Point symbolic witnesses into `.Self` written inside an impl decl at the
impl that is being declared. This is tricky because the impl does not
yet exist. So we use a new instruction `ImplSelfWitness` which _will_ be
replaced by the `ImplWitness` once it becomes available. The
`ImplSelfWitness` acts like a symbolic witness, except it does not
perform lookup, since we know which impl we will get a witness from.
This prevents us from finding other impls when performing lookups into
`.Self` in an impl decl, which produces incorrect/incoherent results.
Previously the last decl had to be the definition. Now we allow a
declaration after a definition, so that the user can write a match_first
block last, and put (re-)declarations of impls in it, after the
definitions have already been written elsewhere.
We track the location of the decl that was associated with a match_first
block so that we can correctly point to it in diagnostics when an impl
is written twice in match_first blocks. Since impls may not be
redeclared across an import boundary, we will never have a `SemIR::Impl`
with a match_first from a different file in a redeclaration, so we don't
need to import the location of a previous decl that was in a match_first
for diagnostics. As such we just store a LocId on the `SemIR::Impl`
struct.
When we check an `impl` decl, find the containing `match_first` block,
if any, and store a connection to it in the `SemIR::Impl` structure,
along with the impl's position in that `match_first` block so that we
can sort/prioritize the `SemIR::Impl`s later.
Also, update the `is_final` flag if the `match_first` block is modified
as `final`. But ensure we diagnose trying to put a `final impl`, or a
redeclaration of one, in a `match_first` block. Also diagnose if an impl
is attached to a `match_first` block more than once - either in two
different blocks or in the same block at different positions.
Putting the `match_first` connection on the `SemIR::Impl` structure
means we have to import it, so implement import and add a smoke test for
that, which ensures nothing explodes.
Drop the `scope_stack` entry for the `match_first` block, as it was not
needed. Once we started tracking the `match_first` size on the `Context`
class, it became more straightforward to just store the `match_first`
decl `InstId` in the same place. The `match_first` block is not supposed
to act like a different scope for the purpose of redecls anyhow, so it's
a bit simpler this way.
As part of this, move functions that seem reasonable to make out-of-line
to a separate `_impl.h` header file that is only included where the
explicit instantiation _definition_ is provided.
By using explicit instantiation we can make these templates behave more
like non-template classes in terms of supporting out-of-line definitions
that don't need to be compiled by every translation unit. The set of
eventual instantiations here is fundamentally known, and there tend to
be headers that define a canonical "leaf" type where it makes sense to
trigger the explicit instantiation.
Where we already had a `.cpp` file to put the explicit instantiation
definition, use it. But in some places we didn't have such a `.cpp` file
so this PR adds those.
This also requires that we have precise constraints on APIs that _can't_
be instantiated for specific argument types, as now we don't do this
lazily.
Combined, this appears to reduce the sum of object file sizes in the
`check` directory by almost 40% (122mb -> 74mb) in my measurement.
My actual goal was to improve compile times, but so far I don't have a
great methodology for measuring these... But the object file size
reduction seems to confirm this is a net win and likely represents a
non-trivial improvement in compile time.
Assisted-by: Antigravity with Gemini
Identifying a facet type takes both a self and facet type as a pair, and
then encode the self into the IdentifiedFacetType. This makes a
constraint that requires some _other_ type implements an interface
visible in the IdentifiedFacetType. And it will help to enable facet
types with `where T impls Z` for `T` that is not `.Self` in the future.
IdentifiedFacetTypes are now stored in a CanonicalValueStore instead of
a RelationalValueStore as they key is the combination of self and
(declared) facet type together now.
When the self-type is a facet value (has type FacetType) this is most
straightforward. But when it's a type we need to construct a FacetValue
to construct a specific for a require decl, to replace the generic
binding of the symbolic `Self`, which has type FacetType. To do so, we
make a FacetValue with an empty FacetType (equivalent to TypeType). This
prevents any looking for witnesses through the FacetType, which matches
what you can get from a type directly, requiring witnesses to come from
finding an `impl` decl.
