aboutsummaryrefslogtreecommitdiffstats
path: root/arch/sh/lib
diff options
context:
space:
mode:
authorPaul Mundt <lethal@linux-sh.org>2008-11-18 03:33:48 -0500
committerPaul Mundt <lethal@linux-sh.org>2008-12-22 04:42:53 -0500
commit1aad54a99b6ce316c851ba99b2efe41998cfd37d (patch)
tree287f90771c7da18a0841d40db1c0ce92ea7f5820 /arch/sh/lib
parente9bf51e5ccc7703226c79888603e157066213700 (diff)
sh: Migrate necessary libgcc bits in to arch/sh/lib for SUPERH32.
This moves in the necessary libgcc bits for SUPERH32 and drops the libgcc linking for the regular targets. This in turn allows us to rip out quite a few hacks both in sh_ksyms_32 and arch/sh/Makefile. Signed-off-by: Paul Mundt <lethal@linux-sh.org>
Diffstat (limited to 'arch/sh/lib')
-rw-r--r--arch/sh/lib/Makefile14
-rw-r--r--arch/sh/lib/ashiftrt.S149
-rw-r--r--arch/sh/lib/ashldi3.c29
-rw-r--r--arch/sh/lib/ashlsi3.S193
-rw-r--r--arch/sh/lib/ashrdi3.c31
-rw-r--r--arch/sh/lib/ashrsi3.S185
-rw-r--r--arch/sh/lib/libgcc.h26
-rw-r--r--arch/sh/lib/lshrdi3.c29
-rw-r--r--arch/sh/lib/lshrsi3.S193
-rw-r--r--arch/sh/lib/movmem.S238
-rw-r--r--arch/sh/lib/udiv_qrnnd.S81
-rw-r--r--arch/sh/lib/udivsi3-Os.S147
-rw-r--r--arch/sh/lib/udivsi3.S664
13 files changed, 1978 insertions, 1 deletions
diff --git a/arch/sh/lib/Makefile b/arch/sh/lib/Makefile
index 596421821d08..a30acb8342d9 100644
--- a/arch/sh/lib/Makefile
+++ b/arch/sh/lib/Makefile
@@ -5,6 +5,18 @@
5lib-y = delay.o memset.o memmove.o memchr.o \ 5lib-y = delay.o memset.o memmove.o memchr.o \
6 checksum.o strlen.o div64.o div64-generic.o 6 checksum.o strlen.o div64.o div64-generic.o
7 7
8# Extracted from libgcc
9lib-y += movmem.o ashldi3.o ashrdi3.o lshrdi3.o \
10 ashlsi3.o ashrsi3.o ashiftrt.o lshrsi3.o \
11 udiv_qrnnd.o
12
13udivsi3-y := udivsi3-Os.o
14
15ifneq ($(CONFIG_CC_OPTIMIZE_FOR_SIZE),y)
16udivsi3-$(CONFIG_CPU_SH3) := udivsi3.o
17udivsi3-$(CONFIG_CPU_SH4) := udivsi3.o
18endif
19
8obj-y += io.o 20obj-y += io.o
9 21
10memcpy-y := memcpy.o 22memcpy-y := memcpy.o
@@ -12,6 +24,6 @@ memcpy-$(CONFIG_CPU_SH4) := memcpy-sh4.o
12 24
13lib-$(CONFIG_MMU) += copy_page.o clear_page.o 25lib-$(CONFIG_MMU) += copy_page.o clear_page.o
14lib-$(CONFIG_FUNCTION_TRACER) += mcount.o 26lib-$(CONFIG_FUNCTION_TRACER) += mcount.o
15lib-y += $(memcpy-y) 27lib-y += $(memcpy-y) $(udivsi3-y)
16 28
17EXTRA_CFLAGS += -Werror 29EXTRA_CFLAGS += -Werror
diff --git a/arch/sh/lib/ashiftrt.S b/arch/sh/lib/ashiftrt.S
new file mode 100644
index 000000000000..45ce86558f46
--- /dev/null
+++ b/arch/sh/lib/ashiftrt.S
@@ -0,0 +1,149 @@
1/* Copyright (C) 1994, 1995, 1997, 1998, 1999, 2000, 2001, 2002, 2003,
2 2004, 2005, 2006
3 Free Software Foundation, Inc.
4
5This file is free software; you can redistribute it and/or modify it
6under the terms of the GNU General Public License as published by the
7Free Software Foundation; either version 2, or (at your option) any
8later version.
9
10In addition to the permissions in the GNU General Public License, the
11Free Software Foundation gives you unlimited permission to link the
12compiled version of this file into combinations with other programs,
13and to distribute those combinations without any restriction coming
14from the use of this file. (The General Public License restrictions
15do apply in other respects; for example, they cover modification of
16the file, and distribution when not linked into a combine
17executable.)
18
19This file is distributed in the hope that it will be useful, but
20WITHOUT ANY WARRANTY; without even the implied warranty of
21MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
22General Public License for more details.
23
24You should have received a copy of the GNU General Public License
25along with this program; see the file COPYING. If not, write to
26the Free Software Foundation, 51 Franklin Street, Fifth Floor,
27Boston, MA 02110-1301, USA. */
28
29!! libgcc routines for the Renesas / SuperH SH CPUs.
30!! Contributed by Steve Chamberlain.
31!! sac@cygnus.com
32
33!! ashiftrt_r4_x, ___ashrsi3, ___ashlsi3, ___lshrsi3 routines
34!! recoded in assembly by Toshiyasu Morita
35!! tm@netcom.com
36
37/* SH2 optimizations for ___ashrsi3, ___ashlsi3, ___lshrsi3 and
38 ELF local label prefixes by J"orn Rennecke
39 amylaar@cygnus.com */
40
41 .global __ashiftrt_r4_0
42 .global __ashiftrt_r4_1
43 .global __ashiftrt_r4_2
44 .global __ashiftrt_r4_3
45 .global __ashiftrt_r4_4
46 .global __ashiftrt_r4_5
47 .global __ashiftrt_r4_6
48 .global __ashiftrt_r4_7
49 .global __ashiftrt_r4_8
50 .global __ashiftrt_r4_9
51 .global __ashiftrt_r4_10
52 .global __ashiftrt_r4_11
53 .global __ashiftrt_r4_12
54 .global __ashiftrt_r4_13
55 .global __ashiftrt_r4_14
56 .global __ashiftrt_r4_15
57 .global __ashiftrt_r4_16
58 .global __ashiftrt_r4_17
59 .global __ashiftrt_r4_18
60 .global __ashiftrt_r4_19
61 .global __ashiftrt_r4_20
62 .global __ashiftrt_r4_21
63 .global __ashiftrt_r4_22
64 .global __ashiftrt_r4_23
65 .global __ashiftrt_r4_24
66 .global __ashiftrt_r4_25
67 .global __ashiftrt_r4_26
68 .global __ashiftrt_r4_27
69 .global __ashiftrt_r4_28
70 .global __ashiftrt_r4_29
71 .global __ashiftrt_r4_30
72 .global __ashiftrt_r4_31
73 .global __ashiftrt_r4_32
74
75 .align 1
76__ashiftrt_r4_32:
77__ashiftrt_r4_31:
78 rotcl r4
79 rts
80 subc r4,r4
81__ashiftrt_r4_30:
82 shar r4
83__ashiftrt_r4_29:
84 shar r4
85__ashiftrt_r4_28:
86 shar r4
87__ashiftrt_r4_27:
88 shar r4
89__ashiftrt_r4_26:
90 shar r4
91__ashiftrt_r4_25:
92 shar r4
93__ashiftrt_r4_24:
94 shlr16 r4
95 shlr8 r4
96 rts
97 exts.b r4,r4
98__ashiftrt_r4_23:
99 shar r4
100__ashiftrt_r4_22:
101 shar r4
102__ashiftrt_r4_21:
103 shar r4
104__ashiftrt_r4_20:
105 shar r4
106__ashiftrt_r4_19:
107 shar r4
108__ashiftrt_r4_18:
109 shar r4
110__ashiftrt_r4_17:
111 shar r4
112__ashiftrt_r4_16:
113 shlr16 r4
114 rts
115 exts.w r4,r4
116__ashiftrt_r4_15:
117 shar r4
118__ashiftrt_r4_14:
119 shar r4
120__ashiftrt_r4_13:
121 shar r4
122__ashiftrt_r4_12:
123 shar r4
124__ashiftrt_r4_11:
125 shar r4
126__ashiftrt_r4_10:
127 shar r4
128__ashiftrt_r4_9:
129 shar r4
130__ashiftrt_r4_8:
131 shar r4
132__ashiftrt_r4_7:
133 shar r4
134__ashiftrt_r4_6:
135 shar r4
136__ashiftrt_r4_5:
137 shar r4
138__ashiftrt_r4_4:
139 shar r4
140__ashiftrt_r4_3:
141 shar r4
142__ashiftrt_r4_2:
143 shar r4
144__ashiftrt_r4_1:
145 rts
146 shar r4
147__ashiftrt_r4_0:
148 rts
149 nop
diff --git a/arch/sh/lib/ashldi3.c b/arch/sh/lib/ashldi3.c
new file mode 100644
index 000000000000..beb80f316095
--- /dev/null
+++ b/arch/sh/lib/ashldi3.c
@@ -0,0 +1,29 @@
1#include <linux/module.h>
2
3#include "libgcc.h"
4
5long long __ashldi3(long long u, word_type b)
6{
7 DWunion uu, w;
8 word_type bm;
9
10 if (b == 0)
11 return u;
12
13 uu.ll = u;
14 bm = 32 - b;
15
16 if (bm <= 0) {
17 w.s.low = 0;
18 w.s.high = (unsigned int) uu.s.low << -bm;
19 } else {
20 const unsigned int carries = (unsigned int) uu.s.low >> bm;
21
22 w.s.low = (unsigned int) uu.s.low << b;
23 w.s.high = ((unsigned int) uu.s.high << b) | carries;
24 }
25
26 return w.ll;
27}
28
29EXPORT_SYMBOL(__ashldi3);
