Re: [PATCH 1/2] powerpc: Optimized strncpy for POWER9

[Paul E Murphy <murphyp@linux.ibm.com>]

On 9/2/20 8:20 AM, Tulio Magno Quites Machado Filho wrote:
> Raphael Moreira Zinsly via Libc-alpha <libc-alpha@sourceware.org> writes:
> 
>> diff --git a/sysdeps/powerpc/powerpc64/le/power9/strncpy.S b/sysdeps/powerpc/powerpc64/le/power9/strncpy.S
>> new file mode 100644
>> index 0000000000..cde68384d4
>> --- /dev/null
>> +++ b/sysdeps/powerpc/powerpc64/le/power9/strncpy.S
>> @@ -0,0 +1,276 @@
>> ...
>> +/* Implements the function
>> +
>> +   char * [r3] strncpy (char *dest [r3], const char *src [r4], size_t n [r5])
>> +
>> +   The implementation can load bytes past a null terminator, but only
>> +   up to the next 16B boundary, so it never crosses a page.  */
>> +
>> +.machine power9
> 
> I don't think Binutils 2.26 supports .machine power9.  Likewise for all P9
> instructions.  However, current glibc is expected to work with Binutils 2.26
> (ppc64le), i.e. builds with Binutils 2.26 should not fail.
> 
> So, we either need to change this code (e.g. similar to strcmp) or we need
> to bump the Binutils requirements.
> The last time Binutils requirements was bumped was in 2017, so I think it's safe
> to do this now.
> 
> Let me prepare a patch proposing this.

There are at least 5 uses of .machine power9 throughout glibc today.  I 
agree with bumping at least the ppc64le requirements to match.