diff options
Diffstat (limited to 'arch/powerpc/lib')
-rw-r--r-- | arch/powerpc/lib/Makefile | 9 | ||||
-rw-r--r-- | arch/powerpc/lib/checksum.S | 225 | ||||
-rw-r--r-- | arch/powerpc/lib/checksum64.S | 229 | ||||
-rw-r--r-- | arch/powerpc/lib/copy32.S | 543 | ||||
-rw-r--r-- | arch/powerpc/lib/copypage.S | 121 | ||||
-rw-r--r-- | arch/powerpc/lib/copyuser.S | 576 | ||||
-rw-r--r-- | arch/powerpc/lib/div64.S | 58 | ||||
-rw-r--r-- | arch/powerpc/lib/e2a.c | 108 | ||||
-rw-r--r-- | arch/powerpc/lib/memcpy.S | 172 | ||||
-rw-r--r-- | arch/powerpc/lib/rheap.c | 693 | ||||
-rw-r--r-- | arch/powerpc/lib/sstep.c | 141 | ||||
-rw-r--r-- | arch/powerpc/lib/strcase.c | 23 | ||||
-rw-r--r-- | arch/powerpc/lib/string.S | 203 | ||||
-rw-r--r-- | arch/powerpc/lib/usercopy.c | 41 |
14 files changed, 3142 insertions, 0 deletions
diff --git a/arch/powerpc/lib/Makefile b/arch/powerpc/lib/Makefile new file mode 100644 index 000000000000..347f9798e433 --- /dev/null +++ b/arch/powerpc/lib/Makefile | |||
@@ -0,0 +1,9 @@ | |||
1 | # | ||
2 | # Makefile for ppc-specific library files.. | ||
3 | # | ||
4 | |||
5 | obj-y := strcase.o string.o | ||
6 | obj-$(CONFIG_PPC32) += div64.o copy32.o checksum.o | ||
7 | obj-$(CONFIG_PPC64) += copypage.o copyuser.o memcpy.o usercopy.o \ | ||
8 | sstep.o checksum64.o | ||
9 | obj-$(CONFIG_PPC_ISERIES) += e2a.o | ||
diff --git a/arch/powerpc/lib/checksum.S b/arch/powerpc/lib/checksum.S new file mode 100644 index 000000000000..7874e8a80455 --- /dev/null +++ b/arch/powerpc/lib/checksum.S | |||
@@ -0,0 +1,225 @@ | |||
1 | /* | ||
2 | * This file contains assembly-language implementations | ||
3 | * of IP-style 1's complement checksum routines. | ||
4 | * | ||
5 | * Copyright (C) 1995-1996 Gary Thomas (gdt@linuxppc.org) | ||
6 | * | ||
7 | * This program is free software; you can redistribute it and/or | ||
8 | * modify it under the terms of the GNU General Public License | ||
9 | * as published by the Free Software Foundation; either version | ||
10 | * 2 of the License, or (at your option) any later version. | ||
11 | * | ||
12 | * Severely hacked about by Paul Mackerras (paulus@cs.anu.edu.au). | ||
13 | */ | ||
14 | |||
15 | #include <linux/sys.h> | ||
16 | #include <asm/processor.h> | ||
17 | #include <asm/errno.h> | ||
18 | #include <asm/ppc_asm.h> | ||
19 | |||
20 | .text | ||
21 | |||
22 | /* | ||
23 | * ip_fast_csum(buf, len) -- Optimized for IP header | ||
24 | * len is in words and is always >= 5. | ||
25 | */ | ||
26 | _GLOBAL(ip_fast_csum) | ||
27 | lwz r0,0(r3) | ||
28 | lwzu r5,4(r3) | ||
29 | addic. r4,r4,-2 | ||
30 | addc r0,r0,r5 | ||
31 | mtctr r4 | ||
32 | blelr- | ||
33 | 1: lwzu r4,4(r3) | ||
34 | adde r0,r0,r4 | ||
35 | bdnz 1b | ||
36 | addze r0,r0 /* add in final carry */ | ||
37 | rlwinm r3,r0,16,0,31 /* fold two halves together */ | ||
38 | add r3,r0,r3 | ||
39 | not r3,r3 | ||
40 | srwi r3,r3,16 | ||
41 | blr | ||
42 | |||
43 | /* | ||
44 | * Compute checksum of TCP or UDP pseudo-header: | ||
45 | * csum_tcpudp_magic(saddr, daddr, len, proto, sum) | ||
46 | */ | ||
47 | _GLOBAL(csum_tcpudp_magic) | ||
48 | rlwimi r5,r6,16,0,15 /* put proto in upper half of len */ | ||
49 | addc r0,r3,r4 /* add 4 32-bit words together */ | ||
50 | adde r0,r0,r5 | ||
51 | adde r0,r0,r7 | ||
52 | addze r0,r0 /* add in final carry */ | ||
53 | rlwinm r3,r0,16,0,31 /* fold two halves together */ | ||
54 | add r3,r0,r3 | ||
55 | not r3,r3 | ||
56 | srwi r3,r3,16 | ||
57 | blr | ||
58 | |||
59 | /* | ||
60 | * computes the checksum of a memory block at buff, length len, | ||
61 | * and adds in "sum" (32-bit) | ||
62 | * | ||
63 | * csum_partial(buff, len, sum) | ||
64 | */ | ||
65 | _GLOBAL(csum_partial) | ||
66 | addic r0,r5,0 | ||
67 | subi r3,r3,4 | ||
68 | srwi. r6,r4,2 | ||
69 | beq 3f /* if we're doing < 4 bytes */ | ||
70 | andi. r5,r3,2 /* Align buffer to longword boundary */ | ||
71 | beq+ 1f | ||
72 | lhz r5,4(r3) /* do 2 bytes to get aligned */ | ||
73 | addi r3,r3,2 | ||
74 | subi r4,r4,2 | ||
75 | addc r0,r0,r5 | ||
76 | srwi. r6,r4,2 /* # words to do */ | ||
77 | beq 3f | ||
78 | 1: mtctr r6 | ||
79 | 2: lwzu r5,4(r3) /* the bdnz has zero overhead, so it should */ | ||
80 | adde r0,r0,r5 /* be unnecessary to unroll this loop */ | ||
81 | bdnz 2b | ||
82 | andi. r4,r4,3 | ||
83 | 3: cmpwi 0,r4,2 | ||
84 | blt+ 4f | ||
85 | lhz r5,4(r3) | ||
86 | addi r3,r3,2 | ||
87 | subi r4,r4,2 | ||
88 | adde r0,r0,r5 | ||
89 | 4: cmpwi 0,r4,1 | ||
90 | bne+ 5f | ||
91 | lbz r5,4(r3) | ||
92 | slwi r5,r5,8 /* Upper byte of word */ | ||
93 | adde r0,r0,r5 | ||
94 | 5: addze r3,r0 /* add in final carry */ | ||
95 | blr | ||
96 | |||
97 | /* | ||
98 | * Computes the checksum of a memory block at src, length len, | ||
99 | * and adds in "sum" (32-bit), while copying the block to dst. | ||
100 | * If an access exception occurs on src or dst, it stores -EFAULT | ||
101 | * to *src_err or *dst_err respectively, and (for an error on | ||
102 | * src) zeroes the rest of dst. | ||
103 | * | ||
104 | * csum_partial_copy_generic(src, dst, len, sum, src_err, dst_err) | ||
105 | */ | ||
106 | _GLOBAL(csum_partial_copy_generic) | ||
107 | addic r0,r6,0 | ||
108 | subi r3,r3,4 | ||
109 | subi r4,r4,4 | ||
110 | srwi. r6,r5,2 | ||
111 | beq 3f /* if we're doing < 4 bytes */ | ||
112 | andi. r9,r4,2 /* Align dst to longword boundary */ | ||
113 | beq+ 1f | ||
114 | 81: lhz r6,4(r3) /* do 2 bytes to get aligned */ | ||
115 | addi r3,r3,2 | ||
116 | subi r5,r5,2 | ||
117 | 91: sth r6,4(r4) | ||
118 | addi r4,r4,2 | ||
119 | addc r0,r0,r6 | ||
120 | srwi. r6,r5,2 /* # words to do */ | ||
121 | beq 3f | ||
122 | 1: srwi. r6,r5,4 /* # groups of 4 words to do */ | ||
123 | beq 10f | ||
124 | mtctr r6 | ||
125 | 71: lwz r6,4(r3) | ||
126 | 72: lwz r9,8(r3) | ||
127 | 73: lwz r10,12(r3) | ||
128 | 74: lwzu r11,16(r3) | ||
129 | adde r0,r0,r6 | ||
130 | 75: stw r6,4(r4) | ||
131 | adde r0,r0,r9 | ||
132 | 76: stw r9,8(r4) | ||
133 | adde r0,r0,r10 | ||
134 | 77: stw r10,12(r4) | ||
135 | adde r0,r0,r11 | ||
136 | 78: stwu r11,16(r4) | ||
137 | bdnz 71b | ||
138 | 10: rlwinm. r6,r5,30,30,31 /* # words left to do */ | ||
139 | beq 13f | ||
140 | mtctr r6 | ||
141 | 82: lwzu r9,4(r3) | ||
142 | 92: stwu r9,4(r4) | ||
143 | adde r0,r0,r9 | ||
144 | bdnz 82b | ||
145 | 13: andi. r5,r5,3 | ||
146 | 3: cmpwi 0,r5,2 | ||
147 | blt+ 4f | ||
148 | 83: lhz r6,4(r3) | ||
149 | addi r3,r3,2 | ||
150 | subi r5,r5,2 | ||
151 | 93: sth r6,4(r4) | ||
152 | addi r4,r4,2 | ||
153 | adde r0,r0,r6 | ||
154 | 4: cmpwi 0,r5,1 | ||
155 | bne+ 5f | ||
156 | 84: lbz r6,4(r3) | ||
157 | 94: stb r6,4(r4) | ||
158 | slwi r6,r6,8 /* Upper byte of word */ | ||
159 | adde r0,r0,r6 | ||
160 | 5: addze r3,r0 /* add in final carry */ | ||
161 | blr | ||
162 | |||
163 | /* These shouldn't go in the fixup section, since that would | ||
164 | cause the ex_table addresses to get out of order. */ | ||
165 | |||
166 | src_error_4: | ||
167 | mfctr r6 /* update # bytes remaining from ctr */ | ||
168 | rlwimi r5,r6,4,0,27 | ||
169 | b 79f | ||
170 | src_error_1: | ||
171 | li r6,0 | ||
172 | subi r5,r5,2 | ||
173 | 95: sth r6,4(r4) | ||
174 | addi r4,r4,2 | ||
175 | 79: srwi. r6,r5,2 | ||
176 | beq 3f | ||
177 | mtctr r6 | ||
178 | src_error_2: | ||
179 | li r6,0 | ||
180 | 96: stwu r6,4(r4) | ||
181 | bdnz 96b | ||
182 | 3: andi. r5,r5,3 | ||
183 | beq src_error | ||
184 | src_error_3: | ||
185 | li r6,0 | ||
186 | mtctr r5 | ||
187 | addi r4,r4,3 | ||
188 | 97: stbu r6,1(r4) | ||
189 | bdnz 97b | ||
190 | src_error: | ||
191 | cmpwi 0,r7,0 | ||
192 | beq 1f | ||
193 | li r6,-EFAULT | ||
194 | stw r6,0(r7) | ||
195 | 1: addze r3,r0 | ||
196 | blr | ||
197 | |||
198 | dst_error: | ||
199 | cmpwi 0,r8,0 | ||
200 | beq 1f | ||
201 | li r6,-EFAULT | ||
202 | stw r6,0(r8) | ||
203 | 1: addze r3,r0 | ||
204 | blr | ||
205 | |||
206 | .section __ex_table,"a" | ||
207 | .long 81b,src_error_1 | ||
208 | .long 91b,dst_error | ||
209 | .long 71b,src_error_4 | ||
210 | .long 72b,src_error_4 | ||
211 | .long 73b,src_error_4 | ||
212 | .long 74b,src_error_4 | ||
213 | .long 75b,dst_error | ||
214 | .long 76b,dst_error | ||
215 | .long 77b,dst_error | ||
216 | .long 78b,dst_error | ||
217 | .long 82b,src_error_2 | ||
218 | .long 92b,dst_error | ||
219 | .long 83b,src_error_3 | ||
220 | .long 93b,dst_error | ||
221 | .long 84b,src_error_3 | ||
222 | .long 94b,dst_error | ||
223 | .long 95b,dst_error | ||
224 | .long 96b,dst_error | ||
225 | .long 97b,dst_error | ||
diff --git a/arch/powerpc/lib/checksum64.S b/arch/powerpc/lib/checksum64.S new file mode 100644 index 000000000000..ef96c6c58efc --- /dev/null +++ b/arch/powerpc/lib/checksum64.S | |||
@@ -0,0 +1,229 @@ | |||
1 | /* | ||
2 | * This file contains assembly-language implementations | ||
3 | * of IP-style 1's complement checksum routines. | ||
4 | * | ||
5 | * Copyright (C) 1995-1996 Gary Thomas (gdt@linuxppc.org) | ||
6 | * | ||
7 | * This program is free software; you can redistribute it and/or | ||
8 | * modify it under the terms of the GNU General Public License | ||
9 | * as published by the Free Software Foundation; either version | ||
10 | * 2 of the License, or (at your option) any later version. | ||
11 | * | ||
12 | * Severely hacked about by Paul Mackerras (paulus@cs.anu.edu.au). | ||
13 | */ | ||
14 | |||
15 | #include <linux/sys.h> | ||
16 | #include <asm/processor.h> | ||
17 | #include <asm/errno.h> | ||
18 | #include <asm/ppc_asm.h> | ||
19 | |||
20 | /* | ||
21 | * ip_fast_csum(r3=buf, r4=len) -- Optimized for IP header | ||
22 | * len is in words and is always >= 5. | ||
23 | * | ||
24 | * In practice len == 5, but this is not guaranteed. So this code does not | ||
25 | * attempt to use doubleword instructions. | ||
26 | */ | ||
27 | _GLOBAL(ip_fast_csum) | ||
28 | lwz r0,0(r3) | ||
29 | lwzu r5,4(r3) | ||
30 | addic. r4,r4,-2 | ||
31 | addc r0,r0,r5 | ||
32 | mtctr r4 | ||
33 | blelr- | ||
34 | 1: lwzu r4,4(r3) | ||
35 | adde r0,r0,r4 | ||
36 | bdnz 1b | ||
37 | addze r0,r0 /* add in final carry */ | ||
38 | rldicl r4,r0,32,0 /* fold two 32-bit halves together */ | ||
39 | add r0,r0,r4 | ||
40 | srdi r0,r0,32 | ||
41 | rlwinm r3,r0,16,0,31 /* fold two halves together */ | ||
42 | add r3,r0,r3 | ||
43 | not r3,r3 | ||
44 | srwi r3,r3,16 | ||
45 | blr | ||
46 | |||
47 | /* | ||
48 | * Compute checksum of TCP or UDP pseudo-header: | ||
49 | * csum_tcpudp_magic(r3=saddr, r4=daddr, r5=len, r6=proto, r7=sum) | ||
50 | * No real gain trying to do this specially for 64 bit, but | ||
51 | * the 32 bit addition may spill into the upper bits of | ||
52 | * the doubleword so we still must fold it down from 64. | ||
53 | */ | ||
54 | _GLOBAL(csum_tcpudp_magic) | ||
55 | rlwimi r5,r6,16,0,15 /* put proto in upper half of len */ | ||
56 | addc r0,r3,r4 /* add 4 32-bit words together */ | ||
57 | adde r0,r0,r5 | ||
58 | adde r0,r0,r7 | ||
59 | rldicl r4,r0,32,0 /* fold 64 bit value */ | ||
60 | add r0,r4,r0 | ||
61 | srdi r0,r0,32 | ||
62 | rlwinm r3,r0,16,0,31 /* fold two halves together */ | ||
63 | add r3,r0,r3 | ||
64 | not r3,r3 | ||
65 | srwi r3,r3,16 | ||
66 | blr | ||
67 | |||
68 | /* | ||
69 | * Computes the checksum of a memory block at buff, length len, | ||
70 | * and adds in "sum" (32-bit). | ||
71 | * | ||
72 | * This code assumes at least halfword alignment, though the length | ||
73 | * can be any number of bytes. The sum is accumulated in r5. | ||
74 | * | ||
75 | * csum_partial(r3=buff, r4=len, r5=sum) | ||
76 | */ | ||
77 | _GLOBAL(csum_partial) | ||
78 | subi r3,r3,8 /* we'll offset by 8 for the loads */ | ||
79 | srdi. r6,r4,3 /* divide by 8 for doubleword count */ | ||
80 | addic r5,r5,0 /* clear carry */ | ||
81 | beq 3f /* if we're doing < 8 bytes */ | ||
82 | andi. r0,r3,2 /* aligned on a word boundary already? */ | ||
83 | beq+ 1f | ||
84 | lhz r6,8(r3) /* do 2 bytes to get aligned */ | ||
85 | addi r3,r3,2 | ||
86 | subi r4,r4,2 | ||
87 | addc r5,r5,r6 | ||
88 | srdi. r6,r4,3 /* recompute number of doublewords */ | ||
