fix(mv): 修复mv命令不能移动文件以覆盖已有文件的问题

[xboHodx]

• 在 rename 前检查目标路径是否存在文件并删除
• 改进 truncate_complete_page 函数，从页面回收器中移除当前页面，避免后续访问已释放的inode
• 在脏页回写函数中增加清理页面映射、清除脏位的逻辑
• lru脏页刷新时收集孤儿页面并处理
• rename_file_in_current_dir 和 move_to_another_dir函数中增加在缓存中更新 inode 的名称和父目录引用的逻辑

[Copilot]

Pull Request Overview

This PR fixes a critical issue with the mv command not being able to move files to overwrite existing files. The main changes focus on improving file system consistency and memory management when handling file operations.

• Adds file overwrite functionality by checking and deleting target files before rename operations
• Improves memory management by removing pages from the page reclaimer when truncating and handling orphaned pages
• Updates inode metadata (name and parent directory references) in the cache during file moves

Reviewed Changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 5 comments.

File	Description
kernel/src/mm/truncate.rs	Removes pages from page reclaimer to prevent access to freed inodes
kernel/src/mm/page.rs	Refactors page writeback with improved error handling and orphaned page cleanup
kernel/src/filesystem/vfs/syscall/rename_utils.rs	Adds target file deletion before rename to enable overwrite functionality
kernel/src/filesystem/fat/fs.rs	Updates cached inode metadata during rename operations to maintain consistency

[Copilot]

The code calls find() on the new_parent_inode and then immediately calls do_unlink_at() which likely performs another lookup. Consider storing the result of find() to avoid duplicate lookups.

Suggested change

	if new_parent_inode.find(new_filename).is_ok() {
	let target_exists = new_parent_inode.find(new_filename).is_ok();
	if target_exists {

[Copilot]

The vma.lock_irqsave() is called twice in this function for the same VMA - once here and once earlier at line 340. Consider storing the VMA guard to avoid redundant lock acquisitions.

Suggested change

	if let Ok(virt) = vma.lock_irqsave().page_address(page_index) {
	if let Ok(virt) = vma_guard.page_address(page_index) {

[Copilot]

[nitpick] The magic number 0 is used as a parameter to get_entry(). Consider defining a named constant to clarify what this parameter represents.

Suggested change

	if let Some(entry) = mapper.get_entry(virt, 0) {
	if let Some(entry) = mapper.get_entry(virt, GET_ENTRY_DEFAULT_FLAGS) {

[Copilot]

Using raw pointer comparison to check if two Arc references point to the same object is unsafe and unclear. Consider using Arc::ptr_eq() or comparing the underlying data addresses in a safer way.

Suggested change

	return self.rename_file_in_current_dir(old_name, new_name);
	let self_arc = self.0.lock().self_ref.upgrade();
	let other_arc = other.0.lock().self_ref.upgrade();
	if let (Some(self_arc), Some(other_arc)) = (&self_arc, &other_arc) {
	if Arc::ptr_eq(self_arc, other_arc) {
	return self.rename_file_in_current_dir(old_name, new_name);
	}

[Copilot]

Using raw pointer comparison to check if two Arc references point to the same object is unsafe and unclear. Consider using Arc::ptr_eq() or comparing the underlying data addresses in a safer way.

Suggested change

	return self.rename_file_in_current_dir(old_name, new_name);
	// Use Arc::ptr_eq to safely check if both directories are the same
	let self_arc = self.0.lock().self_ref.upgrade();
	let other_arc = other.0.lock().self_ref.upgrade();
	if let (Some(self_arc), Some(other_arc)) = (&self_arc, &other_arc) {
	if Arc::ptr_eq(self_arc, other_arc) {
	return self.rename_file_in_current_dir(old_name, new_name);
	}

[fslongjin]

这里这样删除可能会导致内存泄漏，因为：
页面还在page cache里面，然后这里就把他从page reclaimer里面删掉了。这样不合理。
要删除页面请参考shrink_list函数。

并且同步的删除page cache其实性能很差。我认为一种可行的解决方案就是，文件删除之后，标记其page cache为dead的了，然后页面回收器的线程异步回收。