Kernel compatibility : ext2's dtime field?
RFC 9/11 Kernel compatibility : ext2's dtime field?
Now, another question. In ext2, what is the point of the dtime field? Personaly, I'm never setting it, because, well, if an inode is removed, it's removed, and nobody is supposed to access it again; and anyways, since no syscall allows seeing it, the dtime seems to me like nothing but an information leakage. But I doubt you would have put a useless data in the filesystem, so I'm likely overlooking something; if so, what is it?
Quoted 7 lines · Emmanuel Colbus
Now, another question. In ext2, what is the point of the dtime field? Personaly, I'm never setting it, because, well, if an inode is removed, it's removed, and nobody is supposed to access it again; and anyways, since no syscall allows seeing it, the dtime seems to me like nothing but an information leakage. But I doubt you would have put a useless data in the filesystem, so I'm likely overlooking something; if so, what is it?
For ext2, it was useful as a way of being able to do some emergency undelete operations on a file system. These days, it's mostly for debugging purposes.
This field is also used as a linked list in ext3 and ext4 for "orphan inodes" and inodes which are in the process of being truncated or for which should be truncated after a crash (to avoid leaking stale data, for example).
- Ted