gh-126595: fix a crash when calling itertools.count(sys.maxsize)

[picnixz]

• Issue: Failed assertion in itertoolsmodule.c: itertools_count_impl for count(sys.maxsize) #126595

[skirpichev]

LGTM

But I doubt this should be backported.

[picnixz]

But I doubt this should be backported.

Considering it's a bug that makes the interpreter crash, I think this should.

[skirpichev]

Considering it's a bug that makes the interpreter crash

Only for debug builds.

[picnixz]

Yes, but I still think it's better to backport it since it can be used by third-party libraries (let's discuss it on the issue instead).

#126595 (comment)

[picnixz]

Ok, maybe it's better to have a separate PR actually.

[picnixz]

I'm reverting the __repr__ check and delegate it to #126620. The rationale is that there are some free-threaded update to do and I want to ease backports.

[skirpichev]

Yep, this seems correct.

Edit: One nitpick (a slightly smaller diff):

diff --git a/Modules/itertoolsmodule.c b/Modules/itertoolsmodule.c
index 1201fa0949..78fbdcdf77 100644
--- a/Modules/itertoolsmodule.c
+++ b/Modules/itertoolsmodule.c
@@ -3291,6 +3291,9 @@ itertools_count_impl(PyTypeObject *type, PyObject *long_cnt,
                 PyErr_Clear();
                 fast_mode = 0;
             }
+            else if (cnt == PY_SSIZE_T_MAX) {
+                fast_mode = 0;
+            }
         }
     } else {
         cnt = 0;

[picnixz]

I think it will be more likely to have cnt equal to maxsize rather than having an error. Which is why I put that test first.

[skirpichev]

I think it will be more likely to have cnt equal to maxsize rather than having an error.

Well, in terms of probability - it's more likely to have an error (this happens on countable subset of integers), rather having cnt equal to some fixed value ;)

[picnixz]

Actually I was wrong. I thought that the error only happened in bad situations but the overflow exception happens when you put something > maxsize so your fix is better!

[vstinner]

LGTM

[rhettinger]

+1 for backporting

[miss-islington-app]

Thanks @picnixz for the PR, and @vstinner for merging it 🌮🎉.. I'm working now to backport this PR to: 3.12, 3.13.
🐍🍒⛏🤖

[bedevere-app]

GH-126739 is a backport of this pull request to the 3.13 branch.

[miss-islington-app]

Thanks @picnixz for the PR, and @vstinner for merging it 🌮🎉.. I'm working now to backport this PR to: 3.12.
🐍🍒⛏🤖

[vstinner]

I don't know what is going on with the 3.12 backport, it seems to be stuck.

[vstinner]

Merged, thanks @picnixz for the fix and thanks @devdanzin for the bug report.

[bedevere-app]

GH-126740 is a backport of this pull request to the 3.12 branch.