Bump Buildah to v1.18.0, c/storage to v1.24.0

Update to Buildah v1.18.0 and c/storage to v1.24

Signed-off-by: TomSweeneyRedHat <tsweeney@redhat.com>

9 files changed, 104 insertions, 1913 deletions

diff --git a/vendor/github.com/Microsoft/go-winio/archive/tar/LICENSE b/vendor/github.com/Microsoft/go-winio/archive/tar/LICENSE
deleted file mode 100644
index 744875676..000000000
--- a/vendor/github.com/Microsoft/go-winio/archive/tar/LICENSE
+++ /dev/null
@@ -1,27 +0,0 @@
-Copyright (c) 2012 The Go Authors. All rights reserved.
-
-Redistribution and use in source and binary forms, with or without
-modification, are permitted provided that the following conditions are
-met:
-
-   * Redistributions of source code must retain the above copyright
-notice, this list of conditions and the following disclaimer.
-   * Redistributions in binary form must reproduce the above
-copyright notice, this list of conditions and the following disclaimer
-in the documentation and/or other materials provided with the
-distribution.
-   * Neither the name of Google Inc. nor the names of its
-contributors may be used to endorse or promote products derived from
-this software without specific prior written permission.
-
-THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
-"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
-LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
-A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
-OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
-SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
-LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
-DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
-THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
-(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
-OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
diff --git a/vendor/github.com/Microsoft/go-winio/archive/tar/common.go b/vendor/github.com/Microsoft/go-winio/archive/tar/common.go
deleted file mode 100644
index 0378401c0..000000000
--- a/vendor/github.com/Microsoft/go-winio/archive/tar/common.go
+++ /dev/null
@@ -1,344 +0,0 @@
-// Copyright 2009 The Go Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE file.
-
-// Package tar implements access to tar archives.
-// It aims to cover most of the variations, including those produced
-// by GNU and BSD tars.
-//
-// References:
-//   http://www.freebsd.org/cgi/man.cgi?query=tar&sektion=5
-//   http://www.gnu.org/software/tar/manual/html_node/Standard.html
-//   http://pubs.opengroup.org/onlinepubs/9699919799/utilities/pax.html
-package tar
-
-import (
-	"bytes"
-	"errors"
-	"fmt"
-	"os"
-	"path"
-	"time"
-)
-
-const (
-	blockSize = 512
-
-	// Types
-	TypeReg           = '0'    // regular file
-	TypeRegA          = '\x00' // regular file
-	TypeLink          = '1'    // hard link
-	TypeSymlink       = '2'    // symbolic link
-	TypeChar          = '3'    // character device node
-	TypeBlock         = '4'    // block device node
-	TypeDir           = '5'    // directory
-	TypeFifo          = '6'    // fifo node
-	TypeCont          = '7'    // reserved
-	TypeXHeader       = 'x'    // extended header
-	TypeXGlobalHeader = 'g'    // global extended header
-	TypeGNULongName   = 'L'    // Next file has a long name
-	TypeGNULongLink   = 'K'    // Next file symlinks to a file w/ a long name
-	TypeGNUSparse     = 'S'    // sparse file
-)
-
-// A Header represents a single header in a tar archive.
-// Some fields may not be populated.
-type Header struct {
-	Name         string    // name of header file entry
-	Mode         int64     // permission and mode bits
-	Uid          int       // user id of owner
-	Gid          int       // group id of owner
-	Size         int64     // length in bytes
-	ModTime      time.Time // modified time
-	Typeflag     byte      // type of header entry
-	Linkname     string    // target name of link
-	Uname        string    // user name of owner
-	Gname        string    // group name of owner
-	Devmajor     int64     // major number of character or block device
-	Devminor     int64     // minor number of character or block device
-	AccessTime   time.Time // access time
-	ChangeTime   time.Time // status change time
-	CreationTime time.Time // creation time
-	Xattrs       map[string]string
-	Winheaders   map[string]string
-}
-
-// File name constants from the tar spec.
-const (
-	fileNameSize       = 100 // Maximum number of bytes in a standard tar name.
-	fileNamePrefixSize = 155 // Maximum number of ustar extension bytes.
-)
-
-// FileInfo returns an os.FileInfo for the Header.
-func (h *Header) FileInfo() os.FileInfo {
-	return headerFileInfo{h}
-}
-
-// headerFileInfo implements os.FileInfo.
-type headerFileInfo struct {
-	h *Header
-}
-
-func (fi headerFileInfo) Size() int64        { return fi.h.Size }
-func (fi headerFileInfo) IsDir() bool        { return fi.Mode().IsDir() }
-func (fi headerFileInfo) ModTime() time.Time { return fi.h.ModTime }
-func (fi headerFileInfo) Sys() interface{}   { return fi.h }
-
-// Name returns the base name of the file.
-func (fi headerFileInfo) Name() string {
-	if fi.IsDir() {
-		return path.Base(path.Clean(fi.h.Name))
-	}
-	return path.Base(fi.h.Name)
-}
-
-// Mode returns the permission and mode bits for the headerFileInfo.
-func (fi headerFileInfo) Mode() (mode os.FileMode) {
-	// Set file permission bits.
-	mode = os.FileMode(fi.h.Mode).Perm()
-
-	// Set setuid, setgid and sticky bits.
-	if fi.h.Mode&c_ISUID != 0 {
-		// setuid
-		mode |= os.ModeSetuid
-	}
-	if fi.h.Mode&c_ISGID != 0 {
-		// setgid
-		mode |= os.ModeSetgid
-	}
-	if fi.h.Mode&c_ISVTX != 0 {
-		// sticky
-		mode |= os.ModeSticky
-	}
-
-	// Set file mode bits.
-	// clear perm, setuid, setgid and sticky bits.
-	m := os.FileMode(fi.h.Mode) &^ 07777
-	if m == c_ISDIR {
-		// directory
-		mode |= os.ModeDir
-	}
-	if m == c_ISFIFO {
-		// named pipe (FIFO)
-		mode |= os.ModeNamedPipe
-	}
-	if m == c_ISLNK {
-		// symbolic link
-		mode |= os.ModeSymlink
-	}
-	if m == c_ISBLK {
-		// device file
-		mode |= os.ModeDevice
-	}
-	if m == c_ISCHR {
-		// Unix character device
-		mode |= os.ModeDevice
-		mode |= os.ModeCharDevice
-	}
-	if m == c_ISSOCK {
-		// Unix domain socket
-		mode |= os.ModeSocket
-	}
-
-	switch fi.h.Typeflag {
-	case TypeSymlink:
-		// symbolic link
-		mode |= os.ModeSymlink
-	case TypeChar:
-		// character device node
-		mode |= os.ModeDevice
-		mode |= os.ModeCharDevice
-	case TypeBlock:
-		// block device node
-		mode |= os.ModeDevice
-	case TypeDir:
-		// directory
-		mode |= os.ModeDir
-	case TypeFifo:
-		// fifo node
-		mode |= os.ModeNamedPipe
-	}
-
-	return mode
-}
-
-// sysStat, if non-nil, populates h from system-dependent fields of fi.
-var sysStat func(fi os.FileInfo, h *Header) error
-
-// Mode constants from the tar spec.
-const (
-	c_ISUID  = 04000   // Set uid
-	c_ISGID  = 02000   // Set gid
-	c_ISVTX  = 01000   // Save text (sticky bit)
-	c_ISDIR  = 040000  // Directory
-	c_ISFIFO = 010000  // FIFO
-	c_ISREG  = 0100000 // Regular file
-	c_ISLNK  = 0120000 // Symbolic link
-	c_ISBLK  = 060000  // Block special file
-	c_ISCHR  = 020000  // Character special file
-	c_ISSOCK = 0140000 // Socket
-)
-
-// Keywords for the PAX Extended Header
-const (
-	paxAtime        = "atime"
-	paxCharset      = "charset"
-	paxComment      = "comment"
-	paxCtime        = "ctime" // please note that ctime is not a valid pax header.
-	paxCreationTime = "LIBARCHIVE.creationtime"
-	paxGid          = "gid"
-	paxGname        = "gname"
-	paxLinkpath     = "linkpath"
-	paxMtime        = "mtime"
-	paxPath         = "path"
-	paxSize         = "size"
-	paxUid          = "uid"
-	paxUname        = "uname"
-	paxXattr        = "SCHILY.xattr."
-	paxWindows      = "MSWINDOWS."
-	paxNone         = ""
-)
-
-// FileInfoHeader creates a partially-populated Header from fi.
-// If fi describes a symlink, FileInfoHeader records link as the link target.
-// If fi describes a directory, a slash is appended to the name.
-// Because os.FileInfo's Name method returns only the base name of
-// the file it describes, it may be necessary to modify the Name field
-// of the returned header to provide the full path name of the file.
-func FileInfoHeader(fi os.FileInfo, link string) (*Header, error) {
-	if fi == nil {
-		return nil, errors.New("tar: FileInfo is nil")
-	}
-	fm := fi.Mode()
-	h := &Header{
-		Name:    fi.Name(),
-		ModTime: fi.ModTime(),
-		Mode:    int64(fm.Perm()), // or'd with c_IS* constants later
-	}
-	switch {
-	case fm.IsRegular():
-		h.Mode |= c_ISREG
-		h.Typeflag = TypeReg
-		h.Size = fi.Size()
-	case fi.IsDir():
-		h.Typeflag = TypeDir
-		h.Mode |= c_ISDIR
-		h.Name += "/"
-	case fm&os.ModeSymlink != 0:
-		h.Typeflag = TypeSymlink
-		h.Mode |= c_ISLNK
-		h.Linkname = link
-	case fm&os.ModeDevice != 0:
-		if fm&os.ModeCharDevice != 0 {
-			h.Mode |= c_ISCHR
-			h.Typeflag = TypeChar
-		} else {
-			h.Mode |= c_ISBLK
-			h.Typeflag = TypeBlock
-		}
-	case fm&os.ModeNamedPipe != 0:
-		h.Typeflag = TypeFifo
-		h.Mode |= c_ISFIFO
-	case fm&os.ModeSocket != 0:
-		h.Mode |= c_ISSOCK
-	default:
-		return nil, fmt.Errorf("archive/tar: unknown file mode %v", fm)
-	}
-	if fm&os.ModeSetuid != 0 {
-		h.Mode |= c_ISUID
-	}
-	if fm&os.ModeSetgid != 0 {
-		h.Mode |= c_ISGID
-	}
-	if fm&os.ModeSticky != 0 {
-		h.Mode |= c_ISVTX
-	}
-	// If possible, populate additional fields from OS-specific
-	// FileInfo fields.
