Commit Graph

1772 Commits

Author SHA1 Message Date
christos
507caaad52 Add stdatomic.h 2017-10-04 23:43:23 +00:00
he
b7a909847c Replicate fix from the old gcc.old: alpha.d needs pass-instances.def. 2017-09-12 09:25:24 +00:00
mrg
26c43e3543 move the GMP_MACHINE_ARCH definition into external/lgpl3/gmp/Makefile.arch.
soon to be GMP 6 allows us to have only two arm copies.
2017-08-22 09:17:50 +00:00
mrg
bbf2be38c0 put the location of gmp/mpc/mpfr in a variable so we can .old them. 2017-08-04 01:18:39 +00:00
mrg
56cd2f0399 pass the target specific CPUFLAGS and CPPFLAGS for the .c.o rule.
should fix a problem rjs@ reported.
2017-08-03 22:01:29 +00:00
mrg
79ce5b417f remove unused files 2017-08-03 22:00:38 +00:00
maya
8340d18be5 Apply upstream patch:
Incorrect codegen from rdseed intrinsic use (CVE-2017-11671)

We should not expand call arguments in between flags reg setting and
flags reg using instructions, as it may expand with flags reg
clobbering insn (ADD in this case).

Attached patch moves expansion out of the link. Also, change
zero-extension to non-flags reg clobbering sequence in case we perform
zero-extension with and.

2017-03-25  Uros Bizjak
2017-07-28 22:24:27 +00:00
maya
317c1f1a2e Apply upstream patch:
Incorrect codegen from rdseed intrinsic use (CVE-2017-11671)

We should not expand call arguments in between flags reg setting and
flags reg using instructions, as it may expand with flags reg
clobbering insn (ADD in this case).

Attached patch moves expansion out of the link. Also, change
zero-extension to non-flags reg clobbering sequence in case we perform
zero-extension with and.

2017-03-25  Uros Bizjak  <ubizjak@gmail.com>
2017-07-28 22:23:05 +00:00
he
f52ac36319 For alpha, generating alpha.d first needs pass-instances.def re-generated. 2017-07-27 16:05:15 +00:00
joerg
7c8897217a TLS relocations for PIE and non-PIE main executable work the same, so
apply the same relaxation rules and the same static binding rules.
2017-07-25 19:44:25 +00:00
mrg
be7137046e initialise some stack variables to zero so that later checking ensures
they are considered "unfound" instead of some random value.

part 1 in fixing the CI20 kernel build.  this one actually will cause
the build to fail, but it's outputting broken objects right now anyway.
part 2 coming up in the CI20 configuration itself.
2017-07-25 06:43:03 +00:00
mrg
7a0caf96a5 merge GCC 5.4 into gcc.old. 2017-07-23 01:10:50 +00:00
mrg
7d9adb1cf1 import the GCC 5.4 srcs from -current as gcc.old in preparation for GCC 6. 2017-07-23 01:04:23 +00:00
joerg
fc278ede3a One more missing check for DECL_INITIAL being non-NULL. 2017-07-22 20:52:52 +00:00
joerg
688d934e56 PR 51266, PR 52326: Fix GCC specs to always default to -shared-libgcc,
except on Sun2, which just uses plain -lgcc.
2017-07-19 13:10:28 +00:00
joerg
4df4cd4e92 Fix encoding of LSDA entries. .eh_frame references in non-PIC mode can
use plain pointers. For PIC, any references to global objects must be
indirect, but the .eh_frame entries themselve should be pcrel.
2017-07-18 18:02:37 +00:00
joerg
846a441fa1 Detect advance_loc support with readelf, not objdump. The latter is
broken in some situations before binutils 2.28.
2017-07-17 19:55:13 +00:00
joerg
8a8cb2911c Use read-only .eh_frame. 2017-07-17 19:53:50 +00:00
joerg
ef39c7e610 A const declaration with explicit section attribute should create a
read-only section, whether it is initialized or not.
2017-07-17 19:53:10 +00:00
joerg
05cc9c0cb6 Create references as absolute pointers internally for .cfi*.
Do not warn for anonymous relocations in read-only .eh_frame,
they will be turned into relative entries afterwards.
2017-07-17 19:50:23 +00:00
christos
6c406925bb fix the powerpc build, by undefining the EH_SPEC entry first. 2017-07-15 23:52:30 +00:00
joerg
c1a5bf052c Restore --eh-frame-hdr for static binaries. 2017-07-14 23:05:27 +00:00
joerg
a5c3ba3e65 Allocate enough memory. 2017-07-11 21:19:42 +00:00
joerg
e5678be828 Implement __cxa_thread_atexit and __cxa_thread_atexit_impl. This
functions are used for destructors of thread_local objects.

If a pending destructor exists, prevent unloading of shared objects.
Introduce __dl_cxa_refcount interface for this purpose. When the last
reference is gone and the object has been dlclose'd before, the
unloading is finalized.

