| Message ID | 1410903170-31206-2-git-send-email-will.newton@linaro.org |
|---|---|
| State | Accepted |
| Commit | dc6fe23ba65714781a3d26270c132c0ca0ca5beb |
| Headers | show |
On 16 September 2014 14:32, Will Newton <will.newton@linaro.org> wrote: > Building this test on ARM fails because the prototypes for the long > double variants of the math functions are unavailable. > > Add an additional include guard to math.h that enables long double math > function declarations if _LIBC_TEST is defined and define _LIBC_TEST in > stdlib/tst-strtod-round.c. > > ChangeLog: > > 2014-09-16 Will Newton <will.newton@linaro.org> > > * math/math.h: Define long double math functions if > _LIBC_TEST is defined. > * stdlib/tst-strtod-round.c: Define _LIBC_TEST. > --- > math/math.h | 2 +- > stdlib/tst-strtod-round.c | 3 +++ > 2 files changed, 4 insertions(+), 1 deletion(-) Ping? > diff --git a/math/math.h b/math/math.h > index 72ec2ca..dc7ab2d 100644 > --- a/math/math.h > +++ b/math/math.h > @@ -91,7 +91,7 @@ __BEGIN_DECLS > # undef _Mdouble_END_NAMESPACE > # undef __MATH_PRECNAME > > -# if !(defined __NO_LONG_DOUBLE_MATH && defined _LIBC) \ > +# if !(defined __NO_LONG_DOUBLE_MATH && defined _LIBC && !defined _LIBC_TEST) \ > || defined __LDBL_COMPAT > # ifdef __LDBL_COMPAT > > diff --git a/stdlib/tst-strtod-round.c b/stdlib/tst-strtod-round.c > index f334b20..cbde457 100644 > --- a/stdlib/tst-strtod-round.c > +++ b/stdlib/tst-strtod-round.c > @@ -17,6 +17,9 @@ > License along with the GNU C Library; if not, see > <http://www.gnu.org/licenses/>. */ > > +/* Defining _LIBC_TEST ensures long double math functions are > + declared in the headers. */ > +#define _LIBC_TEST 1 > #include <fenv.h> > #include <float.h> > #include <math.h> > -- > 1.9.3 >
On Tue, Sep 16, 2014 at 02:32:50PM -0700, Will Newton wrote: > Building this test on ARM fails because the prototypes for the long > double variants of the math functions are unavailable. > > Add an additional include guard to math.h that enables long double math > function declarations if _LIBC_TEST is defined and define _LIBC_TEST in > stdlib/tst-strtod-round.c. > > ChangeLog: > > 2014-09-16 Will Newton <will.newton@linaro.org> > > * math/math.h: Define long double math functions if > _LIBC_TEST is defined. > * stdlib/tst-strtod-round.c: Define _LIBC_TEST. > --- > math/math.h | 2 +- > stdlib/tst-strtod-round.c | 3 +++ > 2 files changed, 4 insertions(+), 1 deletion(-) > > diff --git a/math/math.h b/math/math.h > index 72ec2ca..dc7ab2d 100644 > --- a/math/math.h > +++ b/math/math.h > @@ -91,7 +91,7 @@ __BEGIN_DECLS > # undef _Mdouble_END_NAMESPACE > # undef __MATH_PRECNAME > > -# if !(defined __NO_LONG_DOUBLE_MATH && defined _LIBC) \ > +# if !(defined __NO_LONG_DOUBLE_MATH && defined _LIBC && !defined _LIBC_TEST) \ Wouldn't it be nicer to read with: # if !(defined __NO_LONG_DOUBLE_MATH && defined _LIBC) || defined _LIBC_TEST || defined __LDBL_COMPAT ... The double negation looks a bit ugly to me. Looks OK otherwise. Siddhesh
On Wed, Sep 24, 2014 at 08:29:15PM +0530, Siddhesh Poyarekar wrote: > On Tue, Sep 16, 2014 at 02:32:50PM -0700, Will Newton wrote: > > Building this test on ARM fails because the prototypes for the long > > double variants of the math functions are unavailable. > > > > Add an additional include guard to math.h that enables long double math > > function declarations if _LIBC_TEST is defined and define _LIBC_TEST in > > stdlib/tst-strtod-round.c. > > > > ChangeLog: > > > > 2014-09-16 Will Newton <will.newton@linaro.org> > > > > * math/math.h: Define long double math functions if > > _LIBC_TEST is defined. > > * stdlib/tst-strtod-round.c: Define _LIBC_TEST. Oh, and this needs a BZ. Siddhesh
