module-hal-detect detects first device only

[cmt]

On recent FreeBSDs (7.2-RC) multi-output/codec chips like HDA capable devices show up as multiple /dev/dsp* devices and likewise as multiple devices in lshal (see attachment). module-hal-detect in pulseaudio 0.9.14 (FreeBSD ports) and 0.9.15 use the first device (index 0) very deliberately, judging from the comment in 0.9.15. On systems like mine where the main output is driven via a device different than dsp0 (dsp1 here) the "right" output device never shows up as a sink. Changing a single line of code in module-hal-detect remedies this, but obviously there was some reason in not using any device beyond index 0, so there must be a better solution than the attaced patch.

[cmt]

output of lshal stripped to the relevant devices

[cmt]

patch to module-hal-detect.c which fixes the issue on my system

[lennart]

I doubt it makes sense to merge this patch: do we really want to initialize all subdevices of a device unconditioanlly? We certainly don't want that on ALSA, so doubt we want that on OSS either.

[cmt]

Perhaps it makes sense to add an option for initializing all subdevices or even specifying which subdevice to use. I have no idea how much flexibility and how much semi-intelligent heuristic is needed here, as I'm using pulseaudio in the simplest way possible...

[lennart]

Sounds reasonable. Would be happy to take a patch for that.

[cmt]

adds option "subdevs" when compiled HAVE_OSS

[cmt]

patch-src_modules_module-hal-detect.2.c (against 0.9.15) should do the job (tested only 0.9.14 backport, as I'm having libtool fun with 0.9.15 but no time to investigate...) Gruß aus Karlsruhe.

[lennart]

Fixed in r0113e72.

Thanks for the patch!