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 /*
23e0070315Sdmick  * Copyright 2007 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 
31dc0093f4Seschrock #include "dis_tables.h"
32dc0093f4Seschrock #include "libdisasm_impl.h"
33dc0093f4Seschrock 
34f7184619SJoshua M. Clulow typedef struct dis_handle_i386 {
35f7184619SJoshua M. Clulow 	int		dhx_mode;
36f7184619SJoshua M. Clulow 	dis86_t		dhx_dis;
37f7184619SJoshua M. Clulow 	uint64_t	dhx_end;
38f7184619SJoshua M. Clulow } dis_handle_i386_t;
39dc0093f4Seschrock 
40dc0093f4Seschrock /*
41dc0093f4Seschrock  * Returns true if we are near the end of a function.  This is a cheap hack at
42dc0093f4Seschrock  * detecting NULL padding between functions.  If we're within a few bytes of the
43dc0093f4Seschrock  * next function, or past the start, then return true.
44dc0093f4Seschrock  */
45dc0093f4Seschrock static int
46dc0093f4Seschrock check_func(void *data)
47dc0093f4Seschrock {
48dc0093f4Seschrock 	dis_handle_t *dhp = data;
49dc0093f4Seschrock 	uint64_t start;
50dc0093f4Seschrock 	size_t len;
51dc0093f4Seschrock 
52dc0093f4Seschrock 	if (dhp->dh_lookup(dhp->dh_data, dhp->dh_addr, NULL, 0, &start, &len)
53dc0093f4Seschrock 	    != 0)
54dc0093f4Seschrock 		return (0);
55dc0093f4Seschrock 
56dc0093f4Seschrock 	if (start < dhp->dh_addr)
57dc0093f4Seschrock 		return (dhp->dh_addr > start + len - 0x10);
58dc0093f4Seschrock 
59dc0093f4Seschrock 	return (1);
60dc0093f4Seschrock }
61dc0093f4Seschrock 
62dc0093f4Seschrock static int
63dc0093f4Seschrock get_byte(void *data)
64dc0093f4Seschrock {
65dc0093f4Seschrock 	uchar_t byte;
66dc0093f4Seschrock 	dis_handle_t *dhp = data;
67dc0093f4Seschrock 
68108ba071Seschrock 	if (dhp->dh_read(dhp->dh_data, dhp->dh_addr, &byte, sizeof (byte)) !=
69108ba071Seschrock 	    sizeof (byte))
70dc0093f4Seschrock 		return (-1);
71dc0093f4Seschrock 
72dc0093f4Seschrock 	dhp->dh_addr++;
73dc0093f4Seschrock 
74dc0093f4Seschrock 	return ((int)byte);
75dc0093f4Seschrock }
76dc0093f4Seschrock 
77dc0093f4Seschrock static int
78dc0093f4Seschrock do_lookup(void *data, uint64_t addr, char *buf, size_t buflen)
79dc0093f4Seschrock {
80dc0093f4Seschrock 	dis_handle_t *dhp = data;
81dc0093f4Seschrock 
82dc0093f4Seschrock 	return (dhp->dh_lookup(dhp->dh_data, addr, buf, buflen, NULL, NULL));
83dc0093f4Seschrock }
84dc0093f4Seschrock 
85f7184619SJoshua M. Clulow static void
86f7184619SJoshua M. Clulow dis_i386_handle_detach(dis_handle_t *dhp)
87f7184619SJoshua M. Clulow {
88f7184619SJoshua M. Clulow 	dis_free(dhp->dh_arch_private, sizeof (dis_handle_i386_t));
89f7184619SJoshua M. Clulow 	dhp->dh_arch_private = NULL;
90f7184619SJoshua M. Clulow }
91f7184619SJoshua M. Clulow 
92f7184619SJoshua M. Clulow static int
93f7184619SJoshua M. Clulow dis_i386_handle_attach(dis_handle_t *dhp)
94dc0093f4Seschrock {
95f7184619SJoshua M. Clulow 	dis_handle_i386_t *dhx;
96dc0093f4Seschrock 
97dc0093f4Seschrock 	/*
98dc0093f4Seschrock 	 * Validate architecture flags
99dc0093f4Seschrock 	 */
100f7184619SJoshua M. Clulow 	if (dhp->dh_flags & ~(DIS_X86_SIZE16 | DIS_X86_SIZE32 | DIS_X86_SIZE64 |
101e0070315Sdmick 	    DIS_OCTAL | DIS_NOIMMSYM)) {
102dc0093f4Seschrock 		(void) dis_seterrno(E_DIS_INVALFLAG);
103f7184619SJoshua M. Clulow 		return (-1);
104dc0093f4Seschrock 	}
105dc0093f4Seschrock 
106dc0093f4Seschrock 	/*
107dc0093f4Seschrock 	 * Create and initialize the internal structure
