wasmtime / PR #8489 refactor(cranelift-codegen): remove `... · git-wasmtime

Stream: git-wasmtime

Topic: wasmtime / PR #8489 refactor(cranelift-codegen): remove `...

Wasmtime GitHub notifications bot (Apr 26 2024 at 10:02):

JonasKruckenberg opened PR #8489 from JonasKruckenberg:no_once_lock to bytecodealliance:main:

This removes the dependency on std::sync::SpinLock by lifting the state out of a static and into the Callee struct.

This PR also removes the call_conv parameter since it was not used by any of the trait implementations.

Wasmtime GitHub notifications bot (Apr 26 2024 at 10:02):

JonasKruckenberg requested elliottt for a review on PR #8489.

Wasmtime GitHub notifications bot (Apr 26 2024 at 10:02):

JonasKruckenberg requested wasmtime-compiler-reviewers for a review on PR #8489.

Wasmtime GitHub notifications bot (Apr 28 2024 at 21:24):

JonasKruckenberg edited PR #8489.

Wasmtime GitHub notifications bot (Apr 28 2024 at 21:25):

JonasKruckenberg edited PR #8489:

This removes the dependency on std::sync::OnceLock by lifting the state out of a static and into the Callee struct.

This PR also removes the call_conv parameter since it was not used by any of the trait implementations.

Wasmtime GitHub notifications bot (May 02 2024 at 17:50):

alexcrichton commented on PR #8489:

ping @elliottt (or other cranelift folks on this), I don't know enough about MachineEnv myself to know if the lift here is reasonable to apply

Wasmtime GitHub notifications bot (May 02 2024 at 19:05):

cfallin commented on PR #8489:

This could possibly result in a performance degradation in the compiler: it removes the caching of the MachineEnv, which contains lists of registers (heap-allocated Vecs), and recreates it for each function that is compiled. Especially for small functions this could be a measurable/non-negligible portion of the total work.

@JonasKruckenberg, could you say more about the motivation here? And have you measured the performance impact?

Wasmtime GitHub notifications bot (May 02 2024 at 21:41):

JonasKruckenberg commented on PR #8489:

Yeah thats a concern I had too.
The motivation is to support #8341 and since OnceLock is the only thing in cranelift that cant be ported to no_std+alloc or gated behind flags the idea was to replace it by something else.
Adding another dependency on something like spin for no_std targets doesn't seem appropriate either.

I'll do some benchmarking next week and report back :+1:

Wasmtime GitHub notifications bot (May 02 2024 at 22:52):

cfallin commented on PR #8489:

Ah, one other option could be to put it in the MachBackend instead -- if it's the case that it no longer depends on the call_conv. If it does on some backends (x64 might differentiate with fastcall?), we could statically construct the various options. That way, it's constructed once per compiler session, but without the lazy init.

Wasmtime GitHub notifications bot (May 10 2024 at 19:05):

alexcrichton commented on PR #8489:

FWIW I ran this locally and on sightglass it reported a 1-3% slowdown in compiling spidermonkey.wasm, but no changes compiling other benchmarks. (I didn't re-run to see if that number was noise.

Wasmtime GitHub notifications bot (May 10 2024 at 19:05):

alexcrichton edited a comment on PR #8489:

FWIW I ran this locally and on sightglass it reported a 1-3% slowdown in compiling spidermonkey.wasm, but no changes compiling other benchmarks. (I didn't re-run to see if that number was noise)

Wasmtime GitHub notifications bot (May 12 2024 at 21:30):

jameysharp commented on PR #8489:

@uweigand introduced the call_conv parameter on get_machine_env in #6957 with the intent to use it to have different sets of allocatable registers for the tail calling convention than for the system calling convention. We talked about that a little at the Cranelift meeting last week and it sounds to me like he still wants to use it that way, so I don't think we should remove it.

Regarding the main point of this PR though: I believe we can get rid of the need for OnceLock here by changing regalloc2 so MachineEnv holds a pair of PRegSet, which are small and const-evaluatable, instead of the six Vecs. Then a MachineEnv can be constructed in the initializer of a static item (note that Cranelift always sets fixed_stack_slots to an empty Vec, and Vec::new is a const fn). Then we can just return a borrow of a constant static in all of these cases, possibly choosing between several definitions based on the flags or calling convention.

Wasmtime GitHub notifications bot (May 13 2024 at 15:20):

uweigand commented on PR #8489:

@uweigand introduced the call_conv parameter on get_machine_env in #6957 with the intent to use it to have different sets of allocatable registers for the tail calling convention than for the system calling convention. We talked about that a little at the Cranelift meeting last week and it sounds to me like he still wants to use it that way, so I don't think we should remove it.

It's not completely certain, depending on choice of frame pointer register, but in general it would be good to have this option. Allocatable registers in general may depend on calling convention.

Last updated: Nov 22 2024 at 16:03 UTC