From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-io1-xd33.google.com (mail-io1-xd33.google.com [IPv6:2607:f8b0:4864:20::d33]) by sourceware.org (Postfix) with ESMTPS id 7F1EE3858D34 for ; Mon, 1 Apr 2024 13:23:15 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org 7F1EE3858D34 Authentication-Results: sourceware.org; dmarc=pass (p=none dis=none) header.from=gmail.com Authentication-Results: sourceware.org; spf=pass smtp.mailfrom=gmail.com ARC-Filter: OpenARC Filter v1.0.0 sourceware.org 7F1EE3858D34 Authentication-Results: server2.sourceware.org; arc=none smtp.remote-ip=2607:f8b0:4864:20::d33 ARC-Seal: i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1711977798; cv=none; b=fN25jM8tuCU0X2p+QLz3f1WquqRhIO96rturqha32oMmUFpwj1SlOU4mv8PevX3/KqYtXnx+QQ2FGKhqimBwNhGGFrZv6qR4t5eTx7RoSNxoONdwFRh0mHfVo9+ma/8sKhhC1XzcA7vV2oYoRA4uk/DV7WFFGdW4Zw6Fg3xm2Ok= ARC-Message-Signature: i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1711977798; c=relaxed/simple; bh=2qocX2qVjuuEg5r7O9/hY9yCaKkJjwFXYP5J7eS+PJY=; h=DKIM-Signature:MIME-Version:From:Date:Message-ID:Subject:To; b=KtZS32c/7h+OlOkMlV5j876MD/Xd7p+Y8ZRdaO+Vdmork3z0TSehsUWnJY290NpD3erk5ToQ/b4nJIfnRDPt87voojsmH3XGQrHlZmYV5Xwc4r6c0nL4Em61skDVaKt2lqsdw9oBzfxdi6ocnZaCbVyilpvo0QGbYAWHXv2k0Qc= ARC-Authentication-Results: i=1; server2.sourceware.org Received: by mail-io1-xd33.google.com with SMTP id ca18e2360f4ac-7d0d0057c8aso18162039f.0 for ; Mon, 01 Apr 2024 06:23:15 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20230601; t=1711977794; x=1712582594; darn=sourceware.org; h=content-transfer-encoding:cc:to:subject:message-id:date:from :in-reply-to:references:mime-version:from:to:cc:subject:date :message-id:reply-to; bh=2qocX2qVjuuEg5r7O9/hY9yCaKkJjwFXYP5J7eS+PJY=; b=Mgbb9Gb2EsbntEo90PmLgMSDlT9kHzgxgMqZPedTzY88x+z5+iJcsL4Wkgcxfws0i0 aSfo06tNYHx0vFQuipUPiyOgqi3goTK09tMqdANk+Y1bdt0HRcVWrG634ELa9qNdihSt jXlK/xwj6c5yAPzoedH5uKgQ67BhvGeajsGcTAtpXBT4WSyVJDGEhitMKWCkbJhm4FHu lMnCn2giu9BVBgqI/kJuu6xqFo1WaJBP7LEx973YFGde4/E9URTA1xS32x15EyNhtU1/ C5TzVJeeQVUI6LYIAmETaAXVi2TRvVq72TJxm2rhb7XPXNtrmI7IOxoHWHKF0GyaU3OM XR5g== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1711977794; x=1712582594; h=content-transfer-encoding:cc:to:subject:message-id:date:from :in-reply-to:references:mime-version:x-gm-message-state:from:to:cc :subject:date:message-id:reply-to; bh=2qocX2qVjuuEg5r7O9/hY9yCaKkJjwFXYP5J7eS+PJY=; b=waV/SE0fD7I8HwAytWv+IG7PygXufFyDyzv+3Lqtah+15KVqkoz8+tYBoZ6FOasjmo nhwpreKGdCj7ZSFz1AoD3A72ceVZmjBu/vsPaqK3fbkQMXC99M+Gco6DcAFPcTHnZ4e6 qvygCC6yja42IOIMA0KBMt9vra4D86HsQ8odSm9bLSMGe9NoirRazMUWq9q4DIh/4kg0 NlqC2IsKN2CdeZGUNwMZGzgBPStptSra69PZJe7/z7GEdpgmSt7OxGzI/pYW25A5K9BS LK9ie5zIp5c3P3M0UDshgc4HbuyBn6Tn3UpP+hEms7nnsS+YgY9sBRfVOFey1SuStnnc H8LA== X-Forwarded-Encrypted: i=1; AJvYcCXXDLHRd2o88TrqgMJbD3qOBvZG0NnvWndkI2FR6HzlMTncYfE4DdTd0C5WZTS9KTDJrRxseaKfwWF1bvXJrRnfOA3/cHy8Pw== X-Gm-Message-State: AOJu0YwQ7TeP138hCiWtmTVSDjuUuJxd9cLTa4yr+ym8fo/ql+Rv0w5/ dVQHuXt/ieW/yviEPioZCX6ScfV02IUeUFEYdRpTQg+3zu2WddKY6Sxjy2QC3Oj4MAh0LO5JfGA BrHBhV82Bi363ZA6MPGzfi4Zf7R+xDAOymASZZg== X-Google-Smtp-Source: AGHT+IGAPthRYhizFSdND0V8ms4LXHHohjxRT3pmEVY2A8zaI1PRK3yjCHgUP97xM6/dSpW81ySDxWO4oT/TvrfRAHg= X-Received: by 2002:a5e:8614:0:b0:7d0:6bf4:bfe5 with SMTP id z20-20020a5e8614000000b007d06bf4bfe5mr10419333ioj.10.1711977793706; Mon, 01 Apr 2024 06:23:13 -0700 (PDT) MIME-Version: 1.0 References: <877chhe80o.fsf@gentoo.org> In-Reply-To: From: Rui Ueyama Date: Mon, 1 Apr 2024 22:23:02 +0900 Message-ID: Subject: Re: Remove dependency on libjansson To: Luca Boccassi Cc: Sam James , Fangrui Song , Binutils Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable X-Spam-Status: No, score=-0.6 required=5.0 tests=BAYES_00,DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,DKIM_VALID_EF,FREEMAIL_ENVFROM_END_DIGIT,FREEMAIL_FROM,RCVD_IN_DNSWL_NONE,SPF_HELO_NONE,SPF_PASS,TXREP autolearn=ham autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on server2.sourceware.org List-Id: On Mon, Apr 1, 2024 at 9:16=E2=80=AFPM Luca Boccassi wro= te: > > On Mon, 1 Apr 2024 at 12:44, Rui Ueyama wrote: > > > > On Mon, Apr 1, 2024 at 7:38=E2=80=AFPM Luca Boccassi = wrote: > > > > > > On Mon, 1 Apr 2024 at 10:50, Sam James wrote: > > > > > > > > Fangrui Song writes: > > > > > > > > > On Sun, Mar 31, 2024 at 8:31=E2=80=AFPM Rui Ueyama wrote: > > > > >> > > > > >> Hi, > > > > >> > > > > >> The recent xz incident demonstrated that supply chain attacks ar= e a > > > > >> real threat, and dependence on third-party libraries can have > > > > >> significant consequences. > > > > >> > > > > >> In the wake of the incident, I propose we remove the dependency = on > > > > >> libjansson from GNU ld. > > > > >> > > > > >> First of all, why does GNU ld depend on libjansson which is a JS= ON > > > > >> parsing library? GNU ld gained the `--package-metadata` option i= n May > > > > >> 2022 to embed a JSON string into a .note section for package > > > > >> management for Fedora and other Linux distributions. At the same= time, > > > > >> the dependency on libjansson, a library for parsing JSON-format > > > > >> strings, was introduced to validate an argument for that option.