download_data_async()

void MayaFlux::Portal::Graphics::TextureLoom::download_data_async	(	const std::shared_ptr< Core::VKImage > &	image,
		void *	data,
		size_t	size
	)

Download pixel data from a texture without blocking the graphics queue.

Allocates a per-call staging buffer, command buffer, and fence. Records a copy-image-to-buffer op, submits with the fence, and blocks the calling thread on vkWaitForFences until the copy completes. Unlike download_data, this does not call queue.waitIdle, so other graphics work proceeds concurrently.

Intended to be called from a worker thread (e.g. via std::async). Calling from the graphics thread or any thread that must not block will stall that thread for the duration of the copy.

Parameters

image	Source image.
data	Destination host pointer, at least size bytes.
size	Byte count to read.

Definition at line 409 of file TextureLoom.cpp.

{
    if (!is_initialized() || !image || !data) {
        MF_ERROR(Journal::Component::Portal, Journal::Context::ImageProcessing,
            "Invalid parameters for download_data_async");
        return;
    }
 
    auto buffer_service = Registry::BackendRegistry::instance()
                              .get_service<Registry::Service::BufferService>();
    if (!buffer_service || !buffer_service->execute_fenced
        || !buffer_service->wait_fenced || !buffer_service->release_fenced
        || !buffer_service->initialize_buffer || !buffer_service->destroy_buffer
        || !buffer_service->invalidate_range) {
        MF_ERROR(Journal::Component::Portal, Journal::Context::ImageProcessing,
            "download_data_async: BufferService unavailable or incomplete");
        return;
    }
 
    auto staging = std::make_shared<Buffers::VKBuffer>(
        size,
        Buffers::VKBuffer::Usage::STAGING,
        Kakshya::DataModality::IMAGE_COLOR);
 
    buffer_service->initialize_buffer(std::static_pointer_cast<void>(staging));
 
    auto handle = buffer_service->execute_fenced([&](void* cmd_ptr) {
        vk::CommandBuffer cmd(static_cast<VkCommandBuffer>(cmd_ptr));
 
        vk::ImageMemoryBarrier barrier {};
        barrier.oldLayout = image->get_current_layout();
        barrier.newLayout = vk::ImageLayout::eTransferSrcOptimal;
        barrier.srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED;
        barrier.dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED;
        barrier.image = image->get_image();
        barrier.subresourceRange.aspectMask = image->get_aspect_flags();
        barrier.subresourceRange.baseMipLevel = 0;
        barrier.subresourceRange.levelCount = image->get_mip_levels();
        barrier.subresourceRange.baseArrayLayer = 0;
        barrier.subresourceRange.layerCount = image->get_array_layers();
        barrier.srcAccessMask = vk::AccessFlagBits::eShaderRead;
        barrier.dstAccessMask = vk::AccessFlagBits::eTransferRead;
 
        cmd.pipelineBarrier(
            vk::PipelineStageFlagBits::eFragmentShader,
            vk::PipelineStageFlagBits::eTransfer,
            vk::DependencyFlags {}, {}, {}, barrier);
 
        vk::BufferImageCopy region {};
        region.bufferOffset = 0;
        region.bufferRowLength = 0;
        region.bufferImageHeight = 0;
        region.imageSubresource.aspectMask = image->get_aspect_flags();
        region.imageSubresource.mipLevel = 0;
        region.imageSubresource.baseArrayLayer = 0;
        region.imageSubresource.layerCount = image->get_array_layers();
        region.imageOffset = vk::Offset3D { 0, 0, 0 };
        region.imageExtent = vk::Extent3D {
            image->get_width(),
            image->get_height(),
            image->get_depth()
        };
 
        cmd.copyImageToBuffer(
            image->get_image(),
            vk::ImageLayout::eTransferSrcOptimal,
            staging->get_buffer(),
            1, &region);
 
        barrier.oldLayout = vk::ImageLayout::eTransferSrcOptimal;
        barrier.newLayout = vk::ImageLayout::eShaderReadOnlyOptimal;
        barrier.srcAccessMask = vk::AccessFlagBits::eTransferRead;
        barrier.dstAccessMask = vk::AccessFlagBits::eMemoryRead;
 
        cmd.pipelineBarrier(
            vk::PipelineStageFlagBits::eTransfer,
            vk::PipelineStageFlagBits::eFragmentShader,
            vk::DependencyFlags {}, {}, {}, barrier);
    });
 
    if (!handle) {
        MF_ERROR(Journal::Component::Portal, Journal::Context::ImageProcessing,
            "download_data_async: execute_fenced returned null handle");
        buffer_service->destroy_buffer(std::static_pointer_cast<void>(staging));
        return;
    }
 
    image->set_current_layout(vk::ImageLayout::eShaderReadOnlyOptimal);
 
    buffer_service->wait_fenced(handle);
 
    auto& resources = staging->get_buffer_resources();
    buffer_service->invalidate_range(resources.memory, 0, 0);
 
    void* mapped = staging->get_mapped_ptr();
    if (mapped) {
        std::memcpy(data, mapped, size);
    } else {
        MF_ERROR(Journal::Component::Portal, Journal::Context::ImageProcessing,
            "download_data_async: staging buffer has no mapped pointer");
    }
 
    buffer_service->release_fenced(handle);
    buffer_service->destroy_buffer(std::static_pointer_cast<void>(staging));
 
    MF_DEBUG(Journal::Component::Portal, Journal::Context::ImageProcessing,
        "download_data_async: completed {} byte download for {}x{}",
        size, image->get_width(), image->get_height());
}

References cmd, MayaFlux::Registry::BackendRegistry::get_service(), image, MayaFlux::Kakshya::IMAGE_COLOR, MayaFlux::Journal::ImageProcessing, MayaFlux::Registry::BackendRegistry::instance(), is_initialized(), MF_DEBUG, MF_ERROR, MayaFlux::Journal::Portal, and MayaFlux::Buffers::VKBuffer::STAGING.

Here is the call graph for this function: