Skip to contentSkip to conversation

Add support for ECDSA-signed kernel modules

PATCH Superseded Stefan Berger 3 participants 12 messages Apr 8, 2021 – Apr 21, 2021

Subsystems (6)

v2 of 6 Superseded Latest is v6 →

lore
Share & tools
lore
#0 lore raw

v2 cover Add support for ECDSA-signed kernel modules

This series adds support for ECDSA-signed kernel modules. It also attempts to address a kbuild issue where a developer created an ECDSA key for signing kernel modules and then builds an older version of the kernel, when bisecting the kernel for example, that does not support ECDSA keys.

The first patch addresses the kbuild issue of needing to delete that ECDSA key if it is in certs/signing_key.pem and trigger the creation of an RSA key. However, for this to work this patch would have to be backported to previous versions of the kernel but would also only work for the developer if he/she used a stable version of the kernel to which this patch was applied. So whether this patch actually achieves the wanted effect is not always guaranteed.

The 2nd patch adds the support for the ECSDA-signed kernel modules.

This patch depends on the ECDSA support series currently queued here:
https://git.kernel.org/pub/scm/linux/kernel/git/herbert/cryptodev-2.6.git/log/?h=ecc

Stefan

v2:
- Adjustment to ECDSA key detector string in 2/2
- Rephrased cover letter and patch descriptions with Mimi

Stefan Berger (2):
certs: Trigger creation of RSA module signing key if it's not an RSA
key
certs: Add support for using elliptic curve keys for signing modules

 certs/Kconfig                         | 25 +++++++++++++++++++++++++
 certs/Makefile                        | 14 ++++++++++++++
 crypto/asymmetric_keys/pkcs7_parser.c |  4 ++++
 3 files changed, 43 insertions(+)
Signature · 1 line
2.29.2
#1 lore raw

PATCH v2 1/2 certs: Trigger creation of RSA module signing key if it's not an RSA key

Address a kbuild issue where a developer created an ECDSA key for signing kernel modules and then builds an older version of the kernel, when bi- secting the kernel for example, that does not support ECDSA keys.

Trigger the creation of an RSA module signing key if it is not an RSA key.

Fixes cfc411e7fff3 ("Move certificate handling to its own directory")
---
 certs/Makefile | 5 +++++
 1 file changed, 5 insertions(+)
certs/Makefile +5 -0
index e3185c57fbd8..f64bc89ccbf1 100644
@@ -59,6 +59,11 @@ silent_redirect_openssl = 2>/dev/null
 # external private key, because 'make randconfig' might enable such a
 # boolean option and we unfortunately can't make it depend on !RANDCONFIG.
 ifeq ($(CONFIG_MODULE_SIG_KEY),"certs/signing_key.pem")
+
+X509TEXT=$(shell openssl x509 -in $(CONFIG_MODULE_SIG_KEY) -text)
+
+$(if $(findstring rsaEncryption,$(X509TEXT)),,$(shell rm -f $(CONFIG_MODULE_SIG_KEY)))
+
 $(obj)/signing_key.pem: $(obj)/x509.genkey
 	@$(kecho) "###"
 	@$(kecho) "### Now generating an X.509 key pair to be used for signing modules."
Signature · 1 line
2.29.2
maintainer Reviewed-by on v2 1/2
#2 lore raw

Re: [PATCH v2 1/2] certs: Trigger creation of RSA module signing key if it's not an RSA key

Quoted 8 lines · Stefan Berger

Address a kbuild issue where a developer created an ECDSA key for signing kernel modules and then builds an older version of the kernel, when bi- secting the kernel for example, that does not support ECDSA keys.

Trigger the creation of an RSA module signing key if it is not an RSA key.

Fixes cfc411e7fff3 ("Move certificate handling to its own directory")

Thanks, Stefan.

#3 lore raw

Re: [PATCH v2 1/2] certs: Trigger creation of RSA module signing key if it's not an RSA key

Quoted 5 lines · Mimi Zohar
Quoted 4 lines · Stefan Berger

Address a kbuild issue where a developer created an ECDSA key for signing kernel modules and then builds an older version of the kernel, when bi- secting the kernel for example, that does not support ECDSA keys.

