Similar to my patch for HUGE_VAL, this patch eliminates the bits/inf.h
headers and just unconditionally uses the same definitions as the
sysdeps/ieee754 version did (__builtin_inff () for GCC >= 3.3,
otherwise HUGE_VALF), directly in math.h, so removing an unnecessary
level of indirection.
Tested for x86_64.
* math/math.h [__USE_ISOC99] (INFINITY): Define directly here. Do
not include <bits/inf.h>.
* math/Makefile (headers): Remove bits/inf.h.
* bits/inf.h: Remove.
* sysdeps/ieee754/bits/inf.h: Likewise.
There are various bits/huge_val*.h headers to define HUGE_VAL and
related macros. All of them use __builtin_huge_val etc. for GCC 3.3
and later. Then there are various fallbacks, such as using a large
hex float constant for GCC 2.96 and later, or using unions (with or
without compound literals) to construct the bytes of an infinity, with
this last being the reason for having architecture-specific files.
Supporting TS 18661-3 _FloatN / _FloatNx types that have the same
format as other supported types will mean adding more such macros;
needing to add more headers for them doesn't seem very desirable.
The fallbacks based on bytes of the representation of an infinity do
not meet the standard requirements for a constant expression. At
least one of them is also wrong: sysdeps/sh/bits/huge_val.h is
producing a mixed-endian representation which does not match what GCC
does.
This patch eliminates all those headers, defining the macros directly
in math.h. For GCC 3.3 and later, the built-in functions are used as
now. For other compilers, a large constant 1e10000 (with appropriate
suffix) is used. This is like the fallback for GCC 2.96 and later,
but without using hex floats (which have no apparent advantage here).
It is unambiguously valid standard C for all floating-point formats
with infinities, which covers all formats supported by glibc or likely
to be supported by glibc in future (C90 DR#025 said that if a
floating-point format represents infinities, all real values lie
within the range of representable values, so the constraints for
constant expressions are not violated), but may generate compiler
warnings and wouldn't handle the TS 18661-1 FENV_ROUND pragma
correctly. If someone is actually using a compiler with glibc that
does not claim to be GCC 3.3 or later, but which has a better way to
define the HUGE_VAL macros, we can always add compiler conditionals in
with alternative definitions.
I intend to make similar changes for INF and NAN. The SNAN macros
already just use __builtin_nans etc. with no fallback for compilers
not claiming to be GCC 3.3 or later.
Tested for x86_64.
* math/math.h: Do not include bits/huge_val.h, bits/huge_valf.h,
bits/huge_vall.h or bits/huge_val_flt128.h.
(HUGE_VAL): Define directly here.
[__USE_ISOC99] (HUGE_VALF): Likewise.
[__USE_ISOC99] (HUGE_VALL): Likewise.
[__HAVE_FLOAT128 && __GLIBC_USE (IEC_60559_TYPES_EXT)]
(HUGE_VAL_F128): Likewise.
* math/Makefile (headers): Remove bits/huge_val.h,
bits/huge_valf.h, bits/huge_vall.h and bits/huge_val_flt128.h.
* bits/huge_val.h: Remove.
* bits/huge_val_flt128.h: Likewise.
* bits/huge_valf.h: Likewise.
* bits/huge_vall.h: Likewise.
* sysdeps/ia64/bits/huge_vall.h: Likewise.
* sysdeps/ieee754/bits/huge_val.h: Likewise.
* sysdeps/ieee754/bits/huge_valf.h: Likewise.
* sysdeps/m68k/m680x0/bits/huge_vall.h: Likewise.
* sysdeps/sh/bits/huge_val.h: Likewise.
* sysdeps/sparc/bits/huge_vall.h: Likewise.
* sysdeps/x86/bits/huge_vall.h: Likewise.
Since sofini.os terminates .eh_frame section, it should be placed last.
[BZ #22051]
* Makerules (build-module-helper-objlist): Filter out
$(elf-objpfx)sofini.os.
