#include <kernel/kernel.h>
#include <kernel/multiboot.h>
#include <kernel/mem.h>
#include <kernel/serial.h>
#include <kernel/idt.h>
#include <kernel/pic.h>
#include <kernel/pit.h>
#include <kernel/elf.h>
#include <kernel/task.h>
#include <kernel/ps2.h>
#include <kernel/acpi.h>
#include <kernel/pci.h>
#include <kernel/vga.h>
#include <stdint.h>

kernel_ctx_t ctx = {0};
struct idt_entry idt[256] = {0};
struct idtr idtr = {0};
tss_t tss = {0};

extern uint32_t isr_stub_table[32];
extern void pit_isr(void);
extern void ps2_isr(void);
extern void jmp_user_mode(uint32_t esp, uint32_t eip);
extern void flush_tss(uint32_t segment_selector);
extern void task_switch(task_t *task);
extern void syscall(void);
extern void syscall_isr(void);
extern gdt_t gdt[6];
extern uint32_t boot_page_dir[1024];
extern uint32_t boot_page_tab[1024];

#define FD_STDIN  0
#define FD_STDOUT 1
void syscall_write(stack_frame_t r) {
    uint32_t fd  = r.edi;
    uint8_t *buf = (uint8_t *)r.esi;
    int len = r.edx;
    if (fd == FD_STDOUT) {
        for (int i = 0; i < len; i++) {
            vga_putc(buf[i]);
        }
    } 
}

void syscall_read(stack_frame_t r) {
    uint32_t fd  = r.edi;
    uint8_t *buf = (uint8_t *)r.esi;
    uint32_t len = r.edx;
    if (fd == FD_STDOUT) {
        // Not yet
    } else if (fd == FD_STDIN) {
        // for some reason interrupts are disabled
        asm volatile ("sti" ::);
        while (ctx.current_task->stdin[ctx.current_task->stdin_len - 1] != '\n') cpu_relax;
        if (ctx.current_task->stdin_len > 0) ctx.current_task->stdin[--ctx.current_task->stdin_len] = 0;
        memcpy(buf, ctx.current_task->stdin, len);
        for (int i = 0; i < 0x1000; i++) ctx.current_task->stdin[i] = 0;
        ctx.current_task->stdin_len = 0;
    }
}

void syscall_reboot(stack_frame_t r) { 
    (void)r;
    ps2_reboot;
}

void syscall_uptime(stack_frame_t r) {
    uint32_t *low = (uint32_t *)r.esi;
    uint32_t *high = (uint32_t *)r.edi;
    *low = (uint32_t)ctx.ticks;
    *high = (uint32_t)(ctx.ticks >> 32);
}

void exception_handler(int_stack_frame_t r) {
    LOG("Exception 0x%X (0x%X) has occurred\nEAX: 0x%08X\nECX: 0x%08X\nEDX: 0x%08X\nEBX: 0x%08X\nESI: 0x%08X\nEDI: 0x%08X\nEBP: 0x%08X\nESP: 0x%08X\n",
            r.int_no, r.err_code, r.eax, r.ecx, r.edx, r.ebx, r.esi, r.edi, r.ebp, r.esp+24); 
    LOG("\nEIP:     0x%08X\nCS:      0x%08X\nEFLAGS:  0x%08X\n",
            r.eip, r.cs, r.eflags);

    // Page fault
    if (r.int_no == 0xe) {
        uint32_t cr2 = 0;
        asm volatile ("mov %%cr2, %0" : "=r" (cr2));
        LOG("CR2:     0x%08X\n", cr2)

        uint32_t cr3 = 0;
        asm volatile ("mov %%cr3, %0" : "=r" (cr3));
        uint32_t *page_dir = (uint32_t *)(KERNEL_VMA + cr3);

        if ((page_dir[PAGE_DIR_INDEX(cr2)] & 0xFFFFF000) == 0) {
            LOG("Creating new page table\n");
            uint32_t *page_tab = mmap_find_first_free_block();
            if (!page_tab) for (;;) {}
            page_tab[PAGE_TAB_INDEX(cr2)] = (cr2 & 0xFFFFF000) | 7;
            page_dir[PAGE_DIR_INDEX(cr2)] = ((uint32_t)page_tab - KERNEL_VMA) | 7;
        } else {
            uint32_t *page_tab = (uint32_t *)(KERNEL_VMA + (page_dir[PAGE_DIR_INDEX(cr2)] & 0xFFFFF000)); 
            page_tab[PAGE_TAB_INDEX(cr2)] = (cr2 & 0xFFFFF000) | 7;
        }
    }
}

void sleep(int delay) {
    uint64_t end = ctx.ticks + delay;
    while (ctx.ticks < end)
        cpu_relax;
}

