Run Runic after explicit return rule addition

.github/workflows/Check.yml

@@ -0,0 +1,22 @@
+name: Check
+on:
+  push:
+    branches:
+      - 'master'
+      - 'release-'
+    tags:
+      - '*'
+  pull_request:
+jobs:
+  runic:
+    name: Runic formatting
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: julia-actions/setup-julia@v2
+        with:
+          version: "1"
+      - uses: julia-actions/cache@v2
+      - uses: fredrikekre/runic-action@v1
+        with:
+          version: "1"

docs/make.jl

@@ -44,6 +44,7 @@ function main()
             push_preview = true,
         )
     end
+    return
 end
 
 isinteractive() || main()

examples/histogram.jl

@@ -63,12 +63,14 @@ function histogram!(histogram_output, input)
     # Need static block size
     kernel! = histogram_kernel!(backend, (256,))
     kernel!(histogram_output, input, ndrange = size(input))
+    return
 end
 
 function move(backend, input)
     # TODO replace with adapt(backend, input)
     out = KernelAbstractions.allocate(backend, eltype(input), size(input))
     KernelAbstractions.copyto!(backend, out, input)
+    return out
 end
 
 @testset "histogram tests" begin

examples/matmul.jl

@@ -23,6 +23,7 @@ function matmul!(output, a, b)
     backend = KernelAbstractions.get_backend(a)
     kernel! = matmul_kernel!(backend)
     kernel!(output, a, b, ndrange = size(output))
+    return
 end
 
 a = rand!(allocate(backend, Float32, 256, 123))

examples/memcopy.jl

@@ -13,6 +13,7 @@ function mycopy!(A, B)
 
     kernel = copy_kernel!(backend)
     kernel(A, B, ndrange = length(A))
+    return
 end
 
 A = KernelAbstractions.zeros(backend, Float64, 128, 128)

examples/memcopy_static.jl

@@ -13,6 +13,7 @@ function mycopy_static!(A, B)
 
     kernel = copy_kernel!(backend, 32, size(A)) # if size(A) varies this will cause recompilation
     kernel(A, B, ndrange = size(A))
+    return
 end
 
 A = KernelAbstractions.zeros(backend, Float64, 128, 128)

examples/mpi.jl

@@ -9,6 +9,7 @@ function cooperative_test!(req)
         done, _ = MPI.Test(req, MPI.Status)
         yield()
     end
+    return
 end
 
 function cooperative_wait(task::Task)
@@ -17,6 +18,7 @@ function cooperative_wait(task::Task)
         yield()
     end
     wait(task)
+    return
 end
 
 function exchange!(h_send_buf, d_recv_buf, h_recv_buf, src_rank, dst_rank, comm)
@@ -68,6 +70,7 @@ function main(backend)
     cooperative_wait(send_task)
 
     @test all(d_recv_buf .== src_rank)
+    return
 end
 
 main(backend)

examples/naive_transpose.jl

@@ -18,6 +18,7 @@ function naive_transpose!(a, b)
     groupsize = KernelAbstractions.isgpu(backend) ? 256 : 1024
     kernel! = naive_transpose_kernel!(backend, groupsize)
     kernel!(a, b, ndrange = size(a))
+    return
 end
 
 # resolution of grid will be res*res

ext/EnzymeCore07Ext.jl

@@ -22,7 +22,7 @@ function EnzymeRules.forward(
     f = kernel.f
     fwd_kernel = similar(kernel, cpu_fwd)
 
-    fwd_kernel(f, args...; ndrange, workgroupsize)
+    return fwd_kernel(f, args...; ndrange, workgroupsize)
 end
 
 function EnzymeRules.forward(
@@ -36,7 +36,7 @@ function EnzymeRules.forward(
     f = kernel.f
     fwd_kernel = similar(kernel, gpu_fwd)
 
-    fwd_kernel(f, args...; ndrange, workgroupsize)
+    return fwd_kernel(f, args...; ndrange, workgroupsize)
 end
 
 _enzyme_mkcontext(kernel::Kernel{CPU}, ndrange, iterspace, dynamic) =

ext/EnzymeCore08Ext.jl

@@ -23,7 +23,7 @@ function EnzymeRules.forward(
     f = kernel.f
     fwd_kernel = similar(kernel, cpu_fwd)
 
-    fwd_kernel(config, f, args...; ndrange, workgroupsize)
+    return fwd_kernel(config, f, args...; ndrange, workgroupsize)
 end
 
 function EnzymeRules.forward(
@@ -38,7 +38,7 @@ function EnzymeRules.forward(
     f = kernel.f
     fwd_kernel = similar(kernel, gpu_fwd)
 
-    fwd_kernel(config, f, args...; ndrange, workgroupsize)
+    return fwd_kernel(config, f, args...; ndrange, workgroupsize)
 end
 
