On 08/24/2018 05:04 PM, Jason Ekstrand wrote:
On Fri, Aug 24, 2018 at 12:08 AM Tapani Pälli <tapani.pa...@intel.com <mailto:tapani.pa...@intel.com>> wrote:



    On 08/22/2018 05:18 PM, Jason Ekstrand wrote:
     > On Tue, Aug 21, 2018 at 3:27 AM Tapani Pälli
    <tapani.pa...@intel.com <mailto:tapani.pa...@intel.com>
     > <mailto:tapani.pa...@intel.com <mailto:tapani.pa...@intel.com>>>
    wrote:
     >
     >     When adding YUV support, we need to figure out
    implementation-defined
     >     external format identifier.
     >
     >     Signed-off-by: Tapani Pälli <tapani.pa...@intel.com
    <mailto:tapani.pa...@intel.com>
     >     <mailto:tapani.pa...@intel.com <mailto:tapani.pa...@intel.com>>>
     >     ---
     >       src/intel/vulkan/anv_android.c | 99
     >     ++++++++++++++++++++++++++++++++++++++++++
     >       1 file changed, 99 insertions(+)
     >
     >     diff --git a/src/intel/vulkan/anv_android.c
     >     b/src/intel/vulkan/anv_android.c
     >     index 46c41d57861..7d0eb588e2b 100644
     >     --- a/src/intel/vulkan/anv_android.c
     >     +++ b/src/intel/vulkan/anv_android.c
     >     @@ -29,6 +29,8 @@
     >       #include <sync/sync.h>
     >
     >       #include "anv_private.h"
     >     +#include "vk_format_info.h"
     >     +#include "vk_util.h"
     >
     >       static int anv_hal_open(const struct hw_module_t* mod,
    const char*
     >     id, struct hw_device_t** dev);
     >       static int anv_hal_close(struct hw_device_t *dev);
     >     @@ -96,6 +98,103 @@ anv_hal_close(struct hw_device_t *dev)
     >          return -1;
     >       }
     >
     >     +static VkResult
     >     +get_ahw_buffer_format_properties(
     >     +   VkDevice device_h,
     >     +   const struct AHardwareBuffer *buffer,
     >     +   VkAndroidHardwareBufferFormatPropertiesANDROID *pProperties)
     >     +{
     >     +   ANV_FROM_HANDLE(anv_device, device, device_h);
     >     +
     >     +   /* Get a description of buffer contents . */
     >     +   AHardwareBuffer_Desc desc;
     >     +   AHardwareBuffer_describe(buffer, &desc);
     >     +
     >     +   /* Verify description. */
     >     +   uint64_t gpu_usage =
     >     +      AHARDWAREBUFFER_USAGE_GPU_SAMPLED_IMAGE |
     >     +      AHARDWAREBUFFER_USAGE_GPU_COLOR_OUTPUT |
     >     +      AHARDWAREBUFFER_USAGE_GPU_DATA_BUFFER;
     >     +
     >     +   /* "Buffer must be a valid Android hardware buffer object
    with
     >     at least
     >     +    * one of the AHARDWAREBUFFER_USAGE_GPU_* usage flags."
     >     +    */
     >     +   if (!(desc.usage & (gpu_usage)))
     >     +      return VK_ERROR_INVALID_EXTERNAL_HANDLE_KHR;
     >     +
     >     +   /* Fill properties fields based on description. */
     >     +   VkAndroidHardwareBufferFormatPropertiesANDROID *p =
    pProperties;
     >     +
     >     +   p->pNext = NULL;
     >
     >
     > You shouldn't be overwriting pNext.  That's used by the client to
    let
     > them chain in multiple structs to fill out in case Google ever
    extends
     > this extension.  Also, while we're here, it'd be good to throw in an
     > assert that p->sType is the right thing.

    Yes of course, will remove.

     >     +   p->format = vk_format_from_android(desc.format);
     >     +   p->externalFormat = 1; /* XXX */
     >     +
     >     +   const struct anv_format *anv_format =
    anv_get_format(p->format);
     >     +   struct anv_physical_device *physical_device =
     >     +      &device->instance->physicalDevice;
     >     +   const struct gen_device_info *devinfo =
    &physical_device->info;
     >
     >
     > If all you need is devinfo, that's avilable in the device; you don't
     > need to get the physical device for it.  Should be device->info.

    OK

     >     +
     >     +   p->formatFeatures =
     >     +      anv_get_image_format_features(devinfo, p->format,
     >     +                                    anv_format,
     >     VK_IMAGE_TILING_OPTIMAL);
     >     +
     >     +   /* "The formatFeatures member *must* include
     >     +    *  VK_FORMAT_FEATURE_SAMPLED_IMAGE_BIT and at least one of
     >     +    *  VK_FORMAT_FEATURE_MIDPOINT_CHROMA_SAMPLES_BIT or
     >     +    *  VK_FORMAT_FEATURE_COSITED_CHROMA_SAMPLES_BIT"
     >     +    */
     >     +   p->formatFeatures |=
     >     +      VK_FORMAT_FEATURE_MIDPOINT_CHROMA_SAMPLES_BIT;
     >
     >
     > Uh... Why not just throw in SAMPLED_BIT?  For that matter, all of
    the
     > formats you have in your conversion helpers support sampling.  Maybe
     > just replace that with an assert for now.

