On Fri, Jun 27, 2025 at 11:56:49AM -0700, Eric Biggers wrote:
Commit 88c02b3f79a6 ("s390/sha3: Support sha3 performance enhancements") added the field s390_sha_ctx::first_message_part and made it be used by s390_sha_update_blocks(). At the time, s390_sha_update_blocks() was used by all the s390 SHA-1, SHA-2, and SHA-3 algorithms. However, only the initialization functions for SHA-3 were updated, leaving SHA-1 and SHA-2 using first_message_part uninitialized.
This could cause e.g. CPACF_KIMD_SHA_512 | CPACF_KIMD_NIP to be used instead of just CPACF_KIMD_NIP. It's unclear why this didn't cause a problem earlier; this bug was found only when UBSAN detected the uninitialized boolean. Perhaps the CPU ignores CPACF_KIMD_NIP for SHA-1 and SHA-2. Regardless, let's fix this. For now just initialize to false, i.e. don't try to "optimize" the SHA state initialization.
Note: in 6.16, we need to patch SHA-1, SHA-384, and SHA-512. In 6.15 and earlier, we'll also need to patch SHA-224 and SHA-256, as they hadn't yet been librarified (which incidentally fixed this bug).
Fixes: 88c02b3f79a6 ("s390/sha3: Support sha3 performance enhancements") Cc: stable@vger.kernel.org Reported-by: Ingo Franzki ifranzki@linux.ibm.com Closes: https://lore.kernel.org/r/12740696-595c-4604-873e-aefe8b405fbf@linux.ibm.com Signed-off-by: Eric Biggers ebiggers@kernel.org
This is targeting 6.16. I'd prefer to take this through libcrypto-fixes, since the librarification work is also touching this area. But let me know if there's a preference for the crypto tree or the s390 tree instead.
arch/s390/crypto/sha1_s390.c | 1 + arch/s390/crypto/sha512_s390.c | 2 ++ 2 files changed, 3 insertions(+)
Routing this via libcrypto-fixes is fine.
Acked-by: Heiko Carstens hca@linux.ibm.com