Skip to content

Fake PR to test the upstreaming of commit up to gerris/rebase/2025-11-12 #4282

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Open
wants to merge 14 commits into
base: dkm/upstream-base
Choose a base branch
from
Open

Fake PR to test the upstreaming of commit up to gerris/rebase/2025-11-12 #4282

wants to merge 14 commits into from

Conversation

@dkm
Copy link
Member

@dkm dkm commented Nov 12, 2025

This is a fake PR, not meant to be merged. It tries to merge commits to upstream with an upstream base branch dkm/upstream-base.
We're only interested by the CI results.
-- gerris 🦀

@dkm dkm force-pushed the gerris/rebase/2025-11-12 branch from e06bccc to cfa155b Compare November 12, 2025 11:07
@dkm
Copy link
Member Author

dkm commented Nov 12, 2025

@CohenArthur this should be ready for upstream.

@dkm
Copy link
Member Author

dkm commented Nov 12, 2025

ah, not quite. There's a revert in there.

Lucas Ly Ba and others added 14 commits November 12, 2025 12:15
@dkm
gccrs: fix segfault on exported macro
6311097 
An imbricated exported macro leads to a segfault.

gcc/rust/ChangeLog:

	* metadata/rust-export-metadata.cc (ExportContext::emit_macro):
	Change method argument NodeId to AST::MacroRulesDefinition.
	* metadata/rust-export-metadata.h:
	Likewise.
	* util/rust-hir-map.cc (Mappings::insert_exported_macro):
	Insert AST::MacroRulesDefinition instead of NodeId.
	* util/rust-hir-map.h:
	Change methods declarations of exported macros.

gcc/testsuite/ChangeLog:

	* rust/compile/issue-3617.rs: New test.

Signed-off-by: Lucas Ly Ba <[email protected]>
@Polygonalr @dkm
gccrs: Add minus sign compilation for LiteralPattern
9a93563 
GIMPLE output for literalpattern_neg.rs test case:

...
  x = -55;
  RUSTTMP.2 = x;
  if (RUSTTMP.2 == 55) goto <D.113>; else goto <D.114>;
  <D.113>:
  {
    RUSTTMP.1 = 1;
    goto <D.107>;
  }
  <D.114>:
  if (RUSTTMP.2 == -55) goto <D.115>; else goto <D.116>;
  <D.115>:
  {
    RUSTTMP.1 = 0;
    goto <D.107>;
  }
  <D.116>:
  if (1 != 0) goto <D.117>; else goto <D.118>;
  <D.117>:
  {
    RUSTTMP.1 = 1;
    goto <D.107>;
  }
...

gcc/rust/ChangeLog:

	* parse/rust-parse-impl.h (parse_literal_or_range_pattern): Parse minus sign
	properly for LiteralPattern.
	* ast/rust-pattern.h (LiteralPattern): Add has_minus boolean for LiteralPattern.
	* hir/tree/rust-hir-pattern.h (LiteralPattern): Ditto.
	* ast/rust-pattern.cc (LiteralPattern::as_string): Update to include minus sign
	if present.
	* hir/tree/rust-hir.cc (LiteralPattern::as_string): Ditto.
	* hir/rust-ast-lower-pattern.cc (visit(LiteralPattern)): Pass has_minus boolean
	from AST to HIR.
	* backend/rust-compile-pattern.cc (CompilePatternCheckExpr::visit(LiteralPattern)):
	Compile litexpr as negative if minus sign is present.

Signed-off-by: Yap Zhi Heng <[email protected]>
@vishruth-thimmaiah @dkm
gccrs: Fix ICE with non-trailing const defaults
8d7edfa 
When a const generic with a default value is not trailing, emit an
error.

gcc/rust/ChangeLog:

	* parse/rust-parse-impl.h (Parser::parse_generic_params): Emit
	an error when const generics with a default value is not
	trailing.

gcc/testsuite/ChangeLog:

	* rust/compile/const_generics_17.rs: New test.
	* rust/compile/generics14.rs: New test.

Signed-off-by: vishruth-thimmaiah <[email protected]>
@Polygonalr @dkm
gccrs: Fix RangePattern negative literal bounds being treated as po…
ab254f8 
…sitive

GIMPLE output for compile/issue-4242.rs:

...

  x = 1;
  RUSTTMP.2 = x;
  _1 = RUSTTMP.2 >= -55;
  _2 = RUSTTMP.2 < 0;
  _3 = _1 & _2;
  if (_3 != 0) goto <D.112>; else goto <D.113>;
  <D.112>:
  {
    RUSTTMP.1 = 2;
    goto <D.105>;
  }
  <D.113>:
  _4 = RUSTTMP.2 >= -99;
  _5 = RUSTTMP.2 < -55;
  _6 = _4 & _5;
  if (_6 != 0) goto <D.114>; else goto <D.115>;
  <D.114>:
  {
    RUSTTMP.1 = 3;
    goto <D.105>;
  }
...

gcc/rust/ChangeLog:

	* backend/rust-compile-pattern.cc (compile_range_pattern_bound): Set litexpr
	to negative if has_minus is present in the RangePatternBoundLiteral param.

Signed-off-by: Yap Zhi Heng <[email protected]>
@Polygonalr @dkm
gccrs: Implement E0579 error checking in RangePattern compilation
4192f59 
Checks whether upper bound of range is not lower or equal to the lower bound.

gcc/rust/ChangeLog:

	* backend/rust-compile-pattern.cc(compilePatternCheckExpr::visit(RangePattern)):
	Add E0579 check to ensure that lower bound is always below upper bound.

Signed-off-by: Yap Zhi Heng <[email protected]>
@philberty @dkm
gccrs: Add support for binding const generic values to paths
8346ee6 
Const generics bind values which can be accessed like a normal path but the difference
is that they can be true expression values not just type paths. This patch adds support
to resolving a method inference which passes a generic value into the method and fixes
some missed bugs along the way. The tricky part was that there is a case where in the
return position of a method returning a const param type vs the type of the method
there is a special case in the unify rules so that we unify the specified type of the
const param type not the const param itself.

gcc/rust/ChangeLog:

	* backend/rust-compile-resolve-path.cc: handle const param values
	* typecheck/rust-hir-type-check-item.cc: generate const infer vars when required
	* typecheck/rust-type-util.cc (unify_site_and): handle a null param cleanup
	* typecheck/rust-tyty-util.cc (TyVar::get_implicit_const_infer_var): helper interface
	* typecheck/rust-tyty-util.h: update header prototypes
	* typecheck/rust-tyty.cc (BaseType::is_concrete): correctly handle const types
	(ConstParamType::get_name): emit the specified type
	(ConstParamType::is_equal): fix recursion loop
	* typecheck/rust-unify.cc (UnifyRules::go): const infer vars need cleanup too
	* typecheck/rust-unify.h: support base generics

gcc/testsuite/ChangeLog:

	* rust/execute/torture/const-generics-2.rs: New test.

Signed-off-by: Philip Herron <[email protected]>
@lucasly-ba @dkm
gccrs: fix error multiple cfg predicates
4ee7370 
gcc/rust/ChangeLog:

	* ast/rust-ast.cc (Attribute::check_cfg_predicate):
	Make error.

gcc/testsuite/ChangeLog:

	* rust/compile/issue-4267.rs: New test.

Signed-off-by: Lucas Ly Ba <[email protected]>
@lucasly-ba @dkm
gccrs: fix segfault with empty cfg attribute
6abcb83 
gcc/rust/ChangeLog:

	* ast/rust-ast.cc (Attribute::check_cfg_predicate): add cfg path in condition

gcc/testsuite/ChangeLog:

	* rust/compile/issue-4261.rs: New test.

Signed-off-by: Lucas Ly Ba <[email protected]>
@lucasly-ba @dkm
gccrs: fix cfg attribute error with literal predicate
9d008cd 
gcc/rust/ChangeLog:

	* ast/rust-ast.cc (MetaItemLitExpr::check_cfg_predicate): Make error.

gcc/testsuite/ChangeLog:

	* rust/compile/issue-4222.rs: New test.

Signed-off-by: Lucas Ly Ba <[email protected]>
@lucasly-ba @dkm
gccrs: fix ICE on missing pattern in while loop
86d3a33 
Adds a proper check for missing patterns in while expressions.

