February 2010 - Darwin-dev - lists.apple.com

RE: close on USB serial port hangs (read pending) on Snow Leopard
by Matthew Owings 18 Feb '10

18 Feb '10

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thread-index: Acqw61Js285xUhjQRR6bhcYUyjOwQQAB7tlQ Thread-topic: close on USB serial port hangs (read pending) on Snow Leopard We do something similar in our app, except we do not put our serial port reading on a different thread, although we would like to at some point. We have a timer checking to see if there are any bytes to read that fires 10 times a second. If there is nothing to read it returns immediately. In your thread, you could have the thread sleep for a bit if there is no data. -----Original Message----- From: Fritz Anderson [mailto:fritza@manoverboard.org] Sent: Thursday, February 18, 2010 2:40 PM To: Matthew Owings Cc: darwin-dev(a)lists.apple.com Subject: Re: close on USB serial port hangs (read pending) on Snow Leopard On 18 Feb 2010, at 3:36 PM, Matthew Owings wrote: > Might I suggest not doing a blocking read. Then when you need to close > the port, set a flag that tells the thread it needs to end. Then when > the thread has terminated, close the port. I'm perplexed. The nature of reading from a serial device is that the data is intermittent, and in single bytes. If I did a non-blocking read, I'd have to poll, would I not? And get nothing back from virtually all the calls? Wait. Should I be using select()? I admit to being very naive. - F _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: close on USB serial port hangs (read pending) on Snow Leopard
by Fritz Anderson 18 Feb '10

18 Feb '10

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On 18 Feb 2010, at 3:36 PM, Matthew Owings wrote: > Might I suggest not doing a blocking read. Then when you need to close > the port, set a flag that tells the thread it needs to end. Then when > the thread has terminated, close the port. I'm perplexed. The nature of reading from a serial device is that the data is intermittent, and in single bytes. If I did a non-blocking read, I'd have to poll, would I not? And get nothing back from virtually all the calls? Wait. Should I be using select()? I admit to being very naive. — F _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

RE: close on USB serial port hangs (read pending) on Snow Leopard
by Matthew Owings 18 Feb '10

