document openat2(2)
v2 cover document openat2(2)
Patch changelog:
v2:
* Updated based on the new (now-merged) structure layout and semantics.
v1: <https://lore.kernel.org/linux-man/20191206142931.28138-1-cyphar-gVpy/LI/[email protected]/>
* Updated based on Kerrisk's review.
RFC: <https://lore.kernel.org/linux-man/20191003145542.17490-1-cyphar-gVpy/LI/[email protected]/>
Now that openat2(2) has landed in Linus's tree[1], here is the associated man-page changes to document it.
Aleksa Sarai (2):
path_resolution.7: update to mention openat2(2) features openat2.2: document new openat2(2) syscall
man2/open.2 | 17 ++ man2/openat2.2 | 471 +++++++++++++++++++++++++++++++++++++++++ man7/path_resolution.7 | 56 +++-- 3 files changed, 526 insertions(+), 18 deletions(-) create mode 100644 man2/openat2.2
Signature · 1 line
2.25.0
PATCH v2 1/2 path_resolution.7: update to mention openat2(2) features
--- man7/path_resolution.7 | 56 ++++++++++++++++++++++++++++-------------- 1 file changed, 38 insertions(+), 18 deletions(-)
man7/path_resolution.7 +38 -18
@@ -29,30 +29,38 @@ path_resolution \- how a pathname is resolved to a file Some UNIX/Linux system calls have as parameter one or more filenames. A filename (or pathname) is resolved as follows. .SS Step 1: start of the resolution process -If the pathname starts with the \(aq/\(aq character, -the starting lookup directory -is the root directory of the calling process. -(A process inherits its -root directory from its parent. -Usually this will be the root directory -of the file hierarchy. -A process may get a different root directory -by use of the +If the pathname starts with the \(aq/\(aq character, the starting lookup +directory is the root directory of the calling process. +A process inherits its root directory from its parent. +Usually this will be the root directory of the file hierarchy. +A process may get a different root directory by use of the .BR chroot (2) -system call. +system call, or may temporarily use a different root directory by using +.BR openat2 (2) +with the +.B RESOLVE_IN_ROOT +flag set. +.PP A process may get an entirely private mount namespace in case it\(emor one of its ancestors\(emwas started by an invocation of the .BR clone (2) system call that had the .B CLONE_NEWNS -flag set.) +flag set. This handles the \(aq/\(aq part of the pathname. .PP -If the pathname does not start with the \(aq/\(aq character, the -starting lookup directory of the resolution process is the current working -directory of the process. -(This is also inherited from the parent. -It can be changed by use of the +If the pathname does not start with the \(aq/\(aq character, the starting +lookup directory of the resolution process is the current working directory of +the process \(em or in the case of +.BR openat (2)-style +system calls, the +.I dfd +argument (or the current working directory if +.B AT_FDCWD +is passed as the +.I dfd +argument). The current working directory is inherited from the parent, and can +be changed by use of the .BR chdir (2) system call.) .PP
@@ -91,7 +99,7 @@ Upon error, that error is returned. If the result is not a directory, an .B ENOTDIR error is returned. -If the resolution of the symlink is successful and returns a directory, +If the resolution of the symbolic link is successful and returns a directory, we set the current lookup directory to that directory, and go to the next component. Note that the resolution process here can involve recursion if the
@@ -124,6 +132,12 @@ the kernel's pathname-resolution code was reworked to eliminate the use of recursion, so that the only limit that remains is the maximum of 40 resolutions for the entire pathname. +.PP +The resolution of symbolic links during this stage can be blocked by using +.BR openat2 (2), +with the +.B RESOLVE_NO_SYMLINKS +flag set. .SS Step 3: find the final entry The lookup of the final component of the pathname goes just like that of all other components, as described in the previous step,
@@ -145,7 +159,7 @@ The path resolution process will assume that these entries have their conventional meanings, regardless of whether they are actually present in the physical filesystem. .PP -One cannot walk down past the root: "/.." is the same as "/". +One cannot walk up past the root: "/.." is the same as "/". .SS Mount points After a "mount dev path" command, the pathname "path" refers to the root of the filesystem hierarchy on the device "dev", and no
@@ -154,6 +168,12 @@ longer to whatever it referred to earlier. One can walk out of a mounted filesystem: "path/.." refers to the parent directory of "path", outside of the filesystem hierarchy on "dev". +.PP +Traversal of mount points can be blocked by using +.BR openat2 (2), +with the +.B RESOLVE_NO_XDEV +flag set (though note that this also restricts bind mount traversal). .SS Trailing slashes If a pathname ends in a \(aq/\(aq, that forces resolution of the preceding component as in Step 2: it has to exist and resolve to a directory.
Signature · 1 line
2.25.0
Re: [PATCH man-pages v2 1/2] path_resolution.7: update to mention openat2(2) features
Hello Aleksa,
Quoted 1 line · Aleksa Sarai
Signed-off-by Aleksa Sarai <[email protected]>
Thanks. I've applied this patch.
Cheers,
Michael
Quoted 108 lines
--- man7/path_resolution.7 | 56 ++++++++++++++++++++++++++++-------------- 1 file changed, 38 insertions(+), 18 deletions(-)man7/path_resolution.7 +38
-18@@ -29,30 +29,38 @@ path_resolution \- how a pathname is resolved to a file Some UNIX/Linux system calls have as parameter one or more filenames. A filename (or pathname) is resolved as follows. .SS Step 1: start of the resolution process -If the pathname starts with the \(aq/\(aq character, -the starting lookup directory -is the root directory of the calling process. -(A process inherits its -root directory from its parent. -Usually this will be the root directory -of the file hierarchy. -A process may get a different root directory -by use of the +If the pathname starts with the \(aq/\(aq character, the starting lookup +directory is the root directory of the calling process. +A process inherits its root directory from its parent. +Usually this will be the root directory of the file hierarchy. +A process may get a different root directory by use of the .BR chroot (2) -system call. +system call, or may temporarily use a different root directory by using +.BR openat2 (2) +with the +.B RESOLVE_IN_ROOT +flag set. +.PP A process may get an entirely private mount namespace in case it\(emor one of its ancestors\(emwas started by an invocation of the .BR clone (2) system call that had the .B CLONE_NEWNS -flag set.) +flag set. This handles the \(aq/\(aq part of the pathname. .PP -If the pathname does not start with the \(aq/\(aq character, the -starting lookup directory of the resolution process is the current working -directory of the process. -(This is also inherited from the parent. -It can be changed by use of the +If the pathname does not start with the \(aq/\(aq character, the starting +lookup directory of the resolution process is the current working directory of +the process \(em or in the case of +.BR openat (2)-style +system calls, the +.I dfd +argument (or the current working directory if +.B AT_FDCWD +is passed as the +.I dfd +argument). The current working directory is inherited from the parent, and can +be changed by use of the .BR chdir (2) system call.) .PP@@ -91,7 +99,7 @@ Upon error, that error is returned. If the result is not a directory, an .B ENOTDIR error is returned. -If the resolution of the symlink is successful and returns a directory, +If the resolution of the symbolic link is successful and returns a directory, we set the current lookup directory to that directory, and go to the next component. Note that the resolution process here can involve recursion if the@@ -124,6 +132,12 @@ the kernel's pathname-resolution code was reworked to eliminate the use of recursion, so that the only limit that remains is the maximum of 40 resolutions for the entire pathname. +.PP +The resolution of symbolic links during this stage can be blocked by using +.BR openat2 (2), +with the +.B RESOLVE_NO_SYMLINKS +flag set. .SS Step 3: find the final entry The lookup of the final component of the pathname goes just like that of all other components, as described in the previous step,@@ -145,7 +159,7 @@ The path resolution process will assume that these entries have their conventional meanings, regardless of whether they are actually present in the physical filesystem. .PP -One cannot walk down past the root: "/.." is the same as "/". +One cannot walk up past the root: "/.." is the same as "/". .SS Mount points After a "mount dev path" command, the pathname "path" refers to the root of the filesystem hierarchy on the device "dev", and no@@ -154,6 +168,12 @@ longer to whatever it referred to earlier. One can walk out of a mounted filesystem: "path/.." refers to the parent directory of "path", outside of the filesystem hierarchy on "dev". +.PP +Traversal of mount points can be blocked by using +.BR openat2 (2), +with the +.B RESOLVE_NO_XDEV +flag set (though note that this also restricts bind mount traversal). .SS Trailing slashes If a pathname ends in a \(aq/\(aq, that forces resolution of the preceding component as in Step 2: it has to exist and resolve to a directory.
Signature · 3 lines
Michael Kerrisk Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/ Linux/UNIX System Programming Training: http://man7.org/training/
PATCH v2 2/2 openat2.2: document new openat2(2) syscall
In reply to Aleksa SaraiRather than trying to merge the new syscall documentation into open.2 (which would probably result in the man-page being incomprehensible), instead the new syscall gets its own dedicated page with links between open(2) and openat2(2) to avoid duplicating information such as the list of O_* flags or common errors.
