Fix various invalid uses of sig_on()

[jdemeyer]

In order to debug crashes with cypari2 (#26442), I added some checks to cypari2. But these checks are now failing because sig_on() is regularly used badly inside Sage.

As documented in the cysignals documentation, code inside a sig_on()/sig_off() block should not manipulate Python objects. The reason is that an interrupt at that time could put Python objects in an invalid state. It is especially problematic if a garbage collection happens inside a sig_on() block (because that takes a non-trivial amount of time and has a high probability of putting things in an invalid state).

This ticket fixes a few of those cases where garbage collection might happen.

Component: cython

Author: Jeroen Demeyer

Branch/Commit: 9fffd00

Reviewer: Timo Kaufmann

Issue created by migration from https://trac.sagemath.org/ticket/27060

[jdemeyer]

Branch: u/jdemeyer/fix_various_invalid_uses_of_sig_on__

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

ab966ad

Fix various invalid uses of sig_on()

[sagetrac-git]

Commit: ab966ad

[sagetrac-git]

Changed commit from ab966ad to 9fffd00

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

9fffd00

Fix various invalid uses of sig_on()

[embray]

comment:5

Retarging tickets optimistically to the next milestone. If you are responsible for this ticket (either its reporter or owner) and don't believe you are likely to complete this ticket before the next release (8.7) please retarget this ticket's milestone to sage-pending or sage-wishlist.

[timokau]

comment:6

On the macro-scale this LGTM. On the details, the person I'd choose to review this is you. But since you already made the commit...

[timokau]

Reviewer: Timo Kaufmann

[embray]

comment:8

Could you say a little bit more about why these sig_on() are not correct? Is it because they're releasing the GIL inappropriately?

[jdemeyer]

comment:10

Replying to @embray:

Could you say a little bit more about why these sig_on() are not correct? Is it because they're releasing the GIL inappropriately?

I updated the ticket description, I hope it is clearer now. It has nothing to do with the GIL.

[embray]

comment:12

I see what you're saying now--thank you.

[vbraun]

Changed branch from u/jdemeyer/fix_various_invalid_uses_of_sig_on__ to 9fffd00