github/skopeo
1
0
Fork 0
mirror of https://github.com/containers/skopeo.git synced 2025-04-28 03:10:18 +00:00
Code Issues Projects Releases Wiki Activity

Re-add c/image documentation

Browse Source 
... which was for some uknown reason removed in the previous backport.

Signed-off-by: Miloslav Trmač <mitr@redhat.com>
This commit is contained in:
Miloslav Trmač 2020-02-21 16:26:47 +01:00
parent 5c3a5ddf2d
commit 7902be2dc4
9 changed files with 1250 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

66
vendor/github.com/containers/image/docs/atomic-signature-embedded-json.json generated vendored Normal file
View File

@ -0,0 +1,66 @@
{
"title": "JSON embedded in an atomic container signature",
"description": "This schema is a supplement to atomic-signature.md in this directory.\n\nConsumers of the JSON MUST use the processing rules documented in atomic-signature.md, especially the requirements for the 'critical' subjobject.\n\nWhenever this schema and atomic-signature.md, or the github.com/containers/image/signature implementation, differ,\nit is the atomic-signature.md document, or the github.com/containers/image/signature implementation, which governs.\n\nUsers are STRONGLY RECOMMENDED to use the github.com/containeres/image/signature implementation instead of writing\ntheir own, ESPECIALLY when consuming signatures, so that the policy.json format can be shared by all image consumers.\n",
"type": "object",
"required": [
"critical",
"optional"
],
"additionalProperties": false,
"properties": {
"critical": {
"type": "object",
"required": [
"type",
"image",
"identity"
],
"additionalProperties": false,
"properties": {
"type": {
"type": "string",
"enum": [
"atomic container signature"
]
},
"image": {
"type": "object",
"required": [
"docker-manifest-digest"
],
"additionalProperties": false,
"properties": {
"docker-manifest-digest": {
"type": "string"
}
}
},
"identity": {
"type": "object",
"required": [
"docker-reference"
],
"additionalProperties": false,
"properties": {
"docker-reference": {
"type": "string"
}
}
}
}
},
"optional": {
"type": "object",
"description": "All members are optional, but if they are included, they must be valid.",
"additionalProperties": true,
"properties": {
"creator": {
"type": "string"
},
"timestamp": {
"type": "integer"
}
}
}
}
}

28
vendor/github.com/containers/image/docs/containers-certs.d.5.md generated vendored Normal file
View File

@ -0,0 +1,28 @@
% containers-certs.d(5)
# NAME
containers-certs.d - Directory for storing custom container-registry TLS configurations
# DESCRIPTION
A custom TLS configuration for a container registry can be configured by creating a directory under `/etc/containers/certs.d`.
The name of the directory must correspond to the `host:port` of the registry (e.g., `my-registry.com:5000`).
## Directory Structure
A certs directory can contain one or more files with the following extensions:
* `*.crt` files with this extensions will be interpreted as CA certificates
* `*.cert` files with this extensions will be interpreted as client certificates
* `*.key` files with this extensions will be interpreted as client keys
Note that the client certificate-key pair will be selected by the file name (e.g., `client.{cert,key}`).
An examplary setup for a registry running at `my-registry.com:5000` may look as follows:
```
/etc/containers/certs.d/ <- Certificate directory
└── my-registry.com:5000 <- Hostname:port
├── client.cert <- Client certificate
├── client.key <- Client key
└── ca.crt <- Certificate authority that signed the registry certificate
```
# HISTORY
Feb 2019, Originally compiled by Valentin Rothberg <rothberg@redhat.com>

283
vendor/github.com/containers/image/docs/containers-policy.json.5.md generated vendored Normal file
View File

