Use non-zero page size

[JonathanWoollett-Light]

This avoids potential divide by zero error.

Fixes #268

Summary of the PR

Please summarize here why the changes in this PR are needed.

Requirements

Before submitting your PR, please make sure you addressed the following
requirements:

• All commits in this PR are signed (with git commit -s), and the commit
  message has max 60 characters for the summary and max 75 characters for each
  description line.
• All added/changed functionality has a corresponding unit/integration
  test.
• All added/changed public-facing functionality has entries in the "Upcoming
  Release" section of CHANGELOG.md (if no such section exists, please create one).
• Any newly added unsafe code is properly documented.

[Ablu]

This is breaking API and should have a CHANGELOG entry. While I see no problem with the approach, I wonder whether this could be reasoned a bit better in the commit message.

My immediate questions are: Where did this cause major problems? Why is making the user go through the NonZeroUsize conversion worth it?

Is this the only place where we deal with page_sizes? 🤔 Should we convert them all?

[JonathanWoollett-Light]

This is breaking API and should have a CHANGELOG entry. While I see no problem with the approach, I wonder whether this could be reasoned a bit better in the commit message.

I've added more details to the commit message.

My immediate questions are: Where did this cause major problems? Why is making the user go through the NonZeroUsize conversion worth it?

Without this it is possible to trigger a divide by zero error, see https://github.com/rust-vmm/vm-memory/blob/main/src/bitmap/backend/atomic_bitmap.rs#L28

Is this the only place where we deal with page_sizes? 🤔 Should we convert them all?

Are there other places you see potential errors that you think should be fixed?

[Ablu]

My immediate questions are: Where did this cause major problems? Why is making the user go through the NonZeroUsize conversion worth it?

Without this it is possible to trigger a divide by zero error, see https://github.com/rust-vmm/vm-memory/blob/main/src/bitmap/backend/atomic_bitmap.rs#L28

Sure! I was just wondering where that became a problem in "practise" :).

[JonathanWoollett-Light]

Sure! I was just wondering where that became a problem in "practise" :).

I understand your point, in practice I cannot see this occurring outside very exotic circumstances, but it is possible (as its a public function) so it should be mitigated.