Compare commits

...

44 Commits

Author SHA1 Message Date
Sebastiaan van Stijn
852759a7df Merge commit from fork
[24.0] AuthZ plugin security fixes
2024-07-23 21:36:28 +02:00
Jameson Hyde
55fa16e22f If url includes scheme, urlPath will drop hostname, which would not match the auth check
Signed-off-by: Jameson Hyde <jameson.hyde@docker.com>
(cherry picked from commit 754fb8d9d03895ae3ab60d2ad778152b0d835206)
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
(cherry picked from commit 5282cb25d0)
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
2024-07-17 13:10:06 +02:00
Jameson Hyde
f91551869e Authz plugin security fixes for 0-length content and path validation
Signed-off-by: Jameson Hyde <jameson.hyde@docker.com>

fix comments

(cherry picked from commit 9659c3a52bac57e615b5fb49b0652baca448643e)
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
(cherry picked from commit 2ac8a479c5)
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
2024-07-17 13:10:04 +02:00
Sebastiaan van Stijn
e0ce0ff8d5 Merge pull request #47779 from AkihiroSuda/cherrypick-createmountpoint-24
[24.0] mounts/validate: Don't check source exists with CreateMountpoint
2024-04-30 15:43:23 +02:00
Paweł Gronowski
02aa896934 mounts/validate: Don't check source exists with CreateMountpoint
Don't error out when mount source doesn't exist and mounts has
`CreateMountpoint` option enabled.

Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
(cherry picked from commit 05b883bdc8)
Signed-off-by: Akihiro Suda <akihiro.suda.cz@hco.ntt.co.jp>
2024-04-30 20:45:24 +09:00
Paweł Gronowski
e0b4fc1aae Merge pull request #47698 from vvoland/v24.0-47658
[24.0 backport] Fix cases where we are wrapping a nil error
2024-04-09 13:55:20 +02:00
Brian Goff
e264217787 Fix cases where we are wrapping a nil error
This was using `errors.Wrap` when there was no error to wrap, meanwhile
we are supposed to be creating a new error.

Found this while investigating some log corruption issues and
unexpectedly getting a nil reader and a nil error from `getTailReader`.

Signed-off-by: Brian Goff <cpuguy83@gmail.com>
(cherry picked from commit 0a48d26fbc)
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
2024-04-09 10:21:05 +02:00
Sebastiaan van Stijn
e633d64982 Merge pull request #47534 from vvoland/v24.0-47530
[24.0 backport] volume: Don't decrement refcount below 0
2024-03-11 15:58:15 +01:00
Paweł Gronowski
76b88fb5ff volume: Don't decrement refcount below 0
With both rootless and live restore enabled, there's some race condition
which causes the container to be `Unmount`ed before the refcount is
restored.

This makes sure we don't underflow the refcount (uint64) when
decrementing it.

The root cause of this race condition still needs to be investigated and
fixed, but at least this unflakies the `TestLiveRestore`.

Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
(cherry picked from commit 294fc9762e)
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
2024-03-08 12:52:35 +01:00
Bjorn Neergaard
b8bc11af70 Merge pull request #47528 from vvoland/v24.0-47523
[24.0 backport] builder-next: fix missing lock in ensurelayer
2024-03-07 07:08:57 -07:00
Tonis Tiigi
8ffaef6d61 builder-next: fix missing lock in ensurelayer
When this was called concurrently from the moby image
exporter there could be a data race where a layer was
written to the refs map when it was already there.

In that case the reference count got mixed up and on
release only one of these layers was actually released.

Signed-off-by: Tonis Tiigi <tonistiigi@gmail.com>
(cherry picked from commit 37545cc644)
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
2024-03-07 12:30:33 +01:00
Sebastiaan van Stijn
69b5ded97b Merge pull request #47514 from vvoland/v24.0-47498
[24.0 backport] daemon: overlay2: remove world writable permission from the lower file
2024-03-06 21:38:45 +01:00
Jaroslav Jindrak
257d4ec79a daemon: overlay2: remove world writable permission from the lower file
In de2447c, the creation of the 'lower' file was changed from using
os.Create to using ioutils.AtomicWriteFile, which ignores the system's
umask. This means that even though the requested permission in the
source code was always 0666, it was 0644 on systems with default
umask of 0022 prior to de2447c, so the move to AtomicFile potentially
increased the file's permissions.

This is not a security issue because the parent directory does not
allow writes into the file, but it can confuse security scanners on
Linux-based systems into giving false positives.

Signed-off-by: Jaroslav Jindrak <dzejrou@gmail.com>
(cherry picked from commit cadb124ab6)
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
2024-03-06 13:13:39 +01:00
Sebastiaan van Stijn
c593074455 Merge pull request #47338 from vvoland/cache-fix-older-windows-24
[24.0 backport] image/cache: Ignore Build and Revision on Windows
2024-02-16 17:01:14 +01:00
Akihiro Suda
f11b55ffec Merge pull request #47345 from thaJeztah/24.0_backport_seccomp_updates
[24.0 backport] profiles/seccomp: add syscalls for kernel v5.17 - v6.6, match containerd's profile
2024-02-07 19:48:11 +09:00
Paweł Gronowski
b3888ed899 image/cache: Require Major and Minor match for Windows OSVersion
The platform comparison was backported from the branch that vendors
containerd 1.7.

In this branch the vendored containerd version is older and doesn't have
the same comparison logic for Windows specific OSVersion.

Require both major and minor components of Windows OSVersion to match.

Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
2024-02-06 17:52:03 +01:00
Sebastiaan van Stijn
a162f3c0f4 seccomp: add futex_wake syscall (kernel v6.7, libseccomp v2.5.5)
Add this syscall to match the profile in containerd

containerd: a6e52c74fa
libseccomp: 53267af3fb
kernel: 9f6c532f59

    futex: Add sys_futex_wake()

    To complement sys_futex_waitv() add sys_futex_wake(). This syscall
    implements what was previously known as FUTEX_WAKE_BITSET except it
    uses 'unsigned long' for the bitmask and takes FUTEX2 flags.

    The 'unsigned long' allows FUTEX2_SIZE_U64 on 64bit platforms.

Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
(cherry picked from commit d69729e053)
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2024-02-06 15:27:49 +01:00
Sebastiaan van Stijn
4977af06ee seccomp: add futex_wait syscall (kernel v6.7, libseccomp v2.5.5)
Add this syscall to match the profile in containerd

containerd: a6e52c74fa
libseccomp: 53267af3fb
kernel: cb8c4312af

    futex: Add sys_futex_wait()

    To complement sys_futex_waitv()/wake(), add sys_futex_wait(). This
    syscall implements what was previously known as FUTEX_WAIT_BITSET
    except it uses 'unsigned long' for the value and bitmask arguments,
    takes timespec and clockid_t arguments for the absolute timeout and
    uses FUTEX2 flags.

    The 'unsigned long' allows FUTEX2_SIZE_U64 on 64bit platforms.

Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
(cherry picked from commit 10d344d176)
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2024-02-06 15:27:49 +01:00
Sebastiaan van Stijn
f87c7381b4 seccomp: add futex_requeue syscall (kernel v6.7, libseccomp v2.5.5)
Add this syscall to match the profile in containerd

containerd: a6e52c74fa
libseccomp: 53267af3fb
kernel: 0f4b5f9722

    futex: Add sys_futex_requeue()

    Finish off the 'simple' futex2 syscall group by adding
    sys_futex_requeue(). Unlike sys_futex_{wait,wake}() its arguments are
    too numerous to fit into a regular syscall. As such, use struct
    futex_waitv to pass the 'source' and 'destination' futexes to the
    syscall.

    This syscall implements what was previously known as FUTEX_CMP_REQUEUE
    and uses {val, uaddr, flags} for source and {uaddr, flags} for
    destination.

    This design explicitly allows requeueing between different types of
    futex by having a different flags word per uaddr.

Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
(cherry picked from commit df57a080b6)
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2024-02-06 15:27:49 +01:00
Sebastiaan van Stijn
12627c2449 seccomp: add map_shadow_stack syscall (kernel v6.6, libseccomp v2.5.5)
Add this syscall to match the profile in containerd

containerd: a6e52c74fa
libseccomp: 53267af3fb
kernel: c35559f94e

    x86/shstk: Introduce map_shadow_stack syscall

    When operating with shadow stacks enabled, the kernel will automatically
    allocate shadow stacks for new threads, however in some cases userspace
    will need additional shadow stacks. The main example of this is the
    ucontext family of functions, which require userspace allocating and
    pivoting to userspace managed stacks.

    Unlike most other user memory permissions, shadow stacks need to be
    provisioned with special data in order to be useful. They need to be setup
    with a restore token so that userspace can pivot to them via the RSTORSSP
    instruction. But, the security design of shadow stacks is that they
    should not be written to except in limited circumstances. This presents a
    problem for userspace, as to how userspace can provision this special
    data, without allowing for the shadow stack to be generally writable.

Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
(cherry picked from commit 8826f402f9)
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2024-02-06 15:27:49 +01:00
Sebastiaan van Stijn
1ea6c4be57 seccomp: add fchmodat2 syscall (kernel v6.6, libseccomp v2.5.5)
Add this syscall to match the profile in containerd

containerd: a6e52c74fa
libseccomp: 53267af3fb
kernel: 09da082b07

    fs: Add fchmodat2()

    On the userspace side fchmodat(3) is implemented as a wrapper
    function which implements the POSIX-specified interface. This
    interface differs from the underlying kernel system call, which does not
    have a flags argument. Most implementations require procfs [1][2].

    There doesn't appear to be a good userspace workaround for this issue
    but the implementation in the kernel is pretty straight-forward.

    The new fchmodat2() syscall allows to pass the AT_SYMLINK_NOFOLLOW flag,
    unlike existing fchmodat.

Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
(cherry picked from commit 6f242f1a28)
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2024-02-06 15:27:48 +01:00
Sebastiaan van Stijn
4f26525b13 seccomp: add cachestat syscall (kernel v6.5, libseccomp v2.5.5)
Add this syscall to match the profile in containerd

containerd: a6e52c74fa
libseccomp: 53267af3fb
kernel: cf264e1329

    NAME
        cachestat - query the page cache statistics of a file.

    SYNOPSIS
        #include <sys/mman.h>

        struct cachestat_range {
            __u64 off;
            __u64 len;
        };

        struct cachestat {
            __u64 nr_cache;
            __u64 nr_dirty;
            __u64 nr_writeback;
            __u64 nr_evicted;
            __u64 nr_recently_evicted;
        };

        int cachestat(unsigned int fd, struct cachestat_range *cstat_range,
            struct cachestat *cstat, unsigned int flags);

    DESCRIPTION
        cachestat() queries the number of cached pages, number of dirty
        pages, number of pages marked for writeback, number of evicted
        pages, number of recently evicted pages, in the bytes range given by
        `off` and `len`.

Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
(cherry picked from commit 4d0d5ee10d)
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2024-02-06 15:27:48 +01:00
Sebastiaan van Stijn
00f3d189e4 seccomp: add set_mempolicy_home_node syscall (kernel v5.17, libseccomp v2.5.4)
This syscall is gated by CAP_SYS_NICE, matching the profile in containerd.

containerd: a6e52c74fa
libseccomp: d83cb7ac25
kernel: c6018b4b25

    mm/mempolicy: add set_mempolicy_home_node syscall
    This syscall can be used to set a home node for the MPOL_BIND and
    MPOL_PREFERRED_MANY memory policy.  Users should use this syscall after
    setting up a memory policy for the specified range as shown below.

      mbind(p, nr_pages * page_size, MPOL_BIND, new_nodes->maskp,
            new_nodes->size + 1, 0);
      sys_set_mempolicy_home_node((unsigned long)p, nr_pages * page_size,
                    home_node, 0);

    The syscall allows specifying a home node/preferred node from which
    kernel will fulfill memory allocation requests first.
    ...

Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
(cherry picked from commit 1251982cf7)
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2024-02-06 15:27:48 +01:00
Paweł Gronowski
f110a9d310 image/cache: Use Platform from ocispec
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
(cherry picked from commit 2c01d53d96)
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
2024-02-06 14:30:06 +01:00
Paweł Gronowski
b81579adf4 image/cache: Ignore Build and Revision on Windows
The compatibility depends on whether `hyperv` or `process` container
isolation is used.
This fixes cache not being used when building images based on older
Windows versions on a newer Windows host.

Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
(cherry picked from commit 91ea04089b)
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
2024-02-06 13:15:58 +01:00
Sebastiaan van Stijn
9d3a6a2133 Merge pull request #47324 from thaJeztah/24.0_backport_plugin-install-digest
[24.0 backport] plugins: Fix panic when fetching by digest
2024-02-05 11:08:19 +01:00
Laura Brehm
f406728e41 plugins: fix panic installing from repo w/ digest
Only print the tag when the received reference has a tag, if
we can't cast the received tag to a `reference.Tagged` then
skip printing the tag as it's likely a digest.

Fixes panic when trying to install a plugin from a reference
with a digest such as
`vieux/sshfs@sha256:1d3c3e42c12138da5ef7873b97f7f32cf99fb6edde75fa4f0bcf9ed277855811`

Signed-off-by: Laura Brehm <laurabrehm@hey.com>
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2024-02-05 10:11:38 +01:00
Laura Brehm
6f483e7dbb tests: add plugin install test w/ digest
Adds a test case for installing a plugin from a remote in the form
of `plugin-content-trust@sha256:d98f2f8061...`, which is currently
causing the daemon to panic, as we found while running the CLI e2e
tests:

```
docker plugin install registry:5000/plugin-content-trust@sha256:d98f2f806144bf4ba62d4ecaf78fec2f2fe350df5a001f6e3b491c393326aedb
```

Signed-off-by: Laura Brehm <laurabrehm@hey.com>
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2024-02-05 10:11:30 +01:00
Sebastiaan van Stijn
a9f4dd168e Merge pull request #47314 from thaJeztah/24.0_backport_libc8d_fix_startup_data_race
[24.0 backport] libcontainerd/supervisor: fix data race
2024-02-03 14:38:17 +01:00
Cory Snider
d22068f8e3 libcontainerd/supervisor: fix data race
The monitorDaemon() goroutine calls startContainerd() then blocks on
<-daemonWaitCh to wait for it to exit. The startContainerd() function
would (re)initialize the daemonWaitCh so a restarted containerd could be
waited on. This implementation was race-free because startContainerd()
would synchronously initialize the daemonWaitCh before returning. When
the call to start the managed containerd process was moved into the
waiter goroutine, the code to initialize the daemonWaitCh struct field
was also moved into the goroutine. This introduced a race condition.

Move the daemonWaitCh initialization to guarantee that it happens before
the startContainerd() call returns.

Signed-off-by: Cory Snider <csnider@mirantis.com>
(cherry picked from commit dd20bf4862)
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2024-02-03 11:41:08 +01:00
Sebastiaan van Stijn
fca702de7f Merge pull request from GHSA-xw73-rw38-6vjc
[24.0 backport] image/cache: Restrict cache candidates to locally built images
2024-02-01 01:12:23 +01:00
Sebastiaan van Stijn
f78a7726d7 Merge pull request #47281 from thaJeztah/24.0_backport_bump_containerd_binary_1.7.13
[24.0 backport] update containerd binary to v1.7.13
2024-01-31 22:52:15 +01:00
Sebastiaan van Stijn
61afffeeb3 Merge pull request #47270 from thaJeztah/24.0_backport_bump_runc_binary_1.1.12
[24.0 backport] update runc binary to v1.1.12
2024-01-31 22:30:39 +01:00
Sebastiaan van Stijn
b38e74c4e0 Merge pull request #47276 from thaJeztah/24.0_backport_bump_runc_1.1.12
[24.0 backport] vendor: github.com/opencontainers/runc v1.1.12
2024-01-31 22:14:37 +01:00
Sebastiaan van Stijn
dac56638ad update containerd binary to v1.7.13
Update the containerd binary that's used in CI

- full diff: https://github.com/containerd/containerd/compare/v1.7.12...v1.7.13
- release notes: https://github.com/containerd/containerd/releases/tag/v1.7.13

Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
(cherry picked from commit 835cdcac95)
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2024-01-31 22:05:33 +01:00
Sebastiaan van Stijn
20e1af3616 vendor: github.com/opencontainers/runc v1.1.12
- release notes: https://github.com/opencontainers/runc/releases/tag/v1.1.12
- full diff: https://github.com/opencontainers/runc/compare/v1.1.11...v1.1.12

Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
(cherry picked from commit b20dccba5e)
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2024-01-31 21:21:21 +01:00
Sebastiaan van Stijn
858919d399 update runc binary to v1.1.12
Update the runc binary that's used in CI and for the static packages, which
includes a fix for [CVE-2024-21626].

- release notes: https://github.com/opencontainers/runc/releases/tag/v1.1.12
- full diff: https://github.com/opencontainers/runc/compare/v1.1.11...v1.1.12

[CVE-2024-21626]: https://github.com/opencontainers/runc/security/advisories/GHSA-xr7r-f8xq-vfvv

Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
(cherry picked from commit 44bf407d4d)
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
2024-01-31 21:06:29 +01:00
Sebastiaan van Stijn
141ad39e38 Merge pull request #47266 from vvoland/ci-fix-makeps1-templatefail-24
[24.0 backport] hack/make.ps1: Fix go list pattern
2024-01-31 21:02:50 +01:00
Paweł Gronowski
db968c672b hack/make.ps1: Fix go list pattern
The double quotes inside a single quoted string don't need to be
escaped.
Looks like different Powershell versions are treating this differently
and it started failing unexpectedly without any changes on our side.

Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
(cherry picked from commit ecb217cf69)
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
2024-01-31 19:57:19 +01:00
Paweł Gronowski
44e6f3da60 builder/windows: Don't set ArgsEscaped for RUN cache probe
Previously this was done indirectly - the `compare` function didn't
check the `ArgsEscaped`.

Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
(cherry picked from commit 96d461d27e)
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
2024-01-25 16:21:39 +01:00
Paweł Gronowski
8a19bb7193 image/cache: Check image platform
Make sure the cache candidate platform matches the requested.

Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
(cherry picked from commit 877ebbe038)
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
2024-01-25 16:21:37 +01:00
Paweł Gronowski
17af50f46b image/cache: Restrict cache candidates to locally built images
Restrict cache candidates only to images that were built locally.
This doesn't affect builds using `--cache-from`.

Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
(cherry picked from commit 96ac22768a)
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
2024-01-25 16:21:36 +01:00
Paweł Gronowski
ffb63c0bae daemon/imageStore: Mark images built locally
Store additional image property which makes it possible to distinguish
if image was built locally.

Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
(cherry picked from commit c6156dc51b)
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
2024-01-25 16:21:35 +01:00
Paweł Gronowski
593b754d8f image/cache: Compare all config fields
Add checks for some image config fields that were missing.

Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
(cherry picked from commit 537348763f)
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
2024-01-25 16:21:32 +01:00
37 changed files with 689 additions and 124 deletions

View File

@@ -197,7 +197,7 @@ RUN git init . && git remote add origin "https://github.com/containerd/container
# When updating the binary version you may also need to update the vendor
# version to pick up bug fixes or new APIs, however, usually the Go packages
# are built from a commit from the master branch.
ARG CONTAINERD_VERSION=v1.7.12
ARG CONTAINERD_VERSION=v1.7.13
RUN git fetch -q --depth 1 origin "${CONTAINERD_VERSION}" +refs/tags/*:refs/tags/* && git checkout -q FETCH_HEAD
FROM base AS containerd-build
@@ -280,7 +280,7 @@ RUN git init . && git remote add origin "https://github.com/opencontainers/runc.
# that is used. If you need to update runc, open a pull request in the containerd
# project first, and update both after that is merged. When updating RUNC_VERSION,
# consider updating runc in vendor.mod accordingly.
ARG RUNC_VERSION=v1.1.11
ARG RUNC_VERSION=v1.1.12
RUN git fetch -q --depth 1 origin "${RUNC_VERSION}" +refs/tags/*:refs/tags/* && git checkout -q FETCH_HEAD
FROM base AS runc-build

View File

@@ -168,7 +168,7 @@ SHELL ["powershell", "-Command", "$ErrorActionPreference = 'Stop'; $ProgressPref
ARG GO_VERSION=1.20.13
ARG GOTESTSUM_VERSION=v1.8.2
ARG GOWINRES_VERSION=v0.3.1
ARG CONTAINERD_VERSION=v1.7.12
ARG CONTAINERD_VERSION=v1.7.13
# Environment variable notes:
# - GO_VERSION must be consistent with 'Dockerfile' used by Linux.

View File

@@ -22,6 +22,9 @@ func (s *snapshotter) GetDiffIDs(ctx context.Context, key string) ([]layer.DiffI
}
func (s *snapshotter) EnsureLayer(ctx context.Context, key string) ([]layer.DiffID, error) {
s.layerCreateLocker.Lock(key)
defer s.layerCreateLocker.Unlock(key)
diffIDs, err := s.GetDiffIDs(ctx, key)
if err != nil {
return nil, err

View File

@@ -16,6 +16,7 @@ import (
"github.com/docker/docker/pkg/idtools"
"github.com/moby/buildkit/identity"
"github.com/moby/buildkit/snapshot"
"github.com/moby/locker"
"github.com/opencontainers/go-digest"
"github.com/pkg/errors"
bolt "go.etcd.io/bbolt"
@@ -48,10 +49,11 @@ type checksumCalculator interface {
type snapshotter struct {
opt Opt
refs map[string]layer.Layer
db *bolt.DB
mu sync.Mutex
reg graphIDRegistrar
refs map[string]layer.Layer
db *bolt.DB
mu sync.Mutex
reg graphIDRegistrar
layerCreateLocker *locker.Locker
}
// NewSnapshotter creates a new snapshotter
@@ -68,10 +70,11 @@ func NewSnapshotter(opt Opt, prevLM leases.Manager) (snapshot.Snapshotter, lease
}
s := &snapshotter{
opt: opt,
db: db,
refs: map[string]layer.Layer{},
reg: reg,
opt: opt,
db: db,
refs: map[string]layer.Layer{},
reg: reg,
layerCreateLocker: locker.New(),
}
lm := newLeaseManager(s, prevLM)

View File

@@ -15,6 +15,7 @@ import (
"github.com/docker/docker/image"
"github.com/docker/docker/layer"
"github.com/opencontainers/go-digest"
ocispec "github.com/opencontainers/image-spec/specs-go/v1"
)
const (
@@ -89,7 +90,7 @@ type ImageCacheBuilder interface {
type ImageCache interface {
// GetCache returns a reference to a cached image whose parent equals `parent`
// and runconfig equals `cfg`. A cache miss is expected to return an empty ID and a nil error.
GetCache(parentID string, cfg *container.Config) (imageID string, err error)
GetCache(parentID string, cfg *container.Config, platform ocispec.Platform) (imageID string, err error)
}
// Image represents a Docker image used by the builder.

View File

@@ -8,7 +8,6 @@ import (
"net/url"
"os"
"path/filepath"
"runtime"
"sort"
"strings"
"time"
@@ -74,7 +73,7 @@ type copier struct {
source builder.Source
pathCache pathCache
download sourceDownloader
platform *ocispec.Platform
platform ocispec.Platform
// for cleanup. TODO: having copier.cleanup() is error prone and hard to
// follow. Code calling performCopy should manage the lifecycle of its params.
// Copier should take override source as input, not imageMount.
@@ -83,19 +82,7 @@ type copier struct {
}
func copierFromDispatchRequest(req dispatchRequest, download sourceDownloader, imageSource *imageMount) copier {
platform := req.builder.platform
if platform == nil {
// May be nil if not explicitly set in API/dockerfile
platform = &ocispec.Platform{}
}
if platform.OS == "" {
// Default to the dispatch requests operating system if not explicit in API/dockerfile
platform.OS = req.state.operatingSystem
}
if platform.OS == "" {
// This is a failsafe just in case. Shouldn't be hit.
platform.OS = runtime.GOOS
}
platform := req.builder.getPlatform(req.state)
return copier{
source: req.source,

View File

@@ -349,9 +349,16 @@ func dispatchRun(ctx context.Context, d dispatchRequest, c *instructions.RunComm
saveCmd = prependEnvOnCmd(d.state.buildArgs, buildArgs, cmdFromArgs)
}
cacheArgsEscaped := argsEscaped
// ArgsEscaped is not persisted in the committed image on Windows.
// Use the original from previous build steps for cache probing.
if d.state.operatingSystem == "windows" {
cacheArgsEscaped = stateRunConfig.ArgsEscaped
}
runConfigForCacheProbe := copyRunConfig(stateRunConfig,
withCmd(saveCmd),
withArgsEscaped(argsEscaped),
withArgsEscaped(cacheArgsEscaped),
withEntrypointOverride(saveCmd, nil))
if hit, err := d.builder.probeCache(d.state, runConfigForCacheProbe); err != nil || hit {
return err

View File

@@ -5,6 +5,7 @@ import (
"github.com/docker/docker/api/types/container"
"github.com/docker/docker/builder"
ocispec "github.com/opencontainers/image-spec/specs-go/v1"
"github.com/sirupsen/logrus"
)
@@ -12,7 +13,7 @@ import (
// cache.
type ImageProber interface {
Reset(ctx context.Context) error
Probe(parentID string, runConfig *container.Config) (string, error)
Probe(parentID string, runConfig *container.Config, platform ocispec.Platform) (string, error)
}
type resetFunc func(context.Context) (builder.ImageCache, error)
@@ -51,11 +52,11 @@ func (c *imageProber) Reset(ctx context.Context) error {
// Probe checks if cache match can be found for current build instruction.
// It returns the cachedID if there is a hit, and the empty string on miss
func (c *imageProber) Probe(parentID string, runConfig *container.Config) (string, error) {
func (c *imageProber) Probe(parentID string, runConfig *container.Config, platform ocispec.Platform) (string, error) {
if c.cacheBusted {
return "", nil
}
cacheID, err := c.cache.GetCache(parentID, runConfig)
cacheID, err := c.cache.GetCache(parentID, runConfig, platform)
if err != nil {
return "", err
}
@@ -74,6 +75,6 @@ func (c *nopProber) Reset(ctx context.Context) error {
return nil
}
func (c *nopProber) Probe(_ string, _ *container.Config) (string, error) {
func (c *nopProber) Probe(_ string, _ *container.Config, _ ocispec.Platform) (string, error) {
return "", nil
}

