Commit Graph
211 Commits
Author SHA1 Message Date
josh11bandJosh L ea0b0b4b48 Add facet type values and an instruction that produces them (#4460)
Still to do:
* Represent facet type values in a canonical form
* Produce & consume facet type values instead of interface values
* `type` should be associated with a canonical facet type value
* Support `&` on facet type values
* Type check and enforce requirements in facet types

---------

Co-authored-by: Josh L <josh11b@users.noreply.github.com>
2024-11-04 21:43:55 +00:00
Richard Smith 26d7717d60 Insert a value_of_initializer after a call to ImplicitAs where possible. (#4473)
This avoids going through memory when performing an implicit conversion
to a type with a by-copy value representation, such as i32.
2024-11-04 16:06:19 +00:00
Richard Smith 44fe65fbe5 Rename BigInt to IntLiteral. (#4476)
In preparation for changing the type of integer literals to
`IntLiteral`.
2024-11-02 02:09:10 +00:00
Richard Smith db76e81630 Rename IntLiteral to IntValue. (#4475)
This instruction represents integer values, whether they come from
literals or calculations, so it the old name is inaccurate. I also plan
to rename `BigInt` to `IntLiteral` based on recent discussion and this
change aims to avoid confusion stemming from the same name being used
for two different things.

I'm not renaming `FloatLiteral` because recent discussion suggests we
may want distinct `FloatLiteral` versus `FloatValue` representations in
SemIR.
2024-11-02 01:19:39 +00:00
Geoff RomerandRichard Smith ac5cc33da4 Model return slot as parameter in lowering (#4457)
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
2024-11-01 21:48:09 +00:00
Richard Smith 32e5212daa Fix lowering of a conversion from a type with a pointer value representation to a type with a copy value representation. (#4467)
We previously generated a `value_bind` instruction of the wrong type,
resulting in lowering building bad LLVM IR.

Also fix another issue exposed by this change, where we would import
constants without marking their types as complete, and then crash in
lowering while trying to lower them. This happens in particular for the
`FunctionType`s of functions in `ImplDecl`s. Address this by skipping
lowering for constants with incomplete types.
2024-11-01 14:52:57 +00:00
Richard Smith e2ab97672d Fix lowering of specific_functions referring to methods. (#4464)
In this case, the callee may be non-constant because it includes a
reference to `self`, so we need to be able to lower a non-constant
`specific_function`.
2024-10-31 20:07:30 +00:00
Richard Smith df68bf9f71 Switch to using Core.BigInt as the type of the size of a type literal. (#4450)
This removes one of the few ways in which `i32` is special and gets us
closer to removing it as a special case.
2024-10-28 23:35:08 +00:00
Geoff Romer b67d03126e Separate inst kind for out params (#4442) 2024-10-24 15:22:32 +00:00
Richard Smith 2e63da1a40 Move diagnostic kind name to the end of the diagnostic. (#4437)
Also surround it in square brackets rather than parentheses. This
matches the format used by Clang and GCC, and means diagnostics will
still match the `file:line:col: error: ` pattern used by some IDE tools.

Before:
```console
fail_builtins.carbon:11:11: error(AliasRequiresNameRef): alias initializer must be a name reference
```

After:
```console
fail_builtins.carbon:11:11: error: alias initializer must be a name reference [AliasRequiresNameRef]
```

Also tighten up test regex to only match on `STDERR` lines that list a
file name.
2024-10-23 16:56:23 +00:00
Geoff Romer 9266f867f9 Model the return slot as an output parameter (#4432)
Also fix `Param` insts to have meaningful names in pretty-printing, to
help clarify relationship with return slot.
2024-10-23 16:53:34 +00:00
Geoff Romer 223c5cb04b Restructure handling of runtime parameters (#4422)
- Generate runtime indices as part of pattern matching, rather than as a
separate postprocessing/rewriting step.
- In contexts where runtime parameters aren't permitted, avoid emitting
insts for them to begin with, rather than trying to detect the problem
and rewrite the IR to remove them later on.
2024-10-21 19:53:38 +00:00
Jon Ross-Perkins 62c36eceda Support printing the diagnostic kind for verification. (#4425)
This is to help identify which diagnostics we're actually using.

