* [PATCH RFA] -Wdangling-pointer: fix MEM_REF handling [PR109514]
@ 2023-04-14 23:42 Jason Merrill
2023-04-15 1:22 ` Jeff Law
0 siblings, 1 reply; 3+ messages in thread
From: Jason Merrill @ 2023-04-14 23:42 UTC (permalink / raw)
To: gcc-patches
Tested x86_64-pc-linux-gnu, ok for trunk?
-- 8< --
Here we hit the MEM_REF case, with its arg an ADDR_EXPR, but had no handling
for that and wrongly assumed it would be a reference to a local variable.
This patch overhauls the logic for deciding whether the target is something
to warn about so that we only warn if we specifically recognize the target
as non-local. None of the existing tests regress as a result.
PR c++/109514
gcc/ChangeLog:
* gimple-ssa-warn-access.cc (pass_waccess::check_dangling_stores):
Overhaul lhs_ref.ref analysis.
gcc/testsuite/ChangeLog:
* g++.dg/warn/Wdangling-pointer-6.C: New test.
---
gcc/gimple-ssa-warn-access.cc | 50 ++++++++-----------
.../g++.dg/warn/Wdangling-pointer-6.C | 30 +++++++++++
2 files changed, 50 insertions(+), 30 deletions(-)
create mode 100644 gcc/testsuite/g++.dg/warn/Wdangling-pointer-6.C
diff --git a/gcc/gimple-ssa-warn-access.cc b/gcc/gimple-ssa-warn-access.cc
index b3de4b77924..d0d2148c872 100644
--- a/gcc/gimple-ssa-warn-access.cc
+++ b/gcc/gimple-ssa-warn-access.cc
@@ -4528,39 +4528,34 @@ pass_waccess::check_dangling_stores (basic_block bb,
if (!m_ptr_qry.get_ref (lhs, stmt, &lhs_ref, 0))
continue;
- if (auto_var_p (lhs_ref.ref))
- continue;
-
- if (DECL_P (lhs_ref.ref))
+ if (TREE_CODE (lhs_ref.ref) == MEM_REF)
{
- if (!POINTER_TYPE_P (TREE_TYPE (lhs_ref.ref))
- || lhs_ref.deref > 0)
- continue;
+ lhs_ref.ref = TREE_OPERAND (lhs_ref.ref, 0);
+ ++lhs_ref.deref;
}
- else if (TREE_CODE (lhs_ref.ref) == SSA_NAME)
+ if (TREE_CODE (lhs_ref.ref) == ADDR_EXPR)
+ {
+ lhs_ref.ref = TREE_OPERAND (lhs_ref.ref, 0);
+ --lhs_ref.deref;
+ }
+ if (TREE_CODE (lhs_ref.ref) == SSA_NAME)
{
gimple *def_stmt = SSA_NAME_DEF_STMT (lhs_ref.ref);
if (!gimple_nop_p (def_stmt))
/* Avoid looking at or before stores into unknown objects. */
return;
- tree var = SSA_NAME_VAR (lhs_ref.ref);
- if (TREE_CODE (var) == PARM_DECL && DECL_BY_REFERENCE (var))
- /* Avoid by-value arguments transformed into by-reference. */
- continue;
+ lhs_ref.ref = SSA_NAME_VAR (lhs_ref.ref);
+ }
- }
- else if (TREE_CODE (lhs_ref.ref) == MEM_REF)
- {
- tree arg = TREE_OPERAND (lhs_ref.ref, 0);
- if (TREE_CODE (arg) == SSA_NAME)
- {
- gimple *def_stmt = SSA_NAME_DEF_STMT (arg);
- if (!gimple_nop_p (def_stmt))
- return;
- }
- }
+ if (TREE_CODE (lhs_ref.ref) == PARM_DECL
+ && (lhs_ref.deref - DECL_BY_REFERENCE (lhs_ref.ref)) > 0)
+ /* Assignment through a (real) pointer/reference parameter. */;
+ else if (TREE_CODE (lhs_ref.ref) == VAR_DECL
+ && !auto_var_p (lhs_ref.ref))
+ /* Assignment to/through a non-local variable. */;
else
+ /* Something else, don't warn. */
continue;
if (stores.add (lhs_ref.ref))
@@ -4587,13 +4582,8 @@ pass_waccess::check_dangling_stores (basic_block bb,
location_t loc = DECL_SOURCE_LOCATION (rhs_ref.ref);
inform (loc, "%qD declared here", rhs_ref.ref);
- if (DECL_P (lhs_ref.ref))
- loc = DECL_SOURCE_LOCATION (lhs_ref.ref);
- else if (EXPR_HAS_LOCATION (lhs_ref.ref))
- loc = EXPR_LOCATION (lhs_ref.ref);
-
- if (loc != UNKNOWN_LOCATION)
- inform (loc, "%qE declared here", lhs_ref.ref);
+ loc = DECL_SOURCE_LOCATION (lhs_ref.ref);
+ inform (loc, "%qD declared here", lhs_ref.ref);
}
}
diff --git a/gcc/testsuite/g++.dg/warn/Wdangling-pointer-6.C b/gcc/testsuite/g++.dg/warn/Wdangling-pointer-6.C
new file mode 100644
index 00000000000..20d47edecc5
--- /dev/null
+++ b/gcc/testsuite/g++.dg/warn/Wdangling-pointer-6.C
@@ -0,0 +1,30 @@
+// PR c++/109514
+// { dg-do compile { target c++11 } }