= If an > > > > >> argument is not a valid JSON string, ld reports an error. If the > > > > >> library is unavailable, or if `--disable-jansson` was passed to = the > > > > >> configure script, the library will not be linked and the error c= heck > > > > >> will be disabled. By default, the library will be linked if it e= xists. > > > > >> > > > > >> I opposed adding an extra dependency to GNU ld just for string > > > > >> verification purposes because it didn't seem worth adding extra > > > > >> dependency to the linker. LLVM lld and the mold linker also supp= ort > > > > >> the option, but they do not verify if the argument is a valid JS= ON > > > > >> string -- they simply treat it as an opaque string. If libjansso= n is > > > > >> unavailable, even GNU ld doesn't verify arguments. Therefore, th= e > > > > >> verification is not trustworthy, and the reader must be prepared= for a > > > > >> malformed JSON string when reading a .note section. Moreover, > > > > >> verifying a string is straightforward without the feature; you c= an > > > > >> simply `echo` the string to pipe it to `jq` for verification bef= ore > > > > >> passing it to GNU ld. > > > > >> > > > > >> I just checked /usr/bin/ld on Ubuntu 24.04, which is set to be > > > > >> released this month, and the dependency on libjansson was indeed > > > > >> present. > > > > >> > > > > >> How much risk does it pose? Probably not much, as long as the li= brary > > > > >> is maintained properly. However, the stakes are high; if someone= takes > > > > >> control of the library and introduces malicious code, they could > > > > >> execute a Ken Thompson-style supply chain attack. Since GNU ld i= s used > > > > >> to build essentially everything, the attacker could in theory ga= in the > > > > >> power to not just contaminate a specific program such as openssh= , but > > > > >> every executable in an official Linux distribution image. I thin= k the > > > > >> risk is not worth taking. I believe we just should remove the st= ring > > > > >> verification code and the dependency on the library from GNU ld. > > > > >> > > > > >> Rui Ueyama > > > > > > > > > > Thanks for bringing this up again. I support removing the json de= pendency. > > > > > > > > > > I lightly expressed my concern > > > > > https://sourceware.org/pipermail/binutils/2022-May/120846.html an= d > > > > > there might be others unsure about the dependency as well. > > > > > > > > I'd like to hear bluca's take before making up my mind. Note that i= t's > > > > also automagic right now IIRC (enabled if installed, not opt-in). > > > > > > > > But my take on it so far is that it doesn't sound worth it. > > > > > > $ apt-cache rdepends libjansson4 | wc -l > > > 310 > > > > > > Sorry, but this looks to me like a knee-jerk reaction that misses the > > > wood for the trees. The xz issue is that an extremely complex and > > > sophisticated social engineering attack was planned and executed over > > > multiple years. If it hadn't been on xz, it would have been on anothe= r > > > project. Dropping dependencies left and right without thinking will > > > not stop the next attack, as there is no shortage of overworked > > > maintainers to try and abuse. In fact, avoiding common libraries and > > > everyone reimplementing the wheel _increases_ the attack > > > opportunities. > > > > > > The reason for this integration is that finding issues at build time > > > is cheaper than finding them at runtime or release time. We rely on > > > this in Fedora, Debian and Ubuntu to ensure valid data is written (in > > > all packages in the former, in an increasing set of opt-ins in the > > > latters) - in fact, validating input _before_ shipping binaries > > > _improves_ security, rather than decreasing it, given the payload is > > > parsed by a privileged component. > > > > > > So if it was removed upstream, we'd just have to patch it back in > > > downstream in each interested distro - which if you recall, was > > > exactly how sshd was exploited, due to the lack of an upstream common > > > implementation. This is opt-in, so distributions that are not > > > interested can just avoid enabling it, with no extra