Skip to contentSkip to conversation

O_REGULAR flag support for open

PATCH In review Dorjoy Chowdhury 5 participants 26 messages Jan 27 – Feb 18

Subsystems (8)
lore
Share & tools
lore
#0 lore raw

v3 cover O_REGULAR flag support for open

Changes in v3:
- included motivation about O_REGULAR flag in commit message e.g., programs not wanting to be tricked into opening device nodes
- fixed commit message wrongly referencing ENOTREGULAR instead of ENOTREG
- fixed the O_REGULAR flag in arch/parisc/include/uapi/asm/fcntl.h from 060000000 to 0100000000
- added 2 commits converting arch/{mips,sparc}/include/uapi/asm/fcntl.h O_* macros from hex to octal
- v2 is at: https://lore.kernel.org/linux-fsdevel/[email protected]/T/

Changes in v2:
- rename ENOTREGULAR to ENOTREG
- define ENOTREG in uapi/asm-generic/errno.h (instead of errno-base.h) and in arch/*/include/uapi/asm/errno.h files
- override O_REGULAR in arch/{alpha,sparc,parisc}/include/uapi/asm/fcntl.h due to clash with include/uapi/asm-generic/fcntl.h
- I have kept the kselftest but now that O_REGULAR and ENOTREG can have different value on different architectures I am not sure if it's right
- v1 is at: https://lore.kernel.org/linux-fsdevel/[email protected]/T/

Hi,

I came upon this "Ability to only open regular files" uapi feature suggestion from https://uapi-group.org/kernel-features/#ability-to-only-open-regular-files
and thought it would be something I could do as a first patch and get to know the kernel code a bit better.

I am not quite sure if the semantics that I baked into the code for this O_REGULAR flag's behavior when combined with other flags like O_CREAT look good and if there are other places that need the checks. I can fixup my patch according to suggestions for improvement. I did some happy path testing and the O_REGULAR flag seems to work as intended.

Thanks.

Regards,
Dorjoy

Dorjoy Chowdhury (4):
open: new O_REGULAR flag support kselftest/openat2: test for O_REGULAR flag sparc/fcntl.h: convert O_* flag macros from hex to octal mips/fcntl.h: convert O_* flag macros from hex to octal

 arch/alpha/include/uapi/asm/errno.h           |  2 +
 arch/alpha/include/uapi/asm/fcntl.h           |  1 +
 arch/mips/include/uapi/asm/errno.h            |  2 +
 arch/mips/include/uapi/asm/fcntl.h            | 22 +++++------
 arch/parisc/include/uapi/asm/errno.h          |  2 +
 arch/parisc/include/uapi/asm/fcntl.h          |  1 +
 arch/sparc/include/uapi/asm/errno.h           |  2 +
 arch/sparc/include/uapi/asm/fcntl.h           | 35 +++++++++---------
 fs/fcntl.c                                    |  2 +-
 fs/namei.c                                    |  6 +++
 fs/open.c                                     |  4 +-
 include/linux/fcntl.h                         |  2 +-
 include/uapi/asm-generic/errno.h              |  2 +
 include/uapi/asm-generic/fcntl.h              |  4 ++
 tools/arch/alpha/include/uapi/asm/errno.h     |  2 +
 tools/arch/mips/include/uapi/asm/errno.h      |  2 +
 tools/arch/parisc/include/uapi/asm/errno.h    |  2 +
 tools/arch/sparc/include/uapi/asm/errno.h     |  2 +
 tools/include/uapi/asm-generic/errno.h        |  2 +
 .../testing/selftests/openat2/openat2_test.c  | 37 ++++++++++++++++++-
 20 files changed, 102 insertions(+), 32 deletions(-)
Signature · 1 line
2.52.0
#1 lore raw

PATCH v3 1/4 open: new O_REGULAR flag support

This flag indicates the path should be opened if it's a regular file. This is useful to write secure programs that want to avoid being tricked into opening device nodes with special semantics while thinking they operate on regular files.

A corresponding error code ENOTREG has been introduced. For example, if open is called on path /dev/null with O_REGULAR in the flag param, it will return -ENOTREG.

When used in combination with O_CREAT, either the regular file is created, or if the path already exists, it is opened if it's a regular file. Otherwise, -ENOTREG is returned.

-EINVAL is returned when O_REGULAR is combined with O_DIRECTORY (not part of O_TMPFILE) because it doesn't make sense to open a path that is both a directory and a regular file.

---
 arch/alpha/include/uapi/asm/errno.h        | 2 ++
 arch/alpha/include/uapi/asm/fcntl.h        | 1 +
 arch/mips/include/uapi/asm/errno.h         | 2 ++
 arch/parisc/include/uapi/asm/errno.h       | 2 ++
 arch/parisc/include/uapi/asm/fcntl.h       | 1 +
 arch/sparc/include/uapi/asm/errno.h        | 2 ++
 arch/sparc/include/uapi/asm/fcntl.h        | 1 +
 fs/fcntl.c                                 | 2 +-
 fs/namei.c                                 | 6 ++++++
 fs/open.c                                  | 4 +++-
 include/linux/fcntl.h                      | 2 +-
 include/uapi/asm-generic/errno.h           | 2 ++
 include/uapi/asm-generic/fcntl.h           | 4 ++++
 tools/arch/alpha/include/uapi/asm/errno.h  | 2 ++
 tools/arch/mips/include/uapi/asm/errno.h   | 2 ++
 tools/arch/parisc/include/uapi/asm/errno.h | 2 ++
 tools/arch/sparc/include/uapi/asm/errno.h  | 2 ++
 tools/include/uapi/asm-generic/errno.h     | 2 ++
 18 files changed, 38 insertions(+), 3 deletions(-)
arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON	139	/* Memory page has hardware error */
 
+#define ENOTREG		140	/* Not a regular file */
+
 #endif
arch/alpha/include/uapi/asm/fcntl.h +1 -0
index 50bdc8e8a271..4da5a64c23bd 100644
@@ -34,6 +34,7 @@
 
 #define O_PATH		040000000
 #define __O_TMPFILE	0100000000
+#define O_REGULAR	0200000000
 
 #define F_GETLK		7
 #define F_SETLK		8
arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON	168	/* Memory page has hardware error */
 
+#define ENOTREG		169	/* Not a regular file */
+
 #define EDQUOT		1133	/* Quota exceeded */
 
 
arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON	257	/* Memory page has hardware error */
 
+#define ENOTREG		258	/* Not a regular file */
+
 #endif
arch/parisc/include/uapi/asm/fcntl.h +1 -0
index 03dee816cb13..0cc3320fe326 100644
@@ -19,6 +19,7 @@
 
 #define O_PATH		020000000
 #define __O_TMPFILE	040000000
+#define O_REGULAR	0100000000
 
 #define F_GETLK64	8
 #define F_SETLK64	9
arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON	135	/* Memory page has hardware error */
 
+#define ENOTREG		136	/* Not a regular file */
+
 #endif
arch/sparc/include/uapi/asm/fcntl.h +1 -0
index 67dae75e5274..a93d18d2c23e 100644
@@ -37,6 +37,7 @@
 
 #define O_PATH		0x1000000
 #define __O_TMPFILE	0x2000000
+#define O_REGULAR	0x4000000
 
 #define F_GETOWN	5	/*  for sockets. */
 #define F_SETOWN	6	/*  for sockets. */
fs/fcntl.c +1 -1
index f93dbca08435..62ab4ad2b6f5 100644
@@ -1169,7 +1169,7 @@ static int __init fcntl_init(void)
 	 * Exceptions: O_NONBLOCK is a two bit define on parisc; O_NDELAY
 	 * is defined as O_NONBLOCK on some platforms and not on others.
 	 */
-	BUILD_BUG_ON(20 - 1 /* for O_RDONLY being 0 */ !=
+	BUILD_BUG_ON(21 - 1 /* for O_RDONLY being 0 */ !=
 		HWEIGHT32(
 			(VALID_OPEN_FLAGS & ~(O_NONBLOCK | O_NDELAY)) |
 			__FMODE_EXEC));
fs/namei.c +6 -0
index b28ecb699f32..f5504ae4b03c 100644
@@ -4616,6 +4616,10 @@ static int do_open(struct nameidata *nd,
 		if (unlikely(error))
 			return error;
 	}
+
+	if ((open_flag & O_REGULAR) && !d_is_reg(nd->path.dentry))
+		return -ENOTREG;
+
 	if ((nd->flags & LOOKUP_DIRECTORY) && !d_can_lookup(nd->path.dentry))
 		return -ENOTDIR;
 
@@ -4765,6 +4769,8 @@ static int do_o_path(struct nameidata *nd, unsigned flags, struct file *file)
 	struct path path;
 	int error = path_lookupat(nd, flags, &path);
 	if (!error) {
+		if ((file->f_flags & O_REGULAR) && !d_is_reg(path.dentry))
+			return -ENOTREG;
 		audit_inode(nd->name, path.dentry, 0);
 		error = vfs_open(&path, file);
 		path_put(&path);
fs/open.c +3 -1
index 74c4c1462b3e..82153e21907e 100644
@@ -1173,7 +1173,7 @@ struct file *kernel_file_open(const struct path *path, int flags,
 EXPORT_SYMBOL_GPL(kernel_file_open);
 
 #define WILL_CREATE(flags)	(flags & (O_CREAT | __O_TMPFILE))
-#define O_PATH_FLAGS		(O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC)
+#define O_PATH_FLAGS		(O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC | O_REGULAR)
 
 inline struct open_how build_open_how(int flags, umode_t mode)
 {
@@ -1250,6 +1250,8 @@ inline int build_open_flags(const struct open_how *how, struct open_flags *op)
 			return -EINVAL;
 		if (!(acc_mode & MAY_WRITE))
 			return -EINVAL;
+	} else if ((flags & O_DIRECTORY) && (flags & O_REGULAR)) {
+		return -EINVAL;
 	}
 	if (flags & O_PATH) {
 		/* O_PATH only permits certain other flags to be set. */
include/linux/fcntl.h +1 -1
index a332e79b3207..4fd07b0e0a17 100644
@@ -10,7 +10,7 @@
 	(O_RDONLY | O_WRONLY | O_RDWR | O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC | \
 	 O_APPEND | O_NDELAY | O_NONBLOCK | __O_SYNC | O_DSYNC | \
 	 FASYNC	| O_DIRECT | O_LARGEFILE | O_DIRECTORY | O_NOFOLLOW | \
-	 O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE)
+	 O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE | O_REGULAR)
 
 /* List of all valid flags for the how->resolve argument: */
 #define VALID_RESOLVE_FLAGS \
include/uapi/asm-generic/errno.h +2 -0
index 92e7ae493ee3..2216ab9aa32e 100644
@@ -122,4 +122,6 @@
 
 #define EHWPOISON	133	/* Memory page has hardware error */
 
+#define ENOTREG		134	/* Not a regular file */
+
 #endif
include/uapi/asm-generic/fcntl.h +4 -0
index 613475285643..3468b352a575 100644
@@ -88,6 +88,10 @@
 #define __O_TMPFILE	020000000
 #endif
 
+#ifndef O_REGULAR
+#define O_REGULAR	040000000
+#endif
+
 /* a horrid kludge trying to make sure that this will fail on old kernels */
 #define O_TMPFILE (__O_TMPFILE | O_DIRECTORY)
 
tools/arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON	139	/* Memory page has hardware error */
 
+#define ENOTREG		140	/* Not a regular file */
+
 #endif
tools/arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON	168	/* Memory page has hardware error */
 
+#define ENOTREG		169	/* Not a regular file */
+
 #define EDQUOT		1133	/* Quota exceeded */
 
 
tools/arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON	257	/* Memory page has hardware error */
 
+#define ENOTREG		258	/* Not a regular file */
+
 #endif
tools/arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON	135	/* Memory page has hardware error */
 
+#define ENOTREG		136	/* Not a regular file */
+
 #endif
tools/include/uapi/asm-generic/errno.h +2 -0
index 92e7ae493ee3..2216ab9aa32e 100644
@@ -122,4 +122,6 @@
 
 #define EHWPOISON	133	/* Memory page has hardware error */
 
+#define ENOTREG		134	/* Not a regular file */
+
 #endif
Signature · 1 line
2.52.0
#2 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

Quoted 9 lines · Dorjoy Chowdhury

This flag indicates the path should be opened if it's a regular file. This is useful to write secure programs that want to avoid being tricked into opening device nodes with special semantics while thinking they operate on regular files.

A corresponding error code ENOTREG has been introduced. For example, if open is called on path /dev/null with O_REGULAR in the flag param, it will return -ENOTREG.

Quoted 7 lines

When used in combination with O_CREAT, either the regular file is created, or if the path already exists, it is opened if it's a regular file. Otherwise, -ENOTREG is returned.

-EINVAL is returned when O_REGULAR is combined with O_DIRECTORY (not part of O_TMPFILE) because it doesn't make sense to open a path that is both a directory and a regular file.

As you mention in your cover letter, this is something that the UAPI group has asked for in the past[1] and was even discussed at a recent LPC (maybe LPC 2024?) -- thanks for the patch!

In the next posting of this patchset, I would suggest including this information in the *commit message* with a link (commit messages end up in the git history, cover letters are a little harder to search for when doing "git blame").

[1]: https://uapi-group.org/kernel-features/#ability-to-only-open-regular-files

fs/open.c struct file *kernel_file_open(const struct path *path, int flags, · line 1176

Quoted 3 lines

#define WILL_CREATE(flags) (flags & (O_CREAT | __O_TMPFILE))

fs/open.c +1 -1
-#define O_PATH_FLAGS		(O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC)
+#define O_PATH_FLAGS		(O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC | O_REGULAR)

It doesn't really make sense to use this flag with O_PATH -- O_PATH file descriptors do not actually open the target inode and so there is no risk to doing this.

In fact the method of safely opening files while avoiding device inodes on Linux today is to open an O_PATH, then use fstat(2) to check whether it is a regular file, and then re-open the file descriptor through /proc/self/fd/$n. (This is totally race-safe.)

My main reason for pushing back against this it's really quite preferable to avoid expanding the set of O_* flags which work with O_PATH if they don't add much -- O_PATH has really unfortunate behaviour with ignoring other flags and openat2(2) finally fixed that by blocking ignored flag combinations.

Quoted 16 lines

inline struct open_how build_open_how(int flags, umode_t mode) {

fs/open.c +2 -0
@@ -1250,6 +1250,8 @@ inline int build_open_flags(const struct open_how *how, struct open_flags *op)
 			return -EINVAL;
 		if (!(acc_mode & MAY_WRITE))
 			return -EINVAL;
+	} else if ((flags & O_DIRECTORY) && (flags & O_REGULAR)) {
+		return -EINVAL;
 	}
 	if (flags & O_PATH) {
 		/* O_PATH only permits certain other flags to be set. */
include/linux/fcntl.h +0 -0
index a332e79b3207..4fd07b0e0a17 100644

include/linux/fcntl.h line 13

Quoted 5 lines
include/linux/fcntl.h +1 -1
@@ -10,7 +10,7 @@
 	(O_RDONLY | O_WRONLY | O_RDWR | O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC | \
 	 O_APPEND | O_NDELAY | O_NONBLOCK | __O_SYNC | O_DSYNC | \
 	 FASYNC	| O_DIRECT | O_LARGEFILE | O_DIRECTORY | O_NOFOLLOW | \
-	 O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE)
+	 O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE | O_REGULAR)

Legacy open(2)/openat(2) do not reject invalid flag arguments, which means that you cannot trivially add a new security-critical flag to them for two reasons:

* You cannot easily rely on them because old kernels will not return -EINVAL, meaning you cannot be sure that the flag is supported. You can try to test-run it, but the operation needs to be a non-dangerous operation to try (and caching this has its own issues, such as with programs that apply seccomp filters later).

To be fair, since you reject O_DIRECTORY|O_REGULAR there is a relatively easy way to detect this, but the caveats about problems with caching still apply.

* Old programs might pass garbage bits that have been ignored thus far, which means that making them have meaning can break userspace. Given the age of open(2) this is a very hard thing to guarantee and is one of many reasons I wrote openat2(2) and finally added proper flag checking.

This is something your patch doesn't deal with and I don't think can be done in a satisfactory way (because the behaviour relies on more than just the arguments).

For reference, this is why O_TMPFILE includes O_DIRECTORY and requires an O_ACCMODE with write bits -- this combination will fail on old kernels, which allows you to rely on it and also guarantees that no existing older programs passed that flag combination already and happened to work on older kernels. This kind of trick won't work for O_REGULAR, unfortunately.

In my view, this should be an openat2(2)-only API. In addition, I would propose that (instead of burning another O_* flag bit for this as a special-purpose API just for regular files) you could have a mask of which S_IFMT bits should be rejected as a new field in "struct open_how". This would let you reject sockets or device inodes but permit FIFOs and regular files or directories, for instance. This could even be done without a new O_* flag at all (the zero-value how->sfmt_mask would allow everything and so would work well with extensible structs), but we could add an O2_* flag anyway.

include/linux/fcntl.h line 15

Quoted 2 lines
include/linux/fcntl.h +2 -0
+#define ENOTREG		134	/* Not a regular file */
+

We are probably a little too reticent to add new errnos, but in this case I think that there should be some description in the commit or cover letter about why a new errno is needed. ENXIO or EPROTONOSUPPORT/EPROTOTYPE is what you would typically use (yes, they aren't a _perfect_ match but one of the common occurrences in syscall design is to read through errno(7) and figure out what errnos kind of fit what you need to express).

Then to be fair, the existence of ENOTBLK, ENOTDIR, ENOTSOCK, etc. kind of justify the existence of ENOTREG too. Unfortunately, you won't be able to use ENOTREG if you go with my idea of having mask bits in open_how... (And what errno should we use then...? Hm.)

Signature · 2 lines
Aleksa Sarai
https://www.cyphar.com/
#3 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

Quoted 1 line · Aleksa Sarai

In my view, this should be an openat2(2)-only API.

fwiw +1 from me, the O_ flag situation is already terrible even without the validation woes.

I find it most unfortunate the openat2 syscall reuses the O_ namespace. For my taste it would be best closed for business, with all new flag additions using a different space.

I can easily see people passing O_WHATEVER to open and openat by blindly assuming they are supported just based on the name.

that's a side mini-rant, too late to do anything here now

Quoted 9 lines

In addition, I would
propose that (instead of burning another O_* flag bit for this as a special-purpose API just for regular files) you could have a mask of which S_IFMT bits should be rejected as a new field in "struct open_how". This would let you reject sockets or device inodes but permit FIFOs and regular files or directories, for instance. This could even be done without a new O_* flag at all (the zero-value how->sfmt_mask would allow everything and so would work well with extensible structs), but we could add an O2_* flag anyway.

I don't think this works because the vars have overlapping bits:
#define S_IFBLK 0060000
#define S_IFDIR 0040000

So you very much can't select what you want off of a bitmask.

At best the field could be used to select the one type you are fine with.

If one was to pursue the idea, some other defines with unique bits would need to be provided. But even then, semantics should be to only *allow* the bits you are fine with and reject the rest.

But I'm not at all confident this is worth any effort -- with O_DIRECTORY already being there and O_REGULAR proposed, is there a use case which wants something else?

Quoted 4 lines
Quoted 3 lines
(quoted diff) +2 -0
+#define ENOTREG		134	/* Not a regular file */
+

[..]

Quoted 5 lines

Then to be fair, the existence of ENOTBLK, ENOTDIR, ENOTSOCK, etc. kind of justify the existence of ENOTREG too. Unfortunately, you won't be able to use ENOTREG if you go with my idea of having mask bits in open_how... (And what errno should we use then...? Hm.)

The most useful behavior would indicate what was found (e.g., a pipe).

The easiest way to do it would create errnos for all types (EISDIR already exists for one), but I can't seriously propose that.

Going the other way, EBADTYPE or something else reusable would be my idea.

#4 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

Quoted 34 lines · Mateusz Guzik
Quoted 1 line · Aleksa Sarai

In my view, this should be an openat2(2)-only API.

fwiw +1 from me, the O_ flag situation is already terrible even without the validation woes.

I find it most unfortunate the openat2 syscall reuses the O_ namespace. For my taste it would be best closed for business, with all new flag additions using a different space.

I can easily see people passing O_WHATEVER to open and openat by blindly assuming they are supported just based on the name.

that's a side mini-rant, too late to do anything here now

Quoted 9 lines

In addition, I would
propose that (instead of burning another O_* flag bit for this as a special-purpose API just for regular files) you could have a mask of which S_IFMT bits should be rejected as a new field in "struct open_how". This would let you reject sockets or device inodes but permit FIFOs and regular files or directories, for instance. This could even be done without a new O_* flag at all (the zero-value how->sfmt_mask would allow everything and so would work well with extensible structs), but we could add an O2_* flag anyway.

I don't think this works because the vars have overlapping bits:
#define S_IFBLK 0060000
#define S_IFDIR 0040000

So you very much can't select what you want off of a bitmask.

At best the field could be used to select the one type you are fine with.

Quoted 8 lines

If one was to pursue the idea, some other defines with unique bits would need to be provided. But even then, semantics should be to only *allow* the bits you are fine with and reject the rest.

But I'm not at all confident this is worth any effort -- with O_DIRECTORY already being there and O_REGULAR proposed, is there a use case which wants something else?

Good discussion. So should I just rename the O_REGULAR to O2_REGULAR and create a VALID_OPENAT2_FLAGS and no need to do how->sfmt_mask stuff?

Quoted 11 lines
Quoted 4 lines
Quoted 3 lines
(quoted diff) +2 -0
+#define ENOTREG            134     /* Not a regular file */
+

[..]

Quoted 5 lines

Then to be fair, the existence of ENOTBLK, ENOTDIR, ENOTSOCK, etc. kind of justify the existence of ENOTREG too. Unfortunately, you won't be able to use ENOTREG if you go with my idea of having mask bits in open_how... (And what errno should we use then...? Hm.)

Quoted 7 lines

The most useful behavior would indicate what was found (e.g., a pipe).

The easiest way to do it would create errnos for all types (EISDIR already exists for one), but I can't seriously propose that.

Going the other way, EBADTYPE or something else reusable would be my idea.

Good point. Maybe ENOTREG is acceptable too?

Regards,
Dorjoy

#5 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

In reply to Mateusz Guzik
Quoted 9 lines · Mateusz Guzik
Quoted 1 line · Aleksa Sarai

In my view, this should be an openat2(2)-only API.

fwiw +1 from me, the O_ flag situation is already terrible even without the validation woes.

I find it most unfortunate the openat2 syscall reuses the O_ namespace. For my taste it would be best closed for business, with all new flag additions using a different space.

We don't have any openat2(2)-only O_* flags yet, I agree that new flag additions (except for very rare cases where you can make them backward compatible -- such as a hypothetical O_EMPTYPATH) should be O2_* or OEXT_* or something.

Quoted 2 lines

I can easily see people passing O_WHATEVER to open and openat by blindly assuming they are supported just based on the name.

Yeah, if we don't do that it'll lead to confusion. openat2(2) has exclusive rights to the 64-bit flag bits so we could start with those before we need to cross with the O_* flag space.

Quoted 12 lines

that's a side mini-rant, too late to do anything here now

Quoted 9 lines

In addition, I would
propose that (instead of burning another O_* flag bit for this as a special-purpose API just for regular files) you could have a mask of which S_IFMT bits should be rejected as a new field in "struct open_how". This would let you reject sockets or device inodes but permit FIFOs and regular files or directories, for instance. This could even be done without a new O_* flag at all (the zero-value how->sfmt_mask would allow everything and so would work well with extensible structs), but we could add an O2_* flag anyway.

Quoted 5 lines

I don't think this works because the vars have overlapping bits:
#define S_IFBLK 0060000
#define S_IFDIR 0040000

So you very much can't select what you want off of a bitmask.

