[Bug libmudflap/33591] mudflap gives warnings exceeding bounds on valid code, when using readdir(2) on large directories

[tomash.brechko at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33591

Tomash Brechko  changed:

   What|Removed |Added

 CC||tomash.brechko at gmail dot
   ||com

--- Comment #2 from Tomash Brechko  2012-08-03 
10:29:27 UTC ---
Still present in gcc 4.6.3.

[Bug target/55673] New: [SPARC] Transposed before-after in sparc_emit_membar_for_model()

[tomash.brechko at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55673



 Bug #: 55673

   Summary: [SPARC] Transposed before-after in

sparc_emit_membar_for_model()

Classification: Unclassified

   Product: gcc

   Version: 4.8.0

Status: UNCONFIRMED

  Severity: normal

  Priority: P3

 Component: target

AssignedTo: unassig...@gcc.gnu.org

ReportedBy: tomash.brec...@gmail.com





As for other architectures (compare with need_atomic_barrier_p() for instance),

in gcc/config/sparc/sparc.c:sparc_emit_membar_for_model() memory barrier for

ACQUIRE should be emitted _after_ the operation and memory barrier for RELEASE

should be emitted _before_ the operation, i.e.:



diff --git i/gcc/config/sparc/sparc.c w/gcc/config/sparc/sparc.c

index 4e9de98..fdc6374 100644

--- i/gcc/config/sparc/sparc.c

+++ w/gcc/config/sparc/sparc.c

@@ -11188,7 +11188,7 @@ sparc_emit_membar_for_model (enum memmodel model,

   gcc_unreachable ();

 }



-  if (before_after & 1)

+  if (before_after & 2)

 {

   if (model == MEMMODEL_ACQUIRE

   || model == MEMMODEL_ACQ_REL

@@ -11200,7 +11200,7 @@ sparc_emit_membar_for_model (enum memmodel model,

mm |= StoreLoad | StoreStore;

}

 }

-  if (before_after & 2)

+  if (before_after & 1)

 {

   if (model == MEMMODEL_RELEASE

  || model == MEMMODEL_ACQ_REL

[Bug target/55690] New: On some targets thread_fence is not a compiler barrier when memmodel != MEMMODEL_SEQ_CST

[tomash.brechko at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55690



 Bug #: 55690

   Summary: On some targets thread_fence is not a compiler barrier

when memmodel != MEMMODEL_SEQ_CST

Classification: Unclassified

   Product: gcc

   Version: 4.7.3

Status: UNCONFIRMED

  Severity: normal

  Priority: P3

 Component: target

AssignedTo: unassig...@gcc.gnu.org

ReportedBy: tomash.brec...@gmail.com





At least for i386 and ia64 a call to expand_asm_memory_barrier() is needed when

memmodel != MEMMODEL_RELAXED.