diff --git a/arch/sh/lib/ashlsi3.S b/arch/sh/lib/ashlsi3.S
new file mode 100644
index 000000000000..bd47e9b403a5
--- /dev/null
+++ b/arch/sh/lib/ashlsi3.S
@@ -0,0 +1,193 @@
1/* Copyright (C) 1994, 1995, 1997, 1998, 1999, 2000, 2001, 2002, 2003,
2 2004, 2005, 2006
3 Free Software Foundation, Inc.
4
5This file is free software; you can redistribute it and/or modify it
6under the terms of the GNU General Public License as published by the
7Free Software Foundation; either version 2, or (at your option) any
8later version.
9
10In addition to the permissions in the GNU General Public License, the
11Free Software Foundation gives you unlimited permission to link the
12compiled version of this file into combinations with other programs,
13and to distribute those combinations without any restriction coming
14from the use of this file. (The General Public License restrictions
15do apply in other respects; for example, they cover modification of
16the file, and distribution when not linked into a combine
17executable.)
18
19This file is distributed in the hope that it will be useful, but
20WITHOUT ANY WARRANTY; without even the implied warranty of
21MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
22General Public License for more details.
23
24You should have received a copy of the GNU General Public License
25along with this program; see the file COPYING. If not, write to
26the Free Software Foundation, 51 Franklin Street, Fifth Floor,
27Boston, MA 02110-1301, USA. */
28
29!! libgcc routines for the Renesas / SuperH SH CPUs.
30!! Contributed by Steve Chamberlain.
31!! sac@cygnus.com
32
33!! ashiftrt_r4_x, ___ashrsi3, ___ashlsi3, ___lshrsi3 routines
34!! recoded in assembly by Toshiyasu Morita
35!! tm@netcom.com
36
37/* SH2 optimizations for ___ashrsi3, ___ashlsi3, ___lshrsi3 and
38 ELF local label prefixes by J"orn Rennecke
39 amylaar@cygnus.com */
40
41!
42! __ashlsi3
43!
44! Entry:
45!
46! r4: Value to shift
47! r5: Shifts
48!
49! Exit:
50!
51! r0: Result
52!
53! Destroys:
54!
55! (none)
56!
57 .global __ashlsi3
58
59 .align 2
60__ashlsi3:
61 mov #31,r0
62 and r0,r5
63 mova ashlsi3_table,r0
64 mov.b @(r0,r5),r5
65#ifdef __sh1__
66 add r5,r0
67 jmp @r0
68#else
69 braf r5
70#endif
71 mov r4,r0
72
73 .align 2
74ashlsi3_table:
75 .byte ashlsi3_0-ashlsi3_table
76 .byte ashlsi3_1-ashlsi3_table
77 .byte ashlsi3_2-ashlsi3_table
78 .byte ashlsi3_3-ashlsi3_table
79 .byte ashlsi3_4-ashlsi3_table
80 .byte ashlsi3_5-ashlsi3_table
81 .byte ashlsi3_6-ashlsi3_table
82 .byte ashlsi3_7-ashlsi3_table
83 .byte ashlsi3_8-ashlsi3_table
84 .byte ashlsi3_9-ashlsi3_table
85 .byte ashlsi3_10-ashlsi3_table
86 .byte ashlsi3_11-ashlsi3_table
87 .byte ashlsi3_12-ashlsi3_table
88 .byte ashlsi3_13-ashlsi3_table
89 .byte ashlsi3_14-ashlsi3_table
90 .byte ashlsi3_15-ashlsi3_table
91 .byte ashlsi3_16-ashlsi3_table
92 .byte ashlsi3_17-ashlsi3_table
93 .byte ashlsi3_18-ashlsi3_table
94 .byte ashlsi3_19-ashlsi3_table
95 .byte ashlsi3_20-ashlsi3_table
96 .byte ashlsi3_21-ashlsi3_table
97 .byte ashlsi3_22-ashlsi3_table
98 .byte ashlsi3_23-ashlsi3_table
99 .byte ashlsi3_24-ashlsi3_table
100 .byte ashlsi3_25-ashlsi3_table
101 .byte ashlsi3_26-ashlsi3_table
102 .byte ashlsi3_27-ashlsi3_table
103 .byte ashlsi3_28-ashlsi3_table
104 .byte ashlsi3_29-ashlsi3_table
105 .byte ashlsi3_30-ashlsi3_table
106 .byte ashlsi3_31-ashlsi3_table
107
108ashlsi3_6:
109 shll2 r0
110ashlsi3_4:
111 shll2 r0
112ashlsi3_2:
113 rts
114 shll2 r0
115
116ashlsi3_7:
117 shll2 r0
118ashlsi3_5:
119 shll2 r0
120ashlsi3_3:
121 shll2 r0
122ashlsi3_1:
123 rts
124 shll r0
125
126ashlsi3_14:
127 shll2 r0
128ashlsi3_12:
129 shll2 r0
130ashlsi3_10:
131 shll2 r0
132ashlsi3_8:
133 rts
134 shll8 r0
135
136ashlsi3_15:
137 shll2 r0
138ashlsi3_13:
139 shll2 r0
140ashlsi3_11:
141 shll2 r0
142ashlsi3_9:
143 shll8 r0
144 rts
145 shll r0
146
147ashlsi3_22:
148 shll2 r0
149ashlsi3_20:
150 shll2 r0
151ashlsi3_18:
152 shll2 r0
153ashlsi3_16:
154 rts
155 shll16 r0
156
157ashlsi3_23:
158 shll2 r0
159ashlsi3_21:
160 shll2 r0
161ashlsi3_19:
162 shll2 r0
163ashlsi3_17:
164 shll16 r0
165 rts
166 shll r0
167
168ashlsi3_30:
169 shll2 r0
170ashlsi3_28:
171 shll2 r0
172ashlsi3_26:
173 shll2 r0
174ashlsi3_24:
175 shll16 r0
176 rts
177 shll8 r0
178
179ashlsi3_31:
180 shll2 r0
181ashlsi3_29:
182 shll2 r0
183ashlsi3_27:
184 shll2 r0
185ashlsi3_25:
186 shll16 r0
187 shll8 r0
188 rts
189 shll r0
190
191ashlsi3_0:
192 rts
193 nop
diff --git a/arch/sh/lib/ashrdi3.c b/arch/sh/lib/ashrdi3.c
new file mode 100644
index 000000000000..c884a912b660
--- /dev/null
+++ b/arch/sh/lib/ashrdi3.c
@@ -0,0 +1,31 @@
1#include <linux/module.h>
2
3#include "libgcc.h"
4
5long long __ashrdi3(long long u, word_type b)
6{
7 DWunion uu, w;
8 word_type bm;
9
10 if (b == 0)
11 return u;
12
13 uu.ll = u;
14 bm = 32 - b;
15
16 if (bm <= 0) {
17 /* w.s.high = 1..1 or 0..0 */
18 w.s.high =
19 uu.s.high >> 31;
20 w.s.low = uu.s.high >> -bm;
21 } else {
22 const unsigned int carries = (unsigned int) uu.s.high << bm;
23
24 w.s.high = uu.s.high >> b;
25 w.s.low = ((unsigned int) uu.s.low >> b) | carries;
26 }
27
28 return w.ll;
29}
30
31EXPORT_SYMBOL(__ashrdi3);
