Merge pull request dotnet#92325 from dotnet-maestro-bot/merge/release…

…/8.0-rc2-to-release/8.0

[automated] Merge branch 'release/8.0-rc2' => 'release/8.0'

eng/Version.Details.xml

@@ -90,9 +90,9 @@
       <Sha>89be445dd4936157533ad96bafb95f701430653a</Sha>
       <SourceBuild RepoName="cecil" ManagedOnly="true" />
     </Dependency>
-    <Dependency Name="Microsoft.NET.Workload.Emscripten.Current.Manifest-8.0.100.Transport" Version="8.0.0-rtm.23469.3">
+    <Dependency Name="Microsoft.NET.Workload.Emscripten.Current.Manifest-8.0.100.Transport" Version="8.0.0-rtm.23470.1">
       <Uri>https://github.com/dotnet/emsdk</Uri>
-      <Sha>50bf805c8b5ca52abd34fde390609d8a54640246</Sha>
+      <Sha>446eeb331fcbf2f48c14a377601a8ab950ec942e</Sha>
       <SourceBuild RepoName="emsdk" ManagedOnly="true" />
     </Dependency>
     <Dependency Name="Microsoft.SourceBuild.Intermediate.source-build-reference-packages" Version="8.0.0-alpha.1.23469.1">
@@ -398,5 +398,9 @@
       <Uri>https://github.com/NuGet/NuGet.Client</Uri>
       <Sha>8fef55f5a55a3b4f2c96cd1a9b5ddc51d4b927f8</Sha>
     </Dependency>
+    <Dependency Name="Microsoft.Dotnet.Sdk.Internal" Version="8.0.100-rc.2.23470.7">
+      <Uri>https://github.com/dotnet/installer</Uri>
+      <Sha>dbeae1ac71d95355452952059f35960991cb3fd2</Sha>
+    </Dependency>
   </ToolsetDependencies>
 </Dependencies>

eng/Versions.props

@@ -205,8 +205,6 @@
     <GrpcCoreVersion>2.46.3</GrpcCoreVersion>
     <GrpcDotnetClientVersion>2.45.0</GrpcDotnetClientVersion>
     <GrpcToolsVersion>2.45.0</GrpcToolsVersion>
-    <!-- Uncomment to set a fixed version, else the latest is used -->
-    <SdkVersionForWorkloadTesting>8.0.100-rc.1.23415.5</SdkVersionForWorkloadTesting>
     <CompilerPlatformTestingVersion>1.1.2-beta1.23323.1</CompilerPlatformTestingVersion>
     <!-- Docs -->
     <MicrosoftPrivateIntellisenseVersion>8.0.0-preview-20230918.1</MicrosoftPrivateIntellisenseVersion>
@@ -240,7 +238,7 @@
          Note: when the name is updated, make sure to update dependency name in eng/pipelines/common/xplat-setup.yml
                like - DarcDependenciesChanged.Microsoft_NET_Workload_Emscripten_Current_Manifest-8_0_100_Transport
     -->
-    <MicrosoftNETWorkloadEmscriptenCurrentManifest80100TransportVersion>8.0.0-rtm.23469.3</MicrosoftNETWorkloadEmscriptenCurrentManifest80100TransportVersion>
+    <MicrosoftNETWorkloadEmscriptenCurrentManifest80100TransportVersion>8.0.0-rtm.23470.1</MicrosoftNETWorkloadEmscriptenCurrentManifest80100TransportVersion>
     <MicrosoftNETRuntimeEmscriptenVersion>$(MicrosoftNETWorkloadEmscriptenCurrentManifest80100TransportVersion)</MicrosoftNETRuntimeEmscriptenVersion>
     <!-- workloads -->
     <SwixPackageVersion>1.1.87-gba258badda</SwixPackageVersion>
@@ -257,5 +255,8 @@
     <!-- BrowserDebugProxy libs -->
     <MicrosoftExtensionsLoggingVersion>3.1.7</MicrosoftExtensionsLoggingVersion>
     <MicrosoftSymbolStoreVersion>1.0.406601</MicrosoftSymbolStoreVersion>
+    <!-- installer version, for testing workloads -->
+    <MicrosoftDotnetSdkInternalVersion>8.0.100-rc.2.23470.7</MicrosoftDotnetSdkInternalVersion>
+    <SdkVersionForWorkloadTesting>$(MicrosoftDotnetSdkInternalVersion)</SdkVersionForWorkloadTesting>
   </PropertyGroup>
 </Project>

