mesa/src/gfxstream/guest/vulkan_enc/DescriptorSetVirtualization.cpp

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

550 lines
20 KiB
C++
Raw Normal View History

/*
* Copyright 2021 Google LLC
* SPDX-License-Identifier: MIT
*/
#include "DescriptorSetVirtualization.h"
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
#include "Resources.h"
#include "util/log.h"
namespace gfxstream {
namespace vk {
void clearReifiedDescriptorSet(ReifiedDescriptorSet* set) {
set->pool = VK_NULL_HANDLE;
set->setLayout = VK_NULL_HANDLE;
set->poolId = -1;
set->allocationPending = false;
set->allWrites.clear();
set->pendingWriteArrayRanges.clear();
}
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
void initDescriptorWriteTable(const std::vector<VkDescriptorSetLayoutBinding>& layoutBindings,
DescriptorWriteTable& table) {
uint32_t highestBindingNumber = 0;
for (uint32_t i = 0; i < layoutBindings.size(); ++i) {
if (layoutBindings[i].binding > highestBindingNumber) {
highestBindingNumber = layoutBindings[i].binding;
}
}
std::vector<uint32_t> countsEachBinding(highestBindingNumber + 1, 0);
for (uint32_t i = 0; i < layoutBindings.size(); ++i) {
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
countsEachBinding[layoutBindings[i].binding] = layoutBindings[i].descriptorCount;
}
table.resize(countsEachBinding.size());
for (uint32_t i = 0; i < table.size(); ++i) {
table[i].resize(countsEachBinding[i]);
for (uint32_t j = 0; j < countsEachBinding[i]; ++j) {
table[i][j].type = DescriptorWriteType::Empty;
table[i][j].dstArrayElement = 0;
}
}
}
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
static void initializeReifiedDescriptorSet(VkDescriptorPool pool, VkDescriptorSetLayout setLayout,
ReifiedDescriptorSet* set) {
set->pendingWriteArrayRanges.clear();
const auto& layoutInfo = *(as_goldfish_VkDescriptorSetLayout(setLayout)->layoutInfo);
initDescriptorWriteTable(layoutInfo.bindings, set->allWrites);
for (size_t i = 0; i < layoutInfo.bindings.size(); ++i) {
// Bindings can be sparsely defined
const auto& binding = layoutInfo.bindings[i];
uint32_t bindingIndex = binding.binding;
if (set->bindingIsImmutableSampler.size() <= bindingIndex) {
set->bindingIsImmutableSampler.resize(bindingIndex + 1, false);
}
set->bindingIsImmutableSampler[bindingIndex] =
binding.descriptorCount > 0 &&
(binding.descriptorType == VK_DESCRIPTOR_TYPE_SAMPLER ||
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
binding.descriptorType == VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER) &&
binding.pImmutableSamplers;
}
set->pool = pool;
set->setLayout = setLayout;
set->allocationPending = true;
set->bindings = layoutInfo.bindings;
}
bool isDescriptorTypeImageInfo(VkDescriptorType descType) {
return (descType == VK_DESCRIPTOR_TYPE_SAMPLER) ||
(descType == VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER) ||
(descType == VK_DESCRIPTOR_TYPE_SAMPLED_IMAGE) ||
(descType == VK_DESCRIPTOR_TYPE_STORAGE_IMAGE) ||
(descType == VK_DESCRIPTOR_TYPE_INPUT_ATTACHMENT);
}
bool isDescriptorTypeBufferInfo(VkDescriptorType descType) {
return (descType == VK_DESCRIPTOR_TYPE_UNIFORM_BUFFER) ||
(descType == VK_DESCRIPTOR_TYPE_UNIFORM_BUFFER_DYNAMIC) ||
(descType == VK_DESCRIPTOR_TYPE_STORAGE_BUFFER) ||
(descType == VK_DESCRIPTOR_TYPE_STORAGE_BUFFER_DYNAMIC);
}
bool isDescriptorTypeBufferView(VkDescriptorType descType) {
return (descType == VK_DESCRIPTOR_TYPE_UNIFORM_TEXEL_BUFFER) ||
(descType == VK_DESCRIPTOR_TYPE_STORAGE_TEXEL_BUFFER);
}