View File

@@ -10,6 +10,7 @@ import (
"fmt"
"strings"
"github.com/containerd/containerd/platforms"
"github.com/docker/docker/api/types"
"github.com/docker/docker/api/types/backend"
"github.com/docker/docker/api/types/container"
@@ -328,7 +329,7 @@ func getShell(c *container.Config, os string) []string {
}
func (b *Builder) probeCache(dispatchState *dispatchState, runConfig *container.Config) (bool, error) {
cachedID, err := b.imageProber.Probe(dispatchState.imageID, runConfig)
cachedID, err := b.imageProber.Probe(dispatchState.imageID, runConfig, b.getPlatform(dispatchState))
if cachedID == "" || err != nil {
return false, err
}
@@ -388,3 +389,17 @@ func hostConfigFromOptions(options *types.ImageBuildOptions) *container.HostConf
}
return hc
}
func (b *Builder) getPlatform(state *dispatchState) ocispec.Platform {
// May be nil if not explicitly set in API/dockerfile
out := platforms.DefaultSpec()
if b.platform != nil {
out = *b.platform
}
if state.operatingSystem != "" {
out.OS = state.operatingSystem
}
return out
}

View File

@@ -14,6 +14,7 @@ import (
"github.com/docker/docker/image"
"github.com/docker/docker/layer"
"github.com/opencontainers/go-digest"
ocispec "github.com/opencontainers/image-spec/specs-go/v1"
)
// MockBackend implements the builder.Backend interface for unit testing
@@ -111,7 +112,7 @@ type mockImageCache struct {
getCacheFunc func(parentID string, cfg *container.Config) (string, error)
}
func (mic *mockImageCache) GetCache(parentID string, cfg *container.Config) (string, error) {
func (mic *mockImageCache) GetCache(parentID string, cfg *container.Config, _ ocispec.Platform) (string, error) {
if mic.getCacheFunc != nil {
return mic.getCacheFunc(parentID, cfg)
}

View File

@@ -8,7 +8,9 @@ import (
"github.com/docker/docker/api/types/container"
imagetype "github.com/docker/docker/api/types/image"
"github.com/docker/docker/builder"
"github.com/docker/docker/errdefs"
"github.com/docker/docker/image"
ocispec "github.com/opencontainers/image-spec/specs-go/v1"
)
// MakeImageCache creates a stateful image cache.
@@ -29,7 +31,7 @@ type imageCache struct {
c *ImageService
}
func (ic *imageCache) GetCache(parentID string, cfg *container.Config) (imageID string, err error) {
func (ic *imageCache) GetCache(parentID string, cfg *container.Config, platform ocispec.Platform) (imageID string, err error) {
ctx := context.TODO()
if parentID == "" {
@@ -37,8 +39,11 @@ func (ic *imageCache) GetCache(parentID string, cfg *container.Config) (imageID
return "", nil
}
parent, err := ic.c.GetImage(ctx, parentID, imagetype.GetImageOpts{})
parent, err := ic.c.GetImage(ctx, parentID, imagetype.GetImageOpts{Platform: &platform})
if err != nil {
if errdefs.IsNotFound(err) {
return "", nil
}
return "", err
}
@@ -54,8 +59,11 @@ func (ic *imageCache) GetCache(parentID string, cfg *container.Config) (imageID
}
for _, children := range children {
childImage, err := ic.c.GetImage(ctx, children.String(), imagetype.GetImageOpts{})
childImage, err := ic.c.GetImage(ctx, children.String(), imagetype.GetImageOpts{Platform: &platform})
if err != nil {
if errdefs.IsNotFound(err) {
continue
}
return "", err
}

View File

@@ -409,7 +409,7 @@ func (d *Driver) create(id, parent string, opts *graphdriver.CreateOpts) (retErr
return err
}
if lower != "" {
if err := ioutils.AtomicWriteFile(path.Join(dir, lowerFile), []byte(lower), 0o666); err != nil {
if err := ioutils.AtomicWriteFile(path.Join(dir, lowerFile), []byte(lower), 0o644); err != nil {
return err
}
}

View File

@@ -257,6 +257,9 @@ func (i *ImageService) CreateImage(ctx context.Context, config []byte, parent st
return nil, errors.Wrapf(err, "failed to set parent %s", parent)
}
}
if err := i.imageStore.SetBuiltLocally(id); err != nil {
return nil, errors.Wrapf(err, "failed to mark image %s as built locally", id)
}
return i.imageStore.Get(id)
}

View File

@@ -62,6 +62,9 @@ func (i *ImageService) CommitImage(ctx context.Context, c backend.CommitConfig)
if err != nil {
return "", err
}
if err := i.imageStore.SetBuiltLocally(id); err != nil {
return "", err
}
if c.ParentImageID != "" {
if err := i.imageStore.SetParent(id, image.ID(c.ParentImageID)); err != nil {

View File

@@ -66,7 +66,7 @@ func getTailReader(ctx context.Context, r loggerutils.SizeReaderAt, req int) (io
}
if msgLen != binary.BigEndian.Uint32(buf) {
return nil, 0, errdefs.DataLoss(errors.Wrap(err, "log message header and footer indicate different message sizes"))
return nil, 0, errdefs.DataLoss(errors.New("log message header and footer indicate different message sizes"))
}
found++

View File

@@ -15,7 +15,7 @@ set -e
# the binary version you may also need to update the vendor version to pick up
# bug fixes or new APIs, however, usually the Go packages are built from a
# commit from the master branch.
: "${CONTAINERD_VERSION:=v1.7.12}"
: "${CONTAINERD_VERSION:=v1.7.13}"
install_containerd() (
echo "Install containerd version $CONTAINERD_VERSION"

View File

@@ -9,7 +9,7 @@ set -e
# the containerd project first, and update both after that is merged.
#
# When updating RUNC_VERSION, consider updating runc in vendor.mod accordingly
: "${RUNC_VERSION:=v1.1.11}"
: "${RUNC_VERSION:=v1.1.12}"
install_runc() {
RUNC_BUILDTAGS="${RUNC_BUILDTAGS:-"seccomp"}"

View File

@@ -348,7 +348,7 @@ Function Run-UnitTests() {
Function Run-IntegrationTests() {
$escRoot = [Regex]::Escape($root)
$env:DOCKER_INTEGRATION_DAEMON_DEST = $bundlesDir + "\tmp"
$dirs = go list -test -f '{{- if ne .ForTest `"`" -}}{{- .Dir -}}{{- end -}}' .\integration\...
$dirs = go list -test -f '{{- if ne .ForTest "" -}}{{- .Dir -}}{{- end -}}' .\integration\...
ForEach($dir in $dirs) {
# Normalize directory name for using in the test results files.
$normDir = $dir.Trim()

