wasm-bindgen/src/anyref.rs

183 lines
5.8 KiB
Rust
Raw Normal View History

Add experimental support for the `anyref` type This commit adds experimental support to `wasm-bindgen` to emit and leverage the `anyref` native wasm type. This native type is still in a proposal status (the reference-types proposal). The intention of `anyref` is to be able to directly hold JS values in wasm and pass the to imported functions, namely to empower eventual host bindings (now renamed WebIDL bindings) integration where we can skip JS shims altogether for many imports. This commit doesn't actually affect wasm-bindgen's behavior at all as-is, but rather this support requires an opt-in env var to be configured. Once the support is stable in browsers it's intended that this will add a CLI switch for turning on this support, eventually defaulting it to `true` in the far future. The basic strategy here is to take the `stack` and `slab` globals in the generated JS glue and move them into wasm using a table. This new table in wasm is managed at the fringes via injected shims. At `wasm-bindgen`-time the CLI will rewrite exports and imports with shims that actually use `anyref` if needed, performing loads/stores inside the wasm module instead of externally in the wasm module. This should provide a boost over what we have today, but it's not a fantastic strategy long term. We have a more grand vision for `anyref` being a first-class type in the language, but that's on a much longer horizon and this is currently thought to be the best we can do in terms of integration in the near future. The stack/heap JS tables are combined into one wasm table. The stack starts at the end of the table and grows down with a stack pointer (also injected). The heap starts at the end and grows up (state managed in linear memory). The anyref transformation here will hook up various intrinsics in wasm-bindgen to the runtime functionality if the anyref supoprt is enabled. The main tricky treatment here was applied to closures, where we need JS to use a different function pointer than the one Rust gives it to use a JS function pointer empowered with anyref. This works by switching up a bit how descriptors work, embedding the shims to call inside descriptors rather than communicated at runtime. This means that we're accessing constant values in the generated JS and we can just update the constant value accessed.
2018-10-18 08:43:36 -07:00
use std::alloc::{self, Layout};
use std::mem;
2019-05-09 12:59:17 -07:00
use std::ptr;
use std::slice;
use std::vec::Vec;
use std::cell::Cell;
use crate::JsValue;
Add experimental support for the `anyref` type This commit adds experimental support to `wasm-bindgen` to emit and leverage the `anyref` native wasm type. This native type is still in a proposal status (the reference-types proposal). The intention of `anyref` is to be able to directly hold JS values in wasm and pass the to imported functions, namely to empower eventual host bindings (now renamed WebIDL bindings) integration where we can skip JS shims altogether for many imports. This commit doesn't actually affect wasm-bindgen's behavior at all as-is, but rather this support requires an opt-in env var to be configured. Once the support is stable in browsers it's intended that this will add a CLI switch for turning on this support, eventually defaulting it to `true` in the far future. The basic strategy here is to take the `stack` and `slab` globals in the generated JS glue and move them into wasm using a table. This new table in wasm is managed at the fringes via injected shims. At `wasm-bindgen`-time the CLI will rewrite exports and imports with shims that actually use `anyref` if needed, performing loads/stores inside the wasm module instead of externally in the wasm module. This should provide a boost over what we have today, but it's not a fantastic strategy long term. We have a more grand vision for `anyref` being a first-class type in the language, but that's on a much longer horizon and this is currently thought to be the best we can do in terms of integration in the near future. The stack/heap JS tables are combined into one wasm table. The stack starts at the end of the table and grows down with a stack pointer (also injected). The heap starts at the end and grows up (state managed in linear memory). The anyref transformation here will hook up various intrinsics in wasm-bindgen to the runtime functionality if the anyref supoprt is enabled. The main tricky treatment here was applied to closures, where we need JS to use a different function pointer than the one Rust gives it to use a JS function pointer empowered with anyref. This works by switching up a bit how descriptors work, embedding the shims to call inside descriptors rather than communicated at runtime. This means that we're accessing constant values in the generated JS and we can just update the constant value accessed.
