- Use tar.gz as SHA512SUMS does not contain the hash for tar.xz
- Fix CVE-2020-15900: A memory corruption issue was found in Artifex
Ghostscript 9.50 and 9.52. Use of a non-standard PostScript operator
can allow overriding of file access controls. The 'rsearch'
calculation for the 'post' size resulted in a size that was too large,
and could underflow to max uint32_t.
https://www.ghostscript.com/doc/9.53.0/News.htm
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
(cherry picked from commit cae8be20ed)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Buildroot generates Linux systems, so there is no need to have
MS Windows executables in there.
This reduces the target filesystem size by about 600kB:
$ du -hcs usr/lib/python2.7/distutils/command/wininst-*.exe
60K usr/lib/python2.7/distutils/command/wininst-6.0.exe
64K usr/lib/python2.7/distutils/command/wininst-7.1.exe
60K usr/lib/python2.7/distutils/command/wininst-8.0.exe
192K usr/lib/python2.7/distutils/command/wininst-9.0.exe
220K usr/lib/python2.7/distutils/command/wininst-9.0-amd64.exe
596K total
Signed-off-by: Titouan Christophe <titouan.christophe@railnova.eu>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
(cherry picked from commit a7e71716f5)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Buildroot generates Linux systems, so there is no need to have
MS Windows executables in there.
This reduces the target filesystem size by about 2MB:
$ du -hcs usr/lib/python3.8/distutils/command/wininst-*.exe
60K usr/lib/python3.8/distutils/command/wininst-6.0.exe
64K usr/lib/python3.8/distutils/command/wininst-7.1.exe
60K usr/lib/python3.8/distutils/command/wininst-8.0.exe
192K usr/lib/python3.8/distutils/command/wininst-9.0.exe
220K usr/lib/python3.8/distutils/command/wininst-9.0-amd64.exe
188K usr/lib/python3.8/distutils/command/wininst-10.0.exe
220K usr/lib/python3.8/distutils/command/wininst-10.0-amd64.exe
448K usr/lib/python3.8/distutils/command/wininst-14.0.exe
576K usr/lib/python3.8/distutils/command/wininst-14.0-amd64.exe
2,0M total
Signed-off-by: Titouan Christophe <titouan.christophe@railnova.eu>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
(cherry picked from commit 4e97032c82)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Commit 4e60247be6 ("package/xerces: fix
coding style in Config.in") introduced a silly typo in the xerces
package depends on... which effectively made it impossible to enable
this package as the "USE_WCHAR" option does not exist.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
(cherry picked from commit 6e61a4ce04)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
When generating a .pyc file, the original .py source file path is
encoded in it. It is used for various purposes: traceback generation,
.pyc file comparison with its .py source, and code inspection.
By default, the source path used when invoking compileall is encoded in
the .pyc file. Since we use paths relative to TARGET_DIR, we end up with
paths that are only valid when relative to '/' encoded in the installed
.pyc files on the target.
This breaks code inspection at runtime since the original source path
will be invalid unless the code is executed from '/'.
Unfortunately, compileall cannot be forced to use the proper path. It
was not written with cross-compilation usage in mind.
Rework the script to call py_compile.compile() directly with pertinent
options:
- The script now has a new --strip-root argument. This argument is
optional but will always be specified when compiling py files in
buildroot.
- All other (non-optional) arguments are folders in which all
"importable" .py files will be compiled to .pyc.
- Using --strip-root=$(TARGET_DIR), the future runtime path of each .py
file is computed and encoded into the compiled .pyc.
No need to change directory before running the script anymore.
The trickery used to handle error reporting was only applicable with
compileall. Since we implement our own "compileall", error reporting
becomes trivial.
Previously, we had a --force option to tell compileall.compiledir() to
forcibly recompile files if they had changed. Now, we would have to
handle it ourselves. It turns out to not be easy and would need us to
delve into the format of bytecompiled files to extract metadata and
compare it with the expected values, that being even dependent on the
python version being used (fortunately, only two for us: python 2.7 and
the latext 3.x).
