Bug 1870021 - Add some GPU process telemetry probes. r=jrmuizel

Add probes for total number of launch attempts and unstable number of
launch attempts. Port expired CRASH_FALLBACKS probe to glean and renew
it, telling us the number of times we handled each crash fallback with
each method. And finally, add a probe to tell us the status of the GPU
process gfxFeature - this is already available on desktop via the
telemetry environment, but adding a glean probe allows us to gather
the data on Android too.

Differential Revision: https://phabricator.services.mozilla.com/D196866

gfx/ipc/GPUProcessManager.cpp

@@ -24,6 +24,7 @@
 #include "mozilla/dom/ContentParent.h"
 #include "mozilla/gfx/gfxVars.h"
 #include "mozilla/gfx/GPUChild.h"
+#include "mozilla/glean/GleanMetrics.h"
 #include "mozilla/ipc/Endpoint.h"
 #include "mozilla/ipc/ProcessChild.h"
 #include "mozilla/layers/APZCTreeManagerChild.h"
@@ -70,12 +71,6 @@ namespace gfx {
 
 using namespace mozilla::layers;
 
-enum class FallbackType : uint32_t {
-  NONE = 0,
-  DECODINGDISABLED,
-  DISABLED,
-};
-
 static StaticAutoPtr<GPUProcessManager> sSingleton;
 
 GPUProcessManager* GPUProcessManager::Get() { return sSingleton; }
@@ -223,8 +218,11 @@ bool GPUProcessManager::LaunchGPUProcess() {
   auto newTime = TimeStamp::Now();
   if (!IsProcessStable(newTime)) {
     mUnstableProcessAttempts++;
+    mozilla::glean::gpu_process::unstable_launch_attempts.Set(
+        mUnstableProcessAttempts);
   }
   mTotalProcessAttempts++;
+  mozilla::glean::gpu_process::total_launch_attempts.Set(mTotalProcessAttempts);
   mProcessAttemptLastTime = newTime;
   mProcessStable = false;
 
@@ -296,8 +294,11 @@ bool GPUProcessManager::MaybeDisableGPUProcess(const char* aMessage,
 
   gfxPlatform::DisableGPUProcess();
 
-  Telemetry::Accumulate(Telemetry::GPU_PROCESS_CRASH_FALLBACKS,
-                        uint32_t(FallbackType::DISABLED));
+  mozilla::glean::gpu_process::feature_status.Set(
+      gfxConfig::GetFeature(Feature::GPU_PROCESS)
+          .GetStatusAndFailureIdString());
+
+  mozilla::glean::gpu_process::crash_fallbacks.Get("disabled"_ns).Add(1);
 
   DestroyProcess();
   ShutdownVsyncIOThread();
@@ -808,12 +809,11 @@ void GPUProcessManager::OnProcessUnexpectedShutdown(GPUProcessHost* aHost) {
                               layers_gpu_process_max_restarts_with_decoder()) &&
              mDecodeVideoOnGpuProcess) {
     mDecodeVideoOnGpuProcess = false;
-    Telemetry::Accumulate(Telemetry::GPU_PROCESS_CRASH_FALLBACKS,
-                          uint32_t(FallbackType::DECODINGDISABLED));
+    mozilla::glean::gpu_process::crash_fallbacks.Get("decoding_disabled"_ns)
+        .Add(1);
     HandleProcessLost();
   } else {
-    Telemetry::Accumulate(Telemetry::GPU_PROCESS_CRASH_FALLBACKS,
-                          uint32_t(FallbackType::NONE));
+    mozilla::glean::gpu_process::crash_fallbacks.Get("none"_ns).Add(1);
     HandleProcessLost();
   }
 }

gfx/metrics.yaml

@@ -27,6 +27,70 @@ paint:
     expires: never
     telemetry_mirror: PAINT_BUILD_DISPLAYLIST_TIME
 
+gpu_process:
+  feature_status:
+    type: string
+    description: >
+      Current status of the GPU process feature
+    bugs:
+      - https://bugzilla.mozilla.org/show_bug.cgi?id=1870021
+    data_reviews:
+      - https://bugzilla.mozilla.org/show_bug.cgi?id=1870021
+    data_sensitivity:
+      - technical
+    notification_emails:
+      - gfx-telemetry-alerts@mozilla.com
+      - jnicol@mozilla.com
+    expires: never
+  crash_fallbacks:
+    type: labeled_counter
+    description: >
+      How often we use different fallbacks when the GPU process crashes
+    labels:
+      - none
+      - decoding_disabled
+      - disabled
+    bugs:
+      - https://bugzilla.mozilla.org/show_bug.cgi?id=1870021
+    data_reviews:
+      - https://bugzilla.mozilla.org/show_bug.cgi?id=1870021
+    data_sensitivity:
+      - technical
+    notification_emails:
+      - gfx-telemetry-alerts@mozilla.com
+      - jnicol@mozilla.com
+    expires: never
+  total_launch_attempts:
+    type: quantity
+    description: >
+      The number of total GPU process launch attempts.
+    unit: launches
+    bugs:
+      - https://bugzilla.mozilla.org/show_bug.cgi?id=1870021
+    data_reviews:
+      - https://bugzilla.mozilla.org/show_bug.cgi?id=1870021
+    data_sensitivity:
+      - technical
+    notification_emails:
+      - gfx-telemetry-alerts@mozilla.com
+      - jnicol@mozilla.com
+    expires: never
+  unstable_launch_attempts:
+    type: quantity
+    description: >
+      The number of consecutive unstable launch attempts.
+    unit: launches
+    bugs:
+      - https://bugzilla.mozilla.org/show_bug.cgi?id=1870021
+    data_reviews:
+      - https://bugzilla.mozilla.org/show_bug.cgi?id=1870021
+    data_sensitivity:
+      - technical
+    notification_emails:
+      - gfx-telemetry-alerts@mozilla.com
+      - jnicol@mozilla.com
+    expires: never
+
 wr:
   gpu_wait_time:
     type: timing_distribution

gfx/thebes/gfxPlatform.cpp

@@ -935,6 +935,10 @@ void gfxPlatform::Init() {
   }
 #endif
 
+  mozilla::glean::gpu_process::feature_status.Set(
+      gfxConfig::GetFeature(Feature::GPU_PROCESS)
+          .GetStatusAndFailureIdString());
+
   if (gfxConfig::IsEnabled(Feature::GPU_PROCESS)) {
     GPUProcessManager* gpu = GPUProcessManager::Get();
     Unused << gpu->LaunchGPUProcess();

toolkit/components/telemetry/Histograms.json

@@ -1434,20 +1434,6 @@
     "n_buckets": 100,
     "description": "GPU process initialization (excluding XPCOM and fork time) time in milliseconds"
   },
-  "GPU_PROCESS_CRASH_FALLBACKS": {
-    "record_in_processes": ["main", "content"],
-    "products": ["firefox", "fennec"],
-    "alert_emails": [
-      "gfx-telemetry-alerts@mozilla.com",
-      "gsquelart@mozilla.com"
-    ],
-    "expires_in_version": "70",
-    "bug_numbers": [1338011],
-    "kind": "enumerated",
-    "releaseChannelCollection": "opt-out",
-    "n_values": 10,
-    "description": "Count how often we use different fallbacks when the GPU process crashes: None=0, GPUProcessDecodingDisabled=1, GPUProcessDisabled=2"
-  },
   "MEMORY_RESIDENT_FAST": {
     "record_in_processes": ["main", "content"],
     "products": ["firefox", "fennec"],