[v3,2/5] fs: Add a MAY_EXECMOUNT flag to infer the noexec mount property

Commit Message

Mickaël Salaün April 28, 2020, 5:51 p.m. UTC

An LSM doesn't get path information related to an access request to open
an inode.  This new (internal) MAY_EXECMOUNT flag enables an LSM to
check if the underlying mount point of an inode is marked as executable.
This is useful to implement a security policy taking advantage of the
noexec mount option.

This flag is set according to path_noexec(), which checks if a mount
point is mounted with MNT_NOEXEC or if the underlying superblock is
SB_I_NOEXEC.

Signed-off-by: Mickaël Salaün <[e-mail]>
Reviewed-by: Philippe Trébuchet <[e-mail]>
Reviewed-by: Thibaut Sautereau <[e-mail]>
Cc: Aleksa Sarai <[e-mail]>
Cc: Al Viro <[e-mail]>
Cc: Kees Cook <[e-mail]>
---
 fs/namei.c         | 2 ++
 include/linux/fs.h | 2 ++
 2 files changed, 4 insertions(+)

Comments

James Morris May 1, 2020, 4:02 a.m. UTC | #1

On Tue, 28 Apr 2020, Mickaël Salaün wrote:

> An LSM doesn't get path information related to an access request to open
> an inode.  This new (internal) MAY_EXECMOUNT flag enables an LSM to
> check if the underlying mount point of an inode is marked as executable.
> This is useful to implement a security policy taking advantage of the
> noexec mount option.
> 
> This flag is set according to path_noexec(), which checks if a mount
> point is mounted with MNT_NOEXEC or if the underlying superblock is
> SB_I_NOEXEC.
> 
> Signed-off-by: Mickaël Salaün <[e-mail]>
> Reviewed-by: Philippe Trébuchet <[e-mail]>
> Reviewed-by: Thibaut Sautereau <[e-mail]>
> Cc: Aleksa Sarai <[e-mail]>
> Cc: Al Viro <[e-mail]>
> Cc: Kees Cook <[e-mail]>

Are there any existing LSMs which plan to use this aspect?

Mickaël Salaün May 1, 2020, 2:17 p.m. UTC | #2

On 01/05/2020 06:02, James Morris wrote:
> On Tue, 28 Apr 2020, Mickaël Salaün wrote:
> 
>> An LSM doesn't get path information related to an access request to open
>> an inode.  This new (internal) MAY_EXECMOUNT flag enables an LSM to
>> check if the underlying mount point of an inode is marked as executable.
>> This is useful to implement a security policy taking advantage of the
>> noexec mount option.
>>
>> This flag is set according to path_noexec(), which checks if a mount
>> point is mounted with MNT_NOEXEC or if the underlying superblock is
>> SB_I_NOEXEC.
>>
>> Signed-off-by: Mickaël Salaün <[e-mail]>
>> Reviewed-by: Philippe Trébuchet <[e-mail]>
>> Reviewed-by: Thibaut Sautereau <[e-mail]>
>> Cc: Aleksa Sarai <[e-mail]>
>> Cc: Al Viro <[e-mail]>
>> Cc: Kees Cook <[e-mail]>
> 
> Are there any existing LSMs which plan to use this aspect?

This commit message was initially for the first version of O_MAYEXEC,
which extended Yama. The current enforcement implementation is now
directly in the FS subsystem (as requested by Kees Cook). However, this
MAY_EXECMOUNT flag is still used by the current FS implementation and it
could still be useful for LSMs.

Patch

diff --git a/fs/namei.c b/fs/namei.c
index a320371899cf..33b6d372e74a 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -2849,6 +2849,8 @@  static int may_open(const struct path *path, int acc_mode, int flag)
 		break;
 	}
 
+	/* Pass the mount point executability. */
+	acc_mode |= path_noexec(path) ? 0 : MAY_EXECMOUNT;
 	error = inode_permission(inode, MAY_OPEN | acc_mode);
 	if (error)
 		return error;
diff --git a/include/linux/fs.h b/include/linux/fs.h
index f5be4be7c01d..9213147d8636 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -103,6 +103,8 @@  typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
 #define MAY_NOT_BLOCK		0x00000080
 /* the inode is opened with RESOLVE_MAYEXEC */
 #define MAY_OPENEXEC		0x00000100
+/* the mount point is marked as executable */
+#define MAY_EXECMOUNT		0x00000200
 
 /*
  * flags in file.f_mode.  Note that FMODE_READ and FMODE_WRITE must correspond