2018-10-18 08:43:36 -07:00
externs! {
#[link(wasm_import_module = "__wbindgen_anyref_xform__")]
extern "C" {
fn __wbindgen_anyref_table_grow(delta: usize) -> i32;
fn __wbindgen_anyref_table_set_null(idx: usize) -> ();
}
}
pub struct Slab {
data: Vec<usize>,
head: usize,
base: usize,
}
impl Slab {
fn new() -> Slab {
Slab {
data: Vec::new(),
head: 0,
base: 0,
}
}
fn alloc(&mut self) -> usize {
let ret = self.head;
if ret == self.data.len() {
if self.data.len() == self.data.capacity() {
let extra = 128;
2019-05-09 12:59:17 -07:00
let r = unsafe { __wbindgen_anyref_table_grow(extra) };
Add experimental support for the `anyref` type This commit adds experimental support to `wasm-bindgen` to emit and leverage the `anyref` native wasm type. This native type is still in a proposal status (the reference-types proposal). The intention of `anyref` is to be able to directly hold JS values in wasm and pass the to imported functions, namely to empower eventual host bindings (now renamed WebIDL bindings) integration where we can skip JS shims altogether for many imports. This commit doesn't actually affect wasm-bindgen's behavior at all as-is, but rather this support requires an opt-in env var to be configured. Once the support is stable in browsers it's intended that this will add a CLI switch for turning on this support, eventually defaulting it to `true` in the far future. The basic strategy here is to take the `stack` and `slab` globals in the generated JS glue and move them into wasm using a table. This new table in wasm is managed at the fringes via injected shims. At `wasm-bindgen`-time the CLI will rewrite exports and imports with shims that actually use `anyref` if needed, performing loads/stores inside the wasm module instead of externally in the wasm module. This should provide a boost over what we have today, but it's not a fantastic strategy long term. We have a more grand vision for `anyref` being a first-class type in the language, but that's on a much longer horizon and this is currently thought to be the best we can do in terms of integration in the near future. The stack/heap JS tables are combined into one wasm table. The stack starts at the end of the table and grows down with a stack pointer (also injected). The heap starts at the end and grows up (state managed in linear memory). The anyref transformation here will hook up various intrinsics in wasm-bindgen to the runtime functionality if the anyref supoprt is enabled. The main tricky treatment here was applied to closures, where we need JS to use a different function pointer than the one Rust gives it to use a JS function pointer empowered with anyref. This works by switching up a bit how descriptors work, embedding the shims to call inside descriptors rather than communicated at runtime. This means that we're accessing constant values in the generated JS and we can just update the constant value accessed.
2018-10-18 08:43:36 -07:00
if r == -1 {
internal_error("table grow failure")
}
if self.base == 0 {
self.base = r as usize + (super::JSIDX_RESERVED as usize);
} else if self.base + self.data.len() != r as usize {
internal_error("someone else allocated table entires?")
