[issue32635] test_crypt segfaults when using libxcrypt instead of libcrypt
Geoffrey Thomas added the comment: Opened #44751 and https://github.com/python/cpython/pull/27394 . -- ___ Python tracker <https://bugs.python.org/issue32635> ___ ___ Python-bugs-list mailing list Unsubscribe: https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
[issue44751] crypt.h should be in _cryptmodule.c, not in public header
Change by Geoffrey Thomas : -- keywords: +patch pull_requests: +25927 stage: -> patch review pull_request: https://github.com/python/cpython/pull/27394 ___ Python tracker <https://bugs.python.org/issue44751> ___ ___ Python-bugs-list mailing list Unsubscribe: https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
[issue44751] crypt.h should be in _cryptmodule.c, not in public header
New submission from Geoffrey Thomas : In #32635, it was discovered that _cryptmodule.c was missing a dependency on crypt.h, which caused it to segfault when it was missing the proper prototype for crypt. This was fixed by adding an #include to Python.h. This include doesn't need to be in the public header; it only needs to be in _cryptmodule.c. Removing it from the public header is helpful for packagers, because it means that the libpython-dev (or whatever) package doesn't need a dependency on libcrypt-dev, only on the libcrypt runtime library. -- components: C API messages: 398321 nosy: geofft priority: normal severity: normal status: open title: crypt.h should be in _cryptmodule.c, not in public header versions: Python 3.10, Python 3.11, Python 3.7, Python 3.8, Python 3.9 ___ Python tracker <https://bugs.python.org/issue44751> ___ ___ Python-bugs-list mailing list Unsubscribe: https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
[issue32635] test_crypt segfaults when using libxcrypt instead of libcrypt
Geoffrey Thomas added the comment: Sorry to bump an old thread, but - does need to be in the public Python.h header? Or would it be enough to be in _cryptmodule.c, which is the code that uses crypt_r()? I can provide a quick patch to change that if that seems sound. -- nosy: +geofft ___ Python tracker <https://bugs.python.org/issue32635> ___ ___ Python-bugs-list mailing list Unsubscribe: https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
[issue42854] OpenSSL 1.1.1: use SSL_write_ex() and SSL_read_ex()
Geoffrey Thomas added the comment: Christian mentioned on Twitter that this is probably due to a missing argument clinic change from "int" to "Py_ssize_t". I can confirm that fixing that and rerunning argument clinic makes things start to work. I don't have the ability to reopen this bug (I think), can someone reopen it please? -- ___ Python tracker <https://bugs.python.org/issue42854> ___ ___ Python-bugs-list mailing list Unsubscribe: https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
[issue42854] OpenSSL 1.1.1: use SSL_write_ex() and SSL_read_ex()
Geoffrey Thomas added the comment: I am still seeing failures to read responses over 2 GB in Python 3.10b1. I'm working on a reproducer, but I'm getting the same "OverflowError: signed integer is greater than maximum" that I get in 3.9. -- nosy: +geofft ___ Python tracker <https://bugs.python.org/issue42854> ___ ___ Python-bugs-list mailing list Unsubscribe: https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