In addition to describing all of the key flags, information about the extensibility design is provided so that users can better understand why they need to pass sizeof(struct open_how) and how their programs will work across kernels. After some discussions with David Laight, I also included explicit instructions to zero the structure to avoid issues when recompiling with new headers.
--- man2/open.2 | 17 ++ man2/openat2.2 | 471 +++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 488 insertions(+) create mode 100644 man2/openat2.2
man2/open.2 +17 -0
@@ -65,6 +65,10 @@ open, openat, creat \- open and possibly create a file .BI "int openat(int " dirfd ", const char *" pathname ", int " flags ); .BI "int openat(int " dirfd ", const char *" pathname ", int " flags \ ", mode_t " mode ); +.PP +/* Documented separately, in \fBopenat2\fP(2). */ +.BI "int openat2(int " dirfd ", const char *" pathname ", \ +const struct open_how *" how ", size_t " size "); .fi .PP .in -4n
@@ -933,6 +937,15 @@ If is absolute, then .I dirfd is ignored. +.SS openat2(2) +The +.BR openat2 (2) +system call is an extension of +.BR openat (), +with a superset of features. To avoid making this man page too long, the +description of +.BR openat2 (2) +and its features is documented in a separate man page. .SH RETURN VALUE .BR open (), .BR openat (),
@@ -1220,6 +1233,9 @@ SVr4, 4.3BSD, POSIX.1-2001, POSIX.1-2008. .BR openat (): POSIX.1-2008. .PP +.BR openat2 (2) +is Linux-specific. +.PP The .BR O_DIRECT , .BR O_NOATIME ,
@@ -1778,6 +1794,7 @@ is ignored). .BR mknod (2), .BR mmap (2), .BR mount (2), +.BR openat2 (2), .BR open_by_handle_at (2), .BR read (2), .BR socket (2),
man2/openat2.2 new file +471 -0
471 lines not on this page · raw message
Signature · 1 line
2.25.0
Re: [PATCH man-pages v2 2/2] openat2.2: document new openat2(2) syscall
Hello Aleksa,
Quoted 6 lines · Aleksa Sarai
Rather than trying to merge the new syscall documentation into open.2 (which would probably result in the man-page being incomprehensible), instead the new syscall gets its own dedicated page with links between open(2) and openat2(2) to avoid duplicating information such as the list of O_* flags or common errors.
Quoted 7 lines
In addition to describing all of the key flags, information about the extensibility design is provided so that users can better understand why they need to pass
sizeof(struct open_how)and how their programs will work across kernels. After some discussions with David Laight, I also included explicit instructions to zero the structure to avoid issues when recompiling with new headers.>Signed-off-by Aleksa Sarai <[email protected]>
I'm just editing this page, and have a question on one piece.
man2/openat2.2 line 157
Quoted 8 lines
man2/openat2.2 new file +8
-0+Unlike +.BR openat (2), +it is an error to provide +.BR openat2 () +with a +.I mode +which contains bits other than +.IR 0777 ,
This piece appears not to be true, both from my reading of the source code, and from testing (i.e., I wrote a a small program that successfully called openat2() and created a file that had the set-UID, set-GID, and sticky bits set).
Is this a bug in the implementation or a bug in the manual page text?
Thanks,
Michael
Signature · 3 lines
Michael Kerrisk Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/ Linux/UNIX System Programming Training: http://man7.org/training/
Re: [PATCH man-pages v2 2/2] openat2.2: document new openat2(2) syscall
Quoted 33 lines · Michael Kerrisk (man-pages)
Hello Aleksa,
Quoted 6 lines · Aleksa Sarai
Rather than trying to merge the new syscall documentation into open.2 (which would probably result in the man-page being incomprehensible), instead the new syscall gets its own dedicated page with links between open(2) and openat2(2) to avoid duplicating information such as the list of O_* flags or common errors.
Quoted 7 lines
In addition to describing all of the key flags, information about the extensibility design is provided so that users can better understand why they need to pass
sizeof(struct open_how)and how their programs will work across kernels. After some discussions with David Laight, I also included explicit instructions to zero the structure to avoid issues when recompiling with new headers.>Signed-off-by Aleksa Sarai <[email protected]>I'm just editing this page, and have a question on one piece.
man2/openat2.2 line 157
Quoted 8 lines
man2/openat2.2 new file +8
-0+Unlike +.BR openat (2), +it is an error to provide +.BR openat2 () +with a +.I mode +which contains bits other than +.IR 0777 ,This piece appears not to be true, both from my reading of the source code, and from testing (i.e., I wrote a a small program that successfully called
openat2()and created a file that had the set-UID, set-GID, and sticky bits set).
Quoted 1 line
Is this a bug in the implementation or a bug in the manual page text?
My bad -- it's a bug in the manual. The actual check (which does work, there are selftests for this) is:
if (how->mode & ~S_IALLUGO) return -EINVAL;
But when writing the man page I forgot that S_IALLUGO also includes those bits. Do you want me to send an updated version or would you prefer to clean it up?
Signature · 4 lines
Aleksa Sarai Senior Software Engineer (Containers) SUSE Linux GmbH <https://www.cyphar.com/>
Re: [PATCH man-pages v2 2/2] openat2.2: document new openat2(2) syscall
Quoted 39 lines · Aleksa Sarai
Quoted 33 lines · Michael Kerrisk (man-pages)
Hello Aleksa,
Quoted 6 lines · Aleksa Sarai
Rather than trying to merge the new syscall documentation into open.2 (which would probably result in the man-page being incomprehensible), instead the new syscall gets its own dedicated page with links between open(2) and openat2(2) to avoid duplicating information such as the list of O_* flags or common errors.
Quoted 7 lines
In addition to describing all of the key flags, information about the extensibility design is provided so that users can better understand why they need to pass
sizeof(struct open_how)and how their programs will work across kernels. After some discussions with David Laight, I also included explicit instructions to zero the structure to avoid issues when recompiling with new headers.>Signed-off-by Aleksa Sarai <[email protected]>I'm just editing this page, and have a question on one piece.
man2/openat2.2 line 157
Quoted 8 lines
man2/openat2.2 new file +8
-0+Unlike +.BR openat (2), +it is an error to provide +.BR openat2 () +with a +.I mode +which contains bits other than +.IR 0777 ,This piece appears not to be true, both from my reading of the source code, and from testing (i.e., I wrote a a small program that successfully called
openat2()and created a file that had the set-UID, set-GID, and sticky bits set).Quoted 1 line
Is this a bug in the implementation or a bug in the manual page text?
My bad -- it's a bug in the manual. The actual check (which does work, there are selftests for this) is:
Quoted 6 lines
if (how->mode & ~S_IALLUGO) return -EINVAL;But when writing the man page I forgot that
S_IALLUGOalso includes those bits. Do you want me to send an updated version or would you prefer to clean it up?
I'll clean it up.
So, it should say, "bits other than 07777", right?
Thanks,
Michael
Signature · 3 lines
Michael Kerrisk Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/ Linux/UNIX System Programming Training: http://man7.org/training/
Re: [PATCH man-pages v2 2/2] openat2.2: document new openat2(2) syscall
Quoted 43 lines · Michael Kerrisk (man-pages)
Quoted 42 lines · Aleksa Sarai
Quoted 33 lines · Michael Kerrisk (man-pages)
Hello Aleksa,
Quoted 6 lines · Aleksa Sarai
Rather than trying to merge the new syscall documentation into open.2 (which would probably result in the man-page being incomprehensible), instead the new syscall gets its own dedicated page with links between open(2) and openat2(2) to avoid duplicating information such as the list of O_* flags or common errors.
Quoted 7 lines
In addition to describing all of the key flags, information about the extensibility design is provided so that users can better understand why they need to pass
sizeof(struct open_how)and how their programs will work across kernels. After some discussions with David Laight, I also included explicit instructions to zero the structure to avoid issues when recompiling with new headers.>Signed-off-by Aleksa Sarai <[email protected]>I'm just editing this page, and have a question on one piece.
man2/openat2.2 line 157
Quoted 8 lines
man2/openat2.2 new file +8
-0+Unlike +.BR openat (2), +it is an error to provide +.BR openat2 () +with a +.I mode +which contains bits other than +.IR 0777 ,This piece appears not to be true, both from my reading of the source code, and from testing (i.e., I wrote a a small program that successfully called
openat2()and created a file that had the set-UID, set-GID, and sticky bits set).Quoted 1 line
Is this a bug in the implementation or a bug in the manual page text?
My bad -- it's a bug in the manual. The actual check (which does work, there are selftests for this) is:
if (how->mode & ~S_IALLUGO) return -EINVAL;
Quoted 7 lines
Quoted 3 lines
But when writing the man page I forgot that
S_IALLUGOalso includes those bits. Do you want me to send an updated version or would you prefer to clean it up?I'll clean it up.
So, it should say, "bits other than 07777", right?
Yes, that would be correct.
