WIP: pm_layered for cpu/atmega_common #8207

roberthartung · 2017-12-05T08:58:57Z

This is a WIP PR for pm_layered in atmega_common. This has to be discussed on several parts, e.g.:

How to actually disable/enable correct hardware devices?

roberthartung · 2017-12-05T10:25:56Z

cpu/atmega_common/periph/adc.c

@@ -33,6 +36,10 @@ static mutex_t lock = MUTEX_INIT;
 static inline void _prep(void)
 {
    mutex_lock(&lock);
+#ifdef MODULE_PM_LAYERED
+    pm_block(PM_INVALID_ADC);


The define PM_INVALID_ADC defines the mode which the ADC needs to be running.

roberthartung · 2017-12-05T10:26:25Z

cpu/atmega_common/periph/spi.c

+#ifdef MODULE_PM_LAYERED
+    pm_block(PM_INVALID_SPI);
+#endif
+    switch(bus) {


Is there a better way than a switch/case?

roberthartung · 2017-12-05T10:26:43Z

cpu/atmega_common/periph/spi.c


    /* configure as master, with given mode and clock */
    SPSR = (clk >> S2X_SHIFT);
    SPCR = ((1 << SPE) | (1 << MSTR) | mode | (clk & CLK_MASK));
+    //SPCR |= (1 << SPE);


Duplicate line... right?

roberthartung · 2017-12-05T10:27:10Z

cpu/atmega_common/periph/timer.c

@@ -58,6 +62,7 @@ typedef struct {
    void *arg;                  /**< interrupt callback argument */
    uint8_t mode;               /**< remember the configured mode */
    uint8_t isrs;               /**< remember the interrupt state */
+    int8_t num;                /** < hardware timer number */


This is an approach where I map the timer within RIOT to the actual hardware timer. Any better solutions?

roberthartung · 2017-12-05T10:27:23Z

cpu/atmega_common/periph/timer.c

@@ -85,6 +90,113 @@ static ctx_t *ctx[] = {{ NULL }};
 #endif
 /** @} */

+static inline void timer_poweroff(tim_t tim) {
+    switch(ctx[tim].num) {


Again, big switch/case

roberthartung · 2017-12-05T10:27:37Z

cpu/atmega_common/periph/timer.c

+
+    #ifdef MEGA_TIMER0
+    if(ctx[tim].dev == MEGA_TIMER0) {
+        ctx[tim].num = 0;


Maybe there is a better/more elegant way of setting the hardware timer?

roberthartung · 2017-12-05T10:29:06Z

cpu/atmega_common/periph/uart.c


    if (sched_context_switch_request) {
        thread_yield();
    }
 }

+void uart_poweron(uart_t uart) {
+    switch(ctx[uart].num) {
+#ifdef MEGA_UART0


Do we need a differentiation if we have rx/tx enabled or not? See discussion in #7947

miri64 · 2018-03-15T12:09:27Z

I'm not sure why I self-assigned for this... Maybe @MichelRottleuthner @kaspar030 or @haukepetersen want to take a look instead?

stale · 2019-08-10T07:07:50Z

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. If you want me to ignore this issue, please mark it with the "State: don't stale" label. Thank you for your contributions.

roberthartung · 2019-08-19T07:56:37Z

@miri64 Can you remove the label? Now that #11122 is merged, I can finally work on the pm_layered module again (can finally test it with our platform that needs #11122).

miri64 · 2019-08-28T05:24:57Z

@miri64 Can you remove the label?

Stale bot did it already in reaction to your comment ;-)

maribu · 2019-10-15T18:06:21Z

@roberthartung: Feel free to ping me when this PR has been rebased and got into a state a review makes sense. I really would like to see the power management of the ATmega platform improved in RIOT.

roberthartung · 2019-10-16T06:57:28Z

@maribu Thanks! Will work on this as one of the next steps. Are there enough tests for pm already? I should also add my pm test to this PR:

A simple test that consists of an RTT and enters a low-power sleep mode and then wakes up from it using the RTT.

miri64 · 2019-10-16T07:08:29Z

@MrKevinWeiss can probably answer that.

