XFESystem

【Rust】XFE系统，基于Rust和汇编

公开

关注 0 Fork 0 Star 0

返回提交历史

XFEstudio/XFESystem

Record RTX 5090 port validation and verified diagnostic USB

5190c06

XFE工作室室长 <mail@xfegzs.com>

提交于 2026-10-15 17:53

父提交 690d6d5

代码差异

5 个文件 +328 -4

Modified AGENTS.md +6 -0

@@ -61,3 +61,9 @@

61 61 ## 后续开发基线

62 62

63 63

2026-10-15 用户明确要求：将自研内核合并到当前开发分支 `codex/csharp-system`，保留现有桌面、浏览器和文档功能，后续开发全部基于自研内核。默认构建、启动、打包与新功能实现以 `Source/XFESystem.NativeKernel`、`Source/XFESystem.NativeRuntime`、`Source/XFESystem.Platform` 和 `Tools/IndependentKernel` 为基线；可以保留外部编译器、Microsoft .NET GC/EH/BCL 与已标明来源的运行时库。旧 Cosmos 路径仅保留历史参考和显式回归入口，不自动回退为默认内核。迁移必须保留现有上层功能，并用合并后的源码与镜像重新验证，不能沿用旧 Cosmos 来宾结果宣称自研驱动或硬件能力通过。

2026-10-15 RTX 5090 移植检查点：新分支 `codex/rtx5090-native-cuda`，代码提交 `690d6d5`，0.3.0-dev.4。保留 C# 自研内核，原生接入只读 PCI/64 位 BAR、受校验 MCFG/ReBAR 和 UEFI GetBarAttributes 快照、独立 MMIO 所有权/UC/NX/撤销，以及精确 `10DE:2B85` 的只读 FSP 状态采样。发现该设备时会自动在普通编辑器打开 `X:\System\Diagnostics\Gpu.txt`；其 fallback 命令为 `cat /System/Diagnostics/Gpu.txt`。用户已答应在准备好后从 U 盘进行实体测试并提供报告内容或照片；尚未取得这份实体结果。

最终内核 SHA256 `bae23a9b7aae677d228ba46ebad9ed5daff276123493f535f1111ac1f8823e1e`；源镜像 SHA256 `e955131f726e98846c6bf663ec5736f9d7f296f22ee5604f3f8e537c40fe2e73`。同镜像 QEMU 8 vCPU（执行一个 BSP）完整 Bing/Monaco/Markdown/十条终端命令和诊断 RAM 文件读取通过，1335 项宿主检查通过，825 个编译/宿主源码与 Git 提交字节一致。另有受控宿主页表 118 项断言、固件交接 157 项断言及发现检查 107 项通过，均不是实体 GPU 执行证明。WindowsHostOnly 参考程序在本机 RTX 5090 实际执行 SM120 向量计算，65537 元素及双侧 guards 正确、资源清理通过；GuestCudaVerified/GpuRasterizationVerified 始终 false。GSP/RM、显存/GPU VA、通道/计算提交和 CUDA 接口仍未完成，不能说“只差实体测试”或“已支持 CUDA”。固定 NVIDIA 615.78.08 的 RM 工具只把一个实际 msgq.c 编为 freestanding 对象，完整 FSP/GSP 组件仍受依赖闭包/来源网络失败阻断，不能把它当完整 RM 链接。

专用 USB（序列号 `230222336207`，UniqueId `5000000000002AEF`，本次正常容量 `256060514304`）已更新到最终自动显示报告镜像。容量调整后完整 128 MiB 的预期/实际 SHA256 均为 `4ee3657482a139568c3977d0b72734c263195769618c6c59086f046d98d7a4e4`，真实尾部备份 GPT 校验通过；句柄关闭、HelperState completed、独立检查时辅助进程已不存在，U 盘保持离线。未回收到该进程的实际退出码，不把推断退出码声称为观测。完成记录：`build/usb-media/XFEstudio-RTX5090-AutoReport-0.3.0-dev.4-1792057539053/Completion.json`，SHA `c51c823fc0ee9e3d08856bbd851588987d59b0dbe8d84ab8ca2f0d894528d25b`；`Postflight.json` SHA `03ec5dace9d588461dacfaaacec75f1a07ebb8e5e45636a53c368a487c6e2991`。先前同轮手动读取报告版本的已验证写盘 `XFEstudio-RTX5090-Port-0.3.0-dev.4-1792056829405` 保留，并用 `SupersededByWrite.json` 关联最终自动打开报告的写盘。最终发布目录/ZIP `dist/XFESystem-Independent-RAM-0.3.0-dev.4-1792057514091-79011a11` 保留，前一个同版本包也保留。可提交摘要：`docs/validation-results/Rtx5090NativePort20261015.json`。所有操作都不是显卡/主板/U 盘固件刷写；没有重绑或复位宿主 GPU，没有重启宿主。

