Debian Patches
Status for minidlna/1.3.3+dfsg-2
| Patch | Description | Author | Forwarded | Bugs | Origin | Last update |
|---|---|---|---|---|---|---|
| 01-run-instead-of-var-run.patch | Use /run instead of /var/run to store runtime files | Benoît Knecht <benoit.knecht@fsfe.org> | no | 2015-03-01 | ||
| 03-make-sure-the-database-is-closed-after-scanning.patch | Make sure the database is closed after scanning | Benoît Knecht <benoit.knecht@fsfe.org> | no | 2015-03-01 | ||
| 07-fix-multi-artist-album-handling.patch | Ignore artist when looking up existing container This prevents creation of multiple album containers for compilation albums (i.e. one separate album for each artist) when adding files via inotify, which causes the scanner cache to be invalidated all the time. OTOH, different albums with identical names ("Greatest Hits", "Love Songs") are now listed as just a single album in the "Album" container. This might be solved using some heuristic based on music file location (same directory / different directories). The new behaviour is usually the same as encountered when doing a full rescan, because there's a scanner cache that is used during full rescans and that cache ignores artist names as well. |
Florian Will <florian.will@gmail.com> | yes | debian | 2015-03-09 | |
| 08-Fix-testupnpdescgen-build.patch | Fix testupnpdescgen build. | Alexander Gerasiov <a@gerasiov.net> | no | 2020-12-05 | ||
| 10-do-not-close-socket-on-sighup.patch | do-not-close-socket-on-sighup | Alexander GQ Gerasiov <gq@debian.org> | no | 2021-01-31 | ||
| 13-spelling-and-typos.patch | spelling and typos | "Barak A. Pearlmutter" <barak+git@pearlmutter.net> | no | 2022-11-13 | ||
| 14-autoupdate.patch | autoupdate | "Barak A. Pearlmutter" <barak+git@pearlmutter.net> | no | 2022-11-27 | ||
| 15-thumbnails.patch | thumbnails This patch is from https://sourceforge.net/p/minidlna/patches/_discuss/thread/28a9f79d/9aca/attachment/minidlna-1.3.3-thumb.patch by https://sourceforge.net/u/wilderman/profile/ with changes to derived and cache files (configure and such) removed. |
"Barak A. Pearlmutter" <barak+git@pearlmutter.net> | no | 2023-08-21 | ||
| 16-Add-compatibility-with-FFMPEG-7.0.patch | Add compatibility with FFMPEG 7.0 From 243cf2dd27ebcaf4ef1093c79b96077378d52018 Mon Sep 17 00:00:00 2001 channel_layout has been replaced with ch_layout |
Robert-André Mauchin <zebob.m@gmail.com> | no | debian | other, https://sourceforge.net/u/eclipseo/minidlna/ci/243cf2dd27ebcaf4ef1093c79b96077378d52018/ | 2024-05-04 |
| 17-Replace-rm-rf-with-in-process-deletion-code.patch | Replace 'rm -rf' with in-process deletion code. This fixes two edge-case bugs in the code that cleans up minidlna's previous cache directory. Firstly, the 'rm -rf' command wasn't properly quoted, so if a user had configured a more interesting cache directory than the typical /var/cache/minidlna, the rm might have unwanted effects. Secondly, at the time of the call in check_db(), a SIGCHLD handler has been installed which waits for any available process, and also multiple threads are running, meaning that if race conditions work out wrongly, the other thread can reap system()'s subprocess before system() does, causing the wait in glibc's implementation of system() to get ECHILD and system() to report failure. |
Simon Tatham <anakin@pobox.com> | no | 2024-09-15 | ||
| 18-CVE-2026-29508-dff-parser-bounds.patch | Fix memory safety issues in the DFF parser (CVE-2026-29508) _get_dfffileinfo() trusted chunk sizes taken from the file: - the PROP and DIIN chunks were read into variable-length stack arrays of a size taken from the file, without any bound (CVE-2026-29508 for PROP; DIIN has the same problem); - sub-chunks of PROP and DIIN were parsed without checking that they fit in the parent chunk, leading to out-of-bounds reads; - DIAR/DITI sizes below 4 underflowed into huge stack arrays; - a zero DST frame rate caused a division by zero (SIGFPE); - a chunk size that does not fit in off_t made fseeko() seek backwards, so the parser looped forever on the same chunk. Read PROP and DIIN into bounded heap buffers, check every sub-chunk against its parent, guard the divisions and refuse backward seeks. . Also drop the (char *) casts in the FRTE parsing: they caused sign extension and produced bogus durations for DST files. |
Alexander GQ Gerasiov <gq@debian.org> | no | debian | 2026-11-08 | |
| 19-upnpevents-unsubscribe-while-connecting.patch | Do not abort when a subscriber goes away while connecting If a client unsubscribes while the initial event notification is still connecting to its callback URL, upnpevents_removeSubscriber() clears notify->sub, and upnp_event_prepare() then hits assert(obj->sub) and aborts the whole daemon. Any host on the LAN can trigger this with a SUBSCRIBE to an unresponsive callback followed by an UNSUBSCRIBE. Drop such notifications instead. |
Alexander GQ Gerasiov <gq@debian.org> | no | debian | 2026-11-08 | |
| 20-select-del-after-fini.patch | Make select_del() safe after select_fini() and on double delete A child forked by process_fork() calls event_module.fini(), which frees the events table but leaves nevents untouched. When the child later closes its HTTP socket, select_del() dereferences the freed (NULL) table and the child crashes with SIGSEGV whenever the connection is not the last registered event. A second select_del() on the same event would also write to events[-1]. Reset nevents in select_fini() and make select_del() a no-op for events that are not registered. |
Alexander GQ Gerasiov <gq@debian.org> | no | debian | 2026-11-08 | |
| 21-fix-ftbfs-on-hurd.patch | Fix FTBFS on GNU/Hurd GNU/Hurd defines neither PATH_MAX nor MAXPATHLEN. Provide fallbacks in config.h, so every source file gets them. . getifaddr.c treated every non-Linux system with getifaddrs() as BSD and included <net/if_dl.h>, which Hurd does not have. Hurd supports SIOCGIFHWADDR and struct ifreq.ifr_hwaddr, so use the Linux code path. |
Alexander GQ Gerasiov <gq@debian.org> | no | 2026-11-08 | ||
| 22-dsf-parser-bounds.patch | Fix memory safety issues in the DSF parser The DSF reader, like the DFF one (CVE-2026-29508), trusted values taken from the file: - the sampling frequency was used as a divisor without checking it for zero (SIGFPE); - the metadata chunk was malloc()ed and memset() with a size computed from the header, with no upper bound, so a crafted header could make the scanner allocate and touch gigabytes of memory; - GET_DSF_INT32 shifted uint8_t values promoted to int by 24 bits, which is undefined behaviour for bytes >= 0x80. Bound the metadata chunk by the real file size and by 64 MiB, guard the division, do the shifts on uint32_t and check the cover art malloc(). |
Alexander GQ Gerasiov <gq@debian.org> | no | 2026-11-08 | ||
| auto-gitignore | Update .gitignore from Debian packaging branch The Debian packaging git branch contains these updates to the upstream .gitignore file(s). This patch is autogenerated, to provide these updates to users of the official Debian archive view of the package. [dgit (17.0) update-gitignore] |
no |
All known versions for source package 'minidlna'
- 1.3.3+dfsg-2 (sid, forky)
- 1.3.3+dfsg-1.1 (trixie)
- 1.3.0+dfsg-2.2+deb12u1 (bookworm-security, bookworm)
