Dma stm32 fix

[avisconti]

(FIX for STM32 DMA)
I noticed that the STM32 DMA is not used much inside zephyr, expecially the memory-to-peripheral and peripherl-to-memory part. I'm currently implementing the stm32-i2s driver and I'm using the DMA for that purpose. My intention is to use it also inside SPI driver and maybe others.

[codecov-io]

Codecov Report

Merging #8579 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master    #8579   +/-   ##
=======================================
  Coverage   64.17%   64.17%           
=======================================
  Files         429      429           
  Lines       41157    41157           
  Branches     6925     6925           
=======================================
  Hits        26413    26413           
  Misses      11587    11587           
  Partials     3157     3157

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 19d7803...8b8d8ba. Read the comment docs.

[SebastianBoe]

recheck

[avisconti]

@SebastianBoe
same (or similar) error

[avisconti]

Ok, sanitycheck is passed and bug has been closed. If it happens again I will re-open it.

[avisconti]

@erwango
Pls add who else is needed for the review of this part. I noticed that Lee Jones (Linaro) has done some work in the past on stm32 DMA.

[erwango]

LGTM, Can you confirm that tests/drivers/dma is still 'passed' with this change?
(requires to change DMA_DEVICE_NAME to "DMA_2")

[avisconti]

LGTM, Can you confirm that tests/drivers/dma is still 'passed' with this change?
(requires to change DMA_DEVICE_NAME to "DMA_2")

Actually I don't know. How can I check it?
I have to improve my knowledge on the Continuos Integration (CI) system technology.

[erwango]

@avisconti :
Test is not in CI, You have to run the test manually
Go in tests/drivers/dma/loop_transfer
Make the change to "DMA_2"
Build vs a target supporting dma, then flash and run the test, and you will have the verdict.

Then (if you're interested), maybe would be worth to get some STM32 targets running this test in CI, btw...

[avisconti]

@erwango

Test is not in CI, You have to run the test manually

OK, got it.

Go in tests/drivers/dma/loop_transfer
Make the change to "DMA_2"
Build vs a target supporting dma, then flash and run the test, and you will have the verdict.

I had also to add in prj.conf the following
CONFIG_DMA_STM32F4X=y

After that I launched the test on my ArgonKey and it worked:

***** Booting Zephyr OS v1.12.0-453-g286bd85 *****
DMA memory to memory transfer started on DMA_2
Preparing DMA Controller
Starting the transfer and waiting for 1 second
Each RX buffer should contain the full TX buffer string.
TX data: The quick brown fox jumps over the lazy dog
RX data Loop 0: The quick brown fox jumps over the lazy dog
RX data Loop 1: The quick brown fox jumps over the lazy dog
RX data Loop 2: The quick brown fox jumps over the lazy dog
RX data Loop 3: The quick brown fox jumps over the lazy dog
RX data Loop 4: The quick brown fox jumps over the lazy dog
Finished: DMA

[avisconti]

@erwango
Pls note that the DMA loop transfer test does only memory to memory transfers, while my modifications are more concerned with mem-to-periph and periph-to-mem. But it is good to see that my commits didn't break the functionality.

To really test the commits I used the I2S driver I'm writing and tested it on ArgonKey. I tested it in both directions using I2S RX (from Microphone) and I2S TX on 96 board connector watching the transactions with a scope.

[avisconti]

@erwango
Is this patch fine for you then? As you see above the dma test pass as expected.