Modified README.md +3 -1

@@ -16,7 +16,9 @@ XFESystem 默认使用 **C# 自研裸机内核**：页分配与映射、中断

16 16

17 17

0.3.0-dev.3 使用同一最终镜像通过 QEMU 8 虚拟 CPU（实际一个 BSP）的原生 HTTPS Bing、JavaScript DOM 结果 12、Monaco 实际字形/颜色和第四行键盘编辑、Markdown 八项流程及经典终端九次命令验收。1314 项宿主检查通过；同一内核的单 CPU 随机源缺失测试正确拒绝 HTTPS 并保持桌面响应。地址栏后台刷新覆盖输入、裸机默认 LF 导致提示符缩进的问题均有实际复现与修复。ZIP 完整回读、重定位验证和 PowerShell 5.1 启动计划通过，实际 QEMU 交互窗口已打开。范围和交付状态以 [本轮验证摘要](docs/validation-results/IndependentNativeBing20261015.json) 为准；此前 0.3.0-dev.2 的记录保留于 [合并摘要](docs/validation-results/IndependentKernelMerge20261015.json)。

18 18

专用 U 盘现已更新为同一 0.3.0-dev.3 实验镜像：完整 128 MiB 回读、容量调整后的 SHA256 与真实盘尾备份 GPT 校验通过，句柄已关闭，辅助进程退出码 0，设备保持离线。首次管理员提权取消的记录保留，并已关联本次成功重试；详细身份、哈希和 Completion/Postflight 记录见本轮验证摘要。`DeliveryComplete=true` 表示本次实验镜像交付完成，实体启动与输入仍需实测，不代表完整发行版验收。

此前专用 U 盘的 0.3.0-dev.3 实验镜像已完成完整 128 MiB 回读与真实盘尾 GPT 校验；首次管理员提权取消及成功重试记录均保留。该轮 `DeliveryComplete=true` 表示当次实验镜像交付完成，不代表完整发行版验收。

RTX 5090 移植实验位于 `codex/rtx5090-native-cuda`，版本 0.3.0-dev.4。独立内核已接入只读 PCI/64 位 BAR、ACPI/MCFG/ReBAR、UEFI 资源交接、设备映射生命周期与 GB202 FSP 状态探测；发现 `10DE:2B85` 时自动打开 RAM 文件 `X:\System\Diagnostics\Gpu.txt` 供实体拍照。Windows 上的小型 SM120 CUDA 参考程序已实际验证 65537 个结果和保护区，但**XFESystem 内 CUDA、GPU 渲染和呈现仍未实现**。GSP/RM、计算通道与所需接口的实际移植继续按 [原生 CUDA 状态](docs/NativeCudaPort.md) 推进。1335 项宿主检查、同镜像 QEMU Bing/Monaco/Markdown/十条终端命令及 GPU 诊断读取通过；825 项编译/宿主源码与提交字节匹配。专用 U 盘已更新为最终自动显示报告的镜像，完整 128 MiB、调整后的 SHA256 和真实盘尾 GPT 全部校验通过，句柄关闭、辅助进程已退出、保持离线；旧发行目录、ZIP 和写盘记录均保留。完整哈希与尚未通过的硬件范围见 [RTX 5090 验证摘要](docs/validation-results/Rtx5090NativePort20261015.json)。