}
// poor man's `try_reserve_exact` until that's stable
unsafe {
let new_cap = self.data.capacity() + extra;
let size = mem::size_of::<usize>() * new_cap;
let align = mem::align_of::<usize>();
let layout = match Layout::from_size_align(size, align) {
Ok(l) => l,
Err(_) => internal_error("size/align layout failure"),
};
let ptr = alloc::alloc(layout) as *mut usize;
if ptr.is_null() {
internal_error("allocation failure");
}
2019-05-09 12:59:17 -07:00
ptr::copy_nonoverlapping(self.data.as_ptr(), ptr, self.data.len());
let new_vec = Vec::from_raw_parts(ptr, self.data.len(), new_cap);
Add experimental support for the `anyref` type This commit adds experimental support to `wasm-bindgen` to emit and leverage the `anyref` native wasm type. This native type is still in a proposal status (the reference-types proposal). The intention of `anyref` is to be able to directly hold JS values in wasm and pass the to imported functions, namely to empower eventual host bindings (now renamed WebIDL bindings) integration where we can skip JS shims altogether for many imports. This commit doesn't actually affect wasm-bindgen's behavior at all as-is, but rather this support requires an opt-in env var to be configured. Once the support is stable in browsers it's intended that this will add a CLI switch for turning on this support, eventually defaulting it to `true` in the far future. The basic strategy here is to take the `stack` and `slab` globals in the generated JS glue and move them into wasm using a table. This new table in wasm is managed at the fringes via injected shims. At `wasm-bindgen`-time the CLI will rewrite exports and imports with shims that actually use `anyref` if needed, performing loads/stores inside the wasm module instead of externally in the wasm module. This should provide a boost over what we have today, but it's not a fantastic strategy long term. We have a more grand vision for `anyref` being a first-class type in the language, but that's on a much longer horizon and this is currently thought to be the best we can do in terms of integration in the near future. The stack/heap JS tables are combined into one wasm table. The stack starts at the end of the table and grows down with a stack pointer (also injected). The heap starts at the end and grows up (state managed in linear memory). The anyref transformation here will hook up various intrinsics in wasm-bindgen to the runtime functionality if the anyref supoprt is enabled. The main tricky treatment here was applied to closures, where we need JS to use a different function pointer than the one Rust gives it to use a JS function pointer empowered with anyref. This works by switching up a bit how descriptors work, embedding the shims to call inside descriptors rather than communicated at runtime. This means that we're accessing constant values in the generated JS and we can just update the constant value accessed.
2018-10-18 08:43:36 -07:00
let mut old = mem::replace(&mut self.data, new_vec);
old.set_len(0);
}
}
// custom condition to ensure `push` below doesn't call `reserve` in
// optimized builds which pulls in lots of panic infrastructure
if self.data.len() >= self.data.capacity() {
internal_error("push should be infallible now")
}
self.data.push(ret + 1);
}
// usage of `get_mut` thwarts panicking infrastructure in optimized
// builds
match self.data.get_mut(ret) {
Some(slot) => self.head = *slot,
None => internal_error("ret out of bounds"),
}
ret + self.base
}
fn dealloc(&mut self, slot: usize) {
if slot < self.base {
internal_error("free reserved slot");
}
let slot = slot - self.base;
// usage of `get_mut` thwarts panicking infrastructure in optimized
// builds
match self.data.get_mut(slot) {
Some(ptr) => {
*ptr = self.head;
self.head = slot;
}
None => internal_error("slot out of bounds"),
}
}
fn live_count(&self) -> u32 {
let mut free_count = 0;
let mut next = self.head;
while next < self.data.len() {
debug_assert!((free_count as usize) < self.data.len());
free_count += 1;
match self.data.get(next) {
Some(n) => next = *n,
None => internal_error("slot out of bounds"),
};
}
self.data.len() as u32 - free_count
}
Add experimental support for the `anyref` type This commit adds experimental support to `wasm-bindgen` to emit and leverage the `anyref` native wasm type. This native type is still in a proposal status (the reference-types proposal). The intention of `anyref` is to be able to directly hold JS values in wasm and pass the to imported functions, namely to empower eventual host bindings (now renamed WebIDL bindings) integration where we can skip JS shims altogether for many imports. This commit doesn't actually affect wasm-bindgen's behavior at all as-is, but rather this support requires an opt-in env var to be configured. Once the support is stable in browsers it's intended that this will add a CLI switch for turning on this support, eventually defaulting it to `true` in the far future. The basic strategy here is to take the `stack` and `slab` globals in the generated JS glue and move them into wasm using a table. This new table in wasm is managed at the fringes via injected shims. At `wasm-bindgen`-time the CLI will rewrite exports and imports with shims that actually use `anyref` if needed, performing loads/stores inside the wasm module instead of externally in the wasm module. This should provide a boost over what we have today, but it's not a fantastic strategy long term. We have a more grand vision for `anyref` being a first-class type in the language, but that's on a much longer horizon and this is currently thought to be the best we can do in terms of integration in the near future. The stack/heap JS tables are combined into one wasm table. The stack starts at the end of the table and grows down with a stack pointer (also injected). The heap starts at the end and grows up (state managed in linear memory). The anyref transformation here will hook up various intrinsics in wasm-bindgen to the runtime functionality if the anyref supoprt is enabled. The main tricky treatment here was applied to closures, where we need JS to use a different function pointer than the one Rust gives it to use a JS function pointer empowered with anyref. This works by switching up a bit how descriptors work, embedding the shims to call inside descriptors rather than communicated at runtime. This means that we're accessing constant values in the generated JS and we can just update the constant value accessed.
