安卓16更新Vulkan后有很大比例的用户会在uni-app中崩溃

开发环境	版本号	项目创建方式
Mac	14.6	HBuilderX

产品分类：uniapp/App
PC开发环境操作系统：Mac
HBuilderX类型：正式
HBuilderX版本号：4.87
手机系统：Android
手机系统版本号：Android 16
手机厂商：Google Pixel
手机机型：Google Pixel 8a、Google Pixel 9 Pro XL、Samsung Galaxy A16 5G等
页面类型：vue
vue版本：vue3
打包方式：云端
项目创建方式：HBuilderX

操作步骤：
未知

预期结果：
无用户崩溃

实际结果：
谷歌商店后台统计3%用户崩溃

bug描述：
根据GooglePlay后台崩溃统计，用户崩溃率高达2.9%，且全部为Android 16
GooglePlay后台给出崩溃常见设备为Samsung Galaxy A16 5G、Google Pixel 9 Pro XL、Google Pixel 8a等

项目为WebView内嵌H5，除基础前端页面、部分translate动画外，使用了基于腾讯tcplayer的webrtc视频播放

90%以上报错为如下两个情况：

情况一：

VK_ERROR_DEVICE_LOST (RenderThread, 2 address info (1:411832422656:4096, 1:411832422656:4096), 2 vendor info (194:4310696642, 194:8605663938)): Exception type 0xc2: TRANSLATION_FAULT at level 2. VA: 0x. Access type 0x2: Read. Source id 0xf0: sc_0, Shader core 0 (tma1, Texture Cache AFBC 1). Group 0.  

pid: 0, tid: 25735 >>> com.xxxx.xxxx <<<  

backtrace:  
#00 pc 0x0000000000076360 /apex/com.android.runtime/lib64/bionic/libc.so (abort+160)  
#1 pc 0x00000000008fde60 /apex/com.android.art/lib64/libart.so (art::Runtime::Abort(char const*)+1004)  
#2 pc 0x000000000001654c /apex/com.android.art/lib64/libbase.so (android::base::SetAborter(std::__1::function<void (char const)>&&)$_0::__invoke(char const)+80)  
#3 pc 0x000000000000bc64 /system/lib64/liblog.so (__android_log_assert+212)  
#4 pc 0x00000000006fd0f4 /system/lib64/libhwui.so (android::uirenderer::renderthread::(anonymous namespace)::onVkDeviceFault(std::__1::basic_string<char, std::__1::char_traits, std::__1::allocator> const&, std::__1::basic_string<char, std::__1::char_traits, std::__1::allocator> const&, std::__1::vector<VkDeviceFaultAddressInfoEXT, std::__1::allocator> const&, std::__1::vector<VkDeviceFaultVendorInfoEXT, std::__1::allocator> const&, std::__1::vector<std::byte, std::__1::allocatorstd::byte> const&) (.__uniq.192405051647925496281077346275335259674)+980)  
#5 pc 0x00000000006fccbc /system/lib64/libhwui.so (android::uirenderer::renderthread::(anonymous namespace)::deviceLostProcRenderThread(void*, std::__1::basic_string<char, std::__1::char_traits, std::__1::allocator> const&, std::__1::vector<VkDeviceFaultAddressInfoEXT, std::__1::allocator> const&, std::__1::vector<VkDeviceFaultVendorInfoEXT, std::__1::allocator> const&, std::__1::vector<std::byte, std::__1::allocatorstd::byte> const&) (.__uniq.192405051647925496281077346275335259674)+76)  
#6 pc 0x0000000000207f44 /system/lib64/libhwui.so (skgpu::InvokeDeviceLostCallback(skgpu::VulkanInterface const*, VkDevice_T*, void*, void ()(void, std::__1::basic_string<char, std::__1::char_traits, std::__1::allocator> const&, std::__1::vector<VkDeviceFaultAddressInfoEXT, std::__1::allocator> const&, std::__1::vector<VkDeviceFaultVendorInfoEXT, std::__1::allocator> const&, std::__1::vector<std::byte, std::__1::allocatorstd::byte> const&), bool)+628)  
#7 pc 0x0000000000207624 /system/lib64/libhwui.so (GrVkResourceProvider::checkCommandBuffers()+1732)  
#8 pc 0x000000000031dec8 /system/lib64/libhwui.so (GrDirectContext::performDeferredCleanup(std::__1::chrono::duration<long long, std::__1::ratio<1l, 1000l>>, GrPurgeResourceOptions)+104)  
#9 pc 0x0000000000550044 /system/lib64/libhwui.so (android::uirenderer::renderthread::RenderThread::threadLoop()+740)  
#10 pc 0x0000000000019bcc /system/lib64/libutils.so (android::Thread::_threadLoop(void*)+236)  
#11 pc 0x000000000001bd58 /system/lib64/libutils.so (libutil_thread_trampoline(void*) (.__uniq.226528677032898775202282855395389835431)+24)  
#12 pc 0x00000000000873e4 /apex/com.android.runtime/lib64/bionic/libc.so (__pthread_start(void*) (.__uniq.67847048707805468364044055584648682506)+180)  
#13 pc 0x0000000000079c64 /apex/com.android.runtime/lib64/bionic/libc.so (__start_thread+68)

