Intel HDMI Kernel 3.2 regression [Solved]

[Antidieu]

My eMachines laptop has an HDMI port that I plug in to a Sony HDTV and it used to work as expected on kernel 3.0. Then some commit between 3.0 and 3.2 kernels made my laptop unable to output a video signal to any TV or monitor through HDMI port.

The internal panel displays the desktop but my TV (or any external monitor) connected to the laptop shows only a blank screen complaining about an "invalid signal" and it remains blank UNLESS any sound is produced that goes out through HDMI, either by playing music/movie files on a media player or making a system sound. But as soon as the audio signal stops, the external monitor goes blank again with no video at all.

I can confirm the bug affects any distro with a 3.2 kernel or higher, i.e. Mint, Ubuntu, Sabayon, Fedora. I've also been testing vanilla kernels 3.3, 3.4, 3.5 and even 3.6-rc5 to no avail as this bug occurs on kernel 3.2 onwards.

UPDATE: Finally a solution

Finally, I found a real solution to this regression. I made a bug report there

https://bugs.freedesktop.org/show_bug.cgi?id=55556

Takashi Iwai provided a workaround to this issue and a patch to fix it. The workaround can be found in comment 15 and the patch in comment 17.

[lxnay]

You may want to ask Greg KH to review that patch and have it landed in stable kernels releases.

[Antidieu]

You may want to ask Greg KH to review that patch and have it landed in stable kernels releases.

I will but I think he must've already been asked, thanks for the advice.

Anyway, could you guys at Sabayon apply this patch in some minor kernel update? I could do it myself but my poor computer will overheat when compiling a whole kernel only because of a couple of reversed arguments in some checking function calls.

I mean patch is so simple that changes can even be done by hand and you've already applied the Fusion patch set after all. It's not a big deal and users will benefit from it.