On Wed, Dec 18, 2024 at 5:41 AM Paul Moore <paul@xxxxxxxxxxxxxx> wrote: > > On Tue, Dec 17, 2024 at 12:29 PM Paul Moore <paul@xxxxxxxxxxxxxx> wrote: > > On Tue, Dec 17, 2024 at 11:28 AM Paul Moore <paul@xxxxxxxxxxxxxx> wrote: > > > On Tue, Dec 17, 2024 at 4:17 AM Ondrej Mosnacek <omosnace@xxxxxxxxxx> wrote: > > > > Hi Paul, > > > > > > > > It seems that aarch64 builds of the secnext kernel stopped appearing > > > > in [1] some time ago (and kernel-secnext testing results for aarch64 > > > > stopped appearing as well). Were they intentionally stopped or is > > > > there a problem that prevents them from being produced? > > > > > > > > The new Testing-Farm-based selinux-testsuite CI matrix includes > > > > aarch64 + secnext kernel, so it is somewhat affected (until [2] it has > > > > been quietly installing rawhide kernels instead and passing; now it > > > > will fail when no secnext kernel is available). Any chance the aarch64 > > > > builds could be reinstated? > > > > > > [NOTE: adding the selinux list since it is somewhat related] > > > > > > Hi Ondrej, > > > > > > Yes, sadly for some reason Fedora Rawhide, which I use to build the > > > kernel-secnext kernels, started panicking my aarch64 system *hard* > > > after an update some time ago and I was forced to disable the > > > kernel-secnext builds. Unfortunately the issue appeared to be due to > > > some userspace, or compiler change, that affected the entire system > > > and once my build chroot was updated using dnf the entire chroot was > > > bricked. I saw similar problems on my aarch64 dev/test VMs. I dug > > > around on the RH bugzilla and didn't see any reports, nor anything > > > obvious via google searches. I was hoping to go back and try to debug > > > and/or restore the chroot and VM but I haven't had the time. > > > > > > As a reference point, Fedora 41, Alama v9.5, and Ubuntu 24.04 are all > > > working fine on my aarch64 system so I doubt there is some hardware > > > issue. > > > > > > Are you aware of any nasty aarch64 problems in Rawhide, either now or > > > starting back in October? > > > > I'm still interested in hearing if anyone has any info on a > > aarch64/Rawhide breakage in October as it may help me recover my VMs, > > but I just now tried creating a minimal aarch64/Rawhide chroot (only > > installed bash) and it seems to be working now (!). I'm going to try > > to rebuild a new Rawhide chroot that can build kernels and see how it > > goes ... more info later. > > I was able to rebuild the Rawhide chroot and recover the Rawhide VMs > and it looks like everything is back and working again. It will > likely take several days to catch-up with the src rpm backlog, but you > should see secnext kernels starting to appear in the repo in the next > few hours. Thanks for making it work again! The aarch64 secnext CI runs are passing now. I'm not aware of any bug that could have caused the issue, but it's Rawhide where unfortunately even nasty bugs might sometimes appear. -- Ondrej Mosnacek Senior Software Engineer, Linux Security - SELinux kernel Red Hat, Inc.
