devel

devel@lists.centos.org

9 participants
4635 discussions

Bringing glibc 2.34 snapshots to Fedora 35 and CentOS Stream 9
by Florian Weimer 07 Jul '21

07 Jul '21

TL;DR: glibc 2.34 snapshots are coming. libpthread as a separate library creates problems, so we are removing it. There may be some hickups. Full updates with “dnf update” are recommended. We expect that we will soon be able to import glibc upstream snapshots regularly into Fedora 35 and CentOS Stream 9. We did such regular imports for the past couple of Fedora rawhide releases, but the situation will be slightly different, as explained below. The snapshots will also land in CentOS Stream 9, after a delay as the result of a different testing pipeline. glibc 2.33 and earlier accidentally provided an very generic ROP gadget in the statically linked startup code that is present in every program (even dynamically linked ones). We have removed the ROP gadget and moved that particular initialization code into the dynamically linked part, but that means that the interface between the startup code and the dynamically loaded glibc parts had to change. As a result, any program linked against glibc 2.34 will not run with glibc 2.33 or any earlier version. (Some of you may remember the memcpy(a)GLIBC_2.14 issue, it was similar.) This version requirement will be properly reflected in RPM dependencies. glibc 2.34 removes libpthread as a separate library. This is based on the observation that most processes load libpthread anyway. (On this system, I count 141 out of 159 processes that load libpthread.) One particularly thorny issue is that certain NSS modules depend on libpthread, and if the main program is not linked (indirectly) against libpthread, loading such NSS modules effectively loads libpthread via dlopen, which is something that glibc has never supported well. Furthermore, the availability of some pthread_* functions without -lpthread has been a source of confusion to developers, resulting in both overlinking and underlinking. We started the libpthread transition in glibc 2.33 when we added the __libc_single_threaded variable as a replacement for the weak symbol hacks that some libraries use to detect single-threaded processes. (For example, libstdc++ used to do this to avoid using atomic instructions in std::shared_ptr.) Backwards compatibility for dynamically linked binaries should be preserved (as usual), but we know of one issue on ppc64le, where the weak symbol hacks resulted in slightly corrupt binaries. Upstream glibc did not accept the proposed backwards compatibility enhancement so far. Instead, we will rebuild affected distribution binaries with a binutils version which handles weak symbols slightly differently, avoiding the corruption. As we plan to perform these rebuilds before the new glibc lands in the buildroot, the requirement to upgrade to the new binaries before or at the same time of the upgrade to the glibc upstream snapshot will NOT be reflected in RPM dependencies. A full upgrade using “dnf update” will work, however. In addition to the removal of libpthread, I also hope to remove libdl and librt. This means that new GLIBC_2.34 symbols will be added to libc.so.6 (e.g., timer_create(a)GLIBC_2.34) The librt removal in particular will probably not land with the first imported snapshot. This is unfortunate because RPM does not have a way to express this: there's just a blanket Provides: libc.so.6(GLIBC_2.34), which will already be included in the first snapshot that adds the symbol __lib_start_main@@GLIBC_2.34 (whether or not the RPM package includes timer_create(a)GLIBC_2.34 as well). Some tools in the fedpkg/centpkg/mock sphere try to install builds from the Koji buildroot into installations from composes, without upgrading everything to the current buildroot, so it's possible that glibc won't be upgraded to match the requirements of RPMs directly imported from the buildroot. Developers encountered similar issues with glibc snapshot imports (e.g., around the symbol pthread_getattr_np). Our RPM infrastructure does not have per-symbol dependencies, so there isn't much we can do about it at the packaging level. It's a transitory issue during rawhide/CentOS Stream 9 development; the finished release will add all GLIBC_2.34 symbols in one upgrade, so end users won't see it in a Fedora 34 to Fedora 35 upgrade or a CentOS Stream 8 to CentOS Stream 9 upgrade. We think there is value in providing access to these snapshots early, and will try to make the transitions as smooth as possible, within the constraints outlined above. Thanks, Florian

