darrylring opened a new pull request, #21814: URL: https://github.com/apache/nuttx/pull/21814
## Summary While testing the timer capture (specifically with PWM inputs) on the STM32H5, I encountered a number of issues, all fairly small. This PR attempts to resolve them all, though I'm happy to split it into smaller PRs if preferred. 1. For PWM input mode, two channels use the same pin. However, `stm32_cap_getcapture` only actually handled the second channel if the board defined a pin macro `GPIO_TIMx_CHxIN` for it. I didn't do this, and so reading the capture register returns 0xFFFFFFFF and the duty cycle is garbage. 2. Avoid dividing by zero when calculating frequency. The STM32H5 has this fix already, and this is applying it to the other capture drivers. 3. Capture counter is reset on the capture edge, so the count is one less than the actual number of timer ticks. This leads to the capture values being off-by-one, and for frequency and duty cycle measurements being slightly off. 4. `maxtimeout` gets set to `(1 << resolution) - 1` but if resolution is 32, this is UB. Do this calculation in 64-bits instead. This matches what is done in the timer drivers. 5. Properly disable capture channels in `stm32_stop`. ## Impact Previously slightly wrong PWM input values are now corrected. Possibly someone is relying on the slightly incorrect values, or correcting them in software after the fact? Not sure how to judge this impact. `stm32_stop` now actually stops the capture channels. Any application counting on the capture timer to remain running will now fail. ## Testing **NOTE: This touches STM32H7 code and code common to other STM32 families. I can only test STM32H5. @raiden00pl would you be able to help me out here and run a similar test on other boards?** Tested on the NUCLEO-H563ZI board with the `capture` defconfig with the following changes: [nucleo-h563zi-capture.patch](https://github.com/user-attachments/files/34761680/nucleo-h563zi-capture.patch) I have a jumper wire from PB6 (CN10-14, TIM4 CH1) to PE9 (CN10-4, TIM1 CH1), generate PWM on TIM4 channel 1 using the `pwm` NSH builtin, and read the TIM1 channel 1 PWM input using the `cap` NSH builtin: ``` nsh> pwm -f 2000 -d 50 -t 60 & nsh> cap -n 5 ``` **BEFORE: With jumper wire** ``` nsh> pwm -f 2000 -d 50 -t 60 & pwm [4:100] nsh> pwm_main: starting output with frequency: 2000 channel: 1 duty: 00007fff cap -n 5 cap_main: Hardware initialized. Opening the capture device: /dev/cap0 cap_main: Number of samples: 5 pwm duty cycle: 49 % pwm frequency: 2004 Hz pwm edges counting: 6 pwm duty cycle: 49 % pwm frequency: 2004 Hz pwm edges counting: 221 pwm duty cycle: 49 % pwm frequency: 2004 Hz pwm edges counting: 441 pwm duty cycle: 49 % pwm frequency: 2004 Hz pwm edges counting: 661 pwm duty cycle: 49 % pwm frequency: 2004 Hz pwm edges counting: 881 Terminating! ``` **AFTER: With jumper wire** ``` sh> pwm -f 2000 -d 50 -t 60 & pwm [4:100] nsh> pwm_main: starting output with frequency: 2000 channel: 1 duty: 00007fff cap -n 5 cap_main: Hardware initialized. Opening the capture device: /dev/cap0 cap_main: Number of samples: 5 pwm duty cycle: 50 % pwm frequency: 2000 Hz pwm edges counting: 6 pwm duty cycle: 50 % pwm frequency: 2000 Hz pwm edges counting: 225 pwm duty cycle: 50 % pwm frequency: 2000 Hz pwm edges counting: 445 pwm duty cycle: 50 % pwm frequency: 2000 Hz pwm edges counting: 665 pwm duty cycle: 50 % pwm frequency: 2000 Hz pwm edges counting: 885 Terminating! ``` **AFTER: No jumper wire** STM32H5 already has the divide-by-0 fix, so this works fine before as well. ``` nsh> cap -n 5 cap_main: Hardware initialized. Opening the capture device: /dev/cap0 cap_main: Number of samples: 5 pwm duty cycle: 0 % pwm frequency: 0 Hz pwm edges counting: 0 pwm duty cycle: 0 % pwm frequency: 0 Hz pwm edges counting: 0 pwm duty cycle: 0 % pwm frequency: 0 Hz pwm edges counting: 0 pwm duty cycle: 0 % pwm frequency: 0 Hz pwm edges counting: 0 pwm duty cycle: 0 % pwm frequency: 0 Hz pwm edges counting: 0 Terminating! ``` -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