Well, you can filter on S_IFCHR if you want to block both block/char devices, but yeah the overlap is quite unfortunate... (That would also mean blocking directories would also block S_IFBLK -- I remembered there was an overlap but I forgot it coincided with S_IFDIR... Damn wacky APIs.)

Quoted 9 lines

At best the field could be used to select the one type you are fine with.

If one was to pursue the idea, some other defines with unique bits would need to be provided. But even then, semantics should be to only *allow* the bits you are fine with and reject the rest.

But I'm not at all confident this is worth any effort -- with O_DIRECTORY already being there and O_REGULAR proposed, is there a use case which wants something else?

There's also O_NOFOLLOW in a similar vein.

I can see someone wanting to permit FIFOs, regular files, and directories being fine but blocking everything else. None of O_REGULAR, O_DIRECTORY, nor O_NOFOLLOW provide that.

Quoted 10 lines
Quoted 3 lines
Quoted 2 lines
(quoted diff) +2 -0
+#define ENOTREG		134	/* Not a regular file */
+

[..]

Quoted 5 lines

Then to be fair, the existence of ENOTBLK, ENOTDIR, ENOTSOCK, etc. kind of justify the existence of ENOTREG too. Unfortunately, you won't be able to use ENOTREG if you go with my idea of having mask bits in open_how... (And what errno should we use then...? Hm.)

Quoted 4 lines

The most useful behavior would indicate what was found (e.g., a pipe).

The easiest way to do it would create errnos for all types (EISDIR already exists for one), but I can't seriously propose that.

It might be kinda neat from a potential re-use perspective in other APIs but yeah it would be quite wasteful to burn 3-5 errnos for this when we already have ~4 that are logical inverses.

Quoted 2 lines

Going the other way, EBADTYPE or something else reusable would be my idea.

I think that would be reasonable and if you word the error message carefully you can even see it being a fairly generic errno for other places to use.

Signature · 2 lines
Aleksa Sarai
https://www.cyphar.com/
maintainer
#6 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

In reply to Dorjoy Chowdhury
Quoted 270 lines · Dorjoy Chowdhury

This flag indicates the path should be opened if it's a regular file. This is useful to write secure programs that want to avoid being tricked into opening device nodes with special semantics while thinking they operate on regular files.

A corresponding error code ENOTREG has been introduced. For example, if open is called on path /dev/null with O_REGULAR in the flag param, it will return -ENOTREG.

When used in combination with O_CREAT, either the regular file is created, or if the path already exists, it is opened if it's a regular file. Otherwise, -ENOTREG is returned.

-EINVAL is returned when O_REGULAR is combined with O_DIRECTORY (not part of O_TMPFILE) because it doesn't make sense to open a path that is both a directory and a regular file.

---
 arch/alpha/include/uapi/asm/errno.h        | 2 ++
 arch/alpha/include/uapi/asm/fcntl.h        | 1 +
 arch/mips/include/uapi/asm/errno.h         | 2 ++
 arch/parisc/include/uapi/asm/errno.h       | 2 ++
 arch/parisc/include/uapi/asm/fcntl.h       | 1 +
 arch/sparc/include/uapi/asm/errno.h        | 2 ++
 arch/sparc/include/uapi/asm/fcntl.h        | 1 +
 fs/fcntl.c                                 | 2 +-
 fs/namei.c                                 | 6 ++++++
 fs/open.c                                  | 4 +++-
 include/linux/fcntl.h                      | 2 +-
 include/uapi/asm-generic/errno.h           | 2 ++
 include/uapi/asm-generic/fcntl.h           | 4 ++++
 tools/arch/alpha/include/uapi/asm/errno.h  | 2 ++
 tools/arch/mips/include/uapi/asm/errno.h   | 2 ++
 tools/arch/parisc/include/uapi/asm/errno.h | 2 ++
 tools/arch/sparc/include/uapi/asm/errno.h  | 2 ++
 tools/include/uapi/asm-generic/errno.h     | 2 ++
 18 files changed, 38 insertions(+), 3 deletions(-)
arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON	139	/* Memory page has hardware error */
 
+#define ENOTREG		140	/* Not a regular file */
+
 #endif
arch/alpha/include/uapi/asm/fcntl.h +1 -0
index 50bdc8e8a271..4da5a64c23bd 100644
@@ -34,6 +34,7 @@
 
 #define O_PATH		040000000
 #define __O_TMPFILE	0100000000
+#define O_REGULAR	0200000000
 
 #define F_GETLK		7
 #define F_SETLK		8
arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON	168	/* Memory page has hardware error */
 
+#define ENOTREG		169	/* Not a regular file */
+
 #define EDQUOT		1133	/* Quota exceeded */
 
 
arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON	257	/* Memory page has hardware error */
 
+#define ENOTREG		258	/* Not a regular file */
+
 #endif
arch/parisc/include/uapi/asm/fcntl.h +1 -0
index 03dee816cb13..0cc3320fe326 100644
@@ -19,6 +19,7 @@
 
 #define O_PATH		020000000
 #define __O_TMPFILE	040000000
+#define O_REGULAR	0100000000
 
 #define F_GETLK64	8
 #define F_SETLK64	9
arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON	135	/* Memory page has hardware error */
 
+#define ENOTREG		136	/* Not a regular file */
+
 #endif
arch/sparc/include/uapi/asm/fcntl.h +1 -0
index 67dae75e5274..a93d18d2c23e 100644
@@ -37,6 +37,7 @@
 
 #define O_PATH		0x1000000
 #define __O_TMPFILE	0x2000000
+#define O_REGULAR	0x4000000
 
 #define F_GETOWN	5	/*  for sockets. */
 #define F_SETOWN	6	/*  for sockets. */
fs/fcntl.c +1 -1
index f93dbca08435..62ab4ad2b6f5 100644
@@ -1169,7 +1169,7 @@ static int __init fcntl_init(void)
 	 * Exceptions: O_NONBLOCK is a two bit define on parisc; O_NDELAY
 	 * is defined as O_NONBLOCK on some platforms and not on others.
 	 */
-	BUILD_BUG_ON(20 - 1 /* for O_RDONLY being 0 */ !=
+	BUILD_BUG_ON(21 - 1 /* for O_RDONLY being 0 */ !=
 		HWEIGHT32(
 			(VALID_OPEN_FLAGS & ~(O_NONBLOCK | O_NDELAY)) |
 			__FMODE_EXEC));
fs/namei.c +6 -0
index b28ecb699f32..f5504ae4b03c 100644
@@ -4616,6 +4616,10 @@ static int do_open(struct nameidata *nd,
 		if (unlikely(error))
 			return error;
 	}
+
+	if ((open_flag & O_REGULAR) && !d_is_reg(nd->path.dentry))
+		return -ENOTREG;
+
 	if ((nd->flags & LOOKUP_DIRECTORY) && !d_can_lookup(nd->path.dentry))
 		return -ENOTDIR;
 
@@ -4765,6 +4769,8 @@ static int do_o_path(struct nameidata *nd, unsigned flags, struct file *file)
 	struct path path;
 	int error = path_lookupat(nd, flags, &path);
 	if (!error) {
+		if ((file->f_flags & O_REGULAR) && !d_is_reg(path.dentry))
+			return -ENOTREG;
 		audit_inode(nd->name, path.dentry, 0);
 		error = vfs_open(&path, file);
 		path_put(&path);
fs/open.c +3 -1
index 74c4c1462b3e..82153e21907e 100644
@@ -1173,7 +1173,7 @@ struct file *kernel_file_open(const struct path *path, int flags,
 EXPORT_SYMBOL_GPL(kernel_file_open);
 
 #define WILL_CREATE(flags)	(flags & (O_CREAT | __O_TMPFILE))
-#define O_PATH_FLAGS		(O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC)
+#define O_PATH_FLAGS		(O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC | O_REGULAR)
 
 inline struct open_how build_open_how(int flags, umode_t mode)
 {
@@ -1250,6 +1250,8 @@ inline int build_open_flags(const struct open_how *how, struct open_flags *op)
 			return -EINVAL;
 		if (!(acc_mode & MAY_WRITE))
 			return -EINVAL;
+	} else if ((flags & O_DIRECTORY) && (flags & O_REGULAR)) {
+		return -EINVAL;
 	}
 	if (flags & O_PATH) {
 		/* O_PATH only permits certain other flags to be set. */
include/linux/fcntl.h +1 -1
index a332e79b3207..4fd07b0e0a17 100644
@@ -10,7 +10,7 @@
 	(O_RDONLY | O_WRONLY | O_RDWR | O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC | \
 	 O_APPEND | O_NDELAY | O_NONBLOCK | __O_SYNC | O_DSYNC | \
 	 FASYNC	| O_DIRECT | O_LARGEFILE | O_DIRECTORY | O_NOFOLLOW | \
-	 O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE)
+	 O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE | O_REGULAR)
 
 /* List of all valid flags for the how->resolve argument: */
 #define VALID_RESOLVE_FLAGS \
include/uapi/asm-generic/errno.h +2 -0
index 92e7ae493ee3..2216ab9aa32e 100644
@@ -122,4 +122,6 @@
 
 #define EHWPOISON	133	/* Memory page has hardware error */
 
+#define ENOTREG		134	/* Not a regular file */
+
 #endif
include/uapi/asm-generic/fcntl.h +4 -0
index 613475285643..3468b352a575 100644
@@ -88,6 +88,10 @@
 #define __O_TMPFILE	020000000
 #endif
 
+#ifndef O_REGULAR
+#define O_REGULAR	040000000
+#endif
+
 /* a horrid kludge trying to make sure that this will fail on old kernels */
 #define O_TMPFILE (__O_TMPFILE | O_DIRECTORY)
 
tools/arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON	139	/* Memory page has hardware error */
 
+#define ENOTREG		140	/* Not a regular file */
+
 #endif
tools/arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON	168	/* Memory page has hardware error */
 
+#define ENOTREG		169	/* Not a regular file */
+
 #define EDQUOT		1133	/* Quota exceeded */
 
 
tools/arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON	257	/* Memory page has hardware error */
 
+#define ENOTREG		258	/* Not a regular file */
+
 #endif
tools/arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON	135	/* Memory page has hardware error */
 
+#define ENOTREG		136	/* Not a regular file */
+
 #endif
tools/include/uapi/asm-generic/errno.h +0 -0
index 92e7ae493ee3..2216ab9aa32e 100644

tools/include/uapi/asm-generic/errno.h line 127

Quoted 6 lines
tools/include/uapi/asm-generic/errno.h +2 -0
@@ -122,4 +122,6 @@
 
 #define EHWPOISON	133	/* Memory page has hardware error */
 
+#define ENOTREG		134	/* Not a regular file */
+
 #endif

One thing this patch is missing is handling for ->atomic_open(). I imagine most of the filesystems that provide that op can't support O_REGULAR properly (maybe cifs can? idk). What you probably want to do is add in some patches that make all of the atomic_open operations in the kernel return -EINVAL if O_REGULAR is set.

Then, once the basic support is in, you or someone else can go back and implement support for O_REGULAR where possible.

Signature · 1 line
Jeff Layton <[email protected]>
#7 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

Quoted 285 lines · Jeff Layton
Quoted 270 lines · Dorjoy Chowdhury

This flag indicates the path should be opened if it's a regular file. This is useful to write secure programs that want to avoid being tricked into opening device nodes with special semantics while thinking they operate on regular files.

A corresponding error code ENOTREG has been introduced. For example, if open is called on path /dev/null with O_REGULAR in the flag param, it will return -ENOTREG.

When used in combination with O_CREAT, either the regular file is created, or if the path already exists, it is opened if it's a regular file. Otherwise, -ENOTREG is returned.

-EINVAL is returned when O_REGULAR is combined with O_DIRECTORY (not part of O_TMPFILE) because it doesn't make sense to open a path that is both a directory and a regular file.

---
 arch/alpha/include/uapi/asm/errno.h        | 2 ++
 arch/alpha/include/uapi/asm/fcntl.h        | 1 +
 arch/mips/include/uapi/asm/errno.h         | 2 ++
 arch/parisc/include/uapi/asm/errno.h       | 2 ++
 arch/parisc/include/uapi/asm/fcntl.h       | 1 +
 arch/sparc/include/uapi/asm/errno.h        | 2 ++
 arch/sparc/include/uapi/asm/fcntl.h        | 1 +
 fs/fcntl.c                                 | 2 +-
 fs/namei.c                                 | 6 ++++++
 fs/open.c                                  | 4 +++-
 include/linux/fcntl.h                      | 2 +-
 include/uapi/asm-generic/errno.h           | 2 ++
 include/uapi/asm-generic/fcntl.h           | 4 ++++
 tools/arch/alpha/include/uapi/asm/errno.h  | 2 ++
 tools/arch/mips/include/uapi/asm/errno.h   | 2 ++
 tools/arch/parisc/include/uapi/asm/errno.h | 2 ++
 tools/arch/sparc/include/uapi/asm/errno.h  | 2 ++
 tools/include/uapi/asm-generic/errno.h     | 2 ++
 18 files changed, 38 insertions(+), 3 deletions(-)
arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
arch/alpha/include/uapi/asm/fcntl.h +1 -0
index 50bdc8e8a271..4da5a64c23bd 100644
@@ -34,6 +34,7 @@
 
 #define O_PATH               040000000
 #define __O_TMPFILE  0100000000
+#define O_REGULAR    0200000000
 
 #define F_GETLK              7
 #define F_SETLK              8
arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
arch/parisc/include/uapi/asm/fcntl.h +1 -0
index 03dee816cb13..0cc3320fe326 100644
@@ -19,6 +19,7 @@
 
 #define O_PATH               020000000
 #define __O_TMPFILE  040000000
+#define O_REGULAR    0100000000
 
 #define F_GETLK64    8
 #define F_SETLK64    9
arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
arch/sparc/include/uapi/asm/fcntl.h +1 -0
index 67dae75e5274..a93d18d2c23e 100644
@@ -37,6 +37,7 @@
 
 #define O_PATH               0x1000000
 #define __O_TMPFILE  0x2000000
+#define O_REGULAR    0x4000000
 
 #define F_GETOWN     5       /*  for sockets. */
 #define F_SETOWN     6       /*  for sockets. */
fs/fcntl.c +1 -1
index f93dbca08435..62ab4ad2b6f5 100644
@@ -1169,7 +1169,7 @@ static int __init fcntl_init(void)
       * Exceptions: O_NONBLOCK is a two bit define on parisc; O_NDELAY
       * is defined as O_NONBLOCK on some platforms and not on others.
       */
-     BUILD_BUG_ON(20 - 1 /* for O_RDONLY being 0 */ !=
+     BUILD_BUG_ON(21 - 1 /* for O_RDONLY being 0 */ !=
              HWEIGHT32(
                      (VALID_OPEN_FLAGS & ~(O_NONBLOCK | O_NDELAY)) |
                      __FMODE_EXEC));
fs/namei.c +6 -0
index b28ecb699f32..f5504ae4b03c 100644
@@ -4616,6 +4616,10 @@ static int do_open(struct nameidata *nd,
              if (unlikely(error))
                      return error;
      }
+
+     if ((open_flag & O_REGULAR) && !d_is_reg(nd->path.dentry))
+             return -ENOTREG;
+
      if ((nd->flags & LOOKUP_DIRECTORY) && !d_can_lookup(nd->path.dentry))
              return -ENOTDIR;
 
@@ -4765,6 +4769,8 @@ static int do_o_path(struct nameidata *nd, unsigned flags, struct file *file)
      struct path path;
      int error = path_lookupat(nd, flags, &path);
      if (!error) {
+             if ((file->f_flags & O_REGULAR) && !d_is_reg(path.dentry))
+                     return -ENOTREG;
              audit_inode(nd->name, path.dentry, 0);
              error = vfs_open(&path, file);
              path_put(&path);
fs/open.c +3 -1
index 74c4c1462b3e..82153e21907e 100644
@@ -1173,7 +1173,7 @@ struct file *kernel_file_open(const struct path *path, int flags,
 EXPORT_SYMBOL_GPL(kernel_file_open);
 
 #define WILL_CREATE(flags)   (flags & (O_CREAT | __O_TMPFILE))
-#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC)
+#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC | O_REGULAR)
 
 inline struct open_how build_open_how(int flags, umode_t mode)
 {
@@ -1250,6 +1250,8 @@ inline int build_open_flags(const struct open_how *how, struct open_flags *op)
                      return -EINVAL;
              if (!(acc_mode & MAY_WRITE))
                      return -EINVAL;
+     } else if ((flags & O_DIRECTORY) && (flags & O_REGULAR)) {
+             return -EINVAL;
      }
      if (flags & O_PATH) {
              /* O_PATH only permits certain other flags to be set. */
include/linux/fcntl.h +1 -1
index a332e79b3207..4fd07b0e0a17 100644
@@ -10,7 +10,7 @@
      (O_RDONLY | O_WRONLY | O_RDWR | O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC | \
       O_APPEND | O_NDELAY | O_NONBLOCK | __O_SYNC | O_DSYNC | \
       FASYNC | O_DIRECT | O_LARGEFILE | O_DIRECTORY | O_NOFOLLOW | \
-      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE)
+      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE | O_REGULAR)
 
 /* List of all valid flags for the how->resolve argument: */
 #define VALID_RESOLVE_FLAGS \
include/uapi/asm-generic/errno.h +2 -0
index 92e7ae493ee3..2216ab9aa32e 100644
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif
include/uapi/asm-generic/fcntl.h +4 -0
index 613475285643..3468b352a575 100644
@@ -88,6 +88,10 @@
 #define __O_TMPFILE  020000000
 #endif
 
+#ifndef O_REGULAR
+#define O_REGULAR    040000000
+#endif
+
 /* a horrid kludge trying to make sure that this will fail on old kernels */
 #define O_TMPFILE (__O_TMPFILE | O_DIRECTORY)
 
tools/arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
tools/arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
tools/arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
tools/arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
tools/include/uapi/asm-generic/errno.h +0 -0
index 92e7ae493ee3..2216ab9aa32e 100644

tools/include/uapi/asm-generic/errno.h line 127

Quoted 6 lines
tools/include/uapi/asm-generic/errno.h +2 -0
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif

One thing this patch is missing is handling for ->atomic_open(). I imagine most of the filesystems that provide that op can't support O_REGULAR properly (maybe cifs can? idk). What you probably want to do is add in some patches that make all of the atomic_open operations in the kernel return -EINVAL if O_REGULAR is set.

Quoted 2 lines

Then, once the basic support is in, you or someone else can go back and implement support for O_REGULAR where possible.

Thank you for the feedback. I don't quite understand what I need to fix. I thought open system calls always create regular files, so atomic_open probably always creates regular files? Can you please give me some more details as to where I need to fix this and what the actual bug here is that is related to atomic_open? I think I had done some normal testing and when using O_CREAT | O_REGULAR, if the file doesn't exist, the file gets created and the file that gets created is a regular file, so it probably makes sense? Or should the behavior be that if file doesn't exist, -EINVAL is returned and if file exists it is opened if regular, otherwise -ENOTREG is returned?

Regards,
Dorjoy

maintainer
#8 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

Quoted 289 lines · Dorjoy Chowdhury
Quoted 285 lines · Jeff Layton
Quoted 270 lines · Dorjoy Chowdhury

This flag indicates the path should be opened if it's a regular file. This is useful to write secure programs that want to avoid being tricked into opening device nodes with special semantics while thinking they operate on regular files.

A corresponding error code ENOTREG has been introduced. For example, if open is called on path /dev/null with O_REGULAR in the flag param, it will return -ENOTREG.

When used in combination with O_CREAT, either the regular file is created, or if the path already exists, it is opened if it's a regular file. Otherwise, -ENOTREG is returned.

-EINVAL is returned when O_REGULAR is combined with O_DIRECTORY (not part of O_TMPFILE) because it doesn't make sense to open a path that is both a directory and a regular file.

---
 arch/alpha/include/uapi/asm/errno.h        | 2 ++
 arch/alpha/include/uapi/asm/fcntl.h        | 1 +
 arch/mips/include/uapi/asm/errno.h         | 2 ++
 arch/parisc/include/uapi/asm/errno.h       | 2 ++
 arch/parisc/include/uapi/asm/fcntl.h       | 1 +
 arch/sparc/include/uapi/asm/errno.h        | 2 ++
 arch/sparc/include/uapi/asm/fcntl.h        | 1 +
 fs/fcntl.c                                 | 2 +-
 fs/namei.c                                 | 6 ++++++
 fs/open.c                                  | 4 +++-
 include/linux/fcntl.h                      | 2 +-
 include/uapi/asm-generic/errno.h           | 2 ++
 include/uapi/asm-generic/fcntl.h           | 4 ++++
 tools/arch/alpha/include/uapi/asm/errno.h  | 2 ++
 tools/arch/mips/include/uapi/asm/errno.h   | 2 ++
 tools/arch/parisc/include/uapi/asm/errno.h | 2 ++
 tools/arch/sparc/include/uapi/asm/errno.h  | 2 ++
 tools/include/uapi/asm-generic/errno.h     | 2 ++
 18 files changed, 38 insertions(+), 3 deletions(-)
arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
arch/alpha/include/uapi/asm/fcntl.h +1 -0
index 50bdc8e8a271..4da5a64c23bd 100644
@@ -34,6 +34,7 @@
 
 #define O_PATH               040000000
 #define __O_TMPFILE  0100000000
+#define O_REGULAR    0200000000
 
 #define F_GETLK              7
 #define F_SETLK              8
arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
arch/parisc/include/uapi/asm/fcntl.h +1 -0
index 03dee816cb13..0cc3320fe326 100644
@@ -19,6 +19,7 @@
 
 #define O_PATH               020000000
 #define __O_TMPFILE  040000000
+#define O_REGULAR    0100000000
 
 #define F_GETLK64    8
 #define F_SETLK64    9
arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
arch/sparc/include/uapi/asm/fcntl.h +1 -0
index 67dae75e5274..a93d18d2c23e 100644
@@ -37,6 +37,7 @@
 
 #define O_PATH               0x1000000
 #define __O_TMPFILE  0x2000000
+#define O_REGULAR    0x4000000
 
 #define F_GETOWN     5       /*  for sockets. */
 #define F_SETOWN     6       /*  for sockets. */
fs/fcntl.c +1 -1
index f93dbca08435..62ab4ad2b6f5 100644
@@ -1169,7 +1169,7 @@ static int __init fcntl_init(void)
       * Exceptions: O_NONBLOCK is a two bit define on parisc; O_NDELAY
       * is defined as O_NONBLOCK on some platforms and not on others.
       */
-     BUILD_BUG_ON(20 - 1 /* for O_RDONLY being 0 */ !=
+     BUILD_BUG_ON(21 - 1 /* for O_RDONLY being 0 */ !=
              HWEIGHT32(
                      (VALID_OPEN_FLAGS & ~(O_NONBLOCK | O_NDELAY)) |
                      __FMODE_EXEC));
fs/namei.c +6 -0
index b28ecb699f32..f5504ae4b03c 100644
@@ -4616,6 +4616,10 @@ static int do_open(struct nameidata *nd,
              if (unlikely(error))
                      return error;
      }
+
+     if ((open_flag & O_REGULAR) && !d_is_reg(nd->path.dentry))
+             return -ENOTREG;
+
      if ((nd->flags & LOOKUP_DIRECTORY) && !d_can_lookup(nd->path.dentry))
              return -ENOTDIR;
 
@@ -4765,6 +4769,8 @@ static int do_o_path(struct nameidata *nd, unsigned flags, struct file *file)
      struct path path;
      int error = path_lookupat(nd, flags, &path);
      if (!error) {
+             if ((file->f_flags & O_REGULAR) && !d_is_reg(path.dentry))
+                     return -ENOTREG;
              audit_inode(nd->name, path.dentry, 0);
              error = vfs_open(&path, file);
              path_put(&path);
fs/open.c +3 -1
index 74c4c1462b3e..82153e21907e 100644
@@ -1173,7 +1173,7 @@ struct file *kernel_file_open(const struct path *path, int flags,
 EXPORT_SYMBOL_GPL(kernel_file_open);
 
 #define WILL_CREATE(flags)   (flags & (O_CREAT | __O_TMPFILE))
-#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC)
+#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC | O_REGULAR)
 
 inline struct open_how build_open_how(int flags, umode_t mode)
 {
@@ -1250,6 +1250,8 @@ inline int build_open_flags(const struct open_how *how, struct open_flags *op)
                      return -EINVAL;
              if (!(acc_mode & MAY_WRITE))
                      return -EINVAL;
+     } else if ((flags & O_DIRECTORY) && (flags & O_REGULAR)) {
+             return -EINVAL;
      }
      if (flags & O_PATH) {
              /* O_PATH only permits certain other flags to be set. */
include/linux/fcntl.h +1 -1
index a332e79b3207..4fd07b0e0a17 100644
@@ -10,7 +10,7 @@
      (O_RDONLY | O_WRONLY | O_RDWR | O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC | \
       O_APPEND | O_NDELAY | O_NONBLOCK | __O_SYNC | O_DSYNC | \
       FASYNC | O_DIRECT | O_LARGEFILE | O_DIRECTORY | O_NOFOLLOW | \
-      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE)
+      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE | O_REGULAR)
 
 /* List of all valid flags for the how->resolve argument: */
 #define VALID_RESOLVE_FLAGS \
include/uapi/asm-generic/errno.h +2 -0
index 92e7ae493ee3..2216ab9aa32e 100644
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif
include/uapi/asm-generic/fcntl.h +4 -0
index 613475285643..3468b352a575 100644
@@ -88,6 +88,10 @@
 #define __O_TMPFILE  020000000
 #endif
 
+#ifndef O_REGULAR
+#define O_REGULAR    040000000
+#endif
+
 /* a horrid kludge trying to make sure that this will fail on old kernels */
 #define O_TMPFILE (__O_TMPFILE | O_DIRECTORY)
 
tools/arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
tools/arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
tools/arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
tools/arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
tools/include/uapi/asm-generic/errno.h +0 -0
index 92e7ae493ee3..2216ab9aa32e 100644

tools/include/uapi/asm-generic/errno.h line 127

Quoted 6 lines
tools/include/uapi/asm-generic/errno.h +2 -0
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif

One thing this patch is missing is handling for ->atomic_open(). I imagine most of the filesystems that provide that op can't support O_REGULAR properly (maybe cifs can? idk). What you probably want to do is add in some patches that make all of the atomic_open operations in the kernel return -EINVAL if O_REGULAR is set.

