Fix memory leak in si2sa_ZZ in sage/libs/singular/singular.pyx

[sagetrac-mabshoff]

We currently leak on mpz in si2sa_ZZ:

==696== 104 bytes in 13 blocks are definitely lost in loss record 11,644 of 19,410
==696==    at 0x4A1BE1B: malloc (vg_replace_malloc.c:207)
==696==    by 0x6760947: __gmpz_init (in /scratch/mabshoff/release-cycle/sage-3.1.3.final/local/lib/libgmp.so.3.4.1)
==696==    by 0x15E05AAE: __pyx_f_4sage_4libs_8singular_8singular_10Conversion_si2sa_ZZ (singular.cpp:2513)
==696==    by 0x15E07C85: __pyx_f_4sage_4libs_8singular_8singular_10Conversion_si2sa (singular.cpp:4803)
==696==    by 0x1572034D: __pyx_pf_4sage_5rings_10polynomial_28multi_polynomial_libsingular_23MPolynomial_libsingular_coefficients(_object*, _object*) (multi_polynomial_libsingular.cpp:25706)
==696==    by 0x415832: PyObject_Call (abstract.c:1861)
==696==    by 0x15735A21: __pyx_pf_4sage_5rings_10polynomial_28multi_polynomial_libsingular_23MPolynomial_libsingular_gcd(_object*, _object*, _object*) (multi_polynomial_libsingular.cpp:23546)

The attached patch fixes that.

Cheers,

Michael

CC: @malb

Component: memleak

Issue created by migration from https://trac.sagemath.org/ticket/4387

[craigcitro]

comment:1

Attachment: trac_4387.patch.gz

Notice that the mpz_init happens twice -- this is definitely the right fix.

[sagetrac-mabshoff]

comment:2

Merged in Sage 3.2.alpha2

[sagetrac-mabshoff]

comment:3

CCed malb so he knows about the issue.

Cheers,

Michael