On Thu, Feb 26, 2009 at 09:17:28PM -0700, Eric Blake wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > According to Eric Blake on 2/26/2009 4:57 PM: > > I've done some investigation into failures of test-frexpl, test-isnanl, and > > test-vasnprintf-posixt-ldexpl.c:130: assertion failed /bin/sh[9]: 14100596 Abort(coredump) on Irix 6.5. > > > > So, I am working on a patch that looks for all instances of 0.0L / 0.0L in > > the > > testsuite, and replacing them with a call to NaNl() from tests/nan.h; at > > the > > same time, beefing up that macro to use type conversion from double to long > > double on Irix as the only known way to guarantee a NaN. > > Before the patch, gcc gave 9 of 16 tests failing on assertions, all > dealing with NaN, when I ran: > > gnulib-tool --with-tests --test isfinite isnan ceill floorl frexpl isnanl > ldexpl roundl truncl > > After the patch, under gcc, 2 tests still fail (probably due to gcc > mishandling Irix's weird notion of a 128-bit long double formed by adding > two 64-bit doubles together). It reminds me of a previously fixed MaxOS > failure in that area, although I haven't had time to test. But since the > new failures are unrelated to NaN, and since cc passes all 16 tests, I'm > committing the patch below. > > ../../gltests/test-frexpl.c:144: assertion failed > /bin/sh[9]: 341777 Abort(coredump) > FAIL: test-frexpl > ../../gltests/test-ldexpl.c:130: assertion failed > /bin/sh[9]: 338455 Abort(coredump) > FAIL: test-ldexpl > I just gave your updates a spin on IRIX 6.5.30, 6.2 and 5.3.
mips-sgi-irix6.5.30-cc744m all tests pass mips-sgi-irix6.5.30-gcc431: test-ldexpl.c:130: assertion failed /bin/sh[9]: 14100596 Abort(coredump) mips-sgi-irix6.2-cc730 all tests pass mips-sgi-irix6.2-gcc433 all tests pass mips-sgi-irix5.3-cc53 all tests pass mips-sgi-irix5.3-gcc346 all tests pass -tgc