effort required. > > > So please, leave it as-is. Thank you. > > > > I'd like to say that that's not an instinctive reaction to the xz > > incident, as I expressed a concern when the dependency was being > > introduced. I also want to say that I'm not trying to drop > > dependencies left and right without thinking. > > Well, the email did start with: > > > The recent xz incident demonstrated that supply chain attacks are a rea= l threat, and dependence on third-party libraries can have significant cons= equences. > > so it should be understandable why that's the impression it gave me. > This feature has been present in binutils for 2 years now, and AFAIK > caused zero issues so far (probably going to jinx it with this > comment, but still). > > > I understand that the error check would be sometimes useful, but I'd > > like to propose reevaluating its cost and benefit. IIUC, the > > --package-metadata option is, as its name suggests, used by packaging > > tools such as dpkg-dev or rpmbuild. My understanding is that this > > option is not intended to be used directly by developers at large, but > > rather, the JSON string is generated by a very limited set of > > packaging tools from package description files and passed to the > > linker. If that's the case, the string format check itself is an extra > > measure to catch rare string construction bugs, given that JSON > > strings are machine-generated and not manually crafted. > > The metadata comes from each package, so it is very much possible for > errors to appear in one build but not another. And they would be hard > to spot and expensive to fix, as they'd likely only be found when > things go really wrong and the data is needed at runtime, or worse. Correct me if I'm wrong, but I believe the main concern in creating a malformed JSON string is forgetting to escape certain characters when embedding it in a JSON string. Correctly handling it shouldn't be too difficult. I understand that it is sometimes hard to notice an error in the .note section as it's just embedded data sitting there, and errors in the section would often be noticed exactly in the situation when you really need it. But if you are very cautious about the sanity of the .note section, why would you stop here and only focus on catching JSON format errors? If you do not trust your own tool and aim to catch bugs before its generated data goes into production, you should read the .note section back and compare the JSON contents with its package metadata file. There are tons of different ways for the tool to semantically go wrong such as missing field or something like that, and just catching JSON format errors (or, maybe just string escaping bugs) seems insufficient to me. And if you think that that kind of error would be unlikely, then I wonder why you are worried about the JSON format error. > > While we may still do extra checks for catching possible tooling bugs, > > I wonder how difficult it would be to assign this additional > > verification to jq or a similar tool instead of the linker. IIRC, dpkg > > comes with a tool to scan all executables in a directory to find > > packaging errors. Isn't it a better place to run the extra format > > check? Would it be too complex to output the .note section contents > > after linking and pipe them to jq to confirm the embedded string is a > > valid JSON string? > > There's no such thing in dpkg. You might be thinking of Lintian, but > that's an entirely separate tool, that is optional, runs long after > the package is published and distributed, and is only seldomly checked > (most maintainers just ignore it as it's informational only), and > doesn't stop bad builds from being generated and released. It's also > package-format-specific, so again every distribution would have to > reimplement exactly the same check in different ways, which is just > duplication of work for no gain. The intention here was to have bad > input fail the build immediately, with no extra effort required from > distributions or packagers, and it works very well for that. > > This is akin to suggesting that compilers shoudn't output warnings, > but the job should just be left to static analyzers. The earlier > problems are detected, the cheaper they are to fix, and the less > impact they have. This is one of the few truisms of engineering. If a --package-metadata argument is machine-generated, a better analogy would be the relationship between the compiler and the linker, rather than the human programmer and the compiler. The linker sometimes verifies object file contents, but it doesn't disassemble the .text section to catch rare assembler bugs.