Kernel: Implement software context switching and Processor structure

Moving certain globals into a new Processor structure for
each CPU allows us to eventually run an instance of the
scheduler on each CPU.

Kernel/Arch/i386/Boot/boot.S

@@ -252,15 +252,6 @@ apic_ap_start:
  mov %cs, %ax
  mov %ax, %ds
 
- /* Generate a new processor id. This is not the APIC id. We just
- need a way to find ourselves a stack without stomping on other
- APs that may be doing this concurrently. */
- xor %ax, %ax
- mov %ax, %bp
- inc %ax
- lock; xaddw %ax, %ds:(ap_cpu_id - apic_ap_start)(%bp) /* avoid relocation entries */
- mov %ax, %bx
-
  xor %ax, %ax
  mov %ax, %sp
 
@@ -281,14 +272,18 @@ apic_ap_start32:
  mov %ax, %es
  mov %ax, %fs
  mov %ax, %gs
- 
+
  movl $0x8000, %ebp
-
+
+ /* generate a unique ap cpu id (0 means 1st ap, not bsp!) */
+ xorl %eax, %eax
+ incl %eax
+ lock; xaddl %eax, (ap_cpu_id - apic_ap_start)(%ebp) /* avoid relocation entries */
+ movl %eax, %esi
+
  /* find our allocated stack based on the generated id */
- andl 0x0000FFFF, %ebx
- movl %ebx, %esi
- movl (ap_cpu_init_stacks - apic_ap_start)(%ebp, %ebx, 4), %esp
-
+ movl (ap_cpu_init_stacks - apic_ap_start)(%ebp, %eax, 4), %esp
+
  /* check if we support NX and enable it if we do */
  movl $0x80000001, %eax
  cpuid
@@ -319,8 +314,8 @@ apic_ap_start32:
  lgdt (ap_cpu_gdtr_initial2 - apic_ap_start + 0xc0008000)
 
  /* jump above 3GB into our identity mapped area now */
- ljmp $8, $(1f - apic_ap_start + 0xc0008000)
-1:
+ ljmp $8, $(apic_ap_start32_2 - apic_ap_start + 0xc0008000)
+apic_ap_start32_2:
  /* flush the TLB */
  movl %cr3, %eax
  movl %eax, %cr3
@@ -338,13 +333,20 @@ apic_ap_start32:
  movl %eax, %cr0
  movl (ap_cpu_init_cr4 - apic_ap_start)(%ebp), %eax
  movl %eax, %cr4
-
+
+ /* push the Processor pointer this CPU is going to use */
+ movl (ap_cpu_init_processor_info_array - apic_ap_start)(%ebp), %eax
+ addl $0xc0000000, %eax
+ movl 0(%eax, %esi, 4), %eax
+ push %eax
+
+ /* push the cpu id, 0 representing the bsp and call into c++ */
+ incl %esi
+ push %esi
+
  xor %ebp, %ebp
  cld
 
- /* push the arbitrary cpu id, 0 representing the bsp and call into c++ */
- inc %esi
- push %esi
  /* We are in identity mapped P0x8000 and the BSP will unload this code
  once all APs are initialized, so call init_ap but return to our
  infinite loop */
@@ -356,7 +358,7 @@ apic_ap_start32:
 apic_ap_start_size:
  .2byte end_apic_ap_start - apic_ap_start
 ap_cpu_id:
- .2byte 0x0
+ .4byte 0x0
 ap_cpu_gdt:
  /* null */
  .8byte 0x0
@@ -388,6 +390,9 @@ ap_cpu_init_cr3:
 .global ap_cpu_init_cr4
 ap_cpu_init_cr4:
  .4byte 0x0 /* will be set at runtime */
+.global ap_cpu_init_processor_info_array
+ap_cpu_init_processor_info_array:
+ .4byte 0x0 /* will be set at runtime */
 .global ap_cpu_init_stacks
 ap_cpu_init_stacks:
  /* array of allocated stack pointers */