Fnlb was moved to its own repo: fnproject/lb (#702)

* Fnlb was moved to its own repo: fnproject/lb

* Clean up fnlb leftovers

* Newer deps
This commit is contained in:
Denis Makogon
2018-01-23 00:17:29 +02:00
committed by Reed Allman
parent 4ffa3d5005
commit d3be603e54
8310 changed files with 457462 additions and 1749312 deletions

View File

@@ -1,6 +1,6 @@
## OCI Image Layout Specification
* The OCI Image Layout is a slash separated layout of OCI content-addressable blobs and [location-addressable](https://en.wikipedia.org/wiki/Content-addressable_storage#Content-addressed_vs._location-addressed) references (refs).
* The OCI Image Layout is directory structure for OCI content-addressable blobs and [location-addressable](https://en.wikipedia.org/wiki/Content-addressable_storage#Content-addressed_vs._location-addressed) references (refs).
* This layout MAY be used in a variety of different transport mechanisms: archive formats (e.g. tar, zip), shared filesystem environments (e.g. nfs), or networked file fetching (e.g. http, ftp, rsync).
Given an image layout and a ref, a tool can create an [OCI Runtime Specification bundle](https://github.com/opencontainers/runtime-spec/blob/v1.0.0/bundle.md) by:

View File

@@ -22,10 +22,10 @@ const (
// VersionMinor is for functionality in a backwards-compatible manner
VersionMinor = 0
// VersionPatch is for backwards-compatible bug fixes
VersionPatch = 0
VersionPatch = 1
// VersionDev indicates development branch. Releases will be empty string.
VersionDev = "-dev"
VersionDev = ""
)
// Version is the specification version that the package types support.

View File

@@ -1,5 +1,4 @@
vendor/pkg
/runc
contrib/cmd/recvtty/recvtty
Godeps/_workspace/src/github.com/opencontainers/runc
man/man8
release

View File

@@ -1,10 +0,0 @@
approve_by_comment: true
approve_regex: ^LGTM
reject_regex: ^Rejected
reset_on_push: true
author_approval: ignored
reviewers:
teams:
- runc-maintainers
name: default
required: 2

View File

@@ -1,33 +0,0 @@
language: go
go:
- 1.7.x
- 1.8.x
- tip
matrix:
allow_failures:
- go: tip
go_import_path: github.com/opencontainers/runc
# `make ci` uses Docker.
sudo: required
services:
- docker
env:
global:
- BUILDTAGS="seccomp apparmor selinux ambient"
before_install:
- sudo apt-get -qq update
- sudo apt-get install -y libseccomp-dev libapparmor-dev
- go get -u github.com/golang/lint/golint
- go get -u github.com/vbatts/git-validation
- go get -u github.com/mvdan/sh/cmd/shfmt
- env | grep TRAVIS_
script:
- git-validation -run DCO,short-subject -v
- make BUILDTAGS="${BUILDTAGS}"
- make BUILDTAGS="${BUILDTAGS}" clean ci

View File

@@ -1,12 +1,5 @@
## Contribution Guidelines
### Security issues
If you are reporting a security issue, do not create an issue or file a pull
request on GitHub. Instead, disclose the issue responsibly by sending an email
to security@opencontainers.org (which is inhabited only by the maintainers of
the various OCI projects).
### Pull requests are always welcome
We are always thrilled to receive pull requests, and do our best to

View File

@@ -1,60 +1,3 @@
FROM golang:1.8.0
# libseccomp in jessie is not _quite_ new enough -- need backports version
RUN echo 'deb http://httpredir.debian.org/debian jessie-backports main' > /etc/apt/sources.list.d/backports.list
RUN apt-get update && apt-get install -y \
build-essential \
curl \
sudo \
gawk \
iptables \
jq \
pkg-config \
libaio-dev \
libcap-dev \
libprotobuf-dev \
libprotobuf-c0-dev \
libnl-3-dev \
libnet-dev \
libseccomp2/jessie-backports \
libseccomp-dev/jessie-backports \
protobuf-c-compiler \
protobuf-compiler \
python-minimal \
--no-install-recommends \
&& apt-get clean
# Add a dummy user for the rootless integration tests. While runC does
# not require an entry in /etc/passwd to operate, one of the tests uses
# `git clone` -- and `git clone` does not allow you to clone a
# repository if the current uid does not have an entry in /etc/passwd.
RUN useradd -u1000 -m -d/home/rootless -s/bin/bash rootless
# install bats
RUN cd /tmp \
&& git clone https://github.com/sstephenson/bats.git \
&& cd bats \
&& git reset --hard 03608115df2071fff4eaaff1605768c275e5f81f \
&& ./install.sh /usr/local \
&& rm -rf /tmp/bats
# install criu
ENV CRIU_VERSION 3ca8e575b49763030d3ddfec4af190a4c9f9deef
RUN mkdir -p /usr/src/criu \
&& curl -sSL https://github.com/xemul/criu/archive/${CRIU_VERSION}.tar.gz | tar -v -C /usr/src/criu/ -xz --strip-components=1 \
&& cd /usr/src/criu \
&& make install-criu \
&& rm -rf /usr/src/criu
# setup a playground for us to spawn containers in
ENV ROOTFS /busybox
RUN mkdir -p ${ROOTFS} \
&& curl -o- -sSL 'https://github.com/docker-library/busybox/raw/a0558a9006ce0dd6f6ec5d56cfd3f32ebeeb815f/glibc/busybox.tar.xz' | tar xfJC - ${ROOTFS}
COPY script/tmpmount /
WORKDIR /go/src/github.com/opencontainers/runc
ENTRYPOINT ["/tmpmount"]
FROM runc_test
ADD . /go/src/github.com/opencontainers/runc
RUN make

View File

@@ -0,0 +1,77 @@
{
"ImportPath": "github.com/opencontainers/runc",
"GoVersion": "go1.5.3",
"Deps": [
{
"ImportPath": "github.com/Sirupsen/logrus",
"Comment": "v0.7.3-2-g26709e2",
"Rev": "26709e2714106fb8ad40b773b711ebce25b78914"
},
{
"ImportPath": "github.com/codegangsta/cli",
"Comment": "1.2.0-187-gc31a797",
"Rev": "c31a7975863e7810c92e2e288a9ab074f9a88f29"
},
{
"ImportPath": "github.com/coreos/go-systemd/activation",
"Comment": "v4",
"Rev": "b4a58d95188dd092ae20072bac14cece0e67c388"
},
{
"ImportPath": "github.com/coreos/go-systemd/dbus",
"Comment": "v4",
"Rev": "b4a58d95188dd092ae20072bac14cece0e67c388"
},
{
"ImportPath": "github.com/coreos/go-systemd/util",
"Comment": "v4",
"Rev": "b4a58d95188dd092ae20072bac14cece0e67c388"
},
{
"ImportPath": "github.com/docker/docker/pkg/mount",
"Comment": "v1.4.1-4831-g0f5c9d3",
"Rev": "0f5c9d301b9b1cca66b3ea0f9dec3b5317d3686d"
},
{
"ImportPath": "github.com/docker/docker/pkg/symlink",
"Comment": "v1.4.1-4831-g0f5c9d3",
"Rev": "0f5c9d301b9b1cca66b3ea0f9dec3b5317d3686d"
},
{
"ImportPath": "github.com/docker/docker/pkg/term",
"Comment": "v1.4.1-4831-g0f5c9d3",
"Rev": "0f5c9d301b9b1cca66b3ea0f9dec3b5317d3686d"
},
{
"ImportPath": "github.com/docker/go-units",
"Comment": "v0.1.0",
"Rev": "9b001659dd36225e356b4467c465d732e745f53d"
},
{
"ImportPath": "github.com/godbus/dbus",
"Comment": "v3",
"Rev": "c7fdd8b5cd55e87b4e1f4e372cdb1db61dd6c66f"
},
{
"ImportPath": "github.com/golang/protobuf/proto",
"Rev": "f7137ae6b19afbfd61a94b746fda3b3fe0491874"
},
{
"ImportPath": "github.com/opencontainers/runtime-spec/specs-go",
"Comment": "v0.5.0-34-g6e08c69",
"Rev": "6e08c6983ef8c2173f10ca09266907d4e9e71716"
},
{
"ImportPath": "github.com/seccomp/libseccomp-golang",
"Rev": "1b506fc7c24eec5a3693cdcbed40d9c226cfc6a1"
},
{
"ImportPath": "github.com/syndtr/gocapability/capability",
"Rev": "2c00daeb6c3b45114c80ac44119e7b8801fdd852"
},
{
"ImportPath": "github.com/vishvananda/netlink",
"Rev": "1e2e08e8a2dcdacaae3f14ac44c5cfa31361f270"
}
]
}

5
vendor/github.com/opencontainers/runc/Godeps/Readme generated vendored Normal file
View File

@@ -0,0 +1,5 @@
This directory tree is generated automatically by godep.
Please do not edit.
See https://github.com/tools/godep for more information.

View File

@@ -0,0 +1,2 @@
/pkg
/bin

View File

@@ -2,6 +2,8 @@ Michael Crosby <michael@docker.com> (@crosbymichael)
Rohit Jnagal <jnagal@google.com> (@rjnagal)
Victor Marmol <vmarmol@google.com> (@vmarmol)
Mrunal Patel <mpatel@redhat.com> (@mrunalp)
Alexander Morozov <lk4d4@docker.com> (@LK4D4)
Daniel, Dao Quang Minh <dqminh89@gmail.com> (@dqminh)
Andrey Vagin <avagin@virtuozzo.com> (@avagin)
Qiang Huang <h.huangqiang@huawei.com> (@hqhq)
Aleksa Sarai <asarai@suse.de> (@cyphar)

View File

@@ -1,122 +1,57 @@
.PHONY: all shell dbuild man \
localtest localunittest localintegration \
test unittest integration
RUNC_IMAGE=runc_dev
RUNC_TEST_IMAGE=runc_test
PROJECT=github.com/opencontainers/runc
TEST_DOCKERFILE=script/test_Dockerfile
BUILDTAGS=seccomp
RUNC_BUILD_PATH=/go/src/github.com/opencontainers/runc/runc
RUNC_INSTANCE=runc_dev
COMMIT=$(shell git rev-parse HEAD 2> /dev/null || true)
RUNC_LINK=$(CURDIR)/Godeps/_workspace/src/github.com/opencontainers/runc
export GOPATH:=$(CURDIR)/Godeps/_workspace:$(GOPATH)
GO := go
.PHONY=dbuild
SOURCES := $(shell find . 2>&1 | grep -E '.*\.(c|h|go)$$')
PREFIX := $(DESTDIR)/usr/local
BINDIR := $(PREFIX)/sbin
GIT_BRANCH := $(shell git rev-parse --abbrev-ref HEAD 2>/dev/null)
GIT_BRANCH_CLEAN := $(shell echo $(GIT_BRANCH) | sed -e "s/[^[:alnum:]]/-/g")
RUNC_IMAGE := runc_dev$(if $(GIT_BRANCH_CLEAN),:$(GIT_BRANCH_CLEAN))
PROJECT := github.com/opencontainers/runc
BUILDTAGS := seccomp
COMMIT_NO := $(shell git rev-parse HEAD 2> /dev/null || true)
COMMIT := $(if $(shell git status --porcelain --untracked-files=no),"${COMMIT_NO}-dirty","${COMMIT_NO}")
all:
ifneq ($(RUNC_LINK), $(wildcard $(RUNC_LINK)))
ln -sfn $(CURDIR) $(RUNC_LINK)
endif
go build -ldflags "-X main.gitCommit=${COMMIT}" -tags "$(BUILDTAGS)" -o runc .
MAN_DIR := $(CURDIR)/man/man8
MAN_PAGES = $(shell ls $(MAN_DIR)/*.8)
MAN_PAGES_BASE = $(notdir $(MAN_PAGES))
MAN_INSTALL_PATH := ${PREFIX}/share/man/man8/
RELEASE_DIR := $(CURDIR)/release
VERSION := ${shell cat ./VERSION}
SHELL := $(shell command -v bash 2>/dev/null)
.DEFAULT: runc
runc: $(SOURCES)
$(GO) build -buildmode=pie $(EXTRA_FLAGS) -ldflags "-X main.gitCommit=${COMMIT} -X main.version=${VERSION} $(EXTRA_LDFLAGS)" -tags "$(BUILDTAGS)" -o runc .
all: runc recvtty
recvtty: contrib/cmd/recvtty/recvtty
contrib/cmd/recvtty/recvtty: $(SOURCES)
$(GO) build -buildmode=pie $(EXTRA_FLAGS) -ldflags "-X main.gitCommit=${COMMIT} -X main.version=${VERSION} $(EXTRA_LDFLAGS)" -tags "$(BUILDTAGS)" -o contrib/cmd/recvtty/recvtty ./contrib/cmd/recvtty
static: $(SOURCES)
CGO_ENABLED=1 $(GO) build $(EXTRA_FLAGS) -tags "$(BUILDTAGS) cgo static_build" -ldflags "-w -extldflags -static -X main.gitCommit=${COMMIT} -X main.version=${VERSION} $(EXTRA_LDFLAGS)" -o runc .
CGO_ENABLED=1 $(GO) build $(EXTRA_FLAGS) -tags "$(BUILDTAGS) cgo static_build" -ldflags "-w -extldflags -static -X main.gitCommit=${COMMIT} -X main.version=${VERSION} $(EXTRA_LDFLAGS)" -o contrib/cmd/recvtty/recvtty ./contrib/cmd/recvtty
release:
script/release.sh -r release/$(VERSION) -v $(VERSION)
dbuild: runcimage
docker run --rm -v $(CURDIR):/go/src/$(PROJECT) --privileged $(RUNC_IMAGE) make clean all
static:
CGO_ENABLED=1 go build -tags "$(BUILDTAGS) cgo static_build" -ldflags "-w -extldflags -static -X main.gitCommit=${COMMIT}" -o runc .
lint:
$(GO) vet $(allpackages)
$(GO) fmt $(allpackages)
go vet ./...
go fmt ./...
man:
man/md2man-all.sh
runctestimage:
docker build -t $(RUNC_TEST_IMAGE) -f $(TEST_DOCKERFILE) .
runcimage:
test: runctestimage
docker run -e TESTFLAGS -ti --privileged --rm -v $(CURDIR):/go/src/$(PROJECT) $(RUNC_TEST_IMAGE) make localtest
tests/sniffTest
localtest: all
go test -tags "$(BUILDTAGS)" ${TESTFLAGS} -v ./...
dbuild: runctestimage
docker build -t $(RUNC_IMAGE) .
test:
make unittest integration rootlessintegration
localtest:
make localunittest localintegration localrootlessintegration
unittest: runcimage
docker run -e TESTFLAGS -t --privileged --rm -v $(CURDIR):/go/src/$(PROJECT) $(RUNC_IMAGE) make localunittest
localunittest: all
$(GO) test -timeout 3m -tags "$(BUILDTAGS)" ${TESTFLAGS} -v $(allpackages)
integration: runcimage
docker run -e TESTFLAGS -t --privileged --rm -v $(CURDIR):/go/src/$(PROJECT) $(RUNC_IMAGE) make localintegration
localintegration: all
bats -t tests/integration${TESTFLAGS}
rootlessintegration: runcimage
docker run -e TESTFLAGS -t --privileged --rm -v $(CURDIR):/go/src/$(PROJECT) --cap-drop=ALL -u rootless $(RUNC_IMAGE) make localintegration
# FIXME: This should not be separate from rootlessintegration's method of running.
localrootlessintegration: all
sudo -u rootless -H PATH="${PATH}" bats -t tests/integration${TESTFLAGS}
shell: all
docker run -e TESTFLAGS -ti --privileged --rm -v $(CURDIR):/go/src/$(PROJECT) $(RUNC_IMAGE) bash
docker create --name=$(RUNC_INSTANCE) $(RUNC_IMAGE)
docker cp $(RUNC_INSTANCE):$(RUNC_BUILD_PATH) .
docker rm $(RUNC_INSTANCE)
install:
install -D -m0755 runc $(BINDIR)/runc
install-bash:
install -D -m0644 contrib/completions/bash/runc $(PREFIX)/share/bash-completion/completions/runc
install-man:
install -d -m 755 $(MAN_INSTALL_PATH)
install -m 644 $(MAN_PAGES) $(MAN_INSTALL_PATH)
install -D -m0755 runc /usr/local/sbin/runc
uninstall:
rm -f $(BINDIR)/runc
uninstall-bash:
rm -f $(PREFIX)/share/bash-completion/completions/runc
uninstall-man:
rm -f $(addprefix $(MAN_INSTALL_PATH),$(MAN_PAGES_BASE))
rm -f /usr/local/sbin/runc
clean:
rm -f runc
rm -f contrib/cmd/recvtty/recvtty
rm -rf $(RELEASE_DIR)
rm -rf $(MAN_DIR)
rm -f $(RUNC_LINK)
validate:
script/validate-gofmt
$(GO) vet $(allpackages)
go vet ./...
ci: validate test release
# memoize allpackages, so that it's executed only once and only if used
_allpackages = $(shell $(GO) list ./... | grep -v vendor)
allpackages = $(if $(__allpackages),,$(eval __allpackages := $$(_allpackages)))$(__allpackages)
ci: validate localtest