-	if sys, ok := fi.Sys().(*Header); ok {
-		// This FileInfo came from a Header (not the OS). Use the
-		// original Header to populate all remaining fields.
-		h.Uid = sys.Uid
-		h.Gid = sys.Gid
-		h.Uname = sys.Uname
-		h.Gname = sys.Gname
-		h.AccessTime = sys.AccessTime
-		h.ChangeTime = sys.ChangeTime
-		if sys.Xattrs != nil {
-			h.Xattrs = make(map[string]string)
-			for k, v := range sys.Xattrs {
-				h.Xattrs[k] = v
-			}
-		}
-		if sys.Typeflag == TypeLink {
-			// hard link
-			h.Typeflag = TypeLink
-			h.Size = 0
-			h.Linkname = sys.Linkname
-		}
-	}
-	if sysStat != nil {
-		return h, sysStat(fi, h)
-	}
-	return h, nil
-}
-
-var zeroBlock = make([]byte, blockSize)
-
-// POSIX specifies a sum of the unsigned byte values, but the Sun tar uses signed byte values.
-// We compute and return both.
-func checksum(header []byte) (unsigned int64, signed int64) {
-	for i := 0; i < len(header); i++ {
-		if i == 148 {
-			// The chksum field (header[148:156]) is special: it should be treated as space bytes.
-			unsigned += ' ' * 8
-			signed += ' ' * 8
-			i += 7
-			continue
-		}
-		unsigned += int64(header[i])
-		signed += int64(int8(header[i]))
-	}
-	return
-}
-
-type slicer []byte
-
-func (sp *slicer) next(n int) (b []byte) {
-	s := *sp
-	b, *sp = s[0:n], s[n:]
-	return
-}
-
-func isASCII(s string) bool {
-	for _, c := range s {
-		if c >= 0x80 {
-			return false
-		}
-	}
-	return true
-}
-
-func toASCII(s string) string {
-	if isASCII(s) {
-		return s
-	}
-	var buf bytes.Buffer
-	for _, c := range s {
-		if c < 0x80 {
-			buf.WriteByte(byte(c))
-		}
-	}
-	return buf.String()
-}
-
-// isHeaderOnlyType checks if the given type flag is of the type that has no
-// data section even if a size is specified.
-func isHeaderOnlyType(flag byte) bool {
-	switch flag {
-	case TypeLink, TypeSymlink, TypeChar, TypeBlock, TypeDir, TypeFifo:
-		return true
-	default:
-		return false
-	}
-}
diff --git a/vendor/github.com/Microsoft/go-winio/archive/tar/reader.go b/vendor/github.com/Microsoft/go-winio/archive/tar/reader.go
deleted file mode 100644
index e210c618a..000000000
--- a/vendor/github.com/Microsoft/go-winio/archive/tar/reader.go
+++ /dev/null
@@ -1,1002 +0,0 @@
-// Copyright 2009 The Go Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE file.
-
-package tar
-
-// TODO(dsymonds):
-//   - pax extensions
-
-import (
-	"bytes"
-	"errors"
-	"io"
-	"io/ioutil"
-	"math"
-	"os"
-	"strconv"
-	"strings"
-	"time"
-)
-
-var (
-	ErrHeader = errors.New("archive/tar: invalid tar header")
-)
-
-const maxNanoSecondIntSize = 9
-
-// A Reader provides sequential access to the contents of a tar archive.
-// A tar archive consists of a sequence of files.
-// The Next method advances to the next file in the archive (including the first),
-// and then it can be treated as an io.Reader to access the file's data.
-type Reader struct {
-	r       io.Reader
-	err     error
-	pad     int64           // amount of padding (ignored) after current file entry
-	curr    numBytesReader  // reader for current file entry
-	hdrBuff [blockSize]byte // buffer to use in readHeader
-}
-
-type parser struct {
-	err error // Last error seen
-}
-
-// A numBytesReader is an io.Reader with a numBytes method, returning the number
-// of bytes remaining in the underlying encoded data.
-type numBytesReader interface {
-	io.Reader
-	numBytes() int64
-}
-
-// A regFileReader is a numBytesReader for reading file data from a tar archive.
-type regFileReader struct {
-	r  io.Reader // underlying reader
-	nb int64     // number of unread bytes for current file entry
-}
-
-// A sparseFileReader is a numBytesReader for reading sparse file data from a
-// tar archive.
-type sparseFileReader struct {
-	rfr   numBytesReader // Reads the sparse-encoded file data
-	sp    []sparseEntry  // The sparse map for the file
-	pos   int64          // Keeps track of file position
-	total int64          // Total size of the file
-}
-
-// A sparseEntry holds a single entry in a sparse file's sparse map.
-//
-// Sparse files are represented using a series of sparseEntrys.
-// Despite the name, a sparseEntry represents an actual data fragment that
-// references data found in the underlying archive stream. All regions not
-// covered by a sparseEntry are logically filled with zeros.
-//
-// For example, if the underlying raw file contains the 10-byte data:
-//	var compactData = "abcdefgh"
-//
-// And the sparse map has the following entries:
-//	var sp = []sparseEntry{
-//		{offset: 2,  numBytes: 5} // Data fragment for [2..7]
-//		{offset: 18, numBytes: 3} // Data fragment for [18..21]
-//	}
-//
-// Then the content of the resulting sparse file with a "real" size of 25 is:
-//	var sparseData = "\x00"*2 + "abcde" + "\x00"*11 + "fgh" + "\x00"*4
-type sparseEntry struct {
-	offset   int64 // Starting position of the fragment
-	numBytes int64 // Length of the fragment
-}
-
-// Keywords for GNU sparse files in a PAX extended header
-const (
-	paxGNUSparseNumBlocks = "GNU.sparse.numblocks"
-	paxGNUSparseOffset    = "GNU.sparse.offset"
-	paxGNUSparseNumBytes  = "GNU.sparse.numbytes"
-	paxGNUSparseMap       = "GNU.sparse.map"
-	paxGNUSparseName      = "GNU.sparse.name"
-	paxGNUSparseMajor     = "GNU.sparse.major"
-	paxGNUSparseMinor     = "GNU.sparse.minor"
-	paxGNUSparseSize      = "GNU.sparse.size"
-	paxGNUSparseRealSize  = "GNU.sparse.realsize"
-)
-
-// Keywords for old GNU sparse headers
-const (
-	oldGNUSparseMainHeaderOffset               = 386
-	oldGNUSparseMainHeaderIsExtendedOffset     = 482
-	oldGNUSparseMainHeaderNumEntries           = 4
-	oldGNUSparseExtendedHeaderIsExtendedOffset = 504
-	oldGNUSparseExtendedHeaderNumEntries       = 21
-	oldGNUSparseOffsetSize                     = 12
-	oldGNUSparseNumBytesSize                   = 12
-)
-
-// NewReader creates a new Reader reading from r.
-func NewReader(r io.Reader) *Reader { return &Reader{r: r} }
-
-// Next advances to the next entry in the tar archive.
-//
-// io.EOF is returned at the end of the input.
-func (tr *Reader) Next() (*Header, error) {
-	if tr.err != nil {
-		return nil, tr.err
-	}
-
-	var hdr *Header
-	var extHdrs map[string]string
-
-	// Externally, Next iterates through the tar archive as if it is a series of
-	// files. Internally, the tar format often uses fake "files" to add meta
-	// data that describes the next file. These meta data "files" should not
-	// normally be visible to the outside. As such, this loop iterates through
-	// one or more "header files" until it finds a "normal file".
-loop:
-	for {
-		tr.err = tr.skipUnread()
-		if tr.err != nil {
-			return nil, tr.err
-		}
-
-		hdr = tr.readHeader()
-		if tr.err != nil {
-			return nil, tr.err
-		}
-
-		// Check for PAX/GNU special headers and files.
-		switch hdr.Typeflag {
-		case TypeXHeader:
-			extHdrs, tr.err = parsePAX(tr)
-			if tr.err != nil {
-				return nil, tr.err
-			}
-			continue loop // This is a meta header affecting the next header
-		case TypeGNULongName, TypeGNULongLink:
-			var realname []byte
-			realname, tr.err = ioutil.ReadAll(tr)
-			if tr.err != nil {
-				return nil, tr.err
-			}
-
-			// Convert GNU extensions to use PAX headers.
-			if extHdrs == nil {
-				extHdrs = make(map[string]string)
-			}
-			var p parser
-			switch hdr.Typeflag {
-			case TypeGNULongName:
-				extHdrs[paxPath] = p.parseString(realname)
-			case TypeGNULongLink:
-				extHdrs[paxLinkpath] = p.parseString(realname)
-			}
-			if p.err != nil {
-				tr.err = p.err
-				return nil, tr.err
-			}
-			continue loop // This is a meta header affecting the next header
-		default:
-			mergePAX(hdr, extHdrs)
-
-			// Check for a PAX format sparse file
-			sp, err := tr.checkForGNUSparsePAXHeaders(hdr, extHdrs)
-			if err != nil {
-				tr.err = err
-				return nil, err
-			}
-			if sp != nil {
-				// Current file is a PAX format GNU sparse file.
-				// Set the current file reader to a sparse file reader.
-				tr.curr, tr.err = newSparseFileReader(tr.curr, sp, hdr.Size)
-				if tr.err != nil {
-					return nil, tr.err
-				}
-			}
-			break loop // This is a file, so stop
-		}
-	}
-	return hdr, nil
-}
-
-// checkForGNUSparsePAXHeaders checks the PAX headers for GNU sparse headers. If they are found, then
-// this function reads the sparse map and returns it. Unknown sparse formats are ignored, causing the file to
-// be treated as a regular file.
