PW_INIT(3) BSD Library Functions Manual PW_INIT(3)
pw_init, pw_edit, pw_prompt, pw_copy, pw_copyx, pw_scan, pw_error -- utility functions for interactive passwd file updates
System Utilities Library (libutil, -lutil)
pw_edit(int notsetuid, const char *filename);
pw_copy(int ffd, int tfd, struct passwd *pw, struct passwd *old_pw);
pw_copyx(int ffd, int tfd, struct passwd *pw, struct passwd *old_pw, char *errbuf, size_t errbufsz);
pw_scan(char *bp, struct passwd *pw, int *flags);
pw_error(const char *name, int err, int eval);
These functions are designed as conveniences for interactive programs which update the passwd file and do nothing else. They generally han-
dle errors by printing out a message to the standard error stream and possibly aborting the process.
The pw_init() function prepares for a passwd update by unlimiting all resource constraints, disabling core dumps (thus preventing dumping the
contents of the passwd database into a world-readable file), and disabling most signals.
The pw_edit() function runs an editor (named by the environment variable EDITOR, or /usr/bin/vi if EDITOR is not set) on the file filename
(or /etc/ptmp if filename is NULL). If notsetuid is nonzero, pw_edit() will set the effective user and group ID to the real user and group
ID before running the editor.
The pw_prompt() function asks the user whether he or she wants to re-edit the password file; if the answer is no, pw_prompt() deletes the
lock file and exits the process.
The pw_copy() function reads a passwd file from ffd and writes it to tfd, updating the entry corresponding to pw->pw_name with the informa-
tion in pw. If old_pw is not NULL, it checks to make sure the old entry is the same as the one described in old_pw or the process is
aborted. If an entry is not found to match pw, a new entry is appended to the passwd file only if the real user ID is 0. If an error
occurs, pw_copy() will display a message on stderr and call pw_error().
The pw_copyx() function performs the same operation as pw_copy() with the exception of error handling. Upon an error, pw_copyx() will write
an error message into the buffer pointed to by errbuf which has the size errbufsz.
The pw_scan() function accepts in bp a passwd entry as it would be represented in /etc/master.passwd and fills in pw with corresponding val-
ues; string fields in pw will be pointers into bp. Some characters in bp will be overwritten with 0s in order to terminate the strings
pointed to by pw. If flags is non-null, it should be cleared and the following options enabled if required:
_PASSWORD_NOWARN Don't print warnings.
_PASSWORD_OLDFMT Parse bp as an old format entry as found in /etc/passwd.
Upon return it is cleared, and filled in with the following flags:
_PASSWORD_NOUID The uid field of bp is empty.
_PASSWORD_NOGID The gid field of bp is empty.
_PASSWORD_NOCHG The change field of bp is empty.
_PASSWORD_NOEXP The expire field of bp is empty.
The pw_error() function displays an error message, aborts the current passwd update, and exits the current process. If err is non-zero, a
warning message beginning with name is printed for the current value of errno. The process exits with status eval.
The pw_copyx() function returns 1 if the new password entry was successfully written to the destination file, and 0 otherwise.
The pw_scan() function prints a warning message and returns 0 if the string in the bp argument is not a valid passwd string. Otherwise,
pw_scan() returns 1.
August 1, 2004 BSD