[dfsan] Wrap glibc 2.38 __isoc23_* functions #79958

MaskRay · 2024-01-30T07:52:17Z

Fix #79283: test/dfsan/custom.cpp has undefined symbol linker errors
on glibc 2.38 due to lack of wrappers for __isoc23_strtol and
__isoc23_scanf family functions.

Implement these wrappers as aliases to existing wrappers, similar to
https://reviews.llvm.org/D158943 for other sanitizers.

strtol in a user program, whether or not _ISOC2X_SOURCE is defined,
uses the C23 semantics (strtol("0b1", 0, 0) => 1), when
libclang_rt.dfsan.a is built on glibc 2.38+.

Created using spr 1.3.4

llvmbot · 2024-01-30T07:52:51Z

@llvm/pr-subscribers-compiler-rt-sanitizer

Author: Fangrui Song (MaskRay)

Changes

Fix #79283: test/dfsan/custom.cpp has undefined symbol linker errors
on glibc 2.38 due to lack of wrappers for __isoc23_strtol and
__isoc23_scanf family functions.

Implement these wrappers as aliases to existing wrappers, similar to
https://reviews.llvm.org/D158943 for other sanitizers.

strtol in a user program, whether or not _ISOC2X_SOURCE is defined,
uses the C23 semantics (strtol("0b1", 0, 0) => 1), when
libclang_rt.dfsan.a is built on glibc 2.38+.

Full diff: https://github.com/llvm/llvm-project/pull/79958.diff

3 Files Affected:

(modified) compiler-rt/lib/dfsan/dfsan_custom.cpp (+47-154)
(modified) compiler-rt/lib/dfsan/done_abilist.txt (+6)
(modified) compiler-rt/lib/dfsan/libc_ubuntu1404_abilist.txt (+5)

diff --git a/compiler-rt/lib/dfsan/dfsan_custom.cpp b/compiler-rt/lib/dfsan/dfsan_custom.cpp
index 85b796bd6349c..db55958922a6e 100644
--- a/compiler-rt/lib/dfsan/dfsan_custom.cpp
+++ b/compiler-rt/lib/dfsan/dfsan_custom.cpp
@@ -55,6 +55,10 @@ using namespace __dfsan;
 #define DECLARE_WEAK_INTERCEPTOR_HOOK(f, ...) \
 SANITIZER_INTERFACE_ATTRIBUTE SANITIZER_WEAK_ATTRIBUTE void f(__VA_ARGS__);
 
+#define WRAPPER_ALIAS(fun, real)                                          \
+  SANITIZER_INTERFACE_ATTRIBUTE void __dfsw_##fun() ALIAS(__dfsw_##real); \
+  SANITIZER_INTERFACE_ATTRIBUTE void __dfso_##fun() ALIAS(__dfso_##real);
+
 // Async-safe, non-reentrant spin lock.
 class SignalSpinLocker {
  public:
@@ -1197,16 +1201,20 @@ char *__dfso_strcpy(char *dest, const char *src, dfsan_label dst_label,
   *ret_origin = dst_origin;
   return ret;
 }
+}
 
-static long int dfsan_strtol(const char *nptr, char **endptr, int base,
-                             char **tmp_endptr) {
+template <typename Fn>
+static ALWAYS_INLINE auto dfsan_strtol_impl(
+    Fn real, const char *nptr, char **endptr, int base,
+    char **tmp_endptr) -> decltype(real(nullptr, nullptr, 0)) {
   assert(tmp_endptr);
-  long int ret = strtol(nptr, tmp_endptr, base);
+  auto ret = real(nptr, tmp_endptr, base);
   if (endptr)
     *endptr = *tmp_endptr;
   return ret;
 }
 
+extern "C" {
 static void dfsan_strtolong_label(const char *nptr, const char *tmp_endptr,
                                   dfsan_label base_label,
                                   dfsan_label *ret_label) {
@@ -1236,30 +1244,6 @@ static void dfsan_strtolong_origin(const char *nptr, const char *tmp_endptr,
   }
 }
 