Signature · 4 lines
Aleksa Sarai Senior Software Engineer (Containers) SUSE Linux GmbH <https://www.cyphar.com/>
Re: [PATCH man-pages v2 2/2] openat2.2: document new openat2(2) syscall
In reply to Aleksa SaraiHello Aleksa,
Quoted 6 lines · Aleksa Sarai
Rather than trying to merge the new syscall documentation into open.2 (which would probably result in the man-page being incomprehensible), instead the new syscall gets its own dedicated page with links between open(2) and openat2(2) to avoid duplicating information such as the list of O_* flags or common errors.
Quoted 8 lines
In addition to describing all of the key flags, information about the extensibility design is provided so that users can better understand why they need to pass
sizeof(struct open_how)and how their programs will work across kernels. After some discussions with David Laight, I also included explicit instructions to zero the structure to avoid issues when recompiling with new headers.Signed-off-by Aleksa Sarai <[email protected]>
Thanks. I've applied this patch, but also done quite a lot of editing of the page. The current draft is below (and also pushed to Git). Could I ask you to review the page, to see if I injected any error during my edits.
In addition, I've added a number of FIXMEs in comments in the page source. Can you please check these, and let me know your thoughts.
Cheers,
Michael
.\" Copyright (C) 2019 Aleksa Sarai <[email protected]>
.\"
.\" %%%LICENSE_START(VERBATIM)
.\" Permission is granted to make and distribute verbatim copies of this .\" manual provided the copyright notice and this permission notice are .\" preserved on all copies.
.\"
.\" Permission is granted to copy and distribute modified versions of this .\" manual under the conditions for verbatim copying, provided that the .\" entire resulting derived work is distributed under the terms of a .\" permission notice identical to this one.
.\"
.\" Since the Linux kernel and libraries are constantly changing, this .\" manual page may be incorrect or out-of-date. The author(s) assume no .\" responsibility for errors or omissions, or for damages resulting from .\" the use of the information contained herein. The author(s) may not .\" have taken the same level of care in the production of this manual, .\" which is licensed free of charge, as they might when working .\" professionally.
.\"
.\" Formatted or processed versions of this manual, if unaccompanied by .\" the source, must acknowledge the copyright and authors of this work. .\" %%%LICENSE_END
.TH OPENAT2 2 2019-12-20 "Linux" "Linux Programmer's Manual" .SH NAME
openat2 \- open and possibly create a file (extended)
.SH SYNOPSIS
.nf
.B #include <sys/types.h>
.B #include <sys/stat.h>
.B #include <fcntl.h>
.B #include <openat2.h>
.PP
.BI "int openat2(int " dirfd ", const char *" pathname ,
.BI " struct open_how *" how ", size_t " size ");
.fi
.PP
.IR Note :
There is no glibc wrapper for this system call; see NOTES.
.SH DESCRIPTION
The
.BR openat2 ()
system call is an extension of
.BR openat (2)
and provides a superset of its functionality.
.PP
The
.BR openat2 ()
system call opens the file specified by
.IR pathname .
If the specified file does not exist, it may optionally (if
.B O_CREAT
is specified in
.IR how.flags )
be created.
.PP
As with
.BR openat (2),
if
.I pathname
is a relative pathname, then it is interpreted relative to the directory referred to by the file descriptor
.I dirfd
(or the current working directory of the calling process, if .I dirfd
is the special value
.BR AT_FDCWD ).
If
.I pathname
is an absolute pathname, then
.I dirfd
is ignored (unless
.I how.resolve
contains
.BR RESOLVE_IN_ROOT,
in which case
.I pathname
is resolved relative to
.IR dirfd ).
.PP
Rather than taking a single
.I flags
argument, an extensible structure (\fIhow\fP) is passed to allow for future extensions.
The
.I size
argument must be specified as
.IR "sizeof(struct open_how)" .
.\"
.SS The open_how structure
The
.I how
argument specifies how
.I pathname
should be opened, and acts as a superset of the
.IR flags
and
.IR mode
arguments to
.BR openat (2).
This argument is a pointer to a structure of the following form:
.PP
.in +4n
.EX
struct open_how {
u64 flags; /* O_* flags */
u64 mode; /* Mode for O_{CREAT,TMPFILE} */
u64 resolve; /* RESOLVE_* flags */
/* ... */
};.EE
.in
.PP
Any future extensions to
.BR openat2 ()
will be implemented as new fields appended to the above structure, with a zero value in a new field resulting in the kernel behaving as though that extension field was not present.
Therefore, the caller
.I must
zero-fill this structure on
initialization.
(See the "Extensibility" section of the
.B NOTES
for more detail on why this is necessary.)
.PP
The fields of the
.I open_how
structure are as follows:
.TP
.I flags
This field specifies
the file creation and file status flags to use when opening the file. All of the
.B O_*
flags defined for
.BR openat (2)
are valid
.BR openat2 ()
flag values.
.IP
Whereas
.BR openat (2)
ignores unknown bits in its
.I flags
argument,
.BR openat2 ()
returns an error if unknown or conflicting flags are specified in .IR how.flags .
.TP
.I mode
This field specifies the
mode for the new file, with identical semantics to the
.I mode
argument of
.BR openat (2).
.IP
Whereas
.BR openat (2)
ignores bits other than those in the range
.I 07777
in its
.I mode
argument,
.BR openat2 ()
returns an error if
.I how.mode
contains bits other than
.IR 07777 .
Similarly, an error is returned if
.BR openat2 ()
is called with a non-zero
.IR how.mode
and
.IR how.flags
does not contain
.BR O_CREAT
or
.BR O_TMPFILE .
.TP
.I resolve
This is a bit-mask of flags that modify the way in which
.B all
components of
.I pathname
will be resolved.
(See
.BR path_resolution (7)
for background information.)
.IP
The primary use case for these flags is to allow trusted programs to restrict how untrusted paths (or paths inside untrusted directories) are resolved. The full list of
.I resolve
flags is as follows:
.RS
.TP
.B RESOLVE_NO_XDEV
Disallow traversal of mount points during path resolution (including all bind mounts).
.IP
Applications that employ
this flag are encouraged to make its use configurable (unless it is used for a specific security purpose), as bind mounts are very widely used by end-users.
Setting this flag indiscriminately for all uses of
.BR openat2 ()
may result in spurious errors on previously-functional systems. .\" FIXME I find the "previously-functional systems" in the previous .\" sentence a little odd (since openat2() ia new sysycall), so I would .\" like to clarify a little...
.\" Are you referring to the scenario where someone might take an .\" existing application that uses openat() and replaces the uses .\" of openat() with openat2()? In which case, is it correct to .\" understand that you mean that one should not just indiscriminately .\" add the RESOLVE_NO_XDEV flag to all of the openat2() calls? .\" If I'm not on the right track, could you point me in the right .\" direction please.
.TP
.B RESOLVE_NO_SYMLINKS
Disallow resolution of symbolic links during path resolution.
This option implies
.BR RESOLVE_NO_MAGICLINKS .
.IP
If the trailing component (i.e., basename) of
.I pathname
is a symbolic link, and
.I how.flags
contains both
.BR O_PATH
and
.BR O_NOFOLLOW ,
then an
.B O_PATH
file descriptor referencing the symbolic link will be returned. .IP
Applications that employ
this flag are encouraged to make its use configurable (unless it is used for a specific security purpose), as symbolic links are very widely used by end-users.
Setting this flag indiscriminately for all uses of
.BR openat2 ()
may result in spurious errors on previously-functional systems. .TP
.B RESOLVE_NO_MAGICLINKS
Disallow all magic-link resolution during path resolution.
.IP
If the trailing component (i.e., basename) of
.I pathname
is a magic link, and
.I how.flags
contains both
.BR O_PATH
and
.BR O_NOFOLLOW ,
then an
.B O_PATH
file descriptor referencing the magic link will be returned.
.IP
Magic links are symbolic link-like objects that are most notably found in .BR proc (5)
(examples include
.IR /proc/[pid]/exe
and
.IR /proc/[pid]/fd/* ).
Due to the potential danger of unknowingly opening these magic links, it may be
preferable for users to disable their resolution entirely.
.\" FIXME: what specific details in symlink(7) are being referred .\" by the following sentence? It's not clear.
(See
.BR symlink (7)
for more details.)
.TP
.B RESOLVE_BENEATH
Do not permit the path resolution to succeed if any component of the resolution is not a descendant of the directory indicated by
.IR dirfd .
This causes absolute symbolic links (and absolute values of
.IR pathname )
to be rejected.
.IP
Currently, this flag also disables magic-link resolution.
However, this may change in the future.
Therefore, to ensure that magic links are not resolved,
the caller should explicitly specify
.BR RESOLVE_NO_MAGICLINKS .
.TP
.B RESOLVE_IN_ROOT
Treat the directory referred to by
.I dirfd
as the root directory while resolving
.IR pathname .
.\" FIXME I found the following hard to understand (in particular, the .\" meaning of "scoped" is unclear) , and reworded as below. Is it okay?
