Mercurial > pub > dyncall > dyncall

comparison doc/disas_examples/ppc64.elfabi.disas @ 331:74c056b597b7

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
- disassembly example annotations - callconv appendix in doc: * ppc64 chapter * some cleanups for consistency
author Tassilo Philipp
date Sat, 23 Nov 2019 13:51:35 +0100
parents c0390dc85a07
children ead041d93e36
comparison
equal deleted inserted replaced
330:4e6f63b7020e 331:74c056b597b7
40 3c: 4e 80 00 20 blr 40 3c: 4e 80 00 20 blr
41 ... 41 ...
42 48: 80 01 00 01 lwz r0,1(r1) 42 48: 80 01 00 01 lwz r0,1(r1)
43 43
44 000000000000004c <.nonleaf_call>: 44 000000000000004c <.nonleaf_call>:
45 4c: 7c 08 02 a6 mflr r0 45 4c: 7c 08 02 a6 mflr r0 ; |
46 50: fb e1 ff f8 std r31,-8(r1) 46 50: fb e1 ff f8 std r31,-8(r1) ; |
47 54: f8 01 00 10 std r0,16(r1) 47 54: f8 01 00 10 std r0,16(r1) ; | prolog
48 58: f8 21 ff 71 stdu r1,-144(r1) 48 58: f8 21 ff 71 stdu r1,-144(r1) ; |
49 5c: 7c 3f 0b 78 mr r31,r1 49 5c: 7c 3f 0b 78 mr r31,r1 ; use gpr31 as sort of frame pointer, below
50 60: 7c 60 1b 78 mr r0,r3 50 60: 7c 60 1b 78 mr r0,r3 ; in arg 0 -> gpr0
51 64: 7c 8b 23 78 mr r11,r4 51 64: 7c 8b 23 78 mr r11,r4 ; in arg 1 -> gpr11
52 68: 90 1f 00 c0 stw r0,192(r31) 52 68: 90 1f 00 c0 stw r0,192(r31) ; |
53 6c: 91 7f 00 c8 stw r11,200(r31) 53 6c: 91 7f 00 c8 stw r11,200(r31) ; |
54 70: 90 bf 00 d0 stw r5,208(r31) 54 70: 90 bf 00 d0 stw r5,208(r31) ; |
55 74: 90 df 00 d8 stw r6,216(r31) 55 74: 90 df 00 d8 stw r6,216(r31) ; |
56 78: 90 ff 00 e0 stw r7,224(r31) 56 78: 90 ff 00 e0 stw r7,224(r31) ; | all in args -> spill area in prev frame (jump over own frame (144) + linkage area of prev frame (48) = 192)
57 7c: 91 1f 00 e8 stw r8,232(r31) 57 7c: 91 1f 00 e8 stw r8,232(r31) ; |
58 80: 91 3f 00 f0 stw r9,240(r31) 58 80: 91 3f 00 f0 stw r9,240(r31) ; |
59 84: 91 5f 00 f8 stw r10,248(r31) 59 84: 91 5f 00 f8 stw r10,248(r31) ; |
60 88: e8 01 00 00 ld r0,0(r1) 60 88: e8 01 00 00 ld r0,0(r1) ; fetch back-chain ptr (parent frame's sp) from stack of top by prolog -> gpr0, and ...
61 8c: f8 01 ff 11 stdu r0,-240(r1) 61 8c: f8 01 ff 11 stdu r0,-240(r1) ; ... update it further up the stack for alloca(220) - with padding to guarantee alignment
62 90: 39 21 00 70 addi r9,r1,112 62 90: 39 21 00 70 addi r9,r1,112 ; |
63 94: f9 3f 00 70 std r9,112(r31) 63 94: f9 3f 00 70 std r9,112(r31) ; |
64 98: e9 3f 00 70 ld r9,112(r31) 64 98: e9 3f 00 70 ld r9,112(r31) ; |
65 9c: 38 09 00 0f addi r0,r9,15 65 9c: 38 09 00 0f addi r0,r9,15 ; | start of alloca()'d memory -> gpr9, by ...
