host: unify component for all DMA types #1565

tlauda · 2019-06-13T13:28:44Z

Unifies host component for all DMA types.
We don't need to branch on CONFIG_DMA_GW anymore.

Closes #1052.

Signed-off-by: Tomasz Lauda tomasz.lauda@linux.intel.com

DW-DMA should ignore copy call if flags include DMA_COPY_PRELOAD. Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>

Unifies host_copy implementation for all types of DMA. Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>

Ignores request for HDA-DMA configuration if the channel is in active state. Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>

Checks if dma_sg_elem_array buffer is allocated before performing cache writeback or invalidation. Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>

Changes host_buffer component ops to be a subtype of component set_attribute ops. This change allows to unify host implementation across different types of DMA. Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>

Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>

Removes period_bytes field from host_data struct. This field is only used in host_params, so we don't need it globally. Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>

Adds support for one shot copy to DW-DMA. One shot means that transfer isn't circular, so every copy is basically just restart of DMA engine. Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>

dbaluta · 2019-06-13T14:32:41Z

src/audio/host.c

-#if !CONFIG_DMA_GW
 	struct hc_buf host;
+#if !CONFIG_DMA_GW
 	struct hc_buf local;


Here we still branch on CONFIG_DMA_GW. Is this as intended? Because the PR description says we no longer need to branch on this config option.

@tlauda I now see the last patch where the config is removed. So, it looks good to me.

lgirdwood

@mengdonglin @keqiaozhang can this be tested on legacy.

jajanusz

good job

jajanusz · 2019-06-13T20:48:57Z

src/audio/host.c

Codecheck complains that source_elem can be null here.

jajanusz · 2019-06-13T20:49:43Z

src/audio/host.c

Codecheck complains that sink_elem can be null here.

dbaluta · 2019-06-14T05:53:14Z

@tlauda We should be careful to run all regression test we have! Other than that, excellent work!

jajanusz · 2019-06-17T10:53:44Z

@tlauda ping, UT and CodeCheck didn't pass

Unifies host component for all DMA types. We don't need to branch on CONFIG_DMA_GW anymore. Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>

tlauda · 2019-06-24T06:37:06Z

@mengdonglin @keqiaozhang It's been over a week. Can I assume that it's not breaking legacy platforms?

lgirdwood · 2019-06-24T16:16:51Z

@tlauda I have a feeling there has been some national holidays for @mengdonglin and @keqiaozhang last week but some folks seem back today.

Jiangxinx · 2019-06-25T07:58:54Z

I have done sanity test for this PR on apl and byt platform.

env
kernel (topic/sof-dev): 0087d4a
topology: sof-apl-pcm512x.tplg & sof-apl-nocodec.tplg & sof-byt-rt5651.tplg & sof-byt-nocodec.tplg

Test result

test case	byt nocodec	byt rt5651	apl up2 nocodec	apl up2 nocodec(xcc)	apl up2 PCM512x	apl up2 PCM512x(xcc)
repeat 3 times playback over headset	Passed	Passed	Passed	Passed	Passed	Passed
Pause/Resume playback over headset - 10 times	Failed	Failed	Passed	Passed	Passed	Passed
Play an audio for 3 minutes	Passed	Passed	Passed	Passed	Passed	Passed
Change volume during playback.	Passed	Passed	Passed	Passed	Passed	Passed
aplay - playback over HDMI (hw:0, 2)	N/A	N/A	N/A	N/A	Passed	Passed
stop/resume playback over HDMI	N/A	N/A	N/A	N/A	Passed	Passed
repeat 3 times playback over HDMI	N/A	N/A	N/A	N/A	Passed	Passed
Pause/Resume playback over HDMI	N/A	N/A	N/A	N/A	Passed	Passed
Play an audio for 3 minutes over HDMI	N/A	N/A	N/A	N/A	Passed	Passed
Change volume during playback over HDMI.	N/A	N/A	N/A	N/A	Passed	Passed
arecord - capture over headset(hw:0,0) (HDA Analog)	Passed	Passed	Passed	Passed	N/A	N/A
Change volume during capture (hw:0,0).	Passed	Passed	Passed	Passed	N/A	N/A
Pause/Resume capture (hw:0,0)	Failed (#1578)	Failed (#1578)	Passed	Passed	N/A	N/A
Capture a audio for 5 minutes via HDA Analog	Passed	Passed	Passed	Passed	N/A	N/A
repeat 3 times capture	Passed	Passed	Passed	Passed	N/A	N/A
Stop/Resume capture	Passed	Passed	Passed	Passed	N/A	N/A
arecord - capture over DMIC32 (hw:0,6)	N/A	N/A	Passed	Passed	N/A	N/A
Change volume during capture over DMIC32 (hw:0,6) .	N/A	N/A	Passed	Passed	N/A	N/A
Pause/Resume capture over DMIC32 (hw:0,6)	N/A	N/A	Passed	Passed	N/A	N/A
Capture a audio for 5 minutes over DMIC32 (hw:0,6)	N/A	N/A	Passed	Passed	N/A	N/A
repeat 3 times capture over DMIC32 (hw:0,6)	N/A	N/A	Passed	Passed	N/A	N/A
Stop/Resume capture over DMIC32 (hw:0,6)	N/A	N/A	Passed	Passed	N/A	N/A
Do playback and capture at same time	Passed	Passed	Passed	Passed	N/A	N/A
2 aplay_Analogue + Media Playback	Passed	Passed	Failed (#1526)	Failed (#1526)	Failed (#1526)	Failed (#1526)
2 aplay_Analogue + HDMI	N/A	N/A	N/A	N/A	Passed	Passed
Check the audio channel via speaker-test	Passed	Passed	Passed	Passed	Passed	Passed
Check the runtime PM status	N/A	N/A	Passed	Passed	Passed	Passed
Modules reload	Passed	Passed	Passed	Passed	Passed	Passed
Modules reload - 100 times	Passed	Passed	Passed	Passed	Passed	Passed

