lab4_fs中进程的fd table计数问题

见lab4_1分支的proc_file.c:49，即下面列出代码中加粗行。

proc_file_management *init_proc_file_management(void) {
  proc_file_management *pfiles = (proc_file_management *)alloc_page();
  pfiles->cwd = vfs_root_dentry; // by default, cwd is the root
  pfiles->nfiles = 0; 

  for (int fd = 0; fd < MAX_FILES; ++fd) {
    pfiles->opened_files[fd].status = FD_NONE;
     **++pfiles->nfiles;** 
  }
  sprint("FS: created a file management struct for a process.\n");
  return pfiles;
}

按照int nfiles; // the number of opened files a process has的描述，这里不应该进行计数。

这是一个trivial bug，因为nfiles没有被使用到。

另外对nfiles注释修改为the number of files opened by a process感觉更妥当

do_open中寻找空闲fd的部分proc_file.c:90-96

  struct file *pfile;
  for (fd = 0; fd < MAX_FILES; ++fd) {
    pfile = &(current->pfiles->opened_files[fd]);
    if (pfile->status == FD_NONE) break;
  }
  if (pfile->status != FD_NONE)  // no free entry
    panic("do_open: no file entry for current process!\n");

可以修改为

  if (current->pfiles->nfiles >= MAX_FILES) {
    panic("do_open: no file entry for current process!\n");
  }
  struct file *pfile;
  for (fd = 0; fd < MAX_FILES; ++fd) {
    pfile = &(current->pfiles->opened_files[fd]);
    if (pfile->status == FD_NONE) break;
  }

这样nfiles能够被用上，代码也更"优雅"

感谢，下次进行公开仓库更新时会修复

华中科技大学操作系统团队 / riscv-pke

内容风险标识

评论 (3)

华中科技大学操作系统团队 / riscv-pke .gitee-modal { width: 500px !important; }

内容风险标识

lab4_fs中进程的fd table计数问题

评论 (3)

搜索帮助

华中科技大学操作系统团队 / riscv-pke