1 /*
2  * Dwarf pc to source line conversion.
3  * 
4  * Maybe should do the reverse here, but what should the interface look like?
5  * One possibility is to use the Plan 9 line2addr interface:
6  *
7  *	long line2addr(ulong line, ulong basepc)
8  *
9  * which returns the smallest pc > basepc with line number line (ignoring file name).
10  *
11  * The encoding may be small, but it sure isn't simple!
12  */
13 
14 #include <u.h>
15 #include <libc.h>
16 #include <bio.h>
17 #include "elf.h"
18 #include "dwarf.h"
19 
20 #define trace 0
21 
22 enum
23 {
24 	Isstmt = 1<<0,
25 	BasicDwarfBlock = 1<<1,
26 	EndSequence = 1<<2,
27 	PrologueEnd = 1<<3,
28 	EpilogueBegin = 1<<4
29 };
30 
31 typedef struct State State;
32 struct State
33 {
34 	ulong addr;
35 	ulong file;
36 	ulong line;
37 	ulong column;
38 	ulong flags;
39 	ulong isa;
40 };
41 
42 int
43 dwarfpctoline(Dwarf *d, ulong pc, char **cdir, char **dir, char **file, ulong *line, ulong *mtime, ulong *length)
44 {
45 	uchar *prog, *opcount, *end;
46 	ulong off, unit, len, vers, x, start;
47 	int i, first, op, a, l, quantum, isstmt, linebase, linerange, opcodebase, nf;
48 	char *files, *dirs, *s;
49 	DwarfBuf b;
50 	DwarfSym sym;
51 	State emit, cur, reset;
52 	uchar **f, **newf;
53 
54 	f = nil;
55 
56 	if(dwarfaddrtounit(d, pc, &unit) < 0
57 	|| dwarflookuptag(d, unit, TagCompileUnit, &sym) < 0)
58 		return -1;
59 
60 	if(!sym.attrs.have.stmtlist){
61 		werrstr("no line mapping information for 0x%lux", pc);
62 		return -1;
63 	}
64 	off = sym.attrs.stmtlist;
65 	if(off >= d->line.len){
66 		fprint(2, "bad stmtlist\n");
67 		goto bad;
68 	}
69 
70 	if(trace) fprint(2, "unit 0x%lux stmtlist 0x%lux\n", unit, sym.attrs.stmtlist);
71 
72 	memset(&b, 0, sizeof b);
73 	b.d = d;
74 	b.p = d->line.data + off;
75 	b.ep = b.p + d->line.len;
76 	b.addrsize = sym.b.addrsize;	/* should i get this from somewhere else? */
77 
78 	len = dwarfget4(&b);
79 	if(b.p==nil || b.p+len > b.ep || b.p+len < b.p){
80 		fprint(2, "bad len\n");
81 		goto bad;
82 	}
83 
84 	b.ep = b.p+len;
85 	vers = dwarfget2(&b);
86 	if(vers != 2){
87 		werrstr("bad dwarf version 0x%lux", vers);
88 		return -1;
89 	}
90 
91 	len = dwarfget4(&b);
92 	if(b.p==nil || b.p+len > b.ep || b.p+len < b.p){
93 		fprint(2, "another bad len\n");
94 		goto bad;
95 	}
96 	prog = b.p+len;
97 
98 	quantum = dwarfget1(&b);
99 	isstmt = dwarfget1(&b);
100 	linebase = (schar)dwarfget1(&b);
101 	linerange = (schar)dwarfget1(&b);
102 	opcodebase = dwarfget1(&b);
103 
104 	opcount = b.p-1;
105 	dwarfgetnref(&b, opcodebase-1);
106 	if(b.p == nil){
107 		fprint(2, "bad opcode chart\n");
108 		goto bad;
109 	}
110 
111 	/* just skip the files and dirs for now; we'll come back */
112 	dirs = (char*)b.p;
113 	while(b.p!=nil && *b.p!=0)
114 		dwarfgetstring(&b);
115 	dwarfget1(&b);
116 
117 	files = (char*)b.p;
118 	while(b.p!=nil && *b.p!=0){
119 		dwarfgetstring(&b);
120 		dwarfget128(&b);
121 		dwarfget128(&b);
122 		dwarfget128(&b);
123 	}
