[Freedreno] [PATCH v2 09/28] drm/msm/dsi: move min/max PLL rate to phy config

abhinavk at codeaurora.org abhinavk at codeaurora.org
Fri Mar 26 17:53:09 UTC 2021


On 2021-03-24 08:18, Dmitry Baryshkov wrote:
> Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov at linaro.org>
Reviewed-by: Abhinav Kumar <abhinavk at codeaurora.org>
> ---
>  drivers/gpu/drm/msm/dsi/phy/dsi_phy.h           |  3 +++
>  drivers/gpu/drm/msm/dsi/phy/dsi_phy_10nm.c      |  6 ++++--
>  drivers/gpu/drm/msm/dsi/phy/dsi_phy_14nm.c      |  6 ++++--
>  drivers/gpu/drm/msm/dsi/phy/dsi_phy_28nm.c      |  8 ++++++--
>  drivers/gpu/drm/msm/dsi/phy/dsi_phy_28nm_8960.c |  4 ++--
>  drivers/gpu/drm/msm/dsi/phy/dsi_phy_7nm.c       | 12 ++++--------
>  drivers/gpu/drm/msm/dsi/phy/dsi_pll.c           |  8 ++++----
>  drivers/gpu/drm/msm/dsi/phy/dsi_pll.h           |  3 ---
>  8 files changed, 27 insertions(+), 23 deletions(-)
> 
> diff --git a/drivers/gpu/drm/msm/dsi/phy/dsi_phy.h
> b/drivers/gpu/drm/msm/dsi/phy/dsi_phy.h
> index 39abb86446f9..000e4207dabc 100644
> --- a/drivers/gpu/drm/msm/dsi/phy/dsi_phy.h
> +++ b/drivers/gpu/drm/msm/dsi/phy/dsi_phy.h
> @@ -38,6 +38,9 @@ struct msm_dsi_phy_cfg {
>  	struct msm_dsi_phy_ops ops;
>  	const struct msm_dsi_pll_ops pll_ops;
> 
> +	unsigned long	min_pll_rate;
> +	unsigned long	max_pll_rate;
> +
>  	/*
>  	 * Each cell {phy_id, pll_id} of the truth table indicates
>  	 * if the source PLL selection bit should be set for each PHY.
> diff --git a/drivers/gpu/drm/msm/dsi/phy/dsi_phy_10nm.c
> b/drivers/gpu/drm/msm/dsi/phy/dsi_phy_10nm.c
> index dc8ccc994759..5f9d0cfc4e03 100644
> --- a/drivers/gpu/drm/msm/dsi/phy/dsi_phy_10nm.c
> +++ b/drivers/gpu/drm/msm/dsi/phy/dsi_phy_10nm.c
> @@ -864,8 +864,6 @@ static int dsi_pll_10nm_init(struct msm_dsi_phy 
> *phy)
>  	spin_lock_init(&pll_10nm->postdiv_lock);
> 
>  	pll = &pll_10nm->base;
> -	pll->min_rate = 1000000000UL;
> -	pll->max_rate = 3500000000UL;
>  	pll->cfg = phy->cfg;
> 
>  	pll_10nm->vco_delay = 1;
> @@ -1113,6 +1111,8 @@ const struct msm_dsi_phy_cfg dsi_phy_10nm_cfgs = 
> {
>  		.restore_state = dsi_pll_10nm_restore_state,
>  		.set_usecase = dsi_pll_10nm_set_usecase,
>  	},
> +	.min_pll_rate = 1000000000UL,
> +	.max_pll_rate = 3500000000UL,
>  	.io_start = { 0xae94400, 0xae96400 },
>  	.num_dsi_phy = 2,
>  };
> @@ -1138,6 +1138,8 @@ const struct msm_dsi_phy_cfg 
> dsi_phy_10nm_8998_cfgs = {
>  		.restore_state = dsi_pll_10nm_restore_state,
>  		.set_usecase = dsi_pll_10nm_set_usecase,
>  	},
> +	.min_pll_rate = 1000000000UL,
> +	.max_pll_rate = 3500000000UL,
>  	.io_start = { 0xc994400, 0xc996400 },
>  	.num_dsi_phy = 2,
>  	.quirks = DSI_PHY_10NM_QUIRK_OLD_TIMINGS,
> diff --git a/drivers/gpu/drm/msm/dsi/phy/dsi_phy_14nm.c
> b/drivers/gpu/drm/msm/dsi/phy/dsi_phy_14nm.c
> index d78f846cf8e4..8e4528301e5d 100644
> --- a/drivers/gpu/drm/msm/dsi/phy/dsi_phy_14nm.c
> +++ b/drivers/gpu/drm/msm/dsi/phy/dsi_phy_14nm.c
