mirror of
https://sourceware.org/git/glibc.git
synced 2024-11-06 05:10:05 +00:00
8f5ca04bc7
* malloc/malloc.c (_malloc_internal): Performance fix. Move if statement out of loop. * stdio/_itoa.c, stdio/_itoa.h: Complete rewrite. Much faster implementation using GMP functions. Contributed by Torbjorn Granlund and Ulrich Drepper. * stdio/test_rdwr.c: Include <errno.h>. * sysdeps/i386/i586/Implies: New file. New highly optimized string functions for i[345]86. * sysdeps/i386/memchr.S, sysdeps/i386/memcmp.S: New files. * sysdeps/i386/stpcpy.S, sysdeps/i386/stpncpy.S: New files. * sysdeps/i386/strchr.S, sysdeps/i386/strcspn.S: New files. * sysdeps/i386/strpbrk.S, sysdeps/i386/strrchr.S: New files. * sysdeps/i386/strspn.S, sysdeps/i386/i486/strcat.S: New files. * sysdeps/i386/i486/strlen.S, sysdeps/i386/i586/strchr.S: New files. * sysdeps/i386/i586/strlen.S: New file. * sysdeps/i386/memchr.c: Removed. There is now an assembler version. * sysdeps/i386/i586/memcopy.h (WORD_COPY_BWD): Parameters did not correspond to used values. * sysdeps/unix/sysv/linux/nfs/nfs.h: New file. Simply a wrapper around a kernel header file. * sysdeps/unix/sysv/linux/Dist: Add it. * sysdeps/unix/sysv/linux/Makefile [$(subdir)=sunrpc] (headers): Likewise. * sysdeps/unix/sysv/linux/local_lim.h: Rewrite. Instead of defining ourself we use a kernel header file. * sysdeps/unix/sysv/linux/i386/sysdep.h (DO_CALL): Optimize system call handler for i586. * sysdeps/unix/sysv/linux/sys/param.h: Add copyright and clean up. Sat Oct 14 02:52:36 1995 Ulrich Drepper <drepper@ipd.info.uni-karlsruhe.de> * malloc/malloc.c (_malloc_internal): Performance fix. Move if statement out of loop. * stdio/_itoa.c, stdio/_itoa.h: Complete rewrite. Much faster implementation using GMP functions. Contributed by Torbjorn Granlund and Ulrich Drepper. * stdio/test_rdwr.c: Include <errno.h>. * sysdeps/i386/i586/Implies: New file. New highly optimized string functions for i[345]86. * sysdeps/i386/memchr.S, sysdeps/i386/memcmp.S: New files. * sysdeps/i386/stpcpy.S, sysdeps/i386/stpncpy.S: New files. * sysdeps/i386/strchr.S, sysdeps/i386/strcspn.S: New files. * sysdeps/i386/strpbrk.S, sysdeps/i386/strrchr.S: New files. * sysdeps/i386/strspn.S, sysdeps/i386/i486/strcat.S: New files. * sysdeps/i386/i486/strlen.S, sysdeps/i386/i586/strchr.S: New files. * sysdeps/i386/i586/strlen.S: New file. * sysdeps/i386/memchr.c: Removed. There is now an assembler version. * sysdeps/i386/i586/memcopy.h (WORD_COPY_BWD): Parameters did not correspond to used values. * sysdeps/unix/sysv/linux/nfs/nfs.h: New file. Simply a wrapper around a kernel header file. * sysdeps/unix/sysv/linux/Dist: Add it. * sysdeps/unix/sysv/linux/Makefile [$(subdir)=sunrpc] (headers): Likewise. * sysdeps/unix/sysv/linux/local_lim.h: Rewrite. Instead of defining ourself we use a kernel header file. * sysdeps/unix/sysv/linux/i386/sysdep.h (DO_CALL): Optimize system call handler for i586. * sysdeps/unix/sysv/linux/sys/param.h: Add copyright and clean up.
