diff --git a/base/sparse/higherorderfns.jl b/base/sparse/higherorderfns.jl
index 5ccd57e02a2f2..32a255e133281 100644
--- a/base/sparse/higherorderfns.jl
+++ b/base/sparse/higherorderfns.jl
@@ -107,18 +107,6 @@ broadcast(f::Tf, A::SparseMatrixCSC) where {Tf} = _noshapecheck_map(f, A)
     end
     return C
 end
-@inline function broadcast!(f::Tf, dest::SparseVecOrMat, ::Void, As::Vararg{Any,N}) where {Tf,N}
-    if f isa typeof(identity) && N == 1
-        A = As[1]
-        if A isa Number
-            return fill!(dest, A)
-        elseif A isa AbstractArray && Base.axes(dest) == Base.axes(A)
-            return copyto!(dest, A)
-        end
-    end
-    spbroadcast_args!(f, dest, Broadcast.combine_styles(As...), As...)
-    return dest
-end
 
 # the following three similar defs are necessary for type stability in the mixed vector/matrix case
 broadcast(f::Tf, A::SparseVector, Bs::Vararg{SparseVector,N}) where {Tf,N} =
@@ -1015,25 +1003,27 @@ broadcast(f, ::PromoteToSparse, ::Void, ::Void, As::Vararg{Any,N}) where {N} =
 # For broadcast! with ::Any inputs, we need a layer of indirection to determine whether
 # the inputs can be promoted to SparseVecOrMat. If it's just SparseVecOrMat and scalars,
 # we can handle it here, otherwise see below for the promotion machinery.
-function spbroadcast_args!(f::Tf, C, ::SPVM, A::SparseVecOrMat, Bs::Vararg{SparseVecOrMat,N}) where {Tf,N}
-    _aresameshape(C, A, Bs...) && return _noshapecheck_map!(f, C, A, Bs...)
-    Base.Broadcast.check_broadcast_indices(axes(C), A, Bs...)
+function broadcast!(f::Tf, dest::SparseVecOrMat, ::SPVM, A::SparseVecOrMat, Bs::Vararg{SparseVecOrMat,N}) where {Tf,N}
+    if f isa typeof(identity) && N == 0 && Base.axes(dest) == Base.axes(A)
+        return copyto!(dest, A)
+    end
+    _aresameshape(dest, A, Bs...) && return _noshapecheck_map!(f, dest, A, Bs...)
+    Base.Broadcast.check_broadcast_indices(axes(dest), A, Bs...)
     fofzeros = f(_zeros_eltypes(A, Bs...)...)
     fpreszeros = _iszero(fofzeros)
-    return fpreszeros ? _broadcast_zeropres!(f, C, A, Bs...) :
-                        _broadcast_notzeropres!(f, fofzeros, C, A, Bs...)
+    fpreszeros ? _broadcast_zeropres!(f, dest, A, Bs...) :
+                        _broadcast_notzeropres!(f, fofzeros, dest, A, Bs...)
+    return dest
 end
-function spbroadcast_args!(f::Tf, dest, ::SPVM, mixedsrcargs::Vararg{Any,N}) where {Tf,N}
+function broadcast!(f::Tf, dest::SparseVecOrMat, ::SPVM, mixedsrcargs::Vararg{Any,N}) where {Tf,N}
     # mixedsrcargs contains nothing but SparseVecOrMat and scalars
     parevalf, passedsrcargstup = capturescalars(f, mixedsrcargs)
-    return broadcast!(parevalf, dest, passedsrcargstup...)
-end
-function spbroadcast_args!(f::Tf, dest, ::PromoteToSparse, mixedsrcargs::Vararg{Any,N}) where {Tf,N}
-    broadcast!(f, dest, map(_sparsifystructured, mixedsrcargs)...)
+    broadcast!(parevalf, dest, nothing, passedsrcargstup...)
+    return dest
 end
-function spbroadcast_args!(f::Tf, dest, ::Any, mixedsrcargs::Vararg{Any,N}) where {Tf,N}
-    # Fallback. From a performance perspective would it be best to densify?
-    Broadcast._broadcast!(f, dest, mixedsrcargs...)
+function broadcast!(f::Tf, dest::SparseVecOrMat, ::PromoteToSparse, mixedsrcargs::Vararg{Any,N}) where {Tf,N}
+    broadcast!(f, dest, nothing, map(_sparsifystructured, mixedsrcargs)...)
+    return dest
 end
 
 _sparsifystructured(M::AbstractMatrix) = SparseMatrixCSC(M)