xref: /linux/arch/x86/tools/vdso2c.c (revision c17ee635fd3a482b2ad2bf5e269755c2eae5f25e)
1*a76108d0SH. Peter Anvin // SPDX-License-Identifier: GPL-2.0-only
2*a76108d0SH. Peter Anvin /*
3*a76108d0SH. Peter Anvin  * vdso2c - A vdso image preparation tool
4*a76108d0SH. Peter Anvin  * Copyright (c) 2014 Andy Lutomirski and others
5*a76108d0SH. Peter Anvin  *
6*a76108d0SH. Peter Anvin  * vdso2c requires stripped and unstripped input.  It would be trivial
7*a76108d0SH. Peter Anvin  * to fully strip the input in here, but, for reasons described below,
8*a76108d0SH. Peter Anvin  * we need to write a section table.  Doing this is more or less
9*a76108d0SH. Peter Anvin  * equivalent to dropping all non-allocatable sections, but it's
10*a76108d0SH. Peter Anvin  * easier to let objcopy handle that instead of doing it ourselves.
11*a76108d0SH. Peter Anvin  * If we ever need to do something fancier than what objcopy provides,
12*a76108d0SH. Peter Anvin  * it would be straightforward to add here.
13*a76108d0SH. Peter Anvin  *
14*a76108d0SH. Peter Anvin  * We're keep a section table for a few reasons:
15*a76108d0SH. Peter Anvin  *
16*a76108d0SH. Peter Anvin  * The Go runtime had a couple of bugs: it would read the section
17*a76108d0SH. Peter Anvin  * table to try to figure out how many dynamic symbols there were (it
18*a76108d0SH. Peter Anvin  * shouldn't have looked at the section table at all) and, if there
19*a76108d0SH. Peter Anvin  * were no SHT_SYNDYM section table entry, it would use an
20*a76108d0SH. Peter Anvin  * uninitialized value for the number of symbols.  An empty DYNSYM
21*a76108d0SH. Peter Anvin  * table would work, but I see no reason not to write a valid one (and
22*a76108d0SH. Peter Anvin  * keep full performance for old Go programs).  This hack is only
23*a76108d0SH. Peter Anvin  * needed on x86_64.
24*a76108d0SH. Peter Anvin  *
25*a76108d0SH. Peter Anvin  * The bug was introduced on 2012-08-31 by:
26*a76108d0SH. Peter Anvin  * https://code.google.com/p/go/source/detail?r=56ea40aac72b
27*a76108d0SH. Peter Anvin  * and was fixed on 2014-06-13 by:
28*a76108d0SH. Peter Anvin  * https://code.google.com/p/go/source/detail?r=fc1cd5e12595
29*a76108d0SH. Peter Anvin  *
30*a76108d0SH. Peter Anvin  * Binutils has issues debugging the vDSO: it reads the section table to
31*a76108d0SH. Peter Anvin  * find SHT_NOTE; it won't look at PT_NOTE for the in-memory vDSO, which
32*a76108d0SH. Peter Anvin  * would break build-id if we removed the section table.  Binutils
33*a76108d0SH. Peter Anvin  * also requires that shstrndx != 0.  See:
34*a76108d0SH. Peter Anvin  * https://sourceware.org/bugzilla/show_bug.cgi?id=17064
35*a76108d0SH. Peter Anvin  *
36*a76108d0SH. Peter Anvin  * elfutils might not look for PT_NOTE if there is a section table at
37*a76108d0SH. Peter Anvin  * all.  I don't know whether this matters for any practical purpose.
38*a76108d0SH. Peter Anvin  *
39*a76108d0SH. Peter Anvin  * For simplicity, rather than hacking up a partial section table, we
40*a76108d0SH. Peter Anvin  * just write a mostly complete one.  We omit non-dynamic symbols,
41*a76108d0SH. Peter Anvin  * though, since they're rather large.
42*a76108d0SH. Peter Anvin  *
43*a76108d0SH. Peter Anvin  * Once binutils gets fixed, we might be able to drop this for all but
44*a76108d0SH. Peter Anvin  * the 64-bit vdso, since build-id only works in kernel RPMs, and
45*a76108d0SH. Peter Anvin  * systems that update to new enough kernel RPMs will likely update
46*a76108d0SH. Peter Anvin  * binutils in sync.  build-id has never worked for home-built kernel
47*a76108d0SH. Peter Anvin  * RPMs without manual symlinking, and I suspect that no one ever does
48*a76108d0SH. Peter Anvin  * that.
