SCC: Promote sysctl annotations to fields

api/docs/api/v1.SecurityContextConstraints.adoc

@@ -29,11 +29,19 @@ Expand or mouse-over a field for more information about it.
 </summary><div style="margin-left:13px;">- <span title="(string)">[string]</span>:
 </div></details><details><summary><span title="(array) AllowedFlexVolumes is a whitelist of allowed Flexvolumes.  Empty or nil indicates that all Flexvolumes may be used.  This parameter is effective only when the usage of the Flexvolumes is allowed in the &#34;Volumes&#34; field.">allowedFlexVolumes</span>:
 </summary><div style="margin-left:13px;">- <span title="(string) Driver is the name of the Flexvolume driver.">driver</span>:
+</div></details><details><summary><span title="(array) AllowedUnsafeSysctls is a list of explicitly allowed unsafe sysctls, defaults to none. Each entry is either a plain sysctl name or ends in &#34;*&#34; in which case it is considered as a prefix of allowed sysctls. Single * means all unsafe sysctls are allowed. Kubelet has to whitelist all allowed unsafe sysctls explicitly to avoid rejection.
+
+Examples: e.g. &#34;foo/*&#34; allows &#34;foo/bar&#34;, &#34;foo/baz&#34;, etc. e.g. &#34;foo.*&#34; allows &#34;foo.bar&#34;, &#34;foo.baz&#34;, etc.">allowedUnsafeSysctls</span>:
+</summary><div style="margin-left:13px;">- <span title="(string)">[string]</span>:
 </div></details><div style="margin-left:13px;"><span title="(string) APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/api-conventions.md#resources">apiVersion</span>:
 </div><details><summary><span title="(array) DefaultAddCapabilities is the default set of capabilities that will be added to the container unless the pod spec specifically drops the capability.  You may not list a capabiility in both DefaultAddCapabilities and RequiredDropCapabilities.">defaultAddCapabilities</span>:
 </summary><div style="margin-left:13px;">- <span title="(string)">[string]</span>:
 </div></details><div style="margin-left:13px;"><span title="(boolean) DefaultAllowPrivilegeEscalation controls the default setting for whether a process can gain more privileges than its parent process.">defaultAllowPrivilegeEscalation</span>:
-</div><details><summary><span title="(v1.FSGroupStrategyOptions) FSGroup is the strategy that will dictate what fs group is used by the SecurityContext.">fsGroup</span>:
+</div><details><summary><span title="(array) ForbiddenSysctls is a list of explicitly forbidden sysctls, defaults to none. Each entry is either a plain sysctl name or ends in &#34;*&#34; in which case it is considered as a prefix of forbidden sysctls. Single * means all sysctls are forbidden.
+
+Examples: e.g. &#34;foo/*&#34; forbids &#34;foo/bar&#34;, &#34;foo/baz&#34;, etc. e.g. &#34;foo.*&#34; forbids &#34;foo.bar&#34;, &#34;foo.baz&#34;, etc.">forbiddenSysctls</span>:
+</summary><div style="margin-left:13px;">- <span title="(string)">[string]</span>:
+</div></details><details><summary><span title="(v1.FSGroupStrategyOptions) FSGroup is the strategy that will dictate what fs group is used by the SecurityContext.">fsGroup</span>:
 </summary><details><summary>  <span title="(array) Ranges are the allowed ranges of fs groups.  If you would like to force a single fs group then supply a single range with the same start and end.">ranges</span>:
 </summary><div style="margin-left:13px;">  - <span title="(integer) Max is the end of the range, inclusive.">max</span>:
 </div><div style="margin-left:13px;">    <span title="(integer) Min is the start of the range, inclusive.">min</span>:

api/docs/apis-security.openshift.io/v1.SecurityContextConstraints.adoc

@@ -29,11 +29,19 @@ Expand or mouse-over a field for more information about it.
 </summary><div style="margin-left:13px;">- <span title="(string)">[string]</span>:
 </div></details><details><summary><span title="(array) AllowedFlexVolumes is a whitelist of allowed Flexvolumes.  Empty or nil indicates that all Flexvolumes may be used.  This parameter is effective only when the usage of the Flexvolumes is allowed in the &#34;Volumes&#34; field.">allowedFlexVolumes</span>:
 </summary><div style="margin-left:13px;">- <span title="(string) Driver is the name of the Flexvolume driver.">driver</span>:
+</div></details><details><summary><span title="(array) AllowedUnsafeSysctls is a list of explicitly allowed unsafe sysctls, defaults to none. Each entry is either a plain sysctl name or ends in &#34;*&#34; in which case it is considered as a prefix of allowed sysctls. Single * means all unsafe sysctls are allowed. Kubelet has to whitelist all allowed unsafe sysctls explicitly to avoid rejection.
+
+Examples: e.g. &#34;foo/*&#34; allows &#34;foo/bar&#34;, &#34;foo/baz&#34;, etc. e.g. &#34;foo.*&#34; allows &#34;foo.bar&#34;, &#34;foo.baz&#34;, etc.">allowedUnsafeSysctls</span>:
+</summary><div style="margin-left:13px;">- <span title="(string)">[string]</span>:
 </div></details><div style="margin-left:13px;"><span title="(string) APIVersion defines the versioned schema of this representation of an object. Servers should convert recognized schemas to the latest internal value, and may reject unrecognized values. More info: https://git.k8s.io/community/contributors/devel/api-conventions.md#resources">apiVersion</span>:
 </div><details><summary><span title="(array) DefaultAddCapabilities is the default set of capabilities that will be added to the container unless the pod spec specifically drops the capability.  You may not list a capabiility in both DefaultAddCapabilities and RequiredDropCapabilities.">defaultAddCapabilities</span>:
 </summary><div style="margin-left:13px;">- <span title="(string)">[string]</span>:
 </div></details><div style="margin-left:13px;"><span title="(boolean) DefaultAllowPrivilegeEscalation controls the default setting for whether a process can gain more privileges than its parent process.">defaultAllowPrivilegeEscalation</span>:
-</div><details><summary><span title="(v1.FSGroupStrategyOptions) FSGroup is the strategy that will dictate what fs group is used by the SecurityContext.">fsGroup</span>:
+</div><details><summary><span title="(array) ForbiddenSysctls is a list of explicitly forbidden sysctls, defaults to none. Each entry is either a plain sysctl name or ends in &#34;*&#34; in which case it is considered as a prefix of forbidden sysctls. Single * means all sysctls are forbidden.
+
+Examples: e.g. &#34;foo/*&#34; forbids &#34;foo/bar&#34;, &#34;foo/baz&#34;, etc. e.g. &#34;foo.*&#34; forbids &#34;foo.bar&#34;, &#34;foo.baz&#34;, etc.">forbiddenSysctls</span>:
+</summary><div style="margin-left:13px;">- <span title="(string)">[string]</span>:
+</div></details><details><summary><span title="(v1.FSGroupStrategyOptions) FSGroup is the strategy that will dictate what fs group is used by the SecurityContext.">fsGroup</span>:
 </summary><details><summary>  <span title="(array) Ranges are the allowed ranges of fs groups.  If you would like to force a single fs group then supply a single range with the same start and end.">ranges</span>:
 </summary><div style="margin-left:13px;">  - <span title="(integer) Max is the end of the range, inclusive.">max</span>:
 </div><div style="margin-left:13px;">    <span title="(integer) Min is the start of the range, inclusive.">min</span>:

pkg/cmd/server/bootstrappolicy/securitycontextconstraints.go

@@ -94,7 +94,8 @@ func GetBootstrapSecurityContextConstraints(sccNameToAdditionalGroups map[string
 			SupplementalGroups: securityapi.SupplementalGroupsStrategyOptions{
 				Type: securityapi.SupplementalGroupsStrategyRunAsAny,
 			},
-			SeccompProfiles: []string{"*"},
+			SeccompProfiles:      []string{"*"},
+			AllowedUnsafeSysctls: []string{"*"},
 		},
 		// SecurityContextConstraintNonRoot does not allow host access, allocates SELinux labels
 		// and allows the user to request a specific UID or provide the default in the dockerfile.

pkg/oc/lib/describe/describer.go

@@ -1889,6 +1889,8 @@ func describeSecurityContextConstraints(scc *securityapi.SecurityContextConstrai
 		fmt.Fprintf(out, "  Allowed Seccomp Profiles:\t%s\n", stringOrNone(strings.Join(scc.SeccompProfiles, ",")))
 		fmt.Fprintf(out, "  Allowed Volume Types:\t%s\n", fsTypeToString(scc.Volumes))
 		fmt.Fprintf(out, "  Allowed Flexvolumes:\t%s\n", flexVolumesToString(scc.AllowedFlexVolumes))
+		fmt.Fprintf(out, "  Allowed Unsafe Sysctls:\t%s\n", sysctlsToString(scc.AllowedUnsafeSysctls))
+		fmt.Fprintf(out, "  Forbidden Sysctls:\t%s\n", sysctlsToString(scc.ForbiddenSysctls))
 		fmt.Fprintf(out, "  Allow Host Network:\t%t\n", scc.AllowHostNetwork)
 		fmt.Fprintf(out, "  Allow Host Ports:\t%t\n", scc.AllowHostPorts)
 		fmt.Fprintf(out, "  Allow Host PID:\t%t\n", scc.AllowHostPID)
@@ -1964,6 +1966,10 @@ func flexVolumesToString(flexVolumes []securityapi.AllowedFlexVolume) string {
 	return stringOrDefaultValue(strings.Join(volumes, ","), "<all>")
 }
 
+func sysctlsToString(sysctls []string) string {
+	return stringOrNone(strings.Join(sysctls, ","))
+}
+
 func idRangeToString(ranges []securityapi.IDRange) string {
 	formattedString := ""
 	if ranges != nil {

pkg/security/apis/security/types.go

@@ -92,6 +92,26 @@ type SecurityContextConstraints struct {
 	Users []string
 	// The groups that have permission to use this security context constraints
 	Groups []string
+
+	// AllowedUnsafeSysctls is a list of explicitly allowed unsafe sysctls, defaults to none.
+	// Each entry is either a plain sysctl name or ends in "*" in which case it is considered
+	// as a prefix of allowed sysctls. Single * means all unsafe sysctls are allowed.
+	// Kubelet has to whitelist all allowed unsafe sysctls explicitly to avoid rejection.
+	//
+	// Examples:
+	// e.g. "foo/*" allows "foo/bar", "foo/baz", etc.
+	// e.g. "foo.*" allows "foo.bar", "foo.baz", etc.
+	// +optional
+	AllowedUnsafeSysctls []string
+	// ForbiddenSysctls is a list of explicitly forbidden sysctls, defaults to none.
+	// Each entry is either a plain sysctl name or ends in "*" in which case it is considered
+	// as a prefix of forbidden sysctls. Single * means all sysctls are forbidden.
+	//
+	// Examples:
+	// e.g. "foo/*" forbids "foo/bar", "foo/baz", etc.
+	// e.g. "foo.*" forbids "foo.bar", "foo.baz", etc.
+	// +optional
+	ForbiddenSysctls []string
 }
 
 // FS Type gives strong typing to different file systems that are used by volumes.