src/coreclr/gc/gcconfig.h

@@ -83,6 +83,7 @@ class GCConfigStringHolder
     INT_CONFIG   (BGCSpinCount,              "BGCSpinCount",              NULL,                                140,                "Specifies the bgc spin count")                                                           \
     INT_CONFIG   (BGCSpin,                   "BGCSpin",                   NULL,                                2,                  "Specifies the bgc spin time")                                                            \
     INT_CONFIG   (HeapCount,                 "GCHeapCount",               "System.GC.HeapCount",               0,                  "Specifies the number of server GC heaps")                                                 \
+    INT_CONFIG   (MaxHeapCount,              "GCMaxHeapCount",            "System.GC.MaxHeapCount",            0,                  "Specifies the max number of server GC heaps to adjust to")                                                 \
     INT_CONFIG   (Gen0Size,                  "GCgen0size",                NULL,                                0,                  "Specifies the smallest gen0 budget")                                                     \
     INT_CONFIG   (SegmentSize,               "GCSegmentSize",             NULL,                                0,                  "Specifies the managed heap segment size")                                                \
     INT_CONFIG   (LatencyMode,               "GCLatencyMode",             NULL,                                -1,                 "Specifies the GC latency mode - batch, interactive or low latency (note that the same "   \

src/coreclr/gc/gcpriv.h

@@ -402,8 +402,6 @@ struct GCDebugSpinLock {
 #if defined(DYNAMIC_HEAP_COUNT)
     // time in microseconds we wait for the more space lock
     uint64_t msl_wait_time;
-    // number of times we wait for the more space lock
-    uint64_t msl_wait_count;
 #endif //DYNAMIC_HEAP_COUNT
 
     GCDebugSpinLock()
@@ -415,7 +413,7 @@ struct GCDebugSpinLock {
         , num_switch_thread(0), num_wait_longer(0), num_switch_thread_w(0), num_disable_preemptive_w(0)
 #endif
 #if defined(DYNAMIC_HEAP_COUNT)
-        , msl_wait_time(0), msl_wait_count(0)
+        , msl_wait_time(0)
 #endif //DYNAMIC_HEAP_COUNT
     {
     }
@@ -1148,15 +1146,12 @@ class dynamic_data
     //
     // The following 3 fields are updated at the beginning of each GC, if that GC condemns this generation.
     //
-    // The number of GC that condemned this generation. The only difference between this
-    // and collection_count is just that collection_count is maintained for all physical generations
-    // (currently there are 5) whereas this is only updated for logical generations (there are 3).
-    size_t    gc_clock;
-    uint64_t  time_clock;       //time when this gc started
+    size_t    gc_clock; // the gc index
+    uint64_t  time_clock; // time when this gc started
     uint64_t  previous_time_clock; // time when previous gc started
 
     // Updated at the end of a GC, if that GC condemns this generation.
-    size_t    gc_elapsed_time;  // Time it took for the gc to complete
+    size_t    gc_elapsed_time;  // time it took for the gc to complete
 
     //
     // The following fields (and fields in sdata) are initialized during GC init time and do not change.
@@ -1495,6 +1490,8 @@ class mark_queue_t
     void verify_empty();
 };
 
