generate callers based on layout traversal

This commit is contained in:
Folkert 2023-02-20 22:05:35 +01:00
parent 8a443ba8bb
commit 68524ef07e
No known key found for this signature in database
GPG key ID: 1F17F6FFD112B97C
4 changed files with 562 additions and 44 deletions

View file

@ -188,20 +188,28 @@ where
let func_name = FuncName(&bytes);
if let HostExposedLayouts::HostExposed { aliases, .. } = &proc.host_exposed_layouts {
for (_, (symbol, top_level, layout)) in aliases {
match layout {
for (_, hels) in aliases {
match hels.raw_function_layout {
RawFunctionLayout::Function(_, _, _) => {
let it = top_level.arguments.iter().copied();
let bytes =
func_name_bytes_help(*symbol, it, Niche::NONE, top_level.result);
let it = hels.proc_layout.arguments.iter().copied();
let bytes = func_name_bytes_help(
hels.symbol,
it,
Niche::NONE,
hels.proc_layout.result,
);
host_exposed_functions.push((bytes, top_level.arguments));
host_exposed_functions.push((bytes, hels.proc_layout.arguments));
}
RawFunctionLayout::ZeroArgumentThunk(_) => {
let bytes =
func_name_bytes_help(*symbol, [], Niche::NONE, top_level.result);
let bytes = func_name_bytes_help(
hels.symbol,
[],
Niche::NONE,
hels.proc_layout.result,
);
host_exposed_functions.push((bytes, top_level.arguments));
host_exposed_functions.push((bytes, hels.proc_layout.arguments));
}
}
}

View file

@ -39,8 +39,8 @@ use roc_debug_flags::dbg_do;
use roc_debug_flags::ROC_PRINT_LLVM_FN_VERIFICATION;
use roc_module::symbol::{Interns, ModuleId, Symbol};
use roc_mono::ir::{
BranchInfo, CallType, CrashTag, EntryPoint, JoinPointId, ListLiteralElement, ModifyRc,
OptLevel, ProcLayout, SingleEntryPoint,
BranchInfo, CallType, CrashTag, EntryPoint, HostExposedLambdaSet, JoinPointId,
ListLiteralElement, ModifyRc, OptLevel, ProcLayout, SingleEntryPoint,
};
use roc_mono::layout::{
Builtin, InLayout, LambdaName, LambdaSet, Layout, LayoutIds, LayoutInterner, Niche,
@ -3977,7 +3977,8 @@ fn expose_function_to_host_help_c_abi_v2<'a, 'ctx, 'env>(
// In C, this is modelled as a function returning void
(
&params[..],
&param_types[..param_types.len().saturating_sub(1)],
// &param_types[..param_types.len().saturating_sub(1)],
&param_types[..],
)
}
_ => (&params[..], &param_types[..]),
@ -4938,26 +4939,24 @@ fn expose_alias_to_host<'a, 'ctx, 'env>(
mod_solutions: &'a ModSolutions,
proc_name: LambdaName,
alias_symbol: Symbol,
exposed_function_symbol: Symbol,
top_level: ProcLayout<'a>,
layout: RawFunctionLayout<'a>,
hels: &HostExposedLambdaSet<'a>,
) {
let ident_string = proc_name.name().as_str(&env.interns);
let fn_name: String = format!("{}_1", ident_string);
match layout {
match hels.raw_function_layout {
RawFunctionLayout::Function(arguments, closure, result) => {
// define closure size and return value size, e.g.
//
// * roc__mainForHost_1_Update_size() -> i64
// * roc__mainForHost_1_Update_result_size() -> i64
let it = top_level.arguments.iter().copied();
let it = hels.proc_layout.arguments.iter().copied();
let bytes = roc_alias_analysis::func_name_bytes_help(
exposed_function_symbol,
hels.symbol,
it,
Niche::NONE,
top_level.result,
hels.proc_layout.result,
);
let func_name = FuncName(&bytes);
let func_solutions = mod_solutions.func_solutions(func_name).unwrap();
@ -4973,17 +4972,17 @@ fn expose_alias_to_host<'a, 'ctx, 'env>(
function_value_by_func_spec(
env,
*func_spec,
exposed_function_symbol,
top_level.arguments,
hels.symbol,
hels.proc_layout.arguments,
Niche::NONE,
top_level.result,
hels.proc_layout.result,
)
}
None => {
// morphic did not generate a specialization for this function,
// therefore it must actually be unused.
