| 1 |
.\" Copyright 1995, Thomas K. Dyas <[email protected]> |
| 2 |
.\" Copyright 2013-2019, Michael Kerrisk <[email protected]> |
| 3 |
.\" Copyright, the authors of the Linux man-pages project |
| 4 |
.\" |
| 5 |
.\" SPDX-License-Identifier: Linux-man-pages-copyleft |
| 6 |
.\" |
| 7 |
.TH setfsuid 2 2026-02-08 "Linux man-pages 6.18" |
| 8 |
.SH NAME |
| 9 |
setfsuid \- set user identity used for filesystem checks |
| 10 |
.SH LIBRARY |
| 11 |
Standard C library |
| 12 |
.RI ( libc ,\~ \-lc ) |
| 13 |
.SH SYNOPSIS |
| 14 |
.nf |
| 15 |
.B #include <sys/fsuid.h> |
| 16 |
.P |
| 17 |
.BI "[[deprecated]] int setfsuid(uid_t " fsuid ); |
| 18 |
.fi |
| 19 |
.SH DESCRIPTION |
| 20 |
On Linux, a process has both a filesystem user ID and an effective user ID. |
| 21 |
The (Linux-specific) filesystem user ID is used |
| 22 |
for permissions checking when accessing filesystem objects, |
| 23 |
while the effective user ID is used for various other kinds |
| 24 |
of permissions checks (see |
| 25 |
.BR credentials (7)). |
| 26 |
.P |
| 27 |
Normally, the value of the process's filesystem user ID |
| 28 |
is the same as the value of its effective user ID. |
| 29 |
This is so, because whenever a process's effective user ID is changed, |
| 30 |
the kernel also changes the filesystem user ID to be the same as |
| 31 |
the new value of the effective user ID. |
| 32 |
A process can cause the value of its filesystem user ID to diverge |
| 33 |
from its effective user ID by using |
| 34 |
.BR setfsuid () |
| 35 |
to change its filesystem user ID to the value given in |
| 36 |
.IR fsuid . |
| 37 |
.P |
| 38 |
Explicit calls to |
| 39 |
.BR setfsuid () |
| 40 |
and |
| 41 |
.BR setfsgid (2) |
| 42 |
are (were) usually used only by programs such as the Linux NFS server that |
| 43 |
need to change what user and group ID is used for file access without a |
| 44 |
corresponding change in the real and effective user and group IDs. |
| 45 |
A change in the normal user IDs for a program such as the NFS server |
| 46 |
is (was) a security hole that can expose it to unwanted signals. |
| 47 |
(However, |
| 48 |
this issue is historical; |
| 49 |
see below.) |
| 50 |
.P |
| 51 |
.BR setfsuid () |
| 52 |
will succeed only if the caller is the superuser or if |
| 53 |
.I fsuid |
| 54 |
matches either the caller's real user ID, effective user ID, |
| 55 |
saved set-user-ID, or current filesystem user ID. |
| 56 |
.SH RETURN VALUE |
| 57 |
On both success and failure, |
| 58 |
this call returns the previous filesystem user ID of the caller. |
| 59 |
.SH STANDARDS |
| 60 |
Linux. |
| 61 |
.SH HISTORY |
| 62 |
Linux 1.2. |
| 63 |
.\" Linux 1.1.44 |
| 64 |
.\" and in libc since libc 4.7.6. |
| 65 |
.P |
| 66 |
At the time when this system call was introduced, one process |
| 67 |
could send a signal to another process with the same effective user ID. |
| 68 |
This meant that if a privileged process changed its effective user ID |
| 69 |
for the purpose of file permission checking, |
| 70 |
then it could become vulnerable to receiving signals |
| 71 |
sent by another (unprivileged) process with the same user ID. |
| 72 |
The filesystem user ID attribute was thus added to allow a process to |
| 73 |
change its user ID for the purposes of file permission checking without |
| 74 |
at the same time becoming vulnerable to receiving unwanted signals. |
| 75 |
Since Linux 2.0, signal permission handling is different (see |
| 76 |
.BR kill (2)), |
| 77 |
with the result that a process can change its effective user ID |
| 78 |
without being vulnerable to receiving signals from unwanted processes. |
| 79 |
Thus, |
| 80 |
.BR setfsuid () |
| 81 |
is nowadays unneeded and should be avoided in new applications |
| 82 |
(likewise for |
| 83 |
.BR setfsgid (2)). |
| 84 |
.P |
| 85 |
The original Linux |
| 86 |
.BR setfsuid () |
| 87 |
system call supported only 16-bit user IDs. |
| 88 |
Subsequently, Linux 2.4 added |
| 89 |
.BR setfsuid32 () |
| 90 |
supporting 32-bit IDs. |
| 91 |
The glibc |
| 92 |
.BR setfsuid () |
| 93 |
wrapper function transparently deals with the variation across kernel versions. |
| 94 |
.SS C library/kernel differences |
| 95 |
In glibc 2.15 and earlier, |
| 96 |
when the wrapper for this system call determines that the argument can't be |
| 97 |
passed to the kernel without integer truncation (because the kernel |
| 98 |
is old and does not support 32-bit user IDs), |
| 99 |
it will return \-1 and set |
| 100 |
.I errno |
| 101 |
to |
| 102 |
.B EINVAL |
| 103 |
without attempting |
| 104 |
the system call. |
| 105 |
.SH BUGS |
| 106 |
No error indications of any kind are returned to the caller, |
| 107 |
and the fact that both successful and unsuccessful calls return |
| 108 |
the same value makes it impossible to directly determine |
| 109 |
whether the call succeeded or failed. |
| 110 |
Instead, the caller must resort to looking at the return value |
| 111 |
from a further call such as |
| 112 |
.I setfsuid(\-1) |
| 113 |
(which will always fail), in order to determine if a preceding call to |
| 114 |
.BR setfsuid () |
| 115 |
changed the filesystem user ID. |
| 116 |
At the very |
| 117 |
least, |
| 118 |
.B EPERM |
| 119 |
should be returned when the call fails (because the caller lacks the |
| 120 |
.B CAP_SETUID |
| 121 |
capability). |
| 122 |
.SH SEE ALSO |
| 123 |
.BR kill (2), |
| 124 |
.BR setfsgid (2), |
| 125 |
.BR capabilities (7), |
| 126 |
.BR credentials (7) |