//SPDX-License-Identifier: LGPL-3.0-only
//SPDX-FileCopyrightText: Copyright (C) 2026 Catcrafts®

module;

#ifndef CRAFTER_GRAPHICS_WINDOW_DOM
#include "vulkan/vulkan.h"

#endif // !CRAFTER_GRAPHICS_WINDOW_DOM
export module Crafter.Graphics:VulkanBuffer;
#ifndef CRAFTER_GRAPHICS_WINDOW_DOM
import std;
import :Device;

namespace Crafter {
    // Round a host-write flush range outward to nonCoherentAtomSize boundaries —
    // the alignment vkFlushMappedMemoryRanges demands for a sub-buffer range
    // (whole-buffer VK_WHOLE_SIZE calls sidestep it). `atom` is
    // VkPhysicalDeviceLimits::nonCoherentAtomSize (a power of two ≥ 1) and
    // `mappingSize` is the byte size the memory was mapped with (the allocation
    // size). The start rounds down and the end rounds up to atom multiples; the
    // end is then clamped to `mappingSize`. Clamping is what keeps the range
    // valid even when `mappingSize` itself is not atom-aligned: the spec permits
    // a non-atom-multiple size only when offset+size equals the mapping size, and
    // the clamp lands exactly on that exception. Pure math, so it is unit-tested
    // without a device.
    export struct MappedFlushRange { VkDeviceSize offset; VkDeviceSize size; };
    export constexpr MappedFlushRange AlignMappedFlushRange(
            VkDeviceSize offset, VkDeviceSize bytes,
            VkDeviceSize atom, VkDeviceSize mappingSize) {
        VkDeviceSize begin = (offset / atom) * atom;
        VkDeviceSize end   = ((offset + bytes + atom - 1) / atom) * atom;
        if (end > mappingSize) end = mappingSize;
        return { begin, end - begin };
    }

    export class VulkanBufferBase {
        public:
        VkDeviceAddress address;
        std::uint32_t size;
        // Byte size the memory was mapped with — equal to the allocation's
        // VkMemoryRequirements::size, which is ≥ `size` and what ranged flushes
        // clamp their upper bound to. Only meaningful for mapped buffers.
        VkDeviceSize mappedSize = 0;
        VkBuffer buffer = VK_NULL_HANDLE;
        VkDeviceMemory memory;
        // Property flags of the memory type actually chosen by GetMemoryType —
        // not the requested flags. GetMemoryType may land a request without the
        // COHERENT bit on a coherent type (and vice versa), so the flush/
        // invalidate paths gate on this recorded value, never on the request.
        VkMemoryPropertyFlags memoryPropertyFlagsChosen = 0;
        // Byte capacity the buffer was created with, and the usage flags it was
        // created with. Resize reuses the allocation in place when a new
        // request still fits within `capacity` and these immutable-at-create
        // properties match, avoiding a destroy+reallocate.
        std::uint32_t capacity = 0;
        VkBufferUsageFlags2 usageFlagsCreated = 0;
    };

    export template<typename T>
    class VulkanBufferMapped {
        public:
        T* value;
    };
    export class VulkanBufferMappedEmpty {};
    template<typename T, bool Mapped>
    using VulkanBufferMappedConditional =
        std::conditional_t<
            Mapped,
            VulkanBufferMapped<T>,
            VulkanBufferMappedEmpty
        >;

    export template <typename T, bool Mapped> 
    class VulkanBuffer : public VulkanBufferBase, public VulkanBufferMappedConditional<T, Mapped> {
    public:
        VulkanBuffer() = default;
        // `preferredPropertyFlags` are best-effort: the allocation prefers a
        // memory type satisfying them on top of `memoryPropertyFlags`, but
        // falls back to the required flags alone rather than failing. Use it
        // for perf hints (e.g. DEVICE_LOCAL on a mapped buffer) that aren't
        // available on every device — see Device::GetMemoryType.
        void Create(VkBufferUsageFlags2 usageFlags, VkMemoryPropertyFlags memoryPropertyFlags, std::uint32_t count, VkMemoryPropertyFlags preferredPropertyFlags = 0) {
            size = count * sizeof(T);
            capacity = size;
            usageFlagsCreated = usageFlags;