View File

@@ -1,244 +1,144 @@
# runc
[![Build Status](https://jenkins.dockerproject.org/buildStatus/icon?job=runc Master)](https://jenkins.dockerproject.org/job/runc Master)
[![Build Status](https://travis-ci.org/opencontainers/runc.svg?branch=master)](https://travis-ci.org/opencontainers/runc)
[![Go Report Card](https://goreportcard.com/badge/github.com/opencontainers/runc)](https://goreportcard.com/report/github.com/opencontainers/runc)
[![GoDoc](https://godoc.org/github.com/opencontainers/runc?status.svg)](https://godoc.org/github.com/opencontainers/runc)
## runc
## Introduction
`runc` is a CLI tool for spawning and running containers according to the OCF specification.
`runc` is a CLI tool for spawning and running containers according to the OCI specification.
## State of the project
## Releases
Currently `runc` is an implementation of the OCI specification. We are currently sprinting
to have a v1 of the spec out. So the `runc` config format will be constantly changing until
the spec is finalized. However, we encourage you to try out the tool and give feedback.
`runc` depends on and tracks the [runtime-spec](https://github.com/opencontainers/runtime-spec) repository.
We will try to make sure that `runc` and the OCI specification major versions stay in lockstep.
This means that `runc` 1.0.0 should implement the 1.0 version of the specification.
### OCF
You can find official releases of `runc` on the [release](https://github.com/opencontainers/runc/releases) page.
How does `runc` integrate with the Open Container Initiative Specification?
`runc` depends on the types specified in the
[specs](https://github.com/opencontainers/runtime-spec) repository. Whenever the
specification is updated and ready to be versioned `runc` will update its dependency
on the specs repository and support the update spec.
### Security
### Building:
If you wish to report a security issue, please disclose the issue responsibly
to security@opencontainers.org.
## Building
`runc` currently supports the Linux platform with various architecture support.
It must be built with Go version 1.6 or higher in order for some features to function properly.
In order to enable seccomp support you will need to install `libseccomp` on your platform.
> e.g. `libseccomp-devel` for CentOS, or `libseccomp-dev` for Ubuntu
Otherwise, if you do not want to build `runc` with seccomp support you can add `BUILDTAGS=""` when running make.
At the time of writing, runc only builds on the Linux platform.
```bash
# create a 'github.com/opencontainers' in your GOPATH/src
cd github.com/opencontainers
git clone https://github.com/opencontainers/runc
cd runc
make
sudo make install
```
`runc` will be installed to `/usr/local/sbin/runc` on your system.
In order to enable seccomp support you will need to install libseccomp on your platform.
If you do not want to build `runc` with seccomp support you can add `BUILDTAGS=""` when running make.
#### Build Tags
`runc` supports optional build tags for compiling support of various features.
To add build tags to the make option the `BUILDTAGS` variable must be set.
`runc` supports optional build tags for compiling in support for various features.
```bash
make BUILDTAGS='seccomp apparmor'
```
| Build Tag | Feature | Dependency |
|-----------|------------------------------------|-------------|
| seccomp | Syscall filtering | libseccomp |
| selinux | selinux process and mount labeling | <none> |
| apparmor | apparmor profile support | libapparmor |
| ambient | ambient capability support | kernel 4.3 |
### Testing:
### Running the test suite
`runc` currently supports running its test suite via Docker.
To run the suite just type `make test`.
You can run tests for runC by using command:
```bash
make test
# make test
```
There are additional make targets for running the tests outside of a container but this is not recommended as the tests are written with the expectation that they can write and remove anywhere.
Note that test cases are run in Docker container, so you need to install
`docker` first. And test requires mounting cgroups inside container, it's
done by docker now, so you need a docker version newer than 1.8.0-rc2.
You can run a specific test case by setting the `TESTFLAGS` variable.
You can also run specific test cases by:
```bash
# make test TESTFLAGS="-run=SomeTestFunction"
```
### Dependencies Management
### Using:
`runc` uses [vndr](https://github.com/LK4D4/vndr) for dependencies management.
Please refer to [vndr](https://github.com/LK4D4/vndr) for how to add or update
new dependencies.
## Using runc
### Creating an OCI Bundle
In order to use runc you must have your container in the format of an OCI bundle.
If you have Docker installed you can use its `export` method to acquire a root filesystem from an existing Docker container.
To run a container with the id "test", execute `runc start` with the containers id as arg one
in the bundle's root directory:
```bash
# create the top most bundle directory
mkdir /mycontainer
cd /mycontainer
runc start test
/ $ ps
PID USER COMMAND
1 daemon sh
5 daemon sh
/ $
```
# create the rootfs directory
### OCI Container JSON Format:
OCI container JSON format is based on OCI [specs](https://github.com/opencontainers/runtime-spec).
You can generate JSON files by using `runc spec`.
It assumes that the file-system is found in a directory called
`rootfs` and there is a user with uid and gid of `0` defined within that file-system.
### Examples:
#### Using a Docker image (requires version 1.3 or later)
To test using Docker's `busybox` image follow these steps:
* Install `docker` and download the `busybox` image: `docker pull busybox`
* Create a container from that image and export its contents to a tar file:
`docker export $(docker create busybox) > busybox.tar`
* Untar the contents to create your filesystem directory:
```
mkdir rootfs
# export busybox via Docker into the rootfs directory
docker export $(docker create busybox) | tar -C rootfs -xvf -
tar -C rootfs -xf busybox.tar
```
* Create `config.json` by using `runc spec`.
* Execute `runc start` and you should be placed into a shell where you can run `ps`:
```
$ runc start test
/ # ps
PID USER COMMAND
1 root sh
9 root ps
```
After a root filesystem is populated you just generate a spec in the format of a `config.json` file inside your bundle.
`runc` provides a `spec` command to generate a base template spec that you are then able to edit.
To find features and documentation for fields in the spec please refer to the [specs](https://github.com/opencontainers/runtime-spec) repository.
#### Using runc with systemd
```bash
runc spec
```
To use runc with systemd, you can create a unit file
`/usr/lib/systemd/system/minecraft.service` as below (edit your
own Description or WorkingDirectory or service name as you need).
### Running Containers
Assuming you have an OCI bundle from the previous step you can execute the container in two different ways.
The first way is to use the convenience command `run` that will handle creating, starting, and deleting the container after it exits.
```bash
# run as root
cd /mycontainer
runc run mycontainerid
```
If you used the unmodified `runc spec` template this should give you a `sh` session inside the container.
The second way to start a container is using the specs lifecycle operations.
This gives you more power over how the container is created and managed while it is running.
This will also launch the container in the background so you will have to edit the `config.json` to remove the `terminal` setting for the simple examples here.
Your process field in the `config.json` should look like this below with `"terminal": false` and `"args": ["sleep", "5"]`.
```json
"process": {
"terminal": false,
"user": {
"uid": 0,
"gid": 0
},
"args": [
"sleep", "5"
],
"env": [
"PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
"TERM=xterm"
],
"cwd": "/",
"capabilities": {
"bounding": [
"CAP_AUDIT_WRITE",
"CAP_KILL",
"CAP_NET_BIND_SERVICE"
],
"effective": [
"CAP_AUDIT_WRITE",
"CAP_KILL",
"CAP_NET_BIND_SERVICE"
],
"inheritable": [
"CAP_AUDIT_WRITE",
"CAP_KILL",
"CAP_NET_BIND_SERVICE"
],
"permitted": [
"CAP_AUDIT_WRITE",
"CAP_KILL",
"CAP_NET_BIND_SERVICE"
],
"ambient": [
"CAP_AUDIT_WRITE",
"CAP_KILL",
"CAP_NET_BIND_SERVICE"
]
},
"rlimits": [
{
"type": "RLIMIT_NOFILE",
"hard": 1024,
"soft": 1024
}
],
"noNewPrivileges": true
},
```
Now we can go through the lifecycle operations in your shell.
```bash
# run as root
cd /mycontainer
runc create mycontainerid
# view the container is created and in the "created" state
runc list
# start the process inside the container
runc start mycontainerid
# after 5 seconds view that the container has exited and is now in the stopped state
runc list
# now delete the container
runc delete mycontainerid
```
This adds more complexity but allows higher level systems to manage runc and provides points in the containers creation to setup various settings after the container has created and/or before it is deleted.
This is commonly used to setup the container's network stack after `create` but before `start` where the user's defined process will be running.
#### Rootless containers
`runc` has the ability to run containers without root privileges. This is called `rootless`. You need to pass some parameters to `runc` in order to run rootless containers. See below and compare with the previous version. Run the following commands as an ordinary user:
```bash
# Same as the first example
mkdir ~/mycontainer
cd ~/mycontainer
mkdir rootfs
docker export $(docker create busybox) | tar -C rootfs -xvf -
# The --rootless parameter instructs runc spec to generate a configuration for a rootless container, which will allow you to run the container as a non-root user.
runc spec --rootless
# The --root parameter tells runc where to store the container state. It must be writable by the user.
runc --root /tmp/runc run mycontainerid
```
#### Supervisors
`runc` can be used with process supervisors and init systems to ensure that containers are restarted when they exit.
An example systemd unit file looks something like this.
```systemd
```service
[Unit]
Description=Start My Container
Description=Minecraft Build Server
Documentation=http://minecraft.net
After=network.target
[Service]
Type=forking
ExecStart=/usr/local/sbin/runc run -d --pid-file /run/mycontainerid.pid mycontainerid
ExecStopPost=/usr/local/sbin/runc delete mycontainerid
WorkingDirectory=/mycontainer
PIDFile=/run/mycontainerid.pid
CPUQuota=200%
MemoryLimit=1536M
ExecStart=/usr/local/bin/runc start minecraft
Restart=on-failure
WorkingDirectory=/containers/minecraftbuild
[Install]
WantedBy=multi-user.target
```
Make sure you have the bundle's root directory and JSON configs in
your WorkingDirectory, then use systemd commands to start the service:
```bash
systemctl daemon-reload
systemctl start minecraft.service
```
Note that if you use JSON configs by `runc spec`, you need to modify
`config.json` and change `process.terminal` to false so runc won't
create tty, because we can't set terminal from the stdin when using
systemd service.

View File

@@ -1 +0,0 @@
1.0.0-rc4+dev

View File

@@ -7,11 +7,8 @@ import (
"strconv"
"strings"
"github.com/codegangsta/cli"
"github.com/opencontainers/runc/libcontainer"
"github.com/opencontainers/runtime-spec/specs-go"
"github.com/urfave/cli"
"golang.org/x/sys/unix"
)
var checkpointCommand = cli.Command{
@@ -25,50 +22,27 @@ checkpointed.`,
Flags: []cli.Flag{
cli.StringFlag{Name: "image-path", Value: "", Usage: "path for saving criu image files"},
cli.StringFlag{Name: "work-path", Value: "", Usage: "path for saving work files and logs"},
cli.StringFlag{Name: "parent-path", Value: "", Usage: "path for previous criu image files in pre-dump"},
cli.BoolFlag{Name: "leave-running", Usage: "leave the process running after checkpointing"},
cli.BoolFlag{Name: "tcp-established", Usage: "allow open tcp connections"},
cli.BoolFlag{Name: "ext-unix-sk", Usage: "allow external unix sockets"},
cli.BoolFlag{Name: "shell-job", Usage: "allow shell jobs"},
cli.StringFlag{Name: "page-server", Value: "", Usage: "ADDRESS:PORT of the page server"},
cli.BoolFlag{Name: "file-locks", Usage: "handle file locks, for safety"},
cli.BoolFlag{Name: "pre-dump", Usage: "dump container's memory information only, leave the container running after this"},
cli.StringFlag{Name: "manage-cgroups-mode", Value: "", Usage: "cgroups mode: 'soft' (default), 'full' and 'strict'"},
cli.StringSliceFlag{Name: "empty-ns", Usage: "create a namespace, but don't restore its properties"},
cli.BoolFlag{Name: "auto-dedup", Usage: "enable auto deduplication of memory images"},
cli.StringFlag{Name: "manage-cgroups-mode", Value: "", Usage: "cgroups mode: 'soft' (default), 'full' and 'strict'."},
},
Action: func(context *cli.Context) error {
if err := checkArgs(context, 1, exactArgs); err != nil {
return err
}
// XXX: Currently this is untested with rootless containers.
if isRootless() {
return fmt.Errorf("runc checkpoint requires root")
}
Action: func(context *cli.Context) {
container, err := getContainer(context)
if err != nil {
return err
}
status, err := container.Status()
if err != nil {
return err
}
if status == libcontainer.Created {
fatalf("Container cannot be checkpointed in created state")
fatal(err)
}
defer destroy(container)
options := criuOptions(context)
// these are the mandatory criu options for a container
setPageServer(context, options)
setManageCgroupsMode(context, options)
if err := setEmptyNsMask(context, options); err != nil {
return err
}
if err := container.Checkpoint(options); err != nil {
return err
fatal(err)
}
return nil
},
}
@@ -113,22 +87,3 @@ func setManageCgroupsMode(context *cli.Context, options *libcontainer.CriuOpts)
}
}
}
var namespaceMapping = map[specs.LinuxNamespaceType]int{
specs.NetworkNamespace: unix.CLONE_NEWNET,
}
func setEmptyNsMask(context *cli.Context, options *libcontainer.CriuOpts) error {
var nsmask int
for _, ns := range context.StringSlice("empty-ns") {
f, exists := namespaceMapping[specs.LinuxNamespaceType(ns)]
if !exists {
return fmt.Errorf("namespace %q is not supported", ns)
}
nsmask |= f
}
options.EmptyNs = uint32(nsmask)
return nil
}

View File

@@ -1,236 +0,0 @@
/*
* Copyright 2016 SUSE LLC
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package main
import (
"fmt"
"io"
"io/ioutil"
"net"
"os"
"strings"
"github.com/opencontainers/runc/libcontainer"
"github.com/opencontainers/runc/libcontainer/utils"
"github.com/urfave/cli"
)
// version will be populated by the Makefile, read from
// VERSION file of the source code.
var version = ""
// gitCommit will be the hash that the binary was built from
// and will be populated by the Makefile
var gitCommit = ""
const (
usage = `Open Container Initiative contrib/cmd/recvtty
recvtty is a reference implementation of a consumer of runC's --console-socket
API. It has two main modes of operation:
* single: Only permit one terminal to be sent to the socket, which is
then hooked up to the stdio of the recvtty process. This is useful
for rudimentary shell management of a container.
* null: Permit as many terminals to be sent to the socket, but they
are read to /dev/null. This is used for testing, and imitates the
old runC API's --console=/dev/pts/ptmx hack which would allow for a
similar trick. This is probably not what you want to use, unless
you're doing something like our bats integration tests.
To use recvtty, just specify a socket path at which you want to receive
terminals:
$ recvtty [--mode <single|null>] socket.sock
`
)
func bail(err error) {
fmt.Fprintf(os.Stderr, "[recvtty] fatal error: %v\n", err)
os.Exit(1)
}
func handleSingle(path string) error {
// Open a socket.
ln, err := net.Listen("unix", path)
if err != nil {
return err
}
defer ln.Close()
// We only accept a single connection, since we can only really have
// one reader for os.Stdin. Plus this is all a PoC.
conn, err := ln.Accept()
if err != nil {
return err
}
defer conn.Close()
// Close ln, to allow for other instances to take over.
ln.Close()
// Get the fd of the connection.
unixconn, ok := conn.(*net.UnixConn)
if !ok {
return fmt.Errorf("failed to cast to unixconn")
}
socket, err := unixconn.File()
if err != nil {
return err
}
defer socket.Close()
// Get the master file descriptor from runC.
master, err := utils.RecvFd(socket)
if err != nil {
return err
}
if err = libcontainer.SaneTerminal(master); err != nil {
return err
}
// Copy from our stdio to the master fd.
quitChan := make(chan struct{})
go func() {
io.Copy(os.Stdout, master)
quitChan <- struct{}{}
}()
go func() {
io.Copy(master, os.Stdin)
quitChan <- struct{}{}
}()
// Only close the master fd once we've stopped copying.
<-quitChan
master.Close()
return nil
}
func handleNull(path string) error {
// Open a socket.
ln, err := net.Listen("unix", path)
if err != nil {
return err
}
defer ln.Close()
// As opposed to handleSingle we accept as many connections as we get, but
// we don't interact with Stdin at all (and we copy stdout to /dev/null).
for {
conn, err := ln.Accept()
if err != nil {
return err
}
go func(conn net.Conn) {
// Don't leave references lying around.
defer conn.Close()
// Get the fd of the connection.
unixconn, ok := conn.(*net.UnixConn)
if !ok {
return
}
socket, err := unixconn.File()
if err != nil {
return
}
defer socket.Close()
// Get the master file descriptor from runC.
master, err := utils.RecvFd(socket)
if err != nil {
return
}
// Just do a dumb copy to /dev/null.
devnull, err := os.OpenFile("/dev/null", os.O_RDWR, 0)
if err != nil {
// TODO: Handle this nicely.
return
}
io.Copy(devnull, master)
devnull.Close()
}(conn)
}
}
func main() {
app := cli.NewApp()
app.Name = "recvtty"
app.Usage = usage
// Set version to be the same as runC.
var v []string
if version != "" {
v = append(v, version)
}
if gitCommit != "" {
v = append(v, fmt.Sprintf("commit: %s", gitCommit))
}
app.Version = strings.Join(v, "\n")
// Set the flags.
app.Flags = []cli.Flag{
cli.StringFlag{
Name: "mode, m",
Value: "single",
Usage: "Mode of operation (single or null)",
},
cli.StringFlag{
Name: "pid-file",
Value: "",
Usage: "Path to write daemon process ID to",
},
}
app.Action = func(ctx *cli.Context) error {
args := ctx.Args()
if len(args) != 1 {
return fmt.Errorf("need to specify a single socket path")
}
path := ctx.Args()[0]
pidPath := ctx.String("pid-file")
if pidPath != "" {
pid := fmt.Sprintf("%d\n", os.Getpid())
if err := ioutil.WriteFile(pidPath, []byte(pid), 0644); err != nil {
return err
}
}
switch ctx.String("mode") {
case "single":
if err := handleSingle(path); err != nil {
return err
}
case "null":
if err := handleNull(path); err != nil {
return err
}
default:
return fmt.Errorf("need to select a valid mode: %s", ctx.String("mode"))
}
return nil
}
if err := app.Run(os.Args); err != nil {
bail(err)
}
}

View File

@@ -1,805 +0,0 @@
#!/bin/bash
#
# bash completion file for runc command
#
# This script provides completion of:
# - commands and their options
# - filepaths
#
# To enable the completions either:
# - place this file in /usr/share/bash-completion/completions
# or
# - copy this file to e.g. ~/.runc-completion.sh and add the line
# below to your .bashrc after bash completion features are loaded
# . ~/.runc-completion.sh
#
# Configuration:
#
# Note for developers:
# Please arrange options sorted alphabetically by long name with the short
# options immediately following their corresponding long form.
# This order should be applied to lists, alternatives and code blocks.
__runc_previous_extglob_setting=$(shopt -p extglob)
shopt -s extglob
__runc_list_all() {
COMPREPLY=($(compgen -W "$(runc list -q)" -- $cur))
}
__runc_pos_first_nonflag() {
local argument_flags=$1
local counter=$((${subcommand_pos:-${command_pos}} + 1))
while [ $counter -le $cword ]; do
if [ -n "$argument_flags" ] && eval "case '${words[$counter]}' in $argument_flags) true ;; *) false ;; esac"; then
((counter++))
else
case "${words[$counter]}" in
-*) ;;
*)
break
;;
esac
fi
((counter++))
done
echo $counter
}
# Transforms a multiline list of strings into a single line string
# with the words separated by "|".
# This is used to prepare arguments to __runc_pos_first_nonflag().
__runc_to_alternatives() {
local parts=($1)
local IFS='|'
echo "${parts[*]}"
}
# Transforms a multiline list of options into an extglob pattern
# suitable for use in case statements.
__runc_to_extglob() {
local extglob=$(__runc_to_alternatives "$1")
echo "@($extglob)"
}
# Subcommand processing.
# Locates the first occurrence of any of the subcommands contained in the
# first argument. In case of a match, calls the corresponding completion
# function and returns 0.
# If no match is found, 1 is returned. The calling function can then
# continue processing its completion.
#
# TODO if the preceding command has options that accept arguments and an
# argument is equal ot one of the subcommands, this is falsely detected as
# a match.
__runc_subcommands() {
local subcommands="$1"
local counter=$(($command_pos + 1))
while [ $counter -lt $cword ]; do
case "${words[$counter]}" in
$(__runc_to_extglob "$subcommands"))
subcommand_pos=$counter
local subcommand=${words[$counter]}
local completions_func=_runc_${command}_${subcommand}
declare -F $completions_func >/dev/null && $completions_func
return 0
;;
esac
((counter++))
done
return 1
}
# List all Signals
__runc_list_signals() {
COMPREPLY=($(compgen -W "$(for i in $(kill -l | xargs); do echo $i; done | grep SIG)"))
}
# suppress trailing whitespace
__runc_nospace() {
# compopt is not available in ancient bash versions
type compopt &>/dev/null && compopt -o nospace
}
# The list of capabilities is defined in types.go, ALL was added manually.
__runc_complete_capabilities() {
COMPREPLY=($(compgen -W "
ALL
AUDIT_CONTROL
AUDIT_WRITE
AUDIT_READ
BLOCK_SUSPEND
CHOWN
DAC_OVERRIDE
DAC_READ_SEARCH
FOWNER
FSETID
IPC_LOCK
IPC_OWNER
KILL
LEASE
LINUX_IMMUTABLE
MAC_ADMIN
MAC_OVERRIDE
MKNOD
NET_ADMIN
NET_BIND_SERVICE
NET_BROADCAST
NET_RAW
SETFCAP
SETGID
SETPCAP
SETUID
SYS_ADMIN
SYS_BOOT
SYS_CHROOT
SYSLOG
SYS_MODULE
SYS_NICE
SYS_PACCT
SYS_PTRACE
SYS_RAWIO
SYS_RESOURCE
SYS_TIME
SYS_TTY_CONFIG
WAKE_ALARM
" -- "$cur"))
}
_runc_exec() {
local boolean_options="
--help
--no-new-privs
--tty, -t
--detach, -d
"
local options_with_args="
--console
--cwd
--env, -e
--user, -u
--process, -p
--pid-file
--process-label
--apparmor
--cap, -c
"
local all_options="$options_with_args $boolean_options"
case "$prev" in
--cap | -c)
__runc_complete_capabilities
return
;;
--console | --cwd | --process | --apparmor)
case "$cur" in
*:*) ;; # TODO somehow do _filedir for stuff inside the image, if it's already specified (which is also somewhat difficult to determine)
'')
COMPREPLY=($(compgen -W '/' -- "$cur"))
__runc_nospace
;;
/*)
_filedir
__runc_nospace
;;
esac
return
;;
--env | -e)
COMPREPLY=($(compgen -e -- "$cur"))
__runc_nospace
return
;;
$(__runc_to_extglob "$options_with_args"))
return
;;
esac
case "$cur" in
-*)
COMPREPLY=($(compgen -W "$all_options" -- "$cur"))
;;
*)
__runc_list_all
;;
esac
}
# global options that may appear after the runc command
_runc_runc() {
local boolean_options="
$global_boolean_options
--help
--version -v
--debug
"
local options_with_args="
--log
--log-format
--root
--criu
"
case "$prev" in
--log | --root | --criu)
case "$cur" in
*:*) ;; # TODO somehow do _filedir for stuff inside the image, if it's already specified (which is also somewhat difficult to determine)
'')
COMPREPLY=($(compgen -W '/' -- "$cur"))
__runc_nospace
;;
*)
_filedir
__runc_nospace
;;
esac
return
;;
--log-format)
COMPREPLY=($(compgen -W 'text json' -- "$cur"))
return
;;
$(__runc_to_extglob "$options_with_args"))
return
;;
esac
case "$cur" in
-*)
COMPREPLY=($(compgen -W "$boolean_options $options_with_args" -- "$cur"))
;;
*)
local counter=$(__runc_pos_first_nonflag $(__runc_to_extglob "$options_with_args"))
if [ $cword -eq $counter ]; then
COMPREPLY=($(compgen -W "${commands[*]} help" -- "$cur"))
fi
;;
esac
}
_runc_pause() {
local boolean_options="
--help
-h
"
case "$cur" in
-*)
COMPREPLY=($(compgen -W "$boolean_options $options_with_args" -- "$cur"))
;;
*)
__runc_list_all
;;
esac
}
_runc_ps() {
local boolean_options="
--help
-h
"
case "$cur" in
-*)
COMPREPLY=($(compgen -W "$boolean_options $options_with_args" -- "$cur"))
;;
*)
__runc_list_all
;;
esac
}
_runc_delete() {
local boolean_options="
--help
-h
"
case "$cur" in
-*)
COMPREPLY=($(compgen -W "$boolean_options $options_with_args" -- "$cur"))
;;
*)
__runc_list_all
;;
esac
}
_runc_kill() {
local boolean_options="
--help
-h
--all
-a
"
case "$prev" in
"kill")
__runc_list_all
return
;;
*)
__runc_list_signals
return
;;
esac
case "$cur" in
-*)
COMPREPLY=($(compgen -W "$boolean_options $options_with_args" -- "$cur"))
;;
*)
__runc_list_all
;;
esac
}
_runc_events() {
local boolean_options="
--help
--stats
"
local options_with_args="
--interval
"
case "$prev" in
$(__runc_to_extglob "$options_with_args"))
return
;;
esac
case "$cur" in
-*)
COMPREPLY=($(compgen -W "$boolean_options $options_with_args" -- "$cur"))
;;
*)
__runc_list_all
;;
esac
}
_runc_list() {
local boolean_options="
--help
--quiet
-q
"
local options_with_args="
--format
-f
"
case "$prev" in
--format | -f)
COMPREPLY=($(compgen -W 'text json' -- "$cur"))
return
;;
$(__runc_to_extglob "$options_with_args"))
return
;;
esac
case "$cur" in
-*)
COMPREPLY=($(compgen -W "$boolean_options $options_with_args" -- "$cur"))
;;
*)
local counter=$(__runc_pos_first_nonflag $(__runc_to_extglob "$options_with_args"))
;;
esac
}
_runc_spec() {
local boolean_options="
--help
"
local options_with_args="
--bundle
-b
"
case "$prev" in
--bundle | -b)
case "$cur" in
'')
COMPREPLY=($(compgen -W '/' -- "$cur"))
__runc_nospace
;;
/*)
_filedir
__runc_nospace
;;
esac
return
;;
$(__runc_to_extglob "$options_with_args"))
return
;;
esac
case "$cur" in
-*)
COMPREPLY=($(compgen -W "$boolean_options $options_with_args" -- "$cur"))
;;
*)
local counter=$(__runc_pos_first_nonflag $(__runc_to_extglob "$options_with_args"))
;;
esac
}
_runc_run() {
local boolean_options="
--help
--detatch
-d
--no-subreaper
--no-pivot
--no-new-keyring
"
local options_with_args="
--bundle
-b
--console
--pid-file
"
case "$prev" in
--bundle | -b | --console | --pid-file)
case "$cur" in
'')
COMPREPLY=($(compgen -W '/' -- "$cur"))
__runc_nospace
;;
/*)
_filedir
__runc_nospace
;;
esac
return
;;
$(__runc_to_extglob "$options_with_args"))
return
;;
esac
case "$cur" in
-*)
COMPREPLY=($(compgen -W "$boolean_options $options_with_args" -- "$cur"))
;;
*)
__runc_list_all
;;
esac
}
_runc_checkpoint() {
local boolean_options="
--help
-h
--leave-running
--tcp-established
--ext-unix-sk
--shell-job
--file-locks
"
local options_with_args="
--image-path
--work-path
--page-server
--manage-cgroups-mode
"
case "$prev" in
--page-server) ;;
--manage-cgroups-mode)
COMPREPLY=($(compgen -W "soft full strict" -- "$cur"))
return
;;
--image-path | --work-path)
case "$cur" in
*:*) ;; # TODO somehow do _filedir for stuff inside the image, if it's already specified (which is also somewhat difficult to determine)
'')
COMPREPLY=($(compgen -W '/' -- "$cur"))
__runc_nospace
;;
*)
_filedir
__runc_nospace
;;
esac
return
;;
$(__runc_to_extglob "$options_with_args"))
return
;;
esac
case "$cur" in
-*)
COMPREPLY=($(compgen -W "$boolean_options $options_with_args" -- "$cur"))
;;
*)
__runc_list_all
;;
esac
}
_runc_create() {
local boolean_options="
--help
--no-pivot
--no-new-keyring
"
local options_with_args="
--bundle
-b
--console
--pid-file
"
case "$prev" in
--bundle | -b | --console | --pid-file)
case "$cur" in
'')
COMPREPLY=($(compgen -W '/' -- "$cur"))
__runc_nospace
;;
/*)
_filedir
__runc_nospace
;;
esac
return
;;
$(__runc_to_extglob "$options_with_args"))
return
;;
esac
case "$cur" in
-*)
COMPREPLY=($(compgen -W "$boolean_options $options_with_args" -- "$cur"))
;;
*)
__runc_list_all
;;
esac
}
_runc_help() {
local counter=$(__runc_pos_first_nonflag)
if [ $cword -eq $counter ]; then
COMPREPLY=($(compgen -W "${commands[*]}" -- "$cur"))
fi
}
_runc_restore() {
local boolean_options="
--help
--tcp-established
--ext-unix-sk
--shell-job
--file-locks
--detach
-d
--no-subreaper
--no-pivot
"
local options_with_args="
-b
--bundle
--image-path
--work-path
--manage-cgroups-mode
--pid-file
"
local all_options="$options_with_args $boolean_options"
case "$prev" in
--manage-cgroups-mode)
COMPREPLY=($(compgen -W "soft full strict" -- "$cur"))
return
;;
--pid-file | --image-path | --work-path | --bundle | -b)
case "$cur" in
*:*) ;; # TODO somehow do _filedir for stuff inside the image, if it's already specified (which is also somewhat difficult to determine)
'')
COMPREPLY=($(compgen -W '/' -- "$cur"))
__runc_nospace
;;
/*)
_filedir
__runc_nospace
;;
esac
return
;;
$(__runc_to_extglob "$options_with_args"))
return
;;
esac
case "$cur" in
-*)
COMPREPLY=($(compgen -W "$all_options" -- "$cur"))
;;
*)
__runc_list_all
;;
esac
}
_runc_resume() {
local boolean_options="
--help
-h
"
case "$cur" in
-*)
COMPREPLY=($(compgen -W "$boolean_options $options_with_args" -- "$cur"))
;;
*)
__runc_list_all
;;
esac
}
_runc_state() {
local boolean_options="
--help
-h
"
case "$cur" in
-*)
COMPREPLY=($(compgen -W "$boolean_options $options_with_args" -- "$cur"))
;;
*)
__runc_list_all
;;
esac
}
_runc_start() {
local boolean_options="
--help
-h
"
case "$cur" in
-*)
COMPREPLY=($(compgen -W "$boolean_options $options_with_args" -- "$cur"))
;;
*)
__runc_list_all
;;
esac
}
_runc_update() {
local boolean_options="
--help
"
local options_with_args="
--blkio-weight
--cpu-period
--cpu-quota
--cpu-rt-period
--cpu-rt-runtime
--cpu-share
--cpuset-cpus
--cpuset-mems
--kernel-memory
--kernel-memory-tcp
--memory
--memory-reservation
--memory-swap
"
case "$prev" in
$(__runc_to_extglob "$options_with_args"))
return
;;
esac
case "$cur" in
-*)
COMPREPLY=($(compgen -W "$boolean_options $options_with_args" -- "$cur"))
;;
*)
__runc_list_all
;;
esac
}
_runc() {
local previous_extglob_setting=$(shopt -p extglob)
shopt -s extglob
local commands=(
checkpoint
create
delete
events
exec
init
kill
list
pause
ps
restore
resume
run
spec
start
state
update
help
h
)
# These options are valid as global options for all client commands
# and valid as command options for `runc daemon`
local global_boolean_options="
--help -h
--version -v
"
COMPREPLY=()
local cur prev words cword
_get_comp_words_by_ref -n : cur prev words cword
local command='runc' command_pos=0 subcommand_pos
local counter=1
while [ $counter -lt $cword ]; do
case "${words[$counter]}" in
-*) ;;
=)
((counter++))
;;
*)
command="${words[$counter]}"
command_pos=$counter
break
;;
esac
((counter++))
done
local completions_func=_runc_${command}
declare -F $completions_func >/dev/null && $completions_func
eval "$previous_extglob_setting"
return 0
}
eval "$__runc_previous_extglob_setting"
unset __runc_previous_extglob_setting
complete -F _runc runc

View File

@@ -1,74 +0,0 @@
package main
import (
"os"
"github.com/urfave/cli"
)
var createCommand = cli.Command{
Name: "create",
Usage: "create a container",
ArgsUsage: `<container-id>
Where "<container-id>" is your name for the instance of the container that you
are starting. The name you provide for the container instance must be unique on
your host.`,
Description: `The create command creates an instance of a container for a bundle. The bundle
is a directory with a specification file named "` + specConfig + `" and a root
filesystem.
The specification file includes an args parameter. The args parameter is used
to specify command(s) that get run when the container is started. To change the
command(s) that get executed on start, edit the args parameter of the spec. See
"runc spec --help" for more explanation.`,
Flags: []cli.Flag{
cli.StringFlag{
Name: "bundle, b",
Value: "",
Usage: `path to the root of the bundle directory, defaults to the current directory`,
},
cli.StringFlag{
Name: "console-socket",
Value: "",
Usage: "path to an AF_UNIX socket which will receive a file descriptor referencing the master end of the console's pseudoterminal",
},
cli.StringFlag{
Name: "pid-file",
Value: "",
Usage: "specify the file to write the process id to",
},
cli.BoolFlag{
Name: "no-pivot",
Usage: "do not use pivot root to jail process inside rootfs. This should be used whenever the rootfs is on top of a ramdisk",
},
cli.BoolFlag{
Name: "no-new-keyring",
Usage: "do not create a new session keyring for the container. This will cause the container to inherit the calling processes session key",
},
cli.IntFlag{
Name: "preserve-fds",
Usage: "Pass N additional file descriptors to the container (stdio + $LISTEN_FDS + N in total)",
},
},
Action: func(context *cli.Context) error {
if err := checkArgs(context, 1, exactArgs); err != nil {
return err
}
if err := revisePidFile(context); err != nil {
return err
}
spec, err := setupSpec(context)
if err != nil {
return err
}
status, err := startContainer(context, spec, CT_ACT_CREATE, nil)
if err != nil {
return err
}
// exit with the container's exit status so any external supervisor is
// notified of the exit with the correct exit status.
os.Exit(status)
return nil
},
}

View File

@@ -1,90 +1,38 @@
// +build !solaris
package main
import (
"fmt"
"os"
"path/filepath"
"syscall"
"time"
"github.com/codegangsta/cli"
"github.com/opencontainers/runc/libcontainer"
"github.com/urfave/cli"
"golang.org/x/sys/unix"
)
func killContainer(container libcontainer.Container) error {
_ = container.Signal(unix.SIGKILL, false)
for i := 0; i < 100; i++ {
time.Sleep(100 * time.Millisecond)
if err := container.Signal(syscall.Signal(0), false); err != nil {
destroy(container)
return nil
}
}
return fmt.Errorf("container init still running")
}
var deleteCommand = cli.Command{
Name: "delete",
Usage: "delete any resources held by the container often used with detached container",
Usage: "delete any resources held by the container often used with detached containers",
ArgsUsage: `<container-id>
Where "<container-id>" is the name for the instance of the container.
EXAMPLE:
For example, if the container id is "ubuntu01" and runc list currently shows the
status of "ubuntu01" as "stopped" the following will delete resources held for
"ubuntu01" removing "ubuntu01" from the runc list of containers:
status of "ubuntu01" as "destroyed" the following will delete resources held for
"ubuntu01" removing "ubuntu01" from the runc list of containers:
# runc delete ubuntu01`,
Flags: []cli.Flag{
cli.BoolFlag{
Name: "force, f",
Usage: "Forcibly deletes the container if it is still running (uses SIGKILL)",
},
},
Action: func(context *cli.Context) error {
if err := checkArgs(context, 1, exactArgs); err != nil {
return err
}
id := context.Args().First()
force := context.Bool("force")
Action: func(context *cli.Context) {
container, err := getContainer(context)
if err != nil {
if lerr, ok := err.(libcontainer.Error); ok && lerr.Code() == libcontainer.ContainerNotExists {
// if there was an aborted start or something of the sort then the container's directory could exist but
// libcontainer does not see it because the state.json file inside that directory was never created.
path := filepath.Join(context.GlobalString("root"), id)
if e := os.RemoveAll(path); e != nil {
fmt.Fprintf(os.Stderr, "remove %s: %v\n", path, e)
}
if force {
return nil
path := filepath.Join(context.GlobalString("root"), context.Args().First())
if err := os.RemoveAll(path); err == nil {
return
}
}
return err
fatal(err)
}
s, err := container.Status()
if err != nil {
return err
}
switch s {
case libcontainer.Stopped:
destroy(container)
case libcontainer.Created:
return killContainer(container)
default:
if force {
return killContainer(container)
} else {
return fmt.Errorf("cannot delete container %s that is not stopped: %s\n", id, s)
}
}
return nil
destroy(container)
},
}

View File

@@ -4,16 +4,13 @@ package main
import (
"encoding/json"
"fmt"
"os"
"sync"
"time"
"github.com/Sirupsen/logrus"
"github.com/codegangsta/cli"
"github.com/opencontainers/runc/libcontainer"
"github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/sirupsen/logrus"
"github.com/urfave/cli"
)
// event struct for encoding the event data to json.
@@ -23,82 +20,9 @@ type event struct {
Data interface{} `json:"data,omitempty"`
}
// stats is the runc specific stats structure for stability when encoding and decoding stats.
type stats struct {
CPU cpu `json:"cpu"`
Memory memory `json:"memory"`
Pids pids `json:"pids"`
Blkio blkio `json:"blkio"`
Hugetlb map[string]hugetlb `json:"hugetlb"`
}
type hugetlb struct {
Usage uint64 `json:"usage,omitempty"`
Max uint64 `json:"max,omitempty"`
Failcnt uint64 `json:"failcnt"`
}
type blkioEntry struct {
Major uint64 `json:"major,omitempty"`
Minor uint64 `json:"minor,omitempty"`
Op string `json:"op,omitempty"`
Value uint64 `json:"value,omitempty"`
}
type blkio struct {
IoServiceBytesRecursive []blkioEntry `json:"ioServiceBytesRecursive,omitempty"`
IoServicedRecursive []blkioEntry `json:"ioServicedRecursive,omitempty"`
IoQueuedRecursive []blkioEntry `json:"ioQueueRecursive,omitempty"`
IoServiceTimeRecursive []blkioEntry `json:"ioServiceTimeRecursive,omitempty"`
IoWaitTimeRecursive []blkioEntry `json:"ioWaitTimeRecursive,omitempty"`
IoMergedRecursive []blkioEntry `json:"ioMergedRecursive,omitempty"`
IoTimeRecursive []blkioEntry `json:"ioTimeRecursive,omitempty"`
SectorsRecursive []blkioEntry `json:"sectorsRecursive,omitempty"`
}
type pids struct {
Current uint64 `json:"current,omitempty"`
Limit uint64 `json:"limit,omitempty"`
}
type throttling struct {
Periods uint64 `json:"periods,omitempty"`
ThrottledPeriods uint64 `json:"throttledPeriods,omitempty"`
ThrottledTime uint64 `json:"throttledTime,omitempty"`
}
type cpuUsage struct {
// Units: nanoseconds.
Total uint64 `json:"total,omitempty"`
Percpu []uint64 `json:"percpu,omitempty"`
Kernel uint64 `json:"kernel"`
User uint64 `json:"user"`
}
type cpu struct {
Usage cpuUsage `json:"usage,omitempty"`
Throttling throttling `json:"throttling,omitempty"`
}
type memoryEntry struct {
Limit uint64 `json:"limit"`
Usage uint64 `json:"usage,omitempty"`
Max uint64 `json:"max,omitempty"`
Failcnt uint64 `json:"failcnt"`
}
type memory struct {
Cache uint64 `json:"cache,omitempty"`
Usage memoryEntry `json:"usage,omitempty"`
Swap memoryEntry `json:"swap,omitempty"`
Kernel memoryEntry `json:"kernel,omitempty"`
KernelTCP memoryEntry `json:"kernelTCP,omitempty"`
Raw map[string]uint64 `json:"raw,omitempty"`
}
var eventsCommand = cli.Command{
Name: "events",
Usage: "display container events such as OOM notifications, cpu, memory, and IO usage statistics",
Usage: "display container events such as OOM notifications, cpu, memory, IO and network stats",
ArgsUsage: `<container-id>
Where "<container-id>" is the name for the instance of the container.`,
@@ -108,24 +32,10 @@ information is displayed once every 5 seconds.`,
cli.DurationFlag{Name: "interval", Value: 5 * time.Second, Usage: "set the stats collection interval"},
cli.BoolFlag{Name: "stats", Usage: "display the container's stats then exit"},
},
Action: func(context *cli.Context) error {
if err := checkArgs(context, 1, exactArgs); err != nil {
return err
}
Action: func(context *cli.Context) {
container, err := getContainer(context)
if err != nil {
return err
}
duration := context.Duration("interval")
if duration <= 0 {
return fmt.Errorf("duration interval must be greater than 0")
}
status, err := container.Status()
if err != nil {
return err
}
if status == libcontainer.Stopped {
return fmt.Errorf("container with id %s is not running", container.ID())
fatal(err)
}
var (
stats = make(chan *libcontainer.Stats, 1)
@@ -145,12 +55,12 @@ information is displayed once every 5 seconds.`,
if context.Bool("stats") {
s, err := container.Stats()
if err != nil {
return err
fatal(err)
}
events <- &event{Type: "stats", ID: container.ID(), Data: convertLibcontainerStats(s)}
events <- &event{Type: "stats", ID: container.ID(), Data: s}
close(events)
group.Wait()
return nil
return
}
go func() {
for range time.Tick(context.Duration("interval")) {
@@ -164,7 +74,7 @@ information is displayed once every 5 seconds.`,
}()
n, err := container.NotifyOOM()
if err != nil {
return err
fatal(err)
}
for {
select {
@@ -178,7 +88,7 @@ information is displayed once every 5 seconds.`,
n = nil
}
case s := <-stats:
events <- &event{Type: "stats", ID: container.ID(), Data: convertLibcontainerStats(s)}
events <- &event{Type: "stats", ID: container.ID(), Data: s}
}
if n == nil {
close(events)
@@ -186,76 +96,5 @@ information is displayed once every 5 seconds.`,
}
}
group.Wait()
return nil
},
}
func convertLibcontainerStats(ls *libcontainer.Stats) *stats {
cg := ls.CgroupStats
if cg == nil {
return nil
}
var s stats
s.Pids.Current = cg.PidsStats.Current
s.Pids.Limit = cg.PidsStats.Limit
s.CPU.Usage.Kernel = cg.CpuStats.CpuUsage.UsageInKernelmode
s.CPU.Usage.User = cg.CpuStats.CpuUsage.UsageInUsermode
s.CPU.Usage.Total = cg.CpuStats.CpuUsage.TotalUsage
s.CPU.Usage.Percpu = cg.CpuStats.CpuUsage.PercpuUsage
s.CPU.Throttling.Periods = cg.CpuStats.ThrottlingData.Periods
s.CPU.Throttling.ThrottledPeriods = cg.CpuStats.ThrottlingData.ThrottledPeriods
s.CPU.Throttling.ThrottledTime = cg.CpuStats.ThrottlingData.ThrottledTime
s.Memory.Cache = cg.MemoryStats.Cache
s.Memory.Kernel = convertMemoryEntry(cg.MemoryStats.KernelUsage)
s.Memory.KernelTCP = convertMemoryEntry(cg.MemoryStats.KernelTCPUsage)
s.Memory.Swap = convertMemoryEntry(cg.MemoryStats.SwapUsage)
s.Memory.Usage = convertMemoryEntry(cg.MemoryStats.Usage)
s.Memory.Raw = cg.MemoryStats.Stats
s.Blkio.IoServiceBytesRecursive = convertBlkioEntry(cg.BlkioStats.IoServiceBytesRecursive)
s.Blkio.IoServicedRecursive = convertBlkioEntry(cg.BlkioStats.IoServicedRecursive)
s.Blkio.IoQueuedRecursive = convertBlkioEntry(cg.BlkioStats.IoQueuedRecursive)
s.Blkio.IoServiceTimeRecursive = convertBlkioEntry(cg.BlkioStats.IoServiceTimeRecursive)
s.Blkio.IoWaitTimeRecursive = convertBlkioEntry(cg.BlkioStats.IoWaitTimeRecursive)
s.Blkio.IoMergedRecursive = convertBlkioEntry(cg.BlkioStats.IoMergedRecursive)
s.Blkio.IoTimeRecursive = convertBlkioEntry(cg.BlkioStats.IoTimeRecursive)
s.Blkio.SectorsRecursive = convertBlkioEntry(cg.BlkioStats.SectorsRecursive)
s.Hugetlb = make(map[string]hugetlb)
for k, v := range cg.HugetlbStats {
s.Hugetlb[k] = convertHugtlb(v)
}
return &s
}
func convertHugtlb(c cgroups.HugetlbStats) hugetlb {
return hugetlb{
Usage: c.Usage,
Max: c.MaxUsage,
Failcnt: c.Failcnt,
}
}
func convertMemoryEntry(c cgroups.MemoryData) memoryEntry {
return memoryEntry{
Limit: c.Limit,
Usage: c.Usage,
Max: c.MaxUsage,
Failcnt: c.Failcnt,
}
}
func convertBlkioEntry(c []cgroups.BlkioStatEntry) []blkioEntry {
var out []blkioEntry
for _, e := range c {
out = append(out, blkioEntry{
Major: e.Major,
Minor: e.Minor,
Op: e.Op,
Value: e.Value,
})
}
return out
}

View File

@@ -9,30 +9,27 @@ import (
"strconv"
"strings"
"github.com/opencontainers/runc/libcontainer"
"github.com/codegangsta/cli"
"github.com/opencontainers/runc/libcontainer/utils"
"github.com/opencontainers/runtime-spec/specs-go"
"github.com/urfave/cli"
)
var execCommand = cli.Command{
Name: "exec",
Usage: "execute new process inside the container",
ArgsUsage: `<container-id> <command> [command options] || -p process.json <container-id>
ArgsUsage: `<container-id> <container command>
Where "<container-id>" is the name for the instance of the container and
"<command>" is the command to be executed in the container.
"<command>" can't be empty unless a "-p" flag provided.
"<container command>" is the command to be executed in the container.
EXAMPLE:
For example, if the container is configured to run the linux ps command the
following will output a list of processes running in the container:
# runc exec <container-id> ps`,
Flags: []cli.Flag{
cli.StringFlag{
Name: "console-socket",
Usage: "path to an AF_UNIX socket which will receive a file descriptor referencing the master end of the console's pseudoterminal",
Name: "console",
Usage: "specify the pty slave path for use with the container",
},
cli.StringFlag{
Name: "cwd",
@@ -81,25 +78,20 @@ following will output a list of processes running in the container:
Usage: "add a capability to the bounding set for the process",
},
cli.BoolFlag{
Name: "no-subreaper",
Usage: "disable the use of the subreaper used to reap reparented processes",
Hidden: true,
Name: "no-subreaper",
Usage: "disable the use of the subreaper used to reap reparented processes",
},
},
Action: func(context *cli.Context) error {
if err := checkArgs(context, 1, minArgs); err != nil {
return err
}
if err := revisePidFile(context); err != nil {
return err
Action: func(context *cli.Context) {
if os.Geteuid() != 0 {
fatalf("runc should be run as root")
}
status, err := execProcess(context)
if err == nil {
os.Exit(status)
if err != nil {
fatalf("exec failed: %v", err)
}
return fmt.Errorf("exec failed: %v", err)
os.Exit(status)
},
SkipArgReorder: true,
}
func execProcess(context *cli.Context) (int, error) {
@@ -107,17 +99,6 @@ func execProcess(context *cli.Context) (int, error) {
if err != nil {
return -1, err
}
status, err := container.Status()
if err != nil {
return -1, err
}
if status == libcontainer.Stopped {
return -1, fmt.Errorf("cannot exec a container that has stopped")
}
path := context.String("process")
if path == "" && len(context.Args()) == 1 {
return -1, fmt.Errorf("process args cannot be empty")
}
detach := context.Bool("detach")
state, err := container.State()
if err != nil {
@@ -129,13 +110,12 @@ func execProcess(context *cli.Context) (int, error) {
return -1, err
}
r := &runner{
enableSubreaper: false,
enableSubreaper: !context.Bool("no-subreaper"),
shouldDestroy: false,
container: container,
consoleSocket: context.String("console-socket"),
console: context.String("console"),
detach: detach,
pidFile: context.String("pid-file"),
action: CT_ACT_RUN,
}
return r.run(p)
}
@@ -174,17 +154,12 @@ func getProcess(context *cli.Context, bundle string) (*specs.Process, error) {
p.SelinuxLabel = l
}
if caps := context.StringSlice("cap"); len(caps) > 0 {
for _, c := range caps {
p.Capabilities.Bounding = append(p.Capabilities.Bounding, c)
p.Capabilities.Inheritable = append(p.Capabilities.Inheritable, c)
p.Capabilities.Effective = append(p.Capabilities.Effective, c)
p.Capabilities.Permitted = append(p.Capabilities.Permitted, c)
p.Capabilities.Ambient = append(p.Capabilities.Ambient, c)
}
p.Capabilities = caps
}
// append the passed env variables
p.Env = append(p.Env, context.StringSlice("env")...)
for _, e := range context.StringSlice("env") {
p.Env = append(p.Env, e)
}
// set the tty
if context.IsSet("tty") {
p.Terminal = context.Bool("tty")
@@ -208,5 +183,5 @@ func getProcess(context *cli.Context, bundle string) (*specs.Process, error) {
}
p.User.UID = uint32(uid)
}
return p, nil
return &p, nil
}

View File

@@ -1,31 +0,0 @@
package main
import (
"os"
"runtime"
"github.com/opencontainers/runc/libcontainer"
_ "github.com/opencontainers/runc/libcontainer/nsenter"
"github.com/urfave/cli"
)
func init() {
if len(os.Args) > 1 && os.Args[1] == "init" {
runtime.GOMAXPROCS(1)
runtime.LockOSThread()
}
}
var initCommand = cli.Command{
Name: "init",
Usage: `initialize the namespaces and launch the process (do not call it outside of runc)`,
Action: func(context *cli.Context) error {
factory, _ := libcontainer.New("")
if err := factory.StartInitialization(); err != nil {
// as the error is sent back to the parent there is no need to log
// or write it to stderr because the parent process will handle this
os.Exit(1)
}
panic("libcontainer: container init failed to exec")
},
}

View File

@@ -8,78 +8,63 @@ import (
"strings"
"syscall"
"golang.org/x/sys/unix"
"github.com/urfave/cli"
"github.com/codegangsta/cli"
)
var signalMap = map[string]syscall.Signal{
"ABRT": unix.SIGABRT,
"ALRM": unix.SIGALRM,
"BUS": unix.SIGBUS,
"CHLD": unix.SIGCHLD,
"CLD": unix.SIGCLD,
"CONT": unix.SIGCONT,
"FPE": unix.SIGFPE,
"HUP": unix.SIGHUP,
"ILL": unix.SIGILL,
"INT": unix.SIGINT,
"IO": unix.SIGIO,
"IOT": unix.SIGIOT,
"KILL": unix.SIGKILL,
"PIPE": unix.SIGPIPE,
"POLL": unix.SIGPOLL,
"PROF": unix.SIGPROF,
"PWR": unix.SIGPWR,
"QUIT": unix.SIGQUIT,
"SEGV": unix.SIGSEGV,
"STKFLT": unix.SIGSTKFLT,
"STOP": unix.SIGSTOP,
"SYS": unix.SIGSYS,
"TERM": unix.SIGTERM,
"TRAP": unix.SIGTRAP,
"TSTP": unix.SIGTSTP,
"TTIN": unix.SIGTTIN,
"TTOU": unix.SIGTTOU,
"UNUSED": unix.SIGUNUSED,
"URG": unix.SIGURG,
"USR1": unix.SIGUSR1,
"USR2": unix.SIGUSR2,
"VTALRM": unix.SIGVTALRM,
"WINCH": unix.SIGWINCH,
"XCPU": unix.SIGXCPU,
"XFSZ": unix.SIGXFSZ,
"ABRT": syscall.SIGABRT,
"ALRM": syscall.SIGALRM,
"BUS": syscall.SIGBUS,
"CHLD": syscall.SIGCHLD,
"CLD": syscall.SIGCLD,
"CONT": syscall.SIGCONT,
"FPE": syscall.SIGFPE,
"HUP": syscall.SIGHUP,
"ILL": syscall.SIGILL,
"INT": syscall.SIGINT,
"IO": syscall.SIGIO,
"IOT": syscall.SIGIOT,
"KILL": syscall.SIGKILL,
"PIPE": syscall.SIGPIPE,
"POLL": syscall.SIGPOLL,
"PROF": syscall.SIGPROF,
"PWR": syscall.SIGPWR,
"QUIT": syscall.SIGQUIT,
"SEGV": syscall.SIGSEGV,
"STKFLT": syscall.SIGSTKFLT,
"STOP": syscall.SIGSTOP,
"SYS": syscall.SIGSYS,
"TERM": syscall.SIGTERM,
"TRAP": syscall.SIGTRAP,
"TSTP": syscall.SIGTSTP,
"TTIN": syscall.SIGTTIN,
"TTOU": syscall.SIGTTOU,
"UNUSED": syscall.SIGUNUSED,
"URG": syscall.SIGURG,
"USR1": syscall.SIGUSR1,
"USR2": syscall.SIGUSR2,
"VTALRM": syscall.SIGVTALRM,
"WINCH": syscall.SIGWINCH,
"XCPU": syscall.SIGXCPU,
"XFSZ": syscall.SIGXFSZ,
}
var killCommand = cli.Command{
Name: "kill",
Usage: "kill sends the specified signal (default: SIGTERM) to the container's init process",
ArgsUsage: `<container-id> [signal]
ArgsUsage: `<container-id> <signal>
Where "<container-id>" is the name for the instance of the container and
"[signal]" is the signal to be sent to the init process.
EXAMPLE:
"<signal>" is the signal to be sent to the init process.
For example, if the container id is "ubuntu01" the following will send a "KILL"
signal to the init process of the "ubuntu01" container:
# runc kill ubuntu01 KILL`,
Flags: []cli.Flag{
cli.BoolFlag{
Name: "all, a",
Usage: "send the specified signal to all processes inside the container",
},
},
Action: func(context *cli.Context) error {
if err := checkArgs(context, 1, minArgs); err != nil {
return err
}
if err := checkArgs(context, 2, maxArgs); err != nil {
return err
}
Action: func(context *cli.Context) {
container, err := getContainer(context)
if err != nil {
return err
fatal(err)
}
sigstr := context.Args().Get(1)
@@ -89,25 +74,19 @@ signal to the init process of the "ubuntu01" container:
signal, err := parseSignal(sigstr)
if err != nil {
return err
fatal(err)
}
if err := container.Signal(signal, context.Bool("all")); err != nil {
return err
if err := container.Signal(signal); err != nil {
fatal(err)
}
return nil
},
}
func parseSignal(rawSignal string) (syscall.Signal, error) {
s, err := strconv.Atoi(rawSignal)
if err == nil {
sig := syscall.Signal(s)
for _, msig := range signalMap {
if sig == msig {
return sig, nil
}
}
return -1, fmt.Errorf("unknown signal %q", rawSignal)
return syscall.Signal(s), nil
}
signal, ok := signalMap[strings.TrimPrefix(strings.ToUpper(rawSignal), "SIG")]
if !ok {

View File

@@ -1,7 +1,3 @@
# libcontainer
[![GoDoc](https://godoc.org/github.com/opencontainers/runc/libcontainer?status.svg)](https://godoc.org/github.com/opencontainers/runc/libcontainer)
Libcontainer provides a native Go implementation for creating containers
with namespaces, cgroups, capabilities, and filesystem access controls.
It allows you to manage the lifecycle of the container performing additional operations
@@ -20,14 +16,7 @@ the current binary (/proc/self/exe) to be executed as the init process, and use
arg "init", we call the first step process "bootstrap", so you always need a "init"
function as the entry of "bootstrap".
In addition to the go init function the early stage bootstrap is handled by importing
[nsenter](https://github.com/opencontainers/runc/blob/master/libcontainer/nsenter/README.md).
```go
import (
_ "github.com/opencontainers/runc/libcontainer/nsenter"
)
func init() {
if len(os.Args) > 1 && os.Args[1] == "init" {
runtime.GOMAXPROCS(1)
@@ -56,91 +45,25 @@ Once you have an instance of the factory created we can create a configuration
struct describing how the container is to be created. A sample would look similar to this:
```go
defaultMountFlags := unix.MS_NOEXEC | unix.MS_NOSUID | unix.MS_NODEV
defaultMountFlags := syscall.MS_NOEXEC | syscall.MS_NOSUID | syscall.MS_NODEV
config := &configs.Config{
Rootfs: "/your/path/to/rootfs",
Capabilities: &configs.Capabilities{
Bounding: []string{
"CAP_CHOWN",
"CAP_DAC_OVERRIDE",
"CAP_FSETID",
"CAP_FOWNER",
"CAP_MKNOD",
"CAP_NET_RAW",
"CAP_SETGID",
"CAP_SETUID",
"CAP_SETFCAP",
"CAP_SETPCAP",
"CAP_NET_BIND_SERVICE",
"CAP_SYS_CHROOT",
"CAP_KILL",
"CAP_AUDIT_WRITE",
},
Effective: []string{
"CAP_CHOWN",
"CAP_DAC_OVERRIDE",
"CAP_FSETID",
"CAP_FOWNER",
"CAP_MKNOD",
"CAP_NET_RAW",
"CAP_SETGID",
"CAP_SETUID",
"CAP_SETFCAP",
"CAP_SETPCAP",
"CAP_NET_BIND_SERVICE",
"CAP_SYS_CHROOT",
"CAP_KILL",
"CAP_AUDIT_WRITE",
},
Inheritable: []string{
"CAP_CHOWN",
"CAP_DAC_OVERRIDE",
"CAP_FSETID",
"CAP_FOWNER",
"CAP_MKNOD",
"CAP_NET_RAW",
"CAP_SETGID",
"CAP_SETUID",
"CAP_SETFCAP",
"CAP_SETPCAP",
"CAP_NET_BIND_SERVICE",
"CAP_SYS_CHROOT",
"CAP_KILL",
"CAP_AUDIT_WRITE",
},
Permitted: []string{
"CAP_CHOWN",
"CAP_DAC_OVERRIDE",
"CAP_FSETID",
"CAP_FOWNER",
"CAP_MKNOD",
"CAP_NET_RAW",
"CAP_SETGID",
"CAP_SETUID",
"CAP_SETFCAP",
"CAP_SETPCAP",
"CAP_NET_BIND_SERVICE",
"CAP_SYS_CHROOT",
"CAP_KILL",
"CAP_AUDIT_WRITE",
},
Ambient: []string{
"CAP_CHOWN",
"CAP_DAC_OVERRIDE",
"CAP_FSETID",
"CAP_FOWNER",
"CAP_MKNOD",
"CAP_NET_RAW",
"CAP_SETGID",
"CAP_SETUID",
"CAP_SETFCAP",
"CAP_SETPCAP",
"CAP_NET_BIND_SERVICE",
"CAP_SYS_CHROOT",
"CAP_KILL",
"CAP_AUDIT_WRITE",
},
},
Capabilities: []string{
"CAP_CHOWN",
"CAP_DAC_OVERRIDE",
"CAP_FSETID",
"CAP_FOWNER",
"CAP_MKNOD",
"CAP_NET_RAW",
"CAP_SETGID",
"CAP_SETUID",
"CAP_SETFCAP",
"CAP_SETPCAP",
"CAP_NET_BIND_SERVICE",
"CAP_SYS_CHROOT",
"CAP_KILL",
"CAP_AUDIT_WRITE",
},
Namespaces: configs.Namespaces([]configs.Namespace{
{Type: configs.NEWNS},
{Type: configs.NEWUTS},
@@ -154,13 +77,12 @@ config := &configs.Config{
Parent: "system",
Resources: &configs.Resources{
MemorySwappiness: nil,
AllowAllDevices: nil,
AllowAllDevices: false,
AllowedDevices: configs.DefaultAllowedDevices,
},
},
MaskPaths: []string{
"/proc/kcore",
"/sys/firmware",
},
ReadonlyPaths: []string{
"/proc/sys", "/proc/sysrq-trigger", "/proc/irq", "/proc/bus",
@@ -178,14 +100,14 @@ config := &configs.Config{
Source: "tmpfs",
Destination: "/dev",
Device: "tmpfs",
Flags: unix.MS_NOSUID | unix.MS_STRICTATIME,
Flags: syscall.MS_NOSUID | syscall.MS_STRICTATIME,
Data: "mode=755",
},
{
Source: "devpts",
Destination: "/dev/pts",
Device: "devpts",
Flags: unix.MS_NOSUID | unix.MS_NOEXEC,
Flags: syscall.MS_NOSUID | syscall.MS_NOEXEC,
Data: "newinstance,ptmxmode=0666,mode=0620,gid=5",
},
{
@@ -205,7 +127,7 @@ config := &configs.Config{
Source: "sysfs",
Destination: "/sys",
Device: "sysfs",
Flags: defaultMountFlags | unix.MS_RDONLY,
Flags: defaultMountFlags | syscall.MS_RDONLY,
},
},
UidMappings: []configs.IDMap{
@@ -231,7 +153,7 @@ config := &configs.Config{
},
Rlimits: []configs.Rlimit{
{
Type: unix.RLIMIT_NOFILE,
Type: syscall.RLIMIT_NOFILE,
Hard: uint64(1025),
Soft: uint64(1025),
},
@@ -262,10 +184,10 @@ process := &libcontainer.Process{
Stderr: os.Stderr,
}
err := container.Run(process)
err := container.Start(process)
if err != nil {
container.Destroy()
logrus.Fatal(err)
container.Destroy()
return
}
@@ -297,15 +219,6 @@ container.Resume()
// send signal to container's init process.
container.Signal(signal)
// update container resource constraints.
container.Set(config)
// get current status of the container.
status, err := container.Status()
// get current container's state information.
state, err := container.State()
```

View File

@@ -71,6 +71,7 @@ that are required for executing a container's process.
| /dev/tty | 0666 | rwm |
| /dev/random | 0666 | rwm |
| /dev/urandom | 0666 | rwm |
| /dev/fuse | 0666 | rwm |
**ptmx**
@@ -89,7 +90,7 @@ in tmpfs.
After `/dev/null` has been setup we check for any external links between
the container's io, STDIN, STDOUT, STDERR. If the container's io is pointing
to `/dev/null` outside the container we close and `dup2` the `/dev/null`
to `/dev/null` outside the container we close and `dup2` the the `/dev/null`
that is local to the container's rootfs.
@@ -296,7 +297,7 @@ a container.
| -------------- | ------------------------------------------------------------------ |
| Get processes | Return all the pids for processes running inside a container |
| Get Stats | Return resource statistics for the container as a whole |
| Wait | Waits on the container's init process ( pid 1 ) |
| Wait | Wait waits on the container's init process ( pid 1 ) |
| Wait Process | Wait on any of the container's processes returning the exit status |
| Destroy | Kill the container's init process and remove any filesystem state |
| Signal | Send a signal to the container's init process |

View File

@@ -7,7 +7,6 @@ package apparmor
// #include <stdlib.h>
import "C"
import (
"fmt"
"io/ioutil"
"os"
"unsafe"
@@ -33,7 +32,7 @@ func ApplyProfile(name string) error {
cName := C.CString(name)
defer C.free(unsafe.Pointer(cName))
if _, err := C.aa_change_onexec(cName); err != nil {
return fmt.Errorf("apparmor failed to apply profile: %s", err)
return err
}
return nil
}

View File

@@ -7,11 +7,10 @@ import (
"os"
"strings"
"github.com/opencontainers/runc/libcontainer/configs"
"github.com/syndtr/gocapability/capability"
)
const allCapabilityTypes = capability.CAPS | capability.BOUNDS | capability.AMBS
const allCapabilityTypes = capability.CAPS | capability.BOUNDS
var capabilityMap map[string]capability.Cap
@@ -31,84 +30,40 @@ func init() {
}
}
func newContainerCapList(capConfig *configs.Capabilities) (*containerCapabilities, error) {
bounding := []capability.Cap{}
for _, c := range capConfig.Bounding {
func newCapWhitelist(caps []string) (*whitelist, error) {
l := []capability.Cap{}
for _, c := range caps {
v, ok := capabilityMap[c]
if !ok {
return nil, fmt.Errorf("unknown capability %q", c)
}
bounding = append(bounding, v)
}
effective := []capability.Cap{}
for _, c := range capConfig.Effective {
v, ok := capabilityMap[c]
if !ok {
return nil, fmt.Errorf("unknown capability %q", c)
}
effective = append(effective, v)
}
inheritable := []capability.Cap{}
for _, c := range capConfig.Inheritable {
v, ok := capabilityMap[c]
if !ok {
return nil, fmt.Errorf("unknown capability %q", c)
}
inheritable = append(inheritable, v)
}
permitted := []capability.Cap{}
for _, c := range capConfig.Permitted {
v, ok := capabilityMap[c]
if !ok {
return nil, fmt.Errorf("unknown capability %q", c)
}
permitted = append(permitted, v)
}
ambient := []capability.Cap{}
for _, c := range capConfig.Ambient {
v, ok := capabilityMap[c]
if !ok {
return nil, fmt.Errorf("unknown capability %q", c)
}
ambient = append(ambient, v)
l = append(l, v)
}
pid, err := capability.NewPid(os.Getpid())
if err != nil {
return nil, err
}
return &containerCapabilities{
bounding: bounding,
effective: effective,
inheritable: inheritable,
permitted: permitted,
ambient: ambient,
pid: pid,
return &whitelist{
keep: l,
pid: pid,
}, nil
}
type containerCapabilities struct {
pid capability.Capabilities
bounding []capability.Cap
effective []capability.Cap
inheritable []capability.Cap
permitted []capability.Cap
ambient []capability.Cap
type whitelist struct {
pid capability.Capabilities
keep []capability.Cap
}
// ApplyBoundingSet sets the capability bounding set to those specified in the whitelist.
func (c *containerCapabilities) ApplyBoundingSet() error {
c.pid.Clear(capability.BOUNDS)
c.pid.Set(capability.BOUNDS, c.bounding...)
return c.pid.Apply(capability.BOUNDS)
// dropBoundingSet drops the capability bounding set to those specified in the whitelist.
func (w *whitelist) dropBoundingSet() error {
w.pid.Clear(capability.BOUNDS)
w.pid.Set(capability.BOUNDS, w.keep...)
return w.pid.Apply(capability.BOUNDS)
}
// Apply sets all the capabilities for the current process in the config.
func (c *containerCapabilities) ApplyCaps() error {
c.pid.Clear(allCapabilityTypes)
c.pid.Set(capability.BOUNDS, c.bounding...)
c.pid.Set(capability.PERMITTED, c.permitted...)
c.pid.Set(capability.INHERITABLE, c.inheritable...)
c.pid.Set(capability.EFFECTIVE, c.effective...)
c.pid.Set(capability.AMBIENT, c.ambient...)
return c.pid.Apply(allCapabilityTypes)
// drop drops all capabilities for the current process except those specified in the whitelist.
func (w *whitelist) drop() error {
w.pid.Clear(allCapabilityTypes)
w.pid.Set(allCapabilityTypes, w.keep...)
return w.pid.Apply(allCapabilityTypes)
}

View File

@@ -27,9 +27,9 @@ type Manager interface {
// Destroys the cgroup set
Destroy() error
// The option func SystemdCgroups() and Cgroupfs() require following attributes:
// NewCgroupManager() and LoadCgroupManager() require following attributes:
// Paths map[string]string
// Cgroups *configs.Cgroup
// Cgroups *cgroups.Cgroup
// Paths maps cgroup subsystem to path at which it is mounted.
// Cgroups specifies specific cgroup settings for the various subsystems
@@ -37,7 +37,7 @@ type Manager interface {
// restore the object later.
GetPaths() map[string]string
// Sets the cgroup as configured.
// Set the cgroup as configured.
Set(container *configs.Config) error
}

View File

@@ -9,6 +9,7 @@ import (
"io/ioutil"
"os"
"path/filepath"
"strconv"
"sync"
"github.com/opencontainers/runc/libcontainer/cgroups"
@@ -32,6 +33,7 @@ var (
&FreezerGroup{},
&NameGroup{GroupName: "name=systemd", Join: true},
}
CgroupProcesses = "cgroup.procs"
HugePageSizes, _ = cgroups.GetHugePageSize()
)
@@ -104,8 +106,6 @@ func (m *Manager) Apply(pid int) (err error) {
if m.Cgroups == nil {
return nil
}
m.mu.Lock()
defer m.mu.Unlock()
var c = m.Cgroups
@@ -114,8 +114,8 @@ func (m *Manager) Apply(pid int) (err error) {
return err
}
m.Paths = make(map[string]string)
if c.Paths != nil {
paths := make(map[string]string)
for name, path := range c.Paths {
_, err := d.path(name)
if err != nil {
@@ -124,30 +124,32 @@ func (m *Manager) Apply(pid int) (err error) {
}
return err
}
m.Paths[name] = path
paths[name] = path
}
m.Paths = paths
return cgroups.EnterPid(m.Paths, pid)
}
m.mu.Lock()
defer m.mu.Unlock()
paths := make(map[string]string)
for _, sys := range subsystems {
if err := sys.Apply(d); err != nil {
return err
}
// TODO: Apply should, ideally, be reentrant or be broken up into a separate
// create and join phase so that the cgroup hierarchy for a container can be
// created then join consists of writing the process pids to cgroup.procs
p, err := d.path(sys.Name())
if err != nil {
// The non-presence of the devices subsystem is
// considered fatal for security reasons.
if cgroups.IsNotFound(err) && sys.Name() != "devices" {
if cgroups.IsNotFound(err) {
continue
}
return err
}
m.Paths[sys.Name()] = p
if err := sys.Apply(d); err != nil {
return err
}
paths[sys.Name()] = p
}
m.Paths = paths
return nil
}
@@ -188,15 +190,18 @@ func (m *Manager) GetStats() (*cgroups.Stats, error) {
}
func (m *Manager) Set(container *configs.Config) error {
// If Paths are set, then we are just joining cgroups paths
// and there is no need to set any values.
if m.Cgroups.Paths != nil {
return nil
}
paths := m.GetPaths()
for _, sys := range subsystems {
path := paths[sys.Name()]
// Generate fake cgroup data.
d, err := getCgroupData(container.Cgroups, -1)
if err != nil {
return err
}
// Get the path, but don't error out if the cgroup wasn't found.
path, err := d.path(sys.Name())
if err != nil && !cgroups.IsNotFound(err) {
return err
}
if err := sys.Set(path, container.Cgroups); err != nil {
return err
}
@@ -213,8 +218,14 @@ func (m *Manager) Set(container *configs.Config) error {
// Freeze toggles the container's freezer cgroup depending on the state
// provided
func (m *Manager) Freeze(state configs.FreezerState) error {
paths := m.GetPaths()
dir := paths["freezer"]
d, err := getCgroupData(m.Cgroups, 0)
if err != nil {
return err
}
dir, err := d.path("freezer")
if err != nil {
return err
}
prevState := m.Cgroups.Resources.Freezer
m.Cgroups.Resources.Freezer = state
freezer, err := subsystems.Get("freezer")
@@ -230,13 +241,28 @@ func (m *Manager) Freeze(state configs.FreezerState) error {
}
func (m *Manager) GetPids() ([]int, error) {
paths := m.GetPaths()
return cgroups.GetPids(paths["devices"])
dir, err := getCgroupPath(m.Cgroups)
if err != nil {
return nil, err
}
return cgroups.GetPids(dir)
}
func (m *Manager) GetAllPids() ([]int, error) {
paths := m.GetPaths()
return cgroups.GetAllPids(paths["devices"])
dir, err := getCgroupPath(m.Cgroups)
if err != nil {
return nil, err
}
return cgroups.GetAllPids(dir)
}
func getCgroupPath(c *configs.Cgroup) (string, error) {
d, err := getCgroupData(c, 0)
if err != nil {
return "", err
}
return d.path("devices")
}
func getCgroupData(c *configs.Cgroup, pid int) (*cgroupData, error) {
@@ -267,8 +293,25 @@ func getCgroupData(c *configs.Cgroup, pid int) (*cgroupData, error) {
}, nil
}
func (raw *cgroupData) parentPath(subsystem, mountpoint, root string) (string, error) {
// Use GetThisCgroupDir instead of GetInitCgroupDir, because the creating
// process could in container and shared pid namespace with host, and
// /proc/1/cgroup could point to whole other world of cgroups.
initPath, err := cgroups.GetThisCgroupDir(subsystem)
if err != nil {
return "", err
}
// This is needed for nested containers, because in /proc/self/cgroup we
// see pathes from host, which don't exist in container.
relDir, err := filepath.Rel(root, initPath)
if err != nil {
return "", err
}
return filepath.Join(mountpoint, relDir), nil
}
func (raw *cgroupData) path(subsystem string) (string, error) {
mnt, err := cgroups.FindCgroupMountpoint(subsystem)
mnt, root, err := cgroups.FindCgroupMountpointAndRoot(subsystem)
// If we didn't mount the subsystem, there is no point we make the path.
if err != nil {
return "", err
@@ -276,14 +319,11 @@ func (raw *cgroupData) path(subsystem string) (string, error) {
// If the cgroup name/path is absolute do not look relative to the cgroup of the init process.
if filepath.IsAbs(raw.innerPath) {
// Sometimes subsystems can be mounted together as 'cpu,cpuacct'.
// Sometimes subsystems can be mounted togethger as 'cpu,cpuacct'.
return filepath.Join(raw.root, filepath.Base(mnt), raw.innerPath), nil
}
// Use GetOwnCgroupPath instead of GetInitCgroupPath, because the creating
// process could in container and shared pid namespace with host, and
// /proc/1/cgroup could point to whole other world of cgroups.
parentPath, err := cgroups.GetOwnCgroupPath(subsystem)
parentPath, err := raw.parentPath(subsystem, mnt, root)
if err != nil {
return "", err
}
@@ -299,7 +339,7 @@ func (raw *cgroupData) join(subsystem string) (string, error) {
if err := os.MkdirAll(path, 0755); err != nil {
return "", err
}
if err := cgroups.WriteCgroupProc(path, raw.pid); err != nil {
if err := writeFile(path, CgroupProcesses, strconv.Itoa(raw.pid)); err != nil {
return "", err
}
return path, nil
@@ -309,7 +349,7 @@ func writeFile(dir, file, data string) error {
// Normally dir should not be empty, one case is that cgroup subsystem
// is not mounted, we will get empty dir, and we want it fail here.
if dir == "" {
return fmt.Errorf("no such directory for %s", file)
return fmt.Errorf("no such directory for %s.", file)
}
if err := ioutil.WriteFile(filepath.Join(dir, file), []byte(data), 0700); err != nil {
return fmt.Errorf("failed to write %v to %v: %v", data, file, err)
@@ -332,8 +372,8 @@ func removePath(p string, err error) error {
return nil
}
func CheckCpushares(path string, c uint64) error {
var cpuShares uint64
func CheckCpushares(path string, c int64) error {
var cpuShares int64
if c == 0 {
return nil

View File

@@ -22,59 +22,21 @@ func (s *CpuGroup) Name() string {
func (s *CpuGroup) Apply(d *cgroupData) error {
// We always want to join the cpu group, to allow fair cpu scheduling
// on a container basis
path, err := d.path("cpu")
_, err := d.join("cpu")
if err != nil && !cgroups.IsNotFound(err) {
return err
}
return s.ApplyDir(path, d.config, d.pid)
}
func (s *CpuGroup) ApplyDir(path string, cgroup *configs.Cgroup, pid int) error {
// This might happen if we have no cpu cgroup mounted.
// Just do nothing and don't fail.
if path == "" {
return nil
}
if err := os.MkdirAll(path, 0755); err != nil {
return err
}
// We should set the real-Time group scheduling settings before moving
// in the process because if the process is already in SCHED_RR mode
// and no RT bandwidth is set, adding it will fail.
if err := s.SetRtSched(path, cgroup); err != nil {
return err
}
// because we are not using d.join we need to place the pid into the procs file
// unlike the other subsystems
if err := cgroups.WriteCgroupProc(path, pid); err != nil {
return err
}
return nil
}
func (s *CpuGroup) SetRtSched(path string, cgroup *configs.Cgroup) error {
if cgroup.Resources.CpuRtPeriod != 0 {
if err := writeFile(path, "cpu.rt_period_us", strconv.FormatUint(cgroup.Resources.CpuRtPeriod, 10)); err != nil {
return err
}
}
if cgroup.Resources.CpuRtRuntime != 0 {
if err := writeFile(path, "cpu.rt_runtime_us", strconv.FormatInt(cgroup.Resources.CpuRtRuntime, 10)); err != nil {
return err
}
}
return nil
}
func (s *CpuGroup) Set(path string, cgroup *configs.Cgroup) error {
if cgroup.Resources.CpuShares != 0 {
if err := writeFile(path, "cpu.shares", strconv.FormatUint(cgroup.Resources.CpuShares, 10)); err != nil {
if err := writeFile(path, "cpu.shares", strconv.FormatInt(cgroup.Resources.CpuShares, 10)); err != nil {
return err
}
}
if cgroup.Resources.CpuPeriod != 0 {
if err := writeFile(path, "cpu.cfs_period_us", strconv.FormatUint(cgroup.Resources.CpuPeriod, 10)); err != nil {
if err := writeFile(path, "cpu.cfs_period_us", strconv.FormatInt(cgroup.Resources.CpuPeriod, 10)); err != nil {
return err
}
}
@@ -83,8 +45,15 @@ func (s *CpuGroup) Set(path string, cgroup *configs.Cgroup) error {
return err
}
}
if err := s.SetRtSched(path, cgroup); err != nil {
return err
if cgroup.Resources.CpuRtPeriod != 0 {
if err := writeFile(path, "cpu.rt_period_us", strconv.FormatInt(cgroup.Resources.CpuRtPeriod, 10)); err != nil {
return err
}
}
if cgroup.Resources.CpuRtRuntime != 0 {
if err := writeFile(path, "cpu.rt_runtime_us", strconv.FormatInt(cgroup.Resources.CpuRtRuntime, 10)); err != nil {
return err
}
}
return nil