Quoted 8 lines
Quoted 4 lines

Trigger the creation of an RSA module signing key if it is not an RSA key.

Fixes cfc411e7fff3 ("Move certificate handling to its own directory")

Thanks, Stefan.

Via which tree will this go upstream? keyrings?

   Stefan

maintainer
#4 lore raw

Re: [PATCH v2 1/2] certs: Trigger creation of RSA module signing key if it's not an RSA key

Quoted 8 lines · Stefan Berger
Quoted 7 lines · Mimi Zohar
Quoted 6 lines · Stefan Berger

Address a kbuild issue where a developer created an ECDSA key for signing kernel modules and then builds an older version of the kernel, when bi- secting the kernel for example, that does not support ECDSA keys.

Trigger the creation of an RSA module signing key if it is not an RSA key.

Quoted 8 lines
Quoted 6 lines
Quoted 2 lines
Fixes cfc411e7fff3 ("Move certificate handling to its own directory")

Thanks, Stefan.

Via which tree will this go upstream? keyrings?

This patch set originally had a dependency on Nayna's v1 & v2 "ima:
kernel build support for loading the kernel module signing key" patch set and on Herbert's "ecc" branch. With v3, the dependency on Nayna's patch set is gone.

Jarkko, David, Herbert did you want to pick up this patch set or would you prefer that I did? Either way is fine.

thanks,

Mimi

#5 lore raw

PATCH v2 2/2 certs: Add support for using elliptic curve keys for signing modules

In reply to Stefan Berger

Add support for using elliptic curve keys for signing modules. It uses a NIST P384 (secp384r1) key if the user chooses an elliptic curve key and will have ECDSA support built into the kernel.

Note: A developer choosing an ECDSA key for signing modules should still delete the signing key (rm certs/signing_key.*) when building an older version of a kernel that only supports RSA keys. Unless kbuild automati- cally detects and generates a new kernel module key, ECDSA-signed kernel modules will fail signature verification.

---
v2:
- check for ECDSA key by id-ecPublicKey from output line 'Public Key Algorithm: id-ecPublicKey'.

---
 certs/Kconfig                         | 25 +++++++++++++++++++++++++
 certs/Makefile                        |  9 +++++++++
 crypto/asymmetric_keys/pkcs7_parser.c |  4 ++++
 3 files changed, 38 insertions(+)
certs/Kconfig +25 -0
index 48675ad319db..6f8337874ae0 100644
@@ -15,6 +15,31 @@ config MODULE_SIG_KEY
          then the kernel will automatically generate the private key and
          certificate as described in Documentation/admin-guide/module-signing.rst
 
+choice
+	prompt "Type of module signing key to be generated"
+	default MODULE_SIG_KEY_TYPE_RSA
+	help
+	 The type of module signing key type to generate. This option
+	 does not apply if a #PKCS11 URI is used.
+
+config MODULE_SIG_KEY_TYPE_RSA
+	bool "RSA"
+	depends on MODULE_SIG || IMA_APPRAISE_MODSIG
+	help
+	 Use an RSA key for module signing.
+
+config MODULE_SIG_KEY_TYPE_ECDSA
+	bool "ECDSA"
+	select CRYPTO_ECDSA
+	depends on MODULE_SIG || IMA_APPRAISE_MODSIG
+	help
+	 Use an elliptic curve key (NIST P384) for module signing.
+
+	 Note: Remove all ECDSA signing keys, e.g. certs/signing_key.pem,
+	 when falling back to building Linux 5.11 and older kernels.
+
+endchoice
+
 config SYSTEM_TRUSTED_KEYRING
 	bool "Provide system-wide ring of trusted keys"
 	depends on KEYS
certs/Makefile +9 -0
index f64bc89ccbf1..c2fabc288550 100644
@@ -62,7 +62,15 @@ ifeq ($(CONFIG_MODULE_SIG_KEY),"certs/signing_key.pem")
 
 X509TEXT=$(shell openssl x509 -in $(CONFIG_MODULE_SIG_KEY) -text)
 
