f671aeab4e
1997-05-31 02:33 Ulrich Drepper <drepper@cygnus.com> * io/ftwtest-sh: More tests. * misc/tsearch.c: Rewrite tdestroy_recursive. * libio/libio.h: Define bits for libg++-2.8. 1997-05-30 22:21 Thorsten Kukuk <kukuk@vt.uni-paderborn.de> * nis/nss_nis/nis-hosts.c: Same changes as in nss/nss_files/files-hosts.c (Always use inet_pton). * nis/nss_nisplus/nisplus-hosts.c: Likewise. 1997-05-30 09:50 Richard Henderson <rth@tamu.edu> * manual/maint.texi: ECOFF hasn't been tested in ages and I don't plan on doing so ever again. Don't say alpha-linuxecoff is known to work. * elf/dl-lookup.c: Include <alloca.h>. * misc/tsearch.c: Include <string.h>. * posix/execle.c: Include <alloca.h>. * posix/execlp.c: Likewise. * stdio-common/printf_fphex.c: Include <string.h>. * sunrpc/xdr.c: Include <string.h>. * sysdeps/generic/memccpy.c: Include only <string.h>. * sunrpc/clnt_udp.c (clntudp_call): Make fromlen a size_t to fix parameters to network functions. * sunrpc/pmap_rmt.c (clnt_broadcast): Likewise. * sunrpc/svc_tcp.c (svctcp_create): Likewise with `len'. (rendezvous_request): Likewise. * sunrpc/svc_udp.c (svcudp_bufcreate): Likewise. (svcudp_recv): Similar, but go through a local variable. The old code used &xprt->xp_addrlen. Someone should determine if the type of the structure member should be changed instead. * sunrpc/xdr_rec.c (xdrrec_create): Expunge a ptr->int cast warning and transform a nonsense for-loop to a closed form calculation. (fill_input_buf): Another ptr->int warning. * sysdeps/alpha/Makefile: Temporarily turn on -mieee globally. This will last until I figure out how to build a parallel libm_ieee. * sysdeps/alpha/fpu/fclrexcpt.c, * sysdeps/alpha/fpu/fegetenv.c, * sysdeps/alpha/fpu/fegetround.c, * sysdeps/alpha/fpu/feholdexcpt.c, * sysdeps/alpha/fpu/fenvbits.h, * sysdeps/alpha/fpu/fesetenv.c, * sysdeps/alpha/fpu/fesetround.c, * sysdeps/alpha/fpu/feupdateenv.c, * sysdeps/alpha/fpu/fgetexcptflg.c, * sysdeps/alpha/fpu/fraiseexcpt.c, * sysdeps/alpha/fpu/fsetexcptflg.c, * sysdeps/alpha/fpu/ftestexcept.c: New files. There is, btw, a small kernel patch that must be applied as of 2.0.31-pre1 and 2.1.41 for this to work properly. * sysdeps/alpha/fpu/fpu_control.h: Update copyright. Change default FP mode to round to nearest, no exceptions as for Intel. * sysdeps/unix/sysv/linux/alpha/setfpucw.c: Add copyright. (rdfpcr): Use excb rather than trapb. Be more efficient about getting at the fpcr. (wrfpcr): Likewise. (__setfpucw): Reformat. * sysdeps/ieee754/huge_val.h: Change GCC's HUGE_VAL definition to use a single DI-mode integer instead of bytes. This produces better code on Alpha. Similar for HUGE_VALF. * sysdeps/ieee754/nan.h: Similar for NAN. Add NANF and NANL. * sysdeps/libm-ieee754/s_log2.c: Missing */ and weak_alias. * sysdeps/unix/alpha/sysdep.S: Tiny cleanups. * sysdeps/unix/sysv/linux/alpha/syscall.S: Likewise. * sysdeps/unix/sysv/linux/alpha/brk.S: Don't use jmp macro to make sure the stack frame is released the very instruction before the real jump out. * sysdeps/unix/sysv/linux/alpha/ieee_get_fp_control.S: Likewise. * sysdeps/unix/sysv/linux/alpha/ieee_set_fp_control.S: Likewise. * sysdeps/unix/sysv/linux/alpha/gnu/types.h: Undef some bits we may have gotten from <linux/posix_types.h>. * sysdeps/unix/sysv/linux/alpha/ioperm.c: Add definition for Miata. 1997-05-30 13:05 Ulrich Drepper <drepper@cygnus.com> * misc/tst-tsearch.c (mangle_tree): Update casts to prevent warnings. 1997-05-29 14:02 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de> * sysdeps/unix/sysv/linux/sys/ttydefaults.h (CEOL, CSTATUS): Use correct value for _POSIX_VDISABLE. 1997-05-29 13:59 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de> * Makefile (manual/dir-add.texi): Force execution. 1997-05-29 13:57 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de> * sysexits.h: New file. 1997-05-29 13:55 Andreas Schwab <schwab@issan.informatik.uni-dortmund.de> * stdlib/strfmon.c: Don't run past EOS after `='. 1997-02-27 10:34 Richard Henderson <rth@tamu.edu>
108 lines
3.4 KiB
C
108 lines
3.4 KiB
C
/* Copyright (C) 1997 Free Software Foundation, Inc.
