1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
|
diff --git a/lib/x86_64/efi_stub.S b/lib/x86_64/efi_stub.S
index a5e278b..b431255 100644
--- a/lib/x86_64/efi_stub.S
+++ b/lib/x86_64/efi_stub.S
@@ -4,6 +4,7 @@
* Copyright (C) 2007 Intel Corp
* Bibo Mao <bibo.mao@intel.com>
* Huang Ying <ying.huang@intel.com>
+ * Copyright (C) 2012 Felipe Contreras <felipe.contreras@gmail.com>
*/
#if !defined(HAVE_USE_MS_ABI)
@@ -15,152 +16,94 @@
*
* Basically here are the conversion rules:
* a) our function pointer is in %rdi
- * b) ELF gives us 8-byte aligned %rsp, so we need to pad out to 16-byte
- * alignment.
- * c) inside each call thunker, we can only adjust the stack by
- * multiples of 16 bytes. "offset" below refers to however much
- * we allocate inside a thunker.
- * d) rsi through r8 (elf) aka rcx through r9 (ms) require stack space
+ * b) rsi through r8 (elf) aka rcx through r9 (ms) require stack space
* on the MS side even though it's not getting used at all.
- * e) arguments are as follows: (elf -> ms)
+ * c) 8(%rsp) is always aligned to 16 in ELF, so %rsp is shifted 8 bytes extra
+ * d) arguments are as follows: (elf -> ms)
* 1) rdi -> rcx (32 saved)
* 2) rsi -> rdx (32 saved)
- * 3) rdx -> r8 ( 32 saved)
+ * 3) rdx -> r8 (32 saved)
* 4) rcx -> r9 (32 saved)
- * 5) r8 -> 32(%rsp) (48 saved)
+ * 5) r8 -> 32(%rsp) (32 saved)
* 6) r9 -> 40(%rsp) (48 saved)
- * 7) pad+offset+0(%rsp) -> 48(%rsp) (64 saved)
- * 8) pad+offset+8(%rsp) -> 56(%rsp) (64 saved)
- * 9) pad+offset+16(%rsp) -> 64(%rsp) (80 saved)
- * 10) pad+offset+24(%rsp) -> 72(%rsp) (80 saved)
- * 11) pad+offset+32(%rsp) -> 80(%rsp) (96 saved)
- * 12) pad+offset+40(%rsp) -> 88(%rsp) (96 saved)
- * f) because the first argument we recieve in a thunker is actually the
+ * 7) 8(%rsp) -> 48(%rsp) (48 saved)
+ * 8) 16(%rsp) -> 56(%rsp) (64 saved)
+ * 9) 24(%rsp) -> 64(%rsp) (64 saved)
+ * 10) 32(%rsp) -> 72(%rsp) (80 saved)
+ * e) because the first argument we recieve in a thunker is actually the
* function to be called, arguments are offset as such:
* 0) rdi -> caller
* 1) rsi -> rcx (32 saved)
* 2) rdx -> rdx (32 saved)
* 3) rcx -> r8 (32 saved)
* 4) r8 -> r9 (32 saved)
- * 5) r9 -> 32(%rsp) (48 saved)
- * 6) pad+offset+0(%rsp) -> 40(%rsp) (48 saved)
- * 7) pad+offset+8(%rsp) -> 48(%rsp) (64 saved)
- * 8) pad+offset+16(%rsp) -> 56(%rsp) (64 saved)
- * 9) pad+offset+24(%rsp) -> 64(%rsp) (80 saved)
- * 10) pad+offset+32(%rsp) -> 72(%rsp) (80 saved)
- * 11) pad+offset+40(%rsp) -> 80(%rsp) (96 saved)
- * 12) pad+offset+48(%rsp) -> 88(%rsp) (96 saved)
- * e) arguments need to be moved in opposite order to avoid clobbering
- * f) pad_stack leaves the amount of padding it added in %r11 for functions
- * to use
- * g) efi -> elf calls don't need to pad the stack, because the 16-byte
- * alignment is also always 8-byte aligned.
