xref: /onnv-gate/usr/src/cmd/sgs/tools/common/leb128.c (revision 0:68f95e015346)
1*0Sstevel@tonic-gate /*
2*0Sstevel@tonic-gate  * CDDL HEADER START
3*0Sstevel@tonic-gate  *
4*0Sstevel@tonic-gate  * The contents of this file are subject to the terms of the
5*0Sstevel@tonic-gate  * Common Development and Distribution License, Version 1.0 only
6*0Sstevel@tonic-gate  * (the "License").  You may not use this file except in compliance
7*0Sstevel@tonic-gate  * with the License.
8*0Sstevel@tonic-gate  *
9*0Sstevel@tonic-gate  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
10*0Sstevel@tonic-gate  * or http://www.opensolaris.org/os/licensing.
11*0Sstevel@tonic-gate  * See the License for the specific language governing permissions
12*0Sstevel@tonic-gate  * and limitations under the License.
13*0Sstevel@tonic-gate  *
14*0Sstevel@tonic-gate  * When distributing Covered Code, include this CDDL HEADER in each
15*0Sstevel@tonic-gate  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
16*0Sstevel@tonic-gate  * If applicable, add the following below this CDDL HEADER, with the
17*0Sstevel@tonic-gate  * fields enclosed by brackets "[]" replaced with your own identifying
18*0Sstevel@tonic-gate  * information: Portions Copyright [yyyy] [name of copyright owner]
19*0Sstevel@tonic-gate  *
20*0Sstevel@tonic-gate  * CDDL HEADER END
21*0Sstevel@tonic-gate  */
22*0Sstevel@tonic-gate /*
23*0Sstevel@tonic-gate  * Copyright 2004 Sun Microsystems, Inc.  All rights reserved.
24*0Sstevel@tonic-gate  * Use is subject to license terms.
25*0Sstevel@tonic-gate  */
26*0Sstevel@tonic-gate #pragma ident	"%Z%%M%	%I%	%E% SMI"
27*0Sstevel@tonic-gate 
28*0Sstevel@tonic-gate #include <stdio.h>
29*0Sstevel@tonic-gate #include <dwarf.h>
30*0Sstevel@tonic-gate #include <sys/types.h>
31*0Sstevel@tonic-gate #include <sys/elf.h>
32*0Sstevel@tonic-gate #include <gelf.h>
33*0Sstevel@tonic-gate 
34*0Sstevel@tonic-gate /*
35*0Sstevel@tonic-gate  * Little Endian Base 128 (LEB128) numbers.
36*0Sstevel@tonic-gate  * ----------------------------------------
37*0Sstevel@tonic-gate  *
38*0Sstevel@tonic-gate  * LEB128 is a scheme for encoding integers densely that exploits the
39*0Sstevel@tonic-gate  * assumption that most integers are small in magnitude. (This encoding
40*0Sstevel@tonic-gate  * is equally suitable whether the target machine architecture represents
41*0Sstevel@tonic-gate  * data in big-endian or little- endian
42*0Sstevel@tonic-gate  *
43*0Sstevel@tonic-gate  * Unsigned LEB128 numbers are encoded as follows: start at the low order
44*0Sstevel@tonic-gate  * end of an unsigned integer and chop it into 7-bit chunks. Place each
45*0Sstevel@tonic-gate  * chunk into the low order 7 bits of a byte. Typically, several of the
46*0Sstevel@tonic-gate  * high order bytes will be zero; discard them. Emit the remaining bytes in
47*0Sstevel@tonic-gate  * a stream, starting with the low order byte; set the high order bit on
48*0Sstevel@tonic-gate  * each byte except the last emitted byte. The high bit of zero on the last
49*0Sstevel@tonic-gate  * byte indicates to the decoder that it has encountered the last byte.
50*0Sstevel@tonic-gate  * The integer zero is a special case, consisting of a single zero byte.
