time
wasm_runtime_get_wasm_func_exec_time
And rename os_time_get_boot_microsecond to os_time_get_boot_us.
For shared memory, the max memory size must be defined in advanced. Re-allocation
for growing memory can't be used as it might change the base address, therefore when
OS_ENABLE_HW_BOUND_CHECK is enabled the memory is mmaped, and if the flag is
disabled, the memory is allocated. This change introduces a flag that allows users to use
mmap for reserving memory address space even if the OS_ENABLE_HW_BOUND_CHECK
is disabled.
When the original wasm contains multiple compilation units, the current
logic uses the first one for everything. This commit tries to use a bit more
appropriate ones.
Errors were reported when initializing wasm_val_t values with WASM_I32_VAL like macros.
```
error: missing initializer for member ‘wasm_val_t::__paddings’ [-Werror=missing-field-initializers]
64 | wasm_val_t res = {WASM_INIT_VAL};
```
And rename DEPRECATED to WASM_API_DEPRECATED to avoid using defines with generic names.
Compilation error was reported when `cmake -DWAMR_BUILD_LIBC_WASI=0`
on linux-sgx platform:
```
core/shared/platform/linux-sgx/sgx_socket.c:8:10:
fatal error: libc_errno.h: No such file or directory
8 | #include "libc_errno.h"
| ^~~~~~~~~~~~~~
```
After fixing, both `cmake -DWAMR_BUILD_LIBC_WASI=1` and
`WAMR_BUILD_LIBC_WASI=0` work good.
The content in custom name section is changed after loaded since the strings
are adjusted with '\0' appended, the emitted AOT file then cannot be loaded.
The PR disables changing the content for AOT compiler to resolve it.
And disable emitting custom name section for `wamrc --enable-dump-call-stack`,
instead, use `wamrc --emit-custom-sections=name` to emit it.
`pthread_jit_write_protect_np` is only available on macOS, and
`sys_icache_invalidate` is available on both iOS and macOS and
has no restrictions on ARM architecture.
When using the wasm-c-api and there's a trap, `wasm_func_call()` returns
a `wasm_trap_t *` object. No matter which thread crashes, the trap contains
the stack frames of the main thread.
With this PR, when there's an exception, the stack frames of the thread
where the exception occurs are stored into the thread cluster.
`wasm_func_call()` can then return those stack frames.
Allow to invoke the quick call entry wasm_runtime_quick_invoke_c_api_import to
call the wasm-c-api import functions to speedup the calling process, which reduces
the data copying.
Use `wamrc --invoke-c-api-import` to generate the optimized AOT code, and set
`jit_options->quick_invoke_c_api_import` true in wasm_engine_new when LLVM JIT
is enabled.
In some scenarios there may be lots of callings to AOT/JIT functions from the
host embedder, which expects good performance for the calling process, while
in the current implementation, runtime calls the wasm_runtime_invoke_native
to prepare the array of registers and stacks for the invokeNative assemble code,
and the latter then puts the elements in the array to physical registers and
native stacks and calls the AOT/JIT function, there may be many data copying
and handlings which impact the performance.
This PR registers some quick AOT/JIT entries for some simple wasm signatures,
and let runtime call the entry to directly invoke the AOT/JIT function instead of
calling wasm_runtime_invoke_native, which speedups the calling process.
We may extend the mechanism next to allow the developer to register his quick
AOT/JIT entries to speedup the calling process of invoking the AOT/JIT functions
for some specific signatures.
On macOS, by default, the first 4GB is occupied by the pagezero.
While it can be controlled with link time options, as we are
an library, we usually don't have a control on how to link an
executable.
Add an API to set segue flags for wasm-c-api LLVM JIT mode:
```C
wasm_config_t *
wasm_config_set_segue_flags(wasm_config_t *config, uint32 segue_flags);
```
- Don't allocate the implicit/unused frame when calling the LLVM JIT function
- Don't set exec_env's thread handle and stack boundary in the recursive
calling from host, since they have been set in the first time calling
- Fix frame not freed in llvm_jit_call_func_bytecode
before the change, only support wasm app exit like:
```c
void *thread_routine(void *arg)
{
printf("Enter thread\n");
return NULL;
}
```
if call pthread_exit, it will crash:
```c
void *thread_routine(void *arg)
{
printf("Enter thread\n");
pthread_exit(NULL);
return NULL;
}
```
This commit lets both upstairs work correctly, test pass on stm32f103 mcu.
And refactor the original perf support
- use WAMR_BUILD_LINUX_PERF as the cmake compilation control
- use WASM_ENABLE_LINUX_PERF as the compiler macro
- use `wamrc --enable-linux-perf` to generate aot file which contains fp operations
- use `iwasm --enable-linux-perf` to create perf map for `perf record`
The host embedder may also want to terminate the wasm instance
for single-threading mode, and it should work by setting exception
to the wasm instance.
