ARM: detect BX availibility at build time

The "use BX" option is now a suggestion that BX be used if available.
Use a macro to detect if BX is available at build time. If so, and
the user requested it be used, then use it. Otherwise, error out.

Macro courtesy Khem RAJ:
  http://lists.uclibc.org/pipermail/uclibc/2009-April/042301.html

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@anciens.enib.fr>
Cc: Khem Raj <raj.khem@gmail.com>
Cc: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Cc: Carmelo AMOROSO <carmelo.amoroso@st.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>

ARM: #include <bits/arm_asm.h> where __USE_BX__ is used

The check for __USE_BX__ will be available in bits/arm_asm.h,
so the latter must be included wherever the former is used.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@anciens.enib.fr>
Cc: Khem Raj <raj.khem@gmail.com>
Cc: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Cc: Carmelo AMOROSO <carmelo.amoroso@st.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>

ARM: transform the EABI/OABI choice into a boolean

The CONFIG_ARM_OABI option is never used.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@anciens.enib.fr>
Cc: Khem Raj <raj.khem@gmail.com>
Cc: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Cc: Carmelo AMOROSO <carmelo.amoroso@st.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>

ARM: remove sub-arch/variants selection from menuconfig

Rely on the compiler to be correctly set up to generate
appropriate code for the target variant.

This exposes the Thumb option, as it is no longer auto-selected.
The "Use BX" no longer depends on supported CPU to be selected,
so it now defaults to 'n' as it shall work by default on CPUs
that do not have BX.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@anciens.enib.fr>
Cc: Khem Raj <raj.khem@gmail.com>
Cc: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Cc: Carmelo AMOROSO <carmelo.amoroso@st.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>

ARM: introduce blind options to select & force THUMB mode

Add three new blind options to set use of Thumb mode:
- COMPILE_IN_THUMB_MODE
  - if set, CFLAGS will contain -mthumb
  - if unset, the compiler's default is used
- HAS_THUMB
  - CPUS with Thumb instruction set can select this
  - use of BX depends on this
- FORCE_THUMB
  - CPUs that are Thumb-only must select this
  - this selects: HAS_THUMB, COMPILE_IN_THUMB_MODE and USE_BX

Also, remove leading space in Rules.mak.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@anciens.enib.fr>
Cc: Khem Raj <raj.khem@gmail.com>
Cc: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Cc: Carmelo AMOROSO <carmelo.amoroso@st.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>

ARM: reorder "Use BX" option

"Use BX" is not available on all CPUs, so the option depends on
a correct CPU to be chosen . It is weird that e BX" then appears
_above_ the CPU selection, not below.

Move the "Use BX" after the CPU selection.

Signed-off-by: "Yann E. MORIN" <yann.morin.1998@anciens.enib.fr>
Cc: Khem Raj <raj.khem@gmail.com>
Cc: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Cc: Carmelo AMOROSO <carmelo.amoroso@st.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>

Fix __libc_epoll_pwait compile failure on x86

This prevents "memory input 7 is not directly addressable" errors.

| libc/sysdeps/linux/common/epoll.c: In function '__libc_epoll_pwait':
| libc/sysdeps/linux/common/epoll.c:71:80: error: memory input 7 is not directly addressable
| libc/sysdeps/linux/common/epoll.c:75:86: error: memory input 7 is not directly addressable
| make: *** [libc/sysdeps/linux/common/epoll.o] Error 1
| make: *** Waiting for unfinished jobs....

