Skip to contentSkip to conversation

apparmor: use SHA-256 library API instead of crypto_shash API

PATCH In review Eric Biggers 3 participants 11 messages Apr 28, 2025 – Jun 30, 2025

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 apparmor: use SHA-256 library API instead of crypto_shash API

From: Eric Biggers <[email protected]>

This user of SHA-256 does not support any other algorithm, so the crypto_shash abstraction provides no value. Just use the SHA-256 library API instead, which is much simpler and easier to use.

---

This patch is targeting the apparmor tree for 6.16.

 security/apparmor/Kconfig  |  3 +-
 security/apparmor/crypto.c | 85 ++++++--------------------------------
 2 files changed, 13 insertions(+), 75 deletions(-)
security/apparmor/Kconfig +1 -2
index 64cc3044a42ce..1e3bd44643dac 100644
@@ -57,12 +57,11 @@ config SECURITY_APPARMOR_INTROSPECT_POLICY
 	  cpu is paramount.
 
 config SECURITY_APPARMOR_HASH
 	bool "Enable introspection of sha256 hashes for loaded profiles"
 	depends on SECURITY_APPARMOR_INTROSPECT_POLICY
-	select CRYPTO
-	select CRYPTO_SHA256
+	select CRYPTO_LIB_SHA256
 	default y
 	help
 	  This option selects whether introspection of loaded policy
 	  hashes is available to userspace via the apparmor
 	  filesystem. This option provides a light weight means of
security/apparmor/crypto.c +12 -73

127 lines not on this page · raw message

base-commit: 33035b665157558254b3c21c3f049fd728e72368

Signature · 1 line
2.49.0
#1 lore raw
Quoted 6 lines · Eric Biggers

From: Eric Biggers <[email protected]>

This user of SHA-256 does not support any other algorithm, so the crypto_shash abstraction provides no value. Just use the SHA-256 library API instead, which is much simpler and easier to use.

Quoted 8 lines

---

This patch is targeting the apparmor tree for 6.16.

 security/apparmor/Kconfig  |  3 +-
 security/apparmor/crypto.c | 85 ++++++--------------------------------
 2 files changed, 13 insertions(+), 75 deletions(-)

Any interest in taking this patch through the apparmor or security trees?

- Eric

maintainer
#2 lore raw
Quoted 10 lines · Eric Biggers
Quoted 9 lines · Eric Biggers

From: Eric Biggers <[email protected]>

This user of SHA-256 does not support any other algorithm, so the crypto_shash abstraction provides no value. Just use the SHA-256 library API instead, which is much simpler and easier to use.

---

Quoted 7 lines
Quoted 5 lines

This patch is targeting the apparmor tree for 6.16.

 security/apparmor/Kconfig  |  3 +-
 security/apparmor/crypto.c | 85 ++++++--------------------------------
 2 files changed, 13 insertions(+), 75 deletions(-)

Any interest in taking this patch through the apparmor or security trees?

Something like this would need to go through the AppArmor tree. As a FYI, the AppArmor devs are fairly busy at the moment so it may take a bit for them to get around to this.

Signature · 1 line
paul-moore.com
maintainer
#3 lore raw
Quoted 17 lines · Paul Moore
Quoted 16 lines · Eric Biggers
Quoted 14 lines · Eric Biggers

From: Eric Biggers <[email protected]>

This user of SHA-256 does not support any other algorithm, so the crypto_shash abstraction provides no value. Just use the SHA-256 library API instead, which is much simpler and easier to use.

---

This patch is targeting the apparmor tree for 6.16.

  security/apparmor/Kconfig  |  3 +-
  security/apparmor/crypto.c | 85 ++++++--------------------------------
  2 files changed, 13 insertions(+), 75 deletions(-)
Quoted 6 lines
Quoted 1 line

Any interest in taking this patch through the apparmor or security trees?

Something like this would need to go through the AppArmor tree. As a FYI, the AppArmor devs are fairly busy at the moment so it may take a bit for them to get around to this.

I am going to see how much of the backlog I can get through while traveling replies might get batch because I will be mostly off line but hopefully I can deal with most of it this weekend.

maintainer
#4 lore raw
In reply to Eric Biggers
Quoted 10 lines · Eric Biggers
Quoted 9 lines · Eric Biggers

From: Eric Biggers <[email protected]>

This user of SHA-256 does not support any other algorithm, so the crypto_shash abstraction provides no value. Just use the SHA-256 library API instead, which is much simpler and easier to use.

