summaryrefslogtreecommitdiffstats
path: root/dyngen.h
blob: 2a87c448fd3f22f89c6681f6bff1f0c956736380 (plain)
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
/*
 * dyngen helpers
 * 
 *  Copyright (c) 2003 Fabrice Bellard
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2 of the License, or (at your option) any later version.
 *
 * This library 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
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with this library; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 */

int __op_param1, __op_param2, __op_param3;
#if defined(__sparc__) || defined(__arm__)
  void __op_gen_label1(){}
  void __op_gen_label2(){}
  void __op_gen_label3(){}
#else
  int __op_gen_label1, __op_gen_label2, __op_gen_label3;
#endif
int __op_jmp0, __op_jmp1, __op_jmp2, __op_jmp3;

#ifdef __i386__
static inline void flush_icache_range(unsigned long start, unsigned long stop)
{
}
#endif

#ifdef __x86_64__
static inline void flush_icache_range(unsigned long start, unsigned long stop)
{
}
#endif

#ifdef __s390__
static inline void flush_icache_range(unsigned long start, unsigned long stop)
{
}
#endif

#ifdef __ia64__
static inline void flush_icache_range(unsigned long start, unsigned long stop)
{
    while (start < stop) {
	asm volatile ("fc %0" :: "r"(start));
	start += 32;
    }
    asm volatile (";;sync.i;;srlz.i;;");
}
#endif

#ifdef __powerpc__

#define MIN_CACHE_LINE_SIZE 8 /* conservative value */

static void inline flush_icache_range(unsigned long start, unsigned long stop)
{
    unsigned long p;

    start &= ~(MIN_CACHE_LINE_SIZE - 1);
    stop = (stop + MIN_CACHE_LINE_SIZE - 1) & ~(MIN_CACHE_LINE_SIZE - 1);
    
    for (p = start; p < stop; p += MIN_CACHE_LINE_SIZE) {
        asm volatile ("dcbst 0,%0" : : "r"(p) : "memory");
    }
    asm volatile ("sync" : : : "memory");
    for (p = start; p < stop; p += MIN_CACHE_LINE_SIZE) {
        asm volatile ("icbi 0,%0" : : "r"(p) : "memory");
    }
    asm volatile ("sync" : : : "memory");
    asm volatile ("isync" : : : "memory");
}
#endif

#ifdef __alpha__
static inline void flush_icache_range(unsigned long start, unsigned long stop)
{
    asm ("imb");
}
#endif

#ifdef __sparc__

static void inline flush_icache_range(unsigned long start, unsigned long stop)
{
	unsigned long p;

	p = start & ~(8UL - 1UL);
	stop = (stop + (8UL - 1UL)) & ~(8UL - 1UL);

	for (; p < stop; p += 8)
		__asm__ __volatile__("flush\t%0" : : "r" (p));
}

#endif

#ifdef __arm__
static inline void flush_icache_range(unsigned long start, unsigned long stop)
{
    register unsigned long _beg __asm ("a1") = start;
    register unsigned long _end __asm ("a2") = stop;
    register unsigned long _flg __asm ("a3") = 0;
    __asm __volatile__ ("swi 0x9f0002" : : "r" (_beg), "r" (_end), "r" (_flg));
}
#endif

#ifdef __mc68000
#include <asm/cachectl.h>
static inline void flush_icache_range(unsigned long start, unsigned long stop)
{
    cacheflush(start,FLUSH_SCOPE_LINE,FLUSH_CACHE_BOTH,stop-start+16);
}
#endif

#ifdef __alpha__

register int gp asm("$29");

static inline void immediate_ldah(void *p, int val) {
    uint32_t *dest = p;
    long high = ((val >> 16) + ((val >> 15) & 1)) & 0xffff;

    *dest &= ~0xffff;
    *dest |= high;
    *dest |= 31 << 16;
}
static inline void immediate_lda(void *dest, int val) {
    *(uint16_t *) dest = val;
}
void fix_bsr(void *p, int offset) {
    uint32_t *dest = p;
    *dest &= ~((1 << 21) - 1);
    *dest |= (offset >> 2) & ((1 << 21) - 1);
}