@ -0,0 +1,283 @@
% CONTAINERS-POLICY.JSON(5) policy.json Man Page
% Miloslav Trmač
% September 2016
# NAME
containers-policy.json - syntax for the signature verification policy file
## DESCRIPTION
Signature verification policy files are used to specify policy, e.g. trusted keys,
applicable when deciding whether to accept an image, or individual signatures of that image, as valid.
The default policy is stored (unless overridden at compile-time) at `/etc/containers/policy.json`;
applications performing verification may allow using a different policy instead.
## FORMAT
The signature verification policy file, usually called `policy.json`,
uses a JSON format. Unlike some other JSON files, its parsing is fairly strict:
unrecognized, duplicated or otherwise invalid fields cause the entire file,
and usually the entire operation, to be rejected.
The purpose of the policy file is to define a set of *policy requirements* for a container image,
usually depending on its location (where it is being pulled from) or otherwise defined identity.
Policy requirements can be defined for:
- An individual *scope* in a *transport*.
The *transport* values are the same as the transport prefixes when pushing/pulling images (e.g. `docker:`, `atomic:`),
and *scope* values are defined by each transport; see below for more details.
Usually, a scope can be defined to match a single image, and various prefixes of
such a most specific scope define namespaces of matching images.
- A default policy for a single transport, expressed using an empty string as a scope
- A global default policy.
If multiple policy requirements match a given image, only the requirements from the most specific match apply,
the more general policy requirements definitions are ignored.
This is expressed in JSON using the top-level syntax
```js
{
"default": [/* policy requirements: global default */]
"transports": {
transport_name: {
"": [/* policy requirements: default for transport $transport_name */],
scope_1: [/* policy requirements: default for $scope_1 in $transport_name */],
scope_2: [/*…*/]
/*…*/
},
transport_name_2: {/*…*/}
/*…*/
}
}
```
The global `default` set of policy requirements is mandatory; all of the other fields
(`transports` itself, any specific transport, the transport-specific default, etc.) are optional.
<!-- NOTE: Keep this in sync with transports/transports.go! -->
## Supported transports and their scopes
### `atomic:`
The `atomic:` transport refers to images in an Atomic Registry.
Supported scopes use the form _hostname_[`:`_port_][`/`_namespace_[`/`_imagestream_ [`:`_tag_]]],
i.e. either specifying a complete name of a tagged image, or prefix denoting
a host/namespace/image stream.
*Note:* The _hostname_ and _port_ refer to the Docker registry host and port (the one used
e.g. for `docker pull`), _not_ to the OpenShift API host and port.
### `dir:`
The `dir:` transport refers to images stored in local directories.
Supported scopes are paths of directories (either containing a single image or
subdirectories possibly containing images).
*Note:* The paths must be absolute and contain no symlinks. Paths violating these requirements may be silently ignored.
The top-level scope `"/"` is forbidden; use the transport default scope `""`,
for consistency with other transports.
### `docker:`
The `docker:` transport refers to images in a registry implementing the "Docker Registry HTTP API V2".
Scopes matching individual images are named Docker references *in the fully expanded form*, either
using a tag or digest. For example, `docker.io/library/busybox:latest` (*not* `busybox:latest`).
More general scopes are prefixes of individual-image scopes, and specify a repository (by omitting the tag or digest),
a repository namespace, or a registry host (by only specifying the host name).
### `oci:`
The `oci:` transport refers to images in directories compliant with "Open Container Image Layout Specification".
Supported scopes use the form _directory_`:`_tag_, and _directory_ referring to
a directory containing one or more tags, or any of the parent directories.
*Note:* See `dir:` above for semantics and restrictions on the directory paths, they apply to `oci:` equivalently.
### `tarball:`
The `tarball:` transport refers to tarred up container root filesystems.
Scopes are ignored.
## Policy Requirements
Using the mechanisms above, a set of policy requirements is looked up. The policy requirements
are represented as a JSON array of individual requirement objects. For an image to be accepted,
*all* of the requirements must be satisfied simulatenously.
The policy requirements can also be used to decide whether an individual signature is accepted (= is signed by a recognized key of a known author);
in that case some requirements may apply only to some signatures, but each signature must be accepted by *at least one* requirement object.
The following requirement objects are supported:
### `insecureAcceptAnything`
A simple requirement with the following syntax
```json
{"type":"insecureAcceptAnything"}
```
This requirement accepts any image (but note that other requirements in the array still apply).
When deciding to accept an individual signature, this requirement does not have any effect; it does *not* cause the signature to be accepted, though.
This is useful primarily for policy scopes where no signature verification is required;
because the array of policy requirements must not be empty, this requirement is used
to represent the lack of requirements explicitly.
### `reject`
A simple requirement with the following syntax:
```json
{"type":"reject"}
```
This requirement rejects every image, and every signature.
### `signedBy`
This requirement requires an image to be signed with an expected identity, or accepts a signature if it is using an expected identity and key.
```js
{
"type": "signedBy",
"keyType": "GPGKeys", /* The only currently supported value */
"keyPath": "/path/to/local/keyring/file",
"keyData": "base64-encoded-keyring-data",
"signedIdentity": identity_requirement
}
```
<!-- Later: other keyType values -->
Exactly one of `keyPath` and `keyData` must be present, containing a GPG keyring of one or more public keys. Only signatures made by these keys are accepted.
The `signedIdentity` field, a JSON object, specifies what image identity the signature claims about the image.
One of the following alternatives are supported:
- The identity in the signature must exactly match the image identity. Note that with this, referencing an image by digest (with a signature claiming a _repository_`:`_tag_ identity) will fail.
```json
{"type":"matchExact"}
```
- If the image identity carries a tag, the identity in the signature must exactly match;
if the image identity uses a digest reference, the identity in the signature must be in the same repository as the image identity (using any tag).
(Note that with images identified using digest references, the digest from the reference is validated even before signature verification starts.)
```json
{"type":"matchRepoDigestOrExact"}
```
- The identity in the signature must be in the same repository as the image identity. This is useful e.g. to pull an image using the `:latest` tag when the image is signed with a tag specifing an exact image version.
```json
{"type":"matchRepository"}
```
- The identity in the signature must exactly match a specified identity.
This is useful e.g. when locally mirroring images signed using their public identity.
```js
{
"type": "exactReference",
"dockerReference": docker_reference_value
}
```
- The identity in the signature must be in the same repository as a specified identity.
This combines the properties of `matchRepository` and `exactReference`.
```js
{
"type": "exactRepository",
"dockerRepository": docker_repository_value
}
```
If the `signedIdentity` field is missing, it is treated as `matchRepoDigestOrExact`.
*Note*: `matchExact`, `matchRepoDigestOrExact` and `matchRepository` can be only used if a Docker-like image identity is
provided by the transport. In particular, the `dir:` and `oci:` transports can be only
used with `exactReference` or `exactRepository`.
<!-- ### `signedBaseLayer` -->
## Examples
It is *strongly* recommended to set the `default` policy to `reject`, and then
selectively allow individual transports and scopes as desired.
### A reasonably locked-down system
(Note that the `/*`…`*/` comments are not valid in JSON, and must not be used in real policies.)
```js
{
"default": [{"type": "reject"}], /* Reject anything not explicitly allowed */
"transports": {
"docker": {
/* Allow installing images from a specific repository namespace, without cryptographic verification.
This namespace includes images like openshift/hello-openshift and openshift/origin. */
"docker.io/openshift": [{"type": "insecureAcceptAnything"}],
/* Similarly, allow installing the “official” busybox images. Note how the fully expanded
form, with the explicit /library/, must be used. */
"docker.io/library/busybox": [{"type": "insecureAcceptAnything"}]
/* Other docker: images use the global default policy and are rejected */
},
"dir": {
"": [{"type": "insecureAcceptAnything"}] /* Allow any images originating in local directories */
},
"atomic": {
/* The common case: using a known key for a repository or set of repositories */
"hostname:5000/myns/official": [
{
"type": "signedBy",
"keyType": "GPGKeys",
"keyPath": "/path/to/official-pubkey.gpg"
}
],
/* A more complex example, for a repository which contains a mirror of a third-party product,
which must be signed-off by local IT */
"hostname:5000/vendor/product": [
{ /* Require the image to be signed by the original vendor, using the vendor's repository location. */
"type": "signedBy",
"keyType": "GPGKeys",
"keyPath": "/path/to/vendor-pubkey.gpg",
"signedIdentity": {
"type": "exactRepository",
"dockerRepository": "vendor-hostname/product/repository"
}
},
{ /* Require the image to _also_ be signed by a local reviewer. */
"type": "signedBy",
"keyType": "GPGKeys",
"keyPath": "/path/to/reviewer-pubkey.gpg"
}
]
}
}
}
```
### Completely disable security, allow all images, do not trust any signatures
```json
{
"default": [{"type": "insecureAcceptAnything"}]
}
```
## SEE ALSO
atomic(1)
## HISTORY
August 2018, Rename to containers-policy.json(5) by Valentin Rothberg <vrothberg@suse.com>
September 2016, Originally compiled by Miloslav Trmač <mitr@redhat.com>

