xref: /titanic_51/usr/src/lib/libdisasm/common/libdisasm.c (revision 0472e113e9ad4a95dbf46a1d97075472136a1e7c)
1dc0093f4Seschrock /*
2dc0093f4Seschrock  * CDDL HEADER START
3dc0093f4Seschrock  *
4dc0093f4Seschrock  * The contents of this file are subject to the terms of the
5dc0093f4Seschrock  * Common Development and Distribution License (the "License").
6dc0093f4Seschrock  * You may not use this file except in compliance with the License.
7dc0093f4Seschrock  *
8dc0093f4Seschrock  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9dc0093f4Seschrock  * or http://www.opensolaris.org/os/licensing.
10dc0093f4Seschrock  * See the License for the specific language governing permissions
11dc0093f4Seschrock  * and limitations under the License.
12dc0093f4Seschrock  *
13dc0093f4Seschrock  * When distributing Covered Code, include this CDDL HEADER in each
14dc0093f4Seschrock  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15dc0093f4Seschrock  * If applicable, add the following below this CDDL HEADER, with the
16dc0093f4Seschrock  * fields enclosed by brackets "[]" replaced with your own identifying
17dc0093f4Seschrock  * information: Portions Copyright [yyyy] [name of copyright owner]
18dc0093f4Seschrock  *
19dc0093f4Seschrock  * CDDL HEADER END
20dc0093f4Seschrock  */
21dc0093f4Seschrock 
22dc0093f4Seschrock /*
23dc0093f4Seschrock  * Copyright 2006 Sun Microsystems, Inc.  All rights reserved.
24dc0093f4Seschrock  * Use is subject to license terms.
25f7184619SJoshua M. Clulow  * Copyright 2012 Joshua M. Clulow <josh@sysmgr.org>
26f7184619SJoshua M. Clulow  * Copyright 2015 Nexenta Systems, Inc.  All rights reserved.
27dc0093f4Seschrock  */
28dc0093f4Seschrock 
29dc0093f4Seschrock #include <libdisasm.h>
30dc0093f4Seschrock #include <stdlib.h>
31dc0093f4Seschrock #ifdef DIS_STANDALONE
32dc0093f4Seschrock #include <mdb/mdb_modapi.h>
33f7184619SJoshua M. Clulow #define	_MDB
34f7184619SJoshua M. Clulow #include <mdb/mdb_io.h>
35f7184619SJoshua M. Clulow #else
36f7184619SJoshua M. Clulow #include <stdio.h>
37dc0093f4Seschrock #endif
38dc0093f4Seschrock 
39f7184619SJoshua M. Clulow #include "libdisasm_impl.h"
40f7184619SJoshua M. Clulow 
41dc0093f4Seschrock static int _dis_errno;
42dc0093f4Seschrock 
43dc0093f4Seschrock /*
44f7184619SJoshua M. Clulow  * If we're building the standalone library, then we only want to
45f7184619SJoshua M. Clulow  * include support for disassembly of the native architecture.
46f7184619SJoshua M. Clulow  * The regular shared library should include support for all
47f7184619SJoshua M. Clulow  * architectures.
48f7184619SJoshua M. Clulow  */
49f7184619SJoshua M. Clulow #if !defined(DIS_STANDALONE) || defined(__i386) || defined(__amd64)
50f7184619SJoshua M. Clulow extern dis_arch_t dis_arch_i386;
51f7184619SJoshua M. Clulow #endif
52f7184619SJoshua M. Clulow #if !defined(DIS_STANDALONE) || defined(__sparc)
53f7184619SJoshua M. Clulow extern dis_arch_t dis_arch_sparc;
54f7184619SJoshua M. Clulow #endif
55*0472e113SJosef 'Jeff' Sipek #if !defined(DIS_STANDALONE) || defined(__s390) || defined(__s390x)
56*0472e113SJosef 'Jeff' Sipek extern dis_arch_t dis_arch_s390;
57*0472e113SJosef 'Jeff' Sipek #endif
58f7184619SJoshua M. Clulow 
59f7184619SJoshua M. Clulow static dis_arch_t *dis_archs[] = {
60f7184619SJoshua M. Clulow #if !defined(DIS_STANDALONE) || defined(__i386) || defined(__amd64)
61f7184619SJoshua M. Clulow 	&dis_arch_i386,
62f7184619SJoshua M. Clulow #endif
63f7184619SJoshua M. Clulow #if !defined(DIS_STANDALONE) || defined(__sparc)
64f7184619SJoshua M. Clulow 	&dis_arch_sparc,
65f7184619SJoshua M. Clulow #endif
66*0472e113SJosef 'Jeff' Sipek #if !defined(DIS_STANDALONE) || defined(__s390) || defined(__s390x)
67*0472e113SJosef 'Jeff' Sipek 	&dis_arch_s390,
68*0472e113SJosef 'Jeff' Sipek #endif
69f7184619SJoshua M. Clulow 	NULL
70f7184619SJoshua M. Clulow };
71f7184619SJoshua M. Clulow 
72f7184619SJoshua M. Clulow /*
73dc0093f4Seschrock  * For the standalone library, we need to link against mdb's malloc/free.