Signed-off-by: Phil Blundell <pb@pbcl.net>
Signed-off-by: Jason Woodward <jason.woodward@timesys.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>

nptl/Makefile.in: Fix the build break caused by UCLIBC_CTOR_DTOR enabling

Extra / somehow does not match the target and complains that
$(top_builddir)libpthread/nptl/sysdeps/pthread/crti.o has no rules to
build

Signed-off-by: Khem Raj <raj.khem@gmail.com>

buildsys: fix pregen target (!NPTL with LOCALE)

When NPTL is not enabled, and LOCALE support is used, compilation fails
with the following error:

-------------------------------------------------------------------------
  MKDIR include/bits
  GEN include/bits/sysnum.h
make[1]: *** No rule to make target `../..//include/bits/uClibc_ctype.h', needed by `../../extra/locale/gen_wc8bit.c'.  Stop.
make: *** [pregen] Error 2
-------------------------------------------------------------------------

The problem is that "headers" is not listed as prerequisite of pregen.
These have been changed multiple times recently:

o Austin: order headers before pregen-headers
1f5c73c9f8a98b1d2a35cea868a585c97ab0e436
o Khem: pregen: Fix the parallel build problem in pregen target
046035ecde92262d96eff2192ba3cda716f04909
o Khem: pregen-headers: Add new target which depends on headers.
ef18cfe8ebab25f5ef92e81956f50e2dc57df602
o Carmelo: build: Fix infinite loop when no threading support is enabled
ff5e4de7088fe8d34812c2a1e604bf04be713606
In all of these commits headers was actually as prerequisite

In commit 9381d622e2411a35a5fd73a5a573eb269e2dd9c9 [nptl: fix buildsys] by Bernhard, headers was removed,
causing the problem reported above.

Signed-off-by: Filippo Arcidiacono <filippo.arcidiacono@st.com>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>

libubacktrace: Fix ASNEEDED emission

In master the variable is called SHARED_LIBNAME and not
SHARED_MAJORNAME.

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

NPTL: Fix CTOR_DTOR

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

libc: make common longjmp usable with NPTL

Call _longjmp_unwind conditionally under NPTL config option,
making longjmp usable with NPTL.

Signed-off-by: Filippo Arcidiacono <filippo.arcidiacono@st.com>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>

sh: use the common implementation for longjmp

Signed-off-by: Filippo Arcidiacono <filippo.arcidiacono@st.com>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>

sh: fix build in !NPTL case

Fix build issue due to missing symbols in !NPTL case:
- vfork
- _longjmp_unwind

Signed-off-by: Filippo Arcidiacono <filippo.arcidiacono@st.com>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>

Correct a bug when remapping textrel segments on nommu

From: Alan Davis <adavis@ti.com>

On C6X, when trying to execute a program that has a textrel DSO, it
fails to load. The telltale line in the LD_DEBUG output is:

_dl_get_ready_to_run:779:       file='';  needed by './a.out'

The corresponding DT_NEEDED entry has 'libc.so.0', but here the
filename is empty. This is what is happening in
_dl_elf_shared_library():

First, map all segments according to their permissions. Text gets
initially mapped read-only.

Then, parse the dynamic information. The dynamic table is in RW but
some of the tags may point to RO. For example, DT_NEEDED points to a
string in .dynstr which is in RO. These pointers get computed
according to the loadmap from the original mapping.

Then, in response to a DT_TEXTREL tag, the RO segment gets remapped,
thereby invaliding anything that points to it, in particular certain
dynamic tags such as DT_NEEDED.

The following patch re-parses the dynamic info after the remapping so
as to re-compute any invalid pointers.

Signed-off-by: Alan Davis <adavis@ti.com>
Signed-off-by: Bernd Schmidt <bernds@codesourcery.com>

unwind.h: Move to libc/sysdeps from nptl sysdeps

Signed-off-by: Khem Raj <raj.khem@gmail.com>

epoll: redo cancellation to match master

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

epoll.c: add cancellation to epoll_[p]wait()

While there, fix epoll_pwait syscall, it takes 6 arguments

Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

libm: fix building out-of-tree

Thanks to Thierry Reding for noticing!

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

wordexp: fix mishandled return value of pipe call

and use pipe2 if available while at it.

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

linux_specific: handle accept4 and pipe2

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

NPTL: needs madvise internally

This fixes 8d09a50a044638fde2ed3e1a1c4d3c7c5a3cce5c since NPTL also uses
madvise internally.