Note that driver/testdata still has tests which don't pass this flag,
and so continue to test the kind-less (default) behavior.
2024-10-18 22:33:56 +00:00
Jon Ross-PerkinsandRichard Smith b5a837aa89 Refactor modifier formatting to remove string passing. (#4418)
I'm taking the approach of making DiagnosticBase an API so that we can
pass similar diagnostics as parameters. An alternative would be to do
the function_ref approach we've done elsewhere, but these felt more
boilerplate to me.

Note I'm also modifying messages here. Let me know if you'd like
different changes and/or just keeping current formatting (keeping
current formatting would also allow removing some of the templating I've
added, but it felt helpful putting explicit tokens where possible). But
also, things like "`protected` not allowed on `interface` declaration at
file scope" were part of the phrasing issue, I think.

---------

Co-authored-by: Richard Smith <richard@metafoo.co.uk>
2024-10-17 22:47:40 +00:00
Richard Smith a02dfe0226 Superficial support for Core.BigInt type (#4414)
Add a `Core.BigInt` type and a corresponding builtin type in the
toolchain. See [corresponding section of the
design](https://docs.carbon-lang.dev/docs/design/expressions/literals.html#defined-types).

So far this type is not used for anything, and there is no way to create
an instance of it.
2024-10-16 23:27:01 +00:00
David BlaikieandRichard Smith dfed743de2 Add vtable pointers to class layout (#4407)
A small step to virtual functions - adding vtable pointers to the
layout, but not initializing or otherwise using them at this stage.

A few open design questions I'd love feedback on:

* Is this the right/good enough SemIR representation for now? This patch
adds a `is_dynamic` attribute to `SemIR::Class` and populates/flags it
based on the flag of the base class, or if any virtual function is
declared in the class (or, at least that's my intent). Some other
options include:
* Each `Class` could store a `ClassId` (or `TypeId`?) of the (possibly
indirect, possibly self) base class that is the first one that is
dynamic/has a vtable pointer
* Could make the property narrower, like `has vtable pointer` and have
it `true` only on the type that introduces the vtable - then derived
classes would have to walk their base classes to check if they're the
one that needs to define the vtable pointer or not
* Should the vtable be the first element in the type? If there's a
non-dynamic base type, we could have a layout that's `{<non-dynamic base
type>, vtable ptr, <derived members>}`? Derived types would still be
able to uniquely identify where their vtable pointer is just fine... -
and the vtable pointer is, in a sense, a member of that intermediate
type, so it does seem a bit strange to force it to the front - but I
guess it's probably more efficient in some ways?

Open to any other suggestions/advice/thoughts on the direction, etc.

---------

Co-authored-by: Richard Smith <richard@metafoo.co.uk>
2024-10-16 21:26:17 +00:00
Geoff RomerandJon Ross-Perkins 9d942f4633 Generate parameter pattern-match IR from pattern IR (#4388)
Also propagate the pattern IR along with the pattern-match IR, and use
it where appropriate.

Strictly speaking, some parts of the pattern-match IR are allocated
eagerly, while traversing the pattern's parse tree, but they still
aren't actually emitted until we traverse the associated pattern insts.

---------

Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
2024-10-16 19:15:29 +00:00
Richard Smith 851ef2c517 Initial, very rough lowering for calls to specific functions and specific function declarations. (#4399)
Generate declarations of specific functions on demand. Definitions are
not emitted yet, and I'm using a temporary, known-broken scheme for name
mangling.
2024-10-10 23:37:16 +00:00
Richard Smith 1a1bfd2eb2 Track and resolve the specific callee in a call to a generic function (#4395)
Add a new `specific_function` instruction that represents a generic
function plus its deduced argument list as a callee in a function call.
The new instruction can only appear as the immediate operand of a call
instruction, so we give it a builtin placeholder type.

