Re: [FAIL bisect] Sound card probe error

[Ulf Hansson]

On 14 May 2018 at 14:22, Sylwester Nawrocki <s.nawrocki@xxxxxxxxxxx> wrote:
> Hi,
>
> On 05/14/2018 12:17 PM, Krzysztof Kozlowski wrote:
>
>> Bisected to:
>> 8c123c14bbba4add148536b6d47a9226deda2f7a is the first bad commit
>> commit 8c123c14bbba4add148536b6d47a9226deda2f7a
>> Author: Ulf Hansson <ulf.hansson@xxxxxxxxxx>
>> Date:   Thu Apr 26 10:53:06 2018 +0200
>>
>>     driver core: Respect all error codes from dev_pm_domain_attach()
>>
>>     The limitation of being able to check only for -EPROBE_DEFER from
>>     dev_pm_domain_attach() has been removed. Hence let's respect all error
>>     codes and bail out accordingly.
>>
>>     Signed-off-by: Ulf Hansson <ulf.hansson@xxxxxxxxxx>
>>     Acked-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
>>     Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@xxxxxxxxx>
>
> The DRM driver creates the HDMI codec but its registration fails, due
> to some missing clocks. The clock are missing because the exynos5-subcmu
> driver probing fails.
>
> [    0.678578] exynos5-subcmu: probe of GSC failed with error -17
> [    0.679250] exynos5-subcmu: probe of MFC failed with error -17
> [    0.679992] exynos5-subcmu: probe of DISP failed with error -17
>
> The exynos5-subcmu driver before registering platform device makes
> of_genpd_add_device() call, so in platform_drv_probe() dev_pm_domain_attach()
> call will fail, as dev->pm_domain is already set. Previously the error
> was masked by code removed in the above commit and platform_drv_probe()
> was could complete successfully.

Thanks for providing this information!

I was about to tell that I overlooked the fact that there is more than
one way to assign PM domain pointers to devices. To fix the problem,
dev_pm_domain_attach() should return 0, not -EEXIST when it finds and
existing PM domain pointer.

Does the below patch fix the problem?

From: Ulf Hansson <ulf.hansson@xxxxxxxxxx>
Date: Mon, 14 May 2018 14:26:16 +0200
Subject: [PATCH] PM / Domains: Don't return -EEXIST at attach when PM domain
 exists

As dev_pm_domain_attach() isn't the only way to assign PM domain pointers
to devices, clearly we must allow a device to have the pointer already
being assigned. For this reason, return 0 instead of -EEXIST.

Reported-by: Krzysztof Kozlowski <krzk@xxxxxxxxxx>
Signed-off-by: Ulf Hansson <ulf.hansson@xxxxxxxxxx>
---
 drivers/base/power/common.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/base/power/common.c b/drivers/base/power/common.c
index 5e4b481..390868c 100644
--- a/drivers/base/power/common.c
+++ b/drivers/base/power/common.c
@@ -106,7 +106,7 @@ int dev_pm_domain_attach(struct device *dev, bool power_on)
        int ret;

        if (dev->pm_domain)
-               return -EEXIST;
+               return 0;

        ret = acpi_dev_pm_attach(dev, power_on);
        if (!ret)
-- 
2.7.4