66 a0: 78 00 e1 02 rldicl r0,r0,60,4 66 a0: 78 00 e1 02 rldicl r0,r0,60,4 ; | ... using gpr0 as helper to align to 16b, leaving at least 112b at top of stack
67 a4: 78 00 26 e4 rldicr r0,r0,4,59 67 a4: 78 00 26 e4 rldicr r0,r0,4,59 ; |
68 a8: f8 1f 00 70 std r0,112(r31) 68 a8: f8 1f 00 70 std r0,112(r31) ; |
69 ac: e9 3f 00 70 ld r9,112(r31) 69 ac: e9 3f 00 70 ld r9,112(r31) ; |
70 b0: 38 00 00 4c li r0,76 70 b0: 38 00 00 4c li r0,76 ; 'L' -> gpr0, and ...
71 b4: 98 09 00 00 stb r0,0(r9) 71 b4: 98 09 00 00 stb r0,0(r9) ; ... store in local area (of alloca()'d space)
72 b8: 80 1f 00 c8 lwz r0,200(r31) 72 b8: 80 1f 00 c8 lwz r0,200(r31) ; prep arg 0 (from prev frame's spill area), ...
73 bc: 7c 08 07 b4 extsw r8,r0 73 bc: 7c 08 07 b4 extsw r8,r0 ; ... -> gpr8 (w/ sign extension b/c int param in 64bit reg)
74 c0: 80 1f 00 d0 lwz r0,208(r31) 74 c0: 80 1f 00 d0 lwz r0,208(r31) ; prep arg 1 (from prev frame's spill area), ...
75 c4: 7c 07 07 b4 extsw r7,r0 75 c4: 7c 07 07 b4 extsw r7,r0 ; ... -> gpr7
76 c8: 80 1f 00 d8 lwz r0,216(r31) 76 c8: 80 1f 00 d8 lwz r0,216(r31) ; prep arg 2 (from prev frame's spill area), ...
77 cc: 7c 06 07 b4 extsw r6,r0 77 cc: 7c 06 07 b4 extsw r6,r0 ; ... -> gpr6
78 d0: 80 1f 00 e0 lwz r0,224(r31) 78 d0: 80 1f 00 e0 lwz r0,224(r31) ; prep arg 3 (from prev frame's spill area), ...
79 d4: 7c 09 07 b4 extsw r9,r0 79 d4: 7c 09 07 b4 extsw r9,r0 ; ... -> gpr9
80 d8: 80 1f 00 e8 lwz r0,232(r31) 80 d8: 80 1f 00 e8 lwz r0,232(r31) ; prep arg 4 (from prev frame's spill area), ...
81 dc: 7c 0b 07 b4 extsw r11,r0 81 dc: 7c 0b 07 b4 extsw r11,r0 ; ... -> gpr11
82 e0: 80 1f 00 f0 lwz r0,240(r31) 82 e0: 80 1f 00 f0 lwz r0,240(r31) ; prep arg 5 (from prev frame's spill area), ...
83 e4: 7c 0a 07 b4 extsw r10,r0 83 e4: 7c 0a 07 b4 extsw r10,r0 ; ... -> gpr10
84 e8: 80 1f 00 f8 lwz r0,248(r31) 84 e8: 80 1f 00 f8 lwz r0,248(r31) ; prep arg 6 (from prev frame's spill area), ...
