Hi Reinhard, On Sun, Jan 25, 2026 at 05:26:22PM +0100, Salvatore Bonaccorso wrote: > Control: tags -1 + upstream > Control: forwarded -1 https://tracker.ceph.com/issues/73886 > https://lore.kernel.org/ceph-devel/[email protected]/ > > Hi Reinhard, > > On Sun, Jan 25, 2026 at 11:22:52AM +0100, Reinhard Eilmsteiner wrote: > > OK, I just restarted > > > > FYI, I have create this script to build kernels: > > #!/bin/bash > > > > cd /home/eilmrein/code/linux-stable > > git checkout $1 > > cp /boot/config-$(uname -r) .config > > yes '' | make localmodconfig > > ./scripts/config --module CEPH_FS > > ./scripts/config --module CEPH_LIB > > ./scripts/config --enable CEPH_FSCACHE > > ./scripts/config --enable CEPH_FS_POSIX_ACL > > make savedefconfig > > mv defconfig arch/x86/configs/my_defconfig > > make my_defconfig > > make -j $(nproc) bindeb-pkg > > > > After having run it with v6.12.58 as argument, I installed the resulting > > kernel using > > sudo dpkg -i linux-image-6.12.58_6.12.58-16_amd64.deb > > > > Booted into it and did a quick ls on one of the cephfs, which I have systemd > > automount for me - however it also fails when I try to mount it manually , I > > get > > natchez➜ codels /remote/vdl/files01 > > [1] 2647 killed ls --color=tty /remote/vdl/files01 > > > > , did the requested git bisect bad > > > > and as it is the immediate successor, I now ran this right after making the > > current kernel as bad. > > > > git bisect start > > # status: waiting for both good and bad commits > > # good: [8a243ecde1f6447b8e237f2c1c67c0bb67d16d67] Linux 6.12.57 > > git bisect good 8a243ecde1f6447b8e237f2c1c67c0bb67d16d67 > > # status: waiting for bad commit, 1 good commit known > > # bad: [567bd8cbc2fe6b28b78864cbbbc41b0d405eb83c] Linux 6.12.63 > > git bisect bad 567bd8cbc2fe6b28b78864cbbbc41b0d405eb83c > > # bad: [7475d784169c7df48b0c55525fb862e06674d63c] Linux 6.12.58 > > git bisect bad 7475d784169c7df48b0c55525fb862e06674d63c > > This is still not the bisect procedure :). > > The next ommit pointed out by the procedure to bisec would be > 850c7f0537cc5a37ed012592907920637cc548b6 . You need to build with that > commit up a kernel, boot into it, test if ceph mount works or not and > based on the result mark it with 'git bisect bad' or 'git bisect > good'. Git will then calculate you the next commit to test, and you > interate until you get 'the first bad commit' result. We need to have > that (or worst case a range, if you needed to skip one). > > But in meahwhile I have found an upstream repot which matches your > report: > > https://tracker.ceph.com/issues/73886 > https://lore.kernel.org/ceph-devel/[email protected]/ > > So it would be great if you instead could built the current kernel in > trixie with this patch added on top. > > The procedure for this one is documented in > https://kernel-team.pages.debian.net/kernel-handbook/ch-common-tasks.html#id-1.6.6.4 > but let me know if you need help with that.
Let's shortcut that step, I have built *test* packages at https://people.debian.org/~carnil/tmp/linux/1125405-cephfs/ (if you want to trust them, unsigned, but I have put a sha256sum file for those signed with my gpg key in the Debian keyring). Please report back if that fixes the problem. Regards, Salvatore