89 | beq 3f /* any left? */ | ||
90 | 1: mtctr r6 | ||
91 | 2: ldu r6,8(r3) /* main sum loop */ | ||
92 | adde r5,r5,r6 | ||
93 | bdnz 2b | ||
94 | andi. r4,r4,7 /* compute bytes left to sum after doublewords */ | ||
95 | 3: cmpwi 0,r4,4 /* is at least a full word left? */ | ||
96 | blt 4f | ||
97 | lwz r6,8(r3) /* sum this word */ | ||
98 | addi r3,r3,4 | ||
99 | subi r4,r4,4 | ||
100 | adde r5,r5,r6 | ||
101 | 4: cmpwi 0,r4,2 /* is at least a halfword left? */ | ||
102 | blt+ 5f | ||
103 | lhz r6,8(r3) /* sum this halfword */ | ||
104 | addi r3,r3,2 | ||
105 | subi r4,r4,2 | ||
106 | adde r5,r5,r6 | ||
107 | 5: cmpwi 0,r4,1 /* is at least a byte left? */ | ||
108 | bne+ 6f | ||
109 | lbz r6,8(r3) /* sum this byte */ | ||
110 | slwi r6,r6,8 /* this byte is assumed to be the upper byte of a halfword */ | ||
111 | adde r5,r5,r6 | ||
112 | 6: addze r5,r5 /* add in final carry */ | ||
113 | rldicl r4,r5,32,0 /* fold two 32-bit halves together */ | ||
114 | add r3,r4,r5 | ||
115 | srdi r3,r3,32 | ||
116 | blr | ||
117 | |||
118 | /* | ||
119 | * Computes the checksum of a memory block at src, length len, | ||
120 | * and adds in "sum" (32-bit), while copying the block to dst. | ||
121 | * If an access exception occurs on src or dst, it stores -EFAULT | ||
122 | * to *src_err or *dst_err respectively, and (for an error on | ||
123 | * src) zeroes the rest of dst. | ||
124 | * | ||
125 | * This code needs to be reworked to take advantage of 64 bit sum+copy. | ||
126 | * However, due to tokenring halfword alignment problems this will be very | ||
127 | * tricky. For now we'll leave it until we instrument it somehow. | ||
128 | * | ||
129 | * csum_partial_copy_generic(r3=src, r4=dst, r5=len, r6=sum, r7=src_err, r8=dst_err) | ||
130 | */ | ||
131 | _GLOBAL(csum_partial_copy_generic) | ||
132 | addic r0,r6,0 | ||
133 | subi r3,r3,4 | ||
134 | subi r4,r4,4 | ||
135 | srwi. r6,r5,2 | ||
136 | beq 3f /* if we're doing < 4 bytes */ | ||
137 | andi. r9,r4,2 /* Align dst to longword boundary */ | ||
138 | beq+ 1f | ||
139 | 81: lhz r6,4(r3) /* do 2 bytes to get aligned */ | ||
140 | addi r3,r3,2 | ||
141 | subi r5,r5,2 | ||
142 | 91: sth r6,4(r4) | ||
143 | addi r4,r4,2 | ||
144 | addc r0,r0,r6 | ||
145 | srwi. r6,r5,2 /* # words to do */ | ||
146 | beq 3f | ||
147 | 1: mtctr r6 | ||
148 | 82: lwzu r6,4(r3) /* the bdnz has zero overhead, so it should */ | ||
149 | 92: stwu r6,4(r4) /* be unnecessary to unroll this loop */ | ||
150 | adde r0,r0,r6 | ||
151 | bdnz 82b | ||
152 | andi. r5,r5,3 | ||
153 | 3: cmpwi 0,r5,2 | ||
154 | blt+ 4f | ||
155 | 83: lhz r6,4(r3) | ||
156 | addi r3,r3,2 | ||
157 | subi r5,r5,2 | ||
158 | 93: sth r6,4(r4) | ||
159 | addi r4,r4,2 | ||
160 | adde r0,r0,r6 | ||
161 | 4: cmpwi 0,r5,1 | ||
162 | bne+ 5f | ||
163 | 84: lbz r6,4(r3) | ||
164 | 94: stb r6,4(r4) | ||
165 | slwi r6,r6,8 /* Upper byte of word */ | ||
166 | adde r0,r0,r6 | ||
167 | 5: addze r3,r0 /* add in final carry (unlikely with 64-bit regs) */ | ||
168 | rldicl r4,r3,32,0 /* fold 64 bit value */ | ||
169 | add r3,r4,r3 | ||
170 | srdi r3,r3,32 | ||
171 | blr | ||
172 | |||
173 | /* These shouldn't go in the fixup section, since that would | ||
174 | cause the ex_table addresses to get out of order. */ | ||
175 | |||
176 | .globl src_error_1 | ||
177 | src_error_1: | ||
178 | li r6,0 | ||
179 | subi r5,r5,2 | ||
180 | 95: sth r6,4(r4) | ||
181 | addi r4,r4,2 | ||
182 | srwi. r6,r5,2 | ||
183 | beq 3f | ||
184 | mtctr r6 | ||
185 | .globl src_error_2 | ||
186 | src_error_2: | ||
187 | li r6,0 | ||
188 | 96: stwu r6,4(r4) | ||
189 | bdnz 96b | ||
190 | 3: andi. r5,r5,3 | ||
191 | beq src_error | ||
192 | .globl src_error_3 | ||
193 | src_error_3: | ||
194 | li r6,0 | ||
195 | mtctr r5 | ||
196 | addi r4,r4,3 | ||
197 | 97: stbu r6,1(r4) | ||
198 | bdnz 97b | ||
199 | .globl src_error | ||
200 | src_error: | ||
201 | cmpdi 0,r7,0 | ||
202 | beq 1f | ||
203 | li r6,-EFAULT | ||
204 | stw r6,0(r7) | ||
205 | 1: addze r3,r0 | ||
206 | blr | ||
207 | |||
208 | .globl dst_error | ||
209 | dst_error: | ||
210 | cmpdi 0,r8,0 | ||
211 | beq 1f | ||
212 | li r6,-EFAULT | ||
213 | stw r6,0(r8) | ||
214 | 1: addze r3,r0 | ||
215 | blr | ||
216 | |||
217 | .section __ex_table,"a" | ||
218 | .align 3 | ||
219 | .llong 81b,src_error_1 | ||
220 | .llong 91b,dst_error | ||
221 | .llong 82b,src_error_2 | ||
222 | .llong 92b,dst_error | ||
223 | .llong 83b,src_error_3 | ||
224 | .llong 93b,dst_error | ||
225 | .llong 84b,src_error_3 | ||
226 | .llong 94b,dst_error | ||
227 | .llong 95b,dst_error | ||
228 | .llong 96b,dst_error | ||
229 | .llong 97b,dst_error | ||
diff --git a/arch/powerpc/lib/copy32.S b/arch/powerpc/lib/copy32.S new file mode 100644 index 000000000000..420a912198a2 --- /dev/null +++ b/arch/powerpc/lib/copy32.S | |||
@@ -0,0 +1,543 @@ | |||
1 | /* | ||
2 | * Memory copy functions for 32-bit PowerPC. | ||
3 | * | ||
4 | * Copyright (C) 1996-2005 Paul Mackerras. | ||
5 | * | ||
6 | * This program is free software; you can redistribute it and/or | ||
7 | * modify it under the terms of the GNU General Public License | ||
8 | * as published by the Free Software Foundation; either version | ||
9 | * 2 of the License, or (at your option) any later version. | ||
10 | */ | ||
11 | #include <linux/config.h> | ||
12 | #include <asm/processor.h> | ||
13 | #include <asm/cache.h> | ||
14 | #include <asm/errno.h> | ||
15 | #include <asm/ppc_asm.h> | ||
16 | |||
17 | #define COPY_16_BYTES \ | ||
18 | lwz r7,4(r4); \ | ||
19 | lwz r8,8(r4); \ | ||
20 | lwz r9,12(r4); \ | ||
21 | lwzu r10,16(r4); \ | ||
22 | stw r7,4(r6); \ | ||
23 | stw r8,8(r6); \ | ||
24 | stw r9,12(r6); \ | ||
25 | stwu r10,16(r6) | ||
26 | |||
27 | #define COPY_16_BYTES_WITHEX(n) \ | ||
28 | 8 ## n ## 0: \ | ||
29 | lwz r7,4(r4); \ | ||
30 | 8 ## n ## 1: \ | ||
31 | lwz r8,8(r4); \ | ||
32 | 8 ## n ## 2: \ | ||
33 | lwz r9,12(r4); \ | ||
34 | 8 ## n ## 3: \ | ||
35 | lwzu r10,16(r4); \ | ||
36 | 8 ## n ## 4: \ | ||
37 | stw r7,4(r6); \ | ||
38 | 8 ## n ## 5: \ | ||
39 | stw r8,8(r6); \ | ||
40 | 8 ## n ## 6: \ | ||
41 | stw r9,12(r6); \ | ||
42 | 8 ## n ## 7: \ | ||
43 | stwu r10,16(r6) | ||
44 | |||
45 | #define COPY_16_BYTES_EXCODE(n) \ | ||
46 | 9 ## n ## 0: \ | ||
47 | addi r5,r5,-(16 * n); \ | ||
48 | b 104f; \ | ||
49 | 9 ## n ## 1: \ | ||
50 | addi r5,r5,-(16 * n); \ | ||
51 | b 105f; \ | ||
52 | .section __ex_table,"a"; \ | ||
53 | .align 2; \ | ||
54 | .long 8 ## n ## 0b,9 ## n ## 0b; \ | ||
55 | .long 8 ## n ## 1b,9 ## n ## 0b; \ | ||
56 | .long 8 ## n ## 2b,9 ## n ## 0b; \ | ||
57 | .long 8 ## n ## 3b,9 ## n ## 0b; \ | ||
58 | .long 8 ## n ## 4b,9 ## n ## 1b; \ | ||
59 | .long 8 ## n ## 5b,9 ## n ## 1b; \ | ||
60 | .long 8 ## n ## 6b,9 ## n ## 1b; \ | ||
61 | .long 8 ## n ## 7b,9 ## n ## 1b; \ | ||
62 | .text | ||
63 | |||
64 | .text | ||
65 | .stabs "arch/powerpc/lib/",N_SO,0,0,0f | ||
66 | .stabs "copy32.S",N_SO,0,0,0f | ||
67 | 0: | ||
68 | |||
69 | CACHELINE_BYTES = L1_CACHE_LINE_SIZE | ||
70 | LG_CACHELINE_BYTES = LG_L1_CACHE_LINE_SIZE | ||
71 | CACHELINE_MASK = (L1_CACHE_LINE_SIZE-1) | ||
72 | |||
73 | /* | ||
74 | * Use dcbz on the complete cache lines in the destination | ||
75 | * to set them to zero. This requires that the destination | ||
76 | * area is cacheable. -- paulus | ||
77 | */ | ||
78 | _GLOBAL(cacheable_memzero) | ||
79 | mr r5,r4 | ||
80 | li r4,0 | ||
81 | addi r6,r3,-4 | ||
82 | cmplwi 0,r5,4 | ||
83 | blt 7f | ||
84 | stwu r4,4(r6) | ||
85 | beqlr | ||
86 | andi. r0,r6,3 | ||
87 | add r5,r0,r5 | ||
88 | subf r6,r0,r6 | ||
89 | clrlwi r7,r6,32-LG_CACHELINE_BYTES | ||
90 | add r8,r7,r5 | ||
91 | srwi r9,r8,LG_CACHELINE_BYTES | ||
92 | addic. r9,r9,-1 /* total number of complete cachelines */ | ||
93 | ble 2f | ||
94 | xori r0,r7,CACHELINE_MASK & ~3 | ||
95 | srwi. r0,r0,2 | ||
96 | beq 3f | ||
97 | mtctr r0 | ||
98 | 4: stwu r4,4(r6) | ||
99 | bdnz 4b | ||
100 | 3: mtctr r9 | ||
101 | li r7,4 | ||
102 | #if !defined(CONFIG_8xx) | ||
103 | 10: dcbz r7,r6 | ||
104 | #else | ||
105 | 10: stw r4, 4(r6) | ||
106 | stw r4, 8(r6) | ||
107 | stw r4, 12(r6) | ||
108 | stw r4, 16(r6) | ||
109 | #if CACHE_LINE_SIZE >= 32 | ||
110 | stw r4, 20(r6) | ||
111 | stw r4, 24(r6) | ||
112 | stw r4, 28(r6) | ||
113 | stw r4, 32(r6) | ||
114 | #endif /* CACHE_LINE_SIZE */ | ||
115 | #endif | ||
116 | addi r6,r6,CACHELINE_BYTES | ||
117 | bdnz 10b | ||
118 | clrlwi r5,r8,32-LG_CACHELINE_BYTES | ||
119 | addi r5,r5,4 | ||
120 | 2: srwi r0,r5,2 | ||
121 | mtctr r0 | ||
122 | bdz 6f | ||
123 | 1: stwu r4,4(r6) | ||
124 | bdnz 1b | ||
125 | 6: andi. r5,r5,3 | ||
126 | 7: cmpwi 0,r5,0 | ||
127 | beqlr | ||
128 | mtctr r5 | ||
129 | addi r6,r6,3 | ||
130 | 8: stbu r4,1(r6) | ||
131 | bdnz 8b | ||
132 | blr | ||
133 | |||
134 | _GLOBAL(memset) | ||
135 | rlwimi r4,r4,8,16,23 | ||
136 | rlwimi r4,r4,16,0,15 | ||
137 | addi r6,r3,-4 | ||
138 | cmplwi 0,r5,4 | ||
139 | blt 7f | ||
140 | stwu r4,4(r6) | ||
141 | beqlr | ||
142 | andi. r0,r6,3 | ||
143 | add r5,r0,r5 | ||
144 | subf r6,r0,r6 | ||
145 | srwi r0,r5,2 | ||
146 | mtctr r0 | ||
147 | bdz 6f | ||
148 | 1: stwu r4,4(r6) | ||
149 | bdnz 1b | ||
150 | 6: andi. r5,r5,3 | ||
151 | 7: cmpwi 0,r5,0 | ||
152 | beqlr | ||
153 | mtctr r5 | ||
154 | addi r6,r6,3 | ||
155 | 8: stbu r4,1(r6) | ||
156 | bdnz 8b | ||
157 | blr | ||
158 | |||
159 | /* | ||
160 | * This version uses dcbz on the complete cache lines in the | ||
161 | * destination area to reduce memory traffic. This requires that | ||
162 | * the destination area is cacheable. | ||
163 | * We only use this version if the source and dest don't overlap. | ||
164 | * -- paulus. | ||
165 | */ | ||
166 | _GLOBAL(cacheable_memcpy) | ||
167 | add r7,r3,r5 /* test if the src & dst overlap */ | ||
168 | add r8,r4,r5 | ||
169 | cmplw 0,r4,r7 | ||
170 | cmplw 1,r3,r8 | ||
171 | crand 0,0,4 /* cr0.lt &= cr1.lt */ | ||
172 | blt memcpy /* if regions overlap */ | ||
173 | |||
174 | addi r4,r4,-4 | ||
175 | addi r6,r3,-4 | ||
176 | neg r0,r3 | ||
177 | andi. r0,r0,CACHELINE_MASK /* # bytes to start of cache line */ | ||
178 | beq 58f | ||
179 | |||
180 | cmplw 0,r5,r0 /* is this more than total to do? */ | ||
181 | blt 63f /* if not much to do */ | ||
182 | andi. r8,r0,3 /* get it word-aligned first */ | ||
183 | subf r5,r0,r5 | ||
184 | mtctr r8 | ||
185 | beq+ 61f | ||
186 | 70: lbz r9,4(r4) /* do some bytes */ | ||
187 | stb r9,4(r6) | ||
188 | addi r4,r4,1 | ||
189 | addi r6,r6,1 | ||
190 | bdnz 70b | ||
191 | 61: srwi. r0,r0,2 | ||
192 | mtctr r0 | ||
193 | beq 58f | ||
194 | 72: lwzu r9,4(r4) /* do some words */ | ||
195 | stwu r9,4(r6) | ||
196 | bdnz 72b | ||
197 | |||
198 | 58: srwi. r0,r5,LG_CACHELINE_BYTES /* # complete cachelines */ | ||
199 | clrlwi r5,r5,32-LG_CACHELINE_BYTES | ||
200 | li r11,4 | ||
201 | mtctr r0 | ||
202 | beq 63f | ||
203 | 53: | ||
204 | #if !defined(CONFIG_8xx) | ||
205 | dcbz r11,r6 | ||
206 | #endif | ||
207 | COPY_16_BYTES | ||
208 | #if L1_CACHE_LINE_SIZE >= 32 | ||
209 | COPY_16_BYTES | ||
210 | #if L1_CACHE_LINE_SIZE >= 64 | ||
211 | COPY_16_BYTES | ||
212 | COPY_16_BYTES | ||
213 | #if L1_CACHE_LINE_SIZE >= 128 | ||
214 | COPY_16_BYTES | ||
215 | COPY_16_BYTES | ||
216 | COPY_16_BYTES | ||
217 | COPY_16_BYTES | ||
218 | #endif | ||
219 | #endif | ||
220 | #endif | ||
221 | bdnz 53b | ||
222 | |||
223 | 63: srwi. r0,r5,2 | ||
224 | mtctr r0 | ||
225 | beq 64f | ||
226 | 30: lwzu r0,4(r4) | ||
227 | stwu r0,4(r6) | ||
228 | bdnz 30b | ||
229 | |||
230 | 64: andi. r0,r5,3 | ||
231 | mtctr r0 | ||
232 | beq+ 65f | ||
233 | 40: lbz r0,4(r4) | ||
234 | stb r0,4(r6) | ||
235 | addi r4,r4,1 | ||
236 | addi r6,r6,1 | ||
237 | bdnz 40b | ||
238 | 65: blr | ||
239 | |||
240 | _GLOBAL(memmove) | ||
241 | cmplw 0,r3,r4 | ||
242 | bgt backwards_memcpy | ||
243 | /* fall through */ | ||
244 | |||
245 | _GLOBAL(memcpy) | ||
246 | srwi. r7,r5,3 | ||
247 | addi r6,r3,-4 | ||
248 | addi r4,r4,-4 | ||
249 | beq 2f /* if less than 8 bytes to do */ | ||
