cleaned up some task details

src/arch/x86/gdt.zig

@@ -107,22 +107,11 @@ pub fn setKernelStack(esp0: usize) void {
     tss.esp0 = esp0;
 }
 
-////
-// Load the GDT into the system registers (defined in assembly).
-//
-// Arguments:
-//     gdtr: Pointer to the GDTR.
-//
 extern fn loadGDT(gdtr: *const GDTRegister) void;
 
-////
-// Initialize the Global Descriptor Table.
-//
 pub fn initialize() void {
-    // tty.step("Setting up the Global Descriptor Table");
-
     // Initialize GDT.
-    loadGDT(&gdtr);
+    loadGDT(&gdtr); //asm routine
 
     // Initialize TSS.
     // const tss_entry = makeEntry(@ptrToInt(&tss), @sizeOf(TSS) - 1, TSS_ACCESS, PROTECTED);

src/arch/x86/paging.zig

@@ -20,13 +20,14 @@ fn pageFault() void {
     while (true) asm volatile ("hlt");
 }
 
-inline fn pageBase(virt: usize) usize {
-    return addr & (~PAGE_SIZE +% 1);
+// TODO: inline these
+fn pageBase(virt: usize) usize {
+    return virt & (~PAGE_SIZE +% 1);
 }
-inline fn pde(virt: usize) *PageEntry {
+fn pde(virt: usize) *PageEntry {
     return &PD[virt >> 22]; //relies on recursive mapping
 }
-inline fn pte(virt: usize) *PageEntry {
+fn pte(virt: usize) *PageEntry {
     return &PT[virt >> 12]; //relies on recursive mapping
 }
 
@@ -38,7 +39,7 @@ pub fn translate(virt: usize) ?usize {
 
 pub fn unmap(virt: usize) void {
     if (translate(virt)) |phys| {
-        mem.free(phys);
+        pmem.free(phys);
     } else {
         kernel.println("can't unmap 0x{x} because it is not mapped.", virt);
     }
@@ -64,7 +65,7 @@ pub fn initialize() void {
     assert(pmem.stack_end < kernel.layout.IDENTITY);
 
     interrupt.register(14, pageFault);
-    setupPaging(@ptrToInt(&pageDirectory[0]));
+    setupPaging(@ptrToInt(&pageDirectory[0])); //asm routine
 }
 
 pub fn introspect() void {

src/console.zig

@@ -3,14 +3,15 @@ usingnamespace @import("index.zig");
 var command: [10]u8 = undefined;
 var command_len: usize = 0;
 
-fn execute(com: []u8) void {
+fn execute(input: []u8) void {
     const eql = std.mem.eql;
-    if (eql(u8, com, "x86paging")) return x86.paging.introspect();
-    if (eql(u8, com, "x86memory")) return x86.pmem.introspect();
-    if (eql(u8, com, "lspci")) return pci.lspci();
-    if (eql(u8, com, "uptime")) return time.uptime();
-    if (eql(u8, com, "topbar")) return topbar();
-    println("{}: command not found", com);
+    if (eql(u8, input, "x86paging")) return x86.paging.introspect();
+    if (eql(u8, input, "x86memory")) return x86.pmem.introspect();
+    if (eql(u8, input, "tasks")) return task.introspect();
+    if (eql(u8, input, "lspci")) return pci.lspci();
+    if (eql(u8, input, "uptime")) return time.uptime();
+    if (eql(u8, input, "topbar")) return topbar();
+    println("{}: command not found", input);
 }
 
 pub fn keypress(char: u8) void {

src/layout.zig

@@ -1,8 +1,8 @@
 //https://wiki.osdev.org/Memory_Map_(x86)
 // virtual memory layout of the kernel
 
-const kiB = 1024; // bytes
-const MiB = 1024 * kiB; // 0x100000
+const kiB = 1024;
+const MiB = 1024 * kiB;
 const GiB = 1024 * MiB;
 
