Skip to contentSkip to conversation

KEYS: encrypted: Instantiate key with user-provided decrypted data

PATCH Superseded Yael Tiomkin 5 participants 16 messages Dec 29, 2021 – Jan 20, 2022

Subsystems (5)

v4 of 5 Superseded Latest is v5 →

lore
Share & tools
lore
#0 lore raw

PATCH v4 1/1 KEYS: encrypted: Instantiate key with user-provided decrypted data

The encrypted.c class supports instantiation of encrypted keys with either an already-encrypted key material, or by generating new key material based on random numbers. This patch defines a new datablob
format: [<format>] <master-key name> <decrypted data length> <decrypted data> that allows to instantiate encrypted keys using user-provided decrypted data, and therefore allows to perform key encryption from userspace. The decrypted key material will be inaccessible from userspace.

---

Notes:

    v -> v2: fixed compilation error.
    
    v2 -> v3: modified documentation.
    
    v3 -> v4: modified commit message.
 .../security/keys/trusted-encrypted.rst       | 25 ++++++--
 security/keys/encrypted-keys/encrypted.c      | 62 +++++++++++++------
 2 files changed, 63 insertions(+), 24 deletions(-)
Documentation/security/keys/trusted-encrypted.rst +19 -6
index 80d5a5af62a1..f614dad7de12 100644
@@ -107,12 +107,13 @@ Encrypted Keys
 --------------
 
 Encrypted keys do not depend on a trust source, and are faster, as they use AES
-for encryption/decryption. New keys are created from kernel-generated random
-numbers, and are encrypted/decrypted using a specified ‘master’ key. The
-‘master’ key can either be a trusted-key or user-key type. The main disadvantage
-of encrypted keys is that if they are not rooted in a trusted key, they are only
-as secure as the user key encrypting them. The master user key should therefore
-be loaded in as secure a way as possible, preferably early in boot.
+for encryption/decryption. New keys are created either from kernel-generated
+random numbers or user-provided decrypted data, and are encrypted/decrypted
+using a specified ‘master’ key. The ‘master’ key can either be a trusted-key or
+user-key type. The main disadvantage of encrypted keys is that if they are not
+rooted in a trusted key, they are only as secure as the user key encrypting
+them. The master user key should therefore be loaded in as secure a way as
+possible, preferably early in boot.
 
 
 Usage
@@ -199,6 +200,8 @@ Usage::
 
     keyctl add encrypted name "new [format] key-type:master-key-name keylen"
         ring
+    keyctl add encrypted name "new [format] key-type:master-key-name keylen
+        decrypted-data" ring
     keyctl add encrypted name "load hex_blob" ring
     keyctl update keyid "update key-type:master-key-name"
 
@@ -303,6 +306,16 @@ Load an encrypted key "evm" from saved blob::
     82dbbc55be2a44616e4959430436dc4f2a7a9659aa60bb4652aeb2120f149ed197c564e0
     24717c64 5972dcb82ab2dde83376d82b2e3c09ffc
 
+Instantiate an encrypted key "evm" using user-provided decrypted data::
+
+    $ keyctl add encrypted evm "new default user:kmk 32 `cat evm_decrypted_data.blob`" @u
+    794890253
+
+    $ keyctl print 794890253
+    default user:kmk 32 2375725ad57798846a9bbd240de8906f006e66c03af53b1b382d
+    bbc55be2a44616e4959430436dc4f2a7a9659aa60bb4652aeb2120f149ed197c564e0247
+    17c64 5972dcb82ab2dde83376d82b2e3c09ffc
+
 Other uses for trusted and encrypted keys, such as for disk and file encryption
 are anticipated.  In particular the new format 'ecryptfs' has been defined
 in order to use encrypted keys to mount an eCryptfs filesystem.  More details
security/keys/encrypted-keys/encrypted.c +44 -18

154 lines not on this page · raw message

Signature · 1 line
2.34.1.448.ga2b2bfdf31-goog
#1 lore raw

+ Jan, Ahmad

Quoted 9 lines · Yael Tiomkin

The encrypted.c class supports instantiation of encrypted keys with either an already-encrypted key material, or by generating new key material based on random numbers. This patch defines a new datablob
format: [<format>] <master-key name> <decrypted data length> <decrypted data> that allows to instantiate encrypted keys using user-provided decrypted data, and therefore allows to perform key encryption from userspace. The decrypted key material will be inaccessible from userspace.

This type of user-space key import feature has already been discussed at large in the context of trusted keys here [1]. So what makes it special in case of encrypted keys such that it isn't a "UNSAFE_IMPORT" or "DEBUGGING_IMPORT" or "DEVELOPMENT_IMPORT", ...?

[1] https://lore.kernel.org/linux-integrity/[email protected]/

-Sumit

Quoted 239 lines

---

Notes:

    v -> v2: fixed compilation error.

    v2 -> v3: modified documentation.

    v3 -> v4: modified commit message.
 .../security/keys/trusted-encrypted.rst       | 25 ++++++--
 security/keys/encrypted-keys/encrypted.c      | 62 +++++++++++++------
 2 files changed, 63 insertions(+), 24 deletions(-)
Documentation/security/keys/trusted-encrypted.rst +19 -6
index 80d5a5af62a1..f614dad7de12 100644
@@ -107,12 +107,13 @@ Encrypted Keys
 --------------
 
 Encrypted keys do not depend on a trust source, and are faster, as they use AES
-for encryption/decryption. New keys are created from kernel-generated random
-numbers, and are encrypted/decrypted using a specified ‘master’ key. The
-‘master’ key can either be a trusted-key or user-key type. The main disadvantage
-of encrypted keys is that if they are not rooted in a trusted key, they are only
-as secure as the user key encrypting them. The master user key should therefore
-be loaded in as secure a way as possible, preferably early in boot.
+for encryption/decryption. New keys are created either from kernel-generated
+random numbers or user-provided decrypted data, and are encrypted/decrypted
+using a specified ‘master’ key. The ‘master’ key can either be a trusted-key or
+user-key type. The main disadvantage of encrypted keys is that if they are not
+rooted in a trusted key, they are only as secure as the user key encrypting
+them. The master user key should therefore be loaded in as secure a way as
+possible, preferably early in boot.
 
 
 Usage
@@ -199,6 +200,8 @@ Usage::
 
     keyctl add encrypted name "new [format] key-type:master-key-name keylen"
         ring
+    keyctl add encrypted name "new [format] key-type:master-key-name keylen
+        decrypted-data" ring
     keyctl add encrypted name "load hex_blob" ring
     keyctl update keyid "update key-type:master-key-name"
 
@@ -303,6 +306,16 @@ Load an encrypted key "evm" from saved blob::
     82dbbc55be2a44616e4959430436dc4f2a7a9659aa60bb4652aeb2120f149ed197c564e0
     24717c64 5972dcb82ab2dde83376d82b2e3c09ffc
 