2018-10-18 08:43:36 -07:00
}
fn internal_error(msg: &str) -> ! {
if cfg!(debug_assertions) {
super::throw_str(msg)
} else {
std::process::abort()
}
Add experimental support for the `anyref` type This commit adds experimental support to `wasm-bindgen` to emit and leverage the `anyref` native wasm type. This native type is still in a proposal status (the reference-types proposal). The intention of `anyref` is to be able to directly hold JS values in wasm and pass the to imported functions, namely to empower eventual host bindings (now renamed WebIDL bindings) integration where we can skip JS shims altogether for many imports. This commit doesn't actually affect wasm-bindgen's behavior at all as-is, but rather this support requires an opt-in env var to be configured. Once the support is stable in browsers it's intended that this will add a CLI switch for turning on this support, eventually defaulting it to `true` in the far future. The basic strategy here is to take the `stack` and `slab` globals in the generated JS glue and move them into wasm using a table. This new table in wasm is managed at the fringes via injected shims. At `wasm-bindgen`-time the CLI will rewrite exports and imports with shims that actually use `anyref` if needed, performing loads/stores inside the wasm module instead of externally in the wasm module. This should provide a boost over what we have today, but it's not a fantastic strategy long term. We have a more grand vision for `anyref` being a first-class type in the language, but that's on a much longer horizon and this is currently thought to be the best we can do in terms of integration in the near future. The stack/heap JS tables are combined into one wasm table. The stack starts at the end of the table and grows down with a stack pointer (also injected). The heap starts at the end and grows up (state managed in linear memory). The anyref transformation here will hook up various intrinsics in wasm-bindgen to the runtime functionality if the anyref supoprt is enabled. The main tricky treatment here was applied to closures, where we need JS to use a different function pointer than the one Rust gives it to use a JS function pointer empowered with anyref. This works by switching up a bit how descriptors work, embedding the shims to call inside descriptors rather than communicated at runtime. This means that we're accessing constant values in the generated JS and we can just update the constant value accessed.
2018-10-18 08:43:36 -07:00
}
// Management of `anyref` is always thread local since an `anyref` value can't
// cross threads in wasm. Indices as a result are always thread-local.
std::thread_local!(pub static HEAP_SLAB: Cell<Slab> = Cell::new(Slab::new()));
Add experimental support for the `anyref` type This commit adds experimental support to `wasm-bindgen` to emit and leverage the `anyref` native wasm type. This native type is still in a proposal status (the reference-types proposal). The intention of `anyref` is to be able to directly hold JS values in wasm and pass the to imported functions, namely to empower eventual host bindings (now renamed WebIDL bindings) integration where we can skip JS shims altogether for many imports. This commit doesn't actually affect wasm-bindgen's behavior at all as-is, but rather this support requires an opt-in env var to be configured. Once the support is stable in browsers it's intended that this will add a CLI switch for turning on this support, eventually defaulting it to `true` in the far future. The basic strategy here is to take the `stack` and `slab` globals in the generated JS glue and move them into wasm using a table. This new table in wasm is managed at the fringes via injected shims. At `wasm-bindgen`-time the CLI will rewrite exports and imports with shims that actually use `anyref` if needed, performing loads/stores inside the wasm module instead of externally in the wasm module. This should provide a boost over what we have today, but it's not a fantastic strategy long term. We have a more grand vision for `anyref` being a first-class type in the language, but that's on a much longer horizon and this is currently thought to be the best we can do in terms of integration in the near future. The stack/heap JS tables are combined into one wasm table. The stack starts at the end of the table and grows down with a stack pointer (also injected). The heap starts at the end and grows up (state managed in linear memory). The anyref transformation here will hook up various intrinsics in wasm-bindgen to the runtime functionality if the anyref supoprt is enabled. The main tricky treatment here was applied to closures, where we need JS to use a different function pointer than the one Rust gives it to use a JS function pointer empowered with anyref. This works by switching up a bit how descriptors work, embedding the shims to call inside descriptors rather than communicated at runtime. This means that we're accessing constant values in the generated JS and we can just update the constant value accessed.