82
image/cache/cache.go vendored
View File

@@ -10,7 +10,9 @@ import (
"github.com/docker/docker/dockerversion"
"github.com/docker/docker/image"
"github.com/docker/docker/layer"
ocispec "github.com/opencontainers/image-spec/specs-go/v1"
"github.com/pkg/errors"
"github.com/sirupsen/logrus"
)
// NewLocal returns a local image cache, based on parent chain
@@ -26,8 +28,8 @@ type LocalImageCache struct {
}
// GetCache returns the image id found in the cache
func (lic *LocalImageCache) GetCache(imgID string, config *containertypes.Config) (string, error) {
return getImageIDAndError(getLocalCachedImage(lic.store, image.ID(imgID), config))
func (lic *LocalImageCache) GetCache(imgID string, config *containertypes.Config, platform ocispec.Platform) (string, error) {
return getImageIDAndError(getLocalCachedImage(lic.store, image.ID(imgID), config, platform))
}
// New returns an image cache, based on history objects
@@ -51,8 +53,8 @@ func (ic *ImageCache) Populate(image *image.Image) {
}
// GetCache returns the image id found in the cache
func (ic *ImageCache) GetCache(parentID string, cfg *containertypes.Config) (string, error) {
imgID, err := ic.localImageCache.GetCache(parentID, cfg)
func (ic *ImageCache) GetCache(parentID string, cfg *containertypes.Config, platform ocispec.Platform) (string, error) {
imgID, err := ic.localImageCache.GetCache(parentID, cfg, platform)
if err != nil {
return "", err
}
@@ -215,7 +217,23 @@ func getImageIDAndError(img *image.Image, err error) (string, error) {
// of the image with imgID, that had the same config when it was
// created. nil is returned if a child cannot be found. An error is
// returned if the parent image cannot be found.
func getLocalCachedImage(imageStore image.Store, imgID image.ID, config *containertypes.Config) (*image.Image, error) {
func getLocalCachedImage(imageStore image.Store, imgID image.ID, config *containertypes.Config, platform ocispec.Platform) (*image.Image, error) {
if config == nil {
return nil, nil
}
isBuiltLocally := func(id image.ID) bool {
builtLocally, err := imageStore.IsBuiltLocally(id)
if err != nil {
logrus.WithFields(logrus.Fields{
"error": err,
"id": id,
}).Warn("failed to check if image was built locally")
return false
}
return builtLocally
}
// Loop on the children of the given image and check the config
getMatch := func(siblings []image.ID) (*image.Image, error) {
var match *image.Image
@@ -225,6 +243,26 @@ func getLocalCachedImage(imageStore image.Store, imgID image.ID, config *contain
return nil, fmt.Errorf("unable to find image %q", id)
}
if !isBuiltLocally(id) {
continue
}
imgPlatform := ocispec.Platform{
Architecture: img.Architecture,
OS: img.OS,
OSVersion: img.OSVersion,
OSFeatures: img.OSFeatures,
Variant: img.Variant,
}
// Discard old linux/amd64 images with empty platform.
if imgPlatform.OS == "" && imgPlatform.Architecture == "" {
continue
}
if !comparePlatform(platform, imgPlatform) {
continue
}
if compare(&img.ContainerConfig, config) {
// check for the most up to date match
if match == nil || match.Created.Before(img.Created) {
@@ -238,11 +276,29 @@ func getLocalCachedImage(imageStore image.Store, imgID image.ID, config *contain
// In this case, this is `FROM scratch`, which isn't an actual image.
if imgID == "" {
images := imageStore.Map()
var siblings []image.ID
for id, img := range images {
if img.Parent == imgID {
siblings = append(siblings, id)
if img.Parent != "" {
continue
}
if !isBuiltLocally(id) {
continue
}
// Do a quick initial filter on the Cmd to avoid adding all
// non-local images with empty parent to the siblings slice and
// performing a full config compare.
//
// config.Cmd is set to the current Dockerfile instruction so we
// check it against the img.ContainerConfig.Cmd which is the
// command of the last layer.
if !strSliceEqual(img.ContainerConfig.Cmd, config.Cmd) {
continue
}
siblings = append(siblings, id)
}
return getMatch(siblings)
}
@@ -251,3 +307,15 @@ func getLocalCachedImage(imageStore image.Store, imgID image.ID, config *contain
siblings := imageStore.Children(imgID)
return getMatch(siblings)
}
func strSliceEqual(a, b []string) bool {
if len(a) != len(b) {
return false
}
for i := 0; i < len(a); i++ {
if a[i] != b[i] {
return false
}
}
return true
}

195
image/cache/compare.go vendored
View File

@@ -1,45 +1,106 @@
package cache // import "github.com/docker/docker/image/cache"
import (
"strings"
"github.com/containerd/containerd/platforms"
"github.com/docker/docker/api/types/container"
ocispec "github.com/opencontainers/image-spec/specs-go/v1"
)
// compare two Config struct. Do not compare the "Image" nor "Hostname" fields
// If OpenStdin is set, then it differs
func compare(a, b *container.Config) bool {
if a == nil || b == nil ||
a.OpenStdin || b.OpenStdin {
return false
}
if a.AttachStdout != b.AttachStdout ||
a.AttachStderr != b.AttachStderr ||
a.User != b.User ||
a.OpenStdin != b.OpenStdin ||
a.Tty != b.Tty {
return false
}
// TODO: Remove once containerd image service directly uses the ImageCache and
// LocalImageCache structs.
func CompareConfig(a, b *container.Config) bool {
return compare(a, b)
}
if len(a.Cmd) != len(b.Cmd) ||
len(a.Env) != len(b.Env) ||
len(a.Labels) != len(b.Labels) ||
len(a.ExposedPorts) != len(b.ExposedPorts) ||
len(a.Entrypoint) != len(b.Entrypoint) ||
len(a.Volumes) != len(b.Volumes) {
return false
}
func comparePlatform(builderPlatform, imagePlatform ocispec.Platform) bool {
// On Windows, only check the Major and Minor versions.
// The Build and Revision compatibility depends on whether `process` or
// `hyperv` isolation used.
//
// Fixes https://github.com/moby/moby/issues/47307
if builderPlatform.OS == "windows" && imagePlatform.OS == builderPlatform.OS {
// OSVersion format is:
// Major.Minor.Build.Revision
builderParts := strings.Split(builderPlatform.OSVersion, ".")
imageParts := strings.Split(imagePlatform.OSVersion, ".")
for i := 0; i < len(a.Cmd); i++ {
if a.Cmd[i] != b.Cmd[i] {
return false
// Major and minor must match.
for i := 0; i < 2; i++ {
if len(builderParts) > i && len(imageParts) > i && builderParts[i] != imageParts[i] {
return false
}
}
if len(builderParts) >= 3 && len(imageParts) >= 3 {
// Keep only Major & Minor.
builderParts[0] = imageParts[0]
builderParts[1] = imageParts[1]
imagePlatform.OSVersion = strings.Join(builderParts, ".")
}
}
return platforms.Only(builderPlatform).Match(imagePlatform)
}
// compare two Config struct. Do not container-specific fields:
// - Image
// - Hostname
// - Domainname
// - MacAddress
func compare(a, b *container.Config) bool {
if a == nil || b == nil {
return false
}
if len(a.Env) != len(b.Env) {
return false
}
if len(a.Cmd) != len(b.Cmd) {
return false
}
if len(a.Entrypoint) != len(b.Entrypoint) {
return false
}
if len(a.Shell) != len(b.Shell) {
return false
}
if len(a.ExposedPorts) != len(b.ExposedPorts) {
return false
}
if len(a.Volumes) != len(b.Volumes) {
return false
}
if len(a.Labels) != len(b.Labels) {
return false
}
if len(a.OnBuild) != len(b.OnBuild) {
return false
}
for i := 0; i < len(a.Env); i++ {
if a.Env[i] != b.Env[i] {
return false
}
}
for k, v := range a.Labels {
if v != b.Labels[k] {
for i := 0; i < len(a.OnBuild); i++ {
if a.OnBuild[i] != b.OnBuild[i] {
return false
}
}
for i := 0; i < len(a.Cmd); i++ {
if a.Cmd[i] != b.Cmd[i] {
return false
}
}
for i := 0; i < len(a.Entrypoint); i++ {
if a.Entrypoint[i] != b.Entrypoint[i] {
return false
}
}
for i := 0; i < len(a.Shell); i++ {
if a.Shell[i] != b.Shell[i] {
return false
}
}
@@ -48,16 +109,84 @@ func compare(a, b *container.Config) bool {
return false
}
}
for i := 0; i < len(a.Entrypoint); i++ {
if a.Entrypoint[i] != b.Entrypoint[i] {
return false
}
}
for key := range a.Volumes {
if _, exists := b.Volumes[key]; !exists {
return false
}
}
for k, v := range a.Labels {
if v != b.Labels[k] {
return false
}
}
if a.AttachStdin != b.AttachStdin {
return false
}
if a.AttachStdout != b.AttachStdout {
return false
}
if a.AttachStderr != b.AttachStderr {
return false
}
if a.NetworkDisabled != b.NetworkDisabled {
return false
}
if a.Tty != b.Tty {
return false
}
if a.OpenStdin != b.OpenStdin {
return false
}
if a.StdinOnce != b.StdinOnce {
return false
}
if a.ArgsEscaped != b.ArgsEscaped {
return false
}
if a.User != b.User {
return false
}
if a.WorkingDir != b.WorkingDir {
return false
}
if a.StopSignal != b.StopSignal {
return false
}
if (a.StopTimeout == nil) != (b.StopTimeout == nil) {
return false
}
if a.StopTimeout != nil && b.StopTimeout != nil {
if *a.StopTimeout != *b.StopTimeout {
return false
}
}
if (a.Healthcheck == nil) != (b.Healthcheck == nil) {
return false
}
if a.Healthcheck != nil && b.Healthcheck != nil {
if a.Healthcheck.Interval != b.Healthcheck.Interval {
return false
}
if a.Healthcheck.StartPeriod != b.Healthcheck.StartPeriod {
return false
}
if a.Healthcheck.Timeout != b.Healthcheck.Timeout {
return false
}
if a.Healthcheck.Retries != b.Healthcheck.Retries {
return false
}
if len(a.Healthcheck.Test) != len(b.Healthcheck.Test) {
return false
}
for i := 0; i < len(a.Healthcheck.Test); i++ {
if a.Healthcheck.Test[i] != b.Healthcheck.Test[i] {
return false
}
}
}
return true
}