20 22

21 23

持久存储、实体网络、GPU/CUDA、USB HID、多屏、SMP、独立用户进程隔离及长期稳定性尚未完成。实体启动/输入兼容尚未验收。浏览器没有 Worker、完整语言服务器、WebAssembly、图片/背景图或完整 CSS/SVG/shaping 支持，Bing 的页面布局仍不完整。Jint 的反射/动态代码路径仍有 NativeAOT 裁剪和动态代码警告，当前验收不证明全部 Jint/.NET API 兼容。TLS 尚无吊销查询、CT、公共后缀表及外部 distrust 元数据。后台编辑、独立导航与字体快照已接入自研运行时；实际 240 Hz 与完整预览一秒目标仍未通过。

22 24

Modified THIRD_PARTY.md +1 -1

@@ -89,4 +89,4 @@ Cosmos 3.0.94（[上游](https://github.com/CosmosOS/Cosmos/tree/v3.0.94)，BSD-

89 89

90 90

组件存在或可以在宿主运行，不证明在 XFESystem 中完成初始化。真实来宾网络、TLS、存储、GPU/CUDA、独立程序和性能是否通过，以对应镜像的验收记录为准。

91 91

NVIDIA 原生命令协议组件参考 `open-gpu-kernel-modules` 615.78.08 的公开命令头文件及队列发布算法，固定提交为 `ca2d03e6fafd4d4b56b383497d931c25b283bfcf`。13 个来源文件的 SHA256、Git blob 和官方链接见 [NvidiaSources.lock.json](Tools/Gpu/NvidiaSources.lock.json)，完整 MIT 许可和相关版权见 [NvidiaPublicHeaders.txt](Tools/Gpu/Licenses/NvidiaPublicHeaders.txt)。新增 [RM 前置工具](Tools/Gpu/RmPort/Readme.md) 另锁定少量实际官方源码、逐文件许可及裸机组件编译结果；这些源码只在受限宿主构建缓存中，没有把完整 RM、GSP 或 CUDA 用户态组件链接入系统。独立内核现已接入只读 PCI/固件资源发现和 GB202 FSP 状态探测，仍没有客体 GPU/CUDA 执行通过记录。Windows [DriverOracle](Tools/Gpu/DriverOracle/Readme.md) 使用系统已有驱动完成小型 SM120 参考计算，始终标记 WindowsHostOnly，不能作为客体加速证据。UEFI PCI I/O ABI 参考 edk2-stable202608（BSD-2-Clause-Patent），资源交接和解析为本项目实现；没有捆绑新的 EDK II 驱动或 NVIDIA 固件二进制。完整范围见 [原生 CUDA 移植状态](docs/NativeCudaPort.md)。

NVIDIA 原生命令协议组件参考 `open-gpu-kernel-modules` 615.78.08 的公开命令头文件及队列发布算法，固定提交为 `ca2d03e6fafd4d4b56b383497d931c25b283bfcf`。13 个来源文件的 SHA256、Git blob 和官方链接见 [NvidiaSources.lock.json](Tools/Gpu/NvidiaSources.lock.json)，完整 MIT 许可和相关版权见 [NvidiaPublicHeaders.txt](Tools/Gpu/Licenses/NvidiaPublicHeaders.txt)。新增 [RM 前置工具](Tools/Gpu/RmPort/README.md) 另锁定少量实际官方源码、逐文件许可及裸机组件编译结果；这些源码只在受限宿主构建缓存中，没有把完整 RM、GSP 或 CUDA 用户态组件链接入系统。独立内核现已接入只读 PCI/固件资源发现和 GB202 FSP 状态探测，仍没有客体 GPU/CUDA 执行通过记录。Windows [DriverOracle](Tools/Gpu/DriverOracle/Readme.md) 使用系统已有驱动完成小型 SM120 参考计算，始终标记 WindowsHostOnly，不能作为客体加速证据。UEFI PCI I/O ABI 参考 edk2-stable202608（BSD-2-Clause-Patent），资源交接和解析为本项目实现；没有捆绑新的 EDK II 驱动或 NVIDIA 固件二进制。完整范围见 [原生 CUDA 移植状态](docs/NativeCudaPort.md)。