bool isDescriptorTypeInlineUniformBlock(VkDescriptorType descType) {
return descType == VK_DESCRIPTOR_TYPE_INLINE_UNIFORM_BLOCK_EXT;
}
bool isDescriptorTypeAccelerationStructure(VkDescriptorType descType) {
return descType == VK_DESCRIPTOR_TYPE_ACCELERATION_STRUCTURE_KHR;
}
void doEmulatedDescriptorWrite(const VkWriteDescriptorSet* write, ReifiedDescriptorSet* toWrite) {
VkDescriptorType descType = write->descriptorType;
uint32_t dstBinding = write->dstBinding;
uint32_t dstArrayElement = write->dstArrayElement;
uint32_t descriptorCount = write->descriptorCount;
DescriptorWriteTable& table = toWrite->allWrites;
uint32_t arrOffset = dstArrayElement;
if (isDescriptorTypeImageInfo(descType)) {
uint32_t i = 0;
while (i < descriptorCount) {
assert(dstBinding < table.size());
if (arrOffset >= table[dstBinding].size()) {
++dstBinding;
arrOffset = 0;
continue;
}
auto& entry = table[dstBinding][arrOffset];
entry.imageInfo = write->pImageInfo[i];
entry.type = DescriptorWriteType::ImageInfo;
entry.descriptorType = descType;
++i;
++arrOffset;
}
} else if (isDescriptorTypeBufferInfo(descType)) {
uint32_t i = 0;
while (i < descriptorCount) {
assert(dstBinding < table.size());
if (arrOffset >= table[dstBinding].size()) {
++dstBinding;
arrOffset = 0;
continue;
}
auto& entry = table[dstBinding][arrOffset];
entry.bufferInfo = write->pBufferInfo[i];
entry.type = DescriptorWriteType::BufferInfo;
entry.descriptorType = descType;
++i;
++arrOffset;
}
} else if (isDescriptorTypeBufferView(descType)) {
uint32_t i = 0;
while (i < descriptorCount) {
assert(dstBinding < table.size());
if (arrOffset >= table[dstBinding].size()) {
++dstBinding;
arrOffset = 0;
continue;
}
auto& entry = table[dstBinding][arrOffset];
entry.bufferView = write->pTexelBufferView[i];
entry.type = DescriptorWriteType::BufferView;
entry.descriptorType = descType;
++i;
++arrOffset;
}
} else if (isDescriptorTypeInlineUniformBlock(descType)) {
const VkWriteDescriptorSetInlineUniformBlock* descInlineUniformBlock =
static_cast<const VkWriteDescriptorSetInlineUniformBlock*>(write->pNext);
while (descInlineUniformBlock &&
descInlineUniformBlock->sType !=
VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET_INLINE_UNIFORM_BLOCK) {
descInlineUniformBlock = static_cast<const VkWriteDescriptorSetInlineUniformBlock*>(
descInlineUniformBlock->pNext);
}
if (!descInlineUniformBlock) {
mesa_loge("%s: did not find inline uniform block\n", __func__);
return;
}
auto& entry = table[dstBinding][0];
entry.inlineUniformBlock = *descInlineUniformBlock;
entry.inlineUniformBlockBuffer.assign(
static_cast<const uint8_t*>(descInlineUniformBlock->pData),
static_cast<const uint8_t*>(descInlineUniformBlock->pData) +
descInlineUniformBlock->dataSize);
entry.type = DescriptorWriteType::InlineUniformBlock;
entry.descriptorType = descType;
entry.dstArrayElement = dstArrayElement;
} else if (isDescriptorTypeAccelerationStructure(descType)) {
// TODO
// Look for pNext inline uniform block or acceleration structure.
// Append new DescriptorWrite entry that holds the buffer
mesa_logw("%s: Ignoring emulated write for descriptor type 0x%x\n", __func__, descType);
}
}
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
void doEmulatedDescriptorCopy(const VkCopyDescriptorSet* copy, const ReifiedDescriptorSet* src,
ReifiedDescriptorSet* dst) {
const DescriptorWriteTable& srcTable = src->allWrites;
DescriptorWriteTable& dstTable = dst->allWrites;
// src/dst may be the same descriptor set, so we need to create a temporary array for that case.