View File

@@ -1,11 +1,15 @@
package cache // import "github.com/docker/docker/image/cache"
import (
"runtime"
"testing"
"github.com/docker/docker/api/types/container"
"github.com/docker/docker/api/types/strslice"
"github.com/docker/go-connections/nat"
ocispec "github.com/opencontainers/image-spec/specs-go/v1"
"gotest.tools/v3/assert"
is "gotest.tools/v3/assert/cmp"
)
// Just to make life easier
@@ -124,3 +128,73 @@ func TestCompare(t *testing.T) {
}
}
}
func TestPlatformCompare(t *testing.T) {
for _, tc := range []struct {
name string
builder ocispec.Platform
image ocispec.Platform
expected bool
}{
{
name: "same os and arch",
builder: ocispec.Platform{Architecture: "amd64", OS: runtime.GOOS},
image: ocispec.Platform{Architecture: "amd64", OS: runtime.GOOS},
expected: true,
},
{
name: "same os different arch",
builder: ocispec.Platform{Architecture: "amd64", OS: runtime.GOOS},
image: ocispec.Platform{Architecture: "arm64", OS: runtime.GOOS},
expected: false,
},
{
name: "same os smaller host variant",
builder: ocispec.Platform{Variant: "v7", Architecture: "arm", OS: runtime.GOOS},
image: ocispec.Platform{Variant: "v8", Architecture: "arm", OS: runtime.GOOS},
expected: false,
},
{
name: "same os higher host variant",
builder: ocispec.Platform{Variant: "v8", Architecture: "arm", OS: runtime.GOOS},
image: ocispec.Platform{Variant: "v7", Architecture: "arm", OS: runtime.GOOS},
expected: true,
},
{
// Test for https://github.com/moby/moby/issues/47307
name: "different build and revision",
builder: ocispec.Platform{Architecture: "amd64", OS: "windows", OSVersion: "10.0.22621"},
image: ocispec.Platform{Architecture: "amd64", OS: "windows", OSVersion: "10.0.17763.5329"},
expected: true,
},
{
name: "different revision",
builder: ocispec.Platform{Architecture: "amd64", OS: "windows", OSVersion: "10.0.17763.1234"},
image: ocispec.Platform{Architecture: "amd64", OS: "windows", OSVersion: "10.0.17763.5329"},
expected: true,
},
{
name: "different major",
builder: ocispec.Platform{Architecture: "amd64", OS: "windows", OSVersion: "11.0.17763.5329"},
image: ocispec.Platform{Architecture: "amd64", OS: "windows", OSVersion: "10.0.17763.5329"},
expected: false,
},
{
name: "different minor same osver",
builder: ocispec.Platform{Architecture: "amd64", OS: "windows", OSVersion: "10.0.17763.5329"},
image: ocispec.Platform{Architecture: "amd64", OS: "windows", OSVersion: "10.1.17763.5329"},
expected: false,
},
{
name: "different arch same osver",
builder: ocispec.Platform{Architecture: "arm64", OS: "windows", OSVersion: "10.0.17763.5329"},
image: ocispec.Platform{Architecture: "amd64", OS: "windows", OSVersion: "10.0.17763.5329"},
expected: false,
},
} {
tc := tc
t.Run(tc.name, func(t *testing.T) {
assert.Check(t, is.Equal(comparePlatform(tc.builder, tc.image), tc.expected))
})
}
}

View File

@@ -2,6 +2,7 @@ package image // import "github.com/docker/docker/image"
import (
"fmt"
"os"
"sync"
"time"
@@ -24,6 +25,8 @@ type Store interface {
GetParent(id ID) (ID, error)
SetLastUpdated(id ID) error
GetLastUpdated(id ID) (time.Time, error)
SetBuiltLocally(id ID) error
IsBuiltLocally(id ID) (bool, error)
Children(id ID) []ID
Map() map[ID]*Image
Heads() map[ID]*Image
@@ -295,6 +298,23 @@ func (is *store) GetLastUpdated(id ID) (time.Time, error) {
return time.Parse(time.RFC3339Nano, string(bytes))
}
// SetBuiltLocally sets whether image can be used as a builder cache
func (is *store) SetBuiltLocally(id ID) error {
return is.fs.SetMetadata(id.Digest(), "builtLocally", []byte{1})
}
// IsBuiltLocally returns whether image can be used as a builder cache
func (is *store) IsBuiltLocally(id ID) (bool, error) {
bytes, err := is.fs.GetMetadata(id.Digest(), "builtLocally")
if err != nil || len(bytes) == 0 {
if errors.Is(err, os.ErrNotExist) {
err = nil
}
return false, err
}
return bytes[0] == 1, nil
}
func (is *store) Children(id ID) []ID {
is.RLock()
defer is.RUnlock()

View File

@@ -117,6 +117,49 @@ func TestPluginInstall(t *testing.T) {
assert.NilError(t, err)
})
t.Run("with digest", func(t *testing.T) {
defer setupTest(t)()
reg := registry.NewV2(t)
defer reg.Close()
name := "test-" + strings.ToLower(t.Name())
repo := path.Join(registry.DefaultURL, name+":latest")
err := plugin.Create(ctx, client, repo)
assert.NilError(t, err)
rdr, err := client.PluginPush(ctx, repo, "")
assert.NilError(t, err)
defer rdr.Close()
buf := &strings.Builder{}
assert.NilError(t, err)
var digest string
assert.NilError(t, jsonmessage.DisplayJSONMessagesStream(rdr, buf, 0, false, func(j jsonmessage.JSONMessage) {
if j.Aux != nil {
var r types.PushResult
assert.NilError(t, json.Unmarshal(*j.Aux, &r))
digest = r.Digest
}
}), buf)
err = client.PluginRemove(ctx, repo, types.PluginRemoveOptions{Force: true})
assert.NilError(t, err)
rdr, err = client.PluginInstall(ctx, repo, types.PluginInstallOptions{
Disabled: true,
RemoteRef: repo + "@" + digest,
})
assert.NilError(t, err)
defer rdr.Close()
_, err = io.Copy(io.Discard, rdr)
assert.NilError(t, err)
_, _, err = client.PluginInspectWithRaw(ctx, repo)
assert.NilError(t, err)
})
t.Run("with htpasswd", func(t *testing.T) {
defer setupTest(t)()

View File

@@ -185,12 +185,13 @@ func (r *remote) startContainerd() error {
runtime.LockOSThread()
defer runtime.UnlockOSThread()
err := cmd.Start()
startedCh <- err
if err != nil {
startedCh <- err
return
}
r.daemonWaitCh = make(chan struct{})
startedCh <- nil
// Reap our child when needed
if err := cmd.Wait(); err != nil {
r.logger.WithError(err).Errorf("containerd did not exit successfully")

View File

@@ -7,6 +7,8 @@ import (
"io"
"mime"
"net/http"
"net/url"
"regexp"
"strings"
"github.com/docker/docker/pkg/ioutils"
@@ -52,10 +54,23 @@ type Ctx struct {
authReq *Request
}
func isChunked(r *http.Request) bool {
// RFC 7230 specifies that content length is to be ignored if Transfer-Encoding is chunked
if strings.EqualFold(r.Header.Get("Transfer-Encoding"), "chunked") {
return true
}
for _, v := range r.TransferEncoding {
if strings.EqualFold(v, "chunked") {
return true
}
}
return false
}
// AuthZRequest authorized the request to the docker daemon using authZ plugins
func (ctx *Ctx) AuthZRequest(w http.ResponseWriter, r *http.Request) error {
var body []byte
if sendBody(ctx.requestURI, r.Header) && r.ContentLength > 0 && r.ContentLength < maxBodySize {
if sendBody(ctx.requestURI, r.Header) && (r.ContentLength > 0 || isChunked(r)) && r.ContentLength < maxBodySize {
var err error
body, r.Body, err = drainBody(r.Body)
if err != nil {
@@ -108,7 +123,6 @@ func (ctx *Ctx) AuthZResponse(rm ResponseModifier, r *http.Request) error {
if sendBody(ctx.requestURI, rm.Header()) {
ctx.authReq.ResponseBody = rm.RawBody()
}
for _, plugin := range ctx.plugins {
logrus.Debugf("AuthZ response using plugin %s", plugin.Name())
@@ -146,10 +160,26 @@ func drainBody(body io.ReadCloser) ([]byte, io.ReadCloser, error) {
return nil, newBody, err
}
func isAuthEndpoint(urlPath string) (bool, error) {
// eg www.test.com/v1.24/auth/optional?optional1=something&optional2=something (version optional)
matched, err := regexp.MatchString(`^[^\/]*\/(v\d[\d\.]*\/)?auth.*`, urlPath)
if err != nil {
return false, err
}
return matched, nil
}
// sendBody returns true when request/response body should be sent to AuthZPlugin
func sendBody(url string, header http.Header) bool {
func sendBody(inURL string, header http.Header) bool {
u, err := url.Parse(inURL)
// Assume no if the URL cannot be parsed - an empty request will still be forwarded to the plugin and should be rejected
if err != nil {
return false
}
// Skip body for auth endpoint
if strings.HasSuffix(url, "/auth") {
isAuth, err := isAuthEndpoint(u.Path)
if isAuth || err != nil {
return false
}