(build-shlib-objlist): Append $(elf-objpfx)sofini.os if it is
needed.
This patch completes the ucontext.h namespace fixes by fixing issues
related to the use of struct sigcontext as mcontext_t, and inclusion
of <bits/sigcontext.h> even when struct sigcontext is not so used.
Inclusion of <bits/sigcontext.h> by <sys/ucontext.h> is removed; the
way to get the sigcontext structure is by including <signal.h> (in a
context where __USE_MISC is defined); the sysdeps/generic version of
sys/ucontext.h keeps the inclusion by necessity, with a comment about
how this is not namespace-clean, but the only configuration that used
it, MicroBlaze, gets its own version of the header in this patch.
Where mcontext_t was typedefed to struct sigcontext, the contents of
struct sigcontext are inserted (with appropriate namespace handling to
prefix fields with __ when __USE_MISC is not defined); review should
check that this has been done correctly in each case, whether the
definition of struct sigcontext comes from glibc headers or from the
Linux kernel. This changes C++ name mangling on affected
architectures (which do not include x86_64/x86).
Tested for x86_64, and with build-many-glibcs.py.
2017-08-14 Joseph Myers <joseph@codesourcery.com>
[BZ #21457]
* sysdeps/arm/sys/ucontext.h: Do not include <bits/sigcontext.h>.
* sysdeps/generic/sys/ucontext.h: Add comment about use of struct
sigcontext and namespace requirements.
* sysdeps/i386/sys/ucontext.h: Do not include <bits/sigcontext.h>.
* sysdeps/m68k/sys/ucontext.h: Likewise.
* sysdeps/mips/sys/ucontext.h: Likewise. Include <bits/types.h>.
* sysdeps/unix/sysv/linux/aarch64/sys/ucontext.h: Do not include
<bits/sigcontext.h>.
(__ctx): Define earlier.
(mcontext_t): Define structure contents rather than using struct
sigcontext.
* sysdeps/unix/sysv/linux/aarch64/ucontext_i.sym (oEXTENSION): Use
__glibc_reserved1 instead of __reserved.
* sysdeps/unix/sysv/linux/alpha/sys/ucontext.h: Do not include
<bits/sigcontext.h>.
(__ctx): Define earlier.
(mcontext_t): Define structure contents rather than using struct
sigcontext.
* sysdeps/unix/sysv/linux/alpha/ucontext-offsets.sym: Use
mcontext_t instead of struct sigcontext.
* sysdeps/unix/sysv/linux/arm/sys/ucontext.h: Do not include
<bits/sigcontext.h>.
(__ctx): Define earlier.
(mcontext_t): Define structure contents rather than using struct
sigcontext.
* sysdeps/unix/sysv/linux/hppa/sys/ucontext.h: Do not include
<bits/sigcontext.h>.
(__ctx): Define earlier.
(mcontext_t): Define structure contents rather than using struct
sigcontext.
* sysdeps/unix/sysv/linux/ia64/makecontext.c (__makecontext): Use
mcontext_t instead of struct sigcontext.
* sysdeps/unix/sysv/linux/ia64/sigcontext-offsets.sym: Use
mcontext_t instead of struct sigcontext.
* sysdeps/unix/sysv/linux/ia64/sys/ucontext.h: Do not include
<bits/sigcontext.h>.
(__ctx): New macro.
(struct __ia64_fpreg_mcontext): New type.
(mcontext_t): Define structure contents rather than using struct
sigcontext.
(_SC_GR0_OFFSET): Use mcontext_t instead of struct sigcontext.
(uc_sigmask): Define using __ctx.
(uc_stack): Likewise.
* sysdeps/unix/sysv/linux/ia64/sys/procfs.h: Include
<bits/sigcontext.h>.
* sysdeps/unix/sysv/linux/ia64/sys/ptrace.h: Likewise.