View File

@@ -106,13 +106,13 @@ func TestCpuStats(t *testing.T) {
defer helper.cleanup()
const (
nrPeriods = 2000
nrThrottled = 200
throttledTime = uint64(18446744073709551615)
kNrPeriods = 2000
kNrThrottled = 200
kThrottledTime = uint64(18446744073709551615)
)
cpuStatContent := fmt.Sprintf("nr_periods %d\n nr_throttled %d\n throttled_time %d\n",
nrPeriods, nrThrottled, throttledTime)
kNrPeriods, kNrThrottled, kThrottledTime)
helper.writeFileContents(map[string]string{
"cpu.stat": cpuStatContent,
})
@@ -125,9 +125,9 @@ func TestCpuStats(t *testing.T) {
}
expectedStats := cgroups.ThrottlingData{
Periods: nrPeriods,
ThrottledPeriods: nrThrottled,
ThrottledTime: throttledTime}
Periods: kNrPeriods,
ThrottledPeriods: kNrThrottled,
ThrottledTime: kThrottledTime}
expectThrottlingDataEquals(t, expectedStats, actualStats.CpuStats.ThrottlingData)
}
@@ -161,49 +161,3 @@ func TestInvalidCpuStat(t *testing.T) {
t.Fatal("Expected failed stat parsing.")
}
}
func TestCpuSetRtSchedAtApply(t *testing.T) {
helper := NewCgroupTestUtil("cpu", t)
defer helper.cleanup()
const (
rtRuntimeBefore = 0
rtRuntimeAfter = 5000
rtPeriodBefore = 0
rtPeriodAfter = 7000
)
helper.writeFileContents(map[string]string{
"cpu.rt_runtime_us": strconv.Itoa(rtRuntimeBefore),
"cpu.rt_period_us": strconv.Itoa(rtPeriodBefore),
})
helper.CgroupData.config.Resources.CpuRtRuntime = rtRuntimeAfter
helper.CgroupData.config.Resources.CpuRtPeriod = rtPeriodAfter
cpu := &CpuGroup{}
if err := cpu.ApplyDir(helper.CgroupPath, helper.CgroupData.config, 1234); err != nil {
t.Fatal(err)
}
rtRuntime, err := getCgroupParamUint(helper.CgroupPath, "cpu.rt_runtime_us")
if err != nil {
t.Fatalf("Failed to parse cpu.rt_runtime_us - %s", err)
}
if rtRuntime != rtRuntimeAfter {
t.Fatal("Got the wrong value, set cpu.rt_runtime_us failed.")
}
rtPeriod, err := getCgroupParamUint(helper.CgroupPath, "cpu.rt_period_us")
if err != nil {
t.Fatalf("Failed to parse cpu.rt_period_us - %s", err)
}
if rtPeriod != rtPeriodAfter {
t.Fatal("Got the wrong value, set cpu.rt_period_us failed.")
}
pid, err := getCgroupParamUint(helper.CgroupPath, "cgroup.procs")
if err != nil {
t.Fatalf("Failed to parse cgroup.procs - %s", err)
}
if pid != 1234 {
t.Fatal("Got the wrong value, set cgroup.procs failed.")
}
}

View File

@@ -8,6 +8,7 @@ import (
"io/ioutil"
"os"
"path/filepath"
"strconv"
"github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/opencontainers/runc/libcontainer/configs"
@@ -57,34 +58,16 @@ func (s *CpusetGroup) ApplyDir(dir string, cgroup *configs.Cgroup, pid int) erro
if dir == "" {
return nil
}
mountInfo, err := ioutil.ReadFile("/proc/self/mountinfo")
root, err := getCgroupRoot()
if err != nil {
return err
}
root := filepath.Dir(cgroups.GetClosestMountpointAncestor(dir, string(mountInfo)))
// 'ensureParent' start with parent because we don't want to
// explicitly inherit from parent, it could conflict with
// 'cpuset.cpu_exclusive'.
if err := s.ensureParent(filepath.Dir(dir), root); err != nil {
if err := s.ensureParent(dir, root); err != nil {
return err
}
if err := os.MkdirAll(dir, 0755); err != nil {
return err
}
// We didn't inherit cpuset configs from parent, but we have
// to ensure cpuset configs are set before moving task into the
// cgroup.
// The logic is, if user specified cpuset configs, use these
// specified configs, otherwise, inherit from parent. This makes
// cpuset configs work correctly with 'cpuset.cpu_exclusive', and
// keep backward compatbility.
if err := s.ensureCpusAndMems(dir, cgroup); err != nil {
return err
}
// because we are not using d.join we need to place the pid into the procs file
// unlike the other subsystems
if err := cgroups.WriteCgroupProc(dir, pid); err != nil {
if err := writeFile(dir, "cgroup.procs", strconv.Itoa(pid)); err != nil {
return err
}
@@ -154,10 +137,3 @@ func (s *CpusetGroup) copyIfNeeded(current, parent string) error {
func (s *CpusetGroup) isEmpty(b []byte) bool {
return len(bytes.Trim(b, "\n")) == 0
}
func (s *CpusetGroup) ensureCpusAndMems(path string, cgroup *configs.Cgroup) error {
if err := s.Set(path, cgroup); err != nil {
return err
}
return s.copyIfNeeded(path, filepath.Dir(path))
}

View File

@@ -43,23 +43,21 @@ func (s *DevicesGroup) Set(path string, cgroup *configs.Cgroup) error {
}
return nil
}
if cgroup.Resources.AllowAllDevices != nil {
if *cgroup.Resources.AllowAllDevices == false {
if err := writeFile(path, "devices.deny", "a"); err != nil {
return err
}
for _, dev := range cgroup.Resources.AllowedDevices {
if err := writeFile(path, "devices.allow", dev.CgroupString()); err != nil {
return err
}
}
return nil
}
if err := writeFile(path, "devices.allow", "a"); err != nil {
if !cgroup.Resources.AllowAllDevices {
if err := writeFile(path, "devices.deny", "a"); err != nil {
return err
}
for _, dev := range cgroup.Resources.AllowedDevices {
if err := writeFile(path, "devices.allow", dev.CgroupString()); err != nil {
return err
}
}
return nil
}
if err := writeFile(path, "devices.allow", "a"); err != nil {
return err
}
for _, dev := range cgroup.Resources.DeniedDevices {

View File

@@ -40,8 +40,8 @@ func TestDevicesSetAllow(t *testing.T) {
helper.writeFileContents(map[string]string{
"devices.deny": "a",
})
allowAllDevices := false
helper.CgroupData.config.Resources.AllowAllDevices = &allowAllDevices
helper.CgroupData.config.Resources.AllowAllDevices = false
helper.CgroupData.config.Resources.AllowedDevices = allowedDevices
devices := &DevicesGroup{}
if err := devices.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
@@ -56,19 +56,6 @@ func TestDevicesSetAllow(t *testing.T) {
if value != allowedList {
t.Fatal("Got the wrong value, set devices.allow failed.")
}
// When AllowAllDevices is nil, devices.allow file should not be modified.
helper.CgroupData.config.Resources.AllowAllDevices = nil
if err := devices.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
t.Fatal(err)
}
value, err = getCgroupParamString(helper.CgroupPath, "devices.allow")
if err != nil {
t.Fatalf("Failed to parse devices.allow - %s", err)
}
if value != allowedList {
t.Fatal("devices policy shouldn't have changed on AllowedAllDevices=nil.")
}
}
func TestDevicesSetDeny(t *testing.T) {
@@ -79,8 +66,7 @@ func TestDevicesSetDeny(t *testing.T) {
"devices.allow": "a",
})
allowAllDevices := true
helper.CgroupData.config.Resources.AllowAllDevices = &allowAllDevices
helper.CgroupData.config.Resources.AllowAllDevices = true
helper.CgroupData.config.Resources.DeniedDevices = deniedDevices
devices := &DevicesGroup{}
if err := devices.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {

View File

@@ -5,23 +5,13 @@ package fs
import (
"bufio"
"fmt"
"io/ioutil"
"os"
"path/filepath"
"strconv"
"strings"
"syscall" // only for Errno
"github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/opencontainers/runc/libcontainer/configs"
"golang.org/x/sys/unix"
)
const (
cgroupKernelMemoryLimit = "memory.kmem.limit_in_bytes"
cgroupMemorySwapLimit = "memory.memsw.limit_in_bytes"
cgroupMemoryLimit = "memory.limit_in_bytes"
)
type MemoryGroup struct {
@@ -35,23 +25,20 @@ func (s *MemoryGroup) Apply(d *cgroupData) (err error) {
path, err := d.path("memory")
if err != nil && !cgroups.IsNotFound(err) {
return err
} else if path == "" {
return nil
}
if memoryAssigned(d.config) {
if _, err := os.Stat(path); os.IsNotExist(err) {
if path != "" {
if err := os.MkdirAll(path, 0755); err != nil {
return err
}
// Only enable kernel memory accouting when this cgroup
// is created by libcontainer, otherwise we might get
// error when people use `cgroupsPath` to join an existed
// cgroup whose kernel memory is not initialized.
if err := EnableKernelMemoryAccounting(path); err != nil {
return err
}
}
// We have to set kernel memory here, as we can't change it once
// processes have been attached.
if err := s.SetKernelMemory(path, d.config); err != nil {
return err
}
}
defer func() {
if err != nil {
os.RemoveAll(path)
@@ -67,57 +54,21 @@ func (s *MemoryGroup) Apply(d *cgroupData) (err error) {
return nil
}
func EnableKernelMemoryAccounting(path string) error {
// Check if kernel memory is enabled
// We have to limit the kernel memory here as it won't be accounted at all
// until a limit is set on the cgroup and limit cannot be set once the
// cgroup has children, or if there are already tasks in the cgroup.
for _, i := range []int64{1, -1} {
if err := setKernelMemory(path, i); err != nil {
func (s *MemoryGroup) SetKernelMemory(path string, cgroup *configs.Cgroup) error {
// This has to be done separately because it has special constraints (it
// can't be done after there are processes attached to the cgroup).
if cgroup.Resources.KernelMemory > 0 {
if err := writeFile(path, "memory.kmem.limit_in_bytes", strconv.FormatInt(cgroup.Resources.KernelMemory, 10)); err != nil {
return err
}
}
return nil
}
func setKernelMemory(path string, kernelMemoryLimit int64) error {
if path == "" {
return fmt.Errorf("no such directory for %s", cgroupKernelMemoryLimit)
}
if !cgroups.PathExists(filepath.Join(path, cgroupKernelMemoryLimit)) {
// kernel memory is not enabled on the system so we should do nothing
return nil
}
if err := ioutil.WriteFile(filepath.Join(path, cgroupKernelMemoryLimit), []byte(strconv.FormatInt(kernelMemoryLimit, 10)), 0700); err != nil {
// Check if the error number returned by the syscall is "EBUSY"
// The EBUSY signal is returned on attempts to write to the
// memory.kmem.limit_in_bytes file if the cgroup has children or
// once tasks have been attached to the cgroup
if pathErr, ok := err.(*os.PathError); ok {
if errNo, ok := pathErr.Err.(syscall.Errno); ok {
if errNo == unix.EBUSY {
return fmt.Errorf("failed to set %s, because either tasks have already joined this cgroup or it has children", cgroupKernelMemoryLimit)
}
}
}
return fmt.Errorf("failed to write %v to %v: %v", kernelMemoryLimit, cgroupKernelMemoryLimit, err)
}
return nil
}
func setMemoryAndSwap(path string, cgroup *configs.Cgroup) error {
// If the memory update is set to -1 we should also
// set swap to -1, it means unlimited memory.
if cgroup.Resources.Memory == -1 {
// Only set swap if it's enabled in kernel
if cgroups.PathExists(filepath.Join(path, cgroupMemorySwapLimit)) {
cgroup.Resources.MemorySwap = -1
}
}
// When memory and swap memory are both set, we need to handle the cases
// for updating container.
if cgroup.Resources.Memory != 0 && cgroup.Resources.MemorySwap != 0 {
if cgroup.Resources.Memory != 0 && cgroup.Resources.MemorySwap > 0 {
memoryUsage, err := getMemoryData(path, "")
if err != nil {
return err
@@ -126,29 +77,29 @@ func setMemoryAndSwap(path string, cgroup *configs.Cgroup) error {
// When update memory limit, we should adapt the write sequence
// for memory and swap memory, so it won't fail because the new
// value and the old value don't fit kernel's validation.
if cgroup.Resources.MemorySwap == -1 || memoryUsage.Limit < uint64(cgroup.Resources.MemorySwap) {
if err := writeFile(path, cgroupMemorySwapLimit, strconv.FormatInt(cgroup.Resources.MemorySwap, 10)); err != nil {
if memoryUsage.Limit < uint64(cgroup.Resources.MemorySwap) {
if err := writeFile(path, "memory.memsw.limit_in_bytes", strconv.FormatInt(cgroup.Resources.MemorySwap, 10)); err != nil {
return err
}
if err := writeFile(path, cgroupMemoryLimit, strconv.FormatInt(cgroup.Resources.Memory, 10)); err != nil {
if err := writeFile(path, "memory.limit_in_bytes", strconv.FormatInt(cgroup.Resources.Memory, 10)); err != nil {
return err
}
} else {
if err := writeFile(path, cgroupMemoryLimit, strconv.FormatInt(cgroup.Resources.Memory, 10)); err != nil {
if err := writeFile(path, "memory.limit_in_bytes", strconv.FormatInt(cgroup.Resources.Memory, 10)); err != nil {
return err
}
if err := writeFile(path, cgroupMemorySwapLimit, strconv.FormatInt(cgroup.Resources.MemorySwap, 10)); err != nil {
if err := writeFile(path, "memory.memsw.limit_in_bytes", strconv.FormatInt(cgroup.Resources.MemorySwap, 10)); err != nil {
return err
}
}
} else {
if cgroup.Resources.Memory != 0 {
if err := writeFile(path, cgroupMemoryLimit, strconv.FormatInt(cgroup.Resources.Memory, 10)); err != nil {
if err := writeFile(path, "memory.limit_in_bytes", strconv.FormatInt(cgroup.Resources.Memory, 10)); err != nil {
return err
}
}
if cgroup.Resources.MemorySwap != 0 {
if err := writeFile(path, cgroupMemorySwapLimit, strconv.FormatInt(cgroup.Resources.MemorySwap, 10)); err != nil {
if cgroup.Resources.MemorySwap > 0 {
if err := writeFile(path, "memory.memsw.limit_in_bytes", strconv.FormatInt(cgroup.Resources.MemorySwap, 10)); err != nil {
return err
}
}
@@ -162,18 +113,11 @@ func (s *MemoryGroup) Set(path string, cgroup *configs.Cgroup) error {
return err
}
if cgroup.Resources.KernelMemory != 0 {
if err := setKernelMemory(path, cgroup.Resources.KernelMemory); err != nil {
return err
}
}
if cgroup.Resources.MemoryReservation != 0 {
if err := writeFile(path, "memory.soft_limit_in_bytes", strconv.FormatInt(cgroup.Resources.MemoryReservation, 10)); err != nil {
return err
}
}
if cgroup.Resources.KernelMemoryTCP != 0 {
if err := writeFile(path, "memory.kmem.tcp.limit_in_bytes", strconv.FormatInt(cgroup.Resources.KernelMemoryTCP, 10)); err != nil {
return err
@@ -186,12 +130,12 @@ func (s *MemoryGroup) Set(path string, cgroup *configs.Cgroup) error {
}
if cgroup.Resources.MemorySwappiness == nil || int64(*cgroup.Resources.MemorySwappiness) == -1 {
return nil
} else if *cgroup.Resources.MemorySwappiness <= 100 {
if err := writeFile(path, "memory.swappiness", strconv.FormatUint(*cgroup.Resources.MemorySwappiness, 10)); err != nil {
} else if int64(*cgroup.Resources.MemorySwappiness) >= 0 && int64(*cgroup.Resources.MemorySwappiness) <= 100 {
if err := writeFile(path, "memory.swappiness", strconv.FormatInt(*cgroup.Resources.MemorySwappiness, 10)); err != nil {
return err
}
} else {
return fmt.Errorf("invalid value:%d. valid memory swappiness range is 0-100", *cgroup.Resources.MemorySwappiness)
return fmt.Errorf("invalid value:%d. valid memory swappiness range is 0-100", int64(*cgroup.Resources.MemorySwappiness))
}
return nil
@@ -243,14 +187,6 @@ func (s *MemoryGroup) GetStats(path string, stats *cgroups.Stats) error {
}
stats.MemoryStats.KernelTCPUsage = kernelTCPUsage
useHierarchy := strings.Join([]string{"memory", "use_hierarchy"}, ".")
value, err := getCgroupParamUint(path, useHierarchy)
if err != nil {
return err
}
if value == 1 {
stats.MemoryStats.UseHierarchy = true
}
return nil
}
@@ -261,7 +197,7 @@ func memoryAssigned(cgroup *configs.Cgroup) bool {
cgroup.Resources.KernelMemory > 0 ||
cgroup.Resources.KernelMemoryTCP > 0 ||
cgroup.Resources.OomKillDisable ||
(cgroup.Resources.MemorySwappiness != nil && int64(*cgroup.Resources.MemorySwappiness) != -1)
(cgroup.Resources.MemorySwappiness != nil && *cgroup.Resources.MemorySwappiness != -1)
}
func getMemoryData(path, name string) (cgroups.MemoryData, error) {

View File

@@ -12,11 +12,10 @@ import (
const (
memoryStatContents = `cache 512
rss 1024`
memoryUsageContents = "2048\n"
memoryMaxUsageContents = "4096\n"
memoryFailcnt = "100\n"
memoryLimitContents = "8192\n"
memoryUseHierarchyContents = "1\n"
memoryUsageContents = "2048\n"
memoryMaxUsageContents = "4096\n"
memoryFailcnt = "100\n"
memoryLimitContents = "8192\n"
)
func TestMemorySetMemory(t *testing.T) {
@@ -87,6 +86,47 @@ func TestMemorySetMemoryswap(t *testing.T) {
}
}
func TestMemorySetNegativeMemoryswap(t *testing.T) {
helper := NewCgroupTestUtil("memory", t)
defer helper.cleanup()
const (
memoryBefore = 314572800 // 300M
memoryAfter = 524288000 // 500M
memoryswapBefore = 629145600 // 600M
memoryswapAfter = 629145600 // 600M
)
helper.writeFileContents(map[string]string{
"memory.limit_in_bytes": strconv.Itoa(memoryBefore),
"memory.memsw.limit_in_bytes": strconv.Itoa(memoryswapBefore),
})
helper.CgroupData.config.Resources.Memory = memoryAfter
// Negative value means not change
helper.CgroupData.config.Resources.MemorySwap = -1
memory := &MemoryGroup{}
if err := memory.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
t.Fatal(err)
}
value, err := getCgroupParamUint(helper.CgroupPath, "memory.limit_in_bytes")
if err != nil {
t.Fatalf("Failed to parse memory.limit_in_bytes - %s", err)
}
if value != memoryAfter {
t.Fatal("Got the wrong value, set memory.limit_in_bytes failed.")
}
value, err = getCgroupParamUint(helper.CgroupPath, "memory.memsw.limit_in_bytes")
if err != nil {
t.Fatalf("Failed to parse memory.memsw.limit_in_bytes - %s", err)
}
if value != memoryswapAfter {
t.Fatal("Got the wrong value, set memory.memsw.limit_in_bytes failed.")
}
}
func TestMemorySetMemoryLargerThanSwap(t *testing.T) {
helper := NewCgroupTestUtil("memory", t)
defer helper.cleanup()
@@ -190,7 +230,7 @@ func TestMemorySetKernelMemory(t *testing.T) {
helper.CgroupData.config.Resources.KernelMemory = kernelMemoryAfter
memory := &MemoryGroup{}
if err := memory.Set(helper.CgroupPath, helper.CgroupData.config); err != nil {
if err := memory.SetKernelMemory(helper.CgroupPath, helper.CgroupData.config); err != nil {
t.Fatal(err)
}
@@ -236,7 +276,7 @@ func TestMemorySetMemorySwappinessDefault(t *testing.T) {
defer helper.cleanup()
swappinessBefore := 60 //default is 60
swappinessAfter := uint64(0)
swappinessAfter := int64(0)
helper.writeFileContents(map[string]string{
"memory.swappiness": strconv.Itoa(swappinessBefore),
@@ -252,7 +292,7 @@ func TestMemorySetMemorySwappinessDefault(t *testing.T) {
if err != nil {
t.Fatalf("Failed to parse memory.swappiness - %s", err)
}
if value != swappinessAfter {
if int64(value) != swappinessAfter {
t.Fatalf("Got the wrong value (%d), set memory.swappiness = %d failed.", value, swappinessAfter)
}
}
@@ -274,7 +314,6 @@ func TestMemoryStats(t *testing.T) {
"memory.kmem.max_usage_in_bytes": memoryMaxUsageContents,
"memory.kmem.failcnt": memoryFailcnt,
"memory.kmem.limit_in_bytes": memoryLimitContents,
"memory.use_hierarchy": memoryUseHierarchyContents,
})
memory := &MemoryGroup{}
@@ -283,7 +322,7 @@ func TestMemoryStats(t *testing.T) {
if err != nil {
t.Fatal(err)
}
expectedStats := cgroups.MemoryStats{Cache: 512, Usage: cgroups.MemoryData{Usage: 2048, MaxUsage: 4096, Failcnt: 100, Limit: 8192}, SwapUsage: cgroups.MemoryData{Usage: 2048, MaxUsage: 4096, Failcnt: 100, Limit: 8192}, KernelUsage: cgroups.MemoryData{Usage: 2048, MaxUsage: 4096, Failcnt: 100, Limit: 8192}, Stats: map[string]uint64{"cache": 512, "rss": 1024}, UseHierarchy: true}
expectedStats := cgroups.MemoryStats{Cache: 512, Usage: cgroups.MemoryData{Usage: 2048, MaxUsage: 4096, Failcnt: 100, Limit: 8192}, SwapUsage: cgroups.MemoryData{Usage: 2048, MaxUsage: 4096, Failcnt: 100, Limit: 8192}, KernelUsage: cgroups.MemoryData{Usage: 2048, MaxUsage: 4096, Failcnt: 100, Limit: 8192}, Stats: map[string]uint64{"cache": 512, "rss": 1024}}
expectMemoryStatEquals(t, expectedStats, actualStats.MemoryStats)
}
@@ -432,11 +471,11 @@ func TestMemorySetOomControl(t *testing.T) {
defer helper.cleanup()
const (
oomKillDisable = 1 // disable oom killer, default is 0
oom_kill_disable = 1 // disable oom killer, default is 0
)
helper.writeFileContents(map[string]string{
"memory.oom_control": strconv.Itoa(oomKillDisable),
"memory.oom_control": strconv.Itoa(oom_kill_disable),
})
memory := &MemoryGroup{}
@@ -449,7 +488,7 @@ func TestMemorySetOomControl(t *testing.T) {
t.Fatalf("Failed to parse memory.oom_control - %s", err)
}
if value != oomKillDisable {
if value != oom_kill_disable {
t.Fatalf("Got the wrong value, set memory.oom_control failed.")
}
}

View File

@@ -3,8 +3,6 @@
package fs
import (
"strconv"
"github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/opencontainers/runc/libcontainer/configs"
)
@@ -25,8 +23,8 @@ func (s *NetClsGroup) Apply(d *cgroupData) error {
}
func (s *NetClsGroup) Set(path string, cgroup *configs.Cgroup) error {
if cgroup.Resources.NetClsClassid != 0 {
if err := writeFile(path, "net_cls.classid", strconv.FormatUint(uint64(cgroup.Resources.NetClsClassid), 10)); err != nil {
if cgroup.Resources.NetClsClassid != "" {
if err := writeFile(path, "net_cls.classid", cgroup.Resources.NetClsClassid); err != nil {
return err
}
}

View File

@@ -3,13 +3,12 @@
package fs
import (
"strconv"
"testing"
)
const (
classidBefore = 0x100002
classidAfter = 0x100001
classidBefore = "0x100002"
classidAfter = "0x100001"
)
func TestNetClsSetClassid(t *testing.T) {
@@ -17,7 +16,7 @@ func TestNetClsSetClassid(t *testing.T) {
defer helper.cleanup()
helper.writeFileContents(map[string]string{
"net_cls.classid": strconv.FormatUint(classidBefore, 10),
"net_cls.classid": classidBefore,
})
helper.CgroupData.config.Resources.NetClsClassid = classidAfter
@@ -29,7 +28,7 @@ func TestNetClsSetClassid(t *testing.T) {
// As we are in mock environment, we can't get correct value of classid from
// net_cls.classid.
// So. we just judge if we successfully write classid into file
value, err := getCgroupParamUint(helper.CgroupPath, "net_cls.classid")
value, err := getCgroupParamString(helper.CgroupPath, "net_cls.classid")
if err != nil {
t.Fatalf("Failed to parse net_cls.classid - %s", err)
}

View File

@@ -6,9 +6,8 @@ import (
"fmt"
"testing"
"github.com/Sirupsen/logrus"
"github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/sirupsen/logrus"
)
func blkioStatEntryEquals(expected, actual []cgroups.BlkioStatEntry) error {
@@ -85,11 +84,6 @@ func expectMemoryStatEquals(t *testing.T, expected, actual cgroups.MemoryStats)
expectMemoryDataEquals(t, expected.SwapUsage, actual.SwapUsage)
expectMemoryDataEquals(t, expected.KernelUsage, actual.KernelUsage)
if expected.UseHierarchy != actual.UseHierarchy {
logrus.Printf("Expected memory use hiearchy %v, but found %v\n", expected.UseHierarchy, actual.UseHierarchy)
t.Fail()
}
for key, expValue := range expected.Stats {
actValue, ok := actual.Stats[key]
if !ok {

View File

@@ -12,6 +12,7 @@ import (
)
var (
ErrNotSupportStat = errors.New("stats are not supported for subsystem")
ErrNotValidFormat = errors.New("line is not a valid key value format")
)

View File

@@ -1,128 +0,0 @@
// +build linux
package rootless
import (
"fmt"
"github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/opencontainers/runc/libcontainer/cgroups/fs"
"github.com/opencontainers/runc/libcontainer/configs"
"github.com/opencontainers/runc/libcontainer/configs/validate"
)
// TODO: This is copied from libcontainer/cgroups/fs, which duplicates this code
// needlessly. We should probably export this list.
var subsystems = []subsystem{
&fs.CpusetGroup{},
&fs.DevicesGroup{},
&fs.MemoryGroup{},
&fs.CpuGroup{},
&fs.CpuacctGroup{},
&fs.PidsGroup{},
&fs.BlkioGroup{},
&fs.HugetlbGroup{},
&fs.NetClsGroup{},
&fs.NetPrioGroup{},
&fs.PerfEventGroup{},
&fs.FreezerGroup{},
&fs.NameGroup{GroupName: "name=systemd"},
}
type subsystem interface {
// Name returns the name of the subsystem.
Name() string
// Returns the stats, as 'stats', corresponding to the cgroup under 'path'.
GetStats(path string, stats *cgroups.Stats) error
}
// The noop cgroup manager is used for rootless containers, because we currently
// cannot manage cgroups if we are in a rootless setup. This manager is chosen
// by factory if we are in rootless mode. We error out if any cgroup options are
// set in the config -- this may change in the future with upcoming kernel features
// like the cgroup namespace.
type Manager struct {
Cgroups *configs.Cgroup
Paths map[string]string
}
func (m *Manager) Apply(pid int) error {
// If there are no cgroup settings, there's nothing to do.
if m.Cgroups == nil {
return nil
}
// We can't set paths.
// TODO(cyphar): Implement the case where the runner of a rootless container
// owns their own cgroup, which would allow us to set up a
// cgroup for each path.
if m.Cgroups.Paths != nil {
return fmt.Errorf("cannot change cgroup path in rootless container")
}
// We load the paths into the manager.
paths := make(map[string]string)
for _, sys := range subsystems {
name := sys.Name()
path, err := cgroups.GetOwnCgroupPath(name)
if err != nil {
// Ignore paths we couldn't resolve.
continue
}
paths[name] = path
}
m.Paths = paths
return nil
}
func (m *Manager) GetPaths() map[string]string {
return m.Paths
}
func (m *Manager) Set(container *configs.Config) error {
// We have to re-do the validation here, since someone might decide to
// update a rootless container.
return validate.New().Validate(container)
}
func (m *Manager) GetPids() ([]int, error) {
dir, err := cgroups.GetOwnCgroupPath("devices")
if err != nil {
return nil, err
}
return cgroups.GetPids(dir)
}
func (m *Manager) GetAllPids() ([]int, error) {
dir, err := cgroups.GetOwnCgroupPath("devices")
if err != nil {
return nil, err
}
return cgroups.GetAllPids(dir)
}
func (m *Manager) GetStats() (*cgroups.Stats, error) {
// TODO(cyphar): We can make this work if we figure out a way to allow usage
// of cgroups with a rootless container. While this doesn't
// actually require write access to a cgroup directory, the
// statistics are not useful if they can be affected by
// non-container processes.
return nil, fmt.Errorf("cannot get cgroup stats in rootless container")
}
func (m *Manager) Freeze(state configs.FreezerState) error {
// TODO(cyphar): We can make this work if we figure out a way to allow usage
// of cgroups with a rootless container.
return fmt.Errorf("cannot use freezer cgroup in rootless container")
}
func (m *Manager) Destroy() error {
// We don't have to do anything here because we didn't do any setup.
return nil
}

View File

@@ -11,7 +11,6 @@ type ThrottlingData struct {
ThrottledTime uint64 `json:"throttled_time,omitempty"`
}
// CpuUsage denotes the usage of a CPU.
// All CPU stats are aggregate since container inception.
type CpuUsage struct {
// Total CPU time consumed.
@@ -51,8 +50,6 @@ type MemoryStats struct {
KernelUsage MemoryData `json:"kernel_usage,omitempty"`
// usage of kernel TCP memory
KernelTCPUsage MemoryData `json:"kernel_tcp_usage,omitempty"`
// if true, memory usage is accounted for throughout a hierarchy of cgroups.
UseHierarchy bool `json:"use_hierarchy"`
Stats map[string]uint64 `json:"stats,omitempty"`
}

View File

@@ -5,8 +5,10 @@ package systemd
import (
"errors"
"fmt"
"io/ioutil"
"os"
"path/filepath"
"strconv"
"strings"
"sync"
"time"
@@ -65,16 +67,13 @@ var subsystems = subsystemSet{
const (
testScopeWait = 4
testSliceWait = 4
)
var (
connLock sync.Mutex
theConn *systemdDbus.Conn
hasStartTransientUnit bool
hasStartTransientSliceUnit bool
hasTransientDefaultDependencies bool
hasDelegate bool
)
func newProp(name string, units interface{}) systemdDbus.Property {
@@ -147,48 +146,6 @@ func UseSystemd() bool {
// Not critical because of the stop unit logic above.
theConn.StopUnit(scope, "replace", nil)
// Assume StartTransientUnit on a scope allows Delegate
hasDelegate = true
dl := newProp("Delegate", true)
if _, err := theConn.StartTransientUnit(scope, "replace", []systemdDbus.Property{dl}, nil); err != nil {
if dbusError, ok := err.(dbus.Error); ok {
if strings.Contains(dbusError.Name, "org.freedesktop.DBus.Error.PropertyReadOnly") {
hasDelegate = false
}
}
}
// Assume we have the ability to start a transient unit as a slice
// This was broken until systemd v229, but has been back-ported on RHEL environments >= 219
// For details, see: https://bugzilla.redhat.com/show_bug.cgi?id=1370299
hasStartTransientSliceUnit = true
// To ensure simple clean-up, we create a slice off the root with no hierarchy
slice := fmt.Sprintf("libcontainer_%d_systemd_test_default.slice", os.Getpid())
if _, err := theConn.StartTransientUnit(slice, "replace", nil, nil); err != nil {
if _, ok := err.(dbus.Error); ok {
hasStartTransientSliceUnit = false
}
}
for i := 0; i <= testSliceWait; i++ {
if _, err := theConn.StopUnit(slice, "replace", nil); err != nil {
if dbusError, ok := err.(dbus.Error); ok {
if strings.Contains(dbusError.Name, "org.freedesktop.systemd1.NoSuchUnit") {
hasStartTransientSliceUnit = false
break
}
}
} else {
break
}
time.Sleep(time.Millisecond)
}
// Not critical because of the stop unit logic above.
theConn.StopUnit(scope, "replace", nil)
theConn.StopUnit(slice, "replace", nil)
}
return hasStartTransientUnit
}
@@ -222,29 +179,13 @@ func (m *Manager) Apply(pid int) error {
slice = c.Parent
}
properties = append(properties, systemdDbus.PropDescription("libcontainer container "+c.Name))
// if we create a slice, the parent is defined via a Wants=
if strings.HasSuffix(unitName, ".slice") {
// This was broken until systemd v229, but has been back-ported on RHEL environments >= 219
if !hasStartTransientSliceUnit {
return fmt.Errorf("systemd version does not support ability to start a slice as transient unit")
}
properties = append(properties, systemdDbus.PropWants(slice))
} else {
// otherwise, we use Slice=
properties = append(properties, systemdDbus.PropSlice(slice))
}
// only add pid if its valid, -1 is used w/ general slice creation.
if pid != -1 {
properties = append(properties, newProp("PIDs", []uint32{uint32(pid)}))
}
if hasDelegate {
properties = append(properties,
systemdDbus.PropSlice(slice),
systemdDbus.PropDescription("docker container "+c.Name),
newProp("PIDs", []uint32{uint32(pid)}),
// This is only supported on systemd versions 218 and above.
properties = append(properties, newProp("Delegate", true))
}
newProp("Delegate", true),
)
// Always enable accounting, this gets us the same behaviour as the fs implementation,
// plus the kernel has some problems with joining the memory cgroup at a later time.
@@ -260,19 +201,12 @@ func (m *Manager) Apply(pid int) error {
if c.Resources.Memory != 0 {
properties = append(properties,
newProp("MemoryLimit", c.Resources.Memory))
newProp("MemoryLimit", uint64(c.Resources.Memory)))
}
if c.Resources.CpuShares != 0 {
properties = append(properties,
newProp("CPUShares", c.Resources.CpuShares))
}
// cpu.cfs_quota_us and cpu.cfs_period_us are controlled by systemd.
if c.Resources.CpuQuota != 0 && c.Resources.CpuPeriod != 0 {
cpuQuotaPerSecUSec := uint64(c.Resources.CpuQuota*1000000) / c.Resources.CpuPeriod
properties = append(properties,
newProp("CPUQuotaPerSecUSec", cpuQuotaPerSecUSec))
newProp("CPUShares", uint64(c.Resources.CpuShares)))
}
if c.Resources.BlkioWeight != 0 {
@@ -280,15 +214,17 @@ func (m *Manager) Apply(pid int) error {
newProp("BlockIOWeight", uint64(c.Resources.BlkioWeight)))
}
// We have to set kernel memory here, as we can't change it once
// processes have been attached to the cgroup.
if c.Resources.KernelMemory != 0 {
// We need to set kernel memory before processes join cgroup because
// kmem.limit_in_bytes can only be set when the cgroup is empty.
// And swap memory limit needs to be set after memory limit, only
// memory limit is handled by systemd, so it's kind of ugly here.
if c.Resources.KernelMemory > 0 {
if err := setKernelMemory(c); err != nil {
return err
}
}
if _, err := theConn.StartTransientUnit(unitName, "replace", properties, nil); err != nil && !isUnitExists(err) {
if _, err := theConn.StartTransientUnit(unitName, "replace", properties, nil); err != nil {
return err
}
@@ -333,6 +269,15 @@ func (m *Manager) GetPaths() map[string]string {
return paths
}
func writeFile(dir, file, data string) error {
// Normally dir should not be empty, one case is that cgroup subsystem
// is not mounted, we will get empty dir, and we want it fail here.
if dir == "" {
return fmt.Errorf("no such directory for %s.", file)
}
return ioutil.WriteFile(filepath.Join(dir, file), []byte(data), 0700)
}
func join(c *configs.Cgroup, subsystem string, pid int) (string, error) {
path, err := getSubsystemPath(c, subsystem)
if err != nil {
@@ -341,9 +286,10 @@ func join(c *configs.Cgroup, subsystem string, pid int) (string, error) {
if err := os.MkdirAll(path, 0755); err != nil {
return "", err
}
if err := cgroups.WriteCgroupProc(path, pid); err != nil {
if err := writeFile(path, "cgroup.procs", strconv.Itoa(pid)); err != nil {
return "", err
}
return path, nil
}
@@ -353,6 +299,7 @@ func joinCgroups(c *configs.Cgroup, pid int) error {
switch name {
case "name=systemd":
// let systemd handle this
break
case "cpuset":
path, err := getSubsystemPath(c, name)
if err != nil && !cgroups.IsNotFound(err) {
@@ -362,6 +309,7 @@ func joinCgroups(c *configs.Cgroup, pid int) error {
if err := s.ApplyDir(path, c, pid); err != nil {
return err
}
break
default:
_, err := join(c, name, pid)
if err != nil {
@@ -383,10 +331,10 @@ func joinCgroups(c *configs.Cgroup, pid int) error {
return nil
}
// systemd represents slice hierarchy using `-`, so we need to follow suit when
// systemd represents slice heirarchy using `-`, so we need to follow suit when
// generating the path of slice. Essentially, test-a-b.slice becomes
// test.slice/test-a.slice/test-a-b.slice.
func ExpandSlice(slice string) (string, error) {
func expandSlice(slice string) (string, error) {
suffix := ".slice"
// Name has to end with ".slice", but can't be just ".slice".
if len(slice) < len(suffix) || !strings.HasSuffix(slice, suffix) {
@@ -400,10 +348,6 @@ func ExpandSlice(slice string) (string, error) {
var path, prefix string
sliceName := strings.TrimSuffix(slice, suffix)
// if input was -.slice, we should just return root now
if sliceName == "-" {
return "/", nil
}
for _, component := range strings.Split(sliceName, "-") {
// test--a.slice isn't permitted, nor is -test.slice.
if component == "" {
@@ -424,19 +368,17 @@ func getSubsystemPath(c *configs.Cgroup, subsystem string) (string, error) {
return "", err
}
initPath, err := cgroups.GetInitCgroup(subsystem)
initPath, err := cgroups.GetInitCgroupDir(subsystem)
if err != nil {
return "", err
}
// if pid 1 is systemd 226 or later, it will be in init.scope, not the root
initPath = strings.TrimSuffix(filepath.Clean(initPath), "init.scope")
slice := "system.slice"
if c.Parent != "" {
slice = c.Parent
}
slice, err = ExpandSlice(slice)
slice, err = expandSlice(slice)
if err != nil {
return "", err
}
@@ -497,11 +439,6 @@ func (m *Manager) GetStats() (*cgroups.Stats, error) {
}
func (m *Manager) Set(container *configs.Config) error {
// If Paths are set, then we are just joining cgroups paths
// and there is no need to set any values.
if m.Cgroups.Paths != nil {
return nil
}
for _, sys := range subsystems {
// Get the subsystem path, but don't error out for not found cgroups.
path, err := getSubsystemPath(container.Cgroups, sys.Name())
@@ -523,11 +460,7 @@ func (m *Manager) Set(container *configs.Config) error {
}
func getUnitName(c *configs.Cgroup) string {
// by default, we create a scope unless the user explicitly asks for a slice.
if !strings.HasSuffix(c.Name, ".slice") {
return fmt.Sprintf("%s-%s.scope", c.ScopePrefix, c.Name)
}
return c.Name
return fmt.Sprintf("%s-%s.scope", c.ScopePrefix, c.Name)
}
func setKernelMemory(c *configs.Cgroup) error {
@@ -539,15 +472,8 @@ func setKernelMemory(c *configs.Cgroup) error {
if err := os.MkdirAll(path, 0755); err != nil {
return err
}
return fs.EnableKernelMemoryAccounting(path)
}
// isUnitExists returns true if the error is that a systemd unit already exists.
func isUnitExists(err error) bool {
if err != nil {
if dbusError, ok := err.(dbus.Error); ok {
return strings.Contains(dbusError.Name, "org.freedesktop.systemd1.UnitExists")
}
}
return false
// This doesn't get called by manager.Set, so we need to do it here.
s := &fs.MemoryGroup{}
return s.SetKernelMemory(path, c)
}

View File

@@ -16,24 +16,37 @@ import (
"github.com/docker/go-units"
)
const (
cgroupNamePrefix = "name="
CgroupProcesses = "cgroup.procs"
)
const cgroupNamePrefix = "name="
// https://www.kernel.org/doc/Documentation/cgroup-v1/cgroups.txt
// https://www.kernel.org/doc/Documentation/cgroups/cgroups.txt
func FindCgroupMountpoint(subsystem string) (string, error) {
mnt, _, err := FindCgroupMountpointAndRoot(subsystem)
return mnt, err
}
func FindCgroupMountpointAndRoot(subsystem string) (string, string, error) {
// We are not using mount.GetMounts() because it's super-inefficient,
// parsing it directly sped up x10 times because of not using Sscanf.
// It was one of two major performance drawbacks in container start.
if !isSubsystemAvailable(subsystem) {
return "", "", NewNotFoundError(subsystem)
f, err := os.Open("/proc/self/mountinfo")
if err != nil {
return "", err
}
defer f.Close()
scanner := bufio.NewScanner(f)
for scanner.Scan() {
txt := scanner.Text()
fields := strings.Split(txt, " ")
for _, opt := range strings.Split(fields[len(fields)-1], ",") {
if opt == subsystem {
return fields[4], nil
}
}
}
if err := scanner.Err(); err != nil {
return "", err
}
return "", NewNotFoundError(subsystem)
}
func FindCgroupMountpointAndRoot(subsystem string) (string, string, error) {
f, err := os.Open("/proc/self/mountinfo")
if err != nil {
return "", "", err
@@ -57,30 +70,6 @@ func FindCgroupMountpointAndRoot(subsystem string) (string, string, error) {
return "", "", NewNotFoundError(subsystem)
}
func isSubsystemAvailable(subsystem string) bool {
cgroups, err := ParseCgroupFile("/proc/self/cgroup")
if err != nil {
return false
}
_, avail := cgroups[subsystem]
return avail
}
func GetClosestMountpointAncestor(dir, mountinfo string) string {
deepestMountPoint := ""
for _, mountInfoEntry := range strings.Split(mountinfo, "\n") {
mountInfoParts := strings.Fields(mountInfoEntry)
if len(mountInfoParts) < 5 {
continue
}
mountPoint := mountInfoParts[4]
if strings.HasPrefix(mountPoint, deepestMountPoint) && strings.HasPrefix(dir, mountPoint) {
deepestMountPoint = mountPoint
}
}
return deepestMountPoint
}
func FindCgroupMountpointDir() (string, error) {
f, err := os.Open("/proc/self/mountinfo")
if err != nil {
@@ -124,7 +113,7 @@ type Mount struct {
Subsystems []string
}
func (m Mount) GetOwnCgroup(cgroups map[string]string) (string, error) {
func (m Mount) GetThisCgroupDir(cgroups map[string]string) (string, error) {
if len(m.Subsystems) == 0 {
return "", fmt.Errorf("no subsystem for mount")
}
@@ -132,17 +121,16 @@ func (m Mount) GetOwnCgroup(cgroups map[string]string) (string, error) {
return getControllerPath(m.Subsystems[0], cgroups)
}
func getCgroupMountsHelper(ss map[string]bool, mi io.Reader, all bool) ([]Mount, error) {
func getCgroupMountsHelper(ss map[string]bool, mi io.Reader) ([]Mount, error) {
res := make([]Mount, 0, len(ss))
scanner := bufio.NewScanner(mi)
numFound := 0
for scanner.Scan() && numFound < len(ss) {
for scanner.Scan() {
txt := scanner.Text()
sepIdx := strings.Index(txt, " - ")
if sepIdx == -1 {
return nil, fmt.Errorf("invalid mountinfo format")
}
if txt[sepIdx+3:sepIdx+10] == "cgroup2" || txt[sepIdx+3:sepIdx+9] != "cgroup" {
if txt[sepIdx+3:sepIdx+9] != "cgroup" {
continue
}
fields := strings.Split(txt, " ")
@@ -151,16 +139,11 @@ func getCgroupMountsHelper(ss map[string]bool, mi io.Reader, all bool) ([]Mount,
Root: fields[3],
}
for _, opt := range strings.Split(fields[len(fields)-1], ",") {
if !ss[opt] {
continue
}
if strings.HasPrefix(opt, cgroupNamePrefix) {
m.Subsystems = append(m.Subsystems, opt[len(cgroupNamePrefix):])
} else {
m.Subsystems = append(m.Subsystems, opt)
}
if !all {
numFound++
if ss[opt] {
m.Subsystems = append(m.Subsystems, opt)
}
}
res = append(res, m)
@@ -171,28 +154,26 @@ func getCgroupMountsHelper(ss map[string]bool, mi io.Reader, all bool) ([]Mount,
return res, nil
}
// GetCgroupMounts returns the mounts for the cgroup subsystems.
// all indicates whether to return just the first instance or all the mounts.
func GetCgroupMounts(all bool) ([]Mount, error) {
func GetCgroupMounts() ([]Mount, error) {
f, err := os.Open("/proc/self/mountinfo")
if err != nil {
return nil, err
}
defer f.Close()
allSubsystems, err := ParseCgroupFile("/proc/self/cgroup")
all, err := GetAllSubsystems()
if err != nil {
return nil, err
}
allMap := make(map[string]bool)
for s := range allSubsystems {
for _, s := range all {
allMap[s] = true
}
return getCgroupMountsHelper(allMap, f, all)
return getCgroupMountsHelper(allMap, f)
}
// GetAllSubsystems returns all the cgroup subsystems supported by the kernel
// Returns all the cgroup subsystems supported by the kernel
func GetAllSubsystems() ([]string, error) {
f, err := os.Open("/proc/cgroups")
if err != nil {
@@ -204,6 +185,9 @@ func GetAllSubsystems() ([]string, error) {
s := bufio.NewScanner(f)
for s.Scan() {
if err := s.Err(); err != nil {
return nil, err
}
text := s.Text()
if text[0] != '#' {
parts := strings.Fields(text)
@@ -212,14 +196,11 @@ func GetAllSubsystems() ([]string, error) {
}
}
}
if err := s.Err(); err != nil {
return nil, err
}
return subsystems, nil
}
// GetOwnCgroup returns the relative path to the cgroup docker is running in.
func GetOwnCgroup(subsystem string) (string, error) {
// Returns the relative path to the cgroup docker is running in.
func GetThisCgroupDir(subsystem string) (string, error) {
cgroups, err := ParseCgroupFile("/proc/self/cgroup")
if err != nil {
return "", err
@@ -228,16 +209,8 @@ func GetOwnCgroup(subsystem string) (string, error) {
return getControllerPath(subsystem, cgroups)
}
func GetOwnCgroupPath(subsystem string) (string, error) {
cgroup, err := GetOwnCgroup(subsystem)
if err != nil {
return "", err
}
func GetInitCgroupDir(subsystem string) (string, error) {
return getCgroupPathHelper(subsystem, cgroup)
}
func GetInitCgroup(subsystem string) (string, error) {
cgroups, err := ParseCgroupFile("/proc/1/cgroup")
if err != nil {
return "", err
@@ -246,33 +219,8 @@ func GetInitCgroup(subsystem string) (string, error) {
return getControllerPath(subsystem, cgroups)
}
func GetInitCgroupPath(subsystem string) (string, error) {
cgroup, err := GetInitCgroup(subsystem)
if err != nil {
return "", err
}
return getCgroupPathHelper(subsystem, cgroup)
}
func getCgroupPathHelper(subsystem, cgroup string) (string, error) {
mnt, root, err := FindCgroupMountpointAndRoot(subsystem)
if err != nil {
return "", err
}
// This is needed for nested containers, because in /proc/self/cgroup we
// see pathes from host, which don't exist in container.
relCgroup, err := filepath.Rel(root, cgroup)
if err != nil {
return "", err
}
return filepath.Join(mnt, relCgroup), nil
}
func readProcsFile(dir string) ([]int, error) {
f, err := os.Open(filepath.Join(dir, CgroupProcesses))
f, err := os.Open(filepath.Join(dir, "cgroup.procs"))
if err != nil {
return nil, err
}
@@ -295,8 +243,6 @@ func readProcsFile(dir string) ([]int, error) {
return out, nil
}
// ParseCgroupFile parses the given cgroup file, typically from
// /proc/<pid>/cgroup, into a map of subgroups to cgroup names.
func ParseCgroupFile(path string) (map[string]string, error) {
f, err := os.Open(path)
if err != nil {
@@ -304,35 +250,21 @@ func ParseCgroupFile(path string) (map[string]string, error) {
}
defer f.Close()
return parseCgroupFromReader(f)
}
// helper function for ParseCgroupFile to make testing easier
func parseCgroupFromReader(r io.Reader) (map[string]string, error) {
s := bufio.NewScanner(r)
s := bufio.NewScanner(f)
cgroups := make(map[string]string)
for s.Scan() {
text := s.Text()
// from cgroups(7):
// /proc/[pid]/cgroup
// ...
// For each cgroup hierarchy ... there is one entry
// containing three colon-separated fields of the form:
// hierarchy-ID:subsystem-list:cgroup-path
parts := strings.SplitN(text, ":", 3)
if len(parts) < 3 {
return nil, fmt.Errorf("invalid cgroup entry: must contain at least two colons: %v", text)
if err := s.Err(); err != nil {
return nil, err
}
text := s.Text()
parts := strings.Split(text, ":")
for _, subs := range strings.Split(parts[1], ",") {
cgroups[subs] = parts[2]
}
}
if err := s.Err(); err != nil {
return nil, err
}
return cgroups, nil
}
@@ -359,7 +291,8 @@ func PathExists(path string) bool {
func EnterPid(cgroupPaths map[string]string, pid int) error {
for _, path := range cgroupPaths {
if PathExists(path) {
if err := WriteCgroupProc(path, pid); err != nil {
if err := ioutil.WriteFile(filepath.Join(path, "cgroup.procs"),
[]byte(strconv.Itoa(pid)), 0700); err != nil {
return err
}
}
@@ -428,7 +361,7 @@ func GetAllPids(path string) ([]int, error) {
// collect pids from all sub-cgroups
err := filepath.Walk(path, func(p string, info os.FileInfo, iErr error) error {
dir, file := filepath.Split(p)
if file != CgroupProcesses {
if file != "cgroup.procs" {
return nil
}
if iErr != nil {
@@ -443,20 +376,3 @@ func GetAllPids(path string) ([]int, error) {
})
return pids, err
}
// WriteCgroupProc writes the specified pid into the cgroup's cgroup.procs file
func WriteCgroupProc(dir string, pid int) error {
// Normally dir should not be empty, one case is that cgroup subsystem
// is not mounted, we will get empty dir, and we want it fail here.
if dir == "" {
return fmt.Errorf("no such directory for %s", CgroupProcesses)
}
// Dont attach any pid to the cgroup if -1 is specified as a pid
if pid != -1 {
if err := ioutil.WriteFile(filepath.Join(dir, CgroupProcesses), []byte(strconv.Itoa(pid)), 0700); err != nil {
return fmt.Errorf("failed to write %v to %v: %v", pid, CgroupProcesses, err)
}
}
return nil
}

