mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
* [PATCH v2] crypto: aes - Fix undesired override of some optimized AES modes
@ 2026-09-25 20:23 Eric Biggers
  0 siblings, 0 replies; only message in thread
From: Eric Biggers @ 2026-09-25 20:23 UTC (permalink / raw)
  To: linux-crypto
  Cc: linux-kernel, Ard Biesheuvel, Jason A . Donenfeld, Herbert Xu,
	Eric Biggers

The new library APIs for AES encryption modes were wired up to the
traditional crypto API via crypto/aes.c.  However, for now the kernel is
still in a transitional state where various architectures still have
architecture-optimized implementations of AES modes in arch/*/crypto/,
wired up to the traditional crypto API only.  Because of that, the
crypto/aes.c algorithms were given a cra_priority of only 110 to prevent
them from overriding arch/*/crypto/ in the traditional crypto API.

However, because of how the traditional crypto API works, the
cra_priority trick doesn't work in cases where the relevant algorithm
isn't directly implemented by arch/*/crypto/ but rather is provided by a
template instance using other code in arch/*/crypto/.

For example, x86 doesn't have its own "ccm(aes)" but rather relies on
the "ccm" template constructing it from the x86-optimized "ctr(aes)".
The existence of the library-based "ccm(aes)" prevents that, even though
its priority is lower than what the template would produce.

Thus, "ccm(aes)" ends up using the slower single-block AES code.

Therefore, skip wiring up the relevant library-based code to the
traditional crypto API on architectures where this problem can occur, as
determined by what exists in arch/*/crypto/ for each architecture.

This is ugly, but it's also temporary: these conditions will go away as
architecture-optimized implementations of AES modes are migrated into
the library.  But until then, we need to prevent performance regressions
by ensuring that the optimized code continues to be used.

Note: "xts(aes)" is left alone.  Though the "xts" template can use
"ecb(aes)" as an inner algorithm, in practice this isn't very efficient
and a dedicated "xts(aes)" is already provided in all the important
cases anyway.  (This omission is also consistent with the fact that the
library isn't planned to provide a similar ECB-to-XTS "adapter".)

Fixes: 20df21a482aa ("crypto: aes - Add CBC and CBC-CTS support using library")
Fixes: 8ca62072faa1 ("crypto: aes - Add GCM support using library")
Fixes: f70ad727d1d6 ("crypto: aes - Add CCM support using library")
Signed-off-by: Eric Biggers <ebiggers@kernel.org>
---

This patch is intended to be taken through libcrypto-fixes

v2: Fixed PowerPC config option, and resent as standalone patch

 crypto/aes.c | 39 ++++++++++++++++++++++++++++++++++++---
 1 file changed, 36 insertions(+), 3 deletions(-)

diff --git a/crypto/aes.c b/crypto/aes.c
index 94791f481e984..5046b887ac9a0 100644
--- a/crypto/aes.c
+++ b/crypto/aes.c
@@ -637,7 +637,17 @@ static struct skcipher_alg skcipher_algs[] = {
 		.decrypt = crypto_aes_cbc_decrypt,
 	},
 #endif
-#if IS_ENABLED(CONFIG_CRYPTO_CTS)
+#if IS_ENABLED(CONFIG_CRYPTO_CTS) && \
+	/*
+	 * Skip registering this when it might block a "better" implementation
+	 * from being instantiated via the "cts" template wrapping an arch-
+	 * optimized "cbc(aes)" that hasn't yet been migrated into the library.
+	 */ \
+	!(IS_ENABLED(CONFIG_ARM) || \
+	  IS_ENABLED(CONFIG_ARM64) || \
+	  IS_ENABLED(CONFIG_PPC) || \
+	  IS_ENABLED(CONFIG_S390) || \
+	  IS_ENABLED(CONFIG_SPARC))
 	{
 		.base.cra_name = "cts(cbc(aes))",
 		.base.cra_driver_name = "cts-cbc-aes-lib",
@@ -980,7 +990,18 @@ static __maybe_unused int crypto_aes_ccm_decrypt(struct aead_request *req)
 }
 
 static struct aead_alg aead_algs[] = {
-#if IS_ENABLED(CONFIG_CRYPTO_GCM)
+#if IS_ENABLED(CONFIG_CRYPTO_GCM) && \
+	/*
+	 * Skip registering these when they might block "better" implementations
+	 * from being instantiated via the corresponding templates using
+	 * arch-optimized code that hasn't yet been migrated into the library.
+	 */ \
+	!(IS_ENABLED(CONFIG_ARM) || \
+	  IS_ENABLED(CONFIG_ARM64) || \
+	  IS_ENABLED(CONFIG_PPC) || \
+	  IS_ENABLED(CONFIG_RISCV) || \
+	  IS_ENABLED(CONFIG_S390) || \
+	  IS_ENABLED(CONFIG_SPARC))
 	{
 		.base.cra_name = "gcm(aes)",
 		.base.cra_driver_name = "gcm-aes-lib",
@@ -1012,7 +1033,19 @@ static struct aead_alg aead_algs[] = {
 		.chunksize = AES_BLOCK_SIZE,
 	},
 #endif /* CONFIG_CRYPTO_GCM */
-#if IS_ENABLED(CONFIG_CRYPTO_CCM)
+#if IS_ENABLED(CONFIG_CRYPTO_CCM) && \
+	/*
+	 * Skip registering this when it might block a "better" implementation
+	 * from being instantiated via the "ccm" template wrapping an arch-
+	 * optimized "ctr(aes)" that hasn't yet been migrated into the library.
+	 */ \
+	!(IS_ENABLED(CONFIG_ARM) || \
+	  IS_ENABLED(CONFIG_ARM64) || \
+	  IS_ENABLED(CONFIG_PPC) || \
+	  IS_ENABLED(CONFIG_RISCV) || \
+	  IS_ENABLED(CONFIG_S390) || \
+	  IS_ENABLED(CONFIG_SPARC) || \
+	  IS_ENABLED(CONFIG_X86))
 	{
 		.base.cra_name = "ccm(aes)",
 		.base.cra_driver_name = "ccm-aes-lib",

base-commit: 93f51579e7df248780214094418f205253383cc5
-- 
2.55.0


^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2026-09-25 20:24 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-09-25 20:23 [PATCH v2] crypto: aes - Fix undesired override of some optimized AES modes Eric Biggers

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®