// An example is our closure callers
panic!("morphic did not specialize {:?}", exposed_function_symbol);
panic!("morphic did not specialize {:?}", hels.symbol);
}
};
@ -5237,16 +5236,14 @@ pub fn build_proc<'a, 'ctx, 'env>(
/* no host, or exposing types is not supported */
}
Binary | BinaryDev => {
for (alias_name, (generated_function, top_level, layout)) in aliases.iter() {
for (alias_name, hels) in aliases.iter() {
expose_alias_to_host(
env,
layout_interner,
mod_solutions,
proc.name,
*alias_name,
*generated_function,
*top_level,
*layout,
hels,
)
}
}

View file

@ -34,7 +34,7 @@ use roc_std::RocDec;
use roc_target::TargetInfo;
use roc_types::subs::{
instantiate_rigids, storage_copy_var_to, Content, ExhaustiveMark, FlatType, RedundantMark,
StorageSubs, Subs, Variable, VariableSubsSlice,
StorageSubs, Subs, Variable, VariableSubsSlice,
};
use std::collections::HashMap;
use ven_pretty::{BoxAllocator, DocAllocator, DocBuilder};
@ -305,12 +305,21 @@ pub struct Proc<'a> {
pub host_exposed_layouts: HostExposedLayouts<'a>,
}
#[derive(Clone, Debug, PartialEq, Eq)]
pub struct HostExposedLambdaSet<'a> {
pub id: LambdaSetId,
/// Symbol of the exposed function
pub symbol: Symbol,
pub proc_layout: ProcLayout<'a>,
pub raw_function_layout: RawFunctionLayout<'a>,
}
#[derive(Clone, Debug, PartialEq, Eq)]
pub enum HostExposedLayouts<'a> {
NotHostExposed,
HostExposed {
rigids: BumpMap<Lowercase, InLayout<'a>>,
aliases: BumpMap<Symbol, (Symbol, ProcLayout<'a>, RawFunctionLayout<'a>)>,
aliases: BumpMap<Symbol, HostExposedLambdaSet<'a>>,
},
}
@ -3068,25 +3077,126 @@ fn specialize_external_help<'a>(
);
match specialization_result {
Ok((proc, layout)) => {
Ok((mut proc, layout)) => {
let top_level = ProcLayout::from_raw_named(env.arena, name, layout);
if procs.is_module_thunk(name.name()) {
debug_assert!(top_level.arguments.is_empty());
}
// layouts that are (transitively) used in the type of `mainForHost`.
let host_exposed_layouts = top_level
.arguments
.iter()
.copied()
.chain([top_level.result]);
if !host_exposed_aliases.is_empty() {
// layouts that are (transitively) used in the type of `mainForHost`.