250 | andi. r0,r6,3 /* get dest word aligned */ | ||
251 | mtctr r7 | ||
252 | bne 5f | ||
253 | 1: lwz r7,4(r4) | ||
254 | lwzu r8,8(r4) | ||
255 | stw r7,4(r6) | ||
256 | stwu r8,8(r6) | ||
257 | bdnz 1b | ||
258 | andi. r5,r5,7 | ||
259 | 2: cmplwi 0,r5,4 | ||
260 | blt 3f | ||
261 | lwzu r0,4(r4) | ||
262 | addi r5,r5,-4 | ||
263 | stwu r0,4(r6) | ||
264 | 3: cmpwi 0,r5,0 | ||
265 | beqlr | ||
266 | mtctr r5 | ||
267 | addi r4,r4,3 | ||
268 | addi r6,r6,3 | ||
269 | 4: lbzu r0,1(r4) | ||
270 | stbu r0,1(r6) | ||
271 | bdnz 4b | ||
272 | blr | ||
273 | 5: subfic r0,r0,4 | ||
274 | mtctr r0 | ||
275 | 6: lbz r7,4(r4) | ||
276 | addi r4,r4,1 | ||
277 | stb r7,4(r6) | ||
278 | addi r6,r6,1 | ||
279 | bdnz 6b | ||
280 | subf r5,r0,r5 | ||
281 | rlwinm. r7,r5,32-3,3,31 | ||
282 | beq 2b | ||
283 | mtctr r7 | ||
284 | b 1b | ||
285 | |||
286 | _GLOBAL(backwards_memcpy) | ||
287 | rlwinm. r7,r5,32-3,3,31 /* r0 = r5 >> 3 */ | ||
288 | add r6,r3,r5 | ||
289 | add r4,r4,r5 | ||
290 | beq 2f | ||
291 | andi. r0,r6,3 | ||
292 | mtctr r7 | ||
293 | bne 5f | ||
294 | 1: lwz r7,-4(r4) | ||
295 | lwzu r8,-8(r4) | ||
296 | stw r7,-4(r6) | ||
297 | stwu r8,-8(r6) | ||
298 | bdnz 1b | ||
299 | andi. r5,r5,7 | ||
300 | 2: cmplwi 0,r5,4 | ||
301 | blt 3f | ||
302 | lwzu r0,-4(r4) | ||
303 | subi r5,r5,4 | ||
304 | stwu r0,-4(r6) | ||
305 | 3: cmpwi 0,r5,0 | ||
306 | beqlr | ||
307 | mtctr r5 | ||
308 | 4: lbzu r0,-1(r4) | ||
309 | stbu r0,-1(r6) | ||
310 | bdnz 4b | ||
311 | blr | ||
312 | 5: mtctr r0 | ||
313 | 6: lbzu r7,-1(r4) | ||
314 | stbu r7,-1(r6) | ||
315 | bdnz 6b | ||
316 | subf r5,r0,r5 | ||
317 | rlwinm. r7,r5,32-3,3,31 | ||
318 | beq 2b | ||
319 | mtctr r7 | ||
320 | b 1b | ||
321 | |||
322 | _GLOBAL(__copy_tofrom_user) | ||
323 | addi r4,r4,-4 | ||
324 | addi r6,r3,-4 | ||
325 | neg r0,r3 | ||
326 | andi. r0,r0,CACHELINE_MASK /* # bytes to start of cache line */ | ||
327 | beq 58f | ||
328 | |||
329 | cmplw 0,r5,r0 /* is this more than total to do? */ | ||
330 | blt 63f /* if not much to do */ | ||
331 | andi. r8,r0,3 /* get it word-aligned first */ | ||
332 | mtctr r8 | ||
333 | beq+ 61f | ||
334 | 70: lbz r9,4(r4) /* do some bytes */ | ||
335 | 71: stb r9,4(r6) | ||
336 | addi r4,r4,1 | ||
337 | addi r6,r6,1 | ||
338 | bdnz 70b | ||
339 | 61: subf r5,r0,r5 | ||
340 | srwi. r0,r0,2 | ||
341 | mtctr r0 | ||
342 | beq 58f | ||
343 | 72: lwzu r9,4(r4) /* do some words */ | ||
344 | 73: stwu r9,4(r6) | ||
345 | bdnz 72b | ||
346 | |||
347 | .section __ex_table,"a" | ||
348 | .align 2 | ||
349 | .long 70b,100f | ||
350 | .long 71b,101f | ||
351 | .long 72b,102f | ||
352 | .long 73b,103f | ||
353 | .text | ||
354 | |||
355 | 58: srwi. r0,r5,LG_CACHELINE_BYTES /* # complete cachelines */ | ||
356 | clrlwi r5,r5,32-LG_CACHELINE_BYTES | ||
357 | li r11,4 | ||
358 | beq 63f | ||
359 | |||
360 | #ifdef CONFIG_8xx | ||
361 | /* Don't use prefetch on 8xx */ | ||
362 | mtctr r0 | ||
363 | li r0,0 | ||
364 | 53: COPY_16_BYTES_WITHEX(0) | ||
365 | bdnz 53b | ||
366 | |||
367 | #else /* not CONFIG_8xx */ | ||
368 | /* Here we decide how far ahead to prefetch the source */ | ||
369 | li r3,4 | ||
370 | cmpwi r0,1 | ||
371 | li r7,0 | ||
372 | ble 114f | ||
373 | li r7,1 | ||
374 | #if MAX_COPY_PREFETCH > 1 | ||
375 | /* Heuristically, for large transfers we prefetch | ||
376 | MAX_COPY_PREFETCH cachelines ahead. For small transfers | ||
377 | we prefetch 1 cacheline ahead. */ | ||
378 | cmpwi r0,MAX_COPY_PREFETCH | ||
379 | ble 112f | ||
380 | li r7,MAX_COPY_PREFETCH | ||
381 | 112: mtctr r7 | ||
382 | 111: dcbt r3,r4 | ||
383 | addi r3,r3,CACHELINE_BYTES | ||
384 | bdnz 111b | ||
385 | #else | ||
386 | dcbt r3,r4 | ||
387 | addi r3,r3,CACHELINE_BYTES | ||
388 | #endif /* MAX_COPY_PREFETCH > 1 */ | ||
389 | |||
390 | 114: subf r8,r7,r0 | ||
391 | mr r0,r7 | ||
392 | mtctr r8 | ||
393 | |||
394 | 53: dcbt r3,r4 | ||
395 | 54: dcbz r11,r6 | ||
396 | .section __ex_table,"a" | ||
397 | .align 2 | ||
398 | .long 54b,105f | ||
399 | .text | ||
400 | /* the main body of the cacheline loop */ | ||
401 | COPY_16_BYTES_WITHEX(0) | ||
402 | #if L1_CACHE_LINE_SIZE >= 32 | ||
403 | COPY_16_BYTES_WITHEX(1) | ||
404 | #if L1_CACHE_LINE_SIZE >= 64 | ||
405 | COPY_16_BYTES_WITHEX(2) | ||
406 | COPY_16_BYTES_WITHEX(3) | ||
407 | #if L1_CACHE_LINE_SIZE >= 128 | ||
408 | COPY_16_BYTES_WITHEX(4) | ||
409 | COPY_16_BYTES_WITHEX(5) | ||
410 | COPY_16_BYTES_WITHEX(6) | ||
411 | COPY_16_BYTES_WITHEX(7) | ||
412 | #endif | ||
413 | #endif | ||
414 | #endif | ||
415 | bdnz 53b | ||
416 | cmpwi r0,0 | ||
417 | li r3,4 | ||
418 | li r7,0 | ||
419 | bne 114b | ||
420 | #endif /* CONFIG_8xx */ | ||
421 | |||
422 | 63: srwi. r0,r5,2 | ||
423 | mtctr r0 | ||
424 | beq 64f | ||
425 | 30: lwzu r0,4(r4) | ||
426 | 31: stwu r0,4(r6) | ||
427 | bdnz 30b | ||
428 | |||
429 | 64: andi. r0,r5,3 | ||
430 | mtctr r0 | ||
431 | beq+ 65f | ||
432 | 40: lbz r0,4(r4) | ||
433 | 41: stb r0,4(r6) | ||
434 | addi r4,r4,1 | ||
435 | addi r6,r6,1 | ||
436 | bdnz 40b | ||
437 | 65: li r3,0 | ||
438 | blr | ||
439 | |||
440 | /* read fault, initial single-byte copy */ | ||
441 | 100: li r9,0 | ||
442 | b 90f | ||
443 | /* write fault, initial single-byte copy */ | ||
444 | 101: li r9,1 | ||
445 | 90: subf r5,r8,r5 | ||
446 | li r3,0 | ||
447 | b 99f | ||
448 | /* read fault, initial word copy */ | ||
449 | 102: li r9,0 | ||
450 | b 91f | ||
451 | /* write fault, initial word copy */ | ||
452 | 103: li r9,1 | ||
453 | 91: li r3,2 | ||
454 | b 99f | ||
455 | |||
456 | /* | ||
457 | * this stuff handles faults in the cacheline loop and branches to either | ||
458 | * 104f (if in read part) or 105f (if in write part), after updating r5 | ||
459 | */ | ||
460 | COPY_16_BYTES_EXCODE(0) | ||
461 | #if L1_CACHE_LINE_SIZE >= 32 | ||
462 | COPY_16_BYTES_EXCODE(1) | ||
463 | #if L1_CACHE_LINE_SIZE >= 64 | ||
464 | COPY_16_BYTES_EXCODE(2) | ||
465 | COPY_16_BYTES_EXCODE(3) | ||
466 | #if L1_CACHE_LINE_SIZE >= 128 | ||
467 | COPY_16_BYTES_EXCODE(4) | ||
468 | COPY_16_BYTES_EXCODE(5) | ||
469 | COPY_16_BYTES_EXCODE(6) | ||
470 | COPY_16_BYTES_EXCODE(7) | ||
471 | #endif | ||
472 | #endif | ||
473 | #endif | ||
474 | |||
475 | /* read fault in cacheline loop */ | ||
476 | 104: li r9,0 | ||
477 | b 92f | ||
478 | /* fault on dcbz (effectively a write fault) */ | ||
479 | /* or write fault in cacheline loop */ | ||
480 | 105: li r9,1 | ||
481 | 92: li r3,LG_CACHELINE_BYTES | ||
482 | mfctr r8 | ||
483 | add r0,r0,r8 | ||
484 | b 106f | ||
485 | /* read fault in final word loop */ | ||
486 | 108: li r9,0 | ||
487 | b 93f | ||
488 | /* write fault in final word loop */ | ||
489 | 109: li r9,1 | ||
490 | 93: andi. r5,r5,3 | ||
491 | li r3,2 | ||
492 | b 99f | ||
493 | /* read fault in final byte loop */ | ||
494 | 110: li r9,0 | ||
495 | b 94f | ||
496 | /* write fault in final byte loop */ | ||
497 | 111: li r9,1 | ||
498 | 94: li r5,0 | ||
499 | li r3,0 | ||
500 | /* | ||
501 | * At this stage the number of bytes not copied is | ||
502 | * r5 + (ctr << r3), and r9 is 0 for read or 1 for write. | ||
503 | */ | ||
504 | 99: mfctr r0 | ||
505 | 106: slw r3,r0,r3 | ||
506 | add. r3,r3,r5 | ||
507 | beq 120f /* shouldn't happen */ | ||
508 | cmpwi 0,r9,0 | ||
509 | bne 120f | ||
510 | /* for a read fault, first try to continue the copy one byte at a time */ | ||
511 | mtctr r3 | ||
512 | 130: lbz r0,4(r4) | ||
513 | 131: stb r0,4(r6) | ||
514 | addi r4,r4,1 | ||
515 | addi r6,r6,1 | ||
516 | bdnz 130b | ||
517 | /* then clear out the destination: r3 bytes starting at 4(r6) */ | ||
518 | 132: mfctr r3 | ||
519 | srwi. r0,r3,2 | ||
520 | li r9,0 | ||
521 | mtctr r0 | ||
522 | beq 113f | ||
523 | 112: stwu r9,4(r6) | ||
524 | bdnz 112b | ||
525 | 113: andi. r0,r3,3 | ||
526 | mtctr r0 | ||
527 | beq 120f | ||
528 | 114: stb r9,4(r6) | ||
529 | addi r6,r6,1 | ||
530 | bdnz 114b | ||
531 | 120: blr | ||
532 | |||
533 | .section __ex_table,"a" | ||
534 | .align 2 | ||
535 | .long 30b,108b | ||
536 | .long 31b,109b | ||
537 | .long 40b,110b | ||
538 | .long 41b,111b | ||
539 | .long 130b,132b | ||
540 | .long 131b,120b | ||
541 | .long 112b,120b | ||
542 | .long 114b,120b | ||
543 | .text | ||
diff --git a/arch/powerpc/lib/copypage.S b/arch/powerpc/lib/copypage.S new file mode 100644 index 000000000000..733d61618bbf --- /dev/null +++ b/arch/powerpc/lib/copypage.S | |||
@@ -0,0 +1,121 @@ | |||
1 | /* | ||
2 | * arch/ppc64/lib/copypage.S | ||
3 | * | ||
4 | * Copyright (C) 2002 Paul Mackerras, IBM Corp. | ||
5 | * | ||
6 | * This program is free software; you can redistribute it and/or | ||
7 | * modify it under the terms of the GNU General Public License | ||
8 | * as published by the Free Software Foundation; either version | ||
9 | * 2 of the License, or (at your option) any later version. | ||
10 | */ | ||
11 | #include <asm/processor.h> | ||
12 | #include <asm/ppc_asm.h> | ||
13 | |||
14 | _GLOBAL(copy_page) | ||
15 | std r31,-8(1) | ||
16 | std r30,-16(1) | ||
17 | std r29,-24(1) | ||
18 | std r28,-32(1) | ||
19 | std r27,-40(1) | ||
20 | std r26,-48(1) | ||
21 | std r25,-56(1) | ||
22 | std r24,-64(1) | ||
23 | std r23,-72(1) | ||
24 | std r22,-80(1) | ||
25 | std r21,-88(1) | ||
26 | std r20,-96(1) | ||
27 | li r5,4096/32 - 1 | ||
28 | addi r3,r3,-8 | ||
29 | li r12,5 | ||
30 | 0: addi r5,r5,-24 | ||
31 | mtctr r12 | ||
32 | ld r22,640(4) | ||
33 | ld r21,512(4) | ||
34 | ld r20,384(4) | ||
35 | ld r11,256(4) | ||
36 | ld r9,128(4) | ||
37 | ld r7,0(4) | ||
38 | ld r25,648(4) | ||
39 | ld r24,520(4) | ||
40 | ld r23,392(4) | ||
41 | ld r10,264(4) | ||
42 | ld r8,136(4) | ||
43 | ldu r6,8(4) | ||
44 | cmpwi r5,24 | ||
45 | 1: std r22,648(3) | ||
46 | std r21,520(3) | ||
47 | std r20,392(3) | ||
48 | std r11,264(3) | ||
49 | std r9,136(3) | ||
50 | std r7,8(3) | ||
51 | ld r28,648(4) | ||
52 | ld r27,520(4) | ||
53 | ld r26,392(4) | ||
54 | ld r31,264(4) | ||
55 | ld r30,136(4) | ||
56 | ld r29,8(4) | ||
57 | std r25,656(3) | ||
58 | std r24,528(3) | ||
59 | std r23,400(3) | ||
60 | std r10,272(3) | ||
61 | std r8,144(3) | ||
62 | std r6,16(3) | ||
63 | ld r22,656(4) | ||
64 | ld r21,528(4) | ||
65 | ld r20,400(4) | ||
66 | ld r11,272(4) | ||
67 | ld r9,144(4) | ||
68 | ld r7,16(4) | ||
69 | std r28,664(3) | ||
70 | std r27,536(3) | ||
71 | std r26,408(3) | ||
72 | std r31,280(3) | ||
73 | std r30,152(3) | ||
74 | stdu r29,24(3) | ||
75 | ld r25,664(4) | ||
76 | ld r24,536(4) | ||
77 | ld r23,408(4) | ||
78 | ld r10,280(4) | ||
79 | ld r8,152(4) | ||
80 | ldu r6,24(4) | ||
81 | bdnz 1b | ||
82 | std r22,648(3) | ||
83 | std r21,520(3) | ||
84 | std r20,392(3) | ||
85 | std r11,264(3) | ||
86 | std r9,136(3) | ||
87 | std r7,8(3) | ||
88 | addi r4,r4,640 | ||
89 | addi r3,r3,648 | ||
90 | bge 0b | ||
91 | mtctr r5 | ||
92 | ld r7,0(4) | ||
93 | ld r8,8(4) | ||
94 | ldu r9,16(4) | ||
95 | 3: ld r10,8(4) | ||
96 | std r7,8(3) | ||
97 | ld r7,16(4) | ||
98 | std r8,16(3) | ||
99 | ld r8,24(4) | ||
100 | std r9,24(3) | ||
101 | ldu r9,32(4) | ||
102 | stdu r10,32(3) | ||
103 | bdnz 3b | ||
104 | 4: ld r10,8(4) | ||
105 | std r7,8(3) | ||
106 | std r8,16(3) | ||
107 | std r9,24(3) | ||
108 | std r10,32(3) | ||
109 | 9: ld r20,-96(1) | ||
110 | ld r21,-88(1) | ||
111 | ld r22,-80(1) | ||
112 | ld r23,-72(1) | ||
113 | ld r24,-64(1) | ||
114 | ld r25,-56(1) | ||
115 | ld r26,-48(1) | ||
116 | ld r27,-40(1) | ||
117 | ld r28,-32(1) | ||
118 | ld r29,-24(1) | ||
119 | ld r30,-16(1) | ||
120 | ld r31,-8(1) | ||
121 | blr | ||
diff --git a/arch/powerpc/lib/copyuser.S b/arch/powerpc/lib/copyuser.S new file mode 100644 index 000000000000..a0b3fbbd6fb1 --- /dev/null +++ b/arch/powerpc/lib/copyuser.S | |||
@@ -0,0 +1,576 @@ | |||
1 | /* | ||
2 | * arch/ppc64/lib/copyuser.S | ||
3 | * | ||
4 | * Copyright (C) 2002 Paul Mackerras, IBM Corp. | ||
5 | * | ||
6 | * This program is free software; you can redistribute it and/or | ||
7 | * modify it under the terms of the GNU General Public License | ||
8 | * as published by the Free Software Foundation; either version | ||
