disable for sparc/mips, need to analyze later

avoid some compiler warnings

Patch seen here:
http://lists.uclibc.org/pipermail/uclibc/2015-April/048892.html

just use 4k pages for microblaze, the config are missing and never committed into uClibc

h8300 has no NPTL

disable not available CFLAGS for h8300

resolve merge conflicts

use static directories

The list might not exist on target when cross-compiling.

getconf.c: undef VERSION

might come in via nptl so undef it before redefining it

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

ARC: enable IPv6 in defconfigs

These days IPv6 is used more and more in different software
packages. And so we're adding IPv6 support by default in uClibc
for ARC cores.

Signed-off-by: Alexey Brodkin <Alexey.Brodkin@synopsys.com>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: Anton Kolesov <akolesov@synopsys.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

h8300: Makefile update

Signed-off-by: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

h8300: Add new feature

Signed-off-by: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

h8300: headers update

- wire up new kernel

Signed-off-by: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

h8300: Assembly functions

- remove symbol prefix
- new startup
- new clone syscall support

Signed-off-by: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

h8300: 64bit integer support

Signed-off-by: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

h8300: config update

- New toolchain
- Add new flags

Signed-off-by: Yoshinori Sato <ysato@users.sourceforge.jp>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

implement experimental pure-sh testsuite runner and generation

add thanks file

siginfo: add signal info for seccomp related SIGSYS

uClibc doesn't define signal info for the SIGSYS signal which is issued
in case of hitting a syscall prohibited by seccomp.
This is sad as it makes debugging seccomp filter policies impossible on
some architectures (at least ARM and PowerPC, maybe also others) which
do not coincidentally set si_value.sival_int as the syscall number.

To fix this, import the definitions and macros needed from glibc.

Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

test/silly: Extend include path.

When attempting to build uClibc under buildroot, including building the
tests, the silly tests don't currently compile, a result of attempting
to build using a compiler that does not yet have an installed version of
uClibc available.  The error is a missing header file, specifically
atomic.h.

Taking inspiration from the nptl tests, I have extended the EXTRA_CFLAGS
variable to add the required include paths.  The tests can now be built
under buildroot.

Signed-off-by: Andrew Burgess <andrew.burgess@embecosm.com>
Acked-by: Vineet Gupta <vgupta@synopsys.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

_scanf.c: Implement 'm' modifier for 'c' and '[' conversions.

The current code implements the 'm' modifier only for 's'
conversions and would cause a segfault if it was used for 'c'
or '[' conversions. This patch extends the code to cover these
cases too.

The original version could write scanned data outside the passed buffer
because index i used in the '[' conversion handling block was clobbered.

Signed-off-by: Will Newton <will.newton@imgtec.com>
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

nptl_db: fix ommitting td_ta_setconcurrency

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

init.c (__nptl_initial_report_events): New variable.

(__pthread_initialize_minimal_internal): Initialize pd->report_events
to that.

This patch helps NPTL report TD_CREATE event, so that GDB could catch the
event and update its thread_list.
Link: http://lists.uclibc.org/pipermail/uclibc/2015-April/048921.html
[shengyong:
 - original patch from glibc: commit 7d9d8bd18906fdd17364f372b160d7ab896ce909
 - context adjust
 - update nptl_db/ChangeLog]

Signed-off-by: Roland McGrath <roland@gnu.org>
Signed-off-by: Sheng Yong <shengyong1@huawei.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

buildsys: pregen depends on xlocale

pt-initfini eventually depends on xlocale

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

strftime: comment on %0xY %+nY %-nY

Would fix: date -u +%4Y%2m%2d%2H%2M%2S

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

buildsys: swap V=1 with V=2 command printing

Previously V=1 did print abbreviated commands and V=2 the full commands.

Kbuild-based build-systems behave in the opposite way and this is
apparently confusing or inconvenient for users so swap our V handling to
be in line with kbuild (and automake as far as V=0 / V=1 is concerned).