177
vendor/github.com/containers/image/docs/containers-registries.conf.5.md generated vendored Normal file
View File

@ -0,0 +1,177 @@
% CONTAINERS-REGISTRIES.CONF(5) System-wide registry configuration file
% Brent Baude
% Aug 2017
# NAME
containers-registries.conf - Syntax of System Registry Configuration File
# DESCRIPTION
The CONTAINERS-REGISTRIES configuration file is a system-wide configuration
file for container image registries. The file format is TOML.
By default, the configuration file is located at `/etc/containers/registries.conf`.
# FORMATS
## VERSION 2
VERSION 2 is the latest format of the `registries.conf` and is currently in
beta. This means in general VERSION 1 should be used in production environments
for now.
### GLOBAL SETTINGS
`unqualified-search-registries`
: An array of _host_[`:`_port_] registries to try when pulling an unqualified image, in order.
### NAMESPACED `[[registry]]` SETTINGS
The bulk of the configuration is represented as an array of `[[registry]]`
TOML tables; the settings may therefore differ among different registries
as well as among different namespaces/repositories within a registry.
#### Choosing a `[[registry]]` TOML table
Given an image name, a single `[[registry]]` TOML table is chosen based on its `prefix` field.
`prefix`
: A prefix of the user-specified image name, i.e. using one of the following formats:
- _host_[`:`_port_]
- _host_[`:`_port_]`/`_namespace_[`/`_namespace_…]
- _host_[`:`_port_]`/`_namespace_[`/`_namespace_…]`/`_repo_
- _host_[`:`_port_]`/`_namespace_[`/`_namespace_…]`/`_repo_(`:`_tag|`@`_digest_)
The user-specified image name must start with the specified `prefix` (and continue
with the appropriate separator) for a particular `[[registry]]` TOML table to be
considered; (only) the TOML table with the longest match is used.
As a special case, the `prefix` field can be missing; if so, it defaults to the value
of the `location` field (described below).
#### Per-namespace settings
`insecure`
: `true` or `false`.
By default, container runtimes require TLS when retrieving images from a registry.
If `insecure` is set to `true`, unencrypted HTTP as well as TLS connections with untrusted
certificates are allowed.
`blocked`
: `true` or `false`.
If `true`, pulling images with matching names is forbidden.
#### Remapping and mirroring registries
The user-specified image reference is, primarily, a "logical" image name, always used for naming
the image. By default, the image reference also directly specifies the registry and repository
to use, but the following options can be used to redirect the underlying accesses
to different registry servers or locations (e.g. to support configurations with no access to the
internet without having to change `Dockerfile`s, or to add redundancy).
`location`
: Accepts the same format as the `prefix` field, and specifies the physical location
of the `prefix`-rooted namespace.
By default, this equal to `prefix` (in which case `prefix` can be omitted and the
`[[registry]]` TOML table can only specify `location`).
Example: Given
```
prefix = "example.com/foo"
location = "internal-registry-for-example.net/bar"
```
requests for the image `example.com/foo/myimage:latest` will actually work with the
`internal-registry-for-example.net/bar/myimage:latest` image.
`mirror`
: An array of TOML tables specifiying (possibly-partial) mirrors for the
`prefix`-rooted namespace.
The mirrors are attempted in the specified order; the first one that can be
contacted and contains the image will be used (and if none of the mirrors contains the image,
the primary location specified by the `registry.location` field, or using the unmodified
user-specified reference, is tried last).
Each TOML table in the `mirror` array can contain the following fields, with the same semantics
as if specified in the `[[registry]]` TOML table directly:
- `location`
- `insecure`
`mirror-by-digest-only`
: `true` or `false`.
If `true`, mirrors will only be used during pulling if the image reference includes a digest.
Referencing an image by digest ensures that the same is always used
(whereas referencing an image by a tag may cause different registries to return
different images if the tag mapping is out of sync).
Note that if this is `true`, images referenced by a tag will only use the primary
registry, failing if that registry is not accessible.
*Note*: Redirection and mirrors are currently processed only when reading images, not when pushing
to a registry; that may change in the future.
### EXAMPLE
```
unqualified-search-registries = ["example.com"]
[[registry]]
prefix = "example.com/foo"
insecure = false
blocked = false
location = "internal-registry-for-example.com/bar"
[[registry.mirror]]
location = "example-mirror-0.local/mirror-for-foo"
[[registry.mirror]]
location = "example-mirror-1.local/mirrors/foo"
insecure = true
```
Given the above, a pull of `example.com/foo/image:latest` will try:
1. `example-mirror-0.local/mirror-for-foo/image:latest`
2. `example-mirror-1.local/mirrors/foo/image:latest`
3. `internal-registry-for-example.net/bar/myimage:latest`
in order, and use the first one that exists.
## VERSION 1
VERSION 1 can be used as alternative to the VERSION 2, but it does not support
using registry mirrors, longest-prefix matches, or location rewriting.
The TOML format is used to build a simple list of registries under three
categories: `registries.search`, `registries.insecure`, and `registries.block`.
You can list multiple registries using a comma separated list.
Search registries are used when the caller of a container runtime does not fully specify the
container image that they want to execute. These registries are prepended onto the front
of the specified container image until the named image is found at a registry.
Note that insecure registries can be used for any registry, not just the registries listed
under search.
The `registries.insecure` and `registries.block` lists have the same meaning as the
`insecure` and `blocked` fields in VERSION 2.
### EXAMPLE
The following example configuration defines two searchable registries, one
insecure registry, and two blocked registries.
```
[registries.search]
registries = ['registry1.com', 'registry2.com']
[registries.insecure]
registries = ['registry3.com']
[registries.block]
registries = ['registry.untrusted.com', 'registry.unsafe.com']
```
# HISTORY
Mar 2019, Added additional configuration format by Sascha Grunert <sgrunert@suse.com>
Aug 2018, Renamed to containers-registries.conf(5) by Valentin Rothberg <vrothberg@suse.com>
Jun 2018, Updated by Tom Sweeney <tsweeney@redhat.com>
Aug 2017, Originally compiled by Brent Baude <bbaude@redhat.com>