 // zig fmt: off
@@ -11,7 +11,7 @@ pub const KERNEL            = 1 * MiB;
 pub const IDENTITY          = 4 * MiB; // 0->4MiB
 
 pub const HEAP              = 8 * MiB;
-pub const HEAP_END          = 0x1000000;
-pub const USER_STACKS       = 0x1000000;
-pub const USER_STACKS_END   = 0x10000000;
+pub const HEAP_END          = 0x01000000;
+pub const USER_STACKS       = 0x01000000;
+pub const USER_STACKS_END   = 0x02000000;
 // zig fmt: on

src/main.zig

@@ -24,17 +24,9 @@ export fn kmain(magic: u32, info: *const multiboot.MultibootInfo) noreturn {
     println("--- x86 initialization ---");
     x86.x86_main(info);
     println("--- core initialization ---");
-    pci.scan();
+    // pci.scan();
     vmem.initialize();
-
-    // var a = vmem.allocate(u32) catch unreachable;
-    // println("a={}", &a);
-    // const b = vmem.allocate(VGAEntry) catch unreachable;
-    // println("b={x}", &b);
-
-    const t = task.Task.new(@ptrToInt(topbar)) catch unreachable;
-    println("task={x}", &t);
-
+    task.initialize() catch unreachable;
     console.initialize();
 
     while (true) asm volatile ("hlt");

src/task.zig

@@ -1,42 +1,52 @@
 pub usingnamespace @import("index.zig");
-// var tasks = Array(?*Task).init(&mem.allocator);
+const TASK_MAX = 1024;
+var tasks = [1]?*Task{null} ** TASK_MAX;
 
 const STACK_SIZE = x86.PAGE_SIZE; // Size of thread stacks.
+var tid_counter: u16 = 1;
 
 pub const Task = struct {
-    // context: isr.Context,
-
-    ////
-    // Create a new thread inside the current process.
-    // NOTE: Do not call this function directly. Use Process.createThread instead.
-    //
-    // Arguments:
-    //     entry_point: The entry point of the new thread.
-    //
-    // Returns:
-    //     Pointer to the new thread structure.
-    //
     tid: u16,
+    stack_top: usize,
+    entrypoint: usize,
+    // context: isr.Context,
+    //cr3: usize,
 
-    pub fn stack(tid: u16) usize {
-        const stack = layout.USER_STACKS + (2 * (tid - 1) * STACK_SIZE);
-        assert(stack < layout.USER_STACKS_END);
-        return stack;
+    pub fn new(entrypoint: usize) !*Task {
+        // Allocate and initialize the thread structure.
+        var t = try vmem.allocate(Task);
+
+        t.entrypoint = entrypoint;
+        t.tid = tid_counter;
+        tid_counter +%= 1;
+        assert(tid_counter != 0); //overflow
+
+        t.stack_top = try vmem.malloc(STACK_SIZE);
+        assert(t.stack_top < layout.USER_STACKS_END);
+
+        tasks[t.tid] = t;
+        return t;
     }
 
-    pub fn new(entry_point: usize) !*Task {
-        // assert(scheduler.current_process == process);
-
-        // map the stack
-
-        // Allocate and initialize the thread structure.
-        var this = try vmem.allocate(Task);
-        this.tid = 4;
-
-        return this;
+    pub fn destroy(self: *Task) void {
+        tasks[self.tid] = null;
+        vmem.free(self.stack_top);
+        vmem.free(@ptrToInt(self));
     }
 };
 
+pub fn initialize() !void {
+    const t = try Task.new(0x0);
+    println("task=0x{x}", t.stack_top);
+}
+
+pub fn introspect() void {
+    for (tasks) |t| {
+        if (t == null) continue;
+        println("{}", t);
+    }
+}
+
 // fn initContext(entry_point: usize, stack: usize) isr.Context {
 //     // Insert a trap return address to destroy the thread on return.
 //     var stack_top = @intToPtr(*usize, stack + STACK_SIZE - @sizeOf(usize));

src/vga.zig

@@ -10,7 +10,7 @@ pub var vga = VGA{
     .vram = @intToPtr([*]VGAEntry, 0xb8000)[0..0x4000],
     .cursor = 80 * 2,
     .foreground = Color.Black,
-    .background = Color.Brown,
+    .background = Color.Green,
 };
 
 // Color codes.

src/vmem.zig

@@ -14,13 +14,17 @@ pub fn available() usize {
     return stack_index * x86.PAGE_SIZE;
 }
 
-pub fn allocate(comptime T: type) !*T {
-    assert(@sizeOf(T) < x86.PAGE_SIZE); // this allocator only support 1:1 mapping
+pub fn malloc(size: usize) !usize {
     if (available() == 0) return error.OutOfMemory;
     stack_index -= 1;
     var vaddr: usize = stack[stack_index];
     try x86.paging.mmap(vaddr, null);
-    return @intToPtr(*T, vaddr);
+    return vaddr;
+}
+
+pub fn allocate(comptime T: type) !*T {
+    assert(@sizeOf(T) < x86.PAGE_SIZE); // this allocator only support 1:1 mapping
+    return @intToPtr(*T, try malloc(@sizeOf(T)));
 }
 
 pub fn free(address: usize) void {