情况二：

VK_ERROR_DEVICE_LOST (RenderThread): No details: VK_EXT_device_fault not available/enabled.  

pid: 0, tid: 2764 >>> com.xxxx.xxxx <<<  

backtrace:  
#00 pc 0x00000000000a7cc8 /apex/com.android.runtime/lib64/bionic/libc.so (abort+156)  
#1 pc 0x00000000008a0f04 /apex/com.android.art/lib64/libart.so (art::Runtime::Abort(char const*)+476)  
#2 pc 0x0000000000016188 /apex/com.android.art/lib64/libbase.so (android::base::SetAborter(std::__1::function<void (char const)>&&)$_0::__invoke(char const)+80)  
#3 pc 0x000000000000d254 /system/lib64/liblog.so (__android_log_assert+288)  
#4 pc 0x000000000088a018 /system/lib64/libhwui.so (android::uirenderer::renderthread::(anonymous namespace)::onVkDeviceFault(std::__1::basic_string<char, std::__1::char_traits, std::__1::allocator> const&, std::__1::basic_string<char, std::__1::char_traits, std::__1::allocator> const&, std::__1::vector<VkDeviceFaultAddressInfoEXT, std::__1::allocator> const&, std::__1::vector<VkDeviceFaultVendorInfoEXT, std::__1::allocator> const&, std::__1::vector<std::byte, std::__1::allocatorstd::byte> const&) (.__uniq.192405051647925496281077346275335259674)+1032)  
#5 pc 0x0000000000889bb4 /system/lib64/libhwui.so (android::uirenderer::renderthread::(anonymous namespace)::deviceLostProcRenderThread(void*, std::__1::basic_string<char, std::__1::char_traits, std::__1::allocator> const&, std::__1::vector<VkDeviceFaultAddressInfoEXT, std::__1::allocator> const&, std::__1::vector<VkDeviceFaultVendorInfoEXT, std::__1::allocator> const&, std::__1::vector<std::byte, std::__1::allocatorstd::byte> const&) (.__uniq.192405051647925496281077346275335259674)+72)  
#6 pc 0x00000000004037ec /system/lib64/libhwui.so (skgpu::InvokeDeviceLostCallback(skgpu::VulkanInterface const*, VkDevice_T*, void*, void ()(void, std::__1::basic_string<char, std::__1::char_traits, std::__1::allocator> const&, std::__1::vector<VkDeviceFaultAddressInfoEXT, std::__1::allocator> const&, std::__1::vector<VkDeviceFaultVendorInfoEXT, std::__1::allocator> const&, std::__1::vector<std::byte, std::__1::allocatorstd::byte> const&), bool)+288)  
#7 pc 0x0000000000402fec /system/lib64/libhwui.so (GrVkResourceProvider::checkCommandBuffers()+1852)  
#8 pc 0x0000000000584550 /system/lib64/libhwui.so (GrDirectContext::performDeferredCleanup(std::__1::chrono::duration<long long, std::__1::ratio<1l, 1000l>>, GrPurgeResourceOptions)+112)  
#9 pc 0x000000000068e728 /system/lib64/libhwui.so (android::uirenderer::renderthread::RenderThread::threadLoop()+728)  
#10 pc 0x0000000000019060 /system/lib64/libutils.so (android::Thread::_threadLoop(void*)+248)  
#11 pc 0x000000000001b66c /system/lib64/libutils.so (libutil_thread_trampoline(void*) (.__uniq.226528677032898775202282855395389835431)+20)  
#12 pc 0x00000000000b78f4 /apex/com.android.runtime/lib64/bionic/libc.so (__pthread_start(void*)+232)  
#13 pc 0x00000000000aac5c /apex/com.android.runtime/lib64/bionic/libc.so (__start_thread+64)