let mut host_exposed_layouts: Vec<_> = top_level
.arguments
.iter()
.copied()
.chain([top_level.result])
.collect_in(env.arena);
// TODO: In the future, we will generate glue procs here
for in_layout in host_exposed_layouts {
let layout = layout_cache.interner.get(in_layout);
// it is very likely we see the same types across functions, or in multiple arguments
host_exposed_layouts.sort();
host_exposed_layouts.dedup();
let _ = layout;
// TODO: In the future, we will generate glue procs here
for in_layout in host_exposed_layouts {
let layout = layout_cache.interner.get(in_layout);
let all_glue_procs = generate_glue_procs(
env.home,
env.ident_ids,
env.arena,
&mut layout_cache.interner,
env.arena.alloc(layout),
);
// for now, getters are not processed here
let GlueProcs {
getters: _,
extern_names,
} = all_glue_procs;
let mut aliases = BumpMap::default();
for (id, mut raw_function_layout) in extern_names {
let symbol = env.unique_symbol();
let lambda_name = LambdaName::no_niche(symbol);
if false {
raw_function_layout = match raw_function_layout {
RawFunctionLayout::Function(a, mut lambda_set, _) => {
lambda_set.ret = in_layout;
RawFunctionLayout::Function(a, lambda_set, in_layout)
}
RawFunctionLayout::ZeroArgumentThunk(x) => {
RawFunctionLayout::ZeroArgumentThunk(x)
}
};
}
let (key, (_name, top_level, proc)) = generate_host_exposed_function(
env,
procs,
layout_cache,
lambda_name,
raw_function_layout,
);
procs
.specialized
.insert_specialized(symbol, top_level, proc);
let hels = HostExposedLambdaSet {
id,
symbol,
proc_layout: top_level,
raw_function_layout,
};
aliases.insert(key, hels);
}
// pre-glue: generate named callers for as-exposed aliases
for (alias_name, variable) in host_exposed_aliases {
let raw_function_layout = layout_cache
.raw_from_var(env.arena, *variable, env.subs)
.unwrap();
let symbol = env.unique_symbol();
let lambda_name = LambdaName::no_niche(symbol);
let (_lambda_name, (proc_name, proc_layout, proc)) =
generate_host_exposed_function(
env,
procs,
layout_cache,
lambda_name,
raw_function_layout,
);
procs
.specialized
.insert_specialized(proc_name, proc_layout, proc);
let hels = HostExposedLambdaSet {
id: LambdaSetId::default(),
symbol: proc_name,
proc_layout,
raw_function_layout,
};
aliases.insert(*alias_name, hels);
}
match &mut proc.host_exposed_layouts {
HostExposedLayouts::HostExposed { aliases: old, .. } => old.extend(aliases),
hep @ HostExposedLayouts::NotHostExposed => {
*hep = HostExposedLayouts::HostExposed {
aliases,
rigids: Default::default(),
};
}
}
}
}
procs
@ -3221,7 +3331,7 @@ fn generate_host_exposed_function<'a>(
layout: RawFunctionLayout<'a>,
) -> (Symbol, (Symbol, ProcLayout<'a>, Proc<'a>)) {
let symbol = lambda_name.name();
let function_name = env.unique_symbol();
let function_name = symbol;
match layout {
RawFunctionLayout::Function(_, lambda_set, _) => {
@ -3379,7 +3489,7 @@ fn specialize_proc_help<'a>(
let body_var = partial_proc.body_var;
// determine the layout of aliases/rigids exposed to the host
let host_exposed_layouts = if host_exposed_variables.is_empty() {
let host_exposed_layouts = if true || host_exposed_variables.is_empty() {
HostExposedLayouts::NotHostExposed
} else {
let mut aliases = BumpMap::new_in(env.arena);
@ -3400,12 +3510,20 @@ fn specialize_proc_help<'a>(
procs
.specialized
.insert_specialized(proc_name, proc_layout, proc);
