PowerPC: PR libgcc/97543, fix 64-bit long double issues I meant to post this to the gcc-patches mailing list last Thursday, but I see I posted this to an internal IBM mailing list.
This patch replaces the previous iterations of this patch: October 22nd, 2020: Message-ID: <20201022220510.ga11...@ibm-toto.the-meissners.org> October 28th, 2020: Message-ID: <20201029004204.ga15...@ibm-toto.the-meissners.org If you use a compiler with long double defaulting to 64-bit instead of 128-bit with IBM extended double, you get linker warnings about mis-matches in the gnu attributes for long double (PR libgcc/97543). Even if the compiler is configured to have long double be 64 bit as the default with the configuration option '--without-long-double-128' you get the warnings. You also get the same issues if you use a compiler with long double defaulting to IEEE 128-bit instead of IBM extended double (PR libgcc/97643). The issue is the way libgcc.a/libgcc.so is built. Right now when building libgcc under Linux, the long double size is set to 128-bits when building libgcc. However, the gnu attributes are set, leading to the warnings. One feature of the current GNU attribute implementation is if you have a shared library (such as libgcc_s.so), the GNU attributes for the shared library is an inclusive OR of all of the modules within the library. This means if any module uses the -mlong-double-128 option and uses long double, the GNU attributes for the library will indicate that it uses 128-bit IBM long doubles. If you have a static library, you will get the warning only if you actually reference a module with the attribute set. This patch does two things: 1) All of the modules that support IBM 128-bit long doubles explicitly set the ABI to IBM extended double. 2) I turned off GNU attributes for building the shared library or for building the IBM 128-bit long double support. I have discussed this patch with Alan Modra, and made several changes based on his suggestions. I have tested this on a little endian power9 system running Linux by building three separate compilers, using the Advance Toolchain AT14.0 which uses GLIBC 2.32: 1) A compiler where the long double default is IBM 128-bit double; 2) A compiler where the long double default is IEEE 128-bit double; (and) 3) A compiler where the long double default is 64-bit. Note, for the IEEE build (#2), the other patches that I will be submitting are needed to enable the full build. For each of the 3 compilers, I then tested some code with long double's and verified that each of the long double options worked without generating warnings. In addition, I have tested this patch on a big endian power8 system running Linux, and there were no regressions. Can I install this patch into the master branch? Since this is a bug for 64-bit long doubles, I would like to back port it to GCC 10, and GCC 9 after a shake-in period. libgcc/ 2020-11-17 Michael Meissner <meiss...@linux.ibm.com> PR libgcc/97543 PR libgcc/97643 * config/rs6000/t-linux (IBM128_STATIC_OBJS): New make variable. (IBM128_SHARED_OBJS): New make variable. (IBM128_OBJS): New make variable. Set all objects to use the explicit IBM format, and disable gnu attributes. (IBM128_CFLAGS): New make variable. (gcc_s_compile): Add -mno-gnu-attribute to all shared library modules. --- libgcc/config/rs6000/t-linux | 22 ++++++++++++++++++++++ 1 file changed, 22 insertions(+) diff --git a/libgcc/config/rs6000/t-linux b/libgcc/config/rs6000/t-linux index ed821947b66..72e9c2770a6 100644 --- a/libgcc/config/rs6000/t-linux +++ b/libgcc/config/rs6000/t-linux @@ -6,3 +6,25 @@ HOST_LIBGCC2_CFLAGS += -mlong-double-128 # smaller and faster libgcc code. Directly specifying -mcmodel=small # would need to take into account targets for which -mcmodel is invalid. HOST_LIBGCC2_CFLAGS += -mno-minimal-toc + +# On the modules that deal with IBM 128-bit values, make sure that TFmode uses +# the IBM extended double format. Also turn off gnu attributes on the static +# modules. +IBM128_STATIC_OBJS = ibm-ldouble$(objext) _powitf2$(objext) \ + ppc64-fp$(objext) _divtc3$(object) _multc3$(object) \ + _fixtfdi$(object) _fixunstfdi$(object) \ + _floatditf$(objext) _floatunsditf$(objext) +IBM128_SHARED_OBJS = $(IBM128_STATIC_OBJS:$(objext):_s$(objext)) +IBM128_OBJS = $(IBM128_STATIC_OBJS) $(IBM128_SHARED_OBJS) + +IBM128_CFLAGS = -Wno-psabi -mabi=ibmlongdouble -mno-gnu-attribute + +$(IBM128_OBJS) : INTERNAL_CFLAGS += $(IBM128_CFLAGS) + +# Turn off gnu attributes for long double size on all of the shared library +# modules, but leave it on for the static modules, except for the functions +# that explicitly process IBM 128-bit floating point. Shared libraries only +# have one gnu attribute for the whole library, and it can lead to warnings if +# somebody changes the long double format. We leave it on for the static +# modules to catch mis-compilation errors. +gcc_s_compile += -mno-gnu-attribute -- 2.22.0 -- Michael Meissner, IBM IBM, M/S 2506R, 550 King Street, Littleton, MA 01460-6245, USA email: meiss...@linux.ibm.com, phone: +1 (978) 899-4797