    Yeah, VK_FORMAT_FEATURE_SAMPLED_IMAGE_BIT is there. Well thing is that
    dEQP checks explicitly that either
    VK_FORMAT_FEATURE_MIDPOINT_CHROMA_SAMPLES_BIT or
    VK_FORMAT_FEATURE_COSITED_CHROMA_SAMPLES_BIT exists (independent of
    surface format). TBH I'm not sure what to do about that.


That's annoying...  Is that in a Khronos CTS test or a Google test? Either way, we should try to get it corrected.  If you don't know how to do that, I can send some e-mails.

But per the Vulkan spec quote above it's not a bug, it says that "formatFeatures member must include sampled bit and at least one of MIDPOINT_CHROMA_SAMPLES or COSITED_CHROMA_SAMPLES. So .. I guess I should limit the supported formats to only those that have anv_format->can_ycbcr() set?


--Jason

     >     +
     >     +   p->samplerYcbcrConversionComponents.r =
     >     VK_COMPONENT_SWIZZLE_IDENTITY;
     >     +   p->samplerYcbcrConversionComponents.g =
     >     VK_COMPONENT_SWIZZLE_IDENTITY;
     >     +   p->samplerYcbcrConversionComponents.b =
     >     VK_COMPONENT_SWIZZLE_IDENTITY;
     >     +   p->samplerYcbcrConversionComponents.a =
     >     VK_COMPONENT_SWIZZLE_IDENTITY;
     >     +
     >     +   p->suggestedYcbcrModel =
     >     VK_SAMPLER_YCBCR_MODEL_CONVERSION_RGB_IDENTITY;
     >     +   p->suggestedYcbcrRange = VK_SAMPLER_YCBCR_RANGE_ITU_FULL;
     >     +   p->suggestedXChromaOffset = VK_CHROMA_LOCATION_COSITED_EVEN;
     >     +   p->suggestedYChromaOffset = VK_CHROMA_LOCATION_COSITED_EVEN;
     >     +
     >     +   return VK_SUCCESS;
     >     +}
     >     +
     >     +VkResult
     >     +anv_GetAndroidHardwareBufferPropertiesANDROID(
     >     +   VkDevice device_h,
     >     +   const struct AHardwareBuffer *buffer,
     >     +   VkAndroidHardwareBufferPropertiesANDROID *pProperties)
     >     +{
     >     +   ANV_FROM_HANDLE(anv_device, dev, device_h);
     >     +   struct anv_physical_device *pdevice =
     >     &dev->instance->physicalDevice;
     >     +
     >     +   VkAndroidHardwareBufferFormatPropertiesANDROID *format_prop =
     >     +      vk_find_struct(pProperties->pNext,
     >     +
     >       ANDROID_HARDWARE_BUFFER_FORMAT_PROPERTIES_ANDROID);
     >     +
     >     +   /* Fill format properties of an Android hardware buffer. */
     >     +   if (format_prop)
     >     +      get_ahw_buffer_format_properties(device_h, buffer,
    format_prop);
     >     +
     >     +   /* Get a description of buffer contents . */
     >     +   AHardwareBuffer_Desc desc;
     >     +   AHardwareBuffer_describe(buffer, &desc);
     >
     >
> I don't see you using this anywyere. get_ahw_bufer_format_properties
     > calls it itself.  An artefact of rebsing perhaps?  Should
    probably drop
     > it if it's not needed.

    Yes this was only used for debugging, will remove.

     >     +
     >     +   const native_handle_t *handle =
     >     +      AHardwareBuffer_getNativeHandle(buffer);
     >     +   int dma_buf = (handle && handle->numFds) ?
    handle->data[0] : -1;
     >     +   if (dma_buf < 0)
     >     +      return VK_ERROR_INVALID_EXTERNAL_HANDLE_KHR;
     >     +
     >     +   /* All memory types. */
     >     +   uint32_t memory_types = (1ull <<
    pdevice->memory.type_count) - 1;
     >     +
     >     +   pProperties->allocationSize = lseek(dma_buf, 0, SEEK_END);
     >     +   pProperties->memoryTypeBits = memory_types;
     >     +
     >     +   return VK_SUCCESS;
     >     +}
     >     +
     >       VkResult
     >       anv_image_from_gralloc(VkDevice device_h,
     >                              const VkImageCreateInfo *base_info,
     >     --
     >     2.14.4
     >
     >     _______________________________________________
     >     mesa-dev mailing list
     > mesa-dev@lists.freedesktop.org
    <mailto:mesa-dev@lists.freedesktop.org>
    <mailto:mesa-dev@lists.freedesktop.org
    <mailto:mesa-dev@lists.freedesktop.org>>
     > https://lists.freedesktop.org/mailman/listinfo/mesa-dev
     >

_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Reply via email to