Currently, ima_appraise_measurement() ignores the EVM status when evm_verifyxattr() returns INTEGRITY_UNKNOWN. If a file has a valid security.ima xattr with type IMA_XATTR_DIGEST or IMA_XATTR_DIGEST_NG, ima_appraise_measurement() returns INTEGRITY_PASS regardless of the EVM status. The problem is that the EVM status is overwritten with the appraisal status.
This patch mitigates the issue by selecting signature verification as the only method allowed for appraisal when EVM is not initialized. Since the new behavior might break user space, it must be turned on by adding the '-evm' suffix to the value of the ima_appraise= kernel option.
Fixes: 2fe5d6def1672 ("ima: integrity appraisal extension") Signed-off-by: Roberto Sassu roberto.sassu@huawei.com Cc: stable@vger.kernel.org --- Documentation/admin-guide/kernel-parameters.txt | 3 ++- security/integrity/ima/ima_appraise.c | 8 ++++++++ 2 files changed, 10 insertions(+), 1 deletion(-)
diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt index 138f6664b2e2..d84a2e612b93 100644 --- a/Documentation/admin-guide/kernel-parameters.txt +++ b/Documentation/admin-guide/kernel-parameters.txt @@ -1585,7 +1585,8 @@ Set number of hash buckets for inode cache.
ima_appraise= [IMA] appraise integrity measurements - Format: { "off" | "enforce" | "fix" | "log" } + Format: { "off" | "enforce" | "fix" | "log" | + "enforce-evm" | "log-evm" } default: "enforce"
ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead. diff --git a/security/integrity/ima/ima_appraise.c b/security/integrity/ima/ima_appraise.c index 5fb7127bbe68..afef06e10fb9 100644 --- a/security/integrity/ima/ima_appraise.c +++ b/security/integrity/ima/ima_appraise.c @@ -18,6 +18,7 @@
#include "ima.h"
+static bool ima_appraise_req_evm __ro_after_init; static int __init default_appraise_setup(char *str) { #ifdef CONFIG_IMA_APPRAISE_BOOTPARAM @@ -28,6 +29,9 @@ static int __init default_appraise_setup(char *str) else if (strncmp(str, "fix", 3) == 0) ima_appraise = IMA_APPRAISE_FIX; #endif + if (strcmp(str, "enforce-evm") == 0 || + strcmp(str, "log-evm") == 0) + ima_appraise_req_evm = true; return 1; }
@@ -245,7 +249,11 @@ int ima_appraise_measurement(enum ima_hooks func, switch (status) { case INTEGRITY_PASS: case INTEGRITY_PASS_IMMUTABLE: + break; case INTEGRITY_UNKNOWN: + if (ima_appraise_req_evm && + xattr_value->type != EVM_IMA_XATTR_DIGSIG) + goto out; break; case INTEGRITY_NOXATTRS: /* No EVM protected xattrs. */ case INTEGRITY_NOLABEL: /* No security.evm xattr. */
On Wed, 2019-05-29 at 15:30 +0200, Roberto Sassu wrote:
Currently, ima_appraise_measurement() ignores the EVM status when evm_verifyxattr() returns INTEGRITY_UNKNOWN. If a file has a valid security.ima xattr with type IMA_XATTR_DIGEST or IMA_XATTR_DIGEST_NG, ima_appraise_measurement() returns INTEGRITY_PASS regardless of the EVM status. The problem is that the EVM status is overwritten with the
appraisal statu
Roberto, your framing of this problem is harsh and misleading. IMA and EVM are intentionally independent of each other and can be configured independently of each other. The intersection of the two is the call to evm_verifyxattr(). INTEGRITY_UNKNOWN is returned for a number of reasons - when EVM is not configured, the EVM hmac key has not yet been loaded, the protected security attribute is unknown, or the file is not in policy.
This patch does not differentiate between any of the above cases, requiring mutable files to always be protected by EVM, when specified as an "ima_appraise=" option on the boot command line.
IMA could be extended to require EVM on a per IMA policy rule basis. Instead of framing allowing IMA file hashes without EVM as a bug that has existed from the very beginning, now that IMA/EVM have matured and is being used, you could frame it as extending IMA or hardening.
