Add new interceptor for regex(3) in NetBSD

[krytarowski]

Summary

Add interceptors for the NetBSD style of regex(3) present inside libc:

• regcomp
• regexec
• regerror
• regfree
• regnsub
• regasub

Add a dedicated test verifying the installed interceptors.

[vitalybuka]

lib/sanitizer_common/sanitizer_common_interceptors.inc

7298

if REAL failed, string may be not null terminated, so strlen will crash

7315

I can't find documentation for regnsub

[krytarowski]

lib/sanitizer_common/sanitizer_common_interceptors.inc

7298

(At least the NetBSD version of) regerror(3) never fails and always returns a string where strlen(3) works.

Actually I can use its value as size to call COMMON_INTERCEPTOR_WRITE_RANGE().

7315

http://netbsd.gw.com/cgi-bin/man-cgi?regnsub++NetBSD-current

[krytarowski]

• enhance patch

[vitalybuka]

LGTM

test/sanitizer_common/TestCases/NetBSD/regex.cc

15

asserts?

[krytarowski]

test/sanitizer_common/TestCases/NetBSD/regex.cc

15

I've decided to go for abort() as it's always compiled-in.

lib/sanitizer_common/sanitizer_common_interceptors.inc

 #define INIT_NETENT \
   COMMON_INTERCEPT_FUNCTION(getnetent); \
   COMMON_INTERCEPT_FUNCTION(getnetbyname); \
   COMMON_INTERCEPT_FUNCTION(getnetbyaddr)
 #else
 #define INIT_NETENT
 #endif
 
+#if SANITIZER_INTERCEPT_REGEX
+INTERCEPTOR(int, regcomp, void *preg, const char *pattern, int cflags) {
+  void *ctx;
+  COMMON_INTERCEPTOR_ENTER(ctx, regcomp, preg, pattern, cflags);
+  if (pattern)
+    COMMON_INTERCEPTOR_READ_RANGE(ctx, pattern, REAL(strlen)(pattern) + 1);
+  int res = REAL(regcomp)(preg, pattern, cflags);
+  if (!res)
+    COMMON_INTERCEPTOR_WRITE_RANGE(ctx, preg, struct_regex_sz);
+  return res;
+}
+INTERCEPTOR(int, regexec, const void *preg, const char *string, SIZE_T nmatch,
+            struct __sanitizer_regmatch *pmatch[], int eflags) {
+  void *ctx;
+  COMMON_INTERCEPTOR_ENTER(ctx, regexec, preg, string, nmatch, pmatch, eflags);
+  if (preg)
+    COMMON_INTERCEPTOR_READ_RANGE(ctx, preg, struct_regex_sz);
+  if (string)
+    COMMON_INTERCEPTOR_READ_RANGE(ctx, string, REAL(strlen)(string) + 1);
+  int res = REAL(regexec)(preg, string, nmatch, pmatch, eflags);
+  if (!res && pmatch)
+    COMMON_INTERCEPTOR_WRITE_RANGE(ctx, pmatch, nmatch * struct_regmatch_sz);
+  return res;
+}
+INTERCEPTOR(SIZE_T, regerror, int errcode, const void *preg, char *errbuf,
+            SIZE_T errbuf_size) {
+  void *ctx;
+  COMMON_INTERCEPTOR_ENTER(ctx, regerror, errcode, preg, errbuf, errbuf_size);
+  if (preg)
+    COMMON_INTERCEPTOR_READ_RANGE(ctx, preg, struct_regex_sz);
+  SIZE_T res = REAL(regerror)(errcode, preg, errbuf, errbuf_size);
+  if (errbuf)
+    COMMON_INTERCEPTOR_WRITE_RANGE(ctx, errbuf, REAL(strlen)(errbuf) + 1);
+  return res;
+}
+INTERCEPTOR(void, regfree, const void *preg) {
+  void *ctx;
+  COMMON_INTERCEPTOR_ENTER(ctx, regfree, preg);
+  if (preg)
+    COMMON_INTERCEPTOR_READ_RANGE(ctx, preg, struct_regex_sz);
+  REAL(regfree)(preg);
+}
+INTERCEPTOR(SSIZE_T, regnsub, char *buf, SIZE_T bufsiz, const char *sub,
+            const struct __sanitizer_regmatch *rm, const char *str) {
+  void *ctx;
+  COMMON_INTERCEPTOR_ENTER(ctx, regnsub, buf, bufsiz, sub, rm, str);
+  if (sub)
+    COMMON_INTERCEPTOR_READ_RANGE(ctx, sub, REAL(strlen)(sub) + 1);
+  // The implementation demands and hardcodes 10 elements
+  if (rm)
+    COMMON_INTERCEPTOR_READ_RANGE(ctx, rm, 10 * struct_regmatch_sz);
+  if (str)
+    COMMON_INTERCEPTOR_READ_RANGE(ctx, str, REAL(strlen)(str) + 1);
+  SSIZE_T res = REAL(regnsub)(buf, bufsiz, sub, rm, str);
+  if (res > 0 && buf)
+    COMMON_INTERCEPTOR_WRITE_RANGE(ctx, buf, REAL(strlen)(buf) + 1);
+  return res;
+}
+INTERCEPTOR(SSIZE_T, regasub, char **buf, const char *sub,
+            const struct __sanitizer_regmatch *rm, const char *sstr) {
+  void *ctx;
+  COMMON_INTERCEPTOR_ENTER(ctx, regasub, buf, sub, rm, sstr);
+  if (sub)
+    COMMON_INTERCEPTOR_READ_RANGE(ctx, sub, REAL(strlen)(sub) + 1);
+  // Hardcode 10 elements as this is hardcoded size
+  if (rm)
+    COMMON_INTERCEPTOR_READ_RANGE(ctx, rm, 10 * struct_regmatch_sz);
+  if (sstr)
+    COMMON_INTERCEPTOR_READ_RANGE(ctx, sstr, REAL(strlen)(sstr) + 1);
+  SSIZE_T res = REAL(regasub)(buf, sub, rm, sstr);
+  if (res > 0 && buf) {
+    COMMON_INTERCEPTOR_WRITE_RANGE(ctx, buf, sizeof(char *));
+    COMMON_INTERCEPTOR_WRITE_RANGE(ctx, *buf, REAL(strlen)(*buf) + 1);
+  }
+  return res;
+}
+#define INIT_REGEX                                                             \
+  COMMON_INTERCEPT_FUNCTION(regcomp);                                          \
+  COMMON_INTERCEPT_FUNCTION(regexec);                                          \
+  COMMON_INTERCEPT_FUNCTION(regerror);                                         \
+  COMMON_INTERCEPT_FUNCTION(regfree);                                          \
+  COMMON_INTERCEPT_FUNCTION(regnsub);                                          \
+  COMMON_INTERCEPT_FUNCTION(regasub);
+#else
+#define INIT_REGEX
+#endif
+
 static void InitializeCommonInterceptors() {
   static u64 metadata_mem[sizeof(MetadataHashMap) / sizeof(u64) + 1];
   interceptor_metadata_map = new((void *)&metadata_mem) MetadataHashMap();
 