Loggers (e.g. glog) usually come with instrumentation to add timestamp
and other information when reporting. That results in the timestamp
being reported twice, making the output confusing.
According to the specification:
```
When instantiating a module which is expected to run
with `wasi-threads`, the WASI host must first allocate shared memories to
satisfy the module's imports.
```
Currently, if a test from the spec is executed while having the `multi-module`
feature enabled, WAMR fails with `WASM module load failed: unknown import`.
That happens because spec tests use memory like this:
`(memory (export "memory") (import "foo" "bar") 1 1 shared)`
and WAMR tries to find a registered module named `foo`.
At the moment, there is no specific module name that can be used to identify
that the memory is imported because using WASI threads:
https://github.com/WebAssembly/wasi-threads/issues/33,
so this PR only avoids treating the submodule dependency not being found
as a failure.
It's possible to set both `atim` and `atim_now` in the `fstflags`
parameter. Same goes for `mtin` and `mtim_now`. However, it's
ambiguous which time should be set in these two cases. This commit
checks this and returns `EINVAL`.
A wasm module can be either a command or a reactor, so it can export
either `_start` or `_initialize`. Currently, if a command module is run,
`iwasm` still looks for `_initialize`, resulting in the warning:
`can not find an export 0 named _initialize in the module`.
Change to look for `_initialize` only if `_start` not found to resolve the issue.
This fixes bug #2880. Zephyr 3.2 made changes to how headers are reference (see [release notes](https://docs.zephyrproject.org/latest/releases/release-notes-3.2.html)). Work item [49578](https://github.com/zephyrproject-rtos/zephyr/issues/49578) deprecated the old headers names.
The current WAMR codebase references these old headers, thus causing compile errors with
current versions of Zephyr.
This update adds #ifdefs around the header names. With this change, compiling with Zephyr 3.2.0
and above will use the new header files. Prior versions will use the existing code.
This commit adds a check to `fd_advise`. If the fd is a directory,
return `ebadf`. This brings iwasm in line with Wasmtime's behavior.
WASI folks have stated that fd_advise should not work on directories
as this is a Linux-specific behavior:
https://github.com/bytecodealliance/wasmtime/issues/6505#issuecomment-1574122949
- Fix op_br_table arity type check when the dest block is loop block
- Fix op_drop issue when the stack is polymorphic and it is to drop
an ANY type value in the stack
* Empty names are spec-wise valid.
* As we ignore unknown custom sections anyway, it's safe to
accept empty names here.
* Currently, the problem is not exposed on our CI because
the wabt version used there is a bit old.
For shared memory, runtime should get the memories pointer from
module_inst first, then get memory instance from memories array,
and then get the fields of the memory instance.
Support new a wasm_config_t, set allocation and linux_perf_support
options to it, and then pass it to wasm_engine_new_with_config to
new an engine with private configuration.
The JSON evidence is allocated on the module instance heap, but no API
was given to dispose of this memory buffer. The sample mentions using
the function free, which behaves differently depending on the
execution context.
This fix provides a new function called librats_dispose_evidence_json,
enabling freeing the JSON evidence directly from the Wasm app.
Change WASMMemoryInstance's field is_shared_memory's type from bool
to uint8 whose size is fixed, so as to make WASMMemoryInstance's size
and layout fixed and not break AOT ABI.
See discussion in https://github.com/bytecodealliance/wasm-micro-runtime/pull/2682.
The popped reachable block may be if block whose else branch hasn't been
translated, and should push the params for the else block if there are.
And use LLVMDisposeMessage to free memory allocated in is_win_platform.
Currently, `data.drop` instruction is implemented by directly modifying the
underlying module. It breaks use cases where you have multiple instances
sharing a single loaded module. `elem.drop` has the same problem too.
This PR fixes the issue by keeping track of which data/elem segments have
been dropped by using bitmaps for each module instances separately, and
add a sample to demonstrate the issue and make the CI run it.
Also add a missing check of dropped elements to the fast-jit `table.init`.
Fixes: https://github.com/bytecodealliance/wasm-micro-runtime/issues/2735
Fixes: https://github.com/bytecodealliance/wasm-micro-runtime/issues/2772
CryptGenRandom is deprecated by Microsoft and may be removed in future
releases. They recommend to use the next generation API instead. See
https://learn.microsoft.com/en-us/windows/win32/seccng/cng-portal for
more details. Also, refactor the random functions to return error codes
rather than aborting the program if they fail.
