crit: add memory page content retrieval features

This commit adds the MemoryReader struct with methods to retrieve
the memory pages associated with a process.

Signed-off-by: Kouame Behouba Manasse <behouba@gmail.com>

crit/mempages.go

@@ -0,0 +1,144 @@
+package crit
+
+import (
+	"bytes"
+	"errors"
+	"fmt"
+	"os"
+	"path/filepath"
+
+	"github.com/checkpoint-restore/go-criu/v6/crit/images/pagemap"
+)
+
+var sysPageSize = os.Getpagesize()
+
+// MemoryReader is a struct used to retrieve
+// the content of memory associated with a specific process ID (pid).
+// New instances should be created with NewMemoryReader()
+type MemoryReader struct {
+	checkpointDir  string
+	pid            uint32
+	pagesID        uint32
+	pageSize       int
+	pagemapEntries []*pagemap.PagemapEntry
+}
+
+// NewMemoryReader creates a new instance of MemoryReader with all the fields populated
+func NewMemoryReader(checkpointDir string, pid uint32, pageSize int) (*MemoryReader, error) {
+	if pageSize == 0 {
+		pageSize = sysPageSize
+	}
+
+	// Check if the given page size is a positive power of 2, otherwise return an error
+	if (pageSize & (pageSize - 1)) != 0 {
+		return nil, errors.New("page size should be a positive power of 2")
+	}
+
+	pagemapImg, err := getImg(filepath.Join(checkpointDir, fmt.Sprintf("pagemap-%d.img", pid)), &pagemap.PagemapHead{})
+	if err != nil {
+		return nil, err
+	}
+
+	pagesID := pagemapImg.Entries[0].Message.(*pagemap.PagemapHead).GetPagesId()
+
+	pagemapEntries := make([]*pagemap.PagemapEntry, 0)
+
+	for _, entry := range pagemapImg.Entries[1:] {
+		pagemapEntries = append(pagemapEntries, entry.Message.(*pagemap.PagemapEntry))
+	}
+
+	return &MemoryReader{
+		checkpointDir:  checkpointDir,
+		pid:            pid,
+		pageSize:       pageSize,
+		pagesID:        pagesID,
+		pagemapEntries: pagemapEntries,
+	}, nil
+}
+
+// GetMemPages retrieves the content of memory pages
+// associated with a given process ID (pid).
+// It retrieves the memory content within the
+// specified range defined by the start and end addresses.
+func (mr *MemoryReader) GetMemPages(start, end uint64) (*bytes.Buffer, error) {
+	size := end - start
+
+	startPage := start / uint64(mr.pageSize)
+	endPage := end / uint64(mr.pageSize)
+
+	var buffer bytes.Buffer
+
+	for pageNumber := startPage; pageNumber <= endPage; pageNumber++ {
+		var page []byte = nil
+
+		pageMem, err := mr.getPage(pageNumber)
+		if err != nil {
+			return nil, err
+		}
+
+		if pageMem != nil {
+			page = pageMem
+		} else {
+			page = bytes.Repeat([]byte("\x00"), int(mr.pageSize))
+		}
+
+		var nSkip, nRead uint64
+
+		if pageNumber == startPage {
+			nSkip = start - pageNumber*uint64(mr.pageSize)
+			if startPage == endPage {
+				nRead = size
+			} else {
+				nRead = uint64(mr.pageSize) - nSkip
+			}
+		} else if pageNumber == endPage {
+			nSkip = 0
+			nRead = end - pageNumber*uint64(mr.pageSize)
+		} else {
+			nSkip = 0
+			nRead = uint64(mr.pageSize)
+		}
+
+		if _, err := buffer.Write(page[nSkip : nSkip+nRead]); err != nil {
+			return nil, err
+		}
+	}
+
+	return &buffer, nil
+}
+
+// getPage retrieves a memory page from the pages.img file.
+func (mr *MemoryReader) getPage(pageNo uint64) ([]byte, error) {
+	var offset uint64 = 0
+
+	// Iterate over pagemap entries to find the corresponding page
+	for _, m := range mr.pagemapEntries {
+		found := false
+		for i := 0; i < int(*m.NrPages); i++ {
+			if m.GetVaddr()+uint64(i)*uint64(mr.pageSize) == pageNo*uint64(mr.pageSize) {
+				found = true
+				break
+			}
+			offset += uint64(mr.pageSize)
+		}
+
+		if !found {
+			continue
+		}
+		f, err := os.Open(filepath.Join(mr.checkpointDir, fmt.Sprintf("pages-%d.img", mr.pagesID)))
+		if err != nil {
+			return nil, err
+		}
+
+		defer f.Close()
+
+		buff := make([]byte, mr.pageSize)
+
+		if _, err := f.ReadAt(buff, int64(offset)); err != nil {
+			return nil, err
+		}
+
+		return buff, nil
+	}
+	return nil, nil
+}

