Fedora Account System
Red Hat Associate
Red Hat Customer
Today's rawhide updates broke checkpoint/restore in podman/criu. I bisected the package updates, and the trigger is the kernel. works: kernel-core-6.13.0-0.rc0.20241119git158f238aa69d.2.fc42.x86_64 fails: kernel-core-6.13.0-0.rc0.20241125git9f16d5e6f220.8.fc42.x86_64 Reproducible: Always Steps to Reproduce: podman run --name=test1 -d quay.io/libpod/busybox sleep 1000 podman container checkpoint test1 Actual Results: 2024-11-26T16:11:08.428787Z: CRIU checkpointing failed -52. Please check CRIU logfile /var/lib/containers/storage/overlay-containers/2a3d740ede2eed92a3f963902dcb12bdd4e476ee2f3f73ffc86dfee71a5f9fbe/userdata/dump.log Error: `/usr/bin/crun checkpoint --image-path /var/lib/containers/storage/overlay-containers/2a3d740ede2eed92a3f963902dcb12bdd4e476ee2f3f73ffc86dfee71a5f9fbe/userdata/checkpoint --work-path /var/lib/containers/storage/overlay-containers/2a3d740ede2eed92a3f963902dcb12bdd4e476ee2f3f73ffc86dfee71a5f9fbe/userdata 2a3d740ede2eed92a3f963902dcb12bdd4e476ee2f3f73ffc86dfee71a5f9fbe` failed: exit status 1 Full criu log [1] ends like this: (00.019461) vdso: Parsing at 7f13e5a1e000 7f13e5a20000 (00.019479) vdso: PT_LOAD p_vaddr: 0 (00.019482) vdso: DT_HASH: 120 (00.019483) vdso: DT_STRTAB: 340 (00.019485) vdso: DT_SYMTAB: 1d8 (00.019486) vdso: DT_STRSZ: 9c (00.019488) vdso: DT_SYMENT: 18 (00.019489) vdso: nbucket 3 nchain f bucket 7f13e5a1e128 chain 7f13e5a1e134 (00.019492) Error (criu/vdso.c:381): vdso: Unexpected rt vDSO area bounds (00.019494) Error (criu/vdso.c:613): vdso: Failed to fill self vdso symtable (00.019496) Error (criu/kerndat.c:1906): kerndat_vdso_fill_symtable failed when initializing kerndat. (00.019588) Found mmap_min_addr 0x10000 (00.019668) files stat: fs/nr_open 1073741816 [1] https://artifacts.dev.testing-farm.io/4ff7f22a-1bc3-4c0b-944c-ff705a67615d/work-podman-miscrmewa1yc/plans/cockpit-podman/podman-misc/execute/data/guest/default-0/test/browser/other-1/data/TestApplication-testCheckpointRestore-dump.log Expected Results: checkpointing works Please reassign to the kernel if that's an actual regression, and not just criu needing an update to kernel changes.
Lokesh, FYI. That's the regression discussed in https://github.com/containers/podman/pull/24682#issuecomment-2500237227 ff., and it affects podman.
We opened the following GitHub issue to track this problem upstream: https://github.com/checkpoint-restore/criu/issues/2532
This error has been fixed upstream in CRIU with the following pull request: https://github.com/checkpoint-restore/criu/pull/2539. We have updated the CRIU package in Fedora Rawhide with this patch.
@rstoyano Can you have a look at the failures in https://bodhi.fedoraproject.org/updates/FEDORA-2024-317bb280cd and fix or waive them so this update can get into the stable repo? Thanks.
Thank you for pointing this out! I waived this test to pass CI.
Confirmed fixed in current rawhide. Thanks!