On Do, 2022-03-17 at 10:10 +1200, Herbert Xu wrote: > On Wed, Mar 16, 2022 at 08:23:24AM +0100, Ard Biesheuvel wrote: > > > > According to the bisect log in the other thread, > > adad556efcdd42a1d9e060cb is the culprit, which does not seem > > surprising, at is would result in the SIMD skcipher being > > encapsulated > > to not be available yet when the SIMD helper tries to take a > > reference > > to it. > > It's supposed to work because any use of an algorithm prior to > the tests starting will automatically trigger the test right away. > I confirmed this by booting the kernel in qemu with AES_ARM_BS=y > and it successfully registered those algorithms and passed the > self-tests. > > Someone else has already sent me a complete kconfig file which > hopefully should reproduce the crash for me. It seems you already did. I can confirm that [1] fixes the crash for me, as I had CRYPTO_CBC disabled. After enabling it, aes_init() doesn't hit the error path anymore, hiding the BUG() again. [1] https://lore.kernel.org/all/YjJq0RLIHvN7YWaT@xxxxxxxxxxxxxxxxxxx/ regards Philipp
