[PATCH] UefiCpuPkg/CpuFeature: Introduce FirstX to indicate 1st unit in parent scope.

["Ni, Ray" <ray.ni@intel.com>]

REF: https://bugzilla.tianocore.org/show_bug.cgi?id=1584

The flow of CPU feature initialization logic is:
1. BSP calls GetConfigDataFunc() for each thread/AP;
2. Each thread/AP calls SupportFunc() to detect its own capability;
3. BSP calls InitializeFunc() for each thead/AP.

There is a design gap in step #3. For a package scope feature that only
requires one thread of each package does the initialization operation,
what InitializeFunc() currently does is to do the initialization
operation only CPU physical location Core# is 0.
But in certain platform, Core#0 might be disabled in hardware level
which results the certain package scope feature isn't initialized at
all.

The patch adds a new field FistX to indicate the CPU's location in
its parent scope.
FirstX.Package is set for all APs/threads under first package;
FirstX.Core is set for all APs/threads under first core of each
package;
FirstX.Thread is set for the AP/thread of each core.

Signed-off-by: Ray Ni <ray.ni@intel.com>
Cc: Eric Dong <eric.dong@intel.com>
Cc: Star Zeng <star.zeng@intel.com>
Cc: Michael D Kinney <michael.d.kinney@intel.com>
---
 .../Include/Library/RegisterCpuFeaturesLib.h  | 19 +++++
 .../CpuFeaturesInitialize.c                   | 74 +++++++++++++++++++
 2 files changed, 93 insertions(+)

diff --git a/UefiCpuPkg/Include/Library/RegisterCpuFeaturesLib.h b/UefiCpuPkg/Include/Library/RegisterCpuFeaturesLib.h
index 6f964027be..7c1e122ed7 100644
--- a/UefiCpuPkg/Include/Library/RegisterCpuFeaturesLib.h
+++ b/UefiCpuPkg/Include/Library/RegisterCpuFeaturesLib.h
@@ -78,6 +78,20 @@
 #define CPU_FEATURE_END                             MAX_UINT32
 /// @}
 
+///
+/// The bit field to indicate whether the CPU is the first in its parent scope.
+///
+typedef struct {
+  UINT32 Thread : 1;
+  UINT32 Core : 1;
+  UINT32 Module : 1;
+  UINT32 Tile : 1;
+  UINT32 Die : 1;
+  UINT32 Package : 1;
+  UINT32 Reserved : 26;
+} REGISTER_CPU_FEATURE_FIRST_X;
+  
+
 ///
 /// CPU Information passed into the SupportFunc and InitializeFunc of the
 /// RegisterCpuFeature() library function.  This structure contains information
@@ -88,6 +102,11 @@ typedef struct {
   /// The package that the CPU resides
   ///
   EFI_PROCESSOR_INFORMATION            ProcessorInfo;
+
+  ///
+  /// The bit flag indicating whether the CPU is the first Thread/Core/Module/Tile/Die/Package in its parent scope.
+  ///
+  REGISTER_CPU_FEATURE_FIRST_X         FirstX;
   ///
   /// The Display Family of the CPU computed from CPUID leaf CPUID_VERSION_INFO
   ///
diff --git a/UefiCpuPkg/Library/RegisterCpuFeaturesLib/CpuFeaturesInitialize.c b/UefiCpuPkg/Library/RegisterCpuFeaturesLib/CpuFeaturesInitialize.c
index aff7ad600c..c7858bc3a8 100644
--- a/UefiCpuPkg/Library/RegisterCpuFeaturesLib/CpuFeaturesInitialize.c
+++ b/UefiCpuPkg/Library/RegisterCpuFeaturesLib/CpuFeaturesInitialize.c
@@ -110,6 +110,9 @@ CpuInitDataInitialize (
   EFI_CPU_PHYSICAL_LOCATION            *Location;
   BOOLEAN                              *CoresVisited;
   UINTN                                Index;
+  UINT32                               PackageIndex;
+  UINT32                               CoreIndex;
+  UINT32                               FirstIndex;
   ACPI_CPU_DATA                        *AcpiCpuData;
   CPU_STATUS_INFORMATION               *CpuStatus;
   UINT32                               *ValidCoreCountPerPackage;
@@ -239,6 +242,77 @@ CpuInitDataInitialize (
   ASSERT (CpuFeaturesData->CpuFlags.CoreSemaphoreCount != NULL);
   CpuFeaturesData->CpuFlags.PackageSemaphoreCount = AllocateZeroPool (sizeof (UINT32) * CpuStatus->PackageCount * CpuStatus->MaxCoreCount * CpuStatus->MaxThreadCount);
   ASSERT (CpuFeaturesData->CpuFlags.PackageSemaphoreCount != NULL);
+
+  //
+  // Initialize CpuFeaturesData->InitOrder[].CpuInfo.FirstX
+  //
+  
+  //
+  // Set FirstX.Package for each thread belonging to the first package.
+  //
+  FirstIndex = MAX_UINT32;
+  for (ProcessorNumber = 0; ProcessorNumber < NumberOfCpus; ProcessorNumber++) {
+    Location = &CpuFeaturesData->InitOrder[ProcessorNumber].CpuInfo.ProcessorInfo.Location;
+    FirstIndex = MIN (Location->Package, FirstIndex);
+  }
+  for (ProcessorNumber = 0; ProcessorNumber < NumberOfCpus; ProcessorNumber++) {
+    Location = &CpuFeaturesData->InitOrder[ProcessorNumber].CpuInfo.ProcessorInfo.Location;
+    if (Location->Package == FirstIndex) {
+      CpuFeaturesData->InitOrder[ProcessorNumber].CpuInfo.FirstX.Package = 1;
+    }
+  }
+
+  //
+  // Set FirstX.Die/Tile/Module for each thread assuming:
+  //  single Die under each package, single Tile under each Die, single Module under each Tile
+  //
+  for (ProcessorNumber = 0; ProcessorNumber < NumberOfCpus; ProcessorNumber++) {
+    CpuFeaturesData->InitOrder[ProcessorNumber].CpuInfo.FirstX.Die = 1;
+    CpuFeaturesData->InitOrder[ProcessorNumber].CpuInfo.FirstX.Tile = 1;
+    CpuFeaturesData->InitOrder[ProcessorNumber].CpuInfo.FirstX.Module = 1;
+  }
+
+  for (PackageIndex = 0; PackageIndex < CpuStatus->PackageCount; PackageIndex++) {
+    //
+    // Set FirstX.Core for each thread belonging to the first core of each package.
+    //
+    FirstIndex = MAX_UINT32;
+    for (ProcessorNumber = 0; ProcessorNumber < NumberOfCpus; ProcessorNumber++) {
+      Location = &CpuFeaturesData->InitOrder[ProcessorNumber].CpuInfo.ProcessorInfo.Location;
+      if (Location->Package == PackageIndex) {
+        FirstIndex = MIN (Location->Core, FirstIndex);
+      }
+    }
+
+    for (ProcessorNumber = 0; ProcessorNumber < NumberOfCpus; ProcessorNumber++) {
+      Location = &CpuFeaturesData->InitOrder[ProcessorNumber].CpuInfo.ProcessorInfo.Location;
+      if (Location->Package == PackageIndex && Location->Core == FirstIndex) {
+        CpuFeaturesData->InitOrder[ProcessorNumber].CpuInfo.FirstX.Core = 1;
+      }
+    }
+  }
+
+  for (PackageIndex = 0; PackageIndex < CpuStatus->PackageCount; PackageIndex++) {
+    for (CoreIndex = 0; CoreIndex < CpuStatus->MaxCoreCount; CoreIndex++) {
+      //
+      // Set FirstX.Thread for the first thread of each core.
+      //
+      FirstIndex = MAX_UINT32;
+      for (ProcessorNumber = 0; ProcessorNumber < NumberOfCpus; ProcessorNumber++) {
+        Location = &CpuFeaturesData->InitOrder[ProcessorNumber].CpuInfo.ProcessorInfo.Location;
+        if (Location->Package == PackageIndex && Location->Core == CoreIndex) {
+          FirstIndex = MIN (Location->Thread, FirstIndex);
+        }
+      }
+
+      for (ProcessorNumber = 0; ProcessorNumber < NumberOfCpus; ProcessorNumber++) {
+        Location = &CpuFeaturesData->InitOrder[ProcessorNumber].CpuInfo.ProcessorInfo.Location;
+        if (Location->Package == PackageIndex && Location->Core == CoreIndex && Location->Thread == FirstIndex) {
+          CpuFeaturesData->InitOrder[ProcessorNumber].CpuInfo.FirstX.Thread = 1;
+        }
+      }
+    }
+  }
 }
 
 /**
-- 
2.21.0.windows.1