Hi Ard, On Thu, Apr 17, 2025 at 2:58 PM Ard Biesheuvel <ardb@xxxxxxxxxx> wrote: > > (cc Eric) > > On Thu, 17 Apr 2025 at 08:49, Qingfang Deng <dqfext@xxxxxxxxx> wrote: > > > > From: Qingfang Deng <qingfang.deng@xxxxxxxxxxxxxxx> > > > > Add a scalar implementation of GHASH for RISC-V using the Zbc (carry-less > > multiplication) and Zbb (bit-manipulation) extensions. This implementation > > is adapted from OpenSSL but rewritten in plain C for clarity. > > > > Unlike the OpenSSL one that rely on bit-reflection of the data, this > > version uses a pre-computed (reflected and multiplied) key, inspired by > > the approach used in Intel's CLMUL driver, to avoid reflections during > > runtime. > > > > Signed-off-by: Qingfang Deng <qingfang.deng@xxxxxxxxxxxxxxx> > > What is the use case for this? AIUI, the scalar AES instructions were > never implemented by anyone, so how do you expect this to be used in > practice? The use case _is_ AES-GCM, as you mentioned. Without this, computing GHASH can take a considerable amount of CPU time (monitored by perf). > ... > > +static __always_inline __uint128_t get_unaligned_be128(const u8 *p) > > +{ > > + __uint128_t val; > > +#ifdef CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS > > CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS means that get_unaligned_xxx() > helpers are cheap. Casting a void* to an aligned type is still UB as > per the C standard. Technically an unaligned access is UB but this pattern is widely used in networking code. > > So better to drop the #ifdef entirely, and just use the > get_unaligned_be64() helpers for both cases. Currently those helpers won't generate rev8 instructions, even if HAVE_EFFICIENT_UNALIGNED_ACCESS and RISCV_ISA_ZBB is set, so I have to implement my own version of this to reduce the number of instructions, and to align with the original OpenSSL implementation. > > (same below) > > Also, do you need to test for int128 support? Or is that guaranteed > for all compilers that are supported by the RISC-V port? I believe int128 support is available for all 64-bit targets.