128
vendor/github.com/containers/image/docs/containers-registries.d.5.md generated vendored Normal file
View File

@ -0,0 +1,128 @@
% CONTAINERS-REGISTRIES.D(5) Registries.d Man Page
% Miloslav Trmač
% August 2016
# NAME
containers-registries.d - Directory for various registries configurations
# DESCRIPTION
The registries configuration directory contains configuration for various registries
(servers storing remote container images), and for content stored in them,
so that the configuration does not have to be provided in command-line options over and over for every command,
and so that it can be shared by all users of containers/image.
By default (unless overridden at compile-time), the registries configuration directory is `/etc/containers/registries.d`;
applications may allow using a different directory instead.
## Directory Structure
The directory may contain any number of files with the extension `.yaml`,
each using the YAML format. Other than the mandatory extension, names of the files
don’t matter.
The contents of these files are merged together; to have a well-defined and easy to understand
behavior, there can be only one configuration section describing a single namespace within a registry
(in particular there can be at most one one `default-docker` section across all files,
and there can be at most one instance of any key under the the `docker` section;
these sections are documented later).
Thus, it is forbidden to have two conflicting configurations for a single registry or scope,
and it is also forbidden to split a configuration for a single registry or scope across
more than one file (even if they are not semantically in conflict).
## Registries, Scopes and Search Order
Each YAML file must contain a “YAML mapping” (key-value pairs). Two top-level keys are defined:
- `default-docker` is the _configuration section_ (as documented below)
for registries implementing "Docker Registry HTTP API V2".
This key is optional.
- `docker` is a mapping, using individual registries implementing "Docker Registry HTTP API V2",
or namespaces and individual images within these registries, as keys;
the value assigned to any such key is a _configuration section_.
This key is optional.
Scopes matching individual images are named Docker references *in the fully expanded form*, either
using a tag or digest. For example, `docker.io/library/busybox:latest` (*not* `busybox:latest`).
More general scopes are prefixes of individual-image scopes, and specify a repository (by omitting the tag or digest),
a repository namespace, or a registry host (and a port if it differs from the default).
Note that if a registry is accessed using a hostname+port configuration, the port-less hostname
is _not_ used as parent scope.
When searching for a configuration to apply for an individual container image, only
the configuration for the most-precisely matching scope is used; configuration using
more general scopes is ignored. For example, if _any_ configuration exists for
`docker.io/library/busybox`, the configuration for `docker.io` is ignored
(even if some element of the configuration is defined for `docker.io` and not for `docker.io/library/busybox`).
## Individual Configuration Sections
A single configuration section is selected for a container image using the process
described above. The configuration section is a YAML mapping, with the following keys:
- `sigstore-staging` defines an URL of of the signature storage, used for editing it (adding or deleting signatures).
This key is optional; if it is missing, `sigstore` below is used.
- `sigstore` defines an URL of the signature storage.
This URL is used for reading existing signatures,
and if `sigstore-staging` does not exist, also for adding or removing them.
This key is optional; if it is missing, no signature storage is defined (no signatures
are download along with images, adding new signatures is possible only if `sigstore-staging` is defined).
## Examples
### Using Containers from Various Origins
The following demonstrates how to to consume and run images from various registries and namespaces:
```yaml
docker:
registry.database-supplier.com:
sigstore: https://sigstore.database-supplier.com
distribution.great-middleware.org:
sigstore: https://security-team.great-middleware.org/sigstore
docker.io/web-framework:
sigstore: https://sigstore.web-framework.io:8080
```
### Developing and Signing Containers, Staging Signatures
For developers in `example.com`:
- Consume most container images using the public servers also used by clients.
- Use a separate sigure storage for an container images in a namespace corresponding to the developers' department, with a staging storage used before publishing signatures.
- Craft an individual exception for a single branch a specific developer is working on locally.
```yaml
docker:
registry.example.com:
sigstore: https://registry-sigstore.example.com
registry.example.com/mydepartment:
sigstore: https://sigstore.mydepartment.example.com
sigstore-staging: file:///mnt/mydepartment/sigstore-staging
registry.example.com/mydepartment/myproject:mybranch:
sigstore: http://localhost:4242/sigstore
sigstore-staging: file:///home/useraccount/webroot/sigstore
```
### A Global Default
If a company publishes its products using a different domain, and different registry hostname for each of them, it is still possible to use a single signature storage server
without listing each domain individually. This is expected to rarely happen, usually only for staging new signatures.
```yaml
default-docker:
sigstore-staging: file:///mnt/company/common-sigstore-staging
```
# AUTHORS
Miloslav Trmač <mitr@redhat.com>

241
vendor/github.com/containers/image/docs/containers-signature.5.md generated vendored Normal file
View File

