Add in compiling compatibility header files and header file for

obtaining the Linux kernel version headers being used for NPTL.

2 files changed, 525 insertions, 0 deletions

diff --git a/libpthread/nptl/sysdeps/unix/sysv/linux/hp-timing.h b/libpthread/nptl/sysdeps/unix/sysv/linux/hp-timing.h
new file mode 100644
index 000000000..099342db8
--- /dev/null
+++ b/libpthread/nptl/sysdeps/unix/sysv/linux/hp-timing.h
@@ -0,0 +1,83 @@
+/* High precision, low overhead timing functions.  Generic version.
+   Copyright (C) 1998, 2000 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by Ulrich Drepper <drepper@cygnus.com>, 1998.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#ifndef _HP_TIMING_H
+#define _HP_TIMING_H	1
+
+
+/* There are no generic definitions for the times.  We could write something
+   using the `gettimeofday' system call where available but the overhead of
+   the system call might be too high.
+
+   In case a platform supports timers in the hardware the following macros
+   and types must be defined:
+
+   - HP_TIMING_AVAIL: test for availability.
+
+   - HP_TIMING_INLINE: this macro is non-zero if the functionality is not
+     implemented using function calls but instead uses some inlined code
+     which might simply consist of a few assembler instructions.  We have to
+     know this since we might want to use the macros here in places where we
+     cannot make function calls.
+
+   - hp_timing_t: This is the type for variables used to store the time
+     values.
+
+   - HP_TIMING_ZERO: clear `hp_timing_t' object.
+
+   - HP_TIMING_NOW: place timestamp for current time in variable given as
+     parameter.
+
+   - HP_TIMING_DIFF_INIT: do whatever is necessary to be able to use the
+     HP_TIMING_DIFF macro.
+
+   - HP_TIMING_DIFF: compute difference between two times and store it
+     in a third.  Source and destination might overlap.
+
+   - HP_TIMING_ACCUM: add time difference to another variable.  This might
+     be a bit more complicated to implement for some platforms as the
+     operation should be thread-safe and 64bit arithmetic on 32bit platforms
+     is not.
+
+   - HP_TIMING_ACCUM_NT: this is the variant for situations where we know
+     there are no threads involved.
+
+   - HP_TIMING_PRINT: write decimal representation of the timing value into
+     the given string.  This operation need not be inline even though
+     HP_TIMING_INLINE is specified.
+
+*/
+
+/* Provide dummy definitions.  */
+#define HP_TIMING_AVAIL		(0)
+#define HP_TIMING_INLINE	(0)
+typedef int hp_timing_t;
+#define HP_TIMING_ZERO(Var)
+#define HP_TIMING_NOW(var)
+#define HP_TIMING_DIFF_INIT()
+#define HP_TIMING_DIFF(Diff, Start, End)
+#define HP_TIMING_ACCUM(Sum, Diff)
+#define HP_TIMING_ACCUM_NT(Sum, Diff)
+#define HP_TIMING_PRINT(Buf, Len, Val)
+
+/* Since this implementation is not available we tell the user about it.  */
+#define HP_TIMING_NONAVAIL	1
+
+#endif	/* hp-timing.h */
diff --git a/libpthread/nptl/sysdeps/unix/sysv/linux/kernel-features.h b/libpthread/nptl/sysdeps/unix/sysv/linux/kernel-features.h
new file mode 100644
index 000000000..ef8240b00
--- /dev/null
+++ b/libpthread/nptl/sysdeps/unix/sysv/linux/kernel-features.h
@@ -0,0 +1,442 @@
+/* Set flags signalling availability of kernel features based on given
+   kernel version number.
