Re: [PATCH v3,3/3] drm/mediatek: dp: Add the audio divider to mtk_dp_data struct

From: AngeloGioacchino Del Regno
Date: Thu Jul 20 2023 - 08:49:37 EST


Il 20/07/23 14:29, Alexandre Mergnat ha scritto:


On 20/07/2023 14:08, AngeloGioacchino Del Regno wrote:
Il 20/07/23 14:07, Alexandre Mergnat ha scritto:


On 20/07/2023 13:54, AngeloGioacchino Del Regno wrote:
Il 20/07/23 12:14, Alexandre Mergnat ha scritto:


On 20/07/2023 10:26, Shuijing Li wrote:
Due to the difference of HW, different dividers need to be set.

Signed-off-by: Shuijing Li <shuijing.li@xxxxxxxxxxxx>
Signed-off-by: Jitao Shi <jitao.shi@xxxxxxxxxxxx>
---
Changes in v3:
Separate these two things into two different patches.
per suggestion from the previous thread:
https://lore.kernel.org/lkml/e2ad22bcba31797f38a12a488d4246a01bf0cb2e.camel@xxxxxxxxxxxx/
Changes in v2:
- change the variables' name to be more descriptive
- add a comment that describes the function of mtk_dp_audio_sample_arrange
- reduce indentation by doing the inverse check
- add a definition of some bits
- add support for mediatek, mt8188-edp-tx
per suggestion from the previous thread:
https://lore.kernel.org/lkml/ac0fcec9-a2fe-06cc-c727-189ef7babe9c@xxxxxxxxxxxxx/
---
  drivers/gpu/drm/mediatek/mtk_dp.c     | 7 ++++++-
  drivers/gpu/drm/mediatek/mtk_dp_reg.h | 1 +
  2 files changed, 7 insertions(+), 1 deletion(-)

...
b/drivers/gpu/drm/mediatek/mtk_dp_reg.h
index f38d6ff12afe..6d7f0405867e 100644
--- a/drivers/gpu/drm/mediatek/mtk_dp_reg.h
+++ b/drivers/gpu/drm/mediatek/mtk_dp_reg.h
@@ -162,6 +162,7 @@
  #define AUDIO_M_CODE_MULT_DIV_SEL_DP_ENC0_P0_MUL_2    (1 << 8)
  #define AUDIO_M_CODE_MULT_DIV_SEL_DP_ENC0_P0_MUL_4    (2 << 8)
  #define AUDIO_M_CODE_MULT_DIV_SEL_DP_ENC0_P0_MUL_8    (3 << 8)
+#define MT8188_AUDIO_M_CODE_MULT_DIV_SEL_DP_ENC0_P0_DIV_2    (4 << 8)

IMO, it's a bit weird to have SoC specific define in the generic header.
Are you sure this bit is only available for MT8188 ?


Eh, the P0_DIV2 bit is 5<<8 for MT8195, while for 8188 it's 4<<8, clearly :-)


Ok then, to avoid this kind of issue for other SoCs in the future, is that make sense for you to do a SoC specific header file beside the generic one?


For just one definition? That's a bit overkill :-)


You're right, but we must start somewhere ^^, and show the proper way for future patches. Actually, I gave my Reviewed-by because it's only one definition. This will be fixed later (I hope).


I'm confident that *if* and when "a bunch" of SoC-specific definitions will
appear, those will be splitted in different headers. :-)

  #define AUDIO_M_CODE_MULT_DIV_SEL_DP_ENC0_P0_DIV_2    (5 << 8)
  #define AUDIO_M_CODE_MULT_DIV_SEL_DP_ENC0_P0_DIV_4    (6 << 8)
  #define AUDIO_M_CODE_MULT_DIV_SEL_DP_ENC0_P0_DIV_8    (7 << 8)

Reviewed-by: Alexandre Mergnat <amergnat@xxxxxxxxxxxx>