---

Quoted 8 lines
Quoted 5 lines

This patch is targeting the apparmor tree for 6.16.

  security/apparmor/Kconfig  |  3 +-
  security/apparmor/crypto.c | 85 ++++++--------------------------------
  2 files changed, 13 insertions(+), 75 deletions(-)

Any interest in taking this patch through the apparmor or security trees?

I can take it through my tree

#5 lore raw
Quoted 13 lines · John Johansen
Quoted 12 lines · Eric Biggers
Quoted 11 lines · Eric Biggers

From: Eric Biggers <[email protected]>

This user of SHA-256 does not support any other algorithm, so the crypto_shash abstraction provides no value. Just use the SHA-256 library API instead, which is much simpler and easier to use.

---

This patch is targeting the apparmor tree for 6.16.

Quoted 7 lines
Quoted 6 lines
Quoted 3 lines
  security/apparmor/Kconfig  |  3 +-
  security/apparmor/crypto.c | 85 ++++++--------------------------------
  2 files changed, 13 insertions(+), 75 deletions(-)

Any interest in taking this patch through the apparmor or security trees?

I can take it through my tree

Thanks! I notice this isn't in v6.16-rc1. Do you have a pull request planned?

- Eric

maintainer
#6 lore raw
Quoted 18 lines · Eric Biggers
Quoted 17 lines · John Johansen
Quoted 16 lines · Eric Biggers
Quoted 14 lines · Eric Biggers

From: Eric Biggers <[email protected]>

This user of SHA-256 does not support any other algorithm, so the crypto_shash abstraction provides no value. Just use the SHA-256 library API instead, which is much simpler and easier to use.

---

This patch is targeting the apparmor tree for 6.16.

   security/apparmor/Kconfig  |  3 +-
   security/apparmor/crypto.c | 85 ++++++--------------------------------
   2 files changed, 13 insertions(+), 75 deletions(-)
Quoted 6 lines
Quoted 3 lines
Quoted 2 lines

Any interest in taking this patch through the apparmor or security trees?

I can take it through my tree

Thanks! I notice this isn't in v6.16-rc1. Do you have a pull request planned?

Hey Eric,

sorry I have been sick and didn't get a 6.16 pull request out. I am slowly trying to dig my way out of the backlog, which is several weeks deeo. I might get together a small PR of bug fixes before the 6.17 merge window but the bulk of what is in apparmor-next will be waiting to merge in 6.17 now.

#7 lore raw
Quoted 26 lines · John Johansen
Quoted 18 lines · Eric Biggers
Quoted 17 lines · John Johansen
Quoted 16 lines · Eric Biggers
Quoted 14 lines · Eric Biggers

From: Eric Biggers <[email protected]>

This user of SHA-256 does not support any other algorithm, so the crypto_shash abstraction provides no value. Just use the SHA-256 library API instead, which is much simpler and easier to use.

---

This patch is targeting the apparmor tree for 6.16.

   security/apparmor/Kconfig  |  3 +-
   security/apparmor/crypto.c | 85 ++++++--------------------------------
   2 files changed, 13 insertions(+), 75 deletions(-)
Quoted 6 lines
Quoted 3 lines
Quoted 2 lines

Any interest in taking this patch through the apparmor or security trees?

I can take it through my tree

Thanks! I notice this isn't in v6.16-rc1. Do you have a pull request planned?

Quoted 6 lines

Hey Eric,

sorry I have been sick and didn't get a 6.16 pull request out. I am slowly trying to dig my way out of the backlog, which is several weeks deeo. I might get together a small PR of bug fixes before the 6.17 merge window but the bulk of what is in apparmor-next will be waiting to merge in 6.17 now.

Hope you're feeling better! Actually, would you mind if instead I took this patch (with your ack) through the libcrypto-next tree for 6.17? Otherwise there will be a silent merge conflict after I apply
https://lore.kernel.org/r/[email protected]/

- Eric

maintainer
#8 lore raw
Quoted 27 lines · Eric Biggers
Quoted 26 lines · John Johansen
Quoted 24 lines · Eric Biggers
Quoted 13 lines · John Johansen
Quoted 12 lines · Eric Biggers
Quoted 11 lines · Eric Biggers

From: Eric Biggers <[email protected]>

This user of SHA-256 does not support any other algorithm, so the crypto_shash abstraction provides no value. Just use the SHA-256 library API instead, which is much simpler and easier to use.