diff --git a/arch/sh/lib/ashrsi3.S b/arch/sh/lib/ashrsi3.S
new file mode 100644
index 000000000000..6f3cf46b77c2
--- /dev/null
+++ b/arch/sh/lib/ashrsi3.S
@@ -0,0 +1,185 @@
1/* Copyright (C) 1994, 1995, 1997, 1998, 1999, 2000, 2001, 2002, 2003,
2 2004, 2005, 2006
3 Free Software Foundation, Inc.
4
5This file is free software; you can redistribute it and/or modify it
6under the terms of the GNU General Public License as published by the
7Free Software Foundation; either version 2, or (at your option) any
8later version.
9
10In addition to the permissions in the GNU General Public License, the
11Free Software Foundation gives you unlimited permission to link the
12compiled version of this file into combinations with other programs,
13and to distribute those combinations without any restriction coming
14from the use of this file. (The General Public License restrictions
15do apply in other respects; for example, they cover modification of
16the file, and distribution when not linked into a combine
17executable.)
18
19This file is distributed in the hope that it will be useful, but
20WITHOUT ANY WARRANTY; without even the implied warranty of
21MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
22General Public License for more details.
23
24You should have received a copy of the GNU General Public License
25along with this program; see the file COPYING. If not, write to
26the Free Software Foundation, 51 Franklin Street, Fifth Floor,
27Boston, MA 02110-1301, USA. */
28
29!! libgcc routines for the Renesas / SuperH SH CPUs.
30!! Contributed by Steve Chamberlain.
31!! sac@cygnus.com
32
33!! ashiftrt_r4_x, ___ashrsi3, ___ashlsi3, ___lshrsi3 routines
34!! recoded in assembly by Toshiyasu Morita
35!! tm@netcom.com
36
37/* SH2 optimizations for ___ashrsi3, ___ashlsi3, ___lshrsi3 and
38 ELF local label prefixes by J"orn Rennecke
39 amylaar@cygnus.com */
40
41!
42! __ashrsi3
43!
44! Entry:
45!
46! r4: Value to shift
47! r5: Shifts
48!
49! Exit:
50!
51! r0: Result
52!
53! Destroys:
54!
55! (none)
56!
57
58 .global __ashrsi3
59
60 .align 2
61__ashrsi3:
62 mov #31,r0
63 and r0,r5
64 mova ashrsi3_table,r0
65 mov.b @(r0,r5),r5
66#ifdef __sh1__
67 add r5,r0
68 jmp @r0
69#else
70 braf r5
71#endif
72 mov r4,r0
73
74 .align 2
75ashrsi3_table:
76 .byte ashrsi3_0-ashrsi3_table
77 .byte ashrsi3_1-ashrsi3_table
78 .byte ashrsi3_2-ashrsi3_table
79 .byte ashrsi3_3-ashrsi3_table
80 .byte ashrsi3_4-ashrsi3_table
81 .byte ashrsi3_5-ashrsi3_table
82 .byte ashrsi3_6-ashrsi3_table
83 .byte ashrsi3_7-ashrsi3_table
84 .byte ashrsi3_8-ashrsi3_table
85 .byte ashrsi3_9-ashrsi3_table
86 .byte ashrsi3_10-ashrsi3_table
87 .byte ashrsi3_11-ashrsi3_table
88 .byte ashrsi3_12-ashrsi3_table
89 .byte ashrsi3_13-ashrsi3_table
90 .byte ashrsi3_14-ashrsi3_table
91 .byte ashrsi3_15-ashrsi3_table
92 .byte ashrsi3_16-ashrsi3_table
93 .byte ashrsi3_17-ashrsi3_table
94 .byte ashrsi3_18-ashrsi3_table
95 .byte ashrsi3_19-ashrsi3_table
96 .byte ashrsi3_20-ashrsi3_table
97 .byte ashrsi3_21-ashrsi3_table
98 .byte ashrsi3_22-ashrsi3_table
99 .byte ashrsi3_23-ashrsi3_table
100 .byte ashrsi3_24-ashrsi3_table
101 .byte ashrsi3_25-ashrsi3_table
102 .byte ashrsi3_26-ashrsi3_table
103 .byte ashrsi3_27-ashrsi3_table
104 .byte ashrsi3_28-ashrsi3_table
105 .byte ashrsi3_29-ashrsi3_table
106 .byte ashrsi3_30-ashrsi3_table
107 .byte ashrsi3_31-ashrsi3_table
108
109ashrsi3_31:
110 rotcl r0
111 rts
112 subc r0,r0
113
114ashrsi3_30:
115 shar r0
116ashrsi3_29:
117 shar r0
118ashrsi3_28:
119 shar r0
120ashrsi3_27:
121 shar r0
122ashrsi3_26:
123 shar r0
124ashrsi3_25:
125 shar r0
126ashrsi3_24:
127 shlr16 r0
128 shlr8 r0
129 rts
130 exts.b r0,r0
131
132ashrsi3_23:
133 shar r0
134ashrsi3_22:
135 shar r0
136ashrsi3_21:
137 shar r0
138ashrsi3_20:
139 shar r0
140ashrsi3_19:
141 shar r0
142ashrsi3_18:
143 shar r0
144ashrsi3_17:
145 shar r0
146ashrsi3_16:
147 shlr16 r0
148 rts
149 exts.w r0,r0
150
151ashrsi3_15:
152 shar r0
153ashrsi3_14:
154 shar r0
155ashrsi3_13:
156 shar r0
157ashrsi3_12:
158 shar r0
159ashrsi3_11:
160 shar r0
161ashrsi3_10:
162 shar r0
163ashrsi3_9:
164 shar r0
165ashrsi3_8:
166 shar r0
167ashrsi3_7:
168 shar r0
169ashrsi3_6:
170 shar r0
171ashrsi3_5:
172 shar r0
173ashrsi3_4:
174 shar r0
175ashrsi3_3:
176 shar r0
177ashrsi3_2:
178 shar r0
179ashrsi3_1:
180 rts
181 shar r0
182
183ashrsi3_0:
184 rts
185 nop
diff --git a/arch/sh/lib/libgcc.h b/arch/sh/lib/libgcc.h
new file mode 100644
index 000000000000..3f19d1c5d942
--- /dev/null
+++ b/arch/sh/lib/libgcc.h
@@ -0,0 +1,26 @@
1#ifndef __ASM_LIBGCC_H
2#define __ASM_LIBGCC_H
3
4#include <asm/byteorder.h>
5
6typedef int word_type __attribute__ ((mode (__word__)));
7
8#ifdef __BIG_ENDIAN
9struct DWstruct {
10 int high, low;
11};
12#elif defined(__LITTLE_ENDIAN)
13struct DWstruct {
14 int low, high;
15};
16#else
17#error I feel sick.
