fix: upgrade hwloc

[jmg-duarte]

When running rust-fil-proofs under Rust 1.81 the following warning is issued:

warning: the following packages contain code that will be rejected by a future version of Rust: bitflags v0.7.0
note: to see what the problems were, use the option `--future-incompat-report`, or run `cargo report future-incompatibilities --id 1`

The package depending on bitflags 0.7.0 is hwloc, which has been replaced by hwloc2.
Understandably, hwloc (the original library) has suffered changes as well, which are reflected in hwloc2.

The main one being the removal of allowed_cpuset, according to the hwloc upgrading guide (https://www.open-mpi.org/projects/hwloc/doc/v2.8.0/a00374.php#upgrade_to_api_2x_allowed) these cpuset should be enough given that the INCLUDE_DISALLOWED flag hasn't been passed (and as far as I know, it isn't a default).

This PR attempts to address these issues.

[cryptonemo]

Hello @jmg-duarte -- thanks for submitting this, we'll take a look!

[jmg-duarte]

@cryptonemo I just noticed the stacked::vanilla::cores::tests::test_cores is failing

I can't quite understand why since the docs I pointed too seem to indicate it should work?
Either way, I'm keen on getting this merged, I may need some pointers where to look for references though.

Thanks!

[cryptonemo]

@cryptonemo I just noticed the stacked::vanilla::cores::tests::test_cores is failing

I can't quite understand why since the docs I pointed too seem to indicate it should work? Either way, I'm keen on getting this merged, I may need some pointers where to look for references though.

Thanks!

That particular test can fail if other tests are running at the same time. CI should isolate it properly.

[jmg-duarte]

If hwloc2 does not cut it, the alternative I found is hwlocality, it requires more changes but at least it is maintained 😓

[cryptonemo]

If hwloc2 does not cut it, the alternative I found is hwlocality, it requires more changes but at least it is maintained 😓

CI aside, what we need for this to 'cut it' is to determine if there's any noticeable (negative) performance impact on multicore-sdr sealing (in particular for production sector sizes of 32/64GiB). Gathering that data will take some time.

[cryptonemo]

My testing of production sectors checks out(!) @DrPeterVanNostrand @vmx Any inputs, or good to go?

[vmx]

@cryptonemo Thanks for testing it out. I think then it's good to go.