aliases.insert(*alias_name, (proc_name, proc_layout, raw_function_layout));
let hels = HostExposedLambdaSet {
id: LambdaSetId::default(),
symbol: proc_name,
proc_layout,
raw_function_layout,
};
aliases.insert(*alias_name, hels);
}
HostExposedLayouts::HostExposed {
rigids: BumpMap::new_in(env.arena),
aliases,
aliases: Default::default(),
}
};
@ -11129,3 +11247,344 @@ where
remainder: env.arena.alloc(switch),
}
}
pub struct GlueLayouts<'a> {
pub getters: std::vec::Vec<(Symbol, ProcLayout<'a>)>,
}
type GlueProcId = u16;
pub struct GlueProc<'a> {
pub name: Symbol,
pub proc_layout: ProcLayout<'a>,
pub proc: Proc<'a>,
}
pub struct GlueProcs<'a> {
pub getters: Vec<'a, (Layout<'a>, Vec<'a, GlueProc<'a>>)>,
pub extern_names: Vec<'a, (LambdaSetId, RawFunctionLayout<'a>)>,
}
#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, Default)]
pub struct LambdaSetId(pub u32);
impl LambdaSetId {
#[must_use]
pub fn next(self) -> Self {
debug_assert!(self.0 < u32::MAX);
Self(self.0 + 1)
}
pub const fn invalid() -> Self {
Self(u32::MAX)
}
}
pub fn generate_glue_procs<'a, 'i, I>(
home: ModuleId,
ident_ids: &mut IdentIds,
arena: &'a Bump,
layout_interner: &'i mut I,
layout: &'a Layout<'a>,
) -> GlueProcs<'a>
where
I: LayoutInterner<'a>,
{
let mut answer = GlueProcs {
getters: Vec::new_in(arena),
extern_names: Vec::new_in(arena),
};
let mut lambda_set_id = LambdaSetId(0);
let mut stack: Vec<'a, Layout<'a>> = Vec::from_iter_in([*layout], arena);
let mut next_unique_id = 0;
macro_rules! handle_struct_field_layouts {
($field_layouts: expr) => {{
if $field_layouts.iter().any(|l| {
layout_interner
.get(*l)
.has_varying_stack_size(layout_interner, arena)
}) {
let procs = generate_glue_procs_for_struct_fields(
layout_interner,
home,
&mut next_unique_id,
ident_ids,
arena,
layout,
$field_layouts,
);
answer.getters.push((*layout, procs));
}
for in_layout in $field_layouts.iter().rev() {
stack.push(layout_interner.get(*in_layout));
}
}};
}
macro_rules! handle_tag_field_layouts {
($tag_id:expr, $union_layout:expr, $field_layouts: expr) => {{
if $field_layouts.iter().any(|l| {
layout_interner
.get(*l)
.has_varying_stack_size(layout_interner, arena)
}) {
let procs = generate_glue_procs_for_tag_fields(
layout_interner,
home,
&mut next_unique_id,
ident_ids,
arena,
$tag_id,
layout,
$union_layout,
$field_layouts,
);
answer.getters.push((*layout, procs));
}
for in_layout in $field_layouts.iter().rev() {
stack.push(layout_interner.get(*in_layout));
}
}};
}
while let Some(layout) = stack.pop() {
match layout {
Layout::Builtin(builtin) => match builtin {
Builtin::Int(_)
| Builtin::Float(_)
| Builtin::Bool
| Builtin::Decimal
| Builtin::Str => { /* do nothing */ }
Builtin::List(element) => stack.push(layout_interner.get(element)),
},
Layout::Struct { field_layouts, .. } => {
handle_struct_field_layouts!(field_layouts);
}
Layout::Boxed(boxed) => {
stack.push(layout_interner.get(boxed));
}
Layout::Union(union_layout) => match union_layout {
UnionLayout::NonRecursive(tags) => {
for in_layout in tags.iter().flat_map(|e| e.iter()) {
stack.push(layout_interner.get(*in_layout));
}
}
UnionLayout::Recursive(tags) => {
for in_layout in tags.iter().flat_map(|e| e.iter()) {
stack.push(layout_interner.get(*in_layout));
}
}
UnionLayout::NonNullableUnwrapped(field_layouts) => {
handle_tag_field_layouts!(0, union_layout, field_layouts);
}