Quoted 2 lines

Then, once the basic support is in, you or someone else can go back and implement support for O_REGULAR where possible.

Quoted 11 lines

Thank you for the feedback. I don't quite understand what I need to fix. I thought open system calls always create regular files, so atomic_open probably always creates regular files? Can you please give me some more details as to where I need to fix this and what the actual bug here is that is related to atomic_open? I think I had done some normal testing and when using O_CREAT | O_REGULAR, if the file doesn't exist, the file gets created and the file that gets created is a regular file, so it probably makes sense? Or should the behavior be that if file doesn't exist, -EINVAL is returned and if file exists it is opened if regular, otherwise -ENOTREG is returned?

atomic_open() is a combination lookup+open for when the dentry isn't present in the dcache. The normal open codepath that you're patching does not get called in this case when ->atomic_open is set for the filesystem. It's mostly used by network filesystems that need to optimize away the lookup since it's wasted round trip, and is often racy anyway. Your patchset doesn't address those filesystems. They will likely end up ignoring O_REGULAR in that case, which is not what you want.

What I was suggesting is that, as an interim step, you find all of the atomic_open operations in the kernel (there are maybe a dozen or so), and just make them return -EINVAL if someone sets O_DIRECTORY. Later, you or someone else can then go back and do a proper implementation of O_REGULAR handling on those filesystems, at least on the ones where it's possible. You will probably also need to similarly patch the open() routines for those filesystems too. Otherwise you'll get inconsistent behavior vs. when the dentry is in the cache.

One note: I think NFS probably can support O_DIRECTORY, since its OPEN call only works on files. We'll need to change how we handle errors from the server when it's set though.

Signature · 1 line
Jeff Layton <[email protected]>
#9 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

Quoted 312 lines · Jeff Layton
Quoted 289 lines · Dorjoy Chowdhury
Quoted 285 lines · Jeff Layton
Quoted 270 lines · Dorjoy Chowdhury

This flag indicates the path should be opened if it's a regular file. This is useful to write secure programs that want to avoid being tricked into opening device nodes with special semantics while thinking they operate on regular files.

A corresponding error code ENOTREG has been introduced. For example, if open is called on path /dev/null with O_REGULAR in the flag param, it will return -ENOTREG.

When used in combination with O_CREAT, either the regular file is created, or if the path already exists, it is opened if it's a regular file. Otherwise, -ENOTREG is returned.

-EINVAL is returned when O_REGULAR is combined with O_DIRECTORY (not part of O_TMPFILE) because it doesn't make sense to open a path that is both a directory and a regular file.

---
 arch/alpha/include/uapi/asm/errno.h        | 2 ++
 arch/alpha/include/uapi/asm/fcntl.h        | 1 +
 arch/mips/include/uapi/asm/errno.h         | 2 ++
 arch/parisc/include/uapi/asm/errno.h       | 2 ++
 arch/parisc/include/uapi/asm/fcntl.h       | 1 +
 arch/sparc/include/uapi/asm/errno.h        | 2 ++
 arch/sparc/include/uapi/asm/fcntl.h        | 1 +
 fs/fcntl.c                                 | 2 +-
 fs/namei.c                                 | 6 ++++++
 fs/open.c                                  | 4 +++-
 include/linux/fcntl.h                      | 2 +-
 include/uapi/asm-generic/errno.h           | 2 ++
 include/uapi/asm-generic/fcntl.h           | 4 ++++
 tools/arch/alpha/include/uapi/asm/errno.h  | 2 ++
 tools/arch/mips/include/uapi/asm/errno.h   | 2 ++
 tools/arch/parisc/include/uapi/asm/errno.h | 2 ++
 tools/arch/sparc/include/uapi/asm/errno.h  | 2 ++
 tools/include/uapi/asm-generic/errno.h     | 2 ++
 18 files changed, 38 insertions(+), 3 deletions(-)
arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
arch/alpha/include/uapi/asm/fcntl.h +1 -0
index 50bdc8e8a271..4da5a64c23bd 100644
@@ -34,6 +34,7 @@
 
 #define O_PATH               040000000
 #define __O_TMPFILE  0100000000
+#define O_REGULAR    0200000000
 
 #define F_GETLK              7
 #define F_SETLK              8
arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
arch/parisc/include/uapi/asm/fcntl.h +1 -0
index 03dee816cb13..0cc3320fe326 100644
@@ -19,6 +19,7 @@
 
 #define O_PATH               020000000
 #define __O_TMPFILE  040000000
+#define O_REGULAR    0100000000
 
 #define F_GETLK64    8
 #define F_SETLK64    9
arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
arch/sparc/include/uapi/asm/fcntl.h +1 -0
index 67dae75e5274..a93d18d2c23e 100644
@@ -37,6 +37,7 @@
 
 #define O_PATH               0x1000000
 #define __O_TMPFILE  0x2000000
+#define O_REGULAR    0x4000000
 
 #define F_GETOWN     5       /*  for sockets. */
 #define F_SETOWN     6       /*  for sockets. */
fs/fcntl.c +1 -1
index f93dbca08435..62ab4ad2b6f5 100644
@@ -1169,7 +1169,7 @@ static int __init fcntl_init(void)
       * Exceptions: O_NONBLOCK is a two bit define on parisc; O_NDELAY
       * is defined as O_NONBLOCK on some platforms and not on others.
       */
-     BUILD_BUG_ON(20 - 1 /* for O_RDONLY being 0 */ !=
+     BUILD_BUG_ON(21 - 1 /* for O_RDONLY being 0 */ !=
              HWEIGHT32(
                      (VALID_OPEN_FLAGS & ~(O_NONBLOCK | O_NDELAY)) |
                      __FMODE_EXEC));
fs/namei.c +6 -0
index b28ecb699f32..f5504ae4b03c 100644
@@ -4616,6 +4616,10 @@ static int do_open(struct nameidata *nd,
              if (unlikely(error))
                      return error;
      }
+
+     if ((open_flag & O_REGULAR) && !d_is_reg(nd->path.dentry))
+             return -ENOTREG;
+
      if ((nd->flags & LOOKUP_DIRECTORY) && !d_can_lookup(nd->path.dentry))
              return -ENOTDIR;
 
@@ -4765,6 +4769,8 @@ static int do_o_path(struct nameidata *nd, unsigned flags, struct file *file)
      struct path path;
      int error = path_lookupat(nd, flags, &path);
      if (!error) {
+             if ((file->f_flags & O_REGULAR) && !d_is_reg(path.dentry))
+                     return -ENOTREG;
              audit_inode(nd->name, path.dentry, 0);
              error = vfs_open(&path, file);
              path_put(&path);
fs/open.c +3 -1
index 74c4c1462b3e..82153e21907e 100644
@@ -1173,7 +1173,7 @@ struct file *kernel_file_open(const struct path *path, int flags,
 EXPORT_SYMBOL_GPL(kernel_file_open);
 
 #define WILL_CREATE(flags)   (flags & (O_CREAT | __O_TMPFILE))
-#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC)
+#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC | O_REGULAR)
 
 inline struct open_how build_open_how(int flags, umode_t mode)
 {
@@ -1250,6 +1250,8 @@ inline int build_open_flags(const struct open_how *how, struct open_flags *op)
                      return -EINVAL;
              if (!(acc_mode & MAY_WRITE))
                      return -EINVAL;
+     } else if ((flags & O_DIRECTORY) && (flags & O_REGULAR)) {
+             return -EINVAL;
      }
      if (flags & O_PATH) {
              /* O_PATH only permits certain other flags to be set. */
include/linux/fcntl.h +1 -1
index a332e79b3207..4fd07b0e0a17 100644
@@ -10,7 +10,7 @@
      (O_RDONLY | O_WRONLY | O_RDWR | O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC | \
       O_APPEND | O_NDELAY | O_NONBLOCK | __O_SYNC | O_DSYNC | \
       FASYNC | O_DIRECT | O_LARGEFILE | O_DIRECTORY | O_NOFOLLOW | \
-      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE)
+      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE | O_REGULAR)
 
 /* List of all valid flags for the how->resolve argument: */
 #define VALID_RESOLVE_FLAGS \
include/uapi/asm-generic/errno.h +2 -0
index 92e7ae493ee3..2216ab9aa32e 100644
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif
include/uapi/asm-generic/fcntl.h +4 -0
index 613475285643..3468b352a575 100644
@@ -88,6 +88,10 @@
 #define __O_TMPFILE  020000000
 #endif
 
+#ifndef O_REGULAR
+#define O_REGULAR    040000000
+#endif
+
 /* a horrid kludge trying to make sure that this will fail on old kernels */
 #define O_TMPFILE (__O_TMPFILE | O_DIRECTORY)
 
tools/arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
tools/arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
tools/arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
tools/arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
tools/include/uapi/asm-generic/errno.h +0 -0
index 92e7ae493ee3..2216ab9aa32e 100644

tools/include/uapi/asm-generic/errno.h line 127

Quoted 6 lines
tools/include/uapi/asm-generic/errno.h +2 -0
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif

One thing this patch is missing is handling for ->atomic_open(). I imagine most of the filesystems that provide that op can't support O_REGULAR properly (maybe cifs can? idk). What you probably want to do is add in some patches that make all of the atomic_open operations in the kernel return -EINVAL if O_REGULAR is set.

Quoted 2 lines

Then, once the basic support is in, you or someone else can go back and implement support for O_REGULAR where possible.

Quoted 11 lines

Thank you for the feedback. I don't quite understand what I need to fix. I thought open system calls always create regular files, so atomic_open probably always creates regular files? Can you please give me some more details as to where I need to fix this and what the actual bug here is that is related to atomic_open? I think I had done some normal testing and when using O_CREAT | O_REGULAR, if the file doesn't exist, the file gets created and the file that gets created is a regular file, so it probably makes sense? Or should the behavior be that if file doesn't exist, -EINVAL is returned and if file exists it is opened if regular, otherwise -ENOTREG is returned?

atomic_open() is a combination lookup+open for when the dentry isn't present in the dcache. The normal open codepath that you're patching does not get called in this case when ->atomic_open is set for the filesystem. It's mostly used by network filesystems that need to optimize away the lookup since it's wasted round trip, and is often racy anyway. Your patchset doesn't address those filesystems. They will likely end up ignoring O_REGULAR in that case, which is not what you want.

Quoted 3 lines

What I was suggesting is that, as an interim step, you find all of the atomic_open operations in the kernel (there are maybe a dozen or so), and just make them return -EINVAL if someone sets O_DIRECTORY. Later,

Sorry, I am just trying to fully understand this. Do you mean to return -EINVAL from all atomic_open implementations in the kernel if both O_REGULAR and O_DIRECTORY are set (or just only if O_REGULAR is set, we need to return -EINVAL)? I am already returning -EINVAL when both these are set from the build_open_flags function, so that should already handle the cases, right? I think after atomic_open get called, all code paths eventually go through the do_open function where I have this check "if ((open_flag & O_REGULAR) && !d_is_reg(nd->path.dentry)) return -ENOTREG". This is right before if ((nd->flags & LOOKUP_DIRECTORY) && !d_can_lookup(nd->path.dentry)) return -ENOTDIR; which I had initially followed. So should I just return -EINVAL from the atomic_open functions too if both O_REGULAR and O_DIRECTORY are set? Sorry if I am misunderstanding this.

Regards,
Dorjoy

#10 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

In reply to Jeff Layton
Quoted 320 lines · Jeff Layton
Quoted 289 lines · Dorjoy Chowdhury
Quoted 285 lines · Jeff Layton
Quoted 270 lines · Dorjoy Chowdhury

This flag indicates the path should be opened if it's a regular file. This is useful to write secure programs that want to avoid being tricked into opening device nodes with special semantics while thinking they operate on regular files.

A corresponding error code ENOTREG has been introduced. For example, if open is called on path /dev/null with O_REGULAR in the flag param, it will return -ENOTREG.

When used in combination with O_CREAT, either the regular file is created, or if the path already exists, it is opened if it's a regular file. Otherwise, -ENOTREG is returned.

-EINVAL is returned when O_REGULAR is combined with O_DIRECTORY (not part of O_TMPFILE) because it doesn't make sense to open a path that is both a directory and a regular file.

---
 arch/alpha/include/uapi/asm/errno.h        | 2 ++
 arch/alpha/include/uapi/asm/fcntl.h        | 1 +
 arch/mips/include/uapi/asm/errno.h         | 2 ++
 arch/parisc/include/uapi/asm/errno.h       | 2 ++
 arch/parisc/include/uapi/asm/fcntl.h       | 1 +
 arch/sparc/include/uapi/asm/errno.h        | 2 ++
 arch/sparc/include/uapi/asm/fcntl.h        | 1 +
 fs/fcntl.c                                 | 2 +-
 fs/namei.c                                 | 6 ++++++
 fs/open.c                                  | 4 +++-
 include/linux/fcntl.h                      | 2 +-
 include/uapi/asm-generic/errno.h           | 2 ++
 include/uapi/asm-generic/fcntl.h           | 4 ++++
 tools/arch/alpha/include/uapi/asm/errno.h  | 2 ++
 tools/arch/mips/include/uapi/asm/errno.h   | 2 ++
 tools/arch/parisc/include/uapi/asm/errno.h | 2 ++
 tools/arch/sparc/include/uapi/asm/errno.h  | 2 ++
 tools/include/uapi/asm-generic/errno.h     | 2 ++
 18 files changed, 38 insertions(+), 3 deletions(-)
arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
arch/alpha/include/uapi/asm/fcntl.h +1 -0
index 50bdc8e8a271..4da5a64c23bd 100644
@@ -34,6 +34,7 @@
 
 #define O_PATH               040000000
 #define __O_TMPFILE  0100000000
+#define O_REGULAR    0200000000
 
 #define F_GETLK              7
 #define F_SETLK              8
arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
arch/parisc/include/uapi/asm/fcntl.h +1 -0
index 03dee816cb13..0cc3320fe326 100644
@@ -19,6 +19,7 @@
 
 #define O_PATH               020000000
 #define __O_TMPFILE  040000000
+#define O_REGULAR    0100000000
 
 #define F_GETLK64    8
 #define F_SETLK64    9
arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
arch/sparc/include/uapi/asm/fcntl.h +1 -0
index 67dae75e5274..a93d18d2c23e 100644
@@ -37,6 +37,7 @@
 
 #define O_PATH               0x1000000
 #define __O_TMPFILE  0x2000000
+#define O_REGULAR    0x4000000
 
 #define F_GETOWN     5       /*  for sockets. */
 #define F_SETOWN     6       /*  for sockets. */
fs/fcntl.c +1 -1
index f93dbca08435..62ab4ad2b6f5 100644
@@ -1169,7 +1169,7 @@ static int __init fcntl_init(void)
       * Exceptions: O_NONBLOCK is a two bit define on parisc; O_NDELAY
       * is defined as O_NONBLOCK on some platforms and not on others.
       */
-     BUILD_BUG_ON(20 - 1 /* for O_RDONLY being 0 */ !=
+     BUILD_BUG_ON(21 - 1 /* for O_RDONLY being 0 */ !=
              HWEIGHT32(
                      (VALID_OPEN_FLAGS & ~(O_NONBLOCK | O_NDELAY)) |
                      __FMODE_EXEC));
fs/namei.c +6 -0
index b28ecb699f32..f5504ae4b03c 100644
@@ -4616,6 +4616,10 @@ static int do_open(struct nameidata *nd,
              if (unlikely(error))
                      return error;
      }
+
+     if ((open_flag & O_REGULAR) && !d_is_reg(nd->path.dentry))
+             return -ENOTREG;
+
      if ((nd->flags & LOOKUP_DIRECTORY) && !d_can_lookup(nd->path.dentry))
              return -ENOTDIR;
 
@@ -4765,6 +4769,8 @@ static int do_o_path(struct nameidata *nd, unsigned flags, struct file *file)
      struct path path;
      int error = path_lookupat(nd, flags, &path);
      if (!error) {
+             if ((file->f_flags & O_REGULAR) && !d_is_reg(path.dentry))
+                     return -ENOTREG;
              audit_inode(nd->name, path.dentry, 0);
              error = vfs_open(&path, file);
              path_put(&path);
fs/open.c +3 -1
index 74c4c1462b3e..82153e21907e 100644
@@ -1173,7 +1173,7 @@ struct file *kernel_file_open(const struct path *path, int flags,
 EXPORT_SYMBOL_GPL(kernel_file_open);
 
 #define WILL_CREATE(flags)   (flags & (O_CREAT | __O_TMPFILE))
-#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC)
+#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC | O_REGULAR)
 
 inline struct open_how build_open_how(int flags, umode_t mode)
 {
@@ -1250,6 +1250,8 @@ inline int build_open_flags(const struct open_how *how, struct open_flags *op)
                      return -EINVAL;
              if (!(acc_mode & MAY_WRITE))
                      return -EINVAL;
+     } else if ((flags & O_DIRECTORY) && (flags & O_REGULAR)) {
+             return -EINVAL;
      }
      if (flags & O_PATH) {
              /* O_PATH only permits certain other flags to be set. */
include/linux/fcntl.h +1 -1
index a332e79b3207..4fd07b0e0a17 100644
@@ -10,7 +10,7 @@
      (O_RDONLY | O_WRONLY | O_RDWR | O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC | \
       O_APPEND | O_NDELAY | O_NONBLOCK | __O_SYNC | O_DSYNC | \
       FASYNC | O_DIRECT | O_LARGEFILE | O_DIRECTORY | O_NOFOLLOW | \
-      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE)
+      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE | O_REGULAR)
 
 /* List of all valid flags for the how->resolve argument: */
 #define VALID_RESOLVE_FLAGS \
include/uapi/asm-generic/errno.h +2 -0
index 92e7ae493ee3..2216ab9aa32e 100644
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif
include/uapi/asm-generic/fcntl.h +4 -0
index 613475285643..3468b352a575 100644
@@ -88,6 +88,10 @@
 #define __O_TMPFILE  020000000
 #endif
 
+#ifndef O_REGULAR
+#define O_REGULAR    040000000
+#endif
+
 /* a horrid kludge trying to make sure that this will fail on old kernels */
 #define O_TMPFILE (__O_TMPFILE | O_DIRECTORY)
 
tools/arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
tools/arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
tools/arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
tools/arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
tools/include/uapi/asm-generic/errno.h +0 -0
index 92e7ae493ee3..2216ab9aa32e 100644

tools/include/uapi/asm-generic/errno.h line 127

Quoted 6 lines
tools/include/uapi/asm-generic/errno.h +2 -0
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif

One thing this patch is missing is handling for ->atomic_open(). I imagine most of the filesystems that provide that op can't support O_REGULAR properly (maybe cifs can? idk). What you probably want to do is add in some patches that make all of the atomic_open operations in the kernel return -EINVAL if O_REGULAR is set.

Quoted 2 lines

Then, once the basic support is in, you or someone else can go back and implement support for O_REGULAR where possible.

Quoted 11 lines

Thank you for the feedback. I don't quite understand what I need to fix. I thought open system calls always create regular files, so atomic_open probably always creates regular files? Can you please give me some more details as to where I need to fix this and what the actual bug here is that is related to atomic_open? I think I had done some normal testing and when using O_CREAT | O_REGULAR, if the file doesn't exist, the file gets created and the file that gets created is a regular file, so it probably makes sense? Or should the behavior be that if file doesn't exist, -EINVAL is returned and if file exists it is opened if regular, otherwise -ENOTREG is returned?

atomic_open() is a combination lookup+open for when the dentry isn't present in the dcache. The normal open codepath that you're patching does not get called in this case when ->atomic_open is set for the filesystem. It's mostly used by network filesystems that need to optimize away the lookup since it's wasted round trip, and is often racy anyway. Your patchset doesn't address those filesystems. They will likely end up ignoring O_REGULAR in that case, which is not what you want.

What I was suggesting is that, as an interim step, you find all of the atomic_open operations in the kernel (there are maybe a dozen or so), and just make them return -EINVAL if someone sets O_DIRECTORY. Later, you or someone else can then go back and do a proper implementation of O_REGULAR handling on those filesystems, at least on the ones where it's possible. You will probably also need to similarly patch the open() routines for those filesystems too. Otherwise you'll get inconsistent behavior vs. when the dentry is in the cache.

Quoted 3 lines

One note: I think NFS probably can support O_DIRECTORY, since its OPEN call only works on files. We'll need to change how we handle errors from the server when it's set though.

So I think you're proposing two separate things or there's a typo:

(1) blocking O_DIRECTORY for ->atomic_open::
(2) blocking O_REGULAR for ->atomic_open::

The (1) point implies that O_DIRECTORY currently doesn't work correctly with atomic open for all filesystems.

Ever since 43b450632676 ("open: return EINVAL for O_DIRECTORY | O_CREAT") O_DIRECTORY with O_CREAT is blocked. It was accidently allowed and completely broken before that.

For O_DIRECTORY without O_CREAT the kernel will pass that down through ->atomic_open:: to the filesystem.

The worry that I see is that a network filesystem via ->atomic_open::
somehow already called open on the server side on something that wasn't a directory. At that point the damage such as side-effects from device opening is already done.

