CVE-2021-41089

Moby is an open-source project created by Docker to enable software containerization. A bug was found in Moby (Docker Engine) where attempting to copy files using docker cp into a specially-crafted container can result in Unix file permission changes for existing files in the host’s filesystem, widening access to others. This bug does not directly allow files to be read, modified, or executed without an additional cooperating process. This bug has been fixed in Moby (Docker Engine) 20.10.9. Users should update to this version as soon as possible. Running containers do not need to be restarted.

Moby is an open-source project created by Docker to enable software containerization. A bug was found in Moby (Docker Engine) where attempting to copy files using docker cp into a specially-crafted container can result in Unix file permission changes for existing files in the host’s filesystem, widening access to others. This bug does not directly allow files to be read, modified, or executed without an additional cooperating process. This bug has been fixed in Moby (Docker Engine) 20.10.9. Users should update to this version as soon as possible. Running containers do not need to be restarted.

Impact

A bug was found in Moby (Docker Engine) where attempting to copy files using docker cp into a specially-crafted container can result in Unix file permission changes for existing files in the host’s filesystem, widening access to others. This bug does not directly allow files to be read, modified, or executed without an additional cooperating process.

Patches

This bug has been fixed in Moby (Docker Engine) 20.10.9. Users should update to this version as soon as possible. Running containers do not need to be restarted.

Workarounds

Ensure you only run trusted containers.

Credits

The Moby project would like to thank Lei Wang and Ruizhi Xiao for responsibly disclosing this issue in accordance with the Moby security policy.

For more information

If you have any questions or comments about this advisory:

• Open an issue
• Email us at  [email protected]  if you think you’ve found a security bug

Docker is an open source project to build, ship and run any application as a lightweight container.

Security Fix(es):

Moby is an open-source project created by Docker to enable software containerization. A bug was found in Moby (Docker Engine) where the data directory (typically /var/lib/docker) contained subdirectories with insufficiently restricted permissions, allowing otherwise unprivileged Linux users to traverse directory contents and execute programs. When containers included executable programs with extended permission bits (such as setuid), unprivileged Linux users could discover and execute those programs. When the UID of an unprivileged Linux user on the host collided with the file owner or group inside a container, the unprivileged Linux user on the host could discover, read, and modify those files. This bug has been fixed in Moby (Docker Engine) 20.10.9. Users should update to this version as soon as possible. Running containers should be stopped and restarted for the permissions to be fixed. For users unable to upgrade limit access to the host to trusted users. Limit access to host volumes to trusted containers.(CVE-2021-41091)

Moby is an open-source project created by Docker to enable software containerization. A bug was found in Moby (Docker Engine) where attempting to copy files using docker cp into a specially-crafted container can result in Unix file permission changes for existing files in the host’s filesystem, widening access to others. This bug does not directly allow files to be read, modified, or executed without an additional cooperating process. This bug has been fixed in Moby (Docker Engine) 20.10.9. Users should update to this version as soon as possible. Running containers do not need to be restarted.(CVE-2021-41089)

Docker CLI is the command line interface for the docker container runtime. A bug was found in the Docker CLI where running docker login my-private-registry.example.com with a misconfigured configuration file (typically ~/.docker/config.json) listing a credsStore or credHelpers that could not be executed would result in any provided credentials being sent to registry-1.docker.io rather than the intended private registry. This bug has been fixed in Docker CLI 20.10.9. Users should update to this version as soon as possible. For users unable to update ensure that any configured credsStore or credHelpers entries in the configuration file reference an installed credential helper that is executable and on the PATH.(CVE-2021-41092)

Moby is an open-source project created by Docker to enable software containerization. A bug was found in Moby (Docker Engine) where attempting to copy files using docker cp into a specially-crafted container can result in Unix file permission changes for existing files in the host’s filesystem, widening access to others. This bug does not directly allow files to be read, modified, or executed without an additional cooperating process. This bug has been fixed in Moby (Docker Engine) 20.10.9. Users should update to this version as soon as possible. Running containers do not need to be restarted.

