Implement a long-standing request to support tuning the size of the
emergency buffer for allocating exceptions after malloc fails, or to
disable that buffer entirely.
It's now possible to disable the dynamic allocation of the buffer and
use a fixed-size static buffer, via --enable-libstdcxx-static-eh-pool.
This is a built-time choice that is baked into libstdc++ and so affects
all code linked against that build of libstdc++.
The size of the pool can be set by --with-libstdcxx-eh-pool-obj-count=N
which is measured in units of sizeof(void*) not bytes. A given exception
type such as std::system_error depends on the target, so giving a size
in bytes wouldn't be portable across 16/32/64-bit targets.
When libstdc++ is configured to use a dynamic buffer, the size of that
buffer can now be tuned at runtime by setting the GLIBCXX_TUNABLES
environment variable (c.f. PR libstdc++/88264). The number of exceptions
to reserve space for is controlled by the "glibcxx.eh_pool.obj_count"
and "glibcxx.eh_pool.obj_size" tunables. The pool will be sized to be
able to allocate obj_count exceptions of size obj_size*sizeof(void*) and
obj_count "dependent" exceptions rethrown by std::rethrow_exception.
With the ability to tune the buffer size, we can reduce the default pool
size on 32-bit and 16-bit targets. Most users never need to throw 1kB
exceptions in parallel from hundreds of threads after malloc is OOM. The
users who do need that can use the tunables to select larger sizes.
The old defaults can be chosen at runtime by setting GLIBCXX_TUNABLES
to:
64-bit: glibcxx.eh_pool.obj_count=64:glibcxx.eh_pool.obj_size=112
32-bit: glibcxx.eh_pool.obj_count=32:glibcxx.eh_pool.obj_size=104
Or approximated by configuring with:
64-bit: --with-libstdcxx-eh-pool-obj-count=252
32-bit: --with-libstdcxx-eh-pool-obj-count=94
libstdc++-v3/ChangeLog:
PR libstdc++/68606
* Makefile.in: Regenerate.
* acinclude.m4 (GLIBCXX_EMERGENCY_EH_ALLOC): New macro.
* configure: Regenerate.
* configure.ac: Use GLIBCXX_EMERGENCY_EH_ALLOC.
* crossconfig.m4: Check for secure_getenv.
* doc/Makefile.in: Regenerate.
* doc/xml/manual/configure.xml: Document new configure options.
* doc/xml/manual/evolution.xml: Document addition of tunables.
* doc/xml/manual/using_exceptions.xml: Document emergency
buffer and tunables.
* doc/html/*: Regenerate.
* include/Makefile.in: Regenerate.
* libsupc++/Makefile.am: Use EH_POOL_FLAGS.
* libsupc++/Makefile.in: Regenerate.
* libsupc++/eh_alloc.cc (EMERGENCY_OBJ_SIZE): Define in units
of sizeof(void*) not including the ABI's exception header.
(EMERGENCY_OBJ_COUNT): Define as target-independent calculation
based on word size.
(MAX_OBJ_COUNT): Define macro for upper limit on pool size.
(pool) [_GLIBCXX_EH_POOL_STATIC]: Use fixed-size buffer.
(pool::buffer_size_in_bytes): New static member function.
(pool::pool): Parse GLIBCXX_TUNABLES environment variable to set
pool size at runtime.
(pool::in_pool): Use std::less<void*> for total order.
(__freeres) [_GLIBCXX_EH_POOL_STATIC]: Do nothing.
(__cxa_free_exception, __cxa_free_dependent_exception): Add
[[unlikely]] attributes.
* po/Makefile.in: Regenerate.
* python/Makefile.in: Regenerate.
* src/Makefile.in: Regenerate.
* src/c++11/Makefile.in: Regenerate.
* src/c++17/Makefile.in: Regenerate.
* src/c++20/Makefile.in: Regenerate.
* src/c++98/Makefile.in: Regenerate.
* src/filesystem/Makefile.in: Regenerate.
* src/libbacktrace/Makefile.in: Regenerate.
* testsuite/Makefile.in: Regenerate.
