On Wed, Jul 15, 2015 at 06:39:30PM +0530, Kausal Malladi wrote: > As per Color Manager design, each driver is responsible to load its > palette color correction and enhancement capabilities in the form of > a DRM blob property, so that user space can query and read. > > This patch loads all CHV platform specific gamma color capabilities > for CRTC into a blob that can be accessible by user space to > query capabilities via DRM property interface. > > Signed-off-by: Shashank Sharma <shashank.sharma@xxxxxxxxx> > Signed-off-by: Kausal Malladi <Kausal.Malladi@xxxxxxxxx> > --- > drivers/gpu/drm/i915/intel_color_manager.c | 48 ++++++++++++++++++++++++++++++ > drivers/gpu/drm/i915/intel_color_manager.h | 4 +++ > 2 files changed, 52 insertions(+) > > diff --git a/drivers/gpu/drm/i915/intel_color_manager.c b/drivers/gpu/drm/i915/intel_color_manager.c > index baa4536..def20d0f 100644 > --- a/drivers/gpu/drm/i915/intel_color_manager.c > +++ b/drivers/gpu/drm/i915/intel_color_manager.c > @@ -27,15 +27,63 @@ > > #include "intel_color_manager.h" > > +int get_chv_pipe_gamma_capabilities(struct drm_device *dev, > + struct drm_palette_caps *palette_caps, struct drm_crtc *crtc) > +{ > + struct drm_property_blob *blob = NULL; > + struct drm_mode_config *config = &dev->mode_config; > + int ret; > + > + /* > + * This function exposes best capability for DeGamma and Gamma > + * For CHV, the DeGamma LUT has 65 entries > + * and the best Gamma capability has 257 entries (CGM unit) > + */ > + palette_caps->version = CHV_PALETTE_STRUCT_VERSION; > + palette_caps->num_samples_before_ctm = > + CHV_DEGAMMA_MAX_VALS; > + palette_caps->num_samples_after_ctm = > + CHV_10BIT_GAMMA_MAX_VALS; > + > + ret = drm_property_replace_global_blob(dev, &blob, We're only calling this once at startup and never actually replacing the capabilities, right? Is there any difference between this and just manually doing a drm_property_create_blob() and passing that blob's ID as the value when we attach the property to the CRTC? It feels a bit strange to be calling 'replace' on a local variable that isn't initialized and is never used/saved. > + sizeof(struct drm_palette_caps), > + (const void *)palette_caps, > + &crtc->base, config->prop_color_capabilities); > + if (ret) { > + DRM_ERROR("Error updating Gamma blob\n"); > + return -EFAULT; > + } > + > + return 0; > +} > + > +int get_pipe_gamma_capabilities(struct drm_device *dev, > + struct drm_palette_caps *palette_caps, struct drm_crtc *crtc) > +{ > + if (IS_CHERRYVIEW(dev)) > + return get_chv_pipe_gamma_capabilities(dev, palette_caps, crtc); > + return -EINVAL; > +} > + > void intel_attach_color_properties_to_crtc(struct drm_device *dev, > struct drm_mode_object *mode_obj) > { > struct drm_mode_config *config = &dev->mode_config; > + struct drm_palette_caps *palette_caps; > + struct drm_crtc *crtc; > + int ret; > > if (IS_CHERRYVIEW(dev)) { > + crtc = obj_to_crtc(mode_obj); > if (config->prop_color_capabilities) > drm_object_attach_property(mode_obj, > config->prop_color_capabilities, 0); > + palette_caps = kzalloc(sizeof(struct drm_palette_caps), > + GFP_KERNEL); > + ret = get_pipe_gamma_capabilities(dev, palette_caps, crtc); Is palette_caps ever freed? I believe the data is copied when a blob is created, so I think you should be free to kfree() this allocation right away. Matt > + if (ret) > + DRM_ERROR("Error getting gamma capability for CHV\n"); > + > if (config->prop_palette_before_ctm) > drm_object_attach_property(mode_obj, > config->prop_palette_before_ctm, 0); > diff --git a/drivers/gpu/drm/i915/intel_color_manager.h b/drivers/gpu/drm/i915/intel_color_manager.h > index 04c921d..51aeb91 100644 > --- a/drivers/gpu/drm/i915/intel_color_manager.h > +++ b/drivers/gpu/drm/i915/intel_color_manager.h > @@ -27,3 +27,7 @@ > #include <drm/drmP.h> > #include <drm/drm_crtc_helper.h> > #include "i915_drv.h" > + > +#define CHV_PALETTE_STRUCT_VERSION 1 > +#define CHV_DEGAMMA_MAX_VALS 65 > +#define CHV_10BIT_GAMMA_MAX_VALS 257 > -- > 2.4.5 > -- Matt Roper Graphics Software Engineer IoTG Platform Enabling & Development Intel Corporation (916) 356-2795 _______________________________________________ dri-devel mailing list dri-devel@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/dri-devel