UnionLayout::NullableWrapped {
other_tags,
nullable_id,
} => {
let tag_ids =
(0..nullable_id).chain(nullable_id + 1..other_tags.len() as u16 + 1);
for (i, field_layouts) in tag_ids.zip(other_tags) {
handle_tag_field_layouts!(i, union_layout, *field_layouts);
}
}
UnionLayout::NullableUnwrapped { other_fields, .. } => {
for in_layout in other_fields.iter().rev() {
stack.push(layout_interner.get(*in_layout));
}
}
},
Layout::LambdaSet(lambda_set) => {
let raw_function_layout =
RawFunctionLayout::Function(*lambda_set.args, lambda_set, lambda_set.ret);
let key = (lambda_set_id, raw_function_layout);
answer.extern_names.push(key);
// this id is used, increment for the next one
lambda_set_id = lambda_set_id.next();
stack.push(layout_interner.get(lambda_set.runtime_representation()));
}
Layout::RecursivePointer(_) => {
/* do nothing, we've already generated for this type through the Union(_) */
}
}
}
answer
}
fn generate_glue_procs_for_struct_fields<'a, 'i, I>(
layout_interner: &'i mut I,
home: ModuleId,
next_unique_id: &mut GlueProcId,
ident_ids: &mut IdentIds,
arena: &'a Bump,
unboxed_struct_layout: &'a Layout<'a>,
field_layouts: &'a [InLayout<'a>],
) -> Vec<'a, GlueProc<'a>>
where
I: LayoutInterner<'a>,
{
let interned_unboxed_struct_layout = layout_interner.insert(*unboxed_struct_layout);
let boxed_struct_layout = Layout::Boxed(interned_unboxed_struct_layout);
let boxed_struct_layout = layout_interner.insert(boxed_struct_layout);
let mut answer = bumpalo::collections::Vec::with_capacity_in(field_layouts.len(), arena);
for (index, field) in field_layouts.iter().enumerate() {
let proc_layout = ProcLayout {
arguments: arena.alloc([boxed_struct_layout]),
result: *field,
niche: Niche::NONE,
};
let symbol = unique_glue_symbol(arena, next_unique_id, home, ident_ids);
let argument = Symbol::new(home, ident_ids.gen_unique());
let unboxed = Symbol::new(home, ident_ids.gen_unique());
let result = Symbol::new(home, ident_ids.gen_unique());
home.register_debug_idents(ident_ids);
let ret_stmt = arena.alloc(Stmt::Ret(result));
let field_get_expr = Expr::StructAtIndex {
index: index as u64,
field_layouts,
structure: unboxed,
};
let field_get_stmt = Stmt::Let(result, field_get_expr, *field, ret_stmt);
let unbox_expr = Expr::ExprUnbox { symbol: argument };
let unbox_stmt = Stmt::Let(
unboxed,
unbox_expr,
interned_unboxed_struct_layout,
arena.alloc(field_get_stmt),
);
let proc = Proc {
name: LambdaName::no_niche(symbol),
args: arena.alloc([(boxed_struct_layout, argument)]),
body: unbox_stmt,
closure_data_layout: None,
ret_layout: *field,
is_self_recursive: SelfRecursive::NotSelfRecursive,
must_own_arguments: false,
host_exposed_layouts: HostExposedLayouts::NotHostExposed,
};
answer.push(GlueProc {
name: symbol,
proc_layout,
proc,
});
}
answer
}
fn unique_glue_symbol(
arena: &Bump,
next_unique_id: &mut GlueProcId,
home: ModuleId,
ident_ids: &mut IdentIds,
) -> Symbol {
let unique_id = *next_unique_id;
*next_unique_id = unique_id + 1;
// Turn unique_id into a Symbol without doing a heap allocation.
use std::fmt::Write;
let mut string = bumpalo::collections::String::with_capacity_in(32, arena);
let _result = write!(
&mut string,
"roc__getter_{}_{}",
"", // then name of the platform `main.roc` is the empty string
unique_id
);
debug_assert_eq!(_result, Ok(())); // This should never fail, but doesn't hurt to debug-check!