Remove RTEMS support from crossconfig.m4 since this code is not used due to
"with_newlib" being "yes".
libstdc++-v3/ChangeLog:
* configure: Regnerate.
* configure.ac (newlib, *-rtems*): Enable TLS support for all RTEMS
targets except bfin, lm32, mips, moxie, or1k, and v850.
For all RTEMS targets, define HAVE_ALIGNED_ALLOC, HAVE_AT_QUICK_EXIT,
HAVE_LINK, HAVE_POLL, HAVE_QUICK_EXIT, HAVE_READLINK, HAVE_SETENV,
HAVE_SLEEP, HAVE_SOCKATMARK, HAVE_STRERROR_L, HAVE_SYMLINK,
HAVE_TRUNCATE, and HAVE_USLEEP.
* crossconfig.m4 (*-rtems*): Remove.
Native mingw builds enable TLS, but crosses don't because we don't use
GCC_CHECK_TLS in the cross-compiler config.
libstdc++-v3/ChangeLog:
* crossconfig.m4: Check for TLS support on mingw.
* configure: Regenerate.
When libstdc++ is enabled, the current high level configuration
bits should apply the same to all versions of VxWorks. Adjust the
config triplets matching rules accordingly.
2010-10-21 Olivier Hainque <hainque@adacore.com>
libstdc++-v3/
* crossconfig.m4: Turn vxworks matcher into vxworks*.
* configure.host: Likewise.
* configure: Regenerate.
This adds the missing std::from_chars overloads for floating-point
types, as required for C++17 conformance.
The implementation is a hack and not intended to be used in the long
term. Rather than parsing the string directly, this determines the
initial portion of the string that matches the pattern determined by the
chars_format parameter, then creates a NTBS to be parsed by strtod (or
strtold or strtof).
Because creating a NTBS requires allocating memory, but std::from_chars
is noexcept, we need to be careful to minimise allocation. Even after
being careful, allocation failure is still possible, and so a
non-conforming std::no_more_memory error code might be returned.
Because strtod et al depend on the current locale, but std::from_chars
does not, we change the current thread's locale to "C" using newlocale
and uselocale before calling strtod, and restore it afterwards.
Because strtod doesn't have the equivalent of a std::chars_format
parameter, it has to examine the input to determine the format in use,
even though the std::from_chars code has already parsed it once (or
twice for large input strings!)
By replacing the use of strtod we could avoid allocation, avoid changing
locale, and use optimised code paths specific to each std::chars_format
case. We would also get more portable behaviour, rather than depending
on the presence of uselocale, and on any bugs or quirks of the target
libc's strtod. Replacing strtod is a project for a later date.
libstdc++-v3/ChangeLog:
* acinclude.m4 (libtool_VERSION): Bump version.
* config.h.in: Regenerate.
* config/abi/pre/gnu.ver: Add GLIBCXX_3.4.29 version and new
exports.
* config/os/gnu-linux/ldbl-extra.ver: Add _GLIBCXX_LDBL_3.4.29
version and new export.
* configure: Regenerate.
* configure.ac: Check for <xlocale.h> and uselocale.
* crossconfig.m4: Add macro or checks for uselocale.
* include/std/charconv (from_chars): Declare overloads for
float, double, and long double.
* src/c++17/Makefile.am: Add new file.
* src/c++17/Makefile.in: Regenerate.
* src/c++17/floating_from_chars.cc: New file.
(from_chars): Define for float, double, and long double.
* testsuite/20_util/from_chars/1_c++20_neg.cc: Prune extra
diagnostics caused by new overloads.
* testsuite/20_util/from_chars/1_neg.cc: Likewise.
* testsuite/20_util/from_chars/2.cc: Check leading '+'.
* testsuite/20_util/from_chars/4.cc: New test.
* testsuite/20_util/from_chars/5.cc: New test.
* testsuite/util/testsuite_abi.cc: Add new symbol versions.
Use the GLIBCXX_CHECK_MATH_DECL macro to check for the full list of
vxworks math decls.
libstdc++-v3/ChangeLog
* crossconfig.m4 (<*-vxworks>): Check for more math decls.
