CODEWITHSUNDEEP
webassembly
Asperger
Asperger

Reputation: 3222

Webassembly growing memory and importing malloc

After reading the specifications I was trying to grow some memory and did the following:

(module
  (table 0 anyfunc)
  (memory $0 1)
  (export "memory" (memory $0))
  (export "f" (func $0))
  (func $0 (param $0 i32) (result i32)
    get_local $0
    grow_memory
  )
)

Now I was hoping that if we input 10 that we have a page size multiple of 11, so something like:

 (memory $0 1)

Becomes 

 (memory $0 11)

I had to add a return value since the stack wasnt emptied. Now I assume that the method returns something other than -1 if it succeeds right? In this case I think it works and I should have more pages.

I know there is also the JS way but I prefer doing all this stuff directly inside WebAssembly. Interesting though is the fact that we can import C functions such as malloc, free, strcopy etc from "Env"..

So my question is: is my approach for growing the memory correct? Should I import the C style modules instead? What is the relation between grow_memory and realloc(), or say the difference?

Upvotes: 1

Views: 2956

Answers (1)

JF Bastien
JF Bastien

Reputation: 6863

grow_memory returns the previous size (in pages) or -1 according to its specification.

This doesn't change the initial page allocation (memory $0 1)! That's a static property of the module, whereas current memory size is a dynamic property of the memory. It's like growing an array which you initially reserved 1 element for. You still initially reserved 1, what's changed is the current size.

You could also use drop to ignore the return value of grow_memory.

malloc / free / realloc are all things you can implement using grow_memory. That opcode is similar to sbrk. Of course, the WebAssembly embedder also uses memory allocation to grant your opcode's request, which may itself be implemented with malloc / free / realloc, or more likely with mmap.

Upvotes: 2

Related Questions