ADD: Tweaks on AMD64's SMP architecture, and almost done with it.

Signed-off-by: Amlal El Mahrouss <amlal.elmahrouss@icloud.com>

7 files changed, 134 insertions, 94 deletions

diff --git a/dev/Kernel/ArchKit/ArchKit.h b/dev/Kernel/ArchKit/ArchKit.h
index af8aef1a..1a3e9c97 100644
--- a/dev/Kernel/ArchKit/ArchKit.h
+++ b/dev/Kernel/ArchKit/ArchKit.h
@@ -46,16 +46,17 @@ namespace Kernel
 	template <typename WordLength>
 	inline Void ke_dma_write(WordLength base, WordLength reg, WordLength value) noexcept
 	{
-		*(volatile WordLength*)((UInt64)base + reg) = value;
+		*(volatile WordLength*)(base + reg) = value;
 	}
 
 	/// @brief read from mapped memory register.
 	/// @param base base address
 	/// @param reg the register.
 	/// @return the value inside the register.
-	inline UInt32 ke_dma_read(UInt32 base, UInt32 reg) noexcept
+	template <typename WordLength>
+	inline UInt32 ke_dma_read(WordLength base, WordLength reg) noexcept
 	{
-		return *(volatile UInt32*)((UInt64)base + reg);
+		return *(volatile WordLength*)((UInt64)base + reg);
 	}
 
 	namespace HAL
diff --git a/dev/Kernel/HALKit/AMD64/HalApplicationProcessor.cc b/dev/Kernel/HALKit/AMD64/HalApplicationProcessor.cc
index 19f3a486..20d512f4 100644
--- a/dev/Kernel/HALKit/AMD64/HalApplicationProcessor.cc
+++ b/dev/Kernel/HALKit/AMD64/HalApplicationProcessor.cc
@@ -4,6 +4,7 @@
 
 ------------------------------------------- */
 
+#include "NewKit/Defines.h"
 #include <Mod/ACPI/ACPIFactoryInterface.h>
 #include <KernelKit/UserProcessScheduler.h>
 #include <HALKit/AMD64/Processor.h>
@@ -36,7 +37,7 @@
 
 namespace Kernel::HAL
 {
-	struct MADT_TABLE;
+	struct PROCESS_APIC_MADT;
 	struct PROCESS_CONTROL_BLOCK;
 
 	struct PROCESS_CONTROL_BLOCK final
@@ -46,9 +47,9 @@ namespace Kernel::HAL
 
 	EXTERN_C Void _hal_spin_core(Void);
 
-	STATIC struct MADT_TABLE* kMADTBlock = nullptr;
-	STATIC Bool				  kSMPAware	 = false;
-	STATIC Int64			  kSMPCount	 = 0;
+	STATIC struct PROCESS_APIC_MADT* kMADTBlock = nullptr;
+	STATIC Bool						 kSMPAware	= false;
+	STATIC Int64					 kSMPCount	= 0;
 
 	STATIC UIntPtr kApicBaseAddress = 0UL;
 
@@ -57,7 +58,7 @@ namespace Kernel::HAL
 	STATIC VoidPtr kRawMADT								   = nullptr;
 
 	/// @brief Multiple APIC Descriptor Table.
-	struct MADT_TABLE final : public SDT
+	struct PROCESS_APIC_MADT final : public SDT
 	{
 		UInt32 Address; // Madt address
 		UInt8  Flags;	// Madt flags
@@ -118,10 +119,15 @@ namespace Kernel::HAL
 	/// @return
 	/***********************************************************************************/
 