Modified docs/NativeCudaPort.md +2 -2

@@ -12,7 +12,7 @@

12 12

13 13

`NvidiaBootProbe` 只对匹配 RTX 5090 且固件范围、整个 BAR0 与 RAM/GOP 等范围检查均通过的设备，读取一页中的 `NV_THERM_I2CS_SCRATCH`（`0x00AD00BC`），然后撤销映射。上游 GB202 FSP 的完成值为完整 32 位 `0xFF`。探测只采样一次，不阻塞等待，不启动 GSP、不写寄存器、不创建通道。状态 5 仅表示观察到 FSP secure boot 完成，不能作为 CUDA 证据。

14 14

生产代码分别导出 `xfe_gpu_discovery_info`、`xfe_gpu_discovery_record` 和 `xfe_gpu_boot_probe_info`，只跨运行时传递整数/字节。应用将本次真实启动快照写入 RAM 文件 `X:\System\Diagnostics\Gpu.txt`。可从终端读取：

生产代码分别导出 `xfe_gpu_discovery_info`、`xfe_gpu_discovery_record` 和 `xfe_gpu_boot_probe_info`，只跨运行时传递整数/字节。应用将本次真实启动快照写入 RAM 文件 `X:\System\Diagnostics\Gpu.txt`。发现实际 `10DE:2B85` 时自动在普通文本编辑器中打开，无需等待 USB HID 或 web/Monaco 初始化；这个实体分支仍待硬件验收。也可从终端读取：

16 16

17 17 ```text

18 18 cat /System/Diagnostics/Gpu.txt

@@ -28,7 +28,7 @@ cat /System/Diagnostics/Gpu.txt

28 28

29 29 ## 官方 RM 的实际编译移植准备

30 30

仍固定 NVIDIA `open-gpu-kernel-modules` 615.78.08 / `ca2d03e6fafd4d4b56b383497d931c25b283bfcf`。`Tools/Gpu/RmPort` 对少量官方源码做 Git blob、SHA256、许可和 freestanding clang 编译检查，缓存有 64 MiB 上限。逐组件结果与真实 undefined 符号必须分别读取；一个队列组件编译成功不等于整个 RM 编译或链接成功。[工具说明](../Tools/Gpu/RmPort/Readme.md)

仍固定 NVIDIA `open-gpu-kernel-modules` 615.78.08 / `ca2d03e6fafd4d4b56b383497d931c25b283bfcf`。`Tools/Gpu/RmPort` 对少量官方源码做 Git blob、SHA256、许可和 freestanding clang 编译检查，缓存有 64 MiB 上限。逐组件结果与真实 undefined 符号必须分别读取；一个队列组件编译成功不等于整个 RM 编译或链接成功。[工具说明](../Tools/Gpu/RmPort/README.md)

32 32

33 33

官方将 OS 无关 RM 与 OS 接口层分开，但完整 RM 发布清单含 1099 个翻译单元。GSP/FSP/RM、GPU VA、DMA 生命周期、上下文/通道和故障回收尚未完成。公开 Copy Engine 编码器仍只消费调用方已经创建的合法通道，没有原生资源提供者，因此未被选作渲染后端。所有既有命令编码/私有内存测试仍只算前置检查。

34 34

Added docs/validation-results/Rtx5090NativePort20261015.json +316 -0

@@ -0,0 +1,316 @@

1 {

2 "Kind": "XfeRtx5090NativePortCheckpoint",

3 "Date": "2026-10-15",

4 "Version": "0.3.0-dev.4",

5 "Status": "AwaitingPhysicalRtx5090ResourceAndFspDiagnostics",

6 "ObjectiveComplete": false,

7 "GuestCudaVerified": false,

8 "GuestGpuExecutionVerified": false,

9 "GpuRasterizationVerified": false,

10 "GpuPresentationVerified": false,

11 "PhysicalHardwareAcceptancePassed": false,

12 "FullReleaseQualified": false,

13 "RamOnly": true,

14 "StorageEnabled": false,

15 "Branch": "codex/rtx5090-native-cuda",

16 "CodeCommit": "690d6d5bae2f3bb02de5d140a99375ce9e19a763",

17 "KernelSha256": "bae23a9b7aae677d228ba46ebad9ed5daff276123493f535f1111ac1f8823e1e",

