[V37,04/29] Enforce module signatures if the kernel is locked down
diff mbox series

Message ID 20190731221617.234725-5-matthewgarrett@google.com
State New
Headers show
Series
  • security: Add support for locking down the kernel
Related show

Commit Message

Matthew Garrett July 31, 2019, 10:15 p.m. UTC
From: David Howells <dhowells@redhat.com>

If the kernel is locked down, require that all modules have valid
signatures that we can verify.

I have adjusted the errors generated:

 (1) If there's no signature (ENODATA) or we can't check it (ENOPKG,
     ENOKEY), then:

     (a) If signatures are enforced then EKEYREJECTED is returned.

     (b) If there's no signature or we can't check it, but the kernel is
	 locked down then EPERM is returned (this is then consistent with
	 other lockdown cases).

 (2) If the signature is unparseable (EBADMSG, EINVAL), the signature fails
     the check (EKEYREJECTED) or a system error occurs (eg. ENOMEM), we
     return the error we got.

Note that the X.509 code doesn't check for key expiry as the RTC might not
be valid or might not have been transferred to the kernel's clock yet.

 [Modified by Matthew Garrett to remove the IMA integration. This will
  be replaced with integration with the IMA architecture policy
  patchset.]

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Matthew Garrett <matthewgarrett@google.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Cc: Jessica Yu <jeyu@kernel.org>
---
 include/linux/security.h     |  1 +
 kernel/module.c              | 37 +++++++++++++++++++++++++++++-------
 security/lockdown/lockdown.c |  1 +
 3 files changed, 32 insertions(+), 7 deletions(-)

Comments

Jessica Yu Aug. 1, 2019, 2:21 p.m. UTC | #1
+++ Matthew Garrett [31/07/19 15:15 -0700]:
>From: David Howells <dhowells@redhat.com>
>
>If the kernel is locked down, require that all modules have valid
>signatures that we can verify.
>
>I have adjusted the errors generated:
>
> (1) If there's no signature (ENODATA) or we can't check it (ENOPKG,
>     ENOKEY), then:
>
>     (a) If signatures are enforced then EKEYREJECTED is returned.
>
>     (b) If there's no signature or we can't check it, but the kernel is
>	 locked down then EPERM is returned (this is then consistent with
>	 other lockdown cases).
>
> (2) If the signature is unparseable (EBADMSG, EINVAL), the signature fails
>     the check (EKEYREJECTED) or a system error occurs (eg. ENOMEM), we
>     return the error we got.
>
>Note that the X.509 code doesn't check for key expiry as the RTC might not
>be valid or might not have been transferred to the kernel's clock yet.
>
> [Modified by Matthew Garrett to remove the IMA integration. This will
>  be replaced with integration with the IMA architecture policy
>  patchset.]
>
>Signed-off-by: David Howells <dhowells@redhat.com>
>Signed-off-by: Matthew Garrett <matthewgarrett@google.com>
>Reviewed-by: Kees Cook <keescook@chromium.org>
>Cc: Jessica Yu <jeyu@kernel.org>
>---
> include/linux/security.h     |  1 +
> kernel/module.c              | 37 +++++++++++++++++++++++++++++-------
> security/lockdown/lockdown.c |  1 +
> 3 files changed, 32 insertions(+), 7 deletions(-)
>
>diff --git a/include/linux/security.h b/include/linux/security.h
>index 54a0532ec12f..8e70063074a1 100644
>--- a/include/linux/security.h
>+++ b/include/linux/security.h
>@@ -103,6 +103,7 @@ enum lsm_event {
>  */
> enum lockdown_reason {
> 	LOCKDOWN_NONE,
>+	LOCKDOWN_MODULE_SIGNATURE,
> 	LOCKDOWN_INTEGRITY_MAX,
> 	LOCKDOWN_CONFIDENTIALITY_MAX,
> };
>diff --git a/kernel/module.c b/kernel/module.c
>index cd8df516666d..318209889e26 100644
>--- a/kernel/module.c
>+++ b/kernel/module.c
>@@ -2771,8 +2771,9 @@ static inline void kmemleak_load_module(const struct module *mod,
> #ifdef CONFIG_MODULE_SIG
> static int module_sig_check(struct load_info *info, int flags)
> {
>-	int err = -ENOKEY;
>+	int err = -ENODATA;
> 	const unsigned long markerlen = sizeof(MODULE_SIG_STRING) - 1;
>+	const char *reason;
> 	const void *mod = info->hdr;
>
> 	/*
>@@ -2787,16 +2788,38 @@ static int module_sig_check(struct load_info *info, int flags)
> 		err = mod_verify_sig(mod, info);
> 	}
>
>-	if (!err) {
>+	switch (err) {
>+	case 0:
> 		info->sig_ok = true;
> 		return 0;
>-	}
>
>-	/* Not having a signature is only an error if we're strict. */
>-	if (err == -ENOKEY && !is_module_sig_enforced())
>-		err = 0;
>+		/* We don't permit modules to be loaded into trusted kernels
>+		 * without a valid signature on them, but if we're not
>+		 * enforcing, certain errors are non-fatal.
>+		 */
>+	case -ENODATA:
>+		reason = "Loading of unsigned module";
>+		goto decide;
>+	case -ENOPKG:
>+		reason = "Loading of module with unsupported crypto";
>+		goto decide;
>+	case -ENOKEY:
>+		reason = "Loading of module with unavailable key";
>+	decide:
>+		if (is_module_sig_enforced()) {
>+			pr_notice("%s is rejected\n", reason);
>+			return -EKEYREJECTED;
>+		}
>
>-	return err;
>+		return security_locked_down(LOCKDOWN_MODULE_SIGNATURE);
>+
>+		/* All other errors are fatal, including nomem, unparseable
>+		 * signatures and signature check failures - even if signatures
>+		 * aren't required.
>+		 */
>+	default:
>+		return err;
>+	}
> }
> #else /* !CONFIG_MODULE_SIG */
> static int module_sig_check(struct load_info *info, int flags)