-func (tr *Reader) checkForGNUSparsePAXHeaders(hdr *Header, headers map[string]string) ([]sparseEntry, error) {
-	var sparseFormat string
-
-	// Check for sparse format indicators
-	major, majorOk := headers[paxGNUSparseMajor]
-	minor, minorOk := headers[paxGNUSparseMinor]
-	sparseName, sparseNameOk := headers[paxGNUSparseName]
-	_, sparseMapOk := headers[paxGNUSparseMap]
-	sparseSize, sparseSizeOk := headers[paxGNUSparseSize]
-	sparseRealSize, sparseRealSizeOk := headers[paxGNUSparseRealSize]
-
-	// Identify which, if any, sparse format applies from which PAX headers are set
-	if majorOk && minorOk {
-		sparseFormat = major + "." + minor
-	} else if sparseNameOk && sparseMapOk {
-		sparseFormat = "0.1"
-	} else if sparseSizeOk {
-		sparseFormat = "0.0"
-	} else {
-		// Not a PAX format GNU sparse file.
-		return nil, nil
-	}
-
-	// Check for unknown sparse format
-	if sparseFormat != "0.0" && sparseFormat != "0.1" && sparseFormat != "1.0" {
-		return nil, nil
-	}
-
-	// Update hdr from GNU sparse PAX headers
-	if sparseNameOk {
-		hdr.Name = sparseName
-	}
-	if sparseSizeOk {
-		realSize, err := strconv.ParseInt(sparseSize, 10, 0)
-		if err != nil {
-			return nil, ErrHeader
-		}
-		hdr.Size = realSize
-	} else if sparseRealSizeOk {
-		realSize, err := strconv.ParseInt(sparseRealSize, 10, 0)
-		if err != nil {
-			return nil, ErrHeader
-		}
-		hdr.Size = realSize
-	}
-
-	// Set up the sparse map, according to the particular sparse format in use
-	var sp []sparseEntry
-	var err error
-	switch sparseFormat {
-	case "0.0", "0.1":
-		sp, err = readGNUSparseMap0x1(headers)
-	case "1.0":
-		sp, err = readGNUSparseMap1x0(tr.curr)
-	}
-	return sp, err
-}
-
-// mergePAX merges well known headers according to PAX standard.
-// In general headers with the same name as those found
-// in the header struct overwrite those found in the header
-// struct with higher precision or longer values. Esp. useful
-// for name and linkname fields.
-func mergePAX(hdr *Header, headers map[string]string) error {
-	for k, v := range headers {
-		switch k {
-		case paxPath:
-			hdr.Name = v
-		case paxLinkpath:
-			hdr.Linkname = v
-		case paxGname:
-			hdr.Gname = v
-		case paxUname:
-			hdr.Uname = v
-		case paxUid:
-			uid, err := strconv.ParseInt(v, 10, 0)
-			if err != nil {
-				return err
-			}
-			hdr.Uid = int(uid)
-		case paxGid:
-			gid, err := strconv.ParseInt(v, 10, 0)
-			if err != nil {
-				return err
-			}
-			hdr.Gid = int(gid)
-		case paxAtime:
-			t, err := parsePAXTime(v)
-			if err != nil {
-				return err
-			}
-			hdr.AccessTime = t
-		case paxMtime:
-			t, err := parsePAXTime(v)
-			if err != nil {
-				return err
-			}
-			hdr.ModTime = t
-		case paxCtime:
-			t, err := parsePAXTime(v)
-			if err != nil {
-				return err
-			}
-			hdr.ChangeTime = t
-		case paxCreationTime:
-			t, err := parsePAXTime(v)
-			if err != nil {
-				return err
-			}
-			hdr.CreationTime = t
-		case paxSize:
-			size, err := strconv.ParseInt(v, 10, 0)
-			if err != nil {
-				return err
-			}
-			hdr.Size = int64(size)
-		default:
-			if strings.HasPrefix(k, paxXattr) {
-				if hdr.Xattrs == nil {
-					hdr.Xattrs = make(map[string]string)
-				}
-				hdr.Xattrs[k[len(paxXattr):]] = v
-			} else if strings.HasPrefix(k, paxWindows) {
-				if hdr.Winheaders == nil {
-					hdr.Winheaders = make(map[string]string)
-				}
-				hdr.Winheaders[k[len(paxWindows):]] = v
-			}
-		}
-	}
-	return nil
-}
-
-// parsePAXTime takes a string of the form %d.%d as described in
-// the PAX specification.
-func parsePAXTime(t string) (time.Time, error) {
-	buf := []byte(t)
-	pos := bytes.IndexByte(buf, '.')
-	var seconds, nanoseconds int64
-	var err error
-	if pos == -1 {
-		seconds, err = strconv.ParseInt(t, 10, 0)
-		if err != nil {
-			return time.Time{}, err
-		}
-	} else {
-		seconds, err = strconv.ParseInt(string(buf[:pos]), 10, 0)
-		if err != nil {
-			return time.Time{}, err
-		}
-		nano_buf := string(buf[pos+1:])
-		// Pad as needed before converting to a decimal.
-		// For example .030 -> .030000000 -> 30000000 nanoseconds
-		if len(nano_buf) < maxNanoSecondIntSize {
-			// Right pad
-			nano_buf += strings.Repeat("0", maxNanoSecondIntSize-len(nano_buf))
-		} else if len(nano_buf) > maxNanoSecondIntSize {
-			// Right truncate
-			nano_buf = nano_buf[:maxNanoSecondIntSize]
-		}
-		nanoseconds, err = strconv.ParseInt(string(nano_buf), 10, 0)
-		if err != nil {
-			return time.Time{}, err
-		}
-	}
-	ts := time.Unix(seconds, nanoseconds)
-	return ts, nil
-}
-
-// parsePAX parses PAX headers.
-// If an extended header (type 'x') is invalid, ErrHeader is returned
-func parsePAX(r io.Reader) (map[string]string, error) {
-	buf, err := ioutil.ReadAll(r)
-	if err != nil {
-		return nil, err
-	}
-	sbuf := string(buf)
-
-	// For GNU PAX sparse format 0.0 support.
-	// This function transforms the sparse format 0.0 headers into sparse format 0.1 headers.
-	var sparseMap bytes.Buffer
-
-	headers := make(map[string]string)
-	// Each record is constructed as
-	//     "%d %s=%s\n", length, keyword, value
-	for len(sbuf) > 0 {
-		key, value, residual, err := parsePAXRecord(sbuf)
-		if err != nil {
-			return nil, ErrHeader
-		}
-		sbuf = residual
-
-		keyStr := string(key)
-		if keyStr == paxGNUSparseOffset || keyStr == paxGNUSparseNumBytes {
-			// GNU sparse format 0.0 special key. Write to sparseMap instead of using the headers map.
-			sparseMap.WriteString(value)
-			sparseMap.Write([]byte{','})
-		} else {
-			// Normal key. Set the value in the headers map.
-			headers[keyStr] = string(value)
-		}
-	}
-	if sparseMap.Len() != 0 {
-		// Add sparse info to headers, chopping off the extra comma
-		sparseMap.Truncate(sparseMap.Len() - 1)
-		headers[paxGNUSparseMap] = sparseMap.String()
-	}
-	return headers, nil
-}
-
-// parsePAXRecord parses the input PAX record string into a key-value pair.
-// If parsing is successful, it will slice off the currently read record and
-// return the remainder as r.
-//
-// A PAX record is of the following form:
-//	"%d %s=%s\n" % (size, key, value)
-func parsePAXRecord(s string) (k, v, r string, err error) {
-	// The size field ends at the first space.
-	sp := strings.IndexByte(s, ' ')
-	if sp == -1 {
-		return "", "", s, ErrHeader
-	}
-
-	// Parse the first token as a decimal integer.
-	n, perr := strconv.ParseInt(s[:sp], 10, 0) // Intentionally parse as native int
-	if perr != nil || n < 5 || int64(len(s)) < n {
-		return "", "", s, ErrHeader
-	}
-
-	// Extract everything between the space and the final newline.
-	rec, nl, rem := s[sp+1:n-1], s[n-1:n], s[n:]
-	if nl != "\n" {
-		return "", "", s, ErrHeader
-	}
-
-	// The first equals separates the key from the value.
-	eq := strings.IndexByte(rec, '=')
-	if eq == -1 {
-		return "", "", s, ErrHeader
-	}
-	return rec[:eq], rec[eq+1:], rem, nil
-}
-
-// parseString parses bytes as a NUL-terminated C-style string.
-// If a NUL byte is not found then the whole slice is returned as a string.
-func (*parser) parseString(b []byte) string {
-	n := 0
-	for n < len(b) && b[n] != 0 {
-		n++
-	}
-	return string(b[0:n])
-}
-
-// parseNumeric parses the input as being encoded in either base-256 or octal.
-// This function may return negative numbers.
-// If parsing fails or an integer overflow occurs, err will be set.
-func (p *parser) parseNumeric(b []byte) int64 {
-	// Check for base-256 (binary) format first.
-	// If the first bit is set, then all following bits constitute a two's
-	// complement encoded number in big-endian byte order.
-	if len(b) > 0 && b[0]&0x80 != 0 {
-		// Handling negative numbers relies on the following identity:
-		//	-a-1 == ^a
-		//
-		// If the number is negative, we use an inversion mask to invert the
-		// data bytes and treat the value as an unsigned number.
-		var inv byte // 0x00 if positive or zero, 0xff if negative
-		if b[0]&0x40 != 0 {
-			inv = 0xff
-		}
-
-		var x uint64
-		for i, c := range b {
-			c ^= inv // Inverts c only if inv is 0xff, otherwise does nothing
-			if i == 0 {
-				c &= 0x7f // Ignore signal bit in first byte
-			}
-			if (x >> 56) > 0 {
-				p.err = ErrHeader // Integer overflow
-				return 0
-			}
-			x = x<<8 | uint64(c)
-		}
-		if (x >> 63) > 0 {
-			p.err = ErrHeader // Integer overflow
-			return 0
-		}
-		if inv == 0xff {
-			return ^int64(x)
-		}
-		return int64(x)
-	}
-
-	// Normal case is base-8 (octal) format.