* sysdeps/unix/sysv/linux/m68k/sys/ucontext.h: Do not include
<bits/sigcontext.h>.
* sysdeps/unix/sysv/linux/microblaze/sys/ucontext.h: New file.
* sysdeps/unix/sysv/linux/mips/sys/ucontext.h: Do not include
<bits/sigcontext.h>.
* sysdeps/unix/sysv/linux/nios2/sys/ucontext.h: Do not include
<bits/sigcontext.h>.
* sysdeps/unix/sysv/linux/powerpc/sys/ucontext.h: Do not include
<bits/sigcontext.h>.
* sysdeps/unix/sysv/linux/s390/sys/ucontext.h: Do not include
<bits/sigcontext.h>.
* sysdeps/unix/sysv/linux/sh/sys/ucontext.h: Do not include
<bits/sigcontext.h>.
* sysdeps/unix/sysv/linux/sparc/sys/ucontext.h: Do not include
<bits/sigcontext.h>.
* sysdeps/unix/sysv/linux/tile/sys/ucontext.h: Do not include
<bits/sigcontext.h>.
(__ctx): Define earlier.
(mcontext_t): Define structure contents rather than using struct
sigcontext.
* sysdeps/unix/sysv/linux/x86/sys/ucontext.h: Do not include
<bits/sigcontext.h>. Include <bits/types.h>.
* conform/Makefile (test-xfail-XPG42/signal.h/conform): Remove.
(test-xfail-XPG42/sys/wait.h/conform): Likewise.
(test-xfail-XPG42/ucontext.h/conform): Likewise.
(test-xfail-UNIX98/signal.h/conform): Likewise.
(test-xfail-UNIX98/sys/wait.h/conform): Likewise.
(test-xfail-UNIX98/ucontext.h/conform): Likewise.
(test-xfail-XOPEN2K/signal.h/conform): Likewise.
(test-xfail-XOPEN2K/sys/wait.h/conform): Likewise.
(test-xfail-XOPEN2K/ucontext.h/conform): Likewise.
(test-xfail-POSIX2008/signal.h/conform): Likewise.
(test-xfail-POSIX2008/sys/wait.h/conform): Likewise.
(test-xfail-XOPEN2K8/signal.h/conform): Likewise.
(test-xfail-XOPEN2K8/sys/wait.h/conform): Likewise.
Clean up calls to malloc_printerr and trim its argument list.
This also removes a few bits of work done before calling
malloc_printerr (such as unlocking operations).
The tunable/environment variable still enables the lightweight
additional malloc checking, but mallopt (M_CHECK_ACTION)
no longer has any effect.
On very large multi-processor systems, creating hundreds of threads
runs into a test time out. The tests do not seem to benefit from
massive over-scheduling.
As shown by build bot failures
<https://sourceware.org/ml/libc-testresults/2017-q3/msg00349.html> the
m68k bits/mathinline.h is not namespace-clean: it fails to compile if
the user has defined macros f or l before it is included, because of
expansions of those arguments to __inline_functions. This patch
changes the __inline_functions definitions to take not the suffix but
a macro that concatenates it with the function name, to avoid the
spurious macro expansions.
Tested for m68k with build-many-glibcs.py.
[BZ #22035]
* sysdeps/m68k/m680x0/fpu/bits/mathinline.h (__inline_functions):
Define to take a second argument that is a macro that
concatentates a suffix, not the suffix itself.
(__CONCAT_d): New macro.
(__CONCAT_f): Likewise.
(__CONCAT_l): Likewise.
Fix a commit cc25c8b4c119 ("New pthread rwlock that is more scalable.")
regression and prevent uncontrolled stack space usage from happening
when a 5-, 6- or 7-argument syscall wrapper is placed in a loop.