Ideally, __cxa_thread_atexit_impl wouldn't exist, but libstdc++ insists
on providing __cxa_thread_atexit as direct wrapper without further
patching.
2017-07-11 15:21:31 +00:00
chs
3c2bb6eab4 copy the CTF section too, if there is one. 2017-07-06 02:34:00 +00:00
joerg
f2faadd888 Mark a few more things as not-for-clang. While the sanitizers would
likely build, there are ordering issues relative to the libc++ build
right now.
2017-06-29 18:52:23 +00:00
joerg
ff6447e3f3 Honor HOST_CXX. 2017-06-26 17:38:10 +00:00
mrg
781c4b6a03 native-gcc for all platforms (only m68000 has a real change.) 2017-06-17 17:12:03 +00:00
kamil
fc24321608 Don't include <sys/user.h> on NetBSD in GDB
This header in this context is freebsdism.
2017-06-15 23:46:30 +00:00
kamil
767ace0885 Detach <sys/user.h> from sanitizer_procmaps_netbsd.cc (GCC)
This header in this context is freebsdism.

Sponsored by <The NetBSD Foundation>
2017-06-14 12:16:27 +00:00
mrg
b0309df642 regen after new binutils 2017-06-08 19:21:54 +00:00
flxd
9c7eabb4c9 Fix PR port-vax/51761 as suggested by Paul Koning on port-vax list.
Installation (install.ram, -Os) on my VS4000 is possible without SCSI timeouts
again.
Other variable-length bit field instructions should be checked for correct
constraints, too!
2017-06-08 15:28:27 +00:00
christos
dd61a7a6c2 remove bogus indirect call for profiling (mcount) 2017-05-29 03:51:24 +00:00
sevan
ec28d4a50d Fix building GCC 5 with clang 4.0
Obtained from FreeBSD ports r432958
https://svnweb.freebsd.org/ports?view=revision&revision=432958
"doesn't compile with recent versions of libc++, because it attempts to redefine abort():
/usr/include/c++/v1/functional:1398:2: error: no member named 'fancy_abort' in namespace 'std::__1'; did you mean simply 'fancy_abort'?
        _VSTD::abort();
        ^~~~~~~
/usr/include/c++/v1/__config:383:15: note: expanded from macro '_VSTD'
#define _VSTD std::_LIBCPP_NAMESPACE"
2017-05-17 22:06:10 +00:00
christos
7ef1d3a265 Teach me about the NetBSD-CORE ELF AuxV note. 2017-03-29 15:28:42 +00:00
chs
7e30e94394 provide a common softfloat fenv implemenation and use it for softfloat builds.
restore ABI compatibility with previous releases for ieeefp.h on sh3.
add namespace.h protection for all the fenv interfaces.
use MKSOFTFLOAT on sh3 instead of assuming softfloat.
standardize on comparing MKSOFTFLOAT with "no".
remove the arm-specific softfloat fenv code (which also had several bugs).
fix logic errors in the arm hardfloat feraiseexcept() and feupdateenv().
2017-03-22 23:11:07 +00:00
christos
5978bc7047 need .WAIT before *san since they depend on libc++ 2017-03-07 16:26:37 +00:00
joerg
6fbd2a0be4 Switch from __ABICALLS__ to __mips_abicalls like upstream GCC does in
the generic MIPS target logic.
2017-02-25 21:16:50 +00:00
chs
1f54905796 revert rev 1.2 (disabling split wide types) because it's no longer necessary
after rev 1.6 of m68k.md (which itself reverted a local patch that hadn't
been necessary for over a decade... see gcc bugzilla 12792).
discussed with christos and confirmed no regressions with the atf tests.
2017-02-23 18:43:01 +00:00
rin
d362232877 sorry, fix comment... 2017-02-14 00:20:20 +00:00
rin
78470a5a42 PR port-vax/51967 GCC 5.4 aborts on vax due to SIGILL for cc1
As a workaround, compile dse.c with -O0.
2017-02-13 22:31:09 +00:00
kamil
77d313fc36 Detach libpthread_dbg(3) from gdb(1) and gdbtui (gdb.old)
libpthread_dbg(3) is a remnant library from the M:N thread model
(pre-NetBSD-5.0) API to introspect threads within a process and for use
of debuggers.

Currently in the 1:1 model it's not used in GDB neither in LLDB and it's
not either planned to be used. It's current function to read pthread_t
structures is realizable within a regular debugger capable to
instrospect objects within a tracee (GDB, LLDB...).

pthread_dbg(3) is going to be removed from the base distribution and moved
to pkgsrc.

Patch reviewed by <christos>

Sponsored by <The NetBSD Foundation>
2017-02-08 01:32:37 +00:00
kamil
e361313d33 Detach libpthread_dbg(3) from gdb(1) and gdbtui
libpthread_dbg(3) is a remnant library from the M:N thread model
(pre-NetBSD-5.0) API to introspect threads within a process and for use
of debuggers.

Currently in the 1:1 model it's not used in GDB neither in LLDB and it's
not either planned to be used. It's current function to read pthread_t
structures is realizable within a regular debugger capable to
instrospect objects within a tracee (GDB, LLDB...).

pthread_dbg(3) is going to be removed from the base distribution and moved
to pkgsrc.

