On Wed, Jun 17, 2015 at 08:23:40AM +0100, Dave Gordon wrote: > On 15/06/15 21:09, Chris Wilson wrote: > > On Mon, Jun 15, 2015 at 07:36:19PM +0100, Dave Gordon wrote: > >> From: Alex Dai <[email protected]> > >> > >> i915_gem_object_write() is a generic function to copy data from a plain > >> linear buffer to a paged gem object. > >> > >> We will need this for the microcontroller firmware loading support code. > >> > >> Issue: VIZ-4884 > >> Signed-off-by: Alex Dai <[email protected]> > >> Signed-off-by: Dave Gordon <[email protected]> > >> --- > >> drivers/gpu/drm/i915/i915_drv.h | 2 ++ > >> drivers/gpu/drm/i915/i915_gem.c | 28 ++++++++++++++++++++++++++++ > >> 2 files changed, 30 insertions(+) > >> > >> diff --git a/drivers/gpu/drm/i915/i915_drv.h > >> b/drivers/gpu/drm/i915/i915_drv.h > >> index 611fbd8..9094c06 100644 > >> --- a/drivers/gpu/drm/i915/i915_drv.h > >> +++ b/drivers/gpu/drm/i915/i915_drv.h > >> @@ -2713,6 +2713,8 @@ void *i915_gem_object_alloc(struct drm_device *dev); > >> void i915_gem_object_free(struct drm_i915_gem_object *obj); > >> void i915_gem_object_init(struct drm_i915_gem_object *obj, > >> const struct drm_i915_gem_object_ops *ops); > >> +int i915_gem_object_write(struct drm_i915_gem_object *obj, > >> + const void *data, size_t size); > >> struct drm_i915_gem_object *i915_gem_alloc_object(struct drm_device *dev, > >> size_t size); > >> void i915_init_vm(struct drm_i915_private *dev_priv, > >> diff --git a/drivers/gpu/drm/i915/i915_gem.c > >> b/drivers/gpu/drm/i915/i915_gem.c > >> index be35f04..75d63c2 100644 > >> --- a/drivers/gpu/drm/i915/i915_gem.c > >> +++ b/drivers/gpu/drm/i915/i915_gem.c > >> @@ -5392,3 +5392,31 @@ bool i915_gem_obj_is_pinned(struct > >> drm_i915_gem_object *obj) > >> return false; > >> } > >> > >> +/* Fill the @obj with the @size amount of @data */ > >> +int i915_gem_object_write(struct drm_i915_gem_object *obj, > >> + const void *data, size_t size) > >> +{ > >> + struct sg_table *sg; > >> + size_t bytes; > >> + int ret; > >> + > >> + ret = i915_gem_object_get_pages(obj); > >> + if (ret) > >> + return ret; > >> + > >> + i915_gem_object_pin_pages(obj); > > > > You don't set the object into the CPU domain, or instead manually handle > > the domain flushing. You don't handle objects that cannot be written > > directly by the CPU, nor do you handle objects whose representation in > > memory is not linear. > > -Chris > > No we don't handle just any random gem object, but we do return an error > code for any types not supported. However, as we don't really need the > full generality of writing into a gem object of any type, I will replace > this function with one that combines the allocation of a new object > (which will therefore definitely be of the correct type, in the correct > domain, etc) and filling it with the data to be preserved.
Domain handling is required for all gem objects, and the resulting bugs if you don't for one-off objects are absolutely no fun to track down. -Daniel -- Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