18#endif
19
20typedef union
21{
22 struct DWstruct s;
23 long long ll;
24} DWunion;
25
26#endif /* __ASM_LIBGCC_H */
diff --git a/arch/sh/lib/lshrdi3.c b/arch/sh/lib/lshrdi3.c
new file mode 100644
index 000000000000..dcf8d6810b7c
--- /dev/null
+++ b/arch/sh/lib/lshrdi3.c
@@ -0,0 +1,29 @@
1#include <linux/module.h>
2
3#include "libgcc.h"
4
5long long __lshrdi3(long long u, word_type b)
6{
7 DWunion uu, w;
8 word_type bm;
9
10 if (b == 0)
11 return u;
12
13 uu.ll = u;
14 bm = 32 - b;
15
16 if (bm <= 0) {
17 w.s.high = 0;
18 w.s.low = (unsigned int) uu.s.high >> -bm;
19 } else {
20 const unsigned int carries = (unsigned int) uu.s.high << bm;
21
22 w.s.high = (unsigned int) uu.s.high >> b;
23 w.s.low = ((unsigned int) uu.s.low >> b) | carries;
24 }
25
26 return w.ll;
27}
28
29EXPORT_SYMBOL(__lshrdi3);
diff --git a/arch/sh/lib/lshrsi3.S b/arch/sh/lib/lshrsi3.S
new file mode 100644
index 000000000000..1e7aaa557130
--- /dev/null
+++ b/arch/sh/lib/lshrsi3.S
@@ -0,0 +1,193 @@
1/* Copyright (C) 1994, 1995, 1997, 1998, 1999, 2000, 2001, 2002, 2003,
2 2004, 2005, 2006
3 Free Software Foundation, Inc.
4
5This file is free software; you can redistribute it and/or modify it
6under the terms of the GNU General Public License as published by the
7Free Software Foundation; either version 2, or (at your option) any
8later version.
9
10In addition to the permissions in the GNU General Public License, the
11Free Software Foundation gives you unlimited permission to link the
12compiled version of this file into combinations with other programs,
13and to distribute those combinations without any restriction coming
14from the use of this file. (The General Public License restrictions
15do apply in other respects; for example, they cover modification of
16the file, and distribution when not linked into a combine
17executable.)
18
19This file is distributed in the hope that it will be useful, but
20WITHOUT ANY WARRANTY; without even the implied warranty of
21MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
22General Public License for more details.
23
24You should have received a copy of the GNU General Public License
25along with this program; see the file COPYING. If not, write to
26the Free Software Foundation, 51 Franklin Street, Fifth Floor,
27Boston, MA 02110-1301, USA. */
28
29!! libgcc routines for the Renesas / SuperH SH CPUs.
30!! Contributed by Steve Chamberlain.
31!! sac@cygnus.com
32
33!! ashiftrt_r4_x, ___ashrsi3, ___ashlsi3, ___lshrsi3 routines
34!! recoded in assembly by Toshiyasu Morita
35!! tm@netcom.com
36
37/* SH2 optimizations for ___ashrsi3, ___ashlsi3, ___lshrsi3 and
38 ELF local label prefixes by J"orn Rennecke
39 amylaar@cygnus.com */
40
41!
42! __lshrsi3
43!
44! Entry:
45!
46! r4: Value to shift
47! r5: Shifts
48!
49! Exit:
50!
51! r0: Result
52!
53! Destroys:
54!
55! (none)
56!
57 .global __lshrsi3
58
59 .align 2
60__lshrsi3:
61 mov #31,r0
62 and r0,r5
63 mova lshrsi3_table,r0
64 mov.b @(r0,r5),r5
65#ifdef __sh1__
66 add r5,r0
67 jmp @r0
68#else
69 braf r5
70#endif
71 mov r4,r0
72
73 .align 2
74lshrsi3_table:
75 .byte lshrsi3_0-lshrsi3_table
76 .byte lshrsi3_1-lshrsi3_table
77 .byte lshrsi3_2-lshrsi3_table
78 .byte lshrsi3_3-lshrsi3_table
79 .byte lshrsi3_4-lshrsi3_table
80 .byte lshrsi3_5-lshrsi3_table
81 .byte lshrsi3_6-lshrsi3_table
82 .byte lshrsi3_7-lshrsi3_table
83 .byte lshrsi3_8-lshrsi3_table
84 .byte lshrsi3_9-lshrsi3_table
85 .byte lshrsi3_10-lshrsi3_table
86 .byte lshrsi3_11-lshrsi3_table
87 .byte lshrsi3_12-lshrsi3_table
88 .byte lshrsi3_13-lshrsi3_table
89 .byte lshrsi3_14-lshrsi3_table
90 .byte lshrsi3_15-lshrsi3_table
91 .byte lshrsi3_16-lshrsi3_table
92 .byte lshrsi3_17-lshrsi3_table
93 .byte lshrsi3_18-lshrsi3_table
94 .byte lshrsi3_19-lshrsi3_table
95 .byte lshrsi3_20-lshrsi3_table
96 .byte lshrsi3_21-lshrsi3_table
97 .byte lshrsi3_22-lshrsi3_table
98 .byte lshrsi3_23-lshrsi3_table
99 .byte lshrsi3_24-lshrsi3_table
100 .byte lshrsi3_25-lshrsi3_table
101 .byte lshrsi3_26-lshrsi3_table
102 .byte lshrsi3_27-lshrsi3_table
103 .byte lshrsi3_28-lshrsi3_table
104 .byte lshrsi3_29-lshrsi3_table
105 .byte lshrsi3_30-lshrsi3_table
106 .byte lshrsi3_31-lshrsi3_table
107
108lshrsi3_6:
109 shlr2 r0
110lshrsi3_4:
111 shlr2 r0
112lshrsi3_2:
113 rts
114 shlr2 r0
115
116lshrsi3_7:
117 shlr2 r0
118lshrsi3_5:
119 shlr2 r0
120lshrsi3_3:
121 shlr2 r0
122lshrsi3_1:
123 rts
124 shlr r0
125
126lshrsi3_14:
127 shlr2 r0
128lshrsi3_12:
129 shlr2 r0
130lshrsi3_10:
131 shlr2 r0
132lshrsi3_8:
133 rts
134 shlr8 r0
135
136lshrsi3_15:
137 shlr2 r0
138lshrsi3_13:
139 shlr2 r0
140lshrsi3_11:
141 shlr2 r0
142lshrsi3_9:
143 shlr8 r0
144 rts
145 shlr r0
146
147lshrsi3_22:
148 shlr2 r0
149lshrsi3_20:
150 shlr2 r0
151lshrsi3_18:
152 shlr2 r0
153lshrsi3_16:
154 rts
155 shlr16 r0
156
157lshrsi3_23:
158 shlr2 r0
159lshrsi3_21:
160 shlr2 r0
161lshrsi3_19:
162 shlr2 r0
163lshrsi3_17:
164 shlr16 r0
165 rts
166 shlr r0
167
168lshrsi3_30:
169 shlr2 r0
170lshrsi3_28:
171 shlr2 r0
172lshrsi3_26:
173 shlr2 r0
174lshrsi3_24:
175 shlr16 r0
176 rts
177 shlr8 r0
178
179lshrsi3_31:
180 shlr2 r0
181lshrsi3_29:
182 shlr2 r0
183lshrsi3_27:
184 shlr2 r0
185lshrsi3_25:
186 shlr16 r0
187 shlr8 r0
188 rts
189 shlr r0
190
191lshrsi3_0:
192 rts
193 nop
diff --git a/arch/sh/lib/movmem.S b/arch/sh/lib/movmem.S
new file mode 100644
index 000000000000..62075f6bc67c
--- /dev/null
+++ b/arch/sh/lib/movmem.S
@@ -0,0 +1,238 @@
1/* Copyright (C) 1994, 1995, 1997, 1998, 1999, 2000, 2001, 2002, 2003,
2 2004, 2005, 2006
3 Free Software Foundation, Inc.
4
5This file is free software; you can redistribute it and/or modify it
6under the terms of the GNU General Public License as published by the
7Free Software Foundation; either version 2, or (at your option) any
8later version.
9
10In addition to the permissions in the GNU General Public License, the
11Free Software Foundation gives you unlimited permission to link the
12compiled version of this file into combinations with other programs,
13and to distribute those combinations without any restriction coming
14from the use of this file. (The General Public License restrictions
15do apply in other respects; for example, they cover modification of
16the file, and distribution when not linked into a combine
17executable.)
18
19This file is distributed in the hope that it will be useful, but
20WITHOUT ANY WARRANTY; without even the implied warranty of
21MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
22General Public License for more details.
23
24You should have received a copy of the GNU General Public License
25along with this program; see the file COPYING. If not, write to
26the Free Software Foundation, 51 Franklin Street, Fifth Floor,
27Boston, MA 02110-1301, USA. */
28
29!! libgcc routines for the Renesas / SuperH SH CPUs.
30!! Contributed by Steve Chamberlain.
