[PATCH 0/2] ima: don't measure/appraise files on configfs

[PATCH 0/2] ima: don't measure/appraise files on configfs

[Frederick Lawler]

When IMA appraises/measures a file configfs, it has an opportunity to
hold onto the ima_iint_mutex while another file from a different file
system is attempting to appraise/measure. This may cause a cyclical
lock dependency.

Add configfs to the default don't measure/appraise lists, similarly to
other pseudo file systems.

Link: https://lore.kernel.org/all/6a77c7cd.b50370da.49fe0.0031.GAE@google.com/#t

Signed-off-by: Frederick Lawler <fred@cloudflare.com>
---
Frederick Lawler (2):
      configfs: move CONFIGFS_MAGIC definition to magic.h
      ima: don't measure/appraise files on configfs

 fs/configfs/mount.c                 | 4 +---
 include/uapi/linux/magic.h          | 1 +
 security/integrity/ima/ima_policy.c | 7 ++++++-
 3 files changed, 8 insertions(+), 4 deletions(-)
---
base-commit: f1e10b10874051e4d99911dae0dd7b75a9f8ae66
change-id: 20260818-configfs-a248c410b083

Best regards,
-- 
Frederick Lawler <fred@cloudflare.com>

[PATCH 1/2] configfs: move CONFIGFS_MAGIC definition to magic.h

[Frederick Lawler]

IMA shouldn't measure or appraise configfs, but currently does because
it's missing from the default exclusion policies. Move CONFIGFS_MAGIC to
magic.h to expose the file system's magic to IMA, as well as other userland
applications.

Suggested-by: Mimi Zohar <zohar@linux.ibm.com>
Signed-off-by: Frederick Lawler <fred@cloudflare.com>
---
 fs/configfs/mount.c        | 4 +---
 include/uapi/linux/magic.h | 1 +
 2 files changed, 2 insertions(+), 3 deletions(-)

diff --git a/fs/configfs/mount.c b/fs/configfs/mount.c
index 4929f343118946eaa55a539db4192e9c6621a8dc..d8cac1cbf3bd573a0549f2f044714c9d669205a4 100644
--- a/fs/configfs/mount.c
+++ b/fs/configfs/mount.c
@@ -9,6 +9,7 @@
  */
 
 #include <linux/fs.h>
+#include <linux/magic.h>
 #include <linux/module.h>
 #include <linux/mount.h>
 #include <linux/fs_context.h>
@@ -19,9 +20,6 @@
 #include <linux/configfs.h>
 #include "configfs_internal.h"
 
-/* Random magic number */
-#define CONFIGFS_MAGIC 0x62656570
-
 static struct vfsmount *configfs_mount = NULL;
 struct kmem_cache *configfs_dir_cachep;
 static int configfs_mnt_count = 0;
diff --git a/include/uapi/linux/magic.h b/include/uapi/linux/magic.h
index 4f2da935a76ccae6e1c3310b2fae5379186548b8..d5b9958ddfaae90b270b66fe7853c5a3d49ce739 100644
--- a/include/uapi/linux/magic.h
+++ b/include/uapi/linux/magic.h
@@ -8,6 +8,7 @@
 #define AUTOFS_SUPER_MAGIC	0x0187
 #define CEPH_SUPER_MAGIC	0x00c36400
 #define CODA_SUPER_MAGIC	0x73757245
+#define CONFIGFS_MAGIC		0x62656570	/* some random number */
 #define CRAMFS_MAGIC		0x28cd3d45	/* some random number */
 #define CRAMFS_MAGIC_WEND	0x453dcd28	/* magic number with the wrong endianess */
 #define DEBUGFS_MAGIC          0x64626720

-- 
2.43.0

Re: [PATCH 1/2] configfs: move CONFIGFS_MAGIC definition to magic.h

[Breno Leitao]

On Tue, Aug 18, 2026 at 02:35:15PM -0500, Frederick Lawler wrote:
> IMA shouldn't measure or appraise configfs, but currently does because
> it's missing from the default exclusion policies. Move CONFIGFS_MAGIC to
> magic.h to expose the file system's magic to IMA, as well as other userland
> applications.
> 
> Suggested-by: Mimi Zohar <zohar@linux.ibm.com>
> Signed-off-by: Frederick Lawler <fred@cloudflare.com>

