Fixes the following security issue:
- CVE-2021-23336: Web cache poisoning via django.utils.http.limited_parse_qsl()
Django contains a copy of urllib.parse.parse_qsl() which was added to
backport some security fixes. A further security fix has been issued
recently such that parse_qsl() no longer allows using ; as a query
parameter separator by default. Django now includes this fix. See
bpo-42967 for further details.
For more details, see the advisory:
https://www.djangoproject.com/weblog/2021/feb/19/security-releases/
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
(cherry picked from commit 82abd78a01)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Fixes the following security issues:
- CVE-2021-23841: Null pointer deref in X509_issuer_and_serial_hash()
The OpenSSL public API function X509_issuer_and_serial_hash() attempts to
create a unique hash value based on the issuer and serial number data
contained within an X509 certificate. However it fails to correctly
handle any errors that may occur while parsing the issuer field (which
might occur if the issuer field is maliciously constructed). This may
subsequently result in a NULL pointer deref and a crash leading to a
potential denial of service attack.
The function X509_issuer_and_serial_hash() is never directly called by
OpenSSL itself so applications are only vulnerable if they use this
function directly and they use it on certificates that may have been
obtained from untrusted sources.
- CVE-2021-23839: Incorrect SSLv2 rollback protection
OpenSSL 1.0.2 supports SSLv2. If a client attempts to negotiate SSLv2
with a server that is configured to support both SSLv2 and more recent SSL
and TLS versions then a check is made for a version rollback attack when
unpadding an RSA signature. Clients that support SSL or TLS versions
greater than SSLv2 are supposed to use a special form of padding. A
server that supports greater than SSLv2 is supposed to reject connection
attempts from a client where this special form of padding is present,
because this indicates that a version rollback has occurred (i.e. both
client and server support greater than SSLv2, and yet this is the version
that is being requested).
The implementation of this padding check inverted the logic so that the
connection attempt is accepted if the padding is present, and rejected if
it is absent. This means that such as server will accept a connection if
a version rollback attack has occurred. Further the server will
erroneously reject a connection if a normal SSLv2 connection attempt is
made.
OpenSSL 1.1.1 does not have SSLv2 support and therefore is not vulnerable
to this issue. The underlying error is in the implementation of the
RSA_padding_check_SSLv23() function. This also affects the
RSA_SSLV23_PADDING padding mode used by various other functions. Although
1.1.1 does not support SSLv2 the RSA_padding_check_SSLv23() function still
exists, as does the RSA_SSLV23_PADDING padding mode. Applications that
directly call that function or use that padding mode will encounter this
issue. However since there is no support for the SSLv2 protocol in 1.1.1
this is considered a bug and not a security issue in that version.
- CVE-2021-23840: Integer overflow in CipherUpdate
Calls to EVP_CipherUpdate, EVP_EncryptUpdate and EVP_DecryptUpdate may
overflow the output length argument in some cases where the input length
is close to the maximum permissable length for an integer on the platform.
In such cases the return value from the function call will be 1
(indicating success), but the output length value will be negative. This
could cause applications to behave incorrectly or crash.
For more details, see the advisory:
https://www.openssl.org/news/secadv/20210216.txt
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
(cherry picked from commit 4745a484a6)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Fixes:
https://gitlab.com/buildroot.org/buildroot/-/jobs/1019385940
FAIL: test_run (tests.core.test_timezone.TestGlibcNonDefaultLimitedTimezone)
----------------------------------------------------------------------
Traceback (most recent call last):
File "/builds/buildroot.org/buildroot/support/testing/tests/core/test_timezone.py", line 66, in test_run
self.assertEqual(tz[0].strip(), "EST")
AssertionError: '' != 'EST'
Commit 7868289fd5 (package/zic: bump version to 2020f) bumped the zic
version to 2020f, which changed the default output format from the classic
"fat" format to the new "slim" format:
6ba6f2117b
The slim format is unfortunately not supported by glibc < 2.28 or uClibc, so
explicitly request the classic "fat" format.
