#15671: sig_error() without sig_on()
-------------------------------------+-------------------------------------
Reporter: zimmerma | Owner:
Type: defect | Status: needs_work
Priority: major | Milestone: sage-6.4
Component: cython | Resolution:
Keywords: | Merged in:
Authors: Peter Bruin, | Reviewers:
Jeroen Demeyer | Work issues:
Report Upstream: N/A | Commit:
Branch: | 84ef4e5fc1bb75cc74ea9147e36efc243dc1a6ab
u/jdemeyer/ticket/15671 | Stopgaps:
Dependencies: |
-------------------------------------+-------------------------------------
Comment (by pbruin):
Replying to [comment:16 jdemeyer]:
> We should indeed put `sig_on()`/`sig_off()` in many more places, but I'm
afraid people will worry about the performance impact (there is a very
small but still measurable slow-down with `sig_on()`).
Indeed, it seems that at least in the first run we explicitly do not use
`sig_on()`:
{{{
#!python
cdef ModuleElement _mul_long(self, long n):
"""
Fast path for multiplying a C long.
...
"""
cdef Integer x = <Integer>PY_NEW(Integer)
if mpz_size(self.value) > 100000:
sig_on()
mpz_mul_si(x.value, self.value, n)
sig_off()
else:
mpz_mul_si(x.value, self.value, n)
return x
}}}
In the second one, `fast_tp_new()` does not check the result of an
allocation:
{{{
#!python
new = PyObject_MALLOC( sizeof_Integer )
...
memcpy(new, (<void*>global_dummy_Integer), sizeof_Integer )
}}}
The signal happens in `memcpy()`; I guess `PyObject_MALLOC()` returned
`NULL`.
(Both fragments are from `integer.pyx`.)
--
Ticket URL: <http://trac.sagemath.org/ticket/15671#comment:18>
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 unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/sage-trac.
For more options, visit https://groups.google.com/d/optout.
