Memleak in singular.pyx

[jpflori]

Using the following piece of code makes the memory footprint of sage
grow indefinitely:

sage: K = GF(1<<50,'t') 
sage: R.<x,y> = PolynomialRing(K)
sage: a = K.random_element()
sage: while 1: 
....:      R(a)
....:

The memleak happens when different si2sa_* functions are called.

See http://groups.google.com/group/sage-support/browse_thread/thread/9a8e887df34a8e9a for further discussion.

CC: @jpflori @burcin

Component: memleak

Keywords: libsingular

Author: Jean-Pierre Flori

Reviewer: Mariah Lenox, Jonathan Bober

Merged: sage-4.7.2.alpha1

Issue created by migration from https://trac.sagemath.org/ticket/11468

[jpflori]

Changed keywords from none to libsingular

[jpflori]

comment:2

Calling gc.collect() just after the creation prevents the memory problem.

But it does not if called later.

[jpflori]

comment:4

I finally found the memleaks in different si2sa_* functions.

Potential fix provided.

[jpflori]

Author: Jean-Pierre Flori

[sagetrac-mariah]

comment:8

The patch does not seem to fix the reported problem. I applied the patch to sage-4.7.1.alpha2, did 'sage -b', yet I still see memory
increasing when I run the code in the ticket description.

[jpflori]

comment:9

I just retested it and it seems to fix the memleak for me:

no increase in memory footprint after 30 minutes.

Which code did you run ?

The one in the tickect description or the one on sage-support post ?

Because there are other memleaks involved when using EllipticCurve class.

See #11495 and #11521 for example.

[sagetrac-mariah]

comment:10

Replying to @jpflori:

I just retested it and it seems to fix the memleak for me:

no increase in memory footprint after 30 minutes.

Which code did you run ?

The one in the tickect description or the one on sage-support post ?

I used the code in the ticket description.

I will try again.

[sagetrac-mariah]

Reviewer: Mariah Lenox

[sagetrac-mariah]

comment:11

I must have forgotten to do 'sage -b' when I tried before. Apologies
for the noise.

This time when I applied the patch (and did 'sage -b') the code in the
ticket description ran for 30 CPU minutes without an increase in memory.
I also did 'make testlong' and all tests passed.

Positive review!

[jdemeyer]

comment:12

The commit message

Trac 114666666x memleaks in singular.pyx

looks a bit odd...

Also, if you change the "copyright" message, at least make it such that it looks like http://sagemath.org/doc/developer/conventions.html#headings-of-sage-library-code-files

[jpflori]

Attachment: trac_11468-memleaks_singular.patch.gz

New version with correct commit message and copyright.

[jpflori]

comment:14

Just apply:

trac_11468-memleaks_singular.patch

[sagetrac-bober]

comment:15

This patch looks good to me. I don't get a leak with either the example or the code I was running that lead me to this ticket, and the changes that the patch makes seem simple and sensible.

[sagetrac-bober]

Changed reviewer from Mariah Lenox to Mariah Lenox, Jonathan Bober

[jdemeyer]

Merged: sage-4.7.2.alpha1