@ -0,0 +1,241 @@
% container-signature(5) Container signature format
% Miloslav Trmač
% March 2017
# Container signature format
This document describes the format of container signatures,
as implemented by the `github.com/containers/image/signature` package.
Most users should be able to consume these signatures by using the `github.com/containers/image/signature` package
(preferably through the higher-level `signature.PolicyContext` interface)
without having to care about the details of the format described below.
This documentation exists primarily for maintainers of the package
and to allow independent reimplementations.
## High-level overview
The signature provides an end-to-end authenticated claim that a container image
has been approved by a specific party (e.g. the creator of the image as their work,
an automated build system as a result of an automated build,
a company IT department approving the image for production) under a specified _identity_
(e.g. an OS base image / specific application, with a specific version).
A container signature consists of a cryptographic signature which identifies
and authenticates who signed the image, and carries as a signed payload a JSON document.
The JSON document identifies the image being signed, claims a specific identity of the
image and if applicable, contains other information about the image.
The signatures do not modify the container image (the layers, configuration, manifest, …);
e.g. their presence does not change the manifest digest used to identify the image in
docker/distribution servers; rather, the signatures are associated with an immutable image.
An image can have any number of signatures so signature distribution systems SHOULD support
associating more than one signature with an image.
## The cryptographic signature
As distributed, the container signature is a blob which contains a cryptographic signature
in an industry-standard format, carrying a signed JSON payload (i.e. the blob contains both the
JSON document and a signature of the JSON document; it is not a “detached signature” with
independent blobs containing the JSON document and a cryptographic signature).
Currently the only defined cryptographic signature format is an OpenPGP signature (RFC 4880),
but others may be added in the future. (The blob does not contain metadata identifying the
cryptographic signature format. It is expected that most formats are sufficiently self-describing
that this is not necessary and the configured expected public key provides another indication
of the expected cryptographic signature format. Such metadata may be added in the future for
newly added cryptographic signature formats, if necessary.)
Consumers of container signatures SHOULD verify the cryptographic signature
against one or more trusted public keys
(e.g. defined in a [policy.json signature verification policy file](policy.json.md))
before parsing or processing the JSON payload in _any_ way,
in particular they SHOULD stop processing the container signature
if the cryptographic signature verification fails, without even starting to process the JSON payload.
(Consumers MAY extract identification of the signing key and other metadata from the cryptographic signature,
and the JSON payload, without verifying the signature, if the purpose is to allow managing the signature blobs,
e.g. to list the authors and image identities of signatures associated with a single container image;
if so, they SHOULD design the output of such processing to minimize the risk of users considering the output trusted
or in any way usable for making policy decisions about the image.)
### OpenPGP signature verification
When verifying a cryptographic signature in the OpenPGP format,
the consumer MUST verify at least the following aspects of the signature
(like the `github.com/containers/image/signature` package does):
- The blob MUST be a “Signed Message” as defined RFC 4880 section 11.3.
(e.g. it MUST NOT be an unsigned “Literal Message”, or any other non-signature format).
- The signature MUST have been made by an expected key trusted for the purpose (and the specific container image).
- The signature MUST be correctly formed and pass the cryptographic validation.
- The signature MUST correctly authenticate the included JSON payload
(in particular, the parsing of the JSON payload MUST NOT start before the complete payload has been cryptographically authenticated).
- The signature MUST NOT be expired.
The consumer SHOULD have tests for its verification code which verify that signatures failing any of the above are rejected.
## JSON processing and forward compatibility
The payload of the cryptographic signature is a JSON document (RFC 7159).
Consumers SHOULD parse it very strictly,
refusing any signature which violates the expected format (e.g. missing members, incorrect member types)
or can be interpreted ambiguously (e.g. a duplicated member in a JSON object).
Any violations of the JSON format or of other requirements in this document MAY be accepted if the JSON document can be recognized
to have been created by a known-incorrect implementation (see [`optional.creator`](#optionalcreator) below)
and if the semantics of the invalid document, as created by such an implementation, is clear.
The top-level value of the JSON document MUST be a JSON object with exactly two members, `critical` and `optional`,
each a JSON object.
The `critical` object MUST contain a `type` member identifying the document as a container signature
(as defined [below](#criticaltype))
and signature consumers MUST reject signatures which do not have this member or in which this member does not have the expected value.
To ensure forward compatibility (allowing older signature consumers to correctly
accept or reject signatures created at a later date, with possible extensions to this format),
consumers MUST reject the signature if the `critical` object, or _any_ of its subobjects,
contain _any_ member or data value which is unrecognized, unsupported, invalid, or in any other way unexpected.
At a minimum, this includes unrecognized members in a JSON object, or incorrect types of expected members.
For the same reason, consumers SHOULD accept any members with unrecognized names in the `optional` object,
and MAY accept signatures where the object member is recognized but unsupported, or the value of the member is unsupported.
Consumers still SHOULD reject signatures where a member of an `optional` object is supported but the value is recognized as invalid.
## JSON data format
An example of the full format follows, with detailed description below.
To reiterate, consumers of the signature SHOULD perform successful cryptographic verification,
and MUST reject unexpected data in the `critical` object, or in the top-level object, as described above.
```json
{
"critical": {
"type": "atomic container signature",
"image": {
"docker-manifest-digest": "sha256:817a12c32a39bbe394944ba49de563e085f1d3c5266eb8e9723256bc4448680e"
},
"identity": {
"docker-reference": "docker.io/library/busybox:latest"
}
},
"optional": {
"creator": "some software package v1.0.1-35",
"timestamp": 1483228800,
}
}
```
### `critical`
This MUST be a JSON object which contains data critical to correctly evaluating the validity of a signature.
Consumers MUST reject any signature where the `critical` object contains any unrecognized, unsupported, invalid or in any other way unexpected member or data.
### `critical.type`
This MUST be a string with a string value exactly equal to `atomic container signature` (three words, including the spaces).
Signature consumers MUST reject signatures which do not have this member or this member does not have exactly the expected value.
(The consumers MAY support signatures with a different value of the `type` member, if any is defined in the future;
if so, the rest of the JSON document is interpreted according to rules defining that value of `critical.type`,
not by this document.)
### `critical.image`
This MUST be a JSON object which identifies the container image this signature applies to.
Consumers MUST reject any signature where the `critical.image` object contains any unrecognized, unsupported, invalid or in any other way unexpected member or data.
(Currently only the `docker-manifest-digest` way of identifying a container image is defined;
alternatives to this may be defined in the future,
but existing consumers are required to reject signatures which use formats they do not support.)
### `critical.image.docker-manifest-digest`
This MUST be a JSON string, in the `github.com/opencontainers/go-digest.Digest` string format.
The value of this member MUST match the manifest of the signed container image, as implemented in the docker/distribution manifest addressing system.
The consumer of the signature SHOULD verify the manifest digest against a fully verified signature before processing the contents of the image manifest in any other way
(e.g. parsing the manifest further or downloading layers of the image).
Implementation notes:
* A single container image manifest may have several valid manifest digest values, using different algorithms.
* For “signed” [docker/distribution schema 1](https://github.com/docker/distribution/blob/master/docs/spec/manifest-v2-1.md) manifests,
the manifest digest applies to the payload of the JSON web signature, not to the raw manifest blob.
### `critical.identity`
This MUST be a JSON object which identifies the claimed identity of the image (usually the purpose of the image, or the application, along with a version information),
as asserted by the author of the signature.
Consumers MUST reject any signature where the `critical.identity` object contains any unrecognized, unsupported, invalid or in any other way unexpected member or data.
(Currently only the `docker-reference` way of claiming an image identity/purpose is defined;
alternatives to this may be defined in the future,
but existing consumers are required to reject signatures which use formats they do not support.)
### `critical.identity.docker-reference`
This MUST be a JSON string, in the `github.com/docker/distribution/reference` string format,
and using the same normalization semantics (where e.g. `busybox:latest` is equivalent to `docker.io/library/busybox:latest`).
If the normalization semantics allows multiple string representations of the claimed identity with equivalent meaning,
the `critical.identity.docker-reference` member SHOULD use the fully explicit form (including the full host name and namespaces).
The value of this member MUST match the image identity/purpose expected by the consumer of the image signature and the image
(again, accounting for the `docker/distribution/reference` normalization semantics).
In the most common case, this means that the `critical.identity.docker-reference` value must be equal to the docker/distribution reference used to refer to or download the image.
However, depending on the specific application, users or system administrators may accept less specific matches
(e.g. ignoring the tag value in the signature when pulling the `:latest` tag or when referencing an image by digest),
or they may require `critical.identity.docker-reference` values with a completely different namespace to the reference used to refer to/download the image
(e.g. requiring a `critical.identity.docker-reference` value which identifies the image as coming from a supplier when fetching it from a company-internal mirror of approved images).
The software performing this verification SHOULD allow the users to define such a policy using the [policy.json signature verification policy file format](policy.json.md).
The `critical.identity.docker-reference` value SHOULD contain either a tag or digest;
in most cases, it SHOULD use a tag rather than a digest. (See also the default [`matchRepoDigestOrExact` matching semantics in `policy.json`](policy.json.md#signedby).)
### `optional`
This MUST be a JSON object.
Consumers SHOULD accept any members with unrecognized names in the `optional` object,
and MAY accept a signature where the object member is recognized but unsupported, or the value of the member is valid but unsupported.
Consumers still SHOULD reject any signature where a member of an `optional` object is supported but the value is recognized as invalid.
### `optional.creator`
If present, this MUST be a JSON string, identifying the name and version of the software which has created the signature.
The contents of this string is not defined in detail; however each implementation creating container signatures:
- SHOULD define the contents to unambiguously define the software in practice (e.g. it SHOULD contain the name of the software, not only the version number)
- SHOULD use a build and versioning process which ensures that the contents of this string (e.g. an included version number)
changes whenever the format or semantics of the generated signature changes in any way;
it SHOULD not be possible for two implementations which use a different format or semantics to have the same `optional.creator` value
- SHOULD use a format which is reasonably easy to parse in software (perhaps using a regexp),
and which makes it easy enough to recognize a range of versions of a specific implementation
(e.g. the version of the implementation SHOULD NOT be only a git hash, because they don’t have an easily defined ordering;
the string should contain a version number, or at least a date of the commit).
Consumers of container signatures MAY recognize specific values or sets of values of `optional.creator`
(perhaps augmented with `optional.timestamp`),
and MAY change their processing of the signature based on these values
(usually to acommodate violations of this specification in past versions of the signing software which cannot be fixed retroactively),
as long as the semantics of the invalid document, as created by such an implementation, is clear.
If consumers of signatures do change their behavior based on the `optional.creator` value,
they SHOULD take care that the way they process the signatures is not inconsistent with
strictly validating signature consumers.
(I.e. it is acceptable for a consumer to accept a signature based on a specific `optional.creator` value
if other implementations would completely reject the signature,
but it would be very undesirable for the two kinds of implementations to accept the signature in different
and inconsistent situations.)
### `optional.timestamp`
If present, this MUST be a JSON number, which is representable as a 64-bit integer, and identifies the time when the signature was created
as the number of seconds since the UNIX epoch (Jan 1 1970 00:00 UTC).