+Instantiate an encrypted key "evm" using user-provided decrypted data::
+
+    $ keyctl add encrypted evm "new default user:kmk 32 `cat evm_decrypted_data.blob`" @u
+    794890253
+
+    $ keyctl print 794890253
+    default user:kmk 32 2375725ad57798846a9bbd240de8906f006e66c03af53b1b382d
+    bbc55be2a44616e4959430436dc4f2a7a9659aa60bb4652aeb2120f149ed197c564e0247
+    17c64 5972dcb82ab2dde83376d82b2e3c09ffc
+
 Other uses for trusted and encrypted keys, such as for disk and file encryption
 are anticipated.  In particular the new format 'ecryptfs' has been defined
 in order to use encrypted keys to mount an eCryptfs filesystem.  More details
security/keys/encrypted-keys/encrypted.c +44 -19
index 87432b35d771..baf6fba5e05e 100644
@@ -159,6 +159,7 @@ static int valid_master_desc(const char *new_desc, const char *orig_desc)
  *
  * datablob format:
  * new [<format>] <master-key name> <decrypted data length>
+ * new [<format>] <master-key name> <decrypted data length> <decrypted data>
  * load [<format>] <master-key name> <decrypted data length>
  *     <encrypted iv + data>
  * update <new-master-key name>
@@ -170,7 +171,7 @@ static int valid_master_desc(const char *new_desc, const char *orig_desc)
  */
 static int datablob_parse(char *datablob, const char **format,
                          char **master_desc, char **decrypted_datalen,
-                         char **hex_encoded_iv)
+                         char **hex_encoded_iv, char **decrypted_data)
 {
        substring_t args[MAX_OPT_ARGS];
        int ret = -EINVAL;
@@ -231,6 +232,8 @@ static int datablob_parse(char *datablob, const char **format,
                                "when called from .update method\n", keyword);
                        break;
                }
+               *decrypted_data = strsep(&datablob, " \t");
+
                ret = 0;
                break;
        case Opt_load:
@@ -595,7 +598,8 @@ static int derived_key_decrypt(struct encrypted_key_payload *epayload,
 static struct encrypted_key_payload *encrypted_key_alloc(struct key *key,
                                                         const char *format,
                                                         const char *master_desc,
-                                                        const char *datalen)
+                                                        const char *datalen,
+                                                        const char *decrypted_data)
 {
        struct encrypted_key_payload *epayload = NULL;
        unsigned short datablob_len;
@@ -604,6 +608,7 @@ static struct encrypted_key_payload *encrypted_key_alloc(struct key *key,
        unsigned int encrypted_datalen;
        unsigned int format_len;
        long dlen;
+       int i;
        int ret;
 
        ret = kstrtol(datalen, 10, &dlen);
@@ -613,6 +618,20 @@ static struct encrypted_key_payload *encrypted_key_alloc(struct key *key,
        format_len = (!format) ? strlen(key_format_default) : strlen(format);
        decrypted_datalen = dlen;
        payload_datalen = decrypted_datalen;
+
+       if (decrypted_data) {
+               if (strlen(decrypted_data) != decrypted_datalen) {
+                       pr_err("encrypted key: decrypted data provided does not match decrypted data length provided\n");
+                       return ERR_PTR(-EINVAL);
+               }
+               for (i = 0; i < strlen(decrypted_data); i++) {
+                       if (!isalnum(decrypted_data[i])) {
+                               pr_err("encrypted key: decrypted data provided must be alphanumeric\n");
+                               return ERR_PTR(-EINVAL);
+                       }
+               }
+       }
+
        if (format) {
                if (!strcmp(format, key_format_ecryptfs)) {
                        if (dlen != ECRYPTFS_MAX_KEY_BYTES) {
@@ -740,13 +759,14 @@ static void __ekey_init(struct encrypted_key_payload *epayload,
 /*
  * encrypted_init - initialize an encrypted key
  *
- * For a new key, use a random number for both the iv and data
- * itself.  For an old key, decrypt the hex encoded data.
+ * For a new key, use either a random number or user-provided decrypted data in
+ * case it is provided. A random number is used for the iv in both cases. For
+ * an old key, decrypt the hex encoded data.
  */
 static int encrypted_init(struct encrypted_key_payload *epayload,
                          const char *key_desc, const char *format,
                          const char *master_desc, const char *datalen,
-                         const char *hex_encoded_iv)
+                         const char *hex_encoded_iv, const char *decrypted_data)
 {
        int ret = 0;
 
@@ -760,21 +780,26 @@ static int encrypted_init(struct encrypted_key_payload *epayload,
        }
 
        __ekey_init(epayload, format, master_desc, datalen);
-       if (!hex_encoded_iv) {
-               get_random_bytes(epayload->iv, ivsize);
-
-               get_random_bytes(epayload->decrypted_data,
-                                epayload->decrypted_datalen);
-       } else
+       if (hex_encoded_iv) {
                ret = encrypted_key_decrypt(epayload, format, hex_encoded_iv);
+       } else if (decrypted_data) {
+               get_random_bytes(epayload->iv, ivsize);
+               memcpy(epayload->decrypted_data, decrypted_data,
+                               epayload->decrypted_datalen);
+       } else {
+               get_random_bytes(epayload->iv, ivsize);
+               get_random_bytes(epayload->decrypted_data, epayload->decrypted_datalen);
+       }
        return ret;
 }
 
 /*
  * encrypted_instantiate - instantiate an encrypted key
  *
- * Decrypt an existing encrypted datablob or create a new encrypted key
- * based on a kernel random number.
+ * Instantiates the key:
+ * - by decrypting an existing encrypted datablob, or
+ * - by creating a new encrypted key based on a kernel random number, or
+ * - using provided decrypted data.
  *
  * On success, return 0. Otherwise return errno.
  */
@@ -787,6 +812,7 @@ static int encrypted_instantiate(struct key *key,
        char *master_desc = NULL;
        char *decrypted_datalen = NULL;
        char *hex_encoded_iv = NULL;
+       char *decrypted_data = NULL;
        size_t datalen = prep->datalen;
        int ret;
 
@@ -799,18 +825,18 @@ static int encrypted_instantiate(struct key *key,
        datablob[datalen] = 0;
        memcpy(datablob, prep->data, datalen);
        ret = datablob_parse(datablob, &format, &master_desc,
-                            &decrypted_datalen, &hex_encoded_iv);
+                            &decrypted_datalen, &hex_encoded_iv, &decrypted_data);
        if (ret < 0)
                goto out;
 
        epayload = encrypted_key_alloc(key, format, master_desc,
-                                      decrypted_datalen);
+                                      decrypted_datalen, decrypted_data);
        if (IS_ERR(epayload)) {
                ret = PTR_ERR(epayload);
                goto out;
        }
        ret = encrypted_init(epayload, key->description, format, master_desc,
-                            decrypted_datalen, hex_encoded_iv);
+                            decrypted_datalen, hex_encoded_iv, decrypted_data);
        if (ret < 0) {
                kfree_sensitive(epayload);
                goto out;
@@ -860,7 +886,7 @@ static int encrypted_update(struct key *key, struct key_preparsed_payload *prep)
 
        buf[datalen] = 0;
        memcpy(buf, prep->data, datalen);
-       ret = datablob_parse(buf, &format, &new_master_desc, NULL, NULL);
+       ret = datablob_parse(buf, &format, &new_master_desc, NULL, NULL, NULL);
        if (ret < 0)
                goto out;
 
@@ -869,7 +895,7 @@ static int encrypted_update(struct key *key, struct key_preparsed_payload *prep)
                goto out;
 
        new_epayload = encrypted_key_alloc(key, epayload->format,
-                                          new_master_desc, epayload->datalen);
+                                          new_master_desc, epayload->datalen, NULL);
        if (IS_ERR(new_epayload)) {
                ret = PTR_ERR(new_epayload);
                goto out;
--

2.34.1.448.ga2b2bfdf31-goog

maintainer
#2 lore raw

Hi Sumit,

Quoted 13 lines · Sumit Garg

+ Jan, Ahmad

Quoted 9 lines · Yael Tiomkin

The encrypted.c class supports instantiation of encrypted keys with either an already-encrypted key material, or by generating new key material based on random numbers. This patch defines a new datablob
format: [<format>] <master-key name> <decrypted data length> <decrypted data> that allows to instantiate encrypted keys using user-provided decrypted data, and therefore allows to perform key encryption from userspace. The decrypted key material will be inaccessible from userspace.