18 Feb '10

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thread-index: Acqw1aX8L3QVgbutSl29sUWea9a53gADDW5A Thread-topic: close on USB serial port hangs (read pending) on Snow Leopard Might I suggest not doing a blocking read. Then when you need to close the port, set a flag that tells the thread it needs to end. Then when the thread has terminated, close the port. -----Original Message----- From: darwin-dev-bounces+matt=qwizdom.com(a)lists.apple.com [mailto:darwin-dev-bounces+matt=qwizdom.com@lists.apple.com] On Behalf Of darwin-dev-request(a)lists.apple.com Sent: Thursday, February 18, 2010 12:05 PM To: darwin-dev(a)lists.apple.com Subject: Darwin-dev Digest, Vol 7, Issue 25 Send Darwin-dev mailing list submissions to darwin-dev(a)lists.apple.com To subscribe or unsubscribe via the World Wide Web, visit http://lists.apple.com/mailman/listinfo/darwin-dev or, via email, send a message with subject or body 'help' to darwin-dev-request(a)lists.apple.com You can reach the person managing the list at darwin-dev-owner(a)lists.apple.com When replying, please edit your Subject line so it is more specific than "Re: Contents of Darwin-dev digest..." Today's Topics: 1. close on USB serial port hangs (read pending) on Snow Leopard (Fritz Anderson) ---------------------------------------------------------------------- Message: 1 Date: Thu, 18 Feb 2010 13:31:49 -0600 From: Fritz Anderson <fritza(a)manoverboard.org> Subject: close on USB serial port hangs (read pending) on Snow Leopard To: "darwin-dev(a)lists.apple.com list" <darwin-dev(a)lists.apple.com> Message-ID: <F2CD52CE-1D55-4260-AE9B-968FF7ABA22E(a)manoverboard.org> Content-Type: text/plain; charset=windows-1252 I have an application (compiled against the 10.4 Universal SDK, GCC 4.0) that hangs upon close() on a serial port when running on 10.6, but not on earlier OSes. There is a blocking read() pending on the file descriptor in another thread. The device is an FTDI USB-to-serial port discovered through the IO registry as kIOSerialBSDRS232Type and opened on its callout /dev entry. I create a thread (using the Cocoa NSThread API) on which I make a blocking read(). The close() is called on the main thread. The close() never returns, and the Xcode debugger can't break into the application. A sample of the hung application is at the end of this message. The NOCANCEL part of the read() implementation is suggestive to me, but I don't know what to do about it. Opening is done as: inputHandle = open([inputPath fileSystemRepresentation], O_RDWR | O_NOCTTY | O_NONBLOCK); There is no relevant entry in the "All Messages" mode of Console.app. This worked on 10.3 (no longer supported), 10.4, and 10.5. The hang occurs on 10.6. The close() is done upon application shutdown, so I suppose I could just skip it, but that's unclean, and inhibits reuse of my code. I've Googled various combinations of FTDI, read, close, read$NOCANCEL, and hang, and if there was any help, it was buried in the noise. I dug up the Kernel (etc.) release notes for Snow Leopard, and didn't find anything related to BSD I/O. Asking the doc window for "read release notes," "BSD release notes," etc., yielded nothing obvious. I'm sure I'm doing something wrong that was caught with the revised system, but I don't know where to begin. Will someone please point me the right way, either technique or document? - F ==== Sampling process 19076 for 3 seconds with 1 millisecond of run time between samples Sampling completed, processing symbols... Analysis of sampling AuthPayX (First Data) Demo (pid 19076) every 1 millisecond Call graph: 2618 Thread_804995 DispatchQueue_1: com.apple.main-thread (serial) 2618 start 2618 _start 2618 main 2618 NSApplicationMain 2618 -[NSApplication run] 2618 -[NSApplication nextEventMatchingMask:untilDate:inMode:dequeue:] 2618 _DPSNextEvent 2618 _NSHandleCarbonMenuEvent 2618 _HandleMenuSelection 2618 _HandleMenuSelection2 2618 MenuSelectCore(MenuData*, Point, double, unsigned long, OpaqueMenuRef**, unsigned short*) 2618 FinishMenuSelection(SelectionData*, MenuResult*, MenuResult*) 2618 SendMenuItemSelectedEvent 2618 SendMenuCommandWithContextAndModifiers 2618 SendHICommandEvent(unsigned long, HICommand const*, unsigned long, unsigned long, unsigned char, void const*, OpaqueEventTargetRef*, OpaqueEventTargetRef*, OpaqueEventRef**) 2618 SendEventToEventTarget 2618 SendEventToEventTargetInternal(OpaqueEventRef*, OpaqueEventTargetRef*, HandlerCallRec*) 2618 DispatchEventToHandlers(EventTargetRec*, OpaqueEventRef*, HandlerCallRec*) 2618 NSSLMMenuEventHandler 2618 -[NSCarbonMenuImpl _carbonCommandProcessEvent:handlerCallRef:] 2618 -[NSMenuItem _internalPerformActionThroughMenuIfPossible] 2618 -[NSMenu _internalPerformActionForItemAtIndex:] 2618 -[NSMenu performActionForItemAtIndex:] 2618 -[NSCarbonMenuImpl performActionWithHighlightingForItemAtIndex:] 2618 -[NSMenuItem _corePerformAction] 2618 -[NSApplication sendAction:to:from:] 2618 -[NSApplication terminate:] 2618 -[NSNotificationCenter postNotificationName:object:] 2618 -[NSNotificationCenter postNotificationName:object:userInfo:] 2618 _CFXNotificationPostNotification 2618 __CFXNotificationPost 2618 _nsnote_callback 2618 -[AppController applicationWillTerminate:] 2618 -[I3010Terminal dealloc] 2618 -[SerialPort close] 2618 close 2618 Thread_805034 DispatchQueue_2: com.apple.libdispatch-manager (serial) 2618 start_wqthread 2618 _pthread_wqthread 2618 _dispatch_worker_thread2 2618 _dispatch_queue_invoke 2618 _dispatch_mgr_invoke 2618 kevent 2618 Thread_805059 2618 thread_start 2618 _pthread_start 2618 __CFSocketManager 2618 select$DARWIN_EXTSN 2618 Thread_805283 2618 thread_start 2618 _pthread_start 2618 __NSThread__main__ 2618 -[NSThread main] 2618 -[AbstractPort readLoop:] 2618 -[SerialPort blockingRead:size:] 2618 read$NOCANCEL$UNIX2003 Total number in stack (recursive counted multiple, when >=5): Sort by top of stack, same collapsed (when >= 5): close 2618 kevent 2618 read$NOCANCEL$UNIX2003 2618 select$DARWIN_EXTSN 2618 Sample analysis of process 19076 written to file /dev/stdout ------------------------------ _______________________________________________ Darwin-dev mailing list Darwin-dev(a)lists.apple.com http://lists.apple.com/mailman/listinfo/darwin-dev End of Darwin-dev Digest, Vol 7, Issue 25 ***************************************** _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

close on USB serial port hangs (read pending) on Snow Leopard
by Fritz Anderson 18 Feb '10

