This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH, PR 43257] Mangle DECL_ASSEMBLER_NAME early in IPA-SRA
Hi,
On Wed, Mar 10, 2010 at 10:56:24AM +0100, Richard Guenther wrote:
> On Tue, 9 Mar 2010, Martin Jambor wrote:
> > the patch below makes IPA-SRA mangle the assembler name of a function
> > decl before changing its type to avoid potential conflicts with other
> > functions.
> >
> > Bootstrapped and regression tested on x86_64-linux. OK for trunk?
>
> Hm. Please look at tree.c:free_lang_data_in_cgraph - that seems to
> fixup some diagnostic issues.
I see.
> But on the whole thing I wonder if
> we shouldn't simply make sure that functions with a cgraph node
> have an assembler name set in cgraph_finalize_function (or
> cgraph_analyze_function if we want to delay it until after
> cgraph_finalize_compilation_unit)? Thus, replicate the
> if (need_assembler_name_p (t)) from tree.c:free_lang_data_in_cgraph
> in cgraph_analyze_function (and maybe similar in
> varpool_analyze_pending_decls).
>
> Can you try doing just the cgraph part in cgraph_analyze_function?
I am not sure if I understand what you mean. Do you want just the
declaration of the function to have its assembler name assigned at
that time or do you think we should duplicate the whole declaration
discovery machinery there as well? Doing the former was easy, it does
fix the issue and the (bootstrapped and tested) patch is below. On an
i686, the memory consumption as measured by maxmem2.sh rose only very
slightly, from 272220 kB to 272268 kB.
Thanks,
Martin
2010-03-11 Martin Jambor <mjambor@suse.cz>
PR tree-optimization/43257
* tree.c (assign_assembler_name_if_neeeded): New function.
(free_lang_data_in_cgraph): Assembler name assignment moved to
the above new function.
* tree.h (assign_assembler_name_if_neeeded): Declare.
* cgraphunit.c (cgraph_finalize_function): Create an assembler name for
the function if needed.
* testsuite/g++.dg/torture/pr43257.C: New test.
Index: mine/gcc/testsuite/g++.dg/torture/pr43257.C
===================================================================
--- /dev/null
+++ mine/gcc/testsuite/g++.dg/torture/pr43257.C
@@ -0,0 +1,30 @@
+/* { dg-do assemble } */
+
+class A {};
+class B {};
+
+static void *func (int n)
+{
+ void *p;
+ if (p == 0) throw ::A ();
+}
+
+static void *func (int n, B const &)
+{
+ try {
+ return func (n);
+ }
+ catch (::A const &) {
+ }
+ return func (n);
+}
+
+void *f1 (int n)
+{
+ return func (n, B());
+}
+
+void *f2 (int n)
+{
+ return func (n, B());
+}
Index: mine/gcc/cgraphunit.c
===================================================================
--- mine.orig/gcc/cgraphunit.c
+++ mine/gcc/cgraphunit.c
@@ -523,6 +523,8 @@ cgraph_finalize_function (tree decl, boo
if (cgraph_decide_is_function_needed (node, decl))
cgraph_mark_needed_node (node);
+ assign_assembler_name_if_neeeded (node->decl);
+
/* Since we reclaim unreachable nodes at the end of every language
level unit, we need to be conservative about possible entry points
there. */
Index: mine/gcc/tree.c
===================================================================
--- mine.orig/gcc/tree.c
+++ mine/gcc/tree.c
@@ -4831,6 +4831,33 @@ find_decls_types_in_var (struct varpool_
find_decls_types (v->decl, fld);
}
+/* If T needs an assembler name, have one created for it. */
+
+void
+assign_assembler_name_if_neeeded (tree t)
+{
+ if (need_assembler_name_p (t))
+ {
+ /* When setting DECL_ASSEMBLER_NAME, the C++ mangler may emit
+ diagnostics that use input_location to show locus
+ information. The problem here is that, at this point,
+ input_location is generally anchored to the end of the file
+ (since the parser is long gone), so we don't have a good
+ position to pin it to.
+
+ To alleviate this problem, this uses the location of T's
+ declaration. Examples of this are
+ testsuite/g++.dg/template/cond2.C and
+ testsuite/g++.dg/template/pr35240.C. */
+ location_t saved_location = input_location;
+ input_location = DECL_SOURCE_LOCATION (t);
+
+ decl_assembler_name (t);
+
+ input_location = saved_location;
+ }
+}
+
/* Free language specific information for every operand and expression
in every node of the call graph. This process operates in three stages:
@@ -4880,26 +4907,7 @@ free_lang_data_in_cgraph (void)
now because free_lang_data_in_decl will invalidate data needed
for mangling. This breaks mangling on interdependent decls. */
for (i = 0; VEC_iterate (tree, fld.decls, i, t); i++)
- if (need_assembler_name_p (t))
- {
- /* When setting DECL_ASSEMBLER_NAME, the C++ mangler may emit
- diagnostics that use input_location to show locus
- information. The problem here is that, at this point,
- input_location is generally anchored to the end of the file
- (since the parser is long gone), so we don't have a good
- position to pin it to.
-
- To alleviate this problem, this uses the location of T's
- declaration. Examples of this are
- testsuite/g++.dg/template/cond2.C and
- testsuite/g++.dg/template/pr35240.C. */
- location_t saved_location = input_location;
- input_location = DECL_SOURCE_LOCATION (t);
-
- decl_assembler_name (t);
-
- input_location = saved_location;
- }
+ assign_assembler_name_if_neeeded (t);
/* Traverse every decl found freeing its language data. */
for (i = 0; VEC_iterate (tree, fld.decls, i, t); i++)
Index: mine/gcc/tree.h
===================================================================
--- mine.orig/gcc/tree.h
+++ mine/gcc/tree.h
@@ -4720,6 +4720,8 @@ extern tree build_low_bits_mask (tree, u
extern tree tree_strip_nop_conversions (tree);
extern tree tree_strip_sign_nop_conversions (tree);
extern tree lhd_gcc_personality (void);
+extern void assign_assembler_name_if_neeeded (tree);
+
/* In cgraph.c */
extern void change_decl_assembler_name (tree, tree);