This patch mitigates the issue by selecting signature verification as the only method allowed for appraisal when EVM is not initialized. Since the new behavior might break user space, it must be turned on by adding the '-evm' suffix to the value of the ima_appraise= kernel option.
Fixes: 2fe5d6def1672 ("ima: integrity appraisal extension") Signed-off-by: Roberto Sassu roberto.sassu@huawei.com Cc: stable@vger.kernel.org
Documentation/admin-guide/kernel-parameters.txt | 3 ++- security/integrity/ima/ima_appraise.c | 8 ++++++++ 2 files changed, 10 insertions(+), 1 deletion(-)
diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt index 138f6664b2e2..d84a2e612b93 100644 --- a/Documentation/admin-guide/kernel-parameters.txt +++ b/Documentation/admin-guide/kernel-parameters.txt @@ -1585,7 +1585,8 @@ Set number of hash buckets for inode cache. ima_appraise= [IMA] appraise integrity measurements
Format: { "off" | "enforce" | "fix" | "log" }
Format: { "off" | "enforce" | "fix" | "log" |"enforce-evm" | "log-evm" }
Is it necessary to define both "enforce-evm" and "log-evm"? Perhaps defining "require-evm" is sufficient.
Mimi
default: "enforce"ima_appraise_tcb [IMA] Deprecated. Use ima_policy= instead. diff --git a/security/integrity/ima/ima_appraise.c b/security/integrity/ima/ima_appraise.c index 5fb7127bbe68..afef06e10fb9 100644 --- a/security/integrity/ima/ima_appraise.c +++ b/security/integrity/ima/ima_appraise.c @@ -18,6 +18,7 @@ #include "ima.h" +static bool ima_appraise_req_evm __ro_after_init; static int __init default_appraise_setup(char *str) { #ifdef CONFIG_IMA_APPRAISE_BOOTPARAM @@ -28,6 +29,9 @@ static int __init default_appraise_setup(char *str) else if (strncmp(str, "fix", 3) == 0) ima_appraise = IMA_APPRAISE_FIX; #endif
- if (strcmp(str, "enforce-evm") == 0 ||
strcmp(str, "log-evm") == 0)ima_appraise_req_evm = true;} @@ -245,7 +249,11 @@ int ima_appraise_measurement(enum ima_hooks func, switch (status) { case INTEGRITY_PASS: case INTEGRITY_PASS_IMMUTABLE:
break;if (ima_appraise_req_evm &&xattr_value->type != EVM_IMA_XATTR_DIGSIG)goto out;
On 5/30/2019 2:00 PM, Mimi Zohar wrote:
On Wed, 2019-05-29 at 15:30 +0200, Roberto Sassu wrote:
Currently, ima_appraise_measurement() ignores the EVM status when evm_verifyxattr() returns INTEGRITY_UNKNOWN. If a file has a valid security.ima xattr with type IMA_XATTR_DIGEST or IMA_XATTR_DIGEST_NG, ima_appraise_measurement() returns INTEGRITY_PASS regardless of the EVM status. The problem is that the EVM status is overwritten with the
appraisal statu
Roberto, your framing of this problem is harsh and misleading. IMA and EVM are intentionally independent of each other and can be configured independently of each other. The intersection of the two is the call to evm_verifyxattr(). INTEGRITY_UNKNOWN is returned for a number of reasons - when EVM is not configured, the EVM hmac key has not yet been loaded, the protected security attribute is unknown, or the file is not in policy.
This patch does not differentiate between any of the above cases, requiring mutable files to always be protected by EVM, when specified as an "ima_appraise=" option on the boot command line.
IMA could be extended to require EVM on a per IMA policy rule basis. Instead of framing allowing IMA file hashes without EVM as a bug that has existed from the very beginning, now that IMA/EVM have matured and is being used, you could frame it as extending IMA or hardening.
I'm seeing it from the perspective of an administrator that manages an already hardened system, and expects that the system only grants access to files with a valid signature/HMAC. That system would not enforce this behavior if EVM keys are removed and the digest in security.ima is set to the actual file digest.
