xref: /titanic_50/usr/src/lib/libdisasm/common/libdisasm.c (revision 038db8fb497789ea9b367f771544442507cede18)
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
55f7184619SJoshua M. Clulow 
56f7184619SJoshua M. Clulow static dis_arch_t *dis_archs[] = {
57f7184619SJoshua M. Clulow #if !defined(DIS_STANDALONE) || defined(__i386) || defined(__amd64)
58f7184619SJoshua M. Clulow 	&dis_arch_i386,
59f7184619SJoshua M. Clulow #endif
60f7184619SJoshua M. Clulow #if !defined(DIS_STANDALONE) || defined(__sparc)
61f7184619SJoshua M. Clulow 	&dis_arch_sparc,
62f7184619SJoshua M. Clulow #endif
63f7184619SJoshua M. Clulow 	NULL
64f7184619SJoshua M. Clulow };
65f7184619SJoshua M. Clulow 
66f7184619SJoshua M. Clulow /*
67dc0093f4Seschrock  * For the standalone library, we need to link against mdb's malloc/free.
68dc0093f4Seschrock  * Otherwise, use the standard malloc/free.
69dc0093f4Seschrock  */
70dc0093f4Seschrock #ifdef DIS_STANDALONE
71dc0093f4Seschrock void *
72dc0093f4Seschrock dis_zalloc(size_t bytes)
73dc0093f4Seschrock {
74dc0093f4Seschrock 	return (mdb_zalloc(bytes, UM_SLEEP));
75dc0093f4Seschrock }
76dc0093f4Seschrock 
77dc0093f4Seschrock void
78dc0093f4Seschrock dis_free(void *ptr, size_t bytes)
79dc0093f4Seschrock {
80dc0093f4Seschrock 	mdb_free(ptr, bytes);
81dc0093f4Seschrock }
82dc0093f4Seschrock #else
83dc0093f4Seschrock void *
84dc0093f4Seschrock dis_zalloc(size_t bytes)
85dc0093f4Seschrock {
86dc0093f4Seschrock 	return (calloc(1, bytes));
87dc0093f4Seschrock }
88dc0093f4Seschrock 
89dc0093f4Seschrock /*ARGSUSED*/
90dc0093f4Seschrock void
91dc0093f4Seschrock dis_free(void *ptr, size_t bytes)
92dc0093f4Seschrock {
93dc0093f4Seschrock 	free(ptr);
94dc0093f4Seschrock }
95dc0093f4Seschrock #endif
96dc0093f4Seschrock 
97dc0093f4Seschrock int
98dc0093f4Seschrock dis_seterrno(int error)
99dc0093f4Seschrock {
100dc0093f4Seschrock 	_dis_errno = error;
101dc0093f4Seschrock 	return (-1);
102dc0093f4Seschrock }
103dc0093f4Seschrock 
104dc0093f4Seschrock int
105dc0093f4Seschrock dis_errno(void)
106dc0093f4Seschrock {
107dc0093f4Seschrock 	return (_dis_errno);
108dc0093f4Seschrock }
109dc0093f4Seschrock 
110dc0093f4Seschrock const char *
111dc0093f4Seschrock dis_strerror(int error)
112dc0093f4Seschrock {
113dc0093f4Seschrock 	switch (error) {
114dc0093f4Seschrock 	case E_DIS_NOMEM:
115dc0093f4Seschrock 		return ("out of memory");
116dc0093f4Seschrock 	case E_DIS_INVALFLAG:
117dc0093f4Seschrock 		return ("invalid flags for this architecture");
118f7184619SJoshua M. Clulow 	case E_DIS_UNSUPARCH:
119f7184619SJoshua M. Clulow 		return ("unsupported machine architecture");
120dc0093f4Seschrock 	default:
121dc0093f4Seschrock 		return ("unknown error");
122dc0093f4Seschrock 	}
123dc0093f4Seschrock }
124f7184619SJoshua M. Clulow 
125f7184619SJoshua M. Clulow void
126f7184619SJoshua M. Clulow dis_set_data(dis_handle_t *dhp, void *data)
127f7184619SJoshua M. Clulow {
128f7184619SJoshua M. Clulow 	dhp->dh_data = data;
129f7184619SJoshua M. Clulow }
130f7184619SJoshua M. Clulow 
131f7184619SJoshua M. Clulow void
132f7184619SJoshua M. Clulow dis_flags_set(dis_handle_t *dhp, int f)
133f7184619SJoshua M. Clulow {
134f7184619SJoshua M. Clulow 	dhp->dh_flags |= f;
135f7184619SJoshua M. Clulow }
136f7184619SJoshua M. Clulow 
137f7184619SJoshua M. Clulow void