+float median_of_3 (float a, float b, float c);
+
 //class definition of the internal class
 class gc_heap
 {
@@ -2422,6 +2419,7 @@ class gc_heap
 #ifndef USE_REGIONS
     PER_HEAP_METHOD void rearrange_heap_segments(BOOL compacting);
 #endif //!USE_REGIONS
+    PER_HEAP_METHOD void delay_free_segments();
     PER_HEAP_ISOLATED_METHOD void distribute_free_regions();
 #ifdef BACKGROUND_GC
     PER_HEAP_ISOLATED_METHOD void reset_write_watch_for_gc_heap(void* base_address, size_t region_size);
@@ -2597,11 +2595,17 @@ class gc_heap
     // re-initialize a heap in preparation to putting it back into service
     PER_HEAP_METHOD void recommission_heap();
 
+    PER_HEAP_ISOLATED_METHOD size_t get_num_completed_gcs();
+
+    PER_HEAP_ISOLATED_METHOD int calculate_new_heap_count();
+
     // check if we should change the heap count
     PER_HEAP_METHOD void check_heap_count();
 
-    PER_HEAP_METHOD bool prepare_to_change_heap_count (int new_n_heaps);
+    PER_HEAP_ISOLATED_METHOD bool prepare_to_change_heap_count (int new_n_heaps);
     PER_HEAP_METHOD bool change_heap_count (int new_n_heaps);
+
+    PER_HEAP_ISOLATED_METHOD size_t get_msl_wait_time();
 #endif //DYNAMIC_HEAP_COUNT
 #endif //USE_REGIONS
 
@@ -3778,6 +3782,13 @@ class gc_heap
     PER_HEAP_FIELD_MAINTAINED mark*  loh_pinned_queue;
 #endif //FEATURE_LOH_COMPACTION
 
+#ifdef DYNAMIC_HEAP_COUNT
+    PER_HEAP_FIELD_MAINTAINED GCEvent gc_idle_thread_event;
+#ifdef BACKGROUND_GC
+    PER_HEAP_FIELD_MAINTAINED GCEvent bgc_idle_thread_event;
+#endif //BACKGROUND_GC
+#endif //DYNAMIC_HEAP_COUNT
+
     /******************************************/
     // PER_HEAP_FIELD_MAINTAINED_ALLOC fields //
     /******************************************/
@@ -4084,7 +4095,6 @@ class gc_heap
     // These 2 fields' values do not change but are set/unset per GC
     PER_HEAP_ISOLATED_FIELD_SINGLE_GC GCEvent gc_start_event;
     PER_HEAP_ISOLATED_FIELD_SINGLE_GC GCEvent ee_suspend_event;
-    PER_HEAP_ISOLATED_FIELD_SINGLE_GC GCEvent gc_idle_thread_event;
 
     // Also updated on the heap#0 GC thread because that's where we are actually doing the decommit.
     PER_HEAP_ISOLATED_FIELD_SINGLE_GC BOOL gradual_decommit_in_progress_p;
@@ -4163,6 +4173,10 @@ class gc_heap
     PER_HEAP_ISOLATED_FIELD_SINGLE_GC uint8_t* gc_high; // high end of the highest region being condemned
 #endif //USE_REGIONS
 
+#ifdef STRESS_DYNAMIC_HEAP_COUNT
+    PER_HEAP_ISOLATED_FIELD_SINGLE_GC int heaps_in_this_gc;
+#endif //STRESS_DYNAMIC_HEAP_COUNT
+
     /**************************************************/
     // PER_HEAP_ISOLATED_FIELD_SINGLE_GC_ALLOC fields //
     /**************************************************/
@@ -4261,37 +4275,65 @@ class gc_heap
 #endif //USE_REGIONS
 