Framing it as a bug rather than an extension would in my opinion help to convince people about the necessity to switch to the safe mode, if their system is already hardened.
This patch mitigates the issue by selecting signature verification as the only method allowed for appraisal when EVM is not initialized. Since the new behavior might break user space, it must be turned on by adding the '-evm' suffix to the value of the ima_appraise= kernel option.
Fixes: 2fe5d6def1672 ("ima: integrity appraisal extension") Signed-off-by: Roberto Sassu roberto.sassu@huawei.com Cc: stable@vger.kernel.org
Documentation/admin-guide/kernel-parameters.txt | 3 ++- security/integrity/ima/ima_appraise.c | 8 ++++++++ 2 files changed, 10 insertions(+), 1 deletion(-)
diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt index 138f6664b2e2..d84a2e612b93 100644 --- a/Documentation/admin-guide/kernel-parameters.txt +++ b/Documentation/admin-guide/kernel-parameters.txt @@ -1585,7 +1585,8 @@ Set number of hash buckets for inode cache. ima_appraise= [IMA] appraise integrity measurements
Format: { "off" | "enforce" | "fix" | "log" }
Format: { "off" | "enforce" | "fix" | "log" |"enforce-evm" | "log-evm" }Is it necessary to define both "enforce-evm" and "log-evm"? Perhaps defining "require-evm" is sufficient.
ima_appraise= accepts as values modes of operation. I consider the -evm suffix as a modifier of already defined modes.
Roberto
On Mon, 2019-06-03 at 11:25 +0200, Roberto Sassu wrote:
On 5/30/2019 2:00 PM, Mimi Zohar wrote:
On Wed, 2019-05-29 at 15:30 +0200, Roberto Sassu wrote:
Currently, ima_appraise_measurement() ignores the EVM status when evm_verifyxattr() returns INTEGRITY_UNKNOWN. If a file has a valid security.ima xattr with type IMA_XATTR_DIGEST or IMA_XATTR_DIGEST_NG, ima_appraise_measurement() returns INTEGRITY_PASS regardless of the EVM status. The problem is that the EVM status is overwritten with the
appraisal status
Roberto, your framing of this problem is harsh and misleading. IMA and EVM are intentionally independent of each other and can be configured independently of each other. The intersection of the two is the call to evm_verifyxattr(). INTEGRITY_UNKNOWN is returned for a number of reasons - when EVM is not configured, the EVM hmac key has not yet been loaded, the protected security attribute is unknown, or the file is not in policy.
This patch does not differentiate between any of the above cases, requiring mutable files to always be protected by EVM, when specified as an "ima_appraise=" option on the boot command line.
IMA could be extended to require EVM on a per IMA policy rule basis. Instead of framing allowing IMA file hashes without EVM as a bug that has existed from the very beginning, now that IMA/EVM have matured and is being used, you could frame it as extending IMA or hardening.
I'm seeing it from the perspective of an administrator that manages an already hardened system, and expects that the system only grants access to files with a valid signature/HMAC. That system would not enforce this behavior if EVM keys are removed and the digest in security.ima is set to the actual file digest.
Framing it as a bug rather than an extension would in my opinion help to convince people about the necessity to switch to the safe mode, if their system is already hardened.
I don't disagree with you that people should be using EVM to protect IMA hashes. If you claim this is a bug in the design from the very beginning, then there needs some explanation as to why it was upstreamed as it was. My review of this patch provided that context/background.
Mimi
This patch mitigates the issue by selecting signature verification as the only method allowed for appraisal when EVM is not initialized. Since the new behavior might break user space, it must be turned on by adding the '-evm' suffix to the value of the ima_appraise= kernel option.
