Fix deep sleep lock bugs #5220

c1728p9 · 2017-09-28T21:47:54Z

Fix problems which could leave deep sleep locked unintentionally, along with adding tests to verify this behavior is fixed.

c1728p9 · 2017-09-28T23:53:52Z

/morph test

mbed-bot · 2017-09-29T02:59:19Z

Result: SUCCESS

Your command has finished executing! Here's what you wrote!

/morph test

Output

mbed Build Number: 1430

All builds and test passed!

0xc0170

👍 For tests

0xc0170 · 2017-09-29T07:56:23Z

platform/DeepSleepLock.h

+            sleep_manager_lock_deep_sleep();
+        }
+        if (0 == count) {
+            error("DeepSleepLock underflow (< 0)");


Is this error report correct? if it is 0 , it already wrapped around (we increment here) ? The conditions seems like reversed (here and in unlock) ?

It should be. The function core_util_atomic_incr_u16 returns the new value, so if lock (which should only make the value bigger) sets the value to 0 then an overflow has occurred.

OK, I realized after your response (pre vs post condition :) ), what about both being underflow? in the case of 0 here, it is overflow

0xc0170 · 2017-09-29T11:25:24Z

retest uvisor

Add _lock_count to DeepSleepLock and use this to prevent deep sleep from staying locked when the DeepSleepLock objected is destroyed after an unbalanced number of calls to lock and unlock.

Test that DeepSleepLock and Timer lock deep sleep at the correct time and do no leave sleep locked after they are destroted.

Add a critical section to attach_us so setting _function and locking deep sleep are done atomically.

Unlock sleep in CAN and SerialBase. This prevents deep sleep from staying locked if these objects are destroyed without first clearing the callbacks.

c1728p9 · 2017-10-02T16:36:47Z

Fix overflow case error message in DeepSleepLock.h and rebased to master.

c1728p9 · 2017-10-02T16:37:03Z

/morph test-nightly

mbed-bot · 2017-10-03T07:07:10Z

Result: ABORTED

Your command has finished executing! Here's what you wrote!

/morph test-nightly

0xc0170 · 2017-10-03T11:35:38Z

@pan- Can you review please

adbridge · 2017-10-03T11:51:01Z

/morph test-nightly

pan- · 2017-10-03T15:15:33Z

drivers/Ticker.h

@@ -113,12 +114,14 @@ class Ticker : public TimerEvent, private NonCopyable<Ticker> {
     *
     */
    void attach_us(Callback<void()> func, us_timestamp_t t) {
+        core_util_critical_section_enter();


if func == NULL then the lock won't be correctly released.

mbed-bot · 2017-10-03T16:48:20Z

Result: FAILURE

Your command has finished executing! Here's what you wrote!

/morph test-nightly

Output

mbed Build Number: 1511

Test failed!

c1728p9 · 2017-10-03T20:22:21Z

This depends on #5242 for testing to pass

0xc0170 · 2017-10-04T06:01:04Z

/morph test-nightly

mbed-bot · 2017-10-04T06:45:09Z

Result: FAILURE

Your command has finished executing! Here's what you wrote!

/morph test-nightly

Output

mbed Build Number: 1518

Build failed!

0xc0170 · 2017-10-04T07:09:19Z

Another utest failure for IAR and one target HEXIWEAR::IAR::FEATURES-FRAMEWORKS-UTEST-TESTS-UNIT_TESTS-CASE_CONTROL_ASYNC , investigating

The error says: Unknown toolchain for memory statistics IAR

cc @studavekar

Update: there is actually this error:

01:32:39 a'I/O error(28): No space left on device
01:32:39 Traceback (most recent call last):
01:32:39   File "C:\mj\workspace\bm_wrap\1601\mbed-os\tools\test_api.py", line 2114, in build_test_worker
01:32:39     bin_file = build_project(*args, **kwargs)
01:32:39   File "C:\mj\workspace\bm_wrap\1601\mbed-os\tools\build_api.py", line 578, in build_project
01:32:39     cur_result["memory_usage"] = memap_instance.mem_report
01:32:39 AttributeError: 'NoneType' object has no attribute 'mem_report'

adbridge · 2017-10-04T11:52:51Z

I'm going to try kicking this off again!
/morph test-nightly

mbed-bot · 2017-10-05T04:22:54Z

Result: SUCCESS

Your command has finished executing! Here's what you wrote!

/morph test-nightly

Output

mbed Build Number: 1523

All builds and test passed!

c1728p9 mentioned this pull request Sep 28, 2017

Deep sleep lock would overflow (> USHRT_MAX) #5067

Closed

0xc0170 reviewed Sep 29, 2017

View reviewed changes

0xc0170 added the needs: review label Sep 29, 2017

c1728p9 added 4 commits October 2, 2017 11:34

Prevent DeepSleepLock from leaving sleep locked

f0ac234

Add _lock_count to DeepSleepLock and use this to prevent deep sleep from staying locked when the DeepSleepLock objected is destroyed after an unbalanced number of calls to lock and unlock.

Add a test to validate locking

ea46885

Test that DeepSleepLock and Timer lock deep sleep at the correct time and do no leave sleep locked after they are destroted.

Protect Ticker attach with a critical section

87b151c

Add a critical section to attach_us so setting _function and locking deep sleep are done atomically.

Properly unlock sleep in destructor of drivers

f854f3e

Unlock sleep in CAN and SerialBase. This prevents deep sleep from staying locked if these objects are destroyed without first clearing the callbacks.

c1728p9 force-pushed the sleep_lock_fixes branch from a23b8b2 to f854f3e Compare October 2, 2017 16:36

0xc0170 added the release-version: 5.6.2 label Oct 3, 2017

pan- reviewed Oct 3, 2017

View reviewed changes

0xc0170 mentioned this pull request Oct 4, 2017

RTOS : added empty and full functions to Mail and Queue #5158

Merged

0xc0170 added ready for merge and removed needs: review labels Oct 5, 2017

theotherjimmy merged commit 4dff32a into ARMmbed:master Oct 5, 2017

theotherjimmy removed the ready for merge label Oct 5, 2017

Fix deep sleep lock bugs #5220

Fix deep sleep lock bugs #5220

Uh oh!

Conversation

c1728p9 commented Sep 28, 2017

Uh oh!

c1728p9 commented Sep 28, 2017

Uh oh!

mbed-bot commented Sep 29, 2017

Output

Uh oh!

0xc0170 left a comment

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 Sep 30, 2017 • 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!

0xc0170 commented Sep 29, 2017

Uh oh!

c1728p9 commented Oct 2, 2017

Uh oh!

c1728p9 commented Oct 2, 2017

Uh oh!

mbed-bot commented Oct 3, 2017

Uh oh!

0xc0170 commented Oct 3, 2017

Uh oh!

adbridge commented Oct 3, 2017

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mbed-bot commented Oct 3, 2017

Output

Uh oh!

c1728p9 commented Oct 3, 2017

Uh oh!

0xc0170 commented Oct 4, 2017

Uh oh!

mbed-bot commented Oct 4, 2017

Output

Uh oh!

0xc0170 commented Oct 4, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

adbridge commented Oct 4, 2017

Uh oh!

mbed-bot commented Oct 5, 2017

Output

Uh oh!

Uh oh!

0xc0170 Sep 30, 2017 •

edited

Loading

0xc0170 commented Oct 4, 2017 •

edited

Loading