The cause of the problem is the use of `alloca' in regular MIPS/Linux
wrappers to force the use of the frame pointer register in any function
using one or more of these wrappers. Using the frame pointer register
is required so as not to break frame unwinding as the the stack pointer
is lowered within the inline asm used by these wrappers to make room for
the stack arguments, which 5-, 6- and 7-argument syscalls use with the
o32 ABI.
The regular MIPS/Linux wrappers are macros however, expanded inline, and
stack allocations made with `alloca' are not discarded until the return
of the function they are made in. Consequently if called in a loop,
then virtual memory is wasted, and if the loop goes through enough
iterations, then ultimately available memory can get exhausted causing
the program to crash.
Address the issue by replacing the inline code with standalone assembly
functions, which rely on the compiler arranging syscall arguments
according to the o32 function calling convention, which MIPS/Linux
syscalls also use, except for the syscall number passed and the error
flag returned. This way there is no need to fiddle with the stack
pointer anymore and all that has to be handled in the new standalone
functions is the special handling of the syscall number and the error
flag.
Redirect 5-, 6- or 7-argument MIPS16/Linux syscall wrappers to these new
functions as well, so as to avoid an unnecessary double call the
existing wrappers would cause with the new arrangement.
[BZ #21956]
* sysdeps/unix/sysv/linux/mips/mips32/mips16/Makefile
[subdir = misc] (sysdep_routines): Remove `mips16-syscall5',
`mips16-syscall6' and `mips16-syscall7'.
(CFLAGS-mips16-syscall5.c, CFLAGS-mips16-syscall6.c)
(CFLAGS-mips16-syscall7.c): Remove.
* sysdeps/unix/sysv/linux/mips/mips32/mips16/Versions (libc):
Remove `__mips16_syscall5', `__mips16_syscall6' and
`__mips16_syscall7'.
* sysdeps/unix/sysv/linux/mips/mips32/mips16/mips16-syscall0.c
(__mips16_syscall0): Rename `__mips16_syscall_return' to
`__mips_syscall_return'.
* sysdeps/unix/sysv/linux/mips/mips32/mips16/mips16-syscall1.c
(__mips16_syscall1): Likewise.
* sysdeps/unix/sysv/linux/mips/mips32/mips16/mips16-syscall2.c
(__mips16_syscall2): Likewise.
* sysdeps/unix/sysv/linux/mips/mips32/mips16/mips16-syscall3.c
(__mips16_syscall3): Likewise.
* sysdeps/unix/sysv/linux/mips/mips32/mips16/mips16-syscall4.c
(__mips16_syscall4): Likewise.
* sysdeps/unix/sysv/linux/mips/mips32/mips16/mips16-syscall5.c:
Remove.
* sysdeps/unix/sysv/linux/mips/mips32/mips16/mips16-syscall6.c:
Remove.
* sysdeps/unix/sysv/linux/mips/mips32/mips16/mips16-syscall7.c:
Remove.
* sysdeps/unix/sysv/linux/mips/mips32/mips16/mips16-syscall.h
(__mips16_syscall5): Expand to `__mips_syscall5' rather than
`__mips16_syscall5'. Remove prototype.
(__mips16_syscall6): Expand to `__mips_syscall6' rather than
`__mips16_syscall6'. Remove prototype.
(__mips16_syscall7): Expand to `__mips_syscall7' rather than
`__mips16_syscall7'. Remove prototype.
(__nomips16, __mips16_syscall_return): Move to...
* sysdeps/unix/sysv/linux/mips/mips32/sysdep.h
(__nomips16, __mips_syscall_return): ... here.
[__mips16] (INTERNAL_SYSCALL_NCS): Rename
`__mips16_syscall_return' to `__mips_syscall_return'.
[__mips16] (INTERNAL_SYSCALL_MIPS16): Pass `number' to
`internal_syscall##nr'.
[!__mips16] (INTERNAL_SYSCALL): Pass `SYS_ify (name)' to
`internal_syscall##nr'.
(FORCE_FRAME_POINTER): Remove.
(__mips_syscall5): New prototype.