tlauda · 2019-06-25T08:38:28Z

@Jiangxinx To sum up, do we have any regression or not?

Jiangxinx · 2019-06-25T08:56:44Z

Sorry to this mission: This pr have not another regression issues
Pause/Resume playback over headset - 10 times can be also reproduced on daily version.

tlauda added 8 commits June 13, 2019 14:38

dw-dma: do nothing on preload

f6ac097

DW-DMA should ignore copy call if flags include DMA_COPY_PRELOAD. Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>

host: unify host_copy for all DMAs

dadc5a5

Unifies host_copy implementation for all types of DMA. Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>

hda-dma: do not configure DMA during transfer

d918d6f

Ignores request for HDA-DMA configuration if the channel is in active state. Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>

dma: check if sg exists before cache operations

89b4737

Checks if dma_sg_elem_array buffer is allocated before performing cache writeback or invalidation. Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>

host: make host_buffer call a part of set_attribute

f59a06e

Changes host_buffer component ops to be a subtype of component set_attribute ops. This change allows to unify host implementation across different types of DMA. Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>

host: headers cleanup

1dc1d33

Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>

host: remove period_bytes from host_data

705a13a

Removes period_bytes field from host_data struct. This field is only used in host_params, so we don't need it globally. Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>

dw-dma: add support for one shot copy

57552c7

Adds support for one shot copy to DW-DMA. One shot means that transfer isn't circular, so every copy is basically just restart of DMA engine. Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>

tlauda requested review from bardliao, lbetlej, lgirdwood, lyakh, mmaka1, mrajwa, plbossart and xiulipan as code owners June 13, 2019 13:28

dbaluta reviewed Jun 13, 2019

View reviewed changes

lgirdwood approved these changes Jun 13, 2019

View reviewed changes

jajanusz approved these changes Jun 13, 2019

View reviewed changes

jajanusz reviewed Jun 13, 2019

View reviewed changes

src/audio/host.c Outdated

Copy link

Contributor

jajanusz Jun 13, 2019

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Codecheck complains that source_elem can be null here.

jajanusz reviewed Jun 13, 2019

View reviewed changes

src/audio/host.c Outdated

Copy link

Contributor

jajanusz Jun 13, 2019

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Codecheck complains that sink_elem can be null here.

dbaluta approved these changes Jun 14, 2019

View reviewed changes

host: unify component for all DMA types

2825d47

Unifies host component for all DMA types. We don't need to branch on CONFIG_DMA_GW anymore. Signed-off-by: Tomasz Lauda <tomasz.lauda@linux.intel.com>

tlauda force-pushed the topic/host-refactor2 branch from 30ab3f7 to 2825d47 Compare June 24, 2019 06:32

jajanusz merged commit fdd04e9 into thesofproject:master Jun 25, 2019

tlauda deleted the topic/host-refactor2 branch November 22, 2019 11:24

This was referenced Dec 12, 2019

[BUG][FEATURE] add snd_soc_cml_rt1011_rt5682 to kmod_script #2197

Closed

[BUG] need to add support for snd_soc_cml_rt1011_rt5682 module removal in kmod_scripts #2199

Closed

host: unify component for all DMA types #1565

host: unify component for all DMA types #1565

Uh oh!

Conversation

tlauda commented Jun 13, 2019

Uh oh!

dbaluta Jun 13, 2019

Choose a reason for hiding this comment

Uh oh!

dbaluta Jun 13, 2019

Choose a reason for hiding this comment

Uh oh!

lgirdwood left a comment

Choose a reason for hiding this comment

Uh oh!

jajanusz left a comment

Choose a reason for hiding this comment

Uh oh!

jajanusz Jun 13, 2019

Choose a reason for hiding this comment

Uh oh!

jajanusz Jun 13, 2019

Choose a reason for hiding this comment

Uh oh!

dbaluta commented Jun 14, 2019

Uh oh!

jajanusz commented Jun 17, 2019

Uh oh!

tlauda commented Jun 24, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

lgirdwood commented Jun 24, 2019

Uh oh!

Jiangxinx commented Jun 25, 2019

Uh oh!

tlauda commented Jun 25, 2019

Uh oh!

Jiangxinx commented Jun 25, 2019

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

tlauda commented Jun 24, 2019 •

edited

Loading