Fixes: 2fe5d6def1672 ("ima: integrity appraisal extension") Signed-off-by: Roberto Sassu roberto.sassu@huawei.com Cc: stable@vger.kernel.org
On Mon, 2019-06-03 at 11:25 +0200, Roberto Sassu wrote:
On 5/30/2019 2:00 PM, Mimi Zohar wrote:
On Wed, 2019-05-29 at 15:30 +0200, Roberto Sassu wrote:
Currently, ima_appraise_measurement() ignores the EVM status when evm_verifyxattr() returns INTEGRITY_UNKNOWN. If a file has a valid security.ima xattr with type IMA_XATTR_DIGEST or IMA_XATTR_DIGEST_NG, ima_appraise_measurement() returns INTEGRITY_PASS regardless of the EVM status. The problem is that the EVM status is overwritten with the appraisal statu
Roberto, your framing of this problem is harsh and misleading. IMA and EVM are intentionally independent of each other and can be configured independently of each other. The intersection of the two is the call to evm_verifyxattr(). INTEGRITY_UNKNOWN is returned for a number of reasons - when EVM is not configured, the EVM hmac key has not yet been loaded, the protected security attribute is unknown, or the file is not in policy.
This patch does not differentiate between any of the above cases, requiring mutable files to always be protected by EVM, when specified as an "ima_appraise=" option on the boot command line.
IMA could be extended to require EVM on a per IMA policy rule basis. Instead of framing allowing IMA file hashes without EVM as a bug that has existed from the very beginning, now that IMA/EVM have matured and is being used, you could frame it as extending IMA or hardening.
I'm seeing it from the perspective of an administrator that manages an already hardened system, and expects that the system only grants access to files with a valid signature/HMAC. That system would not enforce this behavior if EVM keys are removed and the digest in security.ima is set to the actual file digest.
Framing it as a bug rather than an extension would in my opinion help to convince people about the necessity to switch to the safe mode, if their system is already hardened.
I have a use case for IMA where I use it to enforce immutability of containers. In this use case, the cluster admin places hashes on executables as the image is unpacked so that if an executable file is changed, IMA will cause an execution failure. For this use case, I don't care about the EVM, in fact we don't use it, because the only object is to fail execution if a binary is mutated.
So I can see your use case requires IMA+EVM, but requiring it would cause more complexity for my use case.
James
On Jun 3, 2019, at 9:43 AM, James Bottomley James.Bottomley@HansenPartnership.com wrote:
On Mon, 2019-06-03 at 11:25 +0200, Roberto Sassu wrote:
On 5/30/2019 2:00 PM, Mimi Zohar wrote:
On Wed, 2019-05-29 at 15:30 +0200, Roberto Sassu wrote:
Currently, ima_appraise_measurement() ignores the EVM status when evm_verifyxattr() returns INTEGRITY_UNKNOWN. If a file has a valid security.ima xattr with type IMA_XATTR_DIGEST or IMA_XATTR_DIGEST_NG, ima_appraise_measurement() returns INTEGRITY_PASS regardless of the EVM status. The problem is that the EVM status is overwritten with the appraisal statu
Roberto, your framing of this problem is harsh and misleading. IMA and EVM are intentionally independent of each other and can be configured independently of each other. The intersection of the two is the call to evm_verifyxattr(). INTEGRITY_UNKNOWN is returned for a number of reasons - when EVM is not configured, the EVM hmac key has not yet been loaded, the protected security attribute is unknown, or the file is not in policy.
This patch does not differentiate between any of the above cases, requiring mutable files to always be protected by EVM, when specified as an "ima_appraise=" option on the boot command line.
IMA could be extended to require EVM on a per IMA policy rule basis. Instead of framing allowing IMA file hashes without EVM as a bug that has existed from the very beginning, now that IMA/EVM have matured and is being used, you could frame it as extending IMA or hardening.
I'm seeing it from the perspective of an administrator that manages an already hardened system, and expects that the system only grants access to files with a valid signature/HMAC. That system would not enforce this behavior if EVM keys are removed and the digest in security.ima is set to the actual file digest.
Framing it as a bug rather than an extension would in my opinion help to convince people about the necessity to switch to the safe mode, if their system is already hardened.
I have a use case for IMA where I use it to enforce immutability of containers. In this use case, the cluster admin places hashes on executables as the image is unpacked so that if an executable file is changed, IMA will cause an execution failure. For this use case, I don't care about the EVM, in fact we don't use it, because the only object is to fail execution if a binary is mutated.
