Toward a better definition for i_version
Toward a better definition for i_version
Posted Aug 28, 2022 18:15 UTC (Sun) by NYKevin (subscriber, #129325)In reply to: Toward a better definition for i_version by jlayton
Parent article: Toward a better definition for i_version
Given that i_version is (should be) bumped iff mtime or ctime would be bumped, then why does i_version need to exist? Can't userspace just query mtime and ctime directly?
Posted Aug 28, 2022 22:44 UTC (Sun)
by corbet (editor, #1)
[Link]
The problem with mtime and ctime is that they don't have anything near the required resolution to catch every change. Neil's article, linked in this article, describes those issues in detail.
mtime and ctime