At the end of each file, require definitions for all specific functions
used in that file. Resolve the generic with the argument list to produce
those specific function definitions as needed, and diagnose if the
generic doesn't have a definition available.

A few tests are updated in cases where they declared and used generic
functions but didn't previously provide a function definition.
2024-10-10 20:52:46 +00:00
82937e1a3c Change how to get info for a parameter (#4366)
Updates `SemIR::Function::GetParamFromParamRefId` to return more
information in the form of a new `ParamInfo` struct. This struct has a
method for getting the `NameId` from the name binding instruction. The
callers previously got it from the `Param` instruction, but the plan is
for that instruction to no longer be associated with a name.

---------

Co-authored-by: Josh L <josh11b@users.noreply.github.com>
Co-authored-by: Geoff Romer <gromer@google.com>
2024-10-04 16:30:00 +00:00
Richard Smith 568ad197d1 Track the instruction used to name the type and constraint in an impl. (#4368)
This is necessary in order to have access to the specific versions of
their constant values in a generic impl.

Stub out impl deduction.
2024-10-04 00:06:55 +00:00
Richard Smith 5ab957d012 Make ImplDecls evaluate to themselves. (#4352)
Change ImplDecls so they evaluate to themselves in general, rather than
having a special case in import handling that pretends that they do.
2024-09-27 22:39:56 +00:00
bdbd1079a6 where check stage, step 2: SemIR (#4349)
The check stage now produces SemIR instructions to represent a `where`
clause. It still does not check types.

---------

Co-authored-by: Josh L <josh11b@users.noreply.github.com>
Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
2024-09-27 01:41:56 +00:00
Jon Ross-PerkinsandRichard Smith e7aebbe581 Update basic diagnostic capitalization/punctuation (#4328)
This is a primarily automated change:

- Search & replace for capitalization
-
`(CARBON_DIAGNOSTIC\((?:\n\s+)?\w+,(?:\n\s+)?\s\w+,(?:\n\s+)?\s")([A-Z])`
    - `$1\L$2`
- Search & replace for period
-
`(CARBON_DIAGNOSTIC\((?:\n\s+)?\w+,(?:\n\s+)?\s\w+,(?:\n\s+)?\s"(?:[^)]|\n)+)\.("[,)])`
    - `$1$2`
- Limited search & replace for `ERROR: ` -> `error: ` in streamed things
- Leaving a TODO for command_line because there's more cleanup that can
be done there
- Modify diagnostic_consumer.cpp
    - ERROR -> error
    - WARNING -> warning

---------

Co-authored-by: Richard Smith <richard@metafoo.co.uk>
2024-09-19 21:32:53 +00:00
Richard Smith 2044366652 Support initialization of specific classes from struct literals (#4320)
Add support for initializing types like `GenericClass(i32)` from a
struct literal. A new kind of instruction, `complete_type_witness`, is
added to the class definition to track the object representation type so
that it's visible to the generics machinery. Accesses to the object
representation of a class have all been updated to pass in the class's
`SpecificId` so that the types of the fields of the specific class are
used instead of the types of the fields of the generic class in places
that look at the object representation -- primarily class
initialization.
2024-09-19 19:18:32 +00:00
David BlaikieandJon Ross-Perkins 4631767943 Stop name mangling at an imported package name (#4294)
To ensure the outer package name is not mangled into imported names.

With this change I can successfully link/run a two-file example:
```
package Mod;
fn HelloWorld() {
  Core.Print(42);
}
```
```
import Mod;
fn Run() -> i32 {
  Mod.HelloWorld();
  return 0;
}
```
```
$ carbon compile mod.carbon main.carbon
$ carbon link mod.o main.o --output=a.out
$ ./a.out
42
```
\o/

