On Fri, Nov 28, 2025 at 8:17 AM Xi Ruoyao <[email protected]> wrote: > > The current wording is misleading because even the combination of > -frounding-math and -ftrapping-math still doesn't provide the semantics > required by FENV_ACCESS on. It can confuse people and prompt them to > submit more duplicates of PR 34678. Let's make it clear that we've not > implemented FENV_ACCESS on yet even with -frounding-math > -ftrapping-math.
I think this is an improvement over what we have, so OK unless somebody suggests otherwise over the weekend. Thanks, Richard. > gcc/ > > PR middle-end/34678 > * doc/invoke.texi (-fno-trapping-math): Stop mentioning > FENV_ACCESS. > (-frounding-math): Likewise. > * doc/implement-c.texi: Note that -ftrapping-math > -frounding-math is not enough for "FENV_ACCESS on" semantics. > --- > > Superseds https://gcc.gnu.org/pipermail/gcc-patches/2025-November/702123.html. > > gcc/doc/implement-c.texi | 7 ++++--- > gcc/doc/invoke.texi | 9 --------- > 2 files changed, 4 insertions(+), 12 deletions(-) > > diff --git a/gcc/doc/implement-c.texi b/gcc/doc/implement-c.texi > index 8f0c2ff4bf3..c2c04445248 100644 > --- a/gcc/doc/implement-c.texi > +++ b/gcc/doc/implement-c.texi > @@ -373,9 +373,10 @@ This is subject to change. > @cite{The default state for the @code{FENV_ACCESS} pragma (C99 and C11 > 7.6.1, C23 7.6.2).} > > -This pragma is not implemented, but the default is to ``off'' unless > -@option{-frounding-math} is used and @option{-fno-trapping-math} is not > -in which case it is ``on''. > +This pragma is not implemented and GCC always behave as if it's set to > +``off''; even if both @option{-frounding-math} and @option{-ftrapping-math} > +are effective, GCC still does @emph{not} provide all the semantics required > +by setting this pragma to ``on''. > > @item > @cite{Additional floating-point exceptions, rounding modes, environments, > diff --git a/gcc/doc/invoke.texi b/gcc/doc/invoke.texi > index b1e6d703702..10d0ede5384 100644 > --- a/gcc/doc/invoke.texi > +++ b/gcc/doc/invoke.texi > @@ -16126,11 +16126,6 @@ for math functions. > > The default is @option{-ftrapping-math}. > > -Future versions of GCC may provide finer control of this setting > -using C99's @code{FENV_ACCESS} pragma. This command-line option > -will be used along with @option{-frounding-math} to specify the > -default state for @code{FENV_ACCESS}. > - > @opindex frounding-math > @item -frounding-math > Disable transformations and optimizations that assume default floating-point > @@ -16146,10 +16141,6 @@ The default is @option{-fno-rounding-math}. > > This option is experimental and does not currently guarantee to > disable all GCC optimizations that are affected by rounding mode. > -Future versions of GCC may provide finer control of this setting > -using C99's @code{FENV_ACCESS} pragma. This command-line option > -will be used along with @option{-ftrapping-math} to specify the > -default state for @code{FENV_ACCESS}. > > @opindex fsignaling-nans > @item -fsignaling-nans > -- > 2.52.0 >
