@@ -4,7 +4,37 @@ This file documents all notable changes made to this project since runc 1.0.
44The format is based on [ Keep a Changelog] ( https://keepachangelog.com/en/1.0.0/ ) ,
55and this project adheres to [ Semantic Versioning] ( https://semver.org/spec/v2.0.0.html ) .
66
7- ## [ Unreleased]  
7+ ## [ unreleased]  
8+ 
9+ ## [ 1.2.0-rc.1]  - 2023-08-10 
10+ 
11+ >  How, dear sir, did you cross the flood? By not stopping, friend, and by not
12+ >  straining I crossed the flood
13+ 
14+ This new runc release includes all the changes from the v1.1 patch releases.
15+ 
16+ ### Compatibility  
17+ 
18+ *  This release requires Go 1.20.x or Go 1.19.x (#3718  )
19+ 
20+ ### Added  
21+ 
22+  *  Support idmap mounts as specified in the OCI runtime-spec v1.1.0. Currently
23+    the mount mappings need to be identical to the mappings used in the user
24+    namespace section. (#3717  )
25+  *  The ` runc features `  command is no longer experimental. (#3861  )
26+  *  Support for ` cgroup.kill `  to kill all processes inside a container. (#3135  ,
27+    #3825  )
28+  *  Implement to set a domainname as specified in the OCI runtime-spec v1.1.0.
29+    (#3600  )
30+  *  Add support for umask when exec-ing into a container. (#3661  )
31+  *  libct/cg: support SCHED_IDLE for runc cgroupfs. (#3377  )
32+  *  checkpoint/restore: implement --manage-cgroups-mode ignore. (#3546  )
33+  *  seccomp: refactor flags support; add flags to features, set SPEC_ALLOW by
34+    default. (#3588  )
35+  *  libct/cg/sd: use systemd v240+ new ` MAJOR:* `  syntax. (#3843  )
36+  *  Use github.com/checkpoint-restore/go-criu v6.3.0, which reduces the compiled
37+    binary file. (#3652  )
838
939### Deprecated  
1040
@@ -16,6 +46,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
1646   to kill a container (with SIGKILL) which does not have its own private PID
1747   namespace (so that runc would send SIGKILL to all processes). Now, this is
1848   done automatically. (#3864  , #3825  )
49+  *  ` runc kill `  option ` -a `  is now deprecated. Previously, it had to be specified
50+    to kill a container (with SIGKILL) which does not have its own private PID
51+    namespace (so that runc would send SIGKILL to all processes). Now, this is
52+    done automatically. (#3864  , #3825  )
1953
2054### Changed  
2155
@@ -31,6 +65,24 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
3165   This aligns cgroupv2 root usage more closely with cgroupv1 reporting.
3266   Additionally, report root swap usage as sum of swap and memory usage,
3367   aligned with v1 and existing non-root v2 reporting. (#3933  )
68+  *  When running rootless or using a user namespace, a source filesystem mounted
69+    with ` nodev ` , ` nosuid `  or ` noexec `  can now be used as source of a bind mount
70+    without the same options being set for the mount. (#3805  )
71+  *  libcontainer users that create and kill containers from a daemon process
72+    (so that the container init is a child of that process) must now implement
73+    a proper child reaper in case a container does not have its own private PID
74+    namespace, as documented in ` container.Signal ` . (#3825  )
75+  *  libcontainer: ` container.Signal `  no longer have the second ` all bool ` 
76+    argument; a need to kill all processes is now determined automatically.
77+    (#3825  , #3885  )
78+  *  libct/cg: Remove function EnterPid with no users. (#3797  )
79+  *  libct/seccomp: enable seccomp binary tree optimization. (#3405  )
80+  *  runc run/exec: ignore SIGURG. (#3368  )
81+  *  Remove tun/tap from the default device rules. (#3468  )
82+  *  ` runc --root non-existent-dir list `  now reports an error for non-existent
83+    root directory. (#3374  )
84+  *  libct: Mount: Remove {Pre,Post}mountCmds, were never used and are obsoleted
85+    by more generic container hooks. (#3350  )
3486
3587### Fixed  
3688
@@ -42,6 +94,12 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
4294   code (this was due to how s390x does syscall multiplexing). (#3474  )
4395 *  Remove tun/tap from the default device rules. (#3468  )
4496 *  specconv: avoid mapping "acl" to MS_POSIXACL. (#3739  )
97+  *  libcontainer: fix private PID namespace detection when killing the container.
98+    (#3866  , #3825  )
99+  *  Fix ` READY `  notification sometimes not accepted by systemd. (#3291  , #3293  )
100+  *  Reduce the number of mount/umount syscalls in the host mount namespace.
101+    (#3599  , #2532  )
102+ 
45103
46104## [ 1.1.8]  - 2023-07-20 
47105
0 commit comments