TCPStream crash on iOS ARM64 device

[ghmrs356]

Using on iPhone 5s for the first time, crashes in TCPStream.m in TCPStream -disconnect at line 122. Any idea what is going on?

- (void) disconnect
{
    if( _stream ) {
        LogTo(TCP,@"Disconnect %@",self);
        _stream.delegate = nil;
        [_stream close];
        _stream = nil;
    }
    if( _conn ) {
        [_conn _streamDisconnected: self];
        _conn = nil;  <----------- crashes here
    }
}

Thread 1, Queue : com.apple.main-thread
#0  0x00000001985b5d6c in objc_object::release() ()
#1  0x00000001003700fc in -[TCPStream disconnect] at /Users/markus/Projects/SOFramework/MYNetwork/TCP/TCPStream.m:122
#2  0x000000010035e464 in -[BLIPWriter disconnect] at /Users/markus/Projects/SOFramework/MYNetwork/BLIP/BLIPWriter.m:32
#3  0x000000010036ae38 in -[TCPConnection _streamCanClose:] at /Users/markus/Projects/SOFramework/MYNetwork/TCP/TCPConnection.m:413
#4  0x000000010037025c in -[TCPStream close] at /Users/markus/Projects/SOFramework/MYNetwork/TCP/TCPStream.m:137
#5  0x0000000100370e18 in -[TCPStream stream:handleEvent:] at /Users/markus/Projects/SOFramework/MYNetwork/TCP/TCPStream.m:235
#6  0x000000018c5fc62c in _signalEventSync ()
#7  0x000000018c608754 in _cfstream_solo_signalEventSync ()
#8  0x000000018c5fc184 in _CFStreamSignalEvent ()
#9  0x000000018c1ce8f4 in CoreReadStreamCFStreamSupport::coreStreamReadEvent(__CoreReadStream*, unsigned long) ()
#10 0x000000018c1ce85c in CoreReadStreamClient::coreStreamEventsAvailable(unsigned long) ()
#11 0x000000018c2aca9c in CoreStreamBase::_callClientNow(CoreStreamClient*) ()
#12 0x000000018c1ce5d8 in CoreStreamBase::_streamSetEventAndScheduleDelivery(unsigned long, unsigned char) ()
#13 0x000000018c1ce370 in SocketStream::dispatchSignalFromSocketCallbackUnlocked(SocketStreamSignalHolder*) ()
#14 0x000000018c1cdbec in SocketStream::socketCallback(__CFSocket*, unsigned long, __CFData const*, void const*) ()
#15 0x000000018c1cdadc in SocketStream::_SocketCallBack_stream(__CFSocket*, unsigned long, __CFData const*, void const*, void*) ()
#16 0x000000018c64a26c in __CFSocketPerformV0 ()
#17 0x000000018c64777c in __CFRUNLOOP_IS_CALLING_OUT_TO_A_SOURCE0_PERFORM_FUNCTION__ ()
#18 0x000000018c646ad8 in __CFRunLoopDoSources0 ()
#19 0x000000018c644d70 in __CFRunLoopRun ()
#20 0x000000018c585b78 in CFRunLoopRunSpecific ()
#21 0x0000000191fa7830 in GSEventRunModal ()
#22 0x000000018f5c305c in UIApplicationMain ()
#23 0x00000001000cd304 in main at /Users/markus/Projects/rubiTrack-to-Go/src/app/main.m:44
#24 0x0000000198b8baa0 in start ()
Thread 2, Queue : com.apple.libdispatch-manager
#0  0x0000000198c6dac8 in kevent64 ()
#1  0x0000000198b793d8 in _dispatch_mgr_invoke ()
#2  0x0000000198b71d88 in _dispatch_mgr_thread ()
Thread 5, Queue : (null)
#0  0x0000000198d057a4 in start_wqthread ()
Thread 6 com.apple.CFSocket.private, Queue : (null)
#0  0x0000000198c8676c in select$DARWIN_EXTSN ()
#1  0x000000018c64bc2c in __CFSocketManager ()
#2  0x0000000198d081b0 in _pthread_body ()
#3  0x0000000198d08108 in _pthread_start ()
#4  0x0000000198d057b0 in thread_start ()
Thread 7 com.apple.NSURLConnectionLoader, Queue : (null)
#0  0x0000000198c6dcc0 in mach_msg_trap ()
#1  0x0000000198c6db2c in mach_msg ()
#2  0x000000018c646cac in __CFRunLoopServiceMachPort ()