Add additional InstNamer logic for such empty facet types so they print
as `<typename>.type.facet` if possible instead of as just `facet_value`.
The IdTag knows the type of the Id its tagging and the type of the Id
being used as the tag. This prevents mixing up tagged and untagged ids,
and avoids having to work with untyped integers.
Adds an Untagged marker struct that's used as the tag type in IdTag when
no tag is desired.
The complexity of ConstantIds and TypeIds became a bit visible: TypeIds
are concrete ConstantIds. And ConstantIds have two different tagging
schemes, one for concrete and one for symbolic ids. And ConstantIds are
actually re-cast InstIds with the same index. The LoweredTypeStore needs
to work with tagged TypeIds, but the tags actually come from an InstId
store in ConstantValueStore. Now this is expressed in the type system by
getting the tags for TypeIds from the ConstantValueStore.
ValueStores without an TagId type parameter are now visibly untagged.
IdTag is now only default constructible when it does not have a tag,
which means ValueStore is only default constructible when the TagId is
untagged. This forces tagged value stores to be constructed correctly
with a tag at compile time, and untagged ones to be constructed without.
FixedSizeValueStore has overloads for dealing with tagged and untagged
Ids, since it can't default-construct ValueStore for tagged ids, and no
longer requires passing in default-constructed tags when there is no tag
in the ids.
The impl's body block has to end before we make its ImplDecl
instruction, and the witness instructions come later, so they don't end
up in the body block. Currently they just end up in the enclosing (file,
typically, or class) scope block.
Add a new InstBlockId to Impl for holding witness instructions, and
explicitly insert them into that block. Then include those instructions
into the scope of the Impl for naming, and format them into the Impl
right after the body block.
This is based on #6484
When forming an IdentifiedFacetType, we collect interfaces named by
require decls in named constraints that the facet type refers to. These
interfaces come with a specific, but the require decl is inside an named
constraint which may be generic. So we need the specific being applied
to the containing named constraint to also be applied to the require
decl and its target interfaces.
This uncovered that the facet type in require decls was not being
imported correctly, as it was not being attached to the require decl's
generic. This is fixed by making import of RequireImplsDecl multiphase,
so that the decl instruction exists before we resolve the facet type
within it. And by pointing the generic importing machinery to the
RequireImplsDecl, and from there to the RequireImpls structure to get
the generic id.
Then `ImplStore::GetOrAddLookupBucket` can use an IdentifiedFacetType to
correctly get the interface being impl'd, both in the local and the
imported named constraint case. Which allows us to correctly diagnose
redeclarations in the impl file of an impl of an interface through a
named constraint. And to correctly _not_ diagnose them when the specific
in the generic named constraint differs from other decls.
This fixes various violations of C++'s One Definition Rule, where we
accidentally gave the same static data member multiple definitions in
different translation units. Clang happens to emit such definitions with
weak linkage, which allows us to get away with this without link errors,
but it's still formally incorrect.
Also switch keyword order around for a handful of instances of
`constexpr inline`, per agreement in open discussion.
This happens to reduce the size of a `-c dbg` toolchain binary by 7.2
MiB, presumably by making more of our symbols and especially debug info
discardable.
This is reducing ValueStore inference of types from `using`, and removes
`using ValueType = ...` from affected id types.
I'm adding a number of `using FooStore = ValueStore<FooId, Foo>` because
I think it's a little repetitive otherwise; often 4 cases where I'm
doing this: getter, const getter, member, and getter on `Context`. Note
we also have a number of `-> decltype(auto)` that were added I think
mainly to avoid repeating the type, but I'm not sure whether there'll be
agreement on replacing those and so am not changing them here.
I'm placing these aliases with the value type in general, because I
think it's probably easier to view that way. An alternative would be to
put all the types on `File`, but:
- That would be inconsistent with things like `InstStore`, which are
very `ValueStore`-adjacent and put with their value type.
- `File` would have a _lot_ of using's, and the accessors are already
noisy -- I think it would just make the file harder to skim.
