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
331
332
333
334
335
336
337
338
339
340
341
342
343
344
|
/*--------------------------------------------------------------------*/
/*--- The core dispatch loop, for jumping to a code address. ---*/
/*--- dispatch-amd64-linux.S ---*/
/*--------------------------------------------------------------------*/
/*
This file is part of Valgrind, a dynamic binary instrumentation
framework.
Copyright (C) 2000-2009 Julian Seward
jseward@acm.org
This program is free software; you can redistribute it and/or
modify it under the terms of the GNU General Public License as
published by the Free Software Foundation; either version 2 of the
License, or (at your option) any later version.
This program is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
02111-1307, USA.
The GNU General Public License is contained in the file COPYING.
*/
#include "pub_core_basics_asm.h"
#include "pub_core_dispatch_asm.h"
#include "pub_core_transtab_asm.h"
#include "libvex_guest_offsets.h" /* for OFFSET_amd64_RIP */
/*------------------------------------------------------------*/
/*--- ---*/
/*--- The dispatch loop. VG_(run_innerloop) is used to ---*/
/*--- run all translations except no-redir ones. ---*/
/*--- ---*/
/*------------------------------------------------------------*/
/*----------------------------------------------------*/
/*--- Preamble (set everything up) ---*/
/*----------------------------------------------------*/
/* signature:
UWord VG_(run_innerloop) ( void* guest_state, UWord do_profiling );
*/
.text
.globl VG_(run_innerloop)
.type VG_(run_innerloop), @function
VG_(run_innerloop):
/* %rdi holds guest_state */
/* %rsi holds do_profiling */
/* ----- entry point to VG_(run_innerloop) ----- */
pushq %rbx
pushq %rcx
pushq %rdx
pushq %rsi
pushq %rbp
pushq %r8
pushq %r9
pushq %r10
pushq %r11
pushq %r12
pushq %r13
pushq %r14
pushq %r15
pushq %rdi /* guest_state */
movq VG_(dispatch_ctr)@GOTPCREL(%rip), %r15
movl (%r15), %r15d
pushq %r15
/* 8(%rsp) holds cached copy of guest_state ptr */
/* 0(%rsp) holds cached copy of VG_(dispatch_ctr) */
/* Set up the guest state pointer */
movq %rdi, %rbp
/* fetch %RIP into %rax */
movq OFFSET_amd64_RIP(%rbp), %rax
/* set host FPU control word to the default mode expected
by VEX-generated code. See comments in libvex.h for
more info. */
finit
pushq $0x027F
fldcw (%rsp)
addq $8, %rsp
/* set host SSE control word to the default mode expected
by VEX-generated code. */
pushq $0x1F80
ldmxcsr (%rsp)
addq $8, %rsp
/* set dir flag to known value */
cld
/* fall into main loop (the right one) */
cmpq $0, %rsi
je VG_(run_innerloop__dispatch_unprofiled)
jmp VG_(run_innerloop__dispatch_profiled)
/*NOTREACHED*/
/*----------------------------------------------------*/
/*--- NO-PROFILING (standard) dispatcher ---*/
/*----------------------------------------------------*/
.align 16
.global VG_(run_innerloop__dispatch_unprofiled)
VG_(run_innerloop__dispatch_unprofiled):
/* AT ENTRY: %rax is next guest addr, %rbp is possibly
modified guest state ptr */
/* Has the guest state pointer been messed with? If yes, exit.
Also, set %rcx to be &VG_(tt_fast), some insns before it is
used, in the hope of getting it off the critical path. This
location seems to be optimal on 2.2GHz Athlon64. */
cmpq 8(%rsp), %rbp
movq VG_(tt_fast)@GOTPCREL(%rip), %rcx
jnz gsp_changed
/* save the jump address in the guest state */
movq %rax, OFFSET_amd64_RIP(%rbp)
/* Are we out of timeslice? If yes, defer to scheduler. */
subl $1, 0(%rsp)
jz counter_is_zero
/* try a fast lookup in the translation cache */
movq %rax, %rbx /* next guest addr */
andq $VG_TT_FAST_MASK, %rbx /* entry# */
shlq $4, %rbx /* entry# * sizeof(FastCacheEntry) */
movq 0(%rcx,%rbx,1), %r10 /* .guest */
movq 8(%rcx,%rbx,1), %r11 /* .host */
cmpq %rax, %r10
jnz fast_lookup_failed
/* Found a match. Jump to .host. */
jmp *%r11
ud2 /* persuade insn decoders not to speculate past here */
/* generated code should run, then jump back to
VG_(run_innerloop__dispatch_unprofiled). */
/*NOTREACHED*/
/*----------------------------------------------------*/
/*--- PROFILING dispatcher (can be much slower) ---*/
/*----------------------------------------------------*/
.align 16
.global VG_(run_innerloop__dispatch_profiled)
VG_(run_innerloop__dispatch_profiled):
/* AT ENTRY: %rax is next guest addr, %rbp is possibly
modified guest state ptr */
/* Has the guest state pointer been messed with? If yes, exit.