74dc0093f4Seschrock  * Otherwise, use the standard malloc/free.
75dc0093f4Seschrock  */
76dc0093f4Seschrock #ifdef DIS_STANDALONE
77dc0093f4Seschrock void *
78dc0093f4Seschrock dis_zalloc(size_t bytes)
79dc0093f4Seschrock {
80dc0093f4Seschrock 	return (mdb_zalloc(bytes, UM_SLEEP));
81dc0093f4Seschrock }
82dc0093f4Seschrock 
83dc0093f4Seschrock void
84dc0093f4Seschrock dis_free(void *ptr, size_t bytes)
85dc0093f4Seschrock {
86dc0093f4Seschrock 	mdb_free(ptr, bytes);
87dc0093f4Seschrock }
88dc0093f4Seschrock #else
89dc0093f4Seschrock void *
90dc0093f4Seschrock dis_zalloc(size_t bytes)
91dc0093f4Seschrock {
92dc0093f4Seschrock 	return (calloc(1, bytes));
93dc0093f4Seschrock }
94dc0093f4Seschrock 
95dc0093f4Seschrock /*ARGSUSED*/
96dc0093f4Seschrock void
97dc0093f4Seschrock dis_free(void *ptr, size_t bytes)
98dc0093f4Seschrock {
99dc0093f4Seschrock 	free(ptr);
100dc0093f4Seschrock }
101dc0093f4Seschrock #endif
102dc0093f4Seschrock 
103dc0093f4Seschrock int
104dc0093f4Seschrock dis_seterrno(int error)
105dc0093f4Seschrock {
106dc0093f4Seschrock 	_dis_errno = error;
107dc0093f4Seschrock 	return (-1);
108dc0093f4Seschrock }
109dc0093f4Seschrock 
110dc0093f4Seschrock int
111dc0093f4Seschrock dis_errno(void)
112dc0093f4Seschrock {
113dc0093f4Seschrock 	return (_dis_errno);
114dc0093f4Seschrock }
115dc0093f4Seschrock 
116dc0093f4Seschrock const char *
117dc0093f4Seschrock dis_strerror(int error)
118dc0093f4Seschrock {
119dc0093f4Seschrock 	switch (error) {
120dc0093f4Seschrock 	case E_DIS_NOMEM:
121dc0093f4Seschrock 		return ("out of memory");
122dc0093f4Seschrock 	case E_DIS_INVALFLAG:
123dc0093f4Seschrock 		return ("invalid flags for this architecture");
124f7184619SJoshua M. Clulow 	case E_DIS_UNSUPARCH:
125f7184619SJoshua M. Clulow 		return ("unsupported machine architecture");
126dc0093f4Seschrock 	default:
127dc0093f4Seschrock 		return ("unknown error");
128dc0093f4Seschrock 	}
129dc0093f4Seschrock }
130f7184619SJoshua M. Clulow 
131f7184619SJoshua M. Clulow void
132f7184619SJoshua M. Clulow dis_set_data(dis_handle_t *dhp, void *data)
133f7184619SJoshua M. Clulow {
134f7184619SJoshua M. Clulow 	dhp->dh_data = data;
135f7184619SJoshua M. Clulow }
136f7184619SJoshua M. Clulow 
137f7184619SJoshua M. Clulow void
138f7184619SJoshua M. Clulow dis_flags_set(dis_handle_t *dhp, int f)
139f7184619SJoshua M. Clulow {
140f7184619SJoshua M. Clulow 	dhp->dh_flags |= f;
141f7184619SJoshua M. Clulow }
142f7184619SJoshua M. Clulow 
143f7184619SJoshua M. Clulow void
144f7184619SJoshua M. Clulow dis_flags_clear(dis_handle_t *dhp, int f)
145f7184619SJoshua M. Clulow {
146f7184619SJoshua M. Clulow 	dhp->dh_flags &= ~f;
147f7184619SJoshua M. Clulow }
148f7184619SJoshua M. Clulow 
149f7184619SJoshua M. Clulow void
150f7184619SJoshua M. Clulow dis_handle_destroy(dis_handle_t *dhp)
151f7184619SJoshua M. Clulow {
1520c923cf7SJosef 'Jeff' Sipek 	if (dhp->dh_arch->da_handle_detach != NULL)
153f7184619SJoshua M. Clulow 		dhp->dh_arch->da_handle_detach(dhp);
1540c923cf7SJosef 'Jeff' Sipek 
155f7184619SJoshua M. Clulow 	dis_free(dhp, sizeof (dis_handle_t));