let ident_id = ident_ids.get_or_insert(string.into_bump_str());
Symbol::new(home, ident_id)
}
#[allow(clippy::too_many_arguments)]
fn generate_glue_procs_for_tag_fields<'a, 'i, I>(
layout_interner: &'i mut I,
home: ModuleId,
next_unique_id: &mut GlueProcId,
ident_ids: &mut IdentIds,
arena: &'a Bump,
tag_id: TagIdIntType,
unboxed_struct_layout: &'a Layout<'a>,
union_layout: UnionLayout<'a>,
field_layouts: &'a [InLayout<'a>],
) -> Vec<'a, GlueProc<'a>>
where
I: LayoutInterner<'a>,
{
let interned = layout_interner.insert(*unboxed_struct_layout);
let boxed_struct_layout = Layout::Boxed(interned);
let boxed_struct_layout = layout_interner.insert(boxed_struct_layout);
let mut answer = bumpalo::collections::Vec::with_capacity_in(field_layouts.len(), arena);
for (index, field) in field_layouts.iter().enumerate() {
let proc_layout = ProcLayout {
arguments: arena.alloc([boxed_struct_layout]),
result: *field,
niche: Niche::NONE,
};
let symbol = unique_glue_symbol(arena, next_unique_id, home, ident_ids);
let argument = Symbol::new(home, ident_ids.gen_unique());
let unboxed = Symbol::new(home, ident_ids.gen_unique());
let result = Symbol::new(home, ident_ids.gen_unique());
home.register_debug_idents(ident_ids);
let ret_stmt = arena.alloc(Stmt::Ret(result));
let field_get_expr = Expr::UnionAtIndex {
structure: unboxed,
tag_id,
union_layout,
index: index as u64,
};
let field_get_stmt = Stmt::Let(result, field_get_expr, *field, ret_stmt);
let unbox_expr = Expr::ExprUnbox { symbol: argument };
let unbox_stmt = Stmt::Let(unboxed, unbox_expr, interned, arena.alloc(field_get_stmt));
let proc = Proc {
name: LambdaName::no_niche(symbol),
args: arena.alloc([(boxed_struct_layout, argument)]),
body: unbox_stmt,
closure_data_layout: None,
ret_layout: *field,
is_self_recursive: SelfRecursive::NotSelfRecursive,
must_own_arguments: false,
host_exposed_layouts: HostExposedLayouts::NotHostExposed,
};
answer.push(GlueProc {
name: symbol,
proc_layout,
proc,
});
}
answer
}

View file

@ -2717,6 +2717,60 @@ impl<'a> Layout<'a> {
}
}
pub fn has_varying_stack_size<I>(self, interner: &I, arena: &bumpalo::Bump) -> bool
where
I: LayoutInterner<'a>,
{
let mut stack: Vec<Layout> = bumpalo::collections::Vec::new_in(arena);
stack.push(self);
while let Some(layout) = stack.pop() {
match layout {
Layout::Builtin(builtin) => match builtin {
Builtin::Int(_)
| Builtin::Float(_)
| Builtin::Bool
| Builtin::Decimal
| Builtin::Str
// If there's any layer of indirection (behind a pointer), then it doesn't vary!
| Builtin::List(_) => { /* do nothing */ }
},
// If there's any layer of indirection (behind a pointer), then it doesn't vary!
Layout::Struct { field_layouts, .. } => {
stack.extend(field_layouts.iter().map(|interned| interner.get(*interned)))
}
Layout::Union(tag_union) => match tag_union {
UnionLayout::NonRecursive(tags) | UnionLayout::Recursive(tags) => {
for tag in tags {
stack.extend(tag.iter().map(|interned| interner.get(*interned)));
}
}
UnionLayout::NonNullableUnwrapped(fields) => {
stack.extend(fields.iter().map(|interned| interner.get(*interned)));
}
UnionLayout::NullableWrapped { other_tags, .. } => {
for tag in other_tags {
stack.extend(tag.iter().map(|interned| interner.get(*interned)));
}
}
UnionLayout::NullableUnwrapped { other_fields, .. } => {
stack.extend(other_fields.iter().map(|interned| interner.get(*interned)));
}
},
Layout::LambdaSet(_) => return true,
Layout::Boxed(_) => {
// If there's any layer of indirection (behind a pointer), then it doesn't vary!
}
Layout::RecursivePointer(_) => {
/* do nothing, we've already generated for this type through the Union(_) */
}
}
}
false
}
/// Used to build a `Layout::Struct` where the field name order is irrelevant.
pub fn struct_no_name_order(field_layouts: &'a [InLayout]) -> Self {
if field_layouts.is_empty() {