49*a76108d0SH. Peter Anvin  */
50*a76108d0SH. Peter Anvin 
51*a76108d0SH. Peter Anvin #include <inttypes.h>
52*a76108d0SH. Peter Anvin #include <stdint.h>
53*a76108d0SH. Peter Anvin #include <unistd.h>
54*a76108d0SH. Peter Anvin #include <stdarg.h>
55*a76108d0SH. Peter Anvin #include <stdlib.h>
56*a76108d0SH. Peter Anvin #include <stdio.h>
57*a76108d0SH. Peter Anvin #include <string.h>
58*a76108d0SH. Peter Anvin #include <fcntl.h>
59*a76108d0SH. Peter Anvin #include <err.h>
60*a76108d0SH. Peter Anvin 
61*a76108d0SH. Peter Anvin #include <sys/mman.h>
62*a76108d0SH. Peter Anvin #include <sys/types.h>
63*a76108d0SH. Peter Anvin 
64*a76108d0SH. Peter Anvin #include <tools/le_byteshift.h>
65*a76108d0SH. Peter Anvin 
66*a76108d0SH. Peter Anvin #include <linux/elf.h>
67*a76108d0SH. Peter Anvin #include <linux/types.h>
68*a76108d0SH. Peter Anvin #include <linux/kernel.h>
69*a76108d0SH. Peter Anvin 
70*a76108d0SH. Peter Anvin const char *outfilename;
71*a76108d0SH. Peter Anvin 
72*a76108d0SH. Peter Anvin struct vdso_sym {
73*a76108d0SH. Peter Anvin 	const char *name;
74*a76108d0SH. Peter Anvin 	bool export;
75*a76108d0SH. Peter Anvin };
76*a76108d0SH. Peter Anvin 
77*a76108d0SH. Peter Anvin struct vdso_sym required_syms[] = {
78*a76108d0SH. Peter Anvin 	{"VDSO32_NOTE_MASK", true},
79*a76108d0SH. Peter Anvin 	{"__kernel_vsyscall", true},
80*a76108d0SH. Peter Anvin 	{"__kernel_sigreturn", true},
81*a76108d0SH. Peter Anvin 	{"__kernel_rt_sigreturn", true},
82*a76108d0SH. Peter Anvin 	{"int80_landing_pad", true},
83*a76108d0SH. Peter Anvin 	{"vdso32_rt_sigreturn_landing_pad", true},
84*a76108d0SH. Peter Anvin 	{"vdso32_sigreturn_landing_pad", true},
85*a76108d0SH. Peter Anvin };
86*a76108d0SH. Peter Anvin 
87*a76108d0SH. Peter Anvin __attribute__((format(printf, 1, 2))) __attribute__((noreturn))
88*a76108d0SH. Peter Anvin static void fail(const char *format, ...)
89*a76108d0SH. Peter Anvin {
90*a76108d0SH. Peter Anvin 	va_list ap;
91*a76108d0SH. Peter Anvin 	va_start(ap, format);
92*a76108d0SH. Peter Anvin 	fprintf(stderr, "Error: ");
93*a76108d0SH. Peter Anvin 	vfprintf(stderr, format, ap);
94*a76108d0SH. Peter Anvin 	if (outfilename)
95*a76108d0SH. Peter Anvin 		unlink(outfilename);
96*a76108d0SH. Peter Anvin 	exit(1);
97*a76108d0SH. Peter Anvin 	va_end(ap);
98*a76108d0SH. Peter Anvin }
99*a76108d0SH. Peter Anvin 
100*a76108d0SH. Peter Anvin /*
101*a76108d0SH. Peter Anvin  * Evil macros for little-endian reads and writes
102*a76108d0SH. Peter Anvin  */
103*a76108d0SH. Peter Anvin #define GLE(x, bits, ifnot)						\
104*a76108d0SH. Peter Anvin 	__builtin_choose_expr(						\
105*a76108d0SH. Peter Anvin 		(sizeof(*(x)) == bits/8),				\
106*a76108d0SH. Peter Anvin 		(__typeof__(*(x)))get_unaligned_le##bits(x), ifnot)
107*a76108d0SH. Peter Anvin 
108*a76108d0SH. Peter Anvin extern void bad_get_le(void);
109*a76108d0SH. Peter Anvin #define LAST_GLE(x)							\
110*a76108d0SH. Peter Anvin 	__builtin_choose_expr(sizeof(*(x)) == 1, *(x), bad_get_le())
111*a76108d0SH. Peter Anvin 
112*a76108d0SH. Peter Anvin #define GET_LE(x)							\
113*a76108d0SH. Peter Anvin 	GLE(x, 64, GLE(x, 32, GLE(x, 16, LAST_GLE(x))))
