malloc: Abort on heap corruption, without a backtrace [BZ #21754]

The stack trace printing caused deadlocks and has been itself been
targeted by code execution exploits.

ChangeLog

@@ -1,3 +1,12 @@
+2017-08-30  Florian Weimer  <fweimer@redhat.com>
+
+	[BZ #21754]
+	* malloc/malloc.c (malloc_printerr): Always terminate the process,
+	without printing a backtrace.  Do not leak any information in the
+	error message.
+	* manual/memory.texi (Heap Consistency Checking): Update.
+	* manual/tunables.texi (Memory Allocation Tunables): Likewise.
+
 2017-08-30  Florian Weimer  <fweimer@redhat.com>
 
 	Do not scale NPTL tests with available number of CPUs.

NEWS

@@ -12,6 +12,14 @@ Major new features:
 * Optimized x86-64 asin, atan2, exp, expf, log, pow, atan, sin and tan
   with FMA, contributed by Arjan van de Ven and H.J. Lu from Intel.
 
+* In order to support faster and safer process termination the malloc API
+  family of functions will no longer print a failure address and stack
+  backtrace after detecting heap corruption.  The goal is to minimize the
+  amount of work done after corruption is detected and to avoid potential
+  security issues in continued process execution.  Reducing shutdown time
+  leads to lower overall process restart latency, so there is benefit both
+  from a security and performance perspective.
+
 Deprecated and removed features, and other changes affecting compatibility:
 
 * On GNU/Linux, the obsolete Linux constant PTRACE_SEIZE_DEVEL is no longer

malloc/malloc.c

@@ -1019,7 +1019,8 @@ static void*  _int_realloc(mstate, mchunkptr, INTERNAL_SIZE_T,
 static void*  _int_memalign(mstate, size_t, size_t);
 static void*  _mid_memalign(size_t, size_t, void *);
 
-static void malloc_printerr(int action, const char *str, void *ptr, mstate av);
+static void malloc_printerr(int action, const char *str, void *ptr, mstate av)
+  __attribute__ ((noreturn));
 
 static void* internal_function mem2mem_check(void *p, size_t sz);
 static int internal_function top_check(void);
@@ -5408,24 +5409,8 @@ malloc_printerr (int action, const char *str, void *ptr, mstate ar_ptr)
   if (ar_ptr)
     set_arena_corrupt (ar_ptr);
 
-  if ((action & 5) == 5)
-    __libc_message ((action & 2) ? (do_abort | do_backtrace) : do_message,
-		    "%s\n", str);
-  else if (action & 1)
-    {
-      char buf[2 * sizeof (uintptr_t) + 1];
-
-      buf[sizeof (buf) - 1] = '\0';
-      char *cp = _itoa_word ((uintptr_t) ptr, &buf[sizeof (buf) - 1], 16, 0);
-      while (cp > buf)
-        *--cp = '0';
-
-      __libc_message ((action & 2) ? (do_abort | do_backtrace) : do_message,
-		      "*** Error in `%s': %s: 0x%s ***\n",
-                      __libc_argv[0] ? : "<unknown>", str, cp);
-    }
-  else if (action & 2)
-    abort ();
+  __libc_message (do_abort, "%s\n", str);
+  __builtin_unreachable ();
 }
 
 /* We need a wrapper function for one of the additions of POSIX.  */

manual/memory.texi

@@ -1309,17 +1309,15 @@ The block was already freed.
 
 Another possibility to check for and guard against bugs in the use of
 @code{malloc}, @code{realloc} and @code{free} is to set the environment
-variable @code{MALLOC_CHECK_}.  When @code{MALLOC_CHECK_} is set, a
-special (less efficient) implementation is used which is designed to be
-tolerant against simple errors, such as double calls of @code{free} with
-the same argument, or overruns of a single byte (off-by-one bugs).  Not
-all such errors can be protected against, however, and memory leaks can
-result.  If @code{MALLOC_CHECK_} is set to @code{0}, any detected heap
-corruption is silently ignored; if set to @code{1}, a diagnostic is
-printed on @code{stderr}; if set to @code{2}, @code{abort} is called
-immediately.  This can be useful because otherwise a crash may happen
-much later, and the true cause for the problem is then very hard to
-track down.
+variable @code{MALLOC_CHECK_}.  When @code{MALLOC_CHECK_} is set to a
+non-zero value, a special (less efficient) implementation is used which
+is designed to be tolerant against simple errors, such as double calls
+of @code{free} with the same argument, or overruns of a single byte
+(off-by-one bugs).  Not all such errors can be protected against,
+however, and memory leaks can result.
+
+Any detected heap corruption results in immediate termination of the
+process.
 
 There is one problem with @code{MALLOC_CHECK_}: in SUID or SGID binaries
 it could possibly be exploited since diverging from the normal programs

manual/tunables.texi

@@ -71,27 +71,13 @@ following tunables in the @code{malloc} namespace:
 This tunable supersedes the @env{MALLOC_CHECK_} environment variable and is
 identical in features.
 
-Setting this tunable enables a special (less efficient) memory allocator for
-the malloc family of functions that is designed to be tolerant against simple
-errors such as double calls of free with the same argument, or overruns of a
-single byte (off-by-one bugs). Not all such errors can be protected against,
-however, and memory leaks can result.  The following list describes the values
-that this tunable can take and the effect they have on malloc functionality:
-
-@itemize @bullet
-@item @code{0} Ignore all errors.  The default allocator continues to be in
-use, but all errors are silently ignored.
-@item @code{1} Report errors.  The alternate allocator is selected and heap
-corruption, if detected, is reported as diagnostic messages to @code{stderr}
-and the program continues execution.
-@item @code{2} Abort on errors.  The alternate allocator is selected and if
-heap corruption is detected, the program is ended immediately by calling
-@code{abort}.
-@item @code{3} Fully enabled.  The alternate allocator is selected and is fully
-functional.  That is, if heap corruption is detected, a verbose diagnostic
-message is printed to @code{stderr} and the program is ended by calling
-@code{abort}.
-@end itemize
+Setting this tunable to a non-zero value enables a special (less
+efficient) memory allocator for the malloc family of functions that is
+designed to be tolerant against simple errors such as double calls of
+free with the same argument, or overruns of a single byte (off-by-one
+bugs). Not all such errors can be protected against, however, and memory
+leaks can result.  Any detected heap corruption results in immediate
+termination of the process.
 
 Like @env{MALLOC_CHECK_}, @code{glibc.malloc.check} has a problem in that it
 diverges from normal program behavior by writing to @code{stderr}, which could