Patch reviewed by <christos>

Sponsored by <The NetBSD Foundation>
2017-02-07 22:53:19 +00:00
sevan
cc576e1d8e Update supporting files for components which rely on autoconf to allow systems
introducing since release of software to be recognised. This should hopefully
allow the builds to progress a littles further on systems such as the POWER8
which features a little endian 64-bit PowerPC CPU identified as ppc64le.
2017-02-01 09:26:39 +00:00
christos
e145362bfe use the full attribute syntax 2017-01-17 12:49:26 +00:00
christos
0133d15da2 use the full attribute syntax. 2017-01-17 12:48:39 +00:00
christos
d3da06c622 elide psabi errors on i386 2017-01-15 21:27:38 +00:00
christos
8e7ebb8f8c Handle errors in system headers 2017-01-13 01:14:27 +00:00
christos
78df756bd3 sprinkle __unused 2017-01-12 19:21:09 +00:00
christos
73f4d05a56 revert for now; too many issues. 2017-01-12 19:10:00 +00:00
christos
d0c807d2c9 Our headers are clean, enable warnings caused by them. We want to know
about <ctype_macro>(char) usages.
2017-01-12 17:14:23 +00:00
skrll
3e2d39191a Fix merge botch. arm eabi targets now match properly and we get the
right defaults.
2017-01-11 14:13:05 +00:00
buhrow
d33b37e9a7 Fix problem which prevents cross building on NetBSD systems which predate
NetBSD-6 or non-NetBSD systems.  See discussion on port-arm for details.
2016-12-30 22:40:23 +00:00
kamil
60b1976a76 Revert introduction of FIXME_PR_51139
This change introduced ABI incompatible change with older versions shipped
on NetBSD. This back out code that is currently not working correctly due
to TLS-based std::call_once implementation in GNU libstdc++.

Error when starting gnuchash:

/usr/pkg/lib/libwebkitgtk-1.0.so.0: Undefined symbol "_ZSt15__once_callable" (symnum = 1705)

PR 51139

Reported by <wiz>
2016-12-24 15:48:26 +00:00
kamil
49e3a2363a Add a walkaround for TLS bug in libstdc++ exposed with std::call_once
Currently std::call_once with libstdc++ works only with static linking.
Disable code path using __thread types and introduce FIXME_PR_51139.
Problem discussed in PR 51139

Functional std::call_once is required in LLVM and LLDB codebase.

Example code to test std::call_once:
#include <iostream>
#include <thread>
#include <mutex>
#include <cstdlib>
std::once_flag flag;
int main(int argc, char **argv)
{
        std::call_once(flag, [](){ std::cout << "Simple example: called once\n"; });
        return EXIT_SUCCESS;
}

Sponsored by <The NetBSD Foundation>
2016-12-21 21:55:46 +00:00
rin
36532a0d5a PR port-powerpc/51366 kernel for powerpc/ibm4xx miscompiled by gcc >= 4.8.5
backport fix from upstream:

URL: https://gcc.gnu.org/viewcvs?rev=238789&root=gcc&view=rev
Log:
rs6000: Fix logic for when to emit .machine (PR71216)

The current logic determining whether to use .machine in the generated
asm code puts it there if the compiler is not configured with a default
target cpu, _or_ no -mcpu= was given on the command line.  It should
be "and" instead.

	PR target/71216
	* config/rs6000/rs6000.c (rs6000_file_start): Fix condition for
	when to emit a ".machine" pseudo-op.

Modified:
    branches/gcc-5-branch/gcc/ChangeLog
    branches/gcc-5-branch/gcc/config/rs6000/rs6000.c
2016-12-18 03:00:58 +00:00
rin
e31b62ddaa Skip the initial GP load in function prologue when inserting a breakpoint.
GNU ld for alpha is so clever that the redundant GP load in function
entrypoint is skipped. we must therefore skip initial GP loads; otherwise
breakpoints in function entrypoints can also be skipped.

Reported to upstream (Bug 20969):
  https://sourceware.org/bugzilla/show_bug.cgi?id=20969

