shim: add support for core scheduling

In linux 5.14 and hopefully some backports, core scheduling allows processes to
be co scheduled within the same domain on SMT enabled systems.

Containerd impl sets the core sched domain when launching a shim. This
allows a clean way for each shim(container/pod) to be in its own domain and any
additional containers, (v2 pods) be be launched with the same domain as well as
any exec'd process added to the container.

kernel docs: https://www.kernel.org/doc/html/latest/admin-guide/hw-vuln/core-scheduling.html

For Kata specifically, we will look for SCHED_CORE environment variable
to be set to indicate we shuold create a new schedule core domain.

This is equivalent to the containerd shim's PR: e48bbe8394

Fixes: #4309

Signed-off-by: Eric Ernst <eric_ernst@apple.com>
Signed-off-by: Michael Crosby <michael@thepasture.io>

src/runtime/pkg/containerd-shim-v2/service.go

@@ -10,6 +10,7 @@ import (
 	"io"
 	"os"
 	sysexec "os/exec"
+	goruntime "runtime"
 	"sync"
 	"syscall"
 	"time"
@@ -31,6 +32,7 @@ import (
 	"github.com/kata-containers/kata-containers/src/runtime/pkg/katautils"
 	"github.com/kata-containers/kata-containers/src/runtime/pkg/katautils/katatrace"
 	"github.com/kata-containers/kata-containers/src/runtime/pkg/oci"
+	"github.com/kata-containers/kata-containers/src/runtime/pkg/utils"
 	vc "github.com/kata-containers/kata-containers/src/runtime/virtcontainers"
 	"github.com/kata-containers/kata-containers/src/runtime/virtcontainers/pkg/compatoci"
 	"github.com/kata-containers/kata-containers/src/runtime/virtcontainers/types"
@@ -234,9 +236,19 @@ func (s *service) StartShim(ctx context.Context, opts cdshim.StartOpts) (_ strin
 
 	cmd.ExtraFiles = append(cmd.ExtraFiles, f)
 
+	goruntime.LockOSThread()
+	if os.Getenv("SCHED_CORE") != "" {
+		if err := utils.Create(utils.ProcessGroup); err != nil {
+			return "", errors.Wrap(err, "enable sched core support")
+		}
+	}
+
 	if err := cmd.Start(); err != nil {
 		return "", err
 	}
+
+	goruntime.UnlockOSThread()
+
 	defer func() {
 		if retErr != nil {
 			cmd.Process.Kill()

src/runtime/pkg/utils/schedcore.go

@@ -0,0 +1,36 @@
+// Copyright (c) 2022 Apple Inc.
+//
+// SPDX-License-Identifier: Apache-2.0
+//
+
+package utils
+
+import (
+	"golang.org/x/sys/unix"
+)
+
+// PidType is the type of provided pid value and how it should be treated
+type PidType int
+
+const (
+	pidTypePid            = 0
+	pidTypeThreadGroupId  = 1
+	pidTypeProcessGroupId = 2
+
+	// Pid affects the current pid
+	Pid PidType = pidtypePid
+	// ThreadGroup affects all threads in the group
+	ThreadGroup PidType = pidtypeTgid
+	// ProcessGroup affects all processes in the group
+	ProcessGroup PidType = pidtypePgid
+)
+
+// Create a new sched core domain
+func Create(t PidType) error {
+	return unix.Prctl(unix.PR_SCHED_CORE, unix.PR_SCHED_CORE_CREATE, 0, uintptr(t), 0)
+}
+
+// ShareFrom shares the sched core domain from the provided pid
+func ShareFrom(pid uint64, t PidType) error {
+	return unix.Prctl(unix.PR_SCHED_CORE, unix.PR_SCHED_CORE_SHARE_FROM, uintptr(pid), uintptr(t), 0)
+}