May 2007 - Bluetooth-dev - lists.apple.com

Crash on 10.4 after canceling pairing
by James Blair 29 May '07

29 May '07

site_archiver(a)lists.apple.com Delivered-To: bluetooth-dev(a)lists.apple.com My application is crashing on 10.4.9 (PPC and Intel) after I cancel a pairing operation that is in progress. I do not see similar crashes on 10.3.9. I'm calling the runModal method on an IOBluetoothPairingController object to programmatically display a device pairing dialog. After I enter a passkey and click "Pair", my device prompts me to enter the passkey. At this point, if I cancel the pairing from my computer, my device eventually times out on the pairing request and sends a timeout message back to my computer. This message is what causes the crash. Here is the relevant part of the crash report: Exception: EXC_BAD_ACCESS (0x0001) Codes: KERN_INVALID_ADDRESS (0x0001) at 0x3133009b Thread 0 Crashed: 0 libobjc.A.dylib 0x90a410f8 objc_msgSend + 24 1 com.apple.Bluetooth 0x94ffa9dc -[IOBluetoothDevicePairing sendNotificationToTarget:param:] + 80 2 com.apple.Bluetooth 0x94ffa920 -[IOBluetoothDevicePairing connectionComplete:status:] + 268 3 com.apple.Bluetooth 0x94fcb080 -[IOBluetoothDevice(IOBluetoothDevicePrivate) callConnectionCompleteCallback:status:] + 92 4 com.apple.Bluetooth 0x94fc5994 DeviceConfigAsyncHCINotificationHandler + 536 5 com.apple.Bluetooth 0x94fc56f8 -[IOBluetoothNotification invokeCallbackWithDataRunLoopScheduled:] + 208 6 com.apple.Foundation 0x92bebc04 __NSFireDelayedPerform + 304 7 com.apple.CoreFoundation 0x907f2578 __CFRunLoopDoTimer + 184 8 com.apple.CoreFoundation 0x907deef8 __CFRunLoopRun + 1680 9 com.apple.CoreFoundation 0x907de4ac CFRunLoopRunSpecific + 268 10 com.apple.HIToolbox 0x93298b20 RunCurrentEventLoopInMode + 264 11 com.apple.HIToolbox 0x9337d48c GetNextEventMatchingMask + 400 12 com.apple.HIToolbox 0x9337d238 WNEInternal + 152 13 com.apple.HIToolbox 0x9337d18c WaitNextEvent + 76 This appears to be similar, but not identical, to another problem previously reported here: http://lists.apple.com/archives/bluetooth-dev/2006/Apr/msg00010.html Can anyone offer insight on what I might be doing wrong? Or, is this possibly a bug in Apple's Bluetooth support? Thanks, James Blair Staff Software Engineer National Instruments, http://ni.com/ _______________________________________________ Do not post admin requests to the list. They will be ignored. Bluetooth-dev mailing list (Bluetooth-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/bluetooth-dev/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Sony Sisxaxis device, pairing without pairing, and bluetooth HID connection timeouts
by Blueberry 27 May '07

