[SR-14516] Swift generates subscript.read accessors that allocate! #56868

weissi · 2021-04-21T21:47:35Z


Previous ID	SR-14516
Radar	rdar://problem/77008933
Original Reporter	@weissi
Type	Bug

Attachment: Download

Environment

Apple Swift version 5.4 (swiftlang-1205.0.22.2 clang-1205.0.19.29)

Additional Detail from JIRA


Votes	0
Component/s	Compiler
Labels	Bug, 5.4Regression, Optimizer
Assignee	None
Priority	Medium

md5: 06a0e70a5391143833033eeeb67e35f7

Issue Description:

The following (sorry, this used to be 1700 lines, and creduce reduced this) program allocates 10,000 times:

extension CircularBuffer {
    @inline(never)
    func F() {
        _ = first!     // << this allocates as it calls generic specialization <repro.NIOAny> of repro.CircularBuffer.subscript.read : (repro.CircularBuffer<A>.Index) -> A
    }
}
protocol p {
    init()
}
struct CircularBuffer<d: p>{
    var e: ContiguousArray<d?> = [.init()]
    struct Index: Comparable {
        var f: Int = 0
        var i: Int {
            return f
        }
        static func == (a: Index, j: Index) -> Bool {
                return a.i > j.i
        }
        static func < (aa: Index, j: Index) -> Bool {
                return aa.i < j.i
        }
    }
    func index(after: Index) -> Index {
        return index(after, offsetBy: 1)
    }
    subscript(k: Index) -> d {
        get {
            return e[k.i]!
        }
    }
    var startIndex = Index()
    var endIndex = Index()
}
extension CircularBuffer: Collection {}

enum XY {
    case X
    case Y
}

struct NIOAny: p {
    let av: aw = .ay(())
    enum aw {
        case ax(XY)
        case ay(Any)
        init<az>(a: az) {
            self = .ay(a)
        }
    }
}

let b = CircularBuffer<NIOAny>()
for _ in 0..<10000 {
    b.F()
}

here's how it goes:

b.F calls b.first (which is provided automatically from Collection)
b.first calls the subscript's read accessor which has an unconditional malloc (for coroutines I think) in there 🙁

                       _$s5repro14CircularBufferVyxAC5IndexVyx_GcirAA6NIOAnyV_Tg5:        // generic specialization <repro.NIOAny> of repro.CircularBuffer.subscript.read : (repro.CircularBuffer<A>.Index) -> A
0x0000000100002110         push       rbp                                       ; CODE XREF=_$s5repro14CircularBufferV1FyyFAA6NIOAnyV_Tg5+34
0x0000000100002111         mov        rbp, rsp
0x0000000100002114         push       r15
0x0000000100002116         push       r14
0x0000000100002118         push       r12
0x000000010000211a         push       rbx
0x000000010000211b         sub        rsp, 0x30
0x000000010000211f         mov        r14, rdx
0x0000000100002122         mov        r15, rsi
0x0000000100002125         mov        r12, rdi
0x0000000100002128         mov        edi, 0x21                                 ; argument "size" for method imp___stubs__malloc
0x000000010000212d         call       imp___stubs__malloc                       ; malloc

the read accessors really shouldn't allocate as that defeats many other optimisations.

Full program attached.

Repro:

$ swiftc -O ~/tmp/repro.swift && sudo ~/devel/swift-nio/dev/malloc-aggregation.d -c ./repro
dtrace: system integrity protection is on, some features will not be available


=====
This will collect stack shots of allocations and print it when you exit dtrace.
So go ahead, run your tests and then press Ctrl+C in this window to see the aggregated result
=====
[...]

              libsystem_malloc.dylib`malloc
              repro`specialized CircularBuffer.subscript.read+0x22
              repro`specialized CircularBuffer.F()+0x27
              repro`main+0x7c
              libdyld.dylib`start+0x1
              0x1
            10000

See how we get 10,000 allocations (malloc) through the subscript.read by calling F 10,000 times?

This seems to be a Swift 5.4 regression. We found this issue in the NIO CI which regresses allocations in a few allocation counter tests but only in 5.4.

