5.5.6 regression (stuck at boot) on devices using the sof_hda audio driver + fix

From: Hans de Goede <hdegoede@redhat.com>
To: stable@vger.kernel.org,
	Sathyanarayana Nujella <sathyanarayana.nujella@intel.com>,
	Cezary Rojewski <cezary.rojewski@intel.com>,
	Jaroslav Kysela <perex@perex.cz>
Subject: 5.5.6 regression (stuck at boot) on devices using the sof_hda audio driver + fix
Date: Thu, 27 Feb 2020 20:39:04 +0100	[thread overview]
Message-ID: <e15641c3-2cf8-db66-3eeb-019af4b482db@redhat.com> (raw)

Hi All,

I and various other Fedora users have noticed that Fedora's 5.5.6 build gets stuck
at boot on a Lenovo X1 7th gen, see:
https://bugzilla.redhat.com/show_bug.cgi?id=1772498

This is caused by the addition of this commit to 5.5.6:

24c259557c45 ("ASoC: SOF: Intel: hda: Fix SKL dai count")
https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=linux-5.5.y&id=24c259557c45e817941d3843f82331a477c86a7e

###
ASoC: SOF: Intel: hda: Fix SKL dai count
[ Upstream commit a6947c9d86bcfd61b758b5693eba58defe7fd2ae ]

With fourth pin added for iDisp for skl_dai, update SOF_SKL_DAI_NUM to
account for the change. Without this, dais from the bottom of the list
are skipped. In current state that's the case for 'Alt Analog CPU DAI'.

Fixes: ac42b142cd76 ("ASoC: SOF: Intel: hda: Add iDisp4 DAI")
Signed-off-by: Cezary Rojewski <cezary.rojewski@intel.com>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20200113114054.9716-1-cezary.rojewski@intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
###

Notice the "Fixes: ac42b142cd76 (...)", that commit-id actually
does not exist, the correct commit-id which this fixes is:

e68d6696575e ("ASoC: SOF: Intel: hda: Add iDisp4 DAI")
and that commit is not in 5.5.6, which is causing the problem,
the missing commit makes an array one larger and the fix for the
missing fix which did end up in 5.5.6 and bumps a define which is
used to walk over the array in some places by one so now the
walking is going over the array boundary.

For the Fedora kernels I've fixed this by adding the
"ASoC: SOF: Intel: hda: Add iDisp4 DAI" commit as a downstream
patch for our kernels. I believe that this is probably the best
fix for 5.5.z too.

Regards,

Hans

p.s.

I know that the stable series are partly based on automatically
picking patches now. I wonder if the scripts doing that could be
made smarter wrt rejecting patches with a Fixes tag where the
fixed patch is not present, so where in essence a pre-requisite
of the patch being added is missing ?