Still, this is deemed too complex, and byte-compiling is pretty fast, so
much so that it should be eclipsed by the build duration anyway.
So we just drop support for --force, and instead we always byte-compile.
Signed-off-by: Julien Floret <julien.floret@6wind.com>
Signed-off-by: Robin Jarry <robin.jarry@6wind.com>
[yann.morin.1998@free.fr:
- always byte-compile
- drop --force
- expand commit log to state so and explain why
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
(cherry picked from commit c566f5206a)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Only run code when the script is executed directly (not imported).
Factorize command description by using the script's __doc__ variable.
Fix typo in --force help message.
Signed-off-by: Robin Jarry <robin.jarry@6wind.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
(cherry picked from commit 7b3025f93e)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
This commit adds a patch that passes -fno-builtin to fix the build
with gcc 10. The patch has been submitted upstream.
There are no autobuilder failures, since it's a bootloader package,
and therefore not tested by the autobuilders. However, a build failure
was detected when building toolchains and a minimal ARMv7-M system at
https://gitlab.com/bootlin/toolchains-builder/-/jobs/729359529.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
(cherry picked from commit 1c967e2c8c)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Fixes the following security issues:
- CVE-2020-15166: Denial-of-Service on CURVE/ZAP-protected servers by
unauthenticated clients.
If a raw TCP socket is opened and connected to an endpoint that is fully
configured with CURVE/ZAP, legitimate clients will not be able to exchange
any message. Handshakes complete successfully, and messages are delivered
to the library, but the server application never receives them. For more
information see the security advisory:
https://github.com/zeromq/libzmq/security/advisories/GHSA-25wp-cf8g-938m
- Stack overflow on server running PUB/XPUB socket (CURVE disabled).
The PUB/XPUB subscription store (mtrie) is traversed using recursive
function calls. In the remove (unsubscription) case, the recursive calls
are NOT tail calls, so even with optimizations the stack grows linearly
with the length of a subscription topic. Topics are under the control of
remote clients - they can send a subscription to arbitrary length topics.
An attacker can thus cause a server to create an mtrie sufficiently large
such that, when unsubscribing, traversal will cause a stack overflow. For
more information see the security advisory:
https://github.com/zeromq/libzmq/security/advisories/GHSA-qq65-x72m-9wr8
- Memory leak in PUB server induced by malicious client(s) without CURVE/ZAP.
Messages with metadata are never processed by PUB sockets, but the
metadata is kept referenced in the PUB object and never freed. For more
information see the security advisory:
https://github.com/zeromq/libzmq/security/advisories/GHSA-4p5v-h92w-6wxw
- Memory leak in client induced by malicious server(s) without CURVE/ZAP.
When a pipe processes a delimiter and is already not in active state but
still has an unfinished message, the message is leaked.
For more information see the security advisory:
https://github.com/zeromq/libzmq/security/advisories/GHSA-wfr2-29gj-5w87
- Heap overflow when receiving malformed ZMTP v1 packets (CURVE disabled).
By crafting a packet which is not valid ZMTP v2/v3, and which has two
messages larger than 8192 bytes, the decoder can be tricked into changing
the recorded size of the 8192 bytes static buffer, which then gets
overflown by the next message. The content that gets written in the
overflown memory is entirely decided by the sender.
For more information see the security advisory:
https://github.com/zeromq/libzmq/security/advisories/GHSA-fc3w-qxf5-7hp6
Drop now upstreamed patches, autoreconf and reformat hash file with 2 space
delimiters.
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
(cherry picked from commit fd1ac2e762)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
On aarch64 With the config option "--enable-mpers=check" the configure.ac
script searchs for a 32bit compiler. When a matching compiler is found
in the PATH some compatiblity checks are done. This can fail when the
available kernel headers on host and buildroot target does not match.
Since buildroot does not support 32bit binaries when building for 64bit
architecture (no -m32 option) we can disable this option unconditionally.
When disabling unconditionally also the configuration for toolchain using
MUSL can be removed.
