Commit Graph

32 Commits

Author SHA1 Message Date
landry
5b823cb084 Sync the config/openbsd.h and config/sparc/openbsd64.h patches with
respectively r1.9 and r1.6 of base gcc ones adding #define
NO_IMPLICIT_EXTERN_C. While here add the missing #define
NO_UNSAFE_BUILTINS to gfortran/gcc 4.2, Bump all REVISIONs.
ok sthen@
2012-10-07 20:09:44 +00:00
espie
19a7a697da fix parallel builds. fixincludes doesn't own that directory, so don't
clean it up.

More recent gcc have a cleaner fix, where fixincludes has been moved to
another directory that it truly owns, but in our case, we do not need to
clean up the fake directory as it's just been created.

okay pascal@, naddy@
2012-09-23 10:32:16 +00:00
pascal
13f6a15732 Patch GCC and derived ports to use PIE by default on the same architectures it
is enabled on in base; except for gcc 3.3, which does not have PIE support and
therefore needs to pass -nopie to the linker.

Also, sync powerpc pthread specs with base.  From Brad.

ok naddy@
2012-09-01 00:00:32 +00:00
sthen
c8cbedb21c sync OBSD_LIB_SPEC in the ports tree compilers with the updated one
in base; from Brad
2012-08-28 23:39:12 +00:00
pascal
e27ffe7e0c Sync -pthread semantics with base gcc. From Brad.
ok matthew@
2012-08-20 12:35:41 +00:00
pascal
21fda32a64 Don't hardcode GC_nprocs to 1 in boehm-gc, but get it from sysconf instead,
following a similar change in devel/boehm-gc.
From Brad.

ok kurt@
2012-05-07 10:34:33 +00:00
pascal
7db97b1c58 Define _REENTRANT, not _POSIX_THREADS for -pthread; ok kurt@ 2012-04-23 14:12:18 +00:00
kurt
1805bc8d2b Port boehm-gc rthread support from boehm-gc port. Tested in bulk by espie@ 2012-02-22 12:40:04 +00:00
ajacoutot
546de5306a Now that the system headers provide a proper definition of NULL we
can remove all these NULL patches.

from Brad
ok espie@
2012-02-22 07:48:10 +00:00
espie
022f1ce588 same dance as for 4.6: stop at stage2, oh and use CONFIG_SITE everywhere 2012-01-23 22:20:34 +00:00
naddy
ec58a1093d Merge in fixes for alpha from base (miod@):
builtins.c:
Allow MD backend to prevent the optimization of a bcopy() or memmove() of
size 1 (the size being known at compile-time) into an inline mempcpy()
expansion, which will in turn expand into a byte load and store operation.
This expansion loses precious address alignment information at some point
(because everybody knows that you can read a byte from any address, right?),
and this loses bigtime on strict alignment platforms which lack the ability
to accesse bytes directly, such as alpha (unless compiling with -mbwx and
runnning on a BWX-capable cpu).

config/alpha:
Require alignment of local arrays on word boundaries, and enable
the builtins.c `one-byte memcpy' workaround.

ok espie@
2011-12-03 15:54:05 +00:00
espie
de328a0db8 synch with recent changes in src.
mostly from brad@
2011-11-16 12:32:49 +00:00
espie
2d6d25ab59 add PRAGMA_POP_PUSH_PACK support on i386 and amd64, to allow chromium9 to
work.
Limit it to i386 and amd64 so that it has minimal impact.
As discussed with naddy@/jasper@/sthen@/miod@
2011-02-08 09:37:51 +00:00
kurt
c9c6b32dfc Correct UTHREAD_SP_OFFSET on i386 after layout change in struct
_machdep_state (last release). Fixes classpath build on i386.
2010-10-31 22:38:42 +00:00
naddy
8c7fcd5467 add the bits to actually support the alpha arch, but don't enable yet 2010-10-29 08:50:38 +00:00
jasper
caab56b2bd - resolve conflict of gcj with devel/libffi:
gcj only needs the internal libffi copy to build and links statically against
it. so there is no point in installing the libffi headers and libraries from
the gcc tree.