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

Revert "Do not define unimplemented functions"

This reverts commit bd3eaf83ef1b4954b6c0e7ba8bbdd29b2cd4a833.

They are now implemented.

disabled ARC tests working now

As reported by Alexey Brodkin <Alexey.Brodkin@synopsys.com>
these tests do not fail anymore. After upgrading binutils/gcc
tests compile fine.

buildsys: LT{,.old} CRT prereq

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

buildsys: Tweak pregen wrt headers_dep

Required for !NPTL, !context-funcs for example.

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

ARC: enable more options to satisfy build requirements of applications

As reported by Buildroot autobuilder following options were missing:

 * Libutil stub (UCLIBC_HAS_LIBUTIL option)
   http://autobuild.buildroot.org/results/ce3/ce39eb9b9ece0968563641fb2207099d1a37b191/

 * Program_invocation_name (UCLIBC_HAS_PROGRAM_INVOCATION_NAME=y)
   http://autobuild.buildroot.org/results/154/1546d909e606daefd41b87dece94d642c0fdeba4/
   http://autobuild.buildroot.org/results/bd5/bd54581d7b0cc73bc501072d27e870a443dfce79/

 * Ifaddrd support (UCLIBC_SUPPORT_AI_ADDRCONFIG=y)
   http://autobuild.buildroot.org/results/134/134e78ef1fa87f7fbf26c23ec5dfc68785d79613/

 * Libnsl (UCLIBC_HAS_LIBNSL_STUB=y)
   http://autobuild.buildroot.net/results/331/331ed781b422448205fb9e7c9730ec0c438d6306/
   http://autobuild.buildroot.net/results/402/402d64965ac7ac6e1d4e1990080394958802fe8c/
   http://autobuild.buildroot.net/results/960/9605bac2972d3e3d3fb91947ae6921e89210247b/
   http://autobuild.buildroot.net/results/6b6/6b61ea80a3a6dcead233c4b408eba8b8d647e841/

* UTMP got reworked recently, which breaks packages such as busybox, gdbserver
  etc so enable that too

Enabling mentioned options to make sure more packages could be built with ARC pre-built uClibc tools.

Note UCLIBC_USE_NETLINK is a prerequisite for UCLIBC_SUPPORT_AI_ADDRCONFIG.

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Anton Kolesov <akolesov@synopsys.com>
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

buildsys: pregen depends on xlocale

since ae9e3f46 sched.h pulls in stdlib.h (for malloc() and free()) and
thus also xlocale..

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

arm: Add BX and BXC macros

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

arm: Fix POP_RET for armv4t && interworking

It seems the condition was reversed which lead to e.g. arm-920t being
confused

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

NPTL: Rename a variable

There seems to be a bug in gold with static TLS at least on x86_64 (?)

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

ldso: PRELINK: Remove surplus newline

in early debugging code

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

buildsys: try to merge all constants

Saves a couple of bytes

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

return NULL for realloc(p,0) like glibc

See discussion here about the issue:
https://sourceware.org/bugzilla/show_bug.cgi?id=12547

Fixes testsuite errors.

simplify malloc, remove sbrk support

stick with one malloc implementation for all

Supporting three different malloc implementations seems
a big overhead to be avoided. Just use malloc, which
works for MMU/no-MMU devices just fine.

fix static linking of pthread apps

When compiling python you get duplicate symbol problem.
Seen in the autobuilders of buildroot project.