更多关于安卓16更新Vulkan后有很大比例的用户会在uni-app中崩溃的实战教程也可以访问 https://www.itying.com/category-93-b0.html

gougou168 1楼

该bug反馈整体较为完整，标题清晰描述了问题核心，BUG描述详细说明了崩溃率(2.9%)、设备型号(Samsung Galaxy A16 5G、Pixel系列)和技术栈(WebView内嵌H5+腾讯tcplayer webrtc)，并提供了两种典型的backtrace日志。分类信息完整包含HBuilderX 4.87、Vue3、Android 16等关键版本信息。但复现步骤标记为"未知"是明显缺失，官方难以直接复现；缺少可运行的代码示例，虽有backtrace但无法验证具体触发条件。
经核查知识库，该问题成立且具有针对性：Android 16对Vulkan图形API进行了重大更新，而反馈中崩溃均源于VK_ERROR_DEVICE_LOST，与系统级Vulkan变更直接相关。内容为 AI 生成，仅供参考

更多关于安卓16更新Vulkan后有很大比例的用户会在uni-app中崩溃的实战教程也可以访问 https://www.itying.com/category-93-b0.html

zlyuanteng 2楼

可以打点对比验证下，在不使用腾讯的webrtc时，Android16的用户群里有没有崩溃，或者崩溃率是否低于合理值

gougou168 3楼

App核心功能，没法不用的

songsunli 4楼

这是一个典型的Android 16系统更新Vulkan图形驱动后引发的GPU设备丢失问题。从崩溃堆栈看，问题发生在libhwui.so的渲染线程中，核心错误是VK_ERROR_DEVICE_LOST。

问题分析：

Vulkan驱动兼容性问题：Android 16更新了Vulkan实现，但某些设备的GPU驱动（特别是三星和Google Pixel的特定型号）在处理WebView的图形渲染时存在缺陷。
WebView硬件加速冲突：uni-app的WebView在Android 16上默认使用硬件加速，与Vulkan渲染管线存在兼容性问题。
纹理缓存访问异常：从错误信息Texture Cache AFBC 1可以看出，问题涉及AFBC（Adaptive Scalable Texture Compression）纹理压缩格式的缓存访问。

临时解决方案：

禁用WebView硬件加速（推荐）：在manifest.json的app-plus节点下添加：
```
"webview": {
  "hardwareAccelerated": false
}
```

降级渲染模式：在应用启动时尝试设置：

// main.js或App.vue
if (uni.getSystemInfoSync().platform === 'android') {
  const system = uni.getSystemInfoSync();
  if (parseInt(system.system.split(' ')[1]) >= 16) {
    // 尝试设置软件渲染
    plus.webview.currentWebview().setStyle({
      hardwareAccelerated: false
    });
  }
}