-SANITIZER_INTERFACE_ATTRIBUTE
-long int __dfsw_strtol(const char *nptr, char **endptr, int base,
-                       dfsan_label nptr_label, dfsan_label endptr_label,
-                       dfsan_label base_label, dfsan_label *ret_label) {
-  char *tmp_endptr;
-  long int ret = dfsan_strtol(nptr, endptr, base, &tmp_endptr);
-  dfsan_strtolong_label(nptr, tmp_endptr, base_label, ret_label);
-  return ret;
-}
-
-SANITIZER_INTERFACE_ATTRIBUTE
-long int __dfso_strtol(const char *nptr, char **endptr, int base,
-                       dfsan_label nptr_label, dfsan_label endptr_label,
-                       dfsan_label base_label, dfsan_label *ret_label,
-                       dfsan_origin nptr_origin, dfsan_origin endptr_origin,
-                       dfsan_origin base_origin, dfsan_origin *ret_origin) {
-  char *tmp_endptr;
-  long int ret = dfsan_strtol(nptr, endptr, base, &tmp_endptr);
-  dfsan_strtolong_label(nptr, tmp_endptr, base_label, ret_label);
-  dfsan_strtolong_origin(nptr, tmp_endptr, base_label, ret_label, base_origin,
-                         ret_origin);
-  return ret;
-}
-
 static double dfsan_strtod(const char *nptr, char **endptr, char **tmp_endptr) {
   assert(tmp_endptr);
   double ret = strtod(nptr, tmp_endptr);
@@ -1307,108 +1291,40 @@ double __dfso_strtod(const char *nptr, char **endptr, dfsan_label nptr_label,
   return ret;
 }
 