Note this is the heart of what I'd brought up [on
Discord](https://discord.com/channels/655572317891461132/655578254970716160/1388199282250613019).
This PR still leaves CanonicalValueStore and BlockValueStore as things
to also add parameters to, but I thought it best to try breaking the set
of changes apart by type. Both of those rely on ValueStore, so
ValueStore needs to change first.
ValueStoreChunk and ValueStoreRange are implemented in a way that's
closely tied to ValueStore, and the separation makes for a lot of
additional template parameter passing, which seems easy to make mistakes
on. Combine types in order to make the close association more implicit.
I also considered passing `ValueT` everywhere, as an additional template
parameter. Note I believe the simplification is important. I'll
highlight four notes that I think favor this approach:
- `ValueStore`, with the chunk type in the same file, now has more of
the closely related implementation features in the same file. I think we
probably will want any chunking to continue to be done by `ValueStore`
itself, with related types using the implementation on `ValueStore` and
never creating their own.
- Making `ValueT` a template parameter on `ValueStore` -- my next step
-- will only change a couple lines of code on this type, instead of
sweeping changes. That should make it easier to be confident of the
correctness of those changes.
- Simpler to verify correctness. For example, `ValueStoreChunk` takes a
`ValueType` parameter that it doesn't forward; other functions assume
they can use `IdT::ValueType`. With the changes, this also no longer
benefits from separating out `IdHasValueType`, which was inconsistently
applied to related types (e.g., `ValueStoreRange` didn't use it).
- Template parameters often lead to `sizeof`, where we can't rely on
type checking to catch mistakes.
- The reduction of code is significant, with 8 `template<...>` removed
(including 1 forward declaration for `ValueStoreRange`), and also the
related `requires`. Correspondingly, places specifying template
parameters also decreased.
This avoids reallocating the backing buffer in ValueStore so that
references into the ValueStore are never invalidated when adding new
values. This works especially well since we never delete values from a
ValueStore.
The strategy used is to allocate chunks of a fixed size, and inserting
into each chunk until it is full before allocating the next. The
ValueStore starts with an initial allocated chunk in all cases, so that
there is only a single indirection for adding and accessing values from
this chunk. After it's full, additional chunks are allocated in a
vector, so two indirections are required to add or access values in
these chunks.
This obviates the need for
https://github.com/carbon-language/carbon-lang/pull/5529 as we no longer
need to worry about holding pointers into a ValueStore.
We introduce a Flatten operation for ranges. It flattens a "range over
ranges over Ts" down to a "range over Ts". This allows us to make an
range over the values in the ValueStore from a range over the chunks in
the ValueStore. See
https://doc.rust-lang.org/stable/std/iter/trait.Iterator.html#method.flatten
for inspiration for this name choice. Flatten is used in one other case
where we were writing two levels of for loops to do the same thing.
The `array_ref()` accessor is changed to `values()` and its now a range
(typed as a `ValueStoreRange`) over all values as references (like
ArrayRef was, but without random access).
As pointers to a ValueStore can no longer be invalidated, we remove the
ASAN poisoning feature and support from ValueStore.
This may cause a regression in our compile benchmark of up to 5%, though
that is close to or within the noise of the benchmark. We can look at
ways to optimize things further in the future. Perhaps by tuning the
chunk size further, or by making later chunks larger than earlier
chunks, or other strategies.
With this enabled, entities that live in value stores are poisoned
whenever any action is taken that might invalidate pointers and
references to those options -- in particular, adding another item to
that value store, or attempting to load any entity from an import IR.
Subsequent uses of those pointers or references then trigger an ASan
failure.
This detects latent bugs where the pointer or reference to the entity
would become stale if we got unlucky about when the value store
reallocates, even in cases where the reallocation didn't actually
happen.
This is not enabled by default: it finds a lot of latent bugs, so our
tests don't pass with this option. This PR also includes fixes for a few
of those bugs.