+ * 5) r9 -> 32(%rsp) (32 saved)
+ * 6) 8(%rsp) -> 40(%rsp) (48 saved)
+ * 7) 16(%rsp) -> 48(%rsp) (48 saved)
+ * 8) 24(%rsp) -> 56(%rsp) (64 saved)
+ * 9) 32(%rsp) -> 64(%rsp) (64 saved)
+ * 10) 40(%rsp) -> 72(%rsp) (80 saved)
+ * f) arguments need to be moved in opposite order to avoid clobbering
*/
#define ENTRY(name) \
.globl name; \
name:
-#define out(val) \
- push %rax ; \
- mov val, %rax ; \
- out %al, $128 ; \
- pop %rax
-
-#define pad_stack \
- subq $8, %rsp ; /* must be a multiple of 16 - sizeof(%rip) */ \
- /* stash some handy integers */ \
- mov $0x8, %rax ; \
- mov $0x10, %r10 ; \
- /* see if we need padding */ \
- and %rsp, %rax ; \
- /* store the pad amount in %r11 */ \
- cmovnz %rax, %r11 ; \
- cmovz %r10, %r11 ; \
- /* insert the padding */ \
- subq %r11, %rsp ; \
- /* add the $8 we saved above in %r11 */ \
- addq $8, %r11 ; \
- /* store the pad amount */ \
- mov %r11, (%rsp) ; \
- /* compensate for %rip being stored on the stack by call */ \
- addq $8, %r11
-
-#define unpad_stack \
- /* fetch the pad amount we saved (%r11 has been clobbered) */ \
- mov (%rsp), %r11 ; \
- /* remove the padding */ \
- addq %r11, %rsp
-
ENTRY(efi_call0)
- pad_stack
- subq $32, %rsp
+ subq $40, %rsp
call *%rdi
- addq $32, %rsp
- unpad_stack
+ addq $40, %rsp
ret
ENTRY(efi_call1)
- pad_stack
- subq $32, %rsp
+ subq $40, %rsp
mov %rsi, %rcx
call *%rdi
- addq $32, %rsp
- unpad_stack
+ addq $40, %rsp
ret
ENTRY(efi_call2)
- pad_stack
- subq $32, %rsp
+ subq $40, %rsp
/* mov %rdx, %rdx */
mov %rsi, %rcx
call *%rdi
- addq $32, %rsp
- unpad_stack
+ addq $40, %rsp
ret
ENTRY(efi_call3)
- pad_stack
- subq $32, %rsp
+ subq $40, %rsp
mov %rcx, %r8
/* mov %rdx, %rdx */
mov %rsi, %rcx
call *%rdi
- addq $32, %rsp
- unpad_stack
+ addq $40, %rsp
ret
ENTRY(efi_call4)
- pad_stack
- subq $32, %rsp
+ subq $40, %rsp
mov %r8, %r9
mov %rcx, %r8
/* mov %rdx, %rdx */
mov %rsi, %rcx
call *%rdi
- addq $32, %rsp
- unpad_stack
+ addq $40, %rsp
ret
ENTRY(efi_call5)
- pad_stack
- subq $48, %rsp
+ subq $40, %rsp
mov %r9, 32(%rsp)
mov %r8, %r9
mov %rcx, %r8
/* mov %rdx, %rdx */
mov %rsi, %rcx
call *%rdi
- addq $48, %rsp
- unpad_stack
+ addq $40, %rsp
ret
ENTRY(efi_call6)
- pad_stack
- subq $48, %rsp
- addq $48, %r11
- addq %rsp, %r11
- mov (%r11), %rax
+ subq $56, %rsp
+ mov 56+8(%rsp), %rax
mov %rax, 40(%rsp)
mov %r9, 32(%rsp)
