PSoC 6: rework sleep overrides by Cypress's debug macro #13976

LDong-Arm · 2020-11-27T17:33:54Z

Summary of changes

Fix for PSoC 6 targets mentioned in #13265

According to the discussion here, Cypress has a debug tool that generates the macro CY_CFG_PWR_SYS_IDLE_MODE which intends to disable/limit sleep during debug sessions:

unset or set to CY_CFG_PWR_MODE_DEEPSLEEP: sleep and deep sleep
CY_CFG_PWR_MODE_SLEEP: sleep only
CY_CFG_PWR_MODE_ACTIVE: no sleep

To override sleep behaviours, the previous implementation relies on rtos_attach_idle_hook(), which is a non-API internal function reserved for the user-facing API Kernel::attach_idle_hook() only. Functionally, if an application uses Kernel::attach_idle_hook() (which is a valid use case), the Cypress debug idle hook will get overriden.

As discussed here, this PR moves the sleep overriding to hal_sleep() and hal_deepsleep() implementations.

Impact of changes

Target-specific rework, no noticeable impact to developers.

Migration actions required

None.

Documentation

None.

Pull request type

[x] Patch update (Bug fix / Target update / Docs update / Test update / Refactor)
[] Feature update (New feature / Functionality change / New API)
[] Major update (Breaking change E.g. Return code change / API behaviour change)

Test results

[] No Tests required for this change (E.g docs only update)
[x] Covered by existing mbed-os tests (Greentea or Unittest)
[] Tests / results supplied as part of this PR

Reviewers

@ARMmbed/team-cypress @ARMmbed/mbed-os-core

LDong-Arm · 2020-11-27T17:40:53Z

targets/TARGET_Cypress/TARGET_PSOC6/cy_sleep_api.c

    cyhal_syspm_sleep();
+#endif


Question: As the macro CY_CFG_PWR_SYS_IDLE_MODE is Cypress-specific, maybe it makes more sense to have that defined in cyhal_syspm_sleep() (Cypress SDK code) instead of the Mbed OS HAL implementation? In that case we wouldn't need any tricks at all.

@ARMmbed/team-cypress Or, could you confirm if the macro is still being used by Cypress's debug tool as of today?

That's a good suggestion.
@kyle-cypress please review this proposal and let us know if you'd rather change your SDK.

I don't think we should change cyhal_syspm_sleep. The intent of CY_CFG_PWR_SYS_IDLE_MODE is to control how the RTOS idle task behaves, not to completely disable the sleep driver (even if the two wind up being very similar in practice).

The current proposal is good and easy to follow so I would go with it.

The intent of CY_CFG_PWR_SYS_IDLE_MODE is to control how the RTOS idle task behaves,

Well, as explained in the description, the idle overriding is a feature we provide to the application. Overriding it by drivers/BSPs would create a conflict if the application uses it.

ciarmcom · 2020-11-27T18:00:18Z

@LDong-Arm, thank you for your changes.
@ARMmbed/team-cypress @ARMmbed/mbed-os-maintainers please review.

kyle-cypress · 2020-12-02T15:49:57Z

targets/TARGET_Cypress/TARGET_PSOC6/cy_sleep_api.c

    cyhal_syspm_sleep();
+#endif


I don't think we should change cyhal_syspm_sleep. The intent of CY_CFG_PWR_SYS_IDLE_MODE is to control how the RTOS idle task behaves, not to completely disable the sleep driver (even if the two wind up being very similar in practice).

kyle-cypress · 2020-12-02T15:51:34Z

targets/TARGET_Cypress/TARGET_PSOC6/cy_sleep_api.c

@@ -41,9 +46,14 @@ void hal_deepsleep(void)
    cy_us_ticker_stop();
    cyhal_syspm_deepsleep();
    cy_us_ticker_start();
-#else
+#else // DEVICE_LPTICKER
+    cyhal_syspm_sleep();


This seems like it is adding a lot of complexity to the hal_deepsleep implementation. Wouldn't it be simpler to keep the call to sleep_manager_lock_deep_sleep and let the sleep manager take care of figuring out whether to call hal_sleep or hal_deepsleep?

sleep_manager_lock_deep_sleep() as used before is fine in my opinion. I changed hal_deepsleep() for consistency with the macro-checks in hal_sleep(), but if it's overly complex I can revert this bit.

kyle-cypress · 2020-12-02T15:54:54Z

targets/TARGET_Cypress/TARGET_PSOC6/mbed_overrides.c

-    * Configurator. The default value is system deep sleep.
-    */
-#if (CY_CFG_PWR_SYS_IDLE_MODE == CY_CFG_PWR_MODE_ACTIVE)
-    rtos_attach_idle_hook(&active_idle_hook);


Instead of removing this entirely and adding extra checks to hal_sleep to turn it into a no-op, could this be replaced by a call to sleep_manager_lock_sleep? That would be simpler and I believe it would produce more correct results from the sleep/deepsleep statistics mechanism when that is enabled.

Mbed OS only supports locking deep sleep, not sleep. This was probably the reason an idle hook was used instead. The locking of deep sleep ensures all features work:

mbed-os/platform/include/platform/mbed_power_mgmt.h

Lines 45 to 47 in e77b1d8

* Use locking/unlocking deepsleep for drivers that depend on features that

* are not allowed (=disabled) during the deepsleep. For instance, high frequency

* clocks.

