ACPI AML decompiler w/ CFG recovery and structured pseudocode
Add deterministic corpus regression runner across parse decompile disassemble and JSON
author
vm.fail
date
(Jun 29, 2026, 1:08 AM UTC)
commit
034b4891
034b4891619aa3a5698eb96b490904a061eeddc7
parent
a8aa84ef
a8aa84eff429991b5ae88ad8daaaa18dac268e72
vm.fail
date
(Jun 29, 2026, 1:08 AM UTC)
commit
034b4891
034b4891619aa3a5698eb96b490904a061eeddc7
parent
a8aa84ef
a8aa84eff429991b5ae88ad8daaaa18dac268e72
+154
-3
+2
-2
CMakeLists.txt
+2
-2
CMakeLists.txt
···
41
41
add_executable(tobi_test
42
42
test/main.c test/t_rd.c test/t_nm.c test/t_p.c test/t_ir.c
43
43
test/t_cf.c test/t_da.c test/t_dc.c test/t_js.c test/t_bad.c test/t_cli.c
44
-
test/t_prod.c)
44
+
test/t_prod.c test/t_corpus.c)
45
45
target_link_libraries(tobi_test PRIVATE tobilib)
46
46
target_compile_options(tobi_test PRIVATE ${TOBI_WARN})
47
47
target_compile_definitions(tobi_test PRIVATE
···
52
52
target_link_options(tobi_test PRIVATE ${SAN_FLAGS})
53
53
endif()
54
54
55
-
foreach(t rd nm p ir cf da dc js bad cli prod)
55
+
foreach(t rd nm p ir cf da dc js bad cli prod corpus)
56
56
add_test(NAME ${t} COMMAND tobi_test ${t})
57
57
endforeach()
58
58
+1
-1
test/main.c
+1
-1
test/main.c
···
4
4
struct ent { const char *name; int (*fn)(void); } tests[] = {
5
5
{"rd", t_rd}, {"nm", t_nm}, {"p", t_p}, {"ir", t_ir}, {"cf", t_cf},
6
6
{"da", t_da}, {"dc", t_dc}, {"js", t_js}, {"bad", t_bad}, {"cli", t_cli},
7
-
{"prod", t_prod},
7
+
{"prod", t_prod}, {"corpus", t_corpus},
8
8
};
9
9
if (argc == 2) {
10
10
for (size_t i = 0; i < sizeof(tests) / sizeof(tests[0]); i++) {
+1
test/t.h
+1
test/t.h
+150
test/t_corpus.c
+150
test/t_corpus.c
···
1
+
#include "t.h"
2
+
3
+
#include "cf.h"
4
+
#include "da.h"
5
+
#include "dc.h"
6
+
#include "js.h"
7
+
#include "p.h"
8
+
9
+
#include <dirent.h>
10
+
#include <errno.h>
11
+
#include <sys/stat.h>
12
+
13
+
#ifndef TOBI_SRC_DIR
14
+
#define TOBI_SRC_DIR "."
15
+
#endif
16
+
17
+
static int read_blob(const char *path, uint8_t **data, size_t *len) {
18
+
FILE *fp = fopen(path, "rb");
19
+
if (!fp) {
20
+
fprintf(stderr, "open %s: %s\n", path, strerror(errno));
21
+
return 0;
22
+
}
23
+
if (fseek(fp, 0, SEEK_END) != 0) {
24
+
fclose(fp);
25
+
return 0;
26
+
}
27
+
long n = ftell(fp);
28
+
if (n < 0) {
29
+
fclose(fp);
30
+
return 0;
31
+
}
32
+
rewind(fp);
33
+
*len = (size_t)n;
34
+
*data = malloc(*len ? *len : 1u);
35
+
if (!*data) {
36
+
fclose(fp);
37
+
return 0;
38
+
}
39
+
if (*len && fread(*data, 1, *len, fp) != *len) {
40
+
free(*data);
41
+
fclose(fp);
42
+
return 0;
43
+
}
44
+
fclose(fp);
45
+
return 1;
46
+
}
47
+
48
+
static int stable_emit_blob(const char *label, const uint8_t *data, size_t len, int strict_should_pass) {
49
+
int ok = 0;
50
+
tobi_parse_result a;
51
+
tobi_parse_result b;
52
+
tobi_parse_result strict;
53
+
(void)tobi_parse(data, len, 0, &a);
54
+
(void)tobi_parse(data, len, 0, &b);
55
+
int strict_ok = tobi_parse(data, len, 1, &strict);
56
+
int expected_strict_ok = strict_should_pass && a.diag.len == 0;