114*a76108d0SH. Peter Anvin 
115*a76108d0SH. Peter Anvin #define PLE(x, val, bits, ifnot)					\
116*a76108d0SH. Peter Anvin 	__builtin_choose_expr(						\
117*a76108d0SH. Peter Anvin 		(sizeof(*(x)) == bits/8),				\
118*a76108d0SH. Peter Anvin 		put_unaligned_le##bits((val), (x)), ifnot)
119*a76108d0SH. Peter Anvin 
120*a76108d0SH. Peter Anvin extern void bad_put_le(void);
121*a76108d0SH. Peter Anvin #define LAST_PLE(x, val)						\
122*a76108d0SH. Peter Anvin 	__builtin_choose_expr(sizeof(*(x)) == 1, *(x) = (val), bad_put_le())
123*a76108d0SH. Peter Anvin 
124*a76108d0SH. Peter Anvin #define PUT_LE(x, val)					\
125*a76108d0SH. Peter Anvin 	PLE(x, val, 64, PLE(x, val, 32, PLE(x, val, 16, LAST_PLE(x, val))))
126*a76108d0SH. Peter Anvin 
127*a76108d0SH. Peter Anvin 
128*a76108d0SH. Peter Anvin #define NSYMS ARRAY_SIZE(required_syms)
129*a76108d0SH. Peter Anvin 
130*a76108d0SH. Peter Anvin #define BITSFUNC3(name, bits, suffix) name##bits##suffix
131*a76108d0SH. Peter Anvin #define BITSFUNC2(name, bits, suffix) BITSFUNC3(name, bits, suffix)
132*a76108d0SH. Peter Anvin #define BITSFUNC(name) BITSFUNC2(name, ELF_BITS, )
133*a76108d0SH. Peter Anvin 
134*a76108d0SH. Peter Anvin #define INT_BITS BITSFUNC2(int, ELF_BITS, _t)
135*a76108d0SH. Peter Anvin 
136*a76108d0SH. Peter Anvin #define ELF_BITS_XFORM2(bits, x) Elf##bits##_##x
137*a76108d0SH. Peter Anvin #define ELF_BITS_XFORM(bits, x) ELF_BITS_XFORM2(bits, x)
138*a76108d0SH. Peter Anvin #define ELF(x) ELF_BITS_XFORM(ELF_BITS, x)
139*a76108d0SH. Peter Anvin 
140*a76108d0SH. Peter Anvin #define ELF_BITS 64
141*a76108d0SH. Peter Anvin #include "vdso2c.h"
142*a76108d0SH. Peter Anvin #undef ELF_BITS
143*a76108d0SH. Peter Anvin 
144*a76108d0SH. Peter Anvin #define ELF_BITS 32
145*a76108d0SH. Peter Anvin #include "vdso2c.h"
146*a76108d0SH. Peter Anvin #undef ELF_BITS
147*a76108d0SH. Peter Anvin 
148*a76108d0SH. Peter Anvin static void go(void *raw_addr, size_t raw_len,
149*a76108d0SH. Peter Anvin 	       void *stripped_addr, size_t stripped_len,
150*a76108d0SH. Peter Anvin 	       FILE *outfile, const char *name)
151*a76108d0SH. Peter Anvin {
152*a76108d0SH. Peter Anvin 	Elf64_Ehdr *hdr = (Elf64_Ehdr *)raw_addr;
153*a76108d0SH. Peter Anvin 
154*a76108d0SH. Peter Anvin 	if (hdr->e_ident[EI_CLASS] == ELFCLASS64) {
155*a76108d0SH. Peter Anvin 		go64(raw_addr, raw_len, stripped_addr, stripped_len,
156*a76108d0SH. Peter Anvin 		     outfile, name);
157*a76108d0SH. Peter Anvin 	} else if (hdr->e_ident[EI_CLASS] == ELFCLASS32) {
158*a76108d0SH. Peter Anvin 		go32(raw_addr, raw_len, stripped_addr, stripped_len,
159*a76108d0SH. Peter Anvin 		     outfile, name);
160*a76108d0SH. Peter Anvin 	} else {
161*a76108d0SH. Peter Anvin 		fail("unknown ELF class\n");
162*a76108d0SH. Peter Anvin 	}
163*a76108d0SH. Peter Anvin }
164*a76108d0SH. Peter Anvin 
165*a76108d0SH. Peter Anvin static void map_input(const char *name, void **addr, size_t *len, int prot)
166*a76108d0SH. Peter Anvin {
167*a76108d0SH. Peter Anvin 	off_t tmp_len;
168*a76108d0SH. Peter Anvin 
169*a76108d0SH. Peter Anvin 	int fd = open(name, O_RDONLY);