mov %r8, %r9
@@ -168,20 +111,14 @@ ENTRY(efi_call6)
/* mov %rdx, %rdx */
mov %rsi, %rcx
call *%rdi
- addq $48, %rsp
- unpad_stack
+ addq $56, %rsp
ret
ENTRY(efi_call7)
- pad_stack
- subq $64, %rsp
- addq $64, %r11
- addq $8, %r11
- addq %rsp, %r11
- mov (%r11), %rax
+ subq $56, %rsp
+ mov 56+16(%rsp), %rax
mov %rax, 48(%rsp)
- subq $8, %r11
- mov (%r11), %rax
+ mov 56+8(%rsp), %rax
mov %rax, 40(%rsp)
mov %r9, 32(%rsp)
mov %r8, %r9
@@ -189,23 +126,16 @@ ENTRY(efi_call7)
/* mov %rdx, %rdx */
mov %rsi, %rcx
call *%rdi
- addq $64, %rsp
- unpad_stack
+ addq $56, %rsp
ret
ENTRY(efi_call8)
- pad_stack
- subq $64, %rsp
- addq $64, %r11
- addq $16, %r11
- addq %rsp, %r11
- mov (%r11), %rax
+ subq $72, %rsp
+ mov 72+24(%rsp), %rax
mov %rax, 56(%rsp)
- subq $8, %r11
- mov (%r11), %rax
+ mov 72+16(%rsp), %rax
mov %rax, 48(%rsp)
- subq $8, %r11
- mov (%r11), %rax
+ mov 72+8(%rsp), %rax
mov %rax, 40(%rsp)
mov %r9, 32(%rsp)
mov %r8, %r9
@@ -213,26 +143,18 @@ ENTRY(efi_call8)
/* mov %rdx, %rdx */
mov %rsi, %rcx
call *%rdi
- addq $64, %rsp
- unpad_stack
+ addq $72, %rsp
ret
ENTRY(efi_call9)
- pad_stack
- subq $80, %rsp
- addq $80, %r11
- addq $24, %r11
- addq %rsp, %r11
- mov (%r11), %rax
+ subq $72, %rsp
+ mov 72+32(%rsp), %rax
mov %rax, 64(%rsp)
- subq $8, %r11
- mov (%r11), %rax
+ mov 72+24(%rsp), %rax
mov %rax, 56(%rsp)
- subq $8, %r11
- mov (%r11), %rax
+ mov 72+16(%rsp), %rax
mov %rax, 48(%rsp)
- subq $8, %r11
- mov (%r11), %rax
+ mov 72+8(%rsp), %rax
mov %rax, 40(%rsp)
mov %r9, 32(%rsp)
mov %r8, %r9
@@ -240,29 +162,20 @@ ENTRY(efi_call9)
/* mov %rdx, %rdx */
mov %rsi, %rcx
call *%rdi
- addq $80, %rsp
- unpad_stack
+ addq $72, %rsp
ret
ENTRY(efi_call10)
- pad_stack
- subq $80, %rsp
- addq $80, %r11
- addq $32, %r11
- addq %rsp, %r11
- mov (%r11), %rax
+ subq $88, %rsp
+ mov 88+40(%rsp), %rax
mov %rax, 72(%rsp)
- subq $8, %r11
- mov (%r11), %rax
+ mov 88+32(%rsp), %rax
mov %rax, 64(%rsp)
- subq $8, %r11
- mov (%r11), %rax
+ mov 88+24(%rsp), %rax
mov %rax, 56(%rsp)
- subq $8, %r11
- mov (%r11), %rax
+ mov 88+16(%rsp), %rax
mov %rax, 48(%rsp)
- subq $8, %r11
- mov (%r11), %rax
+ mov 88+8(%rsp), %rax
mov %rax, 40(%rsp)
mov %r9, 32(%rsp)
mov %r8, %r9
@@ -270,8 +183,7 @@ ENTRY(efi_call10)
/* mov %rdx, %rdx */
mov %rsi, %rcx
call *%rdi
- addq $80, %rsp
- unpad_stack
+ addq $88, %rsp
ret
#endif
|