+   Copyright (C) 1999-2003, 2004, 2005 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+/* This file must not contain any C code.  At least it must be protected
+   to allow using the file also in assembler files.  */
+
+#if defined __mips__
+# include <sgidefs.h>
+#endif
+
+#if __UCLIBC__
+# include <linux/version.h>
+# define __LINUX_KERNEL_VERSION		LINUX_VERSION_CODE
+#endif
+
+#ifndef __LINUX_KERNEL_VERSION
+/* We assume the worst; all kernels should be supported.  */
+# define __LINUX_KERNEL_VERSION	0
+#endif
+
+/* We assume for __LINUX_KERNEL_VERSION the same encoding used in
+   linux/version.h.  I.e., the major, minor, and subminor all get a
+   byte with the major number being in the highest byte.  This means
+   we can do numeric comparisons.
+
+   In the following we will define certain symbols depending on
+   whether the describes kernel feature is available in the kernel
+   version given by __LINUX_KERNEL_VERSION.  We are not always exactly
+   recording the correct versions in which the features were
+   introduced.  If somebody cares these values can afterwards be
+   corrected.  Most of the numbers here are set corresponding to
+   2.2.0.  */
+
+/* `getcwd' system call.  */
+#if __LINUX_KERNEL_VERSION >= 131584
+# define __ASSUME_GETCWD_SYSCALL	1
+#endif
+
+/* Real-time signal became usable in 2.1.70.  */
+#if __LINUX_KERNEL_VERSION >= 131398
+# define __ASSUME_REALTIME_SIGNALS	1
+#endif
+
+/* When were the `pread'/`pwrite' syscalls introduced?  */
+#if __LINUX_KERNEL_VERSION >= 131584
+# define __ASSUME_PREAD_SYSCALL		1
+# define __ASSUME_PWRITE_SYSCALL	1
+#endif
+
+/* When was `poll' introduced?  */
+#if __LINUX_KERNEL_VERSION >= 131584
+# define __ASSUME_POLL_SYSCALL		1
+#endif
+
+/* The `lchown' syscall was introduced in 2.1.80.  */
+#if __LINUX_KERNEL_VERSION >= 131408
+# define __ASSUME_LCHOWN_SYSCALL	1
+#endif
+
+/* When did the `setresuid' sysall became available?  */
+#if __LINUX_KERNEL_VERSION >= 131584 && !defined __sparc__
+# define __ASSUME_SETRESUID_SYSCALL	1
+#endif
+
+/* The SIOCGIFNAME ioctl is available starting with 2.1.50.  */
+#if __LINUX_KERNEL_VERSION >= 131408
+# define __ASSUME_SIOCGIFNAME		1
+#endif
+
+/* MSG_NOSIGNAL was at least available with Linux 2.2.0.  */
+#if __LINUX_KERNEL_VERSION >= 131584
+# define __ASSUME_MSG_NOSIGNAL		1
+#endif
+
+/* On x86 another `getrlimit' syscall was added in 2.3.25.  */
+#if __LINUX_KERNEL_VERSION >= 131865 && defined __i386__
+# define __ASSUME_NEW_GETRLIMIT_SYSCALL	1
+#endif
+
+/* On x86 the truncate64/ftruncate64 syscalls were introduced in 2.3.31.  */
+#if __LINUX_KERNEL_VERSION >= 131871 && defined __i386__
+# define __ASSUME_TRUNCATE64_SYSCALL	1
+#endif
+
+/* On x86 the mmap2 syscall was introduced in 2.3.31.  */
+#if __LINUX_KERNEL_VERSION >= 131871 && defined __i386__
+# define __ASSUME_MMAP2_SYSCALL	1
+#endif
+
+/* On x86 the stat64/lstat64/fstat64 syscalls were introduced in 2.3.34.  */
+#if __LINUX_KERNEL_VERSION >= 131874 && defined __i386__
+# define __ASSUME_STAT64_SYSCALL	1
+#endif
+
+/* On sparc and ARM the truncate64/ftruncate64/mmap2/stat64/lstat64/fstat64
+   syscalls were introduced in 2.3.35.  */
+#if __LINUX_KERNEL_VERSION >= 131875 && (defined __sparc__ || defined __arm__)
+# define __ASSUME_TRUNCATE64_SYSCALL	1
+# define __ASSUME_MMAP2_SYSCALL		1
+# define __ASSUME_STAT64_SYSCALL	1
+#endif
+
