fix bug: sys_mmap didn't zero clean the anonymous memory

We can use delay mode or eager mode(in added comment codes)
WRJ's great work!

crate/memory/src/memory_set/handler/delay.rs

@@ -44,6 +44,11 @@ impl<T: FrameAllocator> MemoryHandler for Delay<T> {
         entry.set_target(frame);
         entry.set_present(true);
         entry.update();
+        //init with zero for delay mmap mode
+        let data = pt.get_page_slice_mut(addr);
+        for x in data {
+            *x = 0;
+        }
         true
     }
 }

kernel/src/memory.rs

@@ -114,7 +114,7 @@ impl Drop for KernelStack {
 /// Handle page fault at `addr`.
 /// Return true to continue, false to halt.
 pub fn handle_page_fault(addr: usize) -> bool {
-    // debug!("page fault @ {:#x}", addr);
+    debug!("page fault @ {:#x}", addr);
 
     // This is safe as long as page fault never happens in page fault handler
     unsafe { process_unsafe().vm.handle_page_fault(addr) }

kernel/src/syscall/mem.rs

@@ -46,9 +46,15 @@ pub fn sys_mmap(
             addr,
             addr + len,
             prot.to_attr(),
+//            ByFrame::new(GlobalFrameAlloc),   //eagle mmap mode
             Delay::new(GlobalFrameAlloc),
             "mmap_anon",
         );
+        //init with zero for eagle mmap mode
+//        let data = unsafe { slice::from_raw_parts_mut(addr as *mut u8, len) };
+//        for x in data {
+//            *x = 0;
+//        }
         return Ok(addr);
     } else {
         // only check