NuttX enable SPI DMA per board #12436

dagar · 2019-07-09T01:35:58Z

davids5

Did you check all the DMAMAPs to ensure there was no overlap burred in a nuttx driver?

The 512 and 1024 magic number are not helpful and do not tie to the granularity of the allocator. How about define the block size and add a comment to describe the limitations.

What is the formula for changing the BOARD_DMA_ALLOC_POOL_SIZE ? Can it be driven from the defconfig?

boards/aerotenna/ocpoc/src/CMakeLists.txt

jorge789 · 2019-07-09T21:12:32Z

Tested on PixRacer V4:

Modes Tested
Position Mode: Good.
Mission Plan Mode (Automated): Good.
RTL (Return To Land): Good.

Procedure
Arm and Take off in position mode, after flying for approximately one minute, switched to mission plan mode then make sure that vehicle follows all waypoints as shown in QGC, once completed all waypoint see landing behavior.

Note:
No issues noted, good flight in general.

Logs:
https://review.px4.io/plot_app?log=3ca5d59f-e633-4281-ac5f-21974ae98f80

https://review.px4.io/plot_app?log=e48ab5c9-1a1d-4095-be23-23e55a800ef4

Logs: Comparison to master
https://review.px4.io/plot_app?log=29892ab0-32ee-40d8-a5fa-1636ab1f62df

https://review.px4.io/plot_app?log=b7466023-f8d2-4366-83b6-d4cef75e549f

Tested on Pixhawk 2 Cube V3:

Modes Tested
Position Mode: Good.
Mission Plan Mode (Automated): Good.
RTL (Return To Land): Good.

Procedure
Arm and Take off in position mode, after flying for approximately one minute, switched to mission plan mode then make sure that vehicle follows all waypoints as shown in QGC, once completed all waypoint see landing behavior.

Note:
No issues noted, good flight in general.

Logs:
https://review.px4.io/plot_app?log=4e7bffaf-0bb3-4b09-87e6-4304a56c2669

https://review.px4.io/plot_app?log=f467fd1d-c1ae-447a-ac57-815f05f02cfe

Logs: Comparison to master
https://review.px4.io/plot_app?log=1ee0bdfc-0a9e-4a47-8dec-593e8565d45a

dannyfpv · 2019-07-09T21:51:37Z

Tested on Pixhawk4 v5 f-450
GPS issue, please see pictures below.
Tested on Master firmware and no issues.

@dagar

Junkim3DR · 2019-07-09T22:36:08Z

Tested on Pixhawk 4 Mini V5:

Modes Tested:
Position Mode: Untested.
Altitude Mode: Untested.
Stabilized Mode: Untested
Mission Plan Mode (Automated): Untested
RTL: Untested

Notes:
The vehicle did not have GPS Lock and was not able to be flown.

Screenshot of Issue

Log:

Master: https://review.px4.io/plot_app?log=20ddbd80-9dc5-4702-9ef2-6ac583c01450

Tested on Pixhawk Pro v4:

Modes Tested:

Position Mode: Good.
Altitude Mode: Good.
Stabilized Mode: Good.
Mission Plan Mode (Automated): Good.
RTL: Good.

Procedure:
Arm and Take off in position mode, after flying for approximately one minute, switched to altitude then stabilized mode proceed to switch to mission plan mode then make sure that vehicle follows all waypoints as shown in QGC, once completed all waypoint activate RTL and see landing behaviour.

Notes:
No issues noted, good flight in general.

Logs:

Tested on NXP FMUK66 v3:

Modes Tested:

Position Mode: Good.
Altitude Mode: Good.
Stabilized Mode: Good.
Mission Plan Mode (Automated): Good.
RTL: Good.

Procedure:
Arm and Take off in position mode, after flying for approximately one minute, switched to altitude then stabilized mode proceed to switch to mission plan mode then make sure that vehicle follows all waypoints as shown in QGC, once completed all waypoint activate RTL and see landing behaviour.

Notes:
No issues noted, good flight in general.