Also affects

Swift version 5.4-dev (LLVM 7a20f40c45aca5d, Swift 031b848b7092c06)

In case you're into creduce, this is the interestingness test I used (for Linux)

#!/bin/bash

set -eu

swiftc -O repro.swift
objdump -d repro |  grep -A14 '14CircularBufferVyxAC5IndexVyx_GcirAA6NIOAnyV_Tg5>:' | grep -q malloc

The text was updated successfully, but these errors were encountered:

weissi · 2021-04-21T21:48:55Z

CC @eeckstein & @rjmccall because this seems to be coro related.

weissi · 2021-04-22T10:16:15Z

@swift-ci create

weissi · 2024-05-03T16:22:29Z

2024 edition:

cat repro.swift | swiftc -module-name T -O -emit-assembly - | grep -A1000 '^_$s1T14CircularBufferVyxAC5IndexVyx_Gcir:' | sed -e 1d -e '/^_/q' | grep malloc
	bl	_malloc

still calls malloc on today's Swift 6 preview.

tbkka · 2024-05-06T17:00:47Z

CC: @nate-chandler

weissi · 2025-06-11T08:49:38Z

This particular repro might be fixed in 6.2. If I re-run my original program, I do not see any traces of allocations functions. The following DTrace/ 8000 AWK program prints any function that was called more than 1000 times. No allocations in there (would be malloc_type_malloc or similar):

$ swiftc -O repro.swift && sudo dtrace -n 'pid$target:::entry { @[probefunc] = count(); } :::END { printa("%10@d\t%s\n", @); }' -c ./repro 2>&1 | awk 'BEGIN { FS="\t"} $1 > 1000 { print($1, $2) }'
dtrace: system integrity protection is on, some features will not be available 

dtrace: description 'pid$target:::entry ' matched 37885 probes 
dtrace: pid 67435 has exited 
CPU     ID                    FUNCTION:NAME 
     10000 __swift_destroy_boxed_opaque_existential_0
     10000 destroy for ClosedRange<>.Index
     10000 destroy for NIOAny
     10000 destroy for Optional
     10000 initializeWithCopy for ClosedRange<>.Index
     10000 initializeWithCopy for NIOAny
     10000 initializeWithCopy for Optional
     10000 outlined destroy of NIOAny?
     10000 outlined init with copy of NIOAny?
     10000 specialized CircularBuffer.F()
     10000 storeEnumTagSinglePayload for NIOAny
     10000 swift::metadataimpl::BufferValueWitnesses<swift::metadataimpl::ValueWitnesses<swift::metadataimpl::AggregateBox<>>, true, 0ul, 
     10000 swift::metadataimpl::ValueWitnesses<swift::metadataimpl::AggregateBox<>>::destroy(swift::OpaqueValue*, swift::TargetMetadata<sw
     20000 getEnumTagSinglePayload for NIOAny
     20001 __swift_instantiateConcreteTypeFromMangledName

weissi · 2025-06-11T08:59:49Z

Yes, repro is no longer valid. Expanded (and cleaned it up a little):

extension CircularBuffer {
    @inline(never)
    func HARNESS() {
        let x = first!     // << this allocates as it calls generic specialization <repro.NIOAny> of repro.CircularBuffer.subscript.read : (repro.CircularBuffer<A>.Index) -> A
        precondition(x.isGood())
    }
}
protocol MyProtocol {
    init()
    func isGood() -> Bool
}
struct CircularBuffer<d: MyProtocol>{
    var e: ContiguousArray<d?> = [.init()]
    struct Index: Comparable {
        var f: Int = 0
        var i: Int {
            return f
        }
        static func == (a: Index, j: Index) -> Bool {
                return a.i > j.i
        }
        static func < (aa: Index, j: Index) -> Bool {
                return aa.i < j.i
        }
    }
    func index(after: Index) -> Index {
        return index(after, offsetBy: 1)
    }
    subscript(k: Index) -> d {
        get {
            return e[k.i]!
        }
    }
    var startIndex = Index()
    var endIndex = Index()
}
extension CircularBuffer: Collection {}

enum XY {
    case X
    case Y
}

struct NIOAny: MyProtocol {
    let av: aw = .ay(())
    enum aw {
        case ax(XY)
        case ay(Any)
        init<az>(a: az) {
            self = .ay(a)
        }
    }

