feat: implement fork system call with deep address space cloning (AI)

- Added paging_clone_directory_from(): deep-copies user-space pages so
  parent and child have independent memory. Kernel pages are shared.
- Fixed process_fork() to accept registers_t* for accurate child state,
  and to clone from the parent's page directory (not the kernel's).
- Refactored process_exit() to properly context-switch to next process
  using new process_switch_to_user assembly stub (loads full registers_t
  and performs iret), instead of halting unconditionally.
- Fixed sys_waitpid() to use proper blocking: marks process BLOCKED,
  invokes scheduler, and resumes with exit code when child dies.
- Added SYSCALL_SWITCHED mechanism to prevent syscall_handler from
  clobbering the next process's EAX after a context switch.
- Created fork-test user app that validates fork + waitpid.
- Added docs/fork.md with architecture documentation.

Tested: fork-test creates child, both print messages, parent waits for
child exit (code 7), parent reaps and exits (code 0). hello-world also
verified to still work correctly after the process_exit refactor.

src/paging.c

@@ -299,6 +299,64 @@ uint32_t paging_clone_directory(void) {
     return new_dir_phys;
 }
 
+uint32_t paging_clone_directory_from(uint32_t src_pd_phys) {
+    uint32_t *src_pd = (uint32_t *)src_pd_phys;
+
+    /* Allocate a new page directory */
+    phys_addr_t new_pd_phys = pmm_alloc_page(PMM_ZONE_NORMAL);
+    if (new_pd_phys == 0) {
+        offset_print("  PAGING: cannot allocate page directory for fork\n");
+        return 0;
+    }
+    uint32_t *new_pd = (uint32_t *)new_pd_phys;
+
+    /* Copy all page directory entries (shares kernel mappings) */
+    memcpy(new_pd, src_pd, 4096);
+
+    /* Deep-copy user-space page tables (those with PAGE_USER set) */
+    for (uint32_t i = 0; i < PAGE_ENTRIES; i++) {
+        if (!(src_pd[i] & PAGE_PRESENT)) continue;
+        if (!(src_pd[i] & PAGE_USER))    continue; /* kernel entry, shared */
+
+        uint32_t *src_pt = (uint32_t *)(src_pd[i] & 0xFFFFF000);
+
+        /* Allocate a new page table */
+        phys_addr_t new_pt_phys = pmm_alloc_page(PMM_ZONE_NORMAL);
+        if (new_pt_phys == 0) {
+            offset_print("  PAGING: fork: cannot allocate page table\n");
+            return 0; /* TODO: free partially allocated pages */
+        }
+        uint32_t *new_pt = (uint32_t *)new_pt_phys;
+
+        /* Deep-copy each page in the page table */
+        for (uint32_t j = 0; j < PAGE_ENTRIES; j++) {
+            if (!(src_pt[j] & PAGE_PRESENT)) {
+                new_pt[j] = 0;
+                continue;
+            }
+
+            if (src_pt[j] & PAGE_USER) {
+                /* User page: allocate new physical page and copy content */
+                phys_addr_t old_phys = src_pt[j] & 0xFFFFF000;
+                phys_addr_t new_phys = pmm_alloc_page(PMM_ZONE_NORMAL);
+                if (new_phys == 0) {
+                    offset_print("  PAGING: fork: cannot allocate page\n");
+                    return 0;
+                }
+                memcpy((void *)new_phys, (void *)old_phys, 4096);
+                new_pt[j] = new_phys | (src_pt[j] & 0xFFF);
+            } else {
+                /* Kernel page within a user page table: share directly */
+                new_pt[j] = src_pt[j];
+            }
+        }
+
+        new_pd[i] = new_pt_phys | (src_pd[i] & 0xFFF);
+    }
+
+    return new_pd_phys;
+}
+
 void paging_map_page_in(uint32_t *pd, uint32_t vaddr, uint32_t paddr, uint32_t flags) {
     uint32_t pd_idx = PD_INDEX(vaddr);
     uint32_t pt_idx = PT_INDEX(vaddr);