criu swrk: return child error to caller

In the podman CI we are seeing a weird flake during criu version
detection[1]. The write to the socket just fails with broken pipe.
The logical thing to assume here is that the child exited. However the
current code never reports back the child error from wait nor does it
try to capture the output from it. This fixes both. The cleanup error is
now added to the returned error so the caller sees both.

As errors.Join is used from the std lib bump the minimum go version to
1.20.

[1] containers/podman#18856

Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Signed-off-by: Radostin Stoyanov <rstoyanov@fedoraproject.org>

.github/workflows/main.yml

@@ -13,13 +13,13 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        go-version: [1.18.x, 1.19.x, 1.20.x]
+        go-version: [1.20, 1.21, 1.22]
         criu_branch: [master, criu-dev]
 
     steps:
 
     - name: Checkout
-      uses: actions/checkout@v3
+      uses: actions/checkout@v4
       with:
         # needed for codecov
         fetch-depth: 0
@@ -32,7 +32,7 @@ jobs:
         sudo make -C criu install-criu PREFIX=/usr
 
     - name: Install Go ${{ matrix.go-version }}
-      uses: actions/setup-go@v4
+      uses: actions/setup-go@v5
       with:
         go-version: ${{ matrix.go-version }}
 

go.mod

@@ -1,6 +1,6 @@
 module github.com/checkpoint-restore/go-criu/v7
 
-go 1.18
+go 1.20
 
 require (
 	github.com/spf13/cobra v1.8.0

main.go

@@ -1,11 +1,13 @@
 package criu
 
 import (
+	"bytes"
 	"errors"
 	"fmt"
 	"os"
 	"os/exec"
 	"strconv"
+	"strings"
 	"syscall"
 
 	"github.com/checkpoint-restore/go-criu/v7/rpc"
@@ -17,6 +19,7 @@ type Criu struct {
 	swrkCmd  *exec.Cmd
 	swrkSk   *os.File
 	swrkPath string
+	output   *bytes.Buffer
 }
 
 // MakeCriu returns the Criu object required for most operations
@@ -44,9 +47,12 @@ func (c *Criu) Prepare() error {
 	srv := os.NewFile(uintptr(fds[1]), "criu-xprt-srv")
 	defer srv.Close()
 
+	out := new(bytes.Buffer)
 	args := []string{"swrk", strconv.Itoa(fds[1])}
 	// #nosec G204
 	cmd := exec.Command(c.swrkPath, args...)
+	cmd.Stdout = out
+	cmd.Stderr = out
 
 	err = cmd.Start()
 	if err != nil {
@@ -61,13 +67,20 @@ func (c *Criu) Prepare() error {
 }
 
 // Cleanup cleans up
-func (c *Criu) Cleanup() {
+func (c *Criu) Cleanup() error {
+	var errs []error
 	if c.swrkCmd != nil {
-		c.swrkSk.Close()
+		if err := c.swrkSk.Close(); err != nil {
+			errs = append(errs, err)
+		}
 		c.swrkSk = nil
-		_ = c.swrkCmd.Wait()
+		if err := c.swrkCmd.Wait(); err != nil {
+			errs = append(errs, fmt.Errorf("criu swrk failed: %w (%s)", err, strings.TrimSpace(c.output.String())))
+		}
 		c.swrkCmd = nil
+		c.output = nil
 	}
+	return errors.Join(errs...)
 }
 
 func (c *Criu) sendAndRecv(reqB []byte) ([]byte, int, error) {
@@ -99,7 +112,7 @@ func (c *Criu) doSwrk(reqType rpc.CriuReqType, opts *rpc.CriuOpts, nfy Notify) e
 	return nil
 }
 
-func (c *Criu) doSwrkWithResp(reqType rpc.CriuReqType, opts *rpc.CriuOpts, nfy Notify, features *rpc.CriuFeatures) (*rpc.CriuResp, error) {
+func (c *Criu) doSwrkWithResp(reqType rpc.CriuReqType, opts *rpc.CriuOpts, nfy Notify, features *rpc.CriuFeatures) (_ *rpc.CriuResp, retErr error) {
 	var resp *rpc.CriuResp
 
 	req := rpc.CriuReq{
@@ -121,7 +134,13 @@ func (c *Criu) doSwrkWithResp(reqType rpc.CriuReqType, opts *rpc.CriuOpts, nfy N
 			return nil, err
 		}
 
-		defer c.Cleanup()
+		defer func() {
+			// append any cleanup errors to the returned error
+			err := c.Cleanup()
+			if err != nil {
+				retErr = errors.Join(retErr, err)
+			}
+		}()
 	}
 
 	for {

phaul/client.go

@@ -1,6 +1,7 @@
 package phaul
 
 import (
+	"errors"
 	"fmt"
 
 	"github.com/checkpoint-restore/go-criu/v7"
@@ -55,7 +56,7 @@ func isLastIter(iter int, stats *stats.DumpStatsEntry, prevStats *stats.DumpStat
 }
 
 // Migrate function
-func (pc *Client) Migrate() error {
+func (pc *Client) Migrate() (retErr error) {
 	criu := criu.MakeCriu()
 	psi := rpc.CriuPageServerInfo{
 		Fd: proto.Int32(int32(pc.cfg.Memfd)),
@@ -72,7 +73,13 @@ func (pc *Client) Migrate() error {
 		return err
 	}
 
-	defer criu.Cleanup()
+	defer func() {
+		// append any cleanup errors to the returned error
+		err := criu.Cleanup()
+		if err != nil {
+			retErr = errors.Join(retErr, err)
+		}
+	}()
 
 	imgs, err := preparePhaulImages(pc.cfg.Wdir)
 	if err != nil {