2018-10-18 08:43:36 -07:00
#[no_mangle]
2019-05-09 12:59:17 -07:00
pub extern "C" fn __wbindgen_anyref_table_alloc() -> usize {
HEAP_SLAB
2019-05-09 12:59:17 -07:00
.try_with(|slot| {
let mut slab = slot.replace(Slab::new());
let ret = slab.alloc();
slot.replace(slab);
ret
})
.unwrap_or_else(|_| internal_error("tls access failure"))
Add experimental support for the `anyref` type This commit adds experimental support to `wasm-bindgen` to emit and leverage the `anyref` native wasm type. This native type is still in a proposal status (the reference-types proposal). The intention of `anyref` is to be able to directly hold JS values in wasm and pass the to imported functions, namely to empower eventual host bindings (now renamed WebIDL bindings) integration where we can skip JS shims altogether for many imports. This commit doesn't actually affect wasm-bindgen's behavior at all as-is, but rather this support requires an opt-in env var to be configured. Once the support is stable in browsers it's intended that this will add a CLI switch for turning on this support, eventually defaulting it to `true` in the far future. The basic strategy here is to take the `stack` and `slab` globals in the generated JS glue and move them into wasm using a table. This new table in wasm is managed at the fringes via injected shims. At `wasm-bindgen`-time the CLI will rewrite exports and imports with shims that actually use `anyref` if needed, performing loads/stores inside the wasm module instead of externally in the wasm module. This should provide a boost over what we have today, but it's not a fantastic strategy long term. We have a more grand vision for `anyref` being a first-class type in the language, but that's on a much longer horizon and this is currently thought to be the best we can do in terms of integration in the near future. The stack/heap JS tables are combined into one wasm table. The stack starts at the end of the table and grows down with a stack pointer (also injected). The heap starts at the end and grows up (state managed in linear memory). The anyref transformation here will hook up various intrinsics in wasm-bindgen to the runtime functionality if the anyref supoprt is enabled. The main tricky treatment here was applied to closures, where we need JS to use a different function pointer than the one Rust gives it to use a JS function pointer empowered with anyref. This works by switching up a bit how descriptors work, embedding the shims to call inside descriptors rather than communicated at runtime. This means that we're accessing constant values in the generated JS and we can just update the constant value accessed.
