Disabling rootfs verification corrupts stateful |
|||
Issue descriptionOS: 10973.0.0 What steps will reproduce the problem? Tried on eve with nvme, <redacted device with nvme> and coral with emmc. (1) /usr/share/vboot/bin/make_dev_ssd --remove_rootfs_verification --partitions <2 or 4> (2) reboot (3) try to ssh in and touch /foo What is the expected result? - either ssh fails (stateful is corrupted, device can be pinged but cannot ssh in, python is gone, and so is /usr/local/bin). OR - rootfs verification disabling was partial. Can do "touch /var/lib/foo" but not "touch /foo" even after a "mount -o remount,rw /" Please use labels and text to provide additional information. If this is a regression (i.e., worked before), please consider using the bisect tool (https://www.chromium.org/developers/bisect-builds-py) to help us identify the root cause and more rapidly triage the issue. For graphics-related bugs, please copy/paste the contents of the about:gpu page at the end of this report.
,
Aug 16
Sorry, I meant "/lib/firmware/foo" not "/var/lib/foo". For a machine where roots verification removal was done previously, here's some output: ssh root@kbench-eve Password: localhost ~ # ls bytes-rootfs-prod localhost ~ # touch /foo touch: cannot touch '/foo': Permission denied localhost ~ # mount -o remount,rw / localhost ~ # touch /bar touch: cannot touch '/bar': Permission denied localhost ~ # touch /lib/firmware/foo "touch /foo" should work after rootfs verification is disabled.
,
Aug 16
failure to create files in / has nothing to do with rootfs verification. that is an selinux denial you can see in `dmesg`. whether that's intentional is up for debate. you can file a new bug about that, but otherwise should have nothing to do with stateful corruption. lets focus on that part. how often do you see stateful reset after disabling rootfs verification ?
,
Aug 16
i've filed issue 874980 for the selinux changes if you want to track that
,
Aug 17
@vapier: I've lost repro of the stateful corruption. I had it 2/2 on my A50 and A70 i7+nvme units on 10971.0.0/10973.0.0. I'll update the bug when I have a repro again.
,
Aug 17
|
|||
►
Sign in to add a comment |
|||
Comment 1 by vapier@google.com
, Aug 16