* [PATCH 0/2] fetch2: harden deb/ipk unpack command argument
@ 2026-05-18 14:59 Anders Heimer
2026-05-18 14:59 ` [PATCH 1/2] fetch2: validate deb/ipk data member names Anders Heimer
2026-05-18 14:59 ` [PATCH 2/2] fetch2: validate striplevel parameter Anders Heimer
0 siblings, 2 replies; 3+ messages in thread
From: Anders Heimer @ 2026-05-18 14:59 UTC (permalink / raw)
To: bitbake-devel; +Cc: Anders Heimer
Hi,
This series hardens fetch2 unpack handling in two small, separate steps.
The first patch fixes deb/ipk unpack handling by selecting the data archive
member only from the set of supported data member names. This avoids passing
unsupported ar member names through to the shell command. It also quotes the
package path used by the deb/ipk unpack command.
The second patch is separate hardening for the striplevel parameter,
which is appended to tar arguments before the unpack command is run. It
validates striplevel as a decimal component count before using it.
Regression coverage is added to FetcherLocalTest for supported deb/ipk data
members, unsupported data member names, quoted package filenames, and invalid
striplevel values.
Tests run:
BB_SKIP_NETTESTS=yes bin/bitbake-selftest bb.tests.fetch.FetcherLocalTest
Anders Heimer (2):
fetch2: validate deb/ipk data member names
fetch2: validate striplevel parameter
lib/bb/fetch2/__init__.py | 15 +++++++---
lib/bb/tests/fetch.py | 63 +++++++++++++++++++++++++++++++++++++++
2 files changed, 74 insertions(+), 4 deletions(-)
--
2.34.1
^ permalink raw reply [flat|nested] 3+ messages in thread
* [PATCH 1/2] fetch2: validate deb/ipk data member names
2026-05-18 14:59 [PATCH 0/2] fetch2: harden deb/ipk unpack command argument Anders Heimer
@ 2026-05-18 14:59 ` Anders Heimer
2026-05-18 14:59 ` [PATCH 2/2] fetch2: validate striplevel parameter Anders Heimer
1 sibling, 0 replies; 3+ messages in thread
From: Anders Heimer @ 2026-05-18 14:59 UTC (permalink / raw)
To: bitbake-devel; +Cc: Anders Heimer
The deb/ipk unpack path selects a data archive member from 'ar -t'
output and then passes that member name to a shell command. Previously,
any member beginning with data.tar. was selected.
Only select known deb/ipk data archive member names when datafile is
created. Quote the package path used in the shell command as it can come
from the local fetch path.
Add local fetcher regression coverage for quoted package filenames,
valid compressed data members, and unsupported or unsafe data member
names.
Signed-off-by: Anders Heimer <anders.heimer@est.tech>
---
lib/bb/fetch2/__init__.py | 10 +++++---
lib/bb/tests/fetch.py | 53 +++++++++++++++++++++++++++++++++++++++
2 files changed, 60 insertions(+), 3 deletions(-)
diff --git a/lib/bb/fetch2/__init__.py b/lib/bb/fetch2/__init__.py
index b6cb4c530..dc93e64a9 100644
--- a/lib/bb/fetch2/__init__.py
+++ b/lib/bb/fetch2/__init__.py
@@ -23,6 +23,7 @@ import collections
import subprocess
import pickle
import errno
+import shlex
import bb.utils
import bb.checksum
import bb.process
@@ -1592,16 +1593,19 @@ class FetchMethod(object):
elif file.endswith('.deb') or file.endswith('.ipk'):
output = subprocess.check_output(['ar', '-t', file], preexec_fn=subprocess_setup)
datafile = None
+ valid_datafiles = ('data.tar', 'data.tar.gz', 'data.tar.xz',
+ 'data.tar.zst', 'data.tar.bz2', 'data.tar.lzma')
if output:
for line in output.decode().splitlines():
- if line.startswith('data.tar.') or line == 'data.tar':
+ if line in valid_datafiles:
datafile = line
break
else:
- raise UnpackError("Unable to unpack deb/ipk package - does not contain data.tar* file", urldata.url)
+ raise UnpackError("Unable to unpack deb/ipk package - does not contain supported data.tar* file", urldata.url)
else:
raise UnpackError("Unable to unpack deb/ipk package - could not list contents", urldata.url)
- cmd = 'ar x %s %s && %s -p -f %s && rm %s' % (file, datafile, tar_cmd, datafile, datafile)