Quoted 12 lines

This type of user-space key import feature has already been discussed at large in the context of trusted keys here [1]. So what makes it special in case of encrypted keys such that it isn't a "UNSAFE_IMPORT" or "DEBUGGING_IMPORT" or "DEVELOPMENT_IMPORT", ...?

[1] https://lore.kernel.org/linux-integrity/[email protected]/

-Sumit

Quoted 3 lines

There is a difference between trusted and encrypted keys. So in addition to pointing to the rather long discussion thread, please summarize the conclusion and, assuming you agree, include why in once case it was acceptable and in the other it wasn't to provide userspace key data.

thanks,

Mimi

#3 lore raw

Hi Mimi,

Apologies for the delayed reply as I was on leave for a long new year weekend.

Quoted 24 lines · Mimi Zohar

Hi Sumit,

Quoted 20 lines · Sumit Garg

+ Jan, Ahmad

Quoted 9 lines · Yael Tiomkin

The encrypted.c class supports instantiation of encrypted keys with either an already-encrypted key material, or by generating new key material based on random numbers. This patch defines a new datablob
format: [<format>] <master-key name> <decrypted data length> <decrypted data> that allows to instantiate encrypted keys using user-provided decrypted data, and therefore allows to perform key encryption from userspace. The decrypted key material will be inaccessible from userspace.

This type of user-space key import feature has already been discussed at large in the context of trusted keys here [1]. So what makes it special in case of encrypted keys such that it isn't a "UNSAFE_IMPORT" or "DEBUGGING_IMPORT" or "DEVELOPMENT_IMPORT", ...?

[1] https://lore.kernel.org/linux-integrity/[email protected]/

Quoted 7 lines
Quoted 5 lines

-Sumit

Quoted 3 lines

There is a difference between trusted and encrypted keys.

Yeah I understand the implementation differences.

Quoted 5 lines

So in
addition to pointing to the rather long discussion thread, please summarize the conclusion and, assuming you agree, include why in once case it was acceptable and in the other it wasn't to provide userspace key data.

My major concern with importing user-space key data in *plain* format is that if import is *not* done in a safe (manufacturing or production) environment then the plain key data is susceptible to user-space compromises when the device is in the field.

And it sounds like we are diverting from basic definition [1] of encrypted keys:

"Trusted and Encrypted Keys are two new key types added to the existing kernel key ring service. Both of these new types are variable length symmetric keys, and in both cases all keys are created in the kernel, and **user space sees, stores, and loads** only encrypted blobs."

Also, as Jarrko mentioned earlier the use-case is still not clear to me as well. Isn't user logon keys an alternative option for non-readable user-space keys?

[1] https://www.kernel.org/doc/html/v4.13/security/keys/trusted-encrypted.html

-Sumit

Quoted 5 lines

thanks,

Mimi

#4 lore raw

Hi Sumit,

Quoted 46 lines · Sumit Garg

Hi Mimi,

Apologies for the delayed reply as I was on leave for a long new year weekend.

Quoted 24 lines · Mimi Zohar

Hi Sumit,

Quoted 20 lines · Sumit Garg

+ Jan, Ahmad

Quoted 9 lines · Yael Tiomkin

The encrypted.c class supports instantiation of encrypted keys with either an already-encrypted key material, or by generating new key material based on random numbers. This patch defines a new datablob
format: [<format>] <master-key name> <decrypted data length> <decrypted data> that allows to instantiate encrypted keys using user-provided decrypted data, and therefore allows to perform key encryption from userspace. The decrypted key material will be inaccessible from userspace.

This type of user-space key import feature has already been discussed at large in the context of trusted keys here [1]. So what makes it special in case of encrypted keys such that it isn't a "UNSAFE_IMPORT" or "DEBUGGING_IMPORT" or "DEVELOPMENT_IMPORT", ...?

[1] https://lore.kernel.org/linux-integrity/[email protected]/

Quoted 7 lines
Quoted 5 lines

-Sumit

Quoted 3 lines

There is a difference between trusted and encrypted keys.

Yeah I understand the implementation differences.

Quoted 5 lines

So in
addition to pointing to the rather long discussion thread, please summarize the conclusion and, assuming you agree, include why in once case it was acceptable and in the other it wasn't to provide userspace key data.

Quoted 4 lines

My major concern with importing user-space key data in *plain* format is that if import is *not* done in a safe (manufacturing or production) environment then the plain key data is susceptible to user-space compromises when the device is in the field.

I agree this can happen. Key distribution in any scenario needs to be secure and this could also potentially be an issue if the key is first encrypted and then imported. We can make sure the documentation highlights the safety requirement.

Quoted 9 lines

And it sounds like we are diverting from basic definition [1] of encrypted keys:

"Trusted and Encrypted Keys are two new key types added to the existing kernel key ring service. Both of these new types are variable length symmetric keys, and in both cases all keys are created in the kernel, and **user space sees, stores, and loads** only encrypted blobs."

Quoted 3 lines

Also, as Jarrko mentioned earlier the use-case is still not clear to me as well. Isn't user logon keys an alternative option for non-readable user-space keys?

The goal in this change is to allow key encryption from userspace, using user-provided decrypted data. This cannot be achieved in logon keys, which as you mentioned, are simply non-readable user type keys.

Quoted 10 lines

[1] https://www.kernel.org/doc/html/v4.13/security/keys/trusted-encrypted.html

-Sumit

Quoted 5 lines

thanks,

Mimi

Yael

#5 lore raw

Hi Yael,

Quoted 50 lines · Yael Tiomkin

Hi Sumit,

Quoted 46 lines · Sumit Garg

Hi Mimi,

Apologies for the delayed reply as I was on leave for a long new year weekend.

Quoted 24 lines · Mimi Zohar

Hi Sumit,

Quoted 20 lines · Sumit Garg

+ Jan, Ahmad

Quoted 9 lines · Yael Tiomkin

The encrypted.c class supports instantiation of encrypted keys with either an already-encrypted key material, or by generating new key material based on random numbers. This patch defines a new datablob
format: [<format>] <master-key name> <decrypted data length> <decrypted data> that allows to instantiate encrypted keys using user-provided decrypted data, and therefore allows to perform key encryption from userspace. The decrypted key material will be inaccessible from userspace.

This type of user-space key import feature has already been discussed at large in the context of trusted keys here [1]. So what makes it special in case of encrypted keys such that it isn't a "UNSAFE_IMPORT" or "DEBUGGING_IMPORT" or "DEVELOPMENT_IMPORT", ...?

[1] https://lore.kernel.org/linux-integrity/[email protected]/

Quoted 7 lines
Quoted 5 lines

-Sumit

Quoted 3 lines

There is a difference between trusted and encrypted keys.

Yeah I understand the implementation differences.

Quoted 5 lines

So in
addition to pointing to the rather long discussion thread, please summarize the conclusion and, assuming you agree, include why in once case it was acceptable and in the other it wasn't to provide userspace key data.

Quoted 8 lines
Quoted 4 lines

My major concern with importing user-space key data in *plain* format is that if import is *not* done in a safe (manufacturing or production) environment then the plain key data is susceptible to user-space compromises when the device is in the field.

I agree this can happen. Key distribution in any scenario needs to be secure and this could also potentially be an issue if the key is first encrypted and then imported.

Currently its not the case with encrypted keys. These are random keys generated within the kernel and encrypted with master key within the kernel and then exposed to user-space as encrypted blob only.

Quoted 3 lines

We can make sure the documentation
highlights the safety requirement.

IMO, you should enable this feature as a compile time option. The help text for that config option should highlight the use-case along with a safety warning.