2018-10-18 08:43:36 -07:00
}
#[no_mangle]
2019-05-09 12:59:17 -07:00
pub extern "C" fn __wbindgen_anyref_table_dealloc(idx: usize) {
Add experimental support for the `anyref` type This commit adds experimental support to `wasm-bindgen` to emit and leverage the `anyref` native wasm type. This native type is still in a proposal status (the reference-types proposal). The intention of `anyref` is to be able to directly hold JS values in wasm and pass the to imported functions, namely to empower eventual host bindings (now renamed WebIDL bindings) integration where we can skip JS shims altogether for many imports. This commit doesn't actually affect wasm-bindgen's behavior at all as-is, but rather this support requires an opt-in env var to be configured. Once the support is stable in browsers it's intended that this will add a CLI switch for turning on this support, eventually defaulting it to `true` in the far future. The basic strategy here is to take the `stack` and `slab` globals in the generated JS glue and move them into wasm using a table. This new table in wasm is managed at the fringes via injected shims. At `wasm-bindgen`-time the CLI will rewrite exports and imports with shims that actually use `anyref` if needed, performing loads/stores inside the wasm module instead of externally in the wasm module. This should provide a boost over what we have today, but it's not a fantastic strategy long term. We have a more grand vision for `anyref` being a first-class type in the language, but that's on a much longer horizon and this is currently thought to be the best we can do in terms of integration in the near future. The stack/heap JS tables are combined into one wasm table. The stack starts at the end of the table and grows down with a stack pointer (also injected). The heap starts at the end and grows up (state managed in linear memory). The anyref transformation here will hook up various intrinsics in wasm-bindgen to the runtime functionality if the anyref supoprt is enabled. The main tricky treatment here was applied to closures, where we need JS to use a different function pointer than the one Rust gives it to use a JS function pointer empowered with anyref. This works by switching up a bit how descriptors work, embedding the shims to call inside descriptors rather than communicated at runtime. This means that we're accessing constant values in the generated JS and we can just update the constant value accessed.
2018-10-18 08:43:36 -07:00
if idx < super::JSIDX_RESERVED as usize {
2019-05-09 12:59:17 -07:00
return;
Add experimental support for the `anyref` type This commit adds experimental support to `wasm-bindgen` to emit and leverage the `anyref` native wasm type. This native type is still in a proposal status (the reference-types proposal). The intention of `anyref` is to be able to directly hold JS values in wasm and pass the to imported functions, namely to empower eventual host bindings (now renamed WebIDL bindings) integration where we can skip JS shims altogether for many imports. This commit doesn't actually affect wasm-bindgen's behavior at all as-is, but rather this support requires an opt-in env var to be configured. Once the support is stable in browsers it's intended that this will add a CLI switch for turning on this support, eventually defaulting it to `true` in the far future. The basic strategy here is to take the `stack` and `slab` globals in the generated JS glue and move them into wasm using a table. This new table in wasm is managed at the fringes via injected shims. At `wasm-bindgen`-time the CLI will rewrite exports and imports with shims that actually use `anyref` if needed, performing loads/stores inside the wasm module instead of externally in the wasm module. This should provide a boost over what we have today, but it's not a fantastic strategy long term. We have a more grand vision for `anyref` being a first-class type in the language, but that's on a much longer horizon and this is currently thought to be the best we can do in terms of integration in the near future. The stack/heap JS tables are combined into one wasm table. The stack starts at the end of the table and grows down with a stack pointer (also injected). The heap starts at the end and grows up (state managed in linear memory). The anyref transformation here will hook up various intrinsics in wasm-bindgen to the runtime functionality if the anyref supoprt is enabled. The main tricky treatment here was applied to closures, where we need JS to use a different function pointer than the one Rust gives it to use a JS function pointer empowered with anyref. This works by switching up a bit how descriptors work, embedding the shims to call inside descriptors rather than communicated at runtime. This means that we're accessing constant values in the generated JS and we can just update the constant value accessed.