Hi Matthew!

Apologies if this was addressed in another patch in your series (I've
only skimmed the first few), but what should happen if the kernel is
locked down, but CONFIG_MODULE_SIG=n? Or shouldn't CONFIG_SECURITY_LOCKDOWN_LSM
depend on CONFIG_MODULE_SIG? Otherwise I think we'll end up calling
the empty !CONFIG_MODULE_SIG module_sig_check() stub even though
lockdown is enabled.


Thanks,

Jessica

>diff --git a/security/lockdown/lockdown.c b/security/lockdown/lockdown.c
>index d30c4d254b5f..2c53fd9f5c9b 100644
>--- a/security/lockdown/lockdown.c
>+++ b/security/lockdown/lockdown.c
>@@ -18,6 +18,7 @@ static enum lockdown_reason kernel_locked_down;
>
> static char *lockdown_reasons[LOCKDOWN_CONFIDENTIALITY_MAX+1] = {
> 	[LOCKDOWN_NONE] = "none",
>+	[LOCKDOWN_MODULE_SIGNATURE] = "unsigned module loading",
> 	[LOCKDOWN_INTEGRITY_MAX] = "integrity",
> 	[LOCKDOWN_CONFIDENTIALITY_MAX] = "confidentiality",
> };
>-- 
>2.22.0.770.g0f2c4a37fd-goog
>
Matthew Garrett Aug. 1, 2019, 8:42 p.m. UTC | #2
On Thu, Aug 1, 2019 at 7:22 AM Jessica Yu <jeyu@kernel.org> wrote:
> Apologies if this was addressed in another patch in your series (I've
> only skimmed the first few), but what should happen if the kernel is
> locked down, but CONFIG_MODULE_SIG=n? Or shouldn't CONFIG_SECURITY_LOCKDOWN_LSM
> depend on CONFIG_MODULE_SIG? Otherwise I think we'll end up calling
> the empty !CONFIG_MODULE_SIG module_sig_check() stub even though
> lockdown is enabled.

Hm. Someone could certainly configure their kernel in that way. I'm
not sure that tying CONFIG_SECURITY_LOCKDOWN_LSM to CONFIG_MODULE_SIG
is the right solution, since the new LSM approach means that any other
LSM could also impose the same policy. Perhaps we should just document
this?
Jessica Yu Aug. 8, 2019, 10:01 a.m. UTC | #3
+++ Matthew Garrett [01/08/19 13:42 -0700]:
>On Thu, Aug 1, 2019 at 7:22 AM Jessica Yu <jeyu@kernel.org> wrote:
>> Apologies if this was addressed in another patch in your series (I've
>> only skimmed the first few), but what should happen if the kernel is
>> locked down, but CONFIG_MODULE_SIG=n? Or shouldn't CONFIG_SECURITY_LOCKDOWN_LSM
>> depend on CONFIG_MODULE_SIG? Otherwise I think we'll end up calling
>> the empty !CONFIG_MODULE_SIG module_sig_check() stub even though
>> lockdown is enabled.
>
>Hm. Someone could certainly configure their kernel in that way. I'm
>not sure that tying CONFIG_SECURITY_LOCKDOWN_LSM to CONFIG_MODULE_SIG
>is the right solution, since the new LSM approach means that any other
>LSM could also impose the same policy. Perhaps we should just document
>this?

