Re: [RFC v2 PATCH 02/14] drm/exynos: dsi: delay setting clocks after reset

[Andrzej Hajda <a.hajda@samsung.com>]

Hi YoungJun,

On 04/21/2014 02:28 PM, YoungJun Cho wrote:
> Some phy control registers are not kept after software reset.
> So this patch makes the clocks containing phy control to be set
> after software reset.
>
> Signed-off-by: YoungJun Cho <yj44.cho@samsung.com>
> Acked-by: Inki Dae <inki.dae@samsung.com>
> Acked-by: Kyungmin Park <kyungmin.park@samsung.com>
> ---
>  drivers/gpu/drm/exynos/exynos_drm_dsi.c |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/exynos/exynos_drm_dsi.c b/drivers/gpu/drm/exynos/exynos_drm_dsi.c
> index 956e5f3..2cf1f0b 100644
> --- a/drivers/gpu/drm/exynos/exynos_drm_dsi.c
> +++ b/drivers/gpu/drm/exynos/exynos_drm_dsi.c
> @@ -946,10 +946,10 @@ static irqreturn_t exynos_dsi_irq(int irq, void *dev_id)
>  
>  static int exynos_dsi_init(struct exynos_dsi *dsi)
>  {
> -	exynos_dsi_enable_clock(dsi);
>  	exynos_dsi_reset(dsi);
>  	enable_irq(dsi->irq);
>  	exynos_dsi_wait_for_reset(dsi);
> +	exynos_dsi_enable_clock(dsi);
>  	exynos_dsi_init_link(dsi);
>  
>  	return 0;

I have commented it in the previous version of the patchset. I repeat it
again.
This is a regression, on exynos4210-trats I have 'timeout waiting for
reset' message after dpms off, on.

I will comment your previous answer here to make the discussion easier:
> As I mentioned in description, it came from phy control registers.
> Fortunately Exynos4 SoCs are safe, but the DSIM_PHYCTRL_REG,
> DSIM_PHYTIMING_REG, DSIM_PHYTIMING1_REG and DSIM_PHYTIMING2_REG are
> affected which are used in exynos_dsi_set_pll() for Exynos5 SoCs.
>
> So this patch is required for Exynos5 SoCs.

In the moment this patch is applied exynos_dsi_set_pll do not touch phy
registers you have mentioned.
Your change would be more clear if it will be merged together with the
patch adding PHYCTRL settings.

Anyway, solution is simple - please set PHY registers after reset and
configure clocks before reset to avoid
reset timeouts, is there any reason to not do it this way?

Regards
Andrzej