-Sumit

Quoted 30 lines
Quoted 9 lines

And it sounds like we are diverting from basic definition [1] of encrypted keys:

"Trusted and Encrypted Keys are two new key types added to the existing kernel key ring service. Both of these new types are variable length symmetric keys, and in both cases all keys are created in the kernel, and **user space sees, stores, and loads** only encrypted blobs."

Quoted 3 lines

Also, as Jarrko mentioned earlier the use-case is still not clear to me as well. Isn't user logon keys an alternative option for non-readable user-space keys?

The goal in this change is to allow key encryption from userspace, using user-provided decrypted data. This cannot be achieved in logon keys, which as you mentioned, are simply non-readable user type keys.

Quoted 10 lines

[1] https://www.kernel.org/doc/html/v4.13/security/keys/trusted-encrypted.html

-Sumit

Quoted 5 lines

thanks,

Mimi

Yael

#6 lore raw

Hi Sumit,

Quoted 59 lines · Sumit Garg

Hi Yael,

Quoted 55 lines · Yael Tiomkin

Hi Sumit,

Quoted 50 lines · Sumit Garg

Hi Mimi,

Apologies for the delayed reply as I was on leave for a long new year weekend.

Quoted 24 lines · Mimi Zohar

Hi Sumit,

Quoted 20 lines · Sumit Garg

+ Jan, Ahmad

Quoted 9 lines · Yael Tiomkin

The encrypted.c class supports instantiation of encrypted keys with either an already-encrypted key material, or by generating new key material based on random numbers. This patch defines a new datablob
format: [<format>] <master-key name> <decrypted data length> <decrypted data> that allows to instantiate encrypted keys using user-provided decrypted data, and therefore allows to perform key encryption from userspace. The decrypted key material will be inaccessible from userspace.

This type of user-space key import feature has already been discussed at large in the context of trusted keys here [1]. So what makes it special in case of encrypted keys such that it isn't a "UNSAFE_IMPORT" or "DEBUGGING_IMPORT" or "DEVELOPMENT_IMPORT", ...?

[1] https://lore.kernel.org/linux-integrity/[email protected]/

Quoted 7 lines
Quoted 5 lines

-Sumit

Quoted 3 lines

There is a difference between trusted and encrypted keys.

Yeah I understand the implementation differences.

Quoted 5 lines

So in
addition to pointing to the rather long discussion thread, please summarize the conclusion and, assuming you agree, include why in once case it was acceptable and in the other it wasn't to provide userspace key data.

My major concern with importing user-space key data in *plain* format is that if import is *not* done in a safe (manufacturing or production) environment then the plain key data is susceptible to user-space compromises when the device is in the field.

Quoted 7 lines
Quoted 3 lines

I agree this can happen. Key distribution in any scenario needs to be secure and this could also potentially be an issue if the key is first encrypted and then imported.

Currently its not the case with encrypted keys. These are random keys generated within the kernel and encrypted with master key within the kernel and then exposed to user-space as encrypted blob only.

There are two different ways to create encrypted keys. One is to have them generated within the kernel using random numbers, and the other is by importing them in their encrypted form from user-space. I was referring to the latter in my previous statement.

Quoted 41 lines
Quoted 4 lines

We can make sure the documentation
highlights the safety requirement.

IMO, you should enable this feature as a compile time option. The help text for that config option should highlight the use-case along with a safety warning.

-Sumit

Quoted 30 lines
Quoted 9 lines

And it sounds like we are diverting from basic definition [1] of encrypted keys:

"Trusted and Encrypted Keys are two new key types added to the existing kernel key ring service. Both of these new types are variable length symmetric keys, and in both cases all keys are created in the kernel, and **user space sees, stores, and loads** only encrypted blobs."

Quoted 3 lines

Also, as Jarrko mentioned earlier the use-case is still not clear to me as well. Isn't user logon keys an alternative option for non-readable user-space keys?

The goal in this change is to allow key encryption from userspace, using user-provided decrypted data. This cannot be achieved in logon keys, which as you mentioned, are simply non-readable user type keys.

Quoted 10 lines

[1] https://www.kernel.org/doc/html/v4.13/security/keys/trusted-encrypted.html

-Sumit

Quoted 5 lines

thanks,

Mimi

Yael

#7 lore raw
Quoted 72 lines · Yael Tiomkin

Hi Sumit,

Quoted 59 lines · Sumit Garg

Hi Yael,

Quoted 55 lines · Yael Tiomkin

Hi Sumit,

Quoted 50 lines · Sumit Garg

Hi Mimi,

Apologies for the delayed reply as I was on leave for a long new year weekend.

Quoted 24 lines · Mimi Zohar

Hi Sumit,

Quoted 20 lines · Sumit Garg

+ Jan, Ahmad

Quoted 9 lines · Yael Tiomkin

The encrypted.c class supports instantiation of encrypted keys with either an already-encrypted key material, or by generating new key material based on random numbers. This patch defines a new datablob
format: [<format>] <master-key name> <decrypted data length> <decrypted data> that allows to instantiate encrypted keys using user-provided decrypted data, and therefore allows to perform key encryption from userspace. The decrypted key material will be inaccessible from userspace.

This type of user-space key import feature has already been discussed at large in the context of trusted keys here [1]. So what makes it special in case of encrypted keys such that it isn't a "UNSAFE_IMPORT" or "DEBUGGING_IMPORT" or "DEVELOPMENT_IMPORT", ...?

[1] https://lore.kernel.org/linux-integrity/[email protected]/

Quoted 7 lines
Quoted 5 lines

-Sumit

Quoted 3 lines

There is a difference between trusted and encrypted keys.

Yeah I understand the implementation differences.

Quoted 5 lines

So in
addition to pointing to the rather long discussion thread, please summarize the conclusion and, assuming you agree, include why in once case it was acceptable and in the other it wasn't to provide userspace key data.

My major concern with importing user-space key data in *plain* format is that if import is *not* done in a safe (manufacturing or production) environment then the plain key data is susceptible to user-space compromises when the device is in the field.

Quoted 7 lines
Quoted 3 lines

I agree this can happen. Key distribution in any scenario needs to be secure and this could also potentially be an issue if the key is first encrypted and then imported.

Currently its not the case with encrypted keys. These are random keys generated within the kernel and encrypted with master key within the kernel and then exposed to user-space as encrypted blob only.

Quoted 5 lines

There are two different ways to create encrypted keys. One is to have them generated within the kernel using random numbers, and the other is by importing them in their encrypted form from user-space. I was referring to the latter in my previous statement.

So, from a key distribution security point of view, encrypted key user-space import is **not equal to** plain key user-space import. That's why we need to have a separate compile time option as every device may come with its own threat model and may choose to enable or disable this user-space plain key import feature.

-Sumit

Quoted 41 lines
Quoted 41 lines
Quoted 4 lines

We can make sure the documentation
highlights the safety requirement.

IMO, you should enable this feature as a compile time option. The help text for that config option should highlight the use-case along with a safety warning.

-Sumit

Quoted 30 lines
Quoted 9 lines

And it sounds like we are diverting from basic definition [1] of encrypted keys:

"Trusted and Encrypted Keys are two new key types added to the existing kernel key ring service. Both of these new types are variable length symmetric keys, and in both cases all keys are created in the kernel, and **user space sees, stores, and loads** only encrypted blobs."

Quoted 3 lines

Also, as Jarrko mentioned earlier the use-case is still not clear to me as well. Isn't user logon keys an alternative option for non-readable user-space keys?

The goal in this change is to allow key encryption from userspace, using user-provided decrypted data. This cannot be achieved in logon keys, which as you mentioned, are simply non-readable user type keys.