1 1

Case study: libxml2 CVE fixes in the Red Hat ecosystem
by Carl George 02 Jul '21

02 Jul '21

Since the announcement of the CentOS Linux 8 EOL date in December, I've noticed there has been confusion about how CVE fixes work in CentOS Stream 8. Recently there have been some CVE fixes for the libxml2 package, and I took the opportunity to examine how these fixes flowed through the Red Hat ecosystem. Here are the relevant CVE identifiers and their CVSS v3 scores. CVE-2021-3516 - 7.8 [0] CVE-2021-3517 - 8.6 [1] CVE-2021-3518 - 8.6 [2] CVE-2021-3537 - 7.5 [3] CVE-2021-3541 - 6.5 [4] Here is a timeline of notable events. 2021-05-07 * libxml2-2.9.10-12.fc34 [5] was submitted for Fedora 34, which included fixes for the first four CVEs. 2021-05-10 * libxml2-2.9.10-12.fc34 [5] was released for Fedora 34, fixing the first four CVEs. 2021-05-13 * All five CVE fixes were released upstream as part of version 2.9.11 [6]. * libxml2-2.9.12-1.fc34 [7] was submitted for Fedora 34, which included the fix for the fifth CVE. 2021-05-21 * libxml2-2.9.12-2.fc34 [7] was submitted for Fedora 34, which fixed an unrelated upstream regression [8], but also reset the pending update. * libxml2-2.9.7-11.el8 [9] was released for CentOS Stream 8, fixing all five CVEs. This was a backport update that was unaffected by the upstream regression in 2.9.12. 2021-05-24 * libxml2-2.9.12-2.fc34 [7] was released for Fedora 34, fixing the fifth CVE. 2021-06-29 * libxml2-2.9.7-9.el8_4.2 [10] was released for RHEL 8, fixing all five CVEs. Later that day it was rebuilt [11] and released for CentOS Linux 8. This was a backport update that was unaffected by the upstream regression in 2.9.12. It's important to note that these CVE fixes were not part of a security embargo [12]. That is why CentOS Stream 8 was able to provide them before RHEL 8. If these fixes had been part of an embargo, they would have been released for RHEL 8 first (once the embargo was lifted), then CentOS Stream 8 and CentOS Linux 8 immediately after. Another thing I want to point out is that libxml2-2.9.7-11.el8 and libxml2-2.9.7-9.el8_4.2 are effectively identical. They contain the exact same backported CVE fixes. The only difference is the release field. This can be verified in the exported SRPM commits [13][14]. No additional changes were made to the package source between being released for CentOS Stream 8 and being released for RHEL 8. I hope you enjoyed this deep dive into the lifecycle of these CVE fixes. The key takeaway is that CentOS Stream 8 does get security fixes, and usually gets them much sooner than CentOS Linux 8. [0] https://access.redhat.com/security/cve/CVE-2021-3516 [1] https://access.redhat.com/security/cve/CVE-2021-3517 [2] https://access.redhat.com/security/cve/CVE-2021-3518 [3] https://access.redhat.com/security/cve/CVE-2021-3537 [4] https://access.redhat.com/security/cve/CVE-2021-3541 [5] https://bodhi.fedoraproject.org/updates/FEDORA-2021-e3ed1ba38b [6] https://mail.gnome.org/archives/xml/2021-May/msg00000.html [7] https://bodhi.fedoraproject.org/updates/FEDORA-2021-e8b7e177a4 [8] https://gitlab.gnome.org/GNOME/libxml2/-/issues/255 [9] https://koji.mbox.centos.org/koji/buildinfo?buildID=17568 [10] https://access.redhat.com/errata/RHSA-2021:2569 [11] https://koji.mbox.centos.org/koji/buildinfo?buildID=18244 [12] https://www.redhat.com/en/blog/security-embargoes-red-hat [13] https://git.centos.org/rpms/libxml2/c/6ce3da4b1430e975a40a538aa250775e101e5… [14] https://git.centos.org/rpms/libxml2/c/bc5a009a460cda9e2392f75fff8bf6edae43e… -- Carl George

2 1

CentOS Hyperscale SIG Quarterly Report for 2021Q2
by Davide Cavalca 30 Jun '21

30 Jun '21

Hello, I've published our quarterly update for the Hyperscale SIG activities on the CentOS Blog: https://blog.centos.org/2021/06/centos-hyperscale-sig-quarterly-report-for-… Cheers Davide

1 0

[Storage-SIG] missing centos-release-ceph-pacific in CentOS Stream
by Sandro Bonazzola 30 Jun '21