   INIT_MMAP;
   INIT_MMAP64;
   INIT_TEXTDOMAIN;
   INIT_STRLEN;
static void InitializeCommonInterceptors() {
   INIT_STRLCPY;
   INIT_DEVNAME;
   INIT_DEVNAME_R;
   INIT_FGETLN;
   INIT_STRMODE;
   INIT_TTYENT;
   INIT_PROTOENT;
   INIT_NETENT;
+  INIT_REGEX;
 
   INIT___PRINTF_CHK;
 }

lib/sanitizer_common/sanitizer_platform_interceptors.h

 
 #define SANITIZER_INTERCEPT_DEVNAME (SI_NETBSD || SI_OPENBSD || SI_FREEBSD)
 #define SANITIZER_INTERCEPT_DEVNAME_R (SI_NETBSD || SI_FREEBSD)
 #define SANITIZER_INTERCEPT_FGETLN (SI_NETBSD || SI_FREEBSD)
 #define SANITIZER_INTERCEPT_STRMODE (SI_NETBSD || SI_FREEBSD)
 #define SANITIZER_INTERCEPT_TTYENT SI_NETBSD
 #define SANITIZER_INTERCEPT_PROTOENT SI_NETBSD
 #define SANITIZER_INTERCEPT_NETENT SI_NETBSD
+#define SANITIZER_INTERCEPT_REGEX SI_NETBSD
 
 #endif  // #ifndef SANITIZER_PLATFORM_INTERCEPTORS_H

lib/sanitizer_common/sanitizer_platform_limits_netbsd.h

 
 extern unsigned struct_rlimit_sz;
 extern unsigned struct_utimbuf_sz;
 extern unsigned struct_timespec_sz;
 extern unsigned struct_sembuf_sz;
 
 extern unsigned struct_kevent_sz;
 