-static long long int dfsan_strtoll(const char *nptr, char **endptr, int base,
-                                   char **tmp_endptr) {
-  assert(tmp_endptr);
-  long long int ret = strtoll(nptr, tmp_endptr, base);
-  if (endptr)
-    *endptr = *tmp_endptr;
-  return ret;
-}
-
-SANITIZER_INTERFACE_ATTRIBUTE
-long long int __dfsw_strtoll(const char *nptr, char **endptr, int base,
-                             dfsan_label nptr_label, dfsan_label endptr_label,
-                             dfsan_label base_label, dfsan_label *ret_label) {
-  char *tmp_endptr;
-  long long int ret = dfsan_strtoll(nptr, endptr, base, &tmp_endptr);
-  dfsan_strtolong_label(nptr, tmp_endptr, base_label, ret_label);
-  return ret;
-}
-
-SANITIZER_INTERFACE_ATTRIBUTE
-long long int __dfso_strtoll(const char *nptr, char **endptr, int base,
-                             dfsan_label nptr_label, dfsan_label endptr_label,
-                             dfsan_label base_label, dfsan_label *ret_label,
-                             dfsan_origin nptr_origin,
-                             dfsan_origin endptr_origin,
-                             dfsan_origin base_origin,
-                             dfsan_origin *ret_origin) {
-  char *tmp_endptr;
-  long long int ret = dfsan_strtoll(nptr, endptr, base, &tmp_endptr);
-  dfsan_strtolong_label(nptr, tmp_endptr, base_label, ret_label);
-  dfsan_strtolong_origin(nptr, tmp_endptr, base_label, ret_label, base_origin,
-                         ret_origin);
-  return ret;
-}
-
-static unsigned long int dfsan_strtoul(const char *nptr, char **endptr,
-                                       int base, char **tmp_endptr) {
-  assert(tmp_endptr);
-  unsigned long int ret = strtoul(nptr, tmp_endptr, base);
-  if (endptr)
-    *endptr = *tmp_endptr;
-  return ret;
-}
-
-SANITIZER_INTERFACE_ATTRIBUTE
-unsigned long int __dfsw_strtoul(const char *nptr, char **endptr, int base,
-                       dfsan_label nptr_label, dfsan_label endptr_label,
-                       dfsan_label base_label, dfsan_label *ret_label) {
-  char *tmp_endptr;
-  unsigned long int ret = dfsan_strtoul(nptr, endptr, base, &tmp_endptr);
-  dfsan_strtolong_label(nptr, tmp_endptr, base_label, ret_label);
-  return ret;
-}
-
-SANITIZER_INTERFACE_ATTRIBUTE
-unsigned long int __dfso_strtoul(
-    const char *nptr, char **endptr, int base, dfsan_label nptr_label,
-    dfsan_label endptr_label, dfsan_label base_label, dfsan_label *ret_label,
-    dfsan_origin nptr_origin, dfsan_origin endptr_origin,
-    dfsan_origin base_origin, dfsan_origin *ret_origin) {
-  char *tmp_endptr;
-  unsigned long int ret = dfsan_strtoul(nptr, endptr, base, &tmp_endptr);
-  dfsan_strtolong_label(nptr, tmp_endptr, base_label, ret_label);
-  dfsan_strtolong_origin(nptr, tmp_endptr, base_label, ret_label, base_origin,
-                         ret_origin);
-  return ret;
-}
-
-static long long unsigned int dfsan_strtoull(const char *nptr, char **endptr,
-                                             int base, char **tmp_endptr) {
-  assert(tmp_endptr);
-  long long unsigned int ret = strtoull(nptr, tmp_endptr, base);
-  if (endptr)
-    *endptr = *tmp_endptr;
-  return ret;
-}
-
-SANITIZER_INTERFACE_ATTRIBUTE
-long long unsigned int __dfsw_strtoull(const char *nptr, char **endptr,
-                                       int base, dfsan_label nptr_label,
-                                       dfsan_label endptr_label,
-                                       dfsan_label base_label,
-                                       dfsan_label *ret_label) {
-  char *tmp_endptr;
-  long long unsigned int ret = dfsan_strtoull(nptr, endptr, base, &tmp_endptr);
-  dfsan_strtolong_label(nptr, tmp_endptr, base_label, ret_label);
-  return ret;
-}
-
-SANITIZER_INTERFACE_ATTRIBUTE
-long long unsigned int __dfso_strtoull(
-    const char *nptr, char **endptr, int base, dfsan_label nptr_label,
-    dfsan_label endptr_label, dfsan_label base_label, dfsan_label *ret_label,
-    dfsan_origin nptr_origin, dfsan_origin endptr_origin,
-    dfsan_origin base_origin, dfsan_origin *ret_origin) {
-  char *tmp_endptr;
-  long long unsigned int ret = dfsan_strtoull(nptr, endptr, base, &tmp_endptr);
-  dfsan_strtolong_label(nptr, tmp_endptr, base_label, ret_label);
-  dfsan_strtolong_origin(nptr, tmp_endptr, base_label, ret_label, base_origin,
-                         ret_origin);
-  return ret;
-}
+WRAPPER_ALIAS(__isoc23_strtod, strtod)
+
+#define WRAPPER_STRTO(ret_type, fun)                                     \
+  SANITIZER_INTERFACE_ATTRIBUTE ret_type __dfsw_##fun(                   \
+      const char *nptr, char **endptr, int base, dfsan_label nptr_label, \
+      dfsan_label endptr_label, dfsan_label base_label,                  \
+      dfsan_label *ret_label) {                                          \
+    char *tmp_endptr;                                                    \
+    auto ret = dfsan_strtol_impl(fun, nptr, endptr, base, &tmp_endptr);  \
+    dfsan_strtolong_label(nptr, tmp_endptr, base_label, ret_label);      \
+    return ret;                                                          \
+  }                                                                      \
+  SANITIZER_INTERFACE_ATTRIBUTE ret_type __dfso_##fun(                   \
+      const char *nptr, char **endptr, int base, dfsan_label nptr_label, \
+      dfsan_label endptr_label, dfsan_label base_label,                  \
+      dfsan_label *ret_label, dfsan_origin nptr_origin,                  \
+      dfsan_origin endptr_origin, dfsan_origin base_origin,              \
+      dfsan_origin *ret_origin) {                                        \
+    char *tmp_endptr;                                                    \
+    auto ret = dfsan_strtol_impl(fun, nptr, endptr, base, &tmp_endptr);  \
+    dfsan_strtolong_label(nptr, tmp_endptr, base_label, ret_label);      \
+    dfsan_strtolong_origin(nptr, tmp_endptr, base_label, ret_label,      \
+                           base_origin, ret_origin);                     \
+    return ret;                                                          \
+  }
+
+WRAPPER_STRTO(long, strtol)
+WRAPPER_STRTO(long long, strtoll)
+WRAPPER_STRTO(unsigned long, strtoul)
+WRAPPER_STRTO(unsigned long long, strtoull)
+WRAPPER_ALIAS(__isoc23_strtol, strtol)
+WRAPPER_ALIAS(__isoc23_strtoll, strtoll)
+WRAPPER_ALIAS(__isoc23_strtoul, strtoul)
+WRAPPER_ALIAS(__isoc23_strtoull, strtoull)
 
