Tested on x86_64-darwin, pushed to trunk, thanks,
Iain
--- 8< ---
In select_section we need to determine if a CONSTRUCTOR initialises
an Objective-C/CoreFoundation string which we do by inspecting the
typename. It is possible for a typename to be passed, even if perhaps
in error, with a NULL_TREE id and we should deal with that gracefully.
gcc/ChangeLog:
* config/darwin.cc (machopic_select_section): Do not try to
inspect a typename which has no id.
Signed-off-by: Iain Sandoe <[email protected]>
---
gcc/config/darwin.cc | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/gcc/config/darwin.cc b/gcc/config/darwin.cc
index ba934c26402..7e822a9a852 100644
--- a/gcc/config/darwin.cc
+++ b/gcc/config/darwin.cc
@@ -1839,6 +1839,11 @@ machopic_select_section (tree decl,
if (TREE_CODE (name) == TYPE_DECL)
name = DECL_NAME (name);
+ /* If the type name ID absent, this cannot be an Objective-C or Core
+ Foundation string. */
+ if (!name)
+ return base_section;
+
if (!strcmp (IDENTIFIER_POINTER (name), "__builtin_ObjCString"))
{
if (flag_next_runtime)
--
2.50.1 (Apple Git-155)
