[Bug tree-optimization/39367] [4.4 Regression] ICE at tree-inline.c:1042 with -O
--- Comment #6 from jakub at gcc dot gnu dot org 2009-03-06 12:48 ---
The problem is that mark_all_vars_used only marks blocks used by actual code,
but on the testcase the only reference to a one BLOCK is from TYPE_MAX_VALUE
(TYPE_DOMAIN (type)) containing a PLUS_EXPR with that BLOCK as TREE_BLOCK.
IMHO either we need to talk also types in
remove_unused_locals/mark_all_vars_used, or we should drop TREE_BLOCK stuff
from expressions referenced from types during gimplification (do we actually
ever care about TREE_BLOCK of expressions in TYPE_{MIN,MAX}_VALUE,
TYPE_SIZE{,_UNIT}, DECL_SIZE{,_UNIT} etc.?), or during inlining we should just
ignore remapping of blocks for these expressions. If a BLOCK isn't found for
them, no big deal, just drop it.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39367
[Bug tree-optimization/39367] [4.4 Regression] ICE at tree-inline.c:1042 with -O
--- Comment #7 from rguenth at gcc dot gnu dot org 2009-03-06 12:58 --- Yeah, we should drop TREE_BLOCK from anything not in stmts. Definitely for expressions in the types sizes and domains. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39367
[Bug tree-optimization/39367] [4.4 Regression] ICE at tree-inline.c:1042 with -O
-- rguenth at gcc dot gnu dot org changed: What|Removed |Added CC||hubicka at gcc dot gnu dot ||org Component|c++ |tree-optimization Keywords||ice-on-valid-code Summary|ICE at tree-inline.c:1042 |[4.4 Regression] ICE at |with -O |tree-inline.c:1042 with -O Target Milestone|--- |4.4.0 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39367
[Bug tree-optimization/39367] [4.4 Regression] ICE at tree-inline.c:1042 with -O
--- Comment #2 from rguenth at gcc dot gnu dot org 2009-03-04 21:20 --- Reducing. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39367
[Bug tree-optimization/39367] [4.4 Regression] ICE at tree-inline.c:1042 with -O
--- Comment #3 from rguenth at gcc dot gnu dot org 2009-03-04 22:08 ---
Reduced testcase:
class QScriptEnginePrivate;
class QScriptClassInfo;
namespace QScript {
enum Type { InvalidType };
};
class QScriptValueImpl {
public:
inline QScriptValueImpl();
QScript::Type m_type;
};
namespace QScript {
namespace Ecma {
class Core {
public:
inline QScriptEnginePrivate *engine() const { }
inline QScriptClassInfo *classInfo() const { }
QScriptValueImpl publicPrototype;
};
class Boolean: public Core {
void newBoolean(QScriptValueImpl *result, bool value = false);
};
}
template typename T class Buffer {
public:
inline void reserve(int num);
inline void resize(int s);
T *m_data;
int m_capacity;
int m_size;
};
}
template typename T void QScript::BufferT::resize(int s) {
if (m_capacity s)
reserve (s 1);
}
template typename T void QScript::BufferT::reserve(int x) {
T *new_data = new T[m_capacity];
for (int i=0; im_size; ++i)
new_data[i] = m_data[i];
}
class QScriptObject {
public:
inline void reset();
QScript::BufferQScriptValueImpl m_values;
};
class QScriptEnginePrivate {
public:
inline QScriptObject *allocObject();
inline void newObject(QScriptValueImpl *o, const QScriptValueImpl proto,
QScriptClassInfo *oc = 0);
};
inline void QScriptEnginePrivate::newObject(QScriptValueImpl *o,
const QScriptValueImpl proto,
QScriptClassInfo *oc)
{
QScriptObject *od = allocObject();
od-reset();
}
inline QScriptValueImpl::QScriptValueImpl() : m_type(QScript::InvalidType) { }
inline void QScriptObject::reset() { m_values.resize(0); }
namespace QScript {
namespace Ecma {
void Boolean::newBoolean(QScriptValueImpl *result, bool value)
{
engine()-newObject(result, publicPrototype, classInfo());
}
}
}
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
Status|UNCONFIRMED |NEW
Ever Confirmed|0 |1
Last reconfirmed|-00-00 00:00:00 |2009-03-04 22:08:34
date||
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39367
[Bug tree-optimization/39367] [4.4 Regression] ICE at tree-inline.c:1042 with -O
--- Comment #4 from hjl dot tools at gmail dot com 2009-03-05 00:12 --- This is caused by revision 144529. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39367
[Bug tree-optimization/39367] [4.4 Regression] ICE at tree-inline.c:1042 with -O
--- Comment #5 from hjl dot tools at gmail dot com 2009-03-05 06:04 --- *** Bug 39377 has been marked as a duplicate of this bug. *** -- hjl dot tools at gmail dot com changed: What|Removed |Added CC||dirtyepic at gentoo dot org http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39367