138f7184619SJoshua M. Clulow dis_flags_clear(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_handle_destroy(dis_handle_t *dhp)
145f7184619SJoshua M. Clulow {
146f7184619SJoshua M. Clulow 	dhp->dh_arch->da_handle_detach(dhp);
147f7184619SJoshua M. Clulow 	dis_free(dhp, sizeof (dis_handle_t));
148f7184619SJoshua M. Clulow }
149f7184619SJoshua M. Clulow 
150f7184619SJoshua M. Clulow dis_handle_t *
151f7184619SJoshua M. Clulow dis_handle_create(int flags, void *data, dis_lookup_f lookup_func,
152f7184619SJoshua M. Clulow     dis_read_f read_func)
153f7184619SJoshua M. Clulow {
154f7184619SJoshua M. Clulow 	dis_handle_t *dhp;
155f7184619SJoshua M. Clulow 	dis_arch_t *arch = NULL;
156f7184619SJoshua M. Clulow 	int i;
157f7184619SJoshua M. Clulow 
158f7184619SJoshua M. Clulow 	/* Select an architecture based on flags */
159f7184619SJoshua M. Clulow 	for (i = 0; dis_archs[i] != NULL; i++) {
160f7184619SJoshua M. Clulow 		if (dis_archs[i]->da_supports_flags(flags)) {
161f7184619SJoshua M. Clulow 			arch = dis_archs[i];
162f7184619SJoshua M. Clulow 			break;
163f7184619SJoshua M. Clulow 		}
164f7184619SJoshua M. Clulow 	}
165f7184619SJoshua M. Clulow 	if (arch == NULL) {
166f7184619SJoshua M. Clulow 		(void) dis_seterrno(E_DIS_UNSUPARCH);
167f7184619SJoshua M. Clulow 		return (NULL);
168f7184619SJoshua M. Clulow 	}
169f7184619SJoshua M. Clulow 
170f7184619SJoshua M. Clulow 	if ((dhp = dis_zalloc(sizeof (dis_handle_t))) == NULL) {
171f7184619SJoshua M. Clulow 		(void) dis_seterrno(E_DIS_NOMEM);
172f7184619SJoshua M. Clulow 		return (NULL);
173f7184619SJoshua M. Clulow 	}
174f7184619SJoshua M. Clulow 	dhp->dh_arch = arch;
175f7184619SJoshua M. Clulow 	dhp->dh_lookup = lookup_func;
176f7184619SJoshua M. Clulow 	dhp->dh_read = read_func;
177f7184619SJoshua M. Clulow 	dhp->dh_flags = flags;
178f7184619SJoshua M. Clulow 	dhp->dh_data = data;
179f7184619SJoshua M. Clulow 
180f7184619SJoshua M. Clulow 	/*
181f7184619SJoshua M. Clulow 	 * Allow the architecture-specific code to allocate
182f7184619SJoshua M. Clulow 	 * its private data.
183f7184619SJoshua M. Clulow 	 */
184f7184619SJoshua M. Clulow 	if (arch->da_handle_attach(dhp) != 0) {
185f7184619SJoshua M. Clulow 		dis_free(dhp, sizeof (dis_handle_t));
186f7184619SJoshua M. Clulow 		/* dis errno already set */
187f7184619SJoshua M. Clulow 		return (NULL);
188f7184619SJoshua M. Clulow 	}
189f7184619SJoshua M. Clulow 
190f7184619SJoshua M. Clulow 	return (dhp);
191f7184619SJoshua M. Clulow }
192f7184619SJoshua M. Clulow 
193f7184619SJoshua M. Clulow int
194f7184619SJoshua M. Clulow dis_disassemble(dis_handle_t *dhp, uint64_t addr, char *buf, size_t buflen)
195f7184619SJoshua M. Clulow {
196f7184619SJoshua M. Clulow 	return (dhp->dh_arch->da_disassemble(dhp, addr, buf, buflen));
197f7184619SJoshua M. Clulow }
198f7184619SJoshua M. Clulow 
199*038db8fbSJosef 'Jeff' Sipek /*
200*038db8fbSJosef 'Jeff' Sipek  * On some instruction sets (e.g., x86), we have no choice except to
201*038db8fbSJosef 'Jeff' Sipek  * disassemble everything from the start of the symbol, and stop when we
202*038db8fbSJosef 'Jeff' Sipek  * have reached our instruction address.  If we're not in the middle of a
203*038db8fbSJosef 'Jeff' Sipek  * known symbol, then we return the same address to indicate failure.