Fixes #4162

gcc/rust/ChangeLog:

	* parse/rust-parse-impl.h(Parser<ManagedTokenSource>::parse_while_let_loop_expr):
	Add check for missing pattern.

gcc/testsuite/ChangeLog:
	* rust/compile/issue-4162.rs: New test.

Signed-off-by: Lucas Ly Ba <[email protected]>
@lucasly-ba @dkm
gccrs: make invalid inner attributes show error
77393a3 
gcc/rust/ChangeLog:

	* ast/rust-ast.cc (Attribute::is_derive):
	Change is_derive method with its valid path.
	* util/rust-attribute-values.h:
	Delete redudant derive attribute.
	* util/rust-attributes.cc (AttributeChecker::check_inner_attribute):
	Helper method for check_inner_attributes
	(AttributeChecker::check_inner_attributes):
	Implement method for errors check.
	* util/rust-attributes.h:
	Add methods above in header.

gcc/testsuite/ChangeLog:

	* rust/compile/issue-4212.rs:
	* rust/compile/issue-4219.rs: New test.

Signed-off-by: Lucas Ly Ba <[email protected]>
@philberty @dkm
gccrs: Fix const generics handling on array types
73ed68d 
When we were processing generic const param types on arrays the size type
was overriding the const param decl because of a hirid reference mismatch

Fixes #3879

gcc/rust/ChangeLog:

	* typecheck/rust-hir-type-check-type.cc (TypeCheckType::visit): fix mappings

gcc/testsuite/ChangeLog:

	* rust/compile/const_generics_18.rs: New test.
	* rust/compile/const_generics_19.rs: New test.
	* rust/execute/torture/const-generics-3.rs: New test.
	* rust/execute/torture/const-generics-4.rs: New test.

Signed-off-by: Philip Herron <[email protected]>
@philberty @dkm
gccrs: Support generic constant impl items
a5ca87c 
Impl items can have constants defined which could in turn be generic this was
not supported by gccrs and missed. So for example:

  impl<T> Foo<T> {
    const MAGIC: usize = mem::size_of::<T>();
  }

This is a normal type parameter but in order to setup the generics we need to
create a synthetic TyTy::FnType so we can bind the parent's impl generics to
the type system and it just works like any other generic item at that point.
Then for example we have:

  impl<const N: usize> Foo<N> {
    const VALUE: usize = N;
  }

Again we consistently bind the this const generic parameter the same way so
the lazy evaluation of the generic can take place.

gcc/rust/ChangeLog:

	* backend/rust-compile-item.cc (CompileItem::visit): support the synthetic function consts
	* typecheck/rust-hir-type-check-expr.cc (TypeCheckExpr::Resolve): likewise
	* typecheck/rust-hir-type-check-implitem.cc (TypeCheckImplItem::visit): create the synth
	* typecheck/rust-tyty.h: new flag for synthetic constant

gcc/testsuite/ChangeLog:

	* rust/execute/torture/const-generics-5.rs: New test.
	* rust/execute/torture/const-generics-6.rs: New test.
	* rust/execute/torture/const-generics-7.rs: New test.

Signed-off-by: Philip Herron <[email protected]>
@philberty @dkm
gccrs: Fix segv in errors in type checking an impl item
aa4982d 
When we typecheck a trait impl block item, at the end we validate it
against the trait definition by doing a final unify but if the type check
fails on the the impl item it returns NULL here. The other issue was that
if we fail to resolve the specified lifetime we return error early, this
changes the typechecking to default to an anon lifetime so we can continue
typechecking.

Fixes #4188

gcc/rust/ChangeLog:

	* typecheck/rust-hir-type-check-implitem.cc (TypeCheckImplItemWithTrait::visit): null guard
	* typecheck/rust-hir-type-check.cc (TraitItemReference::get_type_from_fn): default to anon

gcc/testsuite/ChangeLog:

	* rust/compile/issue-4188.rs: New test.

Signed-off-by: Philip Herron <[email protected]>
@dkm dkm force-pushed the gerris/rebase/2025-11-12 branch from cfa155b to aa4982d Compare November 12, 2025 11:15
@dkm
Copy link
Member Author

dkm commented Nov 12, 2025

Ok, done, should be ok now.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

6 participants

@dkm @Polygonalr @vishruth-thimmaiah @philberty @lucasly-ba