-	return p.parseOctal(b)
-}
-
-func (p *parser) parseOctal(b []byte) int64 {
-	// Because unused fields are filled with NULs, we need
-	// to skip leading NULs. Fields may also be padded with
-	// spaces or NULs.
-	// So we remove leading and trailing NULs and spaces to
-	// be sure.
-	b = bytes.Trim(b, " \x00")
-
-	if len(b) == 0 {
-		return 0
-	}
-	x, perr := strconv.ParseUint(p.parseString(b), 8, 64)
-	if perr != nil {
-		p.err = ErrHeader
-	}
-	return int64(x)
-}
-
-// skipUnread skips any unread bytes in the existing file entry, as well as any
-// alignment padding. It returns io.ErrUnexpectedEOF if any io.EOF is
-// encountered in the data portion; it is okay to hit io.EOF in the padding.
-//
-// Note that this function still works properly even when sparse files are being
-// used since numBytes returns the bytes remaining in the underlying io.Reader.
-func (tr *Reader) skipUnread() error {
-	dataSkip := tr.numBytes()      // Number of data bytes to skip
-	totalSkip := dataSkip + tr.pad // Total number of bytes to skip
-	tr.curr, tr.pad = nil, 0
-
-	// If possible, Seek to the last byte before the end of the data section.
-	// Do this because Seek is often lazy about reporting errors; this will mask
-	// the fact that the tar stream may be truncated. We can rely on the
-	// io.CopyN done shortly afterwards to trigger any IO errors.
-	var seekSkipped int64 // Number of bytes skipped via Seek
-	if sr, ok := tr.r.(io.Seeker); ok && dataSkip > 1 {
-		// Not all io.Seeker can actually Seek. For example, os.Stdin implements
-		// io.Seeker, but calling Seek always returns an error and performs
-		// no action. Thus, we try an innocent seek to the current position
-		// to see if Seek is really supported.
-		pos1, err := sr.Seek(0, os.SEEK_CUR)
-		if err == nil {
-			// Seek seems supported, so perform the real Seek.
-			pos2, err := sr.Seek(dataSkip-1, os.SEEK_CUR)
-			if err != nil {
-				tr.err = err
-				return tr.err
-			}
-			seekSkipped = pos2 - pos1
-		}
-	}
-
-	var copySkipped int64 // Number of bytes skipped via CopyN
-	copySkipped, tr.err = io.CopyN(ioutil.Discard, tr.r, totalSkip-seekSkipped)
-	if tr.err == io.EOF && seekSkipped+copySkipped < dataSkip {
-		tr.err = io.ErrUnexpectedEOF
-	}
-	return tr.err
-}
-
-func (tr *Reader) verifyChecksum(header []byte) bool {
-	if tr.err != nil {
-		return false
-	}
-
-	var p parser
-	given := p.parseOctal(header[148:156])
-	unsigned, signed := checksum(header)
-	return p.err == nil && (given == unsigned || given == signed)
-}
-
-// readHeader reads the next block header and assumes that the underlying reader
-// is already aligned to a block boundary.
-//
-// The err will be set to io.EOF only when one of the following occurs:
-//	* Exactly 0 bytes are read and EOF is hit.
-//	* Exactly 1 block of zeros is read and EOF is hit.
-//	* At least 2 blocks of zeros are read.
-func (tr *Reader) readHeader() *Header {
-	header := tr.hdrBuff[:]
-	copy(header, zeroBlock)
-
-	if _, tr.err = io.ReadFull(tr.r, header); tr.err != nil {
-		return nil // io.EOF is okay here
-	}
-
-	// Two blocks of zero bytes marks the end of the archive.
-	if bytes.Equal(header, zeroBlock[0:blockSize]) {
-		if _, tr.err = io.ReadFull(tr.r, header); tr.err != nil {
-			return nil // io.EOF is okay here
-		}
-		if bytes.Equal(header, zeroBlock[0:blockSize]) {
-			tr.err = io.EOF
-		} else {
-			tr.err = ErrHeader // zero block and then non-zero block
-		}
-		return nil
-	}
-
-	if !tr.verifyChecksum(header) {
-		tr.err = ErrHeader
-		return nil
-	}
-
-	// Unpack
-	var p parser
-	hdr := new(Header)
-	s := slicer(header)
-
-	hdr.Name = p.parseString(s.next(100))
-	hdr.Mode = p.parseNumeric(s.next(8))
-	hdr.Uid = int(p.parseNumeric(s.next(8)))
-	hdr.Gid = int(p.parseNumeric(s.next(8)))
-	hdr.Size = p.parseNumeric(s.next(12))
-	hdr.ModTime = time.Unix(p.parseNumeric(s.next(12)), 0)
-	s.next(8) // chksum
-	hdr.Typeflag = s.next(1)[0]
-	hdr.Linkname = p.parseString(s.next(100))
-
-	// The remainder of the header depends on the value of magic.
-	// The original (v7) version of tar had no explicit magic field,
-	// so its magic bytes, like the rest of the block, are NULs.
-	magic := string(s.next(8)) // contains version field as well.
-	var format string
-	switch {
-	case magic[:6] == "ustar\x00": // POSIX tar (1003.1-1988)
-		if string(header[508:512]) == "tar\x00" {
-			format = "star"
-		} else {
-			format = "posix"
-		}
-	case magic == "ustar  \x00": // old GNU tar
-		format = "gnu"
-	}
-
-	switch format {
-	case "posix", "gnu", "star":
-		hdr.Uname = p.parseString(s.next(32))
-		hdr.Gname = p.parseString(s.next(32))
-		devmajor := s.next(8)
-		devminor := s.next(8)
-		if hdr.Typeflag == TypeChar || hdr.Typeflag == TypeBlock {
-			hdr.Devmajor = p.parseNumeric(devmajor)
-			hdr.Devminor = p.parseNumeric(devminor)
-		}
-		var prefix string
-		switch format {
-		case "posix", "gnu":
-			prefix = p.parseString(s.next(155))
-		case "star":
-			prefix = p.parseString(s.next(131))
-			hdr.AccessTime = time.Unix(p.parseNumeric(s.next(12)), 0)
-			hdr.ChangeTime = time.Unix(p.parseNumeric(s.next(12)), 0)
-		}
-		if len(prefix) > 0 {
-			hdr.Name = prefix + "/" + hdr.Name
-		}
-	}
-
-	if p.err != nil {
-		tr.err = p.err
-		return nil
-	}
-
-	nb := hdr.Size
-	if isHeaderOnlyType(hdr.Typeflag) {
-		nb = 0
-	}
-	if nb < 0 {
-		tr.err = ErrHeader
-		return nil
-	}
-
-	// Set the current file reader.
-	tr.pad = -nb & (blockSize - 1) // blockSize is a power of two
-	tr.curr = &regFileReader{r: tr.r, nb: nb}
-
-	// Check for old GNU sparse format entry.
-	if hdr.Typeflag == TypeGNUSparse {
-		// Get the real size of the file.
-		hdr.Size = p.parseNumeric(header[483:495])
-		if p.err != nil {
-			tr.err = p.err
-			return nil
-		}
-
-		// Read the sparse map.
-		sp := tr.readOldGNUSparseMap(header)
-		if tr.err != nil {
-			return nil
-		}
-
-		// Current file is a GNU sparse file. Update the current file reader.
-		tr.curr, tr.err = newSparseFileReader(tr.curr, sp, hdr.Size)
-		if tr.err != nil {
-			return nil
-		}
-	}
-
-	return hdr
-}
-
-// readOldGNUSparseMap reads the sparse map as stored in the old GNU sparse format.
-// The sparse map is stored in the tar header if it's small enough. If it's larger than four entries,
-// then one or more extension headers are used to store the rest of the sparse map.
-func (tr *Reader) readOldGNUSparseMap(header []byte) []sparseEntry {
-	var p parser
-	isExtended := header[oldGNUSparseMainHeaderIsExtendedOffset] != 0
-	spCap := oldGNUSparseMainHeaderNumEntries
-	if isExtended {
-		spCap += oldGNUSparseExtendedHeaderNumEntries
-	}
-	sp := make([]sparseEntry, 0, spCap)
-	s := slicer(header[oldGNUSparseMainHeaderOffset:])
-
-	// Read the four entries from the main tar header
-	for i := 0; i < oldGNUSparseMainHeaderNumEntries; i++ {
-		offset := p.parseNumeric(s.next(oldGNUSparseOffsetSize))
-		numBytes := p.parseNumeric(s.next(oldGNUSparseNumBytesSize))
-		if p.err != nil {
-			tr.err = p.err
-			return nil
-		}
-		if offset == 0 && numBytes == 0 {
-			break
-		}
-		sp = append(sp, sparseEntry{offset: offset, numBytes: numBytes})
-	}
-
-	for isExtended {
-		// There are more entries. Read an extension header and parse its entries.
-		sparseHeader := make([]byte, blockSize)
-		if _, tr.err = io.ReadFull(tr.r, sparseHeader); tr.err != nil {
-			return nil
-		}
-		isExtended = sparseHeader[oldGNUSparseExtendedHeaderIsExtendedOffset] != 0
-		s = slicer(sparseHeader)
-		for i := 0; i < oldGNUSparseExtendedHeaderNumEntries; i++ {
-			offset := p.parseNumeric(s.next(oldGNUSparseOffsetSize))
-			numBytes := p.parseNumeric(s.next(oldGNUSparseNumBytesSize))
-			if p.err != nil {
-				tr.err = p.err
-				return nil
-			}
-			if offset == 0 && numBytes == 0 {
-				break
-			}
-			sp = append(sp, sparseEntry{offset: offset, numBytes: numBytes})
-		}
-	}
-	return sp
-}
-
-// readGNUSparseMap1x0 reads the sparse map as stored in GNU's PAX sparse format
-// version 1.0. The format of the sparse map consists of a series of
-// newline-terminated numeric fields. The first field is the number of entries
-// and is always present. Following this are the entries, consisting of two
-// fields (offset, numBytes). This function must stop reading at the end
-// boundary of the block containing the last newline.