            // Carry usage in the maintenance5 flags2 chain so 64-bit bits
            // (e.g. VK_BUFFER_USAGE_2_MEMORY_DECOMPRESSION_BIT_EXT, bit 35)
            // are not truncated.
            VkBufferUsageFlags2CreateInfo usageFlags2 {
                .sType = VK_STRUCTURE_TYPE_BUFFER_USAGE_FLAGS_2_CREATE_INFO,
                .usage = usageFlags,
            };
            VkBufferCreateInfo bufferCreateInfo {};
			bufferCreateInfo.sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO;
            bufferCreateInfo.pNext = &usageFlags2;
            bufferCreateInfo.usage = 0;
			bufferCreateInfo.size = sizeof(T)*count;
            Device::CheckVkResult(vkCreateBuffer(Device::device, &bufferCreateInfo, nullptr, &buffer));
        
            VkMemoryRequirements memReqs;
            vkGetBufferMemoryRequirements(Device::device, buffer, &memReqs);
            std::uint32_t memoryTypeIndex = Device::GetMemoryType(memReqs.memoryTypeBits, memoryPropertyFlags, preferredPropertyFlags);
            memoryPropertyFlagsChosen = Device::memoryProperties.memoryTypes[memoryTypeIndex].propertyFlags;
            VkMemoryAllocateInfo memAlloc {
                .sType = VK_STRUCTURE_TYPE_MEMORY_ALLOCATE_INFO,
                .allocationSize = memReqs.size,
                .memoryTypeIndex = memoryTypeIndex
            };

            // Only a buffer created for SHADER_DEVICE_ADDRESS may have its
            // address queried (VUID-VkBufferDeviceAddressInfo-buffer-02601);
            // the others keep address 0.
            const bool addressable = (usageFlags & VK_BUFFER_USAGE_2_SHADER_DEVICE_ADDRESS_BIT) != 0;
            VkMemoryAllocateFlagsInfoKHR allocFlagsInfo {
                .sType = VK_STRUCTURE_TYPE_MEMORY_ALLOCATE_FLAGS_INFO_KHR,
                .flags = VK_MEMORY_ALLOCATE_DEVICE_ADDRESS_BIT_KHR,
            };
            if (addressable) {
                memAlloc.pNext = &allocFlagsInfo;
            }
            Device::CheckVkResult(vkAllocateMemory(Device::device, &memAlloc, nullptr, &memory));
        
            Device::CheckVkResult(vkBindBufferMemory(Device::device, buffer, memory, 0));

            VkBufferDeviceAddressInfo addressInfo = {
                .sType = VK_STRUCTURE_TYPE_BUFFER_DEVICE_ADDRESS_INFO,
                .buffer = buffer
            };
            address = addressable ? vkGetBufferDeviceAddress(Device::device, &addressInfo) : 0;

            // Record the allocation's byte size (≥ `size`): a ranged
            // FlushDevice flushes only the dirty sub-range and clamps its
            // rounded-up upper bound to this (the nonCoherentAtomSize-aligned-end
            // exception, see AlignMappedFlushRange).
            mappedSize = memReqs.size;
            if constexpr(Mapped) {
                Device::CheckVkResult(vkMapMemory(Device::device, memory, 0, memReqs.size, 0, reinterpret_cast<void**>(&(VulkanBufferMappedConditional<T, true>::value))));
            }
        }

        void Clear() {
            if constexpr(Mapped) {
                vkUnmapMemory(Device::device, memory);
            }
            vkDestroyBuffer(Device::device, buffer, nullptr);
            vkFreeMemory(Device::device, memory, nullptr);
            buffer = VK_NULL_HANDLE;
        }

