I dont know who needs to hear this bit qBittorrent has a nasty vulnerability ( and there are some older ones too)

qBittorrent, on all platforms, did not verify any SSL certificates in its DownloadManager class from 2010 until October 2024. If it failed to verify a cert, it simply logged an error and proceeded.

To be exploitable, this bug requires either MITM access or DNS spoofing attacks, but under those conditions (seen regularly in some countries), impacts are severe.

The primary impact is single-click RCE for Windows builds from 2015 onward, when prompted to update python the exe is downloaded from a hardcoded URL, executed, and then deleted afterwards.

The secondary impact for all platforms is the update RSS feed can be poisoned with malicious update URLs which the user will open in their browser if they accept the prompt to update. This is browser hijacking and arbitrary exe delivery to a user who would likely trust whatever URL this software sent them to.

The tertiary impact is this means that an older CVE (CVE-2019-13640 https://www.cvedetails.com/cve/CVE-2019-13640/) which allowed remote command execution via shell metacharacters could have been exploited by (government) attackers conducting either MITM or DNS spoofing attacks at the time, instead of only by the author of the feed.

Full write up is here: https://sharpsec.run/rce-vulnerability-in-qbittorrent/

  • AmbiguousProps@lemmy.todayEnglish
    53·
    16 days ago

    If you are running Windows […]

    and

    from June 2015 until the present, affecting v3.2.1 through v5.0.0 inclusive. The behaviour does not appear to be replicated for other OS variants

  • Supermariofan67@programming.devEnglish
    13·
    16 days ago

    Perfect example of a (part of a) security vulnerability being fixed in a commit that doesn’t immediately seem security related and would never be back ported to a stablestale distro

    The code which parses the binary MaxMind database after decompression is well guarded as of 2024 but used to look different, potentially providing more attack surface. There is also an interesting commit where a contributor makes adjustments to the gzip::decompress() function which hints at a stack overflow, as the destination buffer was changed from static allocation on the stack to dynamic allocation on the heap, though it was not exploitable due to checks before it is written to

  • JCPhoenix@beehaw.orgEnglish
    81·
    16 days ago

    Mitigations

    Just use another torrent client. Deluge and Transmission etc do not have this vulnerability.

    Was wondering what the takeaway is here. I updated to 5.0.1. Does that fix all these? If not, guess I’ll try a different torrent client.

    • Majestic@lemmy.mlEnglish
      16·
      16 days ago

      Not really. This requires a sophisticated attacker. I’d suggest updating soon but I doubt most people are at risk. As always verify downloads before running them and check where you should be getting updates and if you’re sent to a sketchy file host try to find updates from the official website instead.

      The biggest risk is MTM interception and replacement of the python executable if you try and use the search tool for the first time. I suppose avoid doing that until you update the client from their website.

  • rolandtb303@lemmy.mlEnglish
    73·
    15 days ago

    i kinda wish that some people who post security vulns stop being so vague and cherry-picky sometimes, like you could have written in your post that it only affects windows and it affects only a certain range of versions of the program. would have clearned things up better imo. interesting to know, though.

    • ShortN0te@lemmy.mlEnglish
      62·
      15 days ago

      You should have read the post more carefully. The CVE affects every OS. Just the first shown example is Windows only.

      Also, the relevant commits are outlined in the first paragraph. This article is not for the stupid user it’s a technical analysis on a few ways to exploit it and for those cases the commits are more relevant than the version. Also saying which versions are affected is not that easy, commits can be backported into an older version by for example the packager.