Re: [PATCH v2 4/7] drm/i915: Try harder to get best m, n, p values with minimal error

[Jani Nikula <jani.nikula@intel.com>]

On Sat, 09 Nov 2013, Shobhit Kumar <shobhit.kumar@intel.com> wrote:
> Basically check for both +ive and -ive deviation from target clock and
> pick the one with minimal error. If we get a direct match, break from
> loop to acheive some optimization.
>
> Signed-off-by: Vijayakumar Balakrishnan <vijayakumar.balakrishnan@intel.com>
> Signed-off-by: Shobhit Kumar <shobhit.kumar@intel.com>
> ---
>  drivers/gpu/drm/i915/intel_dsi_pll.c |   26 ++++++++++++++++++--------
>  1 file changed, 18 insertions(+), 8 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/intel_dsi_pll.c b/drivers/gpu/drm/i915/intel_dsi_pll.c
> index 9f3e6b0..16bc6b2 100644
> --- a/drivers/gpu/drm/i915/intel_dsi_pll.c
> +++ b/drivers/gpu/drm/i915/intel_dsi_pll.c
> @@ -243,22 +243,32 @@ static int dsi_calc_mnp(u32 dsi_clk, struct dsi_mnp *dsi_mnp)
>  	ref_clk = 25000;
>  	target_dsi_clk = dsi_clk;
>  	error = 0xFFFFFFFF;
> +	tmp_error = 0xFFFFFFFF;
>  	calc_m = 0;
>  	calc_p = 0;
>  
>  	for (m = 62; m <= 92; m++) {
>  		for (p = 2; p <= 6; p++) {
> -
> +			/* Find the optimal m and p divisors
> +			with minimal error +/- the required clock */
>  			calc_dsi_clk = (m * ref_clk) / p;
> -			if (calc_dsi_clk >= target_dsi_clk) {
> -				tmp_error = calc_dsi_clk - target_dsi_clk;
> -				if (tmp_error < error) {
> -					error = tmp_error;
> -					calc_m = m;
> -					calc_p = p;
> -				}
> +			if (calc_dsi_clk == target_dsi_clk) {
> +				calc_m = m;
> +				calc_p = p;
> +				error = 0;
> +				break;
> +			} else
> +				tmp_error = abs(target_dsi_clk - calc_dsi_clk);

This is fragile, and only works because of the kernel's implementation
of abs(). The substraction should be done with signed values.

BR,
Jani.

> +
> +			if (tmp_error < error) {
> +				error = tmp_error;
> +				calc_m = m;
> +				calc_p = p;
>  			}
>  		}
> +
> +		if (error == 0)
> +			break;
>  	}
>  
>  	m_seed = lfsr_converts[calc_m - 62];
> -- 
> 1.7.9.5
>
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx

-- 
Jani Nikula, Intel Open Source Technology Center