feat(pageserver): use fnv hash for aux file encoding

[skyzh]

Problem

FNV hash is simple, portable, and stable. This pull request vendors the FNV hash implementation from servo and modified it to use the u128 variant.

replaces #7644

ref #7462

Summary of changes

Checklist before requesting a review

• I have performed a self-review of my code.
• If it is a core feature, I have added thorough tests.
• Do we need to implement analytics? if so did you add the relevant metrics to the dashboard?
• If this PR requires public announcement, mark it with /release-notes label and add several sentences in this section.

Checklist before merging

• Do not forget to reformat commit message to not include the above checklist

[github-actions]

3060 tests run: 2933 passed, 0 failed, 127 skipped (full report)

---

Code coverage* (full report)

• functions: 31.4% (6339 of 20190 functions)
• lines: 47.4% (47964 of 101267 lines)

* collected from Rust tests only

---

_{The comment gets automatically updated with the latest test results
c5f03a8 at 2024-05-14T15:46:38.218Z :recycle:}

[hlinnaka]

We can modify the algorithm to use 128b instead, but I didn't do it in this pull request, as I don't know if using 128b still preserves the properties of the hash algorithm.

128 bit and other length variants exist, see https://en.wikipedia.org/wiki/Fowler%E2%80%93Noll%E2%80%93Vo_hash_function. That particular crate only implements the 64 bit variant but there are others, and it's not hard to roll your own.

I'd assume 64 bits to be enough, though.

[hlinnaka]

Do we need to worry about malicious actors creating intentional collisions here? What happens if there's a ton of collisions? Does it only cause trouble for the single tenant, or the whole pageserver?

[skyzh]

switched to FNV128 using the prime + offset in the wiki, better to get a review again, or I'll merge it later today.

[skyzh]

Okay it still requires a review from code owner 🤣 @koivunej do you have time to take a look? Thanks :)

[skyzh]

Do we need to worry about malicious actors creating intentional collisions here? What happens if there's a ton of collisions? Does it only cause trouble for the single tenant, or the whole pageserver?

Creating intentional collision might cause write amplifications like before John’s one-file-one-delta-record PR, and it will only cause trouble for a single tenant. We can change the encoding of aux file value to delta + image in the future if it indeed becomes a problem.