A vulnerability was found in library/glob.html in the Python 2 and 3 documentation before 2016 has potentially misleading information about whether sorting occurs, as demonstrated by irreproducible cancer-research results. NOTE: the effects of this documentation cross application domains, and thus it is likely that security-relevant code elsewhere is affected. This issue is not a Python implementation bug, and there are no reports that NMR researchers were specifically relying on library/glob.html. In other words, because the older documentation stated "finds all the pathnames matching a specified pattern according to the rules used by the Unix shell," one might have incorrectly inferred that the sorting that occurs in a Unix shell also occurred for glob.glob. There is a workaround in newer versions of Willoughby nmr-data_compilation-p2.py and nmr-data_compilation-p3.py, which call sort() directly. Reference: https://bugs.python.org/issue33275 https://github.com/bminor/bash/blob/ac50fbac377e32b98d2de396f016ea81e8ee9961/pathexp.c#L380 https://github.com/bminor/bash/blob/ac50fbac377e32b98d2de396f016ea81e8ee9961/pathexp.c#L405 https://pubs.acs.org/doi/full/10.1021/acs.orglett.9b03216 https://pubs.acs.org/doi/suppl/10.1021/acs.orglett.9b03216/suppl_file/ol9b03216_si_002.zip https://security.netapp.com/advisory/ntap-20191107-0005/ https://twitter.com/chris_bloke/status/1181997278136958976 https://twitter.com/LucasCMoore/status/1181615421922824192 https://web.archive.org/web/20150822013622/https://docs.python.org/3/library/glob.html https://web.archive.org/web/20150906020027/https://docs.python.org/2.7/library/glob.html https://web.archive.org/web/20160309211341/https://docs.python.org/3/library/glob.html https://web.archive.org/web/20160526201356/https://docs.python.org/2.7/library/glob.html https://www.vice.com/en_us/article/zmjwda/a-code-glitch-may-have-caused-errors-in-more-than-100-published-studies
Created python2 tracking bugs for this issue: Affects: fedora-all [bug 1824450] Created python3 tracking bugs for this issue: Affects: fedora-all [bug 1824454] Created python34 tracking bugs for this issue: Affects: epel-all [bug 1824448] Affects: fedora-all [bug 1824453] Created python35 tracking bugs for this issue: Affects: fedora-all [bug 1824451] Created python36 tracking bugs for this issue: Affects: epel-7 [bug 1824447] Affects: fedora-all [bug 1824452] Created python38 tracking bugs for this issue: Affects: fedora-all [bug 1824455]
Dhananjay, what can I do to stop more EPEL python36 security bugzillas? The package has been retired in EPEL 6 months ago.
This does not seem to me like a real flaw. It was not even a bug actually, but just a slightly misleading documentation. I don't think this is a good use of a CVE, but rather a CVE should be assigned to particular programs that wrongly assume the sorting of that function. Victor, what do you think?
In reply to comment #2: > Dhananjay, what can I do to stop more EPEL python36 security bugzillas? The > package has been retired in EPEL 6 months ago. removed python36 for epel and updated my manifest file.
CVE-2019-17514 should be rejected: the behavior is intentional, it is not a vulnerability. os.listdir() and glob.glob() are not sorted on purpose. See this discussion for more details: https://discuss.python.org/t/a-code-glitch-may-have-caused-errors-in-more-than-100-published-studies/2583 The "fix" was to ensure that the intentional behavior is properly documented: https://docs.python.org/dev/library/os.html#os.listdir "The list is in arbitrary order" https://docs.python.org/dev/library/glob.html#glob.glob "Whether or not the results are sorted depends on the file system." Python issues: * https://bugs.python.org/issue21748 closed as "not a bug" * https://bugs.python.org/issue30461 closed as "rejected" * https://bugs.python.org/issue33275 fixed by documenting the behavior
Closing as NOTABUG according to comment 3 and comment 5.