crit/mempages_test.go

@@ -0,0 +1,200 @@
+package crit
+
+import (
+	"errors"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+
+	"github.com/checkpoint-restore/go-criu/v6/crit/images/pstree"
+)
+
+const (
+	testImgsDir = "test-imgs"
+)
+
+func TestNewMemoryReader(t *testing.T) {
+	pid, err := getTestImgPID(testImgsDir)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	testCases := []struct {
+		name          string
+		dir           string
+		pid           uint32
+		pageSize      int
+		expectedError error
+	}{
+		{
+			name:          "Page size is 0",
+			dir:           testImgsDir,
+			pid:           pid,
+			pageSize:      0,
+			expectedError: nil,
+		},
+		{
+			name:          "Invalid page size",
+			dir:           testImgsDir,
+			pid:           pid,
+			pageSize:      4000,
+			expectedError: errors.New("page size should be a positive power of 2"),
+		},
+		{
+			name:          "Invalid test-imgs directory",
+			dir:           "no test directory",
+			pid:           pid,
+			expectedError: errors.New("no such file or directory"),
+		},
+		{
+			name:          "Valid test-imgs directory, pid and page size",
+			dir:           testImgsDir,
+			pid:           pid,
+			pageSize:      4096,
+			expectedError: nil,
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			mr, err := NewMemoryReader(tc.dir, tc.pid, tc.pageSize)
+			if err != nil && !strings.Contains(err.Error(), tc.expectedError.Error()) {
+				t.Errorf("Expected error: %v, got error: %v", tc.expectedError, err)
+			}
+
+			if mr == nil && tc.expectedError == nil {
+				t.Errorf("MemoryReader creation failed for checkpoint directory: %s and pid: %d", tc.dir, tc.pid)
+			}
+		})
+	}
+}
+
+// TestNewMemoryReader test GetMempages method of MemoryReader.
+func TestGetMemPages(t *testing.T) {
+	type testcase struct {
+		name          string
+		mr            *MemoryReader
+		start         uint64
+		end           uint64
+		expectedError error
+	}
+
+	pid, err := getTestImgPID(testImgsDir)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	// Create a temporary empty memory pages file for testing
+	tmpFilePath := filepath.Join(os.TempDir(), "pages-0.img")
+	tmpFile, err := os.Create(tmpFilePath)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	defer func() {
+		tmpFile.Close()
+		if err := os.Remove(tmpFile.Name()); err != nil {
+			t.Fatal(err)
+		}
+	}()
+
+	mr, err := NewMemoryReader(testImgsDir, pid, sysPageSize)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	testCases := []testcase{
+		{
+			name: "Zero memory area size",
+			mr: &MemoryReader{
+				checkpointDir:  testImgsDir,
+				pid:            pid,
+				pageSize:       sysPageSize,
+				pagesID:        mr.pagesID,
+				pagemapEntries: mr.pagemapEntries,
+			},
+			start:         0,
+			end:           0,
+			expectedError: nil,
+		},
+		{
+			name: "Valid pagemap entry 1",
+			mr: &MemoryReader{
+				checkpointDir:  testImgsDir,
+				pid:            pid,
+				pageSize:       sysPageSize,
+				pagesID:        mr.pagesID,
+				pagemapEntries: mr.pagemapEntries,
+			},
+			start:         mr.pagemapEntries[0].GetVaddr(),
+			end:           mr.pagemapEntries[0].GetVaddr() + uint64(uint32(sysPageSize)*mr.pagemapEntries[0].GetNrPages()),
+			expectedError: nil,
+		},
+		{
+			name: "Valid pagemap entry 2",
+			mr: &MemoryReader{
+				checkpointDir:  testImgsDir,
+				pid:            pid,
+				pageSize:       sysPageSize,
+				pagesID:        mr.pagesID,
+				pagemapEntries: mr.pagemapEntries,
+			},
+			start:         mr.pagemapEntries[1].GetVaddr(),
+			end:           mr.pagemapEntries[1].GetVaddr() + uint64(uint32(sysPageSize)*mr.pagemapEntries[1].GetNrPages()),
+			expectedError: nil,
+		},
+		{
+			name: "Invalid pages file",
+			mr: &MemoryReader{
+				checkpointDir:  testImgsDir,
+				pid:            pid,
+				pageSize:       sysPageSize,
+				pagesID:        mr.pagesID + 1,
+				pagemapEntries: mr.pagemapEntries,
+			},
+			start:         mr.pagemapEntries[0].GetVaddr(),
+			end:           mr.pagemapEntries[0].GetVaddr() + uint64(uint32(sysPageSize)*mr.pagemapEntries[0].GetNrPages()),
+			expectedError: errors.New("no such file or directory"),
+		},
+		{
+			name: "Empty pages file",
+			mr: &MemoryReader{
+				checkpointDir:  os.TempDir(),
+				pid:            pid,
+				pageSize:       sysPageSize,
+				pagesID:        0,
+				pagemapEntries: mr.pagemapEntries,
+			},
+			start:         mr.pagemapEntries[1].GetVaddr(),
+			end:           mr.pagemapEntries[1].GetVaddr() + uint64(uint32(sysPageSize)*mr.pagemapEntries[1].GetNrPages()),
+			expectedError: errors.New("EOF"),
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			buff, err := tc.mr.GetMemPages(tc.start, tc.end)
+			if err != nil && tc.expectedError != nil {
+				if !strings.Contains(err.Error(), tc.expectedError.Error()) {
+					t.Errorf("Expected error: %v, got error: %v", tc.expectedError, err)
+				}
+			}
+
+			if tc.expectedError == nil && buff == nil {
+				t.Errorf("Returned memory chunk is expected to be non-empty")
+			}
+		})
+	}
+}
+
+// helper function to get the PID from the test-imgs directory
+func getTestImgPID(dir string) (uint32, error) {
+	psTreeImg, err := getImg(filepath.Join(dir, "pstree.img"), &pstree.PstreeEntry{})
+	if err != nil {
+		return 0, err
+	}
+	process := psTreeImg.Entries[0].Message.(*pstree.PstreeEntry)
+
+	return process.GetPgid(), nil
+}