156f7184619SJoshua M. Clulow }
157f7184619SJoshua M. Clulow 
158f7184619SJoshua M. Clulow dis_handle_t *
159f7184619SJoshua M. Clulow dis_handle_create(int flags, void *data, dis_lookup_f lookup_func,
160f7184619SJoshua M. Clulow     dis_read_f read_func)
161f7184619SJoshua M. Clulow {
162f7184619SJoshua M. Clulow 	dis_handle_t *dhp;
163f7184619SJoshua M. Clulow 	dis_arch_t *arch = NULL;
164f7184619SJoshua M. Clulow 	int i;
165f7184619SJoshua M. Clulow 
166f7184619SJoshua M. Clulow 	/* Select an architecture based on flags */
167f7184619SJoshua M. Clulow 	for (i = 0; dis_archs[i] != NULL; i++) {
168f7184619SJoshua M. Clulow 		if (dis_archs[i]->da_supports_flags(flags)) {
169f7184619SJoshua M. Clulow 			arch = dis_archs[i];
170f7184619SJoshua M. Clulow 			break;
171f7184619SJoshua M. Clulow 		}
172f7184619SJoshua M. Clulow 	}
173f7184619SJoshua M. Clulow 	if (arch == NULL) {
174f7184619SJoshua M. Clulow 		(void) dis_seterrno(E_DIS_UNSUPARCH);
175f7184619SJoshua M. Clulow 		return (NULL);
176f7184619SJoshua M. Clulow 	}
177f7184619SJoshua M. Clulow 
178f7184619SJoshua M. Clulow 	if ((dhp = dis_zalloc(sizeof (dis_handle_t))) == NULL) {
179f7184619SJoshua M. Clulow 		(void) dis_seterrno(E_DIS_NOMEM);
180f7184619SJoshua M. Clulow 		return (NULL);
181f7184619SJoshua M. Clulow 	}
182f7184619SJoshua M. Clulow 	dhp->dh_arch = arch;
183f7184619SJoshua M. Clulow 	dhp->dh_lookup = lookup_func;
184f7184619SJoshua M. Clulow 	dhp->dh_read = read_func;
185f7184619SJoshua M. Clulow 	dhp->dh_flags = flags;
186f7184619SJoshua M. Clulow 	dhp->dh_data = data;
187f7184619SJoshua M. Clulow 
188f7184619SJoshua M. Clulow 	/*
189f7184619SJoshua M. Clulow 	 * Allow the architecture-specific code to allocate
190f7184619SJoshua M. Clulow 	 * its private data.
191f7184619SJoshua M. Clulow 	 */
1920c923cf7SJosef 'Jeff' Sipek 	if (arch->da_handle_attach != NULL &&
1930c923cf7SJosef 'Jeff' Sipek 	    arch->da_handle_attach(dhp) != 0) {
194f7184619SJoshua M. Clulow 		dis_free(dhp, sizeof (dis_handle_t));
195f7184619SJoshua M. Clulow 		/* dis errno already set */
196f7184619SJoshua M. Clulow 		return (NULL);
197f7184619SJoshua M. Clulow 	}
198f7184619SJoshua M. Clulow 
199f7184619SJoshua M. Clulow 	return (dhp);
200f7184619SJoshua M. Clulow }
201f7184619SJoshua M. Clulow 
202f7184619SJoshua M. Clulow int
203f7184619SJoshua M. Clulow dis_disassemble(dis_handle_t *dhp, uint64_t addr, char *buf, size_t buflen)
204f7184619SJoshua M. Clulow {
205f7184619SJoshua M. Clulow 	return (dhp->dh_arch->da_disassemble(dhp, addr, buf, buflen));
206f7184619SJoshua M. Clulow }
207f7184619SJoshua M. Clulow 
208038db8fbSJosef 'Jeff' Sipek /*
209038db8fbSJosef 'Jeff' Sipek  * On some instruction sets (e.g., x86), we have no choice except to
210038db8fbSJosef 'Jeff' Sipek  * disassemble everything from the start of the symbol, and stop when we
211038db8fbSJosef 'Jeff' Sipek  * have reached our instruction address.  If we're not in the middle of a
212038db8fbSJosef 'Jeff' Sipek  * known symbol, then we return the same address to indicate failure.
