feat(fs): 添加文件预读功能支持 (#1391)

- 新增readahead模块实现文件预读算法
- 修改page_cache模块支持预读标记
- 在File结构体中添加预读状态管理
- 为PageFlags添加PG_READAHEAD标志位

Signed-off-by: xbohodx <[email protected]>

Author: xboHodx

kernel/src/filesystem/page_cache.rs

@@ -62,16 +62,13 @@ impl InnerPageCache {
         self.pages.remove(&offset)
     }
 
-    fn create_pages(&mut self, start_page_index: usize, buf: &[u8]) -> Result<(), SystemError> {
-        assert!(buf.len().is_multiple_of(MMArch::PAGE_SIZE));
-
-        let page_num = buf.len() / MMArch::PAGE_SIZE;
-
-        let len = buf.len();
-        if len == 0 {
+    pub fn create_pages(&mut self, start_page_index: usize, buf: &[u8]) -> Result<(), SystemError> {
+        if buf.is_empty() {
             return Ok(());
         }
 
+        let page_num = ((buf.len() - 1) >> MMArch::PAGE_SHIFT) + 1;
+
         let mut page_manager_guard = page_manager_lock_irqsave();
 
         for i in 0..page_num {
@@ -90,12 +87,15 @@ impl InnerPageCache {
                 &mut LockedFrameAllocator,
             )?;
 
+            let page_len = core::cmp::min(MMArch::PAGE_SIZE, buf.len() - buf_offset);
+
             let mut page_guard = page.write_irqsave();
             unsafe {
-                page_guard.copy_from_slice(&buf[buf_offset..buf_offset + MMArch::PAGE_SIZE]);
+                let dst = page_guard.as_slice_mut();
+                dst[..page_len].copy_from_slice(&buf[buf_offset..buf_offset + page_len]);
             }
 
-            self.add_page(page_index, &page);
+            self.add_page(start_page_index + i, &page);
         }
 
         Ok(())

kernel/src/filesystem/vfs/file.rs

@@ -7,6 +7,7 @@ use system_error::SystemError;
 use super::{FileType, IndexNode, InodeId, Metadata, SpecialNodeData};
 use crate::process::pid::PidPrivateData;
 use crate::{
+    arch::MMArch,
     driver::{
         base::{block::SeekFrom, device::DevicePrivateData},
         tty::tty_device::TtyFilePrivateData,
@@ -18,6 +19,11 @@ use crate::{
     },
     ipc::{kill::kill_process, pipe::PipeFsPrivateData},
     libs::{rwlock::RwLock, spinlock::SpinLock},
+    mm::{
+        page::PageFlags,
+        readahead::{page_cache_async_readahead, page_cache_sync_readahead, FileReadaheadState},
+        MemoryManagementArch,
+    },
     process::{cred::Cred, resource::RLimitID, ProcessControlBlock, ProcessManager, RawPid},
 };
 
@@ -151,6 +157,8 @@ pub struct File {
     close_on_exec: AtomicBool,
     /// owner
     pid: SpinLock<Option<Arc<ProcessControlBlock>>>,
+    /// 预读状态
+    ra_state: SpinLock<FileReadaheadState>,
 }
 
 impl File {
@@ -182,6 +190,7 @@ impl File {
             cred: ProcessManager::current_pcb().cred(),
             close_on_exec: AtomicBool::new(close_on_exec),
             pid: SpinLock::new(None),
+            ra_state: SpinLock::new(FileReadaheadState::new()),
         };
 
         return Ok(f);
@@ -251,6 +260,58 @@ impl File {
         self.do_write(offset, len, buf, false)
     }
 
