Date: Fri, 28 Dec 2018 10:21:01 -0800 From: Steve Kargl <sgk@troutmask.apl.washington.edu> To: Justin Hibbits <chmeeedalf@gmail.com> Cc: freebsd-numerics@freebsd.org Subject: Re: Expose all math *l functions as aliases for the * functions when long double == double Message-ID: <20181228182101.GA26033@troutmask.apl.washington.edu> In-Reply-To: <20181228090036.3ce73ee5@ralga.knownspace> References: <20181227160937.51cfe9a5@ralga.knownspace> <20181227221523.GA5197@troutmask.apl.washington.edu> <20181228090036.3ce73ee5@ralga.knownspace>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Dec 28, 2018 at 09:00:36AM -0600, Justin Hibbits wrote: > On Thu, 27 Dec 2018 14:15:23 -0800 > Steve Kargl <sgk@troutmask.apl.washington.edu> wrote: > > > On Thu, Dec 27, 2018 at 04:09:37PM -0600, Justin Hibbits wrote: > > > When building editors/calligra on powerpc64 it failed with undefined > > > references to erfl and erfcl. Some digging it turns out that these > > > functions aren't being exposed as weak symbols in libm on platforms > > > where long double is 64 bits, because LDBL_MANT_DIG is not defined > > > in the files defining these functions. The attached patch fixes > > > that case. It's only been compile-tested for powerpc64, but should > > > have no negative impact on any other platform. > > > > > > - Justin > > > > Looks OK to me. > > > > Thanks Steve. I figured there would be no issue, but since this is > listed in MAINTAINERS, better to avoid a possibility of post-commit > backlash. > I'm actually surprised that it took 7 years, 9 months to discover the issue with cbrtl() and 4 years, 5 months with erfl(). -- Steve
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20181228182101.GA26033>