+/* I know for sure that getrlimit are in 2.3.35 on powerpc.  */
+#if __LINUX_KERNEL_VERSION >= 131875 && defined __powerpc__
+# define __ASSUME_NEW_GETRLIMIT_SYSCALL	1
+#endif
+
+/* I know for sure that these are in 2.3.35 on powerpc. But PowerPC64 does not
+   support separate 64-bit syscalls, already 64-bit */
+#if __LINUX_KERNEL_VERSION >= 131875 && defined __powerpc__ \
+    && !defined __powerpc64__
+# define __ASSUME_TRUNCATE64_SYSCALL	1
+# define __ASSUME_STAT64_SYSCALL	1
+#endif
+
+/* Linux 2.3.39 introduced 32bit UID/GIDs.  Some platforms had 32
+   bit type all along.  */
+#if __LINUX_KERNEL_VERSION >= 131879 || defined __powerpc__ || defined __mips__
+# define __ASSUME_32BITUIDS		1
+#endif
+
+/* Linux 2.3.39 sparc added setresuid.  */
+#if __LINUX_KERNEL_VERSION >= 131879 && defined __sparc__
+# define __ASSUME_SETRESUID_SYSCALL	1
+#endif
+
+#if __LINUX_KERNEL_VERSION >= 131879
+# define __ASSUME_SETRESGID_SYSCALL	1
+#endif
+
+/* Linux 2.3.39 introduced IPC64.  Except for powerpc.  */
+#if __LINUX_KERNEL_VERSION >= 131879 && !defined __powerpc__
+# define __ASSUME_IPC64		1
+#endif
+
+/* MIPS platforms had IPC64 all along.  */
+#if defined __mips__
+# define __ASSUME_IPC64		1
+#endif
+
+/* We can use the LDTs for threading with Linux 2.3.99 and newer.  */
+#if __LINUX_KERNEL_VERSION >= 131939
+# define __ASSUME_LDT_WORKS		1
+#endif
+
+/* Linux 2.4.0 on PPC introduced a correct IPC64. But PowerPC64 does not
+   support a separate 64-bit sys call, already 64-bit */
+#if __LINUX_KERNEL_VERSION >= 132096 && defined __powerpc__ \
+    && !defined __powerpc64__
+# define __ASSUME_IPC64			1
+#endif
+
+/* SH kernels got stat64, mmap2, and truncate64 during 2.4.0-test.  */
+#if __LINUX_KERNEL_VERSION >= 132096 && defined __sh__
+# define __ASSUME_TRUNCATE64_SYSCALL	1
+# define __ASSUME_MMAP2_SYSCALL		1
+# define __ASSUME_STAT64_SYSCALL	1
+#endif
+
+/* The changed st_ino field appeared in 2.4.0-test6.  But we cannot
+   distinguish this version from other 2.4.0 releases.  Therefore play
+   save and assume it available is for 2.4.1 and up.  However, SH is lame,
+   and still does not have a 64-bit inode field.  */
+#if __LINUX_KERNEL_VERSION >= 132097 && !defined __alpha__ && !defined __sh__
+# define __ASSUME_ST_INO_64_BIT		1
+#endif
+
+/* To support locking of large files a new fcntl() syscall was introduced
+   in 2.4.0-test7.  We test for 2.4.1 for the earliest version we know
+   the syscall is available.  */
+#if __LINUX_KERNEL_VERSION >= 132097 && (defined __i386__ || defined __sparc__)
+# define __ASSUME_FCNTL64		1
+#endif
+
+/* The AT_CLKTCK auxiliary vector entry was introduction in the 2.4.0
+   series.  */
+#if __LINUX_KERNEL_VERSION >= 132097
+# define __ASSUME_AT_CLKTCK		1
+#endif
+
+/* Arm got fcntl64 in 2.4.4, PowerPC and SH have it also in 2.4.4 (I
+   don't know when it got introduced).  But PowerPC64 does not support
+   separate FCNTL64 call, FCNTL is already 64-bit */
+#if __LINUX_KERNEL_VERSION >= 132100 \
+    && (defined __arm__ || defined __powerpc__ || defined __sh__) \
+    && !defined __powerpc64__
+# define __ASSUME_FCNTL64		1
+#endif
+
+/* The getdents64 syscall was introduced in 2.4.0-test7.  We test for
+   2.4.1 for the earliest version we know the syscall is available.  */
+#if __LINUX_KERNEL_VERSION >= 132097
+# define __ASSUME_GETDENTS64_SYSCALL	1
+#endif
+
+/* When did O_DIRECTORY became available?  Early in 2.3 but when?
