On 2022-10-11 14:11, Rodrigo Siqueira wrote:
> Change commit sequence might impact multiple ASICs; for this reason, it
> is better to run it only on DCN32x for the first set of validation.
> After we get more confident about this change, we should expand it to
> all ASICs.
>
> Cc: Nicholas Kazlauskas <[email protected]>
> Cc: Harry Wentland <[email protected]>
> Signed-off-by: Rodrigo Siqueira <[email protected]>
> Co-developed-by: Aurabindo Pillai <[email protected]>
> Signed-off-by: Aurabindo Pillai <[email protected]>
> ---
> drivers/gpu/drm/amd/display/dc/core/dc.c | 16 ++++++++++++++++
> 1 file changed, 16 insertions(+)
>
> diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c
> b/drivers/gpu/drm/amd/display/dc/core/dc.c
> index ad3aafb73cba..21d20900cfe3 100644
> --- a/drivers/gpu/drm/amd/display/dc/core/dc.c
> +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c
> @@ -1913,6 +1913,13 @@ bool dc_commit_state(struct dc *dc, struct dc_state
> *context)
> enum dc_status result = DC_ERROR_UNEXPECTED;
> int i;
>
> + /* TODO: Enable the new commit sequence for all ASICs */
> + if (dc->ctx->asic_id.chip_family == AMDGPU_FAMILY_GC_11_0_0 ||
> + dc->ctx->asic_id.chip_family == AMDGPU_FAMILY_GC_11_0_1) {
Why not check that dc_ctx->dce_version == DCN_VERSION_3_2?
I don't like using the family ID in DC unless we have a good
reason to.
> + result = dc_commit_streams(dc, context->streams,
> context->stream_count);
Wasn't this function initially intended as a crutch for other OSes
that couldn't use dc_commit_state directly for some reasons?
I'm not sure this change is moving into the right direction, at least
not without a much more detailed explanation of why dc_commit_streams
and dc_update_planes_and_streams are the better solution.
Harry
> + return result == DC_OK;
> + }
> +
> if (!context_changed(dc, context))
> return DC_OK;
>
> @@ -3898,6 +3905,15 @@ void dc_commit_updates_for_stream(struct dc *dc,
> struct dc_context *dc_ctx = dc->ctx;
> int i, j;
>
> + /* TODO: Enable the new commit sequence for all ASICs */
> + if (dc->ctx->asic_id.chip_family == AMDGPU_FAMILY_GC_11_0_0 ||
> + dc->ctx->asic_id.chip_family == AMDGPU_FAMILY_GC_11_0_1) {
> + dc_update_planes_and_stream(dc, srf_updates,
> + surface_count, stream,
> + stream_update);
> + return;
> + }
> +
> stream_status = dc_stream_get_status(stream);
> context = dc->current_state;
>