ok martin
2016-12-15 10:01:16 +00:00
skrll
2e2322c9c0 Remove binutils.old as everything has switched. 2016-12-14 21:47:50 +00:00
mrg
899aca2e24 mknative, binutils 2.27, and ia64. 2016-12-12 19:15:15 +00:00
mrg
42a13a4db0 mknative-binutils 2.27 and hppa. 2016-12-11 02:13:34 +00:00
mrg
16e7be53ca regen binutils mknative files for binutils 2.27 on mipsel, mipseb and aarch64. 2016-12-11 01:53:43 +00:00
mrg
30b9303d3c regen mknative-binutils for ppc64 and binutils 2.27 2016-12-10 22:57:41 +00:00
mrg
de41f27e8a rebuild mkantive for ppc and binutils 2.27. 2016-12-10 21:33:53 +00:00
christos
a582531a1a finish arm regeneration 2016-12-08 19:35:10 +00:00
christos
37fdc90f90 regen sun2, earm, earmeb 2016-12-05 22:59:17 +00:00
christos
e33f4e7ec2 Regen sh3 2016-12-04 18:43:07 +00:00
christos
b1a05e3501 switch alpha/sparc/sparc64 to new binutils 2016-12-03 16:01:18 +00:00
christos
5b0871f5ea regen mips64 2016-12-03 00:40:47 +00:00
christos
787b0418af regen m68k 2016-12-02 17:51:07 +00:00
christos
811931c076 on NetBSD we don't have llvm_symbolizer, so try addr2line... 2016-12-01 18:20:25 +00:00
christos
0833d377e3 use the right sysctl to find the main binary name. We don't really need
this since our dl_iterate_phdr DTRT's for objmain, but...
2016-12-01 18:19:19 +00:00
christos
a6a34af0f6 regen for vax 2016-11-27 18:52:05 +00:00
mrg
38ac2f14f5 regen all gdb platforms. 2016-11-20 00:19:55 +00:00
mrg
98abf92201 s/#/$/ in a $NetBSD$ so it actually expands 2016-11-19 09:46:58 +00:00
skrll
0efd40303c mknative re-run 2016-11-19 09:10:33 +00:00
skrll
8d8621a4bd Real mknative re-run 2016-11-19 08:46:03 +00:00
skrll
56b555c863 Mark the alpha options use_secureplt as false... we don't do secure plt
on alpha (yet)
2016-11-19 07:54:19 +00:00
skrll
c7af765c8e Add in netbsd-core.lo for all arm/mips 2016-11-19 07:52:49 +00:00
skrll
f958ae60d4 (simulate) mknative re-run 2016-11-18 12:42:10 +00:00
skrll
3a5e17dcb9 Provide a generic nbsd_skip_solib_resolver and use it in alpha and arm.
Others should follow suit.