109
vendor/github.com/containers/image/docs/containers-transports.5.md generated vendored Normal file
View File

@ -0,0 +1,109 @@
% CONTAINERS-TRANSPORTS(5) Containers Transports Man Page
% Valentin Rothberg
% April 2019
## NAME
containers-transports - description of supported transports for copying and storing container images
## DESCRIPTION
Tools which use the containers/image library, including skopeo(1), buildah(1), podman(1), all share a common syntax for referring to container images in various locations.
The general form of the syntax is _transport:details_, where details are dependent on the specified transport, which are documented below.
### **containers-storage:** [storage-specifier]{image-id|docker-reference[@image-id]}
An image located in a local containers storage.
The format of _docker-reference_ is described in detail in the **docker** transport.
The _storage-specifier_ allows for referencing storage locations on the file system and has the format `[[driver@]root[+run-root][:options]]` where the optional `driver` refers to the storage driver (e.g., overlay or btrfs) and where `root` is an absolute path to the storage's root directory.
The optional `run-root` can be used to specify the run directory of the storage where all temporary writable content is stored.
The optional `options` are a comma-separated list of driver-specific options.
Please refer to containers-storage.conf(5) for further information on the drivers and supported options.
### **dir:**_path_
An existing local directory _path_ storing the manifest, layer tarballs and signatures as individual files.
This is a non-standardized format, primarily useful for debugging or noninvasive container inspection.
### **docker://**_docker-reference_
An image in a registry implementing the "Docker Registry HTTP API V2".
By default, uses the authorization state in `$XDG_RUNTIME_DIR/containers/auth.json`, which is set using podman-login(1).
If the authorization state is not found there, `$HOME/.docker/config.json` is checked, which is set using docker-login(1).
The containers-registries.conf(5) further allows for configuring various settings of a registry.
Note that a _docker-reference_ has the following format: `name[:tag|@digest]`.
While the docker transport does not support both a tag and a digest at the same time some formats like containers-storage do.
Digests can also be used in an image destination as long as the manifest matches the provided digest.
The digest of images can be explored with skopeo-inspect(1).
If `name` does not contain a slash, it is treated as `docker.io/library/name`.
Otherwise, the component before the first slash is checked if it is recognized as a `hostname[:port]` (i.e., it contains either a . or a :, or the component is exactly localhost).
If the first component of name is not recognized as a `hostname[:port]`, `name` is treated as `docker.io/name`.
### **docker-archive:**_path[:docker-reference]_
An image is stored in the docker-save(1) formatted file.
_docker-reference_ is only used when creating such a file, and it must not contain a digest.
It is further possible to copy data to stdin by specifying `docker-archive:/dev/stdin` but note that the used file must be seekable.
### **docker-daemon:**_docker-reference|algo:digest_
An image stored in the docker daemon's internal storage.
The image must be specified as a _docker-reference_ or in an alternative _algo:digest_ format when being used as an image source.
The _algo:digest_ refers to the image ID reported by docker-inspect(1).
### **oci:**_path[:tag]_
An image compliant with the "Open Container Image Layout Specification" at _path_.
Using a _tag_ is optional and allows for storing multiple images at the same _path_.
### **oci-archive:**_path[:tag]_
An image compliant with the "Open Container Image Layout Specification" stored as a tar(1) archive at _path_.
### **ostree:**_docker-reference[@/absolute/repo/path]_
An image in the local ostree(1) repository.
_/absolute/repo/path_ defaults to _/ostree/repo_.
## Examples
The following examples demonstrate how some of the containers transports can be used.
The examples use skopeo-copy(1) for copying container images.
**Copying an image from one registry to another**:
```
$ skopeo copy docker://docker.io/library/alpine:latest docker://localhost:5000/alpine:latest
```
**Copying an image from a running Docker daemon to a directory in the OCI layout**:
```
$ mkdir alpine-oci
$ skopeo copy docker-daemon:alpine:latest oci:alpine-oci
$ tree alpine-oci
test-oci/
├── blobs
│   └── sha256
│   ├── 83ef92b73cf4595aa7fe214ec6747228283d585f373d8f6bc08d66bebab531b7
│   ├── 9a6259e911dcd0a53535a25a9760ad8f2eded3528e0ad5604c4488624795cecc
│   └── ff8df268d29ccbe81cdf0a173076dcfbbea4bb2b6df1dd26766a73cb7b4ae6f7
├── index.json
└── oci-layout
2 directories, 5 files
```
**Copying an image from a registry to the local storage**:
```
$ skopeo copy docker://docker.io/library/alpine:latest containers-storage:alpine:latest
```
## SEE ALSO
docker-login(1), docker-save(1), ostree(1), podman-login(1), skopeo-copy(1), skopeo-inspect(1), tar(1), container-registries.conf(5), containers-storage.conf(5)
## AUTHORS
Miloslav Trmač <mitr@redhat.com>
Valentin Rothberg <rothberg@redhat.com>