18 "ImageSha256": "e955131f726e98846c6bf663ec5736f9d7f296f22ee5604f3f8e537c40fe2e73",

19 "NativeImplementation": [

20 "Versioned integer PCI/BAR discovery ABI in independent C# kernel",

21 "Read-only ACPI/MCFG/ECAM/ReBAR and UEFI PCI I/O resource handoff",

22 "Separate RAM/MMIO ownership, read-only UC/NX, MAXPHYADDR and precise rollback/unmap",

23 "Read-only validated GB202 FSP scratch sample, no GSP boot or command submission",

24 "RAM Gpu.txt with automatic plain-text editor opening on an actual discovered 10DE:2B85"

25 ],

26 "HostTests": {

27 "Passed": true,

28 "Suites": 15,

29 "PassedCount": 1335,

30 "FailedCount": 0,

31 "Receipt": {

32 "Path": "build/CSharp/HostTests-1792057227298/Result.json",

33 "Sha256": "be73380ff96f94e7442a0ff4854144cee5fd09b1dffd3fc5ca09f5d72ce2c2e6"

34 },

35 "HostOnly": true,

36 "GuestExecution": false

37 },

38 "SourceVerification": {

39 "Kind": "XfeCompiledAndHostStagedSourceVerification",

40 "Passed": true,

41 "Files": 825,

42 "Mismatches": [],

43 "KernelSha256": "bae23a9b7aae677d228ba46ebad9ed5daff276123493f535f1111ac1f8823e1e",

44 "ImageSha256": "e955131f726e98846c6bf663ec5736f9d7f296f22ee5604f3f8e537c40fe2e73",

45 "GuestCudaVerified": false,

46 "Receipt": {

47 "Path": "build/Gpu/StagedInputs.json",

48 "Sha256": "5c397f64deaafbed53913a5f690710784217cd6748543628c9004a46d68fa240"

49 }

50 },

51 "PrerequisiteChecks": {

52 "DeviceMapping": {

53 "Assertions": 118,

54 "Receipt": {

55 "Path": "build/Gpu/DeviceMappingChecks-1792055852034-9ac4d2eb/Result.json",

56 "Sha256": "8fc99842242b05d2c1a133d95ca5220004a0896725c75b0c33f0c979d384ee12"

57 }

58 },

59 "GpuFirmware": {

60 "Scenarios": 56,

61 "Assertions": 157,

62 "Receipt": {

63 "Path": "build/Gpu/GpuFirmwareChecks-1792056392556-07ae180f/Result.json",

64 "Sha256": "b75baae52030183e0949318b60dea529f8e8583de0519cfd3bb5208ed877b423"

65 }

66 },

67 "GpuDiscovery": {

68 "Checks": 107,

69 "Receipt": {

70 "Path": "build/IndependentKernel/GpuDiscoveryHost-1792055975816072100/Result.json",

71 "Sha256": "c3b1350ec3a9201b560061265a595727bdf7f8df21c931a871726363b5198950"

72 }

73 },

74 "Scope": "Host controlled memory and pure decoding, not hardware execution"

75 },

76 "Qemu": {

77 "Passed": true,

78 "VirtualCpus": 8,

79 "ExecutingCpus": 1,

80 "PhysicalNvidiaDevicePresent": false,

81 "GuestCudaVerified": false,

82 "Boot": {

83 "Path": "build/IndependentKernel/Proof-1792057253051311900/Result.json",

84 "Sha256": "d3646ef14f07b003ef68a14b55e1723b45e83ab50ae04f5190269b0a9ae62061"

85 },

86 "GpuEvidence": {

87 "Path": "build/IndependentKernel/Proof-1792057253051311900/GpuDiscoveryEvidence.json",

88 "Sha256": "0f866ccb1850e837a9f3cde3c6459d7faf5f1096489bc998d38c5b0744ee4107"