Error is reported when executing `wamrc --target=thumb -o <aot_file> <wasm_file>`:
```
LLVM ERROR: failed to perform tail call elimination on a call site marked musttail
Aborted (core dumped)
```
Set `abi` to "gnu" for the bare-metal target when `abi` is NULL,
or the below `bh_assert` and `bh_memcpy` may deference a NULL
pointer. Error is reported when running wamrc compiled with
`cmake .. -DCMAKE_BUILD_TYPE=Debug`:
```
core/iwasm/compilation/aot_llvm.c:2584:13: runtime error:
null pointer passed as argument 1, which is declared to never be null
```
Add an extra argument `os_file_handle file` for `os_mmap` to support
mapping file from a file fd, and remove `os_get_invalid_handle` from
`posix_file.c` and `win_file.c`, instead, add it in the `platform_internal.h`
files to remove the dependency on libc-wasi.
Signed-off-by: Huang Qi <huangqi3@xiaomi.com>
Heap corruption check in ems memory allocator is enabled by default
to improve the security, but it may impact the performance a lot, this
PR adds cmake variable and compiler flag to enable/disable it.
Returning uint16 from WASI functions is technically correct. However,
the smallest integer type in WASM is int32 and since we don't guarantee
that the upper 16 bits of the result are zero'ed, it can result in
tricky bugs if the language SDK being used in the WASM app does not cast
back immediately to uint16. To prevent this, we directly return uint32
instead, so that the result is well-defined as a 32-bit number.
Set the vendor-sys of bare-metal targets to "-unknown-none-",
and currently only add "thumbxxx" to the bare-metal target list.
Signed-off-by: Huang Qi <huangqi3@xiaomi.com>
The commit fa5e9d72b0 ("Abstract POSIX filesystem functions") introduces
the build warning:
./core/iwasm/libraries/libc-wasi/sandboxed-system-primitives/src/posix.c: In function ‘fd_object_release’:
./core/iwasm/libraries/libc-wasi/sandboxed-system-primitives/src/posix.c:545:20: warning: this statement may fall through [-Wimplicit-fallthrough=]
545 | if (os_is_dir_stream_valid(&fo->directory.handle)) {
| ^
./core/iwasm/libraries/libc-wasi/sandboxed-system-primitives/src/posix.c:549:13: note: here
549 | default:
| ^~~~~~~
Refer to the commit fb4afc7ca4 ("Apply clang-format for core/iwasm compilation and libraries"),
add one line "// Fallthrough." to make compiler happy.
Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
`platform_common.h` already has a declaration for BH_VPRINTF so we can
get rid of the one in `platform_internal.h`. Also add some explicit
casts to avoid MSVC compiler warnings.
UWP apps do not have a console attached so any output to stdout/stderr
is lost. Therefore, provide a default BH_VPRINTF in that case for debug
builds which redirects output to the debugger.
To allow anything to depend on WASI types, including platform-specific
data structures, move the WASI libc filesystem/clock interface into
`platform_api_extension.h`, which leaves just WASI types in
`platform_wasi.h`. And `platform_wasi.h` has been renamed to
`platform_wasi_types.h` to reflect that it only defines types now and no
function declarations. Finally, these changes allow us to remove the
`windows_fdflags` type which was essentially a duplicate of
`__wasi_fdflags_t`.
Most of the WASI filesystem tests require at least creating/deleting a
file to test filesystem functionality so some additional filesystem APIs
have been implemented on Windows so we can test what has been
implemented so far. For those WASI functions which haven't been
implemented, we skip the tests. These will be implemented in a future PR
after which we can remove the relevant filters.
Additionally, in order to run the WASI socket and thread tests, we need
to install the wasi-sdk in CI and build the test source code prior to
running the tests.
`jit_reg_is_const_val` only checks whether the register is a const register and
the const value is stored in the register.
Should use `jit_reg_is_const` instead in the front end.
Reported in #2710.
- Fix potential invalid push param phis and add incoming phis to a un-existed basic block
- Fix potential invalid shift count int rotl/rotr opcodes
- Resize memory_data_size to UINT32_MAX if it is 4G when hw bound check is enabled
- Fix negative linear memory offset is used for 64-bit target it is const and larger than INT32_MAX
To run it locally:
```bash
export TSAN_OPTIONS=suppressions=<path_to_tsan_suppressions.txt>
./test_wamr.sh <your flags> -T tsan
```
An example for wasi-threads would look like:
```bash
export TSAN_OPTIONS=suppressions=<path_to_tsan_suppressions.txt>
./test_wamr.sh -w -s wasi_certification -t fast-interp -T tsan
```
Split memory instance's field `uint32 ref_count` into `bool is_shared_memory`
and `uint16 ref_count`, and lock the memory only when `is_shared_memory`
flag is true, no need to acquire a lock for non-shared memory when shared
memory feature is enabled.