But I suspect that every filesystem implementing ->atomic_open:: just does finish_no_open() and punts to the VFS for the actual open. And the VFS will catch it in do_open() for it actually opens the file. So the only real worry for O_DIRECTORY I see is that there's an fs that handles it wrong.

For (2) it is problematic as there surely are filesystems with ->atomic_open:: that do handle the ~O_CREAT case and return with FMODE_OPENED. So that'll be problematic if the intention is to not trigger an actual open on a non-regular file such as a device/socket/fifo etc. before the VFS had a chance to validate what's going on.

So I'm not excited about having this 70% working and punting on ->atomic_open:: waiting for someone to fix this. One option would be to bypass ->atomic_open:: for OPENAT2_REGULAR without O_CREAT and fallback to racy and pricy lookup + open for now. How problematic would that be? If possible I'd prefer this a lot over merging something that works half-way.

I guess to make that really work you'd need some protocol extension?

maintainer
#11 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

Quoted 337 lines · Christian Brauner
Quoted 320 lines · Jeff Layton
Quoted 289 lines · Dorjoy Chowdhury
Quoted 285 lines · Jeff Layton
Quoted 270 lines · Dorjoy Chowdhury

This flag indicates the path should be opened if it's a regular file. This is useful to write secure programs that want to avoid being tricked into opening device nodes with special semantics while thinking they operate on regular files.

A corresponding error code ENOTREG has been introduced. For example, if open is called on path /dev/null with O_REGULAR in the flag param, it will return -ENOTREG.

When used in combination with O_CREAT, either the regular file is created, or if the path already exists, it is opened if it's a regular file. Otherwise, -ENOTREG is returned.

-EINVAL is returned when O_REGULAR is combined with O_DIRECTORY (not part of O_TMPFILE) because it doesn't make sense to open a path that is both a directory and a regular file.

---
 arch/alpha/include/uapi/asm/errno.h        | 2 ++
 arch/alpha/include/uapi/asm/fcntl.h        | 1 +
 arch/mips/include/uapi/asm/errno.h         | 2 ++
 arch/parisc/include/uapi/asm/errno.h       | 2 ++
 arch/parisc/include/uapi/asm/fcntl.h       | 1 +
 arch/sparc/include/uapi/asm/errno.h        | 2 ++
 arch/sparc/include/uapi/asm/fcntl.h        | 1 +
 fs/fcntl.c                                 | 2 +-
 fs/namei.c                                 | 6 ++++++
 fs/open.c                                  | 4 +++-
 include/linux/fcntl.h                      | 2 +-
 include/uapi/asm-generic/errno.h           | 2 ++
 include/uapi/asm-generic/fcntl.h           | 4 ++++
 tools/arch/alpha/include/uapi/asm/errno.h  | 2 ++
 tools/arch/mips/include/uapi/asm/errno.h   | 2 ++
 tools/arch/parisc/include/uapi/asm/errno.h | 2 ++
 tools/arch/sparc/include/uapi/asm/errno.h  | 2 ++
 tools/include/uapi/asm-generic/errno.h     | 2 ++
 18 files changed, 38 insertions(+), 3 deletions(-)
arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
arch/alpha/include/uapi/asm/fcntl.h +1 -0
index 50bdc8e8a271..4da5a64c23bd 100644
@@ -34,6 +34,7 @@
 
 #define O_PATH               040000000
 #define __O_TMPFILE  0100000000
+#define O_REGULAR    0200000000
 
 #define F_GETLK              7
 #define F_SETLK              8
arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
arch/parisc/include/uapi/asm/fcntl.h +1 -0
index 03dee816cb13..0cc3320fe326 100644
@@ -19,6 +19,7 @@
 
 #define O_PATH               020000000
 #define __O_TMPFILE  040000000
+#define O_REGULAR    0100000000
 
 #define F_GETLK64    8
 #define F_SETLK64    9
arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
arch/sparc/include/uapi/asm/fcntl.h +1 -0
index 67dae75e5274..a93d18d2c23e 100644
@@ -37,6 +37,7 @@
 
 #define O_PATH               0x1000000
 #define __O_TMPFILE  0x2000000
+#define O_REGULAR    0x4000000
 
 #define F_GETOWN     5       /*  for sockets. */
 #define F_SETOWN     6       /*  for sockets. */
fs/fcntl.c +1 -1
index f93dbca08435..62ab4ad2b6f5 100644
@@ -1169,7 +1169,7 @@ static int __init fcntl_init(void)
       * Exceptions: O_NONBLOCK is a two bit define on parisc; O_NDELAY
       * is defined as O_NONBLOCK on some platforms and not on others.
       */
-     BUILD_BUG_ON(20 - 1 /* for O_RDONLY being 0 */ !=
+     BUILD_BUG_ON(21 - 1 /* for O_RDONLY being 0 */ !=
              HWEIGHT32(
                      (VALID_OPEN_FLAGS & ~(O_NONBLOCK | O_NDELAY)) |
                      __FMODE_EXEC));
fs/namei.c +6 -0
index b28ecb699f32..f5504ae4b03c 100644
@@ -4616,6 +4616,10 @@ static int do_open(struct nameidata *nd,
              if (unlikely(error))
                      return error;
      }
+
+     if ((open_flag & O_REGULAR) && !d_is_reg(nd->path.dentry))
+             return -ENOTREG;
+
      if ((nd->flags & LOOKUP_DIRECTORY) && !d_can_lookup(nd->path.dentry))
              return -ENOTDIR;
 
@@ -4765,6 +4769,8 @@ static int do_o_path(struct nameidata *nd, unsigned flags, struct file *file)
      struct path path;
      int error = path_lookupat(nd, flags, &path);
      if (!error) {
+             if ((file->f_flags & O_REGULAR) && !d_is_reg(path.dentry))
+                     return -ENOTREG;
              audit_inode(nd->name, path.dentry, 0);
              error = vfs_open(&path, file);
              path_put(&path);
fs/open.c +3 -1
index 74c4c1462b3e..82153e21907e 100644
@@ -1173,7 +1173,7 @@ struct file *kernel_file_open(const struct path *path, int flags,
 EXPORT_SYMBOL_GPL(kernel_file_open);
 
 #define WILL_CREATE(flags)   (flags & (O_CREAT | __O_TMPFILE))
-#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC)
+#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC | O_REGULAR)
 
 inline struct open_how build_open_how(int flags, umode_t mode)
 {
@@ -1250,6 +1250,8 @@ inline int build_open_flags(const struct open_how *how, struct open_flags *op)
                      return -EINVAL;
              if (!(acc_mode & MAY_WRITE))
                      return -EINVAL;
+     } else if ((flags & O_DIRECTORY) && (flags & O_REGULAR)) {
+             return -EINVAL;
      }
      if (flags & O_PATH) {
              /* O_PATH only permits certain other flags to be set. */
include/linux/fcntl.h +1 -1
index a332e79b3207..4fd07b0e0a17 100644
@@ -10,7 +10,7 @@
      (O_RDONLY | O_WRONLY | O_RDWR | O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC | \
       O_APPEND | O_NDELAY | O_NONBLOCK | __O_SYNC | O_DSYNC | \
       FASYNC | O_DIRECT | O_LARGEFILE | O_DIRECTORY | O_NOFOLLOW | \
-      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE)
+      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE | O_REGULAR)
 
 /* List of all valid flags for the how->resolve argument: */
 #define VALID_RESOLVE_FLAGS \
include/uapi/asm-generic/errno.h +2 -0
index 92e7ae493ee3..2216ab9aa32e 100644
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif
include/uapi/asm-generic/fcntl.h +4 -0
index 613475285643..3468b352a575 100644
@@ -88,6 +88,10 @@
 #define __O_TMPFILE  020000000
 #endif
 
+#ifndef O_REGULAR
+#define O_REGULAR    040000000
+#endif
+
 /* a horrid kludge trying to make sure that this will fail on old kernels */
 #define O_TMPFILE (__O_TMPFILE | O_DIRECTORY)
 
tools/arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
tools/arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
tools/arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
tools/arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
tools/include/uapi/asm-generic/errno.h +0 -0
index 92e7ae493ee3..2216ab9aa32e 100644

tools/include/uapi/asm-generic/errno.h line 127

Quoted 6 lines
tools/include/uapi/asm-generic/errno.h +2 -0
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif

One thing this patch is missing is handling for ->atomic_open(). I imagine most of the filesystems that provide that op can't support O_REGULAR properly (maybe cifs can? idk). What you probably want to do is add in some patches that make all of the atomic_open operations in the kernel return -EINVAL if O_REGULAR is set.

Quoted 2 lines

Then, once the basic support is in, you or someone else can go back and implement support for O_REGULAR where possible.

Quoted 11 lines

Thank you for the feedback. I don't quite understand what I need to fix. I thought open system calls always create regular files, so atomic_open probably always creates regular files? Can you please give me some more details as to where I need to fix this and what the actual bug here is that is related to atomic_open? I think I had done some normal testing and when using O_CREAT | O_REGULAR, if the file doesn't exist, the file gets created and the file that gets created is a regular file, so it probably makes sense? Or should the behavior be that if file doesn't exist, -EINVAL is returned and if file exists it is opened if regular, otherwise -ENOTREG is returned?

atomic_open() is a combination lookup+open for when the dentry isn't present in the dcache. The normal open codepath that you're patching does not get called in this case when ->atomic_open is set for the filesystem. It's mostly used by network filesystems that need to optimize away the lookup since it's wasted round trip, and is often racy anyway. Your patchset doesn't address those filesystems. They will likely end up ignoring O_REGULAR in that case, which is not what you want.

What I was suggesting is that, as an interim step, you find all of the atomic_open operations in the kernel (there are maybe a dozen or so), and just make them return -EINVAL if someone sets O_DIRECTORY. Later, you or someone else can then go back and do a proper implementation of O_REGULAR handling on those filesystems, at least on the ones where it's possible. You will probably also need to similarly patch the open() routines for those filesystems too. Otherwise you'll get inconsistent behavior vs. when the dentry is in the cache.

Quoted 3 lines

One note: I think NFS probably can support O_DIRECTORY, since its OPEN call only works on files. We'll need to change how we handle errors from the server when it's set though.

So I think you're proposing two separate things or there's a typo:

(1) blocking O_DIRECTORY for ->atomic_open::
(2) blocking O_REGULAR for ->atomic_open::

The (1) point implies that O_DIRECTORY currently doesn't work correctly with atomic open for all filesystems.

Ever since 43b450632676 ("open: return EINVAL for O_DIRECTORY | O_CREAT") O_DIRECTORY with O_CREAT is blocked. It was accidently allowed and completely broken before that.

Quoted 9 lines

For O_DIRECTORY without O_CREAT the kernel will pass that down through ->atomic_open:: to the filesystem.

The worry that I see is that a network filesystem via ->atomic_open::
somehow already called open on the server side on something that wasn't a directory. At that point the damage such as side-effects from device opening is already done.

Exactly. I guess you could send an immediate close, but that's not without side effects.

Quoted 14 lines

But I suspect that every filesystem implementing ->atomic_open:: just does finish_no_open() and punts to the VFS for the actual open. And the VFS will catch it in do_open() for it actually opens the file. So the only real worry for O_DIRECTORY I see is that there's an fs that handles it wrong.

For (2) it is problematic as there surely are filesystems with ->atomic_open:: that do handle the ~O_CREAT case and return with FMODE_OPENED. So that'll be problematic if the intention is to not trigger an actual open on a non-regular file such as a device/socket/fifo etc. before the VFS had a chance to validate what's going on.

Quoted 8 lines

So I'm not excited about having this 70% working and punting on ->atomic_open:: waiting for someone to fix this. One option would be to bypass ->atomic_open:: for OPENAT2_REGULAR without O_CREAT and fallback to racy and pricy lookup + open for now. How problematic would that be? If possible I'd prefer this a lot over merging something that works half-way.

I guess to make that really work you'd need some protocol extension?

For NFS, I think we're OK. The OPEN call on NFSv4 only works for regular files, so it should be able to handle O_REGULAR. We just need to rejigger the error handling when it's set (just return an error instead of doing the open of a directory or whatever it is).

The others (at a quick glance):

cifs: I don't see a way to specify an O_REGULAR equivalent to the SMB2_CREATE call and it looks like it can create directories. Maybe SteveF (cc'ed) knows if this is possible?

ceph: I think CEPH_MDS_OP_OPEN might only work for files, in which case O_REGULAR can probably be supported similarly to NFS.

fuse: probably ok? Does finish_no_open() in most cases. May depend on the userland driver though.

gfs2: is ok, it just does finish_no_open() in most cases anyway

vboxsf: does finish_no_open on non-creates, so you could probably just punt to that if O_REGULAR is set.

So, it's probably possible to do this across the board. I'm not sure about cifs though.

Signature · 1 line
Jeff Layton <[email protected]>
maintainer
#12 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

Quoted 379 lines · Jeff Layton
Quoted 337 lines · Christian Brauner
Quoted 320 lines · Jeff Layton
Quoted 289 lines · Dorjoy Chowdhury
Quoted 285 lines · Jeff Layton
Quoted 270 lines · Dorjoy Chowdhury

This flag indicates the path should be opened if it's a regular file. This is useful to write secure programs that want to avoid being tricked into opening device nodes with special semantics while thinking they operate on regular files.

A corresponding error code ENOTREG has been introduced. For example, if open is called on path /dev/null with O_REGULAR in the flag param, it will return -ENOTREG.

When used in combination with O_CREAT, either the regular file is created, or if the path already exists, it is opened if it's a regular file. Otherwise, -ENOTREG is returned.

-EINVAL is returned when O_REGULAR is combined with O_DIRECTORY (not part of O_TMPFILE) because it doesn't make sense to open a path that is both a directory and a regular file.

---
 arch/alpha/include/uapi/asm/errno.h        | 2 ++
 arch/alpha/include/uapi/asm/fcntl.h        | 1 +
 arch/mips/include/uapi/asm/errno.h         | 2 ++
 arch/parisc/include/uapi/asm/errno.h       | 2 ++
 arch/parisc/include/uapi/asm/fcntl.h       | 1 +
 arch/sparc/include/uapi/asm/errno.h        | 2 ++
 arch/sparc/include/uapi/asm/fcntl.h        | 1 +
 fs/fcntl.c                                 | 2 +-
 fs/namei.c                                 | 6 ++++++
 fs/open.c                                  | 4 +++-
 include/linux/fcntl.h                      | 2 +-
 include/uapi/asm-generic/errno.h           | 2 ++
 include/uapi/asm-generic/fcntl.h           | 4 ++++
 tools/arch/alpha/include/uapi/asm/errno.h  | 2 ++
 tools/arch/mips/include/uapi/asm/errno.h   | 2 ++
 tools/arch/parisc/include/uapi/asm/errno.h | 2 ++
 tools/arch/sparc/include/uapi/asm/errno.h  | 2 ++
 tools/include/uapi/asm-generic/errno.h     | 2 ++
 18 files changed, 38 insertions(+), 3 deletions(-)
arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
arch/alpha/include/uapi/asm/fcntl.h +1 -0
index 50bdc8e8a271..4da5a64c23bd 100644
@@ -34,6 +34,7 @@
 
 #define O_PATH               040000000
 #define __O_TMPFILE  0100000000
+#define O_REGULAR    0200000000
 
 #define F_GETLK              7
 #define F_SETLK              8
arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
arch/parisc/include/uapi/asm/fcntl.h +1 -0
index 03dee816cb13..0cc3320fe326 100644
@@ -19,6 +19,7 @@
 
 #define O_PATH               020000000
 #define __O_TMPFILE  040000000
+#define O_REGULAR    0100000000
 
 #define F_GETLK64    8
 #define F_SETLK64    9
arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
arch/sparc/include/uapi/asm/fcntl.h +1 -0
index 67dae75e5274..a93d18d2c23e 100644
@@ -37,6 +37,7 @@
 
 #define O_PATH               0x1000000
 #define __O_TMPFILE  0x2000000
+#define O_REGULAR    0x4000000
 
 #define F_GETOWN     5       /*  for sockets. */
 #define F_SETOWN     6       /*  for sockets. */
fs/fcntl.c +1 -1
index f93dbca08435..62ab4ad2b6f5 100644
@@ -1169,7 +1169,7 @@ static int __init fcntl_init(void)
       * Exceptions: O_NONBLOCK is a two bit define on parisc; O_NDELAY
       * is defined as O_NONBLOCK on some platforms and not on others.
       */
-     BUILD_BUG_ON(20 - 1 /* for O_RDONLY being 0 */ !=
+     BUILD_BUG_ON(21 - 1 /* for O_RDONLY being 0 */ !=
              HWEIGHT32(
                      (VALID_OPEN_FLAGS & ~(O_NONBLOCK | O_NDELAY)) |
                      __FMODE_EXEC));
fs/namei.c +6 -0
index b28ecb699f32..f5504ae4b03c 100644
@@ -4616,6 +4616,10 @@ static int do_open(struct nameidata *nd,
              if (unlikely(error))
                      return error;
      }
+
+     if ((open_flag & O_REGULAR) && !d_is_reg(nd->path.dentry))
+             return -ENOTREG;
+
      if ((nd->flags & LOOKUP_DIRECTORY) && !d_can_lookup(nd->path.dentry))
              return -ENOTDIR;
 
@@ -4765,6 +4769,8 @@ static int do_o_path(struct nameidata *nd, unsigned flags, struct file *file)
      struct path path;
      int error = path_lookupat(nd, flags, &path);
      if (!error) {
+             if ((file->f_flags & O_REGULAR) && !d_is_reg(path.dentry))
+                     return -ENOTREG;
              audit_inode(nd->name, path.dentry, 0);
              error = vfs_open(&path, file);
              path_put(&path);
fs/open.c +3 -1
index 74c4c1462b3e..82153e21907e 100644
@@ -1173,7 +1173,7 @@ struct file *kernel_file_open(const struct path *path, int flags,
 EXPORT_SYMBOL_GPL(kernel_file_open);
 
 #define WILL_CREATE(flags)   (flags & (O_CREAT | __O_TMPFILE))
-#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC)
+#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC | O_REGULAR)
 
 inline struct open_how build_open_how(int flags, umode_t mode)
 {
@@ -1250,6 +1250,8 @@ inline int build_open_flags(const struct open_how *how, struct open_flags *op)
                      return -EINVAL;
              if (!(acc_mode & MAY_WRITE))
                      return -EINVAL;
+     } else if ((flags & O_DIRECTORY) && (flags & O_REGULAR)) {
+             return -EINVAL;
      }
      if (flags & O_PATH) {
              /* O_PATH only permits certain other flags to be set. */
include/linux/fcntl.h +1 -1
index a332e79b3207..4fd07b0e0a17 100644
@@ -10,7 +10,7 @@
      (O_RDONLY | O_WRONLY | O_RDWR | O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC | \
       O_APPEND | O_NDELAY | O_NONBLOCK | __O_SYNC | O_DSYNC | \
       FASYNC | O_DIRECT | O_LARGEFILE | O_DIRECTORY | O_NOFOLLOW | \
-      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE)
+      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE | O_REGULAR)
 
 /* List of all valid flags for the how->resolve argument: */
 #define VALID_RESOLVE_FLAGS \
include/uapi/asm-generic/errno.h +2 -0
index 92e7ae493ee3..2216ab9aa32e 100644
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif
include/uapi/asm-generic/fcntl.h +4 -0
index 613475285643..3468b352a575 100644
@@ -88,6 +88,10 @@
 #define __O_TMPFILE  020000000
 #endif
 
+#ifndef O_REGULAR
+#define O_REGULAR    040000000
+#endif
+
 /* a horrid kludge trying to make sure that this will fail on old kernels */
 #define O_TMPFILE (__O_TMPFILE | O_DIRECTORY)
 
tools/arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
tools/arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
tools/arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
tools/arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
tools/include/uapi/asm-generic/errno.h +0 -0
index 92e7ae493ee3..2216ab9aa32e 100644

tools/include/uapi/asm-generic/errno.h line 127

Quoted 6 lines
tools/include/uapi/asm-generic/errno.h +2 -0
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif

One thing this patch is missing is handling for ->atomic_open(). I imagine most of the filesystems that provide that op can't support O_REGULAR properly (maybe cifs can? idk). What you probably want to do is add in some patches that make all of the atomic_open operations in the kernel return -EINVAL if O_REGULAR is set.

Quoted 2 lines

Then, once the basic support is in, you or someone else can go back and implement support for O_REGULAR where possible.

Quoted 11 lines

Thank you for the feedback. I don't quite understand what I need to fix. I thought open system calls always create regular files, so atomic_open probably always creates regular files? Can you please give me some more details as to where I need to fix this and what the actual bug here is that is related to atomic_open? I think I had done some normal testing and when using O_CREAT | O_REGULAR, if the file doesn't exist, the file gets created and the file that gets created is a regular file, so it probably makes sense? Or should the behavior be that if file doesn't exist, -EINVAL is returned and if file exists it is opened if regular, otherwise -ENOTREG is returned?

atomic_open() is a combination lookup+open for when the dentry isn't present in the dcache. The normal open codepath that you're patching does not get called in this case when ->atomic_open is set for the filesystem. It's mostly used by network filesystems that need to optimize away the lookup since it's wasted round trip, and is often racy anyway. Your patchset doesn't address those filesystems. They will likely end up ignoring O_REGULAR in that case, which is not what you want.

What I was suggesting is that, as an interim step, you find all of the atomic_open operations in the kernel (there are maybe a dozen or so), and just make them return -EINVAL if someone sets O_DIRECTORY. Later, you or someone else can then go back and do a proper implementation of O_REGULAR handling on those filesystems, at least on the ones where it's possible. You will probably also need to similarly patch the open() routines for those filesystems too. Otherwise you'll get inconsistent behavior vs. when the dentry is in the cache.

Quoted 3 lines

One note: I think NFS probably can support O_DIRECTORY, since its OPEN call only works on files. We'll need to change how we handle errors from the server when it's set though.

So I think you're proposing two separate things or there's a typo:

(1) blocking O_DIRECTORY for ->atomic_open::
(2) blocking O_REGULAR for ->atomic_open::

The (1) point implies that O_DIRECTORY currently doesn't work correctly with atomic open for all filesystems.

Ever since 43b450632676 ("open: return EINVAL for O_DIRECTORY | O_CREAT") O_DIRECTORY with O_CREAT is blocked. It was accidently allowed and completely broken before that.

Quoted 9 lines

For O_DIRECTORY without O_CREAT the kernel will pass that down through ->atomic_open:: to the filesystem.

The worry that I see is that a network filesystem via ->atomic_open::
somehow already called open on the server side on something that wasn't a directory. At that point the damage such as side-effects from device opening is already done.

Exactly. I guess you could send an immediate close, but that's not without side effects.

Quoted 14 lines

But I suspect that every filesystem implementing ->atomic_open:: just does finish_no_open() and punts to the VFS for the actual open. And the VFS will catch it in do_open() for it actually opens the file. So the only real worry for O_DIRECTORY I see is that there's an fs that handles it wrong.

For (2) it is problematic as there surely are filesystems with ->atomic_open:: that do handle the ~O_CREAT case and return with FMODE_OPENED. So that'll be problematic if the intention is to not trigger an actual open on a non-regular file such as a device/socket/fifo etc. before the VFS had a chance to validate what's going on.

Quoted 8 lines

So I'm not excited about having this 70% working and punting on ->atomic_open:: waiting for someone to fix this. One option would be to bypass ->atomic_open:: for OPENAT2_REGULAR without O_CREAT and fallback to racy and pricy lookup + open for now. How problematic would that be? If possible I'd prefer this a lot over merging something that works half-way.

I guess to make that really work you'd need some protocol extension?

For NFS, I think we're OK. The OPEN call on NFSv4 only works for regular files, so it should be able to handle O_REGULAR. We just need to rejigger the error handling when it's set (just return an error instead of doing the open of a directory or whatever it is).