31!! sac@cygnus.com
32
33!! ashiftrt_r4_x, ___ashrsi3, ___ashlsi3, ___lshrsi3 routines
34!! recoded in assembly by Toshiyasu Morita
35!! tm@netcom.com
36
37/* SH2 optimizations for ___ashrsi3, ___ashlsi3, ___lshrsi3 and
38 ELF local label prefixes by J"orn Rennecke
39 amylaar@cygnus.com */
40
41 .text
42 .balign 4
43 .global __movmem
44 .global __movstr
45 .set __movstr, __movmem
46 /* This would be a lot simpler if r6 contained the byte count
47 minus 64, and we wouldn't be called here for a byte count of 64. */
48__movmem:
49 sts.l pr,@-r15
50 shll2 r6
51 bsr __movmemSI52+2
52 mov.l @(48,r5),r0
53 .balign 4
54movmem_loop: /* Reached with rts */
55 mov.l @(60,r5),r0
56 add #-64,r6
57 mov.l r0,@(60,r4)
58 tst r6,r6
59 mov.l @(56,r5),r0
60 bt movmem_done
61 mov.l r0,@(56,r4)
62 cmp/pl r6
63 mov.l @(52,r5),r0
64 add #64,r5
65 mov.l r0,@(52,r4)
66 add #64,r4
67 bt __movmemSI52
68! done all the large groups, do the remainder
69! jump to movmem+
70 mova __movmemSI4+4,r0
71 add r6,r0
72 jmp @r0
73movmem_done: ! share slot insn, works out aligned.
74 lds.l @r15+,pr
75 mov.l r0,@(56,r4)
76 mov.l @(52,r5),r0
77 rts
78 mov.l r0,@(52,r4)
79 .balign 4
80
81 .global __movmemSI64
82 .global __movstrSI64
83 .set __movstrSI64, __movmemSI64
84__movmemSI64:
85 mov.l @(60,r5),r0
86 mov.l r0,@(60,r4)
87 .global __movmemSI60
88 .global __movstrSI60
89 .set __movstrSI60, __movmemSI60
90__movmemSI60:
91 mov.l @(56,r5),r0
92 mov.l r0,@(56,r4)
93 .global __movmemSI56
94 .global __movstrSI56
95 .set __movstrSI56, __movmemSI56
96__movmemSI56:
97 mov.l @(52,r5),r0
98 mov.l r0,@(52,r4)
99 .global __movmemSI52
100 .global __movstrSI52
101 .set __movstrSI52, __movmemSI52
102__movmemSI52:
103 mov.l @(48,r5),r0
104 mov.l r0,@(48,r4)
105 .global __movmemSI48
106 .global __movstrSI48
107 .set __movstrSI48, __movmemSI48
108__movmemSI48:
109 mov.l @(44,r5),r0
110 mov.l r0,@(44,r4)
111 .global __movmemSI44
112 .global __movstrSI44
113 .set __movstrSI44, __movmemSI44
114__movmemSI44:
115 mov.l @(40,r5),r0
116 mov.l r0,@(40,r4)
117 .global __movmemSI40
118 .global __movstrSI40
119 .set __movstrSI40, __movmemSI40
120__movmemSI40:
121 mov.l @(36,r5),r0
122 mov.l r0,@(36,r4)
123 .global __movmemSI36
124 .global __movstrSI36
125 .set __movstrSI36, __movmemSI36
126__movmemSI36:
127 mov.l @(32,r5),r0
128 mov.l r0,@(32,r4)
129 .global __movmemSI32
130 .global __movstrSI32
131 .set __movstrSI32, __movmemSI32
132__movmemSI32:
133 mov.l @(28,r5),r0
134 mov.l r0,@(28,r4)
135 .global __movmemSI28
136 .global __movstrSI28
137 .set __movstrSI28, __movmemSI28
138__movmemSI28:
139 mov.l @(24,r5),r0
140 mov.l r0,@(24,r4)
141 .global __movmemSI24
142 .global __movstrSI24
143 .set __movstrSI24, __movmemSI24
144__movmemSI24:
145 mov.l @(20,r5),r0
146 mov.l r0,@(20,r4)
147 .global __movmemSI20
148 .global __movstrSI20
149 .set __movstrSI20, __movmemSI20
150__movmemSI20:
151 mov.l @(16,r5),r0
152 mov.l r0,@(16,r4)
153 .global __movmemSI16
154 .global __movstrSI16
155 .set __movstrSI16, __movmemSI16
156__movmemSI16:
157 mov.l @(12,r5),r0
158 mov.l r0,@(12,r4)
159 .global __movmemSI12
160 .global __movstrSI12
161 .set __movstrSI12, __movmemSI12
162__movmemSI12:
163 mov.l @(8,r5),r0
164 mov.l r0,@(8,r4)
165 .global __movmemSI8
166 .global __movstrSI8
167 .set __movstrSI8, __movmemSI8
168__movmemSI8:
169 mov.l @(4,r5),r0
170 mov.l r0,@(4,r4)
171 .global __movmemSI4
172 .global __movstrSI4
173 .set __movstrSI4, __movmemSI4
174__movmemSI4:
175 mov.l @(0,r5),r0
176 rts
177 mov.l r0,@(0,r4)
178
179 .global __movmem_i4_even
180 .global __movstr_i4_even
181 .set __movstr_i4_even, __movmem_i4_even
182
183 .global __movmem_i4_odd
184 .global __movstr_i4_odd
185 .set __movstr_i4_odd, __movmem_i4_odd
186
187 .global __movmemSI12_i4
188 .global __movstrSI12_i4
189 .set __movstrSI12_i4, __movmemSI12_i4
190
191 .p2align 5
192L_movmem_2mod4_end:
193 mov.l r0,@(16,r4)
194 rts
195 mov.l r1,@(20,r4)
196
197 .p2align 2
198
199__movmem_i4_even:
200 mov.l @r5+,r0
201 bra L_movmem_start_even
202 mov.l @r5+,r1
203
204__movmem_i4_odd:
205 mov.l @r5+,r1
206 add #-4,r4
207 mov.l @r5+,r2
208 mov.l @r5+,r3
209 mov.l r1,@(4,r4)
210 mov.l r2,@(8,r4)
211
212L_movmem_loop:
213 mov.l r3,@(12,r4)
214 dt r6
215 mov.l @r5+,r0
216 bt/s L_movmem_2mod4_end
217 mov.l @r5+,r1
218 add #16,r4
219L_movmem_start_even:
220 mov.l @r5+,r2
221 mov.l @r5+,r3
222 mov.l r0,@r4
223 dt r6
224 mov.l r1,@(4,r4)
225 bf/s L_movmem_loop
226 mov.l r2,@(8,r4)
227 rts
228 mov.l r3,@(12,r4)
229
230 .p2align 4
231__movmemSI12_i4:
232 mov.l @r5,r0
233 mov.l @(4,r5),r1
234 mov.l @(8,r5),r2
235 mov.l r0,@r4
236 mov.l r1,@(4,r4)
237 rts
238 mov.l r2,@(8,r4)
diff --git a/arch/sh/lib/udiv_qrnnd.S b/arch/sh/lib/udiv_qrnnd.S
new file mode 100644
index 000000000000..32b9a36de943
--- /dev/null
+++ b/arch/sh/lib/udiv_qrnnd.S
@@ -0,0 +1,81 @@
1/* Copyright (C) 1994, 1995, 1997, 1998, 1999, 2000, 2001, 2002, 2003,
2 2004, 2005, 2006
3 Free Software Foundation, Inc.
4
5This file is free software; you can redistribute it and/or modify it
6under the terms of the GNU General Public License as published by the
7Free Software Foundation; either version 2, or (at your option) any
8later version.
9
10In addition to the permissions in the GNU General Public License, the
11Free Software Foundation gives you unlimited permission to link the
12compiled version of this file into combinations with other programs,
13and to distribute those combinations without any restriction coming
14from the use of this file. (The General Public License restrictions
15do apply in other respects; for example, they cover modification of
16the file, and distribution when not linked into a combine
17executable.)
18
19This file is distributed in the hope that it will be useful, but
20WITHOUT ANY WARRANTY; without even the implied warranty of
21MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
22General Public License for more details.