9 | * 2 of the License, or (at your option) any later version. | ||
10 | */ | ||
11 | #include <asm/processor.h> | ||
12 | #include <asm/ppc_asm.h> | ||
13 | |||
14 | .align 7 | ||
15 | _GLOBAL(__copy_tofrom_user) | ||
16 | /* first check for a whole page copy on a page boundary */ | ||
17 | cmpldi cr1,r5,16 | ||
18 | cmpdi cr6,r5,4096 | ||
19 | or r0,r3,r4 | ||
20 | neg r6,r3 /* LS 3 bits = # bytes to 8-byte dest bdry */ | ||
21 | andi. r0,r0,4095 | ||
22 | std r3,-24(r1) | ||
23 | crand cr0*4+2,cr0*4+2,cr6*4+2 | ||
24 | std r4,-16(r1) | ||
25 | std r5,-8(r1) | ||
26 | dcbt 0,r4 | ||
27 | beq .Lcopy_page | ||
28 | andi. r6,r6,7 | ||
29 | mtcrf 0x01,r5 | ||
30 | blt cr1,.Lshort_copy | ||
31 | bne .Ldst_unaligned | ||
32 | .Ldst_aligned: | ||
33 | andi. r0,r4,7 | ||
34 | addi r3,r3,-16 | ||
35 | bne .Lsrc_unaligned | ||
36 | srdi r7,r5,4 | ||
37 | 20: ld r9,0(r4) | ||
38 | addi r4,r4,-8 | ||
39 | mtctr r7 | ||
40 | andi. r5,r5,7 | ||
41 | bf cr7*4+0,22f | ||
42 | addi r3,r3,8 | ||
43 | addi r4,r4,8 | ||
44 | mr r8,r9 | ||
45 | blt cr1,72f | ||
46 | 21: ld r9,8(r4) | ||
47 | 70: std r8,8(r3) | ||
48 | 22: ldu r8,16(r4) | ||
49 | 71: stdu r9,16(r3) | ||
50 | bdnz 21b | ||
51 | 72: std r8,8(r3) | ||
52 | beq+ 3f | ||
53 | addi r3,r3,16 | ||
54 | 23: ld r9,8(r4) | ||
55 | .Ldo_tail: | ||
56 | bf cr7*4+1,1f | ||
57 | rotldi r9,r9,32 | ||
58 | 73: stw r9,0(r3) | ||
59 | addi r3,r3,4 | ||
60 | 1: bf cr7*4+2,2f | ||
61 | rotldi r9,r9,16 | ||
62 | 74: sth r9,0(r3) | ||
63 | addi r3,r3,2 | ||
64 | 2: bf cr7*4+3,3f | ||
65 | rotldi r9,r9,8 | ||
66 | 75: stb r9,0(r3) | ||
67 | 3: li r3,0 | ||
68 | blr | ||
69 | |||
70 | .Lsrc_unaligned: | ||
71 | srdi r6,r5,3 | ||
72 | addi r5,r5,-16 | ||
73 | subf r4,r0,r4 | ||
74 | srdi r7,r5,4 | ||
75 | sldi r10,r0,3 | ||
76 | cmpldi cr6,r6,3 | ||
77 | andi. r5,r5,7 | ||
78 | mtctr r7 | ||
79 | subfic r11,r10,64 | ||
80 | add r5,r5,r0 | ||
81 | bt cr7*4+0,28f | ||
82 | |||
83 | 24: ld r9,0(r4) /* 3+2n loads, 2+2n stores */ | ||
84 | 25: ld r0,8(r4) | ||
85 | sld r6,r9,r10 | ||
86 | 26: ldu r9,16(r4) | ||
87 | srd r7,r0,r11 | ||
88 | sld r8,r0,r10 | ||
89 | or r7,r7,r6 | ||
90 | blt cr6,79f | ||
91 | 27: ld r0,8(r4) | ||
92 | b 2f | ||
93 | |||
94 | 28: ld r0,0(r4) /* 4+2n loads, 3+2n stores */ | ||
95 | 29: ldu r9,8(r4) | ||
96 | sld r8,r0,r10 | ||
97 | addi r3,r3,-8 | ||
98 | blt cr6,5f | ||
99 | 30: ld r0,8(r4) | ||
100 | srd r12,r9,r11 | ||
101 | sld r6,r9,r10 | ||
102 | 31: ldu r9,16(r4) | ||
103 | or r12,r8,r12 | ||
104 | srd r7,r0,r11 | ||
105 | sld r8,r0,r10 | ||
106 | addi r3,r3,16 | ||
107 | beq cr6,78f | ||
108 | |||
109 | 1: or r7,r7,r6 | ||
110 | 32: ld r0,8(r4) | ||
111 | 76: std r12,8(r3) | ||
112 | 2: srd r12,r9,r11 | ||
113 | sld r6,r9,r10 | ||
114 | 33: ldu r9,16(r4) | ||
115 | or r12,r8,r12 | ||
116 | 77: stdu r7,16(r3) | ||
117 | srd r7,r0,r11 | ||
118 | sld r8,r0,r10 | ||
119 | bdnz 1b | ||
120 | |||
121 | 78: std r12,8(r3) | ||
122 | or r7,r7,r6 | ||
123 | 79: std r7,16(r3) | ||
124 | 5: srd r12,r9,r11 | ||
125 | or r12,r8,r12 | ||
126 | 80: std r12,24(r3) | ||
127 | bne 6f | ||
128 | li r3,0 | ||
129 | blr | ||
130 | 6: cmpwi cr1,r5,8 | ||
131 | addi r3,r3,32 | ||
132 | sld r9,r9,r10 | ||
133 | ble cr1,.Ldo_tail | ||
134 | 34: ld r0,8(r4) | ||
135 | srd r7,r0,r11 | ||
136 | or r9,r7,r9 | ||
137 | b .Ldo_tail | ||
138 | |||
139 | .Ldst_unaligned: | ||
140 | mtcrf 0x01,r6 /* put #bytes to 8B bdry into cr7 */ | ||
141 | subf r5,r6,r5 | ||
142 | li r7,0 | ||
143 | cmpldi r1,r5,16 | ||
144 | bf cr7*4+3,1f | ||
145 | 35: lbz r0,0(r4) | ||
146 | 81: stb r0,0(r3) | ||
147 | addi r7,r7,1 | ||
148 | 1: bf cr7*4+2,2f | ||
149 | 36: lhzx r0,r7,r4 | ||
150 | 82: sthx r0,r7,r3 | ||
151 | addi r7,r7,2 | ||
152 | 2: bf cr7*4+1,3f | ||
153 | 37: lwzx r0,r7,r4 | ||
154 | 83: stwx r0,r7,r3 | ||
155 | 3: mtcrf 0x01,r5 | ||
156 | add r4,r6,r4 | ||
157 | add r3,r6,r3 | ||
158 | b .Ldst_aligned | ||
159 | |||
160 | .Lshort_copy: | ||
161 | bf cr7*4+0,1f | ||
162 | 38: lwz r0,0(r4) | ||
163 | 39: lwz r9,4(r4) | ||
164 | addi r4,r4,8 | ||
165 | 84: stw r0,0(r3) | ||
166 | 85: stw r9,4(r3) | ||
167 | addi r3,r3,8 | ||
168 | 1: bf cr7*4+1,2f | ||
169 | 40: lwz r0,0(r4) | ||
170 | addi r4,r4,4 | ||
171 | 86: stw r0,0(r3) | ||
172 | addi r3,r3,4 | ||
173 | 2: bf cr7*4+2,3f | ||
174 | 41: lhz r0,0(r4) | ||
175 | addi r4,r4,2 | ||
176 | 87: sth r0,0(r3) | ||
177 | addi r3,r3,2 | ||
178 | 3: bf cr7*4+3,4f | ||
179 | 42: lbz r0,0(r4) | ||
180 | 88: stb r0,0(r3) | ||
181 | 4: li r3,0 | ||
182 | blr | ||
183 | |||
184 | /* | ||
185 | * exception handlers follow | ||
186 | * we have to return the number of bytes not copied | ||
187 | * for an exception on a load, we set the rest of the destination to 0 | ||
188 | */ | ||
189 | |||
190 | 136: | ||
191 | 137: | ||
192 | add r3,r3,r7 | ||
193 | b 1f | ||
194 | 130: | ||
195 | 131: | ||
196 | addi r3,r3,8 | ||
197 | 120: | ||
198 | 122: | ||
199 | 124: | ||
200 | 125: | ||
201 | 126: | ||
202 | 127: | ||
203 | 128: | ||
204 | 129: | ||
205 | 133: | ||
206 | addi r3,r3,8 | ||
207 | 121: | ||
208 | 132: | ||
209 | addi r3,r3,8 | ||
210 | 123: | ||
211 | 134: | ||
212 | 135: | ||
213 | 138: | ||
214 | 139: | ||
215 | 140: | ||
216 | 141: | ||
217 | 142: | ||
218 | |||
219 | /* | ||
220 | * here we have had a fault on a load and r3 points to the first | ||
221 | * unmodified byte of the destination | ||
222 | */ | ||
223 | 1: ld r6,-24(r1) | ||
224 | ld r4,-16(r1) | ||
225 | ld r5,-8(r1) | ||
226 | subf r6,r6,r3 | ||
227 | add r4,r4,r6 | ||
228 | subf r5,r6,r5 /* #bytes left to go */ | ||
229 | |||
230 | /* | ||
231 | * first see if we can copy any more bytes before hitting another exception | ||
232 | */ | ||
233 | mtctr r5 | ||
234 | 43: lbz r0,0(r4) | ||
235 | addi r4,r4,1 | ||
236 | 89: stb r0,0(r3) | ||
237 | addi r3,r3,1 | ||
238 | bdnz 43b | ||
239 | li r3,0 /* huh? all copied successfully this time? */ | ||
240 | blr | ||
241 | |||
242 | /* | ||
243 | * here we have trapped again, need to clear ctr bytes starting at r3 | ||
244 | */ | ||
245 | 143: mfctr r5 | ||
246 | li r0,0 | ||
247 | mr r4,r3 | ||
248 | mr r3,r5 /* return the number of bytes not copied */ | ||
249 | 1: andi. r9,r4,7 | ||
250 | beq 3f | ||
251 | 90: stb r0,0(r4) | ||
252 | addic. r5,r5,-1 | ||
253 | addi r4,r4,1 | ||
254 | bne 1b | ||
255 | blr | ||
256 | 3: cmpldi cr1,r5,8 | ||
257 | srdi r9,r5,3 | ||
258 | andi. r5,r5,7 | ||
259 | blt cr1,93f | ||
260 | mtctr r9 | ||
261 | 91: std r0,0(r4) | ||
262 | addi r4,r4,8 | ||
263 | bdnz 91b | ||
264 | 93: beqlr | ||
265 | mtctr r5 | ||
266 | 92: stb r0,0(r4) | ||
267 | addi r4,r4,1 | ||
268 | bdnz 92b | ||
269 | blr | ||
270 | |||
271 | /* | ||
272 | * exception handlers for stores: we just need to work | ||
273 | * out how many bytes weren't copied | ||
274 | */ | ||
275 | 182: | ||
276 | 183: | ||
277 | add r3,r3,r7 | ||
278 | b 1f | ||
279 | 180: | ||
280 | addi r3,r3,8 | ||
281 | 171: | ||
282 | 177: | ||
283 | addi r3,r3,8 | ||
284 | 170: | ||
285 | 172: | ||
286 | 176: | ||
287 | 178: | ||
288 | addi r3,r3,4 | ||
289 | 185: | ||
290 | addi r3,r3,4 | ||
291 | 173: | ||
292 | 174: | ||
293 | 175: | ||
294 | 179: | ||
295 | 181: | ||
296 | 184: | ||
297 | 186: | ||
298 | 187: | ||
299 | 188: | ||
300 | 189: | ||
301 | 1: | ||
302 | ld r6,-24(r1) | ||
303 | ld r5,-8(r1) | ||
304 | add r6,r6,r5 | ||
305 | subf r3,r3,r6 /* #bytes not copied */ | ||
306 | 190: | ||
307 | 191: | ||
308 | 192: | ||
309 | blr /* #bytes not copied in r3 */ | ||
310 | |||
311 | .section __ex_table,"a" | ||
312 | .align 3 | ||
313 | .llong 20b,120b | ||
314 | .llong 21b,121b | ||
315 | .llong 70b,170b | ||
316 | .llong 22b,122b | ||
317 | .llong 71b,171b | ||
318 | .llong 72b,172b | ||
319 | .llong 23b,123b | ||
320 | .llong 73b,173b | ||
321 | .llong 74b,174b | ||
322 | .llong 75b,175b | ||
323 | .llong 24b,124b | ||
324 | .llong 25b,125b | ||
325 | .llong 26b,126b | ||
326 | .llong 27b,127b | ||
327 | .llong 28b,128b | ||
328 | .llong 29b,129b | ||
329 | .llong 30b,130b | ||
330 | .llong 31b,131b | ||
331 | .llong 32b,132b | ||
332 | .llong 76b,176b | ||
333 | .llong 33b,133b | ||
334 | .llong 77b,177b | ||
335 | .llong 78b,178b | ||
336 | .llong 79b,179b | ||
337 | .llong 80b,180b | ||
338 | .llong 34b,134b | ||
339 | .llong 35b,135b | ||
340 | .llong 81b,181b | ||
341 | .llong 36b,136b | ||
342 | .llong 82b,182b | ||
343 | .llong 37b,137b | ||
344 | .llong 83b,183b | ||
345 | .llong 38b,138b | ||
346 | .llong 39b,139b | ||
347 | .llong 84b,184b | ||
348 | .llong 85b,185b | ||
349 | .llong 40b,140b | ||
350 | .llong 86b,186b | ||
351 | .llong 41b,141b | ||
352 | .llong 87b,187b | ||
353 | .llong 42b,142b | ||
354 | .llong 88b,188b | ||
355 | .llong 43b,143b | ||
356 | .llong 89b,189b | ||
357 | .llong 90b,190b | ||
358 | .llong 91b,191b | ||
359 | .llong 92b,192b | ||
360 | |||
361 | .text | ||
362 | |||
363 | /* | ||
364 | * Routine to copy a whole page of data, optimized for POWER4. | ||
365 | * On POWER4 it is more than 50% faster than the simple loop | ||
366 | * above (following the .Ldst_aligned label) but it runs slightly | ||
367 | * slower on POWER3. | ||
368 | */ | ||
369 | .Lcopy_page: | ||
370 | std r31,-32(1) | ||
371 | std r30,-40(1) | ||
372 | std r29,-48(1) | ||
373 | std r28,-56(1) | ||
374 | std r27,-64(1) | ||
375 | std r26,-72(1) | ||
376 | std r25,-80(1) | ||
377 | std r24,-88(1) | ||
378 | std r23,-96(1) | ||
379 | std r22,-104(1) | ||
380 | std r21,-112(1) | ||
381 | std r20,-120(1) | ||
382 | li r5,4096/32 - 1 | ||
383 | addi r3,r3,-8 | ||
384 | li r0,5 | ||
385 | 0: addi r5,r5,-24 | ||
386 | mtctr r0 | ||
387 | 20: ld r22,640(4) | ||
388 | 21: ld r21,512(4) | ||
389 | 22: ld r20,384(4) | ||
390 | 23: ld r11,256(4) | ||
391 | 24: ld r9,128(4) | ||
392 | 25: ld r7,0(4) | ||
393 | 26: ld r25,648(4) | ||
394 | 27: ld r24,520(4) | ||
395 | 28: ld r23,392(4) | ||
396 | 29: ld r10,264(4) | ||
397 | 30: ld r8,136(4) | ||
398 | 31: ldu r6,8(4) | ||
399 | cmpwi r5,24 | ||
400 | 1: | ||
401 | 32: std r22,648(3) | ||
402 | 33: std r21,520(3) | ||
403 | 34: std r20,392(3) | ||
404 | 35: std r11,264(3) | ||
405 | 36: std r9,136(3) | ||
406 | 37: std r7,8(3) | ||
407 | 38: ld r28,648(4) | ||
408 | 39: ld r27,520(4) | ||
409 | 40: ld r26,392(4) | ||
410 | 41: ld r31,264(4) | ||
411 | 42: ld r30,136(4) | ||
412 | 43: ld r29,8(4) | ||
413 | 44: std r25,656(3) | ||
414 | 45: std r24,528(3) | ||
415 | 46: std r23,400(3) | ||
416 | 47: std r10,272(3) | ||
417 | 48: std r8,144(3) | ||
418 | 49: std r6,16(3) | ||
419 | 50: ld r22,656(4) | ||
420 | 51: ld r21,528(4) | ||
421 | 52: ld r20,400(4) | ||
422 | 53: ld r11,272(4) | ||
423 | 54: ld r9,144(4) | ||
424 | 55: ld r7,16(4) | ||
425 | 56: std r28,664(3) | ||
426 | 57: std r27,536(3) | ||
427 | 58: std r26,408(3) | ||
428 | 59: std r31,280(3) | ||
429 | 60: std r30,152(3) | ||
430 | 61: stdu r29,24(3) | ||
431 | 62: ld r25,664(4) | ||
432 | 63: ld r24,536(4) | ||
433 | 64: ld r23,408(4) | ||
434 | 65: ld r10,280(4) | ||
435 | 66: ld r8,152(4) | ||
436 | 67: ldu r6,24(4) | ||
437 | bdnz 1b | ||
438 | 68: std r22,648(3) | ||
439 | 69: std r21,520(3) | ||
440 | 70: std r20,392(3) | ||
441 | 71: std r11,264(3) | ||
442 | 72: std r9,136(3) | ||
443 | 73: std r7,8(3) | ||
444 | 74: addi r4,r4,640 | ||
445 | 75: addi r3,r3,648 | ||
446 | bge 0b | ||
447 | mtctr r5 | ||
448 | 76: ld r7,0(4) | ||
449 | 77: ld r8,8(4) | ||
450 | 78: ldu r9,16(4) | ||
451 | 3: | ||
452 | 79: ld r10,8(4) | ||
453 | 80: std r7,8(3) | ||
454 | 81: ld r7,16(4) | ||
455 | 82: std r8,16(3) | ||
456 | 83: ld r8,24(4) | ||
457 | 84: std r9,24(3) | ||
458 | 85: ldu r9,32(4) | ||
459 | 86: stdu r10,32(3) | ||
460 | bdnz 3b | ||
461 | 4: | ||
462 | 87: ld r10,8(4) | ||
463 | 88: std r7,8(3) | ||
464 | 89: std r8,16(3) | ||
465 | 90: std r9,24(3) | ||
466 | 91: std r10,32(3) | ||
467 | 9: ld r20,-120(1) | ||
468 | ld r21,-112(1) | ||
469 | ld r22,-104(1) | ||
470 | ld r23,-96(1) | ||
471 | ld r24,-88(1) | ||
472 | ld r25,-80(1) | ||