30 Jun '21

Hi, looking at http://mirror.centos.org/centos/8-stream/extras/x86_64/os/Packages/ I can see only: [image: [ ]] centos-release-ceph-nautilus-1.2-2.el8.noarch.rpm <http://mirror.centos.org/centos/8-stream/extras/x86_64/os/Packages/centos-r…> 2020-05-18 19:53 8.8K [image: [ ]] centos-release-ceph-octopus-1.0-1.el8.noarch.rpm <http://mirror.centos.org/centos/8-stream/extras/x86_64/os/Packages/centos-r…> 2020-05-18 19:53 8.8K but in http://mirror.centos.org/centos/8-stream/storage/x86_64/ there's only Ceph Pacific. Looks like https://koji.mbox.centos.org/koji/buildinfo?buildID=16908 is missing there. Installing it by hand: yum install https://koji.mbox.centos.org/pkgs/packages/centos-release-ceph-pacific/1.0/… I see repo is using $release instead of $stream so looks like it will pick up the wrong repo. -- Sandro Bonazzola MANAGER, SOFTWARE ENGINEERING, EMEA R&D RHV Red Hat EMEA <https://www.redhat.com/> sbonazzo(a)redhat.com <https://www.redhat.com/> *Red Hat respects your work life balance. Therefore there is no need to answer this email out of your office hours.*

2 3

CBS/koji new (opt-in) feature : enabling EPEL repositories as external repos
by Fabian Arrotin 30 Jun '21

30 Jun '21

Based on previous threads on centos-devel and some requests (including from Hyperscale SIG), we had a look at the best way to let some SIG build *against* EPEL for some needed Requires:/BuildRequires: dependencies The first approach was to consider "importing" epel builds/pkgs into cbs, and let SIGs just "tag-build" in their own tag packages, that would also have permitted them to cherry-pick some versions and also ship these pkgs in their SIG repo if needed also a runtime instead of buildtime, but through some conversations, including with EPEL, we considered that it wasn't the best path. Instead, what is now possible is the following : enabling Epel as externalrepo at *build* time so that if your SIG will *rely* on EPEL packages, you can already build against it and you can then ensure that users consuming pkgs from your SIG would have a dependency on epel itself (simple idea is to have your centos-release-$sig pkg having a Requires: on epel-release itself, already available in Extras) The following EPEL repositories can be enabled (on demand) on CBS : - epel 7 (x86_64, ppc64le, aarch64) - epel 8 Everything (x86_64, ppc64le, aarch64) - epel Next 8 Everything (x86_64, ppc64le, aarch64) Worth clarifying that we're talking about plain rpm pkgs, no modularity content at all. (and reminder CBS doesn't have a way to let you build any module -yet- either) This was initially requested and tested/validated with the Hyperscale SIG through this RFE ticket : https://pagure.io/centos-infra/issue/306 , so if you want to "opt-in", feel free to also create ticket and we can enable these external repositories for your next builds PS: just to clarify : if your SIG had to (re)build some EPEL packages for your own tags, nothing will change : you'll still have to rebuild EPEL upstream src.rpm into your tags to see these packages (built on cbs.centos.org koji) be signed and shipped out, without depending on any change (like ENVR bump) on upstream EPEL repositories Kind Regards, -- Fabian Arrotin The CentOS Project | https://www.centos.org gpg key: 17F3B7A1 | twitter: @arrfab

1 0

SIGs building against CentOS Stream
by Rich Bowen 29 Jun '21

29 Jun '21

Do we have a list, somewhere, of which SIGs are currently building against CentOS Stream, and/or are working towards that goal, vs those which are still only building against CentOS Linux?

2 1

CBS/koji feature : enabling centos stream 8 buildroot to have access to missing -devel packages
by Fabian Arrotin 29 Jun '21

29 Jun '21

Hi all SIG members using cbs.centos.org, For a long time there was a need for some SIGs (including Infra SIG) to have access to built-but-unreleased -devel packages for CentOs 8/8-stream (https://wiki.centos.org/FAQ/CentOS8/UnshippedPackages) The only option for SIGs in such situation was to then rebuild src.rpm in cbs, to then be able to consume (as BuildRequires: or just Requires:) these freshly built -devel pkgs for their own packages. Example for such RFE : https://pagure.io/centos-infra/issue/316 As it's now permitted to download directly rpm pkgs from https://koji.mbox.centos.org/koji/, that means that it's also possible (see discussion in ticket) to internally point the CBS builders to the koji.mbox CentOS 8 buildroots, and so be able to automatically grab -devel pkgs built but not pushed to public mirror CDN. So if your SIG needs such feature enabled for your build tags, feel free to also create ticket on infra tracker and we can enable centos8s-buildroot external repo (see https://cbs.centos.org/koji/externalrepoinfo?extrepoID=38) , and it was confirmed to be working for the Hyperscale SIG already. Have a nice day ! -- Fabian Arrotin The CentOS Project | https://www.centos.org gpg key: 17F3B7A1 | twitter: @arrfab

