[PATCH v2 0/1] pkey: Fix for PKEY_VERIFYPROTK ioctl

[PATCH v2 0/1] pkey: Fix for PKEY_VERIFYPROTK ioctl

[Holger Dengler]

The PKEY_VERIFYPROTK ioctl is used to verify protected key blobs. The
verification is mainly done y the called handler implementations. The
following patch 1/1 removes the (broken) length check in the generic API
code. The deep-inspection of the key blob is handler-specific.

The v1 of this series has fixed the length calculation for the keysize, so
that the correct keytype can be derived from it. But this is a violation of
the layering in pkey, where only the handlers should have a deep knowledge
of the key blobs. Therefore, v2 now removes the keysize calculation and the
keytype check, as it is always done by the handler.

The v2 has been rebased to the current master, as it adds a missing length
check for the ioctl request structure. This addresses one comment of the
Sashiko AI review.

Changes since v1:
- Remove the keytype check (instead of fix the length calculation)
- Add stable tag
- Rebase to current master

The patch applies to Linus' master branch.

Holger Dengler (1):
  pkey: Fix for PKEY_VERIFYPROTK iotl

 drivers/s390/crypto/pkey_api.c | 11 +----------
 1 file changed, 1 insertion(+), 10 deletions(-)


base-commit: 502d801f0ab03e4f32f9a33d203154ce84887921
-- 
2.54.0

[PATCH v2 1/1] pkey: Fix for PKEY_VERIFYPROTK iotl

[Holger Dengler]

Remove the keytype calculation based on key length. The verification,
including the keytype check, is the responsibility of the pkey
handler.

The removal also fixes a bug in the key-length calculation, which is
now no longer required.

Cc: stable@kernel.org # 6.12+
Fixes: 8fcc231ce3be ("s390/pkey: Introduce pkey base with handler registry and handler modules")
Signed-off-by: Holger Dengler <dengler@linux.ibm.com>
---
 drivers/s390/crypto/pkey_api.c | 11 +----------
 1 file changed, 1 insertion(+), 10 deletions(-)

diff --git a/drivers/s390/crypto/pkey_api.c b/drivers/s390/crypto/pkey_api.c
index 28e1007005f2..5d8f63f390a8 100644
--- a/drivers/s390/crypto/pkey_api.c
+++ b/drivers/s390/crypto/pkey_api.c
@@ -327,7 +327,6 @@ static int pkey_ioctl_verifyprotk(struct pkey_verifyprotk __user *uvp)
 {
 	struct pkey_verifyprotk kvp;
 	struct protaeskeytoken *t;
-	u32 keytype;
 	u8 *tmpbuf;
 	int rc;
 
@@ -341,14 +340,6 @@ static int pkey_ioctl_verifyprotk(struct pkey_verifyprotk __user *uvp)
 		return -EINVAL;
 	}
 
-	keytype = pkey_aes_bitsize_to_keytype(8 * kvp.protkey.len);
-	if (!keytype) {
-		PKEY_DBF_ERR("%s unknown/unsupported protkey length %u\n",
-			     __func__, kvp.protkey.len);
-		memzero_explicit(&kvp, sizeof(kvp));
-		return -EINVAL;
-	}
-
 	/* build a 'protected key token' from the raw protected key */
 	tmpbuf = kzalloc(sizeof(*t), GFP_KERNEL);
 	if (!tmpbuf) {
@@ -358,7 +349,7 @@ static int pkey_ioctl_verifyprotk(struct pkey_verifyprotk __user *uvp)
 	t = (struct protaeskeytoken *)tmpbuf;
 	t->type = TOKTYPE_NON_CCA;
 	t->version = TOKVER_PROTECTED_KEY;
-	t->keytype = keytype;
+	t->keytype = kvp.protkey.type;
 	t->len = kvp.protkey.len;
 	memcpy(t->protkey, kvp.protkey.protkey, kvp.protkey.len);
 
-- 
2.54.0

Re: [PATCH v2 1/1] pkey: Fix for PKEY_VERIFYPROTK iotl

[Alexander Gordeev]

On Tue, Jun 23, 2026 at 12:20:16PM +0200, Holger Dengler wrote:
> Remove the keytype calculation based on key length. The verification,
> including the keytype check, is the responsibility of the pkey
> handler.
> 
> The removal also fixes a bug in the key-length calculation, which is
> now no longer required.
> 
> Cc: stable@kernel.org # 6.12+
> Fixes: 8fcc231ce3be ("s390/pkey: Introduce pkey base with handler registry and handler modules")
> Signed-off-by: Holger Dengler <dengler@linux.ibm.com>
> ---
>  drivers/s390/crypto/pkey_api.c | 11 +----------
>  1 file changed, 1 insertion(+), 10 deletions(-)

The patch description is too indescriptive.
Could you please suggest a more specific one?

Thanks!