Reviewed-by: Breno Leitao <leitao@debian.org>

> +#define CONFIGFS_MAGIC		0x62656570	/* some random number */

0x62656570 is ASCII "beep", I didn't know that :-)

Re: [PATCH 1/2] configfs: move CONFIGFS_MAGIC definition to magic.h

[Frederick Lawler]

On Wed, Aug 19, 2026 at 02:21:58AM -0700, Breno Leitao wrote:
> On Tue, Aug 18, 2026 at 02:35:15PM -0500, Frederick Lawler wrote:
> > IMA shouldn't measure or appraise configfs, but currently does because
> > it's missing from the default exclusion policies. Move CONFIGFS_MAGIC to
> > magic.h to expose the file system's magic to IMA, as well as other userland
> > applications.
> > 
> > Suggested-by: Mimi Zohar <zohar@linux.ibm.com>
> > Signed-off-by: Frederick Lawler <fred@cloudflare.com>
> 
> Reviewed-by: Breno Leitao <leitao@debian.org>
> 
> > +#define CONFIGFS_MAGIC		0x62656570	/* some random number */
> 
> 0x62656570 is ASCII "beep", I didn't know that :-)

I can't tell if it's a better joke to leave "some random number" as the
comment, or to change the comment to the more interesting fun fact about
the number.

Best,
Fred

[PATCH 2/2] ima: don't measure/appraise files on configfs

[Frederick Lawler]

When IMA appraises/measures a file configfs, it has an opportunity to
hold onto the ima_iint_mutex while another file from a different file
system is attempting to appraise/measure. This may cause a cyclical
lock dependency.

Add configfs to the default don't measure/appraise lists, similarly to
other pseudo file systems.

Reported-by: syzbot+448c2e24b1ceff13ed2a@syzkaller.appspotmail.com
Link: https://lore.kernel.org/all/6a77c7cd.b50370da.49fe0.0031.GAE@google.com/
Suggested-by: Mimi Zohar <zohar@linux.ibm.com>
Signed-off-by: Frederick Lawler <fred@cloudflare.com>
---
 security/integrity/ima/ima_policy.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/security/integrity/ima/ima_policy.c b/security/integrity/ima/ima_policy.c
index f79d07bb63c6fc4ba6fe594140de8d59f57e4f0b..68d9a5e6c232ea0678e9f51f105cebecccccb43e 100644
--- a/security/integrity/ima/ima_policy.c
+++ b/security/integrity/ima/ima_policy.c
@@ -165,7 +165,10 @@ static struct ima_rule_entry dont_measure_rules[] __ro_after_init = {
 	{.action = DONT_MEASURE, .fsmagic = CGROUP2_SUPER_MAGIC,
 	 .flags = IMA_FSMAGIC},
 	{.action = DONT_MEASURE, .fsmagic = NSFS_MAGIC, .flags = IMA_FSMAGIC},
-	{.action = DONT_MEASURE, .fsmagic = EFIVARFS_MAGIC, .flags = IMA_FSMAGIC}
+	{.action = DONT_MEASURE, .fsmagic = EFIVARFS_MAGIC,
+	 .flags = IMA_FSMAGIC},
+	{.action = DONT_MEASURE, .fsmagic = CONFIGFS_MAGIC,
+	 .flags = IMA_FSMAGIC}
 };
 
 static struct ima_rule_entry original_measurement_rules[] __ro_after_init = {
@@ -211,6 +214,8 @@ static struct ima_rule_entry default_appraise_rules[] __ro_after_init = {
 	{.action = DONT_APPRAISE, .fsmagic = EFIVARFS_MAGIC, .flags = IMA_FSMAGIC},
 	{.action = DONT_APPRAISE, .fsmagic = CGROUP_SUPER_MAGIC, .flags = IMA_FSMAGIC},
 	{.action = DONT_APPRAISE, .fsmagic = CGROUP2_SUPER_MAGIC, .flags = IMA_FSMAGIC},
+	{.action = DONT_APPRAISE, .fsmagic = CONFIGFS_MAGIC,
+	 .flags = IMA_FSMAGIC},
 #ifdef CONFIG_IMA_WRITE_POLICY
 	{.action = APPRAISE, .func = POLICY_CHECK,
 	.flags = IMA_FUNC | IMA_DIGSIG_REQUIRED},

-- 
2.43.0

Re: [PATCH 2/2] ima: don't measure/appraise files on configfs

[Mimi Zohar]

On Tue, 2026-08-18 at 14:35 -0500, Frederick Lawler wrote:
> When IMA appraises/measures a file configfs, it has an opportunity to
> hold onto the ima_iint_mutex while another file from a different file
> system is attempting to appraise/measure. This may cause a cyclical
> lock dependency.