use weak to fix f.e. cdrkit static compile. fixes #3

arm: thumb1: Fix conflicting types for _v3

In file included from ./include/sys/syscall.h:33:0,
                 from libc/sysdeps/linux/common/sync_file_range.c:10:
libc/sysdeps/linux/common/sync_file_range.c: In function '__sync_file_range_nocancel':
./include/bits/syscalls.h:144:16: error: conflicting types for '_v3'
   register int _v3 __asm__ ("v3") = _v3tmp;
                ^
./libc/sysdeps/linux/arm/sysdep.h:281:7: note: in expansion of macro 'LOAD_ARGS_7'
       LOAD_ARGS_##nr (args)     \
       ^
./libc/sysdeps/linux/arm/sysdep.h:324:2: note: in expansion of macro 'INTERNAL_SYSCALL_RAW'
  INTERNAL_SYSCALL_RAW(SYS_ify(name), err, nr, args)
  ^
./libc/sysdeps/linux/arm/sysdep.h:256:40: note: in expansion of macro 'INTERNAL_SYSCALL'
   ({ unsigned int _inline_sys_result = INTERNAL_SYSCALL (name, , nr, args); \
                                        ^
libc/sysdeps/linux/common/sync_file_range.c:32:9: note: in expansion of macro 'INLINE_SYSCALL'
  return INLINE_SYSCALL(sync_file_range, 7, fd, 0,
         ^
In file included from ./libpthread/nptl/sysdeps/unix/sysv/linux/arm/sysdep-cancel.h:18:0,
                 from ./include/cancel.h:58,
                 from libc/sysdeps/linux/common/sync_file_range.c:15:
./libc/sysdeps/linux/arm/sysdep.h:280:21: note: previous definition of '_v3' was here
       register int *_v3 __asm__ ("v3") = _sys_buf;  \
...

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

librt: Add missing __dso_handle

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

librt: Refine LIBS

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

librt: honour HAS_STUBS in buildsys

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

include/: ignore sys/random.h symlink

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

librt: Rephrase librt.so library dependencies

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

librt: Fix librt.so depends for !NPTL

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

prelink: handle _begin in a gold-agnostic way

The nostartfiles is redundant but better be safe

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

buildsys: pass -O to ld unless DODEBUG

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

fix compile error when LDSO_RUNPATH_OF_EXECUTABLE is set

remove more of the link_warnings

Only the stub warnings left for now.

remove link warnings

As recently discussed on the pgsql mailinglist, this
warnings are more or less useless and some configure scripts
are failing when these warnings are enabled.

http://www.postgresql.org/message-id/20150320132351.GS3636@alvh.no-ip.org

remove regex old and fnmatch old

for xtensa this is still required to avoid build breakage

libc: Fix page-size in getifaddrs()

TODO: this could need a cleanup..

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

buildsys: Do not build crt upon pregen

No need to build crt when just generating headers

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

buildsys: tweak ucontext_i.h prerequisites

ucontext_i.h too depends on uClibc_config.h.
Update it to look like the other MANGLE generated files and tweak
prereqs

Fixes: make realclean ; make -j libc/misc/internals/__uClibc_main.i

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

buildsys: Fix setting LIBGCC

We were ending up with multiple entries in LIBGCC for recursive child
make instances, avoid that.

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

Revert "librt: Use -nodefaultlibs instead of -nostdlib"

This reverts commit 534f44d53146457b3ca686c47efb9207543b88e1.

I don't think this is wanted.

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

we need -fPIC for the new test for m68k binutils problem

remove sh64 support

No real hardware available. The project for sh64 with sh5 seems
dead since 10 years.

Revert "gcc 4.9.x produces some calls to abort()"

This reverts commit 8d73c06c7a78745a50dc8683fd117aeacd71a04b.

Fixed in a better way by Bernhard in uClibc.

gcc 4.4 does not have __builtin_unreachable

For avr32 we still use gcc 4.4 compiler, so exlude the usage
of __builtin_unreachable here.

disable new tests, atomic.h is missing

merge uClibc master

ARC: don't hard-code ELF_NGREG

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
[updated changelog]

Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

ARC/signal: shield sa_restorer from compiler toggle side-effects

when building uClibc with -O0 (DODEBUG build) the default sigrestorer
had some extra glue code generated for stack manipulation which was
messing up resume from signal path.