        // Like Clear(), but hands the destroy+free to Device's fence-keyed
        // deletion queue instead of doing it immediately. Use when the buffer
        // may still be read by an in-flight frame's GPU work — destroying it
        // now would be a use-after-free, since frames are pipelined up to
        // framesInFlight deep (issue #101). The handle is nulled immediately so
        // this VulkanBuffer no longer owns it; vkFreeMemory (deferred) implicitly
        // unmaps mapped memory, so no explicit vkUnmapMemory is needed here.
        void DeferredClear() {
            Device::EnqueueDeletion(buffer, memory);
            buffer = VK_NULL_HANDLE;
        }

        void Resize(VkBufferUsageFlags2 usageFlags, VkMemoryPropertyFlags memoryPropertyFlags, std::uint32_t count, VkMemoryPropertyFlags preferredPropertyFlags = 0) {
            // Reuse the existing allocation in place when the request still fits
            // and the fixed-at-create properties match: usage flags are
            // immutable after creation, and the memory type already chosen must
            // still satisfy the required property flags. preferredPropertyFlags
            // is a best-effort perf hint and does not affect correctness, so it
            // is intentionally not part of the guard. The buffer handle (and its
            // device address / mapped pointer) is preserved, only `size` shrinks
            // to the new logical extent.
            std::uint32_t requestedSize = count * sizeof(T);
            if(buffer != VK_NULL_HANDLE
                && requestedSize <= capacity
                && usageFlags == usageFlagsCreated
                && (memoryPropertyFlagsChosen & memoryPropertyFlags) == memoryPropertyFlags) {
                size = requestedSize;
                return;
            }
            // Defer the old allocation's destruction: an in-flight frame may
            // still reference it (the #63 hazard, no longer masked by a
            // per-frame wait-idle since #40). DeferredClear nulls the handle,
            // so the Create below starts from a clean slate.
            if(buffer != VK_NULL_HANDLE) {
               DeferredClear();
            }
            Create(usageFlags, memoryPropertyFlags, count, preferredPropertyFlags);
        }

        void Copy(VkCommandBuffer cmd, VulkanBuffer& dst) {
            VkBufferCopy copyRegion = {
                .srcOffset = 0,
                .dstOffset = 0,
                .size = size
            };

            vkCmdCopyBuffer(
                cmd,
                buffer,
                dst.buffer,
                1,
                &copyRegion
            );
        }

        void Copy(VkCommandBuffer cmd, VulkanBuffer& dst, VkAccessFlags srcAccessMask, VkAccessFlags dstAccessMask, VkPipelineStageFlags srcStageMask, VkPipelineStageFlags dstStageMask) {
            Copy(cmd, dst);

            VkBufferMemoryBarrier barrier = {
                .sType = VK_STRUCTURE_TYPE_BUFFER_MEMORY_BARRIER,
                .srcAccessMask = srcAccessMask,
                .dstAccessMask = dstAccessMask,
                .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
                .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
                .buffer = buffer,
                .offset = 0,
                .size = VK_WHOLE_SIZE
            };

            vkCmdPipelineBarrier(
                cmd,
                srcStageMask,
                dstStageMask,
                0,
                0, NULL,
                1, &barrier,
                0, NULL
            );
        }

