some improvements to summarysize, fixes #32881

- 0-field mutable structs take 1 word
- include alignment in object sizes
- take uniqueness into account for Strings
- include union selector bytes for Arrays

base/atomics.jl

@@ -334,7 +334,7 @@ inttype(::Type{Float32}) = Int32
 inttype(::Type{Float64}) = Int64
 
 
-gc_alignment(::Type{T}) where {T} = ccall(:jl_alignment, Cint, (Csize_t,), sizeof(T))
+import ..Base.gc_alignment
 
 # All atomic operations have acquire and/or release semantics, depending on
 # whether the load or store values. Most of the time, this is what one wants

base/reflection.jl

@@ -337,6 +337,9 @@ function datatype_alignment(dt::DataType)
     return Int(alignment & 0x1FF)
 end
 
+gc_alignment(sz::Integer) = Int(ccall(:jl_alignment, Cint, (Csize_t,), sz))
+gc_alignment(T::Type) = gc_alignment(Core.sizeof(T))
+
 """
     Base.datatype_haspadding(dt::DataType) -> Bool
 

base/summarysize.jl

@@ -74,14 +74,29 @@ end
     end
     if isa(obj, UnionAll) || isa(obj, Union)
         # black-list of items that don't have a Core.sizeof
-        return 2 * sizeof(Int)
+        sz = 2 * sizeof(Int)
+    else
+        sz = Core.sizeof(obj)
     end
-    return Core.sizeof(obj)
+    if sz == 0
+        # 0-field mutable structs are not unique
+        return gc_alignment(0)
+    end
+    sz += sizeof(Int) # type tag
+    al = gc_alignment(sz)
+    return (sz + al - 1) & -al
 end
 
 (::SummarySize)(obj::Symbol) = 0
 (::SummarySize)(obj::SummarySize) = 0
-(::SummarySize)(obj::String) = Core.sizeof(Int) + Core.sizeof(obj)
+
+function (ss::SummarySize)(obj::String)
+    key = ccall(:jl_value_ptr, Ptr{Cvoid}, (Any,), obj)
+    haskey(ss.seen, key) ? (return 0) : (ss.seen[key] = true)
+    sz = 2*Core.sizeof(Int) + Core.sizeof(obj)
+    al = gc_alignment(sz)
+    return (sz + al - 1) & -al
+end
 
 function (ss::SummarySize)(obj::DataType)
     key = pointer_from_objref(obj)
@@ -103,12 +118,20 @@ end
 
 function (ss::SummarySize)(obj::Array)
     haskey(ss.seen, obj) ? (return 0) : (ss.seen[obj] = true)
-    headersize = 4*sizeof(Int) + 8 + max(0, ndims(obj)-2)*sizeof(Int)
+    headersize = 5*sizeof(Int) + 8 + max(0, ndims(obj)-2)*sizeof(Int)
+    headersize = (headersize + 64 - 1) & -64
     size::Int = headersize
     datakey = unsafe_convert(Ptr{Cvoid}, obj)
     if !haskey(ss.seen, datakey)
         ss.seen[datakey] = true
-        size += Core.sizeof(obj)
+        dsize = Core.sizeof(obj)
+        if isbitsunion(eltype(obj))
+            # add 1 union selector byte for each element
+            dsize += length(obj)
+        end
+        al = gc_alignment(dsize)
+        dsize = (dsize + al - 1) & -al
+        size += dsize
         if !isbitstype(eltype(obj)) && !isempty(obj)
             push!(ss.frontier_x, obj)
             push!(ss.frontier_i, 1)
@@ -120,7 +143,9 @@ end
 function (ss::SummarySize)(obj::SimpleVector)
     key = pointer_from_objref(obj)
     haskey(ss.seen, key) ? (return 0) : (ss.seen[key] = true)
-    size::Int = Core.sizeof(obj)
+    size::Int = sizeof(Int) + Core.sizeof(obj)
+    al = gc_alignment(size)
+    size = (size + al - 1) & -al
     if !isempty(obj)
         push!(ss.frontier_x, obj)
         push!(ss.frontier_i, 1)

test/misc.jl

@@ -183,7 +183,7 @@ let R = Ref{Any}(nothing), depth = 10^6
         R = Ref{Any}(R)
     end
     R = Core.svec(R, R)
-    @test summarysize(R) == (depth + 4) * sizeof(Ptr)
+    @test summarysize(R) == (2*(depth+1) + 4) * sizeof(Ptr)
 end
 
 # issue #25367 - summarysize with reshaped arrays
@@ -194,6 +194,16 @@ let A = zeros(1000), B = reshape(A, (1,1000))
     @test summarysize(A) > sizeof(A)
 end
 
+# issue #32881
+mutable struct S32881; end
+let s = "abc"
+    @test summarysize([s,s]) < summarysize(["abc","xyz"])
+    @test summarysize("a") > sizeof(Int)+1
+end
+@test summarysize(Vector{Union{Nothing,Missing}}(undef, 16)) < summarysize(Vector{Union{Nothing,Missing}}(undef, 32))
+@test summarysize(Vector{Nothing}(undef, 16)) == summarysize(Vector{Nothing}(undef, 32))
+@test summarysize(S32881()) == sizeof(Int)
+
 # issue #13021
 let ex = try
     Main.x13021 = 0