I think in most use cases, the need to lock regular sleep is comparatively low.

@evedon @kjbracey-arm In terms of functionality/actual use cases, does it make sense to add a lock for regular sleep?

I dont see why anyone would lock sleep, as it is very shallow mode (core system clock is disabled).

As @0xc0170 said, locking normal sleep makes little sense in actual use cases, so we may not add that to the API.

evedon · 2020-12-07T13:20:39Z

targets/TARGET_Cypress/TARGET_PSOC6/cy_sleep_api.c

    cyhal_syspm_sleep();
+#endif


The current proposal is good and easy to follow so I would go with it.

0xc0170 · 2020-12-09T06:13:54Z

@kyle-cypress @LDong-Arm are there any outstanding questions? There were couple of comments and some were answered. I set this to review again as I do not believe this is ready for CI.

LDong-Arm · 2020-12-11T12:31:07Z

We have given our views on the questions, is there any proposal on how we can handle this? @ARMmbed/team-cypress

dustin-crossman · 2020-12-11T19:52:07Z

@0xc0170 @LDong-Arm While we still think the new behavior here will differ slightly from the old behavior, we don't have any other proposals for this. We think it's fine to go ahead with this as-is.

0xc0170 · 2020-12-14T12:02:33Z

Thanks @dustin-crossman , we will proceed

0xc0170 · 2020-12-16T15:13:27Z

CI started

mbed-ci · 2020-12-16T17:40:49Z

Jenkins CI Test : ✔️ SUCCESS

Build Number: 1 | 🔒 Jenkins CI Job | 🌐 Logs & Artifacts

CLICK for Detailed Summary

jobs	Status
jenkins-ci/mbed-os-ci_unittests	✔️
jenkins-ci/mbed-os-ci_cmake-example-ARM	✔️
jenkins-ci/mbed-os-ci_build-greentea-ARM	✔️
jenkins-ci/mbed-os-ci_build-example-ARM	✔️
jenkins-ci/mbed-os-ci_build-greentea-GCC_ARM	✔️
jenkins-ci/mbed-os-ci_cmake-example-GCC_ARM	✔️
jenkins-ci/mbed-os-ci_build-example-GCC_ARM	✔️
jenkins-ci/mbed-os-ci_build-cloud-example-GCC_ARM	✔️
jenkins-ci/mbed-os-ci_build-cloud-example-ARM	✔️
jenkins-ci/mbed-os-ci_greentea-test	✔️
jenkins-ci/mbed-os-ci_cmake-example-test	✔️

LDong-Arm added 2 commits November 27, 2020 17:10

PSoC 6: remove debug sleep lock and hook from mbed_sdk_init

7f0c98d

PSoC 6: enable/disable sleep modes based on CY_CFG_PWR_SYS_IDLE_MODE

b1da342

LDong-Arm commented Nov 27, 2020

View reviewed changes

ciarmcom added the release-type: patch Indentifies a PR as containing just a patch label Nov 27, 2020

ciarmcom requested review from a team November 27, 2020 18:00

ciarmcom added the needs: review label Nov 27, 2020

kyle-cypress reviewed Dec 2, 2020

View reviewed changes

evedon approved these changes Dec 7, 2020

View reviewed changes

mergify bot added needs: CI and removed needs: review labels Dec 7, 2020

0xc0170 added needs: review and removed needs: CI labels Dec 9, 2020

0xc0170 approved these changes Dec 14, 2020

View reviewed changes

0xc0170 added ready for merge and removed needs: review labels Dec 17, 2020

0xc0170 merged commit 89bd565 into ARMmbed:master Dec 17, 2020

mergify bot removed the ready for merge label Dec 17, 2020

mbedmain added release-version: 6.7.0 Release-pending and removed release-type: patch Indentifies a PR as containing just a patch Release-pending labels Jan 25, 2021

	* Use locking/unlocking deepsleep for drivers that depend on features that
	* are not allowed (=disabled) during the deepsleep. For instance, high frequency
	* clocks.

		cyhal_syspm_sleep();
		#endif

		cyhal_syspm_sleep();
		#endif

		cyhal_syspm_sleep();
		#endif

PSoC 6: rework sleep overrides by Cypress's debug macro #13976

PSoC 6: rework sleep overrides by Cypress's debug macro #13976

Uh oh!

Conversation

LDong-Arm commented Nov 27, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary of changes

Impact of changes

Migration actions required

Documentation

Pull request type

Test results

Reviewers

Uh oh!

LDong-Arm Nov 27, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ciarmcom commented Nov 27, 2020

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

LDong-Arm Dec 2, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

0xc0170 commented Dec 9, 2020

Uh oh!

LDong-Arm commented Dec 11, 2020

Uh oh!

dustin-crossman commented Dec 11, 2020

Uh oh!

0xc0170 commented Dec 14, 2020

Uh oh!

0xc0170 commented Dec 16, 2020

Uh oh!

mbed-ci commented Dec 16, 2020

Jenkins CI Test : ✔️ SUCCESS

Build Number: 1 | 🔒 Jenkins CI Job | 🌐 Logs & Artifacts

Uh oh!

Uh oh!

LDong-Arm commented Nov 27, 2020 •

edited

Loading

LDong-Arm Nov 27, 2020 •

edited

Loading

LDong-Arm Dec 2, 2020 •

edited

Loading