213038db8fbSJosef 'Jeff' Sipek  */
214038db8fbSJosef 'Jeff' Sipek static uint64_t
215038db8fbSJosef 'Jeff' Sipek dis_generic_previnstr(dis_handle_t *dhp, uint64_t pc, int n)
216038db8fbSJosef 'Jeff' Sipek {
217038db8fbSJosef 'Jeff' Sipek 	uint64_t *hist, addr, start;
218038db8fbSJosef 'Jeff' Sipek 	int cur, nseen;
219038db8fbSJosef 'Jeff' Sipek 	uint64_t res = pc;
220038db8fbSJosef 'Jeff' Sipek 
221038db8fbSJosef 'Jeff' Sipek 	if (n <= 0)
222038db8fbSJosef 'Jeff' Sipek 		return (pc);
223038db8fbSJosef 'Jeff' Sipek 
224038db8fbSJosef 'Jeff' Sipek 	if (dhp->dh_lookup(dhp->dh_data, pc, NULL, 0, &start, NULL) != 0 ||
225038db8fbSJosef 'Jeff' Sipek 	    start == pc)
226038db8fbSJosef 'Jeff' Sipek 		return (res);
227038db8fbSJosef 'Jeff' Sipek 
228038db8fbSJosef 'Jeff' Sipek 	hist = dis_zalloc(sizeof (uint64_t) * n);
229038db8fbSJosef 'Jeff' Sipek 
230038db8fbSJosef 'Jeff' Sipek 	for (cur = 0, nseen = 0, addr = start; addr < pc; addr = dhp->dh_addr) {
231038db8fbSJosef 'Jeff' Sipek 		hist[cur] = addr;
232038db8fbSJosef 'Jeff' Sipek 		cur = (cur + 1) % n;
233038db8fbSJosef 'Jeff' Sipek 		nseen++;
234038db8fbSJosef 'Jeff' Sipek 
235038db8fbSJosef 'Jeff' Sipek 		/* if we cannot make forward progress, give up */
236038db8fbSJosef 'Jeff' Sipek 		if (dis_disassemble(dhp, addr, NULL, 0) != 0)
237038db8fbSJosef 'Jeff' Sipek 			goto done;
238038db8fbSJosef 'Jeff' Sipek 	}
239038db8fbSJosef 'Jeff' Sipek 
240038db8fbSJosef 'Jeff' Sipek 	if (addr != pc) {
241038db8fbSJosef 'Jeff' Sipek 		/*
242038db8fbSJosef 'Jeff' Sipek 		 * We scanned past %pc, but didn't find an instruction that
243038db8fbSJosef 'Jeff' Sipek 		 * started at %pc.  This means that either the caller specified
244038db8fbSJosef 'Jeff' Sipek 		 * an invalid address, or we ran into something other than code
245038db8fbSJosef 'Jeff' Sipek 		 * during our scan.  Virtually any combination of bytes can be
246038db8fbSJosef 'Jeff' Sipek 		 * construed as a valid Intel instruction, so any non-code bytes
247038db8fbSJosef 'Jeff' Sipek 		 * we encounter will have thrown off the scan.
248038db8fbSJosef 'Jeff' Sipek 		 */
249038db8fbSJosef 'Jeff' Sipek 		goto done;
250038db8fbSJosef 'Jeff' Sipek 	}
251038db8fbSJosef 'Jeff' Sipek 
252038db8fbSJosef 'Jeff' Sipek 	res = hist[(cur + n - MIN(n, nseen)) % n];
253038db8fbSJosef 'Jeff' Sipek 
254038db8fbSJosef 'Jeff' Sipek done:
255038db8fbSJosef 'Jeff' Sipek 	dis_free(hist, sizeof (uint64_t) * n);
256038db8fbSJosef 'Jeff' Sipek 	return (res);
257038db8fbSJosef 'Jeff' Sipek }
258038db8fbSJosef 'Jeff' Sipek 
259038db8fbSJosef 'Jeff' Sipek /*
260038db8fbSJosef 'Jeff' Sipek  * Return the nth previous instruction's address.  Return the same address
261038db8fbSJosef 'Jeff' Sipek  * to indicate failure.