 _enzyme_mkcontext(kernel::Kernel{CPU}, ndrange, iterspace, dynamic) =

src/macros.jl

@@ -103,7 +103,7 @@ function transform_gpu!(def, constargs, force_inbounds)
         Expr(:block, let_constargs...),
         body,
     )
-    return nothing
+    return
 end
 
 # The hard case, transform the function for CPU execution
@@ -138,7 +138,7 @@ function transform_cpu!(def, constargs, force_inbounds)
         Expr(:block, let_constargs...),
         Expr(:block, new_stmts...),
     )
-    return nothing
+    return
 end
 
 struct WorkgroupLoop

test/compiler.jl

@@ -39,7 +39,7 @@ end
 
 function compiler_testsuite(backend, ArrayT)
     kernel = index(CPU(), DynamicSize(), DynamicSize())
-    iterspace = NDRange{1, StaticSize{(128,)}, StaticSize{(8,)}}();
+    iterspace = NDRange{1, StaticSize{(128,)}, StaticSize{(8,)}}()
     ctx = KernelAbstractions.mkcontext(kernel, 1, nothing, iterspace, Val(KernelAbstractions.NoDynamicCheck()))
     @test ndims(ctx) == 1
     @test KernelAbstractions.__index_Global_NTuple(ctx, CartesianIndex(1)) == (1,)
@@ -75,4 +75,5 @@ function compiler_testsuite(backend, ArrayT)
         # test that there is no invoke of overdub
         @test !any(check_for_overdub, CI.code)
     end
+    return
 end

test/convert.jl

@@ -63,4 +63,5 @@ function convert_testsuite(backend, ArrayT)
             @test d_B[:, i + 20] == round.(d_A)
         end
     end
+    return
 end

test/copyto.jl

@@ -17,4 +17,5 @@ function copyto_testsuite(Backend, ArrayT)
 
     @test isapprox(a, Array(A))
     @test isapprox(a, Array(B))
+    return
 end

test/examples.jl

@@ -6,7 +6,7 @@ function find_sources(path::String, sources = String[])
     elseif endswith(path, ".jl")
         push!(sources, path)
     end
-    sources
+    return sources
 end
 
 function examples_testsuite(backend_str)
@@ -27,4 +27,5 @@ function examples_testsuite(backend_str)
         end
 
     end
+    return
 end

test/extensions/enzyme.jl

@@ -11,6 +11,7 @@ function square_caller(A, backend)
     kernel = square!(backend)
     kernel(A, ndrange = size(A))
     KernelAbstractions.synchronize(backend)
+    return
 end
 
 
@@ -23,6 +24,7 @@ function mul_caller(A, B, backend)
     kernel = mul!(backend)
     kernel(A, B, ndrange = size(A))
     KernelAbstractions.synchronize(backend)
+    return
 end
 
 function enzyme_testsuite(backend, ArrayT, supports_reverse = true)
@@ -58,4 +60,5 @@ function enzyme_testsuite(backend, ArrayT, supports_reverse = true)
         @test all(dA .≈ 2:2:128)
 
     end
+    return
 end

test/localmem.jl

@@ -8,7 +8,7 @@ using Test
     end
     I = @index(Global, Linear)
     i = @index(Local, Linear)
-    lmem = @localmem Int (N,) # Ok iff groupsize is static 
+    lmem = @localmem Int (N,) # Ok iff groupsize is static
     @inbounds begin
         lmem[i] = i
         @synchronize
@@ -23,7 +23,7 @@ end
     end
     I = @index(Global, Linear)
     i = @index(Local, Linear)
-    lmem = @localmem Int (N,) # Ok iff groupsize is static 
+    lmem = @localmem Int (N,) # Ok iff groupsize is static
     @inbounds begin
         lmem[i] = i + 3
         for j in 1:2
@@ -47,4 +47,5 @@ function localmem_testsuite(backend, ArrayT)
             @test all(B[49:64] .== 16:-1:1)
         end
     end
+    return
 end

test/nditeration.jl

@@ -30,7 +30,7 @@ function nditeration_testsuite()
 
     function check(idx, offset, offset_x, offset_y, Dim_x, Dim_y)
         N = Dim_x * Dim_y
-        all(p -> p[1] == p[2], zip(idx[(offset * N .+ 1):N], CartesianIndices(((offset_x * Dim_x .+ 1):Dim_x, (offset_y * Dim_y .+ 1):Dim_y))))
+        return all(p -> p[1] == p[2], zip(idx[(offset * N .+ 1):N], CartesianIndices(((offset_x * Dim_x .+ 1):Dim_x, (offset_y * Dim_y .+ 1):Dim_y))))
     end
 