Unexpected chmod of host files via 'docker cp' in Moby Docker Engine in github.com/docker/docker

Updated docker packages fix security vulnerabilities:

A bug was found in Moby (Docker Engine) where attempting to copy files using docker cp into a specially-crafted container can result in Unix file permission changes for existing files in the hosts filesystem, widening access to others. This bug does not directly allow files to be read, modified, or executed without an additional cooperating process (CVE-2021-41089).

A bug was found in Moby (Docker Engine) where the data directory (typically '/var/lib/docker') contained subdirectories with insufficiently restricted permissions, allowing otherwise unprivileged Linux users to traverse directory contents and execute programs. When containers included executable programs with extended permission bits (such as 'setuid'), unprivileged Linux users could discover and execute those programs. When the UID of an unprivileged Linux user on the host collided with the file owner or group inside a container, the unprivileged Linux user on the host could discover, read, and modify those files (CVE-2021-41091).

A bug was found in the Docker CLI where running 'docker login my-private-registry.example.com' with a misconfigured configuration file (typically '~/.docker/config.json') listing a 'credsStore' or 'credHelpers' that could not be executed would result in any provided credentials being sent to 'registry-1.docker.io' rather than the intended private registry (CVE-2021-41092).

This update for containerd, docker, runc fixes the following issues:

Docker was updated to 20.10.9-ce. (bsc#1191355)

See upstream changelog in the packaged /usr/share/doc/packages/docker/CHANGELOG.md.

CVE-2021-41092 CVE-2021-41089 CVE-2021-41091 CVE-2021-41103

container was updated to v1.4.11, to fix CVE-2021-41103. bsc#1191355

• CVE-2021-32760: Fixed that a archive package allows chmod of file outside of unpack target directory (bsc#1188282)

• Install systemd service file as well (bsc#1190826)

Update to runc v1.0.2. Upstream changelog is available from

https://github.com/opencontainers/runc/releases/tag/v1.0.2

• Fixed a failure to set CPU quota period in some cases on cgroup v1.
• Fixed the inability to start a container with the 'adding seccomp filter rule for syscall ...' error, caused by redundant seccomp rules (i.e. those that has action equal to the default one). Such redundant rules are now skipped.
• Made release builds reproducible from now on.
• Fixed a rare debug log race in runc init, which can result in occasional harmful 'failed to decode ...' errors from runc run or exec.
• Fixed the check in cgroup v1 systemd manager if a container needs to be frozen before Set, and add a setting to skip such freeze unconditionally. The previous fix for that issue, done in runc 1.0.1, was not working.

Update to runc v1.0.1. Upstream changelog is available from

https://github.com/opencontainers/runc/releases/tag/v1.0.1

• Fixed occasional runc exec/run failure ('interrupted system call') on an Azure volume.
• Fixed 'unable to find groups ... token too long' error with /etc/group containing lines longer than 64K characters.
• cgroup/systemd/v1: fix leaving cgroup frozen after Set if a parent cgroup is frozen. This is a regression in 1.0.0, not affecting runc itself but some of libcontainer users (e.g Kubernetes).
• cgroupv2: bpf: Ignore inaccessible existing programs in case of permission error when handling replacement of existing bpf cgroup programs. This fixes a regression in 1.0.0, where some SELinux policies would block runc from being able to run entirely.
• cgroup/systemd/v2: don't freeze cgroup on Set.
• cgroup/systemd/v1: avoid unnecessary freeze on Set.

• fix issues with runc under openSUSE MicroOS's SELinux policy. bsc#1187704

Update to runc v1.0.0. Upstream changelog is available from

https://github.com/opencontainers/runc/releases/tag/v1.0.0

! The usage of relative paths for mountpoints will now produce a warning (such configurations are outside of the spec, and in future runc will produce an error when given such configurations).

• cgroupv2: devices: rework the filter generation to produce consistent results with cgroupv1, and always clobber any existing eBPF program(s) to fix runc update and avoid leaking eBPF programs (resulting in errors when managing containers).
• cgroupv2: correctly convert 'number of IOs' statistics in a cgroupv1-compatible way.
• cgroupv2: support larger than 32-bit IO statistics on 32-bit architectures.
• cgroupv2: wait for freeze to finish before returning from the freezing code, optimize the method for checking whether a cgroup is frozen.
• cgroups/systemd: fixed 'retry on dbus disconnect' logic introduced in rc94
• cgroups/systemd: fixed returning 'unit already exists' error from a systemd cgroup manager (regression in rc94)

• cgroupv2: support SkipDevices with systemd driver
• cgroup/systemd: return, not ignore, stop unit error from Destroy
• Make 'runc --version' output sane even when built with go get or otherwise outside of our build scripts.
• cgroups: set SkipDevices during runc update (so we don't modify cgroups at all during runc update).
• cgroup1: blkio: support BFQ weights.
• cgroupv2: set per-device io weights if BFQ IO scheduler is available.