204*038db8fbSJosef 'Jeff' Sipek  */
205*038db8fbSJosef 'Jeff' Sipek static uint64_t
206*038db8fbSJosef 'Jeff' Sipek dis_generic_previnstr(dis_handle_t *dhp, uint64_t pc, int n)
207*038db8fbSJosef 'Jeff' Sipek {
208*038db8fbSJosef 'Jeff' Sipek 	uint64_t *hist, addr, start;
209*038db8fbSJosef 'Jeff' Sipek 	int cur, nseen;
210*038db8fbSJosef 'Jeff' Sipek 	uint64_t res = pc;
211*038db8fbSJosef 'Jeff' Sipek 
212*038db8fbSJosef 'Jeff' Sipek 	if (n <= 0)
213*038db8fbSJosef 'Jeff' Sipek 		return (pc);
214*038db8fbSJosef 'Jeff' Sipek 
215*038db8fbSJosef 'Jeff' Sipek 	if (dhp->dh_lookup(dhp->dh_data, pc, NULL, 0, &start, NULL) != 0 ||
216*038db8fbSJosef 'Jeff' Sipek 	    start == pc)
217*038db8fbSJosef 'Jeff' Sipek 		return (res);
218*038db8fbSJosef 'Jeff' Sipek 
219*038db8fbSJosef 'Jeff' Sipek 	hist = dis_zalloc(sizeof (uint64_t) * n);
220*038db8fbSJosef 'Jeff' Sipek 
221*038db8fbSJosef 'Jeff' Sipek 	for (cur = 0, nseen = 0, addr = start; addr < pc; addr = dhp->dh_addr) {
222*038db8fbSJosef 'Jeff' Sipek 		hist[cur] = addr;
223*038db8fbSJosef 'Jeff' Sipek 		cur = (cur + 1) % n;
224*038db8fbSJosef 'Jeff' Sipek 		nseen++;
225*038db8fbSJosef 'Jeff' Sipek 
226*038db8fbSJosef 'Jeff' Sipek 		/* if we cannot make forward progress, give up */
227*038db8fbSJosef 'Jeff' Sipek 		if (dis_disassemble(dhp, addr, NULL, 0) != 0)
228*038db8fbSJosef 'Jeff' Sipek 			goto done;
229*038db8fbSJosef 'Jeff' Sipek 	}
230*038db8fbSJosef 'Jeff' Sipek 
231*038db8fbSJosef 'Jeff' Sipek 	if (addr != pc) {
232*038db8fbSJosef 'Jeff' Sipek 		/*
233*038db8fbSJosef 'Jeff' Sipek 		 * We scanned past %pc, but didn't find an instruction that
234*038db8fbSJosef 'Jeff' Sipek 		 * started at %pc.  This means that either the caller specified
235*038db8fbSJosef 'Jeff' Sipek 		 * an invalid address, or we ran into something other than code
236*038db8fbSJosef 'Jeff' Sipek 		 * during our scan.  Virtually any combination of bytes can be
237*038db8fbSJosef 'Jeff' Sipek 		 * construed as a valid Intel instruction, so any non-code bytes
238*038db8fbSJosef 'Jeff' Sipek 		 * we encounter will have thrown off the scan.