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

socket: fix stubs

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

commentary typo fix

s/decriptor/descriptor/

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

epoll: unify epoll.h

commentary typo fix

use cancellation (with two 'l') uniformly.

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

Implement epoll_create1 and epoll_pwait system calls.

Note: TODO: This lacks cancellation support.

Signed-off-by: Thierry Reding <thierry.reding@avionic-design.de>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

buildsys: document UCLIBC_EXTRA_CPPFLAGS

Rename EXTRA_CPPFLAGS to UCLIBC_EXTRA_CPPFLAGS and make them override
previous flags (by appending - not prepending - them to the other flags).

lll: avoid shadow warnings

use __ret uniformly to avoid shadow warnings as seen on x86_64.

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

x86_64: add TODO for R_X86_64_RELATIVE

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

libubacktrace: mv unwind.h to include

Fixes compilation if NPTL is off.
I hope this doesn't break ARM EABI but cannot check ATM.

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

Revert "libubacktrace: fix config-symbol handling"

This reverts commit 7a080cd149c7b25d415d76506510d55b34819fc2.

This symbol is in fact different (as it is set no 'n').

add pipe2()

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

socketcalls: add stubs

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

accept4: Implement cancellation

.. and add proper prototype, move it into it's own obj and other such
cleanups.

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

Implement accept4 system call.

Signed-off-by: Thierry Reding <thierry.reding@avionic-design.de>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

libubacktrace: fix config-symbol handling

Config symbals that are not set are empty, not 'n'.

buildsys: Turn on strict aliasing for everyone

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

ldso: commentary typo fix

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

getcwd: Fix memory-leak in error path

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

__uc_malloc: Fix memory-leak in error path

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

libdl: fix local symbol's address handling in dladdr

Fix dladdr to correctly handle local function's address so backtrace_symbols
print only the function address for these function, instead of showing the name
of nearest one.
Indeed the dladdr walk through the hash table to find the nearest symbol, that
doesn't contain local symbols.

Signed-off-by: Filippo Arcidiacono <filippo.arcidiacono@st.com>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>

*/crtn.S: Remove .size directive for _init and _fini

These are split across objects so setting size does not (and never did)
work since the expression cannot be computed at assembly time.

This avoids errors from recent (> 2.21) gas.

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

posix_fadvise64: fix x86 implementation

Commit 73d59554144f429b1cf0d4d7fa7de42bdf59ad92 completely broke
the x86 implementation of posix_fadvise64. It moved the first
the assembly code retn instruction gets missing depending on the

Technically the file has two implementaions for posix_fadvise64,
one when __NR_fadvise64_64 is available, and second one if only
__NR_fadvise64 is there. Fix the #ifdef's to be proper for that.

Signed-off-by: Timo Teräs <timo.teras@iki.fi>
Signed-off-by: Khem Raj <raj.khem@gmail.com>

i386: fix .size of _init/_fini

binutils-2.21 barf on .size that do not evaluate to const, so use
the section size and not a function that is not visible here.

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

x86_64: fix .size of _init/_fini

binutils-2.21 barf on .size that do not evaluate to const, so use the
section size and not a function that is not visible here.

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

Revert "*/crt?.S: remove .size directives"

Removing them generally was not a good idea

This reverts commit 233c504cd940d9802226b6a3a092368b86978f5e.

*/crt?.S: remove .size directives

sed -i -e '/\.size[[:space:]]/d' $(grep -l "\.size" libc/sysdeps/linux/*/crt*.[sSc])

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

libubacktrace: fix backtrace support on arm-eabi, which needs libgcc_eh linked too

Signed-off-by: Henning Heinold <heinold@inf.fu-berlin.de>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>

getaddrinfo.c: fix incorrect check for ERANGE from gethostbyaddr_r

Also, freeaddrinfo(NULL) is ok, no need to check parameted for NULL
before calling it.

Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>

getaddrinfo.c: improve code readability. No functional changes

Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>

string: remove unused variable

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

x86_64: silence warning if !TLS

TODO: fix all other arches

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

buildsys: prettify ssp.c handling

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

madvise is LINUX_SPECIFIC

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

test_nptl: fix expected result for tst-cputimer[123]

Signed-off-by: Filippo Arcidiacono <filippo.arcidiacono@st.com>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>

test_nptl: fix expected result for tst-clock2 test

tst-clock2 should return 0 when _SC_THREAD_CPUTIME option isn't
available, instead of treating it as an error.
Further set the expected ret value as 0 avoiding to hide any real failures
in case of THREAD_CPUTIME feature available.

Signed-off-by: Filippo Arcidiacono <filippo.arcidiacono@st.com>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>

buildsys: make $(LOCAL_INSTALL_PATH) phony

Could depend on the individual files too but this makes sure that
nothing is missed.

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

ether_aton: reject invalid input

fixes PR2227

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

tests: disable ether tests if !HAS_SOCKET

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

inet: add ether_aton testcase

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

sysconf: clock_getres depends on HAS_REALTIME

Bug was introduced in revision a202cf6f.

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

__rt_sigwaitinfo: depends on HAS_REALTIME

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

buildsys: minor fixes in Makefile.arch for C6X

Just a tidy-up by removing commented-out lines.

Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>

buildsys: minor fixes in Makefile.arch for microblaze

Do not include Makefile.commonarch directly from within arch specific
Makefile, as it is already done in parent Maefile.in.

Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>

libubacktrace: enabled for all archs indeed.

Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>

sparc: don't access fp registers when configured for no fpu

Signed-off-by: Austin Foxley <austinf@cetoncorp.com>

libubacktrace: generic implementation based dwarf

Use the initial implementation for SH4 based on dwarf for all archs.
Indeed there are not obvious reason for which it should not work in general.

Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>

Added fts support for traversing UNIX file hierarchies.

It is required by libdwfl in elfutils package.

Signed-off-by: Salvatore Cro <salvatore.cro at st.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>

test: fix tst-tls13 expected result and timeout

tst-tls13 is expected to return 0 when passing.
Set higher timeout to avoid failure on slow archs.

Signed-off-by: Salvatore Cro <salvatore.cro@st.com>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>

dl-startup: fix typos in block comment

Signed-off-by: Kevin Cernekee <cernekee@gmail.com>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>

MIPS LDSO: pass sym_ref parameter to _dl_find_hash() to support PROTECTED symbols

_dl_find_hash() relies on sym_ref parameter to check if the looked-up symbol
is protected. The code fixes a case when _dl_perform_mips_global_got_relocations()
was calling _dl_find_hash() without providing sym_ref parameter.
The bug was causing hangs if a library exporting non-protected symbol was earlier in
link order than library with the same symbol declared as protected.

Signed-off-by: Maksim Rayskiy <mrayskiy@broadcom.com>
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>

buildsys: do_rm ARCH_HEADERS

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

resolv: really fix res_close not to hang with ipv6

Fix goof in previous commit.

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

buildsys: do not use $(and)

make-3.80 does not have $(and) and $(or), so workaround for now.

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

buildsys: fix GNU make v3.80 compatibility again

GNU make 3.80 cannot handle "$(and)" or "$(or)" from commit 18e7136e
(buildsys: use kbuild style).  Replace them with ifeq/ifneq.

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

resolv: fix res_close not to hang with ipv6

Timo Teräs writes:
The memory release loop is missing an obvious counter increment.

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

Support dynamic assignment of DSBT_INDEX

For DSBT targets (C6X only at this point), we'd like to support the case
where the user did not specify --dsbt-index at link time when building a
shared library. The dynamic linker can still assign an index at runtime
and fix up the DSBT_INDEX relocs, at the cost of startup time and memory
space.