23
24You should have received a copy of the GNU General Public License
25along with this program; see the file COPYING. If not, write to
26the Free Software Foundation, 51 Franklin Street, Fifth Floor,
27Boston, MA 02110-1301, USA. */
28
29!! libgcc routines for the Renesas / SuperH SH CPUs.
30!! Contributed by Steve Chamberlain.
31!! sac@cygnus.com
32
33!! ashiftrt_r4_x, ___ashrsi3, ___ashlsi3, ___lshrsi3 routines
34!! recoded in assembly by Toshiyasu Morita
35!! tm@netcom.com
36
37/* SH2 optimizations for ___ashrsi3, ___ashlsi3, ___lshrsi3 and
38 ELF local label prefixes by J"orn Rennecke
39 amylaar@cygnus.com */
40
41 /* r0: rn r1: qn */ /* r0: n1 r4: n0 r5: d r6: d1 */ /* r2: __m */
42 /* n1 < d, but n1 might be larger than d1. */
43 .global __udiv_qrnnd_16
44 .balign 8
45__udiv_qrnnd_16:
46 div0u
47 cmp/hi r6,r0
48 bt .Lots
49 .rept 16
50 div1 r6,r0
51 .endr
52 extu.w r0,r1
53 bt 0f
54 add r6,r0
550: rotcl r1
56 mulu.w r1,r5
57 xtrct r4,r0
58 swap.w r0,r0
59 sts macl,r2
60 cmp/hs r2,r0
61 sub r2,r0
62 bt 0f
63 addc r5,r0
64 add #-1,r1
65 bt 0f
661: add #-1,r1
67 rts
68 add r5,r0
69 .balign 8
70.Lots:
71 sub r5,r0
72 swap.w r4,r1
73 xtrct r0,r1
74 clrt
75 mov r1,r0
76 addc r5,r0
77 mov #-1,r1
78 bf/s 1b
79 shlr16 r1
800: rts
81 nop
diff --git a/arch/sh/lib/udivsi3-Os.S b/arch/sh/lib/udivsi3-Os.S
new file mode 100644
index 000000000000..2bed76587f1e
--- /dev/null
+++ b/arch/sh/lib/udivsi3-Os.S
@@ -0,0 +1,147 @@
1/* Copyright (C) 2006 Free Software Foundation, Inc.
2
3This file is free software; you can redistribute it and/or modify it
4under the terms of the GNU General Public License as published by the
5Free Software Foundation; either version 2, or (at your option) any
6later version.
7
8In addition to the permissions in the GNU General Public License, the
9Free Software Foundation gives you unlimited permission to link the
10compiled version of this file into combinations with other programs,
11and to distribute those combinations without any restriction coming
12from the use of this file. (The General Public License restrictions
13do apply in other respects; for example, they cover modification of
14the file, and distribution when not linked into a combine
15executable.)
16
17This file is distributed in the hope that it will be useful, but
18WITHOUT ANY WARRANTY; without even the implied warranty of
19MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
20General Public License for more details.
21
22You should have received a copy of the GNU General Public License
23along with this program; see the file COPYING. If not, write to
24the Free Software Foundation, 51 Franklin Street, Fifth Floor,
25Boston, MA 02110-1301, USA. */
26
27/* Moderately Space-optimized libgcc routines for the Renesas SH /
28 STMicroelectronics ST40 CPUs.
29 Contributed by J"orn Rennecke joern.rennecke@st.com. */
30
31/* Size: 186 bytes jointly for udivsi3_i4i and sdivsi3_i4i
32 sh4-200 run times:
33 udiv small divisor: 55 cycles
34 udiv large divisor: 52 cycles
35 sdiv small divisor, positive result: 59 cycles
36 sdiv large divisor, positive result: 56 cycles
37 sdiv small divisor, negative result: 65 cycles (*)
38 sdiv large divisor, negative result: 62 cycles (*)
39 (*): r2 is restored in the rts delay slot and has a lingering latency
40 of two more cycles. */
41 .balign 4
42 .global __udivsi3_i4i
43 .global __udivsi3
44 .set __udivsi3, __udivsi3_i4i
45 .type __udivsi3_i4i, @function
46 .type __sdivsi3_i4i, @function
47__udivsi3_i4i:
48 sts pr,r1
49 mov.l r4,@-r15
50 extu.w r5,r0
51 cmp/eq r5,r0
52 swap.w r4,r0
53 shlr16 r4
54 bf/s large_divisor
55 div0u
56 mov.l r5,@-r15
57 shll16 r5
58sdiv_small_divisor:
59 div1 r5,r4
60 bsr div6
61 div1 r5,r4
62 div1 r5,r4
63 bsr div6
64 div1 r5,r4
65 xtrct r4,r0
66 xtrct r0,r4
67 bsr div7
68 swap.w r4,r4
69 div1 r5,r4
70 bsr div7
71 div1 r5,r4
72 xtrct r4,r0
73 mov.l @r15+,r5
74 swap.w r0,r0
75 mov.l @r15+,r4
76 jmp @r1
77 rotcl r0
78div7:
79 div1 r5,r4
80div6:
81 div1 r5,r4; div1 r5,r4; div1 r5,r4
82 div1 r5,r4; div1 r5,r4; rts; div1 r5,r4
83
84divx3:
85 rotcl r0
86 div1 r5,r4
87 rotcl r0
88 div1 r5,r4
89 rotcl r0
90 rts
91 div1 r5,r4
92
93large_divisor:
94 mov.l r5,@-r15
95sdiv_large_divisor:
96 xor r4,r0
97 .rept 4
98 rotcl r0
99 bsr divx3
100 div1 r5,r4
101 .endr
102 mov.l @r15+,r5
103 mov.l @r15+,r4
104 jmp @r1
105 rotcl r0
106
107 .global __sdivsi3_i4i
108 .global __sdivsi3
109 .set __sdivsi3, __sdivsi3_i4i
110__sdivsi3_i4i:
111 mov.l r4,@-r15
112 cmp/pz r5
113 mov.l r5,@-r15
114 bt/s pos_divisor
115 cmp/pz r4
116 neg r5,r5
117 extu.w r5,r0
118 bt/s neg_result
119 cmp/eq r5,r0
120 neg r4,r4
121pos_result:
122 swap.w r4,r0
123 bra sdiv_check_divisor
124 sts pr,r1
125pos_divisor:
126 extu.w r5,r0
127 bt/s pos_result
128 cmp/eq r5,r0
129 neg r4,r4
130neg_result:
131 mova negate_result,r0
132 ;
133 mov r0,r1
134 swap.w r4,r0
135 lds r2,macl
136 sts pr,r2
137sdiv_check_divisor:
138 shlr16 r4
139 bf/s sdiv_large_divisor
140 div0u
141 bra sdiv_small_divisor
142 shll16 r5
143 .balign 4
144negate_result:
145 neg r0,r0
146 jmp @r2
147 sts macl,r2
diff --git a/arch/sh/lib/udivsi3.S b/arch/sh/lib/udivsi3.S
new file mode 100644
index 000000000000..a810fc6f21d5
--- /dev/null
+++ b/arch/sh/lib/udivsi3.S
@@ -0,0 +1,664 @@
1/* Copyright (C) 1994, 1995, 1997, 1998, 1999, 2000, 2001, 2002, 2003,
2 2004, 2005, 2006
3 Free Software Foundation, Inc.
4
5This file is free software; you can redistribute it and/or modify it
6under the terms of the GNU General Public License as published by the
7Free Software Foundation; either version 2, or (at your option) any
8later version.
9
10In addition to the permissions in the GNU General Public License, the
11Free Software Foundation gives you unlimited permission to link the
12compiled version of this file into combinations with other programs,
13and to distribute those combinations without any restriction coming
14from the use of this file. (The General Public License restrictions
15do apply in other respects; for example, they cover modification of
16the file, and distribution when not linked into a combine
17executable.)
18
19This file is distributed in the hope that it will be useful, but
20WITHOUT ANY WARRANTY; without even the implied warranty of
21MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
22General Public License for more details.
23
24You should have received a copy of the GNU General Public License
25along with this program; see the file COPYING. If not, write to
26the Free Software Foundation, 51 Franklin Street, Fifth Floor,
27Boston, MA 02110-1301, USA. */
28
29!! libgcc routines for the Renesas / SuperH SH CPUs.