#endif /* __alpha__ */

#ifdef __arm__

#define ARM_LDR_TABLE_SIZE 1024

typedef struct LDREntry {
    uint8_t *ptr;
    uint32_t *data_ptr;
    unsigned type:2;
} LDREntry;

static LDREntry arm_ldr_table[1024];
static uint32_t arm_data_table[ARM_LDR_TABLE_SIZE];

extern char exec_loop;

static inline void arm_reloc_pc24(uint32_t *ptr, uint32_t insn, int val)
{
    *ptr = (insn & ~0xffffff) | ((insn + ((val - (int)ptr) >> 2)) & 0xffffff);
}

static uint8_t *arm_flush_ldr(uint8_t *gen_code_ptr,
                              LDREntry *ldr_start, LDREntry *ldr_end, 
                              uint32_t *data_start, uint32_t *data_end, 
                              int gen_jmp)
{
    LDREntry *le;
    uint32_t *ptr;
    int offset, data_size, target;
    uint8_t *data_ptr;
    uint32_t insn;
    uint32_t mask;
 
    data_size = (data_end - data_start) << 2;

    if (gen_jmp) {
        /* generate branch to skip the data */
        if (data_size == 0)
            return gen_code_ptr;
        target = (long)gen_code_ptr + data_size + 4;
        arm_reloc_pc24((uint32_t *)gen_code_ptr, 0xeafffffe, target);
        gen_code_ptr += 4;
    }
   
    /* copy the data */
    data_ptr = gen_code_ptr;
    memcpy(gen_code_ptr, data_start, data_size);
    gen_code_ptr += data_size;
    
    /* patch the ldr to point to the data */
    for(le = ldr_start; le < ldr_end; le++) {
        ptr = (uint32_t *)le->ptr;
        offset = ((unsigned long)(le->data_ptr) - (unsigned long)data_start) + 
            (unsigned long)data_ptr - 
            (unsigned long)ptr - 8;
        if (offset < 0) {
            fprintf(stderr, "Negative constant pool offset\n");
            abort();
        }
        switch (le->type) {
          case 0: /* ldr */
            mask = ~0x00800fff;
            if (offset >= 4096) {
                fprintf(stderr, "Bad ldr offset\n");
                abort();
            }
            break;
          case 1: /* ldc */
            mask = ~0x008000ff;
            if (offset >= 1024 ) {
                fprintf(stderr, "Bad ldc offset\n");
                abort();
            }
            break;
          case 2: /* add */
            mask = ~0xfff;
            if (offset >= 1024 ) {
                fprintf(stderr, "Bad add offset\n");
                abort();
            }
            break;
          default:
            fprintf(stderr, "Bad pc relative fixup\n");
            abort();
          }
        insn = *ptr & mask;
        switch (le->type) {
          case 0: /* ldr */
            insn |= offset | 0x00800000;
            break;
          case 1: /* ldc */
            insn |= (offset >> 2) | 0x00800000;
            break;
          case 2: /* add */
            insn |= (offset >> 2) | 0xf00;
            break;
          }
        *ptr = insn;
    }
    return gen_code_ptr;
}

#endif /* __arm__ */

#ifdef __ia64


/* Patch instruction with "val" where "mask" has 1 bits. */
static inline void ia64_patch (uint64_t insn_addr, uint64_t mask, uint64_t val)
{
    uint64_t m0, m1, v0, v1, b0, b1, *b = (uint64_t *) (insn_addr & -16);
#   define insn_mask ((1UL << 41) - 1)
    unsigned long shift;

    b0 = b[0]; b1 = b[1];
    shift = 5 + 41 * (insn_addr % 16); /* 5 template, 3 x 41-bit insns */
    if (shift >= 64) {
	m1 = mask << (shift - 64);
	v1 = val << (shift - 64);
    } else {
	m0 = mask << shift; m1 = mask >> (64 - shift);
	v0 = val  << shift; v1 = val >> (64 - shift);
	b[0] = (b0 & ~m0) | (v0 & m0);
    }
    b[1] = (b1 & ~m1) | (v1 & m1);
}

