Repository navigation
detach ACPI platform dev from pm_domain for Minnowboard - #708
Conversation
bardliao
left a comment
There was a problem hiding this comment.
Verified on BYT and it looks reasonable to me.
|
@juimonen this will allow all tests on the minnowboard but what we really need is for PM to be implemented. Do you think you could take a look into it? |
|
Don't we have a known issue that the MinnowBoard will not survive S3 with audio enabled? |
@plbossart is that so? I had no idea. Then all the more reason to disable PM I suppose |
|
@lgirdwood can you comment here? I think you're the only one who talked to the BIOS folks on this S3 issue. |
|
@ranj063 minnowboard S3 -> S0 does not bring the DSP HW back to D0 state. A BIOS issue afaict. |
|
@lgirdwood @plbossart in that case, is this change acceptable as a means to prevent the DSP from being turn off when all the DAPM widgets get powered down? |
we should do this conditionally on platforms where we know S3 support is broken. I think PM would work without issues on most platforms. I seem to be the only one left with Baytrail devices, we need to get more really. |
@plbossart a quick clarification. should this be platform specific or board specific? Is the minnowboard the only known BYT board to have this problem? How about the T100 that @singalsu has? |
|
S3 exit should work on most platforms, I only know of the MinnowBoard as a problematic case. |
|
|
@plbossart sorry I am a bit confused about this. if I were to do this conditionally only for the minnmowboard, should I match the DMI and then disable PM? |
|
On 3/15/19 6:21 PM, Ranjani Sridharan wrote:
@plbossart <https://github.com/plbossart> sorry I am a bit confused
about this. if I were to do this conditionally only for the
minnmowboard, should I match the DMI and then disable PM?
Yes, that was the idea. Use DMI-based quirks, it's the only way I can
think of.
|
|
@plbossart updated now |
plbossart
left a comment
There was a problem hiding this comment.
Looks good, maybe a comment on the dev_pm_detach_domain() below to help reviewers, and also maybe add in the commit message what issue this really fixes (can't recall if this is related to the rt5651 issues)
There was a problem hiding this comment.
maybe add a comment that the second argument is telling the PM core to leave the DSP on during the detach process, not that it invalidates the detach itself:
@power_off: Used to indicate whether we should power off the device.
A recent change in the SOF driver to power off all DAPM widgets in the audio pipeline when not active causes the DSP on the Minnowboard to be powered off. This is despite the fact that PM is not implemented for the Baytrail platform and we explicitly keep the device usage count > 0 to prevent the system from being runtime suspended. In ACPI platforms (such as BYT), the platform device is attached to a pm_domain. The pm_domain provider is responsible for powering on/off all the devices that are attached it. So when all the DAPM widgets in the card get powered down, the pm_domain provider powers off the DSP as well. Additionally, Minnowboard has a known issue with the BIOS that prevents the DSP from resuming back to D0 when the system resumes from S3 to S0. So, given the above limitations, it is recommended to detach the platform device from the pm_domain to prevent the DSP from being powered off on the Minnowboard. Signed-off-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
|
@plbossart updated now |
Minnowboard has a known issue with the BIOS that prevents the DSP
from resuming back to D0 when the system resumes from S3 to S0.
So, detach the platform device from the pm_domain to prevent
the DSP from entering D3.