Quoted 10 lines

[1] https://www.kernel.org/doc/html/v4.13/security/keys/trusted-encrypted.html

-Sumit

Quoted 5 lines

thanks,

Mimi

Yael

#8 lore raw
Quoted 80 lines · Sumit Garg
Quoted 72 lines · Yael Tiomkin

Hi Sumit,

Quoted 59 lines · Sumit Garg

Hi Yael,

Quoted 55 lines · Yael Tiomkin

Hi Sumit,

Quoted 50 lines · Sumit Garg

Hi Mimi,

Apologies for the delayed reply as I was on leave for a long new year weekend.

Quoted 24 lines · Mimi Zohar

Hi Sumit,

Quoted 20 lines · Sumit Garg

+ Jan, Ahmad

Quoted 9 lines · Yael Tiomkin

The encrypted.c class supports instantiation of encrypted keys with either an already-encrypted key material, or by generating new key material based on random numbers. This patch defines a new datablob
format: [<format>] <master-key name> <decrypted data length> <decrypted data> that allows to instantiate encrypted keys using user-provided decrypted data, and therefore allows to perform key encryption from userspace. The decrypted key material will be inaccessible from userspace.

This type of user-space key import feature has already been discussed at large in the context of trusted keys here [1]. So what makes it special in case of encrypted keys such that it isn't a "UNSAFE_IMPORT" or "DEBUGGING_IMPORT" or "DEVELOPMENT_IMPORT", ...?

[1] https://lore.kernel.org/linux-integrity/[email protected]/

Quoted 7 lines
Quoted 5 lines

-Sumit

Quoted 3 lines

There is a difference between trusted and encrypted keys.

Yeah I understand the implementation differences.

Quoted 5 lines

So in
addition to pointing to the rather long discussion thread, please summarize the conclusion and, assuming you agree, include why in once case it was acceptable and in the other it wasn't to provide userspace key data.

My major concern with importing user-space key data in *plain* format is that if import is *not* done in a safe (manufacturing or production) environment then the plain key data is susceptible to user-space compromises when the device is in the field.

Quoted 7 lines
Quoted 3 lines

I agree this can happen. Key distribution in any scenario needs to be secure and this could also potentially be an issue if the key is first encrypted and then imported.

Currently its not the case with encrypted keys. These are random keys generated within the kernel and encrypted with master key within the kernel and then exposed to user-space as encrypted blob only.

Quoted 5 lines

There are two different ways to create encrypted keys. One is to have them generated within the kernel using random numbers, and the other is by importing them in their encrypted form from user-space. I was referring to the latter in my previous statement.

Quoted 49 lines

So, from a key distribution security point of view, encrypted key user-space import is **not equal to** plain key user-space import. That's why we need to have a separate compile time option as every device may come with its own threat model and may choose to enable or disable this user-space plain key import feature.

-Sumit

Quoted 41 lines
Quoted 41 lines
Quoted 4 lines

We can make sure the documentation
highlights the safety requirement.

IMO, you should enable this feature as a compile time option. The help text for that config option should highlight the use-case along with a safety warning.

-Sumit

Quoted 30 lines
Quoted 9 lines

And it sounds like we are diverting from basic definition [1] of encrypted keys:

"Trusted and Encrypted Keys are two new key types added to the existing kernel key ring service. Both of these new types are variable length symmetric keys, and in both cases all keys are created in the kernel, and **user space sees, stores, and loads** only encrypted blobs."

Quoted 3 lines

Also, as Jarrko mentioned earlier the use-case is still not clear to me as well. Isn't user logon keys an alternative option for non-readable user-space keys?

The goal in this change is to allow key encryption from userspace, using user-provided decrypted data. This cannot be achieved in logon keys, which as you mentioned, are simply non-readable user type keys.

Quoted 10 lines

[1] https://www.kernel.org/doc/html/v4.13/security/keys/trusted-encrypted.html

-Sumit

Quoted 5 lines

thanks,

Mimi

Yael

Hi Sumit,
The encrypted-keys module is already controlled by the CONFIG_ENCRYPTED_KEYS option, which I think might give sufficient granularity to control the behavior. Do you still think a feature dedicated option is needed?

Yael

#9 lore raw
Quoted 130 lines · Yael Tiomkin
Quoted 128 lines · Sumit Garg
Quoted 72 lines · Yael Tiomkin

Hi Sumit,

Quoted 59 lines · Sumit Garg

Hi Yael,

Quoted 55 lines · Yael Tiomkin

Hi Sumit,

Quoted 50 lines · Sumit Garg

Hi Mimi,

Apologies for the delayed reply as I was on leave for a long new year weekend.

Quoted 24 lines · Mimi Zohar

Hi Sumit,

Quoted 20 lines · Sumit Garg

+ Jan, Ahmad

Quoted 9 lines · Yael Tiomkin

The encrypted.c class supports instantiation of encrypted keys with either an already-encrypted key material, or by generating new key material based on random numbers. This patch defines a new datablob
format: [<format>] <master-key name> <decrypted data length> <decrypted data> that allows to instantiate encrypted keys using user-provided decrypted data, and therefore allows to perform key encryption from userspace. The decrypted key material will be inaccessible from userspace.

This type of user-space key import feature has already been discussed at large in the context of trusted keys here [1]. So what makes it special in case of encrypted keys such that it isn't a "UNSAFE_IMPORT" or "DEBUGGING_IMPORT" or "DEVELOPMENT_IMPORT", ...?

[1] https://lore.kernel.org/linux-integrity/[email protected]/

Quoted 7 lines
Quoted 5 lines

-Sumit

Quoted 3 lines

There is a difference between trusted and encrypted keys.

Yeah I understand the implementation differences.

Quoted 5 lines

So in
addition to pointing to the rather long discussion thread, please summarize the conclusion and, assuming you agree, include why in once case it was acceptable and in the other it wasn't to provide userspace key data.

My major concern with importing user-space key data in *plain* format is that if import is *not* done in a safe (manufacturing or production) environment then the plain key data is susceptible to user-space compromises when the device is in the field.

Quoted 7 lines
Quoted 3 lines

I agree this can happen. Key distribution in any scenario needs to be secure and this could also potentially be an issue if the key is first encrypted and then imported.

Currently its not the case with encrypted keys. These are random keys generated within the kernel and encrypted with master key within the kernel and then exposed to user-space as encrypted blob only.

Quoted 5 lines

There are two different ways to create encrypted keys. One is to have them generated within the kernel using random numbers, and the other is by importing them in their encrypted form from user-space. I was referring to the latter in my previous statement.

So, from a key distribution security point of view, encrypted key user-space import is **not equal to** plain key user-space import. That's why we need to have a separate compile time option as every device may come with its own threat model and may choose to enable or disable this user-space plain key import feature.

-Sumit

Quoted 40 lines
Quoted 40 lines
Quoted 4 lines

We can make sure the documentation
highlights the safety requirement.

IMO, you should enable this feature as a compile time option. The help text for that config option should highlight the use-case along with a safety warning.

-Sumit

Quoted 29 lines
Quoted 9 lines

And it sounds like we are diverting from basic definition [1] of encrypted keys:

"Trusted and Encrypted Keys are two new key types added to the existing kernel key ring service. Both of these new types are variable length symmetric keys, and in both cases all keys are created in the kernel, and **user space sees, stores, and loads** only encrypted blobs."

Quoted 3 lines

Also, as Jarrko mentioned earlier the use-case is still not clear to me as well. Isn't user logon keys an alternative option for non-readable user-space keys?