+   Be safe, use 2.3.99.  */
+#if __LINUX_KERNEL_VERSION >= 131939
+# define __ASSUME_O_DIRECTORY		1
+#endif
+
+/* Starting with one of the 2.4.0 pre-releases the Linux kernel passes
+   up the page size information.  */
+#if __LINUX_KERNEL_VERSION >= 132097
+# define __ASSUME_AT_PAGESIZE		1
+#endif
+
+/* Starting with at least 2.4.0 the kernel passes the uid/gid unconditionally
+   up to the child.  */
+#if __LINUX_KERNEL_VERSION >= 132097
+# define __ASSUME_AT_XID		1
+#endif
+
+/* Starting with 2.4.5 kernels PPC passes the AUXV in the standard way
+   and the vfork syscall made it into the official kernel.  */
+#if __LINUX_KERNEL_VERSION >= (132096+5) && defined __powerpc__
+# define __ASSUME_STD_AUXV		1
+# define __ASSUME_VFORK_SYSCALL		1
+#endif
+
+/* Starting with 2.4.5 kernels the mmap2 syscall made it into the official
+   kernel.  But PowerPC64 does not support a separate MMAP2 call.  */
+#if __LINUX_KERNEL_VERSION >= (132096+5) && defined __powerpc__ \
+    && !defined __powerpc64__
+# define __ASSUME_MMAP2_SYSCALL		1
+#endif
+
+/* Starting with 2.4.21 PowerPC implements the new prctl syscall.
+   This allows applications to get/set the Floating Point Exception Mode.  */
+#if __LINUX_KERNEL_VERSION >= (132096+21) && defined __powerpc__
+# define __ASSUME_NEW_PRCTL_SYSCALL		1
+#endif
+
+/* Starting with 2.4.21 the PowerPC32 clone syscall works as expected.  */
+#if __LINUX_KERNEL_VERSION >= (132096+21) && defined __powerpc__ \
+    && !defined __powerpc64__
+# define __ASSUME_FIXED_CLONE_SYSCALL		1
+#endif
+
+/* Starting with 2.4.21 PowerPC64 implements the new rt_sigreturn syscall.
+   The new rt_sigreturn takes an ucontext pointer allowing rt_sigreturn
+   to be used in the set/swapcontext implementation.  */
+#if __LINUX_KERNEL_VERSION >= (132096+21) && defined __powerpc64__
+# define __ASSUME_NEW_RT_SIGRETURN_SYSCALL		1
+#endif
+
+/* On x86, the set_thread_area syscall was introduced in 2.5.29, but its
+   semantics was changed in 2.5.30, and again after 2.5.31.  */
+#if __LINUX_KERNEL_VERSION >= 132384 && defined __i386__
+# define __ASSUME_SET_THREAD_AREA_SYSCALL	1
+#endif
+
+/* The vfork syscall on x86 and arm was definitely available in 2.4.  */
+#if __LINUX_KERNEL_VERSION >= 132097 && (defined __i386__ || defined __arm__)
+# define __ASSUME_VFORK_SYSCALL		1
+#endif
+
+/* There are an infinite number of PA-RISC kernel versions numbered
+   2.4.0.  But they've not really been released as such.  We require
+   and expect the final version here.  */
+#ifdef __hppa__
+# define __ASSUME_32BITUIDS		1
+# define __ASSUME_TRUNCATE64_SYSCALL	1
+# define __ASSUME_MMAP2_SYSCALL		1
+# define __ASSUME_STAT64_SYSCALL	1
+# define __ASSUME_IPC64			1
+# define __ASSUME_ST_INO_64_BIT		1
+# define __ASSUME_FCNTL64		1
+# define __ASSUME_GETDENTS64_SYSCALL	1
+#endif
+
+/* Alpha switched to a 64-bit timeval sometime before 2.2.0.  */
+#if __LINUX_KERNEL_VERSION >= 131584 && defined __alpha__
+# define __ASSUME_TIMEVAL64		1
+#endif
+
+#if defined __mips__ && _MIPS_SIM == _ABIN32
+# define __ASSUME_FCNTL64		1
+#endif
+
+/* The late 2.5 kernels saw a lot of new CLONE_* flags.  Summarize
+   their availability with one define.  The changes were made first
+   for i386 and the have to be done separately for the other archs.
