Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[WIP] Make /proc/sys read-only with carve-outs for some sysctls #3518

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
5 changes: 5 additions & 0 deletions images/base/Dockerfile
Original file line number Diff line number Diff line change
Expand Up @@ -97,6 +97,11 @@ RUN echo "Enabling / Disabling services ... " \
RUN echo "Ensuring /etc/kubernetes/manifests" \
&& mkdir -p /etc/kubernetes/manifests

# Used as mount points for private copies of proc and sys filesystems in entrypoint.
RUN echo "Ensuring /kind/private" \
&& mkdir -p /kind/private/proc /kind/private/sys \
&& chmod 0700 /kind/private /kind/private/proc /kind/private/sys

# shared stage to setup go version for building binaries
# NOTE we will be cross-compiling for performance reasons
# This is also why we start again FROM the same base image but a different
Expand Down
30 changes: 29 additions & 1 deletion images/base/files/usr/local/bin/entrypoint
Original file line number Diff line number Diff line change
Expand Up @@ -172,13 +172,24 @@ fix_mount() {
sync
fi

# Mount sysfs and proc as read-write, on a known, but kind specific location.
# This allows bind mounting, below and is also required to run some workloads
# which need to mount proc and sysfs themselves (this avoids the proc and
# sysfs mounts being "masked", as far as the kernel is concerned).
# XXX, better ref for fs_fully_visible than kernel code?
# https://github.com/torvalds/linux/commit/1b852bceb0d1
log_info 'mounting /kind/private filesystems'
mount -t sysfs -o rw sysfs /kind/private/sys
mount -t proc -o rw proc /kind/private/proc

log_info 'remounting /sys read-only'
# systemd-in-a-container should have read only /sys
# https://systemd.io/CONTAINER_INTERFACE/
# however, we need other things from `docker run --privileged` ...
# and this flag also happens to make /sys rw, amongst other things
#
# This step is ignored when running inside UserNS, because it fails with EACCES.
# This step is ignored when running inside UserNS, because it can fail with
# EACCES.
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Unnecessary change

if ! mount -o remount,ro /sys; then
if [[ -n "$userns" ]]; then
log_info 'UserNS: ignoring mount fail'
Expand All @@ -187,6 +198,23 @@ fix_mount() {
fi
fi

log_info 'making /proc/sys read-only, with known sysctls read-write'
mount --rbind -o ro /proc/sys /proc/sys
# These are the sysctls known to be namespaced in the kernel, list taken from Kubernetes:
# https://github.com/kubernetes/kubernetes/blob/master/staging/src/k8s.io/component-helpers/node/util/sysctl/namespace.go
# In addition the kubelet attempts to set some sysctl to particular settings, we allow those:
# https://github.com/search?q=repo%3Akubernetes/kubernetes%20setupKernelTunables&type=code
for mount_point in \
kernel/shmall kernel/shmmax kernel/shmmni kernel/shm_rmid_forced kernel/msgmax kernel/msgmnb kernel/msgmni \
fs/mqueue \
net \
vm/overcommit_memory vm/panic_on_oom kernel/panic kernel/panic_on_oops \
kernel/keys/root_maxkeys kernel/keys/root_maxbytes; do
if [[ -f /kind/private/proc/sys/"${mount_point}" ]]; then
mount --bind -o rw /kind/private/proc/sys/"${mount_point}" /proc/sys/"${mount_point}"
fi
done
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm not sure about the robustness of this.
I think this should be opt-in.


log_info 'making mounts shared'
# for mount propagation
mount --make-rshared /
Expand Down