+# Support user changing key type
+ifdef CONFIG_MODULE_SIG_KEY_TYPE_ECDSA
+keytype_openssl = -newkey ec -pkeyopt ec_paramgen_curve:secp384r1
+$(if $(findstring id-ecPublicKey,$(X509TEXT)),,$(shell rm -f $(CONFIG_MODULE_SIG_KEY)))
+endif
+
+ifdef CONFIG_MODULE_SIG_KEY_TYPE_RSA
 $(if $(findstring rsaEncryption,$(X509TEXT)),,$(shell rm -f $(CONFIG_MODULE_SIG_KEY)))
+endif
 
 $(obj)/signing_key.pem: $(obj)/x509.genkey
 	@$(kecho) "###"
@@ -77,6 +85,7 @@ $(obj)/signing_key.pem: $(obj)/x509.genkey
 		-batch -x509 -config $(obj)/x509.genkey \
 		-outform PEM -out $(obj)/signing_key.pem \
 		-keyout $(obj)/signing_key.pem \
+		$(keytype_openssl) \
 		$($(quiet)redirect_openssl)
 	@$(kecho) "###"
 	@$(kecho) "### Key pair generated."
crypto/asymmetric_keys/pkcs7_parser.c +4 -0
index 967329e0a07b..2546ec6a0505 100644
@@ -269,6 +269,10 @@ int pkcs7_sig_note_pkey_algo(void *context, size_t hdrlen,
 		ctx->sinfo->sig->pkey_algo = "rsa";
 		ctx->sinfo->sig->encoding = "pkcs1";
 		break;
+	case OID_id_ecdsa_with_sha256:
+		ctx->sinfo->sig->pkey_algo = "ecdsa";
+		ctx->sinfo->sig->encoding = "x962";
+		break;
 	default:
 		printk("Unsupported pkey algo: %u\n", ctx->last_oid);
 		return -ENOPKG;
Signature · 1 line
2.29.2
maintainer Reviewed-by on v2 2/2
#6 lore raw

Re: [PATCH v2 2/2] certs: Add support for using elliptic curve keys for signing modules

Quoted 11 lines · Stefan Berger

Add support for using elliptic curve keys for signing modules. It uses a NIST P384 (secp384r1) key if the user chooses an elliptic curve key and will have ECDSA support built into the kernel.

Note: A developer choosing an ECDSA key for signing modules should still delete the signing key (rm certs/signing_key.*) when building an older version of a kernel that only supports RSA keys. Unless kbuild automati- cally detects and generates a new kernel module key, ECDSA-signed kernel modules will fail signature verification.

Thanks, Stefan.

#7 lore raw

Re: [PATCH v2 2/2] certs: Add support for using elliptic curve keys for signing modules

In reply to Stefan Berger

+++ Stefan Berger [08/04/21 11:24 -0400]:

Quoted 91 lines

Add support for using elliptic curve keys for signing modules. It uses a NIST P384 (secp384r1) key if the user chooses an elliptic curve key and will have ECDSA support built into the kernel.

Note: A developer choosing an ECDSA key for signing modules should still delete the signing key (rm certs/signing_key.*) when building an older version of a kernel that only supports RSA keys. Unless kbuild automati- cally detects and generates a new kernel module key, ECDSA-signed kernel modules will fail signature verification.

---
v2:
- check for ECDSA key by id-ecPublicKey from output line 'Public Key Algorithm: id-ecPublicKey'.
---

certs/Kconfig                         | 25 +++++++++++++++++++++++++
certs/Makefile                        |  9 +++++++++
crypto/asymmetric_keys/pkcs7_parser.c |  4 ++++
3 files changed, 38 insertions(+)
certs/Kconfig +25 -0
index 48675ad319db..6f8337874ae0 100644
@@ -15,6 +15,31 @@ config MODULE_SIG_KEY
         then the kernel will automatically generate the private key and
         certificate as described in Documentation/admin-guide/module-signing.rst
 