-	Void hal_send_start_ipi(UInt32 apic_id, UInt32 vector, UInt32 target)
+	Void hal_send_start_ipi(UInt32 target, UInt32 apic_id)
 	{
-		Kernel::ke_dma_write<UInt32>(target, kAPIC_ICR_High, (apic_id << 24));
-		Kernel::ke_dma_write<UInt32>(target, kAPIC_ICR_Low, kAPIC_SIPI_Vector | vector);
+		Kernel::ke_dma_write<UInt32>(target, kAPIC_ICR_High, apic_id << 24);
+		Kernel::ke_dma_write<UInt32>(target, kAPIC_ICR_Low, 0x00000500 | 0x00004000 | 0x00000000);
+
+		while (Kernel::ke_dma_read<UInt32>(target, kAPIC_ICR_Low) & 0x1000)
+		{
+			;
+		}
 	}
 
 	/***********************************************************************************/
@@ -131,10 +137,15 @@ namespace Kernel::HAL
 	/// @param target
 	/// @return
 	/***********************************************************************************/
-	Void hal_send_sipi(UInt32 apic_id, UInt8 vector, UInt32 target)
+	Void hal_send_sipi(UInt32 target, UInt32 apic_id, UInt8 vector)
 	{
-		Kernel::ke_dma_write<UInt32>(target, kAPIC_ICR_High, (apic_id << 24));
-		Kernel::ke_dma_write<UInt32>(target, kAPIC_ICR_Low, kAPIC_EIPI_Vector | vector);
+		Kernel::ke_dma_write<UInt32>(target, kAPIC_ICR_High, apic_id << 24);
+		Kernel::ke_dma_write<UInt32>(target, kAPIC_ICR_Low, 0x00000600 | 0x00004000 | 0x00000000 | vector);
+
+		while (Kernel::ke_dma_read<UInt32>(target, kAPIC_ICR_Low) & 0x1000)
+		{
+			;
+		}
 	}
 
 	STATIC PROCESS_CONTROL_BLOCK kProcessBlocks[kSchedProcessLimitPerTeam] = {0};
@@ -155,12 +166,7 @@ namespace Kernel::HAL
 
 		auto first_id = kAPICLocales[0];
 
-		hal_send_sipi(first_id, (stack_frame->BP >> 12), kApicBaseAddress);
-
-		HardwareTimer timer(Kernel::Milliseconds(10));
-		timer.Wait();
-
-		hal_send_sipi(first_id, (stack_frame->BP >> 12), kApicBaseAddress);
+		hal_send_sipi(kApicBaseAddress, first_id, (UInt8)(((UIntPtr)stack_frame->BP) >> 12));
 
 		return YES;
 	}
@@ -177,7 +183,12 @@ namespace Kernel::HAL
 	/***********************************************************************************/
 	/// @brief Assembly symbol to bootstrap AP.
 	/***********************************************************************************/
-	EXTERN_C Void hal_ap_start(Void);
+	EXTERN_C Char* hal_ap_blob_start;
+
+	/***********************************************************************************/
+	/// @brief Assembly symbol to bootstrap AP.
+	/***********************************************************************************/
+	EXTERN_C Char* hal_ap_blob_end;
 
 	/***********************************************************************************/
 	/// @brief Fetch and enable SMP scheduler.
@@ -197,7 +208,7 @@ namespace Kernel::HAL
 		auto hw_and_pow_int = PowerFactoryInterface(vendor_ptr);
 		kRawMADT			= hw_and_pow_int.Find(kApicSignature).Leak().Leak();
 
-		kMADTBlock = reinterpret_cast<MADT_TABLE*>(kRawMADT);
+		kMADTBlock = reinterpret_cast<PROCESS_APIC_MADT*>(kRawMADT);
 		kSMPAware  = NO;
 
 		if (kMADTBlock)
@@ -207,10 +218,17 @@ namespace Kernel::HAL
 			kSMPInterrupt = 0;
 			kSMPCount	  = 0;
 
-			kcout << "SMP: Probing Local APICs...\r";
+			kcout << "SMP: Starting APs...\r";
 
 			kApicBaseAddress = kMADTBlock->Address;
 
+			constexpr auto kMemoryAPStart = 0x7C000;
+			Char*		   ptr_ap_code	  = reinterpret_cast<Char*>(kMemoryAPStart);
+			
+			SizeT hal_ap_blob_len = hal_ap_blob_end - hal_ap_blob_start;
+
+			rt_copy_memory((Char*)hal_ap_blob_start, ptr_ap_code, hal_ap_blob_len);
+
 			while (Yes)
 			{
 				if (kMADTBlock->List[index].Type > 9 ||
@@ -228,12 +246,14 @@ namespace Kernel::HAL
 
 					// I'll just make the AP start from scratch here.
 
