sys: Refactor GDT initialization, enhance syscall handling, and improve user mode entry

Finally we got usermode working again, It took awhile but we got it done. A few things we did in this commit:

- Simplified GDT entry initialization in gdt.c for kernel and user segments.
- Fixed TSS structure for task switching.
- Implemented cr2 handling in a page fault.
- Enhanced user stack setup in usermode.c to return the correct RSP.
- Improved syscall implementation in syscall.c, including new syscall numbers.
- Updated syscall entry to correctly handle context switching and argument passing.
- Refactored init.c to demonstrate file operations using syscalls.
- Added new syscalls for file operations in syscalls.h.
- Modified VFS to handle leading slashes in paths correctly.

Signed-off-by: kaguya <vpshinomiya@protonmail.com>

src/syscall/syscall.c

@@ -1,54 +1,72 @@
 #include <stdint.h>
 #include "stdio.h"
+#include "mp/percpu.h"
 #include "fs/vfs.h"
+#include "syscall.h"
 
-#define MSR_EFER   0xC0000080
-#define MSR_STAR   0xC0000081
-#define MSR_LSTAR  0xC0000082
-#define MSR_SFMASK 0xC0000084
+#define MSR_EFER          0xC0000080
+#define MSR_STAR          0xC0000081
+#define MSR_LSTAR         0xC0000082
+#define MSR_SFMASK        0xC0000084
+#define MSR_KERNEL_GSBASE 0xC0000102
 
 #define EFER_SCE (1 << 0)
 
+static struct cpu_local g_cpu_local;
+
+static uint8_t g_syscall_kstack[16384] __attribute__((aligned(16)));
+
 static inline void wrmsr(uint32_t msr, uint64_t value)
 {
-    uint32_t lo = value & 0xFFFFFFFF;
-    uint32_t hi = value >> 32;
-
-    asm volatile("wrmsr"
-                 :
-                 : "c"(msr), "a"(lo), "d"(hi));
+    asm volatile("wrmsr" :: "c"(msr), "a"((uint32_t)value), "d"((uint32_t)(value >> 32)));
 }
 
 static inline uint64_t rdmsr(uint32_t msr)
 {
     uint32_t lo, hi;
-    asm volatile("rdmsr"
-                 : "=a"(lo), "=d"(hi)
-                 : "c"(msr));
+    asm volatile("rdmsr" : "=a"(lo), "=d"(hi) : "c"(msr));
     return ((uint64_t)hi << 32) | lo;
 }
 
-void syscall_handler(uint64_t rax, // syscall num
-                     uint64_t rdi, // arg1
-                     uint64_t rsi, // arg2
-                     uint64_t rdx, // arg3
-                     uint64_t r10, // arg4
-                     uint64_t r8,  // arg5
-                     uint64_t r9)  // arg6
+uint64_t syscall_handler(uint64_t num,
+                     uint64_t arg1, uint64_t arg2, uint64_t arg3,
+                     uint64_t arg4, uint64_t arg5, uint64_t arg6)
 {
-    switch (rax)
-    {
-        case 1: // write (int fd, const void *buf, size_t count);
-            int c = (char)rdi;
-            printf("%c", c);
-            break;
+    (void)arg4; (void)arg5; (void)arg6;
+    switch (num)
+    {   
+        case SYS_READ:
+        {
+            int fd          = (int)arg1;
+            uint8_t* buf    = (uint8_t*)arg2;
+            size_t len      = (size_t)arg3;
 
-        case 2:
-            // exit
-            break;
+            return (uint64_t)VFS_Read(fd, buf, len);
+        }
+
+        case SYS_WRITE:
+        {
+            int fd          = (int)arg1;
+            uint8_t* buf    = (uint8_t*)arg2;
+            size_t len      = (size_t)arg3;
+
+            return (uint64_t)VFS_Write(fd, buf, len);
+        }
+
+        case SYS_OPEN:
+        {
+            const char* path = (const char*)arg1;
+            return (uint64_t)VFS_Open(path);
+        }
+        
+        case SYS_CLOSE:
+        {
+            int fd = (int)arg1;
+            return (uint64_t)VFS_Close(fd);
+        }
 
         default:
-            break;
+            return (uint64_t)-1;
     }
 }
 
@@ -56,21 +74,26 @@ void syscall_init(void)
 {
     extern void syscall_entry(void);
 
-    uint64_t kernel_cs = 0x08;
-    uint64_t user_cs   = 0x23;
+    // ── per-CPU local block ─────────────────────────────────────────────
+    g_cpu_local.self       = (uint64_t)&g_cpu_local;
+    g_cpu_local.kernel_rsp = (uint64_t)(g_syscall_kstack + sizeof(g_syscall_kstack));
 
-    uint64_t star =
-        ((uint64_t)kernel_cs << 32) |
-        ((uint64_t)user_cs << 48);
+    // After swapgs in syscall_entry, GS points here.
+    wrmsr(MSR_KERNEL_GSBASE, (uint64_t)&g_cpu_local);
 
+    // ── STAR ────────────────────────────────────────────────────────────
+    // Bits [47:32]: SYSCALL loads CS = 0x08 (kernel code), SS = 0x10 (kernel data)
+    // Bits [63:48]: SYSRET  loads CS = (base+16)|3 = 0x23, SS = (base+8)|3 = 0x1B
+    //               → base must be 0x10
+    uint64_t star = ((uint64_t)0x08 << 32) | ((uint64_t)0x10 << 48);
     wrmsr(MSR_STAR, star);
 
     wrmsr(MSR_LSTAR, (uint64_t)syscall_entry);
 
     uint64_t efer = rdmsr(MSR_EFER);
-    efer |= (1 << 0); // SYSCALL/SYSRET enable
+    efer |= EFER_SCE;
     wrmsr(MSR_EFER, efer);
 
-    // disable interrupts on syscall entry
+    // Mask IF so interrupts are off during the syscall handler
     wrmsr(MSR_SFMASK, (1 << 9));
-}
+}