* configure: Rebuild.
The C++ headers #undef the functions we are testing for, just in case
they're implemented as macros, so do that in the cross math decl tests
as well.
for libstdc++-v3/ChangeLog
* crossconfig.m4 (GLIBCXX_CHECK_MATH_DECL): Reject macros.
* configure: Rebuild.
When cross-building for vxworks, test for declarations of long double
functions in math.h. We don't normally test for these functions when
cross compiling, because link tests don't work, or ever really, but
not defining them as available causes replacements to be defined in
ways that may cause duplicate definition linker errors if the units
defining both the replacement and the actual implementation are
brought in because of other symbols.
for libstdc++-v3/ChangeLog
* crossconfig.m4 (GLIBCXX_CROSSCONFIG) [*-vxworks*]: Define
long double functions as available if declared by math.h.
(GLIBCXX_CHECK_MATH_DECL, GLIBCXX_CHECK_MATH_DECLS): New.
* configure: Rebuild.
From-SVN: r279731
Check for availability of POSIX sockatmark before using it.
Rename _S_ntoh overloads that are ambiguous when passed an integral type
that is neither uint16_t nor uint32_t.
PR libstdc++/89460
* configure.ac: Check for sockatmark.
* crossconfig.m4: Check for sockatmark.
* config.h.in: Regenerate.
* configure: Regenerate.
* include/experimental/internet (address_v4::_S_hton): Rename
overloaded functions to _S_hton_16 and _S_ntoh_16.
(address_v4::_S_ntoh): Rename to _S_ntoh_16 and _S_ntoh_32.
(basic_endpoint): Adjust calls to _S_hton and _S_ntoh.
* include/experimental/socket (basic_socket::at_mark): Check
_GLIBCXX_HAVE_SOCKATMARK.
From-SVN: r269588
Define and use a new macro with a more descriptive name. Only use the
old macro in <tr1/random.h>.
* acinclude.m4 (GLIBCXX_CHECK_RANDOM_TR1): Replace with ...
(GLIBCXX_CHECK_DEV_RANDOM): New macro with more descriptive name.
Define _GLIBCXX_USE_DEV_RANDOM as well as _GLIBCXX_USE_RANDOM_TR1.
* config.h.in: Regenerate.
* configure: Regenerate.
* configure.ac: Use GLIBCXX_CHECK_DEV_RANDOM instead of
GLIBCXX_CHECK_RANDOM_TR1.
crossconfig.m4: Likewise.
* include/bits/random.h (random_device): Use _GLIBCXX_USE_DEV_RANDOM
instead of _GLIBCXX_USE_RANDOM_TR1.
* testsuite/26_numerics/random/random_device/cons/token.cc: Likewise.
From-SVN: r265197
Enable libstdc++ compilation in AVR targets with AVR-Libc. Most
floating point math functions are already defined in AVR-Libc, so
defines are in place to avoid multiple definition of these functions.
2016-12-06 Felipe Magno de Almeida <felipe@expertisesolutions.com.br>
* crossconfig.m4: Add avr target for cross-compilation.
* configure: Regenerate.
From-SVN: r243309
I noticed that when testing glibc with a cross compiler I got
UNSUPPORTED: nptl/tst-thread_local1
because the libstdc++-v3 configuration for cross compiling defaulted
to __cxa_thread_atexit_impl not being available. This patch fixes
GLIBCXX_CROSSCONFIG to run the same test (for the case covering
targets with glibc) for __cxa_thread_atexit_impl as for native
compilation, just as it runs most of the other tests done for native
compilation (for these targets, it's not possible to build libstdc++
without already having built libc, so link tests are OK).
Tested with no regressions for cross to arm-none-linux-gnueabi.
* crossconfig.m4 (GLIBCXX_CROSSCONFIG) <*-linux* | *-uclinux* |
*-gnu* | *-kfreebsd*-gnu | *-knetbsd*-gnu | *-cygwin*>: Check for
__cxa_thread_atexit_impl.
* configure: Regenerate.
