This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [09/12] Add fesetexcept: powerpc
On 08/12/2016 11:30 AM, Joseph Myers wrote:
> On Fri, 12 Aug 2016, Paul E. Murphy wrote:
>
>>> + n.l = (u.l
>>> + | (excepts & FPSCR_STICKY_BITS)
>>> + | (excepts >> ((31 - FPSCR_VX) - (31 - FPSCR_VXSOFT))
>>> + & FE_INVALID_SOFTWARE));
>>
>> This should also preserve a comment about transforming
>> FE_INVALID into FE_INVALID_SOFTWARE. Otherwise, this patch
>> looks OK to me.
>
> I don't see such a comment in sysdeps/powerpc/fpu/fsetexcptflg.c to
> preserve (all these function implementations follow those of
> fesetexceptflag). I did preserve the comment "Deal with
> FE_INVALID_SOFTWARE not being implemented on some chips." (while trying
> not to duplicate all the completely routine comments in various
> fesetexceptflag implementations that say no more than repeating the plain
> semantics of individual lines of C code).
>
I don't like the naked constants. I am not faulting the patch for them.
But in trying to understand them, I had to do some digging. A similar
snippet shows up in three places in ppc code. fpu/fraiseexcpt.c
seems to contain the only useful comment. Both clear and succinct.
It would be nice to clean up the usage, but such should be done in a
separate patch, and needn't by you.