-//
-// Note that the GNU manual says that numeric values should be encoded in octal
-// format. However, the GNU tar utility itself outputs these values in decimal.
-// As such, this library treats values as being encoded in decimal.
-func readGNUSparseMap1x0(r io.Reader) ([]sparseEntry, error) {
-	var cntNewline int64
-	var buf bytes.Buffer
-	var blk = make([]byte, blockSize)
-
-	// feedTokens copies data in numBlock chunks from r into buf until there are
-	// at least cnt newlines in buf. It will not read more blocks than needed.
-	var feedTokens = func(cnt int64) error {
-		for cntNewline < cnt {
-			if _, err := io.ReadFull(r, blk); err != nil {
-				if err == io.EOF {
-					err = io.ErrUnexpectedEOF
-				}
-				return err
-			}
-			buf.Write(blk)
-			for _, c := range blk {
-				if c == '\n' {
-					cntNewline++
-				}
-			}
-		}
-		return nil
-	}
-
-	// nextToken gets the next token delimited by a newline. This assumes that
-	// at least one newline exists in the buffer.
-	var nextToken = func() string {
-		cntNewline--
-		tok, _ := buf.ReadString('\n')
-		return tok[:len(tok)-1] // Cut off newline
-	}
-
-	// Parse for the number of entries.
-	// Use integer overflow resistant math to check this.
-	if err := feedTokens(1); err != nil {
-		return nil, err
-	}
-	numEntries, err := strconv.ParseInt(nextToken(), 10, 0) // Intentionally parse as native int
-	if err != nil || numEntries < 0 || int(2*numEntries) < int(numEntries) {
-		return nil, ErrHeader
-	}
-
-	// Parse for all member entries.
-	// numEntries is trusted after this since a potential attacker must have
-	// committed resources proportional to what this library used.
-	if err := feedTokens(2 * numEntries); err != nil {
-		return nil, err
-	}
-	sp := make([]sparseEntry, 0, numEntries)
-	for i := int64(0); i < numEntries; i++ {
-		offset, err := strconv.ParseInt(nextToken(), 10, 64)
-		if err != nil {
-			return nil, ErrHeader
-		}
-		numBytes, err := strconv.ParseInt(nextToken(), 10, 64)
-		if err != nil {
-			return nil, ErrHeader
-		}
-		sp = append(sp, sparseEntry{offset: offset, numBytes: numBytes})
-	}
-	return sp, nil
-}
-
-// readGNUSparseMap0x1 reads the sparse map as stored in GNU's PAX sparse format
-// version 0.1. The sparse map is stored in the PAX headers.
-func readGNUSparseMap0x1(extHdrs map[string]string) ([]sparseEntry, error) {
-	// Get number of entries.
-	// Use integer overflow resistant math to check this.
-	numEntriesStr := extHdrs[paxGNUSparseNumBlocks]
-	numEntries, err := strconv.ParseInt(numEntriesStr, 10, 0) // Intentionally parse as native int
-	if err != nil || numEntries < 0 || int(2*numEntries) < int(numEntries) {
-		return nil, ErrHeader
-	}
-
-	// There should be two numbers in sparseMap for each entry.
-	sparseMap := strings.Split(extHdrs[paxGNUSparseMap], ",")
-	if int64(len(sparseMap)) != 2*numEntries {
-		return nil, ErrHeader
-	}
-
-	// Loop through the entries in the sparse map.
-	// numEntries is trusted now.
-	sp := make([]sparseEntry, 0, numEntries)
-	for i := int64(0); i < numEntries; i++ {
-		offset, err := strconv.ParseInt(sparseMap[2*i], 10, 64)
-		if err != nil {
-			return nil, ErrHeader
-		}
-		numBytes, err := strconv.ParseInt(sparseMap[2*i+1], 10, 64)
-		if err != nil {
-			return nil, ErrHeader
-		}
-		sp = append(sp, sparseEntry{offset: offset, numBytes: numBytes})
-	}
-	return sp, nil
-}
-
-// numBytes returns the number of bytes left to read in the current file's entry
-// in the tar archive, or 0 if there is no current file.
-func (tr *Reader) numBytes() int64 {
-	if tr.curr == nil {
-		// No current file, so no bytes
-		return 0
-	}
-	return tr.curr.numBytes()
-}
-
-// Read reads from the current entry in the tar archive.
-// It returns 0, io.EOF when it reaches the end of that entry,
-// until Next is called to advance to the next entry.
-//
-// Calling Read on special types like TypeLink, TypeSymLink, TypeChar,
-// TypeBlock, TypeDir, and TypeFifo returns 0, io.EOF regardless of what
-// the Header.Size claims.
-func (tr *Reader) Read(b []byte) (n int, err error) {
-	if tr.err != nil {
-		return 0, tr.err
-	}
-	if tr.curr == nil {
-		return 0, io.EOF
-	}
-
-	n, err = tr.curr.Read(b)
-	if err != nil && err != io.EOF {
-		tr.err = err
-	}
-	return
-}
-
-func (rfr *regFileReader) Read(b []byte) (n int, err error) {
-	if rfr.nb == 0 {
-		// file consumed
-		return 0, io.EOF
-	}
-	if int64(len(b)) > rfr.nb {
-		b = b[0:rfr.nb]
-	}
-	n, err = rfr.r.Read(b)
-	rfr.nb -= int64(n)
-
-	if err == io.EOF && rfr.nb > 0 {
-		err = io.ErrUnexpectedEOF
-	}
-	return
-}
-
-// numBytes returns the number of bytes left to read in the file's data in the tar archive.
-func (rfr *regFileReader) numBytes() int64 {
-	return rfr.nb
-}
-
-// newSparseFileReader creates a new sparseFileReader, but validates all of the
-// sparse entries before doing so.
-func newSparseFileReader(rfr numBytesReader, sp []sparseEntry, total int64) (*sparseFileReader, error) {
-	if total < 0 {
-		return nil, ErrHeader // Total size cannot be negative
-	}
-
-	// Validate all sparse entries. These are the same checks as performed by
-	// the BSD tar utility.
-	for i, s := range sp {
-		switch {
-		case s.offset < 0 || s.numBytes < 0:
-			return nil, ErrHeader // Negative values are never okay
-		case s.offset > math.MaxInt64-s.numBytes:
-			return nil, ErrHeader // Integer overflow with large length
-		case s.offset+s.numBytes > total:
-			return nil, ErrHeader // Region extends beyond the "real" size
-		case i > 0 && sp[i-1].offset+sp[i-1].numBytes > s.offset:
-			return nil, ErrHeader // Regions can't overlap and must be in order
-		}
-	}
-	return &sparseFileReader{rfr: rfr, sp: sp, total: total}, nil
-}
-
-// readHole reads a sparse hole ending at endOffset.
-func (sfr *sparseFileReader) readHole(b []byte, endOffset int64) int {
-	n64 := endOffset - sfr.pos
-	if n64 > int64(len(b)) {
-		n64 = int64(len(b))
-	}
-	n := int(n64)
-	for i := 0; i < n; i++ {
-		b[i] = 0
-	}
-	sfr.pos += n64
-	return n
-}
-
-// Read reads the sparse file data in expanded form.
-func (sfr *sparseFileReader) Read(b []byte) (n int, err error) {
-	// Skip past all empty fragments.
-	for len(sfr.sp) > 0 && sfr.sp[0].numBytes == 0 {
-		sfr.sp = sfr.sp[1:]
-	}
-
-	// If there are no more fragments, then it is possible that there
-	// is one last sparse hole.
-	if len(sfr.sp) == 0 {
-		// This behavior matches the BSD tar utility.
-		// However, GNU tar stops returning data even if sfr.total is unmet.
-		if sfr.pos < sfr.total {
-			return sfr.readHole(b, sfr.total), nil
-		}
-		return 0, io.EOF
-	}
-
-	// In front of a data fragment, so read a hole.
-	if sfr.pos < sfr.sp[0].offset {
-		return sfr.readHole(b, sfr.sp[0].offset), nil
-	}
-
-	// In a data fragment, so read from it.
-	// This math is overflow free since we verify that offset and numBytes can
-	// be safely added when creating the sparseFileReader.
-	endPos := sfr.sp[0].offset + sfr.sp[0].numBytes // End offset of fragment
-	bytesLeft := endPos - sfr.pos                   // Bytes left in fragment
-	if int64(len(b)) > bytesLeft {
-		b = b[:bytesLeft]
-	}
-
-	n, err = sfr.rfr.Read(b)
-	sfr.pos += int64(n)
-	if err == io.EOF {
-		if sfr.pos < endPos {
-			err = io.ErrUnexpectedEOF // There was supposed to be more data
-		} else if sfr.pos < sfr.total {
-			err = nil // There is still an implicit sparse hole at the end
-		}
-	}
-
-	if sfr.pos == endPos {
-		sfr.sp = sfr.sp[1:] // We are done with this fragment, so pop it
-	}
-	return n, err
-}
-
-// numBytes returns the number of bytes left to read in the sparse file's
-// sparse-encoded data in the tar archive.
-func (sfr *sparseFileReader) numBytes() int64 {
-	return sfr.rfr.numBytes()
-}
diff --git a/vendor/github.com/Microsoft/go-winio/archive/tar/stat_atim.go b/vendor/github.com/Microsoft/go-winio/archive/tar/stat_atim.go
deleted file mode 100644
index cf9cc79c5..000000000
--- a/vendor/github.com/Microsoft/go-winio/archive/tar/stat_atim.go
+++ /dev/null
@@ -1,20 +0,0 @@
-// Copyright 2012 The Go Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE file.
-
-// +build linux dragonfly openbsd solaris
-
-package tar
-
-import (
-	"syscall"
-	"time"
-)
-
-func statAtime(st *syscall.Stat_t) time.Time {
-	return time.Unix(st.Atim.Unix())
-}
-
-func statCtime(st *syscall.Stat_t) time.Time {
-	return time.Unix(st.Ctim.Unix())
-}
diff --git a/vendor/github.com/Microsoft/go-winio/archive/tar/stat_atimespec.go b/vendor/github.com/Microsoft/go-winio/archive/tar/stat_atimespec.go
deleted file mode 100644
index 6f17dbe30..000000000
--- a/vendor/github.com/Microsoft/go-winio/archive/tar/stat_atimespec.go
+++ /dev/null
@@ -1,20 +0,0 @@
-// Copyright 2012 The Go Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE file.