 SANITIZER_INTERFACE_ATTRIBUTE
 time_t __dfsw_time(time_t *t, dfsan_label t_label, dfsan_label *ret_label) {
@@ -2866,31 +2782,8 @@ int __dfso_sscanf(char *str, const char *format, dfsan_label str_label,
   return ret;
 }
 
-SANITIZER_INTERFACE_ATTRIBUTE
-int __dfsw___isoc99_sscanf(char *str, const char *format, dfsan_label str_label,
-                           dfsan_label format_label, dfsan_label *va_labels,
-                           dfsan_label *ret_label, ...) {
-  va_list ap;
-  va_start(ap, ret_label);
-  int ret = scan_buffer(str, ~0ul, format, va_labels, ret_label, nullptr,
-                        nullptr, ap);
-  va_end(ap);
-  return ret;
-}
-
-SANITIZER_INTERFACE_ATTRIBUTE
-int __dfso___isoc99_sscanf(char *str, const char *format, dfsan_label str_label,
-                           dfsan_label format_label, dfsan_label *va_labels,
-                           dfsan_label *ret_label, dfsan_origin str_origin,
-                           dfsan_origin format_origin, dfsan_origin *va_origins,
-                           dfsan_origin *ret_origin, ...) {
-  va_list ap;
-  va_start(ap, ret_origin);
-  int ret = scan_buffer(str, ~0ul, format, va_labels, ret_label, &str_origin,
-                        ret_origin, ap);
-  va_end(ap);
-  return ret;
-}
+WRAPPER_ALIAS(__isoc99_sscanf, sscanf)
+WRAPPER_ALIAS(__isoc23_sscanf, sscanf)
 
 static void BeforeFork() {
   StackDepotLockBeforeFork();
diff --git a/compiler-rt/lib/dfsan/done_abilist.txt b/compiler-rt/lib/dfsan/done_abilist.txt
index c582584d77e45..86a42ee1b4dce 100644
--- a/compiler-rt/lib/dfsan/done_abilist.txt
+++ b/compiler-rt/lib/dfsan/done_abilist.txt
@@ -270,6 +270,11 @@ fun:strtoul=custom
 fun:strtoull=custom
 fun:strcat=custom
 fun:strncat=custom
+fun:__isoc23_strtod=custom
+fun:__isoc23_strtol=custom
+fun:__isoc23_strtoll=custom
+fun:__isoc23_strtoul=custom
+fun:__isoc23_strtoull=custom
 
 # Functions that produce an output that is computed from the input, but is not
 # necessarily data dependent.
@@ -311,6 +316,7 @@ fun:snprintf=custom
 # scanf-like
 fun:sscanf=custom
 fun:__isoc99_sscanf=custom
+fun:__isoc23_sscanf=custom
 
 # TODO: custom
 fun:asprintf=discard
diff --git a/compiler-rt/lib/dfsan/libc_ubuntu1404_abilist.txt b/compiler-rt/lib/dfsan/libc_ubuntu1404_abilist.txt
index 433092e2b27b8..9ffa56a238185 100644
--- a/compiler-rt/lib/dfsan/libc_ubuntu1404_abilist.txt
+++ b/compiler-rt/lib/dfsan/libc_ubuntu1404_abilist.txt
@@ -1,3 +1,8 @@
+fun:__isoc23_sscanf=uninstrumented
+fun:__isoc23_strtol=uninstrumented
+fun:__isoc23_strtoll=uninstrumented
+fun:__isoc23_strtoul=uninstrumented
+fun:__isoc23_strtoull=uninstrumented
 fun:_Exit=uninstrumented
 fun:_IO_adjust_column=uninstrumented
 fun:_IO_adjust_wcolumn=uninstrumented

github-actions · 2024-01-30T07:54:31Z

⚠️ C/C++ code formatter, clang-format found issues in your code. ⚠️

You can test this locally with the following command:

git-clang-format --diff a12390e6202eaf1e5d7810109756e10079b9144b bbb7ed85fe6ac8879417e9484ab95f3ccaf96a3f -- compiler-rt/lib/dfsan/dfsan_custom.cpp