1 0

ANN: Experimental Workstation LiveDVDs from the Hyperscale SIG
by Neal Gompa 29 Jun '21

29 Jun '21

Hello all, Over the past month or so, I've been working on producing CentOS images with Hyperscale content and customizations to showcase the technologies we're working on to the broader community. The goal is to make it easier for the community to get started with the technologies we're working on and to encourage collaboration between the SIG and the wider community on the things we're working on. Today, I'm pleased to announce one of the first deliverables of that effort: experimental live media! The SIG has produced an experimental Hyperscale Workstation Spin that combines CentOS Stream 8 with the Hyperscale SIG content to offer an early preview into some of the work we're doing. This spin is currently available here: * ISO: https://ngompa.fedorapeople.org/centos/hyperscale/CS8-HS-WORK-LIVE-LATESTDE… * SHA512 checksum: https://ngompa.fedorapeople.org/centos/hyperscale/CS8-HS-WORK-LIVE-LATESTDE… There are some limitations and known issues, though: * UEFI boot is not supported. CBS does not support signing kernels such that we can leverage the Secure Boot infrastructure, so UEFI is not supported at all right now. The issue is tracked here: https://pagure.io/centos-infra/issue/307 * The Anaconda installer icon is missing. This will be fixed with the next release of centos-logos. Upstream issue here: https://github.com/CentOS/centos-logos/issues/1 * SELinux denials with systemd-hostnamed and systemd-oomd. This is being investigated, and fixes will become available in a future systemd update provided by the Hyperscale SIG. * SELinux denials with sssd and useradd. This happens with stock CentOS Stream 8, and is not related to anything we're doing. * All non-English locales aren't showing up in Anaconda despite all locales being installed. This is being actively investigated and will hopefully be resolved in a future release. * "dnf upgrade" is broken due to package conflicts with the DNF stack. This is because our custom version of the RPM+DNF stack has not been refreshed to the latest version pushed into CentOS Stream 8. This is ongoing and should be resolved soon. * VDO is broken. This is a consequence of VDO being an out-of-tree kernel module shipped by Red Hat for RHEL kernels only. The Hyperscale SIG is building its kernels based on the ARK tree that's used by Fedora, and thus lacks support for VDO. If you find any other issues, we would definitely love to hear about them! We have a dedicated issue tracker for Hyperscale spins where we would appreciate bugs to be reported: https://pagure.io/centos-sig-hyperscale/spin-bugs The Hyperscale Workstation spin is built on the GNOME desktop. It is my intent to also offer a KDE Plasma flavor once it becomes reasonably possible. This is blocked on KDE Plasma getting rebuilt in EPEL-Next for the updated Qt5 in CentOS Stream 8. Both desktops are capable of leveraging the technologies being developed by the Hyperscale SIG and we intend to use this to showcase that. We hope you enjoy it as much as I enjoyed making this! Thanks and best regards, Neal (on behalf of the CentOS Hyperscale SIG) -- 真実はいつも一つ！/ Always, there's only one truth!

2 1

Zuul CI timeouts on CentOS Stream 9
by Florian Weimer 28 Jun '21

28 Jun '21

It seems that these jobs have a hard timeout of 30 minutes. It is just not possible to build the glibc RPM in under 30 minutes on a single Haswell-era CPU, so these cycles go to waste because every attempt will time out. Is there a way to opt out of this, to conserve resources? Thanks, Florian

3 4

CentOS website pointing at old twitter account.
by Philip Wyett 26 Jun '21

26 Jun '21

Hi all, A trivial one. The project website is still pointing at old twitter account. git diff attached to resolve. Regards Phil -- *** Playing the game for the games own sake. *** WWW: https://kathenas.org Twitter: @kathenasorg Instagram: @kathenasorg IRC: kathenas GPG: 724AA9B52F024C8B

2 1

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

devel