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
(cherry picked from commit 1efb7b9618)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Glibc 2.33 removed `_STAT_VER`. On host systems, which updated to glibc
2.33, building host-fakeroot breaks:
```
In file included from communicate.h:20,
from libfakeroot.c:60:
libfakeroot.c: In function ‘chown’:
libfakeroot.c:99:40: error: ‘_STAT_VER’ undeclared (first use in this function)
99 | #define INT_NEXT_STAT(a,b) NEXT_STAT64(_STAT_VER,a,b)
```
The issue has been discussed on some system package threads, e.g.:
https://bugs.archlinux.org/task/69572https://bugzilla.redhat.com/show_bug.cgi?id=1889862#c13
A patch set was prepared by Ilya Lipnitskiy which included two other
patches not related to the glibc 2.33 compatibility and prepared as
merge request for upstream:
https://www.mail-archive.com/openwrt-devel@lists.openwrt.org/msg57280.html
Upstream accepted the merge request:
https://salsa.debian.org/clint/fakeroot/-/merge_requests/10
Note, that this patch series only contains the necessay patches for glibc
2.33 compatibility.
Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
[Peter: drop patch numbering (PATCH x/y) as pointed out by check-package]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Changes:
* Fix memory-related bugs in the JPEG-2000 codec resulting from
attempting to decode invalid code streams. (#264, #265)
This fix is associated with CVE-2021-26926 and CVE-2021-26927.
* Fix wrong return value under some compilers (#260)
* Fix CVE-2021-3272 heap buffer overflow in jp2_decode (#259)
Signed-off-by: Michael Vetter <jubalh@iodoru.org>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
(cherry picked from commit 72b801010c)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Changes:
* Add JAS_VERSION_MAJOR, JAS_VERSION_MINOR, JAS_VERSION_PATCH for
easier access to the JasPer version.
* Fixes stack overflow bug on Windows, where variable-length
arrays are not available. (#256)
Signed-off-by: Michael Vetter <jubalh@iodoru.org>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
(cherry picked from commit 7a5c61d59b)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Fixes the following security issue:
CVE-2021-27135: xterm through Patch #365 allows remote attackers to cause a
denial of service (segmentation fault) or possibly have unspecified other
impact via a crafted UTF-8 character sequence.
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
(cherry picked from commit fd6f7061ca)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Separate the fields in the hash file by two spaces.
Change the hash of the license:
- Copyright message changed from 2018,2019 to 2019,2020
Signed-off-by: Sergio Prado <sergio.prado@e-labworks.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
(cherry picked from commit 650f0aa3e2)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Fixes the following security issue:
CVE-2020-17525: Remote unauthenticated denial-of-service in Subversion
mod_authz_svn
Subversion's mod_authz_svn module will crash if the server is using
in-repository authz rules with the AuthzSVNReposRelativeAccessFile option
and a client sends a request for a non-existing repository URL.
For more details, see the advisory:
https://subversion.apache.org/security/CVE-2020-17525-advisory.txt
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
(cherry picked from commit 4109401acd)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
First patch is not needed since version 1.8.0 and
f071ec0c26
Indeed, as spotted by upstream when sending them this patch, the
original expressions will not mangle '-mfloat-gprs=double' because the
patterns contain a trailing space.
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
(cherry picked from commit 9670aa2852)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Bugfix release, fixing the following issue:
- Check contexts before importing them to reduce risk of extracted files
escaping context store
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
When a developer has package/pkg-<infra>.mk assigned to him/her in the
DEVELOPERS file, this has 3 implications:
(1) Patches adding new packages using this infrastructure are Cc'ed
to this developer. This is done by the analyze_patch() function,
which matches the regexp r"^\+\$\(eval
\$\((host-)?([^-]*)-package\)\)$" in the patch, i.e where an
added line contains a reference to the infra maintained by the
developer.
(2) Patches touching the package/pkg-<infra>.mk file itself are Cc'ed
to this developer.
(3) Any patch touching a package using this infra are also Cc'ed to
this developer.
Point (3) causes a significant amount of patches to be sent to
developers who have package/pkg-generic.mk and
package/pkg-autotools.mk assigned to them in the DEVELOPERS
file. Basically, all patches touching generic or autotools packages
get CC'ed to such developers, which causes a massive amount of patches
to be received.