473 | ld r26,-72(1) | ||
474 | ld r27,-64(1) | ||
475 | ld r28,-56(1) | ||
476 | ld r29,-48(1) | ||
477 | ld r30,-40(1) | ||
478 | ld r31,-32(1) | ||
479 | li r3,0 | ||
480 | blr | ||
481 | |||
482 | /* | ||
483 | * on an exception, reset to the beginning and jump back into the | ||
484 | * standard __copy_tofrom_user | ||
485 | */ | ||
486 | 100: ld r20,-120(1) | ||
487 | ld r21,-112(1) | ||
488 | ld r22,-104(1) | ||
489 | ld r23,-96(1) | ||
490 | ld r24,-88(1) | ||
491 | ld r25,-80(1) | ||
492 | ld r26,-72(1) | ||
493 | ld r27,-64(1) | ||
494 | ld r28,-56(1) | ||
495 | ld r29,-48(1) | ||
496 | ld r30,-40(1) | ||
497 | ld r31,-32(1) | ||
498 | ld r3,-24(r1) | ||
499 | ld r4,-16(r1) | ||
500 | li r5,4096 | ||
501 | b .Ldst_aligned | ||
502 | |||
503 | .section __ex_table,"a" | ||
504 | .align 3 | ||
505 | .llong 20b,100b | ||
506 | .llong 21b,100b | ||
507 | .llong 22b,100b | ||
508 | .llong 23b,100b | ||
509 | .llong 24b,100b | ||
510 | .llong 25b,100b | ||
511 | .llong 26b,100b | ||
512 | .llong 27b,100b | ||
513 | .llong 28b,100b | ||
514 | .llong 29b,100b | ||
515 | .llong 30b,100b | ||
516 | .llong 31b,100b | ||
517 | .llong 32b,100b | ||
518 | .llong 33b,100b | ||
519 | .llong 34b,100b | ||
520 | .llong 35b,100b | ||
521 | .llong 36b,100b | ||
522 | .llong 37b,100b | ||
523 | .llong 38b,100b | ||
524 | .llong 39b,100b | ||
525 | .llong 40b,100b | ||
526 | .llong 41b,100b | ||
527 | .llong 42b,100b | ||
528 | .llong 43b,100b | ||
529 | .llong 44b,100b | ||
530 | .llong 45b,100b | ||
531 | .llong 46b,100b | ||
532 | .llong 47b,100b | ||
533 | .llong 48b,100b | ||
534 | .llong 49b,100b | ||
535 | .llong 50b,100b | ||
536 | .llong 51b,100b | ||
537 | .llong 52b,100b | ||
538 | .llong 53b,100b | ||
539 | .llong 54b,100b | ||
540 | .llong 55b,100b | ||
541 | .llong 56b,100b | ||
542 | .llong 57b,100b | ||
543 | .llong 58b,100b | ||
544 | .llong 59b,100b | ||
545 | .llong 60b,100b | ||
546 | .llong 61b,100b | ||
547 | .llong 62b,100b | ||
548 | .llong 63b,100b | ||
549 | .llong 64b,100b | ||
550 | .llong 65b,100b | ||
551 | .llong 66b,100b | ||
552 | .llong 67b,100b | ||
553 | .llong 68b,100b | ||
554 | .llong 69b,100b | ||
555 | .llong 70b,100b | ||
556 | .llong 71b,100b | ||
557 | .llong 72b,100b | ||
558 | .llong 73b,100b | ||
559 | .llong 74b,100b | ||
560 | .llong 75b,100b | ||
561 | .llong 76b,100b | ||
562 | .llong 77b,100b | ||
563 | .llong 78b,100b | ||
564 | .llong 79b,100b | ||
565 | .llong 80b,100b | ||
566 | .llong 81b,100b | ||
567 | .llong 82b,100b | ||
568 | .llong 83b,100b | ||
569 | .llong 84b,100b | ||
570 | .llong 85b,100b | ||
571 | .llong 86b,100b | ||
572 | .llong 87b,100b | ||
573 | .llong 88b,100b | ||
574 | .llong 89b,100b | ||
575 | .llong 90b,100b | ||
576 | .llong 91b,100b | ||
diff --git a/arch/powerpc/lib/div64.S b/arch/powerpc/lib/div64.S new file mode 100644 index 000000000000..3527569e9926 --- /dev/null +++ b/arch/powerpc/lib/div64.S | |||
@@ -0,0 +1,58 @@ | |||
1 | /* | ||
2 | * Divide a 64-bit unsigned number by a 32-bit unsigned number. | ||
3 | * This routine assumes that the top 32 bits of the dividend are | ||
4 | * non-zero to start with. | ||
5 | * On entry, r3 points to the dividend, which get overwritten with | ||
6 | * the 64-bit quotient, and r4 contains the divisor. | ||
7 | * On exit, r3 contains the remainder. | ||
8 | * | ||
9 | * Copyright (C) 2002 Paul Mackerras, IBM Corp. | ||
10 | * | ||
11 | * This program is free software; you can redistribute it and/or | ||
12 | * modify it under the terms of the GNU General Public License | ||
13 | * as published by the Free Software Foundation; either version | ||
14 | * 2 of the License, or (at your option) any later version. | ||
15 | */ | ||
16 | #include <asm/ppc_asm.h> | ||
17 | #include <asm/processor.h> | ||
18 | |||
19 | _GLOBAL(__div64_32) | ||
20 | lwz r5,0(r3) # get the dividend into r5/r6 | ||
21 | lwz r6,4(r3) | ||
22 | cmplw r5,r4 | ||
23 | li r7,0 | ||
24 | li r8,0 | ||
25 | blt 1f | ||
26 | divwu r7,r5,r4 # if dividend.hi >= divisor, | ||
27 | mullw r0,r7,r4 # quotient.hi = dividend.hi / divisor | ||
28 | subf. r5,r0,r5 # dividend.hi %= divisor | ||
29 | beq 3f | ||
30 | 1: mr r11,r5 # here dividend.hi != 0 | ||
31 | andis. r0,r5,0xc000 | ||
32 | bne 2f | ||
33 | cntlzw r0,r5 # we are shifting the dividend right | ||
34 | li r10,-1 # to make it < 2^32, and shifting | ||
35 | srw r10,r10,r0 # the divisor right the same amount, | ||
36 | add r9,r4,r10 # rounding up (so the estimate cannot | ||
37 | andc r11,r6,r10 # ever be too large, only too small) | ||
38 | andc r9,r9,r10 | ||
39 | or r11,r5,r11 | ||
40 | rotlw r9,r9,r0 | ||
41 | rotlw r11,r11,r0 | ||
42 | divwu r11,r11,r9 # then we divide the shifted quantities | ||
43 | 2: mullw r10,r11,r4 # to get an estimate of the quotient, | ||
44 | mulhwu r9,r11,r4 # multiply the estimate by the divisor, | ||
45 | subfc r6,r10,r6 # take the product from the divisor, | ||
46 | add r8,r8,r11 # and add the estimate to the accumulated | ||
47 | subfe. r5,r9,r5 # quotient | ||
48 | bne 1b | ||
49 | 3: cmplw r6,r4 | ||
50 | blt 4f | ||
51 | divwu r0,r6,r4 # perform the remaining 32-bit division | ||
52 | mullw r10,r0,r4 # and get the remainder | ||
53 | add r8,r8,r0 | ||
54 | subf r6,r10,r6 | ||
55 | 4: stw r7,0(r3) # return the quotient in *r3 | ||
56 | stw r8,4(r3) | ||
57 | mr r3,r6 # return the remainder in r3 | ||
58 | blr | ||
diff --git a/arch/powerpc/lib/e2a.c b/arch/powerpc/lib/e2a.c new file mode 100644 index 000000000000..d2b834887920 --- /dev/null +++ b/arch/powerpc/lib/e2a.c | |||
@@ -0,0 +1,108 @@ | |||
1 | /* | ||
2 | * arch/ppc64/lib/e2a.c | ||
3 | * | ||
4 | * EBCDIC to ASCII conversion | ||
5 | * | ||
6 | * This function moved here from arch/ppc64/kernel/viopath.c | ||
7 | * | ||
8 | * (C) Copyright 2000-2004 IBM Corporation | ||
9 | * | ||
10 | * This program is free software; you can redistribute it and/or | ||
11 | * modify it under the terms of the GNU General Public License as | ||
12 | * published by the Free Software Foundation; either version 2 of the | ||
13 | * License, or (at your option) anyu later version. | ||
14 | * | ||
15 | * This program is distributed in the hope that it will be useful, but | ||
16 | * WITHOUT ANY WARRANTY; without even the implied warranty of | ||
17 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | ||
18 | * General Public License for more details. | ||
19 | * | ||
20 | * You should have received a copy of the GNU General Public License | ||
21 | * along with this program; if not, write to the Free Software Foundation, | ||
22 | * Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA | ||
23 | * | ||
24 | */ | ||
25 | |||
26 | #include <linux/module.h> | ||
27 | |||
28 | unsigned char e2a(unsigned char x) | ||
29 | { | ||
30 | switch (x) { | ||
31 | case 0xF0: | ||
32 | return '0'; | ||
33 | case 0xF1: | ||
34 | return '1'; | ||
35 | case 0xF2: | ||
36 | return '2'; | ||
37 | case 0xF3: | ||
38 | return '3'; | ||
39 | case 0xF4: | ||
40 | return '4'; | ||
41 | case 0xF5: | ||
42 | return '5'; | ||
43 | case 0xF6: | ||
44 | return '6'; | ||
45 | case 0xF7: | ||
46 | return '7'; | ||
47 | case 0xF8: | ||
48 | return '8'; | ||
49 | case 0xF9: | ||
50 | return '9'; | ||
51 | case 0xC1: | ||
52 | return 'A'; | ||
53 | case 0xC2: | ||
54 | return 'B'; | ||
55 | case 0xC3: | ||
56 | return 'C'; | ||
57 | case 0xC4: | ||
58 | return 'D'; | ||
59 | case 0xC5: | ||
60 | return 'E'; | ||
61 | case 0xC6: | ||
62 | return 'F'; | ||
63 | case 0xC7: | ||
64 | return 'G'; | ||
65 | case 0xC8: | ||
66 | return 'H'; | ||
67 | case 0xC9: | ||
68 | return 'I'; | ||
69 | case 0xD1: | ||
70 | return 'J'; | ||
71 | case 0xD2: | ||
72 | return 'K'; | ||
73 | case 0xD3: | ||
74 | return 'L'; | ||
75 | case 0xD4: | ||
76 | return 'M'; | ||
77 | case 0xD5: | ||
78 | return 'N'; | ||
79 | case 0xD6: | ||
80 | return 'O'; | ||
81 | case 0xD7: | ||
82 | return 'P'; | ||
83 | case 0xD8: | ||
84 | return 'Q'; | ||
85 | case 0xD9: | ||
86 | return 'R'; | ||
87 | case 0xE2: | ||
88 | return 'S'; | ||
89 | case 0xE3: | ||
90 | return 'T'; | ||
91 | case 0xE4: | ||
92 | return 'U'; | ||
93 | case 0xE5: | ||
94 | return 'V'; | ||
95 | case 0xE6: | ||
96 | return 'W'; | ||
97 | case 0xE7: | ||
98 | return 'X'; | ||
99 | case 0xE8: | ||
100 | return 'Y'; | ||
101 | case 0xE9: | ||
102 | return 'Z'; | ||
103 | } | ||
104 | return ' '; | ||
105 | } | ||
106 | EXPORT_SYMBOL(e2a); | ||
107 | |||
108 | |||
diff --git a/arch/powerpc/lib/memcpy.S b/arch/powerpc/lib/memcpy.S new file mode 100644 index 000000000000..9ccacdf5bcb9 --- /dev/null +++ b/arch/powerpc/lib/memcpy.S | |||
@@ -0,0 +1,172 @@ | |||
1 | /* | ||
2 | * arch/ppc64/lib/memcpy.S | ||
3 | * | ||
4 | * Copyright (C) 2002 Paul Mackerras, IBM Corp. | ||
5 | * | ||
6 | * This program is free software; you can redistribute it and/or | ||
7 | * modify it under the terms of the GNU General Public License | ||
8 | * as published by the Free Software Foundation; either version | ||
9 | * 2 of the License, or (at your option) any later version. | ||
10 | */ | ||
11 | #include <asm/processor.h> | ||
12 | #include <asm/ppc_asm.h> | ||
13 | |||
14 | .align 7 | ||
15 | _GLOBAL(memcpy) | ||
16 | mtcrf 0x01,r5 | ||
17 | cmpldi cr1,r5,16 | ||
18 | neg r6,r3 # LS 3 bits = # bytes to 8-byte dest bdry | ||
19 | andi. r6,r6,7 | ||
20 | dcbt 0,r4 | ||
21 | blt cr1,.Lshort_copy | ||
22 | bne .Ldst_unaligned | ||
23 | .Ldst_aligned: | ||
24 | andi. r0,r4,7 | ||
25 | addi r3,r3,-16 | ||
26 | bne .Lsrc_unaligned | ||
27 | srdi r7,r5,4 | ||
28 | ld r9,0(r4) | ||
29 | addi r4,r4,-8 | ||
30 | mtctr r7 | ||
31 | andi. r5,r5,7 | ||
32 | bf cr7*4+0,2f | ||
33 | addi r3,r3,8 | ||
34 | addi r4,r4,8 | ||
35 | mr r8,r9 | ||
36 | blt cr1,3f | ||
37 | 1: ld r9,8(r4) | ||
38 | std r8,8(r3) | ||
39 | 2: ldu r8,16(r4) | ||
40 | stdu r9,16(r3) | ||
41 | bdnz 1b | ||
42 | 3: std r8,8(r3) | ||
43 | beqlr | ||
44 | addi r3,r3,16 | ||
45 | ld r9,8(r4) | ||
46 | .Ldo_tail: | ||
47 | bf cr7*4+1,1f | ||
48 | rotldi r9,r9,32 | ||
49 | stw r9,0(r3) | ||
50 | addi r3,r3,4 | ||
51 | 1: bf cr7*4+2,2f | ||
52 | rotldi r9,r9,16 | ||
53 | sth r9,0(r3) | ||
54 | addi r3,r3,2 | ||
55 | 2: bf cr7*4+3,3f | ||
56 | rotldi r9,r9,8 | ||
57 | stb r9,0(r3) | ||
58 | 3: blr | ||
59 | |||
60 | .Lsrc_unaligned: | ||
61 | srdi r6,r5,3 | ||
62 | addi r5,r5,-16 | ||
63 | subf r4,r0,r4 | ||
64 | srdi r7,r5,4 | ||
65 | sldi r10,r0,3 | ||
66 | cmpdi cr6,r6,3 | ||
67 | andi. r5,r5,7 | ||
68 | mtctr r7 | ||
69 | subfic r11,r10,64 | ||
70 | add r5,r5,r0 | ||
71 | |||
72 | bt cr7*4+0,0f | ||
73 | |||
74 | ld r9,0(r4) # 3+2n loads, 2+2n stores | ||
75 | ld r0,8(r4) | ||
76 | sld r6,r9,r10 | ||
77 | ldu r9,16(r4) | ||
78 | srd r7,r0,r11 | ||
79 | sld r8,r0,r10 | ||
80 | or r7,r7,r6 | ||
81 | blt cr6,4f | ||
82 | ld r0,8(r4) | ||
83 | # s1<< in r8, d0=(s0<<|s1>>) in r7, s3 in r0, s2 in r9, nix in r6 & r12 | ||
84 | b 2f | ||
85 | |||
86 | 0: ld r0,0(r4) # 4+2n loads, 3+2n stores | ||
87 | ldu r9,8(r4) | ||
88 | sld r8,r0,r10 | ||
89 | addi r3,r3,-8 | ||
90 | blt cr6,5f | ||
91 | ld r0,8(r4) | ||
92 | srd r12,r9,r11 | ||
93 | sld r6,r9,r10 | ||
94 | ldu r9,16(r4) | ||
95 | or r12,r8,r12 | ||
96 | srd r7,r0,r11 | ||
97 | sld r8,r0,r10 | ||
98 | addi r3,r3,16 | ||
99 | beq cr6,3f | ||
100 | |||
101 | # d0=(s0<<|s1>>) in r12, s1<< in r6, s2>> in r7, s2<< in r8, s3 in r9 | ||
102 | 1: or r7,r7,r6 | ||
103 | ld r0,8(r4) | ||
104 | std r12,8(r3) | ||
105 | 2: srd r12,r9,r11 | ||
106 | sld r6,r9,r10 | ||
107 | ldu r9,16(r4) | ||
108 | or r12,r8,r12 | ||
109 | stdu r7,16(r3) | ||
110 | srd r7,r0,r11 | ||
111 | sld r8,r0,r10 | ||
112 | bdnz 1b | ||
113 | |||
114 | 3: std r12,8(r3) | ||
115 | or r7,r7,r6 | ||
116 | 4: std r7,16(r3) | ||
117 | 5: srd r12,r9,r11 | ||
118 | or r12,r8,r12 | ||
119 | std r12,24(r3) | ||
120 | beqlr | ||
121 | cmpwi cr1,r5,8 | ||
122 | addi r3,r3,32 | ||
123 | sld r9,r9,r10 | ||
124 | ble cr1,.Ldo_tail | ||
125 | ld r0,8(r4) | ||
126 | srd r7,r0,r11 | ||
127 | or r9,r7,r9 | ||