I even tested it.
2016-11-18 09:52:33 +00:00
joerg
c4414c4118 Merge r1.10 from the main in-tree binutils version:
More detailed error messages for text relocations on ppc code. Tested by
joerg@
2016-11-06 23:46:01 +00:00
joerg
d74117acef Merge r1.11 from the main in-tree binutils version:
Tail calls can use the PLT without a nop after the branch. Since the
callee won't return to the next instruction anyway, it wouldn't work
anyway.
2016-11-06 23:44:44 +00:00
christos
096c8b23c8 follow suit (add missing d) 2016-11-04 15:31:45 +00:00
martin
ed3eda7194 Add NODEBUGLIB here as well (sync with binutils directory) 2016-11-04 13:29:38 +00:00
joerg
374e459041 Tail calls can use the PLT without a nop after the branch. Since the
callee won't return to the next instruction anyway, it wouldn't work
anyway.
2016-11-04 00:19:32 +00:00
christos
f47de02583 More detailed error messages for text relocations on ppc code. Tested by joerg@ 2016-11-03 01:22:59 +00:00
christos
3d63ec494e need an = 2016-11-01 15:58:41 +00:00
christos
24c043b8d9 bump both (belatedly) 2016-11-01 14:02:15 +00:00
christos
ed77187f53 Add NODEBUGLIB, perhaps LIBISPRIVATE is better here? 2016-11-01 14:02:01 +00:00
martin
226fa8b9ee Add a skip_solib_resolver, from Rin Okuyama, fixes single stepping for
shared binaries.
2016-10-29 17:02:06 +00:00
christos
6ffca6ccd9 it might not be a breakpoint, so make the message what it actually is. 2016-10-27 14:30:55 +00:00
skrll
4b15477c73 .mt files are dead 2016-10-27 11:48:24 +00:00
christos
f5b8191dd4 regen x86 2016-10-26 20:39:27 +00:00
christos
944e10e55a restore the initfini array tests 2016-10-26 20:26:20 +00:00
christos
2f15a542cc merge conflicts, disable testsuite 2016-10-26 18:42:52 +00:00
christos
be8325a0f7 for next time 2016-10-26 18:00:56 +00:00
christos
c05b68c164 use the real top 2016-10-26 18:00:46 +00:00
christos
eb46678fb2 use our real top 2016-10-26 17:58:48 +00:00
christos
8cbf5cb7c4 Import binutils-2.27 minus the testsuites 2016-10-26 17:29:32 +00:00
christos
9b818ce2b0 merge conflicts 2016-10-26 17:08:37 +00:00
christos
8450a7c426 Import current binutils 2016-10-26 17:02:27 +00:00
christos
8d3ec8c9f8 Don't fail silently if we can't set a breakpoint 2016-10-24 21:22:33 +00:00
mrg
7b28f87c3c regen. 2016-10-24 03:02:48 +00:00
mrg
647468f389 regen after config.bfd fix. 2016-10-23 18:51:55 +00:00
skrll
f1c7b5d7e5 Sync with external/gpl3/binutils/dist/bfd/config.bfd for mips so we build
in the right vector set
2016-10-23 12:46:57 +00:00
skrll
628b30c419 Sync with external/gpl3/binutils/dist/bfd/config.bfd for arm so we build
in the right vector set
2016-10-23 12:43:01 +00:00
skrll
33576f34ff Trailing whitespace 2016-10-23 07:40:08 +00:00
skrll
074ac3ce32 PR toolchain/51557: gdb can't read core files
Register an iterate_over_regset_sections methods for armnbsd
2016-10-23 07:33:15 +00:00
christos
bbf41bf8ef use LIBGCC_MACHINE_ARCH for the compat build 2016-10-22 20:33:28 +00:00
skrll
a6174a4ba0 Add -lkvm for bsd-kvm.o 2016-10-22 14:57:50 +00:00
skrll
d2624817ef No need for corelow.o here now as it is in COMMON_OBS 2016-10-22 14:38:07 +00:00
skrll
4f5401ddf9 Fix HPPANBSD_SIZEOF_GREGS to match sizeof(struct reg) 2016-10-22 11:04:16 +00:00
christos
f6383fc60c regen and include the proper directory for x86_64 2016-10-22 03:55:03 +00:00
christos
6d0f30fd53 enable tf math (__float128) 2016-10-22 03:33:03 +00:00
christos
5300379ae6 Mark this -z nodelete, so that it does not get unloaded before other shared
libraries do, since they destructors need to still call millicode.
The problem this time was with /usr/pkg/bin/gdk-pixbuf-query-loaders
calling glib2's g_log...
2016-10-21 23:44:09 +00:00
mrg
e7aef7676b basic support for -static -pie: use -dynamic-linker.
doesn't work for all ports (reloc startup issues.)
2016-10-21 07:24:30 +00:00
mrg
a1169008f0 match aarch64* for the ldscripts generation. 2016-10-19 20:55:44 +00:00
mrg
31caae3a77 mknative-binutils for binutils 2.26 and arm64eb 2016-10-19 20:31:28 +00:00
mrg
b54003801f regen again after fixing the armeb-netbsdelf list.
evbarm-armeb builds again.
2016-10-19 19:10:33 +00:00
mrg
4a685cb277 fix the armeb target list. 2016-10-19 19:09:42 +00:00
mrg
8beef5983a regen to remove the accidental freebsd files 2016-10-19 19:02:38 +00:00
christos
bca50078b8 fix typo. 2016-10-18 12:52:32 +00:00
mrg
07ee0c7402 regen mknative for all ports and GCC 5.4 and -current. many ports
were using 5.3 data, and none of them had <stdalign.h>.
2016-10-18 01:32:56 +00:00
mrg
7b202c5d1e fix earmhf and earmhfeb builds. 2016-10-18 00:16:58 +00:00
mrg
c78a23b309 mknative-gdb for GDB 7.12 and coldfire/m68000. untested.
this completes our regularly scheduled mknative-gdb for all ports.
until next time enjoy your flight and have a nice day.
2016-10-17 18:44:43 +00:00
mrg
fbe48ae2ff remove unused files. 2016-10-17 18:32:33 +00:00
mrg
400d1a1689 mknative-gdb for GDB 7.12 and sh3*. untested. 2016-10-17 18:19:06 +00:00
mrg
e110ade75c mknative-gdb for GDB 7.12 for m68k and vax. untested. 2016-10-17 17:38:07 +00:00
mrg
15077f1b3c add some missing casts and make these build. 2016-10-17 17:37:44 +00:00
christos
a417a46e02 regen powerpc64 2016-10-17 14:02:20 +00:00
mrg
c1e10f7755 mknative-gdb for GDB 7.12 and i386 (untested) 2016-10-17 06:26:37 +00:00
mrg
fe0f2a5492 add missing i386bsd_supply_gregset() and i386bsd_collect_gregset() protos 2016-10-17 06:26:20 +00:00
mrg
3caff98623 mknative-gdb for GDB 7,12 and mips64*. untested (but builds.) 2016-10-17 04:15:35 +00:00
mrg
c9be6ee8d6 mknative-gdb for GDB 7.12 and alpha, mipsel and mipseb.
alpha and mipseb build (mipsel should), nothing is tested.
2016-10-16 08:23:05 +00:00
mrg
c03f037c58 define _KMEMUSER so that necessary things become visible always. 2016-10-16 05:12:20 +00:00
mrg
b0273061d2 move the GDB_MACHINE_ARCH setting into a separate file so tools can
access it.  modify it such that it's "arm" or "armeb" for all arm,
but only in new gdb, not gdb.old (avoid changing gdb.old as much
as possible now.)
2016-10-16 04:36:51 +00:00
mrg
47f5dba6de gdb only needs on arm and on armeb set, for all the 32 bit arm ports.
regen arm and armeb for gdb 7.12, and delete the rest.
2016-10-16 04:30:04 +00:00
maya
aa71315a69 this rule shouldn't be specific to little endian 2016-10-15 12:30:55 +00:00
mrg
23a8d08d85 mknative-gdb for GDB 7.12 and sparc, sparc64 and powerpc. sparc
works, the others aren't yet tested but do build.
2016-10-15 03:11:47 +00:00
christos
d1f6f311d5 multiple include protection 2016-10-14 23:49:23 +00:00
mrg
cce0131c9c go back to only installing ldscripts for ports that need them (multi-abi) 2016-10-14 23:27:00 +00:00
christos
b3faa1f1b4 Add new path for arm 2016-10-14 16:38:27 +00:00
christos
f424392a52 regen earm 2016-10-14 16:19:23 +00:00
christos
d11c4303f8 Remove hacks and depend on the files being built to decide wha'ts being
installed.
2016-10-14 15:58:34 +00:00
christos
9ff2fcdd59 fix some aarch64 nits 2016-10-14 14:41:32 +00:00
skrll
f4ea3389b0 We have our own popcount 2016-10-14 14:14:58 +00:00
skrll
e6e32b5f39 Make this compile 2016-10-14 14:12:48 +00:00
mrg
3c9887d690 mknative for gdb 7.12 and sparc*, and earmv4. sparc64 doesn't build yet.
others not yet tested.
2016-10-14 05:37:44 +00:00
skrll
2920ca48d3 mknative for hppa gdb 7.12 2016-10-13 21:09:13 +00:00
martin
a61a76c709 Install the arm ldscripts for big endian variants too 2016-10-13 13:29:39 +00:00
christos
5c86dbca0a And one more level 2016-10-13 01:18:59 +00:00
christos
0c04591cf1 One more level down 2016-10-13 01:10:15 +00:00
christos
bb16d22702 Add more directories 2016-10-13 00:55:20 +00:00
christos
c358683dff Merge conflicts and regen amd64 2016-10-12 20:20:32 +00:00
christos
ba340e457d import gdb-7.12 2016-10-12 16:47:08 +00:00
christos
a5a4af3bd3 re-add all the removed files 2016-10-12 15:36:05 +00:00
christos
4e00368f12 import current gdb-7.10.1 as gdb.old 2016-10-12 15:15:22 +00:00
mrg
4685143ff5 build and install the arm ldscripts. now i can build oabi again. 2016-10-11 07:23:52 +00:00
mrg
9535eb5451 fix the previous - the emul names don't tend to include "32".
tested on shark, though fixes to make compiling for oabi work
remain uncommited.