108dc0093f4Seschrock 	 */
109f7184619SJoshua M. Clulow 	if ((dhx = dis_zalloc(sizeof (dis_handle_i386_t))) == NULL) {
110dc0093f4Seschrock 		(void) dis_seterrno(E_DIS_NOMEM);
111f7184619SJoshua M. Clulow 		return (-1);
112dc0093f4Seschrock 	}
113f7184619SJoshua M. Clulow 	dhp->dh_arch_private = dhx;
114dc0093f4Seschrock 
115dc0093f4Seschrock 	/*
116dc0093f4Seschrock 	 * Initialize x86-specific architecture structure
117dc0093f4Seschrock 	 */
118f7184619SJoshua M. Clulow 	if (dhp->dh_flags & DIS_X86_SIZE16)
119f7184619SJoshua M. Clulow 		dhx->dhx_mode = SIZE16;
120f7184619SJoshua M. Clulow 	else if (dhp->dh_flags & DIS_X86_SIZE64)
121f7184619SJoshua M. Clulow 		dhx->dhx_mode = SIZE64;
122dc0093f4Seschrock 	else
123f7184619SJoshua M. Clulow 		dhx->dhx_mode = SIZE32;
124dc0093f4Seschrock 
125f7184619SJoshua M. Clulow 	if (dhp->dh_flags & DIS_OCTAL)
126f7184619SJoshua M. Clulow 		dhx->dhx_dis.d86_flags = DIS_F_OCTAL;
127dc0093f4Seschrock 
128f7184619SJoshua M. Clulow 	dhx->dhx_dis.d86_sprintf_func = dis_snprintf;
129f7184619SJoshua M. Clulow 	dhx->dhx_dis.d86_get_byte = get_byte;
130f7184619SJoshua M. Clulow 	dhx->dhx_dis.d86_sym_lookup = do_lookup;
131f7184619SJoshua M. Clulow 	dhx->dhx_dis.d86_check_func = check_func;
132dc0093f4Seschrock 
133f7184619SJoshua M. Clulow 	dhx->dhx_dis.d86_data = dhp;
134dc0093f4Seschrock 
135f7184619SJoshua M. Clulow 	return (0);
136dc0093f4Seschrock }
137dc0093f4Seschrock 
138f7184619SJoshua M. Clulow static int
139f7184619SJoshua M. Clulow dis_i386_disassemble(dis_handle_t *dhp, uint64_t addr, char *buf,
140f7184619SJoshua M. Clulow     size_t buflen)
141dc0093f4Seschrock {
142f7184619SJoshua M. Clulow 	dis_handle_i386_t *dhx = dhp->dh_arch_private;
143dc0093f4Seschrock 	dhp->dh_addr = addr;
144dc0093f4Seschrock 
145e0070315Sdmick 	/* DIS_NOIMMSYM might not be set until now, so update */
146e0070315Sdmick 	if (dhp->dh_flags & DIS_NOIMMSYM)
147f7184619SJoshua M. Clulow 		dhx->dhx_dis.d86_flags |= DIS_F_NOIMMSYM;
148e0070315Sdmick 	else
149f7184619SJoshua M. Clulow 		dhx->dhx_dis.d86_flags &= ~DIS_F_NOIMMSYM;
150e0070315Sdmick 
151f7184619SJoshua M. Clulow 	if (dtrace_disx86(&dhx->dhx_dis, dhx->dhx_mode) != 0)
152dc0093f4Seschrock 		return (-1);
153dc0093f4Seschrock 
154dc0093f4Seschrock 	if (buf != NULL)
155f7184619SJoshua M. Clulow 		dtrace_disx86_str(&dhx->dhx_dis, dhx->dhx_mode, addr, buf,
156dc0093f4Seschrock 		    buflen);
157dc0093f4Seschrock 
158dc0093f4Seschrock 	return (0);
159dc0093f4Seschrock }
160dc0093f4Seschrock 
161f7184619SJoshua M. Clulow /* ARGSUSED */
162f7184619SJoshua M. Clulow static int
163f7184619SJoshua M. Clulow dis_i386_max_instrlen(dis_handle_t *dhp)
164e0070315Sdmick {
165f7184619SJoshua M. Clulow 	return (15);
166e0070315Sdmick }
167e0070315Sdmick 
168dc0093f4Seschrock /* ARGSUSED */
169f7184619SJoshua M. Clulow static int
170f7184619SJoshua M. Clulow dis_i386_min_instrlen(dis_handle_t *dhp)
171dc0093f4Seschrock {
172f7184619SJoshua M. Clulow 	return (1);
173dc0093f4Seschrock }
174dc0093f4Seschrock 
175dc0093f4Seschrock /*
176dc0093f4Seschrock  * Return the previous instruction.  On x86, we have no choice except to
177dc0093f4Seschrock  * disassemble everything from the start of the symbol, and stop when we have
178dc0093f4Seschrock  * reached our instruction address.  If we're not in the middle of a known
179dc0093f4Seschrock  * symbol, then we return the same address to indicate failure.