(internal_syscall5): Rewrite to call `__mips_syscall5'.
(__mips_syscall6): New prototype.
(internal_syscall6): Rewrite to call `__mips_syscall6'.
(__mips_syscall7): New prototype.
(internal_syscall7): Rewrite to call `__mips_syscall7'.
* sysdeps/unix/sysv/linux/mips/mips32/mips-syscall5.S: New file.
* sysdeps/unix/sysv/linux/mips/mips32/mips-syscall6.S: New file.
* sysdeps/unix/sysv/linux/mips/mips32/mips-syscall7.S: New file.
* sysdeps/unix/sysv/linux/mips/mips32/Makefile [subdir = misc]
(sysdep_routines): Add libc-do-syscall.
* sysdeps/unix/sysv/linux/mips/mips32/Versions (libc): Add
`__mips_syscall5', `__mips_syscall6' and `__mips_syscall7'.
This patch fixes ia64 failures on thread exit by madvise the required
area taking in consideration its disjoing stacks
(NEED_SEPARATE_REGISTER_STACK). Also the snippet that setup the
madvise call to advertise kernel the area won't be used anymore in
near future is reallocated in allocatestack.c (for consistency to
put all stack management function in one place).
Checked on x86_64-linux-gnu and i686-linux-gnu for sanity (since
it is not expected code changes for architecture that do not
define NEED_SEPARATE_REGISTER_STACK) and also got a report that
it fixes ia64-linux-gnu failures from Sergei Trofimovich
<slyfox@gentoo.org>.
[BZ #21672]
* nptl/allocatestack.c [_STACK_GROWS_DOWN] (setup_stack_prot):
Set to use !NEED_SEPARATE_REGISTER_STACK as well.
(advise_stack_range): New function.
* nptl/pthread_create.c (START_THREAD_DEFN): Move logic to mark
stack non required to advise_stack_range at allocatestack.c
Commit 39e7a5a66828a60008543db5f03f46dc7081668b added stdint.h
to sys/procfs.h, but it is included into signal.h by default and
there is code that does not expect stdint.h to be visible there,
so use __uint64_t instead of uint64_t.
The current bits/math-finite.h approach to defining functions for
different types, involving math.h defining _MSUF_ and _MSUFTO_ for the
function suffixes involved, is not namespace-clean if one of those
suffixes (f, l, f128) is defined as a macro by the user before math.h
is included; too many levels of macro expansion occur. Instead, those
suffixes should appear directly in the expansion of the macro using ##
so they don't get expanded even if defined as macros by the user (that
is, math.h should be defining __REDIRFROM_X and __REDIRTO_X directly
to use those suffixes rather than suffixes being passed as an argument
by macro callers). This patch makes that change.
Tested for x86_64.
[BZ #22028]
* math/math.h [__FINITE_MATH_ONLY__ && __FINITE_MATH_ONLY__ > 0]
(_MSUF_): Remove macro.
[__FINITE_MATH_ONLY__ && __FINITE_MATH_ONLY__ > 0] (_MSUFTO_):
Likewise.
[__FINITE_MATH_ONLY__ && __FINITE_MATH_ONLY__ > 0]
(__REDIRFROM_X): New macro.
[__FINITE_MATH_ONLY__ && __FINITE_MATH_ONLY__ > 0] (__REDIRTO_X):
Likewise.
* math/bits/math-finite.h (__REDIRFROM_X): Remove macro.
(__REDIRTO_X): Likewise.
(__MATH_REDIRCALL): Do not pass _MSUF_ or _MSUFTO_ macro
arguments.
(__MATH_REDIRCALL_2): Likewise.
(__MATH_REDIRCALL_INTERNAL): Likewise.
(__REDIRFROM (lgamma, , _MSUF_)): Likewise.
(__REDIRFROM (gamma, , _MSUF_)): Likweise.
(__REDIRFROM (__gamma, _r_finite, _MSUF_)): Likewise.