---------
Co-authored-by: Chandler Carruth <chandlerc@gmail.com>
Co-authored-by: Carbon Infra Bot <carbon-external-infra@google.com>
TypeInstId is an InstId whose constant value has a type of TypeType.
This includes:
- Type value instructions, the `ClassType` or `IntLiteralType`
instructions.
- Constraint value instructions, which are the `FacetType` and
`TypeType` instructions, each of which also have type TypeType.
TypeInstId encodes in the type system that it is safe to convert the
instruction's value to a TypeId, and CHECKs at construction that this
invariant is maintained.
---------
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
While facets may come with a rewrite for an associated constant, they
are symbolic. A final impl has the ability to provide a concrete value
instead, which allows generic code to use the concrete value in place of
the associated constant's (fully qualified) name.
For instance, instead of `I.Type`, the concrete type `()` can be used if
there is an `impl final [T:! type] T as I where .Type = ()` impl.
This does not yet cache the result of the lookups.
Depends on https://github.com/carbon-language/carbon-lang/pull/5255
---------
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
As of #5087, these terms are no longer synonyms. This change preserves
the original meaning.
---------
Co-authored-by: Josh L <josh11b@users.noreply.github.com>
* Add `RequireCompleteFacetType` and `ResolveFacetTypeImplWitness` to
`check::Context`. Goal was to move code from `impl.cpp` (mostly) without
functional changes.
* Complete type information is cached with the facet type, and is stored
in a `complete_facet_types()` table.
* Main functional change is to diagnose attempts to use a rewrite
constraint on an associated function. Some existing diagnostics have
been updated.
* Remove `check::Context::RequireDefinedType`:
* For class types, use `RequireCompleteType`
* For facet types, use `RequireCompleteFacetType`
* Introduce a `SemIR::SpecificInterface` to hold an interface and
specific id pair.
* Keep the specific interface ids in the impl object.
* Avoid some extra copies in `Dump` functions.
* Future work missing from this PR:
* Resolving for member access or actions that require impl lookup.
* Resolving rewrites constraints that refer to non-concrete values.
* Any support for adding implied constraints that result from a `where`
clause (though TODOs have been added).
---------
Co-authored-by: Josh L <josh11b@users.noreply.github.com>
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
Co-authored-by: Dana Jansens <danakj@orodu.net>
This allows iterating on all values in a store along with the Id for
each value, instead of `llvm::enumerate(store.array_ref())` which would
give you the indices.
While the indices are really the same as the Ids, this provides a
typesafe way to enumerate() over a store.
There's no use for this right now, but I thought I needed this, and it
helped me debug, and it was a pain to write correctly without dangling
references.
---------
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
This is a followup from #4834, I searched for "invalid" uses in our
codebase. This is mostly changing comments, and a couple debug
functions, but shouldn't affect testable behavior.
Note a couple things I'll highlight as not changing (but could) are:
- `ReturnTypeInfo::is_valid`
- `"invalid"` uses in the formatter
- `AddInvalid` for `!has_value` in `inst_fingerprinter` (because the
cases it's called sound invalid-ish)
High level, replacing `Id::Invalid` with `Id::None` and `Id::is_valid`
with `Id::has_value` for clarity, as discussed
[here](https://discord.com/channels/655572317891461132/655578254970716160/1331664574545395794).
The `IntId` refactoring is needed together with `AnyIdBase` because it's
also used with `ValueStore`.
Note, trying to be careful not to rewrite `EnumBase::InvalidIndex`, or
`is_valid` in general (e.g., `IdKind::is_valid`).
I've tried to sequence commits here:
1. Automatic replacements:
- `((?:Id|Index)(?: |::|\(|Base(?:\(|::)))Invalid((?:Index)?\W)` ->
`$1None$2`
- `<invalid>` -> `<none>`
- `InvalidNodeId` -> `NoneNodeId`
- `/\*invalid\*/` -> `/*none*/`
- `id((?:_|\(\))(?:\.|->))is_valid` -> `id$1has_value`
2. Manual edits:
- In `int.h` and `int_test.cpp`
- `IntT` has `is_value`, which I'm renaming to `is_embedded_value`.
