dma: Scatter gather test #44041

teburd · 2022-03-21T17:36:47Z

Adds a scatter gather test against memory to memory transfers.

Based on #43984 so the updated designware driver can be built on both test cases in #43883

DesignWare DMA passes both tests, NXP's eDMA does not seem to pass the scatter gather test and its not clear as to why, maybe @hakehuang wouldn't mind taking a look?

Marked DNM until #43984 is merged

danieldegrasse · 2022-03-21T22:39:47Z

@hakehuang the changes to fsl_edma in MCUXpresso SDK 2.11 (#43826) and the updates to the EDMA driver in #42750 enable this test to pass on an RT1060.

@dleach02 @DerekSnell FYI- These look like the same issues we've been seeing with the EDMA driver, although I didn't investigate too closely (just tried updating the HAL and our EDMA driver)

teburd · 2022-03-21T22:49:37Z

@hakehuang the changes to fsl_edma in MCUXpresso SDK 2.11 (#43826) and the updates to the EDMA driver in #42750 enable this test to pass on an RT1060.

@dleach02 @DerekSnell FYI- These look like the same issues we've been seeing with the EDMA driver, although I didn't investigate too closely (just tried updating the HAL and our EDMA driver)

Thanks for taking a look at this @danieldegrasse and confirming that there are fixes required for this to work on eDMA. I've pushed a change to this PR to disable it for all platforms initially, in the updated designware DMA PR I've enabled it for the appropriate intel adsp platforms. When the NXP changes are merged we can enable it there as well.

hakehuang · 2022-03-22T15:39:38Z

@danieldegrasse, I port the hal fix to this PR, however it does not work at my side for rt1064 board, can you help to double check? maybe my port has problem, thanks in advance.

*** Booting Zephyr OS build zephyr-v3.0.0-1521-g0fba19714e60  ***
Running test suite dma_m2m_sg_test
===================================================================
START - test_dma_m2m_sg
DMA memory to memory transfer started on DMA_0
Preparing DMA Controller
dma block 0 block_size 64, source addr 80000000, dest addr 80000040
set next block pointer to 0x80000b8c
dma block 1 block_size 64, source addr 80000000, dest addr 80000080
Configuring the scatter-gather transfer on channel 0
Starting the transfer on channel 0 and waiting completion
callback status 1
I: channel 0 error status is 0x2

danieldegrasse · 2022-03-22T15:41:15Z

@hakehuang Did you pull in the updates to the EDMA driver in #42750? I saw errors like that when I only had updated the HAL.

hakehuang · 2022-03-22T16:04:55Z

@hakehuang Did you pull in the updates to the EDMA driver in #42750? I saw errors like that when I only had updated the HAL.

Thanks @danieldegrasse , there is one key update at

#ifdef CONFIG_HAS_MCUX_CACHE

#ifdef CONFIG_DMA_MCUX_USE_DTCM_FOR_DMA_DESCRIPTORS

#if DT_NODE_HAS_STATUS(DT_CHOSEN(zephyr_dtcm), okay)
#define EDMA_TCDPOOL_CACHE_ATTR __dtcm_noinit_section
#else /* DT_NODE_HAS_STATUS(DT_CHOSEN(zephyr_dtcm), okay) */
#error Selected DTCM for MCUX DMA descriptors but no DTCM section.
#endif /* DT_NODE_HAS_STATUS(DT_CHOSEN(zephyr_dtcm), okay) */

#elif defined(CONFIG_NOCACHE_MEMORY)
#define EDMA_TCDPOOL_CACHE_ATTR __nocache
#else
#error tcdpool could not be located in cacheable memory, a requirement for proper EDMA operation.
#endif /* CONFIG_DMA_MCUX_USE_DTCM_FOR_DMA_DESCRIPTORS */

#else /* CONFIG_HAS_MCUX_CACHE */

#define EDMA_TCDPOOL_CACHE_ATTR

#endif /* CONFIG_HAS_MCUX_CACHE */

static __aligned(32) EDMA_TCDPOOL_CACHE_ATTR edma_tcd_t
tcdpool[DT_INST_PROP(0, dma_channels)][CONFIG_DMA_TCD_QUEUE_SIZE];

which fixes my issues here

hakehuang

need wait for #42750 merged. others looks OK for me and test pass on my rt series board

laurenmurphyx64 · 2022-03-23T21:31:29Z

I don't think it matters a whole lot, especially considering you just got 2 approvals on this and updating the PR would dismiss them and put you back in review purgatory, but the copyrights for the new files should be 2022 instead of 2021. You might consider fixing the copyright comments in another little PR after this one gets merged.

tests/drivers/dma/scatter_gather/src/test_dma_sg.c

teburd · 2022-03-24T14:32:05Z

@hakehuang please do take a look, should be good to merge now

Adds a scatter gather test against memory to memory transfers. Initially excludes all platforms as they are all failing. Signed-off-by: Tom Burdick <thomas.burdick@intel.com>

ceolin · 2022-03-25T17:21:29Z

tests/drivers/dma/scatter_gather/src/test_dma_sg.c

+	dma_cfg.user_data = (struct device *)dma;
+#else
+	dma_cfg.user_data = NULL;
+#endif /* CONFIG_DMAMUX_STM32 */


Is this conditional really necessary ? I don't see the callback using the data passsed here. The driver should not do anything with the user pointer as well.

