Bug 69676

Created attachment 86317 [details]
alsa-info

Created attachment 86318 [details]
PA debug log

Created attachment 86319 [details]
backtrace

the emulated hda codec does support Jack detection

DMI Information
!!---------------

Manufacturer:      Bochs
Product Name:      Bochs
Product Version:   
Firmware Version:  Bochs

Codec: Generic 1af4 ID 22
Address: 0
AFG Function Id: 0x1 (unsol 0)
Vendor Id: 0x1af40022
Subsystem Id: 0x1af40022
Revision Id: 0x100101




Node 0x03 [Pin Complex] wcaps 0x400101: Stereo
  Control: name="Line Out Phantom Jack", index=0, device=0
  Pincap 0x00000010: OUT
  Pin Default 0x00004010: [Jack] Line Out at Ext N/A
    Conn = Unknown, Color = Green
    DefAssociation = 0x1, Sequence = 0x0
  Pin-ctls: 0x40: OUT
  Connection: 1
     0x02


	control.5 {
		iface CARD
		name 'Line Phantom Jack'
		value true
		comment {
			access read
			type BOOLEAN
			count 1
		}
	}
	control.6 {
		iface CARD
		name 'Line Out Phantom Jack'
		value true
		comment {
			access read
			type BOOLEAN
			count 1
		}
	}

Hi Kalev and thanks for testing, and the detailed bug report!

First, can you verify if it is jack->path or jack->path->port that fails, e g like this:
http://lists.freedesktop.org/archives/pulseaudio-discuss/2013-September/018730.html

Assuming it's the latter (port) that fails, this port should have been added in add_profiles -> pa_alsa_path_set_add_ports -> device_port_alsa_init. I'm not exactly understanding why this does not happen.

Could you add some more instrumentation to this path, like in the attachment, and make a new PA debug log? Thanks!

Created attachment 86341 [details] [review]
more debug info

(In reply to comment #5)
> Hi Kalev and thanks for testing, and the detailed bug report!

Thanks for looking at this, much appreciated!


> First, can you verify if it is jack->path or jack->path->port that fails, e
> g like this:
> http://lists.freedesktop.org/archives/pulseaudio-discuss/2013-September/
> 018730.html

It's the port, yes -- sorry, I should have mentioned that I printed out some of the struct members at the end of the attached backtrace log.


> Assuming it's the latter (port) that fails, this port should have been added
> in add_profiles -> pa_alsa_path_set_add_ports -> device_port_alsa_init. I'm
> not exactly understanding why this does not happen.
> 
> Could you add some more instrumentation to this path, like in the
> attachment, and make a new PA debug log? Thanks!

Done. I added a little bit more instrumentation there, attaching the diff as well for reference.

Created attachment 86355 [details] [review]
more debug info

Created attachment 86356 [details]
PA debug log 2

Thanks. I now saw the end of the backtrace, giving us a hint:

(gdb) p jack->path->name
$5 = 0x6f4740 "analog-output"

Analog output is removed (because analog-output-lineout is there), but it seems its jacks are somehow lingering around anyway.

Giving this a thought, I think the best thing to do is to remove that path a little more at the end of function "pa_alsa_profile_set_probe" where we also drop unsupported paths. This path is supported, but it's superfluous so it's dropped from the path set but not from the profile set.

Created attachment 86358 [details]
Drop unused paths

Could you check if this patch resolves your issue without causing any other issues? Thanks!

(In reply to comment #11)
> Created attachment 86358 [details]
> Drop unused paths

I can confirm this fixes the crash and pulseaudio seems to work fine otherwise.

Thanks David!

The patch has been applied already a long time ago, marking bug as fixed.