.\" Absolute symbolic links and ".." path components will be scoped to
.\" .IR dirfd .
Absolute symbolic links are interpreted relative to
.IR dirfd .
If a prefix component of
.I pathname
equates to
.IR dirfd ,
then an immediately following
.IR ..
component likewise equates to
.IR dirfd
(just as
.I /..
is traditionally equivalent to
.IR / ).
If
.I pathname
is an absolute path, it is also interpreted relative to
.IR dirfd .
.IP
The effect of this flag is as though the calling process had used .BR chroot (2)
to (temporarily) modify its root directory (to the directory referred to by
.IR dirfd ).
However, unlike
.BR chroot (2)
(which changes the filesystem root permanently for a process),
.B RESOLVE_IN_ROOT
allows a program to efficiently restrict path resolution on a per-open basis. .\" FIXME The next piece is unclear (to me). What kind of ".." escape .\" attempts does chroot() not detect that RESOLVE_IN_ROOT does? The
.B RESOLVE_IN_ROOT
flag also has several hardening features
(such as detecting escape attempts during
.I ".."
resolution) which
.BR chroot (2)
does not.
.IP
Currently, this flag also disables magic-link resolution.
However, this may change in the future.
Therefore, to ensure that magic links are not resolved,
the caller should explicitly specify
.BR RESOLVE_NO_MAGICLINKS .
.RE
.IP
If any bits other than those listed above are set in
.IR how.resolve ,
an error is returned.
.SH RETURN VALUE
On success, a new file descriptor is returned.
On error, \-1 is returned, and
.I errno
is set appropriately.
.SH ERRORS
The set of errors returned by
.BR openat2 ()
includes all of the errors returned by
.BR openat (2),
as well as the following additional errors:
.TP
.B E2BIG
An extension that this kernel does not support was specified in .IR how .
(See the "Extensibility" section of
.B NOTES
for more detail on how extensions are handled.)
.TP
.B EAGAIN
.I how.resolve
contains either
.BR RESOLVE_IN_ROOT
or
.BR RESOLVE_BENEATH ,
and the kernel could not ensure that a ".." component didn't escape (due to a race condition or potential attack).
The caller may choose to retry the
.BR openat2 ()
call.
.TP
.B EINVAL
An unknown flag or invalid value was specified in
.IR how .
.TP
.B EINVAL
.I mode
is non-zero, but
.I how.flags
does not contain
.BR O_CREAT
or
.BR O_TMPFILE .
.TP
.B EINVAL
.I size
was smaller than any known version of
.IR "struct open_how" .
.TP
.B ELOOP
.I how.resolve
contains
.BR RESOLVE_NO_SYMLINKS ,
and one of the path components was a symbolic link (or magic link). .TP
.B ELOOP
.I how.resolve
contains
.BR RESOLVE_NO_MAGICLINKS ,
and one of the path components was a magic link.
.TP
.B EXDEV
.I how.resolve
contains either
.BR RESOLVE_IN_ROOT
or
.BR RESOLVE_BENEATH ,
and an escape from the root during path resolution was detected. .TP
.B EXDEV
.I how.resolve
contains
.BR RESOLVE_NO_XDEV ,
and a path component crosses a mount point.
.SH VERSIONS
.BR openat2 ()
first appeared in Linux 5.6.
.SH CONFORMING TO
This system call is Linux-specific.
.PP
The semantics of
.B RESOLVE_BENEATH
were modeled after FreeBSD's
.BR O_BENEATH .
.SH NOTES
Glibc does not provide a wrapper for this system call; call it using .BR syscall (2).
.\"
.SS Extensibility
In order to allow for future extensibility,
.BR openat2 ()
requires the user-space application to specify the size of the
.I open_how
structure that it is passing.
By providing this information, it is possible for
.BR openat2 ()
to provide both forwards- and backwards-compatibility, with
.I size
acting as an implicit version number.
(Because new extension fields will always
be appended, the structure size will always increase.)
This extensibility design is very similar to other system calls such as .BR perf_setattr (2),
.BR perf_event_open (2),
and
.BR clone3 (2).
.PP
If we let
.I usize
be the size of the structure as specified by the user-space application, and .I ksize
be the size of the structure which the kernel supports, then there are three cases to consider:
.IP \(bu 2
If
.IR ksize
equals
.IR usize ,
then there is no version mismatch and
.I how
can be used verbatim.
.IP \(bu
If
.IR ksize
is larger than
.IR usize ,
then there are some extension fields that the kernel supports which the user-space application
is unaware of.
Because a zero value in any added extension field signifies a no-op, the kernel
treats all of the extension fields not provided by the user-space application as having zero values.
This provides backwards-compatibility.
.IP \(bu
If
.IR ksize
is smaller than
.IR usize ,
then there are some extension fields which the user-space application is aware of but which the kernel does not support.
Because any extension field must have its zero values signify a no-op, the kernel can
safely ignore the unsupported extension fields if they are all-zero. If any unsupported extension fields are non-zero, then \-1 is returned and .I errno
is set to
.BR E2BIG .
This provides forwards-compatibility.
.PP
Because the definition of
.I struct open_how
may change in the future (with new fields being added when system headers are updated), user-space applications should zero-fill
.I struct open_how
to ensure that recompiling the program with new headers will not result in spurious errors at runtime.
The simplest way is to use a designated
initializer:
.PP
.in +4n
.EX
struct open_how how = { .flags = O_RDWR,
.resolve = RESOLVE_IN_ROOT };
.EE
.in
.PP
or explicitly using
.BR memset (3)
or similar:
.PP
.in +4n
.EX
struct open_how how;memset(&how, 0, sizeof(how));
how.flags = O_RDWR;
how.resolve = RESOLVE_IN_ROOT;
.EE
.in
.PP
A user-space application that wishes to determine which extensions the running kernel supports can do so by conducting a binary search on .IR size
with a structure which has every byte nonzero (to find the largest value which doesn't produce an error of
.BR E2BIG ).
.SH SEE ALSO
.BR openat (2),
.BR path_resolution (7),
.BR symlink (7)
Signature · 3 lines
Michael Kerrisk Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/ Linux/UNIX System Programming Training: http://man7.org/training/
Re: [PATCH man-pages v2 2/2] openat2.2: document new openat2(2) syscall
Quoted 16 lines · Michael Kerrisk (man-pages)
Hello Aleksa,
Quoted 13 lines · Aleksa Sarai
Rather than trying to merge the new syscall documentation into open.2 (which would probably result in the man-page being incomprehensible), instead the new syscall gets its own dedicated page with links between open(2) and openat2(2) to avoid duplicating information such as the list of O_* flags or common errors.
In addition to describing all of the key flags, information about the extensibility design is provided so that users can better understand why they need to pass
sizeof(struct open_how)and how their programs will work across kernels. After some discussions with David Laight, I also included explicit instructions to zero the structure to avoid issues when recompiling with new headers.
Quoted 6 lines
Quoted 1 line
Signed-off-by Aleksa Sarai <[email protected]>Thanks. I've applied this patch, but also done quite a lot of editing of the page. The current draft is below (and also pushed to Git). Could I ask you to review the page, to see if I injected any error during my edits.
Looks good to me.
Quoted 3 lines
In addition, I've added a number of FIXMEs in comments in the page source. Can you please check these, and let me know your thoughts.
Will do, see below.
Quoted 10 lines
.\" FIXME I find the "previously-functional systems" in the previous .\" sentence a little odd (since
openat2()ia new sysycall), so I would .\" like to clarify a little...
.\" Are you referring to the scenario where someone might take an .\" existing application that usesopenat()and replaces the uses .\" ofopenat()withopenat2()? In which case, is it correct to .\" understand that you mean that one should not just indiscriminately .\" add theRESOLVE_NO_XDEVflag to all of theopenat2()calls? .\" If I'm not on the right track, could you point me in the right .\" direction please.
This is mostly meant as a warning to hopefully avoid applications because the developer didn't realise that system paths may contain symlinks or bind-mounts. For an application which has switched to openat2() and then uses RESOLVE_NO_SYMLINKS for a non-security reason, it's possible that on some distributions (or future versions of a distribution) that their application will stop working because a system path suddenly contains a symlink or is a bind-mount.
This was a concern which was brought up on LWN some time ago. If you can think of a phrasing that makes this more clear, I'd appreciate it.
Quoted 2 lines
.\" FIXME: what specific details in symlink(7) are being referred .\" by the following sentence? It's not clear.
The section on magic-links, but you're right that the sentence ordering is a bit odd. It should probably go after the first sentence.
Quoted 4 lines
.\" FIXME I found the following hard to understand (in particular, the .\" meaning of "scoped" is unclear) , and reworded as below. Is it okay?
.\" Absolute symbolic links and ".." path components will be scoped to
.\" .IR dirfd .