View File

@@ -4,8 +4,6 @@ package cgroups
import (
"bytes"
"fmt"
"reflect"
"strings"
"testing"
)
@@ -93,34 +91,6 @@ const systemdMountinfo = `115 83 0:32 / / rw,relatime - aufs none rw,si=c0bd3d3,
136 117 0:12 /1 /dev/console rw,nosuid,noexec,relatime - devpts none rw,gid=5,mode=620,ptmxmode=000
84 115 0:40 / /tmp rw,relatime - tmpfs none rw`
const cgroup2Mountinfo = `18 64 0:18 / /sys rw,nosuid,nodev,noexec,relatime shared:6 - sysfs sysfs rw,seclabel
19 64 0:4 / /proc rw,nosuid,nodev,noexec,relatime shared:5 - proc proc rw
20 64 0:6 / /dev rw,nosuid shared:2 - devtmpfs devtmpfs rw,seclabel,size=8171204k,nr_inodes=2042801,mode=755
21 18 0:19 / /sys/kernel/security rw,nosuid,nodev,noexec,relatime shared:7 - securityfs securityfs rw
22 20 0:20 / /dev/shm rw,nosuid,nodev shared:3 - tmpfs tmpfs rw,seclabel
23 20 0:21 / /dev/pts rw,nosuid,noexec,relatime shared:4 - devpts devpts rw,seclabel,gid=5,mode=620,ptmxmode=000
24 64 0:22 / /run rw,nosuid,nodev shared:24 - tmpfs tmpfs rw,seclabel,mode=755
25 18 0:23 / /sys/fs/cgroup ro,nosuid,nodev,noexec shared:8 - tmpfs tmpfs ro,seclabel,mode=755
26 25 0:24 / /sys/fs/cgroup/systemd rw,nosuid,nodev,noexec,relatime shared:9 - cgroup2 cgroup rw
27 18 0:25 / /sys/fs/pstore rw,nosuid,nodev,noexec,relatime shared:20 - pstore pstore rw,seclabel
28 18 0:26 / /sys/firmware/efi/efivars rw,nosuid,nodev,noexec,relatime shared:21 - efivarfs efivarfs rw
29 25 0:27 / /sys/fs/cgroup/cpu,cpuacct rw,nosuid,nodev,noexec,relatime shared:10 - cgroup cgroup rw,cpu,cpuacct
30 25 0:28 / /sys/fs/cgroup/memory rw,nosuid,nodev,noexec,relatime shared:11 - cgroup cgroup rw,memory
31 25 0:29 / /sys/fs/cgroup/net_cls,net_prio rw,nosuid,nodev,noexec,relatime shared:12 - cgroup cgroup rw,net_cls,net_prio
32 25 0:30 / /sys/fs/cgroup/blkio rw,nosuid,nodev,noexec,relatime shared:13 - cgroup cgroup rw,blkio
33 25 0:31 / /sys/fs/cgroup/perf_event rw,nosuid,nodev,noexec,relatime shared:14 - cgroup cgroup rw,perf_event
34 25 0:32 / /sys/fs/cgroup/hugetlb rw,nosuid,nodev,noexec,relatime shared:15 - cgroup cgroup rw,hugetlb
35 25 0:33 / /sys/fs/cgroup/freezer rw,nosuid,nodev,noexec,relatime shared:16 - cgroup cgroup rw,freezer
36 25 0:34 / /sys/fs/cgroup/cpuset rw,nosuid,nodev,noexec,relatime shared:17 - cgroup cgroup rw,cpuset
37 25 0:35 / /sys/fs/cgroup/devices rw,nosuid,nodev,noexec,relatime shared:18 - cgroup cgroup rw,devices
38 25 0:36 / /sys/fs/cgroup/pids rw,nosuid,nodev,noexec,relatime shared:19 - cgroup cgroup rw,pids
61 18 0:37 / /sys/kernel/config rw,relatime shared:22 - configfs configfs rw
64 0 253:0 / / rw,relatime shared:1 - ext4 /dev/mapper/fedora_dhcp--16--129-root rw,seclabel,data=ordered
39 18 0:17 / /sys/fs/selinux rw,relatime shared:23 - selinuxfs selinuxfs rw
40 20 0:16 / /dev/mqueue rw,relatime shared:25 - mqueue mqueue rw,seclabel
41 20 0:39 / /dev/hugepages rw,relatime shared:26 - hugetlbfs hugetlbfs rw,seclabel
`
func TestGetCgroupMounts(t *testing.T) {
type testData struct {
mountInfo string
@@ -162,7 +132,7 @@ func TestGetCgroupMounts(t *testing.T) {
}
for _, td := range testTable {
mi := bytes.NewBufferString(td.mountInfo)
cgMounts, err := getCgroupMountsHelper(td.subsystems, mi, false)
cgMounts, err := getCgroupMountsHelper(td.subsystems, mi)
if err != nil {
t.Fatal(err)
}
@@ -215,119 +185,8 @@ func BenchmarkGetCgroupMounts(b *testing.B) {
b.StopTimer()
mi := bytes.NewBufferString(fedoraMountinfo)
b.StartTimer()
if _, err := getCgroupMountsHelper(subsystems, mi, false); err != nil {
if _, err := getCgroupMountsHelper(subsystems, mi); err != nil {
b.Fatal(err)
}
}
}
func TestParseCgroupString(t *testing.T) {
testCases := []struct {
input string
expectedError error
expectedOutput map[string]string
}{
{
// Taken from a CoreOS instance running systemd 225 with CPU/Mem
// accounting enabled in systemd
input: `9:blkio:/
8:freezer:/
7:perf_event:/
6:devices:/system.slice/system-sshd.slice
5:cpuset:/
4:cpu,cpuacct:/system.slice/system-sshd.slice/sshd@126-10.240.0.15:22-xxx.yyy.zzz.aaa:33678.service
3:net_cls,net_prio:/
2:memory:/system.slice/system-sshd.slice/sshd@126-10.240.0.15:22-xxx.yyy.zzz.aaa:33678.service
1:name=systemd:/system.slice/system-sshd.slice/sshd@126-10.240.0.15:22-xxx.yyy.zzz.aaa:33678.service`,
expectedOutput: map[string]string{
"name=systemd": "/system.slice/system-sshd.slice/sshd@126-10.240.0.15:22-xxx.yyy.zzz.aaa:33678.service",
"blkio": "/",
"freezer": "/",
"perf_event": "/",
"devices": "/system.slice/system-sshd.slice",
"cpuset": "/",
"cpu": "/system.slice/system-sshd.slice/sshd@126-10.240.0.15:22-xxx.yyy.zzz.aaa:33678.service",
"cpuacct": "/system.slice/system-sshd.slice/sshd@126-10.240.0.15:22-xxx.yyy.zzz.aaa:33678.service",
"net_cls": "/",
"net_prio": "/",
"memory": "/system.slice/system-sshd.slice/sshd@126-10.240.0.15:22-xxx.yyy.zzz.aaa:33678.service",
},
},
{
input: `malformed input`,
expectedError: fmt.Errorf(`invalid cgroup entry: must contain at least two colons: malformed input`),
},
}
for ndx, testCase := range testCases {
out, err := parseCgroupFromReader(strings.NewReader(testCase.input))
if err != nil {
if testCase.expectedError == nil || testCase.expectedError.Error() != err.Error() {
t.Errorf("%v: expected error %v, got error %v", ndx, testCase.expectedError, err)
}
} else {
if !reflect.DeepEqual(testCase.expectedOutput, out) {
t.Errorf("%v: expected output %v, got error %v", ndx, testCase.expectedOutput, out)
}
}
}
}
func TestIgnoreCgroup2Mount(t *testing.T) {
subsystems := map[string]bool{
"cpuset": true,
"cpu": true,
"cpuacct": true,
"memory": true,
"devices": true,
"freezer": true,
"net_cls": true,
"blkio": true,
"perf_event": true,
"pids": true,
"name=systemd": true,
}
mi := bytes.NewBufferString(cgroup2Mountinfo)
cgMounts, err := getCgroupMountsHelper(subsystems, mi, false)
if err != nil {
t.Fatal(err)
}
for _, m := range cgMounts {
if m.Mountpoint == "/sys/fs/cgroup/systemd" {
t.Errorf("parsed a cgroup2 mount at /sys/fs/cgroup/systemd instead of ignoring it")
}
}
}
const fakeMountInfo = ` 18 24 0:17 / /sys rw,nosuid,nodev,noexec,relatime - sysfs sysfs rw
100 99 1:31 / /foo/bar rw,relatime - fake fake rw,fake
100 99 1:31 / /foo/bar/baz2 rw,relatime - fake fake rw,fake
100 99 1:31 / /foo/bar/baz rw,relatime - fake fake rw,fake
100 99 1:31 / /foo/bar/bazza rw,relatime - fake fake rw,fake
100 99 1:31 / /foo/bar/baz3 rw,relatime - fake fake rw,fake
100 99 1:31 / /foo rw,relatime - fake fake rw,fake
100 99 1:31 / /unrelated rw,relatime - fake fake rw,fake
100 99 1:31 / / rw,relatime - fake fake rw,fake
`
func TestGetClosestMountpointAncestor(t *testing.T) {
testCases := []struct {
input string
mountinfos string
output string
}{
{input: "/foo/bar/baz/a/b/c", mountinfos: fakeMountInfo, output: "/foo/bar/baz"},
{input: "/foo/bar/baz", mountinfos: fakeMountInfo, output: "/foo/bar/baz"},
{input: "/foo/bar/bazza", mountinfos: fakeMountInfo, output: "/foo/bar/bazza"},
{input: "/a/b/c/d", mountinfos: fakeMountInfo, output: "/"},
}
for _, c := range testCases {
mountpoint := GetClosestMountpointAncestor(c.input, c.mountinfos)
if mountpoint != c.output {
t.Errorf("expected %s, got %s", c.output, mountpoint)
}
}
}

View File

@@ -1,3 +1,5 @@
// +build linux freebsd
package configs
type FreezerState string
@@ -20,7 +22,7 @@ type Cgroup struct {
// The path is assumed to be relative to the host system cgroup mountpoint.
Path string `json:"path"`
// ScopePrefix describes prefix for the scope name
// ScopePrefix decribes prefix for the scope name
ScopePrefix string `json:"scope_prefix"`
// Paths represent the absolute cgroups paths to join.
@@ -34,7 +36,7 @@ type Cgroup struct {
type Resources struct {
// If this is true allow access to any kind of device within the container. If false, allow access only to devices explicitly listed in the allowed_devices list.
// Deprecated
AllowAllDevices *bool `json:"allow_all_devices,omitempty"`
AllowAllDevices bool `json:"allow_all_devices,omitempty"`
// Deprecated
AllowedDevices []*Device `json:"allowed_devices,omitempty"`
// Deprecated
@@ -58,19 +60,19 @@ type Resources struct {
KernelMemoryTCP int64 `json:"kernel_memory_tcp"`
// CPU shares (relative weight vs. other containers)
CpuShares uint64 `json:"cpu_shares"`
CpuShares int64 `json:"cpu_shares"`
// CPU hardcap limit (in usecs). Allowed cpu time in a given period.
CpuQuota int64 `json:"cpu_quota"`
// CPU period to be used for hardcapping (in usecs). 0 to use system default.
CpuPeriod uint64 `json:"cpu_period"`
CpuPeriod int64 `json:"cpu_period"`
// How many time CPU will use in realtime scheduling (in usecs).
CpuRtRuntime int64 `json:"cpu_rt_quota"`
CpuRtRuntime int64 `json:"cpu_quota"`
// CPU period to be used for realtime scheduling (in usecs).
CpuRtPeriod uint64 `json:"cpu_rt_period"`
CpuRtPeriod int64 `json:"cpu_period"`
// CPU to use
CpusetCpus string `json:"cpuset_cpus"`
@@ -93,7 +95,7 @@ type Resources struct {
// IO read rate limit per cgroup per device, bytes per second.
BlkioThrottleReadBpsDevice []*ThrottleDevice `json:"blkio_throttle_read_bps_device"`
// IO write rate limit per cgroup per device, bytes per second.
// IO write rate limit per cgroup per divice, bytes per second.
BlkioThrottleWriteBpsDevice []*ThrottleDevice `json:"blkio_throttle_write_bps_device"`
// IO read rate limit per cgroup per device, IO per second.
@@ -112,11 +114,11 @@ type Resources struct {
OomKillDisable bool `json:"oom_kill_disable"`
// Tuning swappiness behaviour per cgroup
MemorySwappiness *uint64 `json:"memory_swappiness"`
MemorySwappiness *int64 `json:"memory_swappiness"`
// Set priority of network traffic for container
NetPrioIfpriomap []*IfPrioMap `json:"net_prio_ifpriomap"`
// Set class identifier for container's network packets
NetClsClassid uint32 `json:"net_cls_classid_u"`
NetClsClassid string `json:"net_cls_classid"`
}

View File

@@ -7,9 +7,7 @@ import (
"os/exec"
"time"
"github.com/opencontainers/runtime-spec/specs-go"
"github.com/sirupsen/logrus"
"github.com/Sirupsen/logrus"
)
type Rlimit struct {
@@ -35,7 +33,7 @@ type Seccomp struct {
Syscalls []*Syscall `json:"syscalls"`
}
// Action is taken upon rule match in Seccomp
// An action to be taken upon rule match in Seccomp
type Action int
const (
@@ -46,7 +44,7 @@ const (
Trace
)
// Operator is a comparison operator to be used when matching syscall arguments in Seccomp
// A comparison operator to be used when matching syscall arguments in Seccomp
type Operator int
const (
@@ -59,7 +57,7 @@ const (
MaskEqualTo
)
// Arg is a rule to match a specific syscall argument in Seccomp
// A rule to match a specific syscall argument in Seccomp
type Arg struct {
Index uint `json:"index"`
Value uint64 `json:"value"`
@@ -67,7 +65,7 @@ type Arg struct {
Op Operator `json:"op"`
}
// Syscall is a rule to match a syscall in Seccomp
// An rule to match a syscall in Seccomp
type Syscall struct {
Name string `json:"name"`
Action Action `json:"action"`
@@ -87,6 +85,11 @@ type Config struct {
// that the parent process dies.
ParentDeathSignal int `json:"parent_death_signal"`
// PivotDir allows a custom directory inside the container's root filesystem to be used as pivot, when NoPivotRoot is not set.
// When a custom PivotDir not set, a temporary dir inside the root filesystem will be used. The pivot dir needs to be writeable.
// This is required when using read only root filesystems. In these cases, a read/writeable path can be (bind) mounted somewhere inside the root filesystem to act as pivot.
PivotDir string `json:"pivot_dir"`
// Path to a directory containing the container's root filesystem.
Rootfs string `json:"rootfs"`
@@ -114,8 +117,8 @@ type Config struct {
Namespaces Namespaces `json:"namespaces"`
// Capabilities specify the capabilities to keep when executing the process inside the container
// All capabilities not specified will be dropped from the processes capability mask
Capabilities *Capabilities `json:"capabilities"`
// All capbilities not specified will be dropped from the processes capability mask
Capabilities []string `json:"capabilities"`
// Networks specifies the container's network setup to be created
Networks []*Network `json:"networks"`
@@ -145,6 +148,10 @@ type Config struct {
// More information about kernel oom score calculation here: https://lwn.net/Articles/317814/
OomScoreAdj int `json:"oom_score_adj"`
// AdditionalGroups specifies the gids that should be added to supplementary groups
// in addition to those that the user belongs to.
AdditionalGroups []string `json:"additional_groups"`
// UidMappings is an array of User ID mappings for User Namespaces
UidMappings []IDMap `json:"uid_mappings"`
@@ -180,13 +187,6 @@ type Config struct {
// Labels are user defined metadata that is stored in the config and populated on the state
Labels []string `json:"labels"`
// NoNewKeyring will not allocated a new session keyring for the container. It will use the
// callers keyring in this case.
NoNewKeyring bool `json:"no_new_keyring"`
// Rootless specifies whether the container is a rootless container.
Rootless bool `json:"rootless"`
}
type Hooks struct {
@@ -201,19 +201,6 @@ type Hooks struct {
Poststop []Hook
}
type Capabilities struct {
// Bounding is the set of capabilities checked by the kernel.
Bounding []string
// Effective is the set of capabilities checked by the kernel.
Effective []string
// Inheritable is the capabilities preserved across execve.
Inheritable []string
// Permitted is the limiting superset for effective capabilities.
Permitted []string
// Ambient is the ambient set of capabilities that are kept.
Ambient []string
}
func (hooks *Hooks) UnmarshalJSON(b []byte) error {
var state struct {
Prestart []CommandHook
@@ -261,14 +248,20 @@ func (hooks Hooks) MarshalJSON() ([]byte, error) {
}
// HookState is the payload provided to a hook on execution.
type HookState specs.State
type HookState struct {
Version string `json:"ociVersion"`
ID string `json:"id"`
Pid int `json:"pid"`
Root string `json:"root"`
BundlePath string `json:"bundlePath"`
}
type Hook interface {
// Run executes the hook with the provided state.
Run(HookState) error
}
// NewFunctionHook will call the provided function when the hook is run.
// NewFunctionHooks will call the provided function when the hook is run.
func NewFunctionHook(f func(HookState) error) FuncHook {
return FuncHook{
run: f,
@@ -291,7 +284,7 @@ type Command struct {
Timeout *time.Duration `json:"timeout"`
}
// NewCommandHook will execute the provided command when the hook is run.
// NewCommandHooks will execute the provided command when the hook is run.
func NewCommandHook(cmd Command) CommandHook {
return CommandHook{
Command: cmd,
@@ -307,38 +300,29 @@ func (c Command) Run(s HookState) error {
if err != nil {
return err
}
var stdout, stderr bytes.Buffer
cmd := exec.Cmd{
Path: c.Path,
Args: c.Args,
Env: c.Env,
Stdin: bytes.NewReader(b),
Stdout: &stdout,
Stderr: &stderr,
}
if err := cmd.Start(); err != nil {
return err
Path: c.Path,
Args: c.Args,
Env: c.Env,
Stdin: bytes.NewReader(b),
}
errC := make(chan error, 1)
go func() {
err := cmd.Wait()
out, err := cmd.CombinedOutput()
if err != nil {
err = fmt.Errorf("error running hook: %v, stdout: %s, stderr: %s", err, stdout.String(), stderr.String())
err = fmt.Errorf("%s: %s", err, out)
}
errC <- err
}()
var timerCh <-chan time.Time
if c.Timeout != nil {
timer := time.NewTimer(*c.Timeout)
defer timer.Stop()
timerCh = timer.C
}
select {
case err := <-errC:
return err
case <-timerCh:
cmd.Process.Kill()
cmd.Wait()
return fmt.Errorf("hook ran past specified timeout of %.1fs", c.Timeout.Seconds())
select {
case err := <-errC:
return err
case <-time.After(*c.Timeout):
cmd.Process.Kill()
cmd.Wait()
return fmt.Errorf("hook ran past specified timeout of %.1fs", c.Timeout.Seconds())
}
}
return <-errC
}

View File

@@ -1,61 +0,0 @@
package configs
import "fmt"
// HostUID gets the translated uid for the process on host which could be
// different when user namespaces are enabled.
func (c Config) HostUID(containerId int) (int, error) {
if c.Namespaces.Contains(NEWUSER) {
if c.UidMappings == nil {
return -1, fmt.Errorf("User namespaces enabled, but no uid mappings found.")
}
id, found := c.hostIDFromMapping(containerId, c.UidMappings)
if !found {
return -1, fmt.Errorf("User namespaces enabled, but no user mapping found.")
}
return id, nil
}
// Return unchanged id.
return containerId, nil
}
// HostRootUID gets the root uid for the process on host which could be non-zero
// when user namespaces are enabled.
func (c Config) HostRootUID() (int, error) {
return c.HostUID(0)
}
// HostGID gets the translated gid for the process on host which could be
// different when user namespaces are enabled.
func (c Config) HostGID(containerId int) (int, error) {
if c.Namespaces.Contains(NEWUSER) {
if c.GidMappings == nil {
return -1, fmt.Errorf("User namespaces enabled, but no gid mappings found.")
}
id, found := c.hostIDFromMapping(containerId, c.GidMappings)
if !found {
return -1, fmt.Errorf("User namespaces enabled, but no group mapping found.")
}
return id, nil
}
// Return unchanged id.
return containerId, nil
}
// HostRootGID gets the root gid for the process on host which could be non-zero
// when user namespaces are enabled.
func (c Config) HostRootGID() (int, error) {
return c.HostGID(0)
}
// Utility function that gets a host ID for a container ID from user namespace map
// if that ID is present in the map.
func (c Config) hostIDFromMapping(containerID int, uMap []IDMap) (int, bool) {
for _, m := range uMap {
if (containerID >= m.ContainerID) && (containerID <= (m.ContainerID + m.Size - 1)) {
hostID := m.HostID + (containerID - m.ContainerID)
return hostID, true
}
}
return -1, false
}

View File

@@ -123,7 +123,7 @@ func TestFuncHookRun(t *testing.T) {
Version: "1",
ID: "1",
Pid: 1,
Bundle: "/bundle",
Root: "root",
}
fHook := configs.NewFunctionHook(func(s configs.HookState) error {
@@ -141,7 +141,7 @@ func TestCommandHookRun(t *testing.T) {
Version: "1",
ID: "1",
Pid: 1,
Bundle: "/bundle",
Root: "root",
}
timeout := time.Second
@@ -164,7 +164,7 @@ func TestCommandHookRunTimeout(t *testing.T) {
Version: "1",
ID: "1",
Pid: 1,
Bundle: "/bundle",
Root: "root",
}
timeout := (10 * time.Millisecond)

View File

@@ -0,0 +1,51 @@
// +build freebsd linux
package configs
import "fmt"
// Gets the root uid for the process on host which could be non-zero
// when user namespaces are enabled.
func (c Config) HostUID() (int, error) {
if c.Namespaces.Contains(NEWUSER) {
if c.UidMappings == nil {
return -1, fmt.Errorf("User namespaces enabled, but no user mappings found.")
}
id, found := c.hostIDFromMapping(0, c.UidMappings)
if !found {
return -1, fmt.Errorf("User namespaces enabled, but no root user mapping found.")
}
return id, nil
}
// Return default root uid 0
return 0, nil
}
// Gets the root gid for the process on host which could be non-zero
// when user namespaces are enabled.
func (c Config) HostGID() (int, error) {
if c.Namespaces.Contains(NEWUSER) {
if c.GidMappings == nil {
return -1, fmt.Errorf("User namespaces enabled, but no gid mappings found.")
}
id, found := c.hostIDFromMapping(0, c.GidMappings)
if !found {
return -1, fmt.Errorf("User namespaces enabled, but no root group mapping found.")
}
return id, nil
}
// Return default root gid 0
return 0, nil
}
// Utility function that gets a host ID for a container ID from user namespace map
// if that ID is present in the map.
func (c Config) hostIDFromMapping(containerID int, uMap []IDMap) (int, bool) {
for _, m := range uMap {
if (containerID >= m.ContainerID) && (containerID <= (m.ContainerID + m.Size - 1)) {
hostID := m.HostID + (containerID - m.ContainerID)
return hostID, true
}
}
return -1, false
}

View File

@@ -1,3 +1,5 @@
// +build linux freebsd
package configs
import (
@@ -8,6 +10,30 @@ import (
"testing"
)
// Checks whether the expected capability is specified in the capabilities.
func contains(expected string, values []string) bool {
for _, v := range values {
if v == expected {
return true
}
}
return false
}
func containsDevice(expected *Device, values []*Device) bool {
for _, d := range values {
if d.Path == expected.Path &&
d.Permissions == expected.Permissions &&
d.FileMode == expected.FileMode &&
d.Major == expected.Major &&
d.Minor == expected.Minor &&
d.Type == expected.Type {
return true
}
}
return false
}
func loadConfig(name string) (*Config, error) {
f, err := os.Open(filepath.Join("../sample_configs", name))
if err != nil {
@@ -63,11 +89,11 @@ func TestRemoveNamespace(t *testing.T) {
}
}
func TestHostRootUIDNoUSERNS(t *testing.T) {
func TestHostUIDNoUSERNS(t *testing.T) {
config := &Config{
Namespaces: Namespaces{},
}
uid, err := config.HostRootUID()
uid, err := config.HostUID()
if err != nil {
t.Fatal(err)
}
@@ -76,7 +102,7 @@ func TestHostRootUIDNoUSERNS(t *testing.T) {
}
}
func TestHostRootUIDWithUSERNS(t *testing.T) {
func TestHostUIDWithUSERNS(t *testing.T) {
config := &Config{
Namespaces: Namespaces{{Type: NEWUSER}},
UidMappings: []IDMap{
@@ -87,7 +113,7 @@ func TestHostRootUIDWithUSERNS(t *testing.T) {
},
},
}
uid, err := config.HostRootUID()
uid, err := config.HostUID()
if err != nil {
t.Fatal(err)
}
@@ -96,11 +122,11 @@ func TestHostRootUIDWithUSERNS(t *testing.T) {
}
}
func TestHostRootGIDNoUSERNS(t *testing.T) {
func TestHostGIDNoUSERNS(t *testing.T) {
config := &Config{
Namespaces: Namespaces{},
}
uid, err := config.HostRootGID()
uid, err := config.HostGID()
if err != nil {
t.Fatal(err)
}
@@ -109,7 +135,7 @@ func TestHostRootGIDNoUSERNS(t *testing.T) {
}
}
func TestHostRootGIDWithUSERNS(t *testing.T) {
func TestHostGIDWithUSERNS(t *testing.T) {
config := &Config{
Namespaces: Namespaces{{Type: NEWUSER}},
GidMappings: []IDMap{
@@ -120,7 +146,7 @@ func TestHostRootGIDWithUSERNS(t *testing.T) {
},
},
}
uid, err := config.HostRootGID()
uid, err := config.HostGID()
if err != nil {
t.Fatal(err)
}

View File

@@ -3,7 +3,7 @@
package configs
var (
// DefaultSimpleDevices are devices that are to be both allowed and created.
// These are devices that are to be both allowed and created.
DefaultSimpleDevices = []*Device{
// /dev/null and zero
{
@@ -107,5 +107,19 @@ var (
Permissions: "rwm",
},
}, DefaultSimpleDevices...)
DefaultAutoCreatedDevices = append([]*Device{}, DefaultSimpleDevices...)
DefaultAutoCreatedDevices = append([]*Device{
{
// /dev/fuse is created but not allowed.
// This is to allow java to work. Because java
// Insists on there being a /dev/fuse
// https://github.com/docker/docker/issues/514
// https://github.com/docker/docker/issues/2393
//
Path: "/dev/fuse",
Type: 'c',
Major: 10,
Minor: 229,
Permissions: "rwm",
},
}, DefaultSimpleDevices...)
)

View File

@@ -1,11 +1,5 @@
package configs
const (
// EXT_COPYUP is a directive to copy up the contents of a directory when
// a tmpfs is mounted over it.
EXT_COPYUP = 1 << iota
)
type Mount struct {
// Source path for the mount.
Source string `json:"source"`
@@ -28,9 +22,6 @@ type Mount struct {
// Relabel source if set, "z" indicates shared, "Z" indicates unshared.
Relabel string `json:"relabel"`
// Extensions are additional flags that are specific to runc.
Extensions int `json:"extensions"`
// Optional Command to be run before Source is mounted.
PremountCmds []Command `json:"premount_cmds"`

View File

@@ -2,19 +2,19 @@
package configs
import "golang.org/x/sys/unix"
import "syscall"
func (n *Namespace) Syscall() int {
return namespaceInfo[n.Type]
}
var namespaceInfo = map[NamespaceType]int{
NEWNET: unix.CLONE_NEWNET,
NEWNS: unix.CLONE_NEWNS,
NEWUSER: unix.CLONE_NEWUSER,
NEWIPC: unix.CLONE_NEWIPC,
NEWUTS: unix.CLONE_NEWUTS,
NEWPID: unix.CLONE_NEWPID,
NEWNET: syscall.CLONE_NEWNET,
NEWNS: syscall.CLONE_NEWNS,
NEWUSER: syscall.CLONE_NEWUSER,
NEWIPC: syscall.CLONE_NEWIPC,
NEWUTS: syscall.CLONE_NEWUTS,
NEWPID: syscall.CLONE_NEWPID,
}
// CloneFlags parses the container's Namespaces options to set the correct

View File

@@ -4,10 +4,12 @@ package configs
func (n *Namespace) Syscall() int {
panic("No namespace syscall support")
return 0
}
// CloneFlags parses the container's Namespaces options to set the correct
// flags on clone, unshare. This function returns flags only for new namespaces.
func (n *Namespaces) CloneFlags() uintptr {
panic("No namespace syscall support")
return uintptr(0)
}

View File

@@ -1,3 +1,5 @@
// +build linux freebsd
package configs
import (
@@ -20,8 +22,8 @@ var (
supportedNamespaces = make(map[NamespaceType]bool)
)
// NsName converts the namespace type to its filename
func NsName(ns NamespaceType) string {
// nsToFile converts the namespace type to its filename
func nsToFile(ns NamespaceType) string {
switch ns {
case NEWNET:
return "net"
@@ -48,7 +50,7 @@ func IsNamespaceSupported(ns NamespaceType) bool {
if ok {
return supported
}
nsFile := NsName(ns)
nsFile := nsToFile(ns)
// if the namespace type is unknown, just return false
if nsFile == "" {
return false
@@ -62,12 +64,12 @@ func IsNamespaceSupported(ns NamespaceType) bool {
func NamespaceTypes() []NamespaceType {
return []NamespaceType{
NEWUSER, // Keep user NS always first, don't move it.
NEWIPC,
NEWUTS,
NEWNET,
NEWPID,
NEWNS,
NEWUTS,
NEWIPC,
NEWUSER,
}
}
@@ -79,7 +81,10 @@ type Namespace struct {
}
func (n *Namespace) GetPath(pid int) string {
return fmt.Sprintf("/proc/%d/ns/%s", pid, NsName(n.Type))
if n.Path != "" {
return n.Path
}
return fmt.Sprintf("/proc/%d/ns/%s", pid, nsToFile(n.Type))
}
func (n *Namespaces) Remove(t NamespaceType) bool {

View File

@@ -1,4 +1,4 @@
// +build !linux
// +build !linux,!freebsd
package configs

View File

@@ -1,117 +0,0 @@
package validate
import (
"fmt"
"os"
"reflect"
"strings"
"github.com/opencontainers/runc/libcontainer/configs"
)
var (
geteuid = os.Geteuid
getegid = os.Getegid
)
func (v *ConfigValidator) rootless(config *configs.Config) error {
if err := rootlessMappings(config); err != nil {
return err
}
if err := rootlessMount(config); err != nil {
return err
}
// Currently, cgroups cannot effectively be used in rootless containers.
// The new cgroup namespace doesn't really help us either because it doesn't
// have nice interactions with the user namespace (we're working with upstream
// to fix this).
if err := rootlessCgroup(config); err != nil {
return err
}
// XXX: We currently can't verify the user config at all, because
// configs.Config doesn't store the user-related configs. So this
// has to be verified by setupUser() in init_linux.go.
return nil
}
func rootlessMappings(config *configs.Config) error {
rootuid, err := config.HostRootUID()
if err != nil {
return fmt.Errorf("failed to get root uid from uidMappings: %v", err)
}
if euid := geteuid(); euid != 0 {
if !config.Namespaces.Contains(configs.NEWUSER) {
return fmt.Errorf("rootless containers require user namespaces")
}
if rootuid != euid {
return fmt.Errorf("rootless containers cannot map container root to a different host user")
}
}
rootgid, err := config.HostRootGID()
if err != nil {
return fmt.Errorf("failed to get root gid from gidMappings: %v", err)
}
// Similar to the above test, we need to make sure that we aren't trying to
// map to a group ID that we don't have the right to be.
if rootgid != getegid() {
return fmt.Errorf("rootless containers cannot map container root to a different host group")
}
// We can only map one user and group inside a container (our own).
if len(config.UidMappings) != 1 || config.UidMappings[0].Size != 1 {
return fmt.Errorf("rootless containers cannot map more than one user")
}
if len(config.GidMappings) != 1 || config.GidMappings[0].Size != 1 {
return fmt.Errorf("rootless containers cannot map more than one group")
}
return nil
}
// cgroup verifies that the user isn't trying to set any cgroup limits or paths.
func rootlessCgroup(config *configs.Config) error {
// Nothing set at all.
if config.Cgroups == nil || config.Cgroups.Resources == nil {
return nil
}
// Used for comparing to the zero value.
left := reflect.ValueOf(*config.Cgroups.Resources)
right := reflect.Zero(left.Type())
// This is all we need to do, since specconv won't add cgroup options in
// rootless mode.
if !reflect.DeepEqual(left.Interface(), right.Interface()) {
return fmt.Errorf("cannot specify resource limits in rootless container")
}
return nil
}
// mount verifies that the user isn't trying to set up any mounts they don't have
// the rights to do. In addition, it makes sure that no mount has a `uid=` or
// `gid=` option that doesn't resolve to root.
func rootlessMount(config *configs.Config) error {
// XXX: We could whitelist allowed devices at this point, but I'm not
// convinced that's a good idea. The kernel is the best arbiter of
// access control.
for _, mount := range config.Mounts {
// Check that the options list doesn't contain any uid= or gid= entries
// that don't resolve to root.
for _, opt := range strings.Split(mount.Data, ",") {
if strings.HasPrefix(opt, "uid=") && opt != "uid=0" {
return fmt.Errorf("cannot specify uid= mount options in rootless containers where argument isn't 0")
}
if strings.HasPrefix(opt, "gid=") && opt != "gid=0" {
return fmt.Errorf("cannot specify gid= mount options in rootless containers where argument isn't 0")
}
}
}
return nil
}

View File

@@ -1,195 +0,0 @@
package validate
import (
"testing"
"github.com/opencontainers/runc/libcontainer/configs"
)
func init() {
geteuid = func() int { return 1337 }
getegid = func() int { return 7331 }
}
func rootlessConfig() *configs.Config {
return &configs.Config{
Rootfs: "/var",
Rootless: true,
Namespaces: configs.Namespaces(
[]configs.Namespace{
{Type: configs.NEWUSER},
},
),
UidMappings: []configs.IDMap{
{
HostID: geteuid(),
ContainerID: 0,
Size: 1,
},
},
GidMappings: []configs.IDMap{
{
HostID: getegid(),
ContainerID: 0,
Size: 1,
},
},
}
}
func TestValidateRootless(t *testing.T) {
validator := New()
config := rootlessConfig()
if err := validator.Validate(config); err != nil {
t.Errorf("Expected error to not occur: %+v", err)
}
}
/* rootlessMappings() */
func TestValidateRootlessUserns(t *testing.T) {
validator := New()
config := rootlessConfig()
config.Namespaces = nil
if err := validator.Validate(config); err == nil {
t.Errorf("Expected error to occur if user namespaces not set")
}
}
func TestValidateRootlessMappingUid(t *testing.T) {
validator := New()
config := rootlessConfig()
config.UidMappings = nil
if err := validator.Validate(config); err == nil {
t.Errorf("Expected error to occur if no uid mappings provided")
}
config = rootlessConfig()
config.UidMappings[0].HostID = geteuid() + 1
if err := validator.Validate(config); err == nil {
t.Errorf("Expected error to occur if geteuid() != mapped uid")
}
config = rootlessConfig()
config.UidMappings[0].Size = 1024
if err := validator.Validate(config); err == nil {
t.Errorf("Expected error to occur if more than one uid mapped")
}
config = rootlessConfig()
config.UidMappings = append(config.UidMappings, configs.IDMap{
HostID: geteuid() + 1,
ContainerID: 0,
Size: 1,
})
if err := validator.Validate(config); err == nil {
t.Errorf("Expected error to occur if more than one uid extent mapped")
}
}
func TestValidateRootlessMappingGid(t *testing.T) {
validator := New()
config := rootlessConfig()
config.GidMappings = nil
if err := validator.Validate(config); err == nil {
t.Errorf("Expected error to occur if no gid mappings provided")
}
config = rootlessConfig()
config.GidMappings[0].HostID = getegid() + 1
if err := validator.Validate(config); err == nil {
t.Errorf("Expected error to occur if getegid() != mapped gid")
}
config = rootlessConfig()
config.GidMappings[0].Size = 1024
if err := validator.Validate(config); err == nil {
t.Errorf("Expected error to occur if more than one gid mapped")
}
config = rootlessConfig()
config.GidMappings = append(config.GidMappings, configs.IDMap{
HostID: getegid() + 1,
ContainerID: 0,
Size: 1,
})
if err := validator.Validate(config); err == nil {
t.Errorf("Expected error to occur if more than one gid extent mapped")
}
}
/* rootlessMount() */
func TestValidateRootlessMountUid(t *testing.T) {
config := rootlessConfig()
validator := New()
config.Mounts = []*configs.Mount{
{
Source: "devpts",
Destination: "/dev/pts",
Device: "devpts",
},
}
if err := validator.Validate(config); err != nil {
t.Errorf("Expected error to not occur when uid= not set in mount options: %+v", err)
}
config.Mounts[0].Data = "uid=5"
if err := validator.Validate(config); err == nil {
t.Errorf("Expected error to occur when setting uid=5 in mount options")
}
config.Mounts[0].Data = "uid=0"
if err := validator.Validate(config); err != nil {
t.Errorf("Expected error to not occur when setting uid=0 in mount options: %+v", err)
}
}
func TestValidateRootlessMountGid(t *testing.T) {
config := rootlessConfig()
validator := New()
config.Mounts = []*configs.Mount{
{
Source: "devpts",
Destination: "/dev/pts",
Device: "devpts",
},
}
if err := validator.Validate(config); err != nil {
t.Errorf("Expected error to not occur when gid= not set in mount options: %+v", err)
}
config.Mounts[0].Data = "gid=5"
if err := validator.Validate(config); err == nil {
t.Errorf("Expected error to occur when setting gid=5 in mount options")
}
config.Mounts[0].Data = "gid=0"
if err := validator.Validate(config); err != nil {
t.Errorf("Expected error to not occur when setting gid=0 in mount options: %+v", err)
}
}
/* rootlessCgroup() */
func TestValidateRootlessCgroup(t *testing.T) {
validator := New()
config := rootlessConfig()
config.Cgroups = &configs.Cgroup{
Resources: &configs.Resources{
PidsLimit: 1337,
},
}
if err := validator.Validate(config); err == nil {
t.Errorf("Expected error to occur if cgroup limits set")
}
}

View File

@@ -7,7 +7,6 @@ import (
"strings"
"github.com/opencontainers/runc/libcontainer/configs"
selinux "github.com/opencontainers/selinux/go-selinux"
)
type Validator interface {
@@ -40,23 +39,12 @@ func (v *ConfigValidator) Validate(config *configs.Config) error {
if err := v.sysctl(config); err != nil {
return err
}
if config.Rootless {
if err := v.rootless(config); err != nil {
return err
}
}
return nil
}
// rootfs validates if the rootfs is an absolute path and is not a symlink
// to the container's root filesystem.
func (v *ConfigValidator) rootfs(config *configs.Config) error {
if _, err := os.Stat(config.Rootfs); err != nil {
if os.IsNotExist(err) {
return fmt.Errorf("rootfs (%s) does not exist", config.Rootfs)
}
return err
}
cleaned, err := filepath.Abs(config.Rootfs)
if err != nil {
return err
@@ -92,10 +80,6 @@ func (v *ConfigValidator) security(config *configs.Config) error {
!config.Namespaces.Contains(configs.NEWNS) {
return fmt.Errorf("unable to restrict sys entries without a private MNT namespace")
}
if config.ProcessLabel != "" && !selinux.GetEnabled() {
return fmt.Errorf("selinux label is specified in config, but selinux is disabled or not supported")
}
return nil
}
@@ -137,11 +121,6 @@ func (v *ConfigValidator) sysctl(config *configs.Config) error {
}
if strings.HasPrefix(s, "net.") {
if config.Namespaces.Contains(configs.NEWNET) {
if path := config.Namespaces.PathOf(configs.NEWNET); path != "" {
if err := checkHostNs(s, path); err != nil {
return err
}
}
continue
} else {
return fmt.Errorf("sysctl %q is not allowed in the hosts network namespace", s)
@@ -152,44 +131,3 @@ func (v *ConfigValidator) sysctl(config *configs.Config) error {
return nil
}
func isSymbolicLink(path string) (bool, error) {
fi, err := os.Lstat(path)
if err != nil {
return false, err
}
return fi.Mode()&os.ModeSymlink == os.ModeSymlink, nil
}
// checkHostNs checks whether network sysctl is used in host namespace.
func checkHostNs(sysctlConfig string, path string) error {
var currentProcessNetns = "/proc/self/ns/net"
// readlink on the current processes network namespace
destOfCurrentProcess, err := os.Readlink(currentProcessNetns)
if err != nil {
return fmt.Errorf("read soft link %q error", currentProcessNetns)
}
// First check if the provided path is a symbolic link
symLink, err := isSymbolicLink(path)
if err != nil {
return fmt.Errorf("could not check that %q is a symlink: %v", path, err)
}
if symLink == false {
// The provided namespace is not a symbolic link,
// it is not the host namespace.
return nil
}
// readlink on the path provided in the struct
destOfContainer, err := os.Readlink(path)
if err != nil {
return fmt.Errorf("read soft link %q error", path)
}
if destOfContainer == destOfCurrentProcess {
return fmt.Errorf("sysctl %q is not allowed in the hosts network namespace", sysctlConfig)
}
return nil
}