View File

@@ -175,8 +175,8 @@ func TestDrainBody(t *testing.T) {
func TestSendBody(t *testing.T) {
var (
url = "nothing.com"
testcases = []struct {
url string
contentType string
expected bool
}{
@@ -220,15 +220,93 @@ func TestSendBody(t *testing.T) {
contentType: "",
expected: false,
},
{
url: "nothing.com/auth",
contentType: "",
expected: false,
},
{
url: "nothing.com/auth",
contentType: "application/json;charset=UTF8",
expected: false,
},
{
url: "nothing.com/auth?p1=test",
contentType: "application/json;charset=UTF8",
expected: false,
},
{
url: "nothing.com/test?p1=/auth",
contentType: "application/json;charset=UTF8",
expected: true,
},
{
url: "nothing.com/something/auth",
contentType: "application/json;charset=UTF8",
expected: true,
},
{
url: "nothing.com/auth/test",
contentType: "application/json;charset=UTF8",
expected: false,
},
{
url: "nothing.com/v1.24/auth/test",
contentType: "application/json;charset=UTF8",
expected: false,
},
{
url: "nothing.com/v1/auth/test",
contentType: "application/json;charset=UTF8",
expected: false,
},
{
url: "www.nothing.com/v1.24/auth/test",
contentType: "application/json;charset=UTF8",
expected: false,
},
{
url: "https://www.nothing.com/v1.24/auth/test",
contentType: "application/json;charset=UTF8",
expected: false,
},
{
url: "http://nothing.com/v1.24/auth/test",
contentType: "application/json;charset=UTF8",
expected: false,
},
{
url: "www.nothing.com/test?p1=/auth",
contentType: "application/json;charset=UTF8",
expected: true,
},
{
url: "http://www.nothing.com/test?p1=/auth",
contentType: "application/json;charset=UTF8",
expected: true,
},
{
url: "www.nothing.com/something/auth",
contentType: "application/json;charset=UTF8",
expected: true,
},
{
url: "https://www.nothing.com/something/auth",
contentType: "application/json;charset=UTF8",
expected: true,
},
}
)
for _, testcase := range testcases {
header := http.Header{}
header.Set("Content-Type", testcase.contentType)
if testcase.url == "" {
testcase.url = "nothing.com"
}
if b := sendBody(url, header); b != testcase.expected {
t.Fatalf("Unexpected Content-Type; Expected: %t, Actual: %t", testcase.expected, b)
if b := sendBody(testcase.url, header); b != testcase.expected {
t.Fatalf("sendBody failed: url: %s, content-type: %s; Expected: %t, Actual: %t", testcase.url, testcase.contentType, testcase.expected, b)
}
}
}

View File

@@ -200,8 +200,13 @@ func withFetchProgress(cs content.Store, out progress.Output, ref reference.Name
switch desc.MediaType {
case ocispec.MediaTypeImageManifest, images.MediaTypeDockerSchema2Manifest:
tn := reference.TagNameOnly(ref)
tagged := tn.(reference.Tagged)
progress.Messagef(out, tagged.Tag(), "Pulling from %s", reference.FamiliarName(ref))
var tagOrDigest string
if tagged, ok := tn.(reference.Tagged); ok {
tagOrDigest = tagged.Tag()
} else {
tagOrDigest = tn.String()
}
progress.Messagef(out, tagOrDigest, "Pulling from %s", reference.FamiliarName(ref))
progress.Messagef(out, "", "Digest: %s", desc.Digest.String())
return nil, nil
case

View File

@@ -64,6 +64,7 @@
"alarm",
"bind",
"brk",
"cachestat",
"capget",
"capset",
"chdir",
@@ -109,6 +110,7 @@
"fchdir",
"fchmod",
"fchmodat",
"fchmodat2",
"fchown",
"fchown32",
"fchownat",
@@ -130,8 +132,11 @@
"ftruncate",
"ftruncate64",
"futex",
"futex_requeue",
"futex_time64",
"futex_wait",
"futex_waitv",
"futex_wake",
"futimesat",
"getcpu",
"getcwd",
@@ -206,6 +211,7 @@
"lstat",
"lstat64",
"madvise",
"map_shadow_stack",
"membarrier",
"memfd_create",
"memfd_secret",
@@ -783,7 +789,8 @@
"names": [
"get_mempolicy",
"mbind",
"set_mempolicy"
"set_mempolicy",
"set_mempolicy_home_node"
],
"action": "SCMP_ACT_ALLOW",
"includes": {

View File

@@ -56,6 +56,7 @@ func DefaultProfile() *Seccomp {
"alarm",
"bind",
"brk",
"cachestat", // kernel v6.5, libseccomp v2.5.5
"capget",
"capset",
"chdir",
@@ -101,6 +102,7 @@ func DefaultProfile() *Seccomp {
"fchdir",
"fchmod",
"fchmodat",
"fchmodat2", // kernel v6.6, libseccomp v2.5.5
"fchown",
"fchown32",
"fchownat",
@@ -122,8 +124,11 @@ func DefaultProfile() *Seccomp {
"ftruncate",
"ftruncate64",
"futex",
"futex_requeue", // kernel v6.7, libseccomp v2.5.5
"futex_time64",
"futex_wait", // kernel v6.7, libseccomp v2.5.5
"futex_waitv",
"futex_wake", // kernel v6.7, libseccomp v2.5.5
"futimesat",
"getcpu",
"getcwd",
@@ -198,6 +203,7 @@ func DefaultProfile() *Seccomp {
"lstat",
"lstat64",
"madvise",
"map_shadow_stack", // kernel v6.6, libseccomp v2.5.5
"membarrier",
"memfd_create",
"memfd_secret",
@@ -771,6 +777,7 @@ func DefaultProfile() *Seccomp {
"get_mempolicy",
"mbind",
"set_mempolicy",
"set_mempolicy_home_node", // kernel v5.17, libseccomp v2.5.4
},
Action: specs.ActAllow,
},

View File

@@ -71,7 +71,7 @@ require (
github.com/morikuni/aec v1.0.0
github.com/opencontainers/go-digest v1.0.0
github.com/opencontainers/image-spec v1.1.0-rc3
github.com/opencontainers/runc v1.1.11
github.com/opencontainers/runc v1.1.12
github.com/opencontainers/runtime-spec v1.1.0-rc.2
github.com/opencontainers/selinux v1.11.0
github.com/pelletier/go-toml v1.9.5

View File

@@ -1147,8 +1147,8 @@ github.com/opencontainers/runc v1.0.0-rc9/go.mod h1:qT5XzbpPznkRYVz/mWwUaVBUv2rm
github.com/opencontainers/runc v1.0.0-rc92/go.mod h1:X1zlU4p7wOlX4+WRCz+hvlRv8phdL7UqbYD+vQwNMmE=
github.com/opencontainers/runc v1.0.0-rc93/go.mod h1:3NOsor4w32B2tC0Zbl8Knk4Wg84SM2ImC1fxBuqJ/H0=
github.com/opencontainers/runc v1.0.2/go.mod h1:aTaHFFwQXuA71CiyxOdFFIorAoemI04suvGRQFzWTD0=
github.com/opencontainers/runc v1.1.11 h1:9LjxyVlE0BPMRP2wuQDRlHV4941Jp9rc3F0+YKimopA=
github.com/opencontainers/runc v1.1.11/go.mod h1:S+lQwSfncpBha7XTy/5lBwWgm5+y5Ma/O44Ekby9FK8=
github.com/opencontainers/runc v1.1.12 h1:BOIssBaW1La0/qbNZHXOOa71dZfZEQOzW7dqQf3phss=
github.com/opencontainers/runc v1.1.12/go.mod h1:S+lQwSfncpBha7XTy/5lBwWgm5+y5Ma/O44Ekby9FK8=
github.com/opencontainers/runtime-spec v0.1.2-0.20190507144316-5b71a03e2700/go.mod h1:jwyrGlmzljRJv/Fgzds9SsS/C5hL+LL3ko9hs6T5lQ0=
github.com/opencontainers/runtime-spec v1.0.1/go.mod h1:jwyrGlmzljRJv/Fgzds9SsS/C5hL+LL3ko9hs6T5lQ0=
github.com/opencontainers/runtime-spec v1.0.2-0.20190207185410-29686dbc5559/go.mod h1:jwyrGlmzljRJv/Fgzds9SsS/C5hL+LL3ko9hs6T5lQ0=