---------

Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
2024-09-18 22:01:08 +00:00
Brymer Meneses 7f930d0f58 Use TupleAccess instead of TupleIndex (#4318)
This change removes the `TupleIndex` instruction, and instead
consolidate it with the `TupleAccess` instruction, per this
[discussion](https://discord.com/channels/655572317891461132/655578254970716160/1271195835975204946).
This change, in turn removes `AnyAggregateIndex`.
2024-09-17 20:26:48 +00:00
Richard Smith 0354efa1fc Rework how we check calls to support deduced implicit parameters (#4302)
Instead of the `call` instruction having a block with one argument per
explicit argument, preceded optionally by `self` and followed optionally
by a return slot, change the `call` to store only the *runtime*
arguments. Store an index on the runtime parameters to make it easier to
determine the correspondence between arguments and parameters in a call.
Compile-time parameters, whether implicit or explicit, are no longer
included in the call argument list. Instead, they're tracked only in the
`specific_id` on the callee.

For calls to generic classes and generic interfaces, it no longer makes
sense to form a `call` instruction, given that the entirety of the
result is determined by the `specific_id`, which is now formed when
checking the call. Instead, the `call` instruction now only models
function calls, and not calls to other kinds of parameterized entity
names, and we create a `class_type` or `interface_type` instead of a
`call` instruction to model these kinds of calls. Notionally the model
here is that we're following the #3720 approach for calls, but for now
we inline the `Call.Op` function when forming SemIR.

We now also track the enclosing specific for a generic class or generic
interface that appears within an enclosing generic. This is necessary in
order for deduction of the inner generic parameters to not get confused
by the outer generic parameters being absent.

In order to not regress diagnostics, the template argument deduction
mechanism has been extended to specify the name of the parameter we're
deducing against when possible, and call arity mismatch errors are now
diagnosed before performing deduction rather than afterwards.
2024-09-13 21:31:43 +00:00
David Blaikie ec63d2be21 Assert that the llvm::Function is created with the same name as requested (#4306)
This will catch some cases of bugs in the name mangling logic - if
within a single file we incorrectly mangle two distinct entities to the
same name, llvm::Function will assign a new name to the second copy
showing one of the two entities should have a distinct name/is missing
something in their mangling.
2024-09-13 00:33:29 +00:00
4845f40dff Switch CARBON_CHECK to a format string API (#4285)
This switches `DCHECK` and `FATAL` as well.

The goal is to reduce the code size impact of these assertions so that
we can keep more of them enabled. Currently, the largest cost I see from
`CHECK` is not the actual check or the cold code itself, but actually
the failure to inline trivial functions due to the presence of the cold
code. This means that our goal isn't to reduce apparent code size in the
final binary but the LLVM IR cost assessed for these routines in the
inliner, which closely correlates with code size but is a bit different.

As discussed in #4283, experimentation shows that a single function call
with a minimal number of arguments is the lowest cost model for these.
This is easily achieved with a format-string API that internally uses
`llvm::formatv`. This PR is essentially the `CHECK` version of #4283.

However, the check macros are substantially harder to make work with
both format strings and streaming because they also take a condition.
Also, unexpectedly, I was very successful at devising a regular
expression based automated rewrite from the streaming to the format
string form with only low 10s of manual fixes. This includes compacting
strings broken up across lines, etc. Given how well that went, I've
prepared this PR which just directly switches to the format string API
and migrate everything to use it.

One nice side-effect is that the format string approach ends up greatly
simplifying the implementation here as well.

