type mismatch: values remaining on stack at end of block about wasmer HOT 4 CLOSED

Not maintainer. Just to answer your question. Wasm is a stack machine with an implicit stack. So an instruction like i32.const 10 pushes a 32-bit integer with a value of 10 onto the stack. An instruction like i32.add pops 2 values off the stack, adds them, and pushes the result back onto the stack. This error can occur if the generated wasm did not "typecheck" (validate) successfully and there are values remaining on the stack. So a good block without any values remaining may look like:

i32.const 1 ; 1 value on the stack
i32.const 2 ; 2 values on the stack
i32.add     ; 1 value on the stack
drop        ; no values on the stack. All good!

from wasmer.

Oh thank you for this information, yeah, I think I have a sense of what's wrong here. I think if I go over the components carefully, I can figure the error out. But I'm also wondering if there are some easier way to ping-poing the issue. When I use wasm2wat on the newhello.wasm, I got more detailed error description of the mismach

root@2299d4e20d1f:/lind-glibc/replace-sysroot/test# wasm2wat newhello.wasm -o newhello.wat
newhello.wasm:00001c6: error: type mismatch at end of function, expected [] but got [i32, i32]

Then is there a way to know what is this function that causes this?
Since I'm building my own libc, and manually locating the error is probably too hard due to the complexity of the codebase.

from wasmer.

Okay I see a good way, while wasm2wat checks the stack issue and give the offset of where the error occurs, we can use wasm-objdump -d to look at the offset position in the file, and know which function is causing the type mismatch

from wasmer.

@yagehu thanks for jumping in...

from wasmer.