now an aarch64 build almost completes!
2016-10-11 07:05:29 +00:00
mrg
ba759bad89 revert previous - was fixed in the mknative. 2016-10-11 04:10:13 +00:00
mrg
de6fd971d9 regen with mostly fixed. 2016-10-11 03:57:15 +00:00
christos
8897404ec7 more HAVE stuff. 2016-10-11 03:09:37 +00:00
christos
be6cc8d4b9 regen and fix aarch64 2016-10-11 03:08:14 +00:00
christos
b885939ba2 fix emulation path 2016-10-11 02:49:38 +00:00
skrll
b51620c841 Remove gcc.old now it is no longer used. 2016-10-09 09:26:45 +00:00
skrll
d99781d3c9 Remove binutils.old now it is no longer used 2016-10-09 08:21:35 +00:00
skrll
cf81abf893 Remove gdb.old not it is unused. 2016-10-09 07:31:17 +00:00
christos
84670d8d34 fix ppc64 2016-10-06 20:29:33 +00:00
christos
bcc92e9b65 fix ssp build on amigappc 2016-10-06 02:58:32 +00:00
christos
0e20c92e52 The NetBSD libc provides the softfloat support which is duplicated in libgcc
and we prefer ours.
2016-10-05 19:24:44 +00:00
christos
93731ee132 regen 2016-10-05 14:30:18 +00:00
christos
cb24a47c16 regen 2016-10-05 14:16:56 +00:00
uwe
c3674760a3 Run configure with MAKE=${TOOL_GMAKE:Q} - otherwise "checking
dependency style" test fails on non-netbsd hosts where nbmake can't
find share/mk.

Ok phone@
2016-09-29 20:08:25 +00:00
christos
e4026cc8ea __NetBSD__ -> SANITIZER_NETBSD 2016-09-22 13:13:09 +00:00
christos
0b7ade5bd2 Avoid definining a prototype for signal(3) from <sys/param.h> because the
INTERCEPTOR macro cannot handle function returns easily.
TODO: fix <sys/param.h> to avoid this kind of name pollution.
2016-09-22 03:43:07 +00:00
christos
1fab549d94 Find bin.h 2016-09-22 01:53:39 +00:00
christos
2e3ca227c0 need <sys/param.h> 2016-09-21 23:18:42 +00:00
christos
9c3dcc6f8b move from /usr/src/usr.sbin/mdsetimage 2016-09-21 21:52:26 +00:00
christos
37b1f1d690 Check the NetBSD version 2016-09-21 21:30:56 +00:00
christos
a75c439d9f add new field 2016-09-21 19:18:01 +00:00
christos
a8b83ade53 remove dup 2016-09-21 16:26:18 +00:00
christos
5cafd780cd fix printf format. 2016-09-21 01:33:42 +00:00
christos
639759bc47 Print more info. 2016-09-20 20:57:45 +00:00
christos
f4001fd30b XXX: manually add uniform_int_dist.h because everyone else has it and
it breaks the build!
2016-09-18 02:03:16 +00:00
christos
d120f5b7a3 undo the attempt to figure out mult clobbering; breaks compilation; revert
to the fsf version
2016-09-17 00:55:40 +00:00
christos
f47e9f5683 turn off split wide types to avoid gcc bug. Arguably this should be fixed
in m68k.md, but I will leave this to someone more qualified.
2016-09-16 20:31:00 +00:00
christos
7f119c296e put the clobber after the instruction. 2016-09-16 20:30:14 +00:00
christos
8514813e35 more errors for text relocations 2016-09-05 14:25:33 +00:00
christos
118ae671e7 sync error with x86_64. 2016-09-05 09:17:35 +00:00
christos
6f1abd5174 gcc generates incorrect jump tables with -fPIE 2016-09-03 15:47:54 +00:00
christos
1eb0ac3f07 regen for alpha and sh now that they have fenv.h 2016-08-26 17:28:54 +00:00
christos
6b98bddc8a the heuristic for the cpu dependent file atomicity.h is unusable for little
endian sh. For big-endian it happens to work because host_cpu!=sh. Elide it
2016-08-26 17:28:32 +00:00
christos
388d69f323 XXX stop gap fix for atomicity.
U __gnu_cxx::__atomic_add(int volatile*, int)
U __gnu_cxx::__exchange_and_add(int volatile*, int)
2016-08-25 13:17:09 +00:00
mrg
56058dff13 canonicalise the netbsd spec functions in NETBSD_SUBTARGET_EXTRA_SPECS:
netbsd_cpp_spec
   netbsd_link_spec
   netbsd_entry_point
   netbsd_endfile_spec