Hi Matthew,

If you're confident that a hard dependency is not the right approach,
then perhaps we could add a comment in the Kconfig (You could take a
look at the comment under MODULE_SIG_ALL in init/Kconfig for an
example)? If someone is configuring the kernel on their own then it'd
be nice to let them know, otherwise having a lockdown kernel without
module signatures would defeat the purpose of lockdown no? :-)

Thank you,

Jessica
Matthew Garrett Aug. 8, 2019, 6:31 p.m. UTC | #4
On Thu, Aug 8, 2019 at 3:01 AM Jessica Yu <jeyu@kernel.org> wrote:
> If you're confident that a hard dependency is not the right approach,
> then perhaps we could add a comment in the Kconfig (You could take a
> look at the comment under MODULE_SIG_ALL in init/Kconfig for an
> example)? If someone is configuring the kernel on their own then it'd
> be nice to let them know, otherwise having a lockdown kernel without
> module signatures would defeat the purpose of lockdown no? :-)

James, what would your preference be here? Jessica is right that not
having CONFIG_MODULE_SIG enabled means lockdown probably doesn't work
as expected, but tying it to the lockdown LSM seems inappropriate when
another LSM could be providing lockdown policy and run into the same
issue. Should this just be mentioned in the CONFIG_MODULE_SIG Kconfig
help?
James Morris Aug. 8, 2019, 10:43 p.m. UTC | #5
On Thu, 8 Aug 2019, Matthew Garrett wrote:

> On Thu, Aug 8, 2019 at 3:01 AM Jessica Yu <jeyu@kernel.org> wrote:
> > If you're confident that a hard dependency is not the right approach,
> > then perhaps we could add a comment in the Kconfig (You could take a
> > look at the comment under MODULE_SIG_ALL in init/Kconfig for an
> > example)? If someone is configuring the kernel on their own then it'd
> > be nice to let them know, otherwise having a lockdown kernel without
> > module signatures would defeat the purpose of lockdown no? :-)
> 
> James, what would your preference be here? Jessica is right that not
> having CONFIG_MODULE_SIG enabled means lockdown probably doesn't work
> as expected, but tying it to the lockdown LSM seems inappropriate when
> another LSM could be providing lockdown policy and run into the same
> issue. Should this just be mentioned in the CONFIG_MODULE_SIG Kconfig
> help?

I agree and yes mention it in the help.  A respin of just this patch is 
fine.

Patch
diff mbox series

diff --git a/include/linux/security.h b/include/linux/security.h
index 54a0532ec12f..8e70063074a1 100644
--- a/include/linux/security.h
+++ b/include/linux/security.h
@@ -103,6 +103,7 @@  enum lsm_event {
  */
 enum lockdown_reason {
 	LOCKDOWN_NONE,
+	LOCKDOWN_MODULE_SIGNATURE,
 	LOCKDOWN_INTEGRITY_MAX,
 	LOCKDOWN_CONFIDENTIALITY_MAX,
 };
diff --git a/kernel/module.c b/kernel/module.c
index cd8df516666d..318209889e26 100644
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -2771,8 +2771,9 @@  static inline void kmemleak_load_module(const struct module *mod,
 #ifdef CONFIG_MODULE_SIG
 static int module_sig_check(struct load_info *info, int flags)
 {
-	int err = -ENOKEY;
+	int err = -ENODATA;
 	const unsigned long markerlen = sizeof(MODULE_SIG_STRING) - 1;
+	const char *reason;
 	const void *mod = info->hdr;
 
 	/*
@@ -2787,16 +2788,38 @@  static int module_sig_check(struct load_info *info, int flags)
 		err = mod_verify_sig(mod, info);
 	}
 
-	if (!err) {
+	switch (err) {
+	case 0:
 		info->sig_ok = true;
 		return 0;
-	}
 
-	/* Not having a signature is only an error if we're strict. */
-	if (err == -ENOKEY && !is_module_sig_enforced())
-		err = 0;
+		/* We don't permit modules to be loaded into trusted kernels
+		 * without a valid signature on them, but if we're not
+		 * enforcing, certain errors are non-fatal.
+		 */
+	case -ENODATA:
+		reason = "Loading of unsigned module";
+		goto decide;
+	case -ENOPKG:
+		reason = "Loading of module with unsupported crypto";
+		goto decide;
+	case -ENOKEY:
+		reason = "Loading of module with unavailable key";
+	decide:
+		if (is_module_sig_enforced()) {
+			pr_notice("%s is rejected\n", reason);
+			return -EKEYREJECTED;
+		}
 
-	return err;
+		return security_locked_down(LOCKDOWN_MODULE_SIGNATURE);
+
+		/* All other errors are fatal, including nomem, unparseable
+		 * signatures and signature check failures - even if signatures
+		 * aren't required.
+		 */
+	default:
+		return err;
+	}
 }
 #else /* !CONFIG_MODULE_SIG */
 static int module_sig_check(struct load_info *info, int flags)
diff --git a/security/lockdown/lockdown.c b/security/lockdown/lockdown.c
index d30c4d254b5f..2c53fd9f5c9b 100644
--- a/security/lockdown/lockdown.c
+++ b/security/lockdown/lockdown.c
@@ -18,6 +18,7 @@  static enum lockdown_reason kernel_locked_down;
 
 static char *lockdown_reasons[LOCKDOWN_CONFIDENTIALITY_MAX+1] = {
 	[LOCKDOWN_NONE] = "none",
+	[LOCKDOWN_MODULE_SIGNATURE] = "unsigned module loading",
 	[LOCKDOWN_INTEGRITY_MAX] = "integrity",
 	[LOCKDOWN_CONFIDENTIALITY_MAX] = "confidentiality",
 };