+ quoted_datafile = shlex.quote(datafile)
+ cmd = 'ar x %s %s && %s -p -f %s && rm %s' % (shlex.quote(file), quoted_datafile, tar_cmd, quoted_datafile, quoted_datafile)
# If 'subdir' param exists, create a dir and use it as destination for unpack cmd
if 'subdir' in urldata.parm:
diff --git a/lib/bb/tests/fetch.py b/lib/bb/tests/fetch.py
index 969e5f876..bece59e62 100644
--- a/lib/bb/tests/fetch.py
+++ b/lib/bb/tests/fetch.py
@@ -16,6 +16,7 @@ import tempfile
import collections
import os
import signal
+import subprocess
import tarfile
from bb.fetch2 import URI
import bb
@@ -740,6 +741,34 @@ class FetcherLocalTest(FetcherTest):
bb.process.run('tar cjf archive.tar.bz2 -C dir .', cwd=self.localsrcdir)
self.d.setVar("FILESPATH", self.localsrcdir)
+ def make_ar_package(self, package_name, data_member="data.tar"):
+ if not shutil.which("ar"):
+ self.skipTest("ar not installed")
+
+ workdir = tempfile.mkdtemp(dir=self.tempdir)
+ payload = os.path.join(workdir, "payload")
+ with open(payload, "w") as f:
+ f.write("payload\n")
+
+ data_path = os.path.join(workdir, data_member)
+ mode = "w:gz" if data_member.endswith(".gz") else "w"
+ with tarfile.open(data_path, mode) as archive:
+ archive.add(payload, arcname="payload")
+
+ with open(os.path.join(workdir, "debian-binary"), "w") as f:
+ f.write("2.0\n")
+
+ control = os.path.join(workdir, "control")
+ with open(control, "w") as f:
+ f.write("Package: fetch-test\nVersion: 1\nArchitecture: all\n")
+ with tarfile.open(os.path.join(workdir, "control.tar"), "w") as archive:
+ archive.add(control, arcname="control")
+
+ package_path = os.path.join(self.localsrcdir, package_name)
+ subprocess.check_call(["ar", "r", package_path, "debian-binary", "control.tar", data_member],
+ cwd=workdir, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
+ return package_name
+
def fetchUnpack(self, uris):
fetcher = bb.fetch.Fetch(uris, self.d)
fetcher.download()
@@ -813,6 +842,30 @@ class FetcherLocalTest(FetcherTest):
tree = self.fetchUnpack(['file://archive.tar.bz2;subdir=bar;striplevel=1'])
self.assertEqual(tree, ['bar/c', 'bar/d', 'bar/subdir/e'])
+ def test_local_deb_quoted_filename(self):
+ package = self.make_ar_package("archive$(id).deb")
+ tree = self.fetchUnpack(['file://%s' % package])
+ self.assertEqual(tree, ['payload'])
+
+ def test_local_ipk_gz_data_member(self):
+ package = self.make_ar_package("archive.ipk", data_member="data.tar.gz")
+ tree = self.fetchUnpack(['file://%s' % package])
+ self.assertEqual(tree, ['payload'])
+
+ def test_local_deb_rejects_unknown_data_member_suffix(self):
+ package = self.make_ar_package("archive.deb", data_member="data.tar.foo")
+ with self.assertRaises(bb.fetch2.UnpackError) as context:
+ self.fetchUnpack(['file://%s' % package])
+
+ self.assertIn("does not contain supported data.tar* file", str(context.exception))
+
+ def test_local_deb_rejects_unsafe_data_member(self):
+ package = self.make_ar_package("archive.deb", data_member="data.tar.xz;id")
+ with self.assertRaises(bb.fetch2.UnpackError) as context:
+ self.fetchUnpack(['file://%s' % package])
+
+ self.assertIn("does not contain supported data.tar* file", str(context.exception))
+
def dummyGitTest(self, suffix):
# Create dummy local Git repo
src_dir = tempfile.mkdtemp(dir=self.tempdir,
--
2.34.1
^ permalink raw reply related [flat|nested] 3+ messages in thread
* [PATCH 2/2] fetch2: validate striplevel parameter
2026-05-18 14:59 [PATCH 0/2] fetch2: harden deb/ipk unpack command argument Anders Heimer
2026-05-18 14:59 ` [PATCH 1/2] fetch2: validate deb/ipk data member names Anders Heimer
@ 2026-05-18 14:59 ` Anders Heimer
1 sibling, 0 replies; 3+ messages in thread
From: Anders Heimer @ 2026-05-18 14:59 UTC (permalink / raw)
To: bitbake-devel; +Cc: Anders Heimer
The striplevel URL parameter is appended to tar_cmd, which is later run
through the shell. Validate it as a decimal count before using it in the
tar arguments.