use this in all netbsd targets.

XXX: 3 ports set this to EXTRA_SPECS instead of SUBTARGET_EXTRA_SPECS.
2016-08-19 05:11:09 +00:00
tls
b139dbeb03 Disable iconv in host-libcpp autoconf: fixes build on Mac OS X.
Thanks to joerg@ for the diagnosis and proposed fix.
2016-08-14 16:04:50 +00:00
scole
6afc3b56dd PR port-ia64/51261
Change JEMALLOC_TINY_MIN_2POW for ia64
2016-08-05 16:25:18 +00:00
mrg
95d9a2c7e1 regen for binutils 2.26.1 and *. 2016-08-02 08:05:05 +00:00
skrll
e084e74ccf Re-run mknative 2016-08-01 11:22:34 +00:00
christos
4f619fea3a regen; Left: earm, armeb, earmeb, riscv64, earmhf, earmhfeb, aarch64. 2016-07-28 14:17:58 +00:00
christos
9180686841 regen amd64. 2016-07-28 12:25:24 +00:00
christos
4a4d5be6eb merge conflicts 2016-07-28 12:25:07 +00:00
christos
09ac01bf76 Import 2.26.1; please see the individual ChangeLogs for bug fixes. 2016-07-28 12:18:51 +00:00
christos
d78413be59 more stack protector 2016-07-21 16:47:50 +00:00
christos
466a5d87d2 don't die printing a name we don't have 2016-07-21 00:10:57 +00:00
mrg
6c93c005b3 find gcov.1 again. 2016-07-07 03:15:58 +00:00
mrg
703c8b8111 missed from previous:
remove old generated versions of man pages.
update the dist versions of man and info pages to include local changes.
2016-07-06 11:37:26 +00:00
mrg
e292c2206e remove old generated versions of man pages.
update the dist versions of man and info pages to include local changes.
2016-07-06 11:21:17 +00:00
mrg
2e7e02f388 avoid the default TARGET_INIT_LIBFUNCS (i think) and XFmode. 2016-07-01 14:48:51 +00:00
mrg
1e2fcaaf47 enable some soft-fp for ia64, needed for libgcc. re-mknative-gcc. 2016-06-30 09:06:35 +00:00
mrg
70e5fb7477 mknative-gcc for GCC 5.4 and ia64. 2016-06-30 01:44:48 +00:00
mlelstv
f72ec5db06 Drop extra argument not used by format string. 2016-06-26 15:46:42 +00:00
mlelstv
ec520c2878 fix build 2016-06-26 14:44:10 +00:00
skrll
a38b18bd73 arm-builtins needs insn-opinit.h
I had this sitting in my tree already, but prompted by Michael Plass on
port-arm
2016-06-17 03:27:50 +00:00
christos
874963f630 Harmonize the error messages about dynamic relocations. 2016-06-15 14:47:23 +00:00
christos
dbe760216f small progress. 2016-06-11 23:10:01 +00:00
christos
fae93e9b6d fix compilation and getdents. 2016-06-11 20:45:07 +00:00
christos
9cc7345716 More fixed from Rin Okuyama
fix waitpid, getdents, explain ptrace, cleanup redefinitions
2016-06-11 13:39:26 +00:00
mrg
4218af5e5e for the eabi targetted compiler, if not compiling libgcc for eabi ensure
that libgcc doesn't try to define __[CD]TOR_LIST__[].
2016-06-10 01:14:00 +00:00
mrg
36b9a4317d add missed file for ppc64 gcc5.4 mknative. 2016-06-10 01:13:05 +00:00
mrg
121c341c36 update powerpc64 info 2016-06-09 23:36:53 +00:00
mrg
2059bccd1a mknative-gcc for powerpc64 and GCC 5.4. 2016-06-09 23:30:18 +00:00
mrg
2303aa3bf1 enable checks for no dot syms on netbsd ppc64, fixes build issues
in libutil where the opd() function conflicts with the .opd section.