View File

@@ -100,7 +100,7 @@ func TestValidateHostnameWithoutUTSNamespace(t *testing.T) {
func TestValidateSecurityWithMaskPaths(t *testing.T) {
config := &configs.Config{
Rootfs: "/var",
MaskPaths: []string{"/proc/kcore"},
MaskPaths: []string{"/proc/kcores"},
Namespaces: configs.Namespaces(
[]configs.Namespace{
{Type: configs.NEWNS},
@@ -136,7 +136,7 @@ func TestValidateSecurityWithROPaths(t *testing.T) {
func TestValidateSecurityWithoutNEWNS(t *testing.T) {
config := &configs.Config{
Rootfs: "/var",
MaskPaths: []string{"/proc/kcore"},
MaskPaths: []string{"/proc/kcores"},
ReadonlyPaths: []string{"/proc/sys"},
}
@@ -148,9 +148,6 @@ func TestValidateSecurityWithoutNEWNS(t *testing.T) {
}
func TestValidateUsernamespace(t *testing.T) {
if _, err := os.Stat("/proc/self/ns/user"); os.IsNotExist(err) {
t.Skip("userns is unsupported")
}
config := &configs.Config{
Rootfs: "/var",
Namespaces: configs.Namespaces(
@@ -201,67 +198,3 @@ func TestValidateSysctl(t *testing.T) {
}
}
}
func TestValidateValidSysctl(t *testing.T) {
sysctl := map[string]string{
"fs.mqueue.ctl": "ctl",
"net.ctl": "ctl",
"kernel.msgmax": "ctl",
}
for k, v := range sysctl {
config := &configs.Config{
Rootfs: "/var",
Sysctl: map[string]string{k: v},
Namespaces: []configs.Namespace{
{
Type: configs.NEWNET,
},
{
Type: configs.NEWIPC,
},
},
}
validator := validate.New()
err := validator.Validate(config)
if err != nil {
t.Errorf("Expected error to not occur with {%s=%s} but got: %q", k, v, err)
}
}
}
func TestValidateSysctlWithSameNs(t *testing.T) {
config := &configs.Config{
Rootfs: "/var",
Sysctl: map[string]string{"net.ctl": "ctl"},
Namespaces: configs.Namespaces(
[]configs.Namespace{
{
Type: configs.NEWNET,
Path: "/proc/self/ns/net",
},
},
),
}
validator := validate.New()
err := validator.Validate(config)
if err == nil {
t.Error("Expected error to occur but it was nil")
}
}
func TestValidateSysctlWithoutNETNamespace(t *testing.T) {
config := &configs.Config{
Rootfs: "/var",
Sysctl: map[string]string{"net.ctl": "ctl"},
Namespaces: []configs.Namespace{},
}
validator := validate.New()
err := validator.Validate(config)
if err == nil {
t.Error("Expected error to occur but it was nil")
}
}

View File

@@ -1,17 +1,15 @@
package libcontainer
import (
"io"
"os"
)
import "io"
// Console represents a pseudo TTY.
type Console interface {
io.ReadWriteCloser
io.ReadWriter
io.Closer
// Path returns the filesystem path to the slave side of the pty.
Path() string
// Fd returns the fd for the master of the pty.
File() *os.File
Fd() uintptr
}

View File

@@ -6,8 +6,8 @@ import (
"errors"
)
// newConsole returns an initialized console that can be used within a container by copying bytes
// NewConsole returns an initalized console that can be used within a container by copying bytes
// from the master side to the slave that is attached as the tty for the container's init process.
func newConsole() (Console, error) {
func NewConsole(uid, gid int) (Console, error) {
return nil, errors.New("libcontainer console is not supported on FreeBSD")
}

View File

@@ -3,21 +3,17 @@ package libcontainer
import (
"fmt"
"os"
"path/filepath"
"syscall"
"unsafe"
"golang.org/x/sys/unix"
"github.com/opencontainers/runc/libcontainer/label"
)
func ConsoleFromFile(f *os.File) Console {
return &linuxConsole{
master: f,
}
}
// newConsole returns an initialized console that can be used within a container by copying bytes
// NewConsole returns an initalized console that can be used within a container by copying bytes
// from the master side to the slave that is attached as the tty for the container's init process.
func newConsole() (Console, error) {
master, err := os.OpenFile("/dev/ptmx", unix.O_RDWR|unix.O_NOCTTY|unix.O_CLOEXEC, 0)
func NewConsole(uid, gid int) (Console, error) {
master, err := os.OpenFile("/dev/ptmx", syscall.O_RDWR|syscall.O_NOCTTY|syscall.O_CLOEXEC, 0)
if err != nil {
return nil, err
}
@@ -28,20 +24,34 @@ func newConsole() (Console, error) {
if err := unlockpt(master); err != nil {
return nil, err
}
if err := os.Chmod(console, 0600); err != nil {
return nil, err
}
if err := os.Chown(console, uid, gid); err != nil {
return nil, err
}
return &linuxConsole{
slavePath: console,
master: master,
}, nil
}
// linuxConsole is a linux pseudo TTY for use within a container.
// newConsoleFromPath is an internal function returning an initialized console for use inside
// a container's MNT namespace.
func newConsoleFromPath(slavePath string) *linuxConsole {
return &linuxConsole{
slavePath: slavePath,
}
}
// linuxConsole is a linux psuedo TTY for use within a container.
type linuxConsole struct {
master *os.File
slavePath string
}
func (c *linuxConsole) File() *os.File {
return c.master
func (c *linuxConsole) Fd() uintptr {
return c.master.Fd()
}
func (c *linuxConsole) Path() string {
@@ -65,29 +75,33 @@ func (c *linuxConsole) Close() error {
// mount initializes the console inside the rootfs mounting with the specified mount label
// and applying the correct ownership of the console.
func (c *linuxConsole) mount() error {
oldMask := unix.Umask(0000)
defer unix.Umask(oldMask)
f, err := os.Create("/dev/console")
func (c *linuxConsole) mount(rootfs, mountLabel string) error {
oldMask := syscall.Umask(0000)
defer syscall.Umask(oldMask)
if err := label.SetFileLabel(c.slavePath, mountLabel); err != nil {
return err
}
dest := filepath.Join(rootfs, "/dev/console")
f, err := os.Create(dest)
if err != nil && !os.IsExist(err) {
return err
}
if f != nil {
f.Close()
}
return unix.Mount(c.slavePath, "/dev/console", "bind", unix.MS_BIND, "")
return syscall.Mount(c.slavePath, dest, "bind", syscall.MS_BIND, "")
}
// dupStdio opens the slavePath for the console and dups the fds to the current
// processes stdio, fd 0,1,2.
func (c *linuxConsole) dupStdio() error {
slave, err := c.open(unix.O_RDWR)
slave, err := c.open(syscall.O_RDWR)
if err != nil {
return err
}
fd := int(slave.Fd())
for _, i := range []int{0, 1, 2} {
if err := unix.Dup3(fd, i, 0); err != nil {
if err := syscall.Dup3(fd, i, 0); err != nil {
return err
}
}
@@ -96,7 +110,7 @@ func (c *linuxConsole) dupStdio() error {
// open is a clone of os.OpenFile without the O_CLOEXEC used to open the pty slave.
func (c *linuxConsole) open(flag int) (*os.File, error) {
r, e := unix.Open(c.slavePath, flag, 0)
r, e := syscall.Open(c.slavePath, flag, 0)
if e != nil {
return nil, &os.PathError{
Op: "open",
@@ -108,7 +122,7 @@ func (c *linuxConsole) open(flag int) (*os.File, error) {
}
func ioctl(fd uintptr, flag, data uintptr) error {
if _, _, err := unix.Syscall(unix.SYS_IOCTL, fd, flag, data); err != 0 {
if _, _, err := syscall.Syscall(syscall.SYS_IOCTL, fd, flag, data); err != 0 {
return err
}
return nil
@@ -118,35 +132,14 @@ func ioctl(fd uintptr, flag, data uintptr) error {
// unlockpt should be called before opening the slave side of a pty.
func unlockpt(f *os.File) error {
var u int32
return ioctl(f.Fd(), unix.TIOCSPTLCK, uintptr(unsafe.Pointer(&u)))
return ioctl(f.Fd(), syscall.TIOCSPTLCK, uintptr(unsafe.Pointer(&u)))
}
// ptsname retrieves the name of the first available pts for the given master.
func ptsname(f *os.File) (string, error) {
n, err := unix.IoctlGetInt(int(f.Fd()), unix.TIOCGPTN)
if err != nil {
var n int32
if err := ioctl(f.Fd(), syscall.TIOCGPTN, uintptr(unsafe.Pointer(&n))); err != nil {
return "", err
}
return fmt.Sprintf("/dev/pts/%d", n), nil
}
// SaneTerminal sets the necessary tty_ioctl(4)s to ensure that a pty pair
// created by us acts normally. In particular, a not-very-well-known default of
// Linux unix98 ptys is that they have +onlcr by default. While this isn't a
// problem for terminal emulators, because we relay data from the terminal we
// also relay that funky line discipline.
func SaneTerminal(terminal *os.File) error {
termios, err := unix.IoctlGetTermios(int(terminal.Fd()), unix.TCGETS)
if err != nil {
return fmt.Errorf("ioctl(tty, tcgets): %s", err.Error())
}
// Set -onlcr so we don't have to deal with \r.
termios.Oflag &^= unix.ONLCR
if err := unix.IoctlSetTermios(int(terminal.Fd()), unix.TCSETS, termios); err != nil {
return fmt.Errorf("ioctl(tty, tcsets): %s", err.Error())
}
return nil
}

View File

@@ -1,11 +0,0 @@
package libcontainer
import (
"errors"
)
// newConsole returns an initialized console that can be used within a container by copying bytes
// from the master side to the slave that is attached as the tty for the container's init process.
func newConsole() (Console, error) {
return nil, errors.New("libcontainer console is not supported on Solaris")
}

View File

@@ -1,11 +1,11 @@
package libcontainer
// newConsole returns an initialized console that can be used within a container
func newConsole() (Console, error) {
// NewConsole returns an initalized console that can be used within a container
func NewConsole(uid, gid int) (Console, error) {
return &windowsConsole{}, nil
}
// windowsConsole is a Windows pseudo TTY for use within a container.
// windowsConsole is a Windows psuedo TTY for use within a container.
type windowsConsole struct {
}

View File

@@ -1,4 +1,4 @@
// Package libcontainer provides a native Go implementation for creating containers
// Libcontainer provides a native Go implementation for creating containers
// with namespaces, cgroups, capabilities, and filesystem access controls.
// It allows you to manage the lifecycle of the container performing additional operations
// after the container is created.
@@ -11,20 +11,24 @@ import (
"github.com/opencontainers/runc/libcontainer/configs"
)
// Status is the status of a container.
// The status of a container.
type Status int
const (
// Created is the status that denotes the container exists but has not been run yet.
// The container exists but has not been run yet
Created Status = iota
// Running is the status that denotes the container exists and is running.
// The container exists and is running.
Running
// Pausing is the status that denotes the container exists, it is in the process of being paused.
// The container exists, it is in the process of being paused.
Pausing
// Paused is the status that denotes the container exists, but all its processes are paused.
// The container exists, but all its processes are paused.
Paused
// Stopped is the status that denotes the container does not have a created or running process.
Stopped
// The container does not exist.
Destroyed
)
func (s Status) String() string {
@@ -37,8 +41,8 @@ func (s Status) String() string {
return "pausing"
case Paused:
return "paused"
case Stopped:
return "stopped"
case Destroyed:
return "destroyed"
default:
return "unknown"
}
@@ -54,7 +58,7 @@ type BaseState struct {
InitProcessPid int `json:"init_process_pid"`
// InitProcessStartTime is the init process start time in clock cycles since boot time.
InitProcessStartTime uint64 `json:"init_process_start"`
InitProcessStartTime string `json:"init_process_start"`
// Created is the unix timestamp for the creation time of the container in UTC
Created time.Time `json:"created"`
@@ -63,7 +67,7 @@ type BaseState struct {
Config configs.Config `json:"config"`
}
// BaseContainer is a libcontainer container object.
// A libcontainer container object.
//
// Each container is thread-safe within the same process. Since a container can
// be destroyed by a separate process, any function may return that the container
@@ -75,14 +79,14 @@ type BaseContainer interface {
// Returns the current status of the container.
//
// errors:
// ContainerNotExists - Container no longer exists,
// ContainerDestroyed - Container no longer exists,
// Systemerror - System error.
Status() (Status, error)
// State returns the current container's state information.
//
// errors:
// SystemError - System error.
// Systemerror - System error.
State() (*State, error)
// Returns the current config of the container.
@@ -91,7 +95,7 @@ type BaseContainer interface {
// Returns the PIDs inside this container. The PIDs are in the namespace of the calling process.
//
// errors:
// ContainerNotExists - Container no longer exists,
// ContainerDestroyed - Container no longer exists,
// Systemerror - System error.
//
// Some of the returned PIDs may no longer refer to processes in the Container, unless
@@ -101,7 +105,7 @@ type BaseContainer interface {
// Returns statistics for the container.
//
// errors:
// ContainerNotExists - Container no longer exists,
// ContainerDestroyed - Container no longer exists,
// Systemerror - System error.
Stats() (*Stats, error)
@@ -110,57 +114,31 @@ type BaseContainer interface {
// We can use this to change resources when containers are running.
//
// errors:
// SystemError - System error.
// Systemerror - System error.
Set(config configs.Config) error
// Start a process inside the container. Returns error if process fails to
// start. You can track process lifecycle with passed Process structure.
//
// errors:
// ContainerNotExists - Container no longer exists,
// ContainerDestroyed - Container no longer exists,
// ConfigInvalid - config is invalid,
// ContainerPaused - Container is paused,
// SystemError - System error.
// Systemerror - System error.
Start(process *Process) (err error)
// Run immediately starts the process inside the container. Returns error if process
// fails to start. It does not block waiting for the exec fifo after start returns but
// opens the fifo after start returns.
//
// errors:
// ContainerNotExists - Container no longer exists,
// ConfigInvalid - config is invalid,
// ContainerPaused - Container is paused,
// SystemError - System error.
Run(process *Process) (err error)
// Destroys the container, if its in a valid state, after killing any
// remaining running processes.
// Destroys the container after killing all running processes.
//
// Any event registrations are removed before the container is destroyed.
// No error is returned if the container is already destroyed.
//
// Running containers must first be stopped using Signal(..).
// Paused containers must first be resumed using Resume(..).
//
// errors:
// ContainerNotStopped - Container is still running,
// ContainerPaused - Container is paused,
// SystemError - System error.
// Systemerror - System error.
Destroy() error
// Signal sends the provided signal code to the container's initial process.
//
// If all is specified the signal is sent to all processes in the container
// including the initial process.
//
// errors:
// SystemError - System error.
Signal(s os.Signal, all bool) error
// Exec signals the container to exec the users process at the end of the init.
//
// errors:
// SystemError - System error.
Exec() error
// Systemerror - System error.
Signal(s os.Signal) error
}

File diff suppressed because it is too large Load Diff

View File

@@ -9,7 +9,6 @@ import (
"github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/opencontainers/runc/libcontainer/configs"
"github.com/opencontainers/runc/libcontainer/system"
)
type mockCgroupManager struct {
@@ -53,7 +52,7 @@ func (m *mockCgroupManager) Freeze(state configs.FreezerState) error {
type mockProcess struct {
_pid int
started uint64
started string
}
func (m *mockProcess) terminate() error {
@@ -64,7 +63,7 @@ func (m *mockProcess) pid() int {
return m._pid
}
func (m *mockProcess) startTime() (uint64, error) {
func (m *mockProcess) startTime() (string, error) {
return m.started, nil
}
@@ -80,11 +79,11 @@ func (m *mockProcess) signal(_ os.Signal) error {
return nil
}
func (m *mockProcess) externalDescriptors() []string {
func (p *mockProcess) externalDescriptors() []string {
return []string{}
}
func (m *mockProcess) setExternalDescriptors(newFds []string) {
func (p *mockProcess) setExternalDescriptors(newFds []string) {
}
func TestGetContainerPids(t *testing.T) {
@@ -135,7 +134,7 @@ func TestGetContainerState(t *testing.T) {
var (
pid = os.Getpid()
expectedMemoryPath = "/sys/fs/cgroup/memory/myid"
expectedNetworkPath = fmt.Sprintf("/proc/%d/ns/net", pid)
expectedNetworkPath = "/networks/fd"
)
container := &linuxContainer{
id: "myid",
@@ -151,7 +150,7 @@ func TestGetContainerState(t *testing.T) {
},
initProcess: &mockProcess{
_pid: pid,
started: 10,
started: "010",
},
cgroupManager: &mockCgroupManager{
pids: []int{1, 2, 3},
@@ -175,8 +174,8 @@ func TestGetContainerState(t *testing.T) {
if state.InitProcessPid != pid {
t.Fatalf("expected pid %d but received %d", pid, state.InitProcessPid)
}
if state.InitProcessStartTime != 10 {
t.Fatalf("expected process start time 10 but received %d", state.InitProcessStartTime)
if state.InitProcessStartTime != "010" {
t.Fatalf("expected process start time 010 but received %s", state.InitProcessStartTime)
}
paths := state.CgroupPaths
if paths == nil {
@@ -217,65 +216,3 @@ func TestGetContainerState(t *testing.T) {
}
}
}
func TestGetContainerStateAfterUpdate(t *testing.T) {
var (
pid = os.Getpid()
)
stat, err := system.Stat(pid)
if err != nil {
t.Fatal(err)
}
container := &linuxContainer{
id: "myid",
config: &configs.Config{
Namespaces: []configs.Namespace{
{Type: configs.NEWPID},
{Type: configs.NEWNS},
{Type: configs.NEWNET},
{Type: configs.NEWUTS},
{Type: configs.NEWIPC},
},
Cgroups: &configs.Cgroup{
Resources: &configs.Resources{
Memory: 1024,
},
},
},
initProcess: &mockProcess{
_pid: pid,
started: stat.StartTime,
},
cgroupManager: &mockCgroupManager{},
}
container.state = &createdState{c: container}
state, err := container.State()
if err != nil {
t.Fatal(err)
}
if state.InitProcessPid != pid {
t.Fatalf("expected pid %d but received %d", pid, state.InitProcessPid)
}
if state.InitProcessStartTime != stat.StartTime {
t.Fatalf("expected process start time %d but received %d", stat.StartTime, state.InitProcessStartTime)
}
if state.Config.Cgroups.Resources.Memory != 1024 {
t.Fatalf("expected Memory to be 1024 but received %q", state.Config.Cgroups.Memory)
}
// Set initProcessStartTime so we fake to be running
container.initProcessStartTime = state.InitProcessStartTime
container.state = &runningState{c: container}
newConfig := container.Config()
newConfig.Cgroups.Resources.Memory = 2048
if err := container.Set(newConfig); err != nil {
t.Fatal(err)
}
state, err = container.State()
if err != nil {
t.Fatal(err)
}
if state.Config.Cgroups.Resources.Memory != 2048 {
t.Fatalf("expected Memory to be 2048 but received %q", state.Config.Cgroups.Memory)
}
}

View File

@@ -1,20 +0,0 @@
package libcontainer
// State represents a running container's state
type State struct {
BaseState
// Platform specific fields below here
}
// A libcontainer container object.
//
// Each container is thread-safe within the same process. Since a container can
// be destroyed by a separate process, any function may return that the container
// was not found.
type Container interface {
BaseContainer
// Methods below here are platform specific
}

View File

@@ -1,13 +1,15 @@
// +build linux freebsd
package libcontainer
// cgroup restoring strategy provided by criu
type cgMode uint32
type cg_mode uint32
const (
CRIU_CG_MODE_SOFT cgMode = 3 + iota // restore cgroup properties if only dir created by criu
CRIU_CG_MODE_FULL // always restore all cgroups and their properties
CRIU_CG_MODE_STRICT // restore all, requiring them to not present in the system
CRIU_CG_MODE_DEFAULT // the same as CRIU_CG_MODE_SOFT
CRIU_CG_MODE_SOFT cg_mode = 3 + iota // restore cgroup properties if only dir created by criu
CRIU_CG_MODE_FULL // always restore all cgroups and their properties
CRIU_CG_MODE_STRICT // restore all, requiring them to not present in the system
CRIU_CG_MODE_DEFAULT // the same as CRIU_CG_MODE_SOFT
)
type CriuPageServerInfo struct {
@@ -23,16 +25,13 @@ type VethPairName struct {
type CriuOpts struct {
ImagesDirectory string // directory for storing image files
WorkDirectory string // directory to cd and write logs/pidfiles/stats to
ParentImage string // direcotry for storing parent image files in pre-dump and dump
LeaveRunning bool // leave container in running state after checkpoint
TcpEstablished bool // checkpoint/restore established TCP connections
ExternalUnixConnections bool // allow external unix connections
ShellJob bool // allow to dump and restore shell jobs
FileLocks bool // handle file locks, for safety
PreDump bool // call criu predump to perform iterative checkpoint
PageServer CriuPageServerInfo // allow to dump to criu page server
VethPairs []VethPairName // pass the veth to criu when restore
ManageCgroupsMode cgMode // dump or restore cgroup mode
ManageCgroupsMode cg_mode // dump or restore cgroup mode
EmptyNs uint32 // don't c/r properties for namespace from this mask
AutoDedup bool // auto deduplication for incremental dumps
}