Quoted 6 lines

The others (at a quick glance):

cifs: I don't see a way to specify an O_REGULAR equivalent to the SMB2_CREATE call and it looks like it can create directories. Maybe SteveF (cc'ed) knows if this is possible?

SMB2 does have this flag:

FILE_NON_DIRECTORY_FILE 0x00000040

If the name of the file being created or opened matches with an existing directory file, the server MUST fail the request with STATUS_FILE_IS_A_DIRECTORY. This flag MUST NOT be used with FILE_DIRECTORY_FILE or the server MUST fail the request with STATUS_INVALID_PARAMETER.

SMB2 can also present named pipes and printer files. Not sure if there is a way to exclude those with this.

Quoted 3 lines

ceph: I think CEPH_MDS_OP_OPEN might only work for files, in which case O_REGULAR can probably be supported similarly to NFS.

Actually I'm wrong here. That op can open a directory. We'll need someone to look at the MDS code and tell us whether this can be done in a non-racy way.

Quoted 10 lines

fuse: probably ok? Does finish_no_open() in most cases. May depend on the userland driver though.

gfs2: is ok, it just does finish_no_open() in most cases anyway

vboxsf: does finish_no_open on non-creates, so you could probably just punt to that if O_REGULAR is set.

So, it's probably possible to do this across the board. I'm not sure about cifs though.

Signature · 1 line
Jeff Layton <[email protected]>
#13 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

In reply to Jeff Layton
Quoted 373 lines · Jeff Layton
Quoted 337 lines · Christian Brauner
Quoted 320 lines · Jeff Layton
Quoted 289 lines · Dorjoy Chowdhury
Quoted 285 lines · Jeff Layton
Quoted 270 lines · Dorjoy Chowdhury

This flag indicates the path should be opened if it's a regular file. This is useful to write secure programs that want to avoid being tricked into opening device nodes with special semantics while thinking they operate on regular files.

A corresponding error code ENOTREG has been introduced. For example, if open is called on path /dev/null with O_REGULAR in the flag param, it will return -ENOTREG.

When used in combination with O_CREAT, either the regular file is created, or if the path already exists, it is opened if it's a regular file. Otherwise, -ENOTREG is returned.

-EINVAL is returned when O_REGULAR is combined with O_DIRECTORY (not part of O_TMPFILE) because it doesn't make sense to open a path that is both a directory and a regular file.

---
 arch/alpha/include/uapi/asm/errno.h        | 2 ++
 arch/alpha/include/uapi/asm/fcntl.h        | 1 +
 arch/mips/include/uapi/asm/errno.h         | 2 ++
 arch/parisc/include/uapi/asm/errno.h       | 2 ++
 arch/parisc/include/uapi/asm/fcntl.h       | 1 +
 arch/sparc/include/uapi/asm/errno.h        | 2 ++
 arch/sparc/include/uapi/asm/fcntl.h        | 1 +
 fs/fcntl.c                                 | 2 +-
 fs/namei.c                                 | 6 ++++++
 fs/open.c                                  | 4 +++-
 include/linux/fcntl.h                      | 2 +-
 include/uapi/asm-generic/errno.h           | 2 ++
 include/uapi/asm-generic/fcntl.h           | 4 ++++
 tools/arch/alpha/include/uapi/asm/errno.h  | 2 ++
 tools/arch/mips/include/uapi/asm/errno.h   | 2 ++
 tools/arch/parisc/include/uapi/asm/errno.h | 2 ++
 tools/arch/sparc/include/uapi/asm/errno.h  | 2 ++
 tools/include/uapi/asm-generic/errno.h     | 2 ++
 18 files changed, 38 insertions(+), 3 deletions(-)
arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
arch/alpha/include/uapi/asm/fcntl.h +1 -0
index 50bdc8e8a271..4da5a64c23bd 100644
@@ -34,6 +34,7 @@
 
 #define O_PATH               040000000
 #define __O_TMPFILE  0100000000
+#define O_REGULAR    0200000000
 
 #define F_GETLK              7
 #define F_SETLK              8
arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
arch/parisc/include/uapi/asm/fcntl.h +1 -0
index 03dee816cb13..0cc3320fe326 100644
@@ -19,6 +19,7 @@
 
 #define O_PATH               020000000
 #define __O_TMPFILE  040000000
+#define O_REGULAR    0100000000
 
 #define F_GETLK64    8
 #define F_SETLK64    9
arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
arch/sparc/include/uapi/asm/fcntl.h +1 -0
index 67dae75e5274..a93d18d2c23e 100644
@@ -37,6 +37,7 @@
 
 #define O_PATH               0x1000000
 #define __O_TMPFILE  0x2000000
+#define O_REGULAR    0x4000000
 
 #define F_GETOWN     5       /*  for sockets. */
 #define F_SETOWN     6       /*  for sockets. */
fs/fcntl.c +1 -1
index f93dbca08435..62ab4ad2b6f5 100644
@@ -1169,7 +1169,7 @@ static int __init fcntl_init(void)
       * Exceptions: O_NONBLOCK is a two bit define on parisc; O_NDELAY
       * is defined as O_NONBLOCK on some platforms and not on others.
       */
-     BUILD_BUG_ON(20 - 1 /* for O_RDONLY being 0 */ !=
+     BUILD_BUG_ON(21 - 1 /* for O_RDONLY being 0 */ !=
              HWEIGHT32(
                      (VALID_OPEN_FLAGS & ~(O_NONBLOCK | O_NDELAY)) |
                      __FMODE_EXEC));
fs/namei.c +6 -0
index b28ecb699f32..f5504ae4b03c 100644
@@ -4616,6 +4616,10 @@ static int do_open(struct nameidata *nd,
              if (unlikely(error))
                      return error;
      }
+
+     if ((open_flag & O_REGULAR) && !d_is_reg(nd->path.dentry))
+             return -ENOTREG;
+
      if ((nd->flags & LOOKUP_DIRECTORY) && !d_can_lookup(nd->path.dentry))
              return -ENOTDIR;
 
@@ -4765,6 +4769,8 @@ static int do_o_path(struct nameidata *nd, unsigned flags, struct file *file)
      struct path path;
      int error = path_lookupat(nd, flags, &path);
      if (!error) {
+             if ((file->f_flags & O_REGULAR) && !d_is_reg(path.dentry))
+                     return -ENOTREG;
              audit_inode(nd->name, path.dentry, 0);
              error = vfs_open(&path, file);
              path_put(&path);
fs/open.c +3 -1
index 74c4c1462b3e..82153e21907e 100644
@@ -1173,7 +1173,7 @@ struct file *kernel_file_open(const struct path *path, int flags,
 EXPORT_SYMBOL_GPL(kernel_file_open);
 
 #define WILL_CREATE(flags)   (flags & (O_CREAT | __O_TMPFILE))
-#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC)
+#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC | O_REGULAR)
 
 inline struct open_how build_open_how(int flags, umode_t mode)
 {
@@ -1250,6 +1250,8 @@ inline int build_open_flags(const struct open_how *how, struct open_flags *op)
                      return -EINVAL;
              if (!(acc_mode & MAY_WRITE))
                      return -EINVAL;
+     } else if ((flags & O_DIRECTORY) && (flags & O_REGULAR)) {
+             return -EINVAL;
      }
      if (flags & O_PATH) {
              /* O_PATH only permits certain other flags to be set. */
include/linux/fcntl.h +1 -1
index a332e79b3207..4fd07b0e0a17 100644
@@ -10,7 +10,7 @@
      (O_RDONLY | O_WRONLY | O_RDWR | O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC | \
       O_APPEND | O_NDELAY | O_NONBLOCK | __O_SYNC | O_DSYNC | \
       FASYNC | O_DIRECT | O_LARGEFILE | O_DIRECTORY | O_NOFOLLOW | \
-      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE)
+      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE | O_REGULAR)
 
 /* List of all valid flags for the how->resolve argument: */
 #define VALID_RESOLVE_FLAGS \
include/uapi/asm-generic/errno.h +2 -0
index 92e7ae493ee3..2216ab9aa32e 100644
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif
include/uapi/asm-generic/fcntl.h +4 -0
index 613475285643..3468b352a575 100644
@@ -88,6 +88,10 @@
 #define __O_TMPFILE  020000000
 #endif
 
+#ifndef O_REGULAR
+#define O_REGULAR    040000000
+#endif
+
 /* a horrid kludge trying to make sure that this will fail on old kernels */
 #define O_TMPFILE (__O_TMPFILE | O_DIRECTORY)
 
tools/arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
tools/arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
tools/arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
tools/arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
tools/include/uapi/asm-generic/errno.h +0 -0
index 92e7ae493ee3..2216ab9aa32e 100644

tools/include/uapi/asm-generic/errno.h line 127

Quoted 6 lines
tools/include/uapi/asm-generic/errno.h +2 -0
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif

One thing this patch is missing is handling for ->atomic_open(). I imagine most of the filesystems that provide that op can't support O_REGULAR properly (maybe cifs can? idk). What you probably want to do is add in some patches that make all of the atomic_open operations in the kernel return -EINVAL if O_REGULAR is set.

Quoted 2 lines

Then, once the basic support is in, you or someone else can go back and implement support for O_REGULAR where possible.

Quoted 11 lines

Thank you for the feedback. I don't quite understand what I need to fix. I thought open system calls always create regular files, so atomic_open probably always creates regular files? Can you please give me some more details as to where I need to fix this and what the actual bug here is that is related to atomic_open? I think I had done some normal testing and when using O_CREAT | O_REGULAR, if the file doesn't exist, the file gets created and the file that gets created is a regular file, so it probably makes sense? Or should the behavior be that if file doesn't exist, -EINVAL is returned and if file exists it is opened if regular, otherwise -ENOTREG is returned?

atomic_open() is a combination lookup+open for when the dentry isn't present in the dcache. The normal open codepath that you're patching does not get called in this case when ->atomic_open is set for the filesystem. It's mostly used by network filesystems that need to optimize away the lookup since it's wasted round trip, and is often racy anyway. Your patchset doesn't address those filesystems. They will likely end up ignoring O_REGULAR in that case, which is not what you want.

What I was suggesting is that, as an interim step, you find all of the atomic_open operations in the kernel (there are maybe a dozen or so), and just make them return -EINVAL if someone sets O_DIRECTORY. Later, you or someone else can then go back and do a proper implementation of O_REGULAR handling on those filesystems, at least on the ones where it's possible. You will probably also need to similarly patch the open() routines for those filesystems too. Otherwise you'll get inconsistent behavior vs. when the dentry is in the cache.

Quoted 3 lines

One note: I think NFS probably can support O_DIRECTORY, since its OPEN call only works on files. We'll need to change how we handle errors from the server when it's set though.

So I think you're proposing two separate things or there's a typo:

(1) blocking O_DIRECTORY for ->atomic_open::
(2) blocking O_REGULAR for ->atomic_open::

The (1) point implies that O_DIRECTORY currently doesn't work correctly with atomic open for all filesystems.

Ever since 43b450632676 ("open: return EINVAL for O_DIRECTORY | O_CREAT") O_DIRECTORY with O_CREAT is blocked. It was accidently allowed and completely broken before that.

Quoted 9 lines

For O_DIRECTORY without O_CREAT the kernel will pass that down through ->atomic_open:: to the filesystem.

The worry that I see is that a network filesystem via ->atomic_open::
somehow already called open on the server side on something that wasn't a directory. At that point the damage such as side-effects from device opening is already done.

Exactly. I guess you could send an immediate close, but that's not without side effects.

Quoted 21 lines

But I suspect that every filesystem implementing ->atomic_open:: just does finish_no_open() and punts to the VFS for the actual open. And the VFS will catch it in do_open() for it actually opens the file. So the only real worry for O_DIRECTORY I see is that there's an fs that handles it wrong.

For (2) it is problematic as there surely are filesystems with ->atomic_open:: that do handle the ~O_CREAT case and return with FMODE_OPENED. So that'll be problematic if the intention is to not trigger an actual open on a non-regular file such as a device/socket/fifo etc. before the VFS had a chance to validate what's going on.

So I'm not excited about having this 70% working and punting on ->atomic_open:: waiting for someone to fix this. One option would be to bypass ->atomic_open:: for OPENAT2_REGULAR without O_CREAT and fallback to racy and pricy lookup + open for now. How problematic would that be? If possible I'd prefer this a lot over merging something that works half-way.

Quoted 7 lines
Quoted 1 line

I guess to make that really work you'd need some protocol extension?

For NFS, I think we're OK. The OPEN call on NFSv4 only works for regular files, so it should be able to handle O_REGULAR. We just need to rejigger the error handling when it's set (just return an error instead of doing the open of a directory or whatever it is).

Thank you for the details. Do you remember which codepath this is? Is this the inode_operations.atomic_open codepath or file_operations.open codepath? I am a bit confused also about where exactly the error handling that needs to be done.

Quoted 9 lines

The others (at a quick glance):

cifs: I don't see a way to specify an O_REGULAR equivalent to the SMB2_CREATE call and it looks like it can create directories. Maybe SteveF (cc'ed) knows if this is possible?

ceph: I think CEPH_MDS_OP_OPEN might only work for files, in which case O_REGULAR can probably be supported similarly to NFS.

Quoted 8 lines

fuse: probably ok? Does finish_no_open() in most cases. May depend on the userland driver though.

gfs2: is ok, it just does finish_no_open() in most cases anyway

vboxsf: does finish_no_open on non-creates, so you could probably just punt to that if O_REGULAR is set.

These are all inode_operations.atomic_open code paths, right? Because you mentioned finish_no_open and I see finish_no_open in the atomic_open code paths as opposed to file_operations.open code paths.

Regards,
Dorjoy

maintainer
#14 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

Quoted 382 lines · Dorjoy Chowdhury
Quoted 373 lines · Jeff Layton
Quoted 337 lines · Christian Brauner
Quoted 320 lines · Jeff Layton
Quoted 289 lines · Dorjoy Chowdhury
Quoted 285 lines · Jeff Layton
Quoted 270 lines · Dorjoy Chowdhury

This flag indicates the path should be opened if it's a regular file. This is useful to write secure programs that want to avoid being tricked into opening device nodes with special semantics while thinking they operate on regular files.

A corresponding error code ENOTREG has been introduced. For example, if open is called on path /dev/null with O_REGULAR in the flag param, it will return -ENOTREG.

When used in combination with O_CREAT, either the regular file is created, or if the path already exists, it is opened if it's a regular file. Otherwise, -ENOTREG is returned.

-EINVAL is returned when O_REGULAR is combined with O_DIRECTORY (not part of O_TMPFILE) because it doesn't make sense to open a path that is both a directory and a regular file.

---
 arch/alpha/include/uapi/asm/errno.h        | 2 ++
 arch/alpha/include/uapi/asm/fcntl.h        | 1 +
 arch/mips/include/uapi/asm/errno.h         | 2 ++
 arch/parisc/include/uapi/asm/errno.h       | 2 ++
 arch/parisc/include/uapi/asm/fcntl.h       | 1 +
 arch/sparc/include/uapi/asm/errno.h        | 2 ++
 arch/sparc/include/uapi/asm/fcntl.h        | 1 +
 fs/fcntl.c                                 | 2 +-
 fs/namei.c                                 | 6 ++++++
 fs/open.c                                  | 4 +++-
 include/linux/fcntl.h                      | 2 +-
 include/uapi/asm-generic/errno.h           | 2 ++
 include/uapi/asm-generic/fcntl.h           | 4 ++++
 tools/arch/alpha/include/uapi/asm/errno.h  | 2 ++
 tools/arch/mips/include/uapi/asm/errno.h   | 2 ++
 tools/arch/parisc/include/uapi/asm/errno.h | 2 ++
 tools/arch/sparc/include/uapi/asm/errno.h  | 2 ++
 tools/include/uapi/asm-generic/errno.h     | 2 ++
 18 files changed, 38 insertions(+), 3 deletions(-)
arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
arch/alpha/include/uapi/asm/fcntl.h +1 -0
index 50bdc8e8a271..4da5a64c23bd 100644
@@ -34,6 +34,7 @@
 
 #define O_PATH               040000000
 #define __O_TMPFILE  0100000000
+#define O_REGULAR    0200000000
 
 #define F_GETLK              7
 #define F_SETLK              8
arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
arch/parisc/include/uapi/asm/fcntl.h +1 -0
index 03dee816cb13..0cc3320fe326 100644
@@ -19,6 +19,7 @@
 
 #define O_PATH               020000000
 #define __O_TMPFILE  040000000
+#define O_REGULAR    0100000000
 
 #define F_GETLK64    8
 #define F_SETLK64    9
arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
arch/sparc/include/uapi/asm/fcntl.h +1 -0
index 67dae75e5274..a93d18d2c23e 100644
@@ -37,6 +37,7 @@
 
 #define O_PATH               0x1000000
 #define __O_TMPFILE  0x2000000
+#define O_REGULAR    0x4000000
 
 #define F_GETOWN     5       /*  for sockets. */
 #define F_SETOWN     6       /*  for sockets. */
fs/fcntl.c +1 -1
index f93dbca08435..62ab4ad2b6f5 100644
@@ -1169,7 +1169,7 @@ static int __init fcntl_init(void)
       * Exceptions: O_NONBLOCK is a two bit define on parisc; O_NDELAY
       * is defined as O_NONBLOCK on some platforms and not on others.
       */
-     BUILD_BUG_ON(20 - 1 /* for O_RDONLY being 0 */ !=
+     BUILD_BUG_ON(21 - 1 /* for O_RDONLY being 0 */ !=
              HWEIGHT32(
                      (VALID_OPEN_FLAGS & ~(O_NONBLOCK | O_NDELAY)) |
                      __FMODE_EXEC));
fs/namei.c +6 -0
index b28ecb699f32..f5504ae4b03c 100644
@@ -4616,6 +4616,10 @@ static int do_open(struct nameidata *nd,
              if (unlikely(error))
                      return error;
      }
+
+     if ((open_flag & O_REGULAR) && !d_is_reg(nd->path.dentry))
+             return -ENOTREG;
+
      if ((nd->flags & LOOKUP_DIRECTORY) && !d_can_lookup(nd->path.dentry))
              return -ENOTDIR;
 
@@ -4765,6 +4769,8 @@ static int do_o_path(struct nameidata *nd, unsigned flags, struct file *file)
      struct path path;
      int error = path_lookupat(nd, flags, &path);
      if (!error) {
+             if ((file->f_flags & O_REGULAR) && !d_is_reg(path.dentry))
+                     return -ENOTREG;
              audit_inode(nd->name, path.dentry, 0);
              error = vfs_open(&path, file);
              path_put(&path);
fs/open.c +3 -1
index 74c4c1462b3e..82153e21907e 100644
@@ -1173,7 +1173,7 @@ struct file *kernel_file_open(const struct path *path, int flags,
 EXPORT_SYMBOL_GPL(kernel_file_open);
 
 #define WILL_CREATE(flags)   (flags & (O_CREAT | __O_TMPFILE))
-#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC)
+#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC | O_REGULAR)
 
 inline struct open_how build_open_how(int flags, umode_t mode)
 {
@@ -1250,6 +1250,8 @@ inline int build_open_flags(const struct open_how *how, struct open_flags *op)
                      return -EINVAL;
              if (!(acc_mode & MAY_WRITE))
                      return -EINVAL;
+     } else if ((flags & O_DIRECTORY) && (flags & O_REGULAR)) {
+             return -EINVAL;
      }
      if (flags & O_PATH) {
              /* O_PATH only permits certain other flags to be set. */
include/linux/fcntl.h +1 -1
index a332e79b3207..4fd07b0e0a17 100644
@@ -10,7 +10,7 @@
      (O_RDONLY | O_WRONLY | O_RDWR | O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC | \
       O_APPEND | O_NDELAY | O_NONBLOCK | __O_SYNC | O_DSYNC | \
       FASYNC | O_DIRECT | O_LARGEFILE | O_DIRECTORY | O_NOFOLLOW | \
-      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE)
+      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE | O_REGULAR)
 
 /* List of all valid flags for the how->resolve argument: */
 #define VALID_RESOLVE_FLAGS \
include/uapi/asm-generic/errno.h +2 -0
index 92e7ae493ee3..2216ab9aa32e 100644
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif
include/uapi/asm-generic/fcntl.h +4 -0
index 613475285643..3468b352a575 100644
@@ -88,6 +88,10 @@
 #define __O_TMPFILE  020000000
 #endif
 
+#ifndef O_REGULAR
+#define O_REGULAR    040000000
+#endif
+
 /* a horrid kludge trying to make sure that this will fail on old kernels */
 #define O_TMPFILE (__O_TMPFILE | O_DIRECTORY)
 
tools/arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
tools/arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
tools/arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
tools/arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
tools/include/uapi/asm-generic/errno.h +0 -0
index 92e7ae493ee3..2216ab9aa32e 100644

tools/include/uapi/asm-generic/errno.h line 127

Quoted 6 lines
tools/include/uapi/asm-generic/errno.h +2 -0
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif

One thing this patch is missing is handling for ->atomic_open(). I imagine most of the filesystems that provide that op can't support O_REGULAR properly (maybe cifs can? idk). What you probably want to do is add in some patches that make all of the atomic_open operations in the kernel return -EINVAL if O_REGULAR is set.

Quoted 2 lines

Then, once the basic support is in, you or someone else can go back and implement support for O_REGULAR where possible.

Quoted 11 lines

Thank you for the feedback. I don't quite understand what I need to fix. I thought open system calls always create regular files, so atomic_open probably always creates regular files? Can you please give me some more details as to where I need to fix this and what the actual bug here is that is related to atomic_open? I think I had done some normal testing and when using O_CREAT | O_REGULAR, if the file doesn't exist, the file gets created and the file that gets created is a regular file, so it probably makes sense? Or should the behavior be that if file doesn't exist, -EINVAL is returned and if file exists it is opened if regular, otherwise -ENOTREG is returned?

atomic_open() is a combination lookup+open for when the dentry isn't present in the dcache. The normal open codepath that you're patching does not get called in this case when ->atomic_open is set for the filesystem. It's mostly used by network filesystems that need to optimize away the lookup since it's wasted round trip, and is often racy anyway. Your patchset doesn't address those filesystems. They will likely end up ignoring O_REGULAR in that case, which is not what you want.

What I was suggesting is that, as an interim step, you find all of the atomic_open operations in the kernel (there are maybe a dozen or so), and just make them return -EINVAL if someone sets O_DIRECTORY. Later, you or someone else can then go back and do a proper implementation of O_REGULAR handling on those filesystems, at least on the ones where it's possible. You will probably also need to similarly patch the open() routines for those filesystems too. Otherwise you'll get inconsistent behavior vs. when the dentry is in the cache.

Quoted 3 lines

One note: I think NFS probably can support O_DIRECTORY, since its OPEN call only works on files. We'll need to change how we handle errors from the server when it's set though.

So I think you're proposing two separate things or there's a typo:

(1) blocking O_DIRECTORY for ->atomic_open::
(2) blocking O_REGULAR for ->atomic_open::

The (1) point implies that O_DIRECTORY currently doesn't work correctly with atomic open for all filesystems.

Ever since 43b450632676 ("open: return EINVAL for O_DIRECTORY | O_CREAT") O_DIRECTORY with O_CREAT is blocked. It was accidently allowed and completely broken before that.

Quoted 9 lines

For O_DIRECTORY without O_CREAT the kernel will pass that down through ->atomic_open:: to the filesystem.

The worry that I see is that a network filesystem via ->atomic_open::
somehow already called open on the server side on something that wasn't a directory. At that point the damage such as side-effects from device opening is already done.

Exactly. I guess you could send an immediate close, but that's not without side effects.

Quoted 21 lines

But I suspect that every filesystem implementing ->atomic_open:: just does finish_no_open() and punts to the VFS for the actual open. And the VFS will catch it in do_open() for it actually opens the file. So the only real worry for O_DIRECTORY I see is that there's an fs that handles it wrong.