So I can see your use case requires IMA+EVM, but requiring it would cause more complexity for my use case.
... and would not work at all for the current NFS IMA implementation.
-- Chuck Lever
On 6/3/2019 3:43 PM, James Bottomley wrote:
On Mon, 2019-06-03 at 11:25 +0200, Roberto Sassu wrote:
On 5/30/2019 2:00 PM, Mimi Zohar wrote:
On Wed, 2019-05-29 at 15:30 +0200, Roberto Sassu wrote:
Currently, ima_appraise_measurement() ignores the EVM status when evm_verifyxattr() returns INTEGRITY_UNKNOWN. If a file has a valid security.ima xattr with type IMA_XATTR_DIGEST or IMA_XATTR_DIGEST_NG, ima_appraise_measurement() returns INTEGRITY_PASS regardless of the EVM status. The problem is that the EVM status is overwritten with the appraisal statu
Roberto, your framing of this problem is harsh and misleading. IMA and EVM are intentionally independent of each other and can be configured independently of each other. The intersection of the two is the call to evm_verifyxattr(). INTEGRITY_UNKNOWN is returned for a number of reasons - when EVM is not configured, the EVM hmac key has not yet been loaded, the protected security attribute is unknown, or the file is not in policy.
This patch does not differentiate between any of the above cases, requiring mutable files to always be protected by EVM, when specified as an "ima_appraise=" option on the boot command line.
IMA could be extended to require EVM on a per IMA policy rule basis. Instead of framing allowing IMA file hashes without EVM as a bug that has existed from the very beginning, now that IMA/EVM have matured and is being used, you could frame it as extending IMA or hardening.
I'm seeing it from the perspective of an administrator that manages an already hardened system, and expects that the system only grants access to files with a valid signature/HMAC. That system would not enforce this behavior if EVM keys are removed and the digest in security.ima is set to the actual file digest.
Framing it as a bug rather than an extension would in my opinion help to convince people about the necessity to switch to the safe mode, if their system is already hardened.
I have a use case for IMA where I use it to enforce immutability of containers. In this use case, the cluster admin places hashes on executables as the image is unpacked so that if an executable file is changed, IMA will cause an execution failure. For this use case, I don't care about the EVM, in fact we don't use it, because the only object is to fail execution if a binary is mutated.
How would you prevent root in the container from updating security.ima?
Roberto
On Mon, 2019-06-03 at 16:29 +0200, Roberto Sassu wrote:
On 6/3/2019 3:43 PM, James Bottomley wrote:
On Mon, 2019-06-03 at 11:25 +0200, Roberto Sassu wrote:
On 5/30/2019 2:00 PM, Mimi Zohar wrote:
On Wed, 2019-05-29 at 15:30 +0200, Roberto Sassu wrote:
Currently, ima_appraise_measurement() ignores the EVM status when evm_verifyxattr() returns INTEGRITY_UNKNOWN. If a file has a valid security.ima xattr with type IMA_XATTR_DIGEST or IMA_XATTR_DIGEST_NG, ima_appraise_measurement() returns INTEGRITY_PASS regardless of the EVM status. The problem is that the EVM status is overwritten with the appraisal statu
Roberto, your framing of this problem is harsh and misleading. IMA and EVM are intentionally independent of each other and can be configured independently of each other. The intersection of the two is the call to evm_verifyxattr(). INTEGRITY_UNKNOWN is returned for a number of reasons - when EVM is not configured, the EVM hmac key has not yet been loaded, the protected security attribute is unknown, or the file is not in policy.
This patch does not differentiate between any of the above cases, requiring mutable files to always be protected by EVM, when specified as an "ima_appraise=" option on the boot command line.
IMA could be extended to require EVM on a per IMA policy rule basis. Instead of framing allowing IMA file hashes without EVM as a bug that has existed from the very beginning, now that IMA/EVM have matured and is being used, you could frame it as extending IMA or hardening.
I'm seeing it from the perspective of an administrator that manages an already hardened system, and expects that the system only grants access to files with a valid signature/HMAC. That system would not enforce this behavior if EVM keys are removed and the digest in security.ima is set to the actual file digest.