MrKevinWeiss · 2019-10-16T11:20:42Z

I haven't touch pm stuff yet in the testing. I am working on a way to do so with the HiL setup but I don't think it would be ready in time.

I can however run some tests manually and can also monitor the current (I think).

I think having a simple (on target) test that turns things off and on is great. You can also try failure cases where you try to turn the power off when something is held on. Or what happens when turning power on twice...

That kind of thing.

The easier the test is to run the easier it is to merge in.

My "helping people" time is currently with the release and esp8266 PR but after I clear those away I would be happy work on how to test this.

benpicco · 2019-12-03T22:18:49Z

A simple test that consists of an RTT and enters a low-power sleep mode and then wakes up from it using the RTT.

tests/periph_pm will do just that.
Unfortunately it uses the RTC instead of RTT, but that should be an easy change.

stale · 2020-06-05T22:23:19Z

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. If you want me to ignore this issue, please mark it with the "State: don't stale" label. Thank you for your contributions.

cpu/atmega_common: adds pm_block/pm_unblock for pm_layered

33f3446

roberthartung commented Dec 5, 2017

View reviewed changes

miri64 self-assigned this Dec 11, 2017

miri64 added Platform: AVR Platform: This PR/issue effects AVR-based platforms Type: new feature The issue requests / The PR implemements a new feature for RIOT labels Dec 11, 2017

miri64 removed their assignment Mar 15, 2018

tcschmidt requested a review from MichelRottleuthner May 25, 2018 10:14

stale bot added the State: stale State: The issue / PR has no activity for >185 days label Aug 10, 2019

stale bot removed the State: stale State: The issue / PR has no activity for >185 days label Aug 19, 2019

benpicco mentioned this pull request Dec 3, 2019

cpu/atmega256rfr2: symbol counter based RTT support #12852

Merged

This was referenced Feb 7, 2020

github: add CODEOWNERS file #13289

Merged

Sleep mode for Arduino #13321

Closed

cpu/atmega_common: Cleanup state flags #13308

Merged

stale bot added the State: stale State: The issue / PR has no activity for >185 days label Jun 5, 2020

stale bot closed this Jul 6, 2020

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

WIP: pm_layered for cpu/atmega_common #8207

WIP: pm_layered for cpu/atmega_common #8207

roberthartung commented Dec 5, 2017

roberthartung Dec 5, 2017

roberthartung Dec 5, 2017

roberthartung Dec 5, 2017

roberthartung Dec 5, 2017

roberthartung Dec 5, 2017

roberthartung Dec 5, 2017

roberthartung Dec 5, 2017

miri64 commented Mar 15, 2018

stale bot commented Aug 10, 2019

roberthartung commented Aug 19, 2019

miri64 commented Aug 28, 2019

maribu commented Oct 15, 2019

roberthartung commented Oct 16, 2019

miri64 commented Oct 16, 2019

MrKevinWeiss commented Oct 16, 2019

benpicco commented Dec 3, 2019

stale bot commented Jun 5, 2020

WIP: pm_layered for cpu/atmega_common #8207

WIP: pm_layered for cpu/atmega_common #8207

Conversation

roberthartung commented Dec 5, 2017

roberthartung Dec 5, 2017

Choose a reason for hiding this comment

roberthartung Dec 5, 2017

Choose a reason for hiding this comment

roberthartung Dec 5, 2017

Choose a reason for hiding this comment

roberthartung Dec 5, 2017

Choose a reason for hiding this comment

roberthartung Dec 5, 2017

Choose a reason for hiding this comment

roberthartung Dec 5, 2017

Choose a reason for hiding this comment

roberthartung Dec 5, 2017

Choose a reason for hiding this comment

miri64 commented Mar 15, 2018

stale bot commented Aug 10, 2019

roberthartung commented Aug 19, 2019

miri64 commented Aug 28, 2019

maribu commented Oct 15, 2019

roberthartung commented Oct 16, 2019

miri64 commented Oct 16, 2019

MrKevinWeiss commented Oct 16, 2019

benpicco commented Dec 3, 2019

stale bot commented Jun 5, 2020