Re: [PATCH] ACPI / LPSS: Don't skip late system PM ops for hibernate on BYT/CHT

From: "Robert R. Howell" <RHowell@uwyo.edu>
To: Hans de Goede <hdegoede@redhat.com>,
	Kai-Heng Feng <kai.heng.feng@canonical.com>,
	"rjw@rjwysocki.net" <rjw@rjwysocki.net>
Cc: "lenb@kernel.org" <lenb@kernel.org>,
	"linux-acpi@vger.kernel.org" <linux-acpi@vger.kernel.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] ACPI / LPSS: Don't skip late system PM ops for hibernate on BYT/CHT
Date: Sun, 7 Apr 2019 20:58:12 +0000	[thread overview]
Message-ID: <b6cd67d7-a4de-0fab-4512-25d732190d17@uwyo.edu> (raw)
In-Reply-To: <eb1462c9-ebef-7bd6-c263-3f4f2e8ba63d@redhat.com>

On 4/3/19 2:54 AM, Hans de Goede wrote:

> 
> Hi,
> 
> On 03-04-19 07:43, Kai-Heng Feng wrote:
>> i2c-designware-platdrv fails to work after the system restored from
>> hibernation:
>> [ 272.775692] i2c_designware 80860F41:00: Unknown Synopsys component type: 0xffffffff
>>
>> Commit 48402cee6889 ("ACPI / LPSS: Resume BYT/CHT I2C controllers from
>> resume_noirq") makes acpi_lpss_{suspend_late,resume_early}() bail early
>> on BYT/CHT as resume_from_noirq is set. This means dw_i2c_plat_resume()
>> doesn't gets called by acpi_lpss_resume_early(), and this causes the
>> issue.
>>
...
> 
> The ordering problem fixed by commit 48402cee6889 can hit hibernate too,
> so I think it would be better to do this instead to fix this problem:
> 
> diff --git a/drivers/acpi/acpi_lpss.c b/drivers/acpi/acpi_lpss.c
> index 1e2a10a06b9d..cf768608437e 100644
...
> If people affected by the problem can give my version of the fix a test,
> then that would be great.
> 
> Regards,
> 
> Hans

I did try applying your (Hans') patch to a 5.0.0 and a 5.1-rc3 kernel, instead of what I had 
been doing previously, which was reverting commit 02e45646d53b.  I tested it on an ASUS T100TA 
and unfortunately with both kernels after resume from hibernation I still get errors like those 
listed below, and sound no longer works.  I also tried applying the patch from Kai-Heng Feng 
(instead of Hans' patch or my revert) on both kernels and again in both cases I get these same errors.

In contrast, on the 5.0.0 kernel when I revert 02e45646d53b, the problem is fixed.  Because 
of changes in adjacent code in 5.1-rc3, I haven't yet found a way to merge the new code 
and the 02e45646d53b revert without causing worse problems. But that's probably because 
I don't really understand the details of that code.

[  120.690428] i2c_designware 80860F41:01: timeout waiting for bus ready
[  120.690442] rt5640 i2c-10EC5640:00: Failed to write 13d = 3600: -110
[  120.712860] i2c_designware 80860F41:01: timeout waiting for bus ready
[  120.737389] i2c_designware 80860F41:01: timeout waiting for bus ready
[  120.759657] i2c_designware 80860F41:01: timeout waiting for bus ready
[  120.781797] i2c_designware 80860F41:01: timeout waiting for bus ready
[  120.804153] i2c_designware 80860F41:01: timeout waiting for bus ready
[  120.826467] i2c_designware 80860F41:01: timeout waiting for bus ready
[  121.965886] i2c_designware 80860F41:01: timeout in disabling adapter
.....
[  132.782634] i2c_designware 80860F41:01: controller timed out
[  133.806538] i2c_designware 80860F41:01: controller timed out
[  134.830991] i2c_designware 80860F41:01: controller timed out
[  135.855183] i2c_designware 80860F41:01: controller timed out

Let me know if I can do any other testing to help.

Bob Howell