Avoid repeatedly initializing the shared memory data when creating the child
thread in lib-pthread or lib-wasi-threads.
Add shared memory lock when accessing some fields of the memory instance
if the memory instance is shared.
Init shared memory's memory_data_size/memory_data_end fields according to
the current page count but not max page count.
Add wasm_runtime_set_mem_bound_check_bytes, and refine the error message
when shared memory flag is found but the feature isn't enabled.
Fixes the Cosmopolitan Libc platform attempting to use `/dev/urandom`
on operating systems that do not have it.
Signed-off-by: G4Vi <gavin@dylibso.com>
This patch enables mapping host directories to guest directories by parsing
the `map_dir_list` argument in API `wasm_runtime_init_wasi` for libc-wasi. It
follows the format `<guest-path>::<host-path>`.
It also adds argument `--map-dir=<guest::host>` argument for `iwasm`
common line tool, and allows to add multiple mappings:
```bash
iwasm --map-dir=<guest-path1::host-path1> --map-dir=<guest-path2::host-path2> ...
```
Implement the necessary os_ filesystem functions to enable successful
WASI initialization on Windows. Some small changes were also required to
the sockets implementation to use the new windows_handle type. The
remaining functions will be implemented in a future PR.
When labels-as-values is enabled in a target which doesn't support
unaligned address access, 16-bit offset is used to store the relative
offset between two opcode labels. But it is a little small and the loader
may report "pre-compiled label offset out of range" error.
Emitting 32-bit data instead to resolve the issue: emit label address in
32-bit target and emit 32-bit relative offset in 64-bit target.
See also:
https://github.com/bytecodealliance/wasm-micro-runtime/issues/2635
To allow non-POSIX platforms such as Windows to support WASI libc
filesystem functionality, create a set of wrapper functions which provide a
platform-agnostic interface to interact with the host filesystem. For now,
the Windows implementation is stubbed but this will be implemented
properly in a future PR. There are no functional changes in this change,
just a reorganization of code to move any direct POSIX references out of
posix.c in the libc implementation into posix_file.c under the shared
POSIX sources.
See https://github.com/bytecodealliance/wasm-micro-runtime/issues/2495 for a
more detailed overview of the plan to port the WASI libc filesystem to Windows.
When doing more investigations related to this PR:
https://github.com/bytecodealliance/wasm-micro-runtime/pull/2619
We found that in some scenarios the constant might not be directly
available to the LLVM IR builder, e.g.:
```
(func $const_ret (result i32)
i32.const -5
)
(func $foo
(i32.shr_u (i32.const -1) (call $const_ret))
(i32.const 31)
)
```
In that case, the right parameter to `i32.shr_u` is not constant, therefore
the `SHIFT_COUNT_MASK` isn't applied. However, when the optimization
is enabled (`--opt-level` is 2 or 3), the optimization passes resolve the
call into constant, and that constant is poisoned, causing the compiler to
resolve the whole function to an exception.
According to the description of `buildPerModuleDefaultPipeline()` and
`buildLTOPreLinkDefaultPipeline()`, it is not allowed to call them with `O0` level.
Use `buildO0DefaultPipeline` instead when the opt-level is 0.
The LLVM zext IR may be inserted after the terminator of a basic block
when popping the arguments of a wasm block. Change to insert the
zext IR before the terminator of the basic block to resolve the issue.
Reported in #2620.
This PR adds the Cosmopolitan Libc platform enabling compatibility with multiple
x86_64 operating systems with the same binary. The platform is similar to the
Linux platform, but for now only x86_64 with interpreter modes are supported.
The only major change to the core is `posix.c/convert_errno()` was rewritten to use
a switch statement. With Cosmopolitan errno values depend on the currently
running operating system, and so they are non-constant and cannot be used in array
designators. However, the `cosmocc` compiler allows non-constant case labels in
switch statements, enabling the new version.
And updated wamr-test-suites script to add `-j <platform>` option. The spec tests
can be ran via `CC=cosmocc ./test_wamr.sh -j cosmopolitan -t classic-interp`
or `CC=cosmocc ./test_wamr.sh -j cosmopolitan -t fast-interp`.
To make it clearer to users when synchronization behaviour is not
supported, return ENOTSUP when O_RSYNC, O_DSYNC or O_SYNC are
respectively not defined. Linux also doesn't support O_RSYNC despite the
O_RSYNC flag being defined.
Fixed a bug in the processing of the br_table_cache opcode that caused out-of-range
references when the label index was greater than the length of the label.