136
vendor/github.com/containers/image/docs/signature-protocols.md generated vendored Normal file
View File

@ -0,0 +1,136 @@
# Signature access protocols
The `github.com/containers/image` library supports signatures implemented as blobs “attached to” an image.
Some image transports (local storage formats and remote procotocols) implement these signatures natively
or trivially; for others, the protocol extensions described below are necessary.
## docker/distribution registries—separate storage
### Usage
Any existing docker/distribution registry, whether or not it natively supports signatures,
can be augmented with separate signature storage by configuring a signature storage URL in [`registries.d`](registries.d.md).
`registries.d` can be configured to use one storage URL for a whole docker/distribution server,
or also separate URLs for smaller namespaces or individual repositories within the server
(which e.g. allows image authors to manage their own signature storage while publishing
the images on the public `docker.io` server).
The signature storage URL defines a root of a path hierarchy.
It can be either a `file:///…` URL, pointing to a local directory structure,
or a `http`/`https` URL, pointing to a remote server.
`file:///` signature storage can be both read and written, `http`/`https` only supports reading.
The same path hierarchy is used in both cases, so the HTTP/HTTPS server can be
a simple static web server serving a directory structure created by writing to a `file:///` signature storage.
(This of course does not prevent other server implementations,
e.g. a HTTP server reading signatures from a database.)
The usual workflow for producing and distributing images using the separate storage mechanism
is to configure the repository in `registries.d` with `sigstore-staging` URL pointing to a private
`file:///` staging area, and a `sigstore` URL pointing to a public web server.
To publish an image, the image author would sign the image as necessary (e.g. using `skopeo copy`),
and then copy the created directory structure from the `file:///` staging area
to a subdirectory of a webroot of the public web server so that they are accessible using the public `sigstore` URL.
The author would also instruct consumers of the image to, or provide a `registries.d` configuration file to,
set up a `sigstore` URL pointing to the public web server.
### Path structure
Given a _base_ signature storage URL configured in `registries.d` as mentioned above,
and a container image stored in a docker/distribution registry using the _fully-expanded_ name
_hostname_`/`_namespaces_`/`_name_{`@`_digest_,`:`_tag_} (e.g. for `docker.io/library/busybox:latest`,
_namespaces_ is `library`, even if the user refers to the image using the shorter syntax as `busybox:latest`),
signatures are accessed using URLs of the form
> _base_`/`_namespaces_`/`_name_`@`_digest-algo_`=`_digest-value_`/signature-`_index_
where _digest-algo_`:`_digest-value_ is a manifest digest usable for referencing the relevant image manifest
(i.e. even if the user referenced the image using a tag,
the signature storage is always disambiguated using digest references).
Note that in the URLs used for signatures,
_digest-algo_ and _digest-value_ are separated using the `=` character,
not `:` like when acessing the manifest using the docker/distribution API.
Within the URL, _index_ is a decimal integer (in the canonical form), starting with 1.
Signatures are stored at URLs with successive _index_ values; to read all of them, start with _index_=1,
and continue reading signatures and increasing _index_ as long as signatures with these _index_ values exist.
Similarly, to add one more signature to an image, find the first _index_ which does not exist, and
then store the new signature using that _index_ value.
There is no way to list existing signatures other than iterating through the successive _index_ values,
and no way to download all of the signatures at once.
### Examples
For a docker/distribution image available as `busybox@sha256:817a12c32a39bbe394944ba49de563e085f1d3c5266eb8e9723256bc4448680e`
(or as `busybox:latest` if the `latest` tag points to to a manifest with the same digest),
and with a `registries.d` configuration specifying a `sigstore` URL `https://example.com/sigstore` for the same image,
the following URLs would be accessed to download all signatures:
> - `https://example.com/sigstore/library/busybox@sha256=817a12c32a39bbe394944ba49de563e085f1d3c5266eb8e9723256bc4448680e/signature-1`
> - `https://example.com/sigstore/library/busybox@sha256=817a12c32a39bbe394944ba49de563e085f1d3c5266eb8e9723256bc4448680e/signature-2`
> - …
For a docker/distribution image available as `example.com/ns1/ns2/ns3/repo@somedigest:digestvalue` and the same
`sigstore` URL, the signatures would be available at
> `https://example.com/sigstore/ns1/ns2/ns3/repo@somedigest=digestvalue/signature-1`
and so on.
## (OpenShift) docker/distribution API extension
As of https://github.com/openshift/origin/pull/12504/ , the OpenShift-embedded registry also provides
an extension of the docker/distribution API which allows simpler access to the signatures,
using only the docker/distribution API endpoint.
This API is not inherently OpenShift-specific (e.g. the client does not need to know the OpenShift API endpoint,
and credentials sufficient to access the docker/distribution API server are sufficient to access signatures as well),
and it is the preferred way implement signature storage in registries.
See https://github.com/openshift/openshift-docs/pull/3556 for the upstream documentation of the API.
To read the signature, any user with access to an image can use the `/extensions/v2/…/signatures/…`
path to read an array of signatures. Use only the signature objects
which have `version` equal to `2`, `type` equal to `atomic`, and read the signature from `content`;
ignore the other fields of the signature object.
To add a single signature, `PUT` a new object with `version` set to `2`, `type` set to `atomic`,
and `content` set to the signature. Also set `name` to an unique name with the form
_digest_`@`_per-image-name_, where _digest_ is an image manifest digest (also used in the URL),
and _per-image-name_ is any unique identifier.
To add more than one signature, add them one at a time. This API does not allow deleting signatures.
Note that because signatures are stored within the cluster-wide image objects,
i.e. different namespaces can not associate different sets of signatures to the same image,
updating signatures requires a cluster-wide access to the `imagesignatures` resource
(by default available to the `system:image-signer` role),
## OpenShift-embedded registries
The OpenShift-embedded registry implements the ordinary docker/distribution API,
and it also exposes images through the OpenShift REST API (available through the “API master” servers).
Note: OpenShift versions 1.5 and later support the above-described [docker/distribution API extension](#openshift-dockerdistribution-api-extension),
which is easier to set up and should usually be preferred.
Continue reading for details on using older versions of OpenShift.
As of https://github.com/openshift/origin/pull/9181,
signatures are exposed through the OpenShift API
(i.e. to access the complete image, it is necessary to use both APIs,
in particular to know the URLs for both the docker/distribution and the OpenShift API master endpoints).
To read the signature, any user with access to an image can use the `imagestreamimages` namespaced
resource to read an `Image` object and its `Signatures` array. Use only the `ImageSignature` objects
which have `Type` equal to `atomic`, and read the signature from `Content`; ignore the other fields of
the `ImageSignature` object.
To add or remove signatures, use the cluster-wide (non-namespaced) `imagesignatures` resource,
with `Type` set to `atomic` and `Content` set to the signature. Signature names must have the form
_digest_`@`_per-image-name_, where _digest_ is an image manifest digest (OpenShift “image name”),
and _per-image-name_ is any unique identifier.
Note that because signatures are stored within the cluster-wide image objects,
i.e. different namespaces can not associate different sets of signatures to the same image,
updating signatures requires a cluster-wide access to the `imagesignatures` resource
(by default available to the `system:image-signer` role),
and deleting signatures is strongly discouraged
(it deletes the signature from all namespaces which contain the same image).