Re: [PATCH v2 1/1] pkey: Fix for PKEY_VERIFYPROTK iotl

[Holger Dengler]

On 6/23/26 13:13, Alexander Gordeev wrote:
> On Tue, Jun 23, 2026 at 12:20:16PM +0200, Holger Dengler wrote:
>> Remove the keytype calculation based on key length. The verification,
>> including the keytype check, is the responsibility of the pkey
>> handler.
>>
>> The removal also fixes a bug in the key-length calculation, which is
>> now no longer required.
>>
>> Cc: stable@kernel.org # 6.12+
>> Fixes: 8fcc231ce3be ("s390/pkey: Introduce pkey base with handler registry and handler modules")
>> Signed-off-by: Holger Dengler <dengler@linux.ibm.com>
>> ---
>>   drivers/s390/crypto/pkey_api.c | 11 +----------
>>   1 file changed, 1 insertion(+), 10 deletions(-)
> 
> The patch description is too indescriptive.
> Could you please suggest a more specific one?

Sure. What about the following:
---
The PKEY_VERIFYPROTK ioctl takes data from user-space and verifies the 
contained protected key. While checking the integrity of the ioctl 
request structure is the responsibility of the generic pkey_api code, 
the verification of the contained protected key is the responsibility of 
the pkey handler.

The keytype verification (based on the calculated bitsize of the key) is 
part of the protected key verification and therefore the responsibility 
of the pkey handler (which already verifies it). Therefore the keytype 
verification is removed from the generic pkey_api code.

As the calculation of the key bitsize is currently wrong, the removal of 
the keytype check in pkey_api also removes this wrong calculation. For 
this reason, the commit is flagged with the Fixes: tag.
---

> 
> Thanks!

-- 
Mit freundlichen Grüßen / Kind regards
Holger Dengler

Re: [PATCH v2 1/1] pkey: Fix for PKEY_VERIFYPROTK iotl

[Ingo Franzki]

On 23.06.2026 12:20, Holger Dengler wrote:
> Remove the keytype calculation based on key length. The verification,
> including the keytype check, is the responsibility of the pkey
> handler.
> 
> The removal also fixes a bug in the key-length calculation, which is
> now no longer required.
> 
> Cc: stable@kernel.org # 6.12+
> Fixes: 8fcc231ce3be ("s390/pkey: Introduce pkey base with handler registry and handler modules")
> Signed-off-by: Holger Dengler <dengler@linux.ibm.com>
> ---
>  drivers/s390/crypto/pkey_api.c | 11 +----------
>  1 file changed, 1 insertion(+), 10 deletions(-)
> 
> diff --git a/drivers/s390/crypto/pkey_api.c b/drivers/s390/crypto/pkey_api.c
> index 28e1007005f2..5d8f63f390a8 100644
> --- a/drivers/s390/crypto/pkey_api.c
> +++ b/drivers/s390/crypto/pkey_api.c
> @@ -327,7 +327,6 @@ static int pkey_ioctl_verifyprotk(struct pkey_verifyprotk __user *uvp)
>  {
>  	struct pkey_verifyprotk kvp;
>  	struct protaeskeytoken *t;
> -	u32 keytype;
>  	u8 *tmpbuf;
>  	int rc;
>  
> @@ -341,14 +340,6 @@ static int pkey_ioctl_verifyprotk(struct pkey_verifyprotk __user *uvp)
>  		return -EINVAL;
>  	}
>  
> -	keytype = pkey_aes_bitsize_to_keytype(8 * kvp.protkey.len);
> -	if (!keytype) {
> -		PKEY_DBF_ERR("%s unknown/unsupported protkey length %u\n",
> -			     __func__, kvp.protkey.len);
> -		memzero_explicit(&kvp, sizeof(kvp));
> -		return -EINVAL;
> -	}
> -
>  	/* build a 'protected key token' from the raw protected key */
>  	tmpbuf = kzalloc(sizeof(*t), GFP_KERNEL);
>  	if (!tmpbuf) {
> @@ -358,7 +349,7 @@ static int pkey_ioctl_verifyprotk(struct pkey_verifyprotk __user *uvp)
>  	t = (struct protaeskeytoken *)tmpbuf;
>  	t->type = TOKTYPE_NON_CCA;
>  	t->version = TOKVER_PROTECTED_KEY;
> -	t->keytype = keytype;
> +	t->keytype = kvp.protkey.type;
>  	t->len = kvp.protkey.len;
>  	memcpy(t->protkey, kvp.protkey.protkey, kvp.protkey.len);
>  

Reviewed-by: Ingo Franzki <ifranzki@linux.ibm.com>

I like your new commit description proposal! 

-- 
Ingo Franzki
eMail: ifranzki@linux.ibm.com  
Linux on IBM Z Development
IBM Campus 1, 71139 Ehningen, Germany

IBM Deutschland Research & Development GmbH
Vorsitzender des Aufsichtsrats: Wolfgang Wendt
Geschäftsführung: David Faller
Sitz der Gesellschaft: Ehningen / Registergericht: Amtsgericht Stuttgart, HRB 243294
IBM DATA Privacy Statement: https://www.ibm.com/privacy/us/en/

Re: [PATCH v2 1/1] pkey: Fix for PKEY_VERIFYPROTK iotl

[Alexander Gordeev]

On Tue, Jun 23, 2026 at 03:25:51PM +0200, Holger Dengler wrote:
> > The patch description is too indescriptive.
> > Could you please suggest a more specific one?
> 
> Sure. What about the following:
...

I rather meant the Subject Line description, but at least Ingo
likes the new commit message better :) Could you send v3, please?

Thanks!

> Mit freundlichen Grüßen / Kind regards
> Holger Dengler