85 ec: 7c 00 07 b4 extsw r0,r0 85 ec: 7c 00 07 b4 extsw r0,r0 ; ... -> gpr0
86 f0: 7d 03 43 78 mr r3,r8 86 f0: 7d 03 43 78 mr r3,r8 ; arg 0
87 f4: 7c e4 3b 78 mr r4,r7 87 f4: 7c e4 3b 78 mr r4,r7 ; arg 1
88 f8: 7c c5 33 78 mr r5,r6 88 f8: 7c c5 33 78 mr r5,r6 ; arg 2
89 fc: 7d 26 4b 78 mr r6,r9 89 fc: 7d 26 4b 78 mr r6,r9 ; arg 3
90 100: 7d 67 5b 78 mr r7,r11 90 100: 7d 67 5b 78 mr r7,r11 ; arg 4
91 104: 7d 48 53 78 mr r8,r10 91 104: 7d 48 53 78 mr r8,r10 ; arg 5
92 108: 7c 09 03 78 mr r9,r0 92 108: 7c 09 03 78 mr r9,r0 ; arg 6
93 10c: 48 00 00 01 bl 10c <.nonleaf_call+0xc0> 93 10c: 48 00 00 01 bl 10c <.nonleaf_call+0xc0> ; call and put return address -> lr
94 110: e8 21 00 00 ld r1,0(r1) 94 110: e8 21 00 00 ld r1,0(r1) ; |
95 114: e8 01 00 10 ld r0,16(r1) 95 114: e8 01 00 10 ld r0,16(r1) ; |
96 118: 7c 08 03 a6 mtlr r0 96 118: 7c 08 03 a6 mtlr r0 ; | epilog
97 11c: eb e1 ff f8 ld r31,-8(r1) 97 11c: eb e1 ff f8 ld r31,-8(r1) ; |
98 120: 4e 80 00 20 blr 98 120: 4e 80 00 20 blr ; |
99 124: 00 00 00 00 .long 0x0 99 124: 00 00 00 00 .long 0x0 ; data
100 128: 00 00 00 01 .long 0x1 100 128: 00 00 00 01 .long 0x1 ; data
101 12c: 80 01 00 01 lwz r0,1(r1) 101 12c: 80 01 00 01 lwz r0,1(r1) ; unsure@@@. data?
102 102
103 0000000000000130 <.main>: 103 0000000000000130 <.main>:
104 130: 7c 08 02 a6 mflr r0 104 130: 7c 08 02 a6 mflr r0 ; | lr -> gpr0
105 134: fb e1 ff f8 std r31,-8(r1) 105 134: fb e1 ff f8 std r31,-8(r1) ; | preseve gpr31 (as used in func as helper addr)
106 138: f8 01 00 10 std r0,16(r1) 106 138: f8 01 00 10 std r0,16(r1) ; | prolog store lr
107 13c: f8 21 ff 81 stdu r1,-128(r1) 107 13c: f8 21 ff 81 stdu r1,-128(r1) ; | open frame
108 140: 7c 3f 0b 78 mr r31,r1 108 140: 7c 3f 0b 78 mr r31,r1 ; use gpr31 as sort of frame pointer, below
109 144: 38 60 00 00 li r3,0 109 144: 38 60 00 00 li r3,0 ; arg 0
110 148: 38 80 00 01 li r4,1 110 148: 38 80 00 01 li r4,1 ; arg 1
111 14c: 38 a0 00 02 li r5,2 111 14c: 38 a0 00 02 li r5,2 ; arg 2
112 150: 38 c0 00 03 li r6,3 112 150: 38 c0 00 03 li r6,3 ; arg 3
113 154: 38 e0 00 04 li r7,4 113 154: 38 e0 00 04 li r7,4 ; arg 4
114 158: 39 00 00 05 li r8,5 114 158: 39 00 00 05 li r8,5 ; arg 5
115 15c: 39 20 00 06 li r9,6 115 15c: 39 20 00 06 li r9,6 ; arg 6
116 160: 39 40 00 07 li r10,7 116 160: 39 40 00 07 li r10,7 ; arg 7
117 164: 48 00 00 01 bl 164 <.main+0x34> 117 164: 48 00 00 01 bl 164 <.main+0x34> ; call and put return address -> lr
118 168: 38 00 00 00 li r0,0 118 168: 38 00 00 00 li r0,0 ; return value ...