+choice
+	prompt "Type of module signing key to be generated"
+	default MODULE_SIG_KEY_TYPE_RSA
+	help
+	 The type of module signing key type to generate. This option
+	 does not apply if a #PKCS11 URI is used.
+
+config MODULE_SIG_KEY_TYPE_RSA
+	bool "RSA"
+	depends on MODULE_SIG || IMA_APPRAISE_MODSIG
+	help
+	 Use an RSA key for module signing.
+
+config MODULE_SIG_KEY_TYPE_ECDSA
+	bool "ECDSA"
+	select CRYPTO_ECDSA
+	depends on MODULE_SIG || IMA_APPRAISE_MODSIG
+	help
+	 Use an elliptic curve key (NIST P384) for module signing.
+
+	 Note: Remove all ECDSA signing keys, e.g. certs/signing_key.pem,
+	 when falling back to building Linux 5.11 and older kernels.
+
+endchoice
+

config SYSTEM_TRUSTED_KEYRING
bool "Provide system-wide ring of trusted keys" depends on KEYS

certs/Makefile +8 -0
index f64bc89ccbf1..c2fabc288550 100644
@@ -62,7 +62,15 @@ ifeq ($(CONFIG_MODULE_SIG_KEY),"certs/signing_key.pem")
 
 X509TEXT=$(shell openssl x509 -in $(CONFIG_MODULE_SIG_KEY) -text)
 
+# Support user changing key type
+ifdef CONFIG_MODULE_SIG_KEY_TYPE_ECDSA
+keytype_openssl = -newkey ec -pkeyopt ec_paramgen_curve:secp384r1
+$(if $(findstring id-ecPublicKey,$(X509TEXT)),,$(shell rm -f $(CONFIG_MODULE_SIG_KEY)))
+endif
+
+ifdef CONFIG_MODULE_SIG_KEY_TYPE_RSA
 $(if $(findstring rsaEncryption,$(X509TEXT)),,$(shell rm -f $(CONFIG_MODULE_SIG_KEY)))
+endif
 

$(obj)/signing_key.pem: $(obj)/x509.genkey

	@$(kecho) "###"
certs/Makefile +1 -0
@@ -77,6 +85,7 @@ $(obj)/signing_key.pem: $(obj)/x509.genkey
 		-batch -x509 -config $(obj)/x509.genkey \
 		-outform PEM -out $(obj)/signing_key.pem \
 		-keyout $(obj)/signing_key.pem \
+		$(keytype_openssl) \
 		$($(quiet)redirect_openssl)
 	@$(kecho) "###"
 	@$(kecho) "### Key pair generated."
crypto/asymmetric_keys/pkcs7_parser.c +0 -0
index 967329e0a07b..2546ec6a0505 100644

crypto/asymmetric_keys/pkcs7_parser.c int pkcs7_sig_note_pkey_algo(void *context, size_t hdrlen, · line 275

Quoted 7 lines
crypto/asymmetric_keys/pkcs7_parser.c +4 -0
@@ -269,6 +269,10 @@ int pkcs7_sig_note_pkey_algo(void *context, size_t hdrlen,
 		ctx->sinfo->sig->pkey_algo = "rsa";
 		ctx->sinfo->sig->encoding = "pkcs1";
 		break;
+	case OID_id_ecdsa_with_sha256:
+		ctx->sinfo->sig->pkey_algo = "ecdsa";
+		ctx->sinfo->sig->encoding = "x962";
+		break;

Hi Stefan,

Does CONFIG_MODULE_SIG_KEY_TYPE_ECDSA have a dependency on MODULE_SIG_SHA256? By default, MODULE_SIG_SHA1 is selected when CONFIG_MODULE_SIG is enabled. I was doing some quick testing and found that when I enabled MODULE_SIG_KEY_TYPE_ECDSA I get a "Unsupported pkey algo: 5" error on module load, which goes away after fixing my config and selecting MODULE_SIG_SHA256.

Thanks,

Jessica

#8 lore raw

Re: [PATCH v2 2/2] certs: Add support for using elliptic curve keys for signing modules

Quoted 18 lines · Jessica Yu

+++ Stefan Berger [08/04/21 11:24 -0400]:

Quoted 6 lines
crypto/asymmetric_keys/pkcs7_parser.c +0 -0
index 967329e0a07b..2546ec6a0505 100644

crypto/asymmetric_keys/pkcs7_parser.c int pkcs7_sig_note_pkey_algo(void *context, · line 276