This is ... *shockingly* effective. Parsing speeds up by more than 3%
with just this change. And checking speeds up by **8%** with this change
alone:
```
BM_CompileAPIFileDenseDecls<Phase::Parse>/256      86.3µs ± 1%  82.9µs ± 1%  -3.94%  (p=0.000 n=17+19)
BM_CompileAPIFileDenseDecls<Phase::Parse>/1024      431µs ± 1%   415µs ± 1%  -3.76%  (p=0.000 n=18+19)
BM_CompileAPIFileDenseDecls<Phase::Parse>/4096     1.77ms ± 1%  1.71ms ± 1%  -3.18%  (p=0.000 n=18+19)
BM_CompileAPIFileDenseDecls<Phase::Parse>/16384    7.44ms ± 1%  7.17ms ± 2%  -3.56%  (p=0.000 n=18+20)
BM_CompileAPIFileDenseDecls<Phase::Parse>/65536    30.7ms ± 1%  29.7ms ± 1%  -3.15%  (p=0.000 n=18+20)
BM_CompileAPIFileDenseDecls<Phase::Parse>/262144    131ms ± 1%   127ms ± 1%  -2.81%  (p=0.000 n=18+18)
BM_CompileAPIFileDenseDecls<Phase::Check>/256       878µs ± 2%   800µs ± 1%  -8.91%  (p=0.000 n=19+20)
BM_CompileAPIFileDenseDecls<Phase::Check>/1024     1.88ms ± 2%  1.72ms ± 1%  -8.56%  (p=0.000 n=19+20)
BM_CompileAPIFileDenseDecls<Phase::Check>/4096     5.78ms ± 2%  5.28ms ± 1%  -8.70%  (p=0.000 n=20+18)
BM_CompileAPIFileDenseDecls<Phase::Check>/16384    21.9ms ± 1%  20.1ms ± 1%  -8.02%  (p=0.000 n=18+20)
BM_CompileAPIFileDenseDecls<Phase::Check>/65536    90.4ms ± 2%  83.1ms ± 1%  -8.04%  (p=0.000 n=19+20)
BM_CompileAPIFileDenseDecls<Phase::Check>/262144    381ms ± 2%   352ms ± 1%  -7.79%  (p=0.000 n=19+19)
```

---------

Co-authored-by: Richard Smith <richard@metafoo.co.uk>
Co-authored-by: josh11b <15258583+josh11b@users.noreply.github.com>
2024-09-12 16:42:08 +00:00
Chandler Carruth 0c8ab663c9 Migrate all CARBON_VLOG to the format string variant. (#4284)
This mostly uses a hilarious set of regular expressions to mechanically
switch all but two uses, and then manually fixed the last two. There
weren't too many.

Also simplifies the `vlog` implementation now that it's all going
through a format string.

This alone has a nice impact on parse and check of about 2% and 1%
respectively. The impact on lex in my timings looks like noise (no
change in instruction count, unlike the other phases).
```
name                                               old cpu/op   new cpu/op   delta
BM_CompileAPIFileDenseDecls<Phase::Lex>/256        39.1µs ± 3%  38.1µs ± 2%  -2.42%  (p=0.000 n=20+19)
BM_CompileAPIFileDenseDecls<Phase::Lex>/1024        187µs ± 3%   183µs ± 1%  -2.30%  (p=0.000 n=20+20)
BM_CompileAPIFileDenseDecls<Phase::Lex>/4096        776µs ± 4%   756µs ± 1%  -2.62%  (p=0.000 n=20+20)
BM_CompileAPIFileDenseDecls<Phase::Lex>/16384      3.36ms ± 1%  3.33ms ± 1%  -0.90%  (p=0.000 n=18+18)
BM_CompileAPIFileDenseDecls<Phase::Lex>/65536      14.4ms ± 2%  14.2ms ± 1%  -1.41%  (p=0.000 n=20+20)
BM_CompileAPIFileDenseDecls<Phase::Lex>/262144     65.7ms ± 1%  65.2ms ± 2%  -0.86%  (p=0.002 n=19+20)
BM_CompileAPIFileDenseDecls<Phase::Parse>/256      87.5µs ± 1%  86.3µs ± 1%  -1.43%  (p=0.000 n=19+20)
BM_CompileAPIFileDenseDecls<Phase::Parse>/1024      438µs ± 2%   431µs ± 1%  -1.54%  (p=0.000 n=19+20)
BM_CompileAPIFileDenseDecls<Phase::Parse>/4096     1.81ms ± 2%  1.77ms ± 1%  -2.12%  (p=0.000 n=20+20)
BM_CompileAPIFileDenseDecls<Phase::Parse>/16384    7.54ms ± 1%  7.43ms ± 1%  -1.44%  (p=0.000 n=19+20)
BM_CompileAPIFileDenseDecls<Phase::Parse>/65536    31.2ms ± 1%  30.6ms ± 1%  -2.03%  (p=0.000 n=20+20)
BM_CompileAPIFileDenseDecls<Phase::Parse>/262144    133ms ± 1%   130ms ± 1%  -1.85%  (p=0.000 n=20+20)
BM_CompileAPIFileDenseDecls<Phase::Check>/256       882µs ± 1%   878µs ± 1%  -0.52%  (p=0.001 n=17+19)
BM_CompileAPIFileDenseDecls<Phase::Check>/1024     1.90ms ± 2%  1.88ms ± 1%  -1.17%  (p=0.000 n=19+19)
BM_CompileAPIFileDenseDecls<Phase::Check>/4096     5.85ms ± 2%  5.76ms ± 1%  -1.43%  (p=0.000 n=20+19)
BM_CompileAPIFileDenseDecls<Phase::Check>/16384    22.2ms ± 2%  21.9ms ± 2%  -1.20%  (p=0.000 n=20+19)
BM_CompileAPIFileDenseDecls<Phase::Check>/65536    91.2ms ± 2%  90.3ms ± 1%  -1.00%  (p=0.000 n=20+19)
BM_CompileAPIFileDenseDecls<Phase::Check>/262144    382ms ± 1%   380ms ± 1%  -0.51%  (p=0.003 n=18+19)
```
2024-09-11 12:11:23 +00:00
Jon Ross-Perkins d4c7743d18 Update tests to use [[@TEST_NAME]] (#4293)
Applies #4278 TEST_NAME substitution to tests. Note I've tried to
structure commits as:

