OpenCores
URL https://opencores.org/ocsvn/openrisc_me/openrisc_me/trunk

Subversion Repositories openrisc_me

[/] [openrisc/] [trunk/] [gnu-src/] [newlib-1.17.0/] [newlib/] [libc/] [machine/] [hppa/] [strncat.S] - Blame information for rev 407

Go to most recent revision | Details | Compare with Previous | View Log

Line No. Rev Author Line
1 148 jeremybenn
/*
2
 *  (c) Copyright 1986 HEWLETT-PACKARD COMPANY
3
 *
4
 *  To anyone who acknowledges that this file is provided "AS IS"
5
 *  without any express or implied warranty:
6
 *      permission to use, copy, modify, and distribute this file
7
 *  for any purpose is hereby granted without fee, provided that
8
 *  the above copyright notice and this notice appears in all
9
 *  copies, and that the name of Hewlett-Packard Company not be
10
 *  used in advertising or publicity pertaining to distribution
11
 *  of the software without specific, written prior permission.
12
 *  Hewlett-Packard Company makes no representations about the
13
 *  suitability of this software for any purpose.
14
 */
15
 
16
/*HPUX_ID:      @(#) $Revision: 1.1 $   */
17
/* strncat(s1,s2,n) : concatonate at most n characters from s2 onto s1 */
18
 
19
#include "DEFS.h"
20
 
21
#define d_addr  r26
22
#define s_addr  r25
23
#define count   r24
24
#define tmp1    r19
25
#define tmp2    r20
26
#define tmp3    r21
27
#define tmp4    r22
28
#define tmp5    arg3
29
#define tmp6    r31
30
#define save    r1
31
#define tmp7    ret1    /* source offset-- reset to orig source addr if not aligned */
32
 
33
 
34
ENTRY(strncat)
35
 
36
        comb,=          r0,s_addr,quit  /* quit if s2=NULL */
37
        copy      d_addr,ret0          /* The return value is the value of d_addr. DELAY SLOT*/
38
 
39
/* First look for end of s1 (d_addr) */
40
 
41
        extru       d_addr,31,2,tmp1   /* Extract the low two bits of the dest address. */
42
        combt,=         tmp1,r0,dont_mask
43
        dep             0,31,2,d_addr   /*set word alignment */
44
        ldwm            4(d_addr),tmp2
45
        sh3add          tmp1,r0,save    /* build mask based on tmp1 */
46
        mtctl           save,11
47
        zvdepi          -2,32,save
48
        or              save,tmp2,tmp2
49
        uxor,nbz        tmp2,r0,save
50
search:
51
        b,n             found_end       /* nullified under uxor conditions above and below */
52
dont_mask:
53
        ldwm            4(d_addr),tmp2
54
        comib,tr        r0,r0,search
55
        uxor,nbz        tmp2,r0,save
56
 
57
found_end:                              /* at this point d_addr points to word */
58
        extru,<>        save,7,8,r0     /* following word with null */
59
        addib,tr,n      -4,d_addr,begin_copy    /*set d_addr to end of s1 */
60
        extru,<>        save,15,8,r0
61
        addib,tr,n      -3,d_addr,begin_copy
62
        extru,<>        save,23,8,r0
63
        addi            -1,d_addr,d_addr
64
        addi            -1,d_addr,d_addr
65
 
66
 
67
begin_copy:
68
        addibt,<,n  -4,count,byteloop     /* If count is <= 4 don't get fancy.*/
69
 
70
        extru       s_addr,31,2,tmp4   /* Extract the low two bits of the source address.*/
71
        extru       d_addr,31,2,tmp5   /* Extract the low two bits of the destination address.*/
72
        add         count,tmp5,count   /* pre increment the count by the byte address so that the count is*/
73
        copy            s_addr,tmp6     /* save original s_addr in case we find null in first word */
74
        copy         s_addr, tmp7       /* save s_addr in case we find null before first store */
75
        comb,<>       tmp5,tmp4,not_aligned /* branch if tmp5<>tmp4. */
76
        dep         0,31,2,s_addr      /* Compute the word address of the source.  DELAY SLOT.*/
77
/* aligned*/
78
        combt,=         tmp5,r0,skip_mask
79
        ldwm        4(0,s_addr),tmp1   /* tmp1 = *s_addr   s_addr += 4 (DELAY SLOT)*/
80
        sh3add          tmp5,r0,save    /* compute mask in save*/
81
        mtctl           save,11
82
        zvdepi          -2,32,save
83
        or              save,tmp1,tmp1  /* or mask with data*/
84
        uxor,nbz        tmp1,r0,save    /* check for null*/
85
        b,n             null1
86
        addibt,<        -4,count,back_porch
87
        stbys,b,m   tmp1,4(0,d_addr)   /* store word (delay slot)*/
88
 