27 May '07

site_archiver(a)lists.apple.com Delivered-To: bluetooth-dev(a)lists.apple.com --- Jenna Fox Way prettier than that Colin Munro guy _______________________________________________ Do not post admin requests to the list. They will be ignored. Bluetooth-dev mailing list (Bluetooth-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/bluetooth-dev/site_archiver%40lists.… Hi, I am writing a set of user space tools to make use of Sony's Sixaxis controller's from Tiger. This device has a few quirks. It doesn't accept incoming connections, it doesn't send valid hid reports until issued a hid feature request asking it to change to its "operational" mode, and it doesn't support the normal bluetooth pairing procedure. Pairing is done like this: The Sixaxis is connected to host via USB cord. Host detects Sixaxis device and sends a get_report feature type request over USB with a report id of 0xF2, sixaxis returns data we do not know the meaning of but is never the less in operational mode now and can be used as a USB HID device after pushing the PS button on the controller. Host then issues a set_report feature type request, with a report id of 0xF5, and a buffer containing the following bytes: 0x01, 0x00, and then six bytes representing the BD_ADDR of the host computer/playstation. This is all implemented and working, the trouble is when trying to turn on the device. The sixaxis will connect to the mac host and seems to go through all the procedures for connecting and initializing. Then one minute and two seconds (on my macbook) after the sixaxis first started the connection, the mac will tell it to disconnect, it obeys, the bluetooth chip in the mac change to idle, and the lights on the Sixaxis go dark, apparently completely shut off. Note that the initial connection takes about two seconds, so it appears to disconnect after precisely one minute of inactivity. This timeout is problematic because we want the device to stay connected so applications and games will be able to connect with it and receive reports. How would this be avoided? If a user space application were to open a hid interface to the sixaxis might that be enough to keep it from being disconnected by inactivity? Would that tool also need to actively request reports every 50 seconds or so to keep it from timing out or merely have a reference to the device in its memory? Also, is it at all an issue that the mac seems to not consider the sixaxis device paired, as its never gone through a traditional passkey based pairing procedure. Would it be worthwhile to try to mess with blued's configuration plist's to convince it that the device is paired, and then kill it to get the daemon to restart and reread those files? I would rather avoid this if its not likely to bring any benefits as requiring a root password and messing with things like that will make me and surely my users squeamish. This email sent to site_archiver(a)lists.apple.com

1 0

Bluetooth and HID devices
by lists＠mice-software.com 27 May '07

27 May '07

site_archiver(a)lists.apple.com Delivered-To: bluetooth-dev(a)lists.apple.com User-agent: Internet Messaging Program (IMP) H3 (4.1.3) Hi, I'm working on a driver for OSX for the Sony PS3 SIXAXIS controller. Thanks, Colin Munro _______________________________________________ Do not post admin requests to the list. They will be ignored. Bluetooth-dev mailing list (Bluetooth-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/bluetooth-dev/site_archiver%40lists.… Currently I have it working alright on USB mode - this is a standard HID device, but requires a getreport to be sent to kick it into life. I do this in the kernel due to the elegance of setting it up at startup of the device, but it also allows me to change the HID descriptor to include the pressure button readings and motion sensors, which are in the data that comes in as a HID report, but not included in Sony's HID descriptor. In addition to this, the device has a Bluetooth mode. The pairing is done over USB, which I have successfully added to my Preference Pane using the standard setReport function from userspace. However, when the controller then connects, it doesn't seem to be seen as a HID device. My question is, if I were to manually connect to the device to accept it's data, is this only possible from userspace? Ideally I'd do it in the kernel so as to expose a new HID device - otherwise, I assume I'm going to need to write a userspace daemon to read the controller, then some sort of virtual kernel driver to expose the relevant HID device. This email sent to site_archiver(a)lists.apple.com

1 0

Re: RFCOMM + SCO connection
by Thomas Engelmeier 22 May '07

22 May '07

site_archiver(a)lists.apple.com Delivered-To: bluetooth-dev(a)lists.apple.com Am 21.05.2007 um 21:03 schrieb Mihnea Galca: Use an (paid) DTS incident if you want an definitive answer... (see developer.apple.com) _______________________________________________ Do not post admin requests to the list. They will be ignored. Bluetooth-dev mailing list (Bluetooth-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/bluetooth-dev/site_archiver%40lists.… Any chance that we will get an answer for this issue ? This email sent to site_archiver(a)lists.apple.com