 #ifdef DYNAMIC_HEAP_COUNT
+    // Sample collection -
+    // 
+    // For every GC, we collect the msl wait time + GC pause duration info and use both to calculate the
+    // throughput cost percentage. We will also be using the wait time and the GC pause duration separately
+    // for other purposes in the future.
+    //
+    // For all gen2 GCs we also keep a separate array currently just for the GC cost. This serves as a backstop
+    // to smooth out the situation when we rarely pick the gen2 GCs in the first array.
     struct dynamic_heap_count_data_t
     {
         static const int sample_size = 3;
 
         struct sample
         {
-            uint64_t    elapsed_between_gcs;    // time between gcs in microseconds
-            uint64_t    gc_elapsed_time;        // time the gc took
-            uint64_t    soh_msl_wait_time;      // time the allocator spent waiting for the soh msl lock
-            uint64_t    uoh_msl_wait_time;      // time the allocator spent waiting for the uoh msl lock
-            size_t      allocating_thread_count;// number of allocating threads
-            size_t      heap_size;
+            uint64_t    elapsed_between_gcs;    // time between gcs in microseconds (this should really be between_pauses)
+            uint64_t    gc_pause_time;          // pause time for this GC
+            uint64_t    msl_wait_time;
         };
 
-        unsigned        sample_index;
+        uint32_t        sample_index;
         sample          samples[sample_size];
+        size_t          prev_num_completed_gcs;
+
+        uint32_t        gen2_sample_index;
+        // This is (gc_elapsed_time / time inbetween this and the last gen2 GC)
+        float           gen2_gc_percents[sample_size];
 
-        float median_percent_overhead;          // estimated overhead of allocator + gc
-        float smoothed_median_percent_overhead; // exponentially smoothed version
-        float percent_heap_space_cost_per_heap; // percent space cost of adding a heap
-        float overhead_reduction_per_step_up;   // percentage effect on overhead of increasing heap count
-        float overhead_increase_per_step_down;  // percentage effect on overhead of decreasing heap count
-        float space_cost_increase_per_step_up;  // percentage effect on space of increasing heap count
-        float space_cost_decrease_per_step_down;// percentage effect on space of decreasing heap count
+        float median_throughput_cost_percent;          // estimated overhead of allocator + gc
+        float smoothed_median_throughput_cost_percent; // exponentially smoothed version
+        float percent_heap_space_cost_per_heap;        // percent space cost of adding a heap
+        float tcp_reduction_per_step_up;               // throughput cost percent effect of increasing heap count
+        float tcp_increase_per_step_down;              // throughput cost percent effect of decreasing heap count
+        float scp_increase_per_step_up;                // space cost percent effect of increasing heap count
+        float scp_decrease_per_step_down;              // space cost percent effect of decreasing heap count
 
         int             new_n_heaps;
+        // the heap count we changed from
+        int             last_n_heaps;
+        // don't start a GC till we see (n_max_heaps - new_n_heaps) number of threads idling
+        VOLATILE(int32_t) idle_thread_count;
+        bool            init_only_p;
+
+        bool            should_change_heap_count;
+        int             heap_count_to_change_to;
+        int             heap_count_change_count;
 #ifdef STRESS_DYNAMIC_HEAP_COUNT
         int             lowest_heap_with_msl_uoh;
 #endif //STRESS_DYNAMIC_HEAP_COUNT
+
+        float get_median_gen2_gc_percent()
+        {
+            return median_of_3 (gen2_gc_percents[0], gen2_gc_percents[1], gen2_gc_percents[2]);
+        }
     };
     PER_HEAP_ISOLATED_FIELD_MAINTAINED dynamic_heap_count_data_t dynamic_heap_count_data;
+    PER_HEAP_ISOLATED_FIELD_MAINTAINED uint64_t last_suspended_end_time;
+    // If the last full GC is blocking, this is that GC's index; for BGC, this is the settings.gc_index
+    // when the BGC ended.
+    PER_HEAP_ISOLATED_FIELD_MAINTAINED size_t gc_index_full_gc_end;
 #endif //DYNAMIC_HEAP_COUNT
 