170*a76108d0SH. Peter Anvin 	if (fd == -1)
171*a76108d0SH. Peter Anvin 		err(1, "open(%s)", name);
172*a76108d0SH. Peter Anvin 
173*a76108d0SH. Peter Anvin 	tmp_len = lseek(fd, 0, SEEK_END);
174*a76108d0SH. Peter Anvin 	if (tmp_len == (off_t)-1)
175*a76108d0SH. Peter Anvin 		err(1, "lseek");
176*a76108d0SH. Peter Anvin 	*len = (size_t)tmp_len;
177*a76108d0SH. Peter Anvin 
178*a76108d0SH. Peter Anvin 	*addr = mmap(NULL, tmp_len, prot, MAP_PRIVATE, fd, 0);
179*a76108d0SH. Peter Anvin 	if (*addr == MAP_FAILED)
180*a76108d0SH. Peter Anvin 		err(1, "mmap");
181*a76108d0SH. Peter Anvin 
182*a76108d0SH. Peter Anvin 	close(fd);
183*a76108d0SH. Peter Anvin }
184*a76108d0SH. Peter Anvin 
185*a76108d0SH. Peter Anvin int main(int argc, char **argv)
186*a76108d0SH. Peter Anvin {
187*a76108d0SH. Peter Anvin 	size_t raw_len, stripped_len;
188*a76108d0SH. Peter Anvin 	void *raw_addr, *stripped_addr;
189*a76108d0SH. Peter Anvin 	FILE *outfile;
190*a76108d0SH. Peter Anvin 	char *name, *tmp;
191*a76108d0SH. Peter Anvin 	int namelen;
192*a76108d0SH. Peter Anvin 
193*a76108d0SH. Peter Anvin 	if (argc != 4) {
194*a76108d0SH. Peter Anvin 		printf("Usage: vdso2c RAW_INPUT STRIPPED_INPUT OUTPUT\n");
195*a76108d0SH. Peter Anvin 		return 1;
196*a76108d0SH. Peter Anvin 	}
197*a76108d0SH. Peter Anvin 
198*a76108d0SH. Peter Anvin 	/*
199*a76108d0SH. Peter Anvin 	 * Figure out the struct name.  If we're writing to a .so file,
200*a76108d0SH. Peter Anvin 	 * generate raw output instead.
201*a76108d0SH. Peter Anvin 	 */
202*a76108d0SH. Peter Anvin 	name = strdup(argv[3]);
203*a76108d0SH. Peter Anvin 	namelen = strlen(name);
204*a76108d0SH. Peter Anvin 	if (namelen >= 3 && !strcmp(name + namelen - 3, ".so")) {
205*a76108d0SH. Peter Anvin 		name = NULL;
206*a76108d0SH. Peter Anvin 	} else {
207*a76108d0SH. Peter Anvin 		tmp = strrchr(name, '/');
208*a76108d0SH. Peter Anvin 		if (tmp)
209*a76108d0SH. Peter Anvin 			name = tmp + 1;
210*a76108d0SH. Peter Anvin 		tmp = strchr(name, '.');
211*a76108d0SH. Peter Anvin 		if (tmp)
212*a76108d0SH. Peter Anvin 			*tmp = '\0';
213*a76108d0SH. Peter Anvin 		for (tmp = name; *tmp; tmp++)
214*a76108d0SH. Peter Anvin 			if (*tmp == '-')
215*a76108d0SH. Peter Anvin 				*tmp = '_';
216*a76108d0SH. Peter Anvin 	}
217*a76108d0SH. Peter Anvin 
218*a76108d0SH. Peter Anvin 	map_input(argv[1], &raw_addr, &raw_len, PROT_READ);
219*a76108d0SH. Peter Anvin 	map_input(argv[2], &stripped_addr, &stripped_len, PROT_READ);
220*a76108d0SH. Peter Anvin 
221*a76108d0SH. Peter Anvin 	outfilename = argv[3];
222*a76108d0SH. Peter Anvin 	outfile = fopen(outfilename, "w");
223*a76108d0SH. Peter Anvin 	if (!outfile)
224*a76108d0SH. Peter Anvin 		err(1, "fopen(%s)", outfilename);
225*a76108d0SH. Peter Anvin 
226*a76108d0SH. Peter Anvin 	go(raw_addr, raw_len, stripped_addr, stripped_len, outfile, name);
227*a76108d0SH. Peter Anvin 
228*a76108d0SH. Peter Anvin 	munmap(raw_addr, raw_len);
229*a76108d0SH. Peter Anvin 	munmap(stripped_addr, stripped_len);
230*a76108d0SH. Peter Anvin 	fclose(outfile);
231*a76108d0SH. Peter Anvin 
232*a76108d0SH. Peter Anvin 	return 0;
233*a76108d0SH. Peter Anvin }
234