So this patch adjusts the getdeveloperlib.py to drop point (3), but
preserves point (1) and (2). Indeed, it makes sense to be Cc'ed on new
package additions (to make a review that they use the package
infrastructure correctly), and it makes sense to be Cc'ed on patches
that touch the infrastructure code itself.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
(cherry picked from commit 38b0560f4e)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Fixes the following security issue:
- CVE-2021-3177: Python 3.x through 3.9.1 has a buffer overflow in
PyCArg_repr in _ctypes/callproc.c, which may lead to remote code execution
in certain Python applications that accept floating-point numbers as
untrusted input, as demonstrated by a 1e300 argument to
c_double.from_param. This occurs because sprintf is used unsafely.
For details, see the advisory:
https://python-security.readthedocs.io/vuln/ctypes-buffer-overflow-pycarg_repr.html
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Fixed the following security issue:
- CVE-2020-6097: An exploitable denial of service vulnerability exists in
the atftpd daemon functionality of atftp 0.7.git20120829-3.1+b1. A
specially crafted sequence of RRQ-Multicast requests trigger an assert()
call resulting in denial-of-service. An attacker can send a sequence of
malicious packets to trigger this vulnerability.
For more details, see the report:
https://talosintelligence.com/vulnerability_reports/TALOS-2020-1029
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
(cherry picked from commit 5b36e91fda)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Fixes the following security issue:
- wpa_supplicant P2P group information processing vulnerability (no CVE yet)
A vulnerability was discovered in how wpa_supplicant processing P2P
(Wi-Fi Direct) group information from active group owners. The actual
parsing of that information validates field lengths appropriately, but
processing of the parsed information misses a length check when storing a
copy of the secondary device types. This can result in writing attacker
controlled data into the peer entry after the area assigned for the
secondary device type. The overflow can result in corrupting pointers
for heap allocations. This can result in an attacker within radio range
of the device running P2P discovery being able to cause unexpected
behavior, including termination of the wpa_supplicant process and
potentially arbitrary code execution.
For more details, see the advisory:
https://w1.fi/security/2020-2/wpa_supplicant-p2p-group-info-processing-vulnerability.txt
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
[yann.morin.1998@free.fr: keep _PATCH near _VERSION and _SITE]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
(cherry picked from commit 74c854bd51)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
When a armv8 target is used in 32bits mode, xenomai fail to detect the
ARM architecture and abord the build. (__ARM_ARCH_7A__ is not defined
for armv8 cpus).
There are no autobuilder failures for this issue since cobalt is never
selected, but the following defconfig:
BR2_arm=y
BR2_cortex_a53=y
BR2_ARM_FPU_NEON_VFPV4=y
BR2_TOOLCHAIN_EXTERNAL=y
BR2_PACKAGE_XENOMAI=y
BR2_PACKAGE_XENOMAI_COBALT=y
This was initialy reproduced using the raspberrypi3_defconfig with
Xenomai package with cobalt selected.
In order to use Xenomai on raspberrypi3 in 32 bits mode, one has to
select BR2_cortex_a7 instead of BR2_cortex_a53 (see a13a388dd4).
See:
https://gitlab.denx.de/Xenomai/xenomai/-/blob/v3.1/lib/cobalt/arch/arm/include/asm/xenomai/features.h#L52
Signed-off-by: Romain Naour <romain.naour@gmail.com>
[yann.morin.1998@free.fr:
- switch to independent conditional 'default y'
- slightly reword the commit log
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
(cherry picked from commit 6490a11018)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Fix CVE-2020-11105: An issue was discovered in USC iLab cereal through
1.3.0. It employs caching of std::shared_ptr values, using the raw
pointer address as a unique identifier. This becomes problematic if an
std::shared_ptr variable goes out of scope and is freed, and a new
std::shared_ptr is allocated at the same address. Serialization fidelity
thereby becomes dependent upon memory layout. In short, serialized
std::shared_ptr variables cannot always be expected to serialize back
into their original values. This can have any number of consequences,
depending on the context within which this manifests.