---

This patch is targeting the apparmor tree for 6.16.

Quoted 7 lines
Quoted 6 lines
Quoted 3 lines
    security/apparmor/Kconfig  |  3 +-
    security/apparmor/crypto.c | 85 ++++++--------------------------------
    2 files changed, 13 insertions(+), 75 deletions(-)

Any interest in taking this patch through the apparmor or security trees?

I can take it through my tree

Thanks! I notice this isn't in v6.16-rc1. Do you have a pull request planned?

Quoted 8 lines
Quoted 6 lines

Hey Eric,

sorry I have been sick and didn't get a 6.16 pull request out. I am slowly trying to dig my way out of the backlog, which is several weeks deeo. I might get together a small PR of bug fixes before the 6.17 merge window but the bulk of what is in apparmor-next will be waiting to merge in 6.17 now.

Hope you're feeling better! Actually, would you mind if instead I took this

I lot, though still generally tired/low on energy

Quoted 4 lines

patch (with your ack) through the libcrypto-next tree for 6.17? Otherwise there will be a silent merge conflict after I apply
https://lore.kernel.org/r/[email protected]/

Avoiding a merge conflict? You have my ACK and blessing I will pull it out of the apparmor tree asap

#9 lore raw
Quoted 36 lines · John Johansen
Quoted 27 lines · Eric Biggers
Quoted 26 lines · John Johansen
Quoted 24 lines · Eric Biggers
Quoted 13 lines · John Johansen
Quoted 12 lines · Eric Biggers
Quoted 11 lines · Eric Biggers

From: Eric Biggers <[email protected]>

This user of SHA-256 does not support any other algorithm, so the crypto_shash abstraction provides no value. Just use the SHA-256 library API instead, which is much simpler and easier to use.

---

This patch is targeting the apparmor tree for 6.16.

Quoted 7 lines
Quoted 6 lines
Quoted 3 lines
    security/apparmor/Kconfig  |  3 +-
    security/apparmor/crypto.c | 85 ++++++--------------------------------
    2 files changed, 13 insertions(+), 75 deletions(-)

Any interest in taking this patch through the apparmor or security trees?

I can take it through my tree

Thanks! I notice this isn't in v6.16-rc1. Do you have a pull request planned?

Quoted 8 lines
Quoted 6 lines

Hey Eric,

sorry I have been sick and didn't get a 6.16 pull request out. I am slowly trying to dig my way out of the backlog, which is several weeks deeo. I might get together a small PR of bug fixes before the 6.17 merge window but the bulk of what is in apparmor-next will be waiting to merge in 6.17 now.

Hope you're feeling better! Actually, would you mind if instead I took this

Quoted 8 lines

I lot, though still generally tired/low on energy

Quoted 4 lines

patch (with your ack) through the libcrypto-next tree for 6.17? Otherwise there will be a silent merge conflict after I apply
https://lore.kernel.org/r/[email protected]/

Avoiding a merge conflict? You have my ACK and blessing I will pull it out of the apparmor tree asap

Thanks, let me know once you've dropped it.

- Eric

maintainer Acked-by on v1 1/1
#10 lore raw
In reply to Eric Biggers
Quoted 7 lines · Eric Biggers

From: Eric Biggers <[email protected]>

This user of SHA-256 does not support any other algorithm, so the crypto_shash abstraction provides no value. Just use the SHA-256 library API instead, which is much simpler and easier to use.

Quoted 160 lines

---

This patch is targeting the apparmor tree for 6.16.

  security/apparmor/Kconfig  |  3 +-
  security/apparmor/crypto.c | 85 ++++++--------------------------------
  2 files changed, 13 insertions(+), 75 deletions(-)
security/apparmor/Kconfig +1 -2
index 64cc3044a42ce..1e3bd44643dac 100644
@@ -57,12 +57,11 @@ config SECURITY_APPARMOR_INTROSPECT_POLICY
  	  cpu is paramount.
 