-
-// +build darwin freebsd netbsd
-
-package tar
-
-import (
-	"syscall"
-	"time"
-)
-
-func statAtime(st *syscall.Stat_t) time.Time {
-	return time.Unix(st.Atimespec.Unix())
-}
-
-func statCtime(st *syscall.Stat_t) time.Time {
-	return time.Unix(st.Ctimespec.Unix())
-}
diff --git a/vendor/github.com/Microsoft/go-winio/archive/tar/stat_unix.go b/vendor/github.com/Microsoft/go-winio/archive/tar/stat_unix.go
deleted file mode 100644
index cb843db4c..000000000
--- a/vendor/github.com/Microsoft/go-winio/archive/tar/stat_unix.go
+++ /dev/null
@@ -1,32 +0,0 @@
-// Copyright 2012 The Go Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE file.
-
-// +build linux darwin dragonfly freebsd openbsd netbsd solaris
-
-package tar
-
-import (
-	"os"
-	"syscall"
-)
-
-func init() {
-	sysStat = statUnix
-}
-
-func statUnix(fi os.FileInfo, h *Header) error {
-	sys, ok := fi.Sys().(*syscall.Stat_t)
-	if !ok {
-		return nil
-	}
-	h.Uid = int(sys.Uid)
-	h.Gid = int(sys.Gid)
-	// TODO(bradfitz): populate username & group.  os/user
-	// doesn't cache LookupId lookups, and lacks group
-	// lookup functions.
-	h.AccessTime = statAtime(sys)
-	h.ChangeTime = statCtime(sys)
-	// TODO(bradfitz): major/minor device numbers?
-	return nil
-}
diff --git a/vendor/github.com/Microsoft/go-winio/archive/tar/writer.go b/vendor/github.com/Microsoft/go-winio/archive/tar/writer.go
deleted file mode 100644
index 30d7e606d..000000000
--- a/vendor/github.com/Microsoft/go-winio/archive/tar/writer.go
+++ /dev/null
@@ -1,444 +0,0 @@
-// Copyright 2009 The Go Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style
-// license that can be found in the LICENSE file.
-
-package tar
-
-// TODO(dsymonds):
-// - catch more errors (no first header, etc.)
-
-import (
-	"bytes"
-	"errors"
-	"fmt"
-	"io"
-	"path"
-	"sort"
-	"strconv"
-	"strings"
-	"time"
-)
-
-var (
-	ErrWriteTooLong    = errors.New("archive/tar: write too long")
-	ErrFieldTooLong    = errors.New("archive/tar: header field too long")
-	ErrWriteAfterClose = errors.New("archive/tar: write after close")
-	errInvalidHeader   = errors.New("archive/tar: header field too long or contains invalid values")
-)
-
-// A Writer provides sequential writing of a tar archive in POSIX.1 format.
-// A tar archive consists of a sequence of files.
-// Call WriteHeader to begin a new file, and then call Write to supply that file's data,
-// writing at most hdr.Size bytes in total.
-type Writer struct {
-	w          io.Writer
-	err        error
-	nb         int64 // number of unwritten bytes for current file entry
-	pad        int64 // amount of padding to write after current file entry
-	closed     bool
-	usedBinary bool            // whether the binary numeric field extension was used
-	preferPax  bool            // use pax header instead of binary numeric header
-	hdrBuff    [blockSize]byte // buffer to use in writeHeader when writing a regular header
-	paxHdrBuff [blockSize]byte // buffer to use in writeHeader when writing a pax header
-}
-
-type formatter struct {
-	err error // Last error seen
-}
-
-// NewWriter creates a new Writer writing to w.
-func NewWriter(w io.Writer) *Writer { return &Writer{w: w, preferPax: true} }
-
-// Flush finishes writing the current file (optional).
-func (tw *Writer) Flush() error {
-	if tw.nb > 0 {
-		tw.err = fmt.Errorf("archive/tar: missed writing %d bytes", tw.nb)
-		return tw.err
-	}
-
-	n := tw.nb + tw.pad
-	for n > 0 && tw.err == nil {
-		nr := n
-		if nr > blockSize {
-			nr = blockSize
-		}
-		var nw int
-		nw, tw.err = tw.w.Write(zeroBlock[0:nr])
-		n -= int64(nw)
-	}
-	tw.nb = 0
-	tw.pad = 0
-	return tw.err
-}
-
-// Write s into b, terminating it with a NUL if there is room.
-func (f *formatter) formatString(b []byte, s string) {
-	if len(s) > len(b) {
-		f.err = ErrFieldTooLong
-		return
-	}
-	ascii := toASCII(s)
-	copy(b, ascii)
-	if len(ascii) < len(b) {
-		b[len(ascii)] = 0
-	}
-}
-
-// Encode x as an octal ASCII string and write it into b with leading zeros.
-func (f *formatter) formatOctal(b []byte, x int64) {
-	s := strconv.FormatInt(x, 8)
-	// leading zeros, but leave room for a NUL.
-	for len(s)+1 < len(b) {
-		s = "0" + s
-	}
-	f.formatString(b, s)
-}
-
-// fitsInBase256 reports whether x can be encoded into n bytes using base-256
-// encoding. Unlike octal encoding, base-256 encoding does not require that the
-// string ends with a NUL character. Thus, all n bytes are available for output.
-//
-// If operating in binary mode, this assumes strict GNU binary mode; which means
-// that the first byte can only be either 0x80 or 0xff. Thus, the first byte is
-// equivalent to the sign bit in two's complement form.
-func fitsInBase256(n int, x int64) bool {
-	var binBits = uint(n-1) * 8
-	return n >= 9 || (x >= -1<<binBits && x < 1<<binBits)
-}
-
-// Write x into b, as binary (GNUtar/star extension).
-func (f *formatter) formatNumeric(b []byte, x int64) {
-	if fitsInBase256(len(b), x) {
-		for i := len(b) - 1; i >= 0; i-- {
-			b[i] = byte(x)
-			x >>= 8
-		}
-		b[0] |= 0x80 // Highest bit indicates binary format
-		return
-	}
-
-	f.formatOctal(b, 0) // Last resort, just write zero
-	f.err = ErrFieldTooLong
-}
-
-var (
-	minTime = time.Unix(0, 0)
-	// There is room for 11 octal digits (33 bits) of mtime.
-	maxTime = minTime.Add((1<<33 - 1) * time.Second)
-)
-
-// WriteHeader writes hdr and prepares to accept the file's contents.
-// WriteHeader calls Flush if it is not the first header.
-// Calling after a Close will return ErrWriteAfterClose.
-func (tw *Writer) WriteHeader(hdr *Header) error {
-	return tw.writeHeader(hdr, true)
-}
-
-// WriteHeader writes hdr and prepares to accept the file's contents.
-// WriteHeader calls Flush if it is not the first header.
-// Calling after a Close will return ErrWriteAfterClose.
-// As this method is called internally by writePax header to allow it to
-// suppress writing the pax header.
-func (tw *Writer) writeHeader(hdr *Header, allowPax bool) error {
-	if tw.closed {
-		return ErrWriteAfterClose
-	}
-	if tw.err == nil {
-		tw.Flush()
-	}
-	if tw.err != nil {
-		return tw.err
-	}
-
-	// a map to hold pax header records, if any are needed
-	paxHeaders := make(map[string]string)
-
-	// TODO(shanemhansen): we might want to use PAX headers for
-	// subsecond time resolution, but for now let's just capture
-	// too long fields or non ascii characters
-
-	var f formatter
-	var header []byte
-
-	// We need to select which scratch buffer to use carefully,
-	// since this method is called recursively to write PAX headers.
-	// If allowPax is true, this is the non-recursive call, and we will use hdrBuff.
-	// If allowPax is false, we are being called by writePAXHeader, and hdrBuff is
-	// already being used by the non-recursive call, so we must use paxHdrBuff.
-	header = tw.hdrBuff[:]
-	if !allowPax {
-		header = tw.paxHdrBuff[:]
-	}
-	copy(header, zeroBlock)
-	s := slicer(header)
-
-	// Wrappers around formatter that automatically sets paxHeaders if the
-	// argument extends beyond the capacity of the input byte slice.
-	var formatString = func(b []byte, s string, paxKeyword string) {
-		needsPaxHeader := paxKeyword != paxNone && len(s) > len(b) || !isASCII(s)
-		if needsPaxHeader {
-			paxHeaders[paxKeyword] = s
-			return
-		}
-		f.formatString(b, s)
-	}
-	var formatNumeric = func(b []byte, x int64, paxKeyword string) {
-		// Try octal first.
-		s := strconv.FormatInt(x, 8)
-		if len(s) < len(b) {
-			f.formatOctal(b, x)
-			return
-		}
-
-		// If it is too long for octal, and PAX is preferred, use a PAX header.
-		if paxKeyword != paxNone && tw.preferPax {
-			f.formatOctal(b, 0)
-			s := strconv.FormatInt(x, 10)
-			paxHeaders[paxKeyword] = s
-			return
-		}
-
-		tw.usedBinary = true
-		f.formatNumeric(b, x)
-	}
-	var formatTime = func(b []byte, t time.Time, paxKeyword string) {
-		var unixTime int64
-		if !t.Before(minTime) && !t.After(maxTime) {
-			unixTime = t.Unix()
-		}
-		formatNumeric(b, unixTime, paxNone)
-
-		// Write a PAX header if the time didn't fit precisely.