128 | b .Ldo_tail | ||
129 | |||
130 | .Ldst_unaligned: | ||
131 | mtcrf 0x01,r6 # put #bytes to 8B bdry into cr7 | ||
132 | subf r5,r6,r5 | ||
133 | li r7,0 | ||
134 | cmpldi r1,r5,16 | ||
135 | bf cr7*4+3,1f | ||
136 | lbz r0,0(r4) | ||
137 | stb r0,0(r3) | ||
138 | addi r7,r7,1 | ||
139 | 1: bf cr7*4+2,2f | ||
140 | lhzx r0,r7,r4 | ||
141 | sthx r0,r7,r3 | ||
142 | addi r7,r7,2 | ||
143 | 2: bf cr7*4+1,3f | ||
144 | lwzx r0,r7,r4 | ||
145 | stwx r0,r7,r3 | ||
146 | 3: mtcrf 0x01,r5 | ||
147 | add r4,r6,r4 | ||
148 | add r3,r6,r3 | ||
149 | b .Ldst_aligned | ||
150 | |||
151 | .Lshort_copy: | ||
152 | bf cr7*4+0,1f | ||
153 | lwz r0,0(r4) | ||
154 | lwz r9,4(r4) | ||
155 | addi r4,r4,8 | ||
156 | stw r0,0(r3) | ||
157 | stw r9,4(r3) | ||
158 | addi r3,r3,8 | ||
159 | 1: bf cr7*4+1,2f | ||
160 | lwz r0,0(r4) | ||
161 | addi r4,r4,4 | ||
162 | stw r0,0(r3) | ||
163 | addi r3,r3,4 | ||
164 | 2: bf cr7*4+2,3f | ||
165 | lhz r0,0(r4) | ||
166 | addi r4,r4,2 | ||
167 | sth r0,0(r3) | ||
168 | addi r3,r3,2 | ||
169 | 3: bf cr7*4+3,4f | ||
170 | lbz r0,0(r4) | ||
171 | stb r0,0(r3) | ||
172 | 4: blr | ||
diff --git a/arch/powerpc/lib/rheap.c b/arch/powerpc/lib/rheap.c new file mode 100644 index 000000000000..42c5de2c898f --- /dev/null +++ b/arch/powerpc/lib/rheap.c | |||
@@ -0,0 +1,693 @@ | |||
1 | /* | ||
2 | * arch/ppc/syslib/rheap.c | ||
3 | * | ||
4 | * A Remote Heap. Remote means that we don't touch the memory that the | ||
5 | * heap points to. Normal heap implementations use the memory they manage | ||
6 | * to place their list. We cannot do that because the memory we manage may | ||
7 | * have special properties, for example it is uncachable or of different | ||
8 | * endianess. | ||
9 | * | ||
10 | * Author: Pantelis Antoniou <panto@intracom.gr> | ||
11 | * | ||
12 | * 2004 (c) INTRACOM S.A. Greece. This file is licensed under | ||
13 | * the terms of the GNU General Public License version 2. This program | ||
14 | * is licensed "as is" without any warranty of any kind, whether express | ||
15 | * or implied. | ||
16 | */ | ||
17 | #include <linux/types.h> | ||
18 | #include <linux/errno.h> | ||
19 | #include <linux/mm.h> | ||
20 | #include <linux/slab.h> | ||
21 | |||
22 | #include <asm/rheap.h> | ||
23 | |||
24 | /* | ||
25 | * Fixup a list_head, needed when copying lists. If the pointers fall | ||
26 | * between s and e, apply the delta. This assumes that | ||
27 | * sizeof(struct list_head *) == sizeof(unsigned long *). | ||
28 | */ | ||
29 | static inline void fixup(unsigned long s, unsigned long e, int d, | ||
30 | struct list_head *l) | ||
31 | { | ||
32 | unsigned long *pp; | ||
33 | |||
34 | pp = (unsigned long *)&l->next; | ||
35 | if (*pp >= s && *pp < e) | ||
36 | *pp += d; | ||
37 | |||
38 | pp = (unsigned long *)&l->prev; | ||
39 | if (*pp >= s && *pp < e) | ||
40 | *pp += d; | ||
41 | } | ||
42 | |||
43 | /* Grow the allocated blocks */ | ||
44 | static int grow(rh_info_t * info, int max_blocks) | ||
45 | { | ||
46 | rh_block_t *block, *blk; | ||
47 | int i, new_blocks; | ||
48 | int delta; | ||
49 | unsigned long blks, blke; | ||
50 | |||
51 | if (max_blocks <= info->max_blocks) | ||
52 | return -EINVAL; | ||
53 | |||
54 | new_blocks = max_blocks - info->max_blocks; | ||
55 | |||
56 | block = kmalloc(sizeof(rh_block_t) * max_blocks, GFP_KERNEL); | ||
57 | if (block == NULL) | ||
58 | return -ENOMEM; | ||
59 | |||
60 | if (info->max_blocks > 0) { | ||
61 | |||
62 | /* copy old block area */ | ||
63 | memcpy(block, info->block, | ||
64 | sizeof(rh_block_t) * info->max_blocks); | ||
65 | |||
66 | delta = (char *)block - (char *)info->block; | ||
67 | |||
68 | /* and fixup list pointers */ | ||
69 | blks = (unsigned long)info->block; | ||
70 | blke = (unsigned long)(info->block + info->max_blocks); | ||
71 | |||
72 | for (i = 0, blk = block; i < info->max_blocks; i++, blk++) | ||
73 | fixup(blks, blke, delta, &blk->list); | ||
74 | |||
75 | fixup(blks, blke, delta, &info->empty_list); | ||
76 | fixup(blks, blke, delta, &info->free_list); | ||
77 | fixup(blks, blke, delta, &info->taken_list); | ||
78 | |||
79 | /* free the old allocated memory */ | ||
80 | if ((info->flags & RHIF_STATIC_BLOCK) == 0) | ||
81 | kfree(info->block); | ||
82 | } | ||
83 | |||
84 | info->block = block; | ||
85 | info->empty_slots += new_blocks; | ||
86 | info->max_blocks = max_blocks; | ||
87 | info->flags &= ~RHIF_STATIC_BLOCK; | ||
88 | |||
89 | /* add all new blocks to the free list */ | ||
90 | for (i = 0, blk = block + info->max_blocks; i < new_blocks; i++, blk++) | ||
91 | list_add(&blk->list, &info->empty_list); | ||
92 | |||
93 | return 0; | ||
94 | } | ||
95 | |||
96 | /* | ||
97 | * Assure at least the required amount of empty slots. If this function | ||
98 | * causes a grow in the block area then all pointers kept to the block | ||
99 | * area are invalid! | ||
100 | */ | ||
101 | static int assure_empty(rh_info_t * info, int slots) | ||
102 | { | ||
103 | int max_blocks; | ||
104 | |||
105 | /* This function is not meant to be used to grow uncontrollably */ | ||
106 | if (slots >= 4) | ||
107 | return -EINVAL; | ||
108 | |||
109 | /* Enough space */ | ||
110 | if (info->empty_slots >= slots) | ||
111 | return 0; | ||
112 | |||
113 | /* Next 16 sized block */ | ||
114 | max_blocks = ((info->max_blocks + slots) + 15) & ~15; | ||
115 | |||
116 | return grow(info, max_blocks); | ||
117 | } | ||
118 | |||
119 | static rh_block_t *get_slot(rh_info_t * info) | ||
120 | { | ||
121 | rh_block_t *blk; | ||
122 | |||
123 | /* If no more free slots, and failure to extend. */ | ||
124 | /* XXX: You should have called assure_empty before */ | ||
125 | if (info->empty_slots == 0) { | ||
126 | printk(KERN_ERR "rh: out of slots; crash is imminent.\n"); | ||
127 | return NULL; | ||
128 | } | ||
129 | |||
130 | /* Get empty slot to use */ | ||
131 | blk = list_entry(info->empty_list.next, rh_block_t, list); | ||
132 | list_del_init(&blk->list); | ||
133 | info->empty_slots--; | ||
134 | |||
135 | /* Initialize */ | ||
136 | blk->start = NULL; | ||
137 | blk->size = 0; | ||
138 | blk->owner = NULL; | ||
139 | |||
140 | return blk; | ||
141 | } | ||
142 | |||
143 | static inline void release_slot(rh_info_t * info, rh_block_t * blk) | ||
144 | { | ||
145 | list_add(&blk->list, &info->empty_list); | ||
146 | info->empty_slots++; | ||
147 | } | ||
148 | |||
149 | static void attach_free_block(rh_info_t * info, rh_block_t * blkn) | ||
150 | { | ||
151 | rh_block_t *blk; | ||
152 | rh_block_t *before; | ||
153 | rh_block_t *after; | ||
154 | rh_block_t *next; | ||
155 | int size; | ||
156 | unsigned long s, e, bs, be; | ||
157 | struct list_head *l; | ||
158 | |||
159 | /* We assume that they are aligned properly */ | ||
160 | size = blkn->size; | ||
161 | s = (unsigned long)blkn->start; | ||
162 | e = s + size; | ||
163 | |||
164 | /* Find the blocks immediately before and after the given one | ||
165 | * (if any) */ | ||
166 | before = NULL; | ||
167 | after = NULL; | ||
168 | next = NULL; | ||
169 | |||
170 | list_for_each(l, &info->free_list) { | ||
171 | blk = list_entry(l, rh_block_t, list); | ||
172 | |||
173 | bs = (unsigned long)blk->start; | ||
174 | be = bs + blk->size; | ||
175 | |||
176 | if (next == NULL && s >= bs) | ||
177 | next = blk; | ||
178 | |||
179 | if (be == s) | ||
180 | before = blk; | ||
181 | |||
182 | if (e == bs) | ||
183 | after = blk; | ||
184 | |||
185 | /* If both are not null, break now */ | ||
186 | if (before != NULL && after != NULL) | ||
187 | break; | ||
188 | } | ||
189 | |||
190 | /* Now check if they are really adjacent */ | ||
191 | if (before != NULL && s != (unsigned long)before->start + before->size) | ||
192 | before = NULL; | ||
193 | |||
194 | if (after != NULL && e != (unsigned long)after->start) | ||
195 | after = NULL; | ||
196 | |||
197 | /* No coalescing; list insert and return */ | ||
198 | if (before == NULL && after == NULL) { | ||
199 | |||
200 | if (next != NULL) | ||
201 | list_add(&blkn->list, &next->list); | ||
202 | else | ||
203 | list_add(&blkn->list, &info->free_list); | ||
204 | |||
205 | return; | ||
206 | } | ||
207 | |||
208 | /* We don't need it anymore */ | ||
209 | release_slot(info, blkn); | ||
210 | |||
211 | /* Grow the before block */ | ||
212 | if (before != NULL && after == NULL) { | ||
213 | before->size += size; | ||
214 | return; | ||
215 | } | ||
216 | |||
217 | /* Grow the after block backwards */ | ||
218 | if (before == NULL && after != NULL) { | ||
219 | after->start = (int8_t *)after->start - size; | ||
220 | after->size += size; | ||
221 | return; | ||
222 | } | ||
223 | |||
224 | /* Grow the before block, and release the after block */ | ||
225 | before->size += size + after->size; | ||
226 | list_del(&after->list); | ||
227 | release_slot(info, after); | ||
228 | } | ||
229 | |||
230 | static void attach_taken_block(rh_info_t * info, rh_block_t * blkn) | ||
231 | { | ||
232 | rh_block_t *blk; | ||
233 | struct list_head *l; | ||
234 | |||
235 | /* Find the block immediately before the given one (if any) */ | ||
236 | list_for_each(l, &info->taken_list) { | ||
237 | blk = list_entry(l, rh_block_t, list); | ||
238 | if (blk->start > blkn->start) { | ||
239 | list_add_tail(&blkn->list, &blk->list); | ||
240 | return; | ||
241 | } | ||
242 | } | ||
243 | |||
244 | list_add_tail(&blkn->list, &info->taken_list); | ||
245 | } | ||
246 | |||
247 | /* | ||
248 | * Create a remote heap dynamically. Note that no memory for the blocks | ||
249 | * are allocated. It will upon the first allocation | ||
250 | */ | ||
251 | rh_info_t *rh_create(unsigned int alignment) | ||
252 | { | ||
253 | rh_info_t *info; | ||
254 | |||
255 | /* Alignment must be a power of two */ | ||
256 | if ((alignment & (alignment - 1)) != 0) | ||
257 | return ERR_PTR(-EINVAL); | ||
258 | |||
259 | info = kmalloc(sizeof(*info), GFP_KERNEL); | ||
260 | if (info == NULL) | ||
261 | return ERR_PTR(-ENOMEM); | ||
262 | |||
263 | info->alignment = alignment; | ||
264 | |||
265 | /* Initially everything as empty */ | ||
266 | info->block = NULL; | ||
267 | info->max_blocks = 0; | ||
268 | info->empty_slots = 0; | ||
269 | info->flags = 0; | ||
270 | |||
271 | INIT_LIST_HEAD(&info->empty_list); | ||
272 | INIT_LIST_HEAD(&info->free_list); | ||
273 | INIT_LIST_HEAD(&info->taken_list); | ||
274 | |||
275 | return info; | ||
276 | } | ||
277 | |||
278 | /* | ||
279 | * Destroy a dynamically created remote heap. Deallocate only if the areas | ||
280 | * are not static | ||
281 | */ | ||
282 | void rh_destroy(rh_info_t * info) | ||
283 | { | ||
284 | if ((info->flags & RHIF_STATIC_BLOCK) == 0 && info->block != NULL) | ||
285 | kfree(info->block); | ||
286 | |||
287 | if ((info->flags & RHIF_STATIC_INFO) == 0) | ||
288 | kfree(info); | ||
289 | } | ||
290 | |||
291 | /* | ||
292 | * Initialize in place a remote heap info block. This is needed to support | ||
293 | * operation very early in the startup of the kernel, when it is not yet safe | ||
294 | * to call kmalloc. | ||
295 | */ | ||
296 | void rh_init(rh_info_t * info, unsigned int alignment, int max_blocks, | ||
297 | rh_block_t * block) | ||
298 | { | ||
299 | int i; | ||
300 | rh_block_t *blk; | ||
301 | |||
302 | /* Alignment must be a power of two */ | ||
303 | if ((alignment & (alignment - 1)) != 0) | ||
304 | return; | ||
305 | |||
306 | info->alignment = alignment; | ||
307 | |||
308 | /* Initially everything as empty */ | ||
309 | info->block = block; | ||
310 | info->max_blocks = max_blocks; | ||
311 | info->empty_slots = max_blocks; | ||
312 | info->flags = RHIF_STATIC_INFO | RHIF_STATIC_BLOCK; | ||
313 | |||
314 | INIT_LIST_HEAD(&info->empty_list); | ||
315 | INIT_LIST_HEAD(&info->free_list); | ||
316 | INIT_LIST_HEAD(&info->taken_list); | ||
317 | |||
318 | /* Add all new blocks to the free list */ | ||
319 | for (i = 0, blk = block; i < max_blocks; i++, blk++) | ||
320 | list_add(&blk->list, &info->empty_list); | ||
321 | } | ||