18 Feb '10

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com I have an application (compiled against the 10.4 Universal SDK, GCC 4.0) that hangs upon close() on a serial port when running on 10.6, but not on earlier OSes. There is a blocking read() pending on the file descriptor in another thread. The device is an FTDI USB-to-serial port discovered through the IO registry as kIOSerialBSDRS232Type and opened on its callout /dev entry. I create a thread (using the Cocoa NSThread API) on which I make a blocking read(). The close() is called on the main thread. The close() never returns, and the Xcode debugger can't break into the application. A sample of the hung application is at the end of this message. The NOCANCEL part of the read() implementation is suggestive to me, but I don't know what to do about it. Opening is done as: inputHandle = open([inputPath fileSystemRepresentation], O_RDWR | O_NOCTTY | O_NONBLOCK); There is no relevant entry in the "All Messages" mode of Console.app. This worked on 10.3 (no longer supported), 10.4, and 10.5. The hang occurs on 10.6. The close() is done upon application shutdown, so I suppose I could just skip it, but that's unclean, and inhibits reuse of my code. I've Googled various combinations of FTDI, read, close, read$NOCANCEL, and hang, and if there was any help, it was buried in the noise. I dug up the Kernel (etc.) release notes for Snow Leopard, and didn't find anything related to BSD I/O. Asking the doc window for "read release notes," "BSD release notes," etc., yielded nothing obvious. I'm sure I'm doing something wrong that was caught with the revised system, but I don't know where to begin. Will someone please point me the right way, either technique or document? — F ==== Sampling process 19076 for 3 seconds with 1 millisecond of run time between samples Sampling completed, processing symbols... Analysis of sampling AuthPayX (First Data) Demo (pid 19076) every 1 millisecond Call graph: 2618 Thread_804995 DispatchQueue_1: com.apple.main-thread (serial) 2618 start 2618 _start 2618 main 2618 NSApplicationMain 2618 -[NSApplication run] 2618 -[NSApplication nextEventMatchingMask:untilDate:inMode:dequeue:] 2618 _DPSNextEvent 2618 _NSHandleCarbonMenuEvent 2618 _HandleMenuSelection 2618 _HandleMenuSelection2 2618 MenuSelectCore(MenuData*, Point, double, unsigned long, OpaqueMenuRef**, unsigned short*) 2618 FinishMenuSelection(SelectionData*, MenuResult*, MenuResult*) 2618 SendMenuItemSelectedEvent 2618 SendMenuCommandWithContextAndModifiers 2618 SendHICommandEvent(unsigned long, HICommand const*, unsigned long, unsigned long, unsigned char, void const*, OpaqueEventTargetRef*, OpaqueEventTargetRef*, OpaqueEventRef**) 2618 SendEventToEventTarget 2618 SendEventToEventTargetInternal(OpaqueEventRef*, OpaqueEventTargetRef*, HandlerCallRec*) 2618 DispatchEventToHandlers(EventTargetRec*, OpaqueEventRef*, HandlerCallRec*) 2618 NSSLMMenuEventHandler 2618 -[NSCarbonMenuImpl _carbonCommandProcessEvent:handlerCallRef:] 2618 -[NSMenuItem _internalPerformActionThroughMenuIfPossible] 2618 -[NSMenu _internalPerformActionForItemAtIndex:] 2618 -[NSMenu performActionForItemAtIndex:] 2618 -[NSCarbonMenuImpl performActionWithHighlightingForItemAtIndex:] 2618 -[NSMenuItem _corePerformAction] 2618 -[NSApplication sendAction:to:from:] 2618 -[NSApplication terminate:] 2618 -[NSNotificationCenter postNotificationName:object:] 2618 -[NSNotificationCenter postNotificationName:object:userInfo:] 2618 _CFXNotificationPostNotification 2618 __CFXNotificationPost 2618 _nsnote_callback 2618 -[AppController applicationWillTerminate:] 2618 -[I3010Terminal dealloc] 2618 -[SerialPort close] 2618 close 2618 Thread_805034 DispatchQueue_2: com.apple.libdispatch-manager (serial) 2618 start_wqthread 2618 _pthread_wqthread 2618 _dispatch_worker_thread2 2618 _dispatch_queue_invoke 2618 _dispatch_mgr_invoke 2618 kevent 2618 Thread_805059 2618 thread_start 2618 _pthread_start 2618 __CFSocketManager 2618 select$DARWIN_EXTSN 2618 Thread_805283 2618 thread_start 2618 _pthread_start 2618 __NSThread__main__ 2618 -[NSThread main] 2618 -[AbstractPort readLoop:] 2618 -[SerialPort blockingRead:size:] 2618 read$NOCANCEL$UNIX2003 Total number in stack (recursive counted multiple, when >=5): Sort by top of stack, same collapsed (when >= 5): close 2618 kevent 2618 read$NOCANCEL$UNIX2003 2618 select$DARWIN_EXTSN 2618 Sample analysis of process 19076 written to file /dev/stdout _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: With two unix socket binds in a row, the second one hangs ...
by Terry Lambert 14 Feb '10