57
+
(void)tobi_cf_recover(a.root, &a.diag);
58
+
(void)tobi_cf_recover(b.root, &b.diag);
59
+
char *aj = tobi_js_emit(&a);
60
+
char *bj = tobi_js_emit(&b);
61
+
char *ad = tobi_dc_emit(a.root, &a.diag);
62
+
char *bd = tobi_dc_emit(b.root, &b.diag);
63
+
char *aa = tobi_da_emit(data, len, &a.meta);
64
+
char *ba = tobi_da_emit(data, len, &b.meta);
65
+
ok = aj != NULL && bj != NULL && ad != NULL && bd != NULL && aa != NULL && ba != NULL &&
66
+
strcmp(aj, bj) == 0 && aj[0] == '{' && strstr(aj, "\"meta\"") != NULL &&
67
+
strstr(aj, "\"ir\"") != NULL && strstr(aj, ",}") == NULL && strstr(aj, ",]") == NULL &&
68
+
ad[0] != '\0' && aa[0] != '\0' && strcmp(ad, bd) == 0 && strcmp(aa, ba) == 0;
69
+
if (expected_strict_ok) {
70
+
ok = ok && strict_ok && !tobi_diag_has_error(&strict.diag);
71
+
} else {
72
+
ok = ok && (!strict_ok || tobi_diag_has_error(&strict.diag));
73
+
}
74
+
if (!ok) {
75
+
fprintf(stderr, "corpus regression failed for %s\n", label);
76
+
}
77
+
free(aa);
78
+
free(ba);
79
+
free(ad);
80
+
free(bd);
81
+
free(aj);
82
+
free(bj);
83
+
tobi_parse_result_free(&a);
84
+
tobi_parse_result_free(&b);
85
+
tobi_parse_result_free(&strict);
86
+
return ok ? 0 : 1;
87
+
}
88
+
89
+
static int stable_emit(const char *path, int strict_should_pass) {
90
+
uint8_t *data = NULL;
91
+
size_t len = 0;
92
+
int rc = 1;
93
+
if (!read_blob(path, &data, &len)) {
94
+
return 1;
95
+
}
96
+
rc = stable_emit_blob(path, data, len, strict_should_pass);
97
+
if (rc == 0 && len > 1) {
98
+
size_t cuts[] = {1u, len / 2u, len - 1u};
99
+
for (size_t i = 0; i < sizeof(cuts) / sizeof(cuts[0]); i++) {
100
+
if (cuts[i] == 0 || cuts[i] >= len) {
101
+
continue;
102
+
}
103
+
char label[900];
104
+
snprintf(label, sizeof(label), "%s truncated:%zu", path, cuts[i]);
105
+
if (stable_emit_blob(label, data, cuts[i], 0) != 0) {
106
+
rc = 1;
107
+
break;
108
+
}
109
+
}
110
+
}
111
+
free(data);
112
+
return rc;
113
+
}
114
+
115
+
static int walk_dir(const char *rel, size_t *count, size_t *bad_count) {
116
+
char dir_path[512];
117
+
snprintf(dir_path, sizeof(dir_path), "%s/%s", TOBI_SRC_DIR, rel);
118
+
DIR *dir = opendir(dir_path);
119
+
T_CHECK(dir != NULL);
120
+
struct dirent *de = NULL;
121
+
while ((de = readdir(dir)) != NULL) {
122
+
if (de->d_name[0] == '.') {
123
+
continue;
124
+
}
125
+
char path[768];
126
+
snprintf(path, sizeof(path), "%s/%s", dir_path, de->d_name);
127
+
struct stat st;
128
+
T_CHECK(stat(path, &st) == 0);
129
+
if (S_ISREG(st.st_mode)) {
130
+
int bad = strstr(de->d_name, "bad") != NULL;
131
+
T_CHECK(stable_emit(path, !bad) == 0);
132
+
(*count)++;
133
+
if (bad) {
134
+
(*bad_count)++;
135
+
}
136
+
}
137
+
}
138
+
T_CHECK(closedir(dir) == 0);
139
+
return 0;
140
+
}
141
+
142
+
int t_corpus(void) {
143
+
size_t count = 0;
144
+
size_t bad_count = 0;
145
+
T_CHECK(walk_dir("sam", &count, &bad_count) == 0);
146
+
T_CHECK(walk_dir("fuzz/seed", &count, &bad_count) == 0);
147
+
T_CHECK(count >= 9);
148
+
T_CHECK(bad_count >= 2);
149
+
return 0;
150
+
}