322 | |||
323 | /* Attach a free memory region, coalesces regions if adjuscent */ | ||
324 | int rh_attach_region(rh_info_t * info, void *start, int size) | ||
325 | { | ||
326 | rh_block_t *blk; | ||
327 | unsigned long s, e, m; | ||
328 | int r; | ||
329 | |||
330 | /* The region must be aligned */ | ||
331 | s = (unsigned long)start; | ||
332 | e = s + size; | ||
333 | m = info->alignment - 1; | ||
334 | |||
335 | /* Round start up */ | ||
336 | s = (s + m) & ~m; | ||
337 | |||
338 | /* Round end down */ | ||
339 | e = e & ~m; | ||
340 | |||
341 | /* Take final values */ | ||
342 | start = (void *)s; | ||
343 | size = (int)(e - s); | ||
344 | |||
345 | /* Grow the blocks, if needed */ | ||
346 | r = assure_empty(info, 1); | ||
347 | if (r < 0) | ||
348 | return r; | ||
349 | |||
350 | blk = get_slot(info); | ||
351 | blk->start = start; | ||
352 | blk->size = size; | ||
353 | blk->owner = NULL; | ||
354 | |||
355 | attach_free_block(info, blk); | ||
356 | |||
357 | return 0; | ||
358 | } | ||
359 | |||
360 | /* Detatch given address range, splits free block if needed. */ | ||
361 | void *rh_detach_region(rh_info_t * info, void *start, int size) | ||
362 | { | ||
363 | struct list_head *l; | ||
364 | rh_block_t *blk, *newblk; | ||
365 | unsigned long s, e, m, bs, be; | ||
366 | |||
367 | /* Validate size */ | ||
368 | if (size <= 0) | ||
369 | return ERR_PTR(-EINVAL); | ||
370 | |||
371 | /* The region must be aligned */ | ||
372 | s = (unsigned long)start; | ||
373 | e = s + size; | ||
374 | m = info->alignment - 1; | ||
375 | |||
376 | /* Round start up */ | ||
377 | s = (s + m) & ~m; | ||
378 | |||
379 | /* Round end down */ | ||
380 | e = e & ~m; | ||
381 | |||
382 | if (assure_empty(info, 1) < 0) | ||
383 | return ERR_PTR(-ENOMEM); | ||
384 | |||
385 | blk = NULL; | ||
386 | list_for_each(l, &info->free_list) { | ||
387 | blk = list_entry(l, rh_block_t, list); | ||
388 | /* The range must lie entirely inside one free block */ | ||
389 | bs = (unsigned long)blk->start; | ||
390 | be = (unsigned long)blk->start + blk->size; | ||
391 | if (s >= bs && e <= be) | ||
392 | break; | ||
393 | blk = NULL; | ||
394 | } | ||
395 | |||
396 | if (blk == NULL) | ||
397 | return ERR_PTR(-ENOMEM); | ||
398 | |||
399 | /* Perfect fit */ | ||
400 | if (bs == s && be == e) { | ||
401 | /* Delete from free list, release slot */ | ||
402 | list_del(&blk->list); | ||
403 | release_slot(info, blk); | ||
404 | return (void *)s; | ||
405 | } | ||
406 | |||
407 | /* blk still in free list, with updated start and/or size */ | ||
408 | if (bs == s || be == e) { | ||
409 | if (bs == s) | ||
410 | blk->start = (int8_t *)blk->start + size; | ||
411 | blk->size -= size; | ||
412 | |||
413 | } else { | ||
414 | /* The front free fragment */ | ||
415 | blk->size = s - bs; | ||
416 | |||
417 | /* the back free fragment */ | ||
418 | newblk = get_slot(info); | ||
419 | newblk->start = (void *)e; | ||
420 | newblk->size = be - e; | ||
421 | |||
422 | list_add(&newblk->list, &blk->list); | ||
423 | } | ||
424 | |||
425 | return (void *)s; | ||
426 | } | ||
427 | |||
428 | void *rh_alloc(rh_info_t * info, int size, const char *owner) | ||
429 | { | ||
430 | struct list_head *l; | ||
431 | rh_block_t *blk; | ||
432 | rh_block_t *newblk; | ||
433 | void *start; | ||
434 | |||
435 | /* Validate size */ | ||
436 | if (size <= 0) | ||
437 | return ERR_PTR(-EINVAL); | ||
438 | |||
439 | /* Align to configured alignment */ | ||
440 | size = (size + (info->alignment - 1)) & ~(info->alignment - 1); | ||
441 | |||
442 | if (assure_empty(info, 1) < 0) | ||
443 | return ERR_PTR(-ENOMEM); | ||
444 | |||
445 | blk = NULL; | ||
446 | list_for_each(l, &info->free_list) { | ||
447 | blk = list_entry(l, rh_block_t, list); | ||
448 | if (size <= blk->size) | ||
449 | break; | ||
450 | blk = NULL; | ||
451 | } | ||
452 | |||
453 | if (blk == NULL) | ||
454 | return ERR_PTR(-ENOMEM); | ||
455 | |||
456 | /* Just fits */ | ||
457 | if (blk->size == size) { | ||
458 | /* Move from free list to taken list */ | ||
459 | list_del(&blk->list); | ||
460 | blk->owner = owner; | ||
461 | start = blk->start; | ||
462 | |||
463 | attach_taken_block(info, blk); | ||
464 | |||
465 | return start; | ||
466 | } | ||
467 | |||
468 | newblk = get_slot(info); | ||
469 | newblk->start = blk->start; | ||
470 | newblk->size = size; | ||
471 | newblk->owner = owner; | ||
472 | |||
473 | /* blk still in free list, with updated start, size */ | ||
474 | blk->start = (int8_t *)blk->start + size; | ||
475 | blk->size -= size; | ||
476 | |||
477 | start = newblk->start; | ||
478 | |||
479 | attach_taken_block(info, newblk); | ||
480 | |||
481 | return start; | ||
482 | } | ||
483 | |||
484 | /* allocate at precisely the given address */ | ||
485 | void *rh_alloc_fixed(rh_info_t * info, void *start, int size, const char *owner) | ||
486 | { | ||
487 | struct list_head *l; | ||
488 | rh_block_t *blk, *newblk1, *newblk2; | ||
489 | unsigned long s, e, m, bs, be; | ||
490 | |||
491 | /* Validate size */ | ||
492 | if (size <= 0) | ||
493 | return ERR_PTR(-EINVAL); | ||
494 | |||
495 | /* The region must be aligned */ | ||
496 | s = (unsigned long)start; | ||
497 | e = s + size; | ||
498 | m = info->alignment - 1; | ||
499 | |||
500 | /* Round start up */ | ||
501 | s = (s + m) & ~m; | ||
502 | |||
503 | /* Round end down */ | ||
504 | e = e & ~m; | ||
505 | |||
506 | if (assure_empty(info, 2) < 0) | ||
507 | return ERR_PTR(-ENOMEM); | ||
508 | |||
509 | blk = NULL; | ||
510 | list_for_each(l, &info->free_list) { | ||
511 | blk = list_entry(l, rh_block_t, list); | ||
512 | /* The range must lie entirely inside one free block */ | ||
513 | bs = (unsigned long)blk->start; | ||
514 | be = (unsigned long)blk->start + blk->size; | ||
515 | if (s >= bs && e <= be) | ||
516 | break; | ||
517 | } | ||
518 | |||
519 | if (blk == NULL) | ||
520 | return ERR_PTR(-ENOMEM); | ||
521 | |||
522 | /* Perfect fit */ | ||
523 | if (bs == s && be == e) { | ||
524 | /* Move from free list to taken list */ | ||
525 | list_del(&blk->list); | ||
526 | blk->owner = owner; | ||
527 | |||
528 | start = blk->start; | ||
529 | attach_taken_block(info, blk); | ||
530 | |||
531 | return start; | ||
532 | |||
533 | } | ||
534 | |||
535 | /* blk still in free list, with updated start and/or size */ | ||
536 | if (bs == s || be == e) { | ||
537 | if (bs == s) | ||
538 | blk->start = (int8_t *)blk->start + size; | ||
539 | blk->size -= size; | ||
540 | |||
541 | } else { | ||
542 | /* The front free fragment */ | ||
543 | blk->size = s - bs; | ||
544 | |||
545 | /* The back free fragment */ | ||
546 | newblk2 = get_slot(info); | ||
547 | newblk2->start = (void *)e; | ||
548 | newblk2->size = be - e; | ||
549 | |||
550 | list_add(&newblk2->list, &blk->list); | ||
551 | } | ||
552 | |||
553 | newblk1 = get_slot(info); | ||
554 | newblk1->start = (void *)s; | ||
555 | newblk1->size = e - s; | ||
556 | newblk1->owner = owner; | ||
557 | |||
558 | start = newblk1->start; | ||
559 | attach_taken_block(info, newblk1); | ||
560 | |||
561 | return start; | ||
562 | } | ||
563 | |||
564 | int rh_free(rh_info_t * info, void *start) | ||
565 | { | ||
566 | rh_block_t *blk, *blk2; | ||
567 | struct list_head *l; | ||
568 | int size; | ||
569 | |||
570 | /* Linear search for block */ | ||
571 | blk = NULL; | ||
572 | list_for_each(l, &info->taken_list) { | ||
573 | blk2 = list_entry(l, rh_block_t, list); | ||
574 | if (start < blk2->start) | ||
575 | break; | ||
576 | blk = blk2; | ||
577 | } | ||
578 | |||
579 | if (blk == NULL || start > (blk->start + blk->size)) | ||
580 | return -EINVAL; | ||
581 | |||
582 | /* Remove from taken list */ | ||
583 | list_del(&blk->list); | ||
584 | |||
585 | /* Get size of freed block */ | ||
586 | size = blk->size; | ||
587 | attach_free_block(info, blk); | ||
588 | |||
589 | return size; | ||
590 | } | ||
591 | |||
592 | int rh_get_stats(rh_info_t * info, int what, int max_stats, rh_stats_t * stats) | ||
593 | { | ||
594 | rh_block_t *blk; | ||
595 | struct list_head *l; | ||
596 | struct list_head *h; | ||
597 | int nr; | ||
598 | |||
599 | switch (what) { | ||
600 | |||
601 | case RHGS_FREE: | ||
602 | h = &info->free_list; | ||
603 | break; | ||
604 | |||
605 | case RHGS_TAKEN: | ||
606 | h = &info->taken_list; | ||
607 | break; | ||
608 | |||
609 | default: | ||
610 | return -EINVAL; | ||
611 | } | ||
612 | |||
613 | /* Linear search for block */ | ||
614 | nr = 0; | ||
615 | list_for_each(l, h) { | ||
616 | blk = list_entry(l, rh_block_t, list); | ||
617 | if (stats != NULL && nr < max_stats) { | ||
618 | stats->start = blk->start; | ||
619 | stats->size = blk->size; | ||
620 | stats->owner = blk->owner; | ||
621 | stats++; | ||
622 | } | ||
623 | nr++; | ||
624 | } | ||
625 | |||
626 | return nr; | ||
627 | } | ||
628 | |||
629 | int rh_set_owner(rh_info_t * info, void *start, const char *owner) | ||
630 | { | ||
631 | rh_block_t *blk, *blk2; | ||
632 | struct list_head *l; | ||
633 | int size; | ||
634 | |||
635 | /* Linear search for block */ | ||
636 | blk = NULL; | ||
637 | list_for_each(l, &info->taken_list) { | ||
638 | blk2 = list_entry(l, rh_block_t, list); | ||
639 | if (start < blk2->start) | ||
640 | break; | ||
641 | blk = blk2; | ||
642 | } | ||
643 | |||
644 | if (blk == NULL || start > (blk->start + blk->size)) | ||
645 | return -EINVAL; | ||
646 | |||
647 | blk->owner = owner; | ||
648 | size = blk->size; | ||
649 | |||
650 | return size; | ||
651 | } | ||
652 | |||
653 | void rh_dump(rh_info_t * info) | ||
654 | { | ||
655 | static rh_stats_t st[32]; /* XXX maximum 32 blocks */ | ||
656 | int maxnr; | ||
657 | int i, nr; | ||
658 | |||
659 | maxnr = sizeof(st) / sizeof(st[0]); | ||
660 | |||
661 | printk(KERN_INFO | ||
662 | "info @0x%p (%d slots empty / %d max)\n", | ||
663 | info, info->empty_slots, info->max_blocks); | ||
664 | |||
665 | printk(KERN_INFO " Free:\n"); | ||
666 | nr = rh_get_stats(info, RHGS_FREE, maxnr, st); | ||
667 | if (nr > maxnr) | ||
668 | nr = maxnr; | ||
669 | for (i = 0; i < nr; i++) | ||
670 | printk(KERN_INFO | ||
671 | " 0x%p-0x%p (%u)\n", | ||
672 | st[i].start, (int8_t *) st[i].start + st[i].size, | ||
673 | st[i].size); | ||
674 | printk(KERN_INFO "\n"); | ||
675 | |||
676 | printk(KERN_INFO " Taken:\n"); | ||
677 | nr = rh_get_stats(info, RHGS_TAKEN, maxnr, st); | ||
678 | if (nr > maxnr) | ||
679 | nr = maxnr; | ||
680 | for (i = 0; i < nr; i++) | ||
681 | printk(KERN_INFO | ||
682 | " 0x%p-0x%p (%u) %s\n", | ||
683 | st[i].start, (int8_t *) st[i].start + st[i].size, | ||
684 | st[i].size, st[i].owner != NULL ? st[i].owner : ""); | ||
685 | printk(KERN_INFO "\n"); | ||
686 | } | ||
687 | |||
688 | void rh_dump_blk(rh_info_t * info, rh_block_t * blk) | ||
689 | { | ||
690 | printk(KERN_INFO | ||
691 | "blk @0x%p: 0x%p-0x%p (%u)\n", | ||
692 | blk, blk->start, (int8_t *) blk->start + blk->size, blk->size); | ||
693 | } | ||
diff --git a/arch/powerpc/lib/sstep.c b/arch/powerpc/lib/sstep.c new file mode 100644 index 000000000000..e79123d1485c --- /dev/null +++ b/arch/powerpc/lib/sstep.c | |||
@@ -0,0 +1,141 @@ | |||
1 | /* | ||
2 | * Single-step support. | ||
3 | * | ||
4 | * Copyright (C) 2004 Paul Mackerras <paulus@au.ibm.com>, IBM | ||
5 | * | ||
6 | * This program is free software; you can redistribute it and/or | ||
7 | * modify it under the terms of the GNU General Public License | ||
8 | * as published by the Free Software Foundation; either version | ||
9 | * 2 of the License, or (at your option) any later version. | ||
10 | */ | ||
11 | #include <linux/kernel.h> | ||
12 | #include <linux/ptrace.h> | ||
13 | #include <asm/sstep.h> | ||
14 | #include <asm/processor.h> | ||
15 | |||
16 | extern char system_call_common[]; | ||
17 | |||
18 | /* Bits in SRR1 that are copied from MSR */ | ||
19 | #define MSR_MASK 0xffffffff87c0ffff | ||
20 | |||
21 | /* | ||
22 | * Determine whether a conditional branch instruction would branch. | ||
23 | */ | ||