180dc0093f4Seschrock  */
181f7184619SJoshua M. Clulow static uint64_t
182f7184619SJoshua M. Clulow dis_i386_previnstr(dis_handle_t *dhp, uint64_t pc, int n)
183dc0093f4Seschrock {
184dc0093f4Seschrock 	uint64_t *hist, addr, start;
185dc0093f4Seschrock 	int cur, nseen;
186dc0093f4Seschrock 	uint64_t res = pc;
187dc0093f4Seschrock 
188d79705c6Seschrock 	if (n <= 0)
189d79705c6Seschrock 		return (pc);
190d79705c6Seschrock 
191dc0093f4Seschrock 	if (dhp->dh_lookup(dhp->dh_data, pc, NULL, 0, &start, NULL) != 0 ||
192dc0093f4Seschrock 	    start == pc)
193dc0093f4Seschrock 		return (res);
194dc0093f4Seschrock 
195dc0093f4Seschrock 	hist = dis_zalloc(sizeof (uint64_t) * n);
196dc0093f4Seschrock 
197dc0093f4Seschrock 	for (cur = 0, nseen = 0, addr = start; addr < pc; addr = dhp->dh_addr) {
198dc0093f4Seschrock 		hist[cur] = addr;
199dc0093f4Seschrock 		cur = (cur + 1) % n;
200dc0093f4Seschrock 		nseen++;
201dc0093f4Seschrock 
202dc0093f4Seschrock 		/* if we cannot make forward progress, give up */
203dc0093f4Seschrock 		if (dis_disassemble(dhp, addr, NULL, 0) != 0)
204dc0093f4Seschrock 			goto done;
205dc0093f4Seschrock 	}
206dc0093f4Seschrock 
207dc0093f4Seschrock 	if (addr != pc) {
208dc0093f4Seschrock 		/*
209dc0093f4Seschrock 		 * We scanned past %pc, but didn't find an instruction that
210dc0093f4Seschrock 		 * started at %pc.  This means that either the caller specified
211dc0093f4Seschrock 		 * an invalid address, or we ran into something other than code
212dc0093f4Seschrock 		 * during our scan.  Virtually any combination of bytes can be
213dc0093f4Seschrock 		 * construed as a valid Intel instruction, so any non-code bytes
214dc0093f4Seschrock 		 * we encounter will have thrown off the scan.
215dc0093f4Seschrock 		 */
216dc0093f4Seschrock 		goto done;
217dc0093f4Seschrock 	}
218dc0093f4Seschrock 
219dc0093f4Seschrock 	res = hist[(cur + n - MIN(n, nseen)) % n];
220dc0093f4Seschrock 
221dc0093f4Seschrock done:
222dc0093f4Seschrock 	dis_free(hist, sizeof (uint64_t) * n);
223dc0093f4Seschrock 	return (res);
224dc0093f4Seschrock }
225702941cdSRichard Lowe 
226f7184619SJoshua M. Clulow static int
227f7184619SJoshua M. Clulow dis_i386_supports_flags(int flags)
228f7184619SJoshua M. Clulow {
229f7184619SJoshua M. Clulow 	int archflags = flags & DIS_ARCH_MASK;
230f7184619SJoshua M. Clulow 
231f7184619SJoshua M. Clulow 	if (archflags == DIS_X86_SIZE16 || archflags == DIS_X86_SIZE32 ||
232f7184619SJoshua M. Clulow 	    archflags == DIS_X86_SIZE64)
233f7184619SJoshua M. Clulow 		return (1);
234f7184619SJoshua M. Clulow 
235f7184619SJoshua M. Clulow 	return (0);
236f7184619SJoshua M. Clulow }
237f7184619SJoshua M. Clulow 
238f7184619SJoshua M. Clulow static int
239f7184619SJoshua M. Clulow dis_i386_instrlen(dis_handle_t *dhp, uint64_t pc)
240702941cdSRichard Lowe {
241702941cdSRichard Lowe 	if (dis_disassemble(dhp, pc, NULL, 0) != 0)
242702941cdSRichard Lowe 		return (-1);
243702941cdSRichard Lowe 
244702941cdSRichard Lowe 	return (dhp->dh_addr - pc);
245702941cdSRichard Lowe }
246f7184619SJoshua M. Clulow 
247f7184619SJoshua M. Clulow dis_arch_t dis_arch_i386 = {
248*b3457a09SJosef 'Jeff' Sipek 	.da_supports_flags	= dis_i386_supports_flags,
249*b3457a09SJosef 'Jeff' Sipek 	.da_handle_attach	= dis_i386_handle_attach,
250*b3457a09SJosef 'Jeff' Sipek 	.da_handle_detach	= dis_i386_handle_detach,
251*b3457a09SJosef 'Jeff' Sipek 	.da_disassemble		= dis_i386_disassemble,
252*b3457a09SJosef 'Jeff' Sipek 	.da_previnstr		= dis_i386_previnstr,
253*b3457a09SJosef 'Jeff' Sipek 	.da_min_instrlen	= dis_i386_min_instrlen,
254*b3457a09SJosef 'Jeff' Sipek 	.da_max_instrlen	= dis_i386_max_instrlen,
255*b3457a09SJosef 'Jeff' Sipek 	.da_instrlen		= dis_i386_instrlen,
256f7184619SJoshua M. Clulow };
257