Follow-up question to DMA

[JackCode0011]

Hello everyone

After my last question about DMA I have a follow-up question.
The code is now running smoothly, except for one detail.

When SM and DMA are collecting data, I use "normal" Micropython to do stuff with the data. In this while-loop, I also check, if SM is still active if sma.active():, I check if DMA is active if dma1.active() or dma2.active(): and a last check, if DMA is really collecting. I use DMA.count with a timedelay and if the difference is 0, DMA is not collecting, otherwise it is. The timedelay gets calculated according to the SM freq, so that it should never happend, that DMA.count is read at the same point.

Now and then I loose the input signals for a short or longer periode. If that happens, the checks I do start a second loop, where it's checking a few more times with a delay between each check, if maybe the signal is back. If not, SM and DMA are closed, reset and started again.

Now my problem: When simulating this, by switching of the device, that delivers the signal, everything is ok and the checks are done, but when I switch it back on, the code is still not able to detect the signal. If a do a hard-reset, everything works again.
Now, after some debugging I'm again pretty sure, that DMA is the problem again.

What I did so far (tried in different orders):
dma.active(0)
dma.config(Trigger=False)
dma.close
reseting with CHAN_ABORT on offset 0x464
'0x1 reset on bit 2 of RESETS, 0x40020000'

After also trying different code it seems, that there is still something in the DMA-settings, that doesn't get cleared and stays on.
Also stopping the code with Thonny or PyCharm doesn't help. It has to be a hard-reset. So something stays in the RP2350 I think.

My question:
Is there another way to reset DMA, which I haven't tried yet?

Greetings

[GitHubsSilverBullet]

dma.close

really wants some brackets: dma.close()

[JackCode0011]

Yes of course. It was a list indicating what had already been tried, but without regard to correct syntax.

[peterhinch]

It's important to shut everything down, otherwise the RP2 can be left in an unstable state. My approach is to run the application in a try...finally bock, ensuring that any exit will shut down cleanly:

obj = MyObject()  # A class using SM and DMA instances
try:
    asyncio.run(application())  # Code that uses obj
finally:
    obj.deinit()  # Shut down

The deinit method looks like this

    def deinit(self):
        self._dma.active(0)
        self._dma.close()
        self._csn.irq(None)  # Disable intarrupt
        self._sm.active(0)  # Stop the SM

[GitHubsSilverBullet]

reseting with CHAN_ABORT on offset 0x464
'0x1 reset on bit 2 of RESETS, 0x40020000'

That doesn't make any sense. 0x4002_0000 is PADS_QSPI_BASE
If you want to do RESETS, that'd be 0x4000_c000 RESETS_BASE using the RESET register at offset 0 and bit 2 to reset the entire DMA system.