remove a couple of tainted defines that are handled differently now.

this, with mknative-gcc as well, enables ppc64 world to build again.
2016-06-09 23:28:22 +00:00
christos
32f250035b Fix syscall argument passing from Ryn Okuyama (with minor changes; hope I did
not make it worse :-)
2016-06-09 14:37:06 +00:00
christos
b95d940b2e Disable glibc workaround for non linux. Rin Okuyama 2016-06-09 14:30:57 +00:00
mrg
a884da9efe add netbsd padding to 'struct addrinfo' copy.
eeew, this is really horrible
2016-06-09 07:38:45 +00:00
mrg
64bc63915b mknative-gcc for sparc64 and GCC 5.4 (and a bit of missed vax, too) 2016-06-09 07:37:49 +00:00
mrg
ed02bbe8b6 mknative-gcc for vax and GCC 5.4. 2016-06-09 07:13:10 +00:00
mrg
4eeb3be5f4 apply -O1 more liberally for vax. fixes build. 2016-06-09 07:11:08 +00:00
mrg
daaf4abadd regen (minor version updates) 2016-06-09 04:42:50 +00:00
mrg
bc64ec48e6 mknative-gcc for all but sparc64, vax and sun2 and GCC 5.4. 2016-06-07 18:53:01 +00:00
mrg
bf2dc96f03 merge GCC 5.4.0. (needs mknative, may not build yet.) 2016-06-07 06:14:16 +00:00
mrg
f30ff5883e import GCC 5.4.0 release. there's not a lot of new info though at
least these 135 (!) GCC PRs have been fixed by the update:

68730 69714 67550 70209 71254 70839 69737 70067 67355 67172 69239
65779 69546 70272 70421 65985 67339 67411 68309 68585 68679 68890
68949 69009 70139 70494 68162 69135 70306 68965 70297 70635 66786
69098 70347 69719 70526 70941 69400 69577 69447 65689 65886 65932
66655 68269 68789 69614 69648 69666 69764 69794 70044 70052 65726
68910 64289 68671 68835 69669 70329 71204 69355 67364 68049 68998
69323 69743 69995 69146 68651 67755 67484 68790 68907 69099 69496
69509 69516 70393 69222 69703 69939 70609 71004 71005 71036 71037
71038 68636 69013 69606 70115 70333 70430 60290 70356 69305 70024
67781 69414 69140 70510 60164 66635 67896 68106 68298 68449 68779
68921 68986 69037 69147 69194 69366 69399 69705 69917 69969 70613
71317 69268 70269 69032 65702 69219 69484 65996 66680 68283 69603
70350 67451 61397
2016-06-07 05:57:41 +00:00
martin
5de48bab33 Build fix for big endian arm, from Rin Okuyama.
Still not working, but other details still under discussion.
2016-06-06 21:06:33 +00:00
christos
5cbcac72ef Don't re-define standard types! 2016-06-05 20:09:00 +00:00
christos
86941a45f0 add glue for unimplemented platforms 2016-06-05 20:08:26 +00:00
christos
e6053aea0a comment out the ones we don't have 2016-06-05 19:17:23 +00:00
christos
543c8f279f Don't play type games with size_t for NetBSD. It is either 32 bits for ILP32
or 64 bits for LP64.
2016-06-05 16:43:10 +00:00
christos
d10a0033d8 add ubsan to the build 2016-06-05 01:17:57 +00:00
christos
8511b0d02a PR/50773: Rin Okuyama: Fix single stepping on dynamically linked binaries
by skipping dynamic linker calls (from OpenBSD)
2016-06-04 16:14:29 +00:00
christos
08cb3c9559 Fix compilation on non-x86 2016-06-03 15:53:18 +00:00
christos
bd2adf4ebf Fix ILP32 build. 2016-06-02 19:54:38 +00:00
christos
63a24122d2 Add ubsan (not enabled, not tested, compiling)
Update tsan (not enable, not tested, not compiling)
Merge common infrastructure.
XXX: Perhaps the common code should be made into a separate library?
Or always require asan?
2016-06-01 22:48:55 +00:00
christos
d94aad05fc Put the sanitizer libraries before everything else, since it needs to be
the first shared object.
2016-06-01 22:27:49 +00:00
christos
214e23d4e1 Put back asan now it works. 2016-06-01 17:07:13 +00:00
christos
860d3f06d6 deal with rename lossage 2016-06-01 17:03:26 +00:00
christos
8c05e0f285 Don't intercept things we don't have. 2016-06-01 17:02:54 +00:00
christos
3711470d09 Handle wrapping the same way as FreeBSD does. 2016-06-01 17:02:19 +00:00
christos
294de9a807 missing space 2016-06-01 04:20:20 +00:00
christos
78fc04143a use the default offsets since gcc hardcodes them (TARGET_ASAN_SHADOW_OFFSET ->
ix86_asan_shadow_offset)
2016-06-01 04:07:51 +00:00
christos
997b5166e9 use the correct array size. 2016-06-01 04:06:15 +00:00