r205086's implementation of prettyprinting vec<>* didn't check for NULL
before attempting to walk the elements of the vec, leading to noise
within gdb:
(gdb) p cfun->cfg->x_entry_block_ptr->preds
$12 = 0x0<error reading variable: Cannot access memory at address 0x4>
The attached patch fixes this issue, giving this output:
(gdb) p cfun->cfg->x_entry_block_ptr->preds
$6 = 0x0
Committed to trunk as r205139.
Index: gcc/ChangeLog
===================================================================
--- gcc/ChangeLog (revision 205138)
+++ gcc/ChangeLog (revision 205139)
@@ -1,3 +1,8 @@
+2013-11-20 David Malcolm <dmalc...@redhat.com>
+
+ * gdbhooks.py (VecPrinter.children): Don't attempt to iterate
+ the children of a NULL pointer.
+
2013-11-20 Robert Suchanek <robert.sucha...@imgtec.com>
* lra.c (lra): Set lra_in_progress before check_rtl call.
Index: gcc/gdbhooks.py
===================================================================
--- gcc/gdbhooks.py (revision 205138)
+++ gcc/gdbhooks.py (revision 205139)
@@ -383,6 +383,8 @@
return '0x%x' % long(self.gdbval)
def children (self):
+ if long(self.gdbval) == 0:
+ return
m_vecpfx = self.gdbval['m_vecpfx']
m_num = m_vecpfx['m_num']
m_vecdata = self.gdbval['m_vecdata']