89 },

90 "BingPassed": true,

91 "DocumentPassed": true,

92 "MonacoPaintAndEditPassed": true,

93 "TerminalCommands": 10,

94 "GpuDiagnosticFileReadbackPassed": true,

95 "GpuDiagnosticScreenshot": {

96 "Path": "build/IndependentKernel/Proof-1792057253051311900/Terminal-GpuDiagnostics.png",

97 "Sha256": "d6f87a0aaaef921ed550b602f1e47f4546ae83da76a2d60229af5e2ca219f774"

98 },

99 "QemuExitCode": 0,

100 "AutomaticOpeningOnPhysicalTargetObserved": false

101 },

102 "WindowsCudaReference": {

103 "Scope": "WindowsHostOnly",

104 "ActualHostExecutionPassed": true,

105 "GuestCudaVerified": false,

106 "DeviceName": "NVIDIA GeForce RTX 5090",

107 "PciBusId": "0000:01:00.0",

108 "ComputeCapability": "12.0",

109 "DriverSupportedCudaVersion": "13.4",

110 "Kernel": "XfeVectorAdd: unsigned 32-bit addition modulo 2^32",

111 "CubinSm": 120,

112 "DeviceAllocationBytes": 786700,

113 "Verification": {

114 "AllElementsMatched": true,

115 "MatchedElements": 65537,

116 "PrefixGuardWords": 32,

117 "SuffixGuardWords": 32,

118 "PrefixGuardPassed": true,

119 "SuffixGuardPassed": true,

120 "ChangedFromPoison": 65537,

121 "Modulo32OverflowElements": 32793,

122 "ExpectedSha256": "6dd3ba2881195af38f51b446dbc93692d77728e2550acb63fe43f95b572bef42",

123 "ActualSha256": "6dd3ba2881195af38f51b446dbc93692d77728e2550acb63fe43f95b572bef42",

124 "OutputBodySha256": "a8d2297db14a4e1370a71959cfb95e7d09e3172377c09dd732336b749e2bf5f9",

125 "InputASha256": "27e2fdf2c12c9484e55449a34a0da427219b6d77c93923501d03fd464365812c",

126 "InputBSha256": "37eb0219fe16aa7e496a38ff322c1858538678187e7a2099fdcf9466b2d02974"

127 },

128 "CleanupPassed": true,

129 "Receipt": {

130 "Path": "build/Gpu/HostOracle-1792055339532-64a79890/Result.json",

131 "Sha256": "1aa6bc5c04aa13f7b3780aa0b48296c8068f66178dbf37f058f747edfd890c7f"

132 }

133 },

134 "OfficialRmPreflight": {

135 "PreflightPassed": true,

136 "CompleteRmCompiled": false,

137 "GuestCudaPassed": false,

138 "Version": "615.78.08",

139 "Commit": "ca2d03e6fafd4d4b56b383497d931c25b283bfcf",

140 "ReferenceFiles": 158,

141 "ReferenceBytes": 4035415,

142 "CaptureComplete": false,

143 "Receipt": {

144 "Path": "build/Gpu/RmPort/Proof-1792057160030421100/Result.json",

145 "Sha256": "7c6512e849ad177d271350c8a9a5c5f0a8bf0e9136bd80b4db80aeae381164f3"

146 },

147 "Components": [

148 {

149 "Source": "src/nvidia/src/libraries/msgq/msgq.c",

150 "Profile": "BareMetalNoOsSelection",

151 "Compiled": false,

152 "Linked": false,

153 "MissingIncludes": [],

154 "UndefinedSymbols": [],

155 "ObjectSha256": null

156 },

157 {

158 "Source": "src/nvidia/src/kernel/gpu/fsp/arch/blackwell/kern_fsp_gb202.c",

159 "Profile": "BareMetalNoOsSelection",

160 "Compiled": false,

161 "Linked": false,

162 "MissingIncludes": [

163 "ctrl/ctrlxxxx.h"

164 ],

165 "UndefinedSymbols": [],

166 "ObjectSha256": null

167 },

168 {

169 "Source": "src/nvidia/src/kernel/gpu/gsp/arch/blackwell/kernel_gsp_gb202.c",

170 "Profile": "BareMetalNoOsSelection",

171 "Compiled": false,

172 "Linked": false,

173 "MissingIncludes": [

174 "core/bin_data.h"