discussed with espie@ kurt@ and upstream libffi maintainer

ok espie@ kurt@
2009-10-10 22:48:28 +00:00
jsg
6da208a332 teach gcc about our kprintf and syslog format attributes,
adapted from the in tree gcc 3.3.5, and already included in
the llvm-gcc port.

ok kurt@ espie@
2009-07-13 23:13:34 +00:00
naddy
6dd742f931 Make gcc 4.2 work, for some value of work, on alpha.
No version bump because the changes only affect alpha and it did not
build on this platform, previously.

From: miod@
2009-07-01 12:43:55 +00:00
robert
62b3092c0e add wchar_t support to gcc 4.2 because some upcoming ports are going to need
it. this is not much different from the in-tree gcc way of doing it.
tested with openoffice and with some other uncommited stuff.
bump needed PKGNAMEs
2009-06-07 01:22:05 +00:00
kurt
1094d158a8 - enable dwarf2 exceptions (w/major bump for libestdc++)
- add gcj support for i386, amd64 and sparc64
- add MODGCC4_LANGS=java support to gcc4.port.mk
okay robert@
2009-06-04 18:24:20 +00:00
espie
ced95bccdd move to gcc 4.2.4, with a lot of work from kurt@ as well.
Doesn't build on alpha yet, but should be on the way...
2009-04-16 18:36:48 +00:00
kurt
746defb0ff Fix an unterminated string in regress tests which caused the regress
tests to bail out early. no pkg bump since no change in package.
2009-03-08 02:30:23 +00:00
robert
980a048fcc Enable sjlj exceptions by default because the other implementation does not
work on OpenBSD, and exceptions are hevaily used by OpenOffice.Org.

Backport PR libstdc++/31481 from GCC repository because this fix is needed
by openoffice:
	PR libstdc++/31481
	* include/ext/type_traits.h (__numeric_traits): Move...
	* include/ext/numeric_traits.h: ... here; fix type of
	__max_digits10.
	* include/ext/pb_ds/detail/type_utils.hpp: Include
	<ext/numeric_traits.h> too.
	* include/tr1/random: Likewise.

Tested with both openoffice and webkit. bump needed PKGNAMEs;
2009-01-18 18:47:05 +00:00
kurt
2016b12bcd - enable _cxa_atexit support
- implement pre-compiled header support using brk/sbrk. idea from drahn@
okay espie@
2008-06-12 01:14:06 +00:00
ajacoutot
3a20c29c43 - activate powerpc support at last
- explicitly disable tls (Thread-local storage) so it is not incorrectly
detected on some arch(s)

ok espie@
2007-06-19 21:23:37 +00:00
espie
4f20f8df6f update to a more recent snapshot. 2007-03-20 10:48:52 +00:00
espie
3cb79e447a revert commit mistake 2007-03-17 22:34:13 +00:00
espie
ceaa955aca new gcc snapshot, this one compiles again on i386 !
NOTE: this needs current -src, in particular current includes
for libgfortran to build correctly (semantics change in __extern inline
that lead to duplicate isalnum symbols otherwise).
2007-03-17 21:58:45 +00:00
espie
a64e971c29 clean-up:
- use PSEUDO_FLAVORS
- fix ownership of various files
- new MULTI_PACKAGES
- slightly newer snapshot
2006-11-25 16:26:46 +00:00
espie
25c7d3930b update to a recent snapshot, and fix some bullshit
can probably compile on amd64 now
2006-10-20 13:43:18 +00:00
kurt
04662e387e correct defines for SIZE_TYPE and PTRDIFF_TYPE which fixes this error:
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=28582
"error: 'operator new' takes type 'size_t' ('unsigned int') as first
parameter". help from kettenis@, okay espie@
2006-10-06 12:42:42 +00:00
espie
9ca5befc7b 4.2 branch of gcc 2006-07-31 09:50:53 +00:00