At Thu, 07 Jun 2012 10:00:24 +0100, Steven Newbury wrote:

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

On 07/06/12 09:47, Takashi Iwai wrote:

...

At the next time you submit a patch, please add Cc to maintainer and subsystem ML. Otherwise it'd be easily overlooked.

I meant to, somehow I managed to send it off before adding the CC, probably too early in the morning!

...

At Thu, 07 Jun 2012 06:52:59 +0100, Steven Newbury wrote:

...

When VGA_SWITCHEROO support is enabled hda_intel initialises the HDMI audio device on the current VGA device. When it's not enabled it only initialises the HDMI device on the default VGA adaptor, this means secondary cards get no audio support which is very unhelpful for multi-seat!

With this patch, when SUPPORT_VGA_SWITCHEROO is disabled hda_intel initialises all HDMI audio devices, not just the default VGA.

Signed-off-by: Steven Newbury steve@snewbury.org.uk

OK, it must be fixed indeed. The bug was introduced due to the change in VGA-switcheroo side -- originally check_hdmi_disabled() really checked the VGA-switcheroo state, so it returned the correct value in the earlier version.

In anyway, your patch looks almost OK, but it'd be better to move ifdef and simplify like below. Could you check whether it works for you?

Looks better, saves the pointless function call. I'll give it spin on my next boot.

Thanks!

Meanwhile I found that the current version is also buggy for some setups with VGA-switcheroo. The current version won't work, for example, when VGA-switcheroo is built-in for systems without switcheroo. I'll need to cook up the fix for 3.5-rc2 in addition to your fix.

Takashi