#3  0x000000018c644e3c in __CFRunLoopRun ()
#4  0x000000018c585b78 in CFRunLoopRunSpecific ()
#5  0x000000018d16ea1c in +[NSURLConnection(Loader) _resourceLoadLoop:] ()
#6  0x000000018d1fc990 in __NSThread__main__ ()
#7  0x0000000198d081b0 in _pthread_body ()
#8  0x0000000198d08108 in _pthread_start ()
Thread 8, Queue : UIDocument File Access
#0  0x0000000198c6dcfc in semaphore_wait_trap ()
#1  0x0000000198b77c08 in _dispatch_semaphore_wait_slow ()
#2  0x000000018d299674 in -[NSFileCoordinator(NSPrivate) _blockOnAccessClaim:] ()
#3  0x000000018d29a1c0 in -[NSFileCoordinator(NSPrivate) _coordinateWritingItemAtURL:options:error:byAccessor:] ()
#4  0x000000018d2984ec in -[NSFileCoordinator coordinateWritingItemAtURL:options:error:byAccessor:] ()
#5  0x000000018fa7f90c in -[UIDocument _coordinateWritingItemAtURL:error:byAccessor:] ()
#6  0x000000018fa801a0 in __59-[UIDocument saveToURL:forSaveOperation:completionHandler:]_block_invoke ()
#7  0x0000000198b7043c in _dispatch_call_block_and_release ()
#8  0x0000000198b703fc in _dispatch_client_callout ()
#9  0x0000000198b766f0 in _dispatch_queue_drain ()
#10 0x0000000198b72890 in _dispatch_queue_invoke ()
#11 0x0000000198b773a4 in _dispatch_root_queue_drain ()
#12 0x0000000198b77634 in _dispatch_worker_thread2 ()
#13 0x0000000198d05918 in _pthread_wqthread ()
Thread 9, Queue : (null)
#0  0x0000000198c86e74 in __workq_kernreturn ()
#1  0x0000000198d0592c in _pthread_wqthread ()
Thread 11, Queue : com.apple.root.default-priority
#0  0x0000000198c6dcfc in semaphore_wait_trap ()
#1  0x0000000198b77c08 in _dispatch_semaphore_wait_slow ()
#2  0x0000000198d24700 in xpc_connection_send_message_with_reply_sync ()
#3  0x000000018f52845c in _reach_server_target_remove ()
#4  0x000000018f5283a0 in __SCNetworkReachabilityServer_targetRemove ()
#5  0x000000018f506c18 in __SCNetworkReachabilityDeallocate ()
#6  0x000000018c581ee4 in CFRelease ()
#7  0x0000000198b703fc in _dispatch_client_callout ()
#8  0x0000000198b773f8 in _dispatch_root_queue_drain ()
#9  0x0000000198b77634 in _dispatch_worker_thread2 ()
#10 0x0000000198d05918 in _pthread_wqthread ()
#11 0x0000000198d057a8 in start_wqthread ()
Thread 12, Queue : (null)
#0  0x0000000198c86e74 in __workq_kernreturn ()

[GoannaGuy]

I just ran into this myself today on 64bit OSX. It manifested for me when I had a server running with a client that connected/disconnected/reconnected rapidly.

It appears to be an ARC related issue. I have a fix for it on my local machine. I'll create a fork and issue a pull request to the maintainer.

[ghmrs356]

That explains why moving my local version from ARC back to manual reference counting fixed the issue.

[GoannaGuy]

Pull request can be found in issue #11

[snej]

You closed that pull request saying that it didn't fix the issue after all. So this is still unexplained?

[GoannaGuy]

Jens,

Actually, I opened a subsequent pull request later yesterday that does appear to fix it completely, which contains the fix from the initial pull request, plus a couple more changes.

-=- John

John Turnipseed johnt@me.com

On Feb 21, 2014, at 11:16:53 AM, Jens Alfke notifications@github.com wrote:

You closed that pull request saying that it didn't fix the issue after all. So this is still unexplained?

— Reply to this email directly or view it on GitHub.