        // Upload `count` host elements from `src` into this buffer as
        // device-local, GPU-read geometry, choosing placement at runtime via
        // the #89 upload strategy (Device::PreferDirectDeviceWrite), and order
        // the upload before (dstStageMask, dstAccessMask) on `cmd`:
        //   ReBAR / UMA (direct) — allocate HOST_VISIBLE with DEVICE_LOCAL as a
        //     best-effort preference (GetMemoryType lands the host-visible
        //     device-local type that the strategy already proved exists), map
        //     transiently, memcpy, flush if the chosen type isn't coherent,
        //     unmap. No staging buffer: it would be pure overhead on a bar where
        //     the device-local heap is itself host-writable.
        //   No / small BAR (staged) — allocate pure DEVICE_LOCAL and copy from
        //     the staging ring on the GPU (CopyThroughStaging).
        // The direct write is only for a fresh allocation, which nothing can be
        // reading yet. A same-size re-upload reuses the allocation (Resize), so
        // the device address stays stable across an in-place AS UPDATE refit
        // that re-calls this, but a frame in flight may still read it: that
        // upload always copies, on the GPU timeline, after those reads.
        // dstStageMask must name every stage that reads the buffer, since the
        // next upload waits for them. Requires !Mapped: the destination must be
        // free to be device-local-only (a persistent map would force
        // HOST_VISIBLE), so the direct path maps just long enough to write.
        void UploadDeviceLocal(VkBufferUsageFlags2 usageFlags, const T* src, std::uint32_t count, VkCommandBuffer cmd, VkAccessFlags dstAccessMask, VkPipelineStageFlags dstStageMask) requires(!Mapped) {
            // Nothing to put anywhere: no buffer of no bytes, no copy of none.
            if (count == 0) return;
            const VkDeviceSize bytes = static_cast<VkDeviceSize>(count) * sizeof(T);
            // Any upload may be followed by one that copies in.
            usageFlags |= VK_BUFFER_USAGE_TRANSFER_DST_BIT;
            const bool direct = Device::PreferDirectDeviceWrite(bytes);
            const VkBuffer previous = buffer;
            if (direct) {
                Resize(usageFlags, VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT, count, VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT);
            } else {
                Resize(usageFlags, VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT, count);
            }
            if (!direct || buffer == previous) {
                CopyThroughStaging(src, 0, count, cmd, dstAccessMask, dstStageMask);
                return;
            }

            void* mapped = nullptr;
            Device::CheckVkResult(vkMapMemory(Device::device, memory, 0, VK_WHOLE_SIZE, 0, &mapped));
            std::memcpy(mapped, src, bytes);
            // Match FlushDevice()'s gate: a non-coherent type needs an
            // explicit flush before the device reads the written range.
            if (!(memoryPropertyFlagsChosen & VK_MEMORY_PROPERTY_HOST_COHERENT_BIT)) {
                VkMappedMemoryRange range {
                    .sType  = VK_STRUCTURE_TYPE_MAPPED_MEMORY_RANGE,
                    .memory = memory,
                    .offset = 0,
                    .size   = VK_WHOLE_SIZE
                };
                vkFlushMappedMemoryRanges(Device::device, 1, &range);
            }
            vkUnmapMemory(Device::device, memory);

            VkBufferMemoryBarrier barrier = {
                .sType = VK_STRUCTURE_TYPE_BUFFER_MEMORY_BARRIER,
                .srcAccessMask = VK_ACCESS_HOST_WRITE_BIT,
                .dstAccessMask = dstAccessMask,
                .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
                .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
                .buffer = buffer,
                .offset = 0,
                .size = VK_WHOLE_SIZE
            };
            vkCmdPipelineBarrier(cmd, VK_PIPELINE_STAGE_HOST_BIT, dstStageMask, 0, 0, NULL, 1, &barrier, 0, NULL);
        }

        // Re-upload only the half-open sub-range [offset, offset+count) of a
        // buffer UploadDeviceLocal made — the dirty-range counterpart of
        // UploadDeviceLocal. `src` points at the first changed element (not the
        // start of the whole array); `offset` is that element's index in this
        // buffer. This never Resizes, so the device address stays stable and the
        // untouched elements keep their last contents. Only the dirty elements
        // are staged and copied, and the barriers cover only that sub-range — so
        // a deforming-mesh refit that nudges a few vertices pays for those
        // vertices, not a full-array re-upload (#119). The buffer is one a frame
        // in flight may still read, so the copy waits for its readers
        // (dstStageMask, as in UploadDeviceLocal) on the GPU, whatever memory
        // type it lives in.
        void UploadDeviceLocalRange(const T* src, std::uint32_t offset, std::uint32_t count, VkCommandBuffer cmd, VkAccessFlags dstAccessMask, VkPipelineStageFlags dstStageMask) requires(!Mapped) {
            CopyThroughStaging(src, offset, count, cmd, dstAccessMask, dstStageMask);
        }