Update to runc v1.0.0~rc95. Upstream changelog is available from https://github.com/opencontainers/runc/releases/tag/v1.0.0-rc95

This release of runc contains a fix…

This update for containerd, docker, runc fixes the following issues:

Docker was updated to 20.10.9-ce. (bsc#1191355)

See upstream changelog in the packaged /usr/share/doc/packages/docker/CHANGELOG.md.

CVE-2021-41092 CVE-2021-41089 CVE-2021-41091 CVE-2021-41103

container was updated to v1.4.11, to fix CVE-2021-41103. bsc#1191355

• CVE-2021-32760: Fixed that a archive package allows chmod of file outside of unpack target directory (bsc#1188282)

• Install systemd service file as well (bsc#1190826)

Update to runc v1.0.2. Upstream changelog is available from

https://github.com/opencontainers/runc/releases/tag/v1.0.2

• Fixed a failure to set CPU quota period in some cases on cgroup v1.
• Fixed the inability to start a container with the 'adding seccomp filter rule for syscall ...' error, caused by redundant seccomp rules (i.e. those that has action equal to the default one). Such redundant rules are now skipped.
• Made release builds reproducible from now on.
• Fixed a rare debug log race in runc init, which can result in occasional harmful 'failed to decode ...' errors from runc run or exec.
• Fixed the check in cgroup v1 systemd manager if a container needs to be frozen before Set, and add a setting to skip such freeze unconditionally. The previous fix for that issue, done in runc 1.0.1, was not working.

Update to runc v1.0.1. Upstream changelog is available from

https://github.com/opencontainers/runc/releases/tag/v1.0.1

• Fixed occasional runc exec/run failure ('interrupted system call') on an Azure volume.
• Fixed 'unable to find groups ... token too long' error with /etc/group containing lines longer than 64K characters.
• cgroup/systemd/v1: fix leaving cgroup frozen after Set if a parent cgroup is frozen. This is a regression in 1.0.0, not affecting runc itself but some of libcontainer users (e.g Kubernetes).
• cgroupv2: bpf: Ignore inaccessible existing programs in case of permission error when handling replacement of existing bpf cgroup programs. This fixes a regression in 1.0.0, where some SELinux policies would block runc from being able to run entirely.
• cgroup/systemd/v2: don't freeze cgroup on Set.
• cgroup/systemd/v1: avoid unnecessary freeze on Set.

• fix issues with runc under openSUSE MicroOS's SELinux policy. bsc#1187704

Update to runc v1.0.0. Upstream changelog is available from

https://github.com/opencontainers/runc/releases/tag/v1.0.0

! The usage of relative paths for mountpoints will now produce a warning (such configurations are outside of the spec, and in future runc will produce an error when given such configurations).

• cgroupv2: devices: rework the filter generation to produce consistent results with cgroupv1, and always clobber any existing eBPF program(s) to fix runc update and avoid leaking eBPF programs (resulting in errors when managing containers).
• cgroupv2: correctly convert 'number of IOs' statistics in a cgroupv1-compatible way.
• cgroupv2: support larger than 32-bit IO statistics on 32-bit architectures.
• cgroupv2: wait for freeze to finish before returning from the freezing code, optimize the method for checking whether a cgroup is frozen.
• cgroups/systemd: fixed 'retry on dbus disconnect' logic introduced in rc94
• cgroups/systemd: fixed returning 'unit already exists' error from a systemd cgroup manager (regression in rc94)

• cgroupv2: support SkipDevices with systemd driver
• cgroup/systemd: return, not ignore, stop unit error from Destroy
• Make 'runc --version' output sane even when built with go get or otherwise outside of our build scripts.
• cgroups: set SkipDevices during runc update (so we don't modify cgroups at all during runc update).
• cgroup1: blkio: support BFQ weights.
• cgroupv2: set per-device io weights if BFQ IO scheduler is available.

