This is the first snapshot after the GCC 9.4 release and nearly
identical with that.
Apart from the version change it only brings a back port for the
C++ front end.
This brings back port for the aarch64 and x86 back end, the RTL
optmizers (5), tree optimizers (5), debug information, Fortran,
the C++ front end and three for libstdc++.
This brings one back port each for the aarch64, powerpc, and x86
back ends, two for the tree optimizers, one for the middle end,
and two for the RTL optimizers, plus one for the Fortran front end.
This brings eight back ports for the tree, inter-procedural (IPO),
link-time (LTO) optimizers, three for the middle end, four for the
low level (RTL) optimizers, six related to debug information, two
for aarch64, one for x86, two related to OMP, five for the C, twentyone
for the C++ front end and 32 for libstdc++.
In other words, if you're using C++ this is a good one to consider
and test.
files/patch-libcpp_lex.c now came in via upstream, so we can (and
actually have to) remove it.
This brings a back port for the aarch64 back end, one for the inter-
procedural optimizers, four for the tree optimizers, and one for the
Fortran front end.
- New functions:
. mpc_sum
. mpc_dot
- Several functions are more robust with a reduced exponent range
(for example corresponding to IEEE 754 binary formats).
- New mpcheck tool for comparison with the native C library (which
is not installed by default).
Bump all directly dependent ports. Do not bump those indirectly
dependent via the lang/gcc* family since their run-time dependencies
and code generated should not be affected.
PR: 249950
Submitted by: wen
This brings a back port for the aarch64 back end and two for the
Fortran front end.
Remove the GRAPHITE option. Graphite still is a largely experimental
feature, so better left to newer versions of GCC such as GCC 10 which
is now our default.
This only brings a minor tweak to the data flow engine.
Plus a back port of part of r565301 | gerald | 2021-02-15 from
lang/gcc11-devel:
Explicitly build --without-zstd such that archivers/zstd isn't pulled
in inadvertedly when present in the build system even though it is not
an explicit dependency. [1]
PR: 253286 [1]
This brings two back ports for the x86 back end and two for the Fortran
front end. It also has back ports FreeBSD/powerpcle support (which we are
not enabling for now, focusing on the GCC 10 series).
This brings two backports for the x86 back end, two for the Fortran
front end, and one for libstdc++.
Fold the PLUGINS option, which has been on by default, into standard
behavior and remove that option. It has not been causing any issues
since we enabled it 11 months ago, nor measurable overhead. [1]
PR: 242644 [1]
This only brings one fix for arm builtins (and two testsuite backports).
Also remove SSP_UNSAFE which was added by r327697 in 2013 to the
pre-pre-...-predecessor of this port and does not appear necessary
any longer.