119 16c: 7c 03 03 78 mr r3,r0 119 16c: 7c 03 03 78 mr r3,r0 ; ... in gpr3
120 170: e8 21 00 00 ld r1,0(r1) 120 170: e8 21 00 00 ld r1,0(r1) ; |
121 174: e8 01 00 10 ld r0,16(r1) 121 174: e8 01 00 10 ld r0,16(r1) ; |
122 178: 7c 08 03 a6 mtlr r0 122 178: 7c 08 03 a6 mtlr r0 ; | epilog
123 17c: eb e1 ff f8 ld r31,-8(r1) 123 17c: eb e1 ff f8 ld r31,-8(r1) ; |
124 180: 4e 80 00 20 blr 124 180: 4e 80 00 20 blr ; |
125 184: 00 00 00 00 .long 0x0 125 184: 00 00 00 00 .long 0x0 ; data
126 188: 00 00 00 01 .long 0x1 126 188: 00 00 00 01 .long 0x1 ; data
127 18c: 80 01 00 01 lwz r0,1(r1) 127 18c: 80 01 00 01 lwz r0,1(r1) ; unsure@@@. data?
128
129
130
131 ; ------------- ints and floats, var args, struct return value (meaning implicit first param), more than 8 params (11, with implicit return value ptr) ----------->
132
133 ; #include <stdlib.h>
134 ; #include <stdarg.h>
135 ;
136 ; void leaf_call(int b, float c, int d, float e, int f, float g, float h, int i, float j)
137 ; {
138 ; }
139 ;
140 ; struct aggr { int x; int y; int z; };
141 ;
142 ; struct aggr nonleaf_call(int a, int b, float c, int d, float e, int f, ...)
143 ; {
144 ; va_list v;
145 ; int i;
146 ; float g, h, j;
147 ; struct aggr st = { b, d, f };
148 ; va_start(v, f);
149 ; g = va_arg(v, float);
150 ; h = va_arg(v, float);
151 ; i = va_arg(v, int);
152 ; h = va_arg(v, float);
153 ; /* use some local data */
154 ; *(char*)alloca(220) = 'L';
155 ; leaf_call(b, c, d, e, f, g, h, i, j);
156 ;
157 ; return st;
158 ; }
159 ;
160 ; int main()
161 ; {
162 ; struct aggr st = nonleaf_call(0, 1, 2.f, 3, 4.f, 5, 6.f, 7.f, 8, 9.f);
163 ; return 0;
164 ; }
165
166
167
168 ; output from freebsd-11.0-ppc64 w/ gcc 4.2.1
169
170 0000000000000000 <.leaf_call>:
171 0: fb e1 ff f8 std r31,-8(r1)
172 4: f8 21 ff c1 stdu r1,-64(r1)
173 8: 7c 3f 0b 78 mr r31,r1
174 c: 7c 69 1b 78 mr r9,r3
175 10: d0 3f 00 78 stfs f1,120(r31)
176 14: 7c ab 2b 78 mr r11,r5
177 18: d0 5f 00 88 stfs f2,136(r31)
178 1c: 7c e8 3b 78 mr r8,r7
179 20: d0 7f 00 98 stfs f3,152(r31)
180 24: d0 9f 00 a0 stfs f4,160(r31)
181 28: 7d 40 53 78 mr r0,r10
182 2c: d0 bf 00 b0 stfs f5,176(r31)
183 30: 91 3f 00 70 stw r9,112(r31)
184 34: 91 7f 00 80 stw r11,128(r31)
185 38: 91 1f 00 90 stw r8,144(r31)
186 3c: 90 1f 00 a8 stw r0,168(r31)
187 40: e8 21 00 00 ld r1,0(r1)
188 44: eb e1 ff f8 ld r31,-8(r1)
189 48: 4e 80 00 20 blr
190 ...
