[PATCH] mini-gmp: pass correct old_size to custom reallocate function

[Niels Möller]

minux <minux.ma at gmail.com> writes:

> Yes, the old_size is supposed to be the size of the originally
> allocated memory block. But the problem is that some old gmp client
> used to rely on the old_size behave as documented:
> https://github.com/ghc/ghc/blob/d874b8c93b737bf26c949ef7bf19fc43e335bd1f/rts/sm/Storage.c#L935

Thanks, that's the first example I've seen of an application using
old_size.

In the alloc function (starting at line 913), it ends with

  SET_ARR_HDR(arr, &stg_ARR_WORDS_info, CCCS, data_size_in_words);
  
  /* and return a ptr to the goods inside the array */
  return arr->payload;

Maybe it would be easy for the realloc function to get the "arr" object
back and get data size from here?

> We couldn't just argue those clients are broken, because they
> implemented realloc exactly according to the docs:
> "The block may be moved if necessary or if desired, and in that case
> the smaller of old_size and new_size bytes must be copied to the new
> location."

Sure, this code uses GMP correctly, according to GMP docs. And mini-gmp
has a documentedly different behavior.

> I consider the risk of passing a slightly smaller old_size in the case
> of mpz_get_str is less risky because most allocations don't use
> old_size for memory management purposes as you have observed. Maybe we should
> document this difference for mini-gmp though.

I don't really like such a subtle difference in documented behavior.
Always passing zero is a less subtle uncompatibility, and will break
applications depending on the GMP behavior in a more obvious way.

If we change mini-gmp to be compatible with GMP here, I'd prefer to go
all the way and make it fully compatible. But we should do that only if
it's useful enough to warrant the extra complexity. ghc is one example,
but it's still unclear to me (i) if it's hard or easy to fix the ghc
allocator to use its own metadata, and (ii) what's the usecase and
context for building ghc with mini-gmp.

Regards,
/Niels

-- 
Niels Möller. PGP-encrypted email is preferred. Keyid 368C6677.
Internet email is subject to wholesale government surveillance.