  config SECURITY_APPARMOR_HASH
  	bool "Enable introspection of sha256 hashes for loaded profiles"
  	depends on SECURITY_APPARMOR_INTROSPECT_POLICY
-	select CRYPTO
-	select CRYPTO_SHA256
+	select CRYPTO_LIB_SHA256
  	default y
  	help
  	  This option selects whether introspection of loaded policy
  	  hashes is available to userspace via the apparmor
  	  filesystem. This option provides a light weight means of
security/apparmor/crypto.c +12 -73
index aad486b2fca65..40e17e153f1e5 100644
@@ -9,115 +9,54 @@
   * Fns to provide a checksum of policy that has been loaded this can be
   * compared to userspace policy compiles to check loaded policy is what
   * it should be.
   */
 
-#include <crypto/hash.h>
+#include <crypto/sha2.h>
 
  #include "include/apparmor.h"
  #include "include/crypto.h"
 
-static unsigned int apparmor_hash_size;
-
-static struct crypto_shash *apparmor_tfm;
-
  unsigned int aa_hash_size(void)
  {
-	return apparmor_hash_size;
+	return SHA256_DIGEST_SIZE;
  }
 
  char *aa_calc_hash(void *data, size_t len)
  {
-	SHASH_DESC_ON_STACK(desc, apparmor_tfm);
  	char *hash;
-	int error;
-
-	if (!apparmor_tfm)
-		return NULL;
 
-	hash = kzalloc(apparmor_hash_size, GFP_KERNEL);
+	hash = kzalloc(SHA256_DIGEST_SIZE, GFP_KERNEL);
  	if (!hash)
  		return ERR_PTR(-ENOMEM);
 
-	desc->tfm = apparmor_tfm;
-
-	error = crypto_shash_init(desc);
-	if (error)
-		goto fail;
-	error = crypto_shash_update(desc, (u8 *) data, len);
-	if (error)
-		goto fail;
-	error = crypto_shash_final(desc, hash);
-	if (error)
-		goto fail;
-
+	sha256(data, len, hash);
  	return hash;
-
-fail:
-	kfree(hash);
-
-	return ERR_PTR(error);
  }
 
  int aa_calc_profile_hash(struct aa_profile *profile, u32 version, void *start,
  			 size_t len)
  {
-	SHASH_DESC_ON_STACK(desc, apparmor_tfm);
-	int error;
+	struct sha256_state state;
  	__le32 le32_version = cpu_to_le32(version);
 
  	if (!aa_g_hash_policy)
  		return 0;
 
-	if (!apparmor_tfm)
-		return 0;
-
-	profile->hash = kzalloc(apparmor_hash_size, GFP_KERNEL);
+	profile->hash = kzalloc(SHA256_DIGEST_SIZE, GFP_KERNEL);
  	if (!profile->hash)
  		return -ENOMEM;
 
-	desc->tfm = apparmor_tfm;
-
-	error = crypto_shash_init(desc);
-	if (error)
-		goto fail;
-	error = crypto_shash_update(desc, (u8 *) &le32_version, 4);
-	if (error)
-		goto fail;
-	error = crypto_shash_update(desc, (u8 *) start, len);
-	if (error)
-		goto fail;
-	error = crypto_shash_final(desc, profile->hash);
-	if (error)
-		goto fail;
-
+	sha256_init(&state);
+	sha256_update(&state, (u8 *)&le32_version, 4);
+	sha256_update(&state, (u8 *)start, len);
+	sha256_final(&state, profile->hash);
  	return 0;
-
-fail:
-	kfree(profile->hash);
-	profile->hash = NULL;
-
-	return error;
  }
 
  static int __init init_profile_hash(void)
  {
-	struct crypto_shash *tfm;
-
-	if (!apparmor_initialized)
-		return 0;
-
-	tfm = crypto_alloc_shash("sha256", 0, 0);
-	if (IS_ERR(tfm)) {
-		int error = PTR_ERR(tfm);
-		AA_ERROR("failed to setup profile sha256 hashing: %d\n", error);
-		return error;
-	}
-	apparmor_tfm = tfm;
-	apparmor_hash_size = crypto_shash_digestsize(apparmor_tfm);
-
-	aa_info_message("AppArmor sha256 policy hashing enabled");
-
+	if (apparmor_initialized)
+		aa_info_message("AppArmor sha256 policy hashing enabled");
  	return 0;
  }
-
  late_initcall(init_profile_hash);

base-commit: 33035b665157558254b3c21c3f049fd728e72368