-		if paxKeyword != "" && tw.preferPax && allowPax && (t.Nanosecond() != 0 || !t.Before(minTime) || !t.After(maxTime)) {
-			paxHeaders[paxKeyword] = formatPAXTime(t)
-		}
-	}
-
-	// keep a reference to the filename to allow to overwrite it later if we detect that we can use ustar longnames instead of pax
-	pathHeaderBytes := s.next(fileNameSize)
-
-	formatString(pathHeaderBytes, hdr.Name, paxPath)
-
-	f.formatOctal(s.next(8), hdr.Mode)               // 100:108
-	formatNumeric(s.next(8), int64(hdr.Uid), paxUid) // 108:116
-	formatNumeric(s.next(8), int64(hdr.Gid), paxGid) // 116:124
-	formatNumeric(s.next(12), hdr.Size, paxSize)     // 124:136
-	formatTime(s.next(12), hdr.ModTime, paxMtime)    // 136:148
-	s.next(8)                                        // chksum (148:156)
-	s.next(1)[0] = hdr.Typeflag                      // 156:157
-
-	formatString(s.next(100), hdr.Linkname, paxLinkpath)
-
-	copy(s.next(8), []byte("ustar\x0000"))          // 257:265
-	formatString(s.next(32), hdr.Uname, paxUname)   // 265:297
-	formatString(s.next(32), hdr.Gname, paxGname)   // 297:329
-	formatNumeric(s.next(8), hdr.Devmajor, paxNone) // 329:337
-	formatNumeric(s.next(8), hdr.Devminor, paxNone) // 337:345
-
-	// keep a reference to the prefix to allow to overwrite it later if we detect that we can use ustar longnames instead of pax
-	prefixHeaderBytes := s.next(155)
-	formatString(prefixHeaderBytes, "", paxNone) // 345:500  prefix
-
-	// Use the GNU magic instead of POSIX magic if we used any GNU extensions.
-	if tw.usedBinary {
-		copy(header[257:265], []byte("ustar  \x00"))
-	}
-
-	_, paxPathUsed := paxHeaders[paxPath]
-	// try to use a ustar header when only the name is too long
-	if !tw.preferPax && len(paxHeaders) == 1 && paxPathUsed {
-		prefix, suffix, ok := splitUSTARPath(hdr.Name)
-		if ok {
-			// Since we can encode in USTAR format, disable PAX header.
-			delete(paxHeaders, paxPath)
-
-			// Update the path fields
-			formatString(pathHeaderBytes, suffix, paxNone)
-			formatString(prefixHeaderBytes, prefix, paxNone)
-		}
-	}
-
-	// The chksum field is terminated by a NUL and a space.
-	// This is different from the other octal fields.
-	chksum, _ := checksum(header)
-	f.formatOctal(header[148:155], chksum) // Never fails
-	header[155] = ' '
-
-	// Check if there were any formatting errors.
-	if f.err != nil {
-		tw.err = f.err
-		return tw.err
-	}
-
-	if allowPax {
-		if !hdr.AccessTime.IsZero() {
-			paxHeaders[paxAtime] = formatPAXTime(hdr.AccessTime)
-		}
-		if !hdr.ChangeTime.IsZero() {
-			paxHeaders[paxCtime] = formatPAXTime(hdr.ChangeTime)
-		}
-		if !hdr.CreationTime.IsZero() {
-			paxHeaders[paxCreationTime] = formatPAXTime(hdr.CreationTime)
-		}
-		for k, v := range hdr.Xattrs {
-			paxHeaders[paxXattr+k] = v
-		}
-		for k, v := range hdr.Winheaders {
-			paxHeaders[paxWindows+k] = v
-		}
-	}
-
-	if len(paxHeaders) > 0 {
-		if !allowPax {
-			return errInvalidHeader
-		}
-		if err := tw.writePAXHeader(hdr, paxHeaders); err != nil {
-			return err
-		}
-	}
-	tw.nb = int64(hdr.Size)
-	tw.pad = (blockSize - (tw.nb % blockSize)) % blockSize
-
-	_, tw.err = tw.w.Write(header)
-	return tw.err
-}
-
-func formatPAXTime(t time.Time) string {
-	sec := t.Unix()
-	usec := t.Nanosecond()
-	s := strconv.FormatInt(sec, 10)
-	if usec != 0 {
-		s = fmt.Sprintf("%s.%09d", s, usec)
-	}
-	return s
-}
-
-// splitUSTARPath splits a path according to USTAR prefix and suffix rules.
-// If the path is not splittable, then it will return ("", "", false).
-func splitUSTARPath(name string) (prefix, suffix string, ok bool) {
-	length := len(name)
-	if length <= fileNameSize || !isASCII(name) {
-		return "", "", false
-	} else if length > fileNamePrefixSize+1 {
-		length = fileNamePrefixSize + 1
-	} else if name[length-1] == '/' {
-		length--
-	}
-
-	i := strings.LastIndex(name[:length], "/")
-	nlen := len(name) - i - 1 // nlen is length of suffix
-	plen := i                 // plen is length of prefix
-	if i <= 0 || nlen > fileNameSize || nlen == 0 || plen > fileNamePrefixSize {
-		return "", "", false
-	}
-	return name[:i], name[i+1:], true
-}
-
-// writePaxHeader writes an extended pax header to the
-// archive.
-func (tw *Writer) writePAXHeader(hdr *Header, paxHeaders map[string]string) error {
-	// Prepare extended header
-	ext := new(Header)
-	ext.Typeflag = TypeXHeader
-	// Setting ModTime is required for reader parsing to
-	// succeed, and seems harmless enough.
-	ext.ModTime = hdr.ModTime
-	// The spec asks that we namespace our pseudo files
-	// with the current pid.  However, this results in differing outputs
-	// for identical inputs.  As such, the constant 0 is now used instead.
-	// golang.org/issue/12358
-	dir, file := path.Split(hdr.Name)
-	fullName := path.Join(dir, "PaxHeaders.0", file)
-
-	ascii := toASCII(fullName)
-	if len(ascii) > 100 {
-		ascii = ascii[:100]
-	}
-	ext.Name = ascii
-	// Construct the body
-	var buf bytes.Buffer
-
-	// Keys are sorted before writing to body to allow deterministic output.
-	var keys []string
-	for k := range paxHeaders {
-		keys = append(keys, k)
-	}
-	sort.Strings(keys)
-
-	for _, k := range keys {
-		fmt.Fprint(&buf, formatPAXRecord(k, paxHeaders[k]))
-	}
-
-	ext.Size = int64(len(buf.Bytes()))
-	if err := tw.writeHeader(ext, false); err != nil {
-		return err
-	}
-	if _, err := tw.Write(buf.Bytes()); err != nil {
-		return err
-	}
-	if err := tw.Flush(); err != nil {
-		return err
-	}
-	return nil
-}
-
-// formatPAXRecord formats a single PAX record, prefixing it with the
-// appropriate length.
-func formatPAXRecord(k, v string) string {
-	const padding = 3 // Extra padding for ' ', '=', and '\n'
-	size := len(k) + len(v) + padding
-	size += len(strconv.Itoa(size))
-	record := fmt.Sprintf("%d %s=%s\n", size, k, v)
-
-	// Final adjustment if adding size field increased the record size.
-	if len(record) != size {
-		size = len(record)
-		record = fmt.Sprintf("%d %s=%s\n", size, k, v)
-	}
-	return record
-}
-
-// Write writes to the current entry in the tar archive.
-// Write returns the error ErrWriteTooLong if more than
-// hdr.Size bytes are written after WriteHeader.
-func (tw *Writer) Write(b []byte) (n int, err error) {
-	if tw.closed {
-		err = ErrWriteAfterClose
-		return
-	}
-	overwrite := false
-	if int64(len(b)) > tw.nb {
-		b = b[0:tw.nb]
-		overwrite = true
-	}
-	n, err = tw.w.Write(b)
-	tw.nb -= int64(n)
-	if err == nil && overwrite {
-		err = ErrWriteTooLong
-		return
-	}
-	tw.err = err
-	return
-}
-
-// Close closes the tar archive, flushing any unwritten
-// data to the underlying writer.
-func (tw *Writer) Close() error {
-	if tw.err != nil || tw.closed {
-		return tw.err
-	}
-	tw.Flush()
-	tw.closed = true
-	if tw.err != nil {
-		return tw.err
-	}
-
-	// trailer: two zero blocks
-	for i := 0; i < 2; i++ {
-		_, tw.err = tw.w.Write(zeroBlock)
-		if tw.err != nil {
-			break
-		}
-	}
-	return tw.err
-}
diff --git a/vendor/github.com/Microsoft/go-winio/backuptar/strconv.go b/vendor/github.com/Microsoft/go-winio/backuptar/strconv.go
new file mode 100644
index 000000000..341609663
--- /dev/null
+++ b/vendor/github.com/Microsoft/go-winio/backuptar/strconv.go
@@ -0,0 +1,68 @@
+package backuptar
+
+import (
+	"archive/tar"
+	"fmt"
+	"strconv"
+	"strings"
+	"time"
+)
+
+// Functions copied from https://github.com/golang/go/blob/master/src/archive/tar/strconv.go
+// as we need to manage the LIBARCHIVE.creationtime PAXRecord manually.
+// Idea taken from containerd which did the same thing.
+
+// parsePAXTime takes a string of the form %d.%d as described in the PAX
+// specification. Note that this implementation allows for negative timestamps,
+// which is allowed for by the PAX specification, but not always portable.
+func parsePAXTime(s string) (time.Time, error) {
+	const maxNanoSecondDigits = 9
+
+	// Split string into seconds and sub-seconds parts.
+	ss, sn := s, ""
+	if pos := strings.IndexByte(s, '.'); pos >= 0 {
+		ss, sn = s[:pos], s[pos+1:]
+	}
+
+	// Parse the seconds.
+	secs, err := strconv.ParseInt(ss, 10, 64)
+	if err != nil {
+		return time.Time{}, tar.ErrHeader
+	}
+	if len(sn) == 0 {
+		return time.Unix(secs, 0), nil // No sub-second values
+	}
+
+	// Parse the nanoseconds.