51*0Sstevel@tonic-gate  *
52*0Sstevel@tonic-gate  * Signed, 2s complement LEB128 numbers are encoded in a similar except
53*0Sstevel@tonic-gate  * that the criterion for discarding high order bytes is not whether they
54*0Sstevel@tonic-gate  * are zero, but whether they consist entirely of sign extension bits.
55*0Sstevel@tonic-gate  * Consider the 32-bit integer -2. The three high level bytes of the number
56*0Sstevel@tonic-gate  * are sign extension, thus LEB128 would represent it as a single byte
57*0Sstevel@tonic-gate  * containing the low order 7 bits, with the high order bit cleared to
58*0Sstevel@tonic-gate  * indicate the end of the byte stream.
59*0Sstevel@tonic-gate  *
60*0Sstevel@tonic-gate  * Note that there is nothing within the LEB128 representation that
61*0Sstevel@tonic-gate  * indicates whether an encoded number is signed or unsigned. The decoder
62*0Sstevel@tonic-gate  * must know what type of number to expect.
63*0Sstevel@tonic-gate  *
64*0Sstevel@tonic-gate  * DWARF Exception Header Encoding
65*0Sstevel@tonic-gate  * -------------------------------
66*0Sstevel@tonic-gate  *
67*0Sstevel@tonic-gate  * The DWARF Exception Header Encoding is used to describe the type of data
68*0Sstevel@tonic-gate  * used in the .eh_frame_hdr section. The upper 4 bits indicate how the
69*0Sstevel@tonic-gate  * value is to be applied. The lower 4 bits indicate the format of the data.
70*0Sstevel@tonic-gate  *
71*0Sstevel@tonic-gate  * DWARF Exception Header value format
72*0Sstevel@tonic-gate  *
73*0Sstevel@tonic-gate  * Name		Value Meaning
74*0Sstevel@tonic-gate  * DW_EH_PE_omit	    0xff No value is present.
75*0Sstevel@tonic-gate  * DW_EH_PE_absptr	    0x00 Value is a void*
76*0Sstevel@tonic-gate  * DW_EH_PE_uleb128	    0x01 Unsigned value is encoded using the
77*0Sstevel@tonic-gate  *				 Little Endian Base 128 (LEB128)
78*0Sstevel@tonic-gate  * DW_EH_PE_udata2	    0x02 A 2 bytes unsigned value.
79*0Sstevel@tonic-gate  * DW_EH_PE_udata4	    0x03 A 4 bytes unsigned value.
80*0Sstevel@tonic-gate  * DW_EH_PE_udata8	    0x04 An 8 bytes unsigned value.
81*0Sstevel@tonic-gate  * DW_EH_PE_signed          0x08 bit on for all signed encodings
82*0Sstevel@tonic-gate  * DW_EH_PE_sleb128	    0x09 Signed value is encoded using the
83*0Sstevel@tonic-gate  *				 Little Endian Base 128 (LEB128)
84*0Sstevel@tonic-gate  * DW_EH_PE_sdata2	    0x0A A 2 bytes signed value.
85*0Sstevel@tonic-gate  * DW_EH_PE_sdata4	    0x0B A 4 bytes signed value.
86*0Sstevel@tonic-gate  * DW_EH_PE_sdata8	    0x0C An 8 bytes signed value.
87*0Sstevel@tonic-gate  *
88*0Sstevel@tonic-gate  * DWARF Exception Header application
89*0Sstevel@tonic-gate  *
90*0Sstevel@tonic-gate  * Name	    Value Meaning
91*0Sstevel@tonic-gate  * DW_EH_PE_absptr	   0x00 Value is used with no modification.
92*0Sstevel@tonic-gate  * DW_EH_PE_pcrel	   0x10 Value is reletive to the location of itself
93*0Sstevel@tonic-gate  * DW_EH_PE_textrel	   0x20
94*0Sstevel@tonic-gate  * DW_EH_PE_datarel	   0x30 Value is reletive to the beginning of the
95*0Sstevel@tonic-gate  *				eh_frame_hdr segment ( segment type
96*0Sstevel@tonic-gate  *			        PT_GNU_EH_FRAME )
97*0Sstevel@tonic-gate  * DW_EH_PE_funcrel        0x40
98*0Sstevel@tonic-gate  * DW_EH_PE_aligned        0x50 value is an aligned void*
99*0Sstevel@tonic-gate  * DW_EH_PE_indirect       0x80 bit to signal indirection after relocation
100*0Sstevel@tonic-gate  * DW_EH_PE_omit	   0xff No value is present.