+    fn file_readahead(&self, offset: usize, len: usize) -> Result<(), SystemError> {
+        let page_cache = match self.inode.page_cache() {
+            Some(page_cahce) => page_cahce,
+            None => return Ok(()),
+        };
+
+        let start_page = offset >> MMArch::PAGE_SHIFT;
+        let end_page = (offset + len - 1) >> MMArch::PAGE_SHIFT;
+
+        let (async_trigger_page, missing_page) = {
+            let page_cache_guard = page_cache.lock_irqsave();
+            let mut async_trigger_page = None;
+            let mut missing_page = None;
+
+            for index in start_page..=end_page {
+                match page_cache_guard.get_page(index) {
+                    Some(page)
+                        if page
+                            .read_irqsave()
+                            .flags()
+                            .contains(PageFlags::PG_READAHEAD) =>
+                    {
+                        async_trigger_page = Some((index, page.clone()));
+                        break;
+                    }
+                    None => {
+                        missing_page = Some(index);
+                        break;
+                    }
+                    _ => {}
+                }
+            }
+            (async_trigger_page, missing_page)
+        };
+
+        if let Some((index, page)) = async_trigger_page {
+            let mut ra_state = self.ra_state.lock().clone();
+            let req_pages = end_page - index + 1;
+            page.write_irqsave().remove_flags(PageFlags::PG_READAHEAD);
+
+            page_cache_async_readahead(&page_cache, &self.inode, &mut ra_state, index, req_pages)?;
+            *self.ra_state.lock() = ra_state;
+        } else if let Some(index) = missing_page {
+            let mut ra_state = self.ra_state.lock().clone();
+            let req_pages = end_page - index + 1;
+
+            page_cache_sync_readahead(&page_cache, &self.inode, &mut ra_state, index, req_pages)?;
+            *self.ra_state.lock() = ra_state;
+        }
+        Ok(())
+    }
+
     pub fn do_read(
         &self,
         offset: usize,
@@ -264,6 +325,10 @@ impl File {
             return Err(SystemError::ENOBUFS);
         }
 
+        if self.file_type == FileType::File && !self.mode().contains(FileMode::O_DIRECT) {
+            self.file_readahead(offset, len)?;
+        }
+
         let len = if self.mode().contains(FileMode::O_DIRECT) {
             self.inode
                 .read_direct(offset, len, buf, self.private_data.lock())
@@ -272,11 +337,15 @@ impl File {
                 .read_at(offset, len, buf, self.private_data.lock())
         }?;
 
+        if len > 0 {
+            let last_page_readed = (offset + len - 1) >> MMArch::PAGE_SHIFT;
+            self.ra_state.lock().prev_index = last_page_readed as i64;
+        }
+
         if update_offset {
             self.offset
                 .fetch_add(len, core::sync::atomic::Ordering::SeqCst);
         }
-
         Ok(len)
     }
 
@@ -522,6 +591,7 @@ impl File {
             cred: self.cred.clone(),
             close_on_exec: AtomicBool::new(self.close_on_exec.load(Ordering::SeqCst)),
             pid: SpinLock::new(None),
+            ra_state: SpinLock::new(self.ra_state.lock().clone()),
         };
         // 调用inode的open方法，让inode知道有新的文件打开了这个inode
         // TODO: reopen is not a good idea for some inodes, need a better design

kernel/src/libs/mod.rs

@@ -31,3 +31,4 @@ pub mod name;
 pub mod decompress;
 
 pub mod pod;
+pub mod ranges;