     /****************************************************/
@@ -4867,7 +4909,6 @@ uint64_t& dd_previous_time_clock (dynamic_data* inst)
     return inst->previous_time_clock;
 }
 
-
 inline
 size_t& dd_gc_clock_interval (dynamic_data* inst)
 {

src/coreclr/jit/emitxarch.cpp

@@ -5485,6 +5485,13 @@ void emitter::emitInsRMW(instruction ins, emitAttr attr, GenTreeStoreInd* storeI
     {
         assert(!src->isContained()); // there must be one non-contained src
 
+        if (addr->isContained() && addr->OperIs(GT_LCL_ADDR))
+        {
+            GenTreeLclVarCommon* lclVar = addr->AsLclVarCommon();
+            emitIns_S_R(ins, attr, src->GetRegNum(), lclVar->GetLclNum(), lclVar->GetLclOffs());
+            return;
+        }
+
         // ind, reg
         id = emitNewInstrAmd(attr, offset);
         emitHandleMemOp(storeInd, id, emitInsModeFormat(ins, IF_ARD_RRD), ins);

src/coreclr/jit/morph.cpp

@@ -10807,8 +10807,6 @@ GenTree* Compiler::fgOptimizeHWIntrinsic(GenTreeHWIntrinsic* node)
         }
 
 #if defined(TARGET_XARCH)
-        case NI_AVX512F_Add:
-        case NI_AVX512BW_Add:
         case NI_AVX512F_And:
         case NI_AVX512DQ_And:
         case NI_AVX512F_AndNot:
@@ -10850,13 +10848,6 @@ GenTree* Compiler::fgOptimizeHWIntrinsic(GenTreeHWIntrinsic* node)
 
             switch (intrinsicId)
             {
-                case NI_AVX512F_Add:
-                case NI_AVX512BW_Add:
-                {
-                    maskIntrinsicId = NI_AVX512F_AddMask;
-                    break;
-                }
-
                 case NI_AVX512F_And:
                 case NI_AVX512DQ_And:
                 {

src/libraries/Microsoft.Extensions.Options.DataAnnotations/src/DataAnnotationValidateOptions.cs

@@ -95,7 +95,8 @@ private static bool TryValidateOptions(object options, string qualifiedName, Lis
 
             foreach (PropertyInfo propertyInfo in options.GetType().GetProperties(BindingFlags.Instance | BindingFlags.Public))
             {
-                if (propertyInfo.GetMethod is null)
+                // Indexers are properties which take parameters. Ignore them.
+                if (propertyInfo.GetMethod is null || propertyInfo.GetMethod.GetParameters().Length > 0)
                 {
                     continue;
                 }

src/libraries/Microsoft.Extensions.Options/tests/SourceGeneration.Unit.Tests/OptionsRuntimeTests.cs

@@ -177,6 +177,30 @@ public void TestValidationWithEnumeration()
                         result2.Failures);
         }
 
+        [ConditionalFact(typeof(PlatformDetection), nameof(PlatformDetection.IsNotBrowser))]
+        public void TestObjectsWithIndexerProperties()
+        {
+            DataAnnotationValidateOptions<MyDictionaryOptions> dataAnnotationValidateOptions1 = new("MyDictionaryOptions");
+            MyDictionaryOptionsOptionsValidator sourceGenOptionsValidator1 = new();
+
+            var options1 = new MyDictionaryOptions();
+            ValidateOptionsResult result1 = sourceGenOptionsValidator1.Validate("MyDictionaryOptions", options1);
+            ValidateOptionsResult result2 = dataAnnotationValidateOptions1.Validate("MyDictionaryOptions", options1);
+
+            Assert.True(result1.Succeeded);
+            Assert.True(result2.Succeeded);
+
+            DataAnnotationValidateOptions<MyListOptions<string>> dataAnnotationValidateOptions2 = new("MyListOptions");
+            MyListOptionsOptionsValidator sourceGenOptionsValidator2 = new();
+
+            var options2 = new MyListOptions<string>() { Prop = "test" };
+            result1 = sourceGenOptionsValidator2.Validate("MyListOptions", options2);
+            result2 = dataAnnotationValidateOptions2.Validate("MyListOptions", options2);
+
+            Assert.True(result1.Succeeded);
+            Assert.True(result2.Succeeded);
+        }
+
         [ConditionalFact(typeof(PlatformDetection), nameof(PlatformDetection.IsNotBrowser))]
         public void TestValidationWithCyclicReferences()
         {
@@ -302,6 +326,12 @@ public partial class MySourceGenOptionsValidator : IValidateOptions<MyOptions>
     {
     }
 