1 0

Re: RFCOMM + SCO connection
by Mihnea Galca 21 May '07

21 May '07

site_archiver(a)lists.apple.com Delivered-To: bluetooth-dev(a)lists.apple.com Importance: Normal User-agent: SquirrelMail/1.4.7-4.fc4 Any chance that we will get an answer for this issue ? Regards, Mihnea Galca > Hello, > > Is it possible to have a rfcomm connection with a bluetooth headset > on top of which to start/stop SCO connection ? > > Usage scenario: a RFCOMM connection has to be maintained to a bluetooth > headset to get button press events, but in the same time, SCO is > activated to play/record sounds using the headset. > > So far, I could make a RFCOMM connection to capture button press events, > but I couldn't start SCO in any way. SCO seems connected to a system audio > device which connects/disconnects the headset when needed. I couldn't find > any method to control SCO from API. > > How can I solve the problem in the above given scenario ? > > Do I need to develop my own SCO layer over RFCOMM and create a new audio > device driver ? Any pointers to do such a thing ? I've tried > IOBluetoothAddSCOAudioDevice, but it's a deprecated function > and the only thing it does, is to create the associated audio device, if > such a device is not already present. My code is the following: > > void* keys[1]; > > keys[0]=[[NSString alloc] initWithCString:"IOAudioControls"]; > > void* values[1]; > > id array=[[NSMutableArray alloc] init]; > > values[0]=array; > > CFDictionaryRef dictionary=CFDictionaryCreate(NULL,keys,values,1, > &kCFTypeDictionaryKeyCallBacks,&kCFTypeDictionaryValueCallBacks); > > IOReturn res=IOBluetoothAddSCOAudioDevice(m_headset,dictionary); > > The operation was successful in the case when the audio device didn't > exist. In the case when the device existed (created using bluetooth > interface in the OS), the function returned a general error. > > Regards, > Mihnea GALCA > _______________________________________________ > Do not post admin requests to the list. They will be ignored. > Bluetooth-dev mailing list (Bluetooth-dev(a)lists.apple.com) > Help/Unsubscribe/Update your Subscription: > http://lists.apple.com/mailman/options/bluetooth-dev/mgalca%40tremend.ro > > This email sent to mgalca(a)tremend.ro > _______________________________________________ Do not post admin requests to the list. They will be ignored. Bluetooth-dev mailing list (Bluetooth-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/bluetooth-dev/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Re: WriteSync / WriteAsync
by Joseph Kelly 10 May '07

10 May '07

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

1 0

Re: WriteSync / WriteAsync
by Joseph Kelly 09 May '07

09 May '07

site_archiver(a)lists.apple.com Delivered-To: bluetooth-dev(a)lists.apple.com On May 9, 2007, at 2:09 PM, Moritz Gmelin wrote: Joe K. _______________________________________________ Do not post admin requests to the list. They will be ignored. Bluetooth-dev mailing list (Bluetooth-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/bluetooth-dev/site_archiver%40lists.… 2. Is Write Async as safe as write sync ? Event when not implementing the callback methods ? If you don't implement the callback, you won't get write error notifications. You'll want those. If you find you have no workaround, then you can "fake" a synchronous operation by issuing the async command, then adding a timer to the current run loop, then run the run loop in a custom mode (i.e. a mode you don't use anywhere else in your program). From your async complete method, you remove the timer from the runloop, and voila. This gives you the added benefit that you can implement a time out using the timer. This email sent to site_archiver(a)lists.apple.com

1 0

WriteSync / WriteAsync
by Moritz Gmelin 09 May '07

09 May '07

site_archiver(a)lists.apple.com Delivered-To: bluetooth-dev(a)lists.apple.com Hi, Regards Moritz _______________________________________________ Do not post admin requests to the list. They will be ignored. Bluetooth-dev mailing list (Bluetooth-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/bluetooth-dev/site_archiver%40lists.… So far, I was using the WriteSync method in my application to send data over an RFComm channel. I just wanted to make sure the stuff was really sent before I moved on. Now I just discovered, that when sending data through an external BT-Dongle (instead of the integrated BT-Dongle). The WriteSync Method was blocking for quite some time. Where the WriteAsync method was not. Using a Blocksize of 400 Bytes I could transfer 16kb/sec (using obex) with writeSync through the internal BT hardware and about 1.6kb/sec wih an external (class 1 BT2.0+EDR) dongle. Using WriteAsync, internal and external both went at 16-20kb/sec. The remote and was always a Nokia e61 phone. 1. Why Is WriteSync so much slower on an external dongle ? 2. Is Write Async as safe as write sync ? Event when not implementing the callback methods ? This email sent to site_archiver(a)lists.apple.com

1 0

Issue with native Mac OS 10.4.9 BlueTooth drivers
by Uwe Schuss 07 May '07

07 May '07

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

1 0