View File

@@ -12,7 +12,6 @@ It has these top-level messages:
CriuPageServerInfo
CriuVethPair
ExtMountMap
JoinNamespace
InheritFd
CgroupRoot
UnixSk
@@ -23,30 +22,21 @@ It has these top-level messages:
CriuFeatures
CriuReq
CriuResp
CriuVersion
*/
package criurpc
import proto "github.com/golang/protobuf/proto"
import fmt "fmt"
import math "math"
// Reference imports to suppress errors if they are not otherwise used.
var _ = proto.Marshal
var _ = fmt.Errorf
var _ = math.Inf
// This is a compile-time assertion to ensure that this generated file
// is compatible with the proto package it is being compiled against.
// A compilation error at this line likely means your copy of the
// proto package needs to be updated.
const _ = proto.ProtoPackageIsVersion2 // please upgrade the proto package
type CriuCgMode int32
const (
CriuCgMode_IGNORE CriuCgMode = 0
CriuCgMode_CG_NONE CriuCgMode = 1
CriuCgMode_NONE CriuCgMode = 1
CriuCgMode_PROPS CriuCgMode = 2
CriuCgMode_SOFT CriuCgMode = 3
CriuCgMode_FULL CriuCgMode = 4
@@ -56,7 +46,7 @@ const (
var CriuCgMode_name = map[int32]string{
0: "IGNORE",
1: "CG_NONE",
1: "NONE",
2: "PROPS",
3: "SOFT",
4: "FULL",
@@ -65,7 +55,7 @@ var CriuCgMode_name = map[int32]string{
}
var CriuCgMode_value = map[string]int32{
"IGNORE": 0,
"CG_NONE": 1,
"NONE": 1,
"PROPS": 2,
"SOFT": 3,
"FULL": 4,
@@ -89,7 +79,6 @@ func (x *CriuCgMode) UnmarshalJSON(data []byte) error {
*x = CriuCgMode(value)
return nil
}
func (CriuCgMode) EnumDescriptor() ([]byte, []int) { return fileDescriptor0, []int{0} }
type CriuReqType int32
@@ -104,21 +93,19 @@ const (
CriuReqType_CPUINFO_DUMP CriuReqType = 7
CriuReqType_CPUINFO_CHECK CriuReqType = 8
CriuReqType_FEATURE_CHECK CriuReqType = 9
CriuReqType_VERSION CriuReqType = 10
)
var CriuReqType_name = map[int32]string{
0: "EMPTY",
1: "DUMP",
2: "RESTORE",
3: "CHECK",
4: "PRE_DUMP",
5: "PAGE_SERVER",
6: "NOTIFY",
7: "CPUINFO_DUMP",
8: "CPUINFO_CHECK",
9: "FEATURE_CHECK",
10: "VERSION",
0: "EMPTY",
1: "DUMP",
2: "RESTORE",
3: "CHECK",
4: "PRE_DUMP",
5: "PAGE_SERVER",
6: "NOTIFY",
7: "CPUINFO_DUMP",
8: "CPUINFO_CHECK",
9: "FEATURE_CHECK",
}
var CriuReqType_value = map[string]int32{
"EMPTY": 0,
@@ -131,7 +118,6 @@ var CriuReqType_value = map[string]int32{
"CPUINFO_DUMP": 7,
"CPUINFO_CHECK": 8,
"FEATURE_CHECK": 9,
"VERSION": 10,
}
func (x CriuReqType) Enum() *CriuReqType {
@@ -150,7 +136,6 @@ func (x *CriuReqType) UnmarshalJSON(data []byte) error {
*x = CriuReqType(value)
return nil
}
func (CriuReqType) EnumDescriptor() ([]byte, []int) { return fileDescriptor0, []int{1} }
type CriuPageServerInfo struct {
Address *string `protobuf:"bytes,1,opt,name=address" json:"address,omitempty"`
@@ -160,10 +145,9 @@ type CriuPageServerInfo struct {
XXX_unrecognized []byte `json:"-"`
}
func (m *CriuPageServerInfo) Reset() { *m = CriuPageServerInfo{} }
func (m *CriuPageServerInfo) String() string { return proto.CompactTextString(m) }
func (*CriuPageServerInfo) ProtoMessage() {}
func (*CriuPageServerInfo) Descriptor() ([]byte, []int) { return fileDescriptor0, []int{0} }
func (m *CriuPageServerInfo) Reset() { *m = CriuPageServerInfo{} }
func (m *CriuPageServerInfo) String() string { return proto.CompactTextString(m) }
func (*CriuPageServerInfo) ProtoMessage() {}
func (m *CriuPageServerInfo) GetAddress() string {
if m != nil && m.Address != nil {
@@ -194,15 +178,14 @@ func (m *CriuPageServerInfo) GetFd() int32 {
}
type CriuVethPair struct {
IfIn *string `protobuf:"bytes,1,req,name=if_in,json=ifIn" json:"if_in,omitempty"`
IfOut *string `protobuf:"bytes,2,req,name=if_out,json=ifOut" json:"if_out,omitempty"`
IfIn *string `protobuf:"bytes,1,req,name=if_in" json:"if_in,omitempty"`
IfOut *string `protobuf:"bytes,2,req,name=if_out" json:"if_out,omitempty"`
XXX_unrecognized []byte `json:"-"`
}
func (m *CriuVethPair) Reset() { *m = CriuVethPair{} }
func (m *CriuVethPair) String() string { return proto.CompactTextString(m) }
func (*CriuVethPair) ProtoMessage() {}
func (*CriuVethPair) Descriptor() ([]byte, []int) { return fileDescriptor0, []int{1} }
func (m *CriuVethPair) Reset() { *m = CriuVethPair{} }
func (m *CriuVethPair) String() string { return proto.CompactTextString(m) }
func (*CriuVethPair) ProtoMessage() {}
func (m *CriuVethPair) GetIfIn() string {
if m != nil && m.IfIn != nil {
@@ -224,10 +207,9 @@ type ExtMountMap struct {
XXX_unrecognized []byte `json:"-"`
}
func (m *ExtMountMap) Reset() { *m = ExtMountMap{} }
func (m *ExtMountMap) String() string { return proto.CompactTextString(m) }
func (*ExtMountMap) ProtoMessage() {}
func (*ExtMountMap) Descriptor() ([]byte, []int) { return fileDescriptor0, []int{2} }
func (m *ExtMountMap) Reset() { *m = ExtMountMap{} }
func (m *ExtMountMap) String() string { return proto.CompactTextString(m) }
func (*ExtMountMap) ProtoMessage() {}
func (m *ExtMountMap) GetKey() string {
if m != nil && m.Key != nil {
@@ -243,49 +225,15 @@ func (m *ExtMountMap) GetVal() string {
return ""
}
type JoinNamespace struct {
Ns *string `protobuf:"bytes,1,req,name=ns" json:"ns,omitempty"`
NsFile *string `protobuf:"bytes,2,req,name=ns_file,json=nsFile" json:"ns_file,omitempty"`
ExtraOpt *string `protobuf:"bytes,3,opt,name=extra_opt,json=extraOpt" json:"extra_opt,omitempty"`
XXX_unrecognized []byte `json:"-"`
}
func (m *JoinNamespace) Reset() { *m = JoinNamespace{} }
func (m *JoinNamespace) String() string { return proto.CompactTextString(m) }
func (*JoinNamespace) ProtoMessage() {}
func (*JoinNamespace) Descriptor() ([]byte, []int) { return fileDescriptor0, []int{3} }
func (m *JoinNamespace) GetNs() string {
if m != nil && m.Ns != nil {
return *m.Ns
}
return ""
}
func (m *JoinNamespace) GetNsFile() string {
if m != nil && m.NsFile != nil {
return *m.NsFile
}
return ""
}
func (m *JoinNamespace) GetExtraOpt() string {
if m != nil && m.ExtraOpt != nil {
return *m.ExtraOpt
}
return ""
}
type InheritFd struct {
Key *string `protobuf:"bytes,1,req,name=key" json:"key,omitempty"`
Fd *int32 `protobuf:"varint,2,req,name=fd" json:"fd,omitempty"`
XXX_unrecognized []byte `json:"-"`
}
func (m *InheritFd) Reset() { *m = InheritFd{} }
func (m *InheritFd) String() string { return proto.CompactTextString(m) }
func (*InheritFd) ProtoMessage() {}
func (*InheritFd) Descriptor() ([]byte, []int) { return fileDescriptor0, []int{4} }
func (m *InheritFd) Reset() { *m = InheritFd{} }
func (m *InheritFd) String() string { return proto.CompactTextString(m) }
func (*InheritFd) ProtoMessage() {}
func (m *InheritFd) GetKey() string {
if m != nil && m.Key != nil {
@@ -307,10 +255,9 @@ type CgroupRoot struct {
XXX_unrecognized []byte `json:"-"`
}
func (m *CgroupRoot) Reset() { *m = CgroupRoot{} }
func (m *CgroupRoot) String() string { return proto.CompactTextString(m) }
func (*CgroupRoot) ProtoMessage() {}
func (*CgroupRoot) Descriptor() ([]byte, []int) { return fileDescriptor0, []int{5} }
func (m *CgroupRoot) Reset() { *m = CgroupRoot{} }
func (m *CgroupRoot) String() string { return proto.CompactTextString(m) }
func (*CgroupRoot) ProtoMessage() {}
func (m *CgroupRoot) GetCtrl() string {
if m != nil && m.Ctrl != nil {
@@ -331,10 +278,9 @@ type UnixSk struct {
XXX_unrecognized []byte `json:"-"`
}
func (m *UnixSk) Reset() { *m = UnixSk{} }
func (m *UnixSk) String() string { return proto.CompactTextString(m) }
func (*UnixSk) ProtoMessage() {}
func (*UnixSk) Descriptor() ([]byte, []int) { return fileDescriptor0, []int{6} }
func (m *UnixSk) Reset() { *m = UnixSk{} }
func (m *UnixSk) String() string { return proto.CompactTextString(m) }
func (*UnixSk) ProtoMessage() {}
func (m *UnixSk) GetInode() uint32 {
if m != nil && m.Inode != nil {
@@ -344,62 +290,51 @@ func (m *UnixSk) GetInode() uint32 {
}
type CriuOpts struct {
ImagesDirFd *int32 `protobuf:"varint,1,req,name=images_dir_fd,json=imagesDirFd" json:"images_dir_fd,omitempty"`
Pid *int32 `protobuf:"varint,2,opt,name=pid" json:"pid,omitempty"`
LeaveRunning *bool `protobuf:"varint,3,opt,name=leave_running,json=leaveRunning" json:"leave_running,omitempty"`
ExtUnixSk *bool `protobuf:"varint,4,opt,name=ext_unix_sk,json=extUnixSk" json:"ext_unix_sk,omitempty"`
TcpEstablished *bool `protobuf:"varint,5,opt,name=tcp_established,json=tcpEstablished" json:"tcp_established,omitempty"`
EvasiveDevices *bool `protobuf:"varint,6,opt,name=evasive_devices,json=evasiveDevices" json:"evasive_devices,omitempty"`
ShellJob *bool `protobuf:"varint,7,opt,name=shell_job,json=shellJob" json:"shell_job,omitempty"`
FileLocks *bool `protobuf:"varint,8,opt,name=file_locks,json=fileLocks" json:"file_locks,omitempty"`
LogLevel *int32 `protobuf:"varint,9,opt,name=log_level,json=logLevel,def=2" json:"log_level,omitempty"`
LogFile *string `protobuf:"bytes,10,opt,name=log_file,json=logFile" json:"log_file,omitempty"`
Ps *CriuPageServerInfo `protobuf:"bytes,11,opt,name=ps" json:"ps,omitempty"`
NotifyScripts *bool `protobuf:"varint,12,opt,name=notify_scripts,json=notifyScripts" json:"notify_scripts,omitempty"`
Root *string `protobuf:"bytes,13,opt,name=root" json:"root,omitempty"`
ParentImg *string `protobuf:"bytes,14,opt,name=parent_img,json=parentImg" json:"parent_img,omitempty"`
TrackMem *bool `protobuf:"varint,15,opt,name=track_mem,json=trackMem" json:"track_mem,omitempty"`
AutoDedup *bool `protobuf:"varint,16,opt,name=auto_dedup,json=autoDedup" json:"auto_dedup,omitempty"`
WorkDirFd *int32 `protobuf:"varint,17,opt,name=work_dir_fd,json=workDirFd" json:"work_dir_fd,omitempty"`
LinkRemap *bool `protobuf:"varint,18,opt,name=link_remap,json=linkRemap" json:"link_remap,omitempty"`
Veths []*CriuVethPair `protobuf:"bytes,19,rep,name=veths" json:"veths,omitempty"`
CpuCap *uint32 `protobuf:"varint,20,opt,name=cpu_cap,json=cpuCap,def=4294967295" json:"cpu_cap,omitempty"`
ForceIrmap *bool `protobuf:"varint,21,opt,name=force_irmap,json=forceIrmap" json:"force_irmap,omitempty"`
ExecCmd []string `protobuf:"bytes,22,rep,name=exec_cmd,json=execCmd" json:"exec_cmd,omitempty"`
ExtMnt []*ExtMountMap `protobuf:"bytes,23,rep,name=ext_mnt,json=extMnt" json:"ext_mnt,omitempty"`
ManageCgroups *bool `protobuf:"varint,24,opt,name=manage_cgroups,json=manageCgroups" json:"manage_cgroups,omitempty"`
CgRoot []*CgroupRoot `protobuf:"bytes,25,rep,name=cg_root,json=cgRoot" json:"cg_root,omitempty"`
RstSibling *bool `protobuf:"varint,26,opt,name=rst_sibling,json=rstSibling" json:"rst_sibling,omitempty"`
InheritFd []*InheritFd `protobuf:"bytes,27,rep,name=inherit_fd,json=inheritFd" json:"inherit_fd,omitempty"`
AutoExtMnt *bool `protobuf:"varint,28,opt,name=auto_ext_mnt,json=autoExtMnt" json:"auto_ext_mnt,omitempty"`
ExtSharing *bool `protobuf:"varint,29,opt,name=ext_sharing,json=extSharing" json:"ext_sharing,omitempty"`
ExtMasters *bool `protobuf:"varint,30,opt,name=ext_masters,json=extMasters" json:"ext_masters,omitempty"`
SkipMnt []string `protobuf:"bytes,31,rep,name=skip_mnt,json=skipMnt" json:"skip_mnt,omitempty"`
EnableFs []string `protobuf:"bytes,32,rep,name=enable_fs,json=enableFs" json:"enable_fs,omitempty"`
UnixSkIno []*UnixSk `protobuf:"bytes,33,rep,name=unix_sk_ino,json=unixSkIno" json:"unix_sk_ino,omitempty"`
ManageCgroupsMode *CriuCgMode `protobuf:"varint,34,opt,name=manage_cgroups_mode,json=manageCgroupsMode,enum=CriuCgMode" json:"manage_cgroups_mode,omitempty"`
GhostLimit *uint32 `protobuf:"varint,35,opt,name=ghost_limit,json=ghostLimit,def=1048576" json:"ghost_limit,omitempty"`
IrmapScanPaths []string `protobuf:"bytes,36,rep,name=irmap_scan_paths,json=irmapScanPaths" json:"irmap_scan_paths,omitempty"`
External []string `protobuf:"bytes,37,rep,name=external" json:"external,omitempty"`
EmptyNs *uint32 `protobuf:"varint,38,opt,name=empty_ns,json=emptyNs" json:"empty_ns,omitempty"`
JoinNs []*JoinNamespace `protobuf:"bytes,39,rep,name=join_ns,json=joinNs" json:"join_ns,omitempty"`
CgroupProps *string `protobuf:"bytes,41,opt,name=cgroup_props,json=cgroupProps" json:"cgroup_props,omitempty"`
CgroupPropsFile *string `protobuf:"bytes,42,opt,name=cgroup_props_file,json=cgroupPropsFile" json:"cgroup_props_file,omitempty"`
CgroupDumpController []string `protobuf:"bytes,43,rep,name=cgroup_dump_controller,json=cgroupDumpController" json:"cgroup_dump_controller,omitempty"`
FreezeCgroup *string `protobuf:"bytes,44,opt,name=freeze_cgroup,json=freezeCgroup" json:"freeze_cgroup,omitempty"`
Timeout *uint32 `protobuf:"varint,45,opt,name=timeout" json:"timeout,omitempty"`
TcpSkipInFlight *bool `protobuf:"varint,46,opt,name=tcp_skip_in_flight,json=tcpSkipInFlight" json:"tcp_skip_in_flight,omitempty"`
WeakSysctls *bool `protobuf:"varint,47,opt,name=weak_sysctls,json=weakSysctls" json:"weak_sysctls,omitempty"`
LazyPages *bool `protobuf:"varint,48,opt,name=lazy_pages,json=lazyPages" json:"lazy_pages,omitempty"`
StatusFd *int32 `protobuf:"varint,49,opt,name=status_fd,json=statusFd" json:"status_fd,omitempty"`
OrphanPtsMaster *bool `protobuf:"varint,50,opt,name=orphan_pts_master,json=orphanPtsMaster" json:"orphan_pts_master,omitempty"`
XXX_unrecognized []byte `json:"-"`
ImagesDirFd *int32 `protobuf:"varint,1,req,name=images_dir_fd" json:"images_dir_fd,omitempty"`
Pid *int32 `protobuf:"varint,2,opt,name=pid" json:"pid,omitempty"`
LeaveRunning *bool `protobuf:"varint,3,opt,name=leave_running" json:"leave_running,omitempty"`
ExtUnixSk *bool `protobuf:"varint,4,opt,name=ext_unix_sk" json:"ext_unix_sk,omitempty"`
TcpEstablished *bool `protobuf:"varint,5,opt,name=tcp_established" json:"tcp_established,omitempty"`
EvasiveDevices *bool `protobuf:"varint,6,opt,name=evasive_devices" json:"evasive_devices,omitempty"`
ShellJob *bool `protobuf:"varint,7,opt,name=shell_job" json:"shell_job,omitempty"`
FileLocks *bool `protobuf:"varint,8,opt,name=file_locks" json:"file_locks,omitempty"`
LogLevel *int32 `protobuf:"varint,9,opt,name=log_level,def=2" json:"log_level,omitempty"`
LogFile *string `protobuf:"bytes,10,opt,name=log_file" json:"log_file,omitempty"`
Ps *CriuPageServerInfo `protobuf:"bytes,11,opt,name=ps" json:"ps,omitempty"`
NotifyScripts *bool `protobuf:"varint,12,opt,name=notify_scripts" json:"notify_scripts,omitempty"`
Root *string `protobuf:"bytes,13,opt,name=root" json:"root,omitempty"`
ParentImg *string `protobuf:"bytes,14,opt,name=parent_img" json:"parent_img,omitempty"`
TrackMem *bool `protobuf:"varint,15,opt,name=track_mem" json:"track_mem,omitempty"`
AutoDedup *bool `protobuf:"varint,16,opt,name=auto_dedup" json:"auto_dedup,omitempty"`
WorkDirFd *int32 `protobuf:"varint,17,opt,name=work_dir_fd" json:"work_dir_fd,omitempty"`
LinkRemap *bool `protobuf:"varint,18,opt,name=link_remap" json:"link_remap,omitempty"`
Veths []*CriuVethPair `protobuf:"bytes,19,rep,name=veths" json:"veths,omitempty"`
CpuCap *uint32 `protobuf:"varint,20,opt,name=cpu_cap,def=4294967295" json:"cpu_cap,omitempty"`
ForceIrmap *bool `protobuf:"varint,21,opt,name=force_irmap" json:"force_irmap,omitempty"`
ExecCmd []string `protobuf:"bytes,22,rep,name=exec_cmd" json:"exec_cmd,omitempty"`
ExtMnt []*ExtMountMap `protobuf:"bytes,23,rep,name=ext_mnt" json:"ext_mnt,omitempty"`
ManageCgroups *bool `protobuf:"varint,24,opt,name=manage_cgroups" json:"manage_cgroups,omitempty"`
CgRoot []*CgroupRoot `protobuf:"bytes,25,rep,name=cg_root" json:"cg_root,omitempty"`
RstSibling *bool `protobuf:"varint,26,opt,name=rst_sibling" json:"rst_sibling,omitempty"`
InheritFd []*InheritFd `protobuf:"bytes,27,rep,name=inherit_fd" json:"inherit_fd,omitempty"`
AutoExtMnt *bool `protobuf:"varint,28,opt,name=auto_ext_mnt" json:"auto_ext_mnt,omitempty"`
ExtSharing *bool `protobuf:"varint,29,opt,name=ext_sharing" json:"ext_sharing,omitempty"`
ExtMasters *bool `protobuf:"varint,30,opt,name=ext_masters" json:"ext_masters,omitempty"`
SkipMnt []string `protobuf:"bytes,31,rep,name=skip_mnt" json:"skip_mnt,omitempty"`
EnableFs []string `protobuf:"bytes,32,rep,name=enable_fs" json:"enable_fs,omitempty"`
UnixSkIno []*UnixSk `protobuf:"bytes,33,rep,name=unix_sk_ino" json:"unix_sk_ino,omitempty"`
ManageCgroupsMode *CriuCgMode `protobuf:"varint,34,opt,name=manage_cgroups_mode,enum=CriuCgMode" json:"manage_cgroups_mode,omitempty"`
GhostLimit *uint32 `protobuf:"varint,35,opt,name=ghost_limit,def=1048576" json:"ghost_limit,omitempty"`
IrmapScanPaths []string `protobuf:"bytes,36,rep,name=irmap_scan_paths" json:"irmap_scan_paths,omitempty"`
External []string `protobuf:"bytes,37,rep,name=external" json:"external,omitempty"`
EmptyNs *uint32 `protobuf:"varint,38,opt,name=empty_ns" json:"empty_ns,omitempty"`
NoSeccomp *bool `protobuf:"varint,39,opt,name=no_seccomp" json:"no_seccomp,omitempty"`
XXX_unrecognized []byte `json:"-"`
}
func (m *CriuOpts) Reset() { *m = CriuOpts{} }
func (m *CriuOpts) String() string { return proto.CompactTextString(m) }
func (*CriuOpts) ProtoMessage() {}
func (*CriuOpts) Descriptor() ([]byte, []int) { return fileDescriptor0, []int{7} }
func (m *CriuOpts) Reset() { *m = CriuOpts{} }
func (m *CriuOpts) String() string { return proto.CompactTextString(m) }
func (*CriuOpts) ProtoMessage() {}
const Default_CriuOpts_LogLevel int32 = 2
const Default_CriuOpts_CpuCap uint32 = 4294967295
@@ -671,79 +606,9 @@ func (m *CriuOpts) GetEmptyNs() uint32 {
return 0
}
func (m *CriuOpts) GetJoinNs() []*JoinNamespace {
if m != nil {
return m.JoinNs
}
return nil
}
func (m *CriuOpts) GetCgroupProps() string {
if m != nil && m.CgroupProps != nil {
return *m.CgroupProps
}
return ""
}
func (m *CriuOpts) GetCgroupPropsFile() string {
if m != nil && m.CgroupPropsFile != nil {
return *m.CgroupPropsFile
}
return ""
}
func (m *CriuOpts) GetCgroupDumpController() []string {
if m != nil {
return m.CgroupDumpController
}
return nil
}
func (m *CriuOpts) GetFreezeCgroup() string {
if m != nil && m.FreezeCgroup != nil {
return *m.FreezeCgroup
}
return ""
}
func (m *CriuOpts) GetTimeout() uint32 {
if m != nil && m.Timeout != nil {
return *m.Timeout
}
return 0
}
func (m *CriuOpts) GetTcpSkipInFlight() bool {
if m != nil && m.TcpSkipInFlight != nil {
return *m.TcpSkipInFlight
}
return false
}
func (m *CriuOpts) GetWeakSysctls() bool {
if m != nil && m.WeakSysctls != nil {
return *m.WeakSysctls
}
return false
}
func (m *CriuOpts) GetLazyPages() bool {
if m != nil && m.LazyPages != nil {
return *m.LazyPages
}
return false
}
func (m *CriuOpts) GetStatusFd() int32 {
if m != nil && m.StatusFd != nil {
return *m.StatusFd
}
return 0
}
func (m *CriuOpts) GetOrphanPtsMaster() bool {
if m != nil && m.OrphanPtsMaster != nil {
return *m.OrphanPtsMaster
func (m *CriuOpts) GetNoSeccomp() bool {
if m != nil && m.NoSeccomp != nil {
return *m.NoSeccomp
}
return false
}
@@ -753,10 +618,9 @@ type CriuDumpResp struct {
XXX_unrecognized []byte `json:"-"`
}
func (m *CriuDumpResp) Reset() { *m = CriuDumpResp{} }
func (m *CriuDumpResp) String() string { return proto.CompactTextString(m) }
func (*CriuDumpResp) ProtoMessage() {}
func (*CriuDumpResp) Descriptor() ([]byte, []int) { return fileDescriptor0, []int{8} }
func (m *CriuDumpResp) Reset() { *m = CriuDumpResp{} }
func (m *CriuDumpResp) String() string { return proto.CompactTextString(m) }
func (*CriuDumpResp) ProtoMessage() {}
func (m *CriuDumpResp) GetRestored() bool {
if m != nil && m.Restored != nil {
@@ -770,10 +634,9 @@ type CriuRestoreResp struct {
XXX_unrecognized []byte `json:"-"`
}
func (m *CriuRestoreResp) Reset() { *m = CriuRestoreResp{} }
func (m *CriuRestoreResp) String() string { return proto.CompactTextString(m) }
func (*CriuRestoreResp) ProtoMessage() {}
func (*CriuRestoreResp) Descriptor() ([]byte, []int) { return fileDescriptor0, []int{9} }
func (m *CriuRestoreResp) Reset() { *m = CriuRestoreResp{} }
func (m *CriuRestoreResp) String() string { return proto.CompactTextString(m) }
func (*CriuRestoreResp) ProtoMessage() {}
func (m *CriuRestoreResp) GetPid() int32 {
if m != nil && m.Pid != nil {
@@ -788,10 +651,9 @@ type CriuNotify struct {
XXX_unrecognized []byte `json:"-"`
}
func (m *CriuNotify) Reset() { *m = CriuNotify{} }
func (m *CriuNotify) String() string { return proto.CompactTextString(m) }
func (*CriuNotify) ProtoMessage() {}
func (*CriuNotify) Descriptor() ([]byte, []int) { return fileDescriptor0, []int{10} }
func (m *CriuNotify) Reset() { *m = CriuNotify{} }
func (m *CriuNotify) String() string { return proto.CompactTextString(m) }
func (*CriuNotify) ProtoMessage() {}
func (m *CriuNotify) GetScript() string {
if m != nil && m.Script != nil {
@@ -811,15 +673,13 @@ func (m *CriuNotify) GetPid() int32 {
// List of features which can queried via
// CRIU_REQ_TYPE__FEATURE_CHECK
type CriuFeatures struct {
MemTrack *bool `protobuf:"varint,1,opt,name=mem_track,json=memTrack" json:"mem_track,omitempty"`
LazyPages *bool `protobuf:"varint,2,opt,name=lazy_pages,json=lazyPages" json:"lazy_pages,omitempty"`
MemTrack *bool `protobuf:"varint,1,opt,name=mem_track" json:"mem_track,omitempty"`
XXX_unrecognized []byte `json:"-"`
}
func (m *CriuFeatures) Reset() { *m = CriuFeatures{} }
func (m *CriuFeatures) String() string { return proto.CompactTextString(m) }
func (*CriuFeatures) ProtoMessage() {}
func (*CriuFeatures) Descriptor() ([]byte, []int) { return fileDescriptor0, []int{11} }
func (m *CriuFeatures) Reset() { *m = CriuFeatures{} }
func (m *CriuFeatures) String() string { return proto.CompactTextString(m) }
func (*CriuFeatures) ProtoMessage() {}
func (m *CriuFeatures) GetMemTrack() bool {
if m != nil && m.MemTrack != nil {
@@ -828,22 +688,15 @@ func (m *CriuFeatures) GetMemTrack() bool {
return false
}
func (m *CriuFeatures) GetLazyPages() bool {
if m != nil && m.LazyPages != nil {
return *m.LazyPages
}
return false
}
type CriuReq struct {
Type *CriuReqType `protobuf:"varint,1,req,name=type,enum=CriuReqType" json:"type,omitempty"`
Opts *CriuOpts `protobuf:"bytes,2,opt,name=opts" json:"opts,omitempty"`
NotifySuccess *bool `protobuf:"varint,3,opt,name=notify_success,json=notifySuccess" json:"notify_success,omitempty"`
NotifySuccess *bool `protobuf:"varint,3,opt,name=notify_success" json:"notify_success,omitempty"`
//
// When set service won't close the connection but
// will wait for more req-s to appear. Works not
// for all request types.
KeepOpen *bool `protobuf:"varint,4,opt,name=keep_open,json=keepOpen" json:"keep_open,omitempty"`
KeepOpen *bool `protobuf:"varint,4,opt,name=keep_open" json:"keep_open,omitempty"`
//
// 'features' can be used to query which features
// are supported by the installed criu/kernel
@@ -852,10 +705,9 @@ type CriuReq struct {
XXX_unrecognized []byte `json:"-"`
}
func (m *CriuReq) Reset() { *m = CriuReq{} }
func (m *CriuReq) String() string { return proto.CompactTextString(m) }
func (*CriuReq) ProtoMessage() {}
func (*CriuReq) Descriptor() ([]byte, []int) { return fileDescriptor0, []int{12} }
func (m *CriuReq) Reset() { *m = CriuReq{} }
func (m *CriuReq) String() string { return proto.CompactTextString(m) }
func (*CriuReq) ProtoMessage() {}
func (m *CriuReq) GetType() CriuReqType {
if m != nil && m.Type != nil {
@@ -899,17 +751,14 @@ type CriuResp struct {
Restore *CriuRestoreResp `protobuf:"bytes,4,opt,name=restore" json:"restore,omitempty"`
Notify *CriuNotify `protobuf:"bytes,5,opt,name=notify" json:"notify,omitempty"`
Ps *CriuPageServerInfo `protobuf:"bytes,6,opt,name=ps" json:"ps,omitempty"`
CrErrno *int32 `protobuf:"varint,7,opt,name=cr_errno,json=crErrno" json:"cr_errno,omitempty"`
CrErrno *int32 `protobuf:"varint,7,opt,name=cr_errno" json:"cr_errno,omitempty"`
Features *CriuFeatures `protobuf:"bytes,8,opt,name=features" json:"features,omitempty"`
CrErrmsg *string `protobuf:"bytes,9,opt,name=cr_errmsg,json=crErrmsg" json:"cr_errmsg,omitempty"`
Version *CriuVersion `protobuf:"bytes,10,opt,name=version" json:"version,omitempty"`
XXX_unrecognized []byte `json:"-"`
}
func (m *CriuResp) Reset() { *m = CriuResp{} }
func (m *CriuResp) String() string { return proto.CompactTextString(m) }
func (*CriuResp) ProtoMessage() {}
func (*CriuResp) Descriptor() ([]byte, []int) { return fileDescriptor0, []int{13} }
func (m *CriuResp) Reset() { *m = CriuResp{} }
func (m *CriuResp) String() string { return proto.CompactTextString(m) }
func (*CriuResp) ProtoMessage() {}
func (m *CriuResp) GetType() CriuReqType {
if m != nil && m.Type != nil {
@@ -967,212 +816,7 @@ func (m *CriuResp) GetFeatures() *CriuFeatures {
return nil
}
func (m *CriuResp) GetCrErrmsg() string {
if m != nil && m.CrErrmsg != nil {
return *m.CrErrmsg
}
return ""
}
func (m *CriuResp) GetVersion() *CriuVersion {
if m != nil {
return m.Version
}
return nil
}
// Answer for criu_req_type.VERSION requests
type CriuVersion struct {
Major *int32 `protobuf:"varint,1,req,name=major" json:"major,omitempty"`
Minor *int32 `protobuf:"varint,2,req,name=minor" json:"minor,omitempty"`
Gitid *string `protobuf:"bytes,3,opt,name=gitid" json:"gitid,omitempty"`
Sublevel *int32 `protobuf:"varint,4,opt,name=sublevel" json:"sublevel,omitempty"`
Extra *int32 `protobuf:"varint,5,opt,name=extra" json:"extra,omitempty"`
Name *string `protobuf:"bytes,6,opt,name=name" json:"name,omitempty"`
XXX_unrecognized []byte `json:"-"`
}
func (m *CriuVersion) Reset() { *m = CriuVersion{} }
func (m *CriuVersion) String() string { return proto.CompactTextString(m) }
func (*CriuVersion) ProtoMessage() {}
func (*CriuVersion) Descriptor() ([]byte, []int) { return fileDescriptor0, []int{14} }
func (m *CriuVersion) GetMajor() int32 {
if m != nil && m.Major != nil {
return *m.Major
}
return 0
}
func (m *CriuVersion) GetMinor() int32 {
if m != nil && m.Minor != nil {
return *m.Minor
}
return 0
}
func (m *CriuVersion) GetGitid() string {
if m != nil && m.Gitid != nil {
return *m.Gitid
}
return ""
}
func (m *CriuVersion) GetSublevel() int32 {
if m != nil && m.Sublevel != nil {
return *m.Sublevel
}
return 0
}
func (m *CriuVersion) GetExtra() int32 {
if m != nil && m.Extra != nil {
return *m.Extra
}
return 0
}
func (m *CriuVersion) GetName() string {
if m != nil && m.Name != nil {
return *m.Name
}
return ""
}
func init() {
proto.RegisterType((*CriuPageServerInfo)(nil), "criu_page_server_info")
proto.RegisterType((*CriuVethPair)(nil), "criu_veth_pair")
proto.RegisterType((*ExtMountMap)(nil), "ext_mount_map")
proto.RegisterType((*JoinNamespace)(nil), "join_namespace")
proto.RegisterType((*InheritFd)(nil), "inherit_fd")
proto.RegisterType((*CgroupRoot)(nil), "cgroup_root")
proto.RegisterType((*UnixSk)(nil), "unix_sk")
proto.RegisterType((*CriuOpts)(nil), "criu_opts")
proto.RegisterType((*CriuDumpResp)(nil), "criu_dump_resp")
proto.RegisterType((*CriuRestoreResp)(nil), "criu_restore_resp")
proto.RegisterType((*CriuNotify)(nil), "criu_notify")
proto.RegisterType((*CriuFeatures)(nil), "criu_features")
proto.RegisterType((*CriuReq)(nil), "criu_req")
proto.RegisterType((*CriuResp)(nil), "criu_resp")
proto.RegisterType((*CriuVersion)(nil), "criu_version")
proto.RegisterEnum("CriuCgMode", CriuCgMode_name, CriuCgMode_value)
proto.RegisterEnum("CriuReqType", CriuReqType_name, CriuReqType_value)
}
func init() { proto.RegisterFile("criurpc.proto", fileDescriptor0) }
var fileDescriptor0 = []byte{
// 1781 bytes of a gzipped FileDescriptorProto
0x1f, 0x8b, 0x08, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02, 0xff, 0x8c, 0x56, 0xdd, 0x72, 0x5b, 0xb7,
0x11, 0x0e, 0x29, 0xfe, 0x1c, 0x82, 0x3f, 0xa6, 0x10, 0xdb, 0x81, 0x93, 0xda, 0x62, 0xe8, 0x28,
0x51, 0x15, 0x97, 0x4d, 0x58, 0x3b, 0xae, 0x33, 0xed, 0x85, 0x47, 0x22, 0x5d, 0x36, 0x92, 0xc8,
0x01, 0x25, 0xcf, 0xe4, 0x0a, 0x73, 0x74, 0x0e, 0x48, 0xc1, 0x3c, 0x7f, 0x05, 0x40, 0x45, 0xf2,
0x83, 0xf4, 0x29, 0xfa, 0x0c, 0x7d, 0x84, 0xbe, 0x4e, 0x6f, 0x3b, 0xbb, 0x00, 0x65, 0x29, 0xc9,
0xb4, 0xbd, 0xc3, 0x7e, 0x58, 0x00, 0xbb, 0xfb, 0xed, 0x0f, 0x48, 0x3b, 0xd2, 0x6a, 0xad, 0x8b,
0x68, 0x50, 0xe8, 0xdc, 0xe6, 0xfd, 0x25, 0x79, 0x00, 0x80, 0x28, 0xc2, 0xa5, 0x14, 0x46, 0xea,
0x4b, 0xa9, 0x85, 0xca, 0x16, 0x39, 0x65, 0xa4, 0x1e, 0xc6, 0xb1, 0x96, 0xc6, 0xb0, 0x52, 0xaf,
0xb4, 0xd7, 0xe0, 0x1b, 0x91, 0x52, 0x52, 0x29, 0x72, 0x6d, 0x59, 0xb9, 0x57, 0xda, 0xab, 0x72,
0x5c, 0xd3, 0x2e, 0xd9, 0x2a, 0x54, 0xcc, 0xb6, 0x10, 0x82, 0x25, 0xed, 0x90, 0xf2, 0x22, 0x66,
0x15, 0x04, 0xca, 0x8b, 0xb8, 0xff, 0x27, 0xd2, 0xc1, 0x87, 0x2e, 0xa5, 0xbd, 0x10, 0x45, 0xa8,
0x34, 0xfd, 0x98, 0x54, 0xd5, 0x42, 0xa8, 0x8c, 0x95, 0x7a, 0xe5, 0xbd, 0x06, 0xaf, 0xa8, 0xc5,
0x24, 0xa3, 0x0f, 0x48, 0x4d, 0x2d, 0x44, 0xbe, 0x86, 0xeb, 0x01, 0xad, 0xaa, 0xc5, 0x74, 0x6d,
0xfb, 0x7f, 0x20, 0x6d, 0x79, 0x65, 0x45, 0x9a, 0xaf, 0x33, 0x2b, 0xd2, 0xb0, 0x80, 0x07, 0x57,
0xf2, 0xda, 0x1f, 0x85, 0x25, 0x20, 0x97, 0x61, 0xe2, 0x8f, 0xc1, 0xb2, 0xff, 0x96, 0x74, 0xde,
0xe5, 0x2a, 0x13, 0x59, 0x98, 0x4a, 0x53, 0x84, 0x91, 0x04, 0xa3, 0x32, 0xe3, 0x0f, 0x95, 0x33,
0x43, 0x3f, 0x21, 0xf5, 0xcc, 0x88, 0x85, 0x4a, 0xa4, 0x3f, 0x57, 0xcb, 0xcc, 0x58, 0x25, 0x92,
0x7e, 0x46, 0x1a, 0xf2, 0xca, 0xea, 0x50, 0xe4, 0x85, 0x45, 0xaf, 0x1a, 0x3c, 0x40, 0x60, 0x5a,
0xd8, 0xfe, 0x80, 0x10, 0x95, 0x5d, 0x48, 0xad, 0xac, 0x58, 0xc4, 0xbf, 0x62, 0x89, 0x73, 0x1d,
0x2e, 0x74, 0xae, 0xbf, 0x20, 0xcd, 0x68, 0xa9, 0xf3, 0x75, 0x21, 0x74, 0x9e, 0x5b, 0x88, 0x5f,
0x64, 0x75, 0xe2, 0xc3, 0x8a, 0x6b, 0x8c, 0x69, 0x68, 0x2f, 0xbc, 0x15, 0xb8, 0xee, 0xef, 0x90,
0xfa, 0x3a, 0x53, 0x57, 0xc2, 0xac, 0xe8, 0x7d, 0x52, 0x55, 0x59, 0x1e, 0x4b, 0x7c, 0xa5, 0xcd,
0x9d, 0xd0, 0xff, 0x57, 0x9b, 0x34, 0x30, 0xa6, 0x79, 0x61, 0x0d, 0xed, 0x93, 0xb6, 0x4a, 0xc3,
0xa5, 0x34, 0x22, 0x56, 0x5a, 0x2c, 0x62, 0xd4, 0xad, 0xf2, 0xa6, 0x03, 0x0f, 0x95, 0x1e, 0xc7,
0x1b, 0x9a, 0xca, 0x1f, 0x68, 0x7a, 0x4a, 0xda, 0x89, 0x0c, 0x2f, 0xa5, 0xd0, 0xeb, 0x2c, 0x53,
0xd9, 0x12, 0x9d, 0x0d, 0x78, 0x0b, 0x41, 0xee, 0x30, 0xfa, 0x84, 0x34, 0x21, 0xfa, 0xde, 0x1a,
0x24, 0x35, 0xe0, 0x10, 0xa0, 0xb3, 0x4c, 0x5d, 0xcd, 0x57, 0xf4, 0x2b, 0x72, 0xcf, 0x46, 0x85,
0x90, 0xc6, 0x86, 0xe7, 0x89, 0x32, 0x17, 0x32, 0x66, 0x55, 0xd4, 0xe9, 0xd8, 0xa8, 0x18, 0x7d,
0x40, 0x41, 0x51, 0x5e, 0x86, 0x46, 0x5d, 0x4a, 0x11, 0xcb, 0x4b, 0x15, 0x49, 0xc3, 0x6a, 0x4e,
0xd1, 0xc3, 0x87, 0x0e, 0x85, 0xf8, 0x9b, 0x0b, 0x99, 0x24, 0xe2, 0x5d, 0x7e, 0xce, 0xea, 0xa8,
0x12, 0x20, 0xf0, 0xd7, 0xfc, 0x9c, 0x3e, 0x26, 0x04, 0x28, 0x13, 0x49, 0x1e, 0xad, 0x0c, 0x0b,
0x9c, 0x35, 0x80, 0x1c, 0x01, 0x40, 0x9f, 0x90, 0x46, 0x92, 0x2f, 0x45, 0x22, 0x2f, 0x65, 0xc2,
0x1a, 0xe0, 0xea, 0xf7, 0xa5, 0x21, 0x0f, 0x92, 0x7c, 0x79, 0x04, 0x10, 0x7d, 0x44, 0x60, 0xed,
0x58, 0x27, 0x2e, 0xb5, 0x93, 0x7c, 0x89, 0xb4, 0x7f, 0x49, 0xca, 0x85, 0x61, 0xcd, 0x5e, 0x69,
0xaf, 0x39, 0x7c, 0x38, 0xf8, 0xd5, 0xc2, 0xe0, 0xe5, 0xc2, 0xd0, 0x5d, 0xd2, 0xc9, 0x72, 0xab,
0x16, 0xd7, 0xc2, 0x44, 0x5a, 0x15, 0xd6, 0xb0, 0x16, 0x5a, 0xd1, 0x76, 0xe8, 0xdc, 0x81, 0xc0,
0x2a, 0x30, 0xce, 0xda, 0x8e, 0x69, 0x64, 0xff, 0x31, 0x21, 0x45, 0xa8, 0x65, 0x66, 0x85, 0x4a,
0x97, 0xac, 0x83, 0x3b, 0x0d, 0x87, 0x4c, 0xd2, 0x25, 0x38, 0x6e, 0x75, 0x18, 0xad, 0x44, 0x2a,
0x53, 0x76, 0xcf, 0x39, 0x8e, 0xc0, 0xb1, 0x4c, 0xe1, 0x6c, 0xb8, 0xb6, 0xb9, 0x88, 0x65, 0xbc,
0x2e, 0x58, 0xd7, 0x39, 0x0e, 0xc8, 0x21, 0x00, 0x40, 0xd3, 0x4f, 0xb9, 0x5e, 0x6d, 0xf8, 0xdf,
0x46, 0x96, 0x1b, 0x00, 0x39, 0xf6, 0x1f, 0x13, 0x92, 0xa8, 0x6c, 0x25, 0xb4, 0x4c, 0xc3, 0x82,
0x51, 0x77, 0x1c, 0x10, 0x0e, 0x00, 0xdd, 0x25, 0x55, 0x28, 0x4e, 0xc3, 0x3e, 0xee, 0x6d, 0xed,
0x35, 0x87, 0xf7, 0x06, 0x77, 0xeb, 0x95, 0xbb, 0x5d, 0xfa, 0x94, 0xd4, 0xa3, 0x62, 0x2d, 0xa2,
0xb0, 0x60, 0xf7, 0x7b, 0xa5, 0xbd, 0xf6, 0xf7, 0xe4, 0xf9, 0xf0, 0xd5, 0xf3, 0x57, 0xdf, 0xbd,
0x1c, 0xbe, 0x7a, 0xc1, 0x6b, 0x51, 0xb1, 0x3e, 0x08, 0x0b, 0xba, 0x43, 0x9a, 0x8b, 0x5c, 0x47,
0x52, 0x28, 0x0d, 0x6f, 0x3d, 0xc0, 0xb7, 0x08, 0x42, 0x13, 0x40, 0x80, 0x04, 0x79, 0x25, 0x23,
0x11, 0xa5, 0x31, 0x7b, 0xd8, 0xdb, 0x02, 0x12, 0x40, 0x3e, 0x48, 0x21, 0x49, 0xea, 0x58, 0xeb,
0x99, 0x65, 0x9f, 0xa0, 0x25, 0x9d, 0xc1, 0x9d, 0xda, 0xe7, 0x35, 0x79, 0x65, 0x8f, 0x33, 0x0b,
0x2c, 0xa4, 0x61, 0x06, 0xfc, 0xb8, 0xf2, 0x32, 0x8c, 0x39, 0x16, 0x1c, 0x7a, 0xe0, 0x40, 0xba,
0x4b, 0xea, 0xd1, 0x12, 0x4b, 0x8f, 0x3d, 0xc2, 0xfb, 0x5a, 0x83, 0x5b, 0xe5, 0xc8, 0x6b, 0xd1,
0x92, 0x03, 0x31, 0x3b, 0xa4, 0xa9, 0x8d, 0x15, 0x46, 0x9d, 0x27, 0x50, 0x07, 0x9f, 0x3a, 0x93,
0xb5, 0xb1, 0x73, 0x87, 0xd0, 0xfd, 0xdb, 0x65, 0xcf, 0x3e, 0xc3, 0xab, 0x9a, 0x83, 0x0f, 0x10,
0x6f, 0xf8, 0xf5, 0x38, 0xa6, 0x3d, 0xd2, 0x42, 0xa6, 0x36, 0x8e, 0xfc, 0xc6, 0xdd, 0x06, 0xd8,
0xc8, 0x19, 0xbf, 0xe3, 0x6a, 0xca, 0x5c, 0x84, 0x1a, 0x9e, 0x7b, 0xec, 0x14, 0xe4, 0x95, 0x9d,
0x3b, 0x64, 0xa3, 0x90, 0x86, 0xc6, 0x4a, 0x6d, 0xd8, 0x93, 0x1b, 0x85, 0x63, 0x87, 0x40, 0x08,
0xcd, 0x4a, 0x15, 0x78, 0xff, 0x8e, 0x0b, 0x21, 0xc8, 0x70, 0x39, 0xb4, 0xaf, 0x2c, 0x3c, 0x4f,
0xa4, 0x58, 0x18, 0xd6, 0xc3, 0xbd, 0xc0, 0x01, 0x63, 0x43, 0xf7, 0x48, 0xd3, 0x57, 0xb2, 0x50,
0x59, 0xce, 0x3e, 0x47, 0x47, 0x82, 0x81, 0xc7, 0x78, 0x63, 0x8d, 0x45, 0x3d, 0xc9, 0x72, 0xfa,
0x67, 0xf2, 0xf1, 0xdd, 0x00, 0x8b, 0x14, 0x9a, 0x50, 0xbf, 0x57, 0xda, 0xeb, 0x0c, 0xdb, 0x2e,
0x3f, 0xa2, 0x25, 0x82, 0x7c, 0xfb, 0x4e, 0xd0, 0x8f, 0xf3, 0x58, 0xc2, 0x43, 0xcb, 0x8b, 0xdc,
0x58, 0x91, 0xa8, 0x54, 0x59, 0xf6, 0x14, 0xb3, 0xa5, 0xfe, 0xed, 0x37, 0xcf, 0xff, 0xf8, 0xe2,
0xe5, 0x77, 0x9c, 0xe0, 0xde, 0x11, 0x6c, 0xd1, 0x3d, 0xd2, 0xc5, 0x44, 0x11, 0x26, 0x0a, 0x33,
0x01, 0xdd, 0xcf, 0xb0, 0x2f, 0xd0, 0xec, 0x0e, 0xe2, 0xf3, 0x28, 0xcc, 0x66, 0x80, 0xd2, 0x4f,
0x21, 0x6f, 0xac, 0xd4, 0x59, 0x98, 0xb0, 0x5d, 0xef, 0x98, 0x97, 0x31, 0xa7, 0xd2, 0xc2, 0x5e,
0x8b, 0xcc, 0xb0, 0x2f, 0xe1, 0x31, 0x5e, 0x47, 0xf9, 0x04, 0x7c, 0xae, 0xbb, 0x51, 0x60, 0xd8,
0x57, 0x3e, 0xbb, 0xef, 0x8e, 0x06, 0x5e, 0x03, 0xf9, 0xc4, 0xd0, 0xcf, 0x49, 0xcb, 0x67, 0x47,
0xa1, 0xf3, 0xc2, 0xb0, 0xdf, 0x62, 0x85, 0xfa, 0x06, 0x3e, 0x03, 0x88, 0xee, 0x93, 0xed, 0xdb,
0x2a, 0xae, 0x93, 0xec, 0xa3, 0xde, 0xbd, 0x5b, 0x7a, 0xd8, 0x51, 0x9e, 0x93, 0x87, 0x5e, 0x37,
0x5e, 0xa7, 0x85, 0x88, 0xf2, 0xcc, 0xea, 0x3c, 0x49, 0xa4, 0x66, 0x5f, 0xa3, 0xf5, 0xf7, 0xdd,
0xee, 0xe1, 0x3a, 0x2d, 0x0e, 0x6e, 0xf6, 0xa0, 0x2b, 0x2f, 0xb4, 0x94, 0xef, 0x37, 0x81, 0x67,
0xcf, 0xf0, 0xf6, 0x96, 0x03, 0x5d, 0x8c, 0x61, 0x42, 0x5b, 0x95, 0x4a, 0x98, 0x95, 0xbf, 0x73,
0xde, 0x7a, 0x91, 0x7e, 0x4d, 0x28, 0xf4, 0x63, 0xcc, 0x0e, 0x95, 0x89, 0x45, 0xa2, 0x96, 0x17,
0x96, 0x0d, 0x30, 0x83, 0xa0, 0x53, 0xcf, 0x57, 0xaa, 0x98, 0x64, 0x63, 0x84, 0xc1, 0xe1, 0x9f,
0x64, 0xb8, 0x12, 0xe6, 0xda, 0x44, 0x36, 0x31, 0xec, 0xf7, 0xa8, 0xd6, 0x04, 0x6c, 0xee, 0x20,
0x6c, 0x1c, 0xe1, 0xfb, 0x6b, 0xec, 0x85, 0x86, 0x7d, 0xe3, 0x1b, 0x47, 0xf8, 0xfe, 0x7a, 0x06,
0x00, 0x36, 0x6b, 0x1b, 0xda, 0xb5, 0x81, 0xba, 0xf8, 0x16, 0xbb, 0x4e, 0xe0, 0x80, 0x71, 0x0c,
0xc1, 0xca, 0x75, 0x71, 0x01, 0xb4, 0x5a, 0xe3, 0xb3, 0x99, 0x0d, 0x9d, 0x29, 0x6e, 0x63, 0x66,
0x8d, 0x4b, 0xe9, 0xfe, 0x33, 0xff, 0x47, 0xc0, 0x50, 0x69, 0x69, 0x0a, 0xa0, 0x5b, 0x4b, 0x63,
0x73, 0x2d, 0x63, 0x9c, 0x97, 0x01, 0xbf, 0x91, 0xfb, 0xbb, 0x64, 0x1b, 0xb5, 0x3d, 0xe0, 0x0e,
0xf8, 0x09, 0xe7, 0x66, 0x1f, 0x2c, 0xfb, 0x2f, 0x49, 0x13, 0xd5, 0x5c, 0x6b, 0xa6, 0x0f, 0x49,
0xcd, 0xf5, 0x6c, 0x3f, 0x7f, 0xbd, 0xf4, 0xcb, 0xd1, 0xd8, 0xff, 0xc1, 0xfd, 0x95, 0xc4, 0x42,
0x86, 0x76, 0xad, 0x9d, 0x9f, 0xa9, 0x4c, 0x05, 0xb6, 0xe3, 0x8d, 0x35, 0xa9, 0x4c, 0x4f, 0x41,
0xfe, 0x59, 0x8c, 0xca, 0x3f, 0x8b, 0x51, 0xff, 0x9f, 0x25, 0x12, 0x78, 0x6b, 0xff, 0x46, 0xfb,
0xa4, 0x62, 0xaf, 0x0b, 0x37, 0xcd, 0x3b, 0xc3, 0xce, 0x60, 0xb3, 0x21, 0x00, 0xe5, 0xb8, 0x47,
0x9f, 0x90, 0x0a, 0x8c, 0x75, 0xbc, 0xa9, 0x39, 0x24, 0x83, 0x9b, 0x41, 0xcf, 0x11, 0xbf, 0x3d,
0x82, 0xd6, 0x51, 0x04, 0xdf, 0xb4, 0xad, 0x3b, 0x23, 0xc8, 0x81, 0x60, 0xf3, 0x4a, 0xca, 0x42,
0xe4, 0x85, 0xcc, 0xfc, 0xe0, 0x0e, 0x00, 0x98, 0x16, 0x32, 0xa3, 0xfb, 0x24, 0xd8, 0x38, 0x87,
0x03, 0xbb, 0xb9, 0xb1, 0x65, 0x83, 0xf2, 0x9b, 0xfd, 0xfe, 0xbf, 0xcb, 0xfe, 0xb3, 0x81, 0x61,
0xfe, 0x7f, 0x3c, 0x60, 0xa4, 0xbe, 0x31, 0x0d, 0xbe, 0x35, 0x01, 0xdf, 0x88, 0xf4, 0x29, 0xa9,
0x00, 0xc5, 0x68, 0xf1, 0xcd, 0xa0, 0xb9, 0x21, 0x9d, 0xe3, 0x26, 0x7d, 0x46, 0xea, 0x9e, 0x59,
0xb4, 0xbb, 0x39, 0xa4, 0x83, 0x5f, 0xd0, 0xcd, 0x37, 0x2a, 0xf4, 0x0b, 0x52, 0x73, 0x8e, 0x7b,
0x47, 0x5a, 0x83, 0x5b, 0xa4, 0x73, 0xbf, 0xe7, 0xe7, 0x7b, 0xed, 0x7f, 0xce, 0xf7, 0x47, 0x40,
0x96, 0x90, 0x5a, 0x67, 0x39, 0xfe, 0x3e, 0xaa, 0xbc, 0x1e, 0xe9, 0x11, 0x88, 0x77, 0x62, 0x16,
0xfc, 0xf7, 0x98, 0x41, 0xf0, 0xdd, 0x35, 0xa9, 0x59, 0xe2, 0x4f, 0xa4, 0xc1, 0x03, 0xbc, 0x27,
0x35, 0x4b, 0x18, 0x73, 0x97, 0x52, 0x1b, 0x95, 0x67, 0xf8, 0x0b, 0x69, 0x6e, 0x1a, 0xaa, 0x07,
0xf9, 0x66, 0xb7, 0xff, 0xf7, 0x12, 0x69, 0xdd, 0xde, 0x81, 0xdf, 0x60, 0x1a, 0xbe, 0xcb, 0xb5,
0xcf, 0x72, 0x27, 0x20, 0xaa, 0xb2, 0x5c, 0xfb, 0x8f, 0xa7, 0x13, 0x00, 0x5d, 0x2a, 0xeb, 0xbf,
0xe6, 0x0d, 0xee, 0x04, 0x28, 0x2b, 0xb3, 0x3e, 0x77, 0x3f, 0xa4, 0x8a, 0x2f, 0x58, 0x2f, 0xc3,
0x09, 0xfc, 0xe9, 0x62, 0x20, 0xab, 0xdc, 0x09, 0xf0, 0x95, 0x81, 0x5e, 0x89, 0xb1, 0x6b, 0x70,
0x5c, 0xef, 0x0b, 0x6f, 0x97, 0x1f, 0x01, 0x94, 0x90, 0xda, 0xe4, 0xcd, 0xc9, 0x94, 0x8f, 0xba,
0x1f, 0xd1, 0x26, 0xa9, 0x1f, 0xbc, 0x11, 0x27, 0xd3, 0x93, 0x51, 0xb7, 0x44, 0x1b, 0xa4, 0x3a,
0xe3, 0xd3, 0xd9, 0xbc, 0x5b, 0xa6, 0x01, 0xa9, 0xcc, 0xa7, 0xe3, 0xd3, 0xee, 0x16, 0xac, 0xc6,
0x67, 0x47, 0x47, 0xdd, 0x0a, 0x9c, 0x9b, 0x9f, 0xf2, 0xc9, 0xc1, 0x69, 0xb7, 0x0a, 0xe7, 0x0e,
0x47, 0xe3, 0xd7, 0x67, 0x47, 0xa7, 0xdd, 0xda, 0xfe, 0x3f, 0x4a, 0xbe, 0x04, 0x37, 0x99, 0x05,
0x37, 0x8d, 0x8e, 0x67, 0xa7, 0x3f, 0x76, 0x3f, 0x82, 0xf3, 0x87, 0x67, 0xc7, 0xb3, 0x6e, 0x09,
0xce, 0xf0, 0xd1, 0xfc, 0x14, 0x1e, 0x2e, 0x83, 0xc6, 0xc1, 0x5f, 0x46, 0x07, 0x3f, 0x74, 0xb7,
0x68, 0x8b, 0x04, 0x33, 0x3e, 0x12, 0xa8, 0x55, 0xa1, 0xf7, 0x48, 0x73, 0xf6, 0xfa, 0xcd, 0x48,
0xcc, 0x47, 0xfc, 0xed, 0x88, 0x77, 0xab, 0xf0, 0xec, 0xc9, 0xf4, 0x74, 0x32, 0xfe, 0xb1, 0x5b,
0xa3, 0x5d, 0xd2, 0x3a, 0x98, 0x9d, 0x4d, 0x4e, 0xc6, 0x53, 0xa7, 0x5e, 0xa7, 0xdb, 0xa4, 0xbd,
0x41, 0xdc, 0x7d, 0x01, 0x40, 0xe3, 0xd1, 0xeb, 0xd3, 0x33, 0x3e, 0xf2, 0x50, 0x03, 0x9e, 0x7e,
0x3b, 0xe2, 0xf3, 0xc9, 0xf4, 0xa4, 0x4b, 0xfe, 0x13, 0x00, 0x00, 0xff, 0xff, 0x5f, 0x2a, 0xaf,
0x49, 0x5b, 0x0d, 0x00, 0x00,
}

View File

@@ -1,5 +1,3 @@
syntax = "proto2";
message criu_page_server_info {
optional string address = 1;
optional int32 port = 2;
@@ -17,12 +15,6 @@ message ext_mount_map {
required string val = 2;
};
message join_namespace {
required string ns = 1;
required string ns_file = 2;
optional string extra_opt = 3;
}
message inherit_fd {
required string key = 1;
required int32 fd = 2;
@@ -39,7 +31,7 @@ message unix_sk {
enum criu_cg_mode {
IGNORE = 0;
CG_NONE = 1;
NONE = 1;
PROPS = 2;
SOFT = 3;
FULL = 4;
@@ -71,13 +63,13 @@ message criu_opts {
optional int32 work_dir_fd = 17;
optional bool link_remap = 18;
repeated criu_veth_pair veths = 19; /* DEPRECATED, use external instead */
repeated criu_veth_pair veths = 19;
optional uint32 cpu_cap = 20 [default = 0xffffffff];
optional bool force_irmap = 21;
repeated string exec_cmd = 22;
repeated ext_mount_map ext_mnt = 23; /* DEPRECATED, use external instead */
repeated ext_mount_map ext_mnt = 23;
optional bool manage_cgroups = 24; /* backward compatibility */
repeated cgroup_root cg_root = 25;
@@ -91,26 +83,14 @@ message criu_opts {
repeated string skip_mnt = 31;
repeated string enable_fs = 32;
repeated unix_sk unix_sk_ino = 33; /* DEPRECATED, use external instead */
repeated unix_sk unix_sk_ino = 33;
optional criu_cg_mode manage_cgroups_mode = 34;
optional uint32 ghost_limit = 35 [default = 0x100000];
repeated string irmap_scan_paths = 36;
repeated string external = 37;
optional uint32 empty_ns = 38;
repeated join_namespace join_ns = 39;
optional string cgroup_props = 41;
optional string cgroup_props_file = 42;
repeated string cgroup_dump_controller = 43;
optional string freeze_cgroup = 44;
optional uint32 timeout = 45;
optional bool tcp_skip_in_flight = 46;
optional bool weak_sysctls = 47;
optional bool lazy_pages = 48;
optional int32 status_fd = 49;
optional bool orphan_pts_master = 50;
optional bool no_seccomp = 39;
}
message criu_dump_resp {
@@ -140,8 +120,6 @@ enum criu_req_type {
CPUINFO_CHECK = 8;
FEATURE_CHECK = 9;
VERSION = 10;
}
/*
@@ -150,7 +128,6 @@ enum criu_req_type {
*/
message criu_features {
optional bool mem_track = 1;
optional bool lazy_pages = 2;
}
/*
@@ -194,16 +171,4 @@ message criu_resp {
optional int32 cr_errno = 7;
optional criu_features features = 8;
optional string cr_errmsg = 9;
optional criu_version version = 10;
}
/* Answer for criu_req_type.VERSION requests */
message criu_version {
required int32 major = 1;
required int32 minor = 2;
optional string gitid = 3;
optional int32 sublevel = 4;
optional int32 extra = 5;
optional string name = 6;
}

View File

@@ -6,16 +6,14 @@ import (
"errors"
"os"
"testing"
"golang.org/x/sys/unix"
)
func TestDeviceFromPathLstatFailure(t *testing.T) {
testError := errors.New("test error")
// Override unix.Lstat to inject error.
unixLstat = func(path string, stat *unix.Stat_t) error {
return testError
// Override os.Lstat to inject error.
osLstat = func(path string) (os.FileInfo, error) {
return nil, testError
}
_, err := DeviceFromPath("", "")

View File

@@ -1,14 +1,16 @@
// +build linux freebsd
package devices
import (
"errors"
"fmt"
"io/ioutil"
"os"
"path/filepath"
"syscall"
"github.com/opencontainers/runc/libcontainer/configs"
"golang.org/x/sys/unix"
)
var (
@@ -17,45 +19,45 @@ var (
// Testing dependencies
var (
unixLstat = unix.Lstat
osLstat = os.Lstat
ioutilReadDir = ioutil.ReadDir
)
// Given the path to a device and its cgroup_permissions(which cannot be easily queried) look up the information about a linux device and return that information as a Device struct.
// Given the path to a device and it's cgroup_permissions(which cannot be easily queried) look up the information about a linux device and return that information as a Device struct.
func DeviceFromPath(path, permissions string) (*configs.Device, error) {
var stat unix.Stat_t
err := unixLstat(path, &stat)
fileInfo, err := osLstat(path)
if err != nil {
return nil, err
}
var (
devNumber = int(stat.Rdev)
major = Major(devNumber)
)
if major == 0 {
return nil, ErrNotADevice
}
var (
devType rune
mode = stat.Mode
devType rune
mode = fileInfo.Mode()
fileModePermissionBits = os.FileMode.Perm(mode)
)
switch {
case mode&unix.S_IFBLK == unix.S_IFBLK:
devType = 'b'
case mode&unix.S_IFCHR == unix.S_IFCHR:
case mode&os.ModeDevice == 0:
return nil, ErrNotADevice
case mode&os.ModeCharDevice != 0:
fileModePermissionBits |= syscall.S_IFCHR
devType = 'c'
default:
fileModePermissionBits |= syscall.S_IFBLK
devType = 'b'
}
stat_t, ok := fileInfo.Sys().(*syscall.Stat_t)
if !ok {
return nil, fmt.Errorf("cannot determine the device number for device %s", path)
}
devNumber := int(stat_t.Rdev)
return &configs.Device{
Type: devType,
Path: path,
Major: major,
Major: Major(devNumber),
Minor: Minor(devNumber),
Permissions: permissions,
FileMode: os.FileMode(mode),
Uid: stat.Uid,
Gid: stat.Gid,
FileMode: fileModePermissionBits,
Uid: stat_t.Uid,
Gid: stat_t.Gid,
}, nil
}
@@ -73,8 +75,7 @@ func getDevices(path string) ([]*configs.Device, error) {
switch {
case f.IsDir():
switch f.Name() {
// ".lxc" & ".lxd-mounts" added to address https://github.com/lxc/lxd/issues/2825
case "pts", "shm", "fd", "mqueue", ".lxc", ".lxd-mounts":
case "pts", "shm", "fd", "mqueue":
continue
default:
sub, err := getDevices(filepath.Join(path, f.Name()))
@@ -93,9 +94,6 @@ func getDevices(path string) ([]*configs.Device, error) {
if err == ErrNotADevice {
continue
}
if os.IsNotExist(err) {
continue
}
return nil, err
}
out = append(out, device)

View File

@@ -1,3 +1,3 @@
// +build !linux
// +build windows
package devices

View File

@@ -2,7 +2,7 @@ package libcontainer
import "io"
// ErrorCode is the API error code type.
// API error code type.
type ErrorCode int
// API error codes.
@@ -56,13 +56,13 @@ func (c ErrorCode) String() string {
}
}
// Error is the API error type.
// API Error type.
type Error interface {
error
// Returns an error if it failed to write the detail of the Error to w.
// The detail of the Error may include the error message and a
// representation of the stack trace.
// Returns a verbose string including the error message
// and a representation of the stack trace suitable for
// printing.
Detail(w io.Writer) error
// Returns the error code for this error.

View File

@@ -4,17 +4,12 @@ import "testing"
func TestErrorCode(t *testing.T) {
codes := map[ErrorCode]string{
IdInUse: "Id already in use",
InvalidIdFormat: "Invalid format",
ContainerPaused: "Container paused",
ConfigInvalid: "Invalid configuration",
SystemError: "System error",
ContainerNotExists: "Container does not exist",
ContainerNotStopped: "Container is not stopped",
ContainerNotRunning: "Container is not running",
ConsoleExists: "Console exists for process",
ContainerNotPaused: "Container is not paused",
NoProcessOps: "No process operations",
IdInUse: "Id already in use",
InvalidIdFormat: "Invalid format",
ContainerPaused: "Container paused",
ConfigInvalid: "Invalid configuration",
SystemError: "System error",
ContainerNotExists: "Container does not exist",
}
for code, expected := range codes {

View File

@@ -10,7 +10,7 @@ type Factory interface {
// between 1 and 1024 characters, inclusive.
//
// The id must not already be in use by an existing container. Containers created using
// a factory with the same path (and filesystem) must have distinct ids.
// a factory with the same path (and file system) must have distinct ids.
//
// Returns the new container with a running process.
//
@@ -28,6 +28,7 @@ type Factory interface {
//
// errors:
// Path does not exist
// Container is stopped
// System error
Load(id string) (Container, error)

View File

@@ -6,42 +6,58 @@ import (
"encoding/json"
"fmt"
"os"
"os/exec"
"path/filepath"
"regexp"
"runtime/debug"
"strconv"
"syscall"
"github.com/docker/docker/pkg/mount"
"github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/opencontainers/runc/libcontainer/cgroups/fs"
"github.com/opencontainers/runc/libcontainer/cgroups/rootless"
"github.com/opencontainers/runc/libcontainer/cgroups/systemd"
"github.com/opencontainers/runc/libcontainer/configs"
"github.com/opencontainers/runc/libcontainer/configs/validate"
"github.com/opencontainers/runc/libcontainer/utils"
"golang.org/x/sys/unix"
)
const (
stateFilename = "state.json"
execFifoFilename = "exec.fifo"
stateFilename = "state.json"
)
var idRegex = regexp.MustCompile(`^[\w+-\.]+$`)
var (
idRegex = regexp.MustCompile(`^[\w-\.]+$`)
maxIdLen = 1024
)
// InitArgs returns an options func to configure a LinuxFactory with the
// provided init binary path and arguments.
// provided init arguments.
func InitArgs(args ...string) func(*LinuxFactory) error {
return func(l *LinuxFactory) (err error) {
if len(args) > 0 {
// Resolve relative paths to ensure that its available
// after directory changes.
if args[0], err = filepath.Abs(args[0]); err != nil {
return newGenericError(err, ConfigInvalid)
return func(l *LinuxFactory) error {
name := args[0]
if filepath.Base(name) == name {
if lp, err := exec.LookPath(name); err == nil {
name = lp
}
} else {
abs, err := filepath.Abs(name)
if err != nil {
return err
}
name = abs
}
l.InitPath = "/proc/self/exe"
l.InitArgs = append([]string{name}, args[1:]...)
return nil
}
}
// InitPath returns an options func to configure a LinuxFactory with the
// provided absolute path to the init binary and arguements.
func InitPath(path string, args ...string) func(*LinuxFactory) error {
return func(l *LinuxFactory) error {
l.InitPath = path
l.InitArgs = args
return nil
}
@@ -72,20 +88,6 @@ func Cgroupfs(l *LinuxFactory) error {
return nil
}
// RootlessCgroups is an options func to configure a LinuxFactory to
// return containers that use the "rootless" cgroup manager, which will
// fail to do any operations not possible to do with an unprivileged user.
// It should only be used in conjunction with rootless containers.
func RootlessCgroups(l *LinuxFactory) error {
l.NewCgroupsManager = func(config *configs.Cgroup, paths map[string]string) cgroups.Manager {
return &rootless.Manager{
Cgroups: config,
Paths: paths,
}
}
return nil
}
// TmpfsRoot is an option func to mount LinuxFactory.Root to tmpfs.
func TmpfsRoot(l *LinuxFactory) error {
mounted, err := mount.Mounted(l.Root)
@@ -93,22 +95,13 @@ func TmpfsRoot(l *LinuxFactory) error {
return err
}
if !mounted {
if err := unix.Mount("tmpfs", l.Root, "tmpfs", 0, ""); err != nil {
if err := syscall.Mount("tmpfs", l.Root, "tmpfs", 0, ""); err != nil {
return err
}
}
return nil
}
// CriuPath returns an option func to configure a LinuxFactory with the
// provided criupath
func CriuPath(criupath string) func(*LinuxFactory) error {
return func(l *LinuxFactory) error {
l.CriuPath = criupath
return nil
}
}
// New returns a linux based container factory based in the root directory and
// configures the factory with the provided option funcs.
func New(root string, options ...func(*LinuxFactory) error) (Factory, error) {
@@ -119,10 +112,10 @@ func New(root string, options ...func(*LinuxFactory) error) (Factory, error) {
}
l := &LinuxFactory{
Root: root,
InitArgs: []string{"/proc/self/exe", "init"},
Validator: validate.New(),
CriuPath: "criu",
}
InitArgs(os.Args[0], "init")(l)
Cgroupfs(l)
for _, opt := range options {
if err := opt(l); err != nil {
@@ -137,6 +130,9 @@ type LinuxFactory struct {
// Root directory for the factory to store state.
Root string
// InitPath is the absolute path to the init binary.
InitPath string
// InitArgs are arguments for calling the init responsibilities for spawning
// a container.
InitArgs []string
@@ -168,19 +164,14 @@ func (l *LinuxFactory) Create(id string, config *configs.Config) (Container, err
} else if !os.IsNotExist(err) {
return nil, newGenericError(err, SystemError)
}
if err := os.MkdirAll(containerRoot, 0711); err != nil {
if err := os.MkdirAll(containerRoot, 0700); err != nil {
return nil, newGenericError(err, SystemError)
}
if err := os.Chown(containerRoot, unix.Geteuid(), unix.Getegid()); err != nil {
return nil, newGenericError(err, SystemError)
}
if config.Rootless {
RootlessCgroups(l)
}
c := &linuxContainer{
id: id,
root: containerRoot,
config: config,
initPath: l.InitPath,
initArgs: l.InitArgs,
criuPath: l.CriuPath,
cgroupManager: l.NewCgroupsManager(config.Cgroups, nil),
@@ -194,7 +185,7 @@ func (l *LinuxFactory) Load(id string) (Container, error) {
return nil, newGenericError(fmt.Errorf("invalid root"), ConfigInvalid)
}
containerRoot := filepath.Join(l.Root, id)
state, err := l.loadState(containerRoot, id)
state, err := l.loadState(containerRoot)
if err != nil {
return nil, err
}
@@ -203,22 +194,18 @@ func (l *LinuxFactory) Load(id string) (Container, error) {
processStartTime: state.InitProcessStartTime,
fds: state.ExternalDescriptors,
}
// We have to use the RootlessManager.
if state.Rootless {
RootlessCgroups(l)
}
c := &linuxContainer{
initProcess: r,
initProcessStartTime: state.InitProcessStartTime,
id: id,
config: &state.Config,
initArgs: l.InitArgs,
criuPath: l.CriuPath,
cgroupManager: l.NewCgroupsManager(state.Config.Cgroups, state.CgroupPaths),
root: containerRoot,
created: state.Created,
initProcess: r,
id: id,
config: &state.Config,
initPath: l.InitPath,
initArgs: l.InitArgs,
criuPath: l.CriuPath,
cgroupManager: l.NewCgroupsManager(state.Config.Cgroups, state.CgroupPaths),
root: containerRoot,
created: state.Created,
}
c.state = &loadedState{c: c}
c.state = &createdState{c: c, s: Created}
if err := c.refreshState(); err != nil {
return nil, err
}
@@ -232,79 +219,55 @@ func (l *LinuxFactory) Type() string {
// StartInitialization loads a container by opening the pipe fd from the parent to read the configuration and state
// This is a low level implementation detail of the reexec and should not be consumed externally
func (l *LinuxFactory) StartInitialization() (err error) {
var (
pipefd, rootfd int
consoleSocket *os.File
envInitPipe = os.Getenv("_LIBCONTAINER_INITPIPE")
envStateDir = os.Getenv("_LIBCONTAINER_STATEDIR")
envConsole = os.Getenv("_LIBCONTAINER_CONSOLE")
)
// Get the INITPIPE.
pipefd, err = strconv.Atoi(envInitPipe)
fdStr := os.Getenv("_LIBCONTAINER_INITPIPE")
pipefd, err := strconv.Atoi(fdStr)
if err != nil {
return fmt.Errorf("unable to convert _LIBCONTAINER_INITPIPE=%s to int: %s", envInitPipe, err)
return fmt.Errorf("error converting env var _LIBCONTAINER_INITPIPE(%q) to an int: %s", fdStr, err)
}
var (
pipe = os.NewFile(uintptr(pipefd), "pipe")
it = initType(os.Getenv("_LIBCONTAINER_INITTYPE"))
)
defer pipe.Close()
// Only init processes have STATEDIR.
rootfd = -1
if it == initStandard {
if rootfd, err = strconv.Atoi(envStateDir); err != nil {
return fmt.Errorf("unable to convert _LIBCONTAINER_STATEDIR=%s to int: %s", envStateDir, err)
}
}
if envConsole != "" {
console, err := strconv.Atoi(envConsole)
if err != nil {
return fmt.Errorf("unable to convert _LIBCONTAINER_CONSOLE=%s to int: %s", envConsole, err)
}
consoleSocket = os.NewFile(uintptr(console), "console-socket")
defer consoleSocket.Close()
}
// clear the current process's environment to clean any libcontainer
// specific env vars.
os.Clearenv()
var i initer
defer func() {
// We have an error during the initialization of the container's init,
// send it back to the parent process in the form of an initError.
if werr := utils.WriteJSON(pipe, syncT{procError}); werr != nil {
fmt.Fprintln(os.Stderr, err)
return
// If container's init successed, syscall.Exec will not return, hence
// this defer function will never be called.
if _, ok := i.(*linuxStandardInit); ok {
// Synchronisation only necessary for standard init.
if err := utils.WriteJSON(pipe, syncT{procError}); err != nil {
panic(err)
}
}
if werr := utils.WriteJSON(pipe, newSystemError(err)); werr != nil {
fmt.Fprintln(os.Stderr, err)
return
if err := utils.WriteJSON(pipe, newSystemError(err)); err != nil {
panic(err)
}
// ensure that this pipe is always closed
pipe.Close()
}()
defer func() {
if e := recover(); e != nil {
err = fmt.Errorf("panic from initialization: %v, %v", e, string(debug.Stack()))
}
}()
i, err := newContainerInit(it, pipe, consoleSocket, rootfd)
i, err = newContainerInit(it, pipe)
if err != nil {
return err
}
// If Init succeeds, syscall.Exec will not return, hence none of the defers will be called.
return i.Init()
}
func (l *LinuxFactory) loadState(root, id string) (*State, error) {
func (l *LinuxFactory) loadState(root string) (*State, error) {
f, err := os.Open(filepath.Join(root, stateFilename))
if err != nil {
if os.IsNotExist(err) {
return nil, newGenericError(fmt.Errorf("container %q does not exist", id), ContainerNotExists)
return nil, newGenericError(err, ContainerNotExists)
}
return nil, newGenericError(err, SystemError)
}
@@ -320,6 +283,8 @@ func (l *LinuxFactory) validateID(id string) error {
if !idRegex.MatchString(id) {
return newGenericError(fmt.Errorf("invalid id format: %v", id), InvalidIdFormat)
}
if len(id) > maxIdLen {
return newGenericError(fmt.Errorf("invalid id format: %v", id), InvalidIdFormat)
}
return nil
}