+// { dg-additional-options "-O2 -Werror=dangling-pointer" }
+
+struct _Rb_tree_node_base {
+ _Rb_tree_node_base *_M_parent;
+};
+struct _Rb_tree_header {
+ _Rb_tree_node_base _M_header;
+ void _M_move_data() { _M_header._M_parent->_M_parent = &_M_header; }
+};
+struct _Rb_tree {
+ _Rb_tree_header _M_impl;
+ _Rb_tree_node_base *&_M_root() { return _M_impl._M_header._M_parent; }
+ _Rb_tree();
+ _Rb_tree &operator=(_Rb_tree &&);
+};
+_Rb_tree &_Rb_tree::operator=(_Rb_tree &&) {
+ if (_M_root())
+ _M_impl._M_move_data();
+ return *this;
+}
+struct set {
+ _Rb_tree _M_t;
+};
+set FilterRs();
+void f() {
+ set rs;
+ rs = FilterRs();
+};
base-commit: f32f7881fb0db085479525b5a23db5dabd990c3b
--
2.31.1
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH RFA] -Wdangling-pointer: fix MEM_REF handling [PR109514]
2023-04-14 23:42 [PATCH RFA] -Wdangling-pointer: fix MEM_REF handling [PR109514] Jason Merrill
@ 2023-04-15 1:22 ` Jeff Law
2023-04-15 1:54 ` Jason Merrill
0 siblings, 1 reply; 3+ messages in thread
From: Jeff Law @ 2023-04-15 1:22 UTC (permalink / raw)
To: Jason Merrill, gcc-patches
On 4/14/23 17:42, Jason Merrill via Gcc-patches wrote:
> Tested x86_64-pc-linux-gnu, ok for trunk?
>
> -- 8< --
>
> Here we hit the MEM_REF case, with its arg an ADDR_EXPR, but had no handling
> for that and wrongly assumed it would be a reference to a local variable.
> This patch overhauls the logic for deciding whether the target is something
> to warn about so that we only warn if we specifically recognize the target
> as non-local. None of the existing tests regress as a result.
>
> PR c++/109514
>
> gcc/ChangeLog:
>
> * gimple-ssa-warn-access.cc (pass_waccess::check_dangling_stores):
> Overhaul lhs_ref.ref analysis.
>
> gcc/testsuite/ChangeLog:
>
> * g++.dg/warn/Wdangling-pointer-6.C: New test.
One might ask if we have a (MEM_REF (ADDR_REF ...)) why we didn't
simplify that by removing the MEM_REF and ADDR_REF wrappers. But on
the assumption there's a good reason to have such nodes, OK for the trunk.
jeff
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH RFA] -Wdangling-pointer: fix MEM_REF handling [PR109514]
2023-04-15 1:22 ` Jeff Law
@ 2023-04-15 1:54 ` Jason Merrill
0 siblings, 0 replies; 3+ messages in thread
From: Jason Merrill @ 2023-04-15 1:54 UTC (permalink / raw)
To: Jeff Law, gcc-patches
On 4/14/23 21:22, Jeff Law wrote:
>
>
> On 4/14/23 17:42, Jason Merrill via Gcc-patches wrote:
>> Tested x86_64-pc-linux-gnu, ok for trunk?
>>
>> -- 8< --
>>
>> Here we hit the MEM_REF case, with its arg an ADDR_EXPR, but had no
>> handling
>> for that and wrongly assumed it would be a reference to a local variable.
>> This patch overhauls the logic for deciding whether the target is
>> something
>> to warn about so that we only warn if we specifically recognize the
>> target
>> as non-local. None of the existing tests regress as a result.
>>
>> PR c++/109514
>>
>> gcc/ChangeLog:
>>
>> * gimple-ssa-warn-access.cc (pass_waccess::check_dangling_stores):
>> Overhaul lhs_ref.ref analysis.
>>
>> gcc/testsuite/ChangeLog:
>>
>> * g++.dg/warn/Wdangling-pointer-6.C: New test.
> One might ask if we have a (MEM_REF (ADDR_REF ...)) why we didn't
> simplify that by removing the MEM_REF and ADDR_REF wrappers.
FYI in this case the type of the MEM_REF is the type of a member of a
member of a member of the variable.
> But on
> the assumption there's a good reason to have such nodes, OK for the trunk.
>
> jeff
>
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2023-04-15 1:54 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-04-14 23:42 [PATCH RFA] -Wdangling-pointer: fix MEM_REF handling [PR109514] Jason Merrill
2023-04-15 1:22 ` Jeff Law
2023-04-15 1:54 ` Jason Merrill
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).