-					hal_send_sipi(kAPICLocales[kSMPCount], ((UIntPtr)hal_ap_start >> 12), kApicBaseAddress);
+					hal_send_start_ipi(kApicBaseAddress, kAPICLocales[kSMPCount]);
 
 					HardwareTimer timer(Kernel::Milliseconds(10));
 					timer.Wait();
 
-					hal_send_sipi(kAPICLocales[kSMPCount], ((UIntPtr)hal_ap_start >> 12), kApicBaseAddress);
+					/// TODO: HAL helper to create an address.
+
+					hal_send_sipi(kApicBaseAddress, kAPICLocales[kSMPCount], (UInt8)(((UIntPtr)ptr_ap_code) >> 12));
 
 					++kSMPCount;
 					break;
diff --git a/dev/Kernel/HALKit/AMD64/HalApplicationProcessorGNU.s b/dev/Kernel/HALKit/AMD64/HalApplicationProcessorGNU.s
new file mode 100644
index 00000000..a8ad3b76
--- /dev/null
+++ b/dev/Kernel/HALKit/AMD64/HalApplicationProcessorGNU.s
@@ -0,0 +1,8 @@
+.data
+
+.global hal_ap_blob_start         /* Export the start symbol */
+.global hal_ap_blob_end           /* Export the end symbol */
+
+hal_ap_blob_start:
+    .incbin "HALKit/AMD64/HalApplicationProcessorStartup.bin"
+hal_ap_blob_end:
diff --git a/dev/Kernel/HALKit/AMD64/HalApplicationProcessorStartup.asm b/dev/Kernel/HALKit/AMD64/HalApplicationProcessorStartup.asm
new file mode 100644
index 00000000..2d2f6fbb
--- /dev/null
+++ b/dev/Kernel/HALKit/AMD64/HalApplicationProcessorStartup.asm
@@ -0,0 +1,77 @@
+;; /*
+;; *	========================================================
+;; *
+;; *	ZKA
+;; * 	Copyright (C) 2024, Theater Quality Corp, all rights reserved., all rights reserved.
+;; *
+;; * 	========================================================
+;; */
+
+[bits 16]
+[org 0x7c000]
+
+hal_ap_start:
+    mov ax, 0x0      
+    mov ss, ax
+    mov esp, 0x7000  
+
+    cli    
+    mov eax, cr0
+    or eax, 1        
+    mov cr0, eax
+    jmp .hal_ap_start_flush       
+.hal_ap_start_flush:
+    mov ax, 0x10     
+    mov ds, ax
+    mov es, ax
+    mov fs, ax
+    mov gs, ax
+    mov ss, ax
+
+    mov eax, cr4
+    or eax, 1 << 5   
+    mov cr4, eax
+
+    mov eax, cr3
+    mov cr3, eax     
+
+    mov ecx, 0xC0000080 
+    rdmsr
+    or eax, 1         
+    wrmsr
+
+    mov eax, cr0
+    or eax, (1 << 31)
+    mov cr0, eax
+
+    jmp 0x08:hal_ap_64bit_entry
+hal_ap_end:
+
+hal_ap_length:
+    dq hal_ap_end - hal_ap_start
+
+[bits 64]
+
+hal_ap_64bit_entry:
+    mov ax, 0x23     
+    mov ds, ax
+    mov es, ax
+    mov fs, ax
+    mov gs, ax
+    mov ss, ax
+    mov rsp, [hal_ap_64bit_entry_stack_end]
+
+    push 0x33                 
+    push qword [hal_ap_64bit_entry_loop]      
+    o64 pushf               
+    push rsp                  
+    push 0x33                 
+
+    o64 iret 
+
+hal_ap_64bit_entry_loop:
+    jmp $
+
+hal_ap_64bit_entry_stack:
+    resb 8196*2
+hal_ap_64bit_entry_stack_end:
\ No newline at end of file
diff --git a/dev/Kernel/HALKit/AMD64/HalCommonAPI.asm b/dev/Kernel/HALKit/AMD64/HalCommonAPI.asm
index 4a516d33..715f4bcf 100644
--- a/dev/Kernel/HALKit/AMD64/HalCommonAPI.asm
+++ b/dev/Kernel/HALKit/AMD64/HalCommonAPI.asm
@@ -80,62 +80,3 @@ mp_system_call_handler:
     o64 sysret
 
 [bits 16]
-
-section .text
-global hal_ap_start
-
-hal_ap_start:
-    mov ax, 0x0      
-    mov ss, ax
-    mov esp, 0x7000  
-
-    cli    
-    mov eax, cr0
-    or eax, 1        
-    mov cr0, eax
-    jmp .flush       
-.flush:
-    mov ax, 0x10     
-    mov ds, ax
-    mov es, ax
-    mov fs, ax
-    mov gs, ax
-    mov ss, ax
-
-    mov eax, cr4
-    or eax, 1 << 5   
-    mov cr4, eax
-
-    mov eax, cr3
-    mov cr3, eax     
-
-    mov ecx, 0xC0000080 
-    rdmsr
-    or eax, 1         
-    wrmsr
-
-    mov eax, cr0
-    or eax, (1 << 31)
-    mov cr0, eax
-
-    jmp 0x08:hal_ap_64bit_entry
-
-[bits 64]
-
-; 64-bit entry point
-section .text
-global hal_ap_64bit_entry
-hal_ap_64bit_entry:
-    mov ax, 0x10     
-    mov ds, ax
-    mov es, ax
-    mov fs, ax
-    mov gs, ax
-    mov ss, ax
-
-    mov rsp, 0x8001000 ; Stack Address for scheduler AP.
-
-    jmp hal_ap_64bit_entry_loop
-
-hal_ap_64bit_entry_loop:
-    jmp $
\ No newline at end of file
diff --git a/dev/Kernel/HALKit/AMD64/Processor.h b/dev/Kernel/HALKit/AMD64/Processor.h
index def26fa9..0db498ec 100644
--- a/dev/Kernel/HALKit/AMD64/Processor.h
+++ b/dev/Kernel/HALKit/AMD64/Processor.h
@@ -192,16 +192,6 @@ namespace Kernel::HAL
 	Void mp_get_cores(VoidPtr vendor_ptr) noexcept;
 
 	/***********************************************************************************/
-
-	Void hal_send_start_ipi(UInt32 apicId, UInt8 vector, UInt32 targetAddress);
-
-	/***********************************************************************************/
-
-	Void hal_send_end_ipi(UInt32 apicId, UInt8 vector, UInt32 targetAddress);
-
-	/***********************************************************************************/
-
-	/***********************************************************************************/
 	/// @brief Do a cpuid to check if MSR exists on CPU.
 	/// @retval true it does exists.
 	/// @retval false it doesn't.
diff --git a/dev/Kernel/HALKit/AMD64/make_ap_blob.sh b/dev/Kernel/HALKit/AMD64/make_ap_blob.sh
new file mode 100755
index 00000000..3f079187
--- /dev/null
+++ b/dev/Kernel/HALKit/AMD64/make_ap_blob.sh
@@ -0,0 +1,3 @@
+# !/bin/sh
+
+nasm -f bin HalApplicationProcessorStartup.asm -o HalApplicationProcessorStartup.bin
\ No newline at end of file