24 | static int branch_taken(unsigned int instr, struct pt_regs *regs) | ||
25 | { | ||
26 | unsigned int bo = (instr >> 21) & 0x1f; | ||
27 | unsigned int bi; | ||
28 | |||
29 | if ((bo & 4) == 0) { | ||
30 | /* decrement counter */ | ||
31 | --regs->ctr; | ||
32 | if (((bo >> 1) & 1) ^ (regs->ctr == 0)) | ||
33 | return 0; | ||
34 | } | ||
35 | if ((bo & 0x10) == 0) { | ||
36 | /* check bit from CR */ | ||
37 | bi = (instr >> 16) & 0x1f; | ||
38 | if (((regs->ccr >> (31 - bi)) & 1) != ((bo >> 3) & 1)) | ||
39 | return 0; | ||
40 | } | ||
41 | return 1; | ||
42 | } | ||
43 | |||
44 | /* | ||
45 | * Emulate instructions that cause a transfer of control. | ||
46 | * Returns 1 if the step was emulated, 0 if not, | ||
47 | * or -1 if the instruction is one that should not be stepped, | ||
48 | * such as an rfid, or a mtmsrd that would clear MSR_RI. | ||
49 | */ | ||
50 | int emulate_step(struct pt_regs *regs, unsigned int instr) | ||
51 | { | ||
52 | unsigned int opcode, rd; | ||
53 | unsigned long int imm; | ||
54 | |||
55 | opcode = instr >> 26; | ||
56 | switch (opcode) { | ||
57 | case 16: /* bc */ | ||
58 | imm = (signed short)(instr & 0xfffc); | ||
59 | if ((instr & 2) == 0) | ||
60 | imm += regs->nip; | ||
61 | regs->nip += 4; | ||
62 | if ((regs->msr & MSR_SF) == 0) | ||
63 | regs->nip &= 0xffffffffUL; | ||
64 | if (instr & 1) | ||
65 | regs->link = regs->nip; | ||
66 | if (branch_taken(instr, regs)) | ||
67 | regs->nip = imm; | ||
68 | return 1; | ||
69 | case 17: /* sc */ | ||
70 | /* | ||
71 | * N.B. this uses knowledge about how the syscall | ||
72 | * entry code works. If that is changed, this will | ||
73 | * need to be changed also. | ||
74 | */ | ||
75 | regs->gpr[9] = regs->gpr[13]; | ||
76 | regs->gpr[11] = regs->nip + 4; | ||
77 | regs->gpr[12] = regs->msr & MSR_MASK; | ||
78 | regs->gpr[13] = (unsigned long) get_paca(); | ||
79 | regs->nip = (unsigned long) &system_call_common; | ||
80 | regs->msr = MSR_KERNEL; | ||
81 | return 1; | ||
82 | case 18: /* b */ | ||
83 | imm = instr & 0x03fffffc; | ||
84 | if (imm & 0x02000000) | ||
85 | imm -= 0x04000000; | ||
86 | if ((instr & 2) == 0) | ||
87 | imm += regs->nip; | ||
88 | if (instr & 1) { | ||
89 | regs->link = regs->nip + 4; | ||
90 | if ((regs->msr & MSR_SF) == 0) | ||
91 | regs->link &= 0xffffffffUL; | ||
92 | } | ||
93 | if ((regs->msr & MSR_SF) == 0) | ||
94 | imm &= 0xffffffffUL; | ||
95 | regs->nip = imm; | ||
96 | return 1; | ||
97 | case 19: | ||
98 | switch (instr & 0x7fe) { | ||
99 | case 0x20: /* bclr */ | ||
100 | case 0x420: /* bcctr */ | ||
101 | imm = (instr & 0x400)? regs->ctr: regs->link; | ||
102 | regs->nip += 4; | ||
103 | if ((regs->msr & MSR_SF) == 0) { | ||
104 | regs->nip &= 0xffffffffUL; | ||
105 | imm &= 0xffffffffUL; | ||
106 | } | ||
107 | if (instr & 1) | ||
108 | regs->link = regs->nip; | ||
109 | if (branch_taken(instr, regs)) | ||
110 | regs->nip = imm; | ||
111 | return 1; | ||
112 | case 0x24: /* rfid, scary */ | ||
113 | return -1; | ||
114 | } | ||
115 | case 31: | ||
116 | rd = (instr >> 21) & 0x1f; | ||
117 | switch (instr & 0x7fe) { | ||
118 | case 0xa6: /* mfmsr */ | ||
119 | regs->gpr[rd] = regs->msr & MSR_MASK; | ||
120 | regs->nip += 4; | ||
121 | if ((regs->msr & MSR_SF) == 0) | ||
122 | regs->nip &= 0xffffffffUL; | ||
123 | return 1; | ||
124 | case 0x164: /* mtmsrd */ | ||
125 | /* only MSR_EE and MSR_RI get changed if bit 15 set */ | ||
126 | /* mtmsrd doesn't change MSR_HV and MSR_ME */ | ||
127 | imm = (instr & 0x10000)? 0x8002: 0xefffffffffffefffUL; | ||
128 | imm = (regs->msr & MSR_MASK & ~imm) | ||
129 | | (regs->gpr[rd] & imm); | ||
130 | if ((imm & MSR_RI) == 0) | ||
131 | /* can't step mtmsrd that would clear MSR_RI */ | ||
132 | return -1; | ||
133 | regs->msr = imm; | ||
134 | regs->nip += 4; | ||
135 | if ((imm & MSR_SF) == 0) | ||
136 | regs->nip &= 0xffffffffUL; | ||
137 | return 1; | ||
138 | } | ||
139 | } | ||
140 | return 0; | ||
141 | } | ||
diff --git a/arch/powerpc/lib/strcase.c b/arch/powerpc/lib/strcase.c new file mode 100644 index 000000000000..36b521091bbc --- /dev/null +++ b/arch/powerpc/lib/strcase.c | |||
@@ -0,0 +1,23 @@ | |||
1 | #include <linux/ctype.h> | ||
2 | |||
3 | int strcasecmp(const char *s1, const char *s2) | ||
4 | { | ||
5 | int c1, c2; | ||
6 | |||
7 | do { | ||
8 | c1 = tolower(*s1++); | ||
9 | c2 = tolower(*s2++); | ||
10 | } while (c1 == c2 && c1 != 0); | ||
11 | return c1 - c2; | ||
12 | } | ||
13 | |||
14 | int strncasecmp(const char *s1, const char *s2, int n) | ||
15 | { | ||
16 | int c1, c2; | ||
17 | |||
18 | do { | ||
19 | c1 = tolower(*s1++); | ||
20 | c2 = tolower(*s2++); | ||
21 | } while ((--n > 0) && c1 == c2 && c1 != 0); | ||
22 | return c1 - c2; | ||
23 | } | ||
diff --git a/arch/powerpc/lib/string.S b/arch/powerpc/lib/string.S new file mode 100644 index 000000000000..15d40e9ef8b1 --- /dev/null +++ b/arch/powerpc/lib/string.S | |||
@@ -0,0 +1,203 @@ | |||
1 | /* | ||
2 | * String handling functions for PowerPC. | ||
3 | * | ||
4 | * Copyright (C) 1996 Paul Mackerras. | ||
5 | * | ||
6 | * This program is free software; you can redistribute it and/or | ||
7 | * modify it under the terms of the GNU General Public License | ||
8 | * as published by the Free Software Foundation; either version | ||
9 | * 2 of the License, or (at your option) any later version. | ||
10 | */ | ||
11 | #include <linux/config.h> | ||
12 | #include <asm/processor.h> | ||
13 | #include <asm/errno.h> | ||
14 | #include <asm/ppc_asm.h> | ||
15 | |||
16 | .text | ||
17 | .stabs "arch/powerpc/lib/",N_SO,0,0,0f | ||
18 | .stabs "string.S",N_SO,0,0,0f | ||
19 | 0: | ||
20 | |||
21 | .section __ex_table,"a" | ||
22 | #ifdef CONFIG_PPC64 | ||
23 | .align 3 | ||
24 | #define EXTBL .llong | ||
25 | #else | ||
26 | .align 2 | ||
27 | #define EXTBL .long | ||
28 | #endif | ||
29 | .text | ||
30 | |||
31 | _GLOBAL(strcpy) | ||
32 | addi r5,r3,-1 | ||
33 | addi r4,r4,-1 | ||
34 | 1: lbzu r0,1(r4) | ||
35 | cmpwi 0,r0,0 | ||
36 | stbu r0,1(r5) | ||
37 | bne 1b | ||
38 | blr | ||
39 | |||
40 | /* This clears out any unused part of the destination buffer, | ||
41 | just as the libc version does. -- paulus */ | ||
42 | _GLOBAL(strncpy) | ||
43 | cmpwi 0,r5,0 | ||
44 | beqlr | ||
45 | mtctr r5 | ||
46 | addi r6,r3,-1 | ||
47 | addi r4,r4,-1 | ||
48 | 1: lbzu r0,1(r4) | ||
49 | cmpwi 0,r0,0 | ||
50 | stbu r0,1(r6) | ||
51 | bdnzf 2,1b /* dec ctr, branch if ctr != 0 && !cr0.eq */ | ||
52 | bnelr /* if we didn't hit a null char, we're done */ | ||
53 | mfctr r5 | ||
54 | cmpwi 0,r5,0 /* any space left in destination buffer? */ | ||
55 | beqlr /* we know r0 == 0 here */ | ||
56 | 2: stbu r0,1(r6) /* clear it out if so */ | ||
57 | bdnz 2b | ||
58 | blr | ||
59 | |||
60 | _GLOBAL(strcat) | ||
61 | addi r5,r3,-1 | ||
62 | addi r4,r4,-1 | ||
63 | 1: lbzu r0,1(r5) | ||
64 | cmpwi 0,r0,0 | ||
65 | bne 1b | ||
66 | addi r5,r5,-1 | ||
67 | 1: lbzu r0,1(r4) | ||
68 | cmpwi 0,r0,0 | ||
69 | stbu r0,1(r5) | ||
70 | bne 1b | ||
71 | blr | ||
72 | |||
73 | _GLOBAL(strcmp) | ||
74 | addi r5,r3,-1 | ||
75 | addi r4,r4,-1 | ||
76 | 1: lbzu r3,1(r5) | ||
77 | cmpwi 1,r3,0 | ||
78 | lbzu r0,1(r4) | ||
79 | subf. r3,r0,r3 | ||
80 | beqlr 1 | ||
81 | beq 1b | ||
82 | blr | ||
83 | |||
84 | _GLOBAL(strlen) | ||
85 | addi r4,r3,-1 | ||
86 | 1: lbzu r0,1(r4) | ||
87 | cmpwi 0,r0,0 | ||
88 | bne 1b | ||
89 | subf r3,r3,r4 | ||
90 | blr | ||
91 | |||
92 | _GLOBAL(memcmp) | ||
93 | cmpwi 0,r5,0 | ||
94 | ble- 2f | ||
95 | mtctr r5 | ||
96 | addi r6,r3,-1 | ||
97 | addi r4,r4,-1 | ||
98 | 1: lbzu r3,1(r6) | ||
99 | lbzu r0,1(r4) | ||
100 | subf. r3,r0,r3 | ||
101 | bdnzt 2,1b | ||
102 | blr | ||
103 | 2: li r3,0 | ||
104 | blr | ||
105 | |||
106 | _GLOBAL(memchr) | ||
107 | cmpwi 0,r5,0 | ||
108 | ble- 2f | ||
109 | mtctr r5 | ||
110 | addi r3,r3,-1 | ||
111 | 1: lbzu r0,1(r3) | ||
112 | cmpw 0,r0,r4 | ||
113 | bdnzf 2,1b | ||
114 | beqlr | ||
115 | 2: li r3,0 | ||
116 | blr | ||
117 | |||
118 | _GLOBAL(__clear_user) | ||
119 | addi r6,r3,-4 | ||
120 | li r3,0 | ||
121 | li r5,0 | ||
122 | cmplwi 0,r4,4 | ||
123 | blt 7f | ||
124 | /* clear a single word */ | ||
125 | 11: stwu r5,4(r6) | ||
126 | beqlr | ||
127 | /* clear word sized chunks */ | ||
128 | andi. r0,r6,3 | ||
129 | add r4,r0,r4 | ||
130 | subf r6,r0,r6 | ||
131 | srwi r0,r4,2 | ||
132 | andi. r4,r4,3 | ||
133 | mtctr r0 | ||
134 | bdz 7f | ||
135 | 1: stwu r5,4(r6) | ||
136 | bdnz 1b | ||
137 | /* clear byte sized chunks */ | ||
138 | 7: cmpwi 0,r4,0 | ||
139 | beqlr | ||
140 | mtctr r4 | ||
141 | addi r6,r6,3 | ||
142 | 8: stbu r5,1(r6) | ||
143 | bdnz 8b | ||
144 | blr | ||
145 | 90: mr r3,r4 | ||
146 | blr | ||
147 | 91: mfctr r3 | ||
148 | slwi r3,r3,2 | ||
149 | add r3,r3,r4 | ||
150 | blr | ||
151 | 92: mfctr r3 | ||
152 | blr | ||
153 | |||
154 | .section __ex_table,"a" | ||
155 | EXTBL 11b,90b | ||
156 | EXTBL 1b,91b | ||
157 | EXTBL 8b,92b | ||
158 | .text | ||
159 | |||
160 | _GLOBAL(__strncpy_from_user) | ||
161 | addi r6,r3,-1 | ||
162 | addi r4,r4,-1 | ||
163 | cmpwi 0,r5,0 | ||
164 | beq 2f | ||
165 | mtctr r5 | ||
166 | 1: lbzu r0,1(r4) | ||
167 | cmpwi 0,r0,0 | ||
168 | stbu r0,1(r6) | ||
169 | bdnzf 2,1b /* dec ctr, branch if ctr != 0 && !cr0.eq */ | ||
170 | beq 3f | ||
171 | 2: addi r6,r6,1 | ||
172 | 3: subf r3,r3,r6 | ||
173 | blr | ||
174 | 99: li r3,-EFAULT | ||
175 | blr | ||
176 | |||
177 | .section __ex_table,"a" | ||
178 | EXTBL 1b,99b | ||
179 | .text | ||
180 | |||
181 | /* r3 = str, r4 = len (> 0), r5 = top (highest addr) */ | ||
182 | _GLOBAL(__strnlen_user) | ||
183 | addi r7,r3,-1 | ||
184 | subf r6,r7,r5 /* top+1 - str */ | ||
185 | cmplw 0,r4,r6 | ||
186 | bge 0f | ||
187 | mr r6,r4 | ||
188 | 0: mtctr r6 /* ctr = min(len, top - str) */ | ||
189 | 1: lbzu r0,1(r7) /* get next byte */ | ||
190 | cmpwi 0,r0,0 | ||
191 | bdnzf 2,1b /* loop if --ctr != 0 && byte != 0 */ | ||
192 | addi r7,r7,1 | ||
193 | subf r3,r3,r7 /* number of bytes we have looked at */ | ||
194 | beqlr /* return if we found a 0 byte */ | ||
195 | cmpw 0,r3,r4 /* did we look at all len bytes? */ | ||
196 | blt 99f /* if not, must have hit top */ | ||
197 | addi r3,r4,1 /* return len + 1 to indicate no null found */ | ||
198 | blr | ||
199 | 99: li r3,0 /* bad address, return 0 */ | ||
200 | blr | ||
201 | |||
202 | .section __ex_table,"a" | ||
203 | EXTBL 1b,99b | ||
diff --git a/arch/powerpc/lib/usercopy.c b/arch/powerpc/lib/usercopy.c new file mode 100644 index 000000000000..5eea6f3c1e03 --- /dev/null +++ b/arch/powerpc/lib/usercopy.c | |||
@@ -0,0 +1,41 @@ | |||
1 | /* | ||
2 | * Functions which are too large to be inlined. | ||
3 | * | ||
4 | * This program is free software; you can redistribute it and/or | ||
5 | * modify it under the terms of the GNU General Public License | ||
6 | * as published by the Free Software Foundation; either version | ||
7 | * 2 of the License, or (at your option) any later version. | ||
8 | */ | ||
9 | #include <linux/module.h> | ||
10 | #include <asm/uaccess.h> | ||
11 | |||
12 | unsigned long copy_from_user(void *to, const void __user *from, unsigned long n) | ||
13 | { | ||
14 | if (likely(access_ok(VERIFY_READ, from, n))) | ||
15 | n = __copy_from_user(to, from, n); | ||
16 | else | ||
17 | memset(to, 0, n); | ||
18 | return n; | ||
19 | } | ||
20 | |||
21 | unsigned long copy_to_user(void __user *to, const void *from, unsigned long n) | ||
22 | { | ||
23 | if (likely(access_ok(VERIFY_WRITE, to, n))) | ||
24 | n = __copy_to_user(to, from, n); | ||
25 | return n; | ||
26 | } | ||
27 | |||
28 | unsigned long copy_in_user(void __user *to, const void __user *from, | ||
29 | unsigned long n) | ||
30 | { | ||
31 | might_sleep(); | ||
32 | if (likely(access_ok(VERIFY_READ, from, n) && | ||
33 | access_ok(VERIFY_WRITE, to, n))) | ||
34 | n =__copy_tofrom_user(to, from, n); | ||
35 | return n; | ||
36 | } | ||
37 | |||
38 | EXPORT_SYMBOL(copy_from_user); | ||
39 | EXPORT_SYMBOL(copy_to_user); | ||
40 | EXPORT_SYMBOL(copy_in_user); | ||
41 | |||