static inline void ia64_patch_imm60 (uint64_t insn_addr, uint64_t val)
{
	ia64_patch(insn_addr,
		   0x011ffffe000UL,
		   (  ((val & 0x0800000000000000UL) >> 23) /* bit 59 -> 36 */
		    | ((val & 0x00000000000fffffUL) << 13) /* bit 0 -> 13 */));
	ia64_patch(insn_addr - 1, 0x1fffffffffcUL, val >> 18);
}

static inline void ia64_imm64 (void *insn, uint64_t val)
{
    /* Ignore the slot number of the relocation; GCC and Intel
       toolchains differed for some time on whether IMM64 relocs are
       against slot 1 (Intel) or slot 2 (GCC).  */
    uint64_t insn_addr = (uint64_t) insn & ~3UL;

    ia64_patch(insn_addr + 2,
	       0x01fffefe000UL,
	       (  ((val & 0x8000000000000000UL) >> 27) /* bit 63 -> 36 */
		| ((val & 0x0000000000200000UL) <<  0) /* bit 21 -> 21 */
		| ((val & 0x00000000001f0000UL) <<  6) /* bit 16 -> 22 */
		| ((val & 0x000000000000ff80UL) << 20) /* bit  7 -> 27 */
		| ((val & 0x000000000000007fUL) << 13) /* bit  0 -> 13 */)
	    );
    ia64_patch(insn_addr + 1, 0x1ffffffffffUL, val >> 22);
}

static inline void ia64_imm60b (void *insn, uint64_t val)
{
    /* Ignore the slot number of the relocation; GCC and Intel
       toolchains differed for some time on whether IMM64 relocs are
       against slot 1 (Intel) or slot 2 (GCC).  */
    uint64_t insn_addr = (uint64_t) insn & ~3UL;

    if (val + ((uint64_t) 1 << 59) >= (1UL << 60))
	fprintf(stderr, "%s: value %ld out of IMM60 range\n",
		__FUNCTION__, (int64_t) val);
    ia64_patch_imm60(insn_addr + 2, val);
}

static inline void ia64_imm22 (void *insn, uint64_t val)
{
    if (val + (1 << 21) >= (1 << 22))
	fprintf(stderr, "%s: value %li out of IMM22 range\n",
		__FUNCTION__, (int64_t)val);
    ia64_patch((uint64_t) insn, 0x01fffcfe000UL,
	       (  ((val & 0x200000UL) << 15) /* bit 21 -> 36 */
		| ((val & 0x1f0000UL) <<  6) /* bit 16 -> 22 */
		| ((val & 0x00ff80UL) << 20) /* bit  7 -> 27 */
		| ((val & 0x00007fUL) << 13) /* bit  0 -> 13 */));
}

/* Like ia64_imm22(), but also clear bits 20-21.  For addl, this has
   the effect of turning "addl rX=imm22,rY" into "addl
   rX=imm22,r0".  */
static inline void ia64_imm22_r0 (void *insn, uint64_t val)
{
    if (val + (1 << 21) >= (1 << 22))
	fprintf(stderr, "%s: value %li out of IMM22 range\n",
		__FUNCTION__, (int64_t)val);
    ia64_patch((uint64_t) insn, 0x01fffcfe000UL | (0x3UL << 20),
	       (  ((val & 0x200000UL) << 15) /* bit 21 -> 36 */
		| ((val & 0x1f0000UL) <<  6) /* bit 16 -> 22 */
		| ((val & 0x00ff80UL) << 20) /* bit  7 -> 27 */
		| ((val & 0x00007fUL) << 13) /* bit  0 -> 13 */));
}

static inline void ia64_imm21b (void *insn, uint64_t val)
{
    if (val + (1 << 20) >= (1 << 21))
	fprintf(stderr, "%s: value %li out of IMM21b range\n",
		__FUNCTION__, (int64_t)val);
    ia64_patch((uint64_t) insn, 0x11ffffe000UL,
	       (  ((val & 0x100000UL) << 16) /* bit 20 -> 36 */
		| ((val & 0x0fffffUL) << 13) /* bit  0 -> 13 */));
}

