Debian Patches
Status for popt/1.19+dfsg-3
| Patch | Description | Author | Forwarded | Bugs | Origin | Last update |
|---|---|---|---|---|---|---|
| fix-reporting-of-test-failures.patch | tests: Fix reporting of test failures After commit a4afb898f7758160acda71d774c7d98d528da273 the testsuite no longer reported test failures. This is because we rely on the exit status code of the script, and currently, this is always 0. If any test fails, we will exit with status code 2. The entire testsuite will continue to run, even after a test has failed. |
Håvard F. Aasen <havard.f.aasen@pfft.no> | yes | 2022-09-27 | ||
| document-flags-used-in-poptGetContext.patch | man-page: Add information for flags used in poptGetContext() | Håvard F. Aasen <havard.f.aasen@pfft.no> | yes | debian | 2023-08-02 | |
| instruct-man-to-run-the-tbl-preprocessor.patch | Instruct man(1) to run the 'tbl' preprocessor This is needed to get set the macro 'TW' with the table width. This was a new warning in groff 1.23.0 Can be tested with the command: LC_ALL=C.UTF-8 MANROFFSEQ='' MANWIDTH=80 \ man --warnings -E UTF-8 -l -Tutf8 -Z popt.3 >/dev/null |
Håvard F. Aasen <havard.f.aasen@pfft.no> | yes | 2024-08-18 | ||
| Fix-CVE-2026-18743-buffer-overflow-s-in-poptConfigFileToS.patch | Fix CVE-2026-18743: buffer overflow(s) in poptConfigFileToString() Fix the reallocation logic thinko, repeated twice in the function: if we need to reallocate, the current maximum value is not interesting in any way. Just allocate double the required size instead. The reproducer test is obviously closely tied to the initial buffer size of 480 and wouldn't work with other sizes, but with the current default this reliably crashes popt under ASAN prior to this fix. Also , the reproducer only exercises the first of these flaws. (cherry picked from commit ea2092edba8d6c1a9acd60d9fd7a587c624d6ffe) |
Panu Matilainen <pmatilai@redhat.com> | no | 2026-08-11 | ||
| Fix-CVE-2026-18739-off-by-one-error-in-poptStuffArgs.patch | Fix CVE-2026-18739: off-by-one error in poptStuffArgs() The poptStuffArgs function only checks whether (con->os - con->optionStack) is equal to POPT_OPTION_DEPTH (10) before incrementing con->os, and does not increment the value by 1 to perform boundary pre-checking, as handleAlias does. Add a new test program as a reproducer for this case. Backported from commit 14c42b415ba0c11640f7d4ee80920452d0287058 |
Yao Zhang <294772273@qq.com> | no | 2026-08-05 | ||
| Bring-in-our-own-allocating-printf-variants.patch | Bring in our own allocating printf() variants These are brought mostly as-is from rpm under a new POPT_ name, just with a slight change return an error on memory alloc failure whereas rpm exits on error. As the author of rvasprintf() and rasprintf() there, I'm at liberty to (re)license them under MIT as well. No functional changes here, just needed for sanity going forward. Backported from commit 4f5a070cd98aade2767677de22661f4854d4a195, 1.19.x does not build as C99 so we can't mix code and variable declarations as in master. |
Panu Matilainen <pmatilai@redhat.com> | no | 2026-08-20 | ||
| Fix-a-bucketful-of-buffer-overflows-in-popt-help-printing.patch | Fix a bucketful of buffer overflows in popt --help printing singleOptionDefaultValue() based the default value string output buffer size on the available line length, without any sort of bounds checking for most data. This was trivial to overflow with longer option names. There was also this baffling ellipsizing of long string arguments, done done based on the rather arbitrary buffer size instead of screen limits. Ellipsizing of long values might be totally reasonable but that belongs to the display code that does all manner of formatting too. Just use our newly introduced allocating sprintf() to get it right for good. Drop the line length argument entirely, it was only used for a bad estimation of the needed buffer size. At the other end, make sure the indent length stays below available columns to avoid the underflow from help printing. Annoyingly we need multiple different test programs for this because the various crashes are quite sensitive to the max length of a single option in the table. There are more broken cases than these, but we can't test everything and these cover the CVE-2026-18839 case. Update the ellipsis test a bit because it no longer makes sense. Backported from commit af6f19e5968d48608a062d6d92d28b7fd248f8ea. |
Panu Matilainen <pmatilai@redhat.com> | no | 2026-08-20 |
All known versions for source package 'popt'
- 1.19+dfsg-3 (forky, sid)
- 1.19+dfsg-2 (trixie)
- 1.19+dfsg-1 (bookworm)