(__REDIRFROM (tgamma, , _MSUF_)): Likewise.
* math/test-finite-macros.c: New file.
* math/Makefile (tests): Add test-finite-macros.
(CFLAGS-test-finite-macros.c): New variable.
Add locale for “Morisyen” which is also called “Mauritian Creole”
and is spoken in Mauritius.
[BZ #21971]
* localedata/SUPPORTED: Add mfe_MU/UTF-8.
* localedata/locales/mfe_MU: New File.
[BZ #21971]
* locale/iso-639.def: add Morisyen.
When signaling nans are enabled (with -fsignaling-nans), the C++ version
of iszero uses the fpclassify macro, which is defined with __MATH_TG.
However, when support for float128 is available, __MATH_TG uses the
builtin __builtin_types_compatible_p, which is only available in C mode.
This patch refactors the C++ version of iszero so that it uses function
overloading to select between the floating-point types, instead of
relying on fpclassify and __MATH_TG.
Tested for powerpc64le, s390x, x86_64, and with build-many-glibcs.py.
[BZ #21930]
* math/math.h [defined __cplusplus && defined __SUPPORT_SNAN__]
(iszero): New C++ implementation that does not use
fpclassify/__MATH_TG/__builtin_types_compatible_p, when
signaling nans are enabled, since __builtin_types_compatible_p
is a C-only feature.
* math/test-math-iszero.cc: When __HAVE_DISTINCT_FLOAT128 is
defined, include ieee754_float128.h for access to the union and
member ieee854_float128.ieee.
[__HAVE_DISTINCT_FLOAT128] (do_test): Call check_float128.
[__HAVE_DISTINCT_FLOAT128] (check_float128): New function.
* sysdeps/powerpc/powerpc64le/Makefile [subdir == math]
(CXXFLAGS-test-math-iszero.cc): Add -mfloat128 to the build
options of test-math-zero on powerpc64le.
Now there are no more assembly wrappers using _LIB_VERSION or
__kernel_standard, the math-svid-compat code can be slighly
simplified. math-svid-compat.h no longer needs __ASSEMBLER__
conditionals, and the _LIB_VERSION variable no longer needs to be
built for static libm, since all references are now in C code that
includes math-svid-compat.h and so gets the macro definition of
_LIB_VERSION to _POSIX_ outside the compat case. This patch makes
those cleanups.
Tested for x86_64, and with build-many-glibcs.py.
* math/math-svid-compat.h [!__ASSEMBLER__]: Make code
unconditional.
* sysdeps/ieee754/s_lib_version.c [!defined SHARED]: Remove
conditional code; define contents only for [LIBM_SVID_COMPAT].
These options are no longer needed since commit
2dba5ce7b8115d6a2789bf279892263621088e74 (<bits/syscall.h>: Use an
arch-independent system call list on Linux).
This commit changes the way the list of SYS_* system call macros is
created on Linux. glibc now contains a list of all known system
calls, and the generated <bits/syscall.h> file defines the SYS_ macro
only if the correspnding __NR_ macro is defined by the kernel headers.
As a result, glibc does not have to be rebuilt to pick up system calls
if the glibc sources already know about them. This means that glibc
can be built with older kernel headers, and if the installed kernel
headers are upgraded afterwards, additional SYS_ macros become
available as long as glibc has a record for those system calls.
When linked statically, TLS initialization is not achieved before
mach_init and alike, so ssp accesses to tcbhead's stack_guard would
crash. We can just avoid using ssp in the few functions needed before
TLS is set up.
* mach/Makefile (CFLAGS-mach_init.o, CFLAGS-RPC_vm_statistics.o,
CFLAGS-RPC_vm_map.o, CFLAGS-RPC_vm_protect.o,
CFLAGS-RPC_i386_set_gdt.o, CFLAGS-RPC_i386_set_ldt.o,
CFLAGS-RPC_task_get_special_port.o): Add $(no-stack-protector).