For (2) it is problematic as there surely are filesystems with ->atomic_open:: that do handle the ~O_CREAT case and return with FMODE_OPENED. So that'll be problematic if the intention is to not trigger an actual open on a non-regular file such as a device/socket/fifo etc. before the VFS had a chance to validate what's going on.

So I'm not excited about having this 70% working and punting on ->atomic_open:: waiting for someone to fix this. One option would be to bypass ->atomic_open:: for OPENAT2_REGULAR without O_CREAT and fallback to racy and pricy lookup + open for now. How problematic would that be? If possible I'd prefer this a lot over merging something that works half-way.

Quoted 7 lines
Quoted 1 line

I guess to make that really work you'd need some protocol extension?

For NFS, I think we're OK. The OPEN call on NFSv4 only works for regular files, so it should be able to handle O_REGULAR. We just need to rejigger the error handling when it's set (just return an error instead of doing the open of a directory or whatever it is).

Quoted 5 lines

Thank you for the details. Do you remember which codepath this is? Is this the inode_operations.atomic_open codepath or file_operations.open codepath? I am a bit confused also about where exactly the error handling that needs to be done.

I was thinking nfs_atomic_open().

Looking now, I think it might actually work OK without changes. It just might not be terribly efficient about it.

If the open_context() call returns -EISDIR or similar, then you really don't need to do the call to nfs_lookup() and the like. You can just return an immediate error when O_REGULAR is set since you know it's not suitable to be opened.

Quoted 14 lines
Quoted 14 lines

The others (at a quick glance):

cifs: I don't see a way to specify an O_REGULAR equivalent to the SMB2_CREATE call and it looks like it can create directories. Maybe SteveF (cc'ed) knows if this is possible?

ceph: I think CEPH_MDS_OP_OPEN might only work for files, in which case O_REGULAR can probably be supported similarly to NFS.

fuse: probably ok? Does finish_no_open() in most cases. May depend on the userland driver though.

gfs2: is ok, it just does finish_no_open() in most cases anyway

Quoted 8 lines
Quoted 3 lines

vboxsf: does finish_no_open on non-creates, so you could probably just punt to that if O_REGULAR is set.

These are all inode_operations.atomic_open code paths, right? Because you mentioned finish_no_open and I see finish_no_open in the atomic_open code paths as opposed to file_operations.open code paths.

Note that this was just a cursory look. Someone will need to do a deeper dive and test these cases.

I think most will end up working ok, since most fall back to doing a finish_no_open(). There may be opportunities to optimize some of these cases though (similarly to how I mentioned with NFS).

Signature · 1 line
Jeff Layton <[email protected]>
#15 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

Quoted 392 lines · Jeff Layton
Quoted 382 lines · Dorjoy Chowdhury
Quoted 373 lines · Jeff Layton
Quoted 337 lines · Christian Brauner
Quoted 320 lines · Jeff Layton
Quoted 289 lines · Dorjoy Chowdhury
Quoted 285 lines · Jeff Layton
Quoted 270 lines · Dorjoy Chowdhury

This flag indicates the path should be opened if it's a regular file. This is useful to write secure programs that want to avoid being tricked into opening device nodes with special semantics while thinking they operate on regular files.

A corresponding error code ENOTREG has been introduced. For example, if open is called on path /dev/null with O_REGULAR in the flag param, it will return -ENOTREG.

When used in combination with O_CREAT, either the regular file is created, or if the path already exists, it is opened if it's a regular file. Otherwise, -ENOTREG is returned.

-EINVAL is returned when O_REGULAR is combined with O_DIRECTORY (not part of O_TMPFILE) because it doesn't make sense to open a path that is both a directory and a regular file.

---
 arch/alpha/include/uapi/asm/errno.h        | 2 ++
 arch/alpha/include/uapi/asm/fcntl.h        | 1 +
 arch/mips/include/uapi/asm/errno.h         | 2 ++
 arch/parisc/include/uapi/asm/errno.h       | 2 ++
 arch/parisc/include/uapi/asm/fcntl.h       | 1 +
 arch/sparc/include/uapi/asm/errno.h        | 2 ++
 arch/sparc/include/uapi/asm/fcntl.h        | 1 +
 fs/fcntl.c                                 | 2 +-
 fs/namei.c                                 | 6 ++++++
 fs/open.c                                  | 4 +++-
 include/linux/fcntl.h                      | 2 +-
 include/uapi/asm-generic/errno.h           | 2 ++
 include/uapi/asm-generic/fcntl.h           | 4 ++++
 tools/arch/alpha/include/uapi/asm/errno.h  | 2 ++
 tools/arch/mips/include/uapi/asm/errno.h   | 2 ++
 tools/arch/parisc/include/uapi/asm/errno.h | 2 ++
 tools/arch/sparc/include/uapi/asm/errno.h  | 2 ++
 tools/include/uapi/asm-generic/errno.h     | 2 ++
 18 files changed, 38 insertions(+), 3 deletions(-)
arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
arch/alpha/include/uapi/asm/fcntl.h +1 -0
index 50bdc8e8a271..4da5a64c23bd 100644
@@ -34,6 +34,7 @@
 
 #define O_PATH               040000000
 #define __O_TMPFILE  0100000000
+#define O_REGULAR    0200000000
 
 #define F_GETLK              7
 #define F_SETLK              8
arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
arch/parisc/include/uapi/asm/fcntl.h +1 -0
index 03dee816cb13..0cc3320fe326 100644
@@ -19,6 +19,7 @@
 
 #define O_PATH               020000000
 #define __O_TMPFILE  040000000
+#define O_REGULAR    0100000000
 
 #define F_GETLK64    8
 #define F_SETLK64    9
arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
arch/sparc/include/uapi/asm/fcntl.h +1 -0
index 67dae75e5274..a93d18d2c23e 100644
@@ -37,6 +37,7 @@
 
 #define O_PATH               0x1000000
 #define __O_TMPFILE  0x2000000
+#define O_REGULAR    0x4000000
 
 #define F_GETOWN     5       /*  for sockets. */
 #define F_SETOWN     6       /*  for sockets. */
fs/fcntl.c +1 -1
index f93dbca08435..62ab4ad2b6f5 100644
@@ -1169,7 +1169,7 @@ static int __init fcntl_init(void)
       * Exceptions: O_NONBLOCK is a two bit define on parisc; O_NDELAY
       * is defined as O_NONBLOCK on some platforms and not on others.
       */
-     BUILD_BUG_ON(20 - 1 /* for O_RDONLY being 0 */ !=
+     BUILD_BUG_ON(21 - 1 /* for O_RDONLY being 0 */ !=
              HWEIGHT32(
                      (VALID_OPEN_FLAGS & ~(O_NONBLOCK | O_NDELAY)) |
                      __FMODE_EXEC));
fs/namei.c +6 -0
index b28ecb699f32..f5504ae4b03c 100644
@@ -4616,6 +4616,10 @@ static int do_open(struct nameidata *nd,
              if (unlikely(error))
                      return error;
      }
+
+     if ((open_flag & O_REGULAR) && !d_is_reg(nd->path.dentry))
+             return -ENOTREG;
+
      if ((nd->flags & LOOKUP_DIRECTORY) && !d_can_lookup(nd->path.dentry))
              return -ENOTDIR;
 
@@ -4765,6 +4769,8 @@ static int do_o_path(struct nameidata *nd, unsigned flags, struct file *file)
      struct path path;
      int error = path_lookupat(nd, flags, &path);
      if (!error) {
+             if ((file->f_flags & O_REGULAR) && !d_is_reg(path.dentry))
+                     return -ENOTREG;
              audit_inode(nd->name, path.dentry, 0);
              error = vfs_open(&path, file);
              path_put(&path);
fs/open.c +3 -1
index 74c4c1462b3e..82153e21907e 100644
@@ -1173,7 +1173,7 @@ struct file *kernel_file_open(const struct path *path, int flags,
 EXPORT_SYMBOL_GPL(kernel_file_open);
 
 #define WILL_CREATE(flags)   (flags & (O_CREAT | __O_TMPFILE))
-#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC)
+#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC | O_REGULAR)
 
 inline struct open_how build_open_how(int flags, umode_t mode)
 {
@@ -1250,6 +1250,8 @@ inline int build_open_flags(const struct open_how *how, struct open_flags *op)
                      return -EINVAL;
              if (!(acc_mode & MAY_WRITE))
                      return -EINVAL;
+     } else if ((flags & O_DIRECTORY) && (flags & O_REGULAR)) {
+             return -EINVAL;
      }
      if (flags & O_PATH) {
              /* O_PATH only permits certain other flags to be set. */
include/linux/fcntl.h +1 -1
index a332e79b3207..4fd07b0e0a17 100644
@@ -10,7 +10,7 @@
      (O_RDONLY | O_WRONLY | O_RDWR | O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC | \
       O_APPEND | O_NDELAY | O_NONBLOCK | __O_SYNC | O_DSYNC | \
       FASYNC | O_DIRECT | O_LARGEFILE | O_DIRECTORY | O_NOFOLLOW | \
-      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE)
+      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE | O_REGULAR)
 
 /* List of all valid flags for the how->resolve argument: */
 #define VALID_RESOLVE_FLAGS \
include/uapi/asm-generic/errno.h +2 -0
index 92e7ae493ee3..2216ab9aa32e 100644
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif
include/uapi/asm-generic/fcntl.h +4 -0
index 613475285643..3468b352a575 100644
@@ -88,6 +88,10 @@
 #define __O_TMPFILE  020000000
 #endif
 
+#ifndef O_REGULAR
+#define O_REGULAR    040000000
+#endif
+
 /* a horrid kludge trying to make sure that this will fail on old kernels */
 #define O_TMPFILE (__O_TMPFILE | O_DIRECTORY)
 
tools/arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
tools/arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
tools/arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
tools/arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
tools/include/uapi/asm-generic/errno.h +0 -0
index 92e7ae493ee3..2216ab9aa32e 100644

tools/include/uapi/asm-generic/errno.h line 127

Quoted 6 lines
tools/include/uapi/asm-generic/errno.h +2 -0
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif

One thing this patch is missing is handling for ->atomic_open(). I imagine most of the filesystems that provide that op can't support O_REGULAR properly (maybe cifs can? idk). What you probably want to do is add in some patches that make all of the atomic_open operations in the kernel return -EINVAL if O_REGULAR is set.

Quoted 2 lines

Then, once the basic support is in, you or someone else can go back and implement support for O_REGULAR where possible.

Quoted 11 lines

Thank you for the feedback. I don't quite understand what I need to fix. I thought open system calls always create regular files, so atomic_open probably always creates regular files? Can you please give me some more details as to where I need to fix this and what the actual bug here is that is related to atomic_open? I think I had done some normal testing and when using O_CREAT | O_REGULAR, if the file doesn't exist, the file gets created and the file that gets created is a regular file, so it probably makes sense? Or should the behavior be that if file doesn't exist, -EINVAL is returned and if file exists it is opened if regular, otherwise -ENOTREG is returned?

atomic_open() is a combination lookup+open for when the dentry isn't present in the dcache. The normal open codepath that you're patching does not get called in this case when ->atomic_open is set for the filesystem. It's mostly used by network filesystems that need to optimize away the lookup since it's wasted round trip, and is often racy anyway. Your patchset doesn't address those filesystems. They will likely end up ignoring O_REGULAR in that case, which is not what you want.

What I was suggesting is that, as an interim step, you find all of the atomic_open operations in the kernel (there are maybe a dozen or so), and just make them return -EINVAL if someone sets O_DIRECTORY. Later, you or someone else can then go back and do a proper implementation of O_REGULAR handling on those filesystems, at least on the ones where it's possible. You will probably also need to similarly patch the open() routines for those filesystems too. Otherwise you'll get inconsistent behavior vs. when the dentry is in the cache.

Quoted 3 lines

One note: I think NFS probably can support O_DIRECTORY, since its OPEN call only works on files. We'll need to change how we handle errors from the server when it's set though.

So I think you're proposing two separate things or there's a typo:

(1) blocking O_DIRECTORY for ->atomic_open::
(2) blocking O_REGULAR for ->atomic_open::

The (1) point implies that O_DIRECTORY currently doesn't work correctly with atomic open for all filesystems.

Ever since 43b450632676 ("open: return EINVAL for O_DIRECTORY | O_CREAT") O_DIRECTORY with O_CREAT is blocked. It was accidently allowed and completely broken before that.

Quoted 9 lines

For O_DIRECTORY without O_CREAT the kernel will pass that down through ->atomic_open:: to the filesystem.

The worry that I see is that a network filesystem via ->atomic_open::
somehow already called open on the server side on something that wasn't a directory. At that point the damage such as side-effects from device opening is already done.

Exactly. I guess you could send an immediate close, but that's not without side effects.

Quoted 21 lines

But I suspect that every filesystem implementing ->atomic_open:: just does finish_no_open() and punts to the VFS for the actual open. And the VFS will catch it in do_open() for it actually opens the file. So the only real worry for O_DIRECTORY I see is that there's an fs that handles it wrong.

For (2) it is problematic as there surely are filesystems with ->atomic_open:: that do handle the ~O_CREAT case and return with FMODE_OPENED. So that'll be problematic if the intention is to not trigger an actual open on a non-regular file such as a device/socket/fifo etc. before the VFS had a chance to validate what's going on.

So I'm not excited about having this 70% working and punting on ->atomic_open:: waiting for someone to fix this. One option would be to bypass ->atomic_open:: for OPENAT2_REGULAR without O_CREAT and fallback to racy and pricy lookup + open for now. How problematic would that be? If possible I'd prefer this a lot over merging something that works half-way.

Quoted 7 lines
Quoted 1 line

I guess to make that really work you'd need some protocol extension?

For NFS, I think we're OK. The OPEN call on NFSv4 only works for regular files, so it should be able to handle O_REGULAR. We just need to rejigger the error handling when it's set (just return an error instead of doing the open of a directory or whatever it is).

Quoted 5 lines

Thank you for the details. Do you remember which codepath this is? Is this the inode_operations.atomic_open codepath or file_operations.open codepath? I am a bit confused also about where exactly the error handling that needs to be done.

I was thinking nfs_atomic_open().

Quoted 8 lines

Looking now, I think it might actually work OK without changes. It just might not be terribly efficient about it.

If the open_context() call returns -EISDIR or similar, then you really don't need to do the call to nfs_lookup() and the like. You can just return an immediate error when O_REGULAR is set since you know it's not suitable to be opened.

Right. And I guess we don't need to worry about O_REGULAR being an unknown flag when it gets sent to the server (not only for NFS / but others as well)?

Quoted 23 lines
Quoted 14 lines
Quoted 14 lines

The others (at a quick glance):

cifs: I don't see a way to specify an O_REGULAR equivalent to the SMB2_CREATE call and it looks like it can create directories. Maybe SteveF (cc'ed) knows if this is possible?

ceph: I think CEPH_MDS_OP_OPEN might only work for files, in which case O_REGULAR can probably be supported similarly to NFS.

fuse: probably ok? Does finish_no_open() in most cases. May depend on the userland driver though.

gfs2: is ok, it just does finish_no_open() in most cases anyway

Quoted 8 lines
Quoted 3 lines

vboxsf: does finish_no_open on non-creates, so you could probably just punt to that if O_REGULAR is set.

These are all inode_operations.atomic_open code paths, right? Because you mentioned finish_no_open and I see finish_no_open in the atomic_open code paths as opposed to file_operations.open code paths.

Quoted 6 lines

Note that this was just a cursory look. Someone will need to do a deeper dive and test these cases.

I think most will end up working ok, since most fall back to doing a finish_no_open(). There may be opportunities to optimize some of these cases though (similarly to how I mentioned with NFS).

I can try to look into these and see if I can implement handling for O_REGULAR flag for these filesystems in the atomic_open code paths. Thanks for the details.

Will I need to modify the corresponding file_operations.open code paths as well along with atomic_open code paths?

Regards,
Dorjoy

maintainer
#16 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

Quoted 402 lines · Dorjoy Chowdhury
Quoted 392 lines · Jeff Layton
Quoted 382 lines · Dorjoy Chowdhury
Quoted 373 lines · Jeff Layton
Quoted 337 lines · Christian Brauner
Quoted 320 lines · Jeff Layton
Quoted 289 lines · Dorjoy Chowdhury
Quoted 285 lines · Jeff Layton
Quoted 270 lines · Dorjoy Chowdhury

This flag indicates the path should be opened if it's a regular file. This is useful to write secure programs that want to avoid being tricked into opening device nodes with special semantics while thinking they operate on regular files.

A corresponding error code ENOTREG has been introduced. For example, if open is called on path /dev/null with O_REGULAR in the flag param, it will return -ENOTREG.

When used in combination with O_CREAT, either the regular file is created, or if the path already exists, it is opened if it's a regular file. Otherwise, -ENOTREG is returned.

-EINVAL is returned when O_REGULAR is combined with O_DIRECTORY (not part of O_TMPFILE) because it doesn't make sense to open a path that is both a directory and a regular file.

---
 arch/alpha/include/uapi/asm/errno.h        | 2 ++
 arch/alpha/include/uapi/asm/fcntl.h        | 1 +
 arch/mips/include/uapi/asm/errno.h         | 2 ++
 arch/parisc/include/uapi/asm/errno.h       | 2 ++
 arch/parisc/include/uapi/asm/fcntl.h       | 1 +
 arch/sparc/include/uapi/asm/errno.h        | 2 ++
 arch/sparc/include/uapi/asm/fcntl.h        | 1 +
 fs/fcntl.c                                 | 2 +-
 fs/namei.c                                 | 6 ++++++
 fs/open.c                                  | 4 +++-
 include/linux/fcntl.h                      | 2 +-
 include/uapi/asm-generic/errno.h           | 2 ++
 include/uapi/asm-generic/fcntl.h           | 4 ++++
 tools/arch/alpha/include/uapi/asm/errno.h  | 2 ++
 tools/arch/mips/include/uapi/asm/errno.h   | 2 ++
 tools/arch/parisc/include/uapi/asm/errno.h | 2 ++
 tools/arch/sparc/include/uapi/asm/errno.h  | 2 ++
 tools/include/uapi/asm-generic/errno.h     | 2 ++
 18 files changed, 38 insertions(+), 3 deletions(-)
arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
arch/alpha/include/uapi/asm/fcntl.h +1 -0
index 50bdc8e8a271..4da5a64c23bd 100644
@@ -34,6 +34,7 @@
 
 #define O_PATH               040000000
 #define __O_TMPFILE  0100000000
+#define O_REGULAR    0200000000
 
 #define F_GETLK              7
 #define F_SETLK              8
arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
arch/parisc/include/uapi/asm/fcntl.h +1 -0
index 03dee816cb13..0cc3320fe326 100644
@@ -19,6 +19,7 @@
 
 #define O_PATH               020000000
 #define __O_TMPFILE  040000000
+#define O_REGULAR    0100000000
 
 #define F_GETLK64    8
 #define F_SETLK64    9
arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
arch/sparc/include/uapi/asm/fcntl.h +1 -0
index 67dae75e5274..a93d18d2c23e 100644
@@ -37,6 +37,7 @@
 
 #define O_PATH               0x1000000
 #define __O_TMPFILE  0x2000000
+#define O_REGULAR    0x4000000
 
 #define F_GETOWN     5       /*  for sockets. */
 #define F_SETOWN     6       /*  for sockets. */
fs/fcntl.c +1 -1
index f93dbca08435..62ab4ad2b6f5 100644
@@ -1169,7 +1169,7 @@ static int __init fcntl_init(void)
       * Exceptions: O_NONBLOCK is a two bit define on parisc; O_NDELAY
       * is defined as O_NONBLOCK on some platforms and not on others.
       */
-     BUILD_BUG_ON(20 - 1 /* for O_RDONLY being 0 */ !=
+     BUILD_BUG_ON(21 - 1 /* for O_RDONLY being 0 */ !=
              HWEIGHT32(
                      (VALID_OPEN_FLAGS & ~(O_NONBLOCK | O_NDELAY)) |
                      __FMODE_EXEC));
fs/namei.c +6 -0
index b28ecb699f32..f5504ae4b03c 100644
@@ -4616,6 +4616,10 @@ static int do_open(struct nameidata *nd,
              if (unlikely(error))
                      return error;
      }
+
+     if ((open_flag & O_REGULAR) && !d_is_reg(nd->path.dentry))
+             return -ENOTREG;
+
      if ((nd->flags & LOOKUP_DIRECTORY) && !d_can_lookup(nd->path.dentry))
              return -ENOTDIR;
 
@@ -4765,6 +4769,8 @@ static int do_o_path(struct nameidata *nd, unsigned flags, struct file *file)
      struct path path;
      int error = path_lookupat(nd, flags, &path);
      if (!error) {
+             if ((file->f_flags & O_REGULAR) && !d_is_reg(path.dentry))
+                     return -ENOTREG;
              audit_inode(nd->name, path.dentry, 0);
              error = vfs_open(&path, file);
              path_put(&path);
fs/open.c +3 -1
index 74c4c1462b3e..82153e21907e 100644
@@ -1173,7 +1173,7 @@ struct file *kernel_file_open(const struct path *path, int flags,
 EXPORT_SYMBOL_GPL(kernel_file_open);
 
 #define WILL_CREATE(flags)   (flags & (O_CREAT | __O_TMPFILE))
-#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC)
+#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC | O_REGULAR)
 
 inline struct open_how build_open_how(int flags, umode_t mode)
 {
@@ -1250,6 +1250,8 @@ inline int build_open_flags(const struct open_how *how, struct open_flags *op)
                      return -EINVAL;
              if (!(acc_mode & MAY_WRITE))
                      return -EINVAL;
+     } else if ((flags & O_DIRECTORY) && (flags & O_REGULAR)) {
+             return -EINVAL;
      }
      if (flags & O_PATH) {
              /* O_PATH only permits certain other flags to be set. */
include/linux/fcntl.h +1 -1
index a332e79b3207..4fd07b0e0a17 100644
@@ -10,7 +10,7 @@
      (O_RDONLY | O_WRONLY | O_RDWR | O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC | \
       O_APPEND | O_NDELAY | O_NONBLOCK | __O_SYNC | O_DSYNC | \
       FASYNC | O_DIRECT | O_LARGEFILE | O_DIRECTORY | O_NOFOLLOW | \
-      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE)
+      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE | O_REGULAR)
 
 /* List of all valid flags for the how->resolve argument: */
 #define VALID_RESOLVE_FLAGS \
include/uapi/asm-generic/errno.h +2 -0
index 92e7ae493ee3..2216ab9aa32e 100644
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif
include/uapi/asm-generic/fcntl.h +4 -0
index 613475285643..3468b352a575 100644
@@ -88,6 +88,10 @@
 #define __O_TMPFILE  020000000
 #endif
 
+#ifndef O_REGULAR
+#define O_REGULAR    040000000
+#endif
+
 /* a horrid kludge trying to make sure that this will fail on old kernels */
 #define O_TMPFILE (__O_TMPFILE | O_DIRECTORY)
 
tools/arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
tools/arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
tools/arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
tools/arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
tools/include/uapi/asm-generic/errno.h +0 -0
index 92e7ae493ee3..2216ab9aa32e 100644

tools/include/uapi/asm-generic/errno.h line 127

Quoted 6 lines
tools/include/uapi/asm-generic/errno.h +2 -0
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif

One thing this patch is missing is handling for ->atomic_open(). I imagine most of the filesystems that provide that op can't support O_REGULAR properly (maybe cifs can? idk). What you probably want to do is add in some patches that make all of the atomic_open operations in the kernel return -EINVAL if O_REGULAR is set.

Quoted 2 lines

Then, once the basic support is in, you or someone else can go back and implement support for O_REGULAR where possible.