// (TODO: Maybe just notice the pointers are the same? can aliasing in any other way happen?)
std::vector<DescriptorWrite> toCopy;
uint32_t currBinding = copy->srcBinding;
uint32_t arrOffset = copy->srcArrayElement;
uint32_t i = 0;
while (i < copy->descriptorCount) {
assert(currBinding < srcTable.size());
if (arrOffset >= srcTable[currBinding].size()) {
++currBinding;
arrOffset = 0;
continue;
}
toCopy.push_back(srcTable[currBinding][arrOffset]);
++i;
++arrOffset;
}
currBinding = copy->dstBinding;
arrOffset = copy->dstArrayElement;
i = 0;
while (i < copy->descriptorCount) {
assert(currBinding < dstTable.size());
if (arrOffset >= dstTable[currBinding].size()) {
++currBinding;
arrOffset = 0;
continue;
}
dstTable[currBinding][arrOffset] = toCopy[i];
++i;
++arrOffset;
}
}
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
void doEmulatedDescriptorImageInfoWriteFromTemplate(VkDescriptorType descType, uint32_t binding,
uint32_t dstArrayElement, uint32_t count,
const VkDescriptorImageInfo* imageInfos,
ReifiedDescriptorSet* set) {
DescriptorWriteTable& table = set->allWrites;
uint32_t currBinding = binding;
uint32_t arrOffset = dstArrayElement;
uint32_t i = 0;
while (i < count) {
assert(currBinding < table.size());
if (arrOffset >= table[currBinding].size()) {
++currBinding;
arrOffset = 0;
continue;
}
auto& entry = table[currBinding][arrOffset];
entry.imageInfo = imageInfos[i];
entry.type = DescriptorWriteType::ImageInfo;
entry.descriptorType = descType;
++i;
++arrOffset;
}
}
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
void doEmulatedDescriptorBufferInfoWriteFromTemplate(VkDescriptorType descType, uint32_t binding,
uint32_t dstArrayElement, uint32_t count,
const VkDescriptorBufferInfo* bufferInfos,
ReifiedDescriptorSet* set) {
DescriptorWriteTable& table = set->allWrites;
uint32_t currBinding = binding;
uint32_t arrOffset = dstArrayElement;
uint32_t i = 0;
while (i < count) {
assert(currBinding < table.size());
if (arrOffset >= table[currBinding].size()) {
++currBinding;
arrOffset = 0;
continue;
}
auto& entry = table[currBinding][arrOffset];
entry.bufferInfo = bufferInfos[i];
entry.type = DescriptorWriteType::BufferInfo;
entry.descriptorType = descType;
++i;
++arrOffset;
}
}
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
void doEmulatedDescriptorBufferViewWriteFromTemplate(VkDescriptorType descType, uint32_t binding,
uint32_t dstArrayElement, uint32_t count,
const VkBufferView* bufferViews,
ReifiedDescriptorSet* set) {
DescriptorWriteTable& table = set->allWrites;
uint32_t currBinding = binding;
uint32_t arrOffset = dstArrayElement;
uint32_t i = 0;
while (i < count) {
assert(currBinding < table.size());
if (arrOffset >= table[currBinding].size()) {
++currBinding;
arrOffset = 0;
continue;
}
auto& entry = table[currBinding][arrOffset];
entry.bufferView = bufferViews[i];
entry.type = DescriptorWriteType::BufferView;
entry.descriptorType = descType;
++i;
++arrOffset;
}
}
void doEmulatedDescriptorInlineUniformBlockFromTemplate(VkDescriptorType descType, uint32_t binding,
uint32_t dstArrayElement, uint32_t count,
const void* pData,
ReifiedDescriptorSet* set) {
DescriptorWriteTable& table = set->allWrites;
auto& entry = table[binding][0];
entry.dstArrayElement = dstArrayElement;
entry.inlineUniformBlockBuffer.assign(static_cast<const uint8_t*>(pData),
static_cast<const uint8_t*>(pData) + count);
entry.type = DescriptorWriteType::InlineUniformBlock;
entry.descriptorType = descType;
}
static bool isBindingFeasibleForAlloc(
const DescriptorPoolAllocationInfo::DescriptorCountInfo& countInfo,
