Re: [PATCH 2/2] bitops: rotate: Add riscv implementation using Zbb extension

[cp0613@linux.alibaba.com]

On Tue, 1 Jul 2025 14:32:14 -0400, yury.norov@gmail.com wrote:

> > static inline u8 variable_rol8(u8 word, unsigned int shift)
> 
> Now that it has assembler inlines, would it help to add the __pure
> qualifier?

Even if the same parameters are passed in, after comparing the disassembly
with and without the __pure qualifier, it is found that it does not help.

> Please do the following:
> 
> 1. Drop the generic_ror8() and keep only ror/l8()
> 2. Add ror/l16, 34 and 64 tests.
> 3. Adjust the 'loop' so that each subtest will take 1-10 ms on your hw.
> 4. Name the function like test_rorl(), and put it next to the test_fns()
>    in lib/test_bitops.c. Refer the 0a2c6664e56f0 for implementation.
> 5. Prepend the series with the benchmark patch, just as with const-eval
>    one, and report before/after for your series. 
> 
> > > Please find attached a test for compile-time ror/rol evaluation.
> > > Please consider prepending your series with it.
> > 
> > Okay, I'll bring it to the next series.
> 
> Still waiting for bloat-o-meter results.

Ok, in the next series I will provide a patch for the benchmark, and then
prepend the const-eval and benchmark patches, and also provide the results
of the bloat-o-meter.

Thank you very much,
Pei