[gcc r11-10241] libstdc++: Check for EOF if extraction avoids buffer overflow [PR106248]

Jonathan Wakely redi@gcc.gnu.org
Wed Sep 7 17:49:25 GMT 2022


https://gcc.gnu.org/g:3d17eaabb3e771472d7bfa1f654564a5db57f5ef

commit r11-10241-g3d17eaabb3e771472d7bfa1f654564a5db57f5ef
Author: Jonathan Wakely <jwakely@redhat.com>
Date:   Tue Jul 12 11:18:47 2022 +0100

    libstdc++: Check for EOF if extraction avoids buffer overflow [PR106248]
    
    In r11-2581-g17abcc77341584 (for LWG 2499) I added overflow checks to
    the pre-C++20 operator>>(istream&, char*) overload.  Those checks can
    cause extraction to stop after filling the buffer, where previously it
    would have tried to extract another character and stopped at EOF. When
    that happens we no longer set eofbit in the stream state, which is
    consistent with the behaviour of the new C++20 overload, but is an
    observable and unexpected change in the C++17 behaviour. What makes it
    worse is that the behaviour change is dependent on optimization, because
    __builtin_object_size is used to detect the buffer size and that only
    works when optimizing.
    
    To avoid the unexpected and optimization-dependent change in behaviour,
    set eofbit manually if we stopped extracting because of the buffer size
    check, but had reached EOF anyway. If the stream's rdstate() != goodbit
    or width() is non-zero and smaller than the buffer, there's nothing to
    do. Otherwise, we filled the buffer and need to check for EOF, and maybe
    set eofbit.
    
    The new check is guarded by #ifdef __OPTIMIZE__ because otherwise
    __builtin_object_size is useless. There's no point compiling and
    emitting dead code that can't be eliminated because we're not
    optimizing.
    
    We could add extra checks that the next character in the buffer is not
    whitespace, to detect the case where we stopped early and prevented a
    buffer overflow that would have happened otherwise. That would allow us
    to assert or set badbit in the stream state when undefined behaviour was
    prevented. However, those extra checks would increase the size of the
    function, potentially reducing the likelihood of it being inlined, and
    so making the buffer size detection less reliable. It seems preferable
    to prevent UB and silently truncate, rather than miss the UB and allow
    the overflow to happen.
    
    libstdc++-v3/ChangeLog:
    
            PR libstdc++/106248
            * include/std/istream [C++17] (operator>>(istream&, char*)):
            Set eofbit if we stopped extracting at EOF.
            * testsuite/27_io/basic_istream/extractors_character/char/pr106248.cc:
            New test.
            * testsuite/27_io/basic_istream/extractors_character/wchar_t/pr106248.cc:
            New test.
    
    (cherry picked from commit 5ae74944af1de032d4a27fad4a2287bd3a2163fd)

Diff:
---
 libstdc++-v3/include/std/istream                   | 31 ++++++++++++++---
 .../extractors_character/char/pr106248.cc          | 40 ++++++++++++++++++++++
 .../extractors_character/wchar_t/pr106248.cc       | 40 ++++++++++++++++++++++
 3 files changed, 106 insertions(+), 5 deletions(-)