View File

@@ -77,16 +77,16 @@ var (
// TestMode is set to true by unit tests that need "fake" cgroupfs.
TestMode bool
cgroupFd int = -1
prepOnce sync.Once
prepErr error
resolveFlags uint64
cgroupRootHandle *os.File
prepOnce sync.Once
prepErr error
resolveFlags uint64
)
func prepareOpenat2() error {
prepOnce.Do(func() {
fd, err := unix.Openat2(-1, cgroupfsDir, &unix.OpenHow{
Flags: unix.O_DIRECTORY | unix.O_PATH,
Flags: unix.O_DIRECTORY | unix.O_PATH | unix.O_CLOEXEC,
})
if err != nil {
prepErr = &os.PathError{Op: "openat2", Path: cgroupfsDir, Err: err}
@@ -97,15 +97,16 @@ func prepareOpenat2() error {
}
return
}
file := os.NewFile(uintptr(fd), cgroupfsDir)
var st unix.Statfs_t
if err = unix.Fstatfs(fd, &st); err != nil {
if err := unix.Fstatfs(int(file.Fd()), &st); err != nil {
prepErr = &os.PathError{Op: "statfs", Path: cgroupfsDir, Err: err}
logrus.Warnf("falling back to securejoin: %s", prepErr)
return
}
cgroupFd = fd
cgroupRootHandle = file
resolveFlags = unix.RESOLVE_BENEATH | unix.RESOLVE_NO_MAGICLINKS
if st.Type == unix.CGROUP2_SUPER_MAGIC {
// cgroupv2 has a single mountpoint and no "cpu,cpuacct" symlinks
@@ -132,7 +133,7 @@ func openFile(dir, file string, flags int) (*os.File, error) {
return openFallback(path, flags, mode)
}
fd, err := unix.Openat2(cgroupFd, relPath,
fd, err := unix.Openat2(int(cgroupRootHandle.Fd()), relPath,
&unix.OpenHow{
Resolve: resolveFlags,
Flags: uint64(flags) | unix.O_CLOEXEC,
@@ -140,20 +141,20 @@ func openFile(dir, file string, flags int) (*os.File, error) {
})
if err != nil {
err = &os.PathError{Op: "openat2", Path: path, Err: err}
// Check if cgroupFd is still opened to cgroupfsDir
// Check if cgroupRootHandle is still opened to cgroupfsDir
// (happens when this package is incorrectly used
// across the chroot/pivot_root/mntns boundary, or
// when /sys/fs/cgroup is remounted).
//
// TODO: if such usage will ever be common, amend this
// to reopen cgroupFd and retry openat2.
fdStr := strconv.Itoa(cgroupFd)
// to reopen cgroupRootHandle and retry openat2.
fdStr := strconv.Itoa(int(cgroupRootHandle.Fd()))
fdDest, _ := os.Readlink("/proc/self/fd/" + fdStr)
if fdDest != cgroupfsDir {
// Wrap the error so it is clear that cgroupFd
// Wrap the error so it is clear that cgroupRootHandle
// is opened to an unexpected/wrong directory.
err = fmt.Errorf("cgroupFd %s unexpectedly opened to %s != %s: %w",
fdStr, fdDest, cgroupfsDir, err)
err = fmt.Errorf("cgroupRootHandle %d unexpectedly opened to %s != %s: %w",
cgroupRootHandle.Fd(), fdDest, cgroupfsDir, err)
}
return nil, err
}

View File

@@ -7,6 +7,7 @@ import (
"fmt"
"os"
"strconv"
_ "unsafe" // for go:linkname
"golang.org/x/sys/unix"
)
@@ -23,9 +24,11 @@ func EnsureProcHandle(fh *os.File) error {
return nil
}
// CloseExecFrom applies O_CLOEXEC to all file descriptors currently open for
// the process (except for those below the given fd value).
func CloseExecFrom(minFd int) error {
type fdFunc func(fd int)
// fdRangeFrom calls the passed fdFunc for each file descriptor that is open in
// the current process.
func fdRangeFrom(minFd int, fn fdFunc) error {
fdDir, err := os.Open("/proc/self/fd")
if err != nil {
return err
@@ -50,15 +53,60 @@ func CloseExecFrom(minFd int) error {
if fd < minFd {
continue
}
// Intentionally ignore errors from unix.CloseOnExec -- the cases where
// this might fail are basically file descriptors that have already
// been closed (including and especially the one that was created when
// os.ReadDir did the "opendir" syscall).
unix.CloseOnExec(fd)
// Ignore the file descriptor we used for readdir, as it will be closed
// when we return.
if uintptr(fd) == fdDir.Fd() {
continue
}
// Run the closure.
fn(fd)
}
return nil
}
// CloseExecFrom sets the O_CLOEXEC flag on all file descriptors greater or
// equal to minFd in the current process.
func CloseExecFrom(minFd int) error {
return fdRangeFrom(minFd, unix.CloseOnExec)
}
//go:linkname runtime_IsPollDescriptor internal/poll.IsPollDescriptor
// In order to make sure we do not close the internal epoll descriptors the Go
// runtime uses, we need to ensure that we skip descriptors that match
// "internal/poll".IsPollDescriptor. Yes, this is a Go runtime internal thing,
// unfortunately there's no other way to be sure we're only keeping the file
// descriptors the Go runtime needs. Hopefully nothing blows up doing this...
func runtime_IsPollDescriptor(fd uintptr) bool //nolint:revive
// UnsafeCloseFrom closes all file descriptors greater or equal to minFd in the
// current process, except for those critical to Go's runtime (such as the
// netpoll management descriptors).
//
// NOTE: That this function is incredibly dangerous to use in most Go code, as
// closing file descriptors from underneath *os.File handles can lead to very
// bad behaviour (the closed file descriptor can be re-used and then any
// *os.File operations would apply to the wrong file). This function is only
// intended to be called from the last stage of runc init.
func UnsafeCloseFrom(minFd int) error {
// We must not close some file descriptors.
return fdRangeFrom(minFd, func(fd int) {
if runtime_IsPollDescriptor(uintptr(fd)) {
// These are the Go runtimes internal netpoll file descriptors.
// These file descriptors are operated on deep in the Go scheduler,
// and closing those files from underneath Go can result in panics.
// There is no issue with keeping them because they are not
// executable and are not useful to an attacker anyway. Also we
// don't have any choice.
return
}
// There's nothing we can do about errors from close(2), and the
// only likely error to be seen is EBADF which indicates the fd was
// already closed (in which case, we got what we wanted).
_ = unix.Close(fd)
})
}
// NewSockPair returns a new unix socket pair
func NewSockPair(name string) (parent *os.File, child *os.File, err error) {
fds, err := unix.Socketpair(unix.AF_LOCAL, unix.SOCK_STREAM|unix.SOCK_CLOEXEC, 0)

2
vendor/modules.txt vendored
View File

@@ -837,7 +837,7 @@ github.com/opencontainers/go-digest/digestset
github.com/opencontainers/image-spec/identity
github.com/opencontainers/image-spec/specs-go
github.com/opencontainers/image-spec/specs-go/v1
# github.com/opencontainers/runc v1.1.11
# github.com/opencontainers/runc v1.1.12
## explicit; go 1.17
github.com/opencontainers/runc/libcontainer/cgroups
github.com/opencontainers/runc/libcontainer/configs

View File

@@ -9,6 +9,7 @@ import (
"os"
"path/filepath"
"reflect"
"runtime/debug"
"strings"
"sync"
@@ -335,8 +336,15 @@ func (v *localVolume) Unmount(id string) error {
// ultimately there's nothing that can be done. If we don't decrement the count
// this volume can never be removed until a daemon restart occurs.
if v.needsMount() {
v.active.count--
logger.WithField("active mounts", v.active).Debug("Decremented active mount count")
// TODO: Remove once the real bug is fixed: https://github.com/moby/moby/issues/46508
if v.active.count > 0 {
v.active.count--
logger.WithField("active mounts", v.active).Debug("Decremented active mount count")
} else {
logger.Error("An attempt to decrement a zero mount count")
logger.Error(string(debug.Stack()))
return nil
}
}
if v.active.count > 0 {

View File

@@ -82,7 +82,9 @@ func (p *linuxParser) validateMountConfigImpl(mnt *mount.Mount, validateBindSour
if err != nil {
return &errMountConfig{mnt, err}
}
if !exists {
createMountpoint := mnt.BindOptions != nil && mnt.BindOptions.CreateMountpoint
if !exists && !createMountpoint {
return &errMountConfig{mnt, errBindSourceDoesNotExist(mnt.Source)}
}
}

View File

@@ -5,6 +5,7 @@ import (
"fmt"
"os"
"path/filepath"
"runtime/debug"
"syscall"
"github.com/containerd/containerd/log"
@@ -83,11 +84,22 @@ func (m *MountPoint) Cleanup() error {
return nil
}
logger := log.G(context.TODO()).WithFields(log.Fields{"active": m.active, "id": m.ID})
// TODO: Remove once the real bug is fixed: https://github.com/moby/moby/issues/46508
if m.active == 0 {
logger.Error("An attempt to decrement a zero mount count")
logger.Error(string(debug.Stack()))
return nil
}
if err := m.Volume.Unmount(m.ID); err != nil {
return errors.Wrapf(err, "error unmounting volume %s", m.Volume.Name())
}
m.active--
logger.Debug("MountPoint.Cleanup Decrement active count")
if m.active == 0 {
m.ID = ""
}