|
|
This file is part of the GNU C Library.
|
|
Contributed by Richard Henderson <rth@tamu.edu>, 1997
|
|
|
|
The GNU C Library is free software; you can redistribute it and/or
|
|
modify it under the terms of the GNU Library General Public License as
|
|
published by the Free Software Foundation; either version 2 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
|
|
Library General Public License for more details.
|
|
|
|
You should have received a copy of the GNU Library General Public
|
|
License along with the GNU C Library; see the file COPYING.LIB. If not,
|
|
write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
|
|
Boston, MA 02111-1307, USA. */
|
|
|
|
/* This file should never be included directly. */
|
|
|
|
#ifndef _FENVBITS_H
|
|
#define _FENVBITS_H 1
|
|
|
|
/* Define the bits representing the exception.
|
|
|
|
Note that these are the bit positions as defined by the OSF/1
|
|
ieee_{get,set}_control_word interface and not by the hardware fpcr.
|
|
|
|
See the Alpha Architecture Handbook section 4.7.7.3 for details,
|
|
but in summary, trap shadows mean the hardware register can acquire
|
|
extra exception bits so for proper IEEE support the tracking has to
|
|
be done in software -- in this case with kernel support.
|
|
|
|
As to why the system call interface isn't in the same format as
|
|
the hardware register, only those crazy folks at DEC can tell you. */
|
|
|
|
enum
|
|
{
|
|
FE_INEXACT = 1UL << 21,
|
|
#define FE_INEXACT FE_INEXACT
|
|
|
|
FE_UNDERFLOW = 1UL << 20,
|
|
#define FE_UNDERFLOW FE_UNDERFLOW
|
|
|
|
FE_OVERFLOW = 1UL << 19,
|
|
#define FE_OVERFLOW FE_OVERFLOW
|
|
|
|
FE_DIVBYZERO = 1UL << 18,
|
|
#define FE_DIVBYZERO FE_DIVBYZERO
|
|
|
|
FE_INVALID = 1UL << 17,
|
|
#define FE_INVALID FE_INVALID
|
|
|
|
FE_ALL_EXCEPT =
|
|
(FE_INEXACT | FE_DIVBYZERO | FE_UNDERFLOW | FE_OVERFLOW | FE_INVALID)
|
|
#define FE_ALL_EXCEPT FE_ALL_EXCEPT
|
|
};
|
|
|
|
|
|
/* Alpha chips support all four defined rouding modes.
|
|
|
|
Note that code must be compiled to use dynamic rounding (/d) instructions
|
|
to see these changes. For gcc this is -mfp-rounding-mode=d; for DEC cc
|
|
this is -fprm d. The default for both is static rounding to nearest.
|
|
|
|
These are shifted down 58 bits from the hardware fpcr because the
|
|
functions are declared to take integers. */
|
|
|
|
enum
|
|
{
|
|
FE_TOWARDSZERO = 0,
|
|
#define FE_TOWARDSZERO FE_TOWARDSZERO
|
|
|
|
FE_DOWNWARD = 1,
|
|
#define FE_DOWNWARD FE_DOWNWARD
|
|
|
|
FE_TONEAREST = 2,
|
|
#define FE_TONEAREST FE_TONEAREST
|
|
|
|
FE_UPWARD = 3,
|
|
#define FE_UPWARD FE_UPWARD
|
|
};
|
|
|
|
|
|
/* Type representing exception flags. */
|
|
typedef unsigned long fexcept_t;
|
|
|
|
/* Type representing floating-point environment. */
|
|
typedef unsigned long fenv_t;
|
|
|
|
/* If the default argument is used we use this value. Note that due to
|
|
architecture-specified page mappings, no user-space pointer will ever
|
|
have its two high bits set. Co-opt one. */
|
|
#define FE_DFL_ENV ((fenv_t *) 0x8800000000000000UL)
|
|
|
|
#ifdef __USE_GNU
|
|
/* Floating-point environment where none of the exceptions are masked. */
|
|
# define FE_NOMASK_ENV ((fenv_t *) 0x880000000000003eUL)
|
|
#endif
|
|
|
|
/* The system calls to talk to the kernel's FP code. */
|
|
extern unsigned long __ieee_get_fp_control(void);
|
|
extern void __ieee_set_fp_control(unsigned long);
|
|
|
|
|
|
#endif /* fenvbits.h */
|