const VkDescriptorSetLayoutBinding& binding) {
if (binding.descriptorCount && (countInfo.type != binding.descriptorType)) {
return false;
}
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
uint32_t availDescriptorCount = countInfo.descriptorCount - countInfo.used;
if (availDescriptorCount < binding.descriptorCount) {
mesa_logi(
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
"%s: Ran out of descriptors of type 0x%x. "
"Wanted %u from layout but "
"we only have %u free (total in pool: %u)\n",
__func__, binding.descriptorType, binding.descriptorCount,
countInfo.descriptorCount - countInfo.used, countInfo.descriptorCount);
return false;
}
return true;
}
static bool isBindingFeasibleForFree(
const DescriptorPoolAllocationInfo::DescriptorCountInfo& countInfo,
const VkDescriptorSetLayoutBinding& binding) {
if (countInfo.type != binding.descriptorType) return false;
if (countInfo.used < binding.descriptorCount) {
mesa_logi(
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
"%s: Was a descriptor set double freed? "
"Ran out of descriptors of type 0x%x. "
"Wanted to free %u from layout but "
"we only have %u used (total in pool: %u)\n",
__func__, binding.descriptorType, binding.descriptorCount, countInfo.used,
countInfo.descriptorCount);
return false;
}
return true;
}
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
static void allocBindingFeasible(const VkDescriptorSetLayoutBinding& binding,
DescriptorPoolAllocationInfo::DescriptorCountInfo& poolState) {
poolState.used += binding.descriptorCount;
}
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
static void freeBindingFeasible(const VkDescriptorSetLayoutBinding& binding,
DescriptorPoolAllocationInfo::DescriptorCountInfo& poolState) {
poolState.used -= binding.descriptorCount;
}
static VkResult validateDescriptorSetAllocation(const VkDescriptorSetAllocateInfo* pAllocateInfo) {
VkDescriptorPool pool = pAllocateInfo->descriptorPool;
DescriptorPoolAllocationInfo* poolInfo = as_goldfish_VkDescriptorPool(pool)->allocInfo;
// Check the number of sets available.
auto setsAvailable = poolInfo->maxSets - poolInfo->usedSets;
if (setsAvailable < pAllocateInfo->descriptorSetCount) {
mesa_logi(
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
"%s: Error: VkDescriptorSetAllocateInfo wants %u sets "
"but we only have %u available. "
"Bailing with VK_ERROR_OUT_OF_POOL_MEMORY.\n",
__func__, pAllocateInfo->descriptorSetCount, setsAvailable);
return VK_ERROR_OUT_OF_POOL_MEMORY;
}
// Perform simulated allocation and error out with
// VK_ERROR_OUT_OF_POOL_MEMORY if it fails.
std::vector<DescriptorPoolAllocationInfo::DescriptorCountInfo> descriptorCountCopy =
poolInfo->descriptorCountInfo;
for (uint32_t i = 0; i < pAllocateInfo->descriptorSetCount; ++i) {
if (!pAllocateInfo->pSetLayouts[i]) {
mesa_logi("%s: Error: Tried to allocate a descriptor set with null set layout.\n",
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
__func__);
return VK_ERROR_INITIALIZATION_FAILED;
}
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
auto setLayoutInfo =
as_goldfish_VkDescriptorSetLayout(pAllocateInfo->pSetLayouts[i])->layoutInfo;
if (!setLayoutInfo) {
return VK_ERROR_INITIALIZATION_FAILED;
}
for (const auto& binding : setLayoutInfo->bindings) {
bool success = false;
for (auto& pool : descriptorCountCopy) {
if (!isBindingFeasibleForAlloc(pool, binding)) continue;
success = true;
allocBindingFeasible(binding, pool);
break;
}
if (!success) {
return VK_ERROR_OUT_OF_POOL_MEMORY;
}
}
}
return VK_SUCCESS;
}
void applyDescriptorSetAllocation(VkDescriptorPool pool, VkDescriptorSetLayout setLayout) {
