| 1 |
.\" Copyright 1994-1995, Daniel Quinlan <[email protected]> |
| 2 |
.\" Copyright 2002-2017, Michael Kerrisk <[email protected]> |
| 3 |
.\" Copyright, the authors of the Linux man-pages project |
| 4 |
.\" |
| 5 |
.\" SPDX-License-Identifier: GPL-3.0-or-later |
| 6 |
.\" |
| 7 |
.TH proc_pid_root 5 2026-02-08 "Linux man-pages 6.18" |
| 8 |
.SH NAME |
| 9 |
/proc/pid/root/ \- symbolic link to root directory |
| 10 |
.SH DESCRIPTION |
| 11 |
.TP |
| 12 |
.IR /proc/ pid /root/ |
| 13 |
UNIX and Linux support the idea of a per-process root of the |
| 14 |
filesystem, set by the |
| 15 |
.BR chroot (2) |
| 16 |
system call. |
| 17 |
This file is a symbolic link that points to the process's |
| 18 |
root directory, and behaves in the same way as |
| 19 |
.IR exe , |
| 20 |
and |
| 21 |
.IR fd/* . |
| 22 |
.IP |
| 23 |
Note however that this file is not merely a symbolic link. |
| 24 |
It provides the same view of the filesystem (including namespaces and the |
| 25 |
set of per-process mounts) as the process itself. |
| 26 |
An example illustrates this point. |
| 27 |
In one terminal, we start a shell in new user and mount namespaces, |
| 28 |
and in that shell we create some new mounts: |
| 29 |
.IP |
| 30 |
.in +4n |
| 31 |
.EX |
| 32 |
.RB $ " PS1=\[aq]sh1# \[aq] unshare \-Urnm" ; |
| 33 |
.RB sh1# " mount \-t tmpfs tmpfs /etc" "; # Mount empty tmpfs at /etc" |
| 34 |
.RB sh1# " mount \-\-bind /usr /dev" "; # Mount /usr at /dev" |
| 35 |
.RB sh1# " echo $$" ; |
| 36 |
27123 |
| 37 |
.EE |
| 38 |
.in |
| 39 |
.IP |
| 40 |
In a second terminal window, in the initial mount namespace, |
| 41 |
we look at the contents of the corresponding mounts in |
| 42 |
the initial and new namespaces: |
| 43 |
.IP |
| 44 |
.in +4n |
| 45 |
.EX |
| 46 |
.RB $ " PS1=\[aq]sh2# \[aq] sudo sh" ; |
| 47 |
.RB sh2# " ls /etc | wc \-l" "; # In initial NS" |
| 48 |
309 |
| 49 |
.RB sh2# " ls /proc/27123/root/etc | wc \-l" "; # /etc in other NS" |
| 50 |
0 # The empty tmpfs dir |
| 51 |
.RB sh2# " ls /dev | wc \-l" "; # In initial NS" |
| 52 |
205 |
| 53 |
.RB sh2# " ls /proc/27123/root/dev | wc \-l" "; # /dev in other NS" |
| 54 |
11 # Actually bind |
| 55 |
# mounted to /usr |
| 56 |
.RB sh2# " ls /usr | wc \-l" "; # /usr in initial NS" |
| 57 |
11 |
| 58 |
.EE |
| 59 |
.in |
| 60 |
.IP |
| 61 |
.\" The following was still true as at kernel 2.6.13 |
| 62 |
In a multithreaded process, the contents of the |
| 63 |
.IR /proc/ pid /root |
| 64 |
symbolic link are not available if the main thread has already terminated |
| 65 |
(typically by calling |
| 66 |
.BR pthread_exit (3)). |
| 67 |
.IP |
| 68 |
Permission to dereference or read |
| 69 |
.RB ( readlink (2)) |
| 70 |
this symbolic link is governed by a ptrace access mode |
| 71 |
.B PTRACE_MODE_READ_FSCREDS |
| 72 |
check; |
| 73 |
see |
| 74 |
.BR ptrace (2). |
| 75 |
.SH SEE ALSO |
| 76 |
.BR proc (5) |