89
chunks:
90
        ldwm            4(0,s_addr),tmp1 /* get a word*/
91
 
92
skip_mask:
93
        uxor,nbz        tmp1,r0,save    /* check for null*/
94
        b,n             align_null1
95
        addibf,<        -4,count,chunks
96
        stbys,b,m   tmp1,4(0,d_addr)   /* store word (delay slot)*/
97
 
98
back_porch:                                /* last word to store*/
99
         addibt,=,n  4,count,done       /* if count = 0 we're, of course, done !*/
100
         ldws        0(s_addr),tmp1     /* load up the back_porch*/
101
         sh3add         count,r0, save  /* setup right mask based on count*/
102
         mtctl          save,r11
103
         zvdepi         -2,32,save      /*save now has left-hand mask*/
104
         uaddcm         r0,save,save    /*form right hand mask */
105
         or             tmp1,save,tmp1  /*and insert data*/
106
         uxor,nbz       tmp1,r0,save    /* check for null*/
107
         b,n            null2
108
         add         d_addr,count,d_addr/* final store address  is +1 too high !*/
109
         b              done
110
         stbys,e        tmp1,0(d_addr)  /* done */
111
 
112
/* Begin non_aligned code. */
113
not_aligned:
114
        sub,>=       tmp5,tmp4,tmp6     /* compute the shift amt.and skip load if tmp5 > tmp4.*/
115
        ldwm         4(0,s_addr),tmp1   /* load up the first word from the source. tmp1 = *s_addr++*/
116
        zdep         tmp6,28,29,tmp4    /* compute the number of bits to shift */
117
        mtctl        tmp4,11            /* load the shift count into cr11 = shift count register.*/
118
        addibt,<,n   -4,count,chkchnk2 /* first step in pre adjustment of count for looping.*/
119
 
120
        ldwm        4(0,s_addr),tmp2    /* get either first or second word from source. */
121
        combt,=         tmp5,r0,skip_mask4 /* don't mask if whole word is valid*/
122
        vshd        tmp1,tmp2,tmp3      /* position data !  (delay slot)*/
123
        sh3add          tmp5,r0,save    /* setup r1*/
124
        mtctl           save,r11        /* setup mask in save*/
125
        zvdepi          -2,32,save
126
        or              save, tmp3, tmp3
127
        mtctl           tmp4,11            /* re-load the shift count into cr11 */
128
 
129
skip_mask4:
130
        uxor,nbz        tmp3, r0, save
131
        b,n             null4           /* special case for first word */
132
        copy            r0, tmp5        /* zero out tmp5 so we don't try to mask again*/
133
        copy            r0, tmp7        /* zero out tmp7 so we don't try to use original s_addr anymore */
134
        b               continue
135
        stbys,b,m   tmp3,4(0,d_addr)    /* store ! */
136
 
137
chunk2:
138
        ldwm        4(0,s_addr),tmp2
139
        vshd        tmp1,tmp2,tmp3
140
 
141
skip_mask2:
142
        uxor,nbz        tmp3, r0, save
143
        b,n             null3
144
        stbys,b,m   tmp3,4(0,d_addr)    /* store ! */
145
 
146
continue:
147
        ldwm        4(0,s_addr),tmp1    /* get 2nd word ! */
148
        vshd        tmp2,tmp1,tmp3      /* position data ! */
149
        uxor,nbz        tmp3, r0, save
150
        b,n             null3
151
 
152
        addibf,<    -8,count,chunk2    /* If count is still >= 8 do another loop.*/
153
        stbys,b,m   tmp3,4(0,d_addr)    /* store !*/
154
 