82
vendor/github.com/containers/image/registries.conf generated vendored Normal file
View File

@ -0,0 +1,82 @@
# For more information on this configuration file, see containers-registries.conf(5).
#
# There are multiple versions of the configuration syntax available, where the
# second iteration is backwards compatible to the first one. Mixing up both
# formats will result in an runtime error.
#
# The initial configuration format looks like this:
#
# Registries to search for images that are not fully-qualified.
# i.e. foobar.com/my_image:latest vs my_image:latest
[registries.search]
registries = []
# Registries that do not use TLS when pulling images or uses self-signed
# certificates.
[registries.insecure]
registries = []
# Blocked Registries, blocks the `docker daemon` from pulling from the blocked registry. If you specify
# "*", then the docker daemon will only be allowed to pull from registries listed above in the search
# registries. Blocked Registries is deprecated because other container runtimes and tools will not use it.
# It is recommended that you use the trust policy file /etc/containers/policy.json to control which
# registries you want to allow users to pull and push from. policy.json gives greater flexibility, and
# supports all container runtimes and tools including the docker daemon, cri-o, buildah ...
# The atomic CLI `atomic trust` can be used to easily configure the policy.json file.
[registries.block]
registries = []
# The second version of the configuration format allows to specify registry
# mirrors:
#
# # An array of host[:port] registries to try when pulling an unqualified image, in order.
# unqualified-search-registries = ["example.com"]
#
# [[registry]]
# # The "prefix" field is used to choose the relevant [[registry]] TOML table;
# # (only) the TOML table with the longest match for the input image name
# # (taking into account namespace/repo/tag/digest separators) is used.
# #
# # If the prefix field is missing, it defaults to be the same as the "location" field.
# prefix = "example.com/foo"
#
# # If true, unencrypted HTTP as well as TLS connections with untrusted
# # certificates are allowed.
# insecure = false
#
# # If true, pulling images with matching names is forbidden.
# blocked = false
#
# # The physical location of the "prefix"-rooted namespace.
# #
# # By default, this equal to "prefix" (in which case "prefix" can be omitted
# # and the [[registry]] TOML table can only specify "location").
# #
# # Example: Given
# # prefix = "example.com/foo"
# # location = "internal-registry-for-example.net/bar"
# # requests for the image example.com/foo/myimage:latest will actually work with the
# # internal-registry-for-example.net/bar/myimage:latest image.
# location = internal-registry-for-example.com/bar"
#
# # (Possibly-partial) mirrors for the "prefix"-rooted namespace.
# #
# # The mirrors are attempted in the specified order; the first one that can be
# # contacted and contains the image will be used (and if none of the mirrors contains the image,
# # the primary location specified by the "registry.location" field, or using the unmodified
# # user-specified reference, is tried last).
# #
# # Each TOML table in the "mirror" array can contain the following fields, with the same semantics
# # as if specified in the [[registry]] TOML table directly:
# # - location
# # - insecure
# [[registry.mirror]]
# location = "example-mirror-0.local/mirror-for-foo"
# [[registry.mirror]]
# location = "example-mirror-1.local/mirrors/foo"
# insecure = true
# # Given the above, a pull of example.com/foo/image:latest will try:
# # 1. example-mirror-0.local/mirror-for-foo/image:latest
# # 2. example-mirror-1.local/mirrors/foo/image:latest
# # 3. internal-registry-for-example.net/bar/myimage:latest
# # in order, and use the first one that exists.