        // Copy `count` elements from `src` to element `offset` of this buffer on
        // the GPU timeline: after the reads by `dstStageMask` recorded earlier
        // or still in flight, and before the ones that follow. The bytes go
        // through this frame's slot of the staging ring first.
        void CopyThroughStaging(const T* src, std::uint32_t offset, std::uint32_t count, VkCommandBuffer cmd, VkAccessFlags dstAccessMask, VkPipelineStageFlags dstStageMask) requires(!Mapped) {
            // A copy or barrier of no bytes is invalid usage.
            if (count == 0) return;
            const VkDeviceSize byteOffset = static_cast<VkDeviceSize>(offset) * sizeof(T);
            const VkDeviceSize bytes      = static_cast<VkDeviceSize>(count) * sizeof(T);

            // Write after read: an execution dependency on the earlier readers.
            VkBufferMemoryBarrier readsDone = {
                .sType = VK_STRUCTURE_TYPE_BUFFER_MEMORY_BARRIER,
                .srcAccessMask = 0,
                .dstAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT,
                .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
                .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
                .buffer = buffer,
                .offset = byteOffset,
                .size = bytes
            };
            vkCmdPipelineBarrier(cmd, dstStageMask, VK_PIPELINE_STAGE_TRANSFER_BIT, 0, 0, NULL, 1, &readsDone, 0, NULL);

            const auto [staging, stagingOffset] = Stage(src, count);
            VkBufferCopy region { .srcOffset = stagingOffset, .dstOffset = byteOffset, .size = bytes };
            vkCmdCopyBuffer(cmd, staging, buffer, 1, &region);

            VkBufferMemoryBarrier copied = {
                .sType = VK_STRUCTURE_TYPE_BUFFER_MEMORY_BARRIER,
                .srcAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT,
                .dstAccessMask = dstAccessMask,
                .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
                .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
                .buffer = buffer,
                .offset = byteOffset,
                .size = bytes
            };
            vkCmdPipelineBarrier(cmd, VK_PIPELINE_STAGE_TRANSFER_BIT, dstStageMask, 0, 0, NULL, 1, &copied, 0, NULL);
        }

        // Put `count` elements in this frame's slot of the staging ring, after
        // whatever this frame staged before, and say where: the slot's buffer
        // and the byte offset. A copy still in flight never reads the slot, as
        // a slot is used again only framesInFlight frames later — the window
        // after which single-queue submission order (the same guarantee the
        // #101 deletion queue relies on) ensures that copy has completed. Each
        // slot keeps its high-water size (issue #120); outgrowing it mid-frame
        // hands the old allocation, with this frame's earlier bytes, to the
        // deletion queue. Without a window no frame passes; after a
        // Device::DrainDeletions (which follows a wait for the queue) nothing
        // reads any slot, so they all start over.
        std::pair<VkBuffer, VkDeviceSize> Stage(const T* src, std::uint32_t count) requires(!Mapped) {
            const std::uint32_t ringSize = Device::framesInFlight ? Device::framesInFlight : 1;
            if (stagingRing.size() < ringSize) {
                stagingRing.resize(ringSize);
                stagingFrame.resize(ringSize, std::numeric_limits<std::uint64_t>::max());
                stagingUsed.resize(ringSize, 0);
            }
            if (stagingDrains != Device::drains) {
                stagingDrains = Device::drains;
                std::ranges::fill(stagingUsed, 0);
            }
            const std::size_t slot = Device::frameCounter % ringSize;
            if (stagingFrame[slot] != Device::frameCounter) {
                stagingFrame[slot] = Device::frameCounter;
                stagingUsed[slot] = 0;
            }
            VulkanBuffer<T, true>& staging = stagingRing[slot];
            const std::uint32_t needed = stagingUsed[slot] + count;
            if (staging.buffer == VK_NULL_HANDLE || static_cast<VkDeviceSize>(needed) * sizeof(T) > staging.capacity) {
                staging.Resize(VK_BUFFER_USAGE_TRANSFER_SRC_BIT, VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT, needed);
                stagingUsed[slot] = 0;
            }
            const std::uint32_t first = stagingUsed[slot];
            std::memcpy(staging.value + first, src, static_cast<std::size_t>(count) * sizeof(T));
            staging.FlushDevice(static_cast<VkDeviceSize>(first) * sizeof(T), static_cast<VkDeviceSize>(count) * sizeof(T));
            stagingUsed[slot] = first + count;
            return { staging.buffer, static_cast<VkDeviceSize>(first) * sizeof(T) };
        }