I may have missed something though.

ceolin · 2022-03-25T17:25:05Z

tests/drivers/dma/scatter_gather/src/test_dma_sg.c

+		dma_block_cfgs[i].source_gather_en = 1U;
+		dma_block_cfgs[i].block_size = XFER_SIZE;
+		dma_block_cfgs[i].source_address = (uint32_t)(tx_data);
+		dma_block_cfgs[i].dest_address = (uint32_t)(rx_data[i]);


Can't these address be 64bit if CONFIG_DMA_64BIT is selected ? Btw, why don't we use uintptr_t for these address instead of having it conditionally set ? (not related with this particular patch, just a question)

ceolin

Minor comments, looks good to me.

teburd requested a review from nashif as a code owner March 21, 2022 17:36

teburd requested a review from hakehuang March 21, 2022 17:36

github-actions bot added the area: Tests Issues related to a particular existing or missing test label Mar 21, 2022

teburd force-pushed the dma_test_sg branch from ecee5d6 to 074c744 Compare March 21, 2022 17:42

teburd added the DNM This PR should not be merged (Do Not Merge) label Mar 21, 2022

teburd force-pushed the dma_test_sg branch from 074c744 to cc8b87b Compare March 21, 2022 19:50

teburd force-pushed the dma_test_sg branch 2 times, most recently from 32dd04e to 0fba197 Compare March 21, 2022 22:47

teburd self-assigned this Mar 21, 2022

teburd force-pushed the dma_test_sg branch from 0fba197 to 11ec4ac Compare March 22, 2022 22:57

hakehuang previously approved these changes Mar 23, 2022

View reviewed changes

teburd requested review from erwango, ceolin, peter-mitsis and laurenmurphyx64 March 23, 2022 15:12

erwango requested a review from FRASTM March 23, 2022 16:10

teburd mentioned this pull request Mar 23, 2022

dma: Update dma_dw from sof #43883

Merged

2 tasks

danieldegrasse previously approved these changes Mar 23, 2022

View reviewed changes

laurenmurphyx64 reviewed Mar 23, 2022

View reviewed changes

tests/drivers/dma/scatter_gather/src/test_dma_sg.c Outdated Show resolved Hide resolved

teburd dismissed stale reviews from danieldegrasse and hakehuang via 5307e72 March 23, 2022 22:14

teburd force-pushed the dma_test_sg branch from 11ec4ac to 5307e72 Compare March 23, 2022 22:14

danieldegrasse previously approved these changes Mar 23, 2022

View reviewed changes

laurenmurphyx64 previously approved these changes Mar 23, 2022

View reviewed changes

hakehuang self-requested a review March 24, 2022 03:03

teburd removed the DNM This PR should not be merged (Do Not Merge) label Mar 24, 2022

teburd dismissed stale reviews from laurenmurphyx64 and danieldegrasse via e9f84eb March 24, 2022 14:31

teburd force-pushed the dma_test_sg branch from 5307e72 to e9f84eb Compare March 24, 2022 14:31

teburd requested review from laurenmurphyx64 and danieldegrasse March 24, 2022 14:31

dma: Scatter gather test

e9f84eb

Adds a scatter gather test against memory to memory transfers. Initially excludes all platforms as they are all failing. Signed-off-by: Tom Burdick <thomas.burdick@intel.com>

hakehuang approved these changes Mar 24, 2022

View reviewed changes

danieldegrasse approved these changes Mar 24, 2022

View reviewed changes

nashif closed this Mar 25, 2022

nashif reopened this Mar 25, 2022

ceolin reviewed Mar 25, 2022

View reviewed changes

ceolin approved these changes Mar 25, 2022

View reviewed changes

carlescufi merged commit 2ec3222 into zephyrproject-rtos:main Mar 28, 2022

teburd deleted the dma_test_sg branch March 28, 2022 21:14

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

dma: Scatter gather test #44041

dma: Scatter gather test #44041

teburd commented Mar 21, 2022 •

edited

Loading

danieldegrasse commented Mar 21, 2022 •

edited

Loading

teburd commented Mar 21, 2022 •

edited

Loading

hakehuang commented Mar 22, 2022

danieldegrasse commented Mar 22, 2022

hakehuang commented Mar 22, 2022 •

edited

Loading

hakehuang left a comment

laurenmurphyx64 commented Mar 23, 2022 •

edited

Loading

teburd commented Mar 24, 2022

ceolin Mar 25, 2022

ceolin Mar 25, 2022

ceolin left a comment

dma: Scatter gather test #44041

dma: Scatter gather test #44041

Conversation

teburd commented Mar 21, 2022 • edited Loading

danieldegrasse commented Mar 21, 2022 • edited Loading

teburd commented Mar 21, 2022 • edited Loading

hakehuang commented Mar 22, 2022

danieldegrasse commented Mar 22, 2022

hakehuang commented Mar 22, 2022 • edited Loading

hakehuang left a comment

Choose a reason for hiding this comment

laurenmurphyx64 commented Mar 23, 2022 • edited Loading

teburd commented Mar 24, 2022

ceolin Mar 25, 2022

Choose a reason for hiding this comment

ceolin Mar 25, 2022

Choose a reason for hiding this comment

ceolin left a comment

Choose a reason for hiding this comment

teburd commented Mar 21, 2022 •

edited

Loading

danieldegrasse commented Mar 21, 2022 •

edited

Loading

teburd commented Mar 21, 2022 •

edited

Loading

hakehuang commented Mar 22, 2022 •

edited

Loading

laurenmurphyx64 commented Mar 23, 2022 •

edited

Loading