Update to runc v1.0.0~rc95. Upstream changelog is available from https://github.com/opencontainers/runc/releases/tag/v1.0.0-rc95

This release of runc contains a fix…

This update for containerd, docker fixes the following issues:

• CVE-2021-41089: Fixed 'cp' can chmod host files (bsc#1191015).
• CVE-2021-41091: Fixed flaw that could lead to data directory traversal in moby (bsc#1191434).
• CVE-2021-41092: Fixed exposed user credentials with a misconfigured configuration file (bsc#1191334).
• CVE-2021-41103: Fixed file access to local users in containerd (bsc#1191121).
• CVE-2021-41190: Fixed OCI manifest and index parsing confusion (bsc#1193273).

These are all security issues fixed in the docker-20.10.9_ce-1.2 package on the GA media of openSUSE Tumbleweed.

These are all security issues fixed in the docker-stable-24.0.9_ce-15.1 package on the GA media of openSUSE Tumbleweed.

This update for containerd, docker, runc fixes the following issues:

Docker was updated to 20.10.9-ce. (bsc#1191355)

See upstream changelog in the packaged /usr/share/doc/packages/docker/CHANGELOG.md.

CVE-2021-41092 CVE-2021-41089 CVE-2021-41091 CVE-2021-41103

container was updated to v1.4.11, to fix CVE-2021-41103. bsc#1191355

• CVE-2021-32760: Fixed that a archive package allows chmod of file outside of unpack target directory (bsc#1188282)

Update to runc v1.0.2. Upstream changelog is available from

https://github.com/opencontainers/runc/releases/tag/v1.0.2

• Fixed a failure to set CPU quota period in some cases on cgroup v1.
• Fixed the inability to start a container with the 'adding seccomp filter rule for syscall ...' error, caused by redundant seccomp rules (i.e. those that has action equal to the default one). Such redundant rules are now skipped.
• Made release builds reproducible from now on.
• Fixed a rare debug log race in runc init, which can result in occasional harmful 'failed to decode ...' errors from runc run or exec.
• Fixed the check in cgroup v1 systemd manager if a container needs to be frozen before Set, and add a setting to skip such freeze unconditionally. The previous fix for that issue, done in runc 1.0.1, was not working.

Update to runc v1.0.1. Upstream changelog is available from

https://github.com/opencontainers/runc/releases/tag/v1.0.1

• Fixed occasional runc exec/run failure ('interrupted system call') on an Azure volume.
• Fixed 'unable to find groups ... token too long' error with /etc/group containing lines longer than 64K characters.
• cgroup/systemd/v1: fix leaving cgroup frozen after Set if a parent cgroup is frozen. This is a regression in 1.0.0, not affecting runc itself but some of libcontainer users (e.g Kubernetes).
• cgroupv2: bpf: Ignore inaccessible existing programs in case of permission error when handling replacement of existing bpf cgroup programs. This fixes a regression in 1.0.0, where some SELinux policies would block runc from being able to run entirely.
• cgroup/systemd/v2: don't freeze cgroup on Set.
• cgroup/systemd/v1: avoid unnecessary freeze on Set.

• fix issues with runc under openSUSE MicroOS's SELinux policy. bsc#1187704

Update to runc v1.0.0. Upstream changelog is available from

https://github.com/opencontainers/runc/releases/tag/v1.0.0

! The usage of relative paths for mountpoints will now produce a warning (such configurations are outside of the spec, and in future runc will produce an error when given such configurations).

• cgroupv2: devices: rework the filter generation to produce consistent results with cgroupv1, and always clobber any existing eBPF program(s) to fix runc update and avoid leaking eBPF programs (resulting in errors when managing containers).
• cgroupv2: correctly convert 'number of IOs' statistics in a cgroupv1-compatible way.
• cgroupv2: support larger than 32-bit IO statistics on 32-bit architectures.
• cgroupv2: wait for freeze to finish before returning from the freezing code, optimize the method for checking whether a cgroup is frozen.
• cgroups/systemd: fixed 'retry on dbus disconnect' logic introduced in rc94
• cgroups/systemd: fixed returning 'unit already exists' error from a systemd cgroup manager (regression in rc94)

• cgroupv2: support SkipDevices with systemd driver
• cgroup/systemd: return, not ignore, stop unit error from Destroy
• Make 'runc --version' output sane even when built with go get or otherwise outside of our build scripts.
• cgroups: set SkipDevices during runc update (so we don't modify cgroups at all during runc update).
• cgroup1: blkio: support BFQ weights.
• cgroupv2: set per-device io weights if BFQ IO scheduler is available.