Also, set %rcx to be &VG_(tt_fast), some insns before it is
used, in the hope of getting it off the critical path. This
location seems to be optimal on 2.2GHz Athlon64. */
cmpq 8(%rsp), %rbp
movq VG_(tt_fast)@GOTPCREL(%rip), %rcx
jnz gsp_changed
/* save the jump address in the guest state */
movq %rax, OFFSET_amd64_RIP(%rbp)
/* Are we out of timeslice? If yes, defer to scheduler. */
subl $1, 0(%rsp)
jz counter_is_zero
/* try a fast lookup in the translation cache */
movq %rax, %rbx
andq $VG_TT_FAST_MASK, %rbx /* entry# */
shlq $4, %rbx /* entry# * sizeof(FastCacheEntry) */
movq 0(%rcx,%rbx,1), %r10 /* .guest */
movq 8(%rcx,%rbx,1), %r11 /* .host */
cmpq %rax, %r10
jnz fast_lookup_failed
/* increment bb profile counter */
movq VG_(tt_fastN)@GOTPCREL(%rip), %rdx
shrq $1, %rbx /* entry# * sizeof(UInt*) */
movq (%rdx,%rbx,1), %rdx
addl $1, (%rdx)
/* Found a match. Jump to .host. */
jmp *%r11
ud2 /* persuade insn decoders not to speculate past here */
/* generated code should run, then jump back to
VG_(run_innerloop__dispatch_profiled). */
/*NOTREACHED*/
/*----------------------------------------------------*/
/*--- exit points ---*/
/*----------------------------------------------------*/
gsp_changed:
/* Someone messed with the gsp. Have to
defer to scheduler to resolve this. dispatch ctr
is not yet decremented, so no need to increment. */
/* %RIP is NOT up to date here. First, need to write
%rax back to %RIP, but without trashing %rbp since
that holds the value we want to return to the scheduler.
Hence use %r15 transiently for the guest state pointer. */
movq 8(%rsp), %r15
movq %rax, OFFSET_amd64_RIP(%r15)
movq %rbp, %rax
jmp run_innerloop_exit
/*NOTREACHED*/
counter_is_zero:
/* %RIP is up to date here */
/* back out decrement of the dispatch counter */
addl $1, 0(%rsp)
movq $VG_TRC_INNER_COUNTERZERO, %rax
jmp run_innerloop_exit
fast_lookup_failed:
/* %RIP is up to date here */
/* back out decrement of the dispatch counter */
addl $1, 0(%rsp)
movq $VG_TRC_INNER_FASTMISS, %rax
jmp run_innerloop_exit
/* All exits from the dispatcher go through here. %rax holds
the return value.
*/
run_innerloop_exit:
/* We're leaving. Check that nobody messed with
%mxcsr or %fpucw. We can't mess with %rax here as it
holds the tentative return value, but any other is OK. */
#if !defined(ENABLE_INNER)
/* This check fails for self-hosting, so skip in that case */
pushq $0
fstcw (%rsp)
cmpl $0x027F, (%rsp)
popq %r15 /* get rid of the word without trashing %eflags */
jnz invariant_violation
#endif
pushq $0
stmxcsr (%rsp)
andl $0xFFFFFFC0, (%rsp) /* mask out status flags */
cmpl $0x1F80, (%rsp)
popq %r15
jnz invariant_violation
/* otherwise we're OK */
jmp run_innerloop_exit_REALLY
invariant_violation:
movq $VG_TRC_INVARIANT_FAILED, %rax
jmp run_innerloop_exit_REALLY
run_innerloop_exit_REALLY:
/* restore VG_(dispatch_ctr) */
popq %r14
movq VG_(dispatch_ctr)@GOTPCREL(%rip), %r15
movl %r14d, (%r15)
popq %rdi
popq %r15
popq %r14
popq %r13
popq %r12
popq %r11
popq %r10
popq %r9
popq %r8
popq %rbp
popq %rsi
popq %rdx
popq %rcx
popq %rbx
ret
.size VG_(run_innerloop), .-VG_(run_innerloop)
/*------------------------------------------------------------*/
/*--- ---*/
/*--- A special dispatcher, for running no-redir ---*/
/*--- translations. Just runs the given translation once. ---*/
/*--- ---*/
/*------------------------------------------------------------*/
/* signature:
void VG_(run_a_noredir_translation) ( UWord* argblock );
*/
/* Run a no-redir translation. argblock points to 4 UWords, 2 to carry args
and 2 to carry results:
0: input: ptr to translation
1: input: ptr to guest state
2: output: next guest PC
3: output: guest state pointer afterwards (== thread return code)
*/
.align 16
.global VG_(run_a_noredir_translation)
.type VG_(run_a_noredir_translation), @function
VG_(run_a_noredir_translation):
/* Save callee-saves regs */
pushq %rbx
pushq %rbp
pushq %r12
pushq %r13
pushq %r14
pushq %r15
pushq %rdi /* we will need it after running the translation */
movq 8(%rdi), %rbp
jmp *0(%rdi)
/*NOTREACHED*/
ud2
/* If the translation has been correctly constructed, we
should resume at the the following label. */
.global VG_(run_a_noredir_translation__return_point)
VG_(run_a_noredir_translation__return_point):
popq %rdi
movq %rax, 16(%rdi)
movq %rbp, 24(%rdi)
popq %r15
popq %r14
popq %r13
popq %r12
popq %rbp
popq %rbx
ret
.size VG_(run_a_noredir_translation), .-VG_(run_a_noredir_translation)
/* Let the linker know we don't need an executable stack */
.section .note.GNU-stack,"",@progbits
/*--------------------------------------------------------------------*/
/*--- end ---*/
/*--------------------------------------------------------------------*/
|