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
(cherry picked from commit 26a46564f3)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Bump to the latest git commit as this will fix the following CVEs:
git log|grep CVE
sox-fmt: validate comments_bytes before use (CVE-2019-13590) [bug #325]
fix possible null pointer deref in lsx_make_lpf() (CVE-2019-8357)
fft4g: bail if size too large (CVE-2019-8356)
fix possible overflow in lsx_(re)valloc() size calculation (CVE-2019-8355)
fix possible buffer size overflow in lsx_make_lpf() (CVE-2019-8354)
xa: validate channel count (CVE-2017-18189)
aiff: fix crash on empty comment chunk (CVE-2017-15642)
adpcm: fix stack overflow with >4 channels (CVE-2017-15372)
flac: fix crash on corrupt metadata (CVE-2017-15371)
wav: ima_adpcm: fix buffer overflow on corrupt input (CVE-2017-15370)
wav: fix crash writing header when channel count >64k (CVE-2017-11359)
hcom: fix crash on input with corrupt dictionary (CVE-2017-11358)
wav: fix crash if channel count is zero (CVE-2017-11332)
- Tweak configuration options due to
6ff0e9322f
- libgsm is now an optional dependency since
e548827ffc
- Add patch to put back --disable-stack-protector
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
(cherry picked from commit b6871f9d93)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Fixes the following security issue:
CVE-2020-28473: The package bottle from 0 and before 0.12.19 are vulnerable
to Web Cache Poisoning by using a vector called parameter cloaking. When
the attacker can separate query parameters using a semicolon (;), they can
cause a difference in the interpretation of the request between the proxy
(running with default configuration) and the server. This can result in
malicious requests being cached as completely safe ones, as the proxy would
usually not see the semicolon as a separator, and therefore would not
include it in a cache key of an unkeyed parameter.
In addition, bottle 0.12.18 fixed a compatibility issue with python 3.8+:
https://github.com/bottlepy/bottle/issues/1181
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
(cherry picked from commit 14cc349d26)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
For instance on risc-v 64 arch the build would otherwise fail because
of undefined ucontext_t because "-DOPENSSL_NO_ASYNC" would not propagate
through to CFLAGS in the Makefile.
Signed-off-by: Yann Sionneau <ysionneau@kalray.eu>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
(cherry picked from commit a00b6354a2)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Commit f4a61d1ae2 (package/pkg-meson.mk avoid host ccache detection)
forced the host C and C++ compilers so that meson does not try to
autodetect ccache, and instead relies on what we provide.
However, this incorrectly used single-expansion of variables in a
package infra.
For traditional builds, this is OK, because the value does not change
across packages.
However, for builds with per-package directories, this value only refers
to the generic path, which ill not exist until the end of the build when
all packages are aggregated in the host-finalize step.
Fix that by postponing the variable evaluation like all the others.
Reported-by: Xogium on IRC
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
(cherry picked from commit 548b8c5412)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
When using the headers from the kernel to be built, with the kernel
set to a custom version, and overriding the kernel sources with
LINUX_OVERRIDE_SRCDIR, the linux-headers package is still trying to
download an archive, and fails to validate its hash.
What is going on under the hood is that, with _OVERRIDE_SRCDIR, the
_VERSION of a package is set to 'custom'. Furthermore, the variable
BR_NO_CHECK_HASH_FOR is recursively expanded, so its value is only
evaluated when it is needed.
For linux-headers, we inherit the values from the linux package, and
the LINUX_HEADERS_VERSION takes the value from the configuration.
Thus we end up with the following situation:
LINUX_VERSION=custom
LINUX_HEADERS_VERSION=5.10 # For example
BR_NO_CHECK_HASH_FOR=... linux-custom.tar.gz ...
And thus the archive downloaded by linux-headers will not match any
exclusion, and since there will most probably not be a hash for it,
the download will fail, as was noticed and reported by Jarkko.
But in this case, what we really want is to really use the headers
from the kernel that we build, we do not even want to attempt a
download at all.
So, when using the headers from the kernel to be built, we also
propagate the LINUX_OVERRIDE_SRCDIR to linux-headers, so that we
also use the headers from the overridden sources.
Furthermore, in that configuration, we explicitly disallow
overriding the linux-headers specifically, as it does not make sense
(even though, if they were overridden to the same location, that'd
be OK, but to simplify the condition, we do not even check for that).
Reported-by: Jarkko Sakkinen <jjs@kapsi.fi>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
(cherry picked from commit b9e7adc152)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