124 	dwarfget1(&b);
125 
126 	/* move on to the program */
127 	if(b.p == nil || b.p > prog){
128 		fprint(2, "bad header\n");
129 		goto bad;
130 	}
131 	b.p = prog;
132 
133 	reset.addr = 0;
134 	reset.file = 1;
135 	reset.line = 1;
136 	reset.column = 0;
137 	reset.flags = isstmt ? Isstmt : 0;
138 	reset.isa = 0;
139 
140 	cur = reset;
141 	emit = reset;
142 	nf = 0;
143 	start = 0;
144 	if(trace) fprint(2, "program @ %lud ... %.*H opbase = %d\n", b.p - d->line.data, b.ep-b.p, b.p, opcodebase);
145 	first = 1;
146 	while(b.p != nil){
147 		op = dwarfget1(&b);
148 		if(trace) fprint(2, "\tline %lud, addr 0x%lux, op %d %.10H", cur.line, cur.addr, op, b.p);
149 		if(op >= opcodebase){
150 			a = (op - opcodebase) / linerange;
151 			l = (op - opcodebase) % linerange + linebase;
152 			cur.line += l;
153 			cur.addr += a * quantum;
154 			if(trace) fprint(2, " +%d,%d\n", a, l);
155 		emit:
156 			if(first){
157 				if(cur.addr > pc){
158 					werrstr("found wrong line mapping 0x%lux for pc 0x%lux", cur.addr, pc);
159 					goto out;
160 				}
161 				first = 0;
162 				start = cur.addr;
163 			}
164 			if(cur.addr > pc)
165 				break;
166 			if(b.p == nil){
167 				werrstr("buffer underflow in line mapping");
168 				goto out;
169 			}
170 			emit = cur;
171 			if(emit.flags & EndSequence){
172 				werrstr("found wrong line mapping 0x%lux-0x%lux for pc 0x%lux", start, cur.addr, pc);
173 				goto out;
174 			}
175 			cur.flags &= ~(BasicDwarfBlock|PrologueEnd|EpilogueBegin);
176 		}else{
177 			switch(op){
178 			case 0:	/* extended op code */
179 				if(trace) fprint(2, " ext");
180 				len = dwarfget128(&b);
181 				end = b.p+len;
182 				if(b.p == nil || end > b.ep || end < b.p || len < 1)
183 					goto bad;
184 				switch(dwarfget1(&b)){
185 				case 1:	/* end sequence */
186 					if(trace) fprint(2, " end\n");
187 					cur.flags |= EndSequence;
188 					goto emit;
189 				case 2:	/* set address */
190 					cur.addr = dwarfgetaddr(&b);
191 					if(trace) fprint(2, " set pc 0x%lux\n", cur.addr);
192 					break;
193 				case 3:	/* define file */
194 					newf = realloc(f, (nf+1)*sizeof(f[0]));
195 					if(newf == nil)
196 						goto out;
197 					f[nf++] = b.p;
198 					s = dwarfgetstring(&b);
199 					dwarfget128(&b);
200 					dwarfget128(&b);
201 					dwarfget128(&b);
202 					if(trace) fprint(2, " def file %s\n", s);
203 					break;
204 				}
205 				if(b.p == nil || b.p > end)
206 					goto bad;
207 				b.p = end;
208 				break;
209 			case 1:	/* emit */
210 				if(trace) fprint(2, " emit\n");
211 				goto emit;
212 			case 2:	/* advance pc */
213 				a = dwarfget128(&b);
214 				if(trace) fprint(2, " advance pc + %lud\n", a*quantum);
215 				cur.addr += a * quantum;
216 				break;
217 			case 3:	/* advance line */
218 				l = dwarfget128s(&b);
219 				if(trace) fprint(2, " advance line + %ld\n", l);
220 				cur.line += l;
221 				break;
222 			case 4:	/* set file */
223 				if(trace) fprint(2, " set file\n");
224 				cur.file = dwarfget128s(&b);