     @testset "linear_iteration" begin
@@ -93,4 +93,5 @@ function nditeration_testsuite()
             @test ndims(ndrange) == 2
         end
     end
+    return
 end

test/print_test.jl

@@ -23,4 +23,5 @@ function printing_testsuite(backend)
         end
         @test true
     end
+    return
 end

test/private.jl

@@ -106,4 +106,5 @@ function private_testsuite(backend, ArrayT)
             end
         end
     end
+    return
 end

test/reflection.jl

@@ -29,6 +29,7 @@ function test_typed_kernel_dynamic(backend, backend_str, ArrayT)
         @test isa(res, Pair{Core.CodeInfo, DataType})
     end
     @test isa(res[1].code, Array{Any, 1})
+    return
 end
 
 function test_typed_kernel_dynamic_no_info(backend, backend_str, ArrayT)
@@ -43,6 +44,7 @@ function test_typed_kernel_dynamic_no_info(backend, backend_str, ArrayT)
         @test isa(res, Pair{Core.CodeInfo, DataType})
     end
     @test isa(res[1].code, Array{Any, 1})
+    return
 end
 
 function test_typed_kernel_static(backend, backend_str, ArrayT)
@@ -59,6 +61,7 @@ function test_typed_kernel_static(backend, backend_str, ArrayT)
         @test isa(res, Pair{Core.CodeInfo, DataType})
     end
     @test isa(res[1].code, Array{Any, 1})
+    return
 end
 
 function test_typed_kernel_no_optimize(backend, backend_str, ArrayT)
@@ -72,6 +75,7 @@ function test_typed_kernel_no_optimize(backend, backend_str, ArrayT)
     res_opt = @ka_code_typed kernel(A, ndrange = size(A))
     # FIXME: Need a better test
     # @test size(res[1].code) < size(res_opt[1].code)
+    return
 end
 
 function test_expr_kernel(backend, backend_str, ArrayT)
@@ -89,6 +93,7 @@ function test_expr_kernel(backend, backend_str, ArrayT)
         @test isa(res, Pair{Core.CodeInfo, DataType})
     end
     @test isa(res[1].code, Array{Any, 1})
+    return
 end
 
 function reflection_testsuite(backend, backend_str, ArrayT)
@@ -97,4 +102,5 @@ function reflection_testsuite(backend, backend_str, ArrayT)
     test_typed_kernel_static(backend, backend_str, ArrayT)
     test_typed_kernel_no_optimize(backend, backend_str, ArrayT)
     test_expr_kernel(backend, backend_str, ArrayT)
+    return
 end

test/specialfunctions.jl

@@ -53,4 +53,5 @@ function specialfunctions_testsuite(Backend)
         synchronize(backend)
         @test Array(cy) ≈ SpecialFunctions.erfc.(x)
     end
+    return
 end

test/test.jl

@@ -154,7 +154,7 @@ function unittest_testsuite(Backend, backend_str, backend_mod, BackendArrayT; sk
     @conditional_testset "Const" skip_tests begin
         let kernel = constarg(Backend(), 8, (1024,))
             # this is poking at internals
-            iterspace = NDRange{1, StaticSize{(128,)}, StaticSize{(8,)}}();
+            iterspace = NDRange{1, StaticSize{(128,)}, StaticSize{(8,)}}()
             ctx = if Backend == CPU
                 KernelAbstractions.mkcontext(kernel, 1, nothing, iterspace, Val(NoDynamicCheck()))
             else
@@ -266,6 +266,7 @@ function unittest_testsuite(Backend, backend_str, backend_mod, BackendArrayT; sk
     function f(KernelAbstractions.@context, a)
         I = @index(Global, Linear)
         a[I] = 1
+        return
     end
     @kernel cpu = false function context_kernel(a)
         f(KernelAbstractions.@context, a)
@@ -324,4 +325,5 @@ function unittest_testsuite(Backend, backend_str, backend_mod, BackendArrayT; sk
         end
     end
 
+    return
 end

test/testsuite.jl

@@ -11,7 +11,7 @@ const Pkg = Base.require(
 )
 
 macro conditional_testset(name, skip_tests, expr)
-    esc(
+    return esc(
         quote
             @testset $name begin
                 if $name ∉ $skip_tests
@@ -91,6 +91,8 @@ function testsuite(backend, backend_str, backend_mod, AT, DAT; skip_tests = Set{
     @conditional_testset "Examples" skip_tests begin
         examples_testsuite(backend_str)
     end
+
+    return
 end
 
 end

test/unroll.jl

@@ -44,4 +44,5 @@ function unroll_testsuite(backend, ArrayT)
     kernel2! = kernel_unroll2!(backend(), 1, 1)
     kernel2!(a)
     synchronize(backend())
+    return
 end