So annotate the function with -Os so that gcc would only generate the
bare min 2 instruction TRAP sequence

Reported-and-Debugged-by: Alexey Brodkin <abrodkin@synopsys.com>
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

ARC: switch to NPTL

Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

Add support for R_ARC_NONE relocations.

The R_ARC_NONE relocation is generated when --gc-sections, used by the
compile time linker, removes some sections.  This is completely normal,
and we can see that all other targets (based on random sampling) have
support for R_*_NONE relocations (named for each target).

Handling R_ARC_NONE involves doing nothing with it, which is nice and
easy.

Signed-off-by: Andrew Burgess <andrew.burgess@embecosm.com>
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

include: update atomic.h from glibc

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

test: Add SIMULATOR_uclibc and SIMULATOR_glibc

where SIMULATOR_glibc is an optional simulator to run host binaries and
SIMULATOR_uclibc is an optional simulator to run target (uClibc)
binaries.

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

utmp: always have at least utmpx

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

NPTL: SH: silence warning

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

SH: add 't' to syscall clobber list

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

ldso: mark _dl_exit as noreturn

Otherwise gcc might not understand that oom() ended control-flow and
might emit an (untaken) reference to abort() in _dl_update_slotinfo()
on e.g. SH4 which breaks linking ld-uClibc.so.
Arguably -ffreestanding should prevent GCC from emitting this
'.global abort' but alas, it does not, which is another bug..

Also mark the function cold to further lower the incoming frequency and
branch probability.

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

test: tweaks

Fix arc4random exclusion for real.
Silence warning about missing prototype for external helper-function in
tst-scandir

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

include: Expand malloc-family comments

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

utmp: favour POSIX utmpx over SVID utmp

Note: _PATH_UTMPX == _PATH_UTMP and the utmp struct is identical to the
utmpx struct so this only changes the external API entrypoints and NOT
the underlying data source.

This saves about 500b (~1300b from previously ~1950) while at it.

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

utmp: Remove unneeded alias

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

mips: switch float_t to float

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

libm: honor NO_LONG_DOUBLE in ldouble_wrappers

Fixes SH build breakage due to attempts to define hidden defs for
unavailable long double functions.

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

libcrypt: Rename some internal macros

to avoid clashes

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

buildsys: fix IS_IN_lib* for .s

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

utmp: indent

indent only, no code changes

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

utmp: Remove unneeded aliases

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

utmp: add _unlocked suffix to internal helpers

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

buildsys: HAS_UTMP (XPG2, SVr4 compat) knob

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

libc: TIME64_COMPAT32 for sparc, mips

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

Revert "utent.c, wtent.c: move functions from utxent.c"

This reverts commit 84135275cfeebc0b233c1c96eeada4d4178a0b18.

This change is said to make systemd deadlock (cannot reproduce this)

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Conflicts:
include/utmp.h

okay, chicken-egg-problem solved really now. (was xtensa no thread failure)

test: add utmpx test

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

libc: add getrandom(2)

Introduce a <sys/random.h> for it.

/* FIXME: aren't there a couple of __restrict and const missing ?
 */
extern int getrandom(void *__buf, size_t count, unsigned int flags)
    __nonnull ((1)) __wur;

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

librt: Use -nodefaultlibs instead of -nostdlib

nostdlib disables linking in startup files too which is not
what we want here since it needs to resolve __dso_handle
which comes from crtbeginS.o, otherwise librt has this
undefined reference to a weak undefined __dso_handle that
shows up as error (with gold linker)
when shared libraries are being built which are
linking in librt

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

nptl/atfork: Hide pthread_atfork in shared versions

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

libc: Fix vfprintf compilation without HAS_FLOATS

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

nptl/arm: Move aeabi_read_tp to uclibc_nonshared.a

Otherwise it creates wrong references from shared libs

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

atexit_old: Do not add it to shared libc

atexit should only be in either uclibc_nonshared.a
shared libc case or libc.a in static build case

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

test: enable libm test log2

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

extra: add clang flags for gen_bits_syscall_h

Closes bugzilla #6008

Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>