b2eaf812b1
!146956 introduced buffer streaming and is used to improve the performance of text rendering. However the solution didn't check the minimum allowed offset for storage buffers. This PR will use the correct offset. This was only visible on Qualcomm devices as other systems don't limit this offset. Detected when looking into: #147023 Pull Request: https://projects.blender.org/blender/blender/pulls/147192
70 lines
2.3 KiB
C++
70 lines
2.3 KiB
C++
/* SPDX-FileCopyrightText: 2022 Blender Authors
|
|
*
|
|
* SPDX-License-Identifier: GPL-2.0-or-later */
|
|
|
|
/** \file
|
|
* \ingroup gpu
|
|
*/
|
|
|
|
#include "vk_streaming_buffer.hh"
|
|
#include "vk_buffer.hh"
|
|
#include "vk_context.hh"
|
|
|
|
namespace blender::gpu {
|
|
VKStreamingBuffer::VKStreamingBuffer(VKBuffer &buffer, VkDeviceSize min_offset_alignment)
|
|
: min_offset_alignment_(min_offset_alignment),
|
|
vk_buffer_dst_(buffer.vk_handle()),
|
|
vk_buffer_size_(buffer.size_in_bytes())
|
|
{
|
|
}
|
|
|
|
VKStreamingBuffer::~VKStreamingBuffer()
|
|
{
|
|
vk_buffer_dst_ = VK_NULL_HANDLE;
|
|
}
|
|
|
|
VkDeviceSize VKStreamingBuffer::update(VKContext &context, const void *data, size_t data_size)
|
|
{
|
|
render_graph::VKRenderGraph &render_graph = context.render_graph();
|
|
const bool allocate_new_buffer = !(host_buffer_.has_value() &&
|
|
data_size <
|
|
host_buffer_.value().get()->size_in_bytes() - offset_);
|
|
if (allocate_new_buffer) {
|
|
host_buffer_.emplace(std::make_unique<VKBuffer>());
|
|
VKBuffer &host_buffer = *host_buffer_.value().get();
|
|
host_buffer.create(vk_buffer_size_,
|
|
VK_BUFFER_USAGE_TRANSFER_SRC_BIT,
|
|
VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT,
|
|
VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT,
|
|
VMA_ALLOCATION_CREATE_MAPPED_BIT |
|
|
VMA_ALLOCATION_CREATE_HOST_ACCESS_SEQUENTIAL_WRITE_BIT,
|
|
0.4f);
|
|
offset_ = 0;
|
|
|
|
render_graph::VKCopyBufferNode::CreateInfo copy_buffer = {
|
|
host_buffer.vk_handle(), vk_buffer_dst(), {0, 0, 0}};
|
|
copy_buffer_handle_ = render_graph.add_node(copy_buffer);
|
|
}
|
|
VKBuffer &host_buffer = *host_buffer_.value().get();
|
|
|
|
VkDeviceSize start_offset = offset_;
|
|
/* Advance the offset to the next possible offset considering the minimum allowed offset
|
|
* alignment. */
|
|
offset_ += data_size;
|
|
if (min_offset_alignment_ > 1) {
|
|
offset_ = ceil_to_multiple_ul(offset_, min_offset_alignment_);
|
|
}
|
|
|
|
memcpy(
|
|
static_cast<void *>(static_cast<uint8_t *>(host_buffer.mapped_memory_get()) + start_offset),
|
|
data,
|
|
data_size);
|
|
|
|
render_graph::VKCopyBufferNode::Data ©_buffer_data = render_graph.get_node_data(
|
|
copy_buffer_handle_);
|
|
copy_buffer_data.region.size += data_size;
|
|
return start_offset;
|
|
}
|
|
|
|
} // namespace blender::gpu
|