225 				break;
226 			case 5:	/* set column */
227 				if(trace) fprint(2, " set column\n");
228 				cur.column = dwarfget128(&b);
229 				break;
230 			case 6:	/* negate stmt */
231 				if(trace) fprint(2, " negate stmt\n");
232 				cur.flags ^= Isstmt;
233 				break;
234 			case 7:	/* set basic block */
235 				if(trace) fprint(2, " set basic block\n");
236 				cur.flags |= BasicDwarfBlock;
237 				break;
238 			case 8:	/* const add pc */
239 				a = (255 - opcodebase) / linerange * quantum;
240 				if(trace) fprint(2, " const add pc + %d\n", a);
241 				cur.addr += a;
242 				break;
243 			case 9:	/* fixed advance pc */
244 				a = dwarfget2(&b);
245 				if(trace) fprint(2, " fixed advance pc + %d\n", a);
246 				cur.addr += a;
247 				break;
248 			case 10:	/* set prologue end */
249 				if(trace) fprint(2, " set prologue end\n");
250 				cur.flags |= PrologueEnd;
251 				break;
252 			case 11:	/* set epilogue begin */
253 				if(trace) fprint(2, " set epilogue begin\n");
254 				cur.flags |= EpilogueBegin;
255 				break;
256 			case 12:	/* set isa */
257 				if(trace) fprint(2, " set isa\n");
258 				cur.isa = dwarfget128(&b);
259 				break;
260 			default:	/* something new - skip it */
261 				if(trace) fprint(2, " unknown %d\n", opcount[op]);
262 				for(i=0; i<opcount[op]; i++)
263 					dwarfget128(&b);
264 				break;
265 			}
266 		}
267 	}
268 	if(b.p == nil)
269 		goto bad;
270 
271 	/* finally!  the data we seek is in "emit" */
272 
273 	if(emit.file == 0){
274 		werrstr("invalid file index in mapping data");
275 		goto out;
276 	}
277 	if(line)
278 		*line = emit.line;
279 
280 	/* skip over first emit.file-2 guys */
281 	b.p = (uchar*)files;
282 	for(i=emit.file-1; i > 0 && b.p!=nil && *b.p!=0; i--){
283 		dwarfgetstring(&b);
284 		dwarfget128(&b);
285 		dwarfget128(&b);
286 		dwarfget128(&b);
287 	}
288 	if(b.p == nil){
289 		werrstr("problem parsing file data second time (cannot happen)");
290 		goto bad;
291 	}
292 	if(*b.p == 0){
293 		if(i >= nf){
294 			werrstr("bad file index in mapping data");
295 			goto bad;
296 		}
297 		b.p = f[i];
298 	}
299 	s = dwarfgetstring(&b);
300 	if(file)
301 		*file = s;
302 	i = dwarfget128(&b);		/* directory */
303 	x = dwarfget128(&b);
304 	if(mtime)
305 		*mtime = x;
306 	x = dwarfget128(&b);
307 	if(length)
308 		*length = x;
309 
310 	/* fetch dir name */
311 	if(cdir)
312 		*cdir = sym.attrs.compdir;
313 
314 	if(dir){
315 		if(i == 0)
316 			*dir = nil;
317 		else{
318 			b.p = (uchar*)dirs;
319 			for(i--; i>0 && b.p!=nil && *b.p!=0; i--)
320 				dwarfgetstring(&b);
321 			if(b.p==nil || *b.p==0){
322 				werrstr("bad directory reference in line mapping");
323 				goto out;		/* can only happen with bad dir index */
324 			}
325 			*dir = dwarfgetstring(&b);
326 		}
327 	}
328 
329 	/* free at last, free at last */
330 	free(f);
331 	return 0;
332 
333 bad:
334 	werrstr("corrupted line mapping for 0x%lux", pc);
335 out:
336 	free(f);
337 	return -1;
338 }