#12313: Fix yet another memory leak caused by caching of coercion data
--------------------------------------------------+-------------------------
Reporter: SimonKing | Owner:
Type: defect | Status:
needs_review
Priority: major | Milestone: sage-5.3
Component: memleak | Resolution:
Keywords: coercion weak dictionary | Work issues:
Report Upstream: N/A | Reviewers: Simon King,
Jean-Pierre Flori, John Perry
Authors: Simon King, Jean-Pierre Flori | Merged in:
Dependencies: #11521, #11599, #12969, #12215 | Stopgaps:
--------------------------------------------------+-------------------------
Comment (by nbruin):
It looks like that garbage collection is a great idea! The error is now
completely reproducible:
{{{
sage -t --verbose -force_lib
"devel/sage/sage/rings/polynomial/pbori.pyx" # Killed/crashed
}}}
I've instrumented the remove function on the `WeakValueDictionary` a
little bit:
{{{#!diff
def remove(wr, selfref=ref(self)):
self = selfref()
if self is not None:
+ if type(wr.key) is tuple:
+ T=[]
+ for t in wr.key:
+ try:
+ T.append(str(t))
+ except:
+ T.append("TYPE: %s"%type(t))
+ print('removal on key %s'%T)
del self.data[wr.key]
self._remove = remove
UserDict.UserDict.__init__(self, *args, **kw)
}}}
to find (now doctests are failing due to the extra output):
{{{
...
***Test Failed*** 18 failures.
removal on key ['Rational Field', "('z', 'y')", '2', 'False', 'Degree
reverse lexicographic term order']
*** glibc detected *** python: double free or corruption (!prev):
0x0000000006baaaa0 ***
...
Unhandled SIGABRT: An abort() occurred in Sage.
}}}
That looks an awful lot like the kind of key that a polynomial ring
constructor would get. So I guess it's just one of those
`WeakCachedFunction` dictionaries again. I suspect that part of polybori
is already torn down when this entry is looked up. It really seems that
the alternatives are:
- all those libraries to keep sufficient references to do equality
testing till the very end
- make a WeakValueDictionary that doesn't need key comparisons on its
callback and use that instead
- convince the release manager that these tickets are important and that
these doctests are not the fault of these tickets. Remove any components
that fail due to these changes.
Option 2 looks quite doable by comparison ... I'd say use a design rather
similar to `MonoDict`. We probably won't beat `WeakValueDict` for standard
performance, but we're making another trade-off: we want safe removal,
even at the price of a little performance. So perhaps two variants,
`WeakCachedFunction` and `SafeWeakCachedFunction`.
--
Ticket URL: <http://trac.sagemath.org/sage_trac/ticket/12313#comment:213>
Sage <http://www.sagemath.org>
Sage: Creating a Viable Open Source Alternative to Magma, Maple, Mathematica,
and MATLAB
--
You received this message because you are subscribed to the Google Groups
"sage-trac" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/sage-trac?hl=en.