View File

@@ -7,13 +7,12 @@ import (
"os"
"path/filepath"
"reflect"
"syscall"
"testing"
"github.com/docker/docker/pkg/mount"
"github.com/opencontainers/runc/libcontainer/configs"
"github.com/opencontainers/runc/libcontainer/utils"
"golang.org/x/sys/unix"
)
func newTestRoot() (string, error) {
@@ -100,7 +99,7 @@ func TestFactoryNewTmpfs(t *testing.T) {
if !found {
t.Fatalf("Factory Root is not listed in mounts list")
}
defer unix.Unmount(root, unix.MNT_DETACH)
defer syscall.Unmount(root, syscall.MNT_DETACH)
}
func TestFactoryLoadNotExists(t *testing.T) {

View File

@@ -1,7 +1,6 @@
package libcontainer
import (
"fmt"
"io"
"text/template"
"time"
@@ -9,6 +8,20 @@ import (
"github.com/opencontainers/runc/libcontainer/stacktrace"
)
type syncType uint8
const (
procReady syncType = iota
procError
procRun
procHooks
procResume
)
type syncT struct {
Type syncType `json:"type"`
}
var errorTemplate = template.Must(template.New("error").Parse(`Timestamp: {{.Timestamp}}
Code: {{.ECode}}
{{if .Message }}
@@ -38,27 +51,14 @@ func newGenericError(err error, c ErrorCode) Error {
}
func newSystemError(err error) Error {
return createSystemError(err, "")
}
func newSystemErrorWithCausef(err error, cause string, v ...interface{}) Error {
return createSystemError(err, fmt.Sprintf(cause, v...))
}
func newSystemErrorWithCause(err error, cause string) Error {
return createSystemError(err, cause)
}
// createSystemError creates the specified error with the correct number of
// stack frames skipped. This is only to be called by the other functions for
// formatting the error.
func createSystemError(err error, cause string) Error {
if le, ok := err.(Error); ok {
return le
}
gerr := &genericError{
Timestamp: time.Now(),
Err: err,
ECode: SystemError,
Cause: cause,
Stack: stacktrace.Capture(2),
Stack: stacktrace.Capture(1),
}
if err != nil {
gerr.Message = err.Error()
@@ -70,17 +70,12 @@ type genericError struct {
Timestamp time.Time
ECode ErrorCode
Err error `json:"-"`
Cause string
Message string
Stack stacktrace.Stacktrace
}
func (e *genericError) Error() string {
if e.Cause == "" {
return e.Message
}
frame := e.Stack.Frames[0]
return fmt.Sprintf("%s:%d: %s caused %q", frame.File, frame.Line, e.Cause, e.Message)
return e.Message
}
func (e *genericError) Code() ErrorCode {

View File

@@ -12,38 +12,3 @@ func TestErrorDetail(t *testing.T) {
t.Fatal(derr)
}
}
func TestErrorWithCode(t *testing.T) {
err := newGenericError(fmt.Errorf("test error"), SystemError)
if code := err.Code(); code != SystemError {
t.Fatalf("expected err code %q but %q", SystemError, code)
}
}
func TestErrorWithError(t *testing.T) {
cc := []struct {
errmsg string
cause string
}{
{
errmsg: "test error",
},
{
errmsg: "test error",
cause: "test",
},
}
for _, v := range cc {
err := newSystemErrorWithCause(fmt.Errorf(v.errmsg), v.cause)
msg := err.Error()
if v.cause == "" && msg != v.errmsg {
t.Fatalf("expected err(%q) equal errmsg(%q)", msg, v.errmsg)
}
if v.cause != "" && msg == v.errmsg {
t.Fatalf("unexpected err(%q) equal errmsg(%q)", msg, v.errmsg)
}
}
}

View File

@@ -6,21 +6,20 @@ import (
"encoding/json"
"fmt"
"io"
"io/ioutil"
"net"
"os"
"strconv"
"strings"
"syscall" // only for Errno
"unsafe"
"syscall"
"github.com/Sirupsen/logrus"
"github.com/opencontainers/runc/libcontainer/cgroups"
"github.com/opencontainers/runc/libcontainer/configs"
"github.com/opencontainers/runc/libcontainer/system"
"github.com/opencontainers/runc/libcontainer/user"
"github.com/opencontainers/runc/libcontainer/utils"
"github.com/sirupsen/logrus"
"github.com/vishvananda/netlink"
"golang.org/x/sys/unix"
)
type initType string
@@ -31,8 +30,7 @@ const (
)
type pid struct {
Pid int `json:"pid"`
PidFirstChild int `json:"pid_first"`
Pid int `json:"pid"`
}
// network is an internal struct used to setup container networks.
@@ -46,29 +44,27 @@ type network struct {
// initConfig is used for transferring parameters from Exec() to Init()
type initConfig struct {
Args []string `json:"args"`
Env []string `json:"env"`
Cwd string `json:"cwd"`
Capabilities *configs.Capabilities `json:"capabilities"`
ProcessLabel string `json:"process_label"`
AppArmorProfile string `json:"apparmor_profile"`
NoNewPrivileges bool `json:"no_new_privileges"`
User string `json:"user"`
AdditionalGroups []string `json:"additional_groups"`
Config *configs.Config `json:"config"`
Networks []*network `json:"network"`
PassedFilesCount int `json:"passed_files_count"`
ContainerId string `json:"containerid"`
Rlimits []configs.Rlimit `json:"rlimits"`
CreateConsole bool `json:"create_console"`
Rootless bool `json:"rootless"`
Args []string `json:"args"`
Env []string `json:"env"`
Cwd string `json:"cwd"`
Capabilities []string `json:"capabilities"`
ProcessLabel string `json:"process_label"`
AppArmorProfile string `json:"apparmor_profile"`
NoNewPrivileges bool `json:"no_new_privileges"`
User string `json:"user"`
Config *configs.Config `json:"config"`
Console string `json:"console"`
Networks []*network `json:"network"`
PassedFilesCount int `json:"passed_files_count"`
ContainerId string `json:"containerid"`
Rlimits []configs.Rlimit `json:"rlimits"`
}
type initer interface {
Init() error
}
func newContainerInit(t initType, pipe *os.File, consoleSocket *os.File, stateDirFD int) (initer, error) {
func newContainerInit(t initType, pipe *os.File) (initer, error) {
var config *initConfig
if err := json.NewDecoder(pipe).Decode(&config); err != nil {
return nil, err
@@ -79,17 +75,13 @@ func newContainerInit(t initType, pipe *os.File, consoleSocket *os.File, stateDi
switch t {
case initSetns:
return &linuxSetnsInit{
pipe: pipe,
consoleSocket: consoleSocket,
config: config,
config: config,
}, nil
case initStandard:
return &linuxStandardInit{
pipe: pipe,
consoleSocket: consoleSocket,
parentPid: unix.Getppid(),
config: config,
stateDirFD: stateDirFD,
pipe: pipe,
parentPid: syscall.Getppid(),
config: config,
}, nil
}
return nil, fmt.Errorf("unknown init type %q", t)
@@ -121,18 +113,16 @@ func finalizeNamespace(config *initConfig) error {
return err
}
capabilities := &configs.Capabilities{}
capabilities := config.Config.Capabilities
if config.Capabilities != nil {
capabilities = config.Capabilities
} else if config.Config.Capabilities != nil {
capabilities = config.Config.Capabilities
}
w, err := newContainerCapList(capabilities)
w, err := newCapWhitelist(capabilities)
if err != nil {
return err
}
// drop capabilities in bounding set before changing user
if err := w.ApplyBoundingSet(); err != nil {
if err := w.dropBoundingSet(); err != nil {
return err
}
// preserve existing capabilities while we change users
@@ -145,54 +135,16 @@ func finalizeNamespace(config *initConfig) error {
if err := system.ClearKeepCaps(); err != nil {
return err
}
if err := w.ApplyCaps(); err != nil {
// drop all other capabilities
if err := w.drop(); err != nil {
return err
}
if config.Cwd != "" {
if err := unix.Chdir(config.Cwd); err != nil {
return fmt.Errorf("chdir to cwd (%q) set in config.json failed: %v", config.Cwd, err)
}
}
return nil
}
// setupConsole sets up the console from inside the container, and sends the
// master pty fd to the config.Pipe (using cmsg). This is done to ensure that
// consoles are scoped to a container properly (see runc#814 and the many
// issues related to that). This has to be run *after* we've pivoted to the new
// rootfs (and the users' configuration is entirely set up).
func setupConsole(socket *os.File, config *initConfig, mount bool) error {
defer socket.Close()
// At this point, /dev/ptmx points to something that we would expect. We
// used to change the owner of the slave path, but since the /dev/pts mount
// can have gid=X set (at the users' option). So touching the owner of the
// slave PTY is not necessary, as the kernel will handle that for us. Note
// however, that setupUser (specifically fixStdioPermissions) *will* change
// the UID owner of the console to be the user the process will run as (so
// they can actually control their console).
console, err := newConsole()
if err != nil {
return err
}
// After we return from here, we don't need the console anymore.
defer console.Close()
linuxConsole, ok := console.(*linuxConsole)
if !ok {
return fmt.Errorf("failed to cast console to *linuxConsole")
}
// Mount the console inside our rootfs.
if mount {
if err := linuxConsole.mount(); err != nil {
if err := syscall.Chdir(config.Cwd); err != nil {
return err
}
}
// While we can access console.master, using the API is a good idea.
if err := utils.SendFd(socket, linuxConsole.File()); err != nil {
return err
}
// Now, dup over all the things.
return linuxConsole.dupStdio()
return nil
}
// syncParentReady sends to the given pipe a JSON payload which indicates that
@@ -200,15 +152,19 @@ func setupConsole(socket *os.File, config *initConfig, mount bool) error {
// indicate that it is cleared to Exec.
func syncParentReady(pipe io.ReadWriter) error {
// Tell parent.
if err := writeSync(pipe, procReady); err != nil {
if err := utils.WriteJSON(pipe, syncT{procReady}); err != nil {
return err
}
// Wait for parent to give the all-clear.
if err := readSync(pipe, procRun); err != nil {
return err
var procSync syncT
if err := json.NewDecoder(pipe).Decode(&procSync); err != nil {
if err == io.EOF {
return fmt.Errorf("parent closed synchronisation channel")
}
if procSync.Type != procRun {
return fmt.Errorf("invalid synchronisation flag from parent")
}
}
return nil
}
@@ -217,15 +173,19 @@ func syncParentReady(pipe io.ReadWriter) error {
// indicate that it is cleared to resume.
func syncParentHooks(pipe io.ReadWriter) error {
// Tell parent.
if err := writeSync(pipe, procHooks); err != nil {
if err := utils.WriteJSON(pipe, syncT{procHooks}); err != nil {
return err
}
// Wait for parent to give the all-clear.
if err := readSync(pipe, procResume); err != nil {
return err
var procSync syncT
if err := json.NewDecoder(pipe).Decode(&procSync); err != nil {
if err == io.EOF {
return fmt.Errorf("parent closed synchronisation channel")
}
if procSync.Type != procResume {
return fmt.Errorf("invalid synchronisation flag from parent")
}
}
return nil
}
@@ -233,76 +193,46 @@ func syncParentHooks(pipe io.ReadWriter) error {
func setupUser(config *initConfig) error {
// Set up defaults.
defaultExecUser := user.ExecUser{
Uid: 0,
Gid: 0,
Uid: syscall.Getuid(),
Gid: syscall.Getgid(),
Home: "/",
}
passwdPath, err := user.GetPasswdPath()
if err != nil {
return err
}
groupPath, err := user.GetGroupPath()
if err != nil {
return err
}
execUser, err := user.GetExecUserPath(config.User, &defaultExecUser, passwdPath, groupPath)
if err != nil {
return err
}
var addGroups []int
if len(config.AdditionalGroups) > 0 {
addGroups, err = user.GetAdditionalGroupsPath(config.AdditionalGroups, groupPath)
if len(config.Config.AdditionalGroups) > 0 {
addGroups, err = user.GetAdditionalGroupsPath(config.Config.AdditionalGroups, groupPath)
if err != nil {
return err
}
}
if config.Rootless {
if execUser.Uid != 0 {
return fmt.Errorf("cannot run as a non-root user in a rootless container")
}
if execUser.Gid != 0 {
return fmt.Errorf("cannot run as a non-root group in a rootless container")
}
// We cannot set any additional groups in a rootless container and thus we
// bail if the user asked us to do so. TODO: We currently can't do this
// earlier, but if libcontainer.Process.User was typesafe this might work.
if len(addGroups) > 0 {
return fmt.Errorf("cannot set any additional groups in a rootless container")
}
}
// before we change to the container's user make sure that the processes STDIO
// is correctly owned by the user that we are switching to.
if err := fixStdioPermissions(config, execUser); err != nil {
if err := fixStdioPermissions(execUser); err != nil {
return err
}
// This isn't allowed in an unprivileged user namespace since Linux 3.19.
// There's nothing we can do about /etc/group entries, so we silently
// ignore setting groups here (since the user didn't explicitly ask us to
// set the group).
if !config.Rootless {
suppGroups := append(execUser.Sgids, addGroups...)
if err := unix.Setgroups(suppGroups); err != nil {
return err
}
suppGroups := append(execUser.Sgids, addGroups...)
if err := syscall.Setgroups(suppGroups); err != nil {
return err
}
if err := system.Setgid(execUser.Gid); err != nil {
return err
}
if err := system.Setuid(execUser.Uid); err != nil {
return err
}
// if we didn't get HOME already, set it based on the user's HOME
if envHome := os.Getenv("HOME"); envHome == "" {
if err := os.Setenv("HOME", execUser.Home); err != nil {
@@ -315,9 +245,9 @@ func setupUser(config *initConfig) error {
// fixStdioPermissions fixes the permissions of PID 1's STDIO within the container to the specified user.
// The ownership needs to match because it is created outside of the container and needs to be
// localized.
func fixStdioPermissions(config *initConfig, u *user.ExecUser) error {
var null unix.Stat_t
if err := unix.Stat("/dev/null", &null); err != nil {
func fixStdioPermissions(u *user.ExecUser) error {
var null syscall.Stat_t
if err := syscall.Stat("/dev/null", &null); err != nil {
return err
}
for _, fd := range []uintptr{
@@ -325,31 +255,15 @@ func fixStdioPermissions(config *initConfig, u *user.ExecUser) error {
os.Stderr.Fd(),
os.Stdout.Fd(),
} {
var s unix.Stat_t
if err := unix.Fstat(int(fd), &s); err != nil {
var s syscall.Stat_t
if err := syscall.Fstat(int(fd), &s); err != nil {
return err
}
// Skip chown of /dev/null if it was used as one of the STDIO fds.
// skip chown of /dev/null if it was used as one of the STDIO fds.
if s.Rdev == null.Rdev {
continue
}
// Skip chown if s.Gid is actually an unmapped gid in the host. While
// this is a bit dodgy if it just so happens that the console _is_
// owned by overflow_gid, there's no way for us to disambiguate this as
// a userspace program.
if _, err := config.Config.HostGID(int(s.Gid)); err != nil {
continue
}
// We only change the uid owner (as it is possible for the mount to
// prefer a different gid, and there's no reason for us to change it).
// The reason why we don't just leave the default uid=X mount setup is
// that users expect to be able to actually use their console. Without
// this code, you couldn't effectively run as a non-root user inside a
// container and also have a console set up.
if err := unix.Fchown(int(fd), u.Uid, int(s.Gid)); err != nil {
if err := syscall.Fchown(int(fd), u.Uid, u.Gid); err != nil {
return err
}
}
@@ -404,58 +318,23 @@ func setupRoute(config *configs.Config) error {
func setupRlimits(limits []configs.Rlimit, pid int) error {
for _, rlimit := range limits {
if err := system.Prlimit(pid, rlimit.Type, unix.Rlimit{Max: rlimit.Hard, Cur: rlimit.Soft}); err != nil {
if err := system.Prlimit(pid, rlimit.Type, syscall.Rlimit{Max: rlimit.Hard, Cur: rlimit.Soft}); err != nil {
return fmt.Errorf("error setting rlimit type %v: %v", rlimit.Type, err)
}
}
return nil
}
const _P_PID = 1
func setOomScoreAdj(oomScoreAdj int, pid int) error {
path := fmt.Sprintf("/proc/%d/oom_score_adj", pid)
type siginfo struct {
si_signo int32
si_errno int32
si_code int32
// below here is a union; si_pid is the only field we use
si_pid int32
// Pad to 128 bytes as detailed in blockUntilWaitable
pad [96]byte
return ioutil.WriteFile(path, []byte(strconv.Itoa(oomScoreAdj)), 0600)
}
// isWaitable returns true if the process has exited false otherwise.
// Its based off blockUntilWaitable in src/os/wait_waitid.go
func isWaitable(pid int) (bool, error) {
si := &siginfo{}
_, _, e := unix.Syscall6(unix.SYS_WAITID, _P_PID, uintptr(pid), uintptr(unsafe.Pointer(si)), unix.WEXITED|unix.WNOWAIT|unix.WNOHANG, 0, 0)
if e != 0 {
return false, os.NewSyscallError("waitid", e)
}
return si.si_pid != 0, nil
}
// isNoChildren returns true if err represents a unix.ECHILD (formerly syscall.ECHILD) false otherwise
func isNoChildren(err error) bool {
switch err := err.(type) {
case syscall.Errno:
if err == unix.ECHILD {
return true
}
case *os.SyscallError:
if err.Err == unix.ECHILD {
return true
}
}
return false
}
// signalAllProcesses freezes then iterates over all the processes inside the
// manager's cgroups sending the signal s to them.
// If s is SIGKILL then it will wait for each process to exit.
// For all other signals it will check if the process is ready to report its
// exit status and only if it is will a wait be performed.
func signalAllProcesses(m cgroups.Manager, s os.Signal) error {
// killCgroupProcesses freezes then iterates over all the processes inside the
// manager's cgroups sending a SIGKILL to each process then waiting for them to
// exit.
func killCgroupProcesses(m cgroups.Manager) error {
var procs []*os.Process
if err := m.Freeze(configs.Frozen); err != nil {
logrus.Warn(err)
@@ -472,31 +351,16 @@ func signalAllProcesses(m cgroups.Manager, s os.Signal) error {
continue
}
procs = append(procs, p)
if err := p.Signal(s); err != nil {
if err := p.Kill(); err != nil {
logrus.Warn(err)
}
}
if err := m.Freeze(configs.Thawed); err != nil {
logrus.Warn(err)
}
for _, p := range procs {
if s != unix.SIGKILL {
if ok, err := isWaitable(p.Pid); err != nil {
if !isNoChildren(err) {
logrus.Warn("signalAllProcesses: ", p.Pid, err)
}
continue
} else if !ok {
// Not ready to report so don't wait
continue
}
}
if _, err := p.Wait(); err != nil {
if !isNoChildren(err) {
logrus.Warn("wait: ", err)
}
logrus.Warn(err)
}
}
return nil

View File

@@ -5,15 +5,13 @@ import (
"bytes"
"io/ioutil"
"os"
"os/exec"
"path/filepath"
"strings"
"syscall"
"testing"
"github.com/opencontainers/runc/libcontainer"
"github.com/opencontainers/runc/libcontainer/configs"
"golang.org/x/sys/unix"
)
func showFile(t *testing.T, fname string) error {
@@ -40,22 +38,7 @@ func showFile(t *testing.T, fname string) error {
return nil
}
func TestUsernsCheckpoint(t *testing.T) {
if _, err := os.Stat("/proc/self/ns/user"); os.IsNotExist(err) {
t.Skip("userns is unsupported")
}
cmd := exec.Command("criu", "check", "--feature", "userns")
if err := cmd.Run(); err != nil {
t.Skip("Unable to c/r a container with userns")
}
testCheckpoint(t, true)
}
func TestCheckpoint(t *testing.T) {
testCheckpoint(t, false)
}
func testCheckpoint(t *testing.T, userns bool) {
if testing.Short() {
return
}
@@ -76,15 +59,9 @@ func testCheckpoint(t *testing.T, userns bool) {
config.Mounts = append(config.Mounts, &configs.Mount{
Destination: "/sys/fs/cgroup",
Device: "cgroup",
Flags: defaultMountFlags | unix.MS_RDONLY,
Flags: defaultMountFlags | syscall.MS_RDONLY,
})
if userns {
config.UidMappings = []configs.IDMap{{HostID: 0, ContainerID: 0, Size: 1000}}
config.GidMappings = []configs.IDMap{{HostID: 0, ContainerID: 0, Size: 1000}}
config.Namespaces = append(config.Namespaces, configs.Namespace{Type: configs.NEWUSER})
}
factory, err := libcontainer.New(root, libcontainer.Cgroupfs)
if err != nil {
@@ -112,7 +89,7 @@ func testCheckpoint(t *testing.T, userns bool) {
Stdout: &stdout,
}
err = container.Run(&pconfig)
err = container.Start(&pconfig)
stdinR.Close()
defer stdinW.Close()
if err != nil {
@@ -129,33 +106,6 @@ func testCheckpoint(t *testing.T, userns bool) {
t.Fatal(err)
}
parentDir, err := ioutil.TempDir("", "criu-parent")
if err != nil {
t.Fatal(err)
}
defer os.RemoveAll(parentDir)
preDumpOpts := &libcontainer.CriuOpts{
ImagesDirectory: parentDir,
WorkDirectory: parentDir,
PreDump: true,
}
preDumpLog := filepath.Join(preDumpOpts.WorkDirectory, "dump.log")
if err := container.Checkpoint(preDumpOpts); err != nil {
showFile(t, preDumpLog)
t.Fatal(err)
}
state, err := container.Status()
if err != nil {
t.Fatal(err)
}
if state != libcontainer.Running {
t.Fatal("Unexpected preDump state: ", state)
}
imagesDir, err := ioutil.TempDir("", "criu")
if err != nil {
t.Fatal(err)
@@ -165,7 +115,6 @@ func testCheckpoint(t *testing.T, userns bool) {
checkpointOpts := &libcontainer.CriuOpts{
ImagesDirectory: imagesDir,
WorkDirectory: imagesDir,
ParentImage: "../criu-parent",
}
dumpLog := filepath.Join(checkpointOpts.WorkDirectory, "dump.log")
restoreLog := filepath.Join(checkpointOpts.WorkDirectory, "restore.log")
@@ -175,12 +124,12 @@ func testCheckpoint(t *testing.T, userns bool) {
t.Fatal(err)
}
state, err = container.Status()
state, err := container.Status()
if err != nil {
t.Fatal(err)
}
if state != libcontainer.Stopped {
if state != libcontainer.Running {
t.Fatal("Unexpected state checkpoint: ", state)
}

View File

@@ -2,7 +2,6 @@ package integration
import (
"bytes"
"encoding/json"
"fmt"
"io/ioutil"
"os"
@@ -11,13 +10,12 @@ import (
"reflect"
"strconv"
"strings"
"syscall"
"testing"
"github.com/opencontainers/runc/libcontainer"
"github.com/opencontainers/runc/libcontainer/cgroups/systemd"
"github.com/opencontainers/runc/libcontainer/configs"
"golang.org/x/sys/unix"
)
func TestExecPS(t *testing.T) {
@@ -40,12 +38,12 @@ func testExecPS(t *testing.T, userns bool) {
defer remove(rootfs)
config := newTemplateConfig(rootfs)
if userns {
config.UidMappings = []configs.IDMap{{HostID: 0, ContainerID: 0, Size: 1000}}
config.GidMappings = []configs.IDMap{{HostID: 0, ContainerID: 0, Size: 1000}}
config.UidMappings = []configs.IDMap{{0, 0, 1000}}
config.GidMappings = []configs.IDMap{{0, 0, 1000}}
config.Namespaces = append(config.Namespaces, configs.Namespace{Type: configs.NEWUSER})
}
buffers, exitCode, err := runContainer(config, "", "ps", "-o", "pid,user,comm")
buffers, exitCode, err := runContainer(config, "", "ps")
if err != nil {
t.Fatalf("%s: %s", buffers, err)
}
@@ -182,14 +180,14 @@ func testRlimit(t *testing.T, userns bool) {
config := newTemplateConfig(rootfs)
if userns {
config.UidMappings = []configs.IDMap{{HostID: 0, ContainerID: 0, Size: 1000}}
config.GidMappings = []configs.IDMap{{HostID: 0, ContainerID: 0, Size: 1000}}
config.UidMappings = []configs.IDMap{{0, 0, 1000}}
config.GidMappings = []configs.IDMap{{0, 0, 1000}}
config.Namespaces = append(config.Namespaces, configs.Namespace{Type: configs.NEWUSER})
}
// ensure limit is lower than what the config requests to test that in a user namespace
// the Setrlimit call happens early enough that we still have permissions to raise the limit.
ok(t, unix.Setrlimit(unix.RLIMIT_NOFILE, &unix.Rlimit{
ok(t, syscall.Setrlimit(syscall.RLIMIT_NOFILE, &syscall.Rlimit{
Max: 1024,
Cur: 1024,
}))
@@ -201,6 +199,17 @@ func testRlimit(t *testing.T, userns bool) {
}
}
func newTestRoot() (string, error) {
dir, err := ioutil.TempDir("", "libcontainer")
if err != nil {
return "", err
}
if err := os.MkdirAll(dir, 0700); err != nil {
return "", err
}
return dir, nil
}
func TestEnter(t *testing.T) {
if testing.Short() {
return
@@ -232,7 +241,7 @@ func TestEnter(t *testing.T) {
Stdin: stdinR,
Stdout: &stdout,
}
err = container.Run(&pconfig)
err = container.Start(&pconfig)
stdinR.Close()
defer stdinW.Close()
ok(t, err)
@@ -250,7 +259,7 @@ func TestEnter(t *testing.T) {
pconfig2.Stdin = stdinR2
pconfig2.Stdout = &stdout2
err = container.Run(&pconfig2)
err = container.Start(&pconfig2)
stdinR2.Close()
defer stdinW2.Close()
ok(t, err)
@@ -321,7 +330,7 @@ func TestProcessEnv(t *testing.T) {
Stdin: nil,
Stdout: &stdout,
}
err = container.Run(&pconfig)
err = container.Start(&pconfig)
ok(t, err)
// Wait for process
@@ -340,57 +349,6 @@ func TestProcessEnv(t *testing.T) {
}
}
func TestProcessEmptyCaps(t *testing.T) {
if testing.Short() {
return
}
root, err := newTestRoot()
ok(t, err)
defer os.RemoveAll(root)
rootfs, err := newRootfs()
ok(t, err)
defer remove(rootfs)
config := newTemplateConfig(rootfs)
config.Capabilities = nil
container, err := factory.Create("test", config)
ok(t, err)
defer container.Destroy()
var stdout bytes.Buffer
pconfig := libcontainer.Process{
Cwd: "/",
Args: []string{"sh", "-c", "cat /proc/self/status"},
Env: standardEnvironment,
Stdin: nil,
Stdout: &stdout,
}
err = container.Run(&pconfig)
ok(t, err)
// Wait for process
waitProcess(&pconfig, t)
outputStatus := string(stdout.Bytes())
lines := strings.Split(outputStatus, "\n")
effectiveCapsLine := ""
for _, l := range lines {
line := strings.TrimSpace(l)
if strings.Contains(line, "CapEff:") {
effectiveCapsLine = line
break
}
}
if effectiveCapsLine == "" {
t.Fatal("Couldn't find effective caps: ", outputStatus)
}
}
func TestProcessCaps(t *testing.T) {
if testing.Short() {
return
@@ -409,20 +367,18 @@ func TestProcessCaps(t *testing.T) {
ok(t, err)
defer container.Destroy()
processCaps := append(config.Capabilities, "CAP_NET_ADMIN")
var stdout bytes.Buffer
pconfig := libcontainer.Process{
Cwd: "/",
Args: []string{"sh", "-c", "cat /proc/self/status"},
Env: standardEnvironment,
Capabilities: processCaps,
Stdin: nil,
Stdout: &stdout,
Capabilities: &configs.Capabilities{},
}
pconfig.Capabilities.Bounding = append(config.Capabilities.Bounding, "CAP_NET_ADMIN")
pconfig.Capabilities.Permitted = append(config.Capabilities.Permitted, "CAP_NET_ADMIN")
pconfig.Capabilities.Effective = append(config.Capabilities.Effective, "CAP_NET_ADMIN")
pconfig.Capabilities.Inheritable = append(config.Capabilities.Inheritable, "CAP_NET_ADMIN")
err = container.Run(&pconfig)
err = container.Start(&pconfig)
ok(t, err)
// Wait for process
@@ -475,6 +431,7 @@ func TestAdditionalGroups(t *testing.T) {
defer remove(rootfs)
config := newTemplateConfig(rootfs)
config.AdditionalGroups = []string{"plugdev", "audio"}
factory, err := libcontainer.New(root, libcontainer.Cgroupfs)
ok(t, err)
@@ -485,14 +442,13 @@ func TestAdditionalGroups(t *testing.T) {
var stdout bytes.Buffer
pconfig := libcontainer.Process{
Cwd: "/",
Args: []string{"sh", "-c", "id", "-Gn"},
Env: standardEnvironment,
Stdin: nil,
Stdout: &stdout,
AdditionalGroups: []string{"plugdev", "audio"},
Cwd: "/",
Args: []string{"sh", "-c", "id", "-Gn"},
Env: standardEnvironment,
Stdin: nil,
Stdout: &stdout,
}
err = container.Run(&pconfig)
err = container.Start(&pconfig)
ok(t, err)
// Wait for process
@@ -552,7 +508,7 @@ func testFreeze(t *testing.T, systemd bool) {
Env: standardEnvironment,
Stdin: stdinR,
}
err = container.Run(pconfig)
err = container.Start(pconfig)
stdinR.Close()
defer stdinW.Close()
ok(t, err)
@@ -657,7 +613,7 @@ func testPids(t *testing.T, systemd bool) {
}
// Enforce a restrictive limit. 64 * /bin/true + 1 * shell should cause this
// to fail reliability.
// to fail reliabily.
config.Cgroups.Resources.PidsLimit = 64
out, _, err := runContainer(config, "", "/bin/sh", "-c", `
/bin/true | /bin/true | /bin/true | /bin/true | /bin/true | /bin/true | bin/true | /bin/true |
@@ -763,7 +719,7 @@ func TestContainerState(t *testing.T) {
Env: standardEnvironment,
Stdin: stdinR,
}
err = container.Run(p)
err = container.Start(p)
if err != nil {
t.Fatal(err)
}
@@ -807,13 +763,7 @@ func TestPassExtraFiles(t *testing.T) {
var stdout bytes.Buffer
pipeout1, pipein1, err := os.Pipe()
if err != nil {
t.Fatal(err)
}
pipeout2, pipein2, err := os.Pipe()
if err != nil {
t.Fatal(err)
}
process := libcontainer.Process{
Cwd: "/",
Args: []string{"sh", "-c", "cd /proc/$$/fd; echo -n *; echo -n 1 >3; echo -n 2 >4"},
@@ -822,7 +772,7 @@ func TestPassExtraFiles(t *testing.T) {
Stdin: nil,
Stdout: &stdout,
}
err = container.Run(&process)
err = container.Start(&process)
if err != nil {
t.Fatal(err)
}
@@ -881,7 +831,7 @@ func TestMountCmds(t *testing.T) {
Source: tmpDir,
Destination: "/tmp",
Device: "bind",
Flags: unix.MS_BIND | unix.MS_REC,
Flags: syscall.MS_BIND | syscall.MS_REC,
PremountCmds: []configs.Command{
{Path: "touch", Args: []string{filepath.Join(tmpDir, "hello")}},
{Path: "touch", Args: []string{filepath.Join(tmpDir, "world")}},
@@ -903,7 +853,7 @@ func TestMountCmds(t *testing.T) {
Args: []string{"sh", "-c", "env"},
Env: standardEnvironment,
}
err = container.Run(&pconfig)
err = container.Start(&pconfig)
if err != nil {
t.Fatal(err)
}
@@ -952,7 +902,7 @@ func TestSysctl(t *testing.T) {
Stdin: nil,
Stdout: &stdout,
}
err = container.Run(&pconfig)
err = container.Start(&pconfig)
ok(t, err)
// Wait for process
@@ -976,7 +926,7 @@ func TestMountCgroupRO(t *testing.T) {
config.Mounts = append(config.Mounts, &configs.Mount{
Destination: "/sys/fs/cgroup",
Device: "cgroup",
Flags: defaultMountFlags | unix.MS_RDONLY,
Flags: defaultMountFlags | syscall.MS_RDONLY,
})
buffers, exitCode, err := runContainer(config, "", "mount")
@@ -1092,7 +1042,7 @@ func TestOomScoreAdj(t *testing.T) {
Stdin: nil,
Stdout: &stdout,
}
err = container.Run(&pconfig)
err = container.Start(&pconfig)
ok(t, err)
// Wait for process
@@ -1109,44 +1059,25 @@ func TestHook(t *testing.T) {
if testing.Short() {
return
}
bundle, err := newTestBundle()
root, err := newTestRoot()
ok(t, err)
defer remove(bundle)
defer os.RemoveAll(root)
rootfs, err := newRootfs()
ok(t, err)
defer remove(rootfs)
config := newTemplateConfig(rootfs)
expectedBundle := bundle
config.Labels = append(config.Labels, fmt.Sprintf("bundle=%s", expectedBundle))
getRootfsFromBundle := func(bundle string) (string, error) {
f, err := os.Open(filepath.Join(bundle, "config.json"))
if err != nil {
return "", err
}
var config configs.Config
if err = json.NewDecoder(f).Decode(&config); err != nil {
return "", err
}
return config.Rootfs, nil
}
expectedBundlePath := "/path/to/bundle/path"
config.Labels = append(config.Labels, fmt.Sprintf("bundle=%s", expectedBundlePath))
config.Hooks = &configs.Hooks{
Prestart: []configs.Hook{
configs.NewFunctionHook(func(s configs.HookState) error {
if s.Bundle != expectedBundle {
t.Fatalf("Expected prestart hook bundlePath '%s'; got '%s'", expectedBundle, s.Bundle)
if s.BundlePath != expectedBundlePath {
t.Fatalf("Expected prestart hook bundlePath '%s'; got '%s'", expectedBundlePath, s.BundlePath)
}
root, err := getRootfsFromBundle(s.Bundle)
if err != nil {
return err
}
f, err := os.Create(filepath.Join(root, "test"))
f, err := os.Create(filepath.Join(s.Root, "test"))
if err != nil {
return err
}
@@ -1155,37 +1086,23 @@ func TestHook(t *testing.T) {
},
Poststart: []configs.Hook{
configs.NewFunctionHook(func(s configs.HookState) error {
if s.Bundle != expectedBundle {
t.Fatalf("Expected poststart hook bundlePath '%s'; got '%s'", expectedBundle, s.Bundle)
if s.BundlePath != expectedBundlePath {
t.Fatalf("Expected poststart hook bundlePath '%s'; got '%s'", expectedBundlePath, s.BundlePath)
}
root, err := getRootfsFromBundle(s.Bundle)
if err != nil {
return err
}
return ioutil.WriteFile(filepath.Join(root, "test"), []byte("hello world"), 0755)
return ioutil.WriteFile(filepath.Join(s.Root, "test"), []byte("hello world"), 0755)
}),
},
Poststop: []configs.Hook{
configs.NewFunctionHook(func(s configs.HookState) error {
if s.Bundle != expectedBundle {
t.Fatalf("Expected poststop hook bundlePath '%s'; got '%s'", expectedBundle, s.Bundle)
if s.BundlePath != expectedBundlePath {
t.Fatalf("Expected poststop hook bundlePath '%s'; got '%s'", expectedBundlePath, s.BundlePath)
}
root, err := getRootfsFromBundle(s.Bundle)
if err != nil {
return err
}
return os.RemoveAll(filepath.Join(root, "test"))
return os.RemoveAll(filepath.Join(s.Root, "test"))
}),
},
}
// write config of json format into config.json under bundle
f, err := os.OpenFile(filepath.Join(bundle, "config.json"), os.O_CREATE|os.O_RDWR, 0644)
ok(t, err)
ok(t, json.NewEncoder(f).Encode(config))
container, err := factory.Create("test", config)
ok(t, err)
@@ -1197,7 +1114,7 @@ func TestHook(t *testing.T) {
Stdin: nil,
Stdout: &stdout,
}
err = container.Run(&pconfig)
err = container.Start(&pconfig)
ok(t, err)
// Wait for process
@@ -1222,7 +1139,7 @@ func TestHook(t *testing.T) {
}
if err := container.Destroy(); err != nil {
t.Fatalf("container destroy %s", err)
t.Fatalf("container destory %s", err)
}
fi, err := os.Stat(filepath.Join(rootfs, "test"))
if err == nil || !os.IsNotExist(err) {
@@ -1251,7 +1168,7 @@ func TestSTDIOPermissions(t *testing.T) {
}
func unmountOp(path string) error {
if err := unix.Unmount(path, unix.MNT_DETACH); err != nil {
if err := syscall.Unmount(path, syscall.MNT_DETACH); err != nil {
return err
}
return nil
@@ -1276,7 +1193,7 @@ func TestRootfsPropagationSlaveMount(t *testing.T) {
defer remove(rootfs)
config := newTemplateConfig(rootfs)
config.RootPropagation = unix.MS_SLAVE | unix.MS_REC
config.RootPropagation = syscall.MS_SLAVE | syscall.MS_REC
// Bind mount a volume
dir1host, err := ioutil.TempDir("", "mnt1host")
@@ -1285,9 +1202,9 @@ func TestRootfsPropagationSlaveMount(t *testing.T) {
// Make this dir a "shared" mount point. This will make sure a
// slave relationship can be established in container.
err = unix.Mount(dir1host, dir1host, "bind", unix.MS_BIND|unix.MS_REC, "")
err = syscall.Mount(dir1host, dir1host, "bind", syscall.MS_BIND|syscall.MS_REC, "")
ok(t, err)
err = unix.Mount("", dir1host, "", unix.MS_SHARED|unix.MS_REC, "")
err = syscall.Mount("", dir1host, "", syscall.MS_SHARED|syscall.MS_REC, "")
ok(t, err)
defer unmountOp(dir1host)
@@ -1295,7 +1212,7 @@ func TestRootfsPropagationSlaveMount(t *testing.T) {
Source: dir1host,
Destination: dir1cont,
Device: "bind",
Flags: unix.MS_BIND | unix.MS_REC})
Flags: syscall.MS_BIND | syscall.MS_REC})
// TODO: systemd specific processing
f := factory
@@ -1314,7 +1231,7 @@ func TestRootfsPropagationSlaveMount(t *testing.T) {
Stdin: stdinR,
}
err = container.Run(pconfig)
err = container.Start(pconfig)
stdinR.Close()
defer stdinW.Close()
ok(t, err)
@@ -1325,7 +1242,7 @@ func TestRootfsPropagationSlaveMount(t *testing.T) {
ok(t, err)
defer os.RemoveAll(dir2host)
err = unix.Mount(dir2host, dir2host, "bind", unix.MS_BIND, "")
err = syscall.Mount(dir2host, dir2host, "bind", syscall.MS_BIND, "")
defer unmountOp(dir2host)
ok(t, err)
@@ -1343,7 +1260,7 @@ func TestRootfsPropagationSlaveMount(t *testing.T) {
Stdout: &stdout2,
}
err = container.Run(pconfig2)
err = container.Start(pconfig2)
stdinR2.Close()
defer stdinW2.Close()
ok(t, err)
@@ -1393,7 +1310,7 @@ func TestRootfsPropagationSharedMount(t *testing.T) {
ok(t, err)
defer remove(rootfs)
config := newTemplateConfig(rootfs)
config.RootPropagation = unix.MS_PRIVATE
config.RootPropagation = syscall.MS_PRIVATE
// Bind mount a volume
dir1host, err := ioutil.TempDir("", "mnt1host")
@@ -1402,9 +1319,9 @@ func TestRootfsPropagationSharedMount(t *testing.T) {
// Make this dir a "shared" mount point. This will make sure a
// shared relationship can be established in container.
err = unix.Mount(dir1host, dir1host, "bind", unix.MS_BIND|unix.MS_REC, "")
err = syscall.Mount(dir1host, dir1host, "bind", syscall.MS_BIND|syscall.MS_REC, "")
ok(t, err)
err = unix.Mount("", dir1host, "", unix.MS_SHARED|unix.MS_REC, "")
err = syscall.Mount("", dir1host, "", syscall.MS_SHARED|syscall.MS_REC, "")
ok(t, err)
defer unmountOp(dir1host)
@@ -1412,7 +1329,7 @@ func TestRootfsPropagationSharedMount(t *testing.T) {
Source: dir1host,
Destination: dir1cont,
Device: "bind",
Flags: unix.MS_BIND | unix.MS_REC})
Flags: syscall.MS_BIND | syscall.MS_REC})
// TODO: systemd specific processing
f := factory
@@ -1431,7 +1348,7 @@ func TestRootfsPropagationSharedMount(t *testing.T) {
Stdin: stdinR,
}
err = container.Run(pconfig)
err = container.Start(pconfig)
stdinR.Close()
defer stdinW.Close()
ok(t, err)
@@ -1451,22 +1368,19 @@ func TestRootfsPropagationSharedMount(t *testing.T) {
stdinR2, stdinW2, err := os.Pipe()
ok(t, err)
// Provide CAP_SYS_ADMIN
processCaps := append(config.Capabilities, "CAP_SYS_ADMIN")
pconfig2 := &libcontainer.Process{
Cwd: "/",
Args: []string{"mount", "--bind", dir2cont, dir2cont},
Env: standardEnvironment,
Stdin: stdinR2,
Stdout: &stdout2,
Capabilities: &configs.Capabilities{},
Capabilities: processCaps,
}
// Provide CAP_SYS_ADMIN
pconfig2.Capabilities.Bounding = append(config.Capabilities.Bounding, "CAP_SYS_ADMIN")
pconfig2.Capabilities.Permitted = append(config.Capabilities.Permitted, "CAP_SYS_ADMIN")
pconfig2.Capabilities.Effective = append(config.Capabilities.Effective, "CAP_SYS_ADMIN")
pconfig2.Capabilities.Inheritable = append(config.Capabilities.Inheritable, "CAP_SYS_ADMIN")
err = container.Run(pconfig2)
err = container.Start(pconfig2)
stdinR2.Close()
defer stdinW2.Close()
ok(t, err)
@@ -1538,7 +1452,7 @@ func TestInitJoinPID(t *testing.T) {
Env: standardEnvironment,
Stdin: stdinR1,
}
err = container1.Run(init1)
err = container1.Start(init1)
stdinR1.Close()
defer stdinW1.Close()
ok(t, err)
@@ -1548,7 +1462,7 @@ func TestInitJoinPID(t *testing.T) {
ok(t, err)
pidns1 := state1.NamespacePaths[configs.NEWPID]
// Run a container inside the existing pidns but with different cgroups
// Start a container inside the existing pidns but with different cgroups
config2 := newTemplateConfig(rootfs)
config2.Namespaces.Add(configs.NEWPID, pidns1)
config2.Cgroups.Path = "integration/test2"
@@ -1564,7 +1478,7 @@ func TestInitJoinPID(t *testing.T) {
Env: standardEnvironment,
Stdin: stdinR2,
}
err = container2.Run(init2)
err = container2.Start(init2)
stdinR2.Close()
defer stdinW2.Close()
ok(t, err)
@@ -1594,7 +1508,7 @@ func TestInitJoinPID(t *testing.T) {
Env: standardEnvironment,
Stdout: buffers.Stdout,
}
err = container1.Run(ps)
err = container1.Start(ps)
ok(t, err)
waitProcess(ps, t)
@@ -1628,8 +1542,8 @@ func TestInitJoinNetworkAndUser(t *testing.T) {
// Execute a long-running container
config1 := newTemplateConfig(rootfs)
config1.UidMappings = []configs.IDMap{{HostID: 0, ContainerID: 0, Size: 1000}}
config1.GidMappings = []configs.IDMap{{HostID: 0, ContainerID: 0, Size: 1000}}
config1.UidMappings = []configs.IDMap{{0, 0, 1000}}
config1.GidMappings = []configs.IDMap{{0, 0, 1000}}
config1.Namespaces = append(config1.Namespaces, configs.Namespace{Type: configs.NEWUSER})
container1, err := newContainer(config1)
ok(t, err)
@@ -1643,7 +1557,7 @@ func TestInitJoinNetworkAndUser(t *testing.T) {
Env: standardEnvironment,
Stdin: stdinR1,
}
err = container1.Run(init1)
err = container1.Start(init1)
stdinR1.Close()
defer stdinW1.Close()
ok(t, err)
@@ -1654,14 +1568,14 @@ func TestInitJoinNetworkAndUser(t *testing.T) {
netns1 := state1.NamespacePaths[configs.NEWNET]
userns1 := state1.NamespacePaths[configs.NEWUSER]
// Run a container inside the existing pidns but with different cgroups
// Start a container inside the existing pidns but with different cgroups
rootfs2, err := newRootfs()
ok(t, err)
defer remove(rootfs2)
config2 := newTemplateConfig(rootfs2)
config2.UidMappings = []configs.IDMap{{HostID: 0, ContainerID: 0, Size: 1000}}
config2.GidMappings = []configs.IDMap{{HostID: 0, ContainerID: 0, Size: 1000}}
config2.UidMappings = []configs.IDMap{{0, 0, 1000}}
config2.GidMappings = []configs.IDMap{{0, 0, 1000}}
config2.Namespaces.Add(configs.NEWNET, netns1)
config2.Namespaces.Add(configs.NEWUSER, userns1)
config2.Cgroups.Path = "integration/test2"
@@ -1677,7 +1591,7 @@ func TestInitJoinNetworkAndUser(t *testing.T) {
Env: standardEnvironment,
Stdin: stdinR2,
}
err = container2.Run(init2)
err = container2.Start(init2)
stdinR2.Close()
defer stdinW2.Close()
ok(t, err)
@@ -1708,52 +1622,3 @@ func TestInitJoinNetworkAndUser(t *testing.T) {
stdinW1.Close()
waitProcess(init1, t)
}
func TestTmpfsCopyUp(t *testing.T) {
if testing.Short() {
return
}
root, err := newTestRoot()
ok(t, err)
defer os.RemoveAll(root)
rootfs, err := newRootfs()
ok(t, err)
defer remove(rootfs)
config := newTemplateConfig(rootfs)
config.Mounts = append(config.Mounts, &configs.Mount{
Source: "tmpfs",
Destination: "/etc",
Device: "tmpfs",
Extensions: configs.EXT_COPYUP,
})
factory, err := libcontainer.New(root, libcontainer.Cgroupfs)
ok(t, err)
container, err := factory.Create("test", config)
ok(t, err)
defer container.Destroy()
var stdout bytes.Buffer
pconfig := libcontainer.Process{
Args: []string{"ls", "/etc/passwd"},
Env: standardEnvironment,
Stdin: nil,
Stdout: &stdout,
}
err = container.Run(&pconfig)
ok(t, err)
// Wait for process
waitProcess(&pconfig, t)
outputLs := string(stdout.Bytes())
// Check that the ls output has /etc/passwd
if !strings.Contains(outputLs, "/etc/passwd") {
t.Fatalf("/etc/passwd not copied up as expected: %v", outputLs)
}
}

