Return-Path: Date: Mon, 20 Apr 2009 00:38:37 +0300 From: Johan Hedberg To: Stuart Pook Cc: BlueZ development Subject: Re: bluez git + Linksys USBBT100 + 2.6.30-rc2 -> Segmentation fault Message-ID: <20090419213837.GA6514@jh-x301> References: <49D89DCD.7090808@pook.es> <49D8E48A.2060807@pook.es> <20090405172212.GB6612@jh-x301> <49EB7950.4000802@pook.es> <20090419200528.GA18068@jh-x301> <49EB93C2.3090702@pook.es> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed In-Reply-To: <49EB93C2.3090702@pook.es> Sender: linux-bluetooth-owner@vger.kernel.org List-ID: Hi Stuart, On Sun, Apr 19, 2009, Stuart Pook wrote: > I did get one error that I had never seen before "Refusing headset": That comes if a headset is trying to connect to you and there's already max_connected_headsets headsets connected (this variable defaults to 1 and can be overrided with the MaxConnected parameter in audio.conf. > bluetoothd[32401]: Audio API: BT_REQUEST <- BT_START_STREAM > bluetoothd[32401]: State changed /org/bluez/32401/hci0/dev_00_1A_45_2F_49_98: HEADSET_STATE_DISCONNECTED -> HEADSET_STATE_CONNECT_IN_PROGRESS > bluetoothd[32401]: Unable to get service record: Connection timed out (110) > bluetoothd[32401]: Unable to get a SCO fd > bluetoothd[32401]: headset_resume_complete: resume failed > bluetoothd[32401]: Audio API: BT_ERROR -> BT_START_STREAM > bluetoothd[32401]: telephony-dummy: device 0xb8f2c550 disconnected > bluetoothd[32401]: State changed /org/bluez/32401/hci0/dev_00_1A_45_2F_49_98: HEADSET_STATE_CONNECT_IN_PROGRESS -> HEADSET_STATE_DISCONNECTED > bluetoothd[32401]: Unix client disconnected (fd=13) > bluetoothd[32401]: Unable to get service record: Connection timed out (110) > Segmentation fault Could you please run bluetoothd through valgrind to get a proper backtrace. This time it should hopefully be more useful than the one we got from the infinite recursion case (with some luck a fix should be just a few minutes away from getting the log). Once we have the valgrind trace I'd also be interested in seeing the output of hcidump for the "Connection timed out" situation. > In fact I get the "Too short (1 bytes) IPC packet from bluetoothd" error > even from bluetoothd 4.36. If this isn't caused by a bug the only possible reason I can think of as a cause for it is a mismatch between the alsa plugin and bluetoothd versions. Btw, you might want to consider trying out pulseaudio 0.9.15. It's worked pretty flawlessly for me with bluez 4.34 and later versions (though I haven't actually tested the mono profiles very much with it). Johan