175 ],

176 "UndefinedSymbols": [],

177 "ObjectSha256": null

178 },

179 {

180 "Source": "src/nvidia/src/kernel/gpu/gsp/message_queue_cpu.c",

181 "Profile": "BareMetalNoOsSelection",

182 "Compiled": false,

183 "Linked": false,

184 "MissingIncludes": [

185 "containers/btree.h"

186 ],

187 "UndefinedSymbols": [],

188 "ObjectSha256": null

189 },

190 {

191 "Source": "src/nvidia/src/libraries/msgq/msgq.c",

192 "Profile": "UpstreamUnixInterfaceNeedsPort",

193 "Compiled": true,

194 "Linked": false,

195 "MissingIncludes": [],

196 "UndefinedSymbols": [

197 "osGpuReadReg032",

198 "osGpuWriteReg032",

199 "portMemCopy",

200 "portMemSet"

201 ],

202 "ObjectSha256": "5824f67b8a9279c083b70a76db71705b2d34343a815819e9cf50da54541057e2"

203 },

204 {

205 "Source": "src/nvidia/src/kernel/gpu/fsp/arch/blackwell/kern_fsp_gb202.c",

206 "Profile": "UpstreamUnixInterfaceNeedsPort",

207 "Compiled": false,

208 "Linked": false,

209 "MissingIncludes": [

210 "ctrl/ctrlxxxx.h"

211 ],

212 "UndefinedSymbols": [],

213 "ObjectSha256": null

214 },

215 {

216 "Source": "src/nvidia/src/kernel/gpu/gsp/arch/blackwell/kernel_gsp_gb202.c",

217 "Profile": "UpstreamUnixInterfaceNeedsPort",

218 "Compiled": false,

219 "Linked": false,

220 "MissingIncludes": [

221 "core/bin_data.h"

222 ],

223 "UndefinedSymbols": [],

224 "ObjectSha256": null

225 },

226 {

227 "Source": "src/nvidia/src/kernel/gpu/gsp/message_queue_cpu.c",

228 "Profile": "UpstreamUnixInterfaceNeedsPort",

229 "Compiled": false,

230 "Linked": false,

231 "MissingIncludes": [

232 "containers/btree.h"

233 ],

234 "UndefinedSymbols": [],

235 "ObjectSha256": null

236 }

237 ],

238

    "NetworkLimitations": "Raw TLS reset and API rate limit; bounded codeload download reached 180 seconds. Partial archive is not a complete source archive."

239 },

240 "Package": {

241 "Kind": "XfeIndependentKernelPackageReceipt",

242 "Passed": true,

243

    "PackageDirectory": "C:\\Users\\XFEstudio\\Desktop\\work\\CSharp\\GitHub\\XFEstudio\\XFESystem\\dist\\XFESystem-Independent-RAM-0.3.0-dev.4-1792057514091-79011a11",

244

    "ZipPath": "C:\\Users\\XFEstudio\\Desktop\\work\\CSharp\\GitHub\\XFEstudio\\XFESystem\\dist\\XFESystem-Independent-RAM-0.3.0-dev.4-1792057514091-79011a11.zip",

245 "ZipSha256": "68b5b347ec1ef95886914bad85405be0499cb068ddc256f0de832990a4874bf1",

246 "ManifestSha256": "963541cadc09b520975b31fa1967e18f7c2db71430ba3724ef08840e8539c969",

247 "ImageSha256": "e955131f726e98846c6bf663ec5736f9d7f296f22ee5604f3f8e537c40fe2e73",

248 "KernelSha256": "bae23a9b7aae677d228ba46ebad9ed5daff276123493f535f1111ac1f8823e1e",