191 54: 80 01 00 01 lwz r0,1(r1)
192
193 0000000000000058 <.nonleaf_call>:
194 58: fb e1 ff f8 std r31,-8(r1) ; |
195 5c: f8 21 ff 91 stdu r1,-112(r1) ; | prolog
196 60: 7c 3f 0b 78 mr r31,r1 ; use gpr31 as sort of frame pointer, below
197 64: 7c 8b 23 78 mr r11,r4 ; in arg 1 (first explicit arg, b/c of struct return value ptr being arg0) -> r11
198 68: 7c a8 2b 78 mr r8,r5 ; in arg 2 -> r8 (free reg, was skipped for float param)
199 6c: d0 3f 00 b8 stfs f1,184(r31) ; | in arg 3 (float) -> prev frame's spill area: 184 = 112 (frame) + 48 (prev frame's linkage area) + 8 (arg 0 = return value ptr) + 16 (first two explicit args)
200 70: d0 5f 00 c8 stfs f2,200(r31) ; | in arg 5 (float) -> prev frame's spill area
201 74: f9 5f 00 d8 std r10,216(r31) ; | in arg 7 (float, also held in gpr reg b/c vararg) -> prev frame's spill area
202 78: 7d 20 4b 78 mr r0,r9 ; | spilling in arg 6 in gpr0 (spilled below)
203 7c: 91 7f 00 a8 stw r11,168(r31) ; | in arg 1 (int) -> prev frame's spill area
204 80: 91 1f 00 b0 stw r8,176(r31) ; | in arg 2 (int) -> prev frame's spill area
205 84: 90 ff 00 c0 stw r7,192(r31) ; | in arg 4 (int) -> prev frame's spill area
206 88: 90 1f 00 d0 stw r0,208(r31) ; / in arg 6 (int) -> prev frame's spill area
207 8c: 80 1f 00 b0 lwz r0,176(r31) ; \
208 90: 90 1f 00 48 stw r0,72(r31) ; |
209 94: 80 1f 00 c0 lwz r0,192(r31) ; |
210 98: 90 1f 00 4c stw r0,76(r31) ; | filling struct with 3 int input args
211 9c: 80 1f 00 d0 lwz r0,208(r31) ; |
212 a0: 90 1f 00 50 stw r0,80(r31) ; |
213 a4: 38 1f 00 d8 addi r0,r31,216 ;
214 a8: f8 1f 00 40 std r0,64(r31) ; .
215 ac: 7f e0 00 08 trap ; .
216 ... ; .
217 b8: 80 01 00 01 lwz r0,1(r1) ;
218
219 00000000000000bc <.main>:
220 bc: 7c 08 02 a6 mflr r0 ; | lr -> gpr0
221 c0: fb e1 ff f8 std r31,-8(r1) ; | preseve gpr31 (as used in func as helper addr)
222 c4: f8 01 00 10 std r0,16(r1) ; | prolog store lr
223 c8: f8 21 ff 41 stdu r1,-192(r1) ; | open frame
224 cc: 7c 3f 0b 78 mr r31,r1 ; use gpr31 as sort of frame pointer, below
225 d0: 39 61 00 30 addi r11,r1,48 ; ptr to param area -> r11
226 d4: e9 22 00 00 ld r9,0(r2) ; prep arg 3 (=explicit arg 2, b/c of implicit return value pointer), ...
227 d8: c1 a9 00 00 lfs f13,0(r9) ; ... load from static data -> f13
228 dc: e9 22 00 08 ld r9,8(r2) ; prep arg 5, ...
229 e0: c1 89 00 00 lfs f12,0(r9) ; ... load from static data -> f12
230 e4: e9 22 00 10 ld r9,16(r2) ; prep arg 7, ...