On Wed, 24 Sep 2014, Siddhesh Poyarekar wrote: > On Wed, Sep 24, 2014 at 08:29:15PM +0530, Siddhesh Poyarekar wrote: > > On Tue, Sep 16, 2014 at 02:32:50PM -0700, Will Newton wrote: > > > Building this test on ARM fails because the prototypes for the long > > > double variants of the math functions are unavailable. > > > > > > Add an additional include guard to math.h that enables long double math > > > function declarations if _LIBC_TEST is defined and define _LIBC_TEST in > > > stdlib/tst-strtod-round.c. > > > > > > ChangeLog: > > > > > > 2014-09-16 Will Newton <will.newton@linaro.org> > > > > > > * math/math.h: Define long double math functions if > > > _LIBC_TEST is defined. > > > * stdlib/tst-strtod-round.c: Define _LIBC_TEST. > > Oh, and this needs a BZ. I don't see why - it's not a bug that was user-visible in a release. User programs, which should never define _LIBC or _LIBC_TEST, will have got the required declarations anyway.
On Wed, Sep 24, 2014 at 03:13:17PM +0000, Joseph S. Myers wrote: > I don't see why - it's not a bug that was user-visible in a release. > User programs, which should never define _LIBC or _LIBC_TEST, will have > got the required declarations anyway. There's a build component in bugzilla that exists for exactly such bugs. It could be argued that it's a test failure and not a build failure, but then most distributions (and people) have the test run as part of their workflow and a test failure may in some cases result in a build failure. Siddhesh
On Wed, 24 Sep 2014, Siddhesh Poyarekar wrote: > On Wed, Sep 24, 2014 at 03:13:17PM +0000, Joseph S. Myers wrote: > > I don't see why - it's not a bug that was user-visible in a release. > > User programs, which should never define _LIBC or _LIBC_TEST, will have > > got the required declarations anyway. > > There's a build component in bugzilla that exists for exactly such > bugs. It could be argued that it's a test failure and not a build > failure, but then most distributions (and people) have the test run as > part of their workflow and a test failure may in some cases result in > a build failure. But it's not a test failure in a release either (and the point of filing a bug when fixing it is for users of releases to be able to find fixed bugs when searching Bugzilla; anyone using git master should be following libc-alpha). It's build warnings (the functions get the right types anyway, being built-in). The use of -Werror=implicit-function-declaration postdates the last release.
On 24 September 2014 20:58, Joseph S. Myers <joseph@codesourcery.com> wrote: > But it's not a test failure in a release either (and the point of filing a > bug when fixing it is for users of releases to be able to find fixed bugs > when searching Bugzilla; anyone using git master should be following > libc-alpha). It's build warnings (the functions get the right types > anyway, being built-in). The use of -Werror=implicit-function-declaration > postdates the last release. Ah, of course. Never mind then. Thanks, Siddhesh
diff --git a/math/math.h b/math/math.h index 72ec2ca..dc7ab2d 100644 --- a/math/math.h +++ b/math/math.h @@ -91,7 +91,7 @@ __BEGIN_DECLS # undef _Mdouble_END_NAMESPACE # undef __MATH_PRECNAME -# if !(defined __NO_LONG_DOUBLE_MATH && defined _LIBC) \ +# if !(defined __NO_LONG_DOUBLE_MATH && defined _LIBC && !defined _LIBC_TEST) \ || defined __LDBL_COMPAT # ifdef __LDBL_COMPAT diff --git a/stdlib/tst-strtod-round.c b/stdlib/tst-strtod-round.c index f334b20..cbde457 100644 --- a/stdlib/tst-strtod-round.c +++ b/stdlib/tst-strtod-round.c @@ -17,6 +17,9 @@ License along with the GNU C Library; if not, see <http://www.gnu.org/licenses/>. */ +/* Defining _LIBC_TEST ensures long double math functions are + declared in the headers. */ +#define _LIBC_TEST 1 #include <fenv.h> #include <float.h> #include <math.h>