        void FlushDevice() requires(Mapped) {
            // Coherent memory needs no explicit flush — host writes are
            // automatically visible to the device.
            if (memoryPropertyFlagsChosen & VK_MEMORY_PROPERTY_HOST_COHERENT_BIT) {
                return;
            }
            VkMappedMemoryRange range = {
                .sType  = VK_STRUCTURE_TYPE_MAPPED_MEMORY_RANGE,
                .memory = memory,
                .offset = 0,
                .size   = VK_WHOLE_SIZE
            };
            vkFlushMappedMemoryRanges(Device::device, 1, &range);
        }

        // Flush only the host writes in [offset, offset+bytes) to the device,
        // instead of the whole buffer. Use after touching a small sub-range
        // (e.g. one descriptor in a multi-KB descriptor heap) so cache
        // maintenance scales with the bytes actually written. The range is
        // rounded outward to nonCoherentAtomSize as the Vulkan spec requires.
        // No-op on coherent memory, same gate as the whole-buffer FlushDevice().
        void FlushDevice(VkDeviceSize offset, VkDeviceSize bytes) requires(Mapped) {
            if (memoryPropertyFlagsChosen & VK_MEMORY_PROPERTY_HOST_COHERENT_BIT) {
                return;
            }
            MappedFlushRange r = AlignMappedFlushRange(
                offset, bytes, Device::nonCoherentAtomSize, mappedSize);
            VkMappedMemoryRange range = {
                .sType  = VK_STRUCTURE_TYPE_MAPPED_MEMORY_RANGE,
                .memory = memory,
                .offset = r.offset,
                .size   = r.size
            };
            vkFlushMappedMemoryRanges(Device::device, 1, &range);
        }

        void FlushDevice(VkCommandBuffer cmd, VkAccessFlags dstAccessMask, VkPipelineStageFlags dstStageMask) requires(Mapped) {
            FlushDevice();
            VkBufferMemoryBarrier barrier = {
                .sType = VK_STRUCTURE_TYPE_BUFFER_MEMORY_BARRIER,
                .srcAccessMask = VK_ACCESS_HOST_WRITE_BIT,
                .dstAccessMask = dstAccessMask,
                .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
                .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
                .buffer = buffer,
                .offset = 0,
                .size = VK_WHOLE_SIZE
            };

            vkCmdPipelineBarrier(
                cmd,
                VK_PIPELINE_STAGE_HOST_BIT,
                dstStageMask,
                0,
                0, NULL,
                1, &barrier,
                0, NULL
            );
        }

        // Ranged variant of FlushDevice(cmd, ...): flushes only the host writes
        // in [offset, offset+bytes) (rounded outward to nonCoherentAtomSize, and
        // a no-op on coherent memory — same gate as the other FlushDevice
        // overloads) and records the HOST->(dstStageMask, dstAccessMask) barrier.
        // Use after writing a sub-range so the cache-flush cost scales with the
        // bytes touched rather than the whole high-water capacity. The barrier
        // itself still spans the whole buffer (VK_WHOLE_SIZE): the execution/
        // visibility dependency is cheap regardless of range, and only the
        // flush's cache maintenance is bandwidth-sensitive.
        void FlushDevice(VkCommandBuffer cmd, VkAccessFlags dstAccessMask, VkPipelineStageFlags dstStageMask, VkDeviceSize offset, VkDeviceSize bytes) requires(Mapped) {
            FlushDevice(offset, bytes);
            VkBufferMemoryBarrier barrier = {
                .sType = VK_STRUCTURE_TYPE_BUFFER_MEMORY_BARRIER,
                .srcAccessMask = VK_ACCESS_HOST_WRITE_BIT,
                .dstAccessMask = dstAccessMask,
                .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
                .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
                .buffer = buffer,
                .offset = 0,
                .size = VK_WHOLE_SIZE
            };

            vkCmdPipelineBarrier(
                cmd,
                VK_PIPELINE_STAGE_HOST_BIT,
                dstStageMask,
                0,
                0, NULL,
                1, &barrier,
                0, NULL
            );
        }

        void FlushHost() requires(Mapped) {
            // Coherent memory needs no explicit invalidate — device writes are
            // automatically visible to the host.
            if (memoryPropertyFlagsChosen & VK_MEMORY_PROPERTY_HOST_COHERENT_BIT) {
                return;
            }
            VkMappedMemoryRange range = {
                .sType  = VK_STRUCTURE_TYPE_MAPPED_MEMORY_RANGE,
                .memory = memory,
                .offset = 0,
                .size   = VK_WHOLE_SIZE
            };
            vkInvalidateMappedMemoryRanges(Device::device, 1, &range);
        }
    
        // Persistent per-frame-in-flight staging ring for the copies Stage
        // records (issue #120). Empty — and so holding zero host-visible
        // allocations — until the first upload that stages: on ReBAR/UMA only
        // re-uploads do, as a first upload writes directly. Sized to
        // Device::framesInFlight so each in-flight frame copies from its own
        // slot; each slot grows to a high-water mark via Resize and is reused
        // across frames rather than reallocated every call — mirroring the TLAS
        // instance/metadata reuse. stagingFrame and stagingUsed say, per slot,
        // which frame last staged into it and how many elements it has staged;
        // stagingDrains, the Device::drains it last saw.
        std::vector<VulkanBuffer<T, true>> stagingRing;
        std::vector<std::uint64_t> stagingFrame;
        std::vector<std::uint32_t> stagingUsed;
        std::uint64_t stagingDrains = 0;

        VulkanBuffer(VulkanBuffer&& other) noexcept {
            *this = std::move(other);
        }

        // Takes `other`'s buffer over. The one this held goes to the deletion
        // queue, as the destructor would send it: a frame in flight may still
        // read it.
        VulkanBuffer& operator=(VulkanBuffer&& other) noexcept {
            if (this == &other) return *this;
            DeferredClear();
            buffer = other.buffer;
            memory = other.memory;
            size = other.size;
            mappedSize = other.mappedSize;
            capacity = other.capacity;
            usageFlagsCreated = other.usageFlagsCreated;
            memoryPropertyFlagsChosen = other.memoryPropertyFlagsChosen;
            other.buffer = VK_NULL_HANDLE;
            address = other.address;
            stagingRing = std::move(other.stagingRing);
            stagingFrame = std::move(other.stagingFrame);
            stagingUsed = std::move(other.stagingUsed);
            stagingDrains = other.stagingDrains;
            if constexpr(Mapped) {
                VulkanBufferMappedConditional<T, true>::value = other.VulkanBufferMappedConditional<T, true>::value;
            }
            return *this;
        }

        // Deferred, like DeferredClear: the owner may go while a frame in
        // flight still reads the buffer, and a static owner goes after the
        // driver has torn down. Call Clear() first to free at once when the
        // GPU is known to be idle. The staging ring's buffers defer the same way.
        ~VulkanBuffer() {
            DeferredClear();
        }

        VulkanBuffer(VulkanBuffer&) = delete;
        VulkanBuffer& operator=(const VulkanBuffer&) = delete;
    };
}
#endif // !CRAFTER_GRAPHICS_WINDOW_DOM