1. Do all the replacements.
2. autoupdate (nothing else) -- this shows incorrect updates.
3. Fix up manually, including autoupdates to get back to original
output.
2024-09-10 22:52:23 +00:00
a548eff0bb Rudimentary name mangling support (#4267)
This seems to be enough to avoid naming collisions for functions in any
of the current test cases (verified by asserting that the name of the
`llvm::Function` matches the name passed to create it - not triggering
LLVM's numbering that happens when names collide)

It currently implements mangling for namespace scopes, class scopes, and
impls.
Nothing generic is mangled yet - haven't looked at how that works,
though evidently it's not covered by existing testing, I guess.

Follow-up change will document the current mangling algorithm in
`toolchain/docs/lower.md`

---------

Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
Co-authored-by: Carbon Infra Bot <carbon-external-infra@google.com>
2024-09-06 23:10:21 +00:00
Jon Ross-PerkinsandDavid Blaikie 1412ecd3f4 Handle unknown lines in DebugInfo (#4252)
Mainly, changes the default from -1 to 0 in DiagnosticLoc, still trying
to keep reusing that. Nothing except for the lowered output is affected,
so I think this is fine.

Also, have lowering consistently call GetDiagnosticLoc.

Pulls in one of the CHECKs suggested from #4251 

Co-authored-by: David Blaikie <dblaikie@gmail.com>
2024-09-04 21:25:02 +00:00
Richard Smith 16fd645663 Add source locations to interface_witness and interface_witness_access instructions. (#4265) 2024-08-28 23:30:00 +00:00
David Blaikie 67b287a2f2 DebugInfo: Add the name as-written, separate from the mangled name (#4253)
There's not much mangling happening yet - but Run -> main (and some
overloading numbering happening, maybe LLVM is doing that 'helpfully'
under the hood?) is enough to demonstrate this improvement/fix.

Ah, here it is:
```
#0  llvm::ValueSymbolTable::makeUniqueName (this=0x50287fe5b6c0, V=0x50287fe827e8, UniqueName="F") at external/_main~llvm_project~llvm-project/llvm/lib/IR/ValueSymbolTable.cpp:45
#1  0x000055555c40f964 in llvm::ValueSymbolTable::reinsertValue (this=0x50287fe5b6c0, V=0x50287fe827e8) at external/_main~llvm_project~llvm-project/llvm/lib/IR/ValueSymbolTable.cpp:100
#2  0x000055555c2a91df in llvm::SymbolTableListTraits<llvm::Function>::addNodeToList (this=0x50287fd16f18, V=0x50287fe827e8) at external/_main~llvm_project~llvm-project/llvm/lib/IR/SymbolTableListTraitsImpl.h:75
#3  0x000055555c2a90e5 in llvm::iplist_impl<llvm::simple_ilist<llvm::Function>, llvm::SymbolTableListTraits<llvm::Function> >::insert (this=0x50287fd16f18, where=..., New=0x50287fe827e8)
    at external/_main~llvm_project~llvm-project/llvm/include/llvm/ADT/ilist.h:166
#4  0x000055555c27fef2 in llvm::iplist_impl<llvm::simple_ilist<llvm::Function>, llvm::SymbolTableListTraits<llvm::Function> >::push_back (this=0x50287fd16f18, val=0x50287fe827e8) at external/_main~llvm_project~llvm-project/llvm/include/llvm/ADT/ilist.h:250
#5  0x000055555c27faeb in llvm::Function::Function (this=0x50287fe827e8, Ty=0x50287fd43058, Linkage=llvm::GlobalValue::ExternalLinkage, AddrSpace=0, name="F", ParentModule=0x50287fd16f00) at external/_main~llvm_project~llvm-project/llvm/lib/IR/Function.cpp:521
#6  0x0000555559441f95 in llvm::Function::Create (Ty=0x50287fd43058, Linkage=llvm::GlobalValue::ExternalLinkage, AddrSpace=0, N="F", M=0x50287fd16f00) at external/_main~llvm_project~llvm-project/llvm/include/llvm/IR/Function.h:175
#7  0x000055555c27ebac in llvm::Function::Create (Ty=0x50287fd43058, Linkage=llvm::GlobalValue::ExternalLinkage, N="F", M=...) at external/_main~llvm_project~llvm-project/llvm/lib/IR/Function.cpp:398
#8  0x0000555558ce7bb5 in Carbon::Lower::FileContext::BuildFunctionDecl (this=0x7fffffffc438, function_id=...) at toolchain/lower/file_context.cpp:257
```

That's where LLVM decides to make a new name (name.number) when asked to
create a new global with the same name as an existing global.

It's not a valid mangling scheme - since the name won't be stable
between different compilations, but it is enough to make
single-compilation code build/run for now.
2024-08-28 19:18:54 +00:00
David Blaikie 2a7c2c5df2 Add test coverage for lowering raw identifiers in function names (#4257) 2024-08-27 19:49:18 +00:00
David Blaikie 4c9021cbc8 Change branch-to-fatal to check condition (#4256) 2024-08-27 18:09:54 +00:00
David Blaikie 5177ebbe2a Fix crash when disabling debug info (#4249)
The source line debug info generation assumed that the function would
have debug info. Check for a non-null di_subprogram_ to ensure we are
emitting debug info for the function.

Rather than checking the di_builder_ - this way if we implement
`nodebug` function attributes, it'll fall out naturally (by creating a
null di_subprogram_) rather than having to come back and change this
from "is debug info enabled" to "is debug info enabled for this
function" later on.
2024-08-26 21:13:39 +00:00
David Blaikie 0ae2a3907e Add line-level debug info (#4247)
Seems to work with lldb ( https://pastebin.com/igKkNECm ), though gdb
has /some/ trouble with the paths (they aren't complete - just using the
filename directly, not providing the working directory - might be some
quick hacks that can help there).
2024-08-26 15:49:31 +00:00
David BlaikieandJon Ross-Perkins c5ada29ba9 Add filename and line number to function debug info metadata (#4243)
Refactors a bunch of the SemIRDiagnosticConverter to be able to use that
from Lower to access source locations there to use in debug info.

I assume some of this is a bit jank/would need to be fixed/improved in
the future - like the context functor that's passed into ConvertLoc?
(not totally clear what that's for/what the debug info will be missing
out on in its absence, I could throw a FIXME in there if you like)

---------

Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
2024-08-23 00:05:26 +00:00
David Blaikie 1e1034ee81 Enable debug info by default (#4232)
Discussed in the toolchain meeting today - we'd like to try having this
on by default and see if the cost isn't too high.