Signed-off-by: Bernd Schmidt <bernds@codesourcery.com>

Fix malloc alignment

In commit 3e0a1f388, Richard tried to fix malloc alignments by using
 alignof (double __attribute_aligned__(sizeof (size_t))).
This doesn't work, since attribute_aligned overrides the alignment
rather than providing a minimum. On C6X, malloc returns four-byte
aligned values rather than the necessary eight-byte alignment.

It's simpler to use a comparison and pick the bigger of the two values,
so that's what I've done.

Signed-off-by: Bernd Schmidt <bernds@codesourcery.com>

Fix nommu handling of DT_TEXTREL

We have a problem with DT_TEXTREL shared libraries on nommu machines.
The dynamic linker's strategy is to map the text segment read-only
first, then look for DT_TEXTREL, and use mprotect to change protections
if necessary. This fails on nommu, since a nommu kernel can decide to
share the memory for private read-only file mappings, and mprotect
doesn't (can't) do anything about this sharing. Existing nommu targets
apparently have no need for this, but on C6X, we may need to assign
library indices at run-time if no --dsbt-index option was passed to the
linker at build time.

Hence, the following patch, which instead of using mprotect, redoes the
mapping with PF_W set.

Signed-off-by: Bernd Schmidt <bernds@codesourcery.com>

utils/ldd: Check for returned pointer from strrchr not the value it holds

Signed-off-by: Kevin Day <thekevinday@gmail.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>

cris: add provide arch-specific vfork implementation

vfork must be called with nothing at all on the stack,
so implementing it via syscall() does not work.

Signed-off-by: Thorsten Glaser <tg@mirbsd.org>
Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>

lutimes.c, stubs.c: fix compiling lutimes, if __NR_utimensat is not defined

while there, provide stubs for functions depending on utimensat syscall.

Reported-by: Sedat Dilek <sedat.dilek@googlemail.com>
Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>
Signed-off-by: Khem Raj <raj.khem@gmail.com>

bump version to 0.9.32-rc3-git

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

release 0.9.32-rc3

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

memalign: include sys/param.h for MAX

libc/stdlib/malloc/memalign.c:22:1: warning: "MAX" redefined In file included from
./libpthread/nptl/sysdeps/unix/sysv/linux/i386/lowlevellock.h:25,
                 from ./include/bits/libc-lock.h:36,
                 from ./include/bits/stdio-lock.h:23,
                 from ./include/bits/uClibc_mutex.h:71,
                 from libc/stdlib/malloc/malloc.h:135,
                 from libc/stdlib/malloc/memalign.c:18:
./include/sys/param.h:75:1: warning: this is the location of the previous definition

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

arm/bits/atomic.h: Include common/bit/atomic.h for thumb1

This restores the behavior for thumb1 builds and yet uses
the new atomic.h for arm and thumb2 modes.

Signed-off-by: Khem Raj <raj.khem@gmail.com>

wctype.h: fix libc_hidden_proto for iswupper and add it for iswspace

Old typo, became visible due to other changes.

Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>

add libc_hidden_proto for wcs[n]casecmp_l

Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>

really fix missing __libc_drand48_data

Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>

Revert "missing prototype of __libc_drand48_data fixed"

This reverts commit e2cea9b9edabe885b9bf2586f9541ffb8fe93f3f.

Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>

missing prototype of __libc_drand48_data fixed

Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>

time.c, time.h: remove unused hidden strftime/strptime

Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>

nanosleep.c: remove duplicated libc_hidden_proto

Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>

ctype.c, ctype.h: remove commented parts that were banned for removal after 0.9.31

Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>

_wctype.c, wctype.h: remove unused isw* and wctype_l hidden functions

Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>

time.c, wchar.h: remove unused hidden wcsftime

Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>

str[n]casecmp.c: fix hidden usage

Provide visible str[n]casecmp[_l], wcs[n]casecmp[_l].

Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>

remove unused hidden functions

Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>

frv/memset.S: add missing libc_hidden_def

Signed-off-by: Peter S. Mazinger <ps.m@gmx.net>