Skip to contentSkip to conversation

security: add a mount option to make caps inheritable, re-enable CAP_SETPCAP

PATCH In review David Madore 2 participants 2 messages Sep 15, 2006 – Sep 19, 2006

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 security: add a mount option to make caps inheritable, re-enable CAP_SETPCAP

Attempt to make capabilities mildly useful, without breaking anything and while still adhering to POSIX.1e semantics:

* add a "inhcaps" mount option (MS_INHCAPS) which provides full executable inheritable and effective sets (we cannot provide finer-grained control over the mask, as fs-independent mount options are only one bit wide each);

* re-enable CAP_SETPCAP which had been disabled following an incorrect analysis of a past sendmail security hole.

---

*** IMPORTANT NOTE ***

This patch IS NOT related (nor compatible) with the one posted a few days ago on this list under the name "new capabilities patch". The latter was much more ambitious and deviated from POSIX.1e semantics. It has been abandoned due to heavy criticism on LKML. *This* patch is much restricted and very simple, and the author hopes that it will be somewhat
consensual: it does not change the (broken) semantics of capabilities in any way, only adds a mount option to enable inheritance; it also does not add any new capabilities, change the size of the capability sets or anything of the sort. It does not add any filesystem support for capabilities (only per-mount support), but should be compatible with patches which add such support.

Note that you need a patched version of mount (not provided...) to use the new "inhcaps" mount option.

 fs/namespace.c             |    5 ++++-
 include/linux/capability.h |    2 +-
 include/linux/fs.h         |    1 +
 include/linux/mount.h      |    1 +
 security/commoncap.c       |   10 ++++++++--
 5 files changed, 15 insertions(+), 4 deletions(-)
fs/namespace.c +4 -1
index fa7ed6a..63b6227 100644
@@ -376,6 +376,7 @@ static int show_vfsmnt(struct seq_file *
 		{ MNT_NOEXEC, ",noexec" },
 		{ MNT_NOATIME, ",noatime" },
 		{ MNT_NODIRATIME, ",nodiratime" },
+		{ MNT_INHCAPS, ",inhcaps" },
 		{ 0, NULL }
 	};
 	struct proc_fs_info *fs_infop;
@@ -1413,9 +1414,11 @@ long do_mount(char *dev_name, char *dir_
 		mnt_flags |= MNT_NOATIME;
 	if (flags & MS_NODIRATIME)
 		mnt_flags |= MNT_NODIRATIME;
+	if (flags & MS_INHCAPS)
+		mnt_flags |= MNT_INHCAPS;
 
 	flags &= ~(MS_NOSUID | MS_NOEXEC | MS_NODEV | MS_ACTIVE |
-		   MS_NOATIME | MS_NODIRATIME);
+		   MS_NOATIME | MS_NODIRATIME | MS_INHCAPS);
 
 	/* ... and get the mountpoint */
 	retval = path_lookup(dir_name, LOOKUP_FOLLOW, &nd);
include/linux/capability.h +1 -1
index 6548b35..e37a268 100644
@@ -312,7 +312,7 @@ #endif
 
 #define CAP_EMPTY_SET       to_cap_t(0)
 #define CAP_FULL_SET        to_cap_t(~0)
-#define CAP_INIT_EFF_SET    to_cap_t(~0 & ~CAP_TO_MASK(CAP_SETPCAP))
+#define CAP_INIT_EFF_SET    to_cap_t(~0)
 #define CAP_INIT_INH_SET    to_cap_t(0)
 
 #define CAP_TO_MASK(x) (1 << (x))
include/linux/fs.h +1 -0
index 555bc19..b41e515 100644
@@ -106,6 +106,7 @@ #define MS_SYNCHRONOUS	16	/* Writes are 
 #define MS_REMOUNT	32	/* Alter flags of a mounted FS */
 #define MS_MANDLOCK	64	/* Allow mandatory locks on an FS */
 #define MS_DIRSYNC	128	/* Directory modifications are synchronous */
+#define MS_INHCAPS	256	/* Inherit capabilities by default */
 #define MS_NOATIME	1024	/* Do not update access times. */
 #define MS_NODIRATIME	2048	/* Do not update directory access times */
 #define MS_BIND		4096
include/linux/mount.h +1 -0
index 403d1a9..b744823 100644
@@ -27,6 +27,7 @@ #define MNT_NODEV	0x02
 #define MNT_NOEXEC	0x04
 #define MNT_NOATIME	0x08
 #define MNT_NODIRATIME	0x10
+#define MNT_INHCAPS	0x20
 
 #define MNT_SHRINKABLE	0x100
 
security/commoncap.c +8 -2
index f50fc29..2488ed5 100644
@@ -13,6 +13,7 @@ #include <linux/init.h>
 #include <linux/kernel.h>
 #include <linux/security.h>
 #include <linux/file.h>
+#include <linux/mount.h>
 #include <linux/mm.h>
 #include <linux/mman.h>
 #include <linux/pagemap.h>
@@ -114,9 +115,14 @@ int cap_bprm_set_security (struct linux_
 	/* Copied from fs/exec.c:prepare_binprm. */
 
 	/* We don't have VFS support for capabilities yet */
-	cap_clear (bprm->cap_inheritable);
+	if (bprm->file->f_vfsmnt->mnt_flags & MNT_INHCAPS) {
+		cap_set_full (bprm->cap_inheritable);
+		cap_set_full (bprm->cap_effective);
+	} else {
+		cap_clear (bprm->cap_inheritable);
+		cap_clear (bprm->cap_effective);
+	}
 	cap_clear (bprm->cap_permitted);
-	cap_clear (bprm->cap_effective);
 
 	/*  To support inheritance of root-permissions and suid-root
 	 *  executables under compatibility mode, we raise all three
Acked-by on v1 1/1
#1 lore raw

Hi!

Quoted 10 lines

Attempt to make capabilities mildly useful, without breaking anything and while still adhering to POSIX.1e semantics:

* add a "inhcaps" mount option (MS_INHCAPS) which provides full executable inheritable and effective sets (we cannot provide finer-grained control over the mask, as fs-independent mount options are only one bit wide each);

* re-enable CAP_SETPCAP which had been disabled following an incorrect analysis of a past sendmail security hole.

Okay, I guess these two should go as a separate patches... Otherwise patch looks okay to me.

Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html