Scoped does broadly mean "interpreted relative to", though the difference is mainly that when I said scoped it's meant to be more of an assertive claim ("the kernel promises to always treat this path inside dirfd"). But "interpreted relative to" is a clearer way of phrasing the semantics, so I'm okay with this change.
Quoted 2 lines
.\" FIXME The next piece is unclear (to me). What kind of ".." escape .\" attempts does
chroot()not detect thatRESOLVE_IN_ROOTdoes?
If the root is moved, you can escape from a chroot(2). But this sentence might not really belong in a man-page since it's describing (important) aspects of the implementation and not the semantics.
Signature · 4 lines
Aleksa Sarai Senior Software Engineer (Containers) SUSE Linux GmbH <https://www.cyphar.com/>
Re: [PATCH man-pages v2 2/2] openat2.2: document new openat2(2) syscall
Hello Aleksa,
Quoted 51 lines · Aleksa Sarai
Quoted 16 lines · Michael Kerrisk (man-pages)
Hello Aleksa,
Quoted 13 lines · Aleksa Sarai
Rather than trying to merge the new syscall documentation into open.2 (which would probably result in the man-page being incomprehensible), instead the new syscall gets its own dedicated page with links between open(2) and openat2(2) to avoid duplicating information such as the list of O_* flags or common errors.
In addition to describing all of the key flags, information about the extensibility design is provided so that users can better understand why they need to pass
sizeof(struct open_how)and how their programs will work across kernels. After some discussions with David Laight, I also included explicit instructions to zero the structure to avoid issues when recompiling with new headers.Quoted 6 lines
Quoted 1 line
Signed-off-by Aleksa Sarai <[email protected]>Thanks. I've applied this patch, but also done quite a lot of editing of the page. The current draft is below (and also pushed to Git). Could I ask you to review the page, to see if I injected any error during my edits.
Looks good to me.
Quoted 3 lines
In addition, I've added a number of FIXMEs in comments in the page source. Can you please check these, and let me know your thoughts.
Will do, see below.
Quoted 10 lines
.\" FIXME I find the "previously-functional systems" in the previous .\" sentence a little odd (since
openat2()ia new sysycall), so I would .\" like to clarify a little...
.\" Are you referring to the scenario where someone might take an .\" existing application that usesopenat()and replaces the uses .\" ofopenat()withopenat2()? In which case, is it correct to .\" understand that you mean that one should not just indiscriminately .\" add theRESOLVE_NO_XDEVflag to all of theopenat2()calls? .\" If I'm not on the right track, could you point me in the right .\" direction please.This is mostly meant as a warning to hopefully avoid applications because the developer didn't realise that system paths may contain symlinks or bind-mounts. For an application which has switched to
openat2()and then usesRESOLVE_NO_SYMLINKSfor a non-security reason, it's possible that on some distributions (or future versions of a distribution) that their application will stop working because a system path suddenly contains a symlink or is a bind-mount.
Quoted 2 lines
This was a concern which was brought up on LWN some time ago. If you can think of a phrasing that makes this more clear, I'd appreciate it.
Thanks. I've made the text:
Applications that employ the RESOLVE_NO_XDEV flag are encouraged to make its use configurable (unless it is used for a specific security purpose), as bind
mounts are widely used by end-users. Setting this
flag indiscriminately—i.e., for purposes not specif‐ ically related to security—for all uses of openat2() may result in spurious errors on previously-func‐ tional systems. This may occur if, for example, a system pathname that is used by an application is modified (e.g., in a new distribution release) so that a pathname component (now) contains a bind mount.
Okay?
Quoted 5 lines
Quoted 2 lines
.\" FIXME: what specific details in symlink(7) are being referred .\" by the following sentence? It's not clear.
The section on magic-links, but you're right that the sentence ordering is a bit odd. It should probably go after the first sentence.
I must admit that I'm still confused. There's only the briefest of mentions of magic links in symlink(7). Perhaps that needs to be fixed?
And, while I think of it, the text just preceding that FIXME says:
Due to the potential danger of unknowingly opening these magic links, it may be preferable for users to disable their resolution entirely.
This sentence reads a little strangely. Could you please give me some concrete examples, and I will try rewording that sentence a bit.
Quoted 10 lines
Quoted 4 lines
.\" FIXME I found the following hard to understand (in particular, the .\" meaning of "scoped" is unclear) , and reworded as below. Is it okay?
.\" Absolute symbolic links and ".." path components will be scoped to
.\" .IR dirfd .Scoped does broadly mean "interpreted relative to", though the difference is mainly that when I said scoped it's meant to be more of an assertive claim ("the kernel promises to always treat this path inside dirfd"). But "interpreted relative to" is a clearer way of phrasing the semantics, so I'm okay with this change.
Okay.
Quoted 6 lines
Quoted 2 lines
.\" FIXME The next piece is unclear (to me). What kind of ".." escape .\" attempts does
chroot()not detect thatRESOLVE_IN_ROOTdoes?If the root is moved, you can escape from a chroot(2). But this sentence might not really belong in a man-page since it's describing (important) aspects of the implementation and not the semantics.
So, should I just remove the sentence?
Thanks,
Michael
Signature · 3 lines
Michael Kerrisk Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/ Linux/UNIX System Programming Training: http://man7.org/training/
Re: [PATCH man-pages v2 2/2] openat2.2: document new openat2(2) syscall
Hi Aleksa,
Ping!
Cheers,
Michael
Quoted 124 lines · Michael Kerrisk (man-pages)
Hello Aleksa,
Quoted 51 lines · Aleksa Sarai
Quoted 16 lines · Michael Kerrisk (man-pages)
Hello Aleksa,
Quoted 13 lines · Aleksa Sarai
Rather than trying to merge the new syscall documentation into open.2 (which would probably result in the man-page being incomprehensible), instead the new syscall gets its own dedicated page with links between open(2) and openat2(2) to avoid duplicating information such as the list of O_* flags or common errors.
In addition to describing all of the key flags, information about the extensibility design is provided so that users can better understand why they need to pass
sizeof(struct open_how)and how their programs will work across kernels. After some discussions with David Laight, I also included explicit instructions to zero the structure to avoid issues when recompiling with new headers.Quoted 6 lines
Quoted 1 line
Signed-off-by Aleksa Sarai <[email protected]>Thanks. I've applied this patch, but also done quite a lot of editing of the page. The current draft is below (and also pushed to Git). Could I ask you to review the page, to see if I injected any error during my edits.
Looks good to me.
Quoted 3 lines
In addition, I've added a number of FIXMEs in comments in the page source. Can you please check these, and let me know your thoughts.
Will do, see below.
Quoted 10 lines
.\" FIXME I find the "previously-functional systems" in the previous .\" sentence a little odd (since
openat2()ia new sysycall), so I would .\" like to clarify a little...
.\" Are you referring to the scenario where someone might take an .\" existing application that usesopenat()and replaces the uses .\" ofopenat()withopenat2()? In which case, is it correct to .\" understand that you mean that one should not just indiscriminately .\" add theRESOLVE_NO_XDEVflag to all of theopenat2()calls? .\" If I'm not on the right track, could you point me in the right .\" direction please.This is mostly meant as a warning to hopefully avoid applications because the developer didn't realise that system paths may contain symlinks or bind-mounts. For an application which has switched to
openat2()and then usesRESOLVE_NO_SYMLINKSfor a non-security reason, it's possible that on some distributions (or future versions of a distribution) that their application will stop working because a system path suddenly contains a symlink or is a bind-mount.Quoted 2 lines
This was a concern which was brought up on LWN some time ago. If you can think of a phrasing that makes this more clear, I'd appreciate it.
Thanks. I've made the text:
Applications that employ the
RESOLVE_NO_XDEVflag are encouraged to make its use configurable (unless it is used for a specific security purpose), as bind
mounts are widely used by end-users. Setting this
flag indiscriminately—i.e., for purposes not specif‐ ically related to security—for all uses ofopenat2()may result in spurious errors on previously-func‐ tional systems. This may occur if, for example, a system pathname that is used by an application is modified (e.g., in a new distribution release) so that a pathname component (now) contains a bind mount.Okay?
Quoted 5 lines
Quoted 2 lines
.\" FIXME: what specific details in symlink(7) are being referred .\" by the following sentence? It's not clear.
The section on magic-links, but you're right that the sentence ordering is a bit odd. It should probably go after the first sentence.
I must admit that I'm still confused. There's only the briefest of mentions of magic links in symlink(7). Perhaps that needs to be fixed?
And, while I think of it, the text just preceding that FIXME says:
Due to the potential danger of unknowingly opening these magic links, it may be preferable for users to disable their resolution entirely.
This sentence reads a little strangely. Could you please give me some concrete examples, and I will try rewording that sentence a bit.
Quoted 10 lines
Quoted 4 lines
.\" FIXME I found the following hard to understand (in particular, the .\" meaning of "scoped" is unclear) , and reworded as below. Is it okay?