30!! Contributed by Steve Chamberlain.
31!! sac@cygnus.com
32
33!! ashiftrt_r4_x, ___ashrsi3, ___ashlsi3, ___lshrsi3 routines
34!! recoded in assembly by Toshiyasu Morita
35!! tm@netcom.com
36
37/* SH2 optimizations for ___ashrsi3, ___ashlsi3, ___lshrsi3 and
38 ELF local label prefixes by J"orn Rennecke
39 amylaar@cygnus.com */
40
41/* This code used shld, thus is not suitable for SH1 / SH2. */
42
43/* Signed / unsigned division without use of FPU, optimized for SH4.
44 Uses a lookup table for divisors in the range -128 .. +128, and
45 div1 with case distinction for larger divisors in three more ranges.
46 The code is lumped together with the table to allow the use of mova. */
47#ifdef CONFIG_CPU_LITTLE_ENDIAN
48#define L_LSB 0
49#define L_LSWMSB 1
50#define L_MSWLSB 2
51#else
52#define L_LSB 3
53#define L_LSWMSB 2
54#define L_MSWLSB 1
55#endif
56
57 .balign 4
58 .global __udivsi3_i4i
59 .global __udivsi3
60 .set __udivsi3, __udivsi3_i4i
61 .type __udivsi3_i4i, @function
62__udivsi3_i4i:
63 mov.w c128_w, r1
64 div0u
65 mov r4,r0
66 shlr8 r0
67 cmp/hi r1,r5
68 extu.w r5,r1
69 bf udiv_le128
70 cmp/eq r5,r1
71 bf udiv_ge64k
72 shlr r0
73 mov r5,r1
74 shll16 r5
75 mov.l r4,@-r15
76 div1 r5,r0
77 mov.l r1,@-r15
78 div1 r5,r0
79 div1 r5,r0
80 bra udiv_25
81 div1 r5,r0
82
83div_le128:
84 mova div_table_ix,r0
85 bra div_le128_2
86 mov.b @(r0,r5),r1
87udiv_le128:
88 mov.l r4,@-r15
89 mova div_table_ix,r0
90 mov.b @(r0,r5),r1
91 mov.l r5,@-r15
92div_le128_2:
93 mova div_table_inv,r0
94 mov.l @(r0,r1),r1
95 mov r5,r0
96 tst #0xfe,r0
97 mova div_table_clz,r0
98 dmulu.l r1,r4
99 mov.b @(r0,r5),r1
100 bt/s div_by_1
101 mov r4,r0
102 mov.l @r15+,r5
103 sts mach,r0
104 /* clrt */
105 addc r4,r0
106 mov.l @r15+,r4
107 rotcr r0
108 rts
109 shld r1,r0
110
111div_by_1_neg:
112 neg r4,r0
113div_by_1:
114 mov.l @r15+,r5
115 rts
116 mov.l @r15+,r4
117
118div_ge64k:
119 bt/s div_r8
120 div0u
121 shll8 r5
122 bra div_ge64k_2
123 div1 r5,r0
124udiv_ge64k:
125 cmp/hi r0,r5
126 mov r5,r1
127 bt udiv_r8
128 shll8 r5
129 mov.l r4,@-r15
130 div1 r5,r0
131 mov.l r1,@-r15
132div_ge64k_2:
133 div1 r5,r0
134 mov.l zero_l,r1
135 .rept 4
136 div1 r5,r0
137 .endr
138 mov.l r1,@-r15
139 div1 r5,r0
140 mov.w m256_w,r1
141 div1 r5,r0
142 mov.b r0,@(L_LSWMSB,r15)
143 xor r4,r0
144 and r1,r0
145 bra div_ge64k_end
146 xor r4,r0
147
148div_r8:
149 shll16 r4
150 bra div_r8_2
151 shll8 r4
152udiv_r8:
153 mov.l r4,@-r15
154 shll16 r4
155 clrt
156 shll8 r4
157 mov.l r5,@-r15
158div_r8_2:
159 rotcl r4
160 mov r0,r1
161 div1 r5,r1
162 mov r4,r0
163 rotcl r0
164 mov r5,r4
165 div1 r5,r1
166 .rept 5
167 rotcl r0; div1 r5,r1
168 .endr
169 rotcl r0
170 mov.l @r15+,r5
171 div1 r4,r1
172 mov.l @r15+,r4
173 rts
174 rotcl r0
175
176 .global __sdivsi3_i4i
177 .global __sdivsi3
178 .set __sdivsi3, __sdivsi3_i4i
179 .type __sdivsi3_i4i, @function
180 /* This is link-compatible with a __sdivsi3 call,
181 but we effectively clobber only r1. */
182__sdivsi3_i4i:
183 mov.l r4,@-r15
184 cmp/pz r5
185 mov.w c128_w, r1
186 bt/s pos_divisor
187 cmp/pz r4
188 mov.l r5,@-r15
189 neg r5,r5
190 bt/s neg_result
191 cmp/hi r1,r5
192 neg r4,r4
193pos_result:
194 extu.w r5,r0
195 bf div_le128
196 cmp/eq r5,r0
197 mov r4,r0
198 shlr8 r0
199 bf/s div_ge64k
200 cmp/hi r0,r5
201 div0u
202 shll16 r5
203 div1 r5,r0
204 div1 r5,r0
205 div1 r5,r0
206udiv_25:
207 mov.l zero_l,r1
208 div1 r5,r0
209 div1 r5,r0
210 mov.l r1,@-r15
211 .rept 3
212 div1 r5,r0
213 .endr
214 mov.b r0,@(L_MSWLSB,r15)
215 xtrct r4,r0
216 swap.w r0,r0
217 .rept 8
218 div1 r5,r0
219 .endr
220 mov.b r0,@(L_LSWMSB,r15)
221div_ge64k_end:
222 .rept 8
223 div1 r5,r0
224 .endr
225 mov.l @r15+,r4 ! zero-extension and swap using LS unit.
226 extu.b r0,r0
227 mov.l @r15+,r5
228 or r4,r0
229 mov.l @r15+,r4
230 rts
231 rotcl r0
232
233div_le128_neg:
234 tst #0xfe,r0
235 mova div_table_ix,r0
236 mov.b @(r0,r5),r1
237 mova div_table_inv,r0
238 bt/s div_by_1_neg
239 mov.l @(r0,r1),r1
240 mova div_table_clz,r0
241 dmulu.l r1,r4
242 mov.b @(r0,r5),r1
243 mov.l @r15+,r5
244 sts mach,r0
245 /* clrt */
246 addc r4,r0
247 mov.l @r15+,r4
248 rotcr r0
249 shld r1,r0
250 rts
251 neg r0,r0
252
253pos_divisor:
254 mov.l r5,@-r15
255 bt/s pos_result
256 cmp/hi r1,r5
257 neg r4,r4
258neg_result:
259 extu.w r5,r0
260 bf div_le128_neg
261 cmp/eq r5,r0
262 mov r4,r0
263 shlr8 r0
264 bf/s div_ge64k_neg
265 cmp/hi r0,r5
266 div0u
267 mov.l zero_l,r1
268 shll16 r5
269 div1 r5,r0
270 mov.l r1,@-r15
271 .rept 7
272 div1 r5,r0
273 .endr
274 mov.b r0,@(L_MSWLSB,r15)
275 xtrct r4,r0
276 swap.w r0,r0
277 .rept 8
278 div1 r5,r0
279 .endr
280 mov.b r0,@(L_LSWMSB,r15)
281div_ge64k_neg_end:
282 .rept 8
283 div1 r5,r0
284 .endr
285 mov.l @r15+,r4 ! zero-extension and swap using LS unit.