The goal in this change is to allow key encryption from userspace, using user-provided decrypted data. This cannot be achieved in logon keys, which as you mentioned, are simply non-readable user type keys.

Quoted 10 lines

[1] https://www.kernel.org/doc/html/v4.13/security/keys/trusted-encrypted.html

-Sumit

Quoted 5 lines

thanks,

Mimi

Quoted 8 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

Yael

Hi Sumit,
The encrypted-keys module is already controlled by the CONFIG_ENCRYPTED_KEYS option, which I think might give sufficient granularity to control the behavior. Do you still think a feature dedicated option is needed?

Since this feature will change security properties for encrypted keys which can be considered as a regression for existing users during uprev. So it needs to be an optional key import feature with a clear user visible warning. Thinking more about this, you can even have a module param to turn it on with a runtime warning as well.

-Sumit

Quoted 1 line

Yael

maintainer
#10 lore raw
In reply to Yael Tiomkin
Quoted 8 lines · Yael Tiomkin

The encrypted.c class supports instantiation of encrypted keys with either an already-encrypted key material, or by generating new key material based on random numbers. This patch defines a new datablob
format: [<format>] <master-key name> <decrypted data length> <decrypted data> that allows to instantiate encrypted keys using user-provided decrypted data, and therefore allows to perform key encryption from userspace. The decrypted key material will be inaccessible from userspace.

The 2nd to last sentence is essentially a tautology but fails to be even that, as you can already "perform key encryption" from user space, just not with arbitrary key material.

It does not elighten any applications of this feature.

/Jarkko

#11 lore raw
Quoted 11 lines · Jarkko Sakkinen
Quoted 8 lines · Yael Tiomkin

The encrypted.c class supports instantiation of encrypted keys with either an already-encrypted key material, or by generating new key material based on random numbers. This patch defines a new datablob
format: [<format>] <master-key name> <decrypted data length> <decrypted data> that allows to instantiate encrypted keys using user-provided decrypted data, and therefore allows to perform key encryption from userspace. The decrypted key material will be inaccessible from userspace.

Quoted 7 lines

The 2nd to last sentence is essentially a tautology but fails to be even that, as you can already "perform key encryption" from user space, just not with arbitrary key material.

It does not elighten any applications of this feature.

/Jarkko

Sure. Please look at the modification below.

The encrypted.c class supports instantiation of encrypted keys with either an already-encrypted key material, or by generating new key material based on random numbers. This patch defines a new datablob
format: [<format>] <master-key name> <decrypted data length> <decrypted data> that allows to inject (and encrypt) user-provided decrypted data. The decrypted key material will be inaccessible from userspace. This feature also acts as a building block for a userspace envelope encryption capability.

Yael

maintainer
#12 lore raw
In reply to Yael Tiomkin
Quoted 11 lines · Yael Tiomkin

The encrypted.c class supports instantiation of encrypted keys with either an already-encrypted key material, or by generating new key material based on random numbers. This patch defines a new datablob
format: [<format>] <master-key name> <decrypted data length> <decrypted data> that allows to instantiate encrypted keys using user-provided decrypted data, and therefore allows to perform key encryption from userspace. The decrypted key material will be inaccessible from userspace.

What is the use case for this?

BR, Jarkko

#13 lore raw
In reply to Yael Tiomkin
Quoted 124 lines · Yael Tiomkin

The encrypted.c class supports instantiation of encrypted keys with either an already-encrypted key material, or by generating new key material based on random numbers. This patch defines a new datablob
format: [<format>] <master-key name> <decrypted data length> <decrypted data> that allows to instantiate encrypted keys using user-provided decrypted data, and therefore allows to perform key encryption from userspace. The decrypted key material will be inaccessible from userspace.

---

Notes:

     v -> v2: fixed compilation error.
     
     v2 -> v3: modified documentation.
     
     v3 -> v4: modified commit message.
  .../security/keys/trusted-encrypted.rst       | 25 ++++++--
  security/keys/encrypted-keys/encrypted.c      | 62 +++++++++++++------
  2 files changed, 63 insertions(+), 24 deletions(-)
Documentation/security/keys/trusted-encrypted.rst +19 -6
index 80d5a5af62a1..f614dad7de12 100644
@@ -107,12 +107,13 @@ Encrypted Keys
  --------------
 
  Encrypted keys do not depend on a trust source, and are faster, as they use AES
-for encryption/decryption. New keys are created from kernel-generated random
-numbers, and are encrypted/decrypted using a specified ‘master’ key. The
-‘master’ key can either be a trusted-key or user-key type. The main disadvantage
-of encrypted keys is that if they are not rooted in a trusted key, they are only
-as secure as the user key encrypting them. The master user key should therefore
-be loaded in as secure a way as possible, preferably early in boot.
+for encryption/decryption. New keys are created either from kernel-generated
+random numbers or user-provided decrypted data, and are encrypted/decrypted
+using a specified ‘master’ key. The ‘master’ key can either be a trusted-key or
+user-key type. The main disadvantage of encrypted keys is that if they are not
+rooted in a trusted key, they are only as secure as the user key encrypting
+them. The master user key should therefore be loaded in as secure a way as
+possible, preferably early in boot.
 
 
  Usage
@@ -199,6 +200,8 @@ Usage::
 
      keyctl add encrypted name "new [format] key-type:master-key-name keylen"
          ring
+    keyctl add encrypted name "new [format] key-type:master-key-name keylen
+        decrypted-data" ring
      keyctl add encrypted name "load hex_blob" ring
      keyctl update keyid "update key-type:master-key-name"
 
@@ -303,6 +306,16 @@ Load an encrypted key "evm" from saved blob::
      82dbbc55be2a44616e4959430436dc4f2a7a9659aa60bb4652aeb2120f149ed197c564e0
      24717c64 5972dcb82ab2dde83376d82b2e3c09ffc
 
+Instantiate an encrypted key "evm" using user-provided decrypted data::
+
+    $ keyctl add encrypted evm "new default user:kmk 32 `cat evm_decrypted_data.blob`" @u
+    794890253
+
+    $ keyctl print 794890253
+    default user:kmk 32 2375725ad57798846a9bbd240de8906f006e66c03af53b1b382d
+    bbc55be2a44616e4959430436dc4f2a7a9659aa60bb4652aeb2120f149ed197c564e0247
+    17c64 5972dcb82ab2dde83376d82b2e3c09ffc
+
  Other uses for trusted and encrypted keys, such as for disk and file encryption
  are anticipated.  In particular the new format 'ecryptfs' has been defined
  in order to use encrypted keys to mount an eCryptfs filesystem.  More details
security/keys/encrypted-keys/encrypted.c +7 -2
index 87432b35d771..baf6fba5e05e 100644
@@ -159,6 +159,7 @@ static int valid_master_desc(const char *new_desc, const char *orig_desc)
   *
   * datablob format:
   * new [<format>] <master-key name> <decrypted data length>
+ * new [<format>] <master-key name> <decrypted data length> <decrypted data>
   * load [<format>] <master-key name> <decrypted data length>
   *     <encrypted iv + data>
   * update <new-master-key name>
@@ -170,7 +171,7 @@ static int valid_master_desc(const char *new_desc, const char *orig_desc)
   */
  static int datablob_parse(char *datablob, const char **format,
  			  char **master_desc, char **decrypted_datalen,
-			  char **hex_encoded_iv)
+			  char **hex_encoded_iv, char **decrypted_data)
  {
  	substring_t args[MAX_OPT_ARGS];
  	int ret = -EINVAL;
@@ -231,6 +232,8 @@ static int datablob_parse(char *datablob, const char **format,
  				"when called from .update method\n", keyword);
  			break;
  		}
+		*decrypted_data = strsep(&datablob, " \t");
+
  		ret = 0;
  		break;
  	case Opt_load:
@@ -595,7 +598,8 @@ static int derived_key_decrypt(struct encrypted_key_payload *epayload,
  static struct encrypted_key_payload *encrypted_key_alloc(struct key *key,
  							 const char *format,
  							 const char *master_desc,
-							 const char *datalen)
+							 const char *datalen,
+							 const char *decrypted_data)
  {
  	struct encrypted_key_payload *epayload = NULL;
  	unsigned short datablob_len;
@@ -604,6 +608,7 @@ static struct encrypted_key_payload *encrypted_key_alloc(struct key *key,
  	unsigned int encrypted_datalen;
  	unsigned int format_len;
  	long dlen;
+	int i;
  	int ret;
 