The locking description is off a bit.

Suggested patch desciption from AI:

IMA measurement of a configfs file causes process_measurement() to hold iint-
>mutex while performing a kernel_read() to hash it, which re-enters configfs's
own file locking (buffer->mutex, frag_sem).

Separately, opening any file with O_TRUNC now causes ima_file_truncate() to take
iint->mutex to reset the cached action flags, while sb_writers is already held
for that mount.

When a configfs-backed nvmet namespace is involved, these two independent lock
chains combine into a cycle:

iint->mutex -> configfs locks -> subsys->lock -> sb_writers -> iint->mutex

Add configfs to the builtin don't measure/appraise rules, similarly to other
pseudo file systems, so IMA never takes iint->mutex for configfs file in the
first place.

> 
> Add configfs to the default don't measure/appraise lists, similarly to
> other pseudo file systems.
> 
> Reported-by: syzbot+448c2e24b1ceff13ed2a@syzkaller.appspotmail.com
> Link: https://lore.kernel.org/all/6a77c7cd.b50370da.49fe0.0031.GAE@google.com/
> Suggested-by: Mimi Zohar <zohar@linux.ibm.com>
> Signed-off-by: Frederick Lawler <fred@cloudflare.com>

Sashiko is now enabled for the linux-integrity mailing list -
https://sashiko.dev/#/?list=org.kernel.vger.linux-integrity.   Please address
Sashiko's 2/2 review.

thanks,

Mimi

> ---
>  security/integrity/ima/ima_policy.c | 7 ++++++-
>  1 file changed, 6 insertions(+), 1 deletion(-)
> 
> diff --git a/security/integrity/ima/ima_policy.c b/security/integrity/ima/ima_policy.c
> index f79d07bb63c6fc4ba6fe594140de8d59f57e4f0b..68d9a5e6c232ea0678e9f51f105cebecccccb43e 100644
> --- a/security/integrity/ima/ima_policy.c
> +++ b/security/integrity/ima/ima_policy.c
> @@ -165,7 +165,10 @@ static struct ima_rule_entry dont_measure_rules[] __ro_after_init = {
>  	{.action = DONT_MEASURE, .fsmagic = CGROUP2_SUPER_MAGIC,
>  	 .flags = IMA_FSMAGIC},
>  	{.action = DONT_MEASURE, .fsmagic = NSFS_MAGIC, .flags = IMA_FSMAGIC},
> -	{.action = DONT_MEASURE, .fsmagic = EFIVARFS_MAGIC, .flags = IMA_FSMAGIC}
> +	{.action = DONT_MEASURE, .fsmagic = EFIVARFS_MAGIC,
> +	 .flags = IMA_FSMAGIC},
> +	{.action = DONT_MEASURE, .fsmagic = CONFIGFS_MAGIC,
> +	 .flags = IMA_FSMAGIC}
>  };
>  
>  static struct ima_rule_entry original_measurement_rules[] __ro_after_init = {
> @@ -211,6 +214,8 @@ static struct ima_rule_entry default_appraise_rules[] __ro_after_init = {
>  	{.action = DONT_APPRAISE, .fsmagic = EFIVARFS_MAGIC, .flags = IMA_FSMAGIC},
>  	{.action = DONT_APPRAISE, .fsmagic = CGROUP_SUPER_MAGIC, .flags = IMA_FSMAGIC},
>  	{.action = DONT_APPRAISE, .fsmagic = CGROUP2_SUPER_MAGIC, .flags = IMA_FSMAGIC},
> +	{.action = DONT_APPRAISE, .fsmagic = CONFIGFS_MAGIC,
> +	 .flags = IMA_FSMAGIC},
>  #ifdef CONFIG_IMA_WRITE_POLICY
>  	{.action = APPRAISE, .func = POLICY_CHECK,
>  	.flags = IMA_FUNC | IMA_DIGSIG_REQUIRED},