feat: support handle very large uid/gid in PAX style tar

This commit add process to read large uid/gid from PAX extensions
to fix very large UIDs/GIDs (>=2097151, limit of USTAR tar) lost
in PAX style tar during unpack.

Fix: alexcrichton#332

Signed-off-by: Qinqi Qu <quqinqi@linux.alibaba.com>

src/archive.rs

@@ -316,8 +316,12 @@ impl<'a> EntriesFields<'a> {
  }
 
  let mut pax_size: Option<u64> = None;
+ let mut pax_uid: Option<u64> = None;
+ let mut pax_gid: Option<u64> = None;
  if let Some(pax_extensions_ref) = &pax_extensions {
- pax_size = pax_extensions_value(pax_extensions_ref, "size");
+ pax_size = pax_extensions_value(pax_extensions_ref, PAX_SIZE);
+ pax_uid = pax_extensions_value(pax_extensions_ref, PAX_UID);
+ pax_gid = pax_extensions_value(pax_extensions_ref, PAX_GID);
  }
 
  let file_pos = self.next;
@@ -327,6 +331,15 @@ impl<'a> EntriesFields<'a> {
  size = pax_size;
  }
  }
+
+ if pax_uid.is_some() {
+ header.set_uid(pax_uid.unwrap());
+ }
+
+ if pax_gid.is_some() {
+ header.set_gid(pax_gid.unwrap());
+ }
+
  let ret = EntryFields {
  size: size,
  header_pos: header_pos,

src/pax.rs

@@ -1,9 +1,39 @@
+#![allow(dead_code)]
 use std::io;
 use std::slice;
 use std::str;
 
 use crate::other;
 
+// Keywords for PAX extended header records.
+pub const PAX_NONE: &str = ""; // Indicates that no PAX key is suitable
+pub const PAX_PATH: &str = "path";
+pub const PAX_LINKPATH: &str = "linkpath";
+pub const PAX_SIZE: &str = "size";
+pub const PAX_UID: &str = "uid";
+pub const PAX_GID: &str = "gid";
+pub const PAX_UNAME: &str = "uname";
+pub const PAX_GNAME: &str = "gname";
+pub const PAX_MTIME: &str = "mtime";
+pub const PAX_ATIME: &str = "atime";
+pub const PAX_CTIME: &str = "ctime"; // Removed from later revision of PAX spec, but was valid
+pub const PAX_CHARSET: &str = "charset"; // Currently unused
+pub const PAX_COMMENT: &str = "comment"; // Currently unused
+
+pub const PAX_SCHILYXATTR: &str = "SCHILY.xattr.";
+
+// Keywords for GNU sparse files in a PAX extended header.
+pub const PAX_GNUSPARSE: &str = "GNU.sparse.";
+pub const PAX_GNUSPARSENUMBLOCKS: &str = "GNU.sparse.numblocks";
+pub const PAX_GNUSPARSEOFFSET: &str = "GNU.sparse.offset";
+pub const PAX_GNUSPARSENUMBYTES: &str = "GNU.sparse.numbytes";
+pub const PAX_GNUSPARSEMAP: &str = "GNU.sparse.map";
+pub const PAX_GNUSPARSENAME: &str = "GNU.sparse.name";
+pub const PAX_GNUSPARSEMAJOR: &str = "GNU.sparse.major";
+pub const PAX_GNUSPARSEMINOR: &str = "GNU.sparse.minor";
+pub const PAX_GNUSPARSESIZE: &str = "GNU.sparse.size";
+pub const PAX_GNUSPARSEREALSIZE: &str = "GNU.sparse.realsize";
+
 /// An iterator over the pax extensions in an archive entry.
 ///
 /// This iterator yields structures which can themselves be parsed into

tests/all.rs

@@ -1488,3 +1488,30 @@ fn ownership_preserving() {
  assert!(ar.unpack(td.path()).is_err());
  }
 }
+
+#[test]
+#[cfg(unix)]
+fn pax_and_gnu_uid_gid() {
+ let tarlist = [tar!("biguid_gnu.tar"), tar!("biguid_pax.tar")];
+
+ for file in &tarlist {
+ let td = t!(TempBuilder::new().prefix("tar-rs").tempdir());
+ let rdr = Cursor::new(file);
+ let mut ar = Archive::new(rdr);
+ ar.set_preserve_ownerships(true);
+
+ if unsafe { libc::getuid() } == 0 {
+ t!(ar.unpack(td.path()));
+ let meta = fs::metadata(td.path().join("test.txt")).unwrap();
+ let uid = std::os::unix::prelude::MetadataExt::uid(&meta);
+ let gid = std::os::unix::prelude::MetadataExt::gid(&meta);
+ // 4294967294 = u32::MAX - 1
+ assert_eq!(uid, 4294967294);
+ assert_eq!(gid, 4294967294);
+ } else {
+ // it's not possible to unpack tar while preserving ownership
+ // without root permissions
+ assert!(ar.unpack(td.path()).is_err());
+ }
+ }
+}