core: emulate six CPU cores

Switch 2 applications are given cores 0..5 (title 4's npdm) and start a
job worker per core. With four emulated cores, 4 and 5 were folded onto
0 and 1, so two cores ran two workers each. Switch 1 titles keep using
cores 0..3; core 3 stays the shared system core.

5 files changed, 13 insertions, 7 deletions

diff --git a/src/core/hardware_properties.h b/src/core/hardware_properties.h
index 3f870becbf..aeea5f858f 100644
--- a/src/core/hardware_properties.h
+++ b/src/core/hardware_properties.h
@@ -15,11 +15,14 @@ namespace Hardware {
 
 constexpr u64 BASE_CLOCK_RATE = 1'020'000'000; // Default CPU Frequency = 1020 MHz
 constexpr u64 CNTFREQ = 19'200'000;            // CNTPCT_EL0 Frequency = 19.2 MHz
-constexpr u32 NUM_CPU_CORES = 4;               // Number of CPU Cores - sync with dynarmic exclusive_monitor.h
+// Switch 1 applications get cores 0..3. Switch 2 applications ask for 0..5 (title 4's npdm), and
+// its job system starts a worker per core, so emulate six: with four, cores 0 and 1 each ran two
+// of its workers. Switch 1 titles leave cores 4 and 5 idle.
+constexpr u32 NUM_CPU_CORES = 6;               // Number of CPU Cores - sync with dynarmic exclusive_monitor.h
 
 // Virtual to Physical core map.
 constexpr std::array<s32, Common::BitSize<u64>()> VirtualToPhysicalCoreMap{
-    0, 1, 2, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
+    0, 1, 2, 3, 4, 5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
     0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3,
 };
 
diff --git a/src/core/hle/kernel/global_scheduler_context.cpp b/src/core/hle/kernel/global_scheduler_context.cpp
index 6d1cd030e0..98d14c8549 100644
--- a/src/core/hle/kernel/global_scheduler_context.cpp
+++ b/src/core/hle/kernel/global_scheduler_context.cpp
@@ -40,12 +40,15 @@ void GlobalSchedulerContext::RemoveThread(KThread* thread) noexcept {
 /// @note This operation happens every 10ms.
 void GlobalSchedulerContext::PreemptThreads(KernelCore& kernel) noexcept {
     // The priority levels at which the global scheduler preempts threads every 10 ms. They are
-    // ordered from Core 0 to Core 3.
+    // ordered from Core 0 to Core 5; core 3 is shared with the system, 4 and 5 are application
+    // cores like 0..2.
     static constexpr std::array<u32, Core::Hardware::NUM_CPU_CORES> per_core{
         59,
         59,
         59,
         63,
+        59,
+        59,
     };
     ASSERT(KScheduler::IsSchedulerLockedByCurrentThread(kernel));
     for (u32 core_id = 0; core_id < per_core.size(); core_id++)
diff --git a/src/core/hle/kernel/kernel.cpp b/src/core/hle/kernel/kernel.cpp
index 4f1d849209..b437dd4b4e 100644
--- a/src/core/hle/kernel/kernel.cpp
+++ b/src/core/hle/kernel/kernel.cpp
@@ -171,7 +171,7 @@ struct KernelCore::Impl {
             schedulers[core_id].reset();
         }
 
-        // Next host thead ID to use, 0-3 IDs represent core threads, >3 represent others
+        // Next host thead ID to use, IDs below NUM_CPU_CORES represent core threads, the rest others
         next_host_thread_id = Core::Hardware::NUM_CPU_CORES;
 
         // Close kernel objects that were not freed on shutdown
diff --git a/src/core/hle/kernel/svc/svc_thread.cpp b/src/core/hle/kernel/svc/svc_thread.cpp
index 9838e68476..7f6b959256 100644
--- a/src/core/hle/kernel/svc/svc_thread.cpp
+++ b/src/core/hle/kernel/svc/svc_thread.cpp
@@ -20,8 +20,8 @@ constexpr bool IsValidVirtualCoreId(int32_t core_id) {
     return (0 <= core_id && core_id < static_cast<int32_t>(Core::Hardware::NUM_CPU_CORES));
 }
 
-// Switch 2 processes may be given virtual cores 4 and 5 (core mask 0x3f). Only four cores are
-// emulated, so fold the extra ones back onto 0..3, both in core ids and in affinity masks.
+// Switch 2 hardware has eight cores. Six are emulated (the most an application is given), so
+// fold any id past them back onto the low cores, both in core ids and in affinity masks.
 constexpr int32_t FoldExtraCoreId(int32_t core_id) {
     constexpr auto num_cores = static_cast<int32_t>(Core::Hardware::NUM_CPU_CORES);
     return core_id >= num_cores && core_id < 2 * num_cores ? core_id - num_cores : core_id;
diff --git a/src/dynarmic/src/dynarmic/interface/exclusive_monitor.h b/src/dynarmic/src/dynarmic/interface/exclusive_monitor.h
index 05b067bdd8..46fce86bda 100644
--- a/src/dynarmic/src/dynarmic/interface/exclusive_monitor.h
+++ b/src/dynarmic/src/dynarmic/interface/exclusive_monitor.h
@@ -82,7 +82,7 @@ private:
 
     static constexpr VAddr RESERVATION_GRANULE_MASK = 0xFFFF'FFFF'FFFF'FFFFull;
     static constexpr VAddr INVALID_EXCLUSIVE_ADDRESS = 0xDEAD'DEAD'DEAD'DEADull;
-    static constexpr size_t MAX_NUM_CPU_CORES = 4; // Sync with src/core/hardware_properties
+    static constexpr size_t MAX_NUM_CPU_CORES = 6; // Sync with src/core/hardware_properties
     boost::container::static_vector<VAddr, MAX_NUM_CPU_CORES> exclusive_addresses;
     boost::container::static_vector<Vector, MAX_NUM_CPU_CORES> exclusive_values;
     SpinLock lock;