262038db8fbSJosef 'Jeff' Sipek  */
263f7184619SJoshua M. Clulow uint64_t
264f7184619SJoshua M. Clulow dis_previnstr(dis_handle_t *dhp, uint64_t pc, int n)
265f7184619SJoshua M. Clulow {
266038db8fbSJosef 'Jeff' Sipek 	if (dhp->dh_arch->da_previnstr == NULL)
267038db8fbSJosef 'Jeff' Sipek 		return (dis_generic_previnstr(dhp, pc, n));
268038db8fbSJosef 'Jeff' Sipek 
269f7184619SJoshua M. Clulow 	return (dhp->dh_arch->da_previnstr(dhp, pc, n));
270f7184619SJoshua M. Clulow }
271f7184619SJoshua M. Clulow 
272f7184619SJoshua M. Clulow int
273f7184619SJoshua M. Clulow dis_min_instrlen(dis_handle_t *dhp)
274f7184619SJoshua M. Clulow {
275f7184619SJoshua M. Clulow 	return (dhp->dh_arch->da_min_instrlen(dhp));
276f7184619SJoshua M. Clulow }
277f7184619SJoshua M. Clulow 
278f7184619SJoshua M. Clulow int
279f7184619SJoshua M. Clulow dis_max_instrlen(dis_handle_t *dhp)
280f7184619SJoshua M. Clulow {
281f7184619SJoshua M. Clulow 	return (dhp->dh_arch->da_max_instrlen(dhp));
282f7184619SJoshua M. Clulow }
283f7184619SJoshua M. Clulow 
284cfc41e72SJosef 'Jeff' Sipek static int
285cfc41e72SJosef 'Jeff' Sipek dis_generic_instrlen(dis_handle_t *dhp, uint64_t pc)
286cfc41e72SJosef 'Jeff' Sipek {
287cfc41e72SJosef 'Jeff' Sipek 	if (dis_disassemble(dhp, pc, NULL, 0) != 0)
288cfc41e72SJosef 'Jeff' Sipek 		return (-1);
289cfc41e72SJosef 'Jeff' Sipek 
290cfc41e72SJosef 'Jeff' Sipek 	return (dhp->dh_addr - pc);
291cfc41e72SJosef 'Jeff' Sipek }
292cfc41e72SJosef 'Jeff' Sipek 
293f7184619SJoshua M. Clulow int
294f7184619SJoshua M. Clulow dis_instrlen(dis_handle_t *dhp, uint64_t pc)
295f7184619SJoshua M. Clulow {
296cfc41e72SJosef 'Jeff' Sipek 	if (dhp->dh_arch->da_instrlen == NULL)
297cfc41e72SJosef 'Jeff' Sipek 		return (dis_generic_instrlen(dhp, pc));
298cfc41e72SJosef 'Jeff' Sipek 
299f7184619SJoshua M. Clulow 	return (dhp->dh_arch->da_instrlen(dhp, pc));
300f7184619SJoshua M. Clulow }
301f7184619SJoshua M. Clulow 
302f7184619SJoshua M. Clulow int
303f7184619SJoshua M. Clulow dis_vsnprintf(char *restrict s, size_t n, const char *restrict format,
304f7184619SJoshua M. Clulow     va_list args)
305f7184619SJoshua M. Clulow {
306f7184619SJoshua M. Clulow #ifdef DIS_STANDALONE
307f7184619SJoshua M. Clulow 	return (mdb_iob_vsnprintf(s, n, format, args));
308f7184619SJoshua M. Clulow #else
309f7184619SJoshua M. Clulow 	return (vsnprintf(s, n, format, args));
310f7184619SJoshua M. Clulow #endif
311f7184619SJoshua M. Clulow }
312f7184619SJoshua M. Clulow 
313f7184619SJoshua M. Clulow int
314f7184619SJoshua M. Clulow dis_snprintf(char *restrict s, size_t n, const char *restrict format, ...)
315f7184619SJoshua M. Clulow {
316f7184619SJoshua M. Clulow 	va_list args;
317f7184619SJoshua M. Clulow 
318f7184619SJoshua M. Clulow 	va_start(args, format);
319f7184619SJoshua M. Clulow 	n = dis_vsnprintf(s, n, format, args);
320f7184619SJoshua M. Clulow 	va_end(args);
321f7184619SJoshua M. Clulow 
322f7184619SJoshua M. Clulow 	return (n);
323f7184619SJoshua M. Clulow }
324