From-SVN: r228695
* crossconfig.m4: Since we know that all TPF builds are cross-
builds and cannot run configuration-time link tests, do not
allow it; just go with known supported linker options.
* configure: Regenerate (called as GLIBCXX_CROSSCONFIG).
From-SVN: r187062
2008-08-23 Paolo Carlini <paolo.carlini@oracle.com>
* acinclude.m4 ([GLIBCXX_CHECK_GETTIMEOFDAY]): Change to use
GCC_TRY_COMPILE_OR_LINK.
([GLIBCXX_CHECK_CLOCK_GETTIME]): Rename to
[GLIBCXX_ENABLE_CLOCK_GETTIME], a versatile configure-time option,
[no] by default, since it includes TRY_LINK checks. librt is also
searched, if requested.
* doc/xml/manual/configure.xml: Document the latter.
* configure.ac: Move the above checks outside native-only.
* configure: Regenerate.
From-SVN: r139576
2008-08-17 Paolo Carlini <paolo.carlini@oracle.com>
* acinclude.m4 ([GLIBCXX_CHECK_LFS], [GLIBCXX_CHECK_POLL],
[GLIBCXX_CHECK_S_ISREG_OR_S_IFREG], [GLIBCXX_CHECK_WRITEV]):
Use GCC_TRY_COMPILE_OR_LINK instead of AC_TRY_LINK, in C++,
with -fno-exceptions as flags.
* configure.ac: Run unconditionally GLIBCXX_CHECK_LFS,
GLIBCXX_CHECK_POLL, GLIBCXX_CHECK_S_ISREG_OR_S_IFREG,
GLIBCXX_CHECK_WRITEV and commonized AC_CHECK_HEADERS.
* crossconfig.m4: Adjust; remove AC_CHECK_HEADERS commonized
in configure.ac.
* configure: Regenerate.
From-SVN: r139168
* configure.ac: Do not check for gconv.h.
* crossconfig.m4 (GLIBCXX_CROSSCONFIG): Do not test for gconv.h or
gconf.h. For glibc and uClibc systems, define
_GLIBCXX_USE_RANDOM_TR1 and HAVE_MMAP and use AC_LC_MESSAGES and
AM_ICONV.
* configure, config.h.in: Regenerate.
From-SVN: r136494
* configure.ac (mips*-sde-elf*): New stanza. Add target-libiberty
to $skipdirs and only disable gprof for newlib. Use the normal
mips*-elf* handling in other respects.
* configure: Regnerate.
gcc/
2007-xx-xx Nigel Stephens <nigel@mips.com>
David Ung <davidu@mips.com>
Thiemo Seufer <ths@mips.com>
Richard Sandiford <richard@codesourcery.com>
* config.gcc (mips*-sde-elf*): Add support for the SDE C libraries.
* configure.ac: Add a mipssde threading type.
* configure: Regenerate.
* config/mips/sdemtk.h: New file.
* config/mips/t-sdemtk: Likewise.
* config/mips/sdemtk.opt: Likewise.
* gthr-mipssde.h: Likewise.
* config/mips/sde.h (FUNCTION_PROFILER): Move to config/mips/sdemtk.h.
* config/mips/mips.h (MIPS_SAVE_REG_FOR_PROFILING_P): New macro.
(MIPS_ICACHE_SYNC): New macro, split from ...
* config/mips/mips.md (clear_cache): ...here.
* config/mips/mips.c (mips_save_reg_p): Check
MIPS_SAVE_REG_FOR_PROFILING_P on profiled functions.
(build_mips16_function_stub): Use targetm.strip_name_encoding.
(build_mips16_call_stub): Likewise.
libstdc++-v3/
2007-xx-xx Thiemo Seufer <ths@mips.com>
* crossconfig.m4 (mips*-sde-elf*): New stanza.
* configure: Regenerate.
From-SVN: r128495
* crossconfig.m4 (djgpp): Pre-detect djgpp-specific results, to
avoid needing to link during cross compiling.
* configure: Regenerated.
From-SVN: r122344