  	ret = kstrtol(datalen, 10, &dlen);
@@ -613,6 +618,20 @@ static struct encrypted_key_payload *encrypted_key_alloc(struct key *key,
  	format_len = (!format) ? strlen(key_format_default) : strlen(format);

security/keys/encrypted-keys/encrypted.c static struct encrypted_key_payload *encrypted_key_alloc(struct key *key, · line 628

Quoted 10 lines
security/keys/encrypted-keys/encrypted.c +8 -0
  	decrypted_datalen = dlen;
  	payload_datalen = decrypted_datalen;
+
+	if (decrypted_data) {
+		if (strlen(decrypted_data) != decrypted_datalen) {
+			pr_err("encrypted key: decrypted data provided does not match decrypted data length provided\n");
+			return ERR_PTR(-EINVAL);
+		}
+		for (i = 0; i < strlen(decrypted_data); i++) {
+			if (!isalnum(decrypted_data[i])) {

User-provided decrypted data may have special characters, commonly used in passwords or key phrases, apart from alphanumeric.  Replace isalnum with !iscntrl() to validate against control characters but allow special characters.

Thanks & Regards,

     - Nayna

#14 lore raw
Quoted 143 lines · Nayna
Quoted 124 lines · Yael Tiomkin

The encrypted.c class supports instantiation of encrypted keys with either an already-encrypted key material, or by generating new key material based on random numbers. This patch defines a new datablob
format: [<format>] <master-key name> <decrypted data length> <decrypted data> that allows to instantiate encrypted keys using user-provided decrypted data, and therefore allows to perform key encryption from userspace. The decrypted key material will be inaccessible from userspace.

---

Notes:

     v -> v2: fixed compilation error.

     v2 -> v3: modified documentation.

     v3 -> v4: modified commit message.
  .../security/keys/trusted-encrypted.rst       | 25 ++++++--
  security/keys/encrypted-keys/encrypted.c      | 62 +++++++++++++------
  2 files changed, 63 insertions(+), 24 deletions(-)
Documentation/security/keys/trusted-encrypted.rst +19 -6
index 80d5a5af62a1..f614dad7de12 100644
@@ -107,12 +107,13 @@ Encrypted Keys
  --------------
 
  Encrypted keys do not depend on a trust source, and are faster, as they use AES
-for encryption/decryption. New keys are created from kernel-generated random
-numbers, and are encrypted/decrypted using a specified ‘master’ key. The
-‘master’ key can either be a trusted-key or user-key type. The main disadvantage
-of encrypted keys is that if they are not rooted in a trusted key, they are only
-as secure as the user key encrypting them. The master user key should therefore
-be loaded in as secure a way as possible, preferably early in boot.
+for encryption/decryption. New keys are created either from kernel-generated
+random numbers or user-provided decrypted data, and are encrypted/decrypted
+using a specified ‘master’ key. The ‘master’ key can either be a trusted-key or
+user-key type. The main disadvantage of encrypted keys is that if they are not
+rooted in a trusted key, they are only as secure as the user key encrypting
+them. The master user key should therefore be loaded in as secure a way as
+possible, preferably early in boot.
 
 
  Usage
@@ -199,6 +200,8 @@ Usage::
 
      keyctl add encrypted name "new [format] key-type:master-key-name keylen"
          ring
+    keyctl add encrypted name "new [format] key-type:master-key-name keylen
+        decrypted-data" ring
      keyctl add encrypted name "load hex_blob" ring
      keyctl update keyid "update key-type:master-key-name"
 
@@ -303,6 +306,16 @@ Load an encrypted key "evm" from saved blob::
      82dbbc55be2a44616e4959430436dc4f2a7a9659aa60bb4652aeb2120f149ed197c564e0
      24717c64 5972dcb82ab2dde83376d82b2e3c09ffc
 
+Instantiate an encrypted key "evm" using user-provided decrypted data::
+
+    $ keyctl add encrypted evm "new default user:kmk 32 `cat evm_decrypted_data.blob`" @u
+    794890253
+
+    $ keyctl print 794890253
+    default user:kmk 32 2375725ad57798846a9bbd240de8906f006e66c03af53b1b382d
+    bbc55be2a44616e4959430436dc4f2a7a9659aa60bb4652aeb2120f149ed197c564e0247
+    17c64 5972dcb82ab2dde83376d82b2e3c09ffc
+
  Other uses for trusted and encrypted keys, such as for disk and file encryption
  are anticipated.  In particular the new format 'ecryptfs' has been defined
  in order to use encrypted keys to mount an eCryptfs filesystem.  More details
security/keys/encrypted-keys/encrypted.c +7 -2
index 87432b35d771..baf6fba5e05e 100644
@@ -159,6 +159,7 @@ static int valid_master_desc(const char *new_desc, const char *orig_desc)
   *
   * datablob format:
   * new [<format>] <master-key name> <decrypted data length>
+ * new [<format>] <master-key name> <decrypted data length> <decrypted data>
   * load [<format>] <master-key name> <decrypted data length>
   *     <encrypted iv + data>
   * update <new-master-key name>
@@ -170,7 +171,7 @@ static int valid_master_desc(const char *new_desc, const char *orig_desc)
   */
  static int datablob_parse(char *datablob, const char **format,
                        char **master_desc, char **decrypted_datalen,
-                       char **hex_encoded_iv)
+                       char **hex_encoded_iv, char **decrypted_data)
  {
      substring_t args[MAX_OPT_ARGS];
      int ret = -EINVAL;
@@ -231,6 +232,8 @@ static int datablob_parse(char *datablob, const char **format,
                              "when called from .update method\n", keyword);
                      break;
              }
+             *decrypted_data = strsep(&datablob, " \t");
+
              ret = 0;
              break;
      case Opt_load:
@@ -595,7 +598,8 @@ static int derived_key_decrypt(struct encrypted_key_payload *epayload,
  static struct encrypted_key_payload *encrypted_key_alloc(struct key *key,
                                                       const char *format,
                                                       const char *master_desc,
-                                                      const char *datalen)
+                                                      const char *datalen,
+                                                      const char *decrypted_data)
  {
      struct encrypted_key_payload *epayload = NULL;
      unsigned short datablob_len;
@@ -604,6 +608,7 @@ static struct encrypted_key_payload *encrypted_key_alloc(struct key *key,
      unsigned int encrypted_datalen;
      unsigned int format_len;
      long dlen;
+     int i;
      int ret;
 
      ret = kstrtol(datalen, 10, &dlen);
@@ -613,6 +618,20 @@ static struct encrypted_key_payload *encrypted_key_alloc(struct key *key,
      format_len = (!format) ? strlen(key_format_default) : strlen(format);

security/keys/encrypted-keys/encrypted.c static struct encrypted_key_payload *encrypted_key_alloc(struct key *key, · line 628

Quoted 10 lines
security/keys/encrypted-keys/encrypted.c +8 -0
      decrypted_datalen = dlen;
      payload_datalen = decrypted_datalen;
+
+     if (decrypted_data) {
+             if (strlen(decrypted_data) != decrypted_datalen) {
+                     pr_err("encrypted key: decrypted data provided does not match decrypted data length provided\n");
+                     return ERR_PTR(-EINVAL);
+             }
+             for (i = 0; i < strlen(decrypted_data); i++) {
+                     if (!isalnum(decrypted_data[i])) {

User-provided decrypted data may have special characters, commonly used in passwords or key phrases, apart from alphanumeric. Replace isalnum with !iscntrl() to validate against control characters but allow special characters.