+   For i386 we pick 2.5.50 as the first version with support.  */
+#if __LINUX_KERNEL_VERSION >= 132402 && defined __i386__
+# define __ASSUME_CLONE_THREAD_FLAGS	1
+#endif
+
+/* These features were surely available with 2.4.12.  */
+#if __LINUX_KERNEL_VERSION >= 132108 && defined __mc68000__
+# define __ASSUME_MMAP2_SYSCALL		1
+# define __ASSUME_TRUNCATE64_SYSCALL	1
+# define __ASSUME_STAT64_SYSCALL	1
+# define __ASSUME_FCNTL64		1
+# define __ASSUME_VFORK_SYSCALL		1
+#endif
+
+/* Beginning with 2.5.63 support for realtime and monotonic clocks and
+   timers based on them is available.  */
+#if __LINUX_KERNEL_VERSION >= 132415
+# define __ASSUME_POSIX_TIMERS		1
+#endif
+
+/* The late 2.5 kernels saw a lot of new CLONE_* flags.  Summarize
+   their availability with one define.  The changes were made first
+   for i386 and the have to be done separately for the other archs.
+   For ia64, s390*, PPC, x86-64 we pick 2.5.64 as the first version
+   with support.  */
+#if __LINUX_KERNEL_VERSION >= 132416 \
+    && (defined __ia64__ || defined __s390__ || defined __powerpc__ \
+	|| defined __x86_64__ || defined __sh__)
+# define __ASSUME_CLONE_THREAD_FLAGS	1
+#endif
+
+/* With kernel 2.4.17 we always have netlink support.  */
+#if __LINUX_KERNEL_VERSION >= (132096+17)
+# define __ASSUME_NETLINK_SUPPORT	1
+#endif
+
+/* The requeue futex functionality was introduced in 2.5.70.  */
+#if __LINUX_KERNEL_VERSION >= 132422
+# define __ASSUME_FUTEX_REQUEUE	1
+#endif
+
+/* The statfs64 syscalls are available in 2.5.74.  */
+#if __LINUX_KERNEL_VERSION >= 132426
+# define __ASSUME_STATFS64	1
+#endif
+
+/* Starting with at least 2.5.74 the kernel passes the setuid-like exec
+   flag unconditionally up to the child.  */
+#if __LINUX_KERNEL_VERSION >= 132426
+# define __ASSUME_AT_SECURE	1
+#endif
+
+/* Starting with the 2.5.75 kernel the kernel fills in the correct value
+   in the si_pid field passed as part of the siginfo_t struct to signal
+   handlers.  */
+#if __LINUX_KERNEL_VERSION >= 132427
+# define __ASSUME_CORRECT_SI_PID	1
+#endif
+
+/* The tgkill syscall was instroduced for i386 in 2.5.75.  For Alpha
+   it was introduced in 2.6.0-test1 which unfortunately cannot be
+   distinguished from 2.6.0.  On x86-64, ppc, and ppc64 it was
+   introduced in 2.6.0-test3. */
+#if (__LINUX_KERNEL_VERSION >= 132427 && defined __i386__) \
+    || (__LINUX_KERNEL_VERSION >= 132609 && defined __alpha__) \
+    || (__LINUX_KERNEL_VERSION >= 132609 && defined __x86_64__) \
+    || (__LINUX_KERNEL_VERSION >= 132609 && defined __powerpc__) \
+    || (__LINUX_KERNEL_VERSION >= 132609 && defined __sh__)
+# define __ASSUME_TGKILL	1
+#endif
+
+/* The utimes syscall has been available for some architectures
+   forever.  For x86 it was introduced after 2.5.75, for x86-64,
+   ppc, and ppc64 it was introduced in 2.6.0-test3.  */
+#if defined __alpha__ || defined __ia64__ || defined __hppa__ \
+    || defined __sparc__ \
+    || (__LINUX_KERNEL_VERSION > 132427 && defined __i386__) \
+    || (__LINUX_KERNEL_VERSION > 132609 && defined __x86_64__) \
+    || (__LINUX_KERNEL_VERSION >= 132609 && defined __powerpc__) \
+    || (__LINUX_KERNEL_VERSION >= 132609 && defined __sh__)
+# define __ASSUME_UTIMES	1
+#endif
+
+// XXX Disabled for now since the semantics we want is not achieved.