Update to runc v1.0.0~rc95. Upstream changelog is available from https://github.com/opencontainers/runc/releases/tag/v1.0.0-rc95

This release of runc contains a fix for CVE-2021-30465, and users are strongly recommended…

This update for containerd, docker, runc fixes the following issues:

Docker was updated to 20.10.9-ce. (bsc#1191355)

See upstream changelog in the packaged /usr/share/doc/packages/docker/CHANGELOG.md.

CVE-2021-41092 CVE-2021-41089 CVE-2021-41091 CVE-2021-41103

container was updated to v1.4.11, to fix CVE-2021-41103. bsc#1191355

• CVE-2021-32760: Fixed that a archive package allows chmod of file outside of unpack target directory (bsc#1188282)

• Install systemd service file as well (bsc#1190826)

Update to runc v1.0.2. Upstream changelog is available from

https://github.com/opencontainers/runc/releases/tag/v1.0.2

• Fixed a failure to set CPU quota period in some cases on cgroup v1.
• Fixed the inability to start a container with the 'adding seccomp filter rule for syscall ...' error, caused by redundant seccomp rules (i.e. those that has action equal to the default one). Such redundant rules are now skipped.
• Made release builds reproducible from now on.
• Fixed a rare debug log race in runc init, which can result in occasional harmful 'failed to decode ...' errors from runc run or exec.
• Fixed the check in cgroup v1 systemd manager if a container needs to be frozen before Set, and add a setting to skip such freeze unconditionally. The previous fix for that issue, done in runc 1.0.1, was not working.

Update to runc v1.0.1. Upstream changelog is available from

https://github.com/opencontainers/runc/releases/tag/v1.0.1

• Fixed occasional runc exec/run failure ('interrupted system call') on an Azure volume.
• Fixed 'unable to find groups ... token too long' error with /etc/group containing lines longer than 64K characters.
• cgroup/systemd/v1: fix leaving cgroup frozen after Set if a parent cgroup is frozen. This is a regression in 1.0.0, not affecting runc itself but some of libcontainer users (e.g Kubernetes).
• cgroupv2: bpf: Ignore inaccessible existing programs in case of permission error when handling replacement of existing bpf cgroup programs. This fixes a regression in 1.0.0, where some SELinux policies would block runc from being able to run entirely.
• cgroup/systemd/v2: don't freeze cgroup on Set.
• cgroup/systemd/v1: avoid unnecessary freeze on Set.

• fix issues with runc under openSUSE MicroOS's SELinux policy. bsc#1187704

Update to runc v1.0.0. Upstream changelog is available from

https://github.com/opencontainers/runc/releases/tag/v1.0.0

! The usage of relative paths for mountpoints will now produce a warning (such configurations are outside of the spec, and in future runc will produce an error when given such configurations).

• cgroupv2: devices: rework the filter generation to produce consistent results with cgroupv1, and always clobber any existing eBPF program(s) to fix runc update and avoid leaking eBPF programs (resulting in errors when managing containers).
• cgroupv2: correctly convert 'number of IOs' statistics in a cgroupv1-compatible way.
• cgroupv2: support larger than 32-bit IO statistics on 32-bit architectures.
• cgroupv2: wait for freeze to finish before returning from the freezing code, optimize the method for checking whether a cgroup is frozen.
• cgroups/systemd: fixed 'retry on dbus disconnect' logic introduced in rc94
• cgroups/systemd: fixed returning 'unit already exists' error from a systemd cgroup manager (regression in rc94)

• cgroupv2: support SkipDevices with systemd driver
• cgroup/systemd: return, not ignore, stop unit error from Destroy
• Make 'runc --version' output sane even when built with go get or otherwise outside of our build scripts.
• cgroups: set SkipDevices during runc update (so we don't modify cgroups at all during runc update).
• cgroup1: blkio: support BFQ weights.
• cgroupv2: set per-device io weights if BFQ IO scheduler is available.

