标签：
【Real and Effective IDs】
At the lowest level of the operating system, the kernel, users and groups aren‘t identified by names, but numbers. The kernel needs to be fast and robust, and data structures better be small, and moving around strings is anything but efficient. So, each user name and group name is mapped to a unique unsigned number, called user and group ID for short, or uid and gid. This mapping is done via the /etc/passwd and /etc/group files, respectively. The user and group ID 0 are commonly called root, but that‘s really just a convention.
Each Unix process has a user ID and a group ID associated with it, and when trying to open a file for writing, for instance, these IDs are used to determine whether the process should be granted access or not. These IDs constitute the effective privilege of the process, because they determine what a process can do and what it cannot. Most of the time, these IDs will be referred to as the effective uid and gid.
What happens when you invoke the passwd utility is that the effective uid of the process is set to 0, i.e. the uid of the root user. As a result, the program is permitted to modify the /etc/passwd file, and can thus replace the encrypted password in your account entry with the new one you just provided.
If you‘re familiar with the passwd utility, you‘ll know that as a normal user, you‘re only allowed to modify the password of your own account; it will not let you modify the password of any other account. So this begs the question, how does it know who invoked it?
That‘s where another pair of user and group ID comes in, called the real uid and gid, respectively. These IDs are used to track who a user really is, i.e. on what account he or she is logged in. This uid value is not changed when you invoke programs such as passwd. So the program simply needs to find out what user name corresponds to its real uid, and refuse to change any other account.
参考：http://www.lst.de/~okir/blackhats/node23.html
标签：
原文地址：http://www.cnblogs.com/tekka/p/4470619.html