auto allocInfo = as_goldfish_VkDescriptorPool(pool)->allocInfo;
auto setLayoutInfo = as_goldfish_VkDescriptorSetLayout(setLayout)->layoutInfo;
++allocInfo->usedSets;
for (const auto& binding : setLayoutInfo->bindings) {
for (auto& countForPool : allocInfo->descriptorCountInfo) {
if (!isBindingFeasibleForAlloc(countForPool, binding)) continue;
allocBindingFeasible(binding, countForPool);
break;
}
}
}
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
void removeDescriptorSetAllocation(VkDescriptorPool pool,
const std::vector<VkDescriptorSetLayoutBinding>& bindings) {
auto allocInfo = as_goldfish_VkDescriptorPool(pool)->allocInfo;
if (0 == allocInfo->usedSets) {
mesa_logi("%s: Warning: a descriptor set was double freed.\n", __func__);
return;
}
--allocInfo->usedSets;
for (const auto& binding : bindings) {
for (auto& countForPool : allocInfo->descriptorCountInfo) {
if (!isBindingFeasibleForFree(countForPool, binding)) continue;
freeBindingFeasible(binding, countForPool);
break;
}
}
}
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
void fillDescriptorSetInfoForPool(VkDescriptorPool pool, VkDescriptorSetLayout setLayout,
VkDescriptorSet set) {
DescriptorPoolAllocationInfo* allocInfo = as_goldfish_VkDescriptorPool(pool)->allocInfo;
ReifiedDescriptorSet* newReified = new ReifiedDescriptorSet;
newReified->poolId = as_goldfish_VkDescriptorSet(set)->underlying;
newReified->allocationPending = true;
as_goldfish_VkDescriptorSet(set)->reified = newReified;
allocInfo->allocedPoolIds.insert(newReified->poolId);
allocInfo->allocedSets.insert(set);
initializeReifiedDescriptorSet(pool, setLayout, newReified);
}
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
VkResult validateAndApplyVirtualDescriptorSetAllocation(
const VkDescriptorSetAllocateInfo* pAllocateInfo, VkDescriptorSet* pSets) {
VkResult validateRes = validateDescriptorSetAllocation(pAllocateInfo);
if (validateRes != VK_SUCCESS) return validateRes;
for (uint32_t i = 0; i < pAllocateInfo->descriptorSetCount; ++i) {
applyDescriptorSetAllocation(pAllocateInfo->descriptorPool, pAllocateInfo->pSetLayouts[i]);
}
VkDescriptorPool pool = pAllocateInfo->descriptorPool;
DescriptorPoolAllocationInfo* allocInfo = as_goldfish_VkDescriptorPool(pool)->allocInfo;
if (allocInfo->freePoolIds.size() < pAllocateInfo->descriptorSetCount) {
mesa_loge(
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
"%s: FATAL: Somehow out of descriptor pool IDs. Wanted %u IDs but only have %u free "
"IDs remaining. The count for maxSets was %u and used was %u\n",
__func__, pAllocateInfo->descriptorSetCount, (uint32_t)allocInfo->freePoolIds.size(),
allocInfo->maxSets, allocInfo->usedSets);
abort();
}
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
for (uint32_t i = 0; i < pAllocateInfo->descriptorSetCount; ++i) {
uint64_t id = allocInfo->freePoolIds.back();
allocInfo->freePoolIds.pop_back();
VkDescriptorSet newSet = new_from_host_VkDescriptorSet((VkDescriptorSet)id);
pSets[i] = newSet;
fillDescriptorSetInfoForPool(pool, pAllocateInfo->pSetLayouts[i], newSet);
}
return VK_SUCCESS;
}
bool removeDescriptorSetFromPool(VkDescriptorSet set, bool usePoolIds) {
ReifiedDescriptorSet* reified = as_goldfish_VkDescriptorSet(set)->reified;
VkDescriptorPool pool = reified->pool;
DescriptorPoolAllocationInfo* allocInfo = as_goldfish_VkDescriptorPool(pool)->allocInfo;
if (usePoolIds) {
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
// Look for the set's pool Id in the pool. If not found, then this wasn't really allocated,
// and bail.