Quoted 8 lines
crypto/asymmetric_keys/pkcs7_parser.c +4 -0
@@ -269,6 +269,10 @@ int pkcs7_sig_note_pkey_algo(void *context, 
 size_t hdrlen,
         ctx->sinfo->sig->pkey_algo = "rsa";
         ctx->sinfo->sig->encoding = "pkcs1";
         break;
+    case OID_id_ecdsa_with_sha256:
+        ctx->sinfo->sig->pkey_algo = "ecdsa";
+        ctx->sinfo->sig->encoding = "x962";
+        break;

Hi Stefan,

Quoted 1 line

Does CONFIG_MODULE_SIG_KEY_TYPE_ECDSA have a dependency on MODULE_SIG_SHA256?

You are right, per the code above it does have a dependency on SHA256. ECDSA is using NIST p384 (secp384r1) for signing and per my tests it can be paired with all the sha hashes once the code above is extended. Now when it comes to module signing, should we pair it with a particular hash? I am not currently aware of a guidance document on this but sha256 and sha384 seem to be good choices these days, so maybe selecting ECDSA module signing should have a 'depends on' on these?

  Stefan

#9 lore raw

Re: [PATCH v2 2/2] certs: Add support for using elliptic curve keys for signing modules

+++ Stefan Berger [20/04/21 17:02 -0400]:

Quoted 20 lines · Jessica Yu
Quoted 18 lines

+++ Stefan Berger [08/04/21 11:24 -0400]:

Quoted 6 lines
crypto/asymmetric_keys/pkcs7_parser.c +0 -0
index 967329e0a07b..2546ec6a0505 100644

crypto/asymmetric_keys/pkcs7_parser.c int pkcs7_sig_note_pkey_algo(void *context, · line 276

Quoted 8 lines
crypto/asymmetric_keys/pkcs7_parser.c +4 -0
@@ -269,6 +269,10 @@ int pkcs7_sig_note_pkey_algo(void *context, 
 size_t hdrlen,
         ctx->sinfo->sig->pkey_algo = "rsa";
         ctx->sinfo->sig->encoding = "pkcs1";
         break;
+    case OID_id_ecdsa_with_sha256:
+        ctx->sinfo->sig->pkey_algo = "ecdsa";
+        ctx->sinfo->sig->encoding = "x962";
+        break;

Hi Stefan,

Quoted 3 lines
Quoted 1 line

Does CONFIG_MODULE_SIG_KEY_TYPE_ECDSA have a dependency on MODULE_SIG_SHA256?

You are right, per the code above it does have a dependency on SHA256. ECDSA is using NIST p384 (secp384r1) for signing and per my tests it can be paired with all the sha hashes once the code above is extended. Now when it comes to module signing, should we pair it with a particular hash? I am not currently aware of a guidance document on this but sha256 and sha384 seem to be good choices these days, so maybe selecting ECDSA module signing should have a 'depends on' on these?

Yeah, I would tack on the 'depends on' until the code above has been extended to cover more sha hashes - because currently if someone builds and signs a bunch of modules with an ECDSA key, they will fail to load if they picked something other than sha256. I am unfortunately not knowledgeable enough to suggest an official guideline on choice of hash, but for now it is reasonable to have a 'depends on' for which hashes the code currently supports, so that users don't run into module loading rejection issues.

Thanks!

Jessica

#10 lore raw

Re: [PATCH v2 2/2] certs: Add support for using elliptic curve keys for signing modules

Quoted 25 lines · Jessica Yu

+++ Stefan Berger [20/04/21 17:02 -0400]:

Quoted 20 lines · Jessica Yu
Quoted 18 lines

+++ Stefan Berger [08/04/21 11:24 -0400]:

Quoted 6 lines
crypto/asymmetric_keys/pkcs7_parser.c +0 -0
index 967329e0a07b..2546ec6a0505 100644

crypto/asymmetric_keys/pkcs7_parser.c int pkcs7_sig_note_pkey_algo(void *context, · line 276

Quoted 8 lines
crypto/asymmetric_keys/pkcs7_parser.c +4 -0
@@ -269,6 +269,10 @@ int pkcs7_sig_note_pkey_algo(void *context, 
 size_t hdrlen,
         ctx->sinfo->sig->pkey_algo = "rsa";
         ctx->sinfo->sig->encoding = "pkcs1";
         break;
+    case OID_id_ecdsa_with_sha256:
+        ctx->sinfo->sig->pkey_algo = "ecdsa";
+        ctx->sinfo->sig->encoding = "x962";
+        break;

Hi Stefan,

Quoted 3 lines
Quoted 1 line

Does CONFIG_MODULE_SIG_KEY_TYPE_ECDSA have a dependency on MODULE_SIG_SHA256?