Quoted 4 lines

Thanks & Regards,

- Nayna

Hi Nayna,
I wonder if we should use isprint() instead?

Yael

#15 lore raw
Quoted 142 lines · Yael Tiomkin
Quoted 141 lines · Nayna
Quoted 124 lines · Yael Tiomkin

The encrypted.c class supports instantiation of encrypted keys with either an already-encrypted key material, or by generating new key material based on random numbers. This patch defines a new datablob
format: [<format>] <master-key name> <decrypted data length> <decrypted data> that allows to instantiate encrypted keys using user-provided decrypted data, and therefore allows to perform key encryption from userspace. The decrypted key material will be inaccessible from userspace.

---

Notes:

      v -> v2: fixed compilation error.

      v2 -> v3: modified documentation.

      v3 -> v4: modified commit message.
   .../security/keys/trusted-encrypted.rst       | 25 ++++++--
   security/keys/encrypted-keys/encrypted.c      | 62 +++++++++++++------
   2 files changed, 63 insertions(+), 24 deletions(-)
Documentation/security/keys/trusted-encrypted.rst +19 -6
index 80d5a5af62a1..f614dad7de12 100644
@@ -107,12 +107,13 @@ Encrypted Keys
   --------------
 
   Encrypted keys do not depend on a trust source, and are faster, as they use AES
-for encryption/decryption. New keys are created from kernel-generated random
-numbers, and are encrypted/decrypted using a specified ‘master’ key. The
-‘master’ key can either be a trusted-key or user-key type. The main disadvantage
-of encrypted keys is that if they are not rooted in a trusted key, they are only
-as secure as the user key encrypting them. The master user key should therefore
-be loaded in as secure a way as possible, preferably early in boot.
+for encryption/decryption. New keys are created either from kernel-generated
+random numbers or user-provided decrypted data, and are encrypted/decrypted
+using a specified ‘master’ key. The ‘master’ key can either be a trusted-key or
+user-key type. The main disadvantage of encrypted keys is that if they are not
+rooted in a trusted key, they are only as secure as the user key encrypting
+them. The master user key should therefore be loaded in as secure a way as
+possible, preferably early in boot.
 
 
   Usage
@@ -199,6 +200,8 @@ Usage::
 
       keyctl add encrypted name "new [format] key-type:master-key-name keylen"
           ring
+    keyctl add encrypted name "new [format] key-type:master-key-name keylen
+        decrypted-data" ring
       keyctl add encrypted name "load hex_blob" ring
       keyctl update keyid "update key-type:master-key-name"
 
@@ -303,6 +306,16 @@ Load an encrypted key "evm" from saved blob::
       82dbbc55be2a44616e4959430436dc4f2a7a9659aa60bb4652aeb2120f149ed197c564e0
       24717c64 5972dcb82ab2dde83376d82b2e3c09ffc
 
+Instantiate an encrypted key "evm" using user-provided decrypted data::
+
+    $ keyctl add encrypted evm "new default user:kmk 32 `cat evm_decrypted_data.blob`" @u
+    794890253
+
+    $ keyctl print 794890253
+    default user:kmk 32 2375725ad57798846a9bbd240de8906f006e66c03af53b1b382d
+    bbc55be2a44616e4959430436dc4f2a7a9659aa60bb4652aeb2120f149ed197c564e0247
+    17c64 5972dcb82ab2dde83376d82b2e3c09ffc
+
   Other uses for trusted and encrypted keys, such as for disk and file encryption
   are anticipated.  In particular the new format 'ecryptfs' has been defined
   in order to use encrypted keys to mount an eCryptfs filesystem.  More details
security/keys/encrypted-keys/encrypted.c +7 -2
index 87432b35d771..baf6fba5e05e 100644
@@ -159,6 +159,7 @@ static int valid_master_desc(const char *new_desc, const char *orig_desc)
    *
    * datablob format:
    * new [<format>] <master-key name> <decrypted data length>
+ * new [<format>] <master-key name> <decrypted data length> <decrypted data>
    * load [<format>] <master-key name> <decrypted data length>
    *     <encrypted iv + data>
    * update <new-master-key name>
@@ -170,7 +171,7 @@ static int valid_master_desc(const char *new_desc, const char *orig_desc)
    */
   static int datablob_parse(char *datablob, const char **format,
                         char **master_desc, char **decrypted_datalen,
-                       char **hex_encoded_iv)
+                       char **hex_encoded_iv, char **decrypted_data)
   {
       substring_t args[MAX_OPT_ARGS];
       int ret = -EINVAL;
@@ -231,6 +232,8 @@ static int datablob_parse(char *datablob, const char **format,
                               "when called from .update method\n", keyword);
                       break;
               }
+             *decrypted_data = strsep(&datablob, " \t");
+
               ret = 0;
               break;
       case Opt_load:
@@ -595,7 +598,8 @@ static int derived_key_decrypt(struct encrypted_key_payload *epayload,
   static struct encrypted_key_payload *encrypted_key_alloc(struct key *key,
                                                        const char *format,
                                                        const char *master_desc,
-                                                      const char *datalen)
+                                                      const char *datalen,
+                                                      const char *decrypted_data)
   {
       struct encrypted_key_payload *epayload = NULL;
       unsigned short datablob_len;
@@ -604,6 +608,7 @@ static struct encrypted_key_payload *encrypted_key_alloc(struct key *key,
       unsigned int encrypted_datalen;
       unsigned int format_len;
       long dlen;
+     int i;
       int ret;
 
       ret = kstrtol(datalen, 10, &dlen);
@@ -613,6 +618,20 @@ static struct encrypted_key_payload *encrypted_key_alloc(struct key *key,
       format_len = (!format) ? strlen(key_format_default) : strlen(format);

security/keys/encrypted-keys/encrypted.c static struct encrypted_key_payload *encrypted_key_alloc(struct key *key, · line 628

Quoted 10 lines
security/keys/encrypted-keys/encrypted.c +8 -0
       decrypted_datalen = dlen;
       payload_datalen = decrypted_datalen;
+
+     if (decrypted_data) {
+             if (strlen(decrypted_data) != decrypted_datalen) {
+                     pr_err("encrypted key: decrypted data provided does not match decrypted data length provided\n");
+                     return ERR_PTR(-EINVAL);
+             }
+             for (i = 0; i < strlen(decrypted_data); i++) {
+                     if (!isalnum(decrypted_data[i])) {

User-provided decrypted data may have special characters, commonly used in passwords or key phrases, apart from alphanumeric. Replace isalnum with !iscntrl() to validate against control characters but allow special characters.

Quoted 6 lines
Quoted 4 lines

Thanks & Regards,

- Nayna

Hi Nayna,
I wonder if we should use isprint() instead?

I have been thinking more about this. Encrypted keys documentation says, "All user level blobs, are displayed and loaded in hex ascii for convenience."

Should we use a similar hex ascii encoding for user-provided data?  Verification would then be isxdigit().

Users should convert their input to hex ascii before passing it to keyctl.

Thanks & Regards,

     - Nayna