+    public class MyDictionaryOptions : Dictionary<string, string> { [Required] public string Prop { get; set; } = "test"; }
+    [OptionsValidator] public partial class MyDictionaryOptionsOptionsValidator : IValidateOptions<MyDictionaryOptions> { }
+
+    public class MyListOptions<T> : List<T> { [Required] public T Prop { get; set; } = default; }
+    [OptionsValidator] public partial class MyListOptionsOptionsValidator : IValidateOptions<MyListOptions<string>> { }
+
 #if NET8_0_OR_GREATER
     public class OptionsUsingNewAttributes
     {

src/mono/wasm/README.md

@@ -350,3 +350,12 @@ npm update --lockfile-version=1
 | Multi-thread      | linux: build only          | none                                 |
 
 * `high resource aot` runs a few specific library tests with AOT, that require more memory to AOT.
+
+
+# Perf pipeline
+
+TBD
+
+## Updates needed
+
+- when the base OS is upgraded, check if the version of node installed in the `eng/pipelines/coreclr/templates/run-performance-job.yml` needs an upgrade too.

src/mono/wasm/Wasm.Build.Tests/Blazor/MiscTests3.cs

@@ -63,7 +63,7 @@ public static class MyDllImports
                 public static extern int cpp_add(int a, int b);
             }}";
 
-        File.WriteAllText(Path.Combine(_projectDir!, "Pages", "MyDllImport.cs"), myDllImportCs);
+        File.WriteAllText(Path.Combine(_projectDir!, "Components", "Pages", "MyDllImport.cs"), myDllImportCs);
 
         AddItemsPropertiesToProject(projectFile, extraItems: @"<NativeFileReference Include=""mylib.cpp"" />");
         BlazorAddRazorButton("cpp_add", """
@@ -144,7 +144,7 @@ public void BugRegression_60479_WithRazorClassLib()
         Assert.Contains(razorClassLibraryFileName, lazyVal.EnumerateObject().Select(jp => jp.Name));
     }
 
-    private void BlazorAddRazorButton(string buttonText, string customCode, string methodName = "test", string razorPage = "Pages/Counter.razor")
+    private void BlazorAddRazorButton(string buttonText, string customCode, string methodName = "test", string razorPage = "Components/Pages/Counter.razor")
     {
         string additionalCode = $$"""
             <p role="{{methodName}}">Output: @outputText</p>

src/mono/wasm/Wasm.Build.Tests/Blazor/WorkloadRequiredTests.cs

@@ -84,7 +84,7 @@ public async Task WorkloadNotRequiredForInvariantGlobalization(string config, bo
         if (invariant)
             AddItemsPropertiesToProject(projectFile, extraProperties: "<InvariantGlobalization>true</InvariantGlobalization>");
 
-        string counterPath = Path.Combine(Path.GetDirectoryName(projectFile)!, "Pages", "Counter.razor");
+        string counterPath = Path.Combine(Path.GetDirectoryName(projectFile)!, "Components", "Pages", "Counter.razor");
         string allText = File.ReadAllText(counterPath);
         string ccText = "currentCount++;";
         if (allText.IndexOf(ccText) < 0)