14 Feb '10

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Please repot it with code to reproduce it. -- Terry I hope this is the right group, perhaps someone will advise me if not. When I switch the order of the binds, it is always the second one that hangs. -- Jay Reynolds Freeman --------------------- Jay_Reynolds_Freeman(a)mac.com http://web.mac.com/jay_reynolds_freeman (personal web site) _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/tlambert%40apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… On Feb 14, 2010, at 2:59 AM, Jay Reynolds Freeman <jay_reynolds_freeman(a)mac.com > wrote: I have a situation in which I am doing two separate binds of completely different Unix sockets ( socket( AF_UNIX, SOCK_STREAM, 0) ), very close together in time. The first one works fine, but the second one hangs; that is, it does not return an error, rather it never returns at all. But when I put a one-second sleep, or an OSMemoryBarrier, between the two binds, all is well. The fact that OSMemoryBarrier fixes the problem (and be it noted that I am running on an 8-core Mac Pro, so real parallelism is possible), makes me suspect that I have a reportable issue, and I will indeed kludge up a test case and file a bug, but I thought I would query the list first, to see if anyone has anything particular to suggest by way of test, analysis, or precedent. I have nosed around on the web a bit, but can't find anything particularly relevant. This email sent to tlambert(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0

With two unix socket binds in a row, the second one hangs ...
by Jay Reynolds Freeman 14 Feb '10

14 Feb '10

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com I hope this is the right group, perhaps someone will advise me if not. I have a situation in which I am doing two separate binds of completely different Unix sockets ( socket( AF_UNIX, SOCK_STREAM, 0) ), very close together in time. The first one works fine, but the second one hangs; that is, it does not return an error, rather it never returns at all. When I switch the order of the binds, it is always the second one that hangs. But when I put a one-second sleep, or an OSMemoryBarrier, between the two binds, all is well. The fact that OSMemoryBarrier fixes the problem (and be it noted that I am running on an 8-core Mac Pro, so real parallelism is possible), makes me suspect that I have a reportable issue, and I will indeed kludge up a test case and file a bug, but I thought I would query the list first, to see if anyone has anything particular to suggest by way of test, analysis, or precedent. I have nosed around on the web a bit, but can't find anything particularly relevant. -- Jay Reynolds Freeman --------------------- Jay_Reynolds_Freeman(a)mac.com http://web.mac.com/jay_reynolds_freeman (personal web site) _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

KUNCExecute
by Cameron Jones 12 Feb '10

12 Feb '10

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thread-index: AcqsLBXuFBdJvEOWk0yWRnRLPuUgBQ== Thread-topic: KUNCExecute User-agent: Microsoft-Entourage/12.23.0.091001 Is there a replacement for the (deprecated?) kUNCExecute call? I need to launch a user-space daemon when my FireWire device is plugged in and I currently use kUNCExecute for that purpose. Thanks. -- Cameron W. Jones ** Please help reduce Internet Junk Email ** ** by keeping this email address confidential ** ** Thank You! ** _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: How to get current process executable
by Michael Ritter 12 Feb '10

12 Feb '10

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Importance: normal Sensitivity: Normal _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: 10.6.2 issues with F_NOCACHE?
by Marc Van Olmen 08 Feb '10

08 Feb '10

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: 10.6.2 issues with F_NOCACHE?
by Chris Sarcone 08 Feb '10

08 Feb '10

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0