> @@ -1078,8 +1078,6 @@ static int dsi_pll_14nm_init(struct msm_dsi_phy 
> *phy)
>  	spin_lock_init(&pll_14nm->postdiv_lock);
> 
>  	pll = &pll_14nm->base;
> -	pll->min_rate = VCO_MIN_RATE;
> -	pll->max_rate = VCO_MAX_RATE;
>  	pll->cfg = phy->cfg;
> 
>  	pll_14nm->vco_delay = 1;
> @@ -1237,6 +1235,8 @@ const struct msm_dsi_phy_cfg dsi_phy_14nm_cfgs = 
> {
>  		.disable_seq = dsi_pll_14nm_disable_seq,
>  		.enable_seq = dsi_pll_14nm_enable_seq,
>  	},
> +	.min_pll_rate = VCO_MIN_RATE,
> +	.max_pll_rate = VCO_MAX_RATE,
>  	.io_start = { 0x994400, 0x996400 },
>  	.num_dsi_phy = 2,
>  };
> @@ -1264,6 +1264,8 @@ const struct msm_dsi_phy_cfg 
> dsi_phy_14nm_660_cfgs = {
>  		.disable_seq = dsi_pll_14nm_disable_seq,
>  		.enable_seq = dsi_pll_14nm_enable_seq,
>  	},
> +	.min_pll_rate = VCO_MIN_RATE,
> +	.max_pll_rate = VCO_MAX_RATE,
>  	.io_start = { 0xc994400, 0xc996000 },
>  	.num_dsi_phy = 2,
>  };
> diff --git a/drivers/gpu/drm/msm/dsi/phy/dsi_phy_28nm.c
> b/drivers/gpu/drm/msm/dsi/phy/dsi_phy_28nm.c
> index bb33261d606d..d267b25e5da0 100644
> --- a/drivers/gpu/drm/msm/dsi/phy/dsi_phy_28nm.c
> +++ b/drivers/gpu/drm/msm/dsi/phy/dsi_phy_28nm.c
> @@ -625,8 +625,6 @@ static int dsi_pll_28nm_init(struct msm_dsi_phy 
> *phy)
>  	}
> 
>  	pll = &pll_28nm->base;
> -	pll->min_rate = VCO_MIN_RATE;
> -	pll->max_rate = VCO_MAX_RATE;
>  	if (phy->cfg->quirks & DSI_PHY_28NM_QUIRK_PHY_LP)
>  		pll_28nm->vco_delay = 1000;
>  	else
> @@ -811,6 +809,8 @@ const struct msm_dsi_phy_cfg dsi_phy_28nm_hpm_cfgs 
> = {
>  		.disable_seq = dsi_pll_28nm_disable_seq,
>  		.enable_seq = dsi_pll_28nm_enable_seq_hpm,
>  	},
> +	.min_pll_rate = VCO_MIN_RATE,
> +	.max_pll_rate = VCO_MAX_RATE,
>  	.io_start = { 0xfd922b00, 0xfd923100 },
>  	.num_dsi_phy = 2,
>  };
> @@ -837,6 +837,8 @@ const struct msm_dsi_phy_cfg 
> dsi_phy_28nm_hpm_famb_cfgs = {
>  		.disable_seq = dsi_pll_28nm_disable_seq,
>  		.enable_seq = dsi_pll_28nm_enable_seq_hpm,
>  	},
> +	.min_pll_rate = VCO_MIN_RATE,
> +	.max_pll_rate = VCO_MAX_RATE,
>  	.io_start = { 0x1a94400, 0x1a96400 },
>  	.num_dsi_phy = 2,
>  };
> @@ -863,6 +865,8 @@ const struct msm_dsi_phy_cfg dsi_phy_28nm_lp_cfgs = 
> {
>  		.disable_seq = dsi_pll_28nm_disable_seq,
>  		.enable_seq = dsi_pll_28nm_enable_seq_lp,
>  	},
> +	.min_pll_rate = VCO_MIN_RATE,
> +	.max_pll_rate = VCO_MAX_RATE,
>  	.io_start = { 0x1a98500 },
>  	.num_dsi_phy = 1,
>  	.quirks = DSI_PHY_28NM_QUIRK_PHY_LP,
> diff --git a/drivers/gpu/drm/msm/dsi/phy/dsi_phy_28nm_8960.c
> b/drivers/gpu/drm/msm/dsi/phy/dsi_phy_28nm_8960.c
> index 79b0842a8dc4..31e7910c6050 100644
> --- a/drivers/gpu/drm/msm/dsi/phy/dsi_phy_28nm_8960.c
> +++ b/drivers/gpu/drm/msm/dsi/phy/dsi_phy_28nm_8960.c
> @@ -508,8 +508,6 @@ static int dsi_pll_28nm_8960_init(struct 
> msm_dsi_phy *phy)
>  	}
> 
>  	pll = &pll_28nm->base;
> -	pll->min_rate = VCO_MIN_RATE;
> -	pll->max_rate = VCO_MAX_RATE;
> 
>  	pll->cfg = phy->cfg;
> 
> @@ -711,6 +709,8 @@ const struct msm_dsi_phy_cfg dsi_phy_28nm_8960_cfgs 
> = {
>  		.disable_seq = dsi_pll_28nm_disable_seq,
>  		.enable_seq = dsi_pll_28nm_enable_seq,
>  	},
> +	.min_pll_rate = VCO_MIN_RATE,
> +	.max_pll_rate = VCO_MAX_RATE,
>  	.io_start = { 0x4700300, 0x5800300 },
>  	.num_dsi_phy = 2,
>  };
> diff --git a/drivers/gpu/drm/msm/dsi/phy/dsi_phy_7nm.c
> b/drivers/gpu/drm/msm/dsi/phy/dsi_phy_7nm.c
> index 44ae495e8fca..4831d6769da7 100644
> --- a/drivers/gpu/drm/msm/dsi/phy/dsi_phy_7nm.c
> +++ b/drivers/gpu/drm/msm/dsi/phy/dsi_phy_7nm.c
> @@ -889,14 +889,6 @@ static int dsi_pll_7nm_init(struct msm_dsi_phy 
> *phy)
>  	spin_lock_init(&pll_7nm->postdiv_lock);
> 
>  	pll = &pll_7nm->base;
> -	pll->min_rate = 1000000000UL;
> -	pll->max_rate = 3500000000UL;
> -	if (phy->cfg->quirks & DSI_PHY_7NM_QUIRK_V4_1) {
> -		pll->min_rate = 600000000UL;
> -		pll->max_rate = (unsigned long)5000000000ULL;
> -		/* workaround for max rate overflowing on 32-bit builds: */
> -		pll->max_rate = max(pll->max_rate, 0xffffffffUL);
> -	}
>  	pll->cfg = phy->cfg;
> 
>  	pll_7nm->vco_delay = 1;
> @@ -1152,6 +1144,8 @@ const struct msm_dsi_phy_cfg dsi_phy_7nm_cfgs = {
>  		.restore_state = dsi_pll_7nm_restore_state,
>  		.set_usecase = dsi_pll_7nm_set_usecase,
>  	},
> +	.min_pll_rate = 600000000UL,
> +	.max_pll_rate = (5000000000ULL < ULONG_MAX) ? 5000000000ULL : 
> ULONG_MAX,
>  	.io_start = { 0xae94400, 0xae96400 },
>  	.num_dsi_phy = 2,
>  	.quirks = DSI_PHY_7NM_QUIRK_V4_1,
> @@ -1178,6 +1172,8 @@ const struct msm_dsi_phy_cfg 
> dsi_phy_7nm_8150_cfgs = {
>  		.restore_state = dsi_pll_7nm_restore_state,
>  		.set_usecase = dsi_pll_7nm_set_usecase,
>  	},
> +	.min_pll_rate = 1000000000UL,
> +	.max_pll_rate = 3500000000UL,
>  	.io_start = { 0xae94400, 0xae96400 },
>  	.num_dsi_phy = 2,
>  };
> diff --git a/drivers/gpu/drm/msm/dsi/phy/dsi_pll.c
> b/drivers/gpu/drm/msm/dsi/phy/dsi_pll.c
> index c7ff0eba0e8b..e607adffe001 100644
> --- a/drivers/gpu/drm/msm/dsi/phy/dsi_pll.c
> +++ b/drivers/gpu/drm/msm/dsi/phy/dsi_pll.c
> @@ -14,10 +14,10 @@ long msm_dsi_pll_helper_clk_round_rate(struct 
> clk_hw *hw,
>  {
>  	struct msm_dsi_pll *pll = hw_clk_to_pll(hw);
> 
> -	if      (rate < pll->min_rate)
> -		return  pll->min_rate;
> -	else if (rate > pll->max_rate)
> -		return  pll->max_rate;
> +	if      (rate < pll->cfg->min_pll_rate)
> +		return  pll->cfg->min_pll_rate;
> +	else if (rate > pll->cfg->max_pll_rate)
> +		return  pll->cfg->max_pll_rate;
>  	else
>  		return rate;
>  }
> diff --git a/drivers/gpu/drm/msm/dsi/phy/dsi_pll.h
> b/drivers/gpu/drm/msm/dsi/phy/dsi_pll.h
> index 4fa73fbcba52..8306911f8318 100644
> --- a/drivers/gpu/drm/msm/dsi/phy/dsi_pll.h
> +++ b/drivers/gpu/drm/msm/dsi/phy/dsi_pll.h
> @@ -18,9 +18,6 @@ struct msm_dsi_pll {
>  	bool		pll_on;
>  	bool		state_saved;
> 
> -	unsigned long	min_rate;
> -	unsigned long	max_rate;
> -
>  	const struct msm_dsi_phy_cfg *cfg;
>  };


More information about the Freedreno mailing list