audio: add aligned limits for component copy function

[andrula-song]

add comp_get_copy_limits_with_lock_aligned API to meet requirement
of some xtensa intrinsics.

Signed-off-by: Andrula Song [email protected]

[lgirdwood]

@singalsu fyi, lets optimise all the stream API calls after this is merged.

[lgirdwood]

@andrula-song @singalsu it would also be good to remove the stream API calls today that are not used as a first step and then align the remaining calls to align on frame or SIMD frames (without the divisions).

[andrula-song]

the frame_aligned set in prepare function would be reset somewhere. Need some time to analyze

[andrula-song]

some error should be nothing about code i commited, trigger test again

[singalsu]

Looks good to me, thanks!

[andrula-song]

since the volume hifi3 implementation version needs 8 bytes align, but the generic version or other components can have different align requirement

@cujomalainey I've answered the question before I close it , but it is pending. Sorry to close too fast.

[andrula-song]

it seems like the codestyle management has become more stringent，
WARNING: Possible repeated word: 'sink'
#262: FILE: src/include/sof/audio/component.h:858:

• • @param[in] sink Sink buffer
    such warings appeared this time but never seen last time.
    and there would be conflict between local commit and git check of the space arround *
    local:
    ERROR: need consistent spacing around '*' (ctx:WxV)
    Wrong core-isa.h is included using gcc #9: FILE: src/audio/component.c:163:
    +void comp_get_copy_limits_frame_aligned(struct comp_buffer __sparse_cache *source,
    github:
    ERROR: "foo * bar" should be "foo *bar"
    apl must support 8-channel, if not, the capture can not work on this platform. #33: FILE: src/audio/component.c:163:
    +void comp_get_copy_limits_frame_aligned(struct comp_buffer __sparse_cache * source,

                                                                       ^
    

[cujomalainey]

since the volume hifi3 implementation version needs 8 bytes align, but the generic version or other components can have different align requirement

@cujomalainey I've answered the question before I close it , but it is pending. Sorry to close too fast.

Ah you were tricked by githubs' review system, i think you have to submit your review first to create the comment then resolve it :)

[marc-hb]

ERROR: "foo * bar" should be "foo *bar"

This should be fixed on your next push thanks to my #5806 (no need to rebase)

Like most linters, checkpatch is not always right, you don't always have to fix every single warning.

Checkpatch is right most of the time, you must TRY to fix every warning.

https://www.kernel.org/doc/html/latest/dev-tools/checkpatch.html

[marc-hb]

ERROR: "foo * bar" should be "foo *bar"
struct comp_buffer __sparse_cache * source

This should be fixed on your next push thanks to my > #5806 (no need to rebase)

I got confused sorry: you must change this code as told.

[lgirdwood]

@wszypelt dont think this should impact the DMIC test, I will rerun CI again to be sure.

[lgirdwood]

SOFCI TEST

[andrula-song]

I recall the commit 08c234b passed all the test cases and Internal Intel CI System/merge/build . but these two commit just changed the code comment and Kconfig(remove the useless item), I don't think modify code comment would lead to those failed cases，trigger test again.

[andrula-song]

SOFCI TEST

[andrula-song]

@wszypelt can you help to check Internal Intel CI System/merge/build test ? I found that many PRs failed this item today.

[andrula-song]

in fact wondering, we expect these functions to also be useful for other components in the future, not only for volume, right?
@lyakh yes, so put this function in audio_stream.h. So other components can call this function with their own byte_align & frame_align_req as in volume_set_alignment function which is called by volume_prepare function.

[wszypelt]

@andrula-song @lgirdwood Of course there was a problem with one machine, it should work fine now. Tests rerun in progress

[andrula-song]

-->Is there some reason for frame_align_req to be inside the #if #else?
@marc-hb Since the byte_align and frame_align_req can be different between these two versions, even though now they are the same, but I think it is better to write it twice.

[andrula-song]

@wszypelt can you help to check Internal Intel CI System/merge/build test again? I don't think the modification would lead to such error, thanks.

[marc-hb]

@wszypelt can you help to check Internal Intel CI System/merge/build test again?

Always provide direct links because they keep changing:
https://sof-ci.01.org/sof-pr-viewer/#/build/PR5266/build9358965

[andrula-song]

@lgirdwood Can you help to review this PR? Thanks.