static inline void ia64_nop_b (void *insn)
{
    ia64_patch((uint64_t) insn, (1UL << 41) - 1, 2UL << 37);
}

static inline void ia64_ldxmov(void *insn, uint64_t val)
{
    if (val + (1 << 21) < (1 << 22))
	ia64_patch((uint64_t) insn, 0x1fff80fe000UL, 8UL << 37);
}

static inline int ia64_patch_ltoff(void *insn, uint64_t val,
				   int relaxable)
{
    if (relaxable && (val + (1 << 21) < (1 << 22))) {
	ia64_imm22_r0(insn, val);
	return 0;
    }
    return 1;
}

struct ia64_fixup {
    struct ia64_fixup *next;
    void *addr;			/* address that needs to be patched */
    long value;
};

#define IA64_PLT(insn, plt_index)			\
do {							\
    struct ia64_fixup *fixup = alloca(sizeof(*fixup));	\
    fixup->next = plt_fixes;				\
    plt_fixes = fixup;					\
    fixup->addr = (insn);				\
    fixup->value = (plt_index);				\
    plt_offset[(plt_index)] = 1;			\
} while (0)

#define IA64_LTOFF(insn, val, relaxable)			\
do {								\
    if (ia64_patch_ltoff(insn, val, relaxable)) {		\
	struct ia64_fixup *fixup = alloca(sizeof(*fixup));	\
	fixup->next = ltoff_fixes;				\
	ltoff_fixes = fixup;					\
	fixup->addr = (insn);					\
	fixup->value = (val);					\
    }								\
} while (0)

static inline void ia64_apply_fixes (uint8_t **gen_code_pp,
				     struct ia64_fixup *ltoff_fixes,
				     uint64_t gp,
				     struct ia64_fixup *plt_fixes,
				     int num_plts,
				     unsigned long *plt_target,
				     unsigned int *plt_offset)
{
    static const uint8_t plt_bundle[] = {
	0x04, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,	/* nop 0; movl r1=GP */
	0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x00, 0x60,

	0x05, 0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00,	/* nop 0; brl IP */
	0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xc0
    };
    uint8_t *gen_code_ptr = *gen_code_pp, *plt_start, *got_start, *vp;
    struct ia64_fixup *fixup;
    unsigned int offset = 0;
    struct fdesc {
	long ip;
	long gp;
    } *fdesc;
    int i;

    if (plt_fixes) {
	plt_start = gen_code_ptr;

	for (i = 0; i < num_plts; ++i) {
	    if (plt_offset[i]) {
		plt_offset[i] = offset;
		offset += sizeof(plt_bundle);

		fdesc = (struct fdesc *) plt_target[i];
		memcpy(gen_code_ptr, plt_bundle, sizeof(plt_bundle));
		ia64_imm64 (gen_code_ptr + 0x02, fdesc->gp);
		ia64_imm60b(gen_code_ptr + 0x12,
			    (fdesc->ip - (long) (gen_code_ptr + 0x10)) >> 4);
		gen_code_ptr += sizeof(plt_bundle);
	    }
	}

	for (fixup = plt_fixes; fixup; fixup = fixup->next)
	    ia64_imm21b(fixup->addr,
			((long) plt_start + plt_offset[fixup->value]
			 - ((long) fixup->addr & ~0xf)) >> 4);
    }

    got_start = gen_code_ptr;

    /* First, create the GOT: */
    for (fixup = ltoff_fixes; fixup; fixup = fixup->next) {
	/* first check if we already have this value in the GOT: */
	for (vp = got_start; vp < gen_code_ptr; ++vp)
	    if (*(uint64_t *) vp == fixup->value)
		break;
	if (vp == gen_code_ptr) {
	    /* Nope, we need to put the value in the GOT: */
	    *(uint64_t *) vp = fixup->value;
	    gen_code_ptr += 8;
	}
	ia64_imm22(fixup->addr, (long) vp - gp);
    }
    /* Keep code ptr aligned. */
    if ((long) gen_code_ptr & 15)
	gen_code_ptr += 8;
    *gen_code_pp = gen_code_ptr;
}

#endif
OpenPOWER on IntegriCloud