The nodebug test is a bit verbose, because it doesn't have the
`--exclude-dump-file-prefix` that test_file would usually add. Is there
a nicer way I could write this test to verify that --no-debug-info does
what it's meant to?
2024-08-22 20:27:45 +00:00
David BlaikieandJon Ross-Perkins ea8ad22a17 Add function debug info descriptions (#4233)
Still doesn't have line tables, so of limited value (at least now
this'll be enough that LLVM really generates debug info into the
resulting object file (whereas with only the compilation unit metadata,
LLVM will consider it empty and avoid emitting any of it)) - but another
step along the path.

This also doesn't attach the right source location to the functions -
I'll do that in a follow-up change because I think it'll require the
majority of the refactoring between driver and check to extract the
essential functionality sem_ir_diagnostic_converter, I think, to allow
retrieving source locations during lowering.

---------

Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
2024-08-22 17:19:30 +00:00
d57aa57215 Add rudimentry debug info metadata emission (#4225)
This adds just the debug info metadata for Compilation Units (the top
level container of debug info) - but without anything in them, LLVM
won't emit them at all, so while this is testable at the IR level, it
isn't observable at the object level until more debug info is added.

A couple of starting points in this patch:
* A flag (`--debug-info`, seems to match the naming/style of other flags
in the carbon driver, though this is different from the naming
conventions of clang/gcc) that enables debug info when lowering. Open to
other names/approaches (on by default? historically debug info's been to
large/expensive to do this, so sticking with that precedent for now).
* Enabling that flag by default in the lowering tests - I do find the
churn on golden tests a bit rough, and adding more features to all the
tests means more churn, but it seems consistent with the approach so far
- keep an eye on this and perhaps revisit this if the churn gets too
annoying

---------

Co-authored-by: Carbon Infra Bot <carbon-external-infra@google.com>
Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
2024-08-20 20:00:35 +00:00
R B 61e87c3a88 Lower global variables (#4228) 2024-08-19 22:43:07 +00:00
Brymer Meneses 34ae2b001b move tuple testdata from index/ to tuple/access (#4226) 2024-08-19 16:47:14 +00:00
Brymer Meneses c353f6bd78 change tuple index (#4218)
This changes the tuple index from tuple[0] to tuple.0 in accordance with
the accepted propsal
https://github.com/carbon-language/carbon-lang/pull/3646

I messed up syncing to trunk on my original PR
https://github.com/carbon-language/carbon-lang/pull/4186, that's why I'm
starting on a blank state. Please let me know if I missed incorporating
a change from my prior PR.
2024-08-15 16:06:02 +00:00
josh11bandJosh L bfe0fc7cc8 Update to a recent LLVM version and the changes to LLVM IR. (#4212)
Co-authored-by: Josh L <josh11b@users.noreply.github.com>
2024-08-12 22:29:33 +00:00
Richard Smith 37a8bfa488 Refactor ReturnTypeInfo and InitRepr. (#4169)
Rename `ReturnInfo` to `ReturnTypeInfo`. Move it and `InitRepr` into
`type_info.h` alongside `ValueRepr`. Replace `ReturnSlot` with
`InitRepr`, and extend `InitRepr` to be able to represent the
incomplete-type case instead of CHECK-failing. Remove `has_return_slot`
from `InitRepr` and instead only provide that as part of
`ReturnTypeInfo`.
2024-07-25 21:41:33 +00:00
Richard Smith a9e835f3dc Remove caching of return slot usage. (#4163)
The caching isn't buying us much, and is adding complexity and
divergence between the codepaths for generic and non-generic functions.

This means we no longer suppress diagnostics for the second or
subsequent time we call a function with an incomplete return type. If we
want to add that back, it might be worth considering moving the
suppression to `TryToCompleteType` and only diagnosing that a type is
incomplete once, regardless of why we're requiring it to be complete.
2024-07-25 20:29:59 +00:00