[SR-8526] Memory leak after switch in release configuration #51046

swift-ci · 2018-08-13T11:02:02Z


Previous ID	SR-8526
Radar	rdar://problem/43888666
Original Reporter	rermolov (JIRA User)
Type	Bug
Status	Closed
Resolution	Done

Environment

Xcode 9.4.1 (9F2000), Xcode 10.0 beta 5 (10L221o)

Swift 4.1/Swift 4.2, Release configuration

iPhone X iOS 11.4 / iPhone X Simulator 12.0

Additional Detail from JIRA


Votes	0
Component/s	Compiler
Labels	Bug, Leak, OptimizedOnly
Assignee	@atrick
Priority	Medium

md5: 78345c898f38d78fa1941b37f88392e2

Issue Description:

What I do:

build code below in Release configuration:

final class Subject {
  init() {
    print("subject init")
  }
  deinit {
    print("!!! subject deinit")
  }
}

final class Holder {
  private let objects: [Subject]
  init(objects: [Subject]) {
    self.objects = objects
  }
}

enum Content {
  case empty
  case withSubject(Subject)
}

final class Graph {
  private let holder: Holder

  init(content: Content) {
    let s: Subject
    switch content {
    case .empty:
      s = Subject()
    case let .withSubject(subject):
      s = subject
    }
    holder = Holder(objects: [s])
    if true {
      print(content)
    }
  }
}

func test() {
  print(Graph(content: .empty))
}

What I expect:

Subject is deallocated eventually

What I receive:

Subject is not deallocated. It seems that the problem is in switch + array + if statement: changing any part of this fixes the problem.

Possible workarounds:

1) Extracting switch to extension of Content

2) Making additional init in Holder, that receives only one Subject

3) Extracting if statement to function with content as argument

belkadan · 2018-08-30T17:21:25Z

@swift-ci create

atrick · 2018-09-01T01:00:26Z

As a build-level workaround, you can add this to your swiftc options (Other Swift Flags):

-Xllvm -enable-destroyhoisting=false

atrick · 2018-09-01T02:34:47Z

PR: [Fix a bug in CopyForwarding. Bailout during destoy hoisting. #19103](#19103

Once the algorithm has begun hoisting destroys globally, there's no
way to cleanly bailout. The previous attempt to bailout could result
in an assert or lost destroy in release mode.

This is continued fall-out from changes in the previous release to
upstream SILGen or mandatory passes, such as PredictableMemOps, that
no longer preserve natural variable lifetimes.

In this case, we end up with SIL like this before CopyForwarding:

bb(%arg)
%local_addr = alloc_stack
store %arg to %local_addr
%payload = switch_enum(%arg)
retain %arg
store %arg to %some_addr
destroy_addr %local_addr
release_value %arg

We're attempting to hoist the destroy_addr to its last use, but can't
because the lifetimes of the alloc_stack (%local_addr) and the value
being stored on the stack (%arg) have become mixed up by an upstream
pass. We actually detect this situation now in order to bail-out of
destroy hoisting. Sadly, the bailout might only partially recover in
the case of interesting control flow, as happens in the test case's
Graph.init function. This triggers an assert, but in release mode it
simply drops the destroy.

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

This issue was closed.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[SR-8526] Memory leak after switch in release configuration #51046

[SR-8526] Memory leak after switch in release configuration #51046

swift-ci commented Aug 13, 2018

belkadan commented Aug 30, 2018

atrick commented Sep 1, 2018

atrick commented Sep 1, 2018

[SR-8526] Memory leak after switch in release configuration #51046

[SR-8526] Memory leak after switch in release configuration #51046

Comments

swift-ci commented Aug 13, 2018

belkadan commented Aug 30, 2018

atrick commented Sep 1, 2018

atrick commented Sep 1, 2018