101*0Sstevel@tonic-gate  *
102*0Sstevel@tonic-gate  */
103*0Sstevel@tonic-gate 
104*0Sstevel@tonic-gate uint64_t
105*0Sstevel@tonic-gate uleb_extract(unsigned char *data, uint64_t *dotp)
106*0Sstevel@tonic-gate {
107*0Sstevel@tonic-gate 	uint64_t	dot = *dotp;
108*0Sstevel@tonic-gate 	uint64_t	res = 0;
109*0Sstevel@tonic-gate 	int		more = 1;
110*0Sstevel@tonic-gate 	int		shift = 0;
111*0Sstevel@tonic-gate 	int		val;
112*0Sstevel@tonic-gate 
113*0Sstevel@tonic-gate 	data += dot;
114*0Sstevel@tonic-gate 
115*0Sstevel@tonic-gate 	while (more) {
116*0Sstevel@tonic-gate 		/*
117*0Sstevel@tonic-gate 		 * Pull off lower 7 bits
118*0Sstevel@tonic-gate 		 */
119*0Sstevel@tonic-gate 		val = (*data) & 0x7f;
120*0Sstevel@tonic-gate 
121*0Sstevel@tonic-gate 		/*
122*0Sstevel@tonic-gate 		 * Add prepend value to head of number.
123*0Sstevel@tonic-gate 		 */
124*0Sstevel@tonic-gate 		res = res | (val << shift);
125*0Sstevel@tonic-gate 
126*0Sstevel@tonic-gate 		/*
127*0Sstevel@tonic-gate 		 * Increment shift & dot pointer
128*0Sstevel@tonic-gate 		 */
129*0Sstevel@tonic-gate 		shift += 7;
130*0Sstevel@tonic-gate 		dot++;
131*0Sstevel@tonic-gate 
132*0Sstevel@tonic-gate 		/*
133*0Sstevel@tonic-gate 		 * Check to see if hi bit is set - if not, this
134*0Sstevel@tonic-gate 		 * is the last byte.
135*0Sstevel@tonic-gate 		 */
136*0Sstevel@tonic-gate 		more = ((*data++) & 0x80) >> 7;
137*0Sstevel@tonic-gate 	}
138*0Sstevel@tonic-gate 	*dotp = dot;
139*0Sstevel@tonic-gate 	return (res);
140*0Sstevel@tonic-gate }
141*0Sstevel@tonic-gate 
142*0Sstevel@tonic-gate int64_t
143*0Sstevel@tonic-gate sleb_extract(unsigned char *data, uint64_t *dotp)
144*0Sstevel@tonic-gate {
145*0Sstevel@tonic-gate 	uint64_t	dot = *dotp;
146*0Sstevel@tonic-gate 	int64_t		res = 0;
147*0Sstevel@tonic-gate 	int		more = 1;
148*0Sstevel@tonic-gate 	int		shift = 0;
149*0Sstevel@tonic-gate 	int		val;
150*0Sstevel@tonic-gate 
151*0Sstevel@tonic-gate 	data += dot;
152*0Sstevel@tonic-gate 
153*0Sstevel@tonic-gate 	while (more) {
154*0Sstevel@tonic-gate 		/*
155*0Sstevel@tonic-gate 		 * Pull off lower 7 bits
156*0Sstevel@tonic-gate 		 */
157*0Sstevel@tonic-gate 		val = (*data) & 0x7f;
158*0Sstevel@tonic-gate 
159*0Sstevel@tonic-gate 		/*
160*0Sstevel@tonic-gate 		 * Add prepend value to head of number.
161*0Sstevel@tonic-gate 		 */
162*0Sstevel@tonic-gate 		res = res | (val << shift);
163*0Sstevel@tonic-gate 
164*0Sstevel@tonic-gate 		/*
165*0Sstevel@tonic-gate 		 * Increment shift & dot pointer
166*0Sstevel@tonic-gate 		 */
167*0Sstevel@tonic-gate 		shift += 7;
168*0Sstevel@tonic-gate 		dot++;
169*0Sstevel@tonic-gate 
170*0Sstevel@tonic-gate 		/*
171*0Sstevel@tonic-gate 		 * Check to see if hi bit is set - if not, this
172*0Sstevel@tonic-gate 		 * is the last byte.
173*0Sstevel@tonic-gate 		 */
174*0Sstevel@tonic-gate 		more = ((*data++) & 0x80) >> 7;
175*0Sstevel@tonic-gate 	}
176*0Sstevel@tonic-gate 	*dotp = dot;
177*0Sstevel@tonic-gate 
178*0Sstevel@tonic-gate 	/*
179*0Sstevel@tonic-gate 	 * Make sure value is properly sign extended.
180*0Sstevel@tonic-gate 	 */
181*0Sstevel@tonic-gate 	res = (res << (64 - shift)) >> (64 - shift);
182*0Sstevel@tonic-gate 
183*0Sstevel@tonic-gate 	return (res);
184*0Sstevel@tonic-gate }
185*0Sstevel@tonic-gate 
186*0Sstevel@tonic-gate uint64_t
187*0Sstevel@tonic-gate dwarf_ehe_extract(unsigned char *data, uint64_t *dotp, uint_t ehe_flags,
188*0Sstevel@tonic-gate     unsigned char *eident, uint64_t pcaddr)
189*0Sstevel@tonic-gate {
190*0Sstevel@tonic-gate 	uint64_t    dot = *dotp;
191*0Sstevel@tonic-gate 	uint_t	    lsb;
192*0Sstevel@tonic-gate 	uint_t	    wordsize;
193*0Sstevel@tonic-gate 	uint_t	    fsize;
194*0Sstevel@tonic-gate 	uint64_t    result;
195*0Sstevel@tonic-gate 
196*0Sstevel@tonic-gate 	if (eident[EI_DATA] == ELFDATA2LSB)
197*0Sstevel@tonic-gate 		lsb = 1;
198*0Sstevel@tonic-gate 	else
199*0Sstevel@tonic-gate 		lsb = 0;
200*0Sstevel@tonic-gate 
201*0Sstevel@tonic-gate 	if (eident[EI_CLASS] == ELFCLASS64)
202*0Sstevel@tonic-gate 		wordsize = 8;
203*0Sstevel@tonic-gate 	else
204*0Sstevel@tonic-gate 		wordsize = 4;
205*0Sstevel@tonic-gate 
206*0Sstevel@tonic-gate 	switch (ehe_flags & 0x0f) {
207*0Sstevel@tonic-gate 	case DW_EH_PE_omit:
208*0Sstevel@tonic-gate 		return (0);
209*0Sstevel@tonic-gate 	case DW_EH_PE_absptr:
210*0Sstevel@tonic-gate 		fsize = wordsize;
211*0Sstevel@tonic-gate 		break;
212*0Sstevel@tonic-gate 	case DW_EH_PE_udata8:
213*0Sstevel@tonic-gate 	case DW_EH_PE_sdata8:
214*0Sstevel@tonic-gate 		fsize = 8;
215*0Sstevel@tonic-gate 		break;
216*0Sstevel@tonic-gate 	case DW_EH_PE_udata4:
217*0Sstevel@tonic-gate 	case DW_EH_PE_sdata4:
218*0Sstevel@tonic-gate 		fsize = 4;
219*0Sstevel@tonic-gate 		break;
220*0Sstevel@tonic-gate 	case DW_EH_PE_udata2:
221*0Sstevel@tonic-gate 	case DW_EH_PE_sdata2:
222*0Sstevel@tonic-gate 		fsize = 2;