- Manual edits to comments in this file.
- `AnyIdBase` and `IdBase`
- Declaration of `is_valid` -> `has_value`, `InvalidIndex` ->
`NoneIndex`.
- In `ids.h` and `ids.cpp`
- `is_valid` -> `has_value`
- `// An explicitly invalid ID.` -> `// An ID with no value.`; similar
for index
- Various math on `InvalidIndex` -> `NoneIndex`
- Various mentions of "valid" in comments
- In `value_store.h`, for `IdT::Invalid`, plus one comment
- In `impl.h` and `tokenized_buffer.h`, we had different initialization
of `::None` values (versus `ids.h` syntax) that I fixed manually.
- Spot checks to compile
- Particularly where `is_valid` replacements didn't catch spots due to
different naming.
3. Autoupdate tests
4. verbose.carbon (NOAUTOUPDATE)
5. Comment spot checks
Note there are probably other mentions of "Invalid" that should be swept
up, but I'd like to argue for merging and separating out remaining
cleanup since this is so sweeping (and likely to hit merge conflicts
from churn). We'll probably have lingering mentions of "invalid" for a
bit regardless, just because there are uses of "invalid" in non-Id APIs.
* Implement ignoring the difference between `Self as` and `as`, as well
as `where` clauses at the end of an `impl` declaration when checking
whether `impl` declarations match, from #3763.
* Allow impl declarations with different constraint ids to match, as
long as the facet type of the constraint has the same interface_id and
specific_id.
* Add some TODOs reflecting future facet type resolution.
---------
Co-authored-by: Josh L <josh11b@users.noreply.github.com>
* Change `InterfaceWitness` -> `ImplWitness`
* Include a `SpecificId` in the `ImplWitness`. This allows the
`InstBlock` it contains to have its own identity, allowing it to be
changed as the impl is processed. Evaluation only updates the specific.
* Create the `ImplWitness` at the start of the impl definition. In the
future, this will be populated with the values of non-function
associated constants. For now, it starts full of invalid instruction
ids.
* Implements the model suggested in #4672 .
Note that the non-SemIR testdata changes are to these file:
* `toolchain/check/testdata/impl/lookup/fail_todo_undefined_impl.carbon`
* `toolchain/check/testdata/struct/import.carbon`
* `toolchain/check/testdata/tuple/import.carbon`
The last two are due to an import of generics bug exposed by this PR,
which will be fixed in a follow-on.
---------
Co-authored-by: Josh L <josh11b@users.noreply.github.com>
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
Note that I left some calls to `is_defined()` where I thought they were
interchangeable.
---------
Co-authored-by: Josh L <josh11b@users.noreply.github.com>
This removes a lot of boilerplate `Print` functions in favor of a
CRTP-based approach that uses a `Label` field as an automatic prefix.
This `Label` is also made available for other purposes, particularly
`IdKind` crash messages in this change. In particular, for
`RequireIdKind` in node_stack.h from using numeric IdKinds (e.g., 5 and
24) to something that will print `IdKind(<label>)` (this came up
recently on #toolchain).
While I'm in here, also doing some other tinkering:
- Moving operators to be `friend` members, to reduce the extra
templating now that the base types are templated.
- Adjusts IntId diagnostics from `int [...]` to `int(...)` for
consistency with other id printing.
- Changes InstBlockId's label from "block" to "inst_block", since we
have multiple blocks now.
- Fixes StructTypeFieldsId to use "struct_type_fields" instead of
"type_block" (from `TypeBlockId`)
- Does some more adjustments from camelCase to snake_case for
consistency
This is for more clearly distinct names, and to make it a clearer
transition from `BuiltinInst` for name conflicts. `FloatType` is also an
instruction, and we have `Carbon::Error` (common/error.h). This avoids
affecting tests, although the name is embedded in the builtin test.
In `LegacyFloatType`, `Legacy` because I was having trouble coming up
with a more appropriate name. I'm not clear this is a `FloatLiteralType`
at present, it needs some work to mirror `IntLiteralType`.