+#if 0
+/* The CLONE_STOPPED flag was introduced in the 2.6.0-test1 series.  */
+#if __LINUX_KERNEL_VERSION >= 132609
+# define __ASSUME_CLONE_STOPPED	1
+#endif
+#endif
+
+/* The fixed version of the posix_fadvise64 syscall appeared in
+   2.6.0-test3.  At least for x86.  Powerpc support appeared in
+   2.6.2, but for 32-bit userspace only.  */
+#if (__LINUX_KERNEL_VERSION >= 132609 && defined __i386__) \
+    || (__LINUX_KERNEL_VERSION >= 132610 && defined __powerpc__ \
+       && !defined __powerpc64__)
+# define __ASSUME_FADVISE64_64_SYSCALL	1
+#endif
+
+/* The PROT_GROWSDOWN/PROT_GROWSUP flags were introduced in the 2.6.0-test
+   series.  */
+#if __LINUX_KERNEL_VERSION >= 132609
+# define __ASSUME_PROT_GROWSUPDOWN	1
+#endif
+
+/* Starting with 2.6.0 PowerPC adds signal/swapcontext support for Vector
+   SIMD (AKA Altivec, VMX) instructions and register state.  This changes
+   the overall size of the sigcontext and adds the swapcontext syscall.  */
+#if __LINUX_KERNEL_VERSION >= 132608 && defined __powerpc__
+# define __ASSUME_SWAPCONTEXT_SYSCALL	1
+#endif
+
+/* The CLONE_DETACHED flag is not necessary in 2.6.2 kernels, it is
+   implied.  */
+#if __LINUX_KERNEL_VERSION >= 132610
+# define __ASSUME_NO_CLONE_DETACHED	1
+#endif
+
+/* Starting with version 2.6.4-rc1 the getdents syscall returns d_type
+   information as well.  */
+#if __LINUX_KERNEL_VERSION >= 132612
+# define __ASSUME_GETDENTS32_D_TYPE	1
+#endif
+
+/* Starting with version 2.5.3, the initial location returned by `brk'
+   after exec is always rounded up to the next page.  */
+#if __LINUX_KERNEL_VERSION >= 132355
+# define __ASSUME_BRK_PAGE_ROUNDED	1
+#endif
+
+/* Starting with version 2.6.9, the waitid system call is available.
+   Except for powerpc and powerpc64, where it is available in 2.6.12.  */
+#if (__LINUX_KERNEL_VERSION >= 0x020609 && !defined __powerpc__) \
+    || (__LINUX_KERNEL_VERSION >= 0x02060c && defined __powerpc__)
+# define __ASSUME_WAITID_SYSCALL	1
+#endif
+
+/* Starting with version 2.6.9, SSI_IEEE_RAISE_EXCEPTION exists.  */
+#if __LINUX_KERNEL_VERSION >= 0x020609 && defined __alpha__
+#define __ASSUME_IEEE_RAISE_EXCEPTION	1
+#endif