* hurd/Makefile (CFLAGS-hurdstartup.o,
CFLAGS-RPC_exec_startup_get_info.o): Add $(no-stack-protector).
libmachuser and libhurduser also need stack_chk_fail_local and they do not
link against libc_nonshared.
* mach/stack_chk_fail_local.c: New file.
* hurd/stack_chk_fail_local.c: New file.
* mach/Machrules ($(interface-library)-routines): Add
stack_chk_fail_local.
* mach/Versions (GLIBC_2.4): Add __stack_chk_fail.
* hurd/Versions (GLIBC_2.4): Add __stack_chk_fail.
Since assembly versions of HAS_CPU_FEATURE and HAS_ARCH_FEATURE have
been removed, assembly versions of index_cpu_* and index_arch_* can
also be removed.
Tested on i686 and x86-64 with and without --disable-multi-arch.
* sysdeps/x86/cpu-features.h [__ASSEMBLER__]
(index_cpu_*, index_arch_*): Removed.
When _Float128 is ABI-equivalent to long double, there is no need for
tgmath.h to have any special _Float128 handling: it's always OK to
call the long double versions of functions for _Float128 arguments in
that case, and the logic to determine return types is generic. Thus,
this patch changes the use of __HAVE_FLOAT128 to
__HAVE_DISTINCT_FLOAT128, as a minor optimization to reduce the size
of the macro expansions in the ABI-equivalent case.
Tested for x86_64.
* math/tgmath.h [__HAVE_FLOAT128]: Change conditional to
[__HAVE_DISTINCT_FLOAT128].
This patch cleans up how bits/math-finite.h handles types that are
ABI-aliases of other types.
For such types, no __*_finite functions exist; instead,
bits/math-finite.h must redirect calls to a the functions for a
canonical choice of type for each floating-point format. (For the
actual public interfaces, symbols need exporting for each type, even
those that are ABI-aliases, because of standard requirements that
programs can declare the functions themselves without including
<math.h>, but that does not apply to __*_finite.)
At present, there is a special-case conditional in bits/math-finite.h
on __MATH_DECLARING_LDOUBLE && defined __NO_LONG_DOUBLE_MATH to handle
redirecting long double function calls to double __*_finite. This
patch replaces this by a more general mechanism. math.h, before each
inclusion of bits/math-finite.h, defines _MSUFTO_ as the suffix to use
on the target of redirection, in addition to the existing _MSUF_.
This way, __MATH_DECLARING_LDOUBLE can go away, as can the special
conditional in bits/math-finite.h. With this patch, math.h is now
prepared for the case of supporting float128 functions as aliases of
long double ones on platforms where long double is binary128, with
_MSUFTO_ appropriately defined for that case, and appropriate _MSUFTO_
definitions can easily be included when supporting _Float32 / _Float64
/ _Float32x / _Float64x (which will always be ABI-aliases of another
type when supported).
Tested for x86_64, and did a compilation test for ARM with
build-many-glibcs.py to cover the long double = double case.
* math/math.h (_MSUFTO_): Define and undefine for each inclusion
of <bits/math-finite.h>.
(__MATH_DECLARING_LDOUBLE): Do not define and undefine for each
inclusion of <bits/math-finite.h>.
* math/bits/math-finite.h (__REDIRTO_X): Do not define
conditionally on [__MATH_DECLARING_LDOUBLE && defined
__NO_LONG_DOUBLE_MATH].
(__MATH_REDIRCALL): Use _MSUFTO_ in __REDIRTO call.
(__MATH_REDIRCALL_2): Likewise.
(__MATH_REDIRCALL_INTERNAL): Likewise.
(__REDIRFROM (lgamma, , _MSUF_)): Likewise.
(__REDIRFROM (gamma, , _MSUF_)): Likewise.
(__REDIRFROM (tgamma, , _MSUF_)): Likewise.