void idt_init(void) {
    for (int i = 0; i < 32; i++) {
        idt_encode_entry(idt, i, isr_stub_table[i], GDT_SEGMENT_SELECTOR(GDT_KERNEL_CODE, RING0), TRAP_GATE_32 | INT_RING0 | INT_PRESENT);
    }
    idt_encode_entry(idt, 0x20, (uint32_t)pit_isr, GDT_SEGMENT_SELECTOR(GDT_KERNEL_CODE, RING0), TRAP_GATE_32 | INT_RING0 | INT_PRESENT);
    idt_encode_entry(idt, 0x21, (uint32_t)ps2_isr, GDT_SEGMENT_SELECTOR(GDT_KERNEL_CODE, RING0), TRAP_GATE_32 | INT_RING0 | INT_PRESENT);
    idt_encode_entry(idt, 0x80, (uint32_t)syscall_isr, GDT_SEGMENT_SELECTOR(GDT_USER_CODE, RING3), TRAP_GATE_32 | INT_RING3 | INT_PRESENT);

    idtr.size = (sizeof(struct idt_entry) * 256) - 1;
    idtr.base = (uint32_t) idt;

    asm volatile ("lidt %0" :: "m"(idtr));
}


void kernel(multiboot_info_t *info) {
    if (!CHECK_FLAG(info->flags, 3))  goto halt;   // Modules
    if (!CHECK_FLAG(info->flags, 6))  goto halt;   // Memory map
    if (!CHECK_FLAG(info->flags, 12)) goto halt;   // VBE data

    if (serial_port_init(COM1)) ctx.log_method = LOG_COM1;
    LOG("Kernel log being sent to COM1\n");

    idt_init();

    ctx.multi_mmap = (multi_mmap_t *)(info->memmapaddress + KERNEL_VMA);
    ctx.multi_mmap_size = info->memmaplength;
    if (!mmap_init()) goto halt;
    LOG("%d bytes of RAM detected\nCreated a %d byte large physical memory map at 0x%08X\n", ctx.mmap_size*BLOCK_SIZE*8, ctx.mmap_size, (uint32_t)ctx.mmap);

    clear_screen();
    acpi_init();

    for (int i = 0; i < 256; i++) {
        for (int j = 0; j < 32; j++) {
            for (int k = 0; k < 8; k++) {
                uint16_t w = pci_read_cfg(i, j, k, 0xA);
                uint8_t class = (uint8_t)(w >> 8);
                uint8_t subclass = (uint8_t)w;
                if (class == 0xFF) continue;
                LOG("PCI Class: %04X | Subclass: %04X\n", class, subclass);
            }
        }
    }

    pic_remap(PIC_1_START, PIC_2_START);
    ps2_init();
    asm volatile ("sti" ::);
    ctx.ticks = 0;
    pit_init();
    
    // Setup TSS
    uint32_t base = (uint32_t)&tss;
    uint32_t limit = base + sizeof(tss_t);
    gdt[5].base_low = base & 0xFFFF;
    gdt[5].base_middle = (base >> 16) & 0xFF;
    gdt[5].base_high = (base >> 24) & 0xFF;
    gdt[5].limit_low = limit & 0xFFFF;
    gdt[5].granularity = (limit >> 16) & 0x0F;
    gdt[5].granularity |= 0x40;
    gdt[5].access = 0x89;
    tss.ss0 = GDT_SEGMENT_SELECTOR(GDT_KERNEL_DATA, RING0);
    uint32_t esp0 = 0;
    asm volatile ("mov %%esp, %0" : "=r" (esp0));
    // Questionable
    tss.esp0 = esp0;
    tss.cs = GDT_SEGMENT_SELECTOR(GDT_USER_CODE, RING3);
    tss.ss = tss.ds = tss.es = tss.fs = tss.gs = GDT_SEGMENT_SELECTOR(GDT_USER_DATA, RING3);
    flush_tss(GDT_SEGMENT_SELECTOR(5, RING3));

    multi_mod_t *init = (multi_mod_t *)(info->moduleaddress + KERNEL_VMA);
    elf_t *elf = (elf_t *)(init->mod_start+KERNEL_VMA);
    task_t task = {0};
    if (!task_init(&task, elf)) goto halt;

    ctx.current_task = &task;
    asm volatile ("mov %0, %%cr3" :: "r"(task.cr3));

    // Set the ring0 CS
    asm volatile (
            "movl $0x174, %%ecx \n\t"
            "movl %0, %%eax \n\t"
            "xor %%edx, %%edx \n\t"
            "wrmsr"
            : 
            : "r"(GDT_SEGMENT_SELECTOR(GDT_KERNEL_CODE, RING0))
            : "eax", "ecx", "edx"
            );
    // Set the ring0 stack
    uint32_t *k_stack = mmap_find_first_free_block();
    asm volatile (
            "movl $0x175, %%ecx \n\t"   
            "movl %0, %%eax \n\t"       
            "xor %%edx, %%edx \n\t"     
            "wrmsr"
            : 
            : "r"((uint32_t)k_stack + 0x1000)
            : "eax", "ecx", "edx"
            );
    // Set the ring0 syscall entry
    asm volatile (
            "movl $0x176, %%ecx \n\t"   
            "movl %0, %%eax \n\t"      
            "xor %%edx, %%edx \n\t"   
            "wrmsr"
            : 
            : "r"(syscall)
            : "eax", "ecx", "edx"
            );

    LOG("Switching to ring 3... EIP: 0x%08X ESP: 0x%08X\n", task.eip, task.esp);
    jmp_user_mode(task.esp, task.eip); 

halt:
    for (;;) {}
}