123 lines
3.0 KiB
ArmAsm
123 lines
3.0 KiB
ArmAsm
# IBM POWER __mpn_addmul_1 -- Multiply a limb vector with a limb and add
|
|
# the result to a second limb vector.
|
|
|
|
# Copyright (C) 1992, 1994 Free Software Foundation, Inc.
|
|
|
|
# This file is part of the GNU MP Library.
|
|
|
|
# The GNU MP 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 MP 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 MP Library; see the file COPYING.LIB. If not, write to
|
|
# the Free Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
|
|
|
|
|
# INPUT PARAMETERS
|
|
# res_ptr r3
|
|
# s1_ptr r4
|
|
# size r5
|
|
# s2_limb r6
|
|
|
|
# The RS/6000 has no unsigned 32x32->64 bit multiplication instruction. To
|
|
# obtain that operation, we have to use the 32x32->64 signed multiplication
|
|
# instruction, and add the appropriate compensation to the high limb of the
|
|
# result. We add the multiplicand if the multiplier has its most significant
|
|
# bit set, and we add the multiplier if the multiplicand has its most
|
|
# significant bit set. We need to preserve the carry flag between each
|
|
# iteration, so we have to compute the compensation carefully (the natural,
|
|
# srai+and doesn't work). Since the POWER architecture has a branch unit
|
|
# we can branch in zero cycles, so that's how we perform the additions.
|
|
|
|
.toc
|
|
.csect .__mpn_addmul_1[PR]
|
|
.align 2
|
|
.globl __mpn_addmul_1
|
|
.globl .__mpn_addmul_1
|
|
.csect __mpn_addmul_1[DS]
|
|
__mpn_addmul_1:
|
|
.long .__mpn_addmul_1[PR], TOC[tc0], 0
|
|
.csect .__mpn_addmul_1[PR]
|
|
.__mpn_addmul_1:
|
|
|
|
cal 3,-4(3)
|
|
l 0,0(4)
|
|
cmpi 0,6,0
|
|
mtctr 5
|
|
mul 9,0,6
|
|
srai 7,0,31
|
|
and 7,7,6
|
|
mfmq 8
|
|
cax 9,9,7
|
|
l 7,4(3)
|
|
a 8,8,7 # add res_limb
|
|
blt Lneg
|
|
Lpos: bdz Lend
|
|
|
|
Lploop: lu 0,4(4)
|
|
stu 8,4(3)
|
|
cmpi 0,0,0
|
|
mul 10,0,6
|
|
mfmq 0
|
|
ae 8,0,9 # low limb + old_cy_limb + old cy
|
|
l 7,4(3)
|
|
aze 10,10 # propagate cy to new cy_limb
|
|
a 8,8,7 # add res_limb
|
|
bge Lp0
|
|
cax 10,10,6 # adjust high limb for negative limb from s1
|
|
Lp0: bdz Lend0
|
|
lu 0,4(4)
|
|
stu 8,4(3)
|
|
cmpi 0,0,0
|
|
mul 9,0,6
|
|
mfmq 0
|
|
ae 8,0,10
|
|
l 7,4(3)
|
|
aze 9,9
|
|
a 8,8,7
|
|
bge Lp1
|
|
cax 9,9,6 # adjust high limb for negative limb from s1
|
|
Lp1: bdn Lploop
|
|
|
|
b Lend
|
|
|
|
Lneg: cax 9,9,0
|
|
bdz Lend
|
|
Lnloop: lu 0,4(4)
|
|
stu 8,4(3)
|
|
cmpi 0,0,0
|
|
mul 10,0,6
|
|
mfmq 7
|
|
ae 8,7,9
|
|
l 7,4(3)
|
|
ae 10,10,0 # propagate cy to new cy_limb
|
|
a 8,8,7 # add res_limb
|
|
bge Ln0
|
|
cax 10,10,6 # adjust high limb for negative limb from s1
|
|
Ln0: bdz Lend0
|
|
lu 0,4(4)
|
|
stu 8,4(3)
|
|
cmpi 0,0,0
|
|
mul 9,0,6
|
|
mfmq 7
|
|
ae 8,7,10
|
|
l 7,4(3)
|
|
ae 9,9,0 # propagate cy to new cy_limb
|
|
a 8,8,7 # add res_limb
|
|
bge Ln1
|
|
cax 9,9,6 # adjust high limb for negative limb from s1
|
|
Ln1: bdn Lnloop
|
|
b Lend
|
|
|
|
Lend0: cal 9,0(10)
|
|
Lend: st 8,4(3)
|
|
aze 3,9
|
|
br
|