kernel/src/libs/ranges.rs

@@ -0,0 +1,120 @@
+use crate::alloc::vec::Vec;
+
+/// 合并连续的整数为范围
+///
+/// ## 参数
+/// - `set`: 整数集合，必须已排序
+///
+/// ## 返回值
+/// - `Vec<(start, count)>`: 范围列表
+///
+/// ## 示例
+/// ```
+/// merge_ranges(&[1, 2, 3, 5, 6, 8])
+/// // 返回 [(1, 3), (5, 2), (8, 1)]
+/// ```
+pub fn merge_ranges(set: &[usize]) -> Vec<(usize, usize)> {
+    if set.is_empty() {
+        return Vec::new();
+    }
+
+    let mut ranges = Vec::new();
+    let mut start = set[0];
+    let mut count = 1;
+
+    for &page_index in &set[1..] {
+        if page_index == start + count {
+            count += 1; // 连续，扩展范围
+        } else {
+            ranges.push((start, count)); // 保存当前范围
+            start = page_index;
+            count = 1;
+        }
+    }
+
+    ranges.push((start, count)); // 最后一个范围
+    ranges
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_empty_input() {
+        // 边界情况：空输入
+        let result = merge_ranges(&[]);
+        assert_eq!(result, vec![]);
+    }
+
+    #[test]
+    fn test_single_page() {
+        // 边界情况：单个页面
+        let result = merge_ranges(&[5]);
+        assert_eq!(result, vec![(5, 1)]);
+    }
+
+    #[test]
+    fn test_fully_consecutive() {
+        // 完全连续的页面
+        let result = merge_ranges(&[1, 2, 3, 4, 5]);
+        assert_eq!(result, vec![(1, 5)]);
+    }
+
+    #[test]
+    fn test_no_consecutive() {
+        // 完全不连续的页面
+        let result = merge_ranges(&[1, 3, 5, 7, 9]);
+        assert_eq!(result, vec![(1, 1), (3, 1), (5, 1), (7, 1), (9, 1)]);
+    }
+
+    #[test]
+    fn test_mixed_ranges() {
+        // 文档示例：混合连续和不连续
+        let result = merge_ranges(&[1, 2, 3, 5, 6, 8]);
+        assert_eq!(result, vec![(1, 3), (5, 2), (8, 1)]);
+    }
+
+    #[test]
+    fn test_two_consecutive_pairs() {
+        // 两对连续的范围
+        let result = merge_ranges(&[0, 1, 10, 11]);
+        assert_eq!(result, vec![(0, 2), (10, 2)]);
+    }
+
+    #[test]
+    fn test_start_from_zero() {
+        // 从 0 开始的连续范围
+        let result = merge_ranges(&[0, 1, 2]);
+        assert_eq!(result, vec![(0, 3)]);
+    }
+
+    #[test]
+    fn test_large_gap() {
+        // 大间隔的不连续页面
+        let result = merge_ranges(&[0, 100, 200]);
+        assert_eq!(result, vec![(0, 1), (100, 1), (200, 1)]);
+    }
+
+    #[test]
+    fn test_long_consecutive_sequence() {
+        // 长连续序列（模拟大文件预读）
+        let input: Vec<usize> = (0..128).collect();
+        let result = merge_ranges(&input);
+        assert_eq!(result, vec![(0, 128)]);
+    }
+
+    #[test]
+    fn test_alternating_pattern() {
+        // 交替的连续和间断模式
+        let result = merge_ranges(&[1, 2, 4, 5, 7, 8, 10]);
+        assert_eq!(result, vec![(1, 2), (4, 2), (7, 2), (10, 1)]);
+    }
+
+    #[test]
+    fn test_real_world_sparse() {
+        // 真实场景：稀疏的页面缓存（部分页面已存在）
+        let result = merge_ranges(&[0, 5, 6, 7, 12, 20, 21]);
+        assert_eq!(result, vec![(0, 1), (5, 3), (12, 1), (20, 2)]);
+    }
+}

kernel/src/mm/mod.rs

@@ -33,6 +33,7 @@ pub mod mmio_buddy;
 pub mod no_init;
 pub mod page;
 pub mod percpu;
+pub mod readahead;
 pub mod syscall;
 pub mod sysfs;
 pub mod truncate;
@@ -554,13 +555,13 @@ pub trait MemoryManagementArch: Clone + Copy + Debug {
 
     /// @brief 读取指定虚拟地址的值，并假设它是类型T的指针
     #[inline(always)]
-    unsafe fn read<T>(address: VirtAddr) -> T {
+    unsafe fn read<T: Sized>(address: VirtAddr) -> T {
         return ptr::read(address.data() as *const T);
     }
 
     /// @brief 将value写入到指定的虚拟地址
     #[inline(always)]
-    unsafe fn write<T>(address: VirtAddr, value: T) {
+    unsafe fn write<T: Sized>(address: VirtAddr, value: T) {
         ptr::write(address.data() as *mut T, value);
     }
 

kernel/src/mm/page.rs

@@ -434,6 +434,7 @@ bitflags! {
         const PG_RECLAIM = 1 << 18;
         const PG_SWAPBACKED = 1 << 19;
         const PG_UNEVICTABLE = 1 << 20;
+        const PG_READAHEAD = 1 << 21;
     }
 }