.\" Absolute symbolic links and ".." path components will be scoped to
.\" .IR dirfd .Scoped does broadly mean "interpreted relative to", though the difference is mainly that when I said scoped it's meant to be more of an assertive claim ("the kernel promises to always treat this path inside dirfd"). But "interpreted relative to" is a clearer way of phrasing the semantics, so I'm okay with this change.
Okay.
Quoted 6 lines
Quoted 2 lines
.\" FIXME The next piece is unclear (to me). What kind of ".." escape .\" attempts does
chroot()not detect thatRESOLVE_IN_ROOTdoes?If the root is moved, you can escape from a chroot(2). But this sentence might not really belong in a man-page since it's describing (important) aspects of the implementation and not the semantics.
So, should I just remove the sentence?
Thanks,
Michael
--
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/
Signature · 3 lines
Michael Kerrisk Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/ Linux/UNIX System Programming Training: http://man7.org/training/
Re: [PATCH man-pages v2 2/2] openat2.2: document new openat2(2) syscall
In reply to Michael Kerrisk (man-pages)Sorry, I could've sworn I responded when you posted this -- comments below. And sorry for not getting back to you before the 5.06 release.
Quoted 68 lines · Michael Kerrisk (man-pages)
Quoted 49 lines · Aleksa Sarai
Quoted 14 lines · Michael Kerrisk (man-pages)
Quoted 13 lines · Aleksa Sarai
Rather than trying to merge the new syscall documentation into open.2 (which would probably result in the man-page being incomprehensible), instead the new syscall gets its own dedicated page with links between open(2) and openat2(2) to avoid duplicating information such as the list of O_* flags or common errors.
In addition to describing all of the key flags, information about the extensibility design is provided so that users can better understand why they need to pass
sizeof(struct open_how)and how their programs will work across kernels. After some discussions with David Laight, I also included explicit instructions to zero the structure to avoid issues when recompiling with new headers.Quoted 6 lines
Quoted 1 line
Signed-off-by Aleksa Sarai <[email protected]>Thanks. I've applied this patch, but also done quite a lot of editing of the page. The current draft is below (and also pushed to Git). Could I ask you to review the page, to see if I injected any error during my edits.
Looks good to me.
Quoted 3 lines
In addition, I've added a number of FIXMEs in comments in the page source. Can you please check these, and let me know your thoughts.
Will do, see below.
Quoted 10 lines
.\" FIXME I find the "previously-functional systems" in the previous .\" sentence a little odd (since
openat2()ia new sysycall), so I would .\" like to clarify a little...
.\" Are you referring to the scenario where someone might take an .\" existing application that usesopenat()and replaces the uses .\" ofopenat()withopenat2()? In which case, is it correct to .\" understand that you mean that one should not just indiscriminately .\" add theRESOLVE_NO_XDEVflag to all of theopenat2()calls? .\" If I'm not on the right track, could you point me in the right .\" direction please.This is mostly meant as a warning to hopefully avoid applications because the developer didn't realise that system paths may contain symlinks or bind-mounts. For an application which has switched to
openat2()and then usesRESOLVE_NO_SYMLINKSfor a non-security reason, it's possible that on some distributions (or future versions of a distribution) that their application will stop working because a system path suddenly contains a symlink or is a bind-mount.Quoted 2 lines
This was a concern which was brought up on LWN some time ago. If you can think of a phrasing that makes this more clear, I'd appreciate it.
Thanks. I've made the text:
Applications that employ the
RESOLVE_NO_XDEVflag are encouraged to make its use configurable (unless it is used for a specific security purpose), as bind
mounts are widely used by end-users. Setting this
flag indiscriminately—i.e., for purposes not specif‐ ically related to security—for all uses ofopenat2()may result in spurious errors on previously-func‐ tional systems. This may occur if, for example, a system pathname that is used by an application is modified (e.g., in a new distribution release) so that a pathname component (now) contains a bind mount.
Quoted 1 line
Okay?
Yup, and the same text should be used for the same warning I gave for RESOLVE_NO_SYMLINKS (for the same reason, because system paths may switch to symlinks -- the prime example being what Arch Linux did several years ago).
Quoted 8 lines
Quoted 5 lines
Quoted 2 lines
.\" FIXME: what specific details in symlink(7) are being referred .\" by the following sentence? It's not clear.
The section on magic-links, but you're right that the sentence ordering is a bit odd. It should probably go after the first sentence.
I must admit that I'm still confused. There's only the briefest of mentions of magic links in symlink(7). Perhaps that needs to be fixed?
It wouldn't hurt to add a longer description of magic-links in symlink(7). I'll send you a small patch to beef up the description (I had planned to include a longer rewrite with the O_EMPTYPATH patches but those require quite a bit more work to land).
Quoted 8 lines
And, while I think of it, the text just preceding that FIXME says:
Due to the potential danger of unknowingly opening these magic links, it may be preferable for users to disable their resolution entirely.
This sentence reads a little strangely. Could you please give me some concrete examples, and I will try rewording that sentence a bit.
The primary example is that certain files (such as tty devices) are best not opened by an unsuspecting program (if you do not have a controlling TTY, and you open such a file that console becomes your controlling TTY unless you use O_NOCTTY).
But more generally, magic-links allow programs to be "beamed" all over the system (bypassing ordinary mount namespace restrictions). Since they are fairly rarely used intentionally by most programs, this is more of a tip to programmers that maybe they should play it safe and disallow magic-links unless they are expecting to have to use them.
Quoted 13 lines
Quoted 10 lines
Quoted 4 lines
.\" FIXME I found the following hard to understand (in particular, the .\" meaning of "scoped" is unclear) , and reworded as below. Is it okay?
.\" Absolute symbolic links and ".." path components will be scoped to
.\" .IR dirfd .Scoped does broadly mean "interpreted relative to", though the difference is mainly that when I said scoped it's meant to be more of an assertive claim ("the kernel promises to always treat this path inside dirfd"). But "interpreted relative to" is a clearer way of phrasing the semantics, so I'm okay with this change.
Okay.
Quoted 8 lines
Quoted 6 lines
Quoted 2 lines
.\" FIXME The next piece is unclear (to me). What kind of ".." escape .\" attempts does
chroot()not detect thatRESOLVE_IN_ROOTdoes?If the root is moved, you can escape from a chroot(2). But this sentence might not really belong in a man-page since it's describing (important) aspects of the implementation and not the semantics.
So, should I just remove the sentence?
Yup, sounds reasonable.
Signature · 4 lines
Aleksa Sarai Senior Software Engineer (Containers) SUSE Linux GmbH <https://www.cyphar.com/>
Re: [PATCH man-pages v2 2/2] openat2.2: document new openat2(2) syscall
Hello Aleksa,
Quoted 2 lines · Aleksa Sarai
Sorry, I could've sworn I responded when you posted this -- comments below. And sorry for not getting back to you before the 5.06 release.
No worries and ahanks for your feedback below.
[...]
Quoted 37 lines
Quoted 37 lines
Quoted 19 lines
Quoted 10 lines
.\" FIXME I find the "previously-functional systems" in the previous .\" sentence a little odd (since
openat2()ia new sysycall), so I would .\" like to clarify a little...
.\" Are you referring to the scenario where someone might take an .\" existing application that usesopenat()and replaces the uses .\" ofopenat()withopenat2()? In which case, is it correct to .\" understand that you mean that one should not just indiscriminately .\" add theRESOLVE_NO_XDEVflag to all of theopenat2()calls? .\" If I'm not on the right track, could you point me in the right .\" direction please.This is mostly meant as a warning to hopefully avoid applications because the developer didn't realise that system paths may contain symlinks or bind-mounts. For an application which has switched to
openat2()and then usesRESOLVE_NO_SYMLINKSfor a non-security reason, it's possible that on some distributions (or future versions of a distribution) that their application will stop working because a system path suddenly contains a symlink or is a bind-mount.Quoted 2 lines
This was a concern which was brought up on LWN some time ago. If you can think of a phrasing that makes this more clear, I'd appreciate it.
Thanks. I've made the text:
Applications that employ the
RESOLVE_NO_XDEVflag are encouraged to make its use configurable (unless it is used for a specific security purpose), as bind
mounts are widely used by end-users. Setting this
flag indiscriminately—i.e., for purposes not specif‐ ically related to security—for all uses ofopenat2()may result in spurious errors on previously-func‐ tional systems. This may occur if, for example, a system pathname that is used by an application is modified (e.g., in a new distribution release) so that a pathname component (now) contains a bind mount.
Quoted 3 lines
Quoted 1 line
Okay?
Yup,
Thanks.
Quoted 4 lines
and the same text should be used for the same warning I gave for
RESOLVE_NO_SYMLINKS(for the same reason, because system paths may switch to symlinks -- the prime example being what Arch Linux did several years ago).
Okay -- I added similar text to RESOLVE_NO_SYMLINKS.
Quoted 6 lines
Quoted 6 lines
Quoted 5 lines
Quoted 2 lines
.\" FIXME: what specific details in symlink(7) are being referred .\" by the following sentence? It's not clear.
