[release-branch.go1.25] archive/tar: set a limit on the size of GNU sparse file 1.0 regions

Sparse files in tar archives contain only the non-zero components
of the file. There are several different encodings for sparse
files. When reading GNU tar pax 1.0 sparse files, archive/tar did
not set a limit on the size of the sparse region data. A malicious
archive containing a large number of sparse blocks could cause
archive/tar to read an unbounded amount of data from the archive
into memory.

Since a malicious input can be highly compressable, a small
compressed input could cause very large allocations.

Cap the size of the sparse block data to the same limit used
for PAX headers (1 MiB).

Thanks to Harshit Gupta (Mr HAX) (https://www.linkedin.com/in/iam-harshit-gupta/)
for reporting this issue.

Fixes CVE-2025-58183
For golang#75677
Fixes golang#75711

Change-Id: I70b907b584a7b8676df8a149a1db728ae681a770
Reviewed-on: https://go-internal-review.googlesource.com/c/go/+/2800
Reviewed-by: Roland Shoemaker <[email protected]>
Reviewed-by: Nicholas Husin <[email protected]>
Reviewed-on: https://go-internal-review.googlesource.com/c/go/+/2987
Reviewed-by: Damien Neil <[email protected]>
Reviewed-on: https://go-review.googlesource.com/c/go/+/709852
TryBot-Bypass: Michael Pratt <[email protected]>
Reviewed-by: Carlos Amedee <[email protected]>
Auto-Submit: Michael Pratt <[email protected]>

Author: neild

src/archive/tar/common.go

@@ -39,6 +39,7 @@ var (
 	errMissData        = errors.New("archive/tar: sparse file references non-existent data")
 	errUnrefData       = errors.New("archive/tar: sparse file contains unreferenced data")
 	errWriteHole       = errors.New("archive/tar: write non-NUL byte in sparse hole")
+	errSparseTooLong   = errors.New("archive/tar: sparse map too long")
 )
 
 type headerError []string

src/archive/tar/reader.go

@@ -531,12 +531,17 @@ func readGNUSparseMap1x0(r io.Reader) (sparseDatas, error) {
 		cntNewline int64
 		buf        bytes.Buffer
 		blk        block
+		totalSize  int
 	)
 
 	// feedTokens copies data in blocks from r into buf until there are
 	// at least cnt newlines in buf. It will not read more blocks than needed.
 	feedTokens := func(n int64) error {
 		for cntNewline < n {
+			totalSize += len(blk)
+			if totalSize > maxSpecialFileSize {
+				return errSparseTooLong
+			}
 			if _, err := mustReadFull(r, blk[:]); err != nil {
 				return err
 			}
@@ -569,8 +574,8 @@ func readGNUSparseMap1x0(r io.Reader) (sparseDatas, error) {
 	}
 
 	// Parse for all member entries.
-	// numEntries is trusted after this since a potential attacker must have
-	// committed resources proportional to what this library used.
+	// numEntries is trusted after this since feedTokens limits the number of
+	// tokens based on maxSpecialFileSize.
 	if err := feedTokens(2 * numEntries); err != nil {
 		return nil, err
 	}

src/archive/tar/reader_test.go

@@ -621,6 +621,11 @@ func TestReader(t *testing.T) {
 			},
 			Format: FormatPAX,
 		}},
+	}, {
+		// Small compressed file that uncompresses to
+		// a file with a very large GNU 1.0 sparse map.
+		file: "testdata/gnu-sparse-many-zeros.tar.bz2",
+		err:  errSparseTooLong,
 	}}
 
 	for _, v := range vectors {