As mentioned in the PR, this seems to no longer be an issue after we
redesigned the TM abnormal edges (in the uninstrumented code path patchset).
Tested with the STAMP benchmark for both 1 and 4 threads, as well as
with the GCC regression suite.
OK for trunk?
commit 462e1c25a81bfb575a9113e4d117c5972c827715
Author: Aldy Hernandez <al...@redhat.com>
Date: Mon Dec 3 13:03:16 2012 -0600
PR rtl-optimization/51771
Revert:
* builtin-attrs.def (ATTR_RETURNS_TWICE, ATTR_TM_NOTHROW_RT_LIST): New.
* gtm-builtins.def (BUILT_IN_TM_START): Add returns-twice attrib.
diff --git a/gcc/builtin-attrs.def b/gcc/builtin-attrs.def
index 545b6fd..d149e26 100644
--- a/gcc/builtin-attrs.def
+++ b/gcc/builtin-attrs.def
@@ -107,7 +107,6 @@ DEF_ATTR_IDENT (ATTR_STRFTIME, "strftime")
DEF_ATTR_IDENT (ATTR_TYPEGENERIC, "type generic")
DEF_ATTR_IDENT (ATTR_TM_REGPARM, "*tm regparm")
DEF_ATTR_IDENT (ATTR_TM_TMPURE, "transaction_pure")
-DEF_ATTR_IDENT (ATTR_RETURNS_TWICE, "returns_twice")
DEF_ATTR_TREE_LIST (ATTR_NOVOPS_LIST, ATTR_NOVOPS, ATTR_NULL, ATTR_NULL)
@@ -253,8 +252,6 @@ DEF_ATTR_TREE_LIST (ATTR_TM_NORETURN_NOTHROW_LIST,
ATTR_TM_REGPARM, ATTR_NULL, ATTR_NORETURN_NOTHROW_LIST)
DEF_ATTR_TREE_LIST (ATTR_TM_CONST_NOTHROW_LIST,
ATTR_TM_REGPARM, ATTR_NULL, ATTR_CONST_NOTHROW_LIST)
-DEF_ATTR_TREE_LIST (ATTR_TM_NOTHROW_RT_LIST,
- ATTR_RETURNS_TWICE, ATTR_NULL, ATTR_TM_NOTHROW_LIST)
/* Same attributes used for BUILT_IN_MALLOC except with TM_PURE thrown in. */
DEF_ATTR_TREE_LIST (ATTR_TMPURE_MALLOC_NOTHROW_LIST,
diff --git a/gcc/gtm-builtins.def b/gcc/gtm-builtins.def
index 171019e..7dcad03 100644
--- a/gcc/gtm-builtins.def
+++ b/gcc/gtm-builtins.def
@@ -1,5 +1,5 @@
DEF_TM_BUILTIN (BUILT_IN_TM_START, "_ITM_beginTransaction",
- BT_FN_UINT32_UINT32_VAR, ATTR_TM_NOTHROW_RT_LIST)
+ BT_FN_UINT32_UINT32_VAR, ATTR_TM_NOTHROW_LIST)
DEF_TM_BUILTIN (BUILT_IN_TM_COMMIT, "_ITM_commitTransaction",
BT_FN_VOID, ATTR_TM_NOTHROW_LIST)
