video_core/codecs: more fixes for VAAPI detection ...

* skip impersonated VAAPI implementaions ("imposter detection")
* place VAAPI priority below CUDA/NVDEC/CUVID

src/video_core/command_classes/codecs/codec.cpp

@@ -17,6 +17,10 @@
 
 extern "C" {
 #include <libavutil/opt.h>
+#ifdef LIBVA_FOUND
+// for querying VAAPI driver information
+#include <libavutil/hwcontext_vaapi.h>
+#endif
 }
 
 namespace Tegra {
@@ -29,6 +33,7 @@ constexpr std::array PREFERRED_GPU_DECODERS = {
     AV_HWDEVICE_TYPE_D3D11VA,
     AV_HWDEVICE_TYPE_DXVA2,
 #elif defined(__linux__)
+    AV_HWDEVICE_TYPE_VAAPI,
     AV_HWDEVICE_TYPE_VDPAU,
 #endif
     // last resort for Linux Flatpak (w/ NVIDIA)
@@ -78,79 +83,18 @@ static std::vector<AVHWDeviceType> ListSupportedContexts() {
     AVHWDeviceType current_device_type = AV_HWDEVICE_TYPE_NONE;
     do {
         current_device_type = av_hwdevice_iterate_types(current_device_type);
-        // filter out VA-API since we will try that first if supported
+        contexts.push_back(current_device_type);
-        if (current_device_type != AV_HWDEVICE_TYPE_VAAPI) {
-            contexts.push_back(current_device_type);
-        }
     } while (current_device_type != AV_HWDEVICE_TYPE_NONE);
     return contexts;
 }
 
-#ifdef LIBVA_FOUND
-// List all the currently loaded Linux modules
-static std::vector<std::string> ListLinuxKernelModules() {
-    using FILEPtr = std::unique_ptr<FILE, decltype(&std::fclose)>;
-    auto module_listing = FILEPtr{fopen("/proc/modules", "rt"), std::fclose};
-    std::vector<std::string> modules{};
-    if (!module_listing) {
-        LOG_WARNING(Service_NVDRV, "Could not open /proc/modules to collect available modules");
-        return modules;
-    }
-    char* buffer = nullptr;
-    size_t buf_len = 0;
-    while (getline(&buffer, &buf_len, module_listing.get()) != -1) {
-        // format for the module listing file (sysfs)
-        // <name> <module_size> <depended_by_count> <depended_by_names> <status> <load_address>
-        auto line = std::string(buffer);
-        // we are only interested in module names
-        auto name_pos = line.find_first_of(" ");
-        if (name_pos == std::string::npos) {
-            continue;
-        }
-        modules.push_back(line.erase(name_pos));
-    }
-    free(buffer);
-    return modules;
-}
-#endif
-
 bool Codec::CreateGpuAvDevice() {
-#if defined(LIBVA_FOUND)
-    static constexpr std::array<const char*, 3> VAAPI_DRIVERS = {
-        "i915",
-        "iHD",
-        "amdgpu",
-    };
-    AVDictionary* hwdevice_options = nullptr;
-    const auto loaded_modules = ListLinuxKernelModules();
-    av_dict_set(&hwdevice_options, "connection_type", "drm", 0);
-    for (const auto& driver : VAAPI_DRIVERS) {
-        // first check if the target driver is loaded in the kernel
-        bool found = std::any_of(loaded_modules.begin(), loaded_modules.end(),
-                                 [&driver](const auto& module) { return module == driver; });
-        if (!found) {
-            LOG_DEBUG(Service_NVDRV, "Kernel driver {} is not loaded, trying the next one", driver);
-            continue;
-        }
-        av_dict_set(&hwdevice_options, "kernel_driver", driver, 0);
-        const int hwdevice_error = av_hwdevice_ctx_create(&av_gpu_decoder, AV_HWDEVICE_TYPE_VAAPI,
-                                                          nullptr, hwdevice_options, 0);
-        if (hwdevice_error >= 0) {
-            LOG_INFO(Service_NVDRV, "Using VA-API with {}", driver);
-            av_dict_free(&hwdevice_options);
-            av_codec_ctx->pix_fmt = AV_PIX_FMT_VAAPI;
-            return true;
-        }
-        LOG_DEBUG(Service_NVDRV, "VA-API av_hwdevice_ctx_create failed {}", hwdevice_error);
-    }
-    LOG_DEBUG(Service_NVDRV, "VA-API av_hwdevice_ctx_create failed for all drivers");
-    av_dict_free(&hwdevice_options);
-#endif
     static constexpr auto HW_CONFIG_METHOD = AV_CODEC_HW_CONFIG_METHOD_HW_DEVICE_CTX;
     static const auto supported_contexts = ListSupportedContexts();
     for (const auto& type : PREFERRED_GPU_DECODERS) {
         if (std::none_of(supported_contexts.begin(), supported_contexts.end(),
                          [&type](const auto& context) { return context == type; })) {
+            LOG_DEBUG(Service_NVDRV, "{} explicitly unsupported", av_hwdevice_get_type_name(type));
             continue;
         }
         const int hwdevice_res = av_hwdevice_ctx_create(&av_gpu_decoder, type, nullptr, nullptr, 0);
@@ -159,6 +103,24 @@ bool Codec::CreateGpuAvDevice() {
                       av_hwdevice_get_type_name(type), hwdevice_res);
             continue;
         }
+#ifdef LIBVA_FOUND
+        if (type == AV_HWDEVICE_TYPE_VAAPI) {
+            // we need to determine if this is an impersonated VAAPI driver
+            AVHWDeviceContext* hwctx =
+                static_cast<AVHWDeviceContext*>(static_cast<void*>(av_gpu_decoder->data));
+            AVVAAPIDeviceContext* vactx = static_cast<AVVAAPIDeviceContext*>(hwctx->hwctx);
+            const char* vendor_name = vaQueryVendorString(vactx->display);
+            if (strstr(vendor_name, "VDPAU backend")) {
+                // VDPAU impersonated VAAPI impl's are super buggy, we need to skip them
+                LOG_DEBUG(Service_NVDRV, "Skipping vdapu impersonated VAAPI driver");
+                continue;
+            } else {
+                // according to some user testing, certain vaapi driver (Intel?) could be buggy
+                // so let's log the driver name which may help the developers/supporters
+                LOG_DEBUG(Service_NVDRV, "Using VAAPI driver: {}", vendor_name);
+            }
+        }
+#endif
         for (int i = 0;; i++) {
             const AVCodecHWConfig* config = avcodec_get_hw_config(av_codec, i);
             if (!config) {