`alloc` wanted features & bugfixes

[ojeda]

Note: there is the "A-rust-for-linux" label in the rust repository: https://github.com/rust-lang/rust/labels/A-rust-for-linux.

Note: In Linux v6.13, Linux started using its own alloc module instead of the Rust standard library alloc crate.

Note: In Linux v6.10, Linux removed its custom fork of the Rust standard library alloc crate.

Features that we would like to see

Required (we almost certainly want them)

Nice to have (not critical, we could workaround if needed, etc.)

Low priority (we will likely not use them in the end)

Done (stabilized, fixed, not needed anymore, etc.)

• Disabling infallible allocations.

  • PR: alloc: Add unstable Cfg feature `no_global_oom_handling rust-lang/rust#84266.

• Arc::try_pin.

  • PR: Added Arc::try_pin rust-lang/rust#85579.

• try_* methods for alloc, usable under no_global_oom_handling.

  • We currently extend Box and Vec via extensions traits and a destructure/rebuild pair of functions, so that we avoid having a custom alloc fork like in the past (Custom alloc, using stable Rust compilers, and more #402).
  • Issue: Tracking issue for try_reserve: RFC 2116 fallible collection allocation rust-lang/rust#48043.
  • Issue in wg-allocators: try_* methods for containers rust-lang/wg-allocators#39.
  • Issue in wg-allocators: Allocator API taking context rust-lang/wg-allocators#138.
  • PR: alloc: add some try_* methods Rust-for-Linux needs  rust-lang/rust#86938.
  • Issue: Tracking Issue for more_fallible_allocation_methods rust-lang/rust#86942.
  • Issue: Consider adding Vec::try_with_capacity(_in) and friends rust-lang/rust#91913.
  • PR: vec: add try_* methods and a try_vec! macro to make Vec usable in without infallible allocation methods rust-lang/rust#95051.
  • PR: alloc: make vec! unavailable under no_global_oom_handling rust-lang/rust#96089.
  • Experiment PR: fallible allocator experiment rust-lang/rust#111970.
  • Zulip: https://rust-lang.zulipchat.com/#narrow/stream/122651-general/topic/Allocator.20API.20for.20fallible.20alloc.20actions/near/359086699.
  • Zulip: https://rust-lang.zulipchat.com/#narrow/stream/197181-t-libs.2Fwg-allocators/topic/Split.20or.20merge.20std.2Falloc.2Fcore.3F/near/359717930.
  • Zulip: https://rust-lang.zulipchat.com/#narrow/channel/219381-t-libs/topic/no_global_oom_handling.20removal/near/498600545.

• Disabling rc and sync.

  • PR: alloc: add unstable cfg features no_rc and no_sync rust-lang/rust#89891.

• Ensure no_global_oom_handling builds are warning-free.

  • Currently we have the in-tree fork, but if in the future we want to use the upstream one, it would be best to have warning-free builds like our fork (for which we use -Dwarnings).
  • PR: alloc: clean and ensure no_global_oom_handling builds are warning-free rust-lang/rust#98652.

---

Bugs that we would like to see fixed

Required (we almost certainly want them)

Nice to have (probably not critical, we could workaround if needed, etc.)

Low priority (we will likely not use them in the end)

Done (stabilized, fixed, or not needed anymore, etc.)

• alloc: RawVec<T, A>::shrink can be in no_global_oom_handling. rust-lang/rust#86755.
• alloc: no_global_oom_handling: disable new()s, pin()s, etc. rust-lang/rust#86810.

[Ericson2314]

Should we try to turn 487d757 into a PR for rust next?

[alex]

That looks good -- it might make sense to split it into multiple pieces though.

…

On Sat, Jul 3, 2021 at 1:00 PM John Ericson ***@***.***> wrote: Should we try to turn 487d757 <487d757> into a PR for rust next? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#408 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAAGBB5E6EUDUQ7FVTJ3QTTV462HANCNFSM47S4VRCQ> .

-- All that is necessary for evil to succeed is for good people to do nothing.

[ojeda]

Yeah, I think many of the new methods can likely be taken without much fuss to upstream -- I started with the easy bits first.

I am not 100% sure about whether upstream would want to use TryReserveError in all the places I used it, though.

[alex]

Arc::try_pin is now merged

[wedsonaf]

Arc::try_pin is now merged

Alex, we are actually deprecating Arc in favor of Ref (see #377). The only place still using it is binder; I have a long series that removes it but it still needs some cleanup before being ready for review (which is taking longer because I'm working on a hw driver now).

[alex]

Good point. Well, it'll be there for the next person. בתאריך יום ה׳, 15 ביולי 2021, 14:25, מאת Wedson Almeida Filho ‏< ***@***.***>:

…

Arc::try_pin is now merged Alex, we are actually deprecating Arc in favor of Ref (see #377 <#377>). The only place still using it is binder; I have a long series that removes it but it still needs some cleanup before being ready for review (which is taking longer because I'm working on a hw driver now). — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#408 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAAGBHRVTHJ3XD4YFT6UFLTX4RYZANCNFSM47S4VRCQ> .