1
0
Fork 0
mirror of https://github.com/denoland/deno.git synced 2024-11-28 16:20:57 -05:00
denoland-deno/ops
Aapo Alasuutari 04ba709b6e
perf(ops): Remove unnecessary fast call fallback options usage (#17585)
Currently fast ops will always check for the alignment of a TypedArray
when getting a slice out of them. A match is then done to ensure that
some slice was received and if not a fallback will be requested.

For Uint8Arrays (and WasmMemory which is equivalent to a Uint8Array) the
alignment will always be okay. Rust probably optimises this away for the
most part (since the Uint8Array check is `x % 1 != 0`), but what it
cannot optimise away is the fast ops path's request for fallback options
parameter.

The extra parameter's cost is likely negligible but V8 will need to
check if a fallback was requested and prepare the fallback call just in
case it was. In the future the lack of a fallback may also enable V8 to
much better optimise the result handling.

For V8 created buffers, it seems like all buffers are actually always
guaranteed to be properly aligned: All buffers seem to always be created
8-byte aligned, and creating a 32 bit array or 64 bit array with a
non-aligned offset from an ArrayBuffer is not allowed. Unfortunately,
Deno FFI cannot give the same guarantees, and it is actually possible
for eg. 32 bit arrays to be created unaligned using it. These arrays
work fine (at least on Linux) so it seems like this is not illegal, it
just means that we cannot remove the alignment checking for 32 bit
arrays.
2023-01-29 19:35:08 +05:30
..
optimizer_tests perf(ops): Remove unnecessary fast call fallback options usage (#17585) 2023-01-29 19:35:08 +05:30
tests/compile_fail fix(ops): disallow memory slices as inputs to async ops (#16738) 2023-01-15 07:40:01 +00:00
attrs.rs chore: use rustfmt imports_granularity option (#17421) 2023-01-14 23:18:58 -05:00
Cargo.toml 1.30.0 (#17532) 2023-01-26 00:15:08 +01:00
deno.rs chore: use rustfmt imports_granularity option (#17421) 2023-01-14 23:18:58 -05:00
fast_call.rs chore: upgrade to Rust 1.67 (#17548) 2023-01-27 10:43:16 -05:00
lib.rs fix(ops): disallow memory slices as inputs to async ops (#16738) 2023-01-15 07:40:01 +00:00
optimizer.rs perf(ops): Remove unnecessary fast call fallback options usage (#17585) 2023-01-29 19:35:08 +05:30
README.md feat(ops): fast calls for Wasm (#16776) 2022-11-27 19:24:28 +05:30

deno_ops

proc_macro for generating highly optimized V8 functions from Deno ops.

// Declare an op.
#[op(fast)]
pub fn op_add(_: &mut OpState, a: i32, b: i32) -> i32 {
  a + b
}

// Register with an extension.
Extension::builder()
  .ops(vec![op_add::decl()])
  .build();

Performance

The macro can optimize away code, short circuit fast paths and generate a Fast API impl.

Cases where code is optimized away:

  • -> () skips serde_v8 and rv.set calls.
  • -> Result<(), E> skips serde_v8 and rv.set calls for Ok() branch.
  • -> ResourceId or -> [int] types will use specialized method like v8::ReturnValue::set_uint32. A fast path for SMI.
  • -> Result<ResourceId, E> or -> Result<[int], E> types will be optimized like above for the Ok() branch.

Fast calls

The macro will infer and try to auto generate V8 fast API call trait impl for sync ops with:

  • arguments: integers, bool, &mut OpState, &[u8], &mut [u8], &[u32], &mut [u32]
  • return_type: integers, bool

The #[op(fast)] attribute should be used to enforce fast call generation at compile time.

Trait gen for async ops & a ZeroCopyBuf equivalent type is planned and will be added soon.

Wasm calls

The #[op(wasm)] attribute should be used for calls expected to be called from Wasm. This enables the fast call generation and allows seamless WasmMemory integration for generic and fast calls.

#[op(wasm)]
pub fn op_args_get(
  offset: i32,
  buffer_offset: i32,
  memory: Option<&[u8]>, // Must be last parameter. Some(..) when entered from Wasm.
) {
  // ...
}