The section on magic-links, but you're right that the sentence ordering is a bit odd. It should probably go after the first sentence.
Quoted 7 lines
Quoted 2 lines
I must admit that I'm still confused. There's only the briefest of mentions of magic links in symlink(7). Perhaps that needs to be fixed?
It wouldn't hurt to add a longer description of magic-links in symlink(7). I'll send you a small patch to beef up the description (I had planned to include a longer rewrite with the
O_EMPTYPATHpatches but those require quite a bit more work to land).
That would be great. Thank you!
Quoted 14 lines
Quoted 8 lines
And, while I think of it, the text just preceding that FIXME says:
Due to the potential danger of unknowingly opening these magic links, it may be preferable for users to disable their resolution entirely.
This sentence reads a little strangely. Could you please give me some concrete examples, and I will try rewording that sentence a bit.
The primary example is that certain files (such as tty devices) are best not opened by an unsuspecting program (if you do not have a controlling TTY, and you open such a file that console becomes your controlling TTY unless you use
O_NOCTTY).
Quoted 5 lines
But more generally, magic-links allow programs to be "beamed" all over the system (bypassing ordinary mount namespace restrictions). Since they are fairly rarely used intentionally by most programs, this is more of a tip to programmers that maybe they should play it safe and disallow magic-links unless they are expecting to have to use them.
I've reworked the text on RESOLVE_NO_MAGICLINKS substantially:
RESOLVE_NO_MAGICLINKS
Disallow all magic-link resolution during path reso‐ lution.
Magic links are symbolic link-like objects that are most notably found in proc(5); examples include
/proc/[pid]/exe and /proc/[pid]/fd/*. (See sym‐
link(7) for more details.)
Unknowingly opening magic links can be risky for some applications. Examples of such risks include the following:
· If the process opening a pathname is a controlling
process that currently has no controlling terminal (see credentials(7)), then opening a magic link inside /proc/[pid]/fd that happens to refer to a terminal would cause the process to acquire a con‐ trolling terminal.
· In a containerized environment, a magic link
inside /proc may refer to an object outside the container, and thus may provide a means to escape from the container.
[The above example derives from https://lwn.net/Articles/796868/]
Because of such risks, an application may prefer to
disable magic link resolution using the
RESOLVE_NO_MAGICLINKS flag.
If the trailing component (i.e., basename) of path‐ name is a magic link, and how.flags contains both O_PATH and O_NOFOLLOW, then an O_PATH file descrip‐ tor referencing the magic link will be returned.
How does the above look?
Also, regarding the last paragraph, I have a question. The text doesn't seem quite to relate to the rest of the discussion. Should it be saying something like:
If the trailing component (i.e., basename) of pathname is a magic link, **how.resolve contains RESOLVE_NO_MAGICLINKS,**
and how.flags contains both O_PATH and O_NOFOLLOW, then an O_PATH file descriptor referencing the magic link will be returned.
?
[...]
Quoted 10 lines
Quoted 8 lines
Quoted 6 lines
Quoted 2 lines
.\" FIXME The next piece is unclear (to me). What kind of ".." escape .\" attempts does
chroot()not detect thatRESOLVE_IN_ROOTdoes?If the root is moved, you can escape from a chroot(2). But this sentence might not really belong in a man-page since it's describing (important) aspects of the implementation and not the semantics.
So, should I just remove the sentence?
Yup, sounds reasonable.
Done.
Thanks,
Michael
Signature · 3 lines
Michael Kerrisk Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/ Linux/UNIX System Programming Training: http://man7.org/training/
Re: [PATCH man-pages v2 2/2] openat2.2: document new openat2(2) syscall
Quoted 41 lines · Michael Kerrisk (man-pages)
Quoted 37 lines
Quoted 37 lines
Quoted 19 lines
Quoted 10 lines
.\" FIXME I find the "previously-functional systems" in the previous .\" sentence a little odd (since
openat2()ia new sysycall), so I would .\" like to clarify a little...
.\" Are you referring to the scenario where someone might take an .\" existing application that usesopenat()and replaces the uses .\" ofopenat()withopenat2()? In which case, is it correct to .\" understand that you mean that one should not just indiscriminately .\" add theRESOLVE_NO_XDEVflag to all of theopenat2()calls? .\" If I'm not on the right track, could you point me in the right .\" direction please.This is mostly meant as a warning to hopefully avoid applications because the developer didn't realise that system paths may contain symlinks or bind-mounts. For an application which has switched to
openat2()and then usesRESOLVE_NO_SYMLINKSfor a non-security reason, it's possible that on some distributions (or future versions of a distribution) that their application will stop working because a system path suddenly contains a symlink or is a bind-mount.Quoted 2 lines
This was a concern which was brought up on LWN some time ago. If you can think of a phrasing that makes this more clear, I'd appreciate it.
Thanks. I've made the text:
Applications that employ the
RESOLVE_NO_XDEVflag are encouraged to make its use configurable (unless it is used for a specific security purpose), as bind
mounts are widely used by end-users. Setting this
flag indiscriminately—i.e., for purposes not specif‐ ically related to security—for all uses ofopenat2()may result in spurious errors on previously-func‐ tional systems. This may occur if, for example, a system pathname that is used by an application is modified (e.g., in a new distribution release) so that a pathname component (now) contains a bind mount.Quoted 3 lines
Quoted 1 line
Okay?
Yup,
Quoted 8 lines
Thanks.
Quoted 4 lines
and the same text should be used for the same warning I gave for
RESOLVE_NO_SYMLINKS(for the same reason, because system paths may switch to symlinks -- the prime example being what Arch Linux did several years ago).Okay -- I added similar text to
RESOLVE_NO_SYMLINKS.
Much appreciated.
Quoted 9 lines
Quoted 9 lines
Quoted 8 lines
Quoted 5 lines
Quoted 2 lines
.\" FIXME: what specific details in symlink(7) are being referred .\" by the following sentence? It's not clear.
The section on magic-links, but you're right that the sentence ordering is a bit odd. It should probably go after the first sentence.
I must admit that I'm still confused. There's only the briefest of mentions of magic links in symlink(7). Perhaps that needs to be fixed?
Quoted 6 lines
Quoted 4 lines
It wouldn't hurt to add a longer description of magic-links in symlink(7). I'll send you a small patch to beef up the description (I had planned to include a longer rewrite with the
O_EMPTYPATHpatches but those require quite a bit more work to land).That would be great. Thank you!
I'll cook something up later this week.
Quoted 54 lines
Quoted 14 lines
Quoted 8 lines
And, while I think of it, the text just preceding that FIXME says:
Due to the potential danger of unknowingly opening these magic links, it may be preferable for users to disable their resolution entirely.
This sentence reads a little strangely. Could you please give me some concrete examples, and I will try rewording that sentence a bit.
The primary example is that certain files (such as tty devices) are best not opened by an unsuspecting program (if you do not have a controlling TTY, and you open such a file that console becomes your controlling TTY unless you use
O_NOCTTY).Quoted 5 lines
But more generally, magic-links allow programs to be "beamed" all over the system (bypassing ordinary mount namespace restrictions). Since they are fairly rarely used intentionally by most programs, this is more of a tip to programmers that maybe they should play it safe and disallow magic-links unless they are expecting to have to use them.
I've reworked the text on
RESOLVE_NO_MAGICLINKSsubstantially:
RESOLVE_NO_MAGICLINKS
Disallow all magic-link resolution during path reso‐ lution.Magic links are symbolic link-like objects that are most notably found in proc(5); examples include
/proc/[pid]/exe and /proc/[pid]/fd/*. (See sym‐
link(7) for more details.)Unknowingly opening magic links can be risky for some applications. Examples of such risks include the following:
· If the process opening a pathname is a controlling
process that currently has no controlling terminal (see credentials(7)), then opening a magic link inside /proc/[pid]/fd that happens to refer to a terminal would cause the process to acquire a con‐ trolling terminal.· In a containerized environment, a magic link
inside /proc may refer to an object outside the container, and thus may provide a means to escape from the container.[The above example derives from https://lwn.net/Articles/796868/]
Because of such risks, an application may prefer to
disable magic link resolution using the
RESOLVE_NO_MAGICLINKSflag.
Quoted 6 lines
If the trailing component (i.e., basename) of path‐ name is a magic link, and how.flags contains both
O_PATHandO_NOFOLLOW, then anO_PATHfile descrip‐ tor referencing the magic link will be returned.How does the above look?
The changes look correct, though you could end up going through procfs even if you weren't resolving a path inside proc directly (since you can bind-mount symlinks or have a symlink to procfs). But I'm not sure if it's necessary to outline all the ways a program could be tricked into doing something unintended.
Quoted 10 lines
Also, regarding the last paragraph, I have a question. The text doesn't seem quite to relate to the rest of the discussion. Should it be saying something like:
If the trailing component (i.e., basename) of pathname is a magic link, **how.resolve contains RESOLVE_NO_MAGICLINKS,**
and how.flags contains bothO_PATHandO_NOFOLLOW, then anO_PATHfile descriptor referencing the magic link will be returned.?