You are right, per the code above it does have a dependency on SHA256. ECDSA is using NIST p384 (secp384r1) for signing and per my tests it can be paired with all the sha hashes once the code above is extended. Now when it comes to module signing, should we pair it with a particular hash? I am not currently aware of a guidance document on this but sha256 and sha384 seem to be good choices these days, so maybe selecting ECDSA module signing should have a 'depends on' on these?

Quoted 8 lines

Yeah, I would tack on the 'depends on' until the code above has been extended to cover more sha hashes - because currently if someone builds and signs a bunch of modules with an ECDSA key, they will fail to load if they picked something other than sha256. I am unfortunately not knowledgeable enough to suggest an official guideline on choice of hash, but for now it is reasonable to have a 'depends on' for which hashes the code currently supports, so that users don't run into module loading rejection issues.

I was going to repost this series now with the additional OIDs supported above and a recommendation to use sha384 in the help text for ECDSA-signed modules, but not enforcing this but instead trusting the user that they will choose a reasonable hash (probably >= sha256).

   Stefan

Quoted 4 lines

Thanks!

Jessica

#11 lore raw

Re: [PATCH v2 2/2] certs: Add support for using elliptic curve keys for signing modules

+++ Stefan Berger [21/04/21 08:54 -0400]:

Quoted 37 lines · Jessica Yu
Quoted 25 lines

+++ Stefan Berger [20/04/21 17:02 -0400]:

Quoted 20 lines · Jessica Yu
Quoted 18 lines

+++ Stefan Berger [08/04/21 11:24 -0400]:

Quoted 6 lines
crypto/asymmetric_keys/pkcs7_parser.c +0 -0
index 967329e0a07b..2546ec6a0505 100644

crypto/asymmetric_keys/pkcs7_parser.c int pkcs7_sig_note_pkey_algo(void · line 276

Quoted 8 lines
crypto/asymmetric_keys/pkcs7_parser.c +4 -0
@@ -269,6 +269,10 @@ int pkcs7_sig_note_pkey_algo(void 
 *context, size_t hdrlen,
         ctx->sinfo->sig->pkey_algo = "rsa";
         ctx->sinfo->sig->encoding = "pkcs1";
         break;
+    case OID_id_ecdsa_with_sha256:
+        ctx->sinfo->sig->pkey_algo = "ecdsa";
+        ctx->sinfo->sig->encoding = "x962";
+        break;

Hi Stefan,

Quoted 3 lines
Quoted 1 line

Does CONFIG_MODULE_SIG_KEY_TYPE_ECDSA have a dependency on MODULE_SIG_SHA256?

You are right, per the code above it does have a dependency on SHA256. ECDSA is using NIST p384 (secp384r1) for signing and per my tests it can be paired with all the sha hashes once the code above is extended. Now when it comes to module signing, should we pair it with a particular hash? I am not currently aware of a guidance document on this but sha256 and sha384 seem to be good choices these days, so maybe selecting ECDSA module signing should have a 'depends on' on these?

Quoted 8 lines

Yeah, I would tack on the 'depends on' until the code above has been extended to cover more sha hashes - because currently if someone builds and signs a bunch of modules with an ECDSA key, they will fail to load if they picked something other than sha256. I am unfortunately not knowledgeable enough to suggest an official guideline on choice of hash, but for now it is reasonable to have a 'depends on' for which hashes the code currently supports, so that users don't run into module loading rejection issues.

Quoted 1 line

I was going to repost this series now with the additional OIDs supported above and a recommendation to use sha384 in the help text for ECDSA-signed modules, but not enforcing this but instead trusting the user that they will choose a reasonable hash (probably >= sha256).

OK, that sounds good to me.

Thanks Stefan!