dlmalloc: account the footprint of the initial heap

[yamt]

While malloc_stats and friends are disabled and unused for wasi-libc, it's neater to be consistent.

Background: My colleagues for some reasons enabled malloc_stats and asked me why it reports negative values.

Note: Depending __heap_base, init_top() adjusts the address for alignment. I think the amount of this adjustment is reported as "used" by malloc_stats. I don't bother to "fix" it.

[sunfishcode]

This looks reasonable to me. It doesn't even look wrong that this doesn't account for init_top adjusting for alignment, because it's about the amount of memory malloc has "obtained" from the system, whether it's used for user data or not.

We may want to #ifdef out the footprint and max_footprint fields altogether by default, to save code size, because we don't expose that functionality in the default wasi-libc configuration. But we can do that separately.

[sbc100]

Do you know why upstream dlmalloc doesn't do this? Is this a bug in dlmalloc?

Can you mark these lines with __wasilibc_unmodified_upstream like we do elsewhere?

[sunfishcode]

This code is code that we added for WASI, so it's already under __wasilibc_unmodified_upstream guards. And we don't expose the footprint functionality publicly, which is why we didn't notice this before.

[yamt]

This looks reasonable to me. It doesn't even look wrong that this doesn't account for init_top adjusting for alignment, because it's about the amount of memory malloc has "obtained" from the system, whether it's used for user data or not.

i agree it isn't necessarily wrong. i mentioned it just because my colleagues will likely have a question.

We may want to #ifdef out the footprint and max_footprint fields altogether by default, to save code size, because we don't expose that functionality in the default wasi-libc configuration. But we can do that separately.

maybe.