155
chkchnk2:
156
        addibt,<,n  4,count,bp_0       /* if we don't have 4 bytes left then do the back porch (bp_0)*/
157
 
158
subchnk2: /* we have less than 8 chars to copy*/
159
 
160
        ldwm        4(0,s_addr),tmp2    /* get next word !*/
161
        combt,=         tmp5,r0,skip_mask3
162
        vshd        tmp1,tmp2,tmp3      /* position data !*/
163
        sh3add          tmp5,r0,save    /* setup r1*/
164
        mtctl           save,r11        /* setup mask in save*/
165
        zvdepi          -2,32,save
166
        or              save, tmp3, tmp3
167
        mtctl           tmp4,11         /* restore shift value again */
168
skip_mask3:
169
        uxor,nbz        tmp3,r0,save
170
        b,n             null3
171
        copy            r0,tmp5   /* zero out tmp5 so null3 does correct alignment */
172
        copy            r0,tmp7   /* zero out tmp7 so we don't use orignal s_addr since no longer valid */
173
        b               bp_1 /* we now have less than 4 bytes to move*/
174
        stbys,b,m   tmp3,4(0,d_addr)    /* store !*/
175
 
176
bp_0:
177
        copy            tmp1,tmp2       /* switch registers for shift process */
178
        addibt,<=,n  4,count,done        /* if count = -4 this implies that count = 0 -> done */
179
 
180
bp_1:
181
        ldwm        4(0,s_addr),tmp1    /* get final word !        */
182
        vshd        tmp2,tmp1,tmp3      /* position data !*/
183
        uxor,nbz        tmp3,r0,save    /* if no-byte-zero */
184
        b,n             bp_null         /* don't goto no_null-find which null instead */
185
no_null:
186
        add         d_addr,count,d_addr /* set up d_addr for stbys,e */
187
        b               done            /* were done*/
188
        stbys,e     tmp3,0(0,d_addr)    /* store the data !*/
189
 
190
/* here we do ye old byte-at-a-time moves.*/
191
align_null1:
192
        b       byteloop
193
        addi    -4,s_addr,s_addr
194
null1:
195
        copy            tmp6,s_addr             /* restore orig s_addr (aligned only) */
196
byteloop:
197
        addibt,=     4,count,done
198
null2:
199
        ldbs,ma     1(s_addr),tmp1
200
encore:
201
        combt,=,n       tmp1,r0, done
202
        stbs,ma     tmp1,1(d_addr)
203
        addibf,=,n  -1,count,encore
204
        ldbs,ma     1(s_addr),tmp1
205
        b,n             done
206
 
207
bp_null:
208
        addi    -4,count,count          /* fudge count 'cause byteloop will re-increment */
209
 
210
null3:  /* not_aligned case reset s_addr and finish byte-wise */
211
        combt,=,n  r0,tmp7,null3a       /* if tmp7 is not valid address then branch below */
212
        b       byteloop                        /* otherwise reset s_addr to tmp7 and finish */
213
        copy    tmp7, s_addr
214
 
215
null3a: /* right shift target */
216
        addibt,<,n  0,tmp6,null3b       /* if left shifting */
217
        sub     r0,tmp6,tmp6            /* do null3b code */
218
        addi    -4,tmp6,tmp6
219
        b       byteloop
220
        add     tmp6,s_addr,s_addr      /* reset s_addr by 4 + shift_amt */
221
 
222
null3b:
223
        subi    -8,tmp6,tmp6
224
        add     tmp5,tmp6,tmp6  /* adjust by the dest offset if this is our first store */
225
        b       byteloop
226
        add     tmp6,s_addr,s_addr      /* adjust s_addr by (8-shift_amt-dest_off) */
227
 
228
null4:
229
        add,>   tmp6,r0,tmp6            /* if left shift */
230
        b,n     null3                   /* then do null3 */
231
        b       byteloop
232
        addi    -4,s_addr,s_addr        /* adj source only by 4 */
233
 
234
done:
235
        bv              0(r2)
236
        stbs            r0,0(d_addr)
237
quit:
238
EXIT(strncat)

powered by: WebSVN 2.1.0

© copyright 1999-2024 OpenCores.org, equivalent to Oliscience, all rights reserved. OpenCores®, registered trademark.