Re: [PATCH 52/89] drm/i915/gen9: Disable WM if corresponding latency is 0

["Ville Syrjälä" <ville.syrjala@linux.intel.com>]

On Thu, Sep 04, 2014 at 12:27:18PM +0100, Damien Lespiau wrote:
> From: Vandana Kannan <vandana.kannan@intel.com>
> 
> According to updated BSpec, If level 1 or any higher level has a value of 0x00,
> that level and any higher levels are unused and the associated watermark
> registers must not be enabled.
> 
> This patch checks for latency 0 for level >=1 and does not enable WM
> corresponding to level m | m>=n, if level n (n != 0) has a 0us latency.
> 
> v2: Satheesh's review comments
> 	- zero-out latency values (for all higher levels if latency of given
> 	level is zero ) in read_wm_latency() function itself
> 
> v3: removed redundant check as per Satheesh's observation.
> v4: rebase on top before merging (Damien)
> 
> Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
> Signed-off-by: Vandana Kannan <vandana.kannan@intel.com>
> Reviewed-by: Satheeshakrishna M <satheeshakrishna.m@intel.com> (v3)
> Cc: Satheeshakrishna M <satheeshakrishna.m@intel.com>
> ---
>  drivers/gpu/drm/i915/intel_pm.c | 14 ++++++++++++--
>  1 file changed, 12 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
> index 16ad008..fdf297f 100644
> --- a/drivers/gpu/drm/i915/intel_pm.c
> +++ b/drivers/gpu/drm/i915/intel_pm.c
> @@ -2253,7 +2253,7 @@ static void intel_read_wm_latency(struct drm_device *dev, uint16_t wm[8])
>  
>  	if (IS_GEN9(dev)) {
>  		uint32_t val;
> -		int ret;
> +		int ret, i;
>  		int level, max_level = ilk_wm_max_level(dev);
>  
>  		/* read the first set of memory latencies[0:3] */
> @@ -2305,12 +2305,22 @@ static void intel_read_wm_latency(struct drm_device *dev, uint16_t wm[8])
>  		 *   we always add 2us there.
>  		 *   - For levels >=1, punit returns 0us latency when they are
>  		 *   disabled, so we respect that and don't add 2us then
> +		 *
> +		 * Additionally, if a level n (n > 1) has a 0us latency, all
> +		 * levels m (m >= n) need to be disabled. We make sure to
> +		 * sanitize the values out of the punit to satisfy this
> +		 * requirement.
>  		 */
>  		wm[0] += 2;
>  		for (level = 1; level <= max_level; level++)
>  			if (wm[level] != 0)
>  				wm[level] += 2;
> +			else {
> +				for (i = level + 1; i <= max_level; i++)
> +					wm[i] = 0;
>  
> +				break;
> +			}

If we're going to be paranoid I think we should disable all higher WM
levels whose latency is lower than any of the lower levels. And I
think we'll want something like dev_priv->wm.max_wm_level instead of
relying on the zero latency tricks. That thing has been on my TODO
list since forever.

>  	} else if (IS_HASWELL(dev) || IS_BROADWELL(dev)) {
>  		uint64_t sskpd = I915_READ64(MCH_SSKPD);
>  
> @@ -3253,7 +3263,7 @@ static bool skl_compute_plane_wm(struct skl_pipe_wm_parameters *p,
>  	uint32_t method1, method2, plane_bytes_per_line;
>  	uint32_t result_bytes;
>  
> -	if (!p->active || !p_params->enabled) {
> +	if (mem_value == 0 || !p->active || !p_params->enabled) {
>  		*res_blocks = PLANE_WM_BLOCKS_DEFAULT;
>  		*res_lines = PLANE_WM_LINES_DEFAULT;
>  		return false;
> -- 
> 1.8.3.1
> 
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx

-- 
Ville Syrjälä
Intel OTC