Re: Regression found (Stop-marking-clocks-as-CLK_IS_CRITICAL)

From: Hans de Goede
Date: Fri Jan 18 2019 - 10:33:36 EST


On 1/17/19 2:16 PM, Dean Wallace wrote:
On 17-01-19, Hans de Goede wrote:
Mogens, Dean, can you please try the SND_SOC_INTEL_CHT_BSW_MAX98090_TI_MACH
driver, without reverting any patches, with the attached patch on top and
see if that helps?

Hi Hans. Just compiled 4.20.3 with your new patch and no other
patches. First impressions are, this is now as stable as it has ever
been in my experiences with this baytrail drama. A quick run down on my

No PLL messages on boot, good.
Played audio using mpd, all fine. Plugged in headphones, switches,
stays stable, no messages.
Unplugged/plugged again few times, no issues.
Ran pavucontrol, stayed stable, no messages.
I can still make it lose the lock and turn on the dalek distortion by
switching profiles in pavucontrol (from stereo output to any other, like
putput + input) but that has always been the case afaik for this
machine. So basically this is now back to working as it was before the
4.18.5 'clk' breakage.

One thing to mention, I don;t normally use UCM files, tho I have in the
past, which is where I assume my asound.state file has come from, which
makes my sound work /shrug. Well I've just tested using the UCM, and
things are a little less stable regarding plugging/unplugging headset a
few times. It loses lock, but, what's different now is after maybe 3-4
seconds it corrects itself. Not sure if it matters, but shows things
are a little more stable with this patch than previously known.

Ok, thank you for testing. Lets wait a bit and then if no problems
show up I will submit the patch upstream.