This is the mail archive of the libc-alpha@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH 1/9] ldbl-128ibm-compat: Create libm-alias-float128.h


On Wed, 6 Jun 2018, Tulio Magno Quites Machado Filho wrote:

> +    __f32addieee128;
> +    __f32divieee128;
> +    __f32mulieee128;
> +    __f32subieee128;
> +    __f64addieee128;
> +    __f64divieee128;
> +    __f64mulieee128;
> +    __f64subieee128;
> +    __f64xaddieee128;
> +    __f64xdivieee128;
> +    __f64xmulieee128;
> +    __f64xsubieee128;

The __f64x* functions shouldn't be needed here.

The __f32* and __f64* ones seem appropriate (for faddl / daddl etc. 
redirection), but in patch 8 you appear to be redirecting instead to the 
f32* / f64* functions (only reserved in TS 18661-3, not in TS 18661-1).

> +    __gamma_productieee128;

__gamma_product is an internal part of libm, not a public exported 
function.  You should start from the *f128 functions that are actually 
exported from libm, and justify any additions to that set carefully 
(generally as functions that are part of the long double API and so need 
*ieee128 versions).

> +    __lgamma_negieee128;
> +    __lgamma_productieee128;

Again, these are internal implementation details, not exported functions.

> +    __x2y2m1ieee128;

Likewise.

-- 
Joseph S. Myers
joseph@codesourcery.com


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]