    @inline(never)
    func isGood() -> Bool {
        switch self.av {
            case .ay(let value):
            if value as? () != nil {
                return true
            } else {
                return false
            }
            default:
            return false
        }
    }
}

let b = CircularBuffer<NIOAny>()
for _ in 0..<10000 {
    b.HARNESS()
}

compile & run

$ swiftc -O repro.swift && sudo dtrace -n 'pid$target:::entry { @[probefunc] = count(); } :::END { printa("%10@d\t%s\n", @); }' -c ./repro 2>&1 | awk 'BEGIN { FS="\t"} $1 > 1000 { print($1, $2) }'
dtrace: system integrity protection is on, some features will not be available 

dtrace: description 'pid$target:::entry ' matched 37889 probes 
dtrace: pid 67721 has exited 
CPU     ID                    FUNCTION:NAME 
     10000 NIOAny.isGood()
     10000 __swift_destroy_boxed_opaque_existential_0
     10000 getEnumTagSinglePayload for NIOAny
     10000 initializeWithCopy for ClosedRange<>.Index
     10000 initializeWithCopy for NIOAny
     10000 initializeWithCopy for Optional
     10000 outlined destroy of NIOAny
     10000 outlined init with copy of NIOAny.aw
     10000 outlined init with copy of NIOAny?
     10000 outlined init with take of Any
     10000 specialized CircularBuffer.HARNESS()
     10000 specialized Collection.first.getter
     10000 storeEnumTagSinglePayload for NIOAny
     10000 swift::TargetExistentialTypeMetadata<swift::InProcess>::getRepresentation() const
     10000 swift::TargetExistentialTypeMetadata<swift::InProcess>::projectValue(swift::OpaqueValue const*) const
     10000 swift::TargetOpaqueExistentialContainer<swift::InProcess>::projectValue() const
     10000 swift::metadataimpl::ValueWitnesses<swift::metadataimpl::AggregateBox<>>::destroy(swift::OpaqueValue*, swift::TargetMetadata<sw
     10000 swift::metadataimpl::ValueWitnesses<swift::metadataimpl::AggregateBox<>>::initializeWithTake(swift::OpaqueValue*, swift::Opaque
     10000 swift_dynamicCast
     10000 tryCastToTuple(swift::OpaqueValue*, swift::TargetMetadata<swift::InProcess> const*, swift::OpaqueValue*, swift::TargetMetadata<
     10001 __swift_instantiateConcreteTypeFromMangledName
     20000 swift::metadataimpl::BufferValueWitnesses<swift::metadataimpl::ValueWitnesses<swift::metadataimpl::AggregateBox<>>, true, 0ul, 
     20000 tryCast(swift::OpaqueValue*, swift::TargetMetadata<swift::InProcess> const*, swift::OpaqueValue*, swift::TargetMetadata<swift::
johannes:/tmp
$ swiftc -version
swift-driver version: 1.120.5 Apple Swift version 6.1.2 (swiftlang-6.1.2.1.2 clang-1700.0.13.5)
Target: arm64-apple-macosx15.0

weissi · 2025-06-11T09:01:43Z

However it is still emitting the allocating read subscript accessor:

$ cat repro.swift | swiftc -module-name T -O -emit-assembly - | grep -A35 '^_$s1T14CircularBufferVyxAC5IndexVyx_Gcir:' | sed -e 1d -e '/^_/q' | grep malloc
	bl	_malloc

swift-ci transferred this issue from apple/swift-issues Apr 25, 2022

AnthonyLatsis added SILOptimizer Area → compiler: SIL optimization passes and removed Optimizer labels Nov 3, 2022

AnthonyLatsis added regression swift 5.4 and removed 5.4 regression labels Nov 20, 2022

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[SR-14516] Swift generates subscript.read accessors that allocate! #56868

[SR-14516] Swift generates subscript.read accessors that allocate! #56868

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

[SR-14516] Swift generates subscript.read accessors that allocate! #56868

[SR-14516] Swift generates subscript.read accessors that allocate! #56868

Comments

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!