Quoted 11 lines

Thank you for the feedback. I don't quite understand what I need to fix. I thought open system calls always create regular files, so atomic_open probably always creates regular files? Can you please give me some more details as to where I need to fix this and what the actual bug here is that is related to atomic_open? I think I had done some normal testing and when using O_CREAT | O_REGULAR, if the file doesn't exist, the file gets created and the file that gets created is a regular file, so it probably makes sense? Or should the behavior be that if file doesn't exist, -EINVAL is returned and if file exists it is opened if regular, otherwise -ENOTREG is returned?

atomic_open() is a combination lookup+open for when the dentry isn't present in the dcache. The normal open codepath that you're patching does not get called in this case when ->atomic_open is set for the filesystem. It's mostly used by network filesystems that need to optimize away the lookup since it's wasted round trip, and is often racy anyway. Your patchset doesn't address those filesystems. They will likely end up ignoring O_REGULAR in that case, which is not what you want.

What I was suggesting is that, as an interim step, you find all of the atomic_open operations in the kernel (there are maybe a dozen or so), and just make them return -EINVAL if someone sets O_DIRECTORY. Later, you or someone else can then go back and do a proper implementation of O_REGULAR handling on those filesystems, at least on the ones where it's possible. You will probably also need to similarly patch the open() routines for those filesystems too. Otherwise you'll get inconsistent behavior vs. when the dentry is in the cache.

Quoted 3 lines

One note: I think NFS probably can support O_DIRECTORY, since its OPEN call only works on files. We'll need to change how we handle errors from the server when it's set though.

So I think you're proposing two separate things or there's a typo:

(1) blocking O_DIRECTORY for ->atomic_open::
(2) blocking O_REGULAR for ->atomic_open::

The (1) point implies that O_DIRECTORY currently doesn't work correctly with atomic open for all filesystems.

Ever since 43b450632676 ("open: return EINVAL for O_DIRECTORY | O_CREAT") O_DIRECTORY with O_CREAT is blocked. It was accidently allowed and completely broken before that.

Quoted 9 lines

For O_DIRECTORY without O_CREAT the kernel will pass that down through ->atomic_open:: to the filesystem.

The worry that I see is that a network filesystem via ->atomic_open::
somehow already called open on the server side on something that wasn't a directory. At that point the damage such as side-effects from device opening is already done.

Exactly. I guess you could send an immediate close, but that's not without side effects.

Quoted 21 lines

But I suspect that every filesystem implementing ->atomic_open:: just does finish_no_open() and punts to the VFS for the actual open. And the VFS will catch it in do_open() for it actually opens the file. So the only real worry for O_DIRECTORY I see is that there's an fs that handles it wrong.

For (2) it is problematic as there surely are filesystems with ->atomic_open:: that do handle the ~O_CREAT case and return with FMODE_OPENED. So that'll be problematic if the intention is to not trigger an actual open on a non-regular file such as a device/socket/fifo etc. before the VFS had a chance to validate what's going on.

So I'm not excited about having this 70% working and punting on ->atomic_open:: waiting for someone to fix this. One option would be to bypass ->atomic_open:: for OPENAT2_REGULAR without O_CREAT and fallback to racy and pricy lookup + open for now. How problematic would that be? If possible I'd prefer this a lot over merging something that works half-way.

Quoted 7 lines
Quoted 1 line

I guess to make that really work you'd need some protocol extension?

For NFS, I think we're OK. The OPEN call on NFSv4 only works for regular files, so it should be able to handle O_REGULAR. We just need to rejigger the error handling when it's set (just return an error instead of doing the open of a directory or whatever it is).

Quoted 5 lines

Thank you for the details. Do you remember which codepath this is? Is this the inode_operations.atomic_open codepath or file_operations.open codepath? I am a bit confused also about where exactly the error handling that needs to be done.

I was thinking nfs_atomic_open().

Quoted 8 lines

Looking now, I think it might actually work OK without changes. It just might not be terribly efficient about it.

If the open_context() call returns -EISDIR or similar, then you really don't need to do the call to nfs_lookup() and the like. You can just return an immediate error when O_REGULAR is set since you know it's not suitable to be opened.

Quoted 4 lines

Right. And I guess we don't need to worry about O_REGULAR being an unknown flag when it gets sent to the server (not only for NFS / but others as well)?

You shouldn't. We don't send POSIX flags in NFSv4 requests. It has its own set of flags. In the case of NFSv4, O_REGULAR is already implied in an OPEN call on the wire. OPEN only operates on regular files.

Quoted 30 lines
Quoted 23 lines
Quoted 14 lines
Quoted 14 lines

The others (at a quick glance):

cifs: I don't see a way to specify an O_REGULAR equivalent to the SMB2_CREATE call and it looks like it can create directories. Maybe SteveF (cc'ed) knows if this is possible?

ceph: I think CEPH_MDS_OP_OPEN might only work for files, in which case O_REGULAR can probably be supported similarly to NFS.

fuse: probably ok? Does finish_no_open() in most cases. May depend on the userland driver though.

gfs2: is ok, it just does finish_no_open() in most cases anyway

Quoted 8 lines
Quoted 3 lines

vboxsf: does finish_no_open on non-creates, so you could probably just punt to that if O_REGULAR is set.

These are all inode_operations.atomic_open code paths, right? Because you mentioned finish_no_open and I see finish_no_open in the atomic_open code paths as opposed to file_operations.open code paths.

Quoted 6 lines

Note that this was just a cursory look. Someone will need to do a deeper dive and test these cases.

I think most will end up working ok, since most fall back to doing a finish_no_open(). There may be opportunities to optimize some of these cases though (similarly to how I mentioned with NFS).

Quoted 7 lines

I can try to look into these and see if I can implement handling for O_REGULAR flag for these filesystems in the atomic_open code paths. Thanks for the details.

Will I need to modify the corresponding file_operations.open code paths as well along with atomic_open code paths?

Probably not.

The main thing to keep in mind is that ->open is used when we already have a dentry for the target of the open. ->atomic_open is used when we don't have one yet or the one we have has failed revalidation.

If you have a valid dentry, then you should be able to satisfy the O_REGULAR check without having to call into ->open at all.

Signature · 1 line
Jeff Layton <[email protected]>
#17 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

Quoted 404 lines · Jeff Layton
Quoted 402 lines · Dorjoy Chowdhury
Quoted 400 lines · Jeff Layton
Quoted 382 lines · Dorjoy Chowdhury
Quoted 373 lines · Jeff Layton
Quoted 337 lines · Christian Brauner
Quoted 320 lines · Jeff Layton
Quoted 289 lines · Dorjoy Chowdhury
Quoted 285 lines · Jeff Layton
Quoted 270 lines · Dorjoy Chowdhury

This flag indicates the path should be opened if it's a regular file. This is useful to write secure programs that want to avoid being tricked into opening device nodes with special semantics while thinking they operate on regular files.

A corresponding error code ENOTREG has been introduced. For example, if open is called on path /dev/null with O_REGULAR in the flag param, it will return -ENOTREG.

When used in combination with O_CREAT, either the regular file is created, or if the path already exists, it is opened if it's a regular file. Otherwise, -ENOTREG is returned.

-EINVAL is returned when O_REGULAR is combined with O_DIRECTORY (not part of O_TMPFILE) because it doesn't make sense to open a path that is both a directory and a regular file.

---
 arch/alpha/include/uapi/asm/errno.h        | 2 ++
 arch/alpha/include/uapi/asm/fcntl.h        | 1 +
 arch/mips/include/uapi/asm/errno.h         | 2 ++
 arch/parisc/include/uapi/asm/errno.h       | 2 ++
 arch/parisc/include/uapi/asm/fcntl.h       | 1 +
 arch/sparc/include/uapi/asm/errno.h        | 2 ++
 arch/sparc/include/uapi/asm/fcntl.h        | 1 +
 fs/fcntl.c                                 | 2 +-
 fs/namei.c                                 | 6 ++++++
 fs/open.c                                  | 4 +++-
 include/linux/fcntl.h                      | 2 +-
 include/uapi/asm-generic/errno.h           | 2 ++
 include/uapi/asm-generic/fcntl.h           | 4 ++++
 tools/arch/alpha/include/uapi/asm/errno.h  | 2 ++
 tools/arch/mips/include/uapi/asm/errno.h   | 2 ++
 tools/arch/parisc/include/uapi/asm/errno.h | 2 ++
 tools/arch/sparc/include/uapi/asm/errno.h  | 2 ++
 tools/include/uapi/asm-generic/errno.h     | 2 ++
 18 files changed, 38 insertions(+), 3 deletions(-)
arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
arch/alpha/include/uapi/asm/fcntl.h +1 -0
index 50bdc8e8a271..4da5a64c23bd 100644
@@ -34,6 +34,7 @@
 
 #define O_PATH               040000000
 #define __O_TMPFILE  0100000000
+#define O_REGULAR    0200000000
 
 #define F_GETLK              7
 #define F_SETLK              8
arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
arch/parisc/include/uapi/asm/fcntl.h +1 -0
index 03dee816cb13..0cc3320fe326 100644
@@ -19,6 +19,7 @@
 
 #define O_PATH               020000000
 #define __O_TMPFILE  040000000
+#define O_REGULAR    0100000000
 
 #define F_GETLK64    8
 #define F_SETLK64    9
arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
arch/sparc/include/uapi/asm/fcntl.h +1 -0
index 67dae75e5274..a93d18d2c23e 100644
@@ -37,6 +37,7 @@
 
 #define O_PATH               0x1000000
 #define __O_TMPFILE  0x2000000
+#define O_REGULAR    0x4000000
 
 #define F_GETOWN     5       /*  for sockets. */
 #define F_SETOWN     6       /*  for sockets. */
fs/fcntl.c +1 -1
index f93dbca08435..62ab4ad2b6f5 100644
@@ -1169,7 +1169,7 @@ static int __init fcntl_init(void)
       * Exceptions: O_NONBLOCK is a two bit define on parisc; O_NDELAY
       * is defined as O_NONBLOCK on some platforms and not on others.
       */
-     BUILD_BUG_ON(20 - 1 /* for O_RDONLY being 0 */ !=
+     BUILD_BUG_ON(21 - 1 /* for O_RDONLY being 0 */ !=
              HWEIGHT32(
                      (VALID_OPEN_FLAGS & ~(O_NONBLOCK | O_NDELAY)) |
                      __FMODE_EXEC));
fs/namei.c +6 -0
index b28ecb699f32..f5504ae4b03c 100644
@@ -4616,6 +4616,10 @@ static int do_open(struct nameidata *nd,
              if (unlikely(error))
                      return error;
      }
+
+     if ((open_flag & O_REGULAR) && !d_is_reg(nd->path.dentry))
+             return -ENOTREG;
+
      if ((nd->flags & LOOKUP_DIRECTORY) && !d_can_lookup(nd->path.dentry))
              return -ENOTDIR;
 
@@ -4765,6 +4769,8 @@ static int do_o_path(struct nameidata *nd, unsigned flags, struct file *file)
      struct path path;
      int error = path_lookupat(nd, flags, &path);
      if (!error) {
+             if ((file->f_flags & O_REGULAR) && !d_is_reg(path.dentry))
+                     return -ENOTREG;
              audit_inode(nd->name, path.dentry, 0);
              error = vfs_open(&path, file);
              path_put(&path);
fs/open.c +3 -1
index 74c4c1462b3e..82153e21907e 100644
@@ -1173,7 +1173,7 @@ struct file *kernel_file_open(const struct path *path, int flags,
 EXPORT_SYMBOL_GPL(kernel_file_open);
 
 #define WILL_CREATE(flags)   (flags & (O_CREAT | __O_TMPFILE))
-#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC)
+#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC | O_REGULAR)
 
 inline struct open_how build_open_how(int flags, umode_t mode)
 {
@@ -1250,6 +1250,8 @@ inline int build_open_flags(const struct open_how *how, struct open_flags *op)
                      return -EINVAL;
              if (!(acc_mode & MAY_WRITE))
                      return -EINVAL;
+     } else if ((flags & O_DIRECTORY) && (flags & O_REGULAR)) {
+             return -EINVAL;
      }
      if (flags & O_PATH) {
              /* O_PATH only permits certain other flags to be set. */
include/linux/fcntl.h +1 -1
index a332e79b3207..4fd07b0e0a17 100644
@@ -10,7 +10,7 @@
      (O_RDONLY | O_WRONLY | O_RDWR | O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC | \
       O_APPEND | O_NDELAY | O_NONBLOCK | __O_SYNC | O_DSYNC | \
       FASYNC | O_DIRECT | O_LARGEFILE | O_DIRECTORY | O_NOFOLLOW | \
-      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE)
+      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE | O_REGULAR)
 
 /* List of all valid flags for the how->resolve argument: */
 #define VALID_RESOLVE_FLAGS \
include/uapi/asm-generic/errno.h +2 -0
index 92e7ae493ee3..2216ab9aa32e 100644
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif
include/uapi/asm-generic/fcntl.h +4 -0
index 613475285643..3468b352a575 100644
@@ -88,6 +88,10 @@
 #define __O_TMPFILE  020000000
 #endif
 
+#ifndef O_REGULAR
+#define O_REGULAR    040000000
+#endif
+
 /* a horrid kludge trying to make sure that this will fail on old kernels */
 #define O_TMPFILE (__O_TMPFILE | O_DIRECTORY)
 
tools/arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
tools/arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
tools/arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
tools/arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
tools/include/uapi/asm-generic/errno.h +0 -0
index 92e7ae493ee3..2216ab9aa32e 100644

tools/include/uapi/asm-generic/errno.h line 127

Quoted 6 lines
tools/include/uapi/asm-generic/errno.h +2 -0
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif

One thing this patch is missing is handling for ->atomic_open(). I imagine most of the filesystems that provide that op can't support O_REGULAR properly (maybe cifs can? idk). What you probably want to do is add in some patches that make all of the atomic_open operations in the kernel return -EINVAL if O_REGULAR is set.

Quoted 2 lines

Then, once the basic support is in, you or someone else can go back and implement support for O_REGULAR where possible.

Quoted 11 lines

Thank you for the feedback. I don't quite understand what I need to fix. I thought open system calls always create regular files, so atomic_open probably always creates regular files? Can you please give me some more details as to where I need to fix this and what the actual bug here is that is related to atomic_open? I think I had done some normal testing and when using O_CREAT | O_REGULAR, if the file doesn't exist, the file gets created and the file that gets created is a regular file, so it probably makes sense? Or should the behavior be that if file doesn't exist, -EINVAL is returned and if file exists it is opened if regular, otherwise -ENOTREG is returned?

atomic_open() is a combination lookup+open for when the dentry isn't present in the dcache. The normal open codepath that you're patching does not get called in this case when ->atomic_open is set for the filesystem. It's mostly used by network filesystems that need to optimize away the lookup since it's wasted round trip, and is often racy anyway. Your patchset doesn't address those filesystems. They will likely end up ignoring O_REGULAR in that case, which is not what you want.

What I was suggesting is that, as an interim step, you find all of the atomic_open operations in the kernel (there are maybe a dozen or so), and just make them return -EINVAL if someone sets O_DIRECTORY. Later, you or someone else can then go back and do a proper implementation of O_REGULAR handling on those filesystems, at least on the ones where it's possible. You will probably also need to similarly patch the open() routines for those filesystems too. Otherwise you'll get inconsistent behavior vs. when the dentry is in the cache.

Quoted 3 lines

One note: I think NFS probably can support O_DIRECTORY, since its OPEN call only works on files. We'll need to change how we handle errors from the server when it's set though.

So I think you're proposing two separate things or there's a typo:

(1) blocking O_DIRECTORY for ->atomic_open::
(2) blocking O_REGULAR for ->atomic_open::

The (1) point implies that O_DIRECTORY currently doesn't work correctly with atomic open for all filesystems.

Ever since 43b450632676 ("open: return EINVAL for O_DIRECTORY | O_CREAT") O_DIRECTORY with O_CREAT is blocked. It was accidently allowed and completely broken before that.

Quoted 9 lines

For O_DIRECTORY without O_CREAT the kernel will pass that down through ->atomic_open:: to the filesystem.

The worry that I see is that a network filesystem via ->atomic_open::
somehow already called open on the server side on something that wasn't a directory. At that point the damage such as side-effects from device opening is already done.

Exactly. I guess you could send an immediate close, but that's not without side effects.

Quoted 21 lines

But I suspect that every filesystem implementing ->atomic_open:: just does finish_no_open() and punts to the VFS for the actual open. And the VFS will catch it in do_open() for it actually opens the file. So the only real worry for O_DIRECTORY I see is that there's an fs that handles it wrong.

For (2) it is problematic as there surely are filesystems with ->atomic_open:: that do handle the ~O_CREAT case and return with FMODE_OPENED. So that'll be problematic if the intention is to not trigger an actual open on a non-regular file such as a device/socket/fifo etc. before the VFS had a chance to validate what's going on.

So I'm not excited about having this 70% working and punting on ->atomic_open:: waiting for someone to fix this. One option would be to bypass ->atomic_open:: for OPENAT2_REGULAR without O_CREAT and fallback to racy and pricy lookup + open for now. How problematic would that be? If possible I'd prefer this a lot over merging something that works half-way.

Quoted 7 lines
Quoted 1 line

I guess to make that really work you'd need some protocol extension?

For NFS, I think we're OK. The OPEN call on NFSv4 only works for regular files, so it should be able to handle O_REGULAR. We just need to rejigger the error handling when it's set (just return an error instead of doing the open of a directory or whatever it is).

Quoted 5 lines

Thank you for the details. Do you remember which codepath this is? Is this the inode_operations.atomic_open codepath or file_operations.open codepath? I am a bit confused also about where exactly the error handling that needs to be done.

I was thinking nfs_atomic_open().

Looking now, I think it might actually work OK without changes. It just might not be terribly efficient about it.

If the open_context() call returns -EISDIR or similar, then you really don't need to do the call to nfs_lookup() and the like. You can just return an immediate error when O_REGULAR is set since you know it's not suitable to be opened.

Quoted 9 lines
Quoted 4 lines

Right. And I guess we don't need to worry about O_REGULAR being an unknown flag when it gets sent to the server (not only for NFS / but others as well)?

You shouldn't. We don't send POSIX flags in NFSv4 requests. It has its own set of flags. In the case of NFSv4, O_REGULAR is already implied in an OPEN call on the wire. OPEN only operates on regular files.

Alright. Thanks for the details again!

Quoted 38 lines
Quoted 30 lines
Quoted 23 lines
Quoted 14 lines
Quoted 14 lines

The others (at a quick glance):

cifs: I don't see a way to specify an O_REGULAR equivalent to the SMB2_CREATE call and it looks like it can create directories. Maybe SteveF (cc'ed) knows if this is possible?

ceph: I think CEPH_MDS_OP_OPEN might only work for files, in which case O_REGULAR can probably be supported similarly to NFS.

fuse: probably ok? Does finish_no_open() in most cases. May depend on the userland driver though.

gfs2: is ok, it just does finish_no_open() in most cases anyway

Quoted 8 lines
Quoted 3 lines

vboxsf: does finish_no_open on non-creates, so you could probably just punt to that if O_REGULAR is set.

These are all inode_operations.atomic_open code paths, right? Because you mentioned finish_no_open and I see finish_no_open in the atomic_open code paths as opposed to file_operations.open code paths.

Quoted 6 lines

Note that this was just a cursory look. Someone will need to do a deeper dive and test these cases.

I think most will end up working ok, since most fall back to doing a finish_no_open(). There may be opportunities to optimize some of these cases though (similarly to how I mentioned with NFS).

Quoted 7 lines

I can try to look into these and see if I can implement handling for O_REGULAR flag for these filesystems in the atomic_open code paths. Thanks for the details.

Will I need to modify the corresponding file_operations.open code paths as well along with atomic_open code paths?

Quoted 8 lines

Probably not.

The main thing to keep in mind is that ->open is used when we already have a dentry for the target of the open. ->atomic_open is used when we don't have one yet or the one we have has failed revalidation.

If you have a valid dentry, then you should be able to satisfy the O_REGULAR check without having to call into ->open at all.

Understood. Thanks!

Regards,
Dorjoy

maintainer
#18 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

In reply to Christian Brauner
Quoted 321 lines · Christian Brauner
Quoted 320 lines · Jeff Layton
Quoted 289 lines · Dorjoy Chowdhury
Quoted 285 lines · Jeff Layton
Quoted 270 lines · Dorjoy Chowdhury

This flag indicates the path should be opened if it's a regular file. This is useful to write secure programs that want to avoid being tricked into opening device nodes with special semantics while thinking they operate on regular files.

A corresponding error code ENOTREG has been introduced. For example, if open is called on path /dev/null with O_REGULAR in the flag param, it will return -ENOTREG.

When used in combination with O_CREAT, either the regular file is created, or if the path already exists, it is opened if it's a regular file. Otherwise, -ENOTREG is returned.

-EINVAL is returned when O_REGULAR is combined with O_DIRECTORY (not part of O_TMPFILE) because it doesn't make sense to open a path that is both a directory and a regular file.