2018-10-18 08:43:36 -07:00
}
// clear this value from the table so while the table slot is un-allocated
// we don't keep around a strong reference to a potentially large object
unsafe {
__wbindgen_anyref_table_set_null(idx);
}
HEAP_SLAB
2019-05-09 12:59:17 -07:00
.try_with(|slot| {
let mut slab = slot.replace(Slab::new());
slab.dealloc(idx);
slot.replace(slab);
})
.unwrap_or_else(|_| internal_error("tls access failure"))
Add experimental support for the `anyref` type This commit adds experimental support to `wasm-bindgen` to emit and leverage the `anyref` native wasm type. This native type is still in a proposal status (the reference-types proposal). The intention of `anyref` is to be able to directly hold JS values in wasm and pass the to imported functions, namely to empower eventual host bindings (now renamed WebIDL bindings) integration where we can skip JS shims altogether for many imports. This commit doesn't actually affect wasm-bindgen's behavior at all as-is, but rather this support requires an opt-in env var to be configured. Once the support is stable in browsers it's intended that this will add a CLI switch for turning on this support, eventually defaulting it to `true` in the far future. The basic strategy here is to take the `stack` and `slab` globals in the generated JS glue and move them into wasm using a table. This new table in wasm is managed at the fringes via injected shims. At `wasm-bindgen`-time the CLI will rewrite exports and imports with shims that actually use `anyref` if needed, performing loads/stores inside the wasm module instead of externally in the wasm module. This should provide a boost over what we have today, but it's not a fantastic strategy long term. We have a more grand vision for `anyref` being a first-class type in the language, but that's on a much longer horizon and this is currently thought to be the best we can do in terms of integration in the near future. The stack/heap JS tables are combined into one wasm table. The stack starts at the end of the table and grows down with a stack pointer (also injected). The heap starts at the end and grows up (state managed in linear memory). The anyref transformation here will hook up various intrinsics in wasm-bindgen to the runtime functionality if the anyref supoprt is enabled. The main tricky treatment here was applied to closures, where we need JS to use a different function pointer than the one Rust gives it to use a JS function pointer empowered with anyref. This works by switching up a bit how descriptors work, embedding the shims to call inside descriptors rather than communicated at runtime. This means that we're accessing constant values in the generated JS and we can just update the constant value accessed.
2018-10-18 08:43:36 -07:00
}
#[no_mangle]
2019-05-09 12:59:17 -07:00
pub unsafe extern "C" fn __wbindgen_drop_anyref_slice(ptr: *mut JsValue, len: usize) {
Add experimental support for the `anyref` type This commit adds experimental support to `wasm-bindgen` to emit and leverage the `anyref` native wasm type. This native type is still in a proposal status (the reference-types proposal). The intention of `anyref` is to be able to directly hold JS values in wasm and pass the to imported functions, namely to empower eventual host bindings (now renamed WebIDL bindings) integration where we can skip JS shims altogether for many imports. This commit doesn't actually affect wasm-bindgen's behavior at all as-is, but rather this support requires an opt-in env var to be configured. Once the support is stable in browsers it's intended that this will add a CLI switch for turning on this support, eventually defaulting it to `true` in the far future. The basic strategy here is to take the `stack` and `slab` globals in the generated JS glue and move them into wasm using a table. This new table in wasm is managed at the fringes via injected shims. At `wasm-bindgen`-time the CLI will rewrite exports and imports with shims that actually use `anyref` if needed, performing loads/stores inside the wasm module instead of externally in the wasm module. This should provide a boost over what we have today, but it's not a fantastic strategy long term. We have a more grand vision for `anyref` being a first-class type in the language, but that's on a much longer horizon and this is currently thought to be the best we can do in terms of integration in the near future. The stack/heap JS tables are combined into one wasm table. The stack starts at the end of the table and grows down with a stack pointer (also injected). The heap starts at the end and grows up (state managed in linear memory). The anyref transformation here will hook up various intrinsics in wasm-bindgen to the runtime functionality if the anyref supoprt is enabled. The main tricky treatment here was applied to closures, where we need JS to use a different function pointer than the one Rust gives it to use a JS function pointer empowered with anyref. This works by switching up a bit how descriptors work, embedding the shims to call inside descriptors rather than communicated at runtime. This means that we're accessing constant values in the generated JS and we can just update the constant value accessed.