diff --git a/libstdc++-v3/include/std/istream b/libstdc++-v3/include/std/istream
index 5ad60dbd709..109564b3c0d 100644
--- a/libstdc++-v3/include/std/istream
+++ b/libstdc++-v3/include/std/istream
@@ -784,7 +784,7 @@ _GLIBCXX_BEGIN_NAMESPACE_VERSION
    *  - if `width()` is greater than zero, `n` is `min(width(), n)`
    *  - otherwise `n` is the number of elements of the array
    *  - (before C++20 the pointer is assumed to point to an array of
-   *  - the largest possible size for an array of `char_type`).
+   *    the largest possible size for an array of `char_type`).
    *
    *  Characters are extracted and stored until one of the following happens:
    *  - `n - 1` characters are stored
@@ -802,19 +802,40 @@ _GLIBCXX_BEGIN_NAMESPACE_VERSION
     inline basic_istream<_CharT, _Traits>&
     operator>>(basic_istream<_CharT, _Traits>& __in, _CharT* __s)
     {
+#ifdef __OPTIMIZE__
+      // Function inlining might make the buffer size known, allowing us to
+      // prevent overflow.
       size_t __n = __builtin_object_size(__s, 0);
-      if (__builtin_expect(__n < sizeof(_CharT), false))
+      if (__n < sizeof(_CharT))
 	{
 	  // There is not even space for the required null terminator.
 	  __glibcxx_assert(__n >= sizeof(_CharT));
+	  // No point calling __istream_extract, but still need to reset width.
 	  __in.width(0);
 	  __in.setstate(ios_base::failbit);
 	}
+      else if (__n != (size_t)-1)
+	{
+	  __n /= sizeof(_CharT);
+	  streamsize __w = __in.width();
+	  std::__istream_extract(__in, __s, __n);
+	  if (__in.good() && (__w <= 0 || __n < __w))
+	    {
+	      // Stopped extracting early to avoid overflowing the buffer,
+	      // but might have stopped anyway (and set eofbit) if at EOF.
+	      const typename _Traits::int_type __c = __in.rdbuf()->sgetc();
+	      const bool __eof = _Traits::eq_int_type(__c, _Traits::eof());
+	      if (__builtin_expect(__eof, true)) // Assume EOF, not overflow.
+		__in.setstate(ios_base::eofbit);
+	    }
+	}
       else
+#endif // __OPTIMIZE
 	{
-	  if (__n == (size_t)-1)
-	    __n = __gnu_cxx::__numeric_traits<streamsize>::__max;
-	  std::__istream_extract(__in, __s, __n / sizeof(_CharT));
+	  // Buffer size is unknown, have to assume it's huge.
+	  streamsize __n = __gnu_cxx::__numeric_traits<streamsize>::__max;
+	  __n /= sizeof(_CharT);
+	  std::__istream_extract(__in, __s, __n);
 	}
       return __in;
     }
diff --git a/libstdc++-v3/testsuite/27_io/basic_istream/extractors_character/char/pr106248.cc b/libstdc++-v3/testsuite/27_io/basic_istream/extractors_character/char/pr106248.cc
new file mode 100644
index 00000000000..6d89a0e5fef
--- /dev/null
+++ b/libstdc++-v3/testsuite/27_io/basic_istream/extractors_character/char/pr106248.cc
@@ -0,0 +1,40 @@
+// { dg-do run }
+
+#include <sstream>
+#include <testsuite_hooks.h>
+
+void
+test_pr106248()
+{
+  char buf[5] = {'x', 'x', 'x', 'x', 'x'};
+  std::string s("  four");
+  std::istringstream in(s);
+  in >> buf;
+#if __cplusplus >= 202002L
+  // Extraction stops because buffer is full.
+  VERIFY( in.good() );
+#else
+  // PR libstdc++/106248
+  // Extraction stops because all input has been consumed and eofbit is set.
+  VERIFY( in.eof() );
+#endif
+  // Extracted string must be null-terminated.
+  VERIFY( buf[4] == '\0' );
+  VERIFY( std::string(buf) == "four" );
+
+  in.clear();
+  in.str(s);
+  for (int i = 0; i < 5; ++i)
+    s[i] = 'x';
+
+  in.width(5);
+  in >> buf;
+  // Extraction stops due to field width, eofbit not set.
+  VERIFY( in.good() );
+  VERIFY( std::string(buf) == "four" );
+}
+
+int main()
+{
+  test_pr106248();
+}
diff --git a/libstdc++-v3/testsuite/27_io/basic_istream/extractors_character/wchar_t/pr106248.cc b/libstdc++-v3/testsuite/27_io/basic_istream/extractors_character/wchar_t/pr106248.cc
new file mode 100644
index 00000000000..7c226600b9e
--- /dev/null
+++ b/libstdc++-v3/testsuite/27_io/basic_istream/extractors_character/wchar_t/pr106248.cc
@@ -0,0 +1,40 @@
+// { dg-do run }
+
+#include <sstream>
+#include <testsuite_hooks.h>
+
+void
+test_pr106248()
+{
+  wchar_t buf[5] = {L'x', L'x', L'x', L'x', L'x'};
+  std::wstring s(L"  four");
+  std::wistringstream in(s);
+  in >> buf;
+#if __cplusplus >= 202002L
+  // Extraction stops because buffer is full.
+  VERIFY( in.good() );
+#else
+  // PR libstdc++/106248
+  // Extraction stops because all input has been consumed and eofbit is set.
+  VERIFY( in.eof() );
+#endif
+  // Extracted string must be null-terminated.
+  VERIFY( buf[4] == L'\0' );
+  VERIFY( std::wstring(buf) == L"four" );
+
+  in.clear();
+  in.str(s);
+  for (int i = 0; i < 5; ++i)
+    s[i] = L'x';
+
+  in.width(5);
+  in >> buf;
+  // Extraction stops due to field width, eofbit not set.
+  VERIFY( in.good() );
+  VERIFY( std::wstring(buf) == L"four" );
+}
+
+int main()
+{
+  test_pr106248();
+}


More information about the Libstdc++-cvs mailing list