---
 arch/alpha/include/uapi/asm/errno.h        | 2 ++
 arch/alpha/include/uapi/asm/fcntl.h        | 1 +
 arch/mips/include/uapi/asm/errno.h         | 2 ++
 arch/parisc/include/uapi/asm/errno.h       | 2 ++
 arch/parisc/include/uapi/asm/fcntl.h       | 1 +
 arch/sparc/include/uapi/asm/errno.h        | 2 ++
 arch/sparc/include/uapi/asm/fcntl.h        | 1 +
 fs/fcntl.c                                 | 2 +-
 fs/namei.c                                 | 6 ++++++
 fs/open.c                                  | 4 +++-
 include/linux/fcntl.h                      | 2 +-
 include/uapi/asm-generic/errno.h           | 2 ++
 include/uapi/asm-generic/fcntl.h           | 4 ++++
 tools/arch/alpha/include/uapi/asm/errno.h  | 2 ++
 tools/arch/mips/include/uapi/asm/errno.h   | 2 ++
 tools/arch/parisc/include/uapi/asm/errno.h | 2 ++
 tools/arch/sparc/include/uapi/asm/errno.h  | 2 ++
 tools/include/uapi/asm-generic/errno.h     | 2 ++
 18 files changed, 38 insertions(+), 3 deletions(-)
arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
arch/alpha/include/uapi/asm/fcntl.h +1 -0
index 50bdc8e8a271..4da5a64c23bd 100644
@@ -34,6 +34,7 @@
 
 #define O_PATH               040000000
 #define __O_TMPFILE  0100000000
+#define O_REGULAR    0200000000
 
 #define F_GETLK              7
 #define F_SETLK              8
arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
arch/parisc/include/uapi/asm/fcntl.h +1 -0
index 03dee816cb13..0cc3320fe326 100644
@@ -19,6 +19,7 @@
 
 #define O_PATH               020000000
 #define __O_TMPFILE  040000000
+#define O_REGULAR    0100000000
 
 #define F_GETLK64    8
 #define F_SETLK64    9
arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
arch/sparc/include/uapi/asm/fcntl.h +1 -0
index 67dae75e5274..a93d18d2c23e 100644
@@ -37,6 +37,7 @@
 
 #define O_PATH               0x1000000
 #define __O_TMPFILE  0x2000000
+#define O_REGULAR    0x4000000
 
 #define F_GETOWN     5       /*  for sockets. */
 #define F_SETOWN     6       /*  for sockets. */
fs/fcntl.c +1 -1
index f93dbca08435..62ab4ad2b6f5 100644
@@ -1169,7 +1169,7 @@ static int __init fcntl_init(void)
       * Exceptions: O_NONBLOCK is a two bit define on parisc; O_NDELAY
       * is defined as O_NONBLOCK on some platforms and not on others.
       */
-     BUILD_BUG_ON(20 - 1 /* for O_RDONLY being 0 */ !=
+     BUILD_BUG_ON(21 - 1 /* for O_RDONLY being 0 */ !=
              HWEIGHT32(
                      (VALID_OPEN_FLAGS & ~(O_NONBLOCK | O_NDELAY)) |
                      __FMODE_EXEC));
fs/namei.c +6 -0
index b28ecb699f32..f5504ae4b03c 100644
@@ -4616,6 +4616,10 @@ static int do_open(struct nameidata *nd,
              if (unlikely(error))
                      return error;
      }
+
+     if ((open_flag & O_REGULAR) && !d_is_reg(nd->path.dentry))
+             return -ENOTREG;
+
      if ((nd->flags & LOOKUP_DIRECTORY) && !d_can_lookup(nd->path.dentry))
              return -ENOTDIR;
 
@@ -4765,6 +4769,8 @@ static int do_o_path(struct nameidata *nd, unsigned flags, struct file *file)
      struct path path;
      int error = path_lookupat(nd, flags, &path);
      if (!error) {
+             if ((file->f_flags & O_REGULAR) && !d_is_reg(path.dentry))
+                     return -ENOTREG;
              audit_inode(nd->name, path.dentry, 0);
              error = vfs_open(&path, file);
              path_put(&path);
fs/open.c +3 -1
index 74c4c1462b3e..82153e21907e 100644
@@ -1173,7 +1173,7 @@ struct file *kernel_file_open(const struct path *path, int flags,
 EXPORT_SYMBOL_GPL(kernel_file_open);
 
 #define WILL_CREATE(flags)   (flags & (O_CREAT | __O_TMPFILE))
-#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC)
+#define O_PATH_FLAGS         (O_DIRECTORY | O_NOFOLLOW | O_PATH | O_CLOEXEC | O_REGULAR)
 
 inline struct open_how build_open_how(int flags, umode_t mode)
 {
@@ -1250,6 +1250,8 @@ inline int build_open_flags(const struct open_how *how, struct open_flags *op)
                      return -EINVAL;
              if (!(acc_mode & MAY_WRITE))
                      return -EINVAL;
+     } else if ((flags & O_DIRECTORY) && (flags & O_REGULAR)) {
+             return -EINVAL;
      }
      if (flags & O_PATH) {
              /* O_PATH only permits certain other flags to be set. */
include/linux/fcntl.h +1 -1
index a332e79b3207..4fd07b0e0a17 100644
@@ -10,7 +10,7 @@
      (O_RDONLY | O_WRONLY | O_RDWR | O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC | \
       O_APPEND | O_NDELAY | O_NONBLOCK | __O_SYNC | O_DSYNC | \
       FASYNC | O_DIRECT | O_LARGEFILE | O_DIRECTORY | O_NOFOLLOW | \
-      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE)
+      O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE | O_REGULAR)
 
 /* List of all valid flags for the how->resolve argument: */
 #define VALID_RESOLVE_FLAGS \
include/uapi/asm-generic/errno.h +2 -0
index 92e7ae493ee3..2216ab9aa32e 100644
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif
include/uapi/asm-generic/fcntl.h +4 -0
index 613475285643..3468b352a575 100644
@@ -88,6 +88,10 @@
 #define __O_TMPFILE  020000000
 #endif
 
+#ifndef O_REGULAR
+#define O_REGULAR    040000000
+#endif
+
 /* a horrid kludge trying to make sure that this will fail on old kernels */
 #define O_TMPFILE (__O_TMPFILE | O_DIRECTORY)
 
tools/arch/alpha/include/uapi/asm/errno.h +2 -0
index 6791f6508632..8bbcaa9024f9 100644
@@ -127,4 +127,6 @@
 
 #define EHWPOISON    139     /* Memory page has hardware error */
 
+#define ENOTREG              140     /* Not a regular file */
+
 #endif
tools/arch/mips/include/uapi/asm/errno.h +2 -0
index c01ed91b1ef4..293c78777254 100644
@@ -126,6 +126,8 @@
 
 #define EHWPOISON    168     /* Memory page has hardware error */
 
+#define ENOTREG              169     /* Not a regular file */
+
 #define EDQUOT               1133    /* Quota exceeded */
 
 
tools/arch/parisc/include/uapi/asm/errno.h +2 -0
index 8cbc07c1903e..442917484f99 100644
@@ -124,4 +124,6 @@
 
 #define EHWPOISON    257     /* Memory page has hardware error */
 
+#define ENOTREG              258     /* Not a regular file */
+
 #endif
tools/arch/sparc/include/uapi/asm/errno.h +2 -0
index 4a41e7835fd5..8dce0bfeab74 100644
@@ -117,4 +117,6 @@
 
 #define EHWPOISON    135     /* Memory page has hardware error */
 
+#define ENOTREG              136     /* Not a regular file */
+
 #endif
tools/include/uapi/asm-generic/errno.h +0 -0
index 92e7ae493ee3..2216ab9aa32e 100644

tools/include/uapi/asm-generic/errno.h line 127

Quoted 6 lines
tools/include/uapi/asm-generic/errno.h +2 -0
@@ -122,4 +122,6 @@
 
 #define EHWPOISON    133     /* Memory page has hardware error */
 
+#define ENOTREG              134     /* Not a regular file */
+
 #endif

One thing this patch is missing is handling for ->atomic_open(). I imagine most of the filesystems that provide that op can't support O_REGULAR properly (maybe cifs can? idk). What you probably want to do is add in some patches that make all of the atomic_open operations in the kernel return -EINVAL if O_REGULAR is set.

Quoted 2 lines

Then, once the basic support is in, you or someone else can go back and implement support for O_REGULAR where possible.

Quoted 11 lines

Thank you for the feedback. I don't quite understand what I need to fix. I thought open system calls always create regular files, so atomic_open probably always creates regular files? Can you please give me some more details as to where I need to fix this and what the actual bug here is that is related to atomic_open? I think I had done some normal testing and when using O_CREAT | O_REGULAR, if the file doesn't exist, the file gets created and the file that gets created is a regular file, so it probably makes sense? Or should the behavior be that if file doesn't exist, -EINVAL is returned and if file exists it is opened if regular, otherwise -ENOTREG is returned?

atomic_open() is a combination lookup+open for when the dentry isn't present in the dcache. The normal open codepath that you're patching does not get called in this case when ->atomic_open is set for the filesystem. It's mostly used by network filesystems that need to optimize away the lookup since it's wasted round trip, and is often racy anyway. Your patchset doesn't address those filesystems. They will likely end up ignoring O_REGULAR in that case, which is not what you want.

What I was suggesting is that, as an interim step, you find all of the atomic_open operations in the kernel (there are maybe a dozen or so), and just make them return -EINVAL if someone sets O_DIRECTORY. Later, you or someone else can then go back and do a proper implementation of O_REGULAR handling on those filesystems, at least on the ones where it's possible. You will probably also need to similarly patch the open() routines for those filesystems too. Otherwise you'll get inconsistent behavior vs. when the dentry is in the cache.

Quoted 6 lines
Quoted 3 lines

One note: I think NFS probably can support O_DIRECTORY, since its OPEN call only works on files. We'll need to change how we handle errors from the server when it's set though.

So I think you're proposing two separate things or there's a typo:

Oops, that was a typo.

Quoted 6 lines

(1) blocking O_DIRECTORY for ->atomic_open::
(2) blocking O_REGULAR for ->atomic_open::

The (1) point implies that O_DIRECTORY currently doesn't work correctly with atomic open for all filesystems.

It looks like NFS at least handles O_DIRECTORY properly. I don't have any reason to believe that O_DIRECTORY handling is broken on the others.

Quoted 33 lines

Ever since 43b450632676 ("open: return EINVAL for O_DIRECTORY | O_CREAT") O_DIRECTORY with O_CREAT is blocked. It was accidently allowed and completely broken before that.

For O_DIRECTORY without O_CREAT the kernel will pass that down through ->atomic_open:: to the filesystem.

The worry that I see is that a network filesystem via ->atomic_open::
somehow already called open on the server side on something that wasn't a directory. At that point the damage such as side-effects from device opening is already done.

But I suspect that every filesystem implementing ->atomic_open:: just does finish_no_open() and punts to the VFS for the actual open. And the VFS will catch it in do_open() for it actually opens the file. So the only real worry for O_DIRECTORY I see is that there's an fs that handles it wrong.

For (2) it is problematic as there surely are filesystems with ->atomic_open:: that do handle the ~O_CREAT case and return with FMODE_OPENED. So that'll be problematic if the intention is to not trigger an actual open on a non-regular file such as a device/socket/fifo etc. before the VFS had a chance to validate what's going on.

So I'm not excited about having this 70% working and punting on ->atomic_open:: waiting for someone to fix this. One option would be to bypass ->atomic_open:: for OPENAT2_REGULAR without O_CREAT and fallback to racy and pricy lookup + open for now. How problematic would that be? If possible I'd prefer this a lot over merging something that works half-way.

I guess to make that really work you'd need some protocol extension?

Signature · 1 line
Jeff Layton <[email protected]>
#19 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

In reply to Dorjoy Chowdhury
Quoted 13 lines · Dorjoy Chowdhury

This flag indicates the path should be opened if it's a regular file. This is useful to write secure programs that want to avoid being tricked into opening device nodes with special semantics while thinking they operate on regular files.

A corresponding error code ENOTREG has been introduced. For example, if open is called on path /dev/null with O_REGULAR in the flag param, it will return -ENOTREG.

When used in combination with O_CREAT, either the regular file is created, or if the path already exists, it is opened if it's a regular file. Otherwise, -ENOTREG is returned.

Quoted 6 lines

-EINVAL is returned when O_REGULAR is combined with O_DIRECTORY (not part of O_TMPFILE) because it doesn't make sense to open a path that is both a directory and a regular file.

---

Yeah, we shouldn't add support for this outside of openat2(). We also shouldn't call this OEXT_* or O2_*. Let's just follow the pattern where we prefix the flag space with the name of the system call OPENAT2_REGULAR.

There's also no real need to make O_DIRECTORY exclusive with OPENAT2_REGULAR. Callers could legimitately want to open a directory or regular file but not anything else. If someone wants to operate on a whole filesystem tree but only wants to interact with regular files and directories and ignore devices, sockets, fifos etc it's very handy to just be able to set both in flags.

Frankly, this shouldn't be a flag at all but we already have O_DIRECTORY in there so no need to move this into a new field.

Add EFTYPE as the errno code. Some of the bsds including macos already have that.

#20 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

Quoted 19 lines · Christian Brauner
Quoted 17 lines · Dorjoy Chowdhury

This flag indicates the path should be opened if it's a regular file. This is useful to write secure programs that want to avoid being tricked into opening device nodes with special semantics while thinking they operate on regular files.

A corresponding error code ENOTREG has been introduced. For example, if open is called on path /dev/null with O_REGULAR in the flag param, it will return -ENOTREG.

When used in combination with O_CREAT, either the regular file is created, or if the path already exists, it is opened if it's a regular file. Otherwise, -ENOTREG is returned.

-EINVAL is returned when O_REGULAR is combined with O_DIRECTORY (not part of O_TMPFILE) because it doesn't make sense to open a path that is both a directory and a regular file.

Quoted 8 lines
Quoted 2 lines

---

Yeah, we shouldn't add support for this outside of openat2(). We also shouldn't call this OEXT_* or O2_*. Let's just follow the pattern where we prefix the flag space with the name of the system call OPENAT2_REGULAR.

Thanks for the feedback. I agree that OPENAT2_REGULAR is better than the other OEXT_*/O2_* options. Right now in the patch, the O_REGULAR took the next slot in all the fcntl files. Should OPENAT2_REGULAR be a bit outside of the 32bits? That way it won't take any of the regular O_* bits and we would only need to define it in include/uapi/asm-generic/fcntl.h file and not need it in arch/*/fcntl.h files. What do you think?

Quoted 7 lines

There's also no real need to make O_DIRECTORY exclusive with OPENAT2_REGULAR. Callers could legimitately want to open a directory or regular file but not anything else. If someone wants to operate on a whole filesystem tree but only wants to interact with regular files and directories and ignore devices, sockets, fifos etc it's very handy to just be able to set both in flags.

Quoted 5 lines

Frankly, this shouldn't be a flag at all but we already have O_DIRECTORY in there so no need to move this into a new field.

Add EFTYPE as the errno code. Some of the bsds including macos already have that.

Great suggestion. Will fixup in v4 submission.

Regards,
Dorjoy

#21 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

In reply to Christian Brauner
Quoted 33 lines · Christian Brauner
Quoted 13 lines · Dorjoy Chowdhury

This flag indicates the path should be opened if it's a regular file. This is useful to write secure programs that want to avoid being tricked into opening device nodes with special semantics while thinking they operate on regular files.

A corresponding error code ENOTREG has been introduced. For example, if open is called on path /dev/null with O_REGULAR in the flag param, it will return -ENOTREG.

When used in combination with O_CREAT, either the regular file is created, or if the path already exists, it is opened if it's a regular file. Otherwise, -ENOTREG is returned.

Quoted 6 lines

-EINVAL is returned when O_REGULAR is combined with O_DIRECTORY (not part of O_TMPFILE) because it doesn't make sense to open a path that is both a directory and a regular file.

---

Yeah, we shouldn't add support for this outside of openat2(). We also shouldn't call this OEXT_* or O2_*. Let's just follow the pattern where we prefix the flag space with the name of the system call OPENAT2_REGULAR.

There's also no real need to make O_DIRECTORY exclusive with OPENAT2_REGULAR. Callers could legimitately want to open a directory or regular file but not anything else. If someone wants to operate on a whole filesystem tree but only wants to interact with regular files and directories and ignore devices, sockets, fifos etc it's very handy to just be able to set both in flags.

Quoted 2 lines

Frankly, this shouldn't be a flag at all but we already have O_DIRECTORY in there so no need to move this into a new field.

You could even say O_NOFOLLOW is kinda like that too.

In my other mail I proposed a bitmask of S_IFMT to reject opening (which would let you allow FIFOs and regular files but block devices, etc). Unfortunately I forgot that S_IFBLK is S_IFCHR|S_IFDIR. This isn't fatal to the idea but it kinda sucks. Grr.

Quoted 2 lines

Add EFTYPE as the errno code. Some of the bsds including macos already have that.

Signature · 2 lines
Aleksa Sarai
https://www.cyphar.com/
#22 lore raw

Re: [PATCH v3 1/4] open: new O_REGULAR flag support

Quoted 38 lines · Aleksa Sarai
Quoted 33 lines · Christian Brauner
Quoted 13 lines · Dorjoy Chowdhury

This flag indicates the path should be opened if it's a regular file. This is useful to write secure programs that want to avoid being tricked into opening device nodes with special semantics while thinking they operate on regular files.

A corresponding error code ENOTREG has been introduced. For example, if open is called on path /dev/null with O_REGULAR in the flag param, it will return -ENOTREG.

When used in combination with O_CREAT, either the regular file is created, or if the path already exists, it is opened if it's a regular file. Otherwise, -ENOTREG is returned.

Quoted 6 lines

-EINVAL is returned when O_REGULAR is combined with O_DIRECTORY (not part of O_TMPFILE) because it doesn't make sense to open a path that is both a directory and a regular file.

---

Yeah, we shouldn't add support for this outside of openat2(). We also shouldn't call this OEXT_* or O2_*. Let's just follow the pattern where we prefix the flag space with the name of the system call OPENAT2_REGULAR.

There's also no real need to make O_DIRECTORY exclusive with OPENAT2_REGULAR. Callers could legimitately want to open a directory or regular file but not anything else. If someone wants to operate on a whole filesystem tree but only wants to interact with regular files and directories and ignore devices, sockets, fifos etc it's very handy to just be able to set both in flags.

Quoted 2 lines

Frankly, this shouldn't be a flag at all but we already have O_DIRECTORY in there so no need to move this into a new field.

Quoted 7 lines

You could even say O_NOFOLLOW is kinda like that too.

In my other mail I proposed a bitmask of S_IFMT to reject opening (which would let you allow FIFOs and regular files but block devices, etc). Unfortunately I forgot that S_IFBLK is S_IFCHR|S_IFDIR. This isn't fatal to the idea but it kinda sucks. Grr.

It is a good suggestion. I guess we can still introduce a new how->sfmt_allow field and have new bits (instead of keeping in sync with S_IF* ones) that allow types and just start with regular file allow bit for now, right? But I guess it would be cumbersome for users as an api to use different bits?

Regards,
Dorjoy

#23 lore raw

PATCH v3 2/4 kselftest/openat2: test for O_REGULAR flag

In reply to Dorjoy Chowdhury

Just a happy path test.

---
 .../testing/selftests/openat2/openat2_test.c  | 37 ++++++++++++++++++-
 1 file changed, 36 insertions(+), 1 deletion(-)
tools/testing/selftests/openat2/openat2_test.c +36 -1
index 0e161ef9e9e4..011ebc9af4e5 100644
@@ -320,8 +320,42 @@ void test_openat2_flags(void)
 	}
 }
 
+#ifndef O_REGULAR
+#define O_REGULAR 040000000
+#endif
+
+#ifndef ENOTREG
+#define ENOTREG 134
+#endif
+
+void test_openat2_o_regular_flag(void)
+{
+	if (!openat2_supported) {
+		ksft_test_result_skip("Skipping %s as openat2 is not supported\n", __func__);
+		return;
+	}
+
+	struct open_how how = {
+		.flags = O_REGULAR | O_RDONLY
+	};
+
+	int fd = sys_openat2(AT_FDCWD, "/dev/null", &how);
+
+	if (fd == ENOENT) {
+		ksft_test_result_skip("Skipping %s as there is no /dev/null\n", __func__);
+		return;
+	}
+
+	if (fd != -ENOTREG) {
+		ksft_test_result_fail("openat2 should return ENOTREG\n");
+		return;
+	}
+
+	ksft_test_result_pass("%s succeeded\n", __func__);
+}
+
 #define NUM_TESTS (NUM_OPENAT2_STRUCT_VARIATIONS * NUM_OPENAT2_STRUCT_TESTS + \
-		   NUM_OPENAT2_FLAG_TESTS)
+		   NUM_OPENAT2_FLAG_TESTS + 1)
 
 int main(int argc, char **argv)
 {
@@ -330,6 +364,7 @@ int main(int argc, char **argv)
 
 	test_openat2_struct();
 	test_openat2_flags();
+	test_openat2_o_regular_flag();
 
 	if (ksft_get_fail_cnt() + ksft_get_error_cnt() > 0)
 		ksft_exit_fail();
Signature · 1 line
2.52.0
#24 lore raw

PATCH v3 3/4 sparc/fcntl.h: convert O_* flag macros from hex to octal

In reply to Dorjoy Chowdhury

Following the convention in include/uapi/asm-generic/fcntl.h and other architecture specific arch/*/include/uapi/asm/fcntl.h files.

---
 arch/sparc/include/uapi/asm/fcntl.h | 36 ++++++++++++++---------------
 1 file changed, 18 insertions(+), 18 deletions(-)
arch/sparc/include/uapi/asm/fcntl.h +18 -18
index a93d18d2c23e..3c16f1a66a6a 100644
@@ -2,23 +2,23 @@
 #ifndef _SPARC_FCNTL_H
 #define _SPARC_FCNTL_H
 
-#define O_APPEND	0x0008
-#define FASYNC		0x0040	/* fcntl, for BSD compatibility */
-#define O_CREAT		0x0200	/* not fcntl */
-#define O_TRUNC		0x0400	/* not fcntl */
-#define O_EXCL		0x0800	/* not fcntl */
-#define O_DSYNC		0x2000	/* used to be O_SYNC, see below */
-#define O_NONBLOCK	0x4000
+#define O_APPEND	0000000010
+#define FASYNC		0000000100	/* fcntl, for BSD compatibility */
+#define O_CREAT		0000001000	/* not fcntl */
+#define O_TRUNC		0000002000	/* not fcntl */
+#define O_EXCL		0000004000	/* not fcntl */
+#define O_DSYNC		0000020000	/* used to be O_SYNC, see below */
+#define O_NONBLOCK	0000040000
 #if defined(__sparc__) && defined(__arch64__)
-#define O_NDELAY	0x0004
+#define O_NDELAY	0000000004
 #else
-#define O_NDELAY	(0x0004 | O_NONBLOCK)
+#define O_NDELAY	(0000000004 | O_NONBLOCK)
 #endif
-#define O_NOCTTY	0x8000	/* not fcntl */
-#define O_LARGEFILE	0x40000
-#define O_DIRECT        0x100000 /* direct disk access hint */
-#define O_NOATIME	0x200000
-#define O_CLOEXEC	0x400000
+#define O_NOCTTY	0000100000	/* not fcntl */
+#define O_LARGEFILE	0001000000
+#define O_DIRECT        0004000000 /* direct disk access hint */
+#define O_NOATIME	0010000000
+#define O_CLOEXEC	0020000000
 /*
  * Before Linux 2.6.33 only O_DSYNC semantics were implemented, but using
  * the O_SYNC flag.  We continue to use the existing numerical value
@@ -32,12 +32,12 @@
  *
  * Note: __O_SYNC must never be used directly.
  */
-#define __O_SYNC	0x800000
+#define __O_SYNC	0040000000
 #define O_SYNC		(__O_SYNC|O_DSYNC)
 
-#define O_PATH		0x1000000
-#define __O_TMPFILE	0x2000000
-#define O_REGULAR	0x4000000
+#define O_PATH		0100000000
+#define __O_TMPFILE	0200000000
+#define O_REGULAR	0400000000
 
 #define F_GETOWN	5	/*  for sockets. */
 #define F_SETOWN	6	/*  for sockets. */
Signature · 1 line
2.52.0
#25 lore raw

PATCH v3 4/4 mips/fcntl.h: convert O_* flag macros from hex to octal

In reply to Dorjoy Chowdhury

Following the convention in include/uapi/asm-generic/fcntl.h and other architecture specific arch/*/include/uapi/asm/fcntl.h files.

---
 arch/mips/include/uapi/asm/fcntl.h | 22 +++++++++++-----------
 1 file changed, 11 insertions(+), 11 deletions(-)
arch/mips/include/uapi/asm/fcntl.h +11 -11
index 0369a38e3d4f..6aa3f49df17e 100644
@@ -11,15 +11,15 @@
 
 #include <asm/sgidefs.h>
 
-#define O_APPEND	0x0008
-#define O_DSYNC		0x0010	/* used to be O_SYNC, see below */
-#define O_NONBLOCK	0x0080
-#define O_CREAT		0x0100	/* not fcntl */
-#define O_TRUNC		0x0200	/* not fcntl */
-#define O_EXCL		0x0400	/* not fcntl */
-#define O_NOCTTY	0x0800	/* not fcntl */
-#define FASYNC		0x1000	/* fcntl, for BSD compatibility */
-#define O_LARGEFILE	0x2000	/* allow large file opens */
+#define O_APPEND	0000010
+#define O_DSYNC		0000020	/* used to be O_SYNC, see below */
+#define O_NONBLOCK	0000200
+#define O_CREAT		0000400	/* not fcntl */
+#define O_TRUNC		0001000	/* not fcntl */
+#define O_EXCL		0002000	/* not fcntl */
+#define O_NOCTTY	0004000	/* not fcntl */
+#define FASYNC		0010000	/* fcntl, for BSD compatibility */
+#define O_LARGEFILE	0020000	/* allow large file opens */
 /*
  * Before Linux 2.6.33 only O_DSYNC semantics were implemented, but using
  * the O_SYNC flag.  We continue to use the existing numerical value
@@ -33,9 +33,9 @@
  *
  * Note: __O_SYNC must never be used directly.
  */
-#define __O_SYNC	0x4000
+#define __O_SYNC	0040000
 #define O_SYNC		(__O_SYNC|O_DSYNC)
-#define O_DIRECT	0x8000	/* direct disk access hint */
+#define O_DIRECT	0100000	/* direct disk access hint */
 
 #define F_GETLK		14
 #define F_SETLK		6
Signature · 1 line
2.52.0