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
|
/*--------------------------------------------------------------------*/
/*--- The core dispatch loop, for jumping to a code address. ---*/
/*--- dispatch-arm64-freebsd.S ---*/
/*--------------------------------------------------------------------*/
/*
This file is part of Valgrind, a dynamic binary instrumentation
framework.
Copyright (C) 2024 Paul Floyd
pjfloyd@wanadoo.fr
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, see <http://www.gnu.org/licenses/>.
The GNU General Public License is contained in the file COPYING.
*/
#include "pub_core_basics_asm.h"
#if defined(VGP_arm64_freebsd)
#include "pub_core_dispatch_asm.h"
#include "pub_core_transtab_asm.h"
#include "libvex_guest_offsets.h" /* for OFFSET_arm_R* */
/*------------------------------------------------------------*/
/*--- ---*/
/*--- The dispatch loop. VG_(disp_run_translations) is ---*/
/*--- used to run all translations, ---*/
/*--- including no-redir ones. ---*/
/*--- ---*/
/*------------------------------------------------------------*/
/*----------------------------------------------------*/
/*--- Entry and preamble (set everything up) ---*/
/*----------------------------------------------------*/
/* signature:
void VG_(disp_run_translations)( UWord* two_words,
void* guest_state,
Addr host_addr );
*/
.text
.global VG_(disp_run_translations)
VG_(disp_run_translations):
/* x0 holds two_words
x1 holds guest_state
x2 holds host_addr
*/
/* Push the callee-saved registers. Unclear if x19/x20 are
callee-saved, but be on the safe side. Note this sequence
maintains 16-alignment of sp. Also save x0 since it will
be needed in the postamble. */
stp x29, x30, [sp, #-16]!
stp x27, x28, [sp, #-16]!
stp x25, x26, [sp, #-16]!
stp x23, x24, [sp, #-16]!
stp x21, x22, [sp, #-16]!
stp x19, x20, [sp, #-16]!
stp x0, xzr, [sp, #-16]!
/* set FPSCR to vex-required default value */
// FIXME
// mov r4, #0
// fmxr fpscr, r4
/* Set up the guest state pointer */
mov x21, x1
/* and jump into the code cache. Chained translations in
the code cache run, until for whatever reason, they can't
continue. When that happens, the translation in question
will jump (or call) to one of the continuation points
VG_(cp_...) below. */
br x2
/* NOTREACHED */
/*----------------------------------------------------*/
/*--- Postamble and exit. ---*/
/*----------------------------------------------------*/
postamble:
/* At this point, r1 and r2 contain two
words to be returned to the caller. r1
holds a TRC value, and r2 optionally may
hold another word (for CHAIN_ME exits, the
address of the place to patch.) */
/* We're leaving. Check that nobody messed with
FPSCR in ways we don't expect. */
// FIXME
// fmrx r4, fpscr
// bic r4, #0xF8000000 /* mask out NZCV and QC */
// bic r4, #0x0000009F /* mask out IDC,IXC,UFC,OFC,DZC,IOC */
// cmp r4, #0
// beq remove_frame /* we're OK */
/* otherwise we have an invariant violation */
// movw r1, #VG_TRC_INVARIANT_FAILED
// movw r2, #0
/* fall through */
remove_frame:
/* Restore int regs, including importantly x0 (two_words),
but not x1 */
ldp x0, xzr, [sp], #16
ldp x19, x20, [sp], #16
ldp x21, x22, [sp], #16
ldp x23, x24, [sp], #16
ldp x25, x26, [sp], #16
ldp x27, x28, [sp], #16
ldp x29, x30, [sp], #16
/* Stash return values */
str x1, [x0, #0]
str x2, [x0, #8]
ret
/*----------------------------------------------------*/
/*--- Continuation points ---*/
/*----------------------------------------------------*/
/* ------ Chain me to slow entry point ------ */
.global VG_(disp_cp_chain_me_to_slowEP)
VG_(disp_cp_chain_me_to_slowEP):
/* We got called. The return address indicates
where the patching needs to happen. Collect
the return address and, exit back to C land,
handing the caller the pair (Chain_me_S, RA) */
mov x1, #VG_TRC_CHAIN_ME_TO_SLOW_EP
mov x2, x30 // 30 == LR
/* 4 = movw x9, disp_cp_chain_me_to_slowEP[15:0]
4 = movk x9, disp_cp_chain_me_to_slowEP[31:16], lsl 16
4 = movk x9, disp_cp_chain_me_to_slowEP[47:32], lsl 32
4 = movk x9, disp_cp_chain_me_to_slowEP[63:48], lsl 48
4 = blr x9
*/
sub x2, x2, #4+4+4+4+4
b postamble
/* ------ Chain me to fast entry point ------ */
.global VG_(disp_cp_chain_me_to_fastEP)
VG_(disp_cp_chain_me_to_fastEP):
/* We got called. The return address indicates
where the patching needs to happen. Collect
the return address and, exit back to C land,
handing the caller the pair (Chain_me_F, RA) */
mov x1, #VG_TRC_CHAIN_ME_TO_FAST_EP
mov x2, x30 // 30 == LR
/* 4 = movw x9, disp_cp_chain_me_to_fastEP[15:0]
4 = movk x9, disp_cp_chain_me_to_fastEP[31:16], lsl 16
4 = movk x9, disp_cp_chain_me_to_fastEP[47:32], lsl 32
4 = movk x9, disp_cp_chain_me_to_fastEP[63:48], lsl 48
4 = blr x9
*/
sub x2, x2, #4+4+4+4+4
b postamble
/* ------ Indirect but boring jump ------ */
.global VG_(disp_cp_xindir)
VG_(disp_cp_xindir):
// Where are we going?
ldr x0, [x21, #OFFSET_arm64_PC]
// stats only
adrp x4, VG_(stats__n_xIndirs_32)
add x4, x4, :lo12:VG_(stats__n_xIndirs_32)
ldr w5, [x4, #0]
add w5, w5, #1
str w5, [x4, #0]
// LIVE: x21 (guest state ptr), x0 (guest address to go to).