231 e8: c8 09 00 00 lfd f0,0(r9) ; ... load from static data -> f0
232 ec: d8 1f 00 a0 stfd f0,160(r31) ; |
233 f0: e8 1f 00 a0 ld r0,160(r31) ; |
234 f4: 7c 09 03 78 mr r9,r0 ; | also hold it in f11 (temporarily, before copying to fpr3 below)
235 f8: 7d 2a 4b 78 mr r10,r9 ; | and gpr10 (instead of skipping that int reg, for straightforward spilling)
236 fc: f8 1f 00 a0 std r0,160(r31) ; | (uses temp space to copy between fpr and gpr regs)
237 100: c8 1f 00 a0 lfd f0,160(r31) ; |
238 104: fd 60 00 90 fmr f11,f0 ; |
239 108: e9 22 00 18 ld r9,24(r2) ; prep arg 8, ...
240 10c: c8 09 00 00 lfd f0,0(r9) ; ... load from static data -> fpr0, and ...
241 110: d8 0b 00 40 stfd f0,64(r11) ; ... "pushed" onto stack (in param area past spill area) and ...
242 114: c9 4b 00 40 lfd f10,64(r11) ; ... also held in f10 (prep, see where it's used below)
243 118: 38 00 00 08 li r0,8 ; arg 9, ...
244 11c: f8 0b 00 48 std r0,72(r11) ; ... "pushed" onto stack
245 120: e9 22 00 20 ld r9,32(r2) ; arg 10 (float, promoted to double), ...
246 124: c8 09 00 00 lfd f0,0(r9) ; ... load from static data -> fpr0, and ...
247 128: d8 0b 00 50 stfd f0,80(r11) ; ... "pushed" onto stack
248 12c: c8 0b 00 50 lfd f0,80(r11) ; ... also held in f0 (prep, see where it's used below), in theory pointless reload of arg10 -> fpr0
249 130: 38 1f 00 90 addi r0,r31,144 ; ptr to return value struct in local space -> gpr0
250 134: 7c 03 03 78 mr r3,r0 ; arg 0 (this is the pointer to the struct return value)
251 138: 38 80 00 00 li r4,0 ; arg 1
252 13c: 38 a0 00 01 li r5,1 ; arg 2
253 140: fc 20 68 90 fmr f1,f13 ; arg 3 (float, in 1st double reg)
254 144: 38 e0 00 03 li r7,3 ; arg 4 (skipping gpr6 b/c of float arg)
255 148: fc 40 60 90 fmr f2,f12 ; arg 5 (float, in 2nd double reg)
256 14c: 39 20 00 05 li r9,5 ; arg 6 (skipping gpr8 b/c of float arg, vararg)
257 150: fc 60 58 90 fmr f3,f11 ; arg 7 (float, in 3rd double reg, promoted to double anyways b/c vararg)
258 154: fc 80 50 90 fmr f4,f10 ; arg 8 (float, in 4th double reg, promoted to double anyways b/c vararg)
259 158: fc a0 00 90 fmr f5,f0 ; arg 10 (float, in 5th double reg, promoted to double anyways b/c vararg)
260 15c: 48 00 00 01 bl 15c <.main+0xa0> ; call and put return address -> lr
261 160: 38 00 00 00 li r0,0 ; return value ...
262 164: 7c 03 03 78 mr r3,r0 ; ... in gpr3
263 168: e8 21 00 00 ld r1,0(r1) ; |
264 16c: e8 01 00 10 ld r0,16(r1) ; |
265 170: 7c 08 03 a6 mtlr r0 ; | epilog
266 174: eb e1 ff f8 ld r31,-8(r1) ; |
267 178: 4e 80 00 20 blr ; |
268 17c: 00 00 00 00 .long 0x0 ; data
269 180: 00 00 00 01 .long 0x1 ; data
270 184: 80 01 00 01 lwz r0,1(r1) ; unsure@@@. data?
128 271
129 ; vim: ft=asm 272 ; vim: ft=asm
130 273