2018-10-18 08:43:36 -07:00
for slot in slice::from_raw_parts_mut(ptr, len) {
__wbindgen_anyref_table_dealloc(slot.idx as usize);
Add experimental support for the `anyref` type This commit adds experimental support to `wasm-bindgen` to emit and leverage the `anyref` native wasm type. This native type is still in a proposal status (the reference-types proposal). The intention of `anyref` is to be able to directly hold JS values in wasm and pass the to imported functions, namely to empower eventual host bindings (now renamed WebIDL bindings) integration where we can skip JS shims altogether for many imports. This commit doesn't actually affect wasm-bindgen's behavior at all as-is, but rather this support requires an opt-in env var to be configured. Once the support is stable in browsers it's intended that this will add a CLI switch for turning on this support, eventually defaulting it to `true` in the far future. The basic strategy here is to take the `stack` and `slab` globals in the generated JS glue and move them into wasm using a table. This new table in wasm is managed at the fringes via injected shims. At `wasm-bindgen`-time the CLI will rewrite exports and imports with shims that actually use `anyref` if needed, performing loads/stores inside the wasm module instead of externally in the wasm module. This should provide a boost over what we have today, but it's not a fantastic strategy long term. We have a more grand vision for `anyref` being a first-class type in the language, but that's on a much longer horizon and this is currently thought to be the best we can do in terms of integration in the near future. The stack/heap JS tables are combined into one wasm table. The stack starts at the end of the table and grows down with a stack pointer (also injected). The heap starts at the end and grows up (state managed in linear memory). The anyref transformation here will hook up various intrinsics in wasm-bindgen to the runtime functionality if the anyref supoprt is enabled. The main tricky treatment here was applied to closures, where we need JS to use a different function pointer than the one Rust gives it to use a JS function pointer empowered with anyref. This works by switching up a bit how descriptors work, embedding the shims to call inside descriptors rather than communicated at runtime. This means that we're accessing constant values in the generated JS and we can just update the constant value accessed.
2018-10-18 08:43:36 -07:00
}
}
// Implementation of `__wbindgen_anyref_heap_live_count` for when we are using
// `anyref` instead of the JS `heap`.
#[no_mangle]
pub unsafe extern "C" fn __wbindgen_anyref_heap_live_count_impl() -> u32 {
HEAP_SLAB
.try_with(|slot| {
let slab = slot.replace(Slab::new());
let count = slab.live_count();
slot.replace(slab);
count
})
.unwrap_or_else(|_| internal_error("tls access failure"))
}
Add experimental support for the `anyref` type This commit adds experimental support to `wasm-bindgen` to emit and leverage the `anyref` native wasm type. This native type is still in a proposal status (the reference-types proposal). The intention of `anyref` is to be able to directly hold JS values in wasm and pass the to imported functions, namely to empower eventual host bindings (now renamed WebIDL bindings) integration where we can skip JS shims altogether for many imports. This commit doesn't actually affect wasm-bindgen's behavior at all as-is, but rather this support requires an opt-in env var to be configured. Once the support is stable in browsers it's intended that this will add a CLI switch for turning on this support, eventually defaulting it to `true` in the far future. The basic strategy here is to take the `stack` and `slab` globals in the generated JS glue and move them into wasm using a table. This new table in wasm is managed at the fringes via injected shims. At `wasm-bindgen`-time the CLI will rewrite exports and imports with shims that actually use `anyref` if needed, performing loads/stores inside the wasm module instead of externally in the wasm module. This should provide a boost over what we have today, but it's not a fantastic strategy long term. We have a more grand vision for `anyref` being a first-class type in the language, but that's on a much longer horizon and this is currently thought to be the best we can do in terms of integration in the near future. The stack/heap JS tables are combined into one wasm table. The stack starts at the end of the table and grows down with a stack pointer (also injected). The heap starts at the end and grows up (state managed in linear memory). The anyref transformation here will hook up various intrinsics in wasm-bindgen to the runtime functionality if the anyref supoprt is enabled. The main tricky treatment here was applied to closures, where we need JS to use a different function pointer than the one Rust gives it to use a JS function pointer empowered with anyref. This works by switching up a bit how descriptors work, embedding the shims to call inside descriptors rather than communicated at runtime. This means that we're accessing constant values in the generated JS and we can just update the constant value accessed.
2018-10-18 08:43:36 -07:00
// see comment in module above this in `link_mem_intrinsics`
#[inline(never)]
2019-05-09 12:59:17 -07:00
pub fn link_intrinsics() {}