286 extu.b r0,r1
287 mov.l @r15+,r5
288 or r4,r1
289div_r8_neg_end:
290 mov.l @r15+,r4
291 rotcl r1
292 rts
293 neg r1,r0
294
295div_ge64k_neg:
296 bt/s div_r8_neg
297 div0u
298 shll8 r5
299 mov.l zero_l,r1
300 .rept 6
301 div1 r5,r0
302 .endr
303 mov.l r1,@-r15
304 div1 r5,r0
305 mov.w m256_w,r1
306 div1 r5,r0
307 mov.b r0,@(L_LSWMSB,r15)
308 xor r4,r0
309 and r1,r0
310 bra div_ge64k_neg_end
311 xor r4,r0
312
313c128_w:
314 .word 128
315
316div_r8_neg:
317 clrt
318 shll16 r4
319 mov r4,r1
320 shll8 r1
321 mov r5,r4
322 .rept 7
323 rotcl r1; div1 r5,r0
324 .endr
325 mov.l @r15+,r5
326 rotcl r1
327 bra div_r8_neg_end
328 div1 r4,r0
329
330m256_w:
331 .word 0xff00
332/* This table has been generated by divtab-sh4.c. */
333 .balign 4
334div_table_clz:
335 .byte 0
336 .byte 1
337 .byte 0
338 .byte -1
339 .byte -1
340 .byte -2
341 .byte -2
342 .byte -2
343 .byte -2
344 .byte -3
345 .byte -3
346 .byte -3
347 .byte -3
348 .byte -3
349 .byte -3
350 .byte -3
351 .byte -3
352 .byte -4
353 .byte -4
354 .byte -4
355 .byte -4
356 .byte -4
357 .byte -4
358 .byte -4
359 .byte -4
360 .byte -4
361 .byte -4
362 .byte -4
363 .byte -4
364 .byte -4
365 .byte -4
366 .byte -4
367 .byte -4
368 .byte -5
369 .byte -5
370 .byte -5
371 .byte -5
372 .byte -5
373 .byte -5
374 .byte -5
375 .byte -5
376 .byte -5
377 .byte -5
378 .byte -5
379 .byte -5
380 .byte -5
381 .byte -5
382 .byte -5
383 .byte -5
384 .byte -5
385 .byte -5
386 .byte -5
387 .byte -5
388 .byte -5
389 .byte -5
390 .byte -5
391 .byte -5
392 .byte -5
393 .byte -5
394 .byte -5
395 .byte -5
396 .byte -5
397 .byte -5
398 .byte -5
399 .byte -5
400 .byte -6
401 .byte -6
402 .byte -6
403 .byte -6
404 .byte -6
405 .byte -6
406 .byte -6
407 .byte -6
408 .byte -6
409 .byte -6
410 .byte -6
411 .byte -6
412 .byte -6
413 .byte -6
414 .byte -6
415 .byte -6
416 .byte -6
417 .byte -6
418 .byte -6
419 .byte -6
420 .byte -6
421 .byte -6
422 .byte -6
423 .byte -6
424 .byte -6
425 .byte -6
426 .byte -6
427 .byte -6
428 .byte -6
429 .byte -6
430 .byte -6
431 .byte -6
432 .byte -6
433 .byte -6
434 .byte -6
435 .byte -6
436 .byte -6
437 .byte -6
438 .byte -6
439 .byte -6
440 .byte -6
441 .byte -6
442 .byte -6
443 .byte -6
444 .byte -6
445 .byte -6
446 .byte -6
447 .byte -6
448 .byte -6
449 .byte -6
450 .byte -6
451 .byte -6
452 .byte -6
453 .byte -6
454 .byte -6
455 .byte -6
456 .byte -6
457 .byte -6
458 .byte -6
459 .byte -6
460 .byte -6
461 .byte -6
462 .byte -6
463/* Lookup table translating positive divisor to index into table of
464 normalized inverse. N.B. the '0' entry is also the last entry of the
465 previous table, and causes an unaligned access for division by zero. */
466div_table_ix:
467 .byte -6
468 .byte -128
469 .byte -128
470 .byte 0
471 .byte -128
472 .byte -64
473 .byte 0
474 .byte 64
475 .byte -128
476 .byte -96
477 .byte -64
478 .byte -32
479 .byte 0
480 .byte 32
481 .byte 64
482 .byte 96
483 .byte -128
484 .byte -112
485 .byte -96
486 .byte -80
487 .byte -64
488 .byte -48
489 .byte -32
490 .byte -16
491 .byte 0
492 .byte 16
493 .byte 32
494 .byte 48
495 .byte 64
496 .byte 80
497 .byte 96
498 .byte 112
499 .byte -128
500 .byte -120
501 .byte -112
502 .byte -104
503 .byte -96
504 .byte -88
505 .byte -80
506 .byte -72
507 .byte -64
508 .byte -56
509 .byte -48
510 .byte -40
511 .byte -32
512 .byte -24
513 .byte -16
514 .byte -8
515 .byte 0
516 .byte 8
517 .byte 16
518 .byte 24
519 .byte 32
520 .byte 40
521 .byte 48
522 .byte 56
523 .byte 64
524 .byte 72
525 .byte 80
526 .byte 88
527 .byte 96
528 .byte 104
529 .byte 112
530 .byte 120
531 .byte -128
532 .byte -124
533 .byte -120
534 .byte -116
535 .byte -112
536 .byte -108
537 .byte -104
538 .byte -100
539 .byte -96
540 .byte -92
541 .byte -88
542 .byte -84
543 .byte -80
544 .byte -76
545 .byte -72
546 .byte -68
547 .byte -64
548 .byte -60
549 .byte -56
550 .byte -52
551 .byte -48
552 .byte -44
553 .byte -40
554 .byte -36
555 .byte -32
556 .byte -28
557 .byte -24
558 .byte -20
559 .byte -16
560 .byte -12
561 .byte -8
562 .byte -4
563 .byte 0
564 .byte 4
565 .byte 8
566 .byte 12
567 .byte 16
568 .byte 20
569 .byte 24
570 .byte 28
571 .byte 32
572 .byte 36
573 .byte 40
574 .byte 44
575 .byte 48
576 .byte 52
577 .byte 56
578 .byte 60
579 .byte 64
580 .byte 68
581 .byte 72
582 .byte 76
583 .byte 80
584 .byte 84
585 .byte 88
586 .byte 92
587 .byte 96
588 .byte 100
589 .byte 104
590 .byte 108
591 .byte 112
592 .byte 116
593 .byte 120
594 .byte 124
595 .byte -128
596/* 1/64 .. 1/127, normalized. There is an implicit leading 1 in bit 32. */
597 .balign 4
598zero_l:
599 .long 0x0
600 .long 0xF81F81F9
601 .long 0xF07C1F08
602 .long 0xE9131AC0
603 .long 0xE1E1E1E2
604 .long 0xDAE6076C
605 .long 0xD41D41D5
606 .long 0xCD856891
607 .long 0xC71C71C8
608 .long 0xC0E07039
609 .long 0xBACF914D
610 .long 0xB4E81B4F
611 .long 0xAF286BCB
612 .long 0xA98EF607
613 .long 0xA41A41A5
614 .long 0x9EC8E952
615 .long 0x9999999A
616 .long 0x948B0FCE
617 .long 0x8F9C18FA
618 .long 0x8ACB90F7
619 .long 0x86186187
620 .long 0x81818182
621 .long 0x7D05F418
622 .long 0x78A4C818
623 .long 0x745D1746
624 .long 0x702E05C1
625 .long 0x6C16C16D
626 .long 0x68168169
627 .long 0x642C8591
628 .long 0x60581606
629 .long 0x5C9882BA
630 .long 0x58ED2309
631div_table_inv:
632 .long 0x55555556
633 .long 0x51D07EAF
634 .long 0x4E5E0A73
635 .long 0x4AFD6A06
636 .long 0x47AE147B
637 .long 0x446F8657
638 .long 0x41414142
639 .long 0x3E22CBCF
640 .long 0x3B13B13C
641 .long 0x38138139
642 .long 0x3521CFB3
643 .long 0x323E34A3
644 .long 0x2F684BDB
645 .long 0x2C9FB4D9
646 .long 0x29E4129F
647 .long 0x27350B89
648 .long 0x24924925
649 .long 0x21FB7813
650 .long 0x1F7047DD
651 .long 0x1CF06ADB
652 .long 0x1A7B9612
653 .long 0x18118119
654 .long 0x15B1E5F8
655 .long 0x135C8114
656 .long 0x11111112
657 .long 0xECF56BF
658 .long 0xC9714FC
659 .long 0xA6810A7
660 .long 0x8421085
661 .long 0x624DD30
662 .long 0x4104105
663 .long 0x2040811
664 /* maximum error: 0.987342 scaled: 0.921875*/
generated by cgit v1.2.2 (git 2.25.0) at 2024-12-01 07:38:16 -0500