Update to runc v1.0.0~rc95. Upstream changelog is available from https://github.com/opencontainers/runc/releases/tag/v1.0.0-rc95

This release of runc contains a fix…

This update for containerd, docker fixes the following issues:

• CVE-2021-41089: Fixed 'cp' can chmod host files (bsc#1191015).
• CVE-2021-41091: Fixed flaw that could lead to data directory traversal in moby (bsc#1191434).
• CVE-2021-41092: Fixed exposed user credentials with a misconfigured configuration file (bsc#1191334).
• CVE-2021-41103: Fixed file access to local users in containerd (bsc#1191121).
• CVE-2021-41190: Fixed OCI manifest and index parsing confusion (bsc#1193273).

This update for containerd, docker fixes the following issues:

• CVE-2021-41089: Fixed 'cp' can chmod host files (bsc#1191015).
• CVE-2021-41091: Fixed flaw that could lead to data directory traversal in moby (bsc#1191434).
• CVE-2021-41092: Fixed exposed user credentials with a misconfigured configuration file (bsc#1191334).
• CVE-2021-41103: Fixed file access to local users in containerd (bsc#1191121).
• CVE-2021-41190: Fixed OCI manifest and index parsing confusion (bsc#1193273).

This update for docker-stable fixes the following issues:

• Include historical changelog data from before the docker-stable fork. The initial changelog entry did technically provide all the necessary information, but our CVE tracking tools do not understand how the package is forked and so it seems that this package does not include fixes for ~12 years of updates. So, include a copy of the original package's changelog up until the fork point. (bsc#1250596)

• Remove git-core recommends on SLE. Most SLE systems have installRecommends=yes by default and thus end up installing git with Docker. bsc#1250508

  This feature is mostly intended for developers ('docker build git://') so most users already have the dependency installed, and the error when git is missing is fairly straightforward (so they can easily figure out what they need to install).

• Backport https://github.com/moby/moby/pull/48517. bsc#1247362

• Update to docker-buildx v0.25.0. Upstream changelog: https://github.com/docker/buildx/releases/tag/v0.25.0

• Do not try to inject SUSEConnect secrets when in Rootless Docker mode, as Docker does not have permission to access the host zypper credentials in this mode (and unprivileged users cannot disable the feature using /etc/docker/suse-secrets-enable.) bsc#1240150

• Initial docker-stable fork, forked from Docker 24.0.7-ce release (packaged on 2024-02-14). The original changelog is included below for historical reference.

This update for docker-stable fixes the following issues:

Note this update contains a already fixed references mostly.

• Remove git-core recommends on SLE to avoid pulling it in unnecessary. (bsc#1250508)

  This feature is mostly intended for developers ('docker build git://') so most users already have the dependency installed, and the error when git is missing is fairly straightforward (so they can easily figure out what they need to install).

• Include historical changelog data from before the docker-stable fork. The initial changelog entry did technically provide all the necessary information, but our CVE tracking tools do not understand how the package is forked and so it seems that this package does not include fixes for ~12 years of updates. So, include a copy of the original package's changelog up until the fork point. bsc#1250596

• Update to docker-buildx v0.25.0. Upstream changelog: https://github.com/docker/buildx/releases/tag/v0.25.0

• Update to Go 1.23 for building now that upstream has switched their 23.0.x LTSS to use Go 1.23.

• Do not try to inject SUSEConnect secrets when in Rootless Docker mode, as Docker does not have permission to access the host zypper credentials in this mode (and unprivileged users cannot disable the feature using /etc/docker/suse-secrets-enable.) bsc#1240150

• Initial docker-stable fork, forked from Docker 24.0.7-ce release (packaged on 2024-02-14). The original changelog is included below for historical reference.

Lei Wang and Ruizhi Xiao discovered that the Moby Docker engine in Docker incorrectly allowed the docker cp command to make permissions changes in the host filesystem in some situations. A local attacker could possibly use to this to expose sensitive information or gain administrative privileges.