Fixed issue with PWM not being freed when the object is destroyed #10074
Conversation
|
@morser499, thank you for your changes. |
| @@ -68,6 +68,7 @@ class PwmOut { | |||
| ~PwmOut() | |||
| { | |||
| core_util_critical_section_enter(); | |||
| pwmout_free(&_pwm); | |||
There was a problem hiding this comment.
@ARMmbed/mbed-os-hal what was the reason not having this earlier ? I recall some discussions in some other drivers around definition of _free and implications for the rest of targets with change like this one.
There was a problem hiding this comment.
I thought it was because not all the targets had this function. If this is the case then CI should catch any problems.
There was a problem hiding this comment.
That is it !
We do not have 100 percent coverage in CI, I am thinking to be on safer side to have this on 5.13 rather - to keep this on master for some time.
|
CI started |
Test run: SUCCESSSummary: 13 of 13 test jobs passed |
| * destructor. The pwmout_init handles multiple calls of constructor. | ||
| */ | ||
| #if DEVICE_SLEEP && DEVICE_LPTICKER | ||
| if (!Cy_SysPm_UnregisterCallback(&obj->pm_callback_handler)) |
There was a problem hiding this comment.
Can you fix the style { on the same line (entire file is consistent)
morser499
force-pushed
the
cy-mbed-os-5.12.0-pwm-free
branch
from
8aaf145 to
51a4713
Compare
|
CI started |
Test run: SUCCESSSummary: 13 of 13 test jobs passed |
…-free Fixed issue with PWM not being freed when the object is destroyed
Description
Fixed issue with PWM not being freed when the object is destroyed. This caused the PWM to continue running after the destructor was called and it could cause a crash if another PWM was initialized.
Pull request type