223*0Sstevel@tonic-gate 		break;
224*0Sstevel@tonic-gate 	case DW_EH_PE_uleb128:
225*0Sstevel@tonic-gate 		return (uleb_extract(data, dotp));
226*0Sstevel@tonic-gate 	case DW_EH_PE_sleb128:
227*0Sstevel@tonic-gate 		return ((uint64_t)sleb_extract(data, dotp));
228*0Sstevel@tonic-gate 	default:
229*0Sstevel@tonic-gate 		return (0);
230*0Sstevel@tonic-gate 	}
231*0Sstevel@tonic-gate 
232*0Sstevel@tonic-gate 	if (lsb) {
233*0Sstevel@tonic-gate 		/*
234*0Sstevel@tonic-gate 		 * Extract unaligned LSB formated data
235*0Sstevel@tonic-gate 		 */
236*0Sstevel@tonic-gate 		uint_t	cnt;
237*0Sstevel@tonic-gate 
238*0Sstevel@tonic-gate 		result = 0;
239*0Sstevel@tonic-gate 		for (cnt = 0; cnt < fsize;
240*0Sstevel@tonic-gate 		    cnt++, dot++) {
241*0Sstevel@tonic-gate 			uint64_t val;
242*0Sstevel@tonic-gate 			val = data[dot];
243*0Sstevel@tonic-gate 			result |= val << (cnt * 8);
244*0Sstevel@tonic-gate 		}
245*0Sstevel@tonic-gate 	} else {
246*0Sstevel@tonic-gate 		/*
247*0Sstevel@tonic-gate 		 * Extract unaligned MSB formated data
248*0Sstevel@tonic-gate 		 */
249*0Sstevel@tonic-gate 		uint_t	cnt;
250*0Sstevel@tonic-gate 		result = 0;
251*0Sstevel@tonic-gate 		for (cnt = 0; cnt < fsize;
252*0Sstevel@tonic-gate 		    cnt++, dot++) {
253*0Sstevel@tonic-gate 			uint64_t	val;
254*0Sstevel@tonic-gate 			val = data[dot];
255*0Sstevel@tonic-gate 			result |= val << ((fsize - cnt - 1) * 8);
256*0Sstevel@tonic-gate 		}
257*0Sstevel@tonic-gate 	}
258*0Sstevel@tonic-gate 	/*
259*0Sstevel@tonic-gate 	 * perform sign extension
260*0Sstevel@tonic-gate 	 */
261*0Sstevel@tonic-gate 	if ((ehe_flags & DW_EH_PE_signed) &&
262*0Sstevel@tonic-gate 	    (fsize < sizeof (uint64_t))) {
263*0Sstevel@tonic-gate 		int64_t	sresult;
264*0Sstevel@tonic-gate 		uint_t	bitshift;
265*0Sstevel@tonic-gate 		sresult = result;
266*0Sstevel@tonic-gate 		bitshift = (sizeof (uint64_t) - fsize) * 8;
267*0Sstevel@tonic-gate 		sresult = (sresult << bitshift) >> bitshift;
268*0Sstevel@tonic-gate 		result = sresult;
269*0Sstevel@tonic-gate 	}
270*0Sstevel@tonic-gate 
271*0Sstevel@tonic-gate 	/*
272*0Sstevel@tonic-gate 	 * If pcrel and we have a value (ie: we've been
273*0Sstevel@tonic-gate 	 * relocated), then adjust the value.
274*0Sstevel@tonic-gate 	 */
275*0Sstevel@tonic-gate 	if (result && (ehe_flags & DW_EH_PE_pcrel)) {
276*0Sstevel@tonic-gate 		result = pcaddr + result;
277*0Sstevel@tonic-gate 	}
278*0Sstevel@tonic-gate 	*dotp = dot;
279*0Sstevel@tonic-gate 	return (result);
280*0Sstevel@tonic-gate }
281