In `ErrorInst`, the suffix `Inst` was discussed as good and similar to
`BuiltinInst` (although I'm trying to get rid of that).
Previously Collect() was used for types that implemented
CollectMemUsage() but otherwise Add() was used. This required the caller
to think about the type of the field and know/decide which method to
use.
Now, the caller always uses Collect() unless they are adding specific
byte values, in which case Add is used. Typically then, Add will only be
used to implement the CollectMemUsage() function.
To do this we require all Collect() methods to be templates so that they
all be a single overload set. The Collect on BumpPtrAllocator is
converted to a template that checks
`std::same_as<llvm::BumpPtrAllocator, T>`.
This is in anticipation of making the integer value store be customized
heavily. I'd like to extract it from the common code when doing that, so
first disentangling them here without any intended change in
functionality or behavior to enable that.
I've tried to update `#include`s to be as minimal as I can and added a
few missing includes spotted in the process.
I've split the test for value store to include what was easy focused on
just the value store templates rather than the unified shared value
stores.
This might surface some opportunities for adding more tests, but for
this PR, just doing the minimal restructuring.
If the bucket is of size zero or one, which is expected to be the common
case, then store it directly. For the remaining cases, use a side table
of lookup buckets.
---------
Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
Update impl handling to more closely match other kinds of declaration,
including support for declaring and defining gneeric `impl`s.
When of performing redeclaration lookup for impls by looking for the
self and constraint type, produce a list of impls rather than a single
impl because it's possible for there to be multiple impls with different
deduced parameters but the same self type and constraint. Following
#3763, only consider impls to be redeclarations if they're spelled the
same, not just if they have the same self and constraint types.
No support yet for impl selection to deduce the arguments of a generic
impl.
---------
Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
Introduces the `BindingPattern` and `SymbolicBindingPattern` insts, and
a separate stack of pattern blocks that they are emitted into. The
intent is to generate the corresponding pattern-matching insts (like
`BindName`) from them in a separate pass, but that is deferred to future
PRs.
See
[here](https://docs.google.com/document/d/1U_vQH17V893J9aF1LJXUnFYBNSs2MjKl4bJPaWCB2zo/edit?usp=sharing&resourcekey=0-w0xGYZ0An31Kpz-wvzSXwQ)
for the design this is based on, but note that during review we have
chosen to deviate from that design by putting the patterns in separate
blocks, and omitting the "forward references" from a `BindingPattern` to
its corresponding `BindName`. This in turn necessitates having separate
inst kinds for symbolic and non-symbolic binding patterns.
---------
Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
This works to leverage the capabilities of the hashtable as much as
possible, for example using the key context in the value stores.
However, there may still be opportunities to refactor more deeply and
use the functionality even better. Hopefully this is at least
a reasonable start and gets us a clean baseline.
On an Arm M1, this is a 15% improvement on my large lexing stress test,
but ends up a wash on my x86-64 server. This is a smaller benefit than
I expected, and it's because we're using a set-of-IDs and looking up
values with a key context for things like identifiers. This pattern has
a surprising tradeoff. The new hashtable uses significantly less memory,
a 10% peak RSS reduction just from the hashtable change. But indirecting
through the vector of values makes growing the hashtable dramatically
less cache-friendly: it causes growth to randomly access every key when
rehashing. On x86, everything gained by the faster hashtable is lost in
even slower growth. And even on Arm, this eats into the benefits.
But I have a plan to tweak how identifiers specifically work to avoid
most of the growth, and so I suspect this is the right tradeoff on the
whole. It gives us significant working set size reduction and we can
likely avoid the regressed operation (growth with rehash) in most cases
by clever reserving and if necessary by adding a hash caching layer to
the table infrastructure.
---------
Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
Add an instruction to hold the witness table, along with a corresponding
type to keep things simpler. Add `check/impl.{h,cpp}` to house the new
logic. No checking of impls against interfaces is performed yet.
---------
Co-authored-by: Jon Ross-Perkins <jperkins@google.com>