Cc: Baruch Siach <baruch@tkos.co.il>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Brandon Maier <brandon.maier@rockwellcollins.com>
Signed-off-by: Heiko Thiery <heiko.thiery@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
(cherry picked from commit bae6142582)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Since commit efa95b19ae
("package/libxml-parser-perl: make host build use correct compiler"),
we pass $(HOST_CONFIGURE_OPTS) when building host-libxml-parser-perl,
in order to use the correct host compiler.
However, this means that LD="$(HOSTLD)" is passed. However, the
host-libxml-parser-perl passes compiler arguments to LD, so it really
assumes that LD is gcc, not ld. For example, it tries to pass
-mtune=generic.
So, let's tell host-libxml-parser-perl that LD is "$(HOSTCC)".
Fixes:
http://autobuild.buildroot.net/results/2ed2e5ccefe9047c597f84d5880de2e8de2bdd94/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Tested-by: Joseph Kogut <joseph.kogut@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
(cherry picked from commit 43a26d7fe4)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
This package uses gcc filename without absolute path, which breaks
the host build if host and target compiler have the same filename.
(Can happen with an external toolchain).
This patch adds the variables for the host as overrides,
as they are otherwise not picked up from the environment.
Signed-off-by: Norbert Lange <nolange79@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
(cherry picked from commit efa95b19ae)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
- Fix CVE-2020-15503: LibRaw before 0.20-RC1 lacks a thumbnail size
range check. This affects decoders/unpack_thumb.cpp,
postprocessing/mem_image.cpp, and utils/thumb_utils.cpp. For example,
malloc(sizeof(libraw_processed_image_t)+T.tlength) occurs without
validating T.tlength.
- zlib is an optional dependency since
b63f017b06
Also update indentation in hash file (two spaces) as well as README.md
hash, no license changes:
- d1975cb0e0
- d38361b76e
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
(cherry picked from commit fd50e0f93f)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Contains fixes for overflows when input chunks are larger than 2 GiB,
an uninitialized data access, and minor correctness and performance
improvements. There does not seem to be any CVEs filed, but there is
a security notice in the release notes at:
https://github.com/google/brotli/releases/tag/v1.0.9
Patch "0001-CMake-Allow-using-BUILD_SHARED_LIBS-to-choose-static.patch"
is rebased against the latest upstream changes.
Signed-off-by: Adrian Perez de Castro <aperez@igalia.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
(cherry picked from commit 777bbd1b07)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
No MINIDLNA_IGNORE_CVES entry is added as no CVE has been assigned to
minidlna. Indeed, CallStranger vulnerability affect(ed) most of the UPnP
stacks (e.g. gupnp, libupnp)
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
(cherry picked from commit 9ab9118831)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Per-package build of apparmor with apache fails on:
/usr/lfs/hdd_v1/rc-buildroot-test/scripts/instance-0/output-1/per-package/apparmor/host/x86_64-buildroot-linux-musl/sysroot/usr/bin/apxs -c mod_apparmor.c -L/usr/lfs/hdd_v1/rc-buildroot-test/scripts/instance-0/output-1/per-package/apparmor/host/bin/../x86_64-buildroot-linux-musl/sysroot/usr/lib -lapparmor
/usr/lfs/hdd_v1/rc-buildroot-test/scripts/instance-0/output-1/per-package/apache/host/x86_64-buildroot-linux-musl/sysroot/usr/bin/../../usr/build-1/libtool --silent --mode=compile /usr/lfs/hdd_v1/rc-buildroot-test/scripts/instance-0/output-1/per-package/apache/host/bin/x86_64-linux-gcc -prefer-pic -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -Os -g2 -I/usr/lfs/hdd_v1/rc-buildroot-test/scripts/instance-0/output-1/per-package/apache/host/x86_64-buildroot-linux-musl/sysroot/usr/include -I/usr/lfs/hdd_v1/rc-buildroot-test/scripts/instance-0/output-1/per-package/apache/host/x86_64-buildroot-linux-musl/sysroot/usr/bin/../../usr/include/apr-1 -I/usr/lfs/hdd_v1/rc-buildroot-test/scripts/instance