On 04. 12. 24, 13:24, Greg Kroah-Hartman wrote:
On Wed, Dec 04, 2024 at 11:45:05AM +0100, Jiri Slaby wrote:
On 04. 12. 24, 11:34, Greg Kroah-Hartman wrote:
On Wed, Dec 04, 2024 at 11:00:34AM +0100, Jiri Slaby wrote:
Hi,
On 03. 12. 24, 15:36, Greg Kroah-Hartman wrote:
6.12-stable review patch. If anyone has any objections, please let me know.
------------------
From: Danny Tsen <dtsen@xxxxxxxxxxxxx>
[ Upstream commit c954b252dee956d33ee59f594710af28fb3037d9 ]
This patch is to fix an issue when simd is not usable that data mismatch
may occur. The fix is to register algs as SIMD modules so that the
algorithm is excecuted when SIMD instructions is usable. Called
gcm_update() to generate the final digest if needed.
A new module rfc4106(gcm(aes)) is also added.
Fixes: cdcecfd9991f ("crypto: p10-aes-gcm - Glue code for AES/GCM stitched implementation")
Signed-off-by: Danny Tsen <dtsen@xxxxxxxxxxxxx>
Signed-off-by: Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx>
Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
---
arch/powerpc/crypto/aes-gcm-p10-glue.c | 141 +++++++++++++++++++++----
1 file changed, 118 insertions(+), 23 deletions(-)
diff --git a/arch/powerpc/crypto/aes-gcm-p10-glue.c b/arch/powerpc/crypto/aes-gcm-p10-glue.c
index f66ad56e765f0..4a029d2fe06ce 100644
--- a/arch/powerpc/crypto/aes-gcm-p10-glue.c
+++ b/arch/powerpc/crypto/aes-gcm-p10-glue.c
...
@@ -281,6 +295,7 @@ static int p10_aes_gcm_crypt(struct aead_request *req, int enc)
/* Finalize hash */
vsx_begin();
+ gcm_update(gctx->iv, hash->Htable);
Now I get:
ERROR: modpost: "gcm_update" [arch/powerpc/crypto/aes-gcm-p10-crypto.ko]
undefined!
Only this:
commit 7aa747edcb266490f93651dd749c69b7eb8541d9
Author: Danny Tsen <dtsen@xxxxxxxxxxxxx>
Date: Mon Sep 23 09:30:38 2024 -0400
crypto: powerpc/p10-aes-gcm - Re-write AES/GCM stitched implementation
added that function...
Ah, thanks, I'll go drop this patch from everywhere.
OK.
Looking at the queue, it looks like a prereq for un-BROKEN-ing the module in
the next patch:
8b6c1e466eec crypto: powerpc/p10-aes-gcm - Add dependency on
CRYPTO_SIMDand re-enable CRYPTO_AES_GCM_P10
No, I don't see a conflict here. Are you sure you are?
Not sure at all about this crypto stuff. But this failing patch
introduces SIMD and the above 8b6c1e466eec adds a dep to SIMD and makes
the module nonBROKEN at the same time. So I assume the failing one is a
prereq to unbreak the module. Maintainers?
--
js
suse labs