+extern unsigned struct_regex_sz;
+extern unsigned struct_regmatch_sz;
+
+struct __sanitizer_regmatch {
+  OFF_T rm_so;
+  OFF_T rm_eo;
+};
+
 union __sanitizer_sigval {
   int sival_int;
   uptr sival_ptr;
 };
 
 struct __sanitizer_sigevent {
   int sigev_notify;
   int sigev_signo;

lib/sanitizer_common/sanitizer_platform_limits_netbsd.cc

 unsigned struct_sigevent_sz = sizeof(struct sigevent);
 unsigned struct_sched_param_sz = sizeof(struct sched_param);
 unsigned struct_sockaddr_sz = sizeof(struct sockaddr);
 unsigned ucontext_t_sz = sizeof(ucontext_t);
 unsigned struct_rlimit_sz = sizeof(struct rlimit);
 unsigned struct_timespec_sz = sizeof(struct timespec);
 unsigned struct_sembuf_sz = sizeof(struct sembuf);
 unsigned struct_kevent_sz = sizeof(struct kevent);
+unsigned struct_regex_sz = sizeof(regex_t);
+unsigned struct_regmatch_sz = sizeof(regmatch_t);
 unsigned struct_utimbuf_sz = sizeof(struct utimbuf);
 unsigned struct_itimerspec_sz = sizeof(struct itimerspec);
 unsigned struct_timex_sz = sizeof(struct timex);
 unsigned struct_msqid_ds_sz = sizeof(struct msqid_ds);
 unsigned struct_mq_attr_sz = sizeof(struct mq_attr);
 unsigned struct_statvfs_sz = sizeof(struct statvfs);
 unsigned struct_sigaltstack_sz = sizeof(stack_t);
 

test/sanitizer_common/TestCases/NetBSD/regex.cc

+// RUN: %clangxx -O0 -g %s -o %t && %run %t 2>&1 | FileCheck %s
+
+#include <err.h>
+#include <regex.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+void test_matched(const regex_t *preg, const char *string) {
+  int rv = regexec(preg, string, 0, NULL, 0);
+  if (!rv)
+    printf("%s: matched\n", string);
+  else if (rv == REG_NOMATCH)
+    printf("%s: not-matched\n", string);
+  else
+    errx(1, "regexec\n");
+}
+
+void test_print_matches(const regex_t *preg, const char *string) {
+  regmatch_t rm[10];
+  int rv = regexec(preg, string, __arraycount(rm), rm, 0);
+  if (!rv) {
+    for (size_t i = 0; i < __arraycount(rm); i++) {
+      // This condition shall be simplified, but verify that the data fields
+      // are accessible.
+      if (rm[i].rm_so == -1 && rm[i].rm_eo == -1)
+        continue;
+      printf("matched[%zu]='%.*s'\n", i, (int)(rm[i].rm_eo - rm[i].rm_so),
+             string + rm[i].rm_so);
+    }
+  } else if (rv == REG_NOMATCH)
+    printf("%s: not-matched\n", string);
+  else
+    errx(1, "regexec\n");
+}
+
+void test_nsub(const regex_t *preg, const char *string) {
+  regmatch_t rm[10];
+  int rv = regexec(preg, string, __arraycount(rm), rm, 0);
+  if (!rv) {
+    char buf[1024];
+    ssize_t ss = regnsub(buf, __arraycount(buf), "\\1xyz", rm, string);
+    if (ss == -1)
+      errx(1, "regnsub");
+    printf("'%s' -> '%s'\n", string, buf);
+  } else if (rv == REG_NOMATCH)
+    printf("%s: not-matched\n", string);
+  else
+    errx(1, "regexec\n");
+}
+
+void test_asub(const regex_t *preg, const char *string) {
+  regmatch_t rm[10];
+  int rv = regexec(preg, string, __arraycount(rm), rm, 0);
+  if (!rv) {
+    char *buf;
+    ssize_t ss = regasub(&buf, "\\1xyz", rm, string);
+    if (ss == -1)
+      errx(1, "regnsub");
+    printf("'%s' -> '%s'\n", string, buf);
+    free(buf);
+  } else if (rv == REG_NOMATCH)
+    printf("%s: not-matched\n", string);
+  else
+    errx(1, "regexec\n");
+}
+
+int main(void) {
+  regex_t regex;
+  int rv;
+
+  printf("regex\n");
+
+  rv = regcomp(&regex, "[[:upper:]]\\([[:upper:]]\\)", 0);
+  if (rv)
+    errx(1, "regcomp");
+
+  test_matched(&regex, "abc");
+  test_matched(&regex, "ABC");
+
+  test_print_matches(&regex, "ABC");
+
+  test_nsub(&regex, "ABC DEF");
+  test_asub(&regex, "GHI JKL");
+
+  regfree(&regex);
+
+  rv = regcomp(&regex, "[[:upp:]]", 0);
+  if (!rv)
+    errx(1, "regcomp finished successfully");
+
+  char errbuf[1024];
+  regerror(rv, &regex, errbuf, sizeof errbuf);
+  printf("error: %s\n", errbuf);
+
+  // CHECK: regex
+  // CHECK: abc: not-matched
+  // CHECK: ABC: matched
+  // CHECK: matched[0]='AB'
+  // CHECK: matched[1]='B'
+  // CHECK: 'ABC DEF' -> 'Bxyz'
+  // CHECK: 'GHI JKL' -> 'Hxyz'
+  // CHECK: error:{{.*}}
+
+  return 0;
+}