================
----------------
daxpedda wrote:
Not that I can tell.
I've tried simply adding a calling function but that doesn't work either:
```c
__attribute__((import_module("env"), import_name("test_slot")))
extern void test_slot(unsigned int foo, ...);
void test_u32(void) {
test_slot(123u, 456u, 789u);
}
```
generates the following input:
```llvm
entry:
call void (i32, ...) @test_slot(i32 noundef 123, i32 noundef 456, i32 noundef
789)
ret void
```
AFAICT we can't properly test it via LLVM IR. The way I tested it and confirmed
it works is by compiling the same code via Clang and then using `wasm-tools
parse -t` to look at the WAT output:
```wat
(func $test_u32 (;2;) (type 1)
(local i64)
global.get $__stack_pointer
i64.const 16
i64.sub
local.set 0
local.get 0
global.set $__stack_pointer
local.get 0
i32.const 789
i32.store offset=8
local.get 0
i32.const 456
i32.store
i32.const 123
local.get 0
call $test_slot
local.get 0
i64.const 16
i64.add
global.set $__stack_pointer
return
)
```
With this patch the `offset` is correctly `8`. Just to make sure I tested it
without the patch which produces an offset of `4`.
If you have any idea how I can test this with the existing LLVM infrastructure
please let me know!
https://github.com/llvm/llvm-project/pull/173580
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