Signed-off-by: Anders Heimer <anders.heimer@est.tech>
---
lib/bb/fetch2/__init__.py | 5 ++++-
lib/bb/tests/fetch.py | 10 ++++++++++
2 files changed, 14 insertions(+), 1 deletion(-)
diff --git a/lib/bb/fetch2/__init__.py b/lib/bb/fetch2/__init__.py
index dc93e64a9..77fc1b15b 100644
--- a/lib/bb/fetch2/__init__.py
+++ b/lib/bb/fetch2/__init__.py
@@ -1545,7 +1545,10 @@ class FetchMethod(object):
if unpack:
tar_cmd = 'tar --extract --no-same-owner'
if 'striplevel' in urldata.parm:
- tar_cmd += ' --strip-components=%s' % urldata.parm['striplevel']
+ striplevel = urldata.parm['striplevel']
+ if not striplevel.isdigit():
+ raise UnpackError("Invalid striplevel parameter: %s" % striplevel, urldata.url)
+ tar_cmd += ' --strip-components=%s' % striplevel
if file.endswith('.tar'):
cmd = '%s -f %s' % (tar_cmd, file)
elif file.endswith('.tgz') or file.endswith('.tar.gz') or file.endswith('.tar.Z'):
diff --git a/lib/bb/tests/fetch.py b/lib/bb/tests/fetch.py
index bece59e62..5850589e0 100644
--- a/lib/bb/tests/fetch.py
+++ b/lib/bb/tests/fetch.py
@@ -866,6 +866,16 @@ class FetcherLocalTest(FetcherTest):
self.assertIn("does not contain supported data.tar* file", str(context.exception))
+ def assertInvalidStriplevel(self, value):
+ with self.assertRaises(bb.fetch2.UnpackError) as context:
+ self.fetchUnpack(['file://archive.tar;subdir=bar;striplevel=%s' % value])
+ self.assertIn("Invalid striplevel parameter", str(context.exception))
+
+ def test_local_striplevel_rejects_invalid_values(self):
+ for value in ("abc", "", "-1", "1\n", "1 2"):
+ with self.subTest(striplevel=repr(value)):
+ self.assertInvalidStriplevel(value)
+
def dummyGitTest(self, suffix):
# Create dummy local Git repo
src_dir = tempfile.mkdtemp(dir=self.tempdir,
--
2.34.1
^ permalink raw reply related [flat|nested] 3+ messages in thread
end of thread, other threads:[~2026-05-18 14:59 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-05-18 14:59 [PATCH 0/2] fetch2: harden deb/ipk unpack command argument Anders Heimer
2026-05-18 14:59 ` [PATCH 1/2] fetch2: validate deb/ipk data member names Anders Heimer
2026-05-18 14:59 ` [PATCH 2/2] fetch2: validate striplevel parameter Anders Heimer
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox