diff mbox

drm/i915/skl+: Fix Watermark calculation for Broxton

Message ID 1442859078-28226-1-git-send-email-mahesh1.kumar@intel.com (mailing list archive)
State New, archived
Headers show

Commit Message

Kumar, Mahesh Sept. 21, 2015, 6:11 p.m. UTC 
In case of Y-Tiling, "plane_blocks_per_line" calculation is different
than X/None-Tiling case.
This patch corrects this calculation according to Bspec.
plane blocks per line = Plane memory format is Y tile ?
		ceiling[4 * plane bytes per line / 512]/4 :
                	ceiling[plane bytes per line / 512]
As per BSpec Don't increment selected "result_blocks" & "result_lines"
in case of BROXTON.

Signed-off-by: Kumar, Mahesh <mahesh1.kumar@intel.com>
---
 drivers/gpu/drm/i915/intel_pm.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

Comments

Matt Roper Oct. 23, 2015, 4:53 p.m. UTC | #1 
On Mon, Sep 21, 2015 at 11:41:18PM +0530, Kumar, Mahesh wrote:
> In case of Y-Tiling, "plane_blocks_per_line" calculation is different
> than X/None-Tiling case.
> This patch corrects this calculation according to Bspec.
> plane blocks per line = Plane memory format is Y tile ?
> 		ceiling[4 * plane bytes per line / 512]/4 :
>                 	ceiling[plane bytes per line / 512]
> As per BSpec Don't increment selected "result_blocks" & "result_lines"
> in case of BROXTON.
> 
> Signed-off-by: Kumar, Mahesh <mahesh1.kumar@intel.com>

Confirmed both changes against bspec.  Note that your first hunk here is
technically a fix for both SKL and BXT, only the second one is a
BXT-specific fix.

Reviewed-by: Matt Roper <matthew.d.roper@intel.com>

> ---
>  drivers/gpu/drm/i915/intel_pm.c | 10 ++++++++--
>  1 file changed, 8 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
> index a1ed920..5cfb5d9 100644
> --- a/drivers/gpu/drm/i915/intel_pm.c
> +++ b/drivers/gpu/drm/i915/intel_pm.c
> @@ -3247,7 +3247,13 @@ static bool skl_compute_plane_wm(const struct drm_i915_private *dev_priv,
>  				 latency);
>  
>  	plane_bytes_per_line = p_params->horiz_pixels * bytes_per_pixel;
> -	plane_blocks_per_line = DIV_ROUND_UP(plane_bytes_per_line, 512);
> +
> +	if (p_params->tiling == I915_FORMAT_MOD_Y_TILED ||
> +	    p_params->tiling == I915_FORMAT_MOD_Yf_TILED) {
> +		plane_blocks_per_line = DIV_ROUND_UP(4 * plane_bytes_per_line, 512);
> +		plane_blocks_per_line /= 4;
> +	} else
> +		plane_blocks_per_line = DIV_ROUND_UP(plane_bytes_per_line, 512);
>  
>  	if (p_params->tiling == I915_FORMAT_MOD_Y_TILED ||
>  	    p_params->tiling == I915_FORMAT_MOD_Yf_TILED) {
> @@ -3277,7 +3283,7 @@ static bool skl_compute_plane_wm(const struct drm_i915_private *dev_priv,
>  	res_blocks = selected_result + 1;
>  	res_lines = DIV_ROUND_UP(selected_result, plane_blocks_per_line);
>  
> -	if (level >= 1 && level <= 7) {
> +	if (level >= 1 && level <= 7 && !IS_BROXTON(dev_priv->dev)) {
>  		if (p_params->tiling == I915_FORMAT_MOD_Y_TILED ||
>  		    p_params->tiling == I915_FORMAT_MOD_Yf_TILED)
>  			res_lines += 4;
> -- 
> 1.9.1
>
Daniel Vetter Nov. 17, 2015, 2:19 p.m. UTC | #2 
On Fri, Oct 23, 2015 at 09:53:35AM -0700, Matt Roper wrote:
> On Mon, Sep 21, 2015 at 11:41:18PM +0530, Kumar, Mahesh wrote:
> > In case of Y-Tiling, "plane_blocks_per_line" calculation is different
> > than X/None-Tiling case.
> > This patch corrects this calculation according to Bspec.
> > plane blocks per line = Plane memory format is Y tile ?
> > 		ceiling[4 * plane bytes per line / 512]/4 :
> >                 	ceiling[plane bytes per line / 512]
> > As per BSpec Don't increment selected "result_blocks" & "result_lines"
> > in case of BROXTON.
> > 
> > Signed-off-by: Kumar, Mahesh <mahesh1.kumar@intel.com>
> 
> Confirmed both changes against bspec.  Note that your first hunk here is
> technically a fix for both SKL and BXT, only the second one is a
> BXT-specific fix.
> 
> Reviewed-by: Matt Roper <matthew.d.roper@intel.com>

This needs to be rebased (p_params is gone).
-Daniel
> 
> > ---
> >  drivers/gpu/drm/i915/intel_pm.c | 10 ++++++++--
> >  1 file changed, 8 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
> > index a1ed920..5cfb5d9 100644
> > --- a/drivers/gpu/drm/i915/intel_pm.c
> > +++ b/drivers/gpu/drm/i915/intel_pm.c
> > @@ -3247,7 +3247,13 @@ static bool skl_compute_plane_wm(const struct drm_i915_private *dev_priv,
> >  				 latency);
> >  
> >  	plane_bytes_per_line = p_params->horiz_pixels * bytes_per_pixel;
> > -	plane_blocks_per_line = DIV_ROUND_UP(plane_bytes_per_line, 512);
> > +
> > +	if (p_params->tiling == I915_FORMAT_MOD_Y_TILED ||
> > +	    p_params->tiling == I915_FORMAT_MOD_Yf_TILED) {
> > +		plane_blocks_per_line = DIV_ROUND_UP(4 * plane_bytes_per_line, 512);
> > +		plane_blocks_per_line /= 4;
> > +	} else
> > +		plane_blocks_per_line = DIV_ROUND_UP(plane_bytes_per_line, 512);
> >  
> >  	if (p_params->tiling == I915_FORMAT_MOD_Y_TILED ||
> >  	    p_params->tiling == I915_FORMAT_MOD_Yf_TILED) {
> > @@ -3277,7 +3283,7 @@ static bool skl_compute_plane_wm(const struct drm_i915_private *dev_priv,
> >  	res_blocks = selected_result + 1;
> >  	res_lines = DIV_ROUND_UP(selected_result, plane_blocks_per_line);
> >  
> > -	if (level >= 1 && level <= 7) {
> > +	if (level >= 1 && level <= 7 && !IS_BROXTON(dev_priv->dev)) {
> >  		if (p_params->tiling == I915_FORMAT_MOD_Y_TILED ||
> >  		    p_params->tiling == I915_FORMAT_MOD_Yf_TILED)
> >  			res_lines += 4;
> > -- 
> > 1.9.1
> > 
> 
> -- 
> Matt Roper
> Graphics Software Engineer
> IoTG Platform Enabling & Development
> Intel Corporation
> (916) 356-2795
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Jani Nikula Nov. 17, 2015, 2:52 p.m. UTC | #3 
On Tue, 17 Nov 2015, Daniel Vetter <daniel@ffwll.ch> wrote:
> On Fri, Oct 23, 2015 at 09:53:35AM -0700, Matt Roper wrote:
>> On Mon, Sep 21, 2015 at 11:41:18PM +0530, Kumar, Mahesh wrote:
>> > In case of Y-Tiling, "plane_blocks_per_line" calculation is different
>> > than X/None-Tiling case.
>> > This patch corrects this calculation according to Bspec.
>> > plane blocks per line = Plane memory format is Y tile ?
>> > 		ceiling[4 * plane bytes per line / 512]/4 :
>> >                 	ceiling[plane bytes per line / 512]
>> > As per BSpec Don't increment selected "result_blocks" & "result_lines"
>> > in case of BROXTON.
>> > 
>> > Signed-off-by: Kumar, Mahesh <mahesh1.kumar@intel.com>
>> 
>> Confirmed both changes against bspec.  Note that your first hunk here is
>> technically a fix for both SKL and BXT, only the second one is a
>> BXT-specific fix.
>> 
>> Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
>
> This needs to be rebased (p_params is gone).

And we already have this for skl

commit 0fda65680e92545caea5be7805a7f0a617fb6c20
Author: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Date:   Fri Feb 27 15:12:35 2015 +0000

    drm/i915/skl: Update watermarks for Y tiling

Mahesh, please check if you still need to make changes for BXT.

BR,
Jani.



> -Daniel
>> 
>> > ---
>> >  drivers/gpu/drm/i915/intel_pm.c | 10 ++++++++--
>> >  1 file changed, 8 insertions(+), 2 deletions(-)
>> > 
>> > diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
>> > index a1ed920..5cfb5d9 100644
>> > --- a/drivers/gpu/drm/i915/intel_pm.c
>> > +++ b/drivers/gpu/drm/i915/intel_pm.c
>> > @@ -3247,7 +3247,13 @@ static bool skl_compute_plane_wm(const struct drm_i915_private *dev_priv,
>> >  				 latency);
>> >  
>> >  	plane_bytes_per_line = p_params->horiz_pixels * bytes_per_pixel;
>> > -	plane_blocks_per_line = DIV_ROUND_UP(plane_bytes_per_line, 512);
>> > +
>> > +	if (p_params->tiling == I915_FORMAT_MOD_Y_TILED ||
>> > +	    p_params->tiling == I915_FORMAT_MOD_Yf_TILED) {
>> > +		plane_blocks_per_line = DIV_ROUND_UP(4 * plane_bytes_per_line, 512);
>> > +		plane_blocks_per_line /= 4;
>> > +	} else
>> > +		plane_blocks_per_line = DIV_ROUND_UP(plane_bytes_per_line, 512);
>> >  
>> >  	if (p_params->tiling == I915_FORMAT_MOD_Y_TILED ||
>> >  	    p_params->tiling == I915_FORMAT_MOD_Yf_TILED) {
>> > @@ -3277,7 +3283,7 @@ static bool skl_compute_plane_wm(const struct drm_i915_private *dev_priv,
>> >  	res_blocks = selected_result + 1;
>> >  	res_lines = DIV_ROUND_UP(selected_result, plane_blocks_per_line);
>> >  
>> > -	if (level >= 1 && level <= 7) {
>> > +	if (level >= 1 && level <= 7 && !IS_BROXTON(dev_priv->dev)) {
>> >  		if (p_params->tiling == I915_FORMAT_MOD_Y_TILED ||
>> >  		    p_params->tiling == I915_FORMAT_MOD_Yf_TILED)
>> >  			res_lines += 4;
>> > -- 
>> > 1.9.1
>> > 
>> 
>> -- 
>> Matt Roper
>> Graphics Software Engineer
>> IoTG Platform Enabling & Development
>> Intel Corporation
>> (916) 356-2795
>> _______________________________________________
>> Intel-gfx mailing list
>> Intel-gfx@lists.freedesktop.org
>> http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Tvrtko Ursulin Nov. 17, 2015, 3:11 p.m. UTC | #4 
On 17/11/15 14:52, Jani Nikula wrote:
> On Tue, 17 Nov 2015, Daniel Vetter <daniel@ffwll.ch> wrote:
>> On Fri, Oct 23, 2015 at 09:53:35AM -0700, Matt Roper wrote:
>>> On Mon, Sep 21, 2015 at 11:41:18PM +0530, Kumar, Mahesh wrote:
>>>> In case of Y-Tiling, "plane_blocks_per_line" calculation is different
>>>> than X/None-Tiling case.
>>>> This patch corrects this calculation according to Bspec.
>>>> plane blocks per line = Plane memory format is Y tile ?
>>>> 		ceiling[4 * plane bytes per line / 512]/4 :
>>>>                  	ceiling[plane bytes per line / 512]
>>>> As per BSpec Don't increment selected "result_blocks" & "result_lines"
>>>> in case of BROXTON.
>>>>
>>>> Signed-off-by: Kumar, Mahesh <mahesh1.kumar@intel.com>
>>>
>>> Confirmed both changes against bspec.  Note that your first hunk here is
>>> technically a fix for both SKL and BXT, only the second one is a
>>> BXT-specific fix.
>>>
>>> Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
>>
>> This needs to be rebased (p_params is gone).
>
> And we already have this for skl
>
> commit 0fda65680e92545caea5be7805a7f0a617fb6c20
> Author: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
> Date:   Fri Feb 27 15:12:35 2015 +0000
>
>      drm/i915/skl: Update watermarks for Y tiling
>
> Mahesh, please check if you still need to make changes for BXT.

Mahesh's patch seems to be on top of that one, so either a new doc 
change or something I've missed back then.

Regards,

Tvrtko
diff mbox

Patch

diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index a1ed920..5cfb5d9 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -3247,7 +3247,13 @@  static bool skl_compute_plane_wm(const struct drm_i915_private *dev_priv,
 				 latency);
 
 	plane_bytes_per_line = p_params->horiz_pixels * bytes_per_pixel;
-	plane_blocks_per_line = DIV_ROUND_UP(plane_bytes_per_line, 512);
+
+	if (p_params->tiling == I915_FORMAT_MOD_Y_TILED ||
+	    p_params->tiling == I915_FORMAT_MOD_Yf_TILED) {
+		plane_blocks_per_line = DIV_ROUND_UP(4 * plane_bytes_per_line, 512);
+		plane_blocks_per_line /= 4;
+	} else
+		plane_blocks_per_line = DIV_ROUND_UP(plane_bytes_per_line, 512);
 
 	if (p_params->tiling == I915_FORMAT_MOD_Y_TILED ||
 	    p_params->tiling == I915_FORMAT_MOD_Yf_TILED) {
@@ -3277,7 +3283,7 @@  static bool skl_compute_plane_wm(const struct drm_i915_private *dev_priv,
 	res_blocks = selected_result + 1;
 	res_lines = DIV_ROUND_UP(selected_result, plane_blocks_per_line);
 
-	if (level >= 1 && level <= 7) {
+	if (level >= 1 && level <= 7 && !IS_BROXTON(dev_priv->dev)) {
 		if (p_params->tiling == I915_FORMAT_MOD_Y_TILED ||
 		    p_params->tiling == I915_FORMAT_MOD_Yf_TILED)
 			res_lines += 4;