Yes, that is what I meant to write -- and I believe that the RESOLVE_NO_SYMLINKS section is missing similar text in the second paragraph (except it should refer to RESOLVE_NO_SYMLINKS, obviously).
Thanks!
Signature · 4 lines
Aleksa Sarai Senior Software Engineer (Containers) SUSE Linux GmbH <https://www.cyphar.com/>
Re: [PATCH man-pages v2 2/2] openat2.2: document new openat2(2) syscall
Hello Aleksa,
On Tue, 14 Apr 2020 at 12:35, Aleksa Sarai <[email protected]> wrote:
[...]
Quoted 11 lines
Quoted 9 lines
Quoted 7 lines
Quoted 2 lines
I must admit that I'm still confused. There's only the briefest of mentions of magic links in symlink(7). Perhaps that needs to be fixed?
It wouldn't hurt to add a longer description of magic-links in symlink(7). I'll send you a small patch to beef up the description (I had planned to include a longer rewrite with the
O_EMPTYPATHpatches but those require quite a bit more work to land).That would be great. Thank you!
I'll cook something up later this week.
Thank you!
[...]
Quoted 38 lines
Quoted 38 lines
I've reworked the text on
RESOLVE_NO_MAGICLINKSsubstantially:
RESOLVE_NO_MAGICLINKS
Disallow all magic-link resolution during path reso‐ lution.Magic links are symbolic link-like objects that are most notably found in proc(5); examples include
/proc/[pid]/exe and /proc/[pid]/fd/*. (See sym‐
link(7) for more details.)Unknowingly opening magic links can be risky for some applications. Examples of such risks include the following:
· If the process opening a pathname is a controlling
process that currently has no controlling terminal (see credentials(7)), then opening a magic link inside /proc/[pid]/fd that happens to refer to a terminal would cause the process to acquire a con‐ trolling terminal.· In a containerized environment, a magic link
inside /proc may refer to an object outside the container, and thus may provide a means to escape from the container.[The above example derives from https://lwn.net/Articles/796868/]
Because of such risks, an application may prefer to
disable magic link resolution using the
RESOLVE_NO_MAGICLINKSflag.If the trailing component (i.e., basename) of path‐ name is a magic link, and how.flags contains both
O_PATHandO_NOFOLLOW, then anO_PATHfile descrip‐ tor referencing the magic link will be returned.
Quoted 7 lines
Quoted 1 line
How does the above look?
The changes look correct, though you could end up going through procfs even if you weren't resolving a path inside proc directly (since you can bind-mount symlinks or have a symlink to procfs). But I'm not sure if it's necessary to outline all the ways a program could be tricked into doing something unintended.
Yes, indeed. These paragraphs are merely intended to give the reader some ideas about what the issues are.
Quoted 4 lines
Quoted 4 lines
Also, regarding the last paragraph, I have a question. The text doesn't seem quite to relate to the rest of the discussion. Should it be saying something like:
Quoted 8 lines
Quoted 6 lines
If the trailing component (i.e., basename) of pathname is a magic link, **how.resolve contains RESOLVE_NO_MAGICLINKS,**
and how.flags contains bothO_PATHandO_NOFOLLOW, then anO_PATHfile descriptor referencing the magic link will be returned.?
Yes, that is what I meant to write --
Good. Fixed.
Quoted 3 lines
and I believe that the
RESOLVE_NO_SYMLINKSsection is missing similar text in the second paragraph (except it should refer toRESOLVE_NO_SYMLINKS, obviously).
Also fixed.
Thanks,
Michael
Signature · 3 lines
Michael Kerrisk Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/ Linux/UNIX System Programming Training: http://man7.org/training/[...]
Re: [PATCH man-pages v2 2/2] openat2.2: document new openat2(2) syscall
In reply to Aleksa SaraiHi Aleksa,
Ping on this piece please:
Quoted 8 lines
Quoted 6 lines
Quoted 4 lines
It wouldn't hurt to add a longer description of magic-links in symlink(7). I'll send you a small patch to beef up the description (I had planned to include a longer rewrite with the
O_EMPTYPATHpatches but those require quite a bit more work to land).That would be great. Thank you!
I'll cook something up later this week.
Thanks,
Michael
PATCH 1/1 symlink.7: document magic-links more completely
In reply to Aleksa SaraiHi Michael,
Sorry for the delay and here is the patch I promised in this thread.
--8<---------------------------------------------------------------------8<--
Traditionally, magic-links have not been a well-understood topic in Linux. This helps clarify some of the terminology used in openat2.2.
--- man7/symlink.7 | 31 ++++++++++++++++++++++--------- 1 file changed, 22 insertions(+), 9 deletions(-)
man7/symlink.7 +22 -9
@@ -84,6 +84,21 @@ as they are implemented on Linux and other systems, are outlined here. It is important that site-local applications also conform to these rules, so that the user interface can be as consistent as possible. +.SS Magic-links +There is a special class of symlink-like objects known as "magic-links" which +can be found in certain pseudo-filesystems such as +.BR proc (5) +(examples include +.IR /proc/[pid]/exe " and " /proc/[pid]/fd/* .) +Unlike normal symlinks, magic-links are not resolved through +pathname-expansion, but instead act as direct references to the kernel's own +representation of a file handle. As such, these magic-links allow users to +access files which cannot be referenced with normal paths (such as unlinked +files still referenced by a running program.) +.PP +Because they can bypass ordinary +.BR mount_namespaces (7)-based +restrictions, magic-links have been used as attack vectors in various exploits. .SS Symbolic link ownership, permissions, and timestamps The owner and group of an existing symbolic link can be changed using
@@ -99,16 +114,14 @@ of a symbolic link can be changed using or .BR lutimes (3). .PP -On Linux, the permissions of a symbolic link are not used -in any operations; the permissions are always -0777 (read, write, and execute for all user categories), .\" Linux does not currently implement an lchmod(2). -and can't be changed. -(Note that there are some "magic" symbolic links in the -.I /proc -directory tree\(emfor example, the -.IR /proc/[pid]/fd/* -files\(emthat have different permissions.) +On Linux, the permissions of an ordinary symbolic link are not used in any +operations; the permissions are always 0777 (read, write, and execute for all +user categories), and can't be changed. +.PP +However, magic-links do not follow this rule. They can have a non-0777 mode, +though this mode is not currently used in any permission checks. + .\" .\" The .\" 4.4BSD
Signature · 1 line
2.26.2
Re: [PATCH] symlink.7: document magic-links more completely
Hello Aleksa,
Quoted 3 lines · Aleksa Sarai
Hi Michael,
Sorry for the delay and here is the patch I promised in this thread.
Thanks!
Patch applied.
Cheers,
Michael
Quoted 61 lines
--8<---------------------------------------------------------------------8<--
Traditionally, magic-links have not been a well-understood topic in Linux. This helps clarify some of the terminology used in openat2.2.
Signed-off-by Aleksa Sarai <[email protected]>--- man7/symlink.7 | 31 ++++++++++++++++++++++--------- 1 file changed, 22 insertions(+), 9 deletions(-)man7/symlink.7 +22
-9@@ -84,6 +84,21 @@ as they are implemented on Linux and other systems, are outlined here. It is important that site-local applications also conform to these rules, so that the user interface can be as consistent as possible. +.SS Magic-links +There is a special class of symlink-like objects known as "magic-links" which +can be found in certain pseudo-filesystems such as +.BR proc (5) +(examples include +.IR /proc/[pid]/exe " and " /proc/[pid]/fd/* .) +Unlike normal symlinks, magic-links are not resolved through +pathname-expansion, but instead act as direct references to the kernel's own +representation of a file handle. As such, these magic-links allow users to +access files which cannot be referenced with normal paths (such as unlinked +files still referenced by a running program.) +.PP +Because they can bypass ordinary +.BR mount_namespaces (7)-based +restrictions, magic-links have been used as attack vectors in various exploits. .SS Symbolic link ownership, permissions, and timestamps The owner and group of an existing symbolic link can be changed using@@ -99,16 +114,14 @@ of a symbolic link can be changed using or .BR lutimes (3). .PP -On Linux, the permissions of a symbolic link are not used -in any operations; the permissions are always -0777 (read, write, and execute for all user categories), .\" Linux does not currently implement an lchmod(2). -and can't be changed. -(Note that there are some "magic" symbolic links in the -.I /proc -directory tree\(emfor example, the -.IR /proc/[pid]/fd/* -files\(emthat have different permissions.) +On Linux, the permissions of an ordinary symbolic link are not used in any +operations; the permissions are always 0777 (read, write, and execute for all +user categories), and can't be changed. +.PP +However, magic-links do not follow this rule. They can have a non-0777 mode, +though this mode is not currently used in any permission checks. + .\" .\" The .\" 4.4BSD
Signature · 3 lines
Michael Kerrisk Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/ Linux/UNIX System Programming Training: http://man7.org/training/