Home | History | Annotate | Line # | Download | only in epiphany
      1       1.1  mrg /* Unsigned 32 bit division optimized for Epiphany.
      2  1.1.1.11  mrg    Copyright (C) 2009-2024 Free Software Foundation, Inc.
      3       1.1  mrg    Contributed by Embecosm on behalf of Adapteva, Inc.
      4       1.1  mrg 
      5       1.1  mrg This file is part of GCC.
      6       1.1  mrg 
      7       1.1  mrg This file is free software; you can redistribute it and/or modify it
      8       1.1  mrg under the terms of the GNU General Public License as published by the
      9       1.1  mrg Free Software Foundation; either version 3, or (at your option) any
     10       1.1  mrg later version.
     11       1.1  mrg 
     12       1.1  mrg This file is distributed in the hope that it will be useful, but
     13       1.1  mrg WITHOUT ANY WARRANTY; without even the implied warranty of
     14       1.1  mrg MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
     15       1.1  mrg General Public License for more details.
     16       1.1  mrg 
     17       1.1  mrg Under Section 7 of GPL version 3, you are granted additional
     18       1.1  mrg permissions described in the GCC Runtime Library Exception, version
     19       1.1  mrg 3.1, as published by the Free Software Foundation.
     20       1.1  mrg 
     21       1.1  mrg You should have received a copy of the GNU General Public License and
     22       1.1  mrg a copy of the GCC Runtime Library Exception along with this program;
     23       1.1  mrg see the files COPYING3 and COPYING.RUNTIME respectively.  If not, see
     24       1.1  mrg <http://www.gnu.org/licenses/>.  */
     25       1.1  mrg 
     26       1.1  mrg #include "epiphany-asm.h"
     27       1.1  mrg 
     28       1.1  mrg 	FSTAB (__udivsi3,T_UINT)
     29       1.1  mrg 	.global SYM(__udivsi3)
     30       1.1  mrg 	.balign 4
     31       1.1  mrg 	HIDDEN_FUNC(__udivsi3)
     32       1.1  mrg SYM(__udivsi3):
     33       1.1  mrg 	sub r3,r0,r1
     34       1.1  mrg 	bltu .Lret0
     35       1.1  mrg 	mov r3,0x95
     36       1.1  mrg 	lsl r12,r3,23 ; 0x4a800000
     37       1.1  mrg 	lsl r3,r3,30 ; 0x40000000
     38       1.1  mrg 	orr r16,r0,r3
     39       1.1  mrg 	orr r2,r1,r3
     40       1.1  mrg 	 fsub r16,r16,r3
     41       1.1  mrg 	fsub r2,r2,r3
     42       1.1  mrg 	 lsr r3,r1,21
     43       1.1  mrg 	lsr r17,r0,21
     44       1.1  mrg 	movt r17,0x4a80
     45       1.1  mrg 	fsub r17,r17,r12
     46       1.1  mrg 	 movt r3,0x4a80
     47       1.1  mrg 	fsub r3,r3,r12
     48       1.1  mrg 	 mov r12,%low(.L0step)
     49       1.1  mrg 	movt r12,%high(.L0step)
     50       1.1  mrg 	mov r21,1
     51       1.1  mrg 	movne r16,r17
     52       1.1  mrg 	lsr r17,r1,21
     53       1.1  mrg 	movne r2,r3
     54       1.1  mrg 	lsr r3,r16,23 ; must mask lower bits of r2 in case op0 was ..
     55       1.1  mrg 	lsr r2,r2,23 ; .. shifted and op1 was not.
     56       1.1  mrg 	sub r3,r3,r2 ; calculate bit number difference.
     57       1.1  mrg 	lsl r1,r1,r3
     58       1.1  mrg 	lsr r16,r1,1
     59       1.1  mrg 	lsl r2,r21,r3
     60       1.1  mrg 	lsl r3,r3,3
     61       1.1  mrg 	sub r12,r12,r3
     62       1.1  mrg 	sub r3,r0,r1
     63       1.1  mrg 	movltu r3,r0
     64       1.1  mrg 	mov r0,0
     65       1.1  mrg 	movgteu r0,r2
     66       1.1  mrg 	lsr r2,r2,1
     67       1.1  mrg 	add r17,r2,r0
     68       1.1  mrg 	sub r1,r3,r16
     69       1.1  mrg 	movgteu r3,r1
     70       1.1  mrg 	movgteu r0,r17
     71       1.1  mrg 	sub r16,r16,1
     72       1.1  mrg 	jr r12
     73       1.1  mrg 	.rep 30
     74       1.1  mrg 	lsl r3,r3,1
     75       1.1  mrg 	sub r1,r3,r16
     76       1.1  mrg 	movgteu r3,r1
     77       1.1  mrg 	.endr
     78       1.1  mrg 	sub r2,r2,1 ; mask result bits from steps ...
     79       1.1  mrg 	and r3,r3,r2
     80       1.1  mrg 	orr r0,r0,r3 ; ... and combine with first bits.
     81       1.1  mrg 	nop
     82       1.1  mrg .L0step:rts
     83       1.1  mrg .Lret0:	mov r0,0
     84       1.1  mrg 	rts
     85       1.1  mrg 	ENDFUNC(__udivsi3)
     86