249 "GuestAcceptancePassed": true,

250 "HostAcceptancePassed": true,

251 "RelocatedPackageVerificationPassed": true,

252 "ZipReadbackPassed": true,

253 "NoRenderFailures": true,

254 "MonacoRasterPaintCompleted": true,

255 "DocumentWorkflowPassed": true,

256 "MonacoKeyboardEditedModel": true,

257 "NetworkAcceptancePassed": true,

258 "BingNativeHttpsPassed": true,

259 "FullCssParityVerified": false,

260 "FullReleaseQualified": false,

261 "FinishedUtc": "2026-10-15T09:45:43.0242189Z",

262 "Receipt": {

263

      "Path": "C:\\Users\\XFEstudio\\Desktop\\work\\CSharp\\GitHub\\XFEstudio\\XFESystem\\dist\\XFESystem-Independent-RAM-0.3.0-dev.4-1792057514091-79011a11.json",

264 "Sha256": "0a91df0d08a85b5bd10dafe97069d623697c1e85a50c9bf0406d5fc2bf6bd0eb"

265 },

266 "WindowsPowerShell51PlanPassed": true,

267 "WindowsPowerShell51PlanLog": {

268 "Path": "build/Gpu/AutoReportWindowsPowerShell51Plan.log",

269 "Sha256": "cf9d6fdb8c835006372bd37cdc4a2dd46e27d9f127fd2de09f5344779352d004"

270 }

271 },

272 "Usb": {

273 "Verified": true,

274 "ImageProfile": "IndependentCSharp128",

275 "Serial": "230222336207",

276 "UniqueId": "5000000000002AEF",

277 "SizeBytes": 256060514304,

278 "SourceImageSha256": "e955131f726e98846c6bf663ec5736f9d7f296f22ee5604f3f8e537c40fe2e73",

279 "Verification": {

280 "ImageBytesVerified": 134217728,

281 "ExpectedAdjustedImageSha256": "4ee3657482a139568c3977d0b72734c263195769618c6c59086f046d98d7a4e4",

282 "ActualAdjustedImageSha256": "4ee3657482a139568c3977d0b72734c263195769618c6c59086f046d98d7a4e4",

283 "ActualLastLba": 500118191,

284 "ActualTailBackupGptVerified": true,

285 "FullImageVerified": true

286 },

287 "Completion": {

288

      "Path": "C:\\Users\\XFEstudio\\Desktop\\work\\CSharp\\GitHub\\XFEstudio\\XFESystem\\build\\usb-media\\XFEstudio-RTX5090-AutoReport-0.3.0-dev.4-1792057539053\\Completion.json",

289 "Sha256": "c51c823fc0ee9e3d08856bbd851588987d59b0dbe8d84ab8ca2f0d894528d25b"

290 },

291 "Postflight": {

292

      "Path": "C:\\Users\\XFEstudio\\Desktop\\work\\CSharp\\GitHub\\XFEstudio\\XFESystem\\build\\usb-media\\XFEstudio-RTX5090-AutoReport-0.3.0-dev.4-1792057539053\\Postflight.json",

293 "Sha256": "03ec5dace9d588461dacfaaacec75f1a07ebb8e5e45636a53c368a487c6e2991"

294 },

295 "HandlesClosed": true,

296 "FinalDiskState": "offline",

297 "HelperProcessPresent": false,

298 "ProcessExitCodeObserved": null

299 },

300 "Remaining": [

301 "Complete matching GSP firmware/RM port and secure boot/RPC handshake",

302 "Create GPU VA/DMA resources and real hardware context/channel",

303 "Verify Copy Engine execution separately from CUDA",

304 "Load SM120 cubin, execute CUDA-compatible program in XFESystem and verify guards/results",

305 "Integrate real GPU compositing/presentation, then measure full preview latency and display cadence"

306 ],

307 "PhysicalTest": {

308 "Ready": true,

309 "UserConfirmedAvailable": true,

310 "AutomaticReportFor": "PCI 10DE:2B85",

311 "ReportPath": "X:\\System\\Diagnostics\\Gpu.txt",

312 "FallbackCommand": "cat /System/Diagnostics/Gpu.txt",

313

    "RequestedEvidence": "Photo of automatically opened full GPU report; no host GPU reset/rebind or reboot performed by agent"

314 },

315 "FinishedUtc": "2026-10-15T09:50:54.272017+00:00"

316 }