[gcc(refs/users/marxin/heads/marxin-gcc-benchmark-branch)] tolerate padding in mbstate_t

Martin Liska marxin@gcc.gnu.org
Mon Mar 30 10:25:22 GMT 2020


https://gcc.gnu.org/g:7a900bcecab71593b731aa7dcad0fa3be8137f5a

commit 7a900bcecab71593b731aa7dcad0fa3be8137f5a
Author: Alexandre Oliva <oliva@adacore.com>
Date:   Thu Jan 23 16:36:34 2020 -0300

    tolerate padding in mbstate_t
    
    Padding in mbstate_t objects may get the memcmp to fail.
    Attempt to avoid the failure with zero initialization.
    
    
    for  libstdc++-v3/ChangeLog
    
            * testsuite/27_io/fpos/mbstate_t/1.cc: Zero-init mbstate_t.

Diff:
---
 libstdc++-v3/ChangeLog                           |  4 ++++
 libstdc++-v3/testsuite/27_io/fpos/mbstate_t/1.cc | 28 ++++++++++++++++++------
 2 files changed, 25 insertions(+), 7 deletions(-)

diff --git a/libstdc++-v3/ChangeLog b/libstdc++-v3/ChangeLog
index 539b0f6e593..ae1af018408 100644
--- a/libstdc++-v3/ChangeLog
+++ b/libstdc++-v3/ChangeLog
@@ -1,3 +1,7 @@
+2020-01-23  Alexandre Oliva <oliva@adacore.com>
+
+	* testsuite/27_io/fpos/mbstate_t/1.cc: Zero-init mbstate_t.
+
 2020-01-23  Jonathan Wakely  <jwakely@redhat.com>
 
 	PR libstdc++/91947
diff --git a/libstdc++-v3/testsuite/27_io/fpos/mbstate_t/1.cc b/libstdc++-v3/testsuite/27_io/fpos/mbstate_t/1.cc
index f92d68ffefa..28fec8e8ef2 100644
--- a/libstdc++-v3/testsuite/27_io/fpos/mbstate_t/1.cc
+++ b/libstdc++-v3/testsuite/27_io/fpos/mbstate_t/1.cc
@@ -28,24 +28,38 @@
 void test01()
 {
   typedef std::mbstate_t state_type;
-  state_type state01 = state_type();
-  state_type state02 = state_type();
+  // Use zero-initialization of the underlying memory so that padding
+  // bytes, if any, stand a better chance of comparing the same.
+  // Zero-initialized memory is guaranteed to be a valid initial
+  // state.  This doesn't quite guarantee that any padding bits won't
+  // be overwritten when copying from other instances that haven't
+  // been fully initialized: this data type is compatible with C, so
+  // it is likely plain old data, but it could have a default ctor
+  // that initializes only the relevant fields, whereas copy-ctor and
+  // operator= could be implemented as a full-object memcpy, including
+  // padding bits, rather than fieldwise copying.  However, since
+  // we're comparing two values copied from the same state_type
+  // instance, if padding bits are copied, we'll get the same for both
+  // of them, and if they aren't, we'll keep the values we initialized
+  // them with, so this should be good.
+  state_type state[2];
+  std::memset(state, 0, sizeof (state));
+
 
   std::streampos pos01(0);
-  std::streampos pos02(0);
 
   // 27.4.3.1 fpos members
   // void state(state_type s);
   // state_type state();
 
   // XXX Need to have better sanity checking for the mbstate_t type,
-  // or whatever the insantiating type for class fpos happens to be
+  // or whatever the instantiating type for class fpos happens to be
   // for streampos, as things like equality operators and assignment
   // operators, increment and deincrement operators need to be in
   // place.
-  pos01.state(state02);
-  state01 = pos01.state();
-  VERIFY( std::memcmp(&state01, &state02, sizeof(state_type)) == 0 );
+  pos01.state(state[1]);
+  state[0] = pos01.state();
+  VERIFY( std::memcmp(&state[0], &state[1], sizeof(state_type)) == 0 );
 }
 
 int main()


More information about the Libstdc++-cvs mailing list