Logs:

dagar · 2019-07-26T17:39:39Z

Added holybro/kakutef7 to the list.

src/drivers/boards/common/board_dma_alloc.h

dagar · 2019-10-07T13:30:21Z

This is currently a problem on stm32f7 with the DCACHE and DMA capable check where the SPI transfer buffer is being incorrectly considered non-DMA capable if it's not sized as a cache line multiple.

dakejahl · 2019-10-09T20:39:00Z

Why don't we just add a check to ensure we always allocate the right amount + padding to get us to the multiple of 4 bytes

The SPI transfer can still just use the actual size of the data it wants. i.e

	/*
	 * Fetch the full set of measurements from the MPU9250 in one pass
	 */
	MPUReport *mpu_report = (MPUReport *)_dma_data_buffer;

	if ((!_magnetometer_only || _mag.is_passthrough()) && _register_wait == 0) {
		if (_whoami == MPU_WHOAMI_9250 || _whoami == MPU_WHOAMI_6500) {
			if (OK != read_reg_range(MPUREG_INT_STATUS, MPU9250_HIGH_BUS_SPEED, _dma_data_buffer, sizeof(MPUReport))) {
				perf_end(_sample_perf);
				return;
			}

		}

		check_registers();

		if (check_duplicate(&mpu_report->accel_x[0])) {
			return;
		}
	}

dagar · 2019-10-13T17:38:21Z

Why don't we just add a check to ensure we always allocate the right amount + padding to get us to the multiple of 4 bytes

The SPI transfer can still just use the actual size of the data it wants. i.e

The problem is the DCACHE alignment check for the buffer end within NuttX.

On stm32f7 spi_exchange() is checking if the buffer is DMA capable.
https://github.com/PX4/NuttX/blob/d8da511082646d83a54c6905daca13f0a1a609f0/arch/arm/src/stm32f7/stm32_spi.c#L1681-L1682

The DMA capable check includes the buffer end.
https://github.com/PX4/NuttX/blob/d8da511082646d83a54c6905daca13f0a1a609f0/arch/arm/src/stm32f7/stm32_dma.c#L928

dakejahl · 2019-10-15T17:11:41Z

So all DMA spi transfers must be multiples of 4 bytes? That seems wrong.

dagar · 2019-10-15T17:27:48Z

So all DMA spi transfers must be multiples of 4 bytes? That seems wrong.

Yes, the check in its current form is bogus.

The memory must be DMA capable (not CCM on F4 or DTCM on F7)
On F7 with DCACHE the buffer must be DCACHE aligned (32 bytes) both start and end (we now have writethrough disabled).

One idea David and I have discussed is to change the notion of "DMA capable" within NuttX, possibly similar to the DMA allocator we have in PX4. Then instead of the DMA capable checking memory range, etc it would be based on originating from this particular allocator.

stale · 2020-01-13T17:38:29Z

This issue has been automatically marked as stale because it has not had recent activity. Thank you for your contributions.

julianoes · 2020-01-23T15:57:39Z

I suppose I should leave these branches around 🤷‍♂️.

dagar added the Admin: Enhancement (improvement) 💡 label Jul 9, 2019

dagar added this to the Release v1.10.0 milestone Jul 9, 2019

dagar assigned dagar and davids5 Jul 9, 2019

dagar mentioned this pull request Jul 9, 2019

[WIP]: start enabling SPI DMA per board (IMUs) #12248

Closed

18 tasks

dagar force-pushed the pr-imu_dma branch 2 times, most recently from 17eb2b7 to 4c209bd Compare July 9, 2019 02:19

dagar requested a review from davids5 July 9, 2019 02:31

dagar changed the title ~~[WIP]: start enabling SPI DMA per board (IMUs)~~ NuttX enable SPI DMA per board Jul 9, 2019

dagar marked this pull request as ready for review July 9, 2019 02:32

dagar added the NuttX (OS) label Jul 9, 2019

dagar requested a review from a team July 9, 2019 03:10

davids5 reviewed Jul 9, 2019

View reviewed changes

boards/aerotenna/ocpoc/src/CMakeLists.txt Outdated Show resolved Hide resolved

dagar force-pushed the pr-imu_dma branch 2 times, most recently from d6c6646 to eacae78 Compare July 9, 2019 19:44

dagar force-pushed the pr-imu_dma branch from eacae78 to 13634cf Compare July 10, 2019 00:37

weekly-digest bot mentioned this pull request Jul 14, 2019

Weekly Digest (7 July, 2019 - 14 July, 2019) #12474

Closed

dagar force-pushed the pr-imu_dma branch from 13634cf to 8654819 Compare July 26, 2019 17:32

weekly-digest bot mentioned this pull request Jul 28, 2019

Weekly Digest (21 July, 2019 - 28 July, 2019) #12569

Closed

dagar force-pushed the pr-imu_dma branch from 8654819 to ee73d00 Compare August 1, 2019 00:42

dakejahl reviewed Aug 2, 2019

View reviewed changes

src/drivers/boards/common/board_dma_alloc.h Show resolved Hide resolved

dagar force-pushed the pr-imu_dma branch from ee73d00 to e05128e Compare August 3, 2019 02:02

dagar force-pushed the pr-imu_dma branch 8 times, most recently from 6bc8c50 to b1d6225 Compare August 6, 2019 05:07

weekly-digest bot mentioned this pull request Aug 11, 2019

Weekly Digest (4 August, 2019 - 11 August, 2019) #12682

Closed

NuttX boards enable SPI DMA for IMUs

f2ed688

dagar force-pushed the pr-imu_dma branch from b1d6225 to f2ed688 Compare August 17, 2019 22:20

weekly-digest bot mentioned this pull request Aug 18, 2019

Weekly Digest (11 August, 2019 - 18 August, 2019) #12733

Closed

dagar mentioned this pull request Aug 18, 2019

[WIP] Dev dshot #12634

Closed

dagar mentioned this pull request Oct 6, 2019

ICM20602 and ICM20608g: new standalone driver (8 kHz gyro, 4 kHz accel) #13103

Merged

weekly-digest bot mentioned this pull request Oct 13, 2019

Weekly Digest (6 October, 2019 - 13 October, 2019) #13171

Closed

weekly-digest bot mentioned this pull request Oct 20, 2019

Weekly Digest (13 October, 2019 - 20 October, 2019) #13236

Closed

stale bot added the stale label Jan 13, 2020

dagar closed this Jan 22, 2020

julianoes deleted the pr-imu_dma branch January 23, 2020 15:57

julianoes restored the pr-imu_dma branch January 23, 2020 15:57

LorenzMeier deleted the pr-imu_dma branch January 18, 2021 14:45

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

NuttX enable SPI DMA per board #12436

NuttX enable SPI DMA per board #12436

dagar commented Jul 9, 2019 •

edited

Loading

davids5 left a comment

jorge789 commented Jul 9, 2019

dannyfpv commented Jul 9, 2019 •

edited by Tony3dr

Loading

Junkim3DR commented Jul 9, 2019 •

edited by Tony3dr

Loading

dagar commented Jul 26, 2019

dagar commented Oct 7, 2019

dakejahl commented Oct 9, 2019 •

edited

Loading

dagar commented Oct 13, 2019

dakejahl commented Oct 15, 2019

dagar commented Oct 15, 2019

stale bot commented Jan 13, 2020

julianoes commented Jan 23, 2020

NuttX enable SPI DMA per board #12436

NuttX enable SPI DMA per board #12436

Conversation

dagar commented Jul 9, 2019 • edited Loading

davids5 left a comment

Choose a reason for hiding this comment

jorge789 commented Jul 9, 2019

dannyfpv commented Jul 9, 2019 • edited by Tony3dr Loading

Junkim3DR commented Jul 9, 2019 • edited by Tony3dr Loading

Tested on Pixhawk 4 Mini V5:

Tested on Pixhawk Pro v4:

Tested on NXP FMUK66 v3:

dagar commented Jul 26, 2019

dagar commented Oct 7, 2019

dakejahl commented Oct 9, 2019 • edited Loading

dagar commented Oct 13, 2019

dakejahl commented Oct 15, 2019

dagar commented Oct 15, 2019

stale bot commented Jan 13, 2020

julianoes commented Jan 23, 2020

dagar commented Jul 9, 2019 •

edited

Loading

dannyfpv commented Jul 9, 2019 •

edited by Tony3dr

Loading

Junkim3DR commented Jul 9, 2019 •

edited by Tony3dr

Loading

dakejahl commented Oct 9, 2019 •

edited

Loading