// We use 6 temporaries:
// x6 (to point at the relevant FastCacheSet),
// x1, x2, x3 (scratch, for swapping entries within a set)
// x4, x5 (other scratch)
/* Try a fast lookup in the translation cache. This is pretty much
a handcoded version of VG_(lookupInFastCache). */
// Compute x6 = VG_TT_FAST_HASH(guest)
lsr x6, x0, #2 // g2 = guest >> 2
eor x6, x6, x6, LSR #VG_TT_FAST_BITS // (g2 >> VG_TT_FAST_BITS) ^ g2
mov x4, #VG_TT_FAST_MASK // VG_TT_FAST_MASK
and x6, x6, x4 // setNo
// Compute x6 = &VG_(tt_fast)[x6]
adrp x4, VG_(tt_fast)
add x4, x4, :lo12:VG_(tt_fast) // &VG_(tt_fast)[0]
add x6, x4, x6, LSL #VG_FAST_CACHE_SET_BITS // &VG_(tt_fast)[setNo]
// LIVE: x21 (guest state ptr), x0 (guest addr), x6 (cache set)
// try way 0
ldp x4, x5, [x6, #FCS_g0] // x4 = .guest0, x5 = .host0
cmp x4, x0 // cmp against .guest0
bne 1f
// hit at way 0
// goto .host0
br x5
/*NOTREACHED*/
1: // try way 1
ldr x4, [x6, #FCS_g1]
cmp x4, x0 // cmp against .guest1
bne 2f
// hit at way 1; swap upwards
ldr x1, [x6, #FCS_g0] // x1 = old .guest0
ldr x2, [x6, #FCS_h0] // x2 = old .host0
ldr x3, [x6, #FCS_h1] // x3 = old .host1
str x0, [x6, #FCS_g0] // new .guest0 = guest
str x3, [x6, #FCS_h0] // new .host0 = old .host1
str x1, [x6, #FCS_g1] // new .guest1 = old .guest0
str x2, [x6, #FCS_h1] // new .host1 = old .host0
// stats only
adrp x4, VG_(stats__n_xIndir_hits1_32)
add x4, x4, :lo12:VG_(stats__n_xIndir_hits1_32)
ldr w5, [x4, #0]
add w5, w5, #1
str w5, [x4, #0]
// goto old .host1 a.k.a. new .host0
br x3
/*NOTREACHED*/
2: // try way 2
ldr x4, [x6, #FCS_g2]
cmp x4, x0 // cmp against .guest2
bne 3f
// hit at way 2; swap upwards
ldr x1, [x6, #FCS_g1]
ldr x2, [x6, #FCS_h1]
ldr x3, [x6, #FCS_h2]
str x0, [x6, #FCS_g1]
str x3, [x6, #FCS_h1]
str x1, [x6, #FCS_g2]
str x2, [x6, #FCS_h2]
// stats only
adrp x4, VG_(stats__n_xIndir_hits2_32)
add x4, x4, :lo12:VG_(stats__n_xIndir_hits2_32)
ldr w5, [x4, #0]
add w5, w5, #1
str w5, [x4, #0]
// goto old .host2 a.k.a. new .host1
br x3
/*NOTREACHED*/
3: // try way 3
ldr x4, [x6, #FCS_g3]
cmp x4, x0 // cmp against .guest3
bne 4f
// hit at way 3; swap upwards
ldr x1, [x6, #FCS_g2]
ldr x2, [x6, #FCS_h2]
ldr x3, [x6, #FCS_h3]
str x0, [x6, #FCS_g2]
str x3, [x6, #FCS_h2]
str x1, [x6, #FCS_g3]
str x2, [x6, #FCS_h3]
// stats only
adrp x4, VG_(stats__n_xIndir_hits3_32)
add x4, x4, :lo12:VG_(stats__n_xIndir_hits3_32)
ldr w5, [x4, #0]
add w5, w5, #1
str w5, [x4, #0]
// goto old .host3 a.k.a. new .host2
br x3
/*NOTREACHED*/
4: // fast lookup failed
adrp x4, VG_(stats__n_xIndir_misses_32)
add x4, x4, :lo12:VG_(stats__n_xIndir_misses_32)
ldr w5, [x4, #0]
add w5, w5, #1
str w5, [x4, #0]
mov x1, #VG_TRC_INNER_FASTMISS
mov x2, #0
b postamble
/* ------ Assisted jump ------ */
.global VG_(disp_cp_xassisted)
VG_(disp_cp_xassisted):
/* x21 contains the TRC */
mov x1, x21
mov x2, #0
b postamble
/* ------ Event check failed ------ */
.global VG_(disp_cp_evcheck_fail)
VG_(disp_cp_evcheck_fail):
mov x1, #VG_TRC_INNER_COUNTERZERO
mov x2, #0
b postamble
.size VG_(disp_run_translations), .-VG_(disp_run_translations)
#endif // defined(VGP_arm64_freebsd)
/* Let the linker know we don't need an executable stack */
MARK_STACK_NO_EXEC
/*--------------------------------------------------------------------*/
/*--- end dispatch-arm64-freebsd.S ---*/
/*--------------------------------------------------------------------*/
|