View File

@@ -7,14 +7,12 @@ import (
"os"
"strconv"
"strings"
"syscall"
"testing"
"time"
"github.com/opencontainers/runc/libcontainer"
"github.com/opencontainers/runc/libcontainer/configs"
"github.com/opencontainers/runc/libcontainer/utils"
"golang.org/x/sys/unix"
)
func TestExecIn(t *testing.T) {
@@ -38,7 +36,7 @@ func TestExecIn(t *testing.T) {
Env: standardEnvironment,
Stdin: stdinR,
}
err = container.Run(process)
err = container.Start(process)
stdinR.Close()
defer stdinW.Close()
ok(t, err)
@@ -53,7 +51,7 @@ func TestExecIn(t *testing.T) {
Stderr: buffers.Stderr,
}
err = container.Run(ps)
err = container.Start(ps)
ok(t, err)
waitProcess(ps, t)
stdinW.Close()
@@ -63,9 +61,6 @@ func TestExecIn(t *testing.T) {
if !strings.Contains(out, "cat") || !strings.Contains(out, "ps") {
t.Fatalf("unexpected running process, output %q", out)
}
if strings.Contains(out, "\r") {
t.Fatalf("unexpected carriage-return in output")
}
}
func TestExecInUsernsRlimit(t *testing.T) {
@@ -91,8 +86,8 @@ func testExecInRlimit(t *testing.T, userns bool) {
config := newTemplateConfig(rootfs)
if userns {
config.UidMappings = []configs.IDMap{{HostID: 0, ContainerID: 0, Size: 1000}}
config.GidMappings = []configs.IDMap{{HostID: 0, ContainerID: 0, Size: 1000}}
config.UidMappings = []configs.IDMap{{0, 0, 1000}}
config.GidMappings = []configs.IDMap{{0, 0, 1000}}
config.Namespaces = append(config.Namespaces, configs.Namespace{Type: configs.NEWUSER})
}
@@ -108,7 +103,7 @@ func testExecInRlimit(t *testing.T, userns bool) {
Env: standardEnvironment,
Stdin: stdinR,
}
err = container.Run(process)
err = container.Start(process)
stdinR.Close()
defer stdinW.Close()
ok(t, err)
@@ -123,10 +118,10 @@ func testExecInRlimit(t *testing.T, userns bool) {
Stderr: buffers.Stderr,
Rlimits: []configs.Rlimit{
// increase process rlimit higher than container rlimit to test per-process limit
{Type: unix.RLIMIT_NOFILE, Hard: 1026, Soft: 1026},
{Type: syscall.RLIMIT_NOFILE, Hard: 1026, Soft: 1026},
},
}
err = container.Run(ps)
err = container.Start(ps)
ok(t, err)
waitProcess(ps, t)
@@ -139,64 +134,6 @@ func testExecInRlimit(t *testing.T, userns bool) {
}
}
func TestExecInAdditionalGroups(t *testing.T) {
if testing.Short() {
return
}
rootfs, err := newRootfs()
ok(t, err)
defer remove(rootfs)
config := newTemplateConfig(rootfs)
container, err := newContainer(config)
ok(t, err)
defer container.Destroy()
// Execute a first process in the container
stdinR, stdinW, err := os.Pipe()
ok(t, err)
process := &libcontainer.Process{
Cwd: "/",
Args: []string{"cat"},
Env: standardEnvironment,
Stdin: stdinR,
}
err = container.Run(process)
stdinR.Close()
defer stdinW.Close()
ok(t, err)
var stdout bytes.Buffer
pconfig := libcontainer.Process{
Cwd: "/",
Args: []string{"sh", "-c", "id", "-Gn"},
Env: standardEnvironment,
Stdin: nil,
Stdout: &stdout,
AdditionalGroups: []string{"plugdev", "audio"},
}
err = container.Run(&pconfig)
ok(t, err)
// Wait for process
waitProcess(&pconfig, t)
stdinW.Close()
waitProcess(process, t)
outputGroups := string(stdout.Bytes())
// Check that the groups output has the groups that we specified
if !strings.Contains(outputGroups, "audio") {
t.Fatalf("Listed groups do not contain the audio group as expected: %v", outputGroups)
}
if !strings.Contains(outputGroups, "plugdev") {
t.Fatalf("Listed groups do not contain the plugdev group as expected: %v", outputGroups)
}
}
func TestExecInError(t *testing.T) {
if testing.Short() {
return
@@ -218,7 +155,7 @@ func TestExecInError(t *testing.T) {
Env: standardEnvironment,
Stdin: stdinR,
}
err = container.Run(process)
err = container.Start(process)
stdinR.Close()
defer func() {
stdinW.Close()
@@ -234,9 +171,9 @@ func TestExecInError(t *testing.T) {
Cwd: "/",
Args: []string{"unexistent"},
Env: standardEnvironment,
Stderr: &out,
Stdout: &out,
}
err = container.Run(unexistent)
err = container.Start(unexistent)
if err == nil {
t.Fatal("Should be an error")
}
@@ -270,7 +207,7 @@ func TestExecInTTY(t *testing.T) {
Env: standardEnvironment,
Stdin: stdinR,
}
err = container.Run(process)
err = container.Start(process)
stdinR.Close()
defer stdinW.Close()
ok(t, err)
@@ -281,43 +218,15 @@ func TestExecInTTY(t *testing.T) {
Args: []string{"ps"},
Env: standardEnvironment,
}
parent, child, err := utils.NewSockPair("console")
if err != nil {
ok(t, err)
}
defer parent.Close()
defer child.Close()
ps.ConsoleSocket = child
type cdata struct {
c libcontainer.Console
err error
}
dc := make(chan *cdata, 1)
go func() {
f, err := utils.RecvFd(parent)
if err != nil {
dc <- &cdata{
err: err,
}
}
libcontainer.SaneTerminal(f)
dc <- &cdata{
c: libcontainer.ConsoleFromFile(f),
}
}()
err = container.Run(ps)
ok(t, err)
data := <-dc
if data.err != nil {
ok(t, data.err)
}
console := data.c
console, err := ps.NewConsole(0)
copy := make(chan struct{})
go func() {
io.Copy(&stdout, console)
close(copy)
}()
ok(t, err)
err = container.Start(ps)
ok(t, err)
select {
case <-time.After(5 * time.Second):
t.Fatal("Waiting for copy timed out")
@@ -329,12 +238,9 @@ func TestExecInTTY(t *testing.T) {
waitProcess(process, t)
out := stdout.String()
if !strings.Contains(out, "cat") || !strings.Contains(out, "ps") {
if !strings.Contains(out, "cat") || !strings.Contains(string(out), "ps") {
t.Fatalf("unexpected running process, output %q", out)
}
if strings.Contains(out, "\r") {
t.Fatalf("unexpected carriage-return in output")
}
}
func TestExecInEnvironment(t *testing.T) {
@@ -358,7 +264,7 @@ func TestExecInEnvironment(t *testing.T) {
Env: standardEnvironment,
Stdin: stdinR,
}
err = container.Run(process)
err = container.Start(process)
stdinR.Close()
defer stdinW.Close()
ok(t, err)
@@ -377,7 +283,7 @@ func TestExecInEnvironment(t *testing.T) {
Stdout: buffers.Stdout,
Stderr: buffers.Stderr,
}
err = container.Run(process2)
err = container.Start(process2)
ok(t, err)
waitProcess(process2, t)
@@ -422,7 +328,7 @@ func TestExecinPassExtraFiles(t *testing.T) {
Env: standardEnvironment,
Stdin: stdinR,
}
err = container.Run(process)
err = container.Start(process)
stdinR.Close()
defer stdinW.Close()
if err != nil {
@@ -431,13 +337,7 @@ func TestExecinPassExtraFiles(t *testing.T) {
var stdout bytes.Buffer
pipeout1, pipein1, err := os.Pipe()
if err != nil {
t.Fatal(err)
}
pipeout2, pipein2, err := os.Pipe()
if err != nil {
t.Fatal(err)
}
inprocess := &libcontainer.Process{
Cwd: "/",
Args: []string{"sh", "-c", "cd /proc/$$/fd; echo -n *; echo -n 1 >3; echo -n 2 >4"},
@@ -446,7 +346,7 @@ func TestExecinPassExtraFiles(t *testing.T) {
Stdin: nil,
Stdout: &stdout,
}
err = container.Run(inprocess)
err = container.Start(inprocess)
if err != nil {
t.Fatal(err)
}
@@ -501,7 +401,7 @@ func TestExecInOomScoreAdj(t *testing.T) {
Env: standardEnvironment,
Stdin: stdinR,
}
err = container.Run(process)
err = container.Start(process)
stdinR.Close()
defer stdinW.Close()
ok(t, err)
@@ -515,7 +415,7 @@ func TestExecInOomScoreAdj(t *testing.T) {
Stdout: buffers.Stdout,
Stderr: buffers.Stderr,
}
err = container.Run(ps)
err = container.Start(ps)
ok(t, err)
waitProcess(ps, t)
@@ -539,8 +439,8 @@ func TestExecInUserns(t *testing.T) {
ok(t, err)
defer remove(rootfs)
config := newTemplateConfig(rootfs)
config.UidMappings = []configs.IDMap{{HostID: 0, ContainerID: 0, Size: 1000}}
config.GidMappings = []configs.IDMap{{HostID: 0, ContainerID: 0, Size: 1000}}
config.UidMappings = []configs.IDMap{{0, 0, 1000}}
config.GidMappings = []configs.IDMap{{0, 0, 1000}}
config.Namespaces = append(config.Namespaces, configs.Namespace{Type: configs.NEWUSER})
container, err := newContainer(config)
ok(t, err)
@@ -556,7 +456,7 @@ func TestExecInUserns(t *testing.T) {
Env: standardEnvironment,
Stdin: stdinR,
}
err = container.Run(process)
err = container.Start(process)
stdinR.Close()
defer stdinW.Close()
ok(t, err)
@@ -576,7 +476,7 @@ func TestExecInUserns(t *testing.T) {
Stdout: buffers.Stdout,
Stderr: os.Stderr,
}
err = container.Run(process2)
err = container.Start(process2)
ok(t, err)
waitProcess(process2, t)
stdinW.Close()

View File

@@ -5,11 +5,10 @@ import (
"runtime"
"testing"
"github.com/Sirupsen/logrus"
"github.com/opencontainers/runc/libcontainer"
"github.com/opencontainers/runc/libcontainer/cgroups/systemd"
_ "github.com/opencontainers/runc/libcontainer/nsenter"
"github.com/sirupsen/logrus"
)
// init runs the libcontainer initialization code because of the busybox style needs
@@ -37,19 +36,19 @@ var (
func TestMain(m *testing.M) {
var (
err error
ret int
ret int = 0
)
logrus.SetOutput(os.Stderr)
logrus.SetLevel(logrus.InfoLevel)
factory, err = libcontainer.New("/run/libctTests", libcontainer.Cgroupfs)
factory, err = libcontainer.New(".", libcontainer.Cgroupfs)
if err != nil {
logrus.Error(err)
os.Exit(1)
}
if systemd.UseSystemd() {
systemdFactory, err = libcontainer.New("/run/libctTests", libcontainer.SystemdCgroups)
systemdFactory, err = libcontainer.New(".", libcontainer.SystemdCgroups)
if err != nil {
logrus.Error(err)
os.Exit(1)

View File

@@ -50,7 +50,7 @@ func TestSeccompDenyGetcwd(t *testing.T) {
Stderr: buffers.Stderr,
}
err = container.Run(pwd)
err = container.Start(pwd)
if err != nil {
t.Fatal(err)
}
@@ -101,8 +101,8 @@ func TestSeccompPermitWriteConditional(t *testing.T) {
Args: []*configs.Arg{
{
Index: 0,
Value: 2,
Op: configs.EqualTo,
Value: 1,
Op: configs.GreaterThan,
},
},
},
@@ -125,7 +125,7 @@ func TestSeccompPermitWriteConditional(t *testing.T) {
Stderr: buffers.Stderr,
}
err = container.Run(dmesg)
err = container.Start(dmesg)
if err != nil {
t.Fatal(err)
}
@@ -162,8 +162,8 @@ func TestSeccompDenyWriteConditional(t *testing.T) {
Args: []*configs.Arg{
{
Index: 0,
Value: 2,
Op: configs.EqualTo,
Value: 1,
Op: configs.GreaterThan,
},
},
},
@@ -186,7 +186,7 @@ func TestSeccompDenyWriteConditional(t *testing.T) {
Stderr: buffers.Stderr,
}
err = container.Run(dmesg)
err = container.Start(dmesg)
if err != nil {
t.Fatal(err)
}

View File

@@ -1,9 +1,9 @@
package integration
import (
"github.com/opencontainers/runc/libcontainer/configs"
"syscall"
"golang.org/x/sys/unix"
"github.com/opencontainers/runc/libcontainer/configs"
)
var standardEnvironment = []string{
@@ -13,97 +13,30 @@ var standardEnvironment = []string{
"TERM=xterm",
}
const defaultMountFlags = unix.MS_NOEXEC | unix.MS_NOSUID | unix.MS_NODEV
const defaultMountFlags = syscall.MS_NOEXEC | syscall.MS_NOSUID | syscall.MS_NODEV
// newTemplateConfig returns a base template for running a container
//
// it uses a network strategy of just setting a loopback interface
// and the default setup for devices
func newTemplateConfig(rootfs string) *configs.Config {
allowAllDevices := false
return &configs.Config{
Rootfs: rootfs,
Capabilities: &configs.Capabilities{
Bounding: []string{
"CAP_CHOWN",
"CAP_DAC_OVERRIDE",
"CAP_FSETID",
"CAP_FOWNER",
"CAP_MKNOD",
"CAP_NET_RAW",
"CAP_SETGID",
"CAP_SETUID",
"CAP_SETFCAP",
"CAP_SETPCAP",
"CAP_NET_BIND_SERVICE",
"CAP_SYS_CHROOT",
"CAP_KILL",
"CAP_AUDIT_WRITE",
},
Permitted: []string{
"CAP_CHOWN",
"CAP_DAC_OVERRIDE",
"CAP_FSETID",
"CAP_FOWNER",
"CAP_MKNOD",
"CAP_NET_RAW",
"CAP_SETGID",
"CAP_SETUID",
"CAP_SETFCAP",
"CAP_SETPCAP",
"CAP_NET_BIND_SERVICE",
"CAP_SYS_CHROOT",
"CAP_KILL",
"CAP_AUDIT_WRITE",
},
Inheritable: []string{
"CAP_CHOWN",
"CAP_DAC_OVERRIDE",
"CAP_FSETID",
"CAP_FOWNER",
"CAP_MKNOD",
"CAP_NET_RAW",
"CAP_SETGID",
"CAP_SETUID",
"CAP_SETFCAP",
"CAP_SETPCAP",
"CAP_NET_BIND_SERVICE",
"CAP_SYS_CHROOT",
"CAP_KILL",
"CAP_AUDIT_WRITE",
},
Ambient: []string{
"CAP_CHOWN",
"CAP_DAC_OVERRIDE",
"CAP_FSETID",
"CAP_FOWNER",
"CAP_MKNOD",
"CAP_NET_RAW",
"CAP_SETGID",
"CAP_SETUID",
"CAP_SETFCAP",
"CAP_SETPCAP",
"CAP_NET_BIND_SERVICE",
"CAP_SYS_CHROOT",
"CAP_KILL",
"CAP_AUDIT_WRITE",
},
Effective: []string{
"CAP_CHOWN",
"CAP_DAC_OVERRIDE",
"CAP_FSETID",
"CAP_FOWNER",
"CAP_MKNOD",
"CAP_NET_RAW",
"CAP_SETGID",
"CAP_SETUID",
"CAP_SETFCAP",
"CAP_SETPCAP",
"CAP_NET_BIND_SERVICE",
"CAP_SYS_CHROOT",
"CAP_KILL",
"CAP_AUDIT_WRITE",
},
Capabilities: []string{
"CAP_CHOWN",
"CAP_DAC_OVERRIDE",
"CAP_FSETID",
"CAP_FOWNER",
"CAP_MKNOD",
"CAP_NET_RAW",
"CAP_SETGID",
"CAP_SETUID",
"CAP_SETFCAP",
"CAP_SETPCAP",
"CAP_NET_BIND_SERVICE",
"CAP_SYS_CHROOT",
"CAP_KILL",
"CAP_AUDIT_WRITE",
},
Namespaces: configs.Namespaces([]configs.Namespace{
{Type: configs.NEWNS},
@@ -116,13 +49,12 @@ func newTemplateConfig(rootfs string) *configs.Config {
Path: "integration/test",
Resources: &configs.Resources{
MemorySwappiness: nil,
AllowAllDevices: &allowAllDevices,
AllowAllDevices: false,
AllowedDevices: configs.DefaultAllowedDevices,
},
},
MaskPaths: []string{
"/proc/kcore",
"/sys/firmware",
},
ReadonlyPaths: []string{
"/proc/sys", "/proc/sysrq-trigger", "/proc/irq", "/proc/bus",
@@ -140,14 +72,14 @@ func newTemplateConfig(rootfs string) *configs.Config {
Source: "tmpfs",
Destination: "/dev",
Device: "tmpfs",
Flags: unix.MS_NOSUID | unix.MS_STRICTATIME,
Flags: syscall.MS_NOSUID | syscall.MS_STRICTATIME,
Data: "mode=755",
},
{
Source: "devpts",
Destination: "/dev/pts",
Device: "devpts",
Flags: unix.MS_NOSUID | unix.MS_NOEXEC,
Flags: syscall.MS_NOSUID | syscall.MS_NOEXEC,
Data: "newinstance,ptmxmode=0666,mode=0620,gid=5",
},
{
@@ -157,20 +89,17 @@ func newTemplateConfig(rootfs string) *configs.Config {
Data: "mode=1777,size=65536k",
Flags: defaultMountFlags,
},
/*
CI is broken on the debian based kernels with this
{
Source: "mqueue",
Destination: "/dev/mqueue",
Device: "mqueue",
Flags: defaultMountFlags,
},
*/
{
Source: "mqueue",
Destination: "/dev/mqueue",
Device: "mqueue",
Flags: defaultMountFlags,
},
{
Source: "sysfs",
Destination: "/sys",
Device: "sysfs",
Flags: defaultMountFlags | unix.MS_RDONLY,
Flags: defaultMountFlags | syscall.MS_RDONLY,
},
},
Networks: []*configs.Network{
@@ -182,7 +111,7 @@ func newTemplateConfig(rootfs string) *configs.Config {
},
Rlimits: []configs.Rlimit{
{
Type: unix.RLIMIT_NOFILE,
Type: syscall.RLIMIT_NOFILE,
Hard: uint64(1025),
Soft: uint64(1025),
},

View File

@@ -2,8 +2,6 @@ package integration
import (
"bytes"
"crypto/md5"
"encoding/hex"
"fmt"
"io/ioutil"
"os"
@@ -13,7 +11,6 @@ import (
"strings"
"syscall"
"testing"
"time"
"github.com/opencontainers/runc/libcontainer"
"github.com/opencontainers/runc/libcontainer/configs"
@@ -65,28 +62,6 @@ func waitProcess(p *libcontainer.Process, t *testing.T) {
}
}
func newTestRoot() (string, error) {
dir, err := ioutil.TempDir("", "libcontainer")
if err != nil {
return "", err
}
if err := os.MkdirAll(dir, 0700); err != nil {
return "", err
}
return dir, nil
}
func newTestBundle() (string, error) {
dir, err := ioutil.TempDir("", "bundle")
if err != nil {
return "", err
}
if err := os.MkdirAll(dir, 0700); err != nil {
return "", err
}
return dir, nil
}
// newRootfs creates a new tmp directory and copies the busybox root filesystem
func newRootfs() (string, error) {
dir, err := ioutil.TempDir("", "")
@@ -117,9 +92,7 @@ func copyBusybox(dest string) error {
}
func newContainer(config *configs.Config) (libcontainer.Container, error) {
h := md5.New()
h.Write([]byte(time.Now().String()))
return newContainerWithName(hex.EncodeToString(h.Sum(nil)), config)
return newContainerWithName("testCT", config)
}
func newContainerWithName(name string, config *configs.Config) (libcontainer.Container, error) {
@@ -150,7 +123,7 @@ func runContainer(config *configs.Config, console string, args ...string) (buffe
Stderr: buffers.Stderr,
}
err = container.Run(process)
err = container.Start(process)
if err != nil {
return buffers, -1, err
}

View File

@@ -1,31 +1,47 @@
// +build linux
package keys
package keyctl
import (
"fmt"
"strconv"
"syscall"
"strings"
"golang.org/x/sys/unix"
"strconv"
"unsafe"
)
const KEYCTL_JOIN_SESSION_KEYRING = 1
const KEYCTL_SETPERM = 5
const KEYCTL_DESCRIBE = 6
type KeySerial uint32
func JoinSessionKeyring(name string) (KeySerial, error) {
sessKeyId, err := unix.KeyctlJoinSessionKeyring(name)
if err != nil {
return 0, fmt.Errorf("could not create session key: %v", err)
var _name *byte = nil
var err error
if len(name) > 0 {
_name, err = syscall.BytePtrFromString(name)
if err != nil {
return KeySerial(0), err
}
}
sessKeyId, _, errn := syscall.Syscall(syscall.SYS_KEYCTL, KEYCTL_JOIN_SESSION_KEYRING, uintptr(unsafe.Pointer(_name)), 0)
if errn != 0 {
return 0, fmt.Errorf("could not create session key: %v", errn)
}
return KeySerial(sessKeyId), nil
}
// ModKeyringPerm modifies permissions on a keyring by reading the current permissions,
// modify permissions on a keyring by reading the current permissions,
// anding the bits with the given mask (clearing permissions) and setting
// additional permission bits
func ModKeyringPerm(ringId KeySerial, mask, setbits uint32) error {
dest, err := unix.KeyctlString(unix.KEYCTL_DESCRIBE, int(ringId))
if err != nil {
dest := make([]byte, 1024)
destBytes := unsafe.Pointer(&dest[0])
if _, _, err := syscall.Syscall6(syscall.SYS_KEYCTL, uintptr(KEYCTL_DESCRIBE), uintptr(ringId), uintptr(destBytes), uintptr(len(dest)), 0, 0); err != 0 {
return err
}
@@ -42,9 +58,10 @@ func ModKeyringPerm(ringId KeySerial, mask, setbits uint32) error {
perm := (uint32(perm64) & mask) | setbits
if err := unix.KeyctlSetperm(int(ringId), perm); err != nil {
if _, _, err := syscall.Syscall(syscall.SYS_KEYCTL, uintptr(KEYCTL_SETPERM), uintptr(ringId), uintptr(perm)); err != 0 {
return err
}
return nil
}

View File

@@ -0,0 +1,80 @@
// +build !selinux !linux
package label
// InitLabels returns the process label and file labels to be used within
// the container. A list of options can be passed into this function to alter
// the labels.
func InitLabels(options []string) (string, string, error) {
return "", "", nil
}
func GenLabels(options string) (string, string, error) {
return "", "", nil
}
func FormatMountLabel(src string, mountLabel string) string {
return src
}
func SetProcessLabel(processLabel string) error {
return nil
}
func GetFileLabel(path string) (string, error) {
return "", nil
}
func SetFileLabel(path string, fileLabel string) error {
return nil
}
func SetFileCreateLabel(fileLabel string) error {
return nil
}
func Relabel(path string, fileLabel string, shared bool) error {
return nil
}
func GetPidLabel(pid int) (string, error) {
return "", nil
}
func Init() {
}
func ReserveLabel(label string) error {
return nil
}
func UnreserveLabel(label string) error {
return nil
}
// DupSecOpt takes a process label and returns security options that
// can be used to set duplicate labels on future container processes
func DupSecOpt(src string) []string {
return nil
}
// DisableSecOpt returns a security opt that can disable labeling
// support for future container processes
func DisableSecOpt() []string {
return nil
}
// Validate checks that the label does not include unexpected options
func Validate(label string) error {
return nil
}
// RelabelNeeded checks whether the user requested a relabel
func RelabelNeeded(label string) bool {
return false
}
// IsShared checks that the label includes a "shared" mark
func IsShared(label string) bool {
return false
}

View File

@@ -0,0 +1,197 @@
// +build selinux,linux
package label
import (
"fmt"
"strings"
"github.com/opencontainers/runc/libcontainer/selinux"
)
// Valid Label Options
var validOptions = map[string]bool{
"disable": true,
"type": true,
"user": true,
"role": true,
"level": true,
}
var ErrIncompatibleLabel = fmt.Errorf("Bad SELinux option z and Z can not be used together")
// InitLabels returns the process label and file labels to be used within
// the container. A list of options can be passed into this function to alter
// the labels. The labels returned will include a random MCS String, that is
// guaranteed to be unique.
func InitLabels(options []string) (string, string, error) {
if !selinux.SelinuxEnabled() {
return "", "", nil
}
processLabel, mountLabel := selinux.GetLxcContexts()
if processLabel != "" {
pcon := selinux.NewContext(processLabel)
mcon := selinux.NewContext(mountLabel)
for _, opt := range options {
if opt == "disable" {
return "", "", nil
}
if i := strings.Index(opt, ":"); i == -1 {
return "", "", fmt.Errorf("Bad label option %q, valid options 'disable' or \n'user, role, level, type' followed by ':' and a value", opt)
}
con := strings.SplitN(opt, ":", 2)
if !validOptions[con[0]] {
return "", "", fmt.Errorf("Bad label option %q, valid options 'disable, user, role, level, type'", con[0])
}
pcon[con[0]] = con[1]
if con[0] == "level" || con[0] == "user" {
mcon[con[0]] = con[1]
}
}
processLabel = pcon.Get()
mountLabel = mcon.Get()
}
return processLabel, mountLabel, nil
}
// DEPRECATED: The GenLabels function is only to be used during the transition to the official API.
func GenLabels(options string) (string, string, error) {
return InitLabels(strings.Fields(options))
}
// FormatMountLabel returns a string to be used by the mount command.
// The format of this string will be used to alter the labeling of the mountpoint.
// The string returned is suitable to be used as the options field of the mount command.
// If you need to have additional mount point options, you can pass them in as
// the first parameter. Second parameter is the label that you wish to apply
// to all content in the mount point.
func FormatMountLabel(src, mountLabel string) string {
if mountLabel != "" {
switch src {
case "":
src = fmt.Sprintf("context=%q", mountLabel)
default:
src = fmt.Sprintf("%s,context=%q", src, mountLabel)
}
}
return src
}
// SetProcessLabel takes a process label and tells the kernel to assign the
// label to the next program executed by the current process.
func SetProcessLabel(processLabel string) error {
if processLabel == "" {
return nil
}
return selinux.Setexeccon(processLabel)
}
// GetProcessLabel returns the process label that the kernel will assign
// to the next program executed by the current process. If "" is returned
// this indicates that the default labeling will happen for the process.
func GetProcessLabel() (string, error) {
return selinux.Getexeccon()
}
// GetFileLabel returns the label for specified path
func GetFileLabel(path string) (string, error) {
return selinux.Getfilecon(path)
}
// SetFileLabel modifies the "path" label to the specified file label
func SetFileLabel(path string, fileLabel string) error {
if selinux.SelinuxEnabled() && fileLabel != "" {
return selinux.Setfilecon(path, fileLabel)
}
return nil
}
// Tell the kernel the label for all files to be created
func SetFileCreateLabel(fileLabel string) error {
if selinux.SelinuxEnabled() {
return selinux.Setfscreatecon(fileLabel)
}
return nil
}
// Change the label of path to the filelabel string.
// It changes the MCS label to s0 if shared is true.
// This will allow all containers to share the content.
func Relabel(path string, fileLabel string, shared bool) error {
if !selinux.SelinuxEnabled() {
return nil
}
if fileLabel == "" {
return nil
}
exclude_paths := map[string]bool{"/": true, "/usr": true, "/etc": true}
if exclude_paths[path] {
return fmt.Errorf("Relabeling of %s is not allowed", path)
}
if shared {
c := selinux.NewContext(fileLabel)
c["level"] = "s0"
fileLabel = c.Get()
}
return selinux.Chcon(path, fileLabel, true)
}
// GetPidLabel will return the label of the process running with the specified pid
func GetPidLabel(pid int) (string, error) {
return selinux.Getpidcon(pid)
}
// Init initialises the labeling system
func Init() {
selinux.SelinuxEnabled()
}
// ReserveLabel will record the fact that the MCS label has already been used.
// This will prevent InitLabels from using the MCS label in a newly created
// container
func ReserveLabel(label string) error {
selinux.ReserveLabel(label)
return nil
}
// UnreserveLabel will remove the reservation of the MCS label.
// This will allow InitLabels to use the MCS label in a newly created
// containers
func UnreserveLabel(label string) error {
selinux.FreeLxcContexts(label)
return nil
}
// DupSecOpt takes an process label and returns security options that
// can be used to set duplicate labels on future container processes
func DupSecOpt(src string) []string {
return selinux.DupSecOpt(src)
}
// DisableSecOpt returns a security opt that can disable labeling
// support for future container processes
func DisableSecOpt() []string {
return selinux.DisableSecOpt()
}
// Validate checks that the label does not include unexpected options
func Validate(label string) error {
if strings.Contains(label, "z") && strings.Contains(label, "Z") {
return ErrIncompatibleLabel
}
return nil
}
// RelabelNeeded checks whether the user requested a relabel
func RelabelNeeded(label string) bool {
return strings.Contains(label, "z") || strings.Contains(label, "Z")
}
// IsShared checks that the label includes a "shared" mark
func IsShared(label string) bool {
return strings.Contains(label, "z")
}

View File

@@ -0,0 +1,143 @@
// +build selinux,linux
package label
import (
"os"
"strings"
"testing"
"github.com/opencontainers/runc/libcontainer/selinux"
)
func TestInit(t *testing.T) {
if selinux.SelinuxEnabled() {
var testNull []string
plabel, mlabel, err := InitLabels(testNull)
if err != nil {
t.Log("InitLabels Failed")
t.Fatal(err)
}
testDisabled := []string{"disable"}
plabel, mlabel, err = InitLabels(testDisabled)
if err != nil {
t.Log("InitLabels Disabled Failed")
t.Fatal(err)
}
if plabel != "" {
t.Log("InitLabels Disabled Failed")
t.Fatal()
}
testUser := []string{"user:user_u", "role:user_r", "type:user_t", "level:s0:c1,c15"}
plabel, mlabel, err = InitLabels(testUser)
if err != nil {
t.Log("InitLabels User Failed")
t.Fatal(err)
}
if plabel != "user_u:user_r:user_t:s0:c1,c15" || mlabel != "user_u:object_r:svirt_sandbox_file_t:s0:c1,c15" {
t.Log("InitLabels User Match Failed")
t.Log(plabel, mlabel)
t.Fatal(err)
}
testBadData := []string{"user", "role:user_r", "type:user_t", "level:s0:c1,c15"}
if _, _, err = InitLabels(testBadData); err == nil {
t.Log("InitLabels Bad Failed")
t.Fatal(err)
}
}
}
func TestDuplicateLabel(t *testing.T) {
secopt := DupSecOpt("system_u:system_r:svirt_lxc_net_t:s0:c1,c2")
t.Log(secopt)
for _, opt := range secopt {
con := strings.SplitN(opt, ":", 3)
if len(con) != 3 || con[0] != "label" {
t.Errorf("Invalid DupSecOpt return value")
continue
}
if con[1] == "user" {
if con[2] != "system_u" {
t.Errorf("DupSecOpt Failed user incorrect")
}
continue
}
if con[1] == "role" {
if con[2] != "system_r" {
t.Errorf("DupSecOpt Failed role incorrect")
}
continue
}
if con[1] == "type" {
if con[2] != "svirt_lxc_net_t" {
t.Errorf("DupSecOpt Failed type incorrect")
}
continue
}
if con[1] == "level" {
if con[2] != "s0:c1,c2" {
t.Errorf("DupSecOpt Failed level incorrect")
}
continue
}
t.Errorf("DupSecOpt Failed invalid field %q", con[1])
}
secopt = DisableSecOpt()
if secopt[0] != "label:disable" {
t.Errorf("DisableSecOpt Failed level incorrect")
}
}
func TestRelabel(t *testing.T) {
testdir := "/tmp/test"
if err := os.Mkdir(testdir, 0755); err != nil {
t.Fatal(err)
}
defer os.RemoveAll(testdir)
label := "system_u:system_r:svirt_sandbox_file_t:s0:c1,c2"
if err := Relabel(testdir, "", true); err != nil {
t.Fatal("Relabel with no label failed: %v", err)
}
if err := Relabel(testdir, label, true); err != nil {
t.Fatal("Relabel shared failed: %v", err)
}
if err := Relabel(testdir, label, false); err != nil {
t.Fatal("Relabel unshared failed: %v", err)
}
if err := Relabel("/etc", label, false); err == nil {
t.Fatal("Relabel /etc succeeded")
}
if err := Relabel("/", label, false); err == nil {
t.Fatal("Relabel / succeeded")
}
if err := Relabel("/usr", label, false); err == nil {
t.Fatal("Relabel /usr succeeded")
}
}
func TestValidate(t *testing.T) {
if err := Validate("zZ"); err != ErrIncompatibleLabel {
t.Fatalf("Expected incompatible error, got %v", err)
}
if err := Validate("Z"); err != nil {
t.Fatal(err)
}
if err := Validate("z"); err != nil {
t.Fatal(err)
}
if err := Validate(""); err != nil {
t.Fatal(err)
}
}
func TestIsShared(t *testing.T) {
if shared := IsShared("Z"); shared {
t.Fatal("Expected label `Z` to not be shared, got %v", shared)
}
if shared := IsShared("z"); !shared {
t.Fatal("Expected label `z` to be shared, got %v", shared)
}
if shared := IsShared("Zz"); !shared {
t.Fatal("Expected label `Zz` to be shared, got %v", shared)
}
}

View File

@@ -3,8 +3,9 @@
package libcontainer
import (
"syscall"
"github.com/vishvananda/netlink/nl"
"golang.org/x/sys/unix"
)
// list of known message types we want to send to bootstrap program
@@ -12,12 +13,13 @@ import (
const (
InitMsg uint16 = 62000
CloneFlagsAttr uint16 = 27281
NsPathsAttr uint16 = 27282
UidmapAttr uint16 = 27283
GidmapAttr uint16 = 27284
SetgroupAttr uint16 = 27285
OomScoreAdjAttr uint16 = 27286
RootlessAttr uint16 = 27287
ConsolePathAttr uint16 = 27282
NsPathsAttr uint16 = 27283
UidmapAttr uint16 = 27284
GidmapAttr uint16 = 27285
SetgroupAttr uint16 = 27286
// When syscall.NLA_HDRLEN is in gccgo, take this out.
syscall_NLA_HDRLEN = (syscall.SizeofNlAttr + syscall.NLA_ALIGNTO - 1) & ^(syscall.NLA_ALIGNTO - 1)
)
type Int32msg struct {
@@ -25,8 +27,7 @@ type Int32msg struct {
Value uint32
}
// Serialize serializes the message.
// Int32msg has the following representation
// int32msg has the following representation
// | nlattr len | nlattr type |
// | uint32 value |
func (msg *Int32msg) Serialize() []byte {
@@ -39,10 +40,10 @@ func (msg *Int32msg) Serialize() []byte {
}
func (msg *Int32msg) Len() int {
return unix.NLA_HDRLEN + 4
return syscall_NLA_HDRLEN + 4
}
// Bytemsg has the following representation
// bytemsg has the following representation
// | nlattr len | nlattr type |
// | value | pad |
type Bytemsg struct {
@@ -52,7 +53,7 @@ type Bytemsg struct {
func (msg *Bytemsg) Serialize() []byte {
l := msg.Len()
buf := make([]byte, (l+unix.NLA_ALIGNTO-1) & ^(unix.NLA_ALIGNTO-1))
buf := make([]byte, (l+syscall.NLA_ALIGNTO-1) & ^(syscall.NLA_ALIGNTO-1))
native := nl.NativeEndian()
native.PutUint16(buf[0:2], uint16(l))
native.PutUint16(buf[2:4], msg.Type)
@@ -61,7 +62,7 @@ func (msg *Bytemsg) Serialize() []byte {
}
func (msg *Bytemsg) Len() int {
return unix.NLA_HDRLEN + len(msg.Value) + 1 // null-terminated
return syscall_NLA_HDRLEN + len(msg.Value) + 1 // null-terminated
}
type Boolmsg struct {
@@ -83,5 +84,5 @@ func (msg *Boolmsg) Serialize() []byte {
}
func (msg *Boolmsg) Len() int {
return unix.NLA_HDRLEN + 1
return syscall_NLA_HDRLEN + 1
}

View File

@@ -7,8 +7,7 @@ import (
"io/ioutil"
"os"
"path/filepath"
"golang.org/x/sys/unix"
"syscall"
)
const oomCgroupName = "memory"
@@ -26,13 +25,13 @@ func registerMemoryEvent(cgDir string, evName string, arg string) (<-chan struct
if err != nil {
return nil, err
}
fd, err := unix.Eventfd(0, unix.EFD_CLOEXEC)
if err != nil {
fd, _, syserr := syscall.RawSyscall(syscall.SYS_EVENTFD2, 0, syscall.FD_CLOEXEC, 0)
if syserr != 0 {
evFile.Close()
return nil, err
return nil, syserr
}
eventfd := os.NewFile(uintptr(fd), "eventfd")
eventfd := os.NewFile(fd, "eventfd")
eventControlPath := filepath.Join(cgDir, "cgroup.event_control")
data := fmt.Sprintf("%d %d %s", eventfd.Fd(), evFile.Fd(), arg)
@@ -44,9 +43,9 @@ func registerMemoryEvent(cgDir string, evName string, arg string) (<-chan struct
ch := make(chan struct{})
go func() {
defer func() {
close(ch)
eventfd.Close()
evFile.Close()
close(ch)
}()
buf := make([]byte, 8)
for {

View File

@@ -8,10 +8,9 @@ import (
"io/ioutil"
"os"
"path/filepath"
"syscall"
"testing"
"time"
"golang.org/x/sys/unix"
)
type notifyFunc func(paths map[string]string) (<-chan struct{}, error)
@@ -53,17 +52,21 @@ func testMemoryNotification(t *testing.T, evName string, notify notifyFunc, targ
t.Fatalf("invalid control data %q: %s", data, err)
}
// dup the eventfd
efd, err := unix.Dup(eventFd)
// re-open the eventfd
efd, err := syscall.Dup(eventFd)
if err != nil {
t.Fatal("unable to dup eventfd:", err)
t.Fatal("unable to reopen eventfd:", err)
}
defer syscall.Close(efd)
if err != nil {
t.Fatal("unable to dup event fd:", err)
}
defer unix.Close(efd)
buf := make([]byte, 8)
binary.LittleEndian.PutUint64(buf, 1)
if _, err := unix.Write(efd, buf); err != nil {
if _, err := syscall.Write(efd, buf); err != nil {
t.Fatal("unable to write to eventfd:", err)
}
@@ -78,7 +81,7 @@ func testMemoryNotification(t *testing.T, evName string, notify notifyFunc, targ
if err := os.RemoveAll(memoryPath); err != nil {
t.Fatal(err)
}
if _, err := unix.Write(efd, buf); err != nil {
if _, err := syscall.Write(efd, buf); err != nil {
t.Fatal("unable to write to eventfd:", err)
}
@@ -89,15 +92,14 @@ func testMemoryNotification(t *testing.T, evName string, notify notifyFunc, targ
t.Fatal("expected no notification to be triggered")
}
case <-time.After(100 * time.Millisecond):
t.Fatal("channel not closed after 100ms")
}
if _, _, err := unix.Syscall(unix.SYS_FCNTL, uintptr(evFd), unix.F_GETFD, 0); err != unix.EBADF {
t.Errorf("expected event control to be closed, but received error %s", err.Error())
if _, _, err := syscall.Syscall(syscall.SYS_FCNTL, uintptr(evFd), syscall.F_GETFD, 0); err != syscall.EBADF {
t.Error("expected event control to be closed")
}
if _, _, err := unix.Syscall(unix.SYS_FCNTL, uintptr(eventFd), unix.F_GETFD, 0); err != unix.EBADF {
t.Errorf("expected event fd to be closed, but received error %s", err.Error())
if _, _, err := syscall.Syscall(syscall.SYS_FCNTL, uintptr(eventFd), syscall.F_GETFD, 0); err != syscall.EBADF {
t.Error("expected event fd to be closed")
}
}

View File

@@ -10,35 +10,16 @@ The `nsenter` package will `import "C"` and it uses [cgo](https://golang.org/cmd
package. In cgo, if the import of "C" is immediately preceded by a comment, that comment,
called the preamble, is used as a header when compiling the C parts of the package.
So every time we import package `nsenter`, the C code function `nsexec()` would be
called. And package `nsenter` is now only imported in `main_unix.go`, so every time
before we call `cmd.Start` on linux, that C code would run.
Because `nsexec()` must be run before the Go runtime in order to use the
Linux kernel namespace, you must `import` this library into a package if
you plan to use `libcontainer` directly. Otherwise Go will not execute
the `nsexec()` constructor, which means that the re-exec will not cause
the namespaces to be joined. You can import it like this:
```go
import _ "github.com/opencontainers/runc/libcontainer/nsenter"
```
`nsexec()` will first get the file descriptor number for the init pipe
from the environment variable `_LIBCONTAINER_INITPIPE` (which was opened
by the parent and kept open across the fork-exec of the `nsexec()` init
process). The init pipe is used to read bootstrap data (namespace paths,
clone flags, uid and gid mappings, and the console path) from the parent
process. `nsexec()` will then call `setns(2)` to join the namespaces
provided in the bootstrap data (if available), `clone(2)` a child process
with the provided clone flags, update the user and group ID mappings, do
some further miscellaneous setup steps, and then send the PID of the
child process to the parent of the `nsexec()` "caller". Finally,
the parent `nsexec()` will exit and the child `nsexec()` process will
return to allow the Go runtime take over.
NOTE: We do both `setns(2)` and `clone(2)` even if we don't have any
CLONE_NEW* clone flags because we must fork a new process in order to
enter the PID namespace.
called. And package `nsenter` is now only imported in Docker execdriver, so every time
before we call `execdriver.Exec()`, that C code would run.
`nsexec()` will first check the environment variable `_LIBCONTAINER_INITPID`
which will give the process of the container that should be joined. Namespaces fd will
be found from `/proc/[pid]/ns` and set by `setns` syscall.
And then get the pipe number from `_LIBCONTAINER_INITPIPE`, error message could
be transfered through it. If tty is added, `_LIBCONTAINER_CONSOLE_PATH` will
have value and start a console for output.
Finally, `nsexec()` will clone a child process , exit the parent process and let
the Go runtime take over.

View File

@@ -1,32 +0,0 @@
#ifndef NSENTER_NAMESPACE_H
#define NSENTER_NAMESPACE_H
#ifndef _GNU_SOURCE
# define _GNU_SOURCE
#endif
#include <sched.h>
/* All of these are taken from include/uapi/linux/sched.h */
#ifndef CLONE_NEWNS
# define CLONE_NEWNS 0x00020000 /* New mount namespace group */
#endif
#ifndef CLONE_NEWCGROUP
# define CLONE_NEWCGROUP 0x02000000 /* New cgroup namespace */
#endif
#ifndef CLONE_NEWUTS
# define CLONE_NEWUTS 0x04000000 /* New utsname namespace */
#endif
#ifndef CLONE_NEWIPC
# define CLONE_NEWIPC 0x08000000 /* New ipc namespace */
#endif
#ifndef CLONE_NEWUSER
# define CLONE_NEWUSER 0x10000000 /* New user namespace */
#endif
#ifndef CLONE_NEWPID
# define CLONE_NEWPID 0x20000000 /* New pid namespace */
#endif
#ifndef CLONE_NEWNET
# define CLONE_NEWNET 0x40000000 /* New network namespace */
#endif
#endif /* NSENTER_NAMESPACE_H */

Some files were not shown because too many files have changed in this diff Show More