On 2025/12/10 上午9:45, Eric Biggers wrote:
On Wed, Dec 10, 2025 at 09:36:06AM +0800, Bibo Mao wrote:
On 2025/12/10 上午7:25, Eric Biggers wrote:
On Tue, Dec 09, 2025 at 10:22:58AM +0800, Bibo Mao wrote:
ECB AES also is added here, its ivsize is zero and name is different
compared with CBC AES algo.
What is the use case for this feature?Currently qemu builtin backend and
openssl afalg only support CBC AES,
it depends on modified qemu and openssl to test this.
Maybe this patch adding ECB AES algo can be skipped now, it is just an
example, the final target is to add SM4 cipher.
There's no need to add useless features. The title of your patchset is
"crypto: virtio: Add ecb aes algo support". So it sounds like the main
point of your patchset is to add a useless feature? If there are
actually unrelated fixes you want, you should send those separately.
yes, will change title of cover letter and remove this patch in next time.
As for SM4 support (which mode?), if you really want that (you
shouldn't), why not use the existing CPU accelerated implementation?
The hardware supports SM4 ECB/CBC/CTL three modes, it depends on the
detail application scenery.
I just notice that openssl removes engines support in recent. The
purpose of use HW accel is that it is faster and can save CPU resource.
However it actually brings some troubles with applications on different
HW platforms, I think HW crypto accel can be used for kernel and some
key user applications, it is not suitable for all general applications
for the present.
Regards
Bibo Mao
- Eric