+	if strings.Trim(sn, "0123456789") != "" {
+		return time.Time{}, tar.ErrHeader
+	}
+	if len(sn) < maxNanoSecondDigits {
+		sn += strings.Repeat("0", maxNanoSecondDigits-len(sn)) // Right pad
+	} else {
+		sn = sn[:maxNanoSecondDigits] // Right truncate
+	}
+	nsecs, _ := strconv.ParseInt(sn, 10, 64) // Must succeed
+	if len(ss) > 0 && ss[0] == '-' {
+		return time.Unix(secs, -1*nsecs), nil // Negative correction
+	}
+	return time.Unix(secs, nsecs), nil
+}
+
+// formatPAXTime converts ts into a time of the form %d.%d as described in the
+// PAX specification. This function is capable of negative timestamps.
+func formatPAXTime(ts time.Time) (s string) {
+	secs, nsecs := ts.Unix(), ts.Nanosecond()
+	if nsecs == 0 {
+		return strconv.FormatInt(secs, 10)
+	}
+
+	// If seconds is negative, then perform correction.
+	sign := ""
+	if secs < 0 {
+		sign = "-"             // Remember sign
+		secs = -(secs + 1)     // Add a second to secs
+		nsecs = -(nsecs - 1e9) // Take that second away from nsecs
+	}
+	return strings.TrimRight(fmt.Sprintf("%s%d.%09d", sign, secs, nsecs), "0")
+}
diff --git a/vendor/github.com/Microsoft/go-winio/backuptar/tar.go b/vendor/github.com/Microsoft/go-winio/backuptar/tar.go
index d6566dbf0..088a43c68 100644
--- a/vendor/github.com/Microsoft/go-winio/backuptar/tar.go
+++ b/vendor/github.com/Microsoft/go-winio/backuptar/tar.go
@@ -3,6 +3,7 @@
 package backuptar
 
 import (
+	"archive/tar"
 	"encoding/base64"
 	"errors"
 	"fmt"
@@ -15,7 +16,6 @@ import (
 	"time"
 
 	"github.com/Microsoft/go-winio"
-	"github.com/Microsoft/go-winio/archive/tar" // until archive/tar supports pax extensions in its interface
 )
 
 const (
@@ -32,11 +32,13 @@ const (
 )
 
 const (
-	hdrFileAttributes        = "fileattr"
-	hdrSecurityDescriptor    = "sd"
-	hdrRawSecurityDescriptor = "rawsd"
-	hdrMountPoint            = "mountpoint"
-	hdrEaPrefix              = "xattr."
+	hdrFileAttributes        = "MSWINDOWS.fileattr"
+	hdrSecurityDescriptor    = "MSWINDOWS.sd"
+	hdrRawSecurityDescriptor = "MSWINDOWS.rawsd"
+	hdrMountPoint            = "MSWINDOWS.mountpoint"
+	hdrEaPrefix              = "MSWINDOWS.xattr."
+
+	hdrCreationTime = "LIBARCHIVE.creationtime"
 )
 
 func writeZeroes(w io.Writer, count int64) error {
@@ -86,16 +88,17 @@ func copySparse(t *tar.Writer, br *winio.BackupStreamReader) error {
 // BasicInfoHeader creates a tar header from basic file information.
 func BasicInfoHeader(name string, size int64, fileInfo *winio.FileBasicInfo) *tar.Header {
 	hdr := &tar.Header{
-		Name:         filepath.ToSlash(name),
-		Size:         size,
-		Typeflag:     tar.TypeReg,
-		ModTime:      time.Unix(0, fileInfo.LastWriteTime.Nanoseconds()),
-		ChangeTime:   time.Unix(0, fileInfo.ChangeTime.Nanoseconds()),
-		AccessTime:   time.Unix(0, fileInfo.LastAccessTime.Nanoseconds()),
-		CreationTime: time.Unix(0, fileInfo.CreationTime.Nanoseconds()),
-		Winheaders:   make(map[string]string),
+		Format:     tar.FormatPAX,
+		Name:       filepath.ToSlash(name),
+		Size:       size,
+		Typeflag:   tar.TypeReg,
+		ModTime:    time.Unix(0, fileInfo.LastWriteTime.Nanoseconds()),
+		ChangeTime: time.Unix(0, fileInfo.ChangeTime.Nanoseconds()),
+		AccessTime: time.Unix(0, fileInfo.LastAccessTime.Nanoseconds()),
+		PAXRecords: make(map[string]string),
 	}
-	hdr.Winheaders[hdrFileAttributes] = fmt.Sprintf("%d", fileInfo.FileAttributes)
+	hdr.PAXRecords[hdrFileAttributes] = fmt.Sprintf("%d", fileInfo.FileAttributes)
+	hdr.PAXRecords[hdrCreationTime] = formatPAXTime(time.Unix(0, fileInfo.CreationTime.Nanoseconds()))
 
 	if (fileInfo.FileAttributes & syscall.FILE_ATTRIBUTE_DIRECTORY) != 0 {
 		hdr.Mode |= c_ISDIR
@@ -155,7 +158,7 @@ func WriteTarFileFromBackupStream(t *tar.Writer, r io.Reader, name string, size
 			if err != nil {
 				return err
 			}
-			hdr.Winheaders[hdrRawSecurityDescriptor] = base64.StdEncoding.EncodeToString(sd)
+			hdr.PAXRecords[hdrRawSecurityDescriptor] = base64.StdEncoding.EncodeToString(sd)
 
 		case winio.BackupReparseData:
 			hdr.Mode |= c_ISLNK
@@ -166,7 +169,7 @@ func WriteTarFileFromBackupStream(t *tar.Writer, r io.Reader, name string, size
 				return err
 			}
 			if rp.IsMountPoint {
-				hdr.Winheaders[hdrMountPoint] = "1"
+				hdr.PAXRecords[hdrMountPoint] = "1"
 			}
 			hdr.Linkname = rp.Target
 
@@ -183,7 +186,7 @@ func WriteTarFileFromBackupStream(t *tar.Writer, r io.Reader, name string, size
 				// Use base64 encoding for the binary value. Note that there
 				// is no way to encode the EA's flags, since their use doesn't
 				// make any sense for persisted EAs.
-				hdr.Winheaders[hdrEaPrefix+ea.Name] = base64.StdEncoding.EncodeToString(ea.Value)
+				hdr.PAXRecords[hdrEaPrefix+ea.Name] = base64.StdEncoding.EncodeToString(ea.Value)
 			}
 
 		case winio.BackupAlternateData, winio.BackupLink, winio.BackupPropertyData, winio.BackupObjectId, winio.BackupTxfsData:
@@ -254,6 +257,7 @@ func WriteTarFileFromBackupStream(t *tar.Writer, r io.Reader, name string, size
 			}
 			if (bhdr.Attributes & winio.StreamSparseAttributes) == 0 {
 				hdr = &tar.Header{
+					Format:     hdr.Format,
 					Name:       name + altName,
 					Mode:       hdr.Mode,
 					Typeflag:   tar.TypeReg,
@@ -296,9 +300,10 @@ func FileInfoFromHeader(hdr *tar.Header) (name string, size int64, fileInfo *win
 		LastAccessTime: syscall.NsecToFiletime(hdr.AccessTime.UnixNano()),
 		LastWriteTime:  syscall.NsecToFiletime(hdr.ModTime.UnixNano()),
 		ChangeTime:     syscall.NsecToFiletime(hdr.ChangeTime.UnixNano()),
-		CreationTime:   syscall.NsecToFiletime(hdr.CreationTime.UnixNano()),
+		// Default to ModTime, we'll pull hdrCreationTime below if present
+		CreationTime: syscall.NsecToFiletime(hdr.ModTime.UnixNano()),
 	}
-	if attrStr, ok := hdr.Winheaders[hdrFileAttributes]; ok {
+	if attrStr, ok := hdr.PAXRecords[hdrFileAttributes]; ok {
 		attr, err := strconv.ParseUint(attrStr, 10, 32)
 		if err != nil {
 			return "", 0, nil, err
@@ -309,6 +314,13 @@ func FileInfoFromHeader(hdr *tar.Header) (name string, size int64, fileInfo *win
 			fileInfo.FileAttributes |= syscall.FILE_ATTRIBUTE_DIRECTORY
 		}
 	}
+	if creationTimeStr, ok := hdr.PAXRecords[hdrCreationTime]; ok {
+		creationTime, err := parsePAXTime(creationTimeStr)
+		if err != nil {
+			return "", 0, nil, err
+		}
+		fileInfo.CreationTime = syscall.NsecToFiletime(creationTime.UnixNano())
+	}
 	return
 }
 
@@ -321,13 +333,13 @@ func WriteBackupStreamFromTarFile(w io.Writer, t *tar.Reader, hdr *tar.Header) (
 	var err error
 	// Maintaining old SDDL-based behavior for backward compatibility.  All new tar headers written
 	// by this library will have raw binary for the security descriptor.
-	if sddl, ok := hdr.Winheaders[hdrSecurityDescriptor]; ok {
+	if sddl, ok := hdr.PAXRecords[hdrSecurityDescriptor]; ok {
 		sd, err = winio.SddlToSecurityDescriptor(sddl)
 		if err != nil {
 			return nil, err
 		}
 	}
-	if sdraw, ok := hdr.Winheaders[hdrRawSecurityDescriptor]; ok {
+	if sdraw, ok := hdr.PAXRecords[hdrRawSecurityDescriptor]; ok {
 		sd, err = base64.StdEncoding.DecodeString(sdraw)
 		if err != nil {
 			return nil, err
@@ -348,7 +360,7 @@ func WriteBackupStreamFromTarFile(w io.Writer, t *tar.Reader, hdr *tar.Header) (
 		}
 	}
 	var eas []winio.ExtendedAttribute
-	for k, v := range hdr.Winheaders {
+	for k, v := range hdr.PAXRecords {
 		if !strings.HasPrefix(k, hdrEaPrefix) {
 			continue
 		}
@@ -380,7 +392,7 @@ func WriteBackupStreamFromTarFile(w io.Writer, t *tar.Reader, hdr *tar.Header) (
 		}
 	}
 	if hdr.Typeflag == tar.TypeSymlink {
-		_, isMountPoint := hdr.Winheaders[hdrMountPoint]
+		_, isMountPoint := hdr.PAXRecords[hdrMountPoint]
 		rp := winio.ReparsePoint{
 			Target:       filepath.FromSlash(hdr.Linkname),
 			IsMountPoint: isMountPoint,