View the diff from clang-format here.

diff --git a/compiler-rt/lib/dfsan/dfsan_custom.cpp b/compiler-rt/lib/dfsan/dfsan_custom.cpp
index 3af26e9f64..14173b2b96 100644
--- a/compiler-rt/lib/dfsan/dfsan_custom.cpp
+++ b/compiler-rt/lib/dfsan/dfsan_custom.cpp
@@ -1204,9 +1204,10 @@ char *__dfso_strcpy(char *dest, const char *src, dfsan_label dst_label,
 }
 
 template <typename Fn>
-static ALWAYS_INLINE auto dfsan_strtol_impl(
-    Fn real, const char *nptr, char **endptr, int base,
-    char **tmp_endptr) -> decltype(real(nullptr, nullptr, 0)) {
+static ALWAYS_INLINE auto dfsan_strtol_impl(Fn real, const char *nptr,
+                                            char **endptr, int base,
+                                            char **tmp_endptr)
+    -> decltype(real(nullptr, nullptr, 0)) {
   assert(tmp_endptr);
   auto ret = real(nptr, tmp_endptr, base);
   if (endptr)

Created using spr 1.3.4

mgorny

Thanks a lot! I can confirm that all tests pass with this patch, for me. I also love the simplification!

Fix llvm#79283: `test/dfsan/custom.cpp` has undefined symbol linker errors on glibc 2.38 due to lack of wrappers for `__isoc23_strtol` and `__isoc23_scanf` family functions. Implement these wrappers as aliases to existing wrappers, similar to https://reviews.llvm.org/D158943 for other sanitizers. `strtol` in a user program, whether or not `_ISOC2X_SOURCE` is defined, uses the C23 semantics (`strtol("0b1", 0, 0)` => 1), when `libclang_rt.dfsan.a` is built on glibc 2.38+. (cherry picked from commit 6485600)

[𝘀𝗽𝗿] initial version

5659e73

Created using spr 1.3.4

MaskRay requested a review from stephan-yichao-zhao January 30, 2024 07:52

llvmbot added compiler-rt compiler-rt:sanitizer labels Jan 30, 2024

MaskRay requested review from browneee and mgorny January 30, 2024 07:52

MaskRay requested a review from vitalybuka January 30, 2024 07:52

clang-format

bbb7ed8

Created using spr 1.3.4

mgorny approved these changes Jan 30, 2024

View reviewed changes

vitalybuka approved these changes Jan 30, 2024

View reviewed changes

MaskRay changed the title ~~[dfsan] Support glibc 2.38 __isoc23_* functions~~ [dfsan] Wrap glibc 2.38 __isoc23_* functions Jan 30, 2024

MaskRay merged commit 6485600 into main Jan 30, 2024
3 of 4 checks passed

MaskRay deleted the users/MaskRay/spr/dfsan-support-glibc-238-__isoc23_-functions branch January 30, 2024 21:58

pointhex mentioned this pull request May 7, 2024

getStyleDiagHandler #91314

Closed

aemerson mentioned this pull request May 9, 2024

release/18.x: [AArc64][GlobalISel] Fix legalizer assert for G_INSERT_VECTOR_ELT - manual merge #91672

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[dfsan] Wrap glibc 2.38 __isoc23_* functions #79958

[dfsan] Wrap glibc 2.38 __isoc23_* functions #79958

MaskRay commented Jan 30, 2024

llvmbot commented Jan 30, 2024

github-actions bot commented Jan 30, 2024 •

edited

Loading

mgorny left a comment

[dfsan] Wrap glibc 2.38 __isoc23_* functions #79958

[dfsan] Wrap glibc 2.38 __isoc23_* functions #79958

Conversation

MaskRay commented Jan 30, 2024

llvmbot commented Jan 30, 2024

github-actions bot commented Jan 30, 2024 • edited Loading

mgorny left a comment

Choose a reason for hiding this comment

github-actions bot commented Jan 30, 2024 •

edited

Loading