239*038db8fbSJosef 'Jeff' Sipek 		 */
240*038db8fbSJosef 'Jeff' Sipek 		goto done;
241*038db8fbSJosef 'Jeff' Sipek 	}
242*038db8fbSJosef 'Jeff' Sipek 
243*038db8fbSJosef 'Jeff' Sipek 	res = hist[(cur + n - MIN(n, nseen)) % n];
244*038db8fbSJosef 'Jeff' Sipek 
245*038db8fbSJosef 'Jeff' Sipek done:
246*038db8fbSJosef 'Jeff' Sipek 	dis_free(hist, sizeof (uint64_t) * n);
247*038db8fbSJosef 'Jeff' Sipek 	return (res);
248*038db8fbSJosef 'Jeff' Sipek }
249*038db8fbSJosef 'Jeff' Sipek 
250*038db8fbSJosef 'Jeff' Sipek /*
251*038db8fbSJosef 'Jeff' Sipek  * Return the nth previous instruction's address.  Return the same address
252*038db8fbSJosef 'Jeff' Sipek  * to indicate failure.
253*038db8fbSJosef 'Jeff' Sipek  */
254f7184619SJoshua M. Clulow uint64_t
255f7184619SJoshua M. Clulow dis_previnstr(dis_handle_t *dhp, uint64_t pc, int n)
256f7184619SJoshua M. Clulow {
257*038db8fbSJosef 'Jeff' Sipek 	if (dhp->dh_arch->da_previnstr == NULL)
258*038db8fbSJosef 'Jeff' Sipek 		return (dis_generic_previnstr(dhp, pc, n));
259*038db8fbSJosef 'Jeff' Sipek 
260f7184619SJoshua M. Clulow 	return (dhp->dh_arch->da_previnstr(dhp, pc, n));
261f7184619SJoshua M. Clulow }
262f7184619SJoshua M. Clulow 
263f7184619SJoshua M. Clulow int
264f7184619SJoshua M. Clulow dis_min_instrlen(dis_handle_t *dhp)
265f7184619SJoshua M. Clulow {
266f7184619SJoshua M. Clulow 	return (dhp->dh_arch->da_min_instrlen(dhp));
267f7184619SJoshua M. Clulow }
268f7184619SJoshua M. Clulow 
269f7184619SJoshua M. Clulow int
270f7184619SJoshua M. Clulow dis_max_instrlen(dis_handle_t *dhp)
271f7184619SJoshua M. Clulow {
272f7184619SJoshua M. Clulow 	return (dhp->dh_arch->da_max_instrlen(dhp));
273f7184619SJoshua M. Clulow }
274f7184619SJoshua M. Clulow 
275f7184619SJoshua M. Clulow int
276f7184619SJoshua M. Clulow dis_instrlen(dis_handle_t *dhp, uint64_t pc)
277f7184619SJoshua M. Clulow {
278f7184619SJoshua M. Clulow 	return (dhp->dh_arch->da_instrlen(dhp, pc));
279f7184619SJoshua M. Clulow }
280f7184619SJoshua M. Clulow 
281f7184619SJoshua M. Clulow int
282f7184619SJoshua M. Clulow dis_vsnprintf(char *restrict s, size_t n, const char *restrict format,
283f7184619SJoshua M. Clulow     va_list args)
284f7184619SJoshua M. Clulow {
285f7184619SJoshua M. Clulow #ifdef DIS_STANDALONE
286f7184619SJoshua M. Clulow 	return (mdb_iob_vsnprintf(s, n, format, args));
287f7184619SJoshua M. Clulow #else
288f7184619SJoshua M. Clulow 	return (vsnprintf(s, n, format, args));
289f7184619SJoshua M. Clulow #endif
290f7184619SJoshua M. Clulow }
291f7184619SJoshua M. Clulow 
292f7184619SJoshua M. Clulow int
293f7184619SJoshua M. Clulow dis_snprintf(char *restrict s, size_t n, const char *restrict format, ...)
294f7184619SJoshua M. Clulow {
295f7184619SJoshua M. Clulow 	va_list args;
296f7184619SJoshua M. Clulow 
297f7184619SJoshua M. Clulow 	va_start(args, format);
298f7184619SJoshua M. Clulow 	n = dis_vsnprintf(s, n, format, args);
299f7184619SJoshua M. Clulow 	va_end(args);
300f7184619SJoshua M. Clulow 
301f7184619SJoshua M. Clulow 	return (n);
302f7184619SJoshua M. Clulow }
303