Framing it as a bug rather than an extension would in my opinion help to convince people about the necessity to switch to the safe mode, if their system is already hardened.
I have a use case for IMA where I use it to enforce immutability of containers. In this use case, the cluster admin places hashes on executables as the image is unpacked so that if an executable file is changed, IMA will cause an execution failure. For this use case, I don't care about the EVM, in fact we don't use it, because the only object is to fail execution if a binary is mutated.
How would you prevent root in the container from updating security.ima?
We don't. We only guarantee immutability for unprivileged containers, so root can't be inside.
James
On 6/3/2019 4:31 PM, James Bottomley wrote:
On Mon, 2019-06-03 at 16:29 +0200, Roberto Sassu wrote:
On 6/3/2019 3:43 PM, James Bottomley wrote:
On Mon, 2019-06-03 at 11:25 +0200, Roberto Sassu wrote:
On 5/30/2019 2:00 PM, Mimi Zohar wrote:
On Wed, 2019-05-29 at 15:30 +0200, Roberto Sassu wrote:
Currently, ima_appraise_measurement() ignores the EVM status when evm_verifyxattr() returns INTEGRITY_UNKNOWN. If a file has a valid security.ima xattr with type IMA_XATTR_DIGEST or IMA_XATTR_DIGEST_NG, ima_appraise_measurement() returns INTEGRITY_PASS regardless of the EVM status. The problem is that the EVM status is overwritten with the appraisal statu
Roberto, your framing of this problem is harsh and misleading. IMA and EVM are intentionally independent of each other and can be configured independently of each other. The intersection of the two is the call to evm_verifyxattr(). INTEGRITY_UNKNOWN is returned for a number of reasons - when EVM is not configured, the EVM hmac key has not yet been loaded, the protected security attribute is unknown, or the file is not in policy.
This patch does not differentiate between any of the above cases, requiring mutable files to always be protected by EVM, when specified as an "ima_appraise=" option on the boot command line.
IMA could be extended to require EVM on a per IMA policy rule basis. Instead of framing allowing IMA file hashes without EVM as a bug that has existed from the very beginning, now that IMA/EVM have matured and is being used, you could frame it as extending IMA or hardening.
I'm seeing it from the perspective of an administrator that manages an already hardened system, and expects that the system only grants access to files with a valid signature/HMAC. That system would not enforce this behavior if EVM keys are removed and the digest in security.ima is set to the actual file digest.
Framing it as a bug rather than an extension would in my opinion help to convince people about the necessity to switch to the safe mode, if their system is already hardened.
I have a use case for IMA where I use it to enforce immutability of containers. In this use case, the cluster admin places hashes on executables as the image is unpacked so that if an executable file is changed, IMA will cause an execution failure. For this use case, I don't care about the EVM, in fact we don't use it, because the only object is to fail execution if a binary is mutated.
How would you prevent root in the container from updating security.ima?
We don't. We only guarantee immutability for unprivileged containers, so root can't be inside.
Ok.
Regarding the new behavior, this must be explicitly enabled by adding ima_appraise=enforce-evm or log-evm to the kernel command line. Otherwise, the current behavior is preserved with this patch. Would this be ok?
Roberto
On Mon, 2019-06-03 at 16:44 +0200, Roberto Sassu wrote:
On 6/3/2019 4:31 PM, James Bottomley wrote:
On Mon, 2019-06-03 at 16:29 +0200, Roberto Sassu wrote:
[...]
How would you prevent root in the container from updating security.ima?
We don't. We only guarantee immutability for unprivileged containers, so root can't be inside.
Ok.
Regarding the new behavior, this must be explicitly enabled by adding ima_appraise=enforce-evm or log-evm to the kernel command line. Otherwise, the current behavior is preserved with this patch. Would this be ok?
Sure, as long as it's an opt-in flag, meaning the behaviour of my kernels on physical cloud systems doesn't change as I upgrade them, I'm fine with that.
James
linux-stable-mirror@lists.linaro.org
-
Chuck Lever -
James Bottomley -
Mimi Zohar -
Roberto Sassu