if (allocInfo->allocedPoolIds.find(reified->poolId) == allocInfo->allocedPoolIds.end()) {
return false;
}
}
const std::vector<VkDescriptorSetLayoutBinding>& bindings = reified->bindings;
removeDescriptorSetAllocation(pool, bindings);
if (usePoolIds) {
allocInfo->freePoolIds.push_back(reified->poolId);
allocInfo->allocedPoolIds.erase(reified->poolId);
}
allocInfo->allocedSets.erase(set);
return true;
}
std::vector<VkDescriptorSet> clearDescriptorPool(VkDescriptorPool pool, bool usePoolIds) {
std::vector<VkDescriptorSet> toClear;
for (auto set : as_goldfish_VkDescriptorPool(pool)->allocInfo->allocedSets) {
toClear.push_back(set);
}
gfxstream: mega-change to support guest Linux WSI with gfxstream This is a mega-change to support Linux guest WSI with gfxstream. We tried to do a branch where every commit was buildable and runnable, but that quickly proved unworkable. So we squashed the branch into a mega-change. Zink provides the GL implementation for Linux guests, so we just needed to implement the proper Vulkan Wayland/X11 WSI entrypoints. The overall strategy to support this is to use Mesa's WSI functions. The Vulkan WSI layer was also considered: https://gitlab.freedesktop.org/mesa/vulkan-wsi-layer But it was less maintained compared to Mesa. The way Mesa common layers communicate with drivers is the through base objects embedded in driver and a common dispatch layer: https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/dispatch.rst https://gitlab.freedesktop.org/mesa/mesa/-/blob/main/docs/vulkan/base-objs.rst Our objects are defined in gfxstream_vk_private.h. Currently, Mesa-derived Vulkan objects just serve as shim to gfxstream Vulkan’s internal handle mapping. Long-term, we can use Mesa-derived objects inside gfxstream guest Vulkan exclusively. The flow is typically inside a Vulkan entrypoint is: - VK_FROM_HANDLE(vk-object) to convert to a gfxstream_vk_obj object - Call ResourceTracker::func(gfxstream_vk_obj->internal) or VkEncoder::func(gfxstream_vk_obj>internal) - Return result A good follow-up cleanup would be to delete older gfxstream objects. For example, we now have struct gfxstream_vk_device and info_VkDevice in ResourceTracker. Most of this logic was auto-generated and included in func_table.cpp. Some vulkan functions were too difficult to auto-generate or required special logic, and these are included in gfxstream_vk_device.cpp. For example, anything that needs to setup the HostConnection requires special handling. Android Blueprint support is added to the parts of Mesa needed to build the Vulkan runtime. One thing to call out it's required to build the guest/vulkan_enc and guest/vulkan files in the same shared library now, when previously have libvulkan_enc.so and libvulkan_ranchu.so was sufficient [otherwise, some weak pointer logic wouldn't work]. A side effect of this is libOpenglSystem must also be a static lib, and so should libandroid_aemu too. That conceptually makes sense and the Meson build had been doing this all a long. We can probably transition everything besides libGLESv1_emulation.so, libGLESv2_emulation.so and libvulkan_ranchu.so to be static. This requires changes in the end2end tests, because since each HostConnection is separate and internal to it's constituent library. Lifetimes need to be managed separately: for example the HostConnection instance created by the end2end tests would not be visible inside libvulkan_ranchu.so anymore. Probably the best solution would be improve the testing facade so a HostConnection represents one virtio-gpu context, while some other entity represents a virtio-gpu device (client-server would work). vk.xml was modified, but change sent to Khronos: https://gitlab.khronos.org/vulkan/vulkan/-/merge_requests/6325 Fuchsia builds still need to be migrated, but they already have Fuchsia Mesa with all the build rules so that shouldn't be too bad. Just need to copy them over the gfxstream/Mesa hybrid. The new command for building Linux guests is: meson amd64-build/ -Dvulkan-drivers="gfxstream" -Dgallium-drivers="" -Dvk-no-nir=true -Dopengl=false Big shout-out to Aaron Ruby, who did most of the gnarly codegen needed to get the function table logic to work. * Run Weston/vkcube on Linux and automotive platform * launch_cvd --gpu_mode=gfxstream vkcube * launch_cvd --gpu_mod=gfxstream_guest_angle * vkcube + 3D Mark Slingshot extreme work with guest ANGLE and GL-VK interop * GfxstreamEnd2EndTests * Some select dEQP tests Aaron Ruby (46): gfxstream: function table: remove entry points that are hand-written. gfxstream: function table: more changes gfxstream: function table: scope internal_arrays to encoder gfxstream: function table: autogenerate compoundType params gfxstream: add handwritten EnumeratePhysicalDeviceGroup entrypoint. gfxstream: function table: handle nested handle arrays gfxstream: function table: adding some handwritten implementations gfxstream: revert some unnecessary changes gfxstream: use vk_object_zalloc/free instead of vk_zalloc/free. gfxstream: revert most gfxstream objects to use vk_object_base gfxstream: function table: handwritten commmand-buffers/pools gfxstream: codegen functionality to handle special param gfxstream: function table: random fixes gfxstream: add vk_command_buffer_ops handlers gfxstream: func_table.py: Codegen support for nested compound type gfxstream: remove handwritten/add autogen entry points gfxstream: add gfxstream_vk_device.cpp gfxstream: query device and instance extensions early gfxstream: func_table: explicit allocation for nested arrays/compound types gfxstream: goldfish_vulkan: fix commandBuffer allocation. gfxstream: meson: Raise api_version in ICD config to 1.1. gfxstream: function table: add more handwritten entries gfxstream: goldfish_vulkan: update VkDescriptorSetAllocateInfo logic gfxstream: function table: NULL check on internal_object dereference gfxstream: function table: Remove POSTPROCESSES handling from functable gfxstream: mesa: Add 'gfxstream' as a -Dvulkan-drivers gfxstream: ResourceTracker: add some allowedExtensions gfxstream: gfxstream_vk_device: add wsi_common_entrypoints gfxstream: Move instance handling into gfxstream_vk_device.cpp gfxstream: ResourceTracker: Enable Linux WSI-related extensions gfxstream: wsi: add wsi_device initialization gfxstream: gfxstream_vk_device: use Mesa common physical device management gfxstream: ResourceTracker: translate mesa objects in user buffer gfxstream: exclude VkSampler and VkDescriptorSet objects from translation gfxstream: Add guest-side external memory support with colorBuffers. gfxstream: function table: Modify semaphoreList inputs to no-op semaphores gfxstream: function table: Allow VK_NULL_HANDLE for free/destroy APIs. gfxstream: cereal: Add VK_EXT_depth_clip_enable as supported feature. gfxstream: vulkan_enc: un-namespace vk_util.h and vk_struct_id.h gfxstream: gfxstream_vk_device.cpp: Support VK_KHR_surface and VK_*_surface gfxstream: vulkan_enc: Add support for Mesa-only extensions. gfxstream: ResourceTracker: Use DEVICE_TYPE_VIRTUAL_GPU always gfxstream: platform: add dma-buf export support with dedicatedBuffer. gfxstream: ResourceTracker: add VK_EXT_depth_clip_enable allowed extension gfxstream: ResourceTracker: external memory via QNX_screen_buffer extension gfxstream: Add VK_QNX_external_memory_screen_buffer to VulkanDispatch Gurchetan Singh (18): gfxstream: mesa: write Android.bp files gfxstream: generate gfxstream_vk_entrypoints.{c, h} gfxstream: vulkan_enc: add gfxstream_vk_private.h (objects) gfxstream: function table: modify function table to use gfxstream_vk_* gfxstream: compiles gfxstream: build system improvements gfxstream: ResourceTracker: don't crash without VkBindImageMemorySwapchainInfoKHR gfxstream: vk.xml: make some vkAcquireImageANDROID params optional gfxstream_vk_device: filter out swapchain maintenance guest side gfxstream: end2end: fixes for End2End tests gfxstream: func_table: custom vkEnumerateInstanceLayerProperties gfxstream: add VK_EXT_DEBUG_UTILS_EXTENSION_NAME into Mesa list gfxstream: clang-format guest code gfxstream: libandroid AEMU static gfxstream: vkEnumerateInstanceVersion gfxstream: vkCreateComputePipeLines gfxstream: make end2end tests happy gfxstream: delete prior vk.xml, vk_icd_gen.py Reviewed-by: Aaron Ruby <aruby@blackberry.com> Acked-by: Yonggang Luo <luoyonggang@gmail.com> Acked-by: Adam Jackson <ajax@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27246>
2023-09-19 08:06:20 -07:00
for (auto set : toClear) {
removeDescriptorSetFromPool(set, usePoolIds);
}
return toClear;
}
} // namespace vk
} // namespace gfxstream