#include <stddef.h>
#include <stdint.h>
#include "vmm/vmm.h"
#include "bootinfo/bootinfo.h"
#include "pmm/pmm.h"
#include "proc/proc.h"
#include "spinlock/spinlock.h"
#include "std/string.h"
#include "kprintf.h"

uint64_t KERNEL_CR3 = 0;
SpinLock spinlock;

uint64_t vmm_current_cr3(void) {
  uint64_t cr3;
  asm volatile("mov %%cr3, %0" : "=r"(cr3));
  return cr3;
}

PgIndex vmm_pageindex(uint64_t vaddr) {
  PgIndex ret;

  ret.pml4 = (vaddr >> 39) & 0x1ff;
  ret.pml3 = (vaddr >> 30) & 0x1ff;
  ret.pml2 = (vaddr >> 21) & 0x1ff;
  ret.pml1 = (vaddr >> 12) & 0x1ff;

  return ret;
}

uint64_t *vmm_nexttable(uint64_t *table, uint64_t ent, bool alloc) {
  uint64_t entry = table[ent];
  uint64_t phys;
  if (entry & VMM_PG_PRESENT) {
    phys = entry & ~0xFFFULL;
  } else {
    if (!alloc) {
      return NULL;
    }

    uint8_t *newphys = pmm_alloc(1);
    phys = (uint64_t)newphys;
    memset(VIRT(phys), 0, VMM_PAGE_SIZE);
    table[ent] = phys | VMM_PG_USER | VMM_PG_RW | VMM_PG_PRESENT;
  }
  return (uint64_t *)((uint8_t *)VIRT(phys));
}

void vmm_map_page(uint64_t cr3phys, uint64_t virtaddr, uint64_t physaddr, uint32_t flags) {
  uint64_t *pml4 = (uint64_t *)VIRT(cr3phys);
  PgIndex pi = vmm_pageindex(virtaddr);

  uint64_t *pml3 = vmm_nexttable(pml4, pi.pml4, true);
  uint64_t *pml2 = vmm_nexttable(pml3, pi.pml3, true);
  uint64_t *pml1 = vmm_nexttable(pml2, pi.pml2, true);
  uint64_t *pte = &pml1[pi.pml1];

  *pte = (physaddr & ~0xFFFULL) | ((uint64_t)flags & 0x7ULL);
}

void vmm_unmap_page(uint64_t cr3phys, uint64_t virtaddr) {
  uint64_t *pml4 = (uint64_t *)VIRT(cr3phys);
  PgIndex pi = vmm_pageindex(virtaddr);
  
  uint64_t *pml3 = vmm_nexttable(pml4, pi.pml4, false);
  uint64_t *pml2 = vmm_nexttable(pml3, pi.pml3, false);
  uint64_t *pml1 = vmm_nexttable(pml2, pi.pml2, false);
  uint64_t *pte = &pml1[pi.pml1];

  *pte &= ~VMM_PG_PRESENT;
}

void vmm_map_range(uint64_t cr3phys, void *virtstart, void *physstart, size_t size, uint32_t flags) {
  if (size % VMM_PAGE_SIZE != 0 || (uint64_t)virtstart % VMM_PAGE_SIZE != 0 || (uint64_t)physstart % VMM_PAGE_SIZE != 0) {
    return;
  }

  spinlock_acquire(&spinlock);
  uint8_t *vaddr = (uint8_t *)virtstart;
  uint8_t *paddr = (uint8_t *)physstart;
  uint8_t *end = (uint8_t *)virtstart + size;
  for (; vaddr < end; vaddr += VMM_PAGE_SIZE, paddr += VMM_PAGE_SIZE) {
    vmm_map_page(cr3phys, (uint64_t)vaddr, (uint64_t)paddr, flags);
  }
  spinlock_release(&spinlock);
}

void vmm_unmap_range(uint64_t cr3phys, void *virtstart, void *physstart, size_t size) {
  if (size % VMM_PAGE_SIZE != 0 || (uint64_t)virtstart % VMM_PAGE_SIZE != 0 || (uint64_t)physstart % VMM_PAGE_SIZE != 0) {
    return;
  }

  spinlock_acquire(&spinlock);
  uint8_t *vaddr = (uint8_t *)virtstart;
  uint8_t *end = vaddr + size;

  for (; vaddr < end; vaddr += VMM_PAGE_SIZE) {
    vmm_unmap_page(cr3phys, (uint64_t)vaddr);
  }
  spinlock_release(&spinlock);
}

void vmm_map_kern(uint64_t targetcr3) {
  uint64_t *kcr3 = (uint64_t *)VIRT(KERNEL_CR3);
  uint64_t *cr3 = (uint64_t *)VIRT(targetcr3);
  for (size_t i = 0; i < 512; i++) {
    cr3[i] = kcr3[i];
  }
}

uint64_t vmm_userproc_pml4_phys(void) {
  uint8_t *cr3phys = pmm_alloc(1);
  uint64_t phys = (uint64_t)cr3phys;
  memset(VIRT(phys), 0, VMM_PAGE_SIZE);

  uint64_t *kcr3 = (uint64_t *)VIRT(KERNEL_CR3);
  uint64_t *pml4 = (uint64_t *)VIRT(phys);
  for (size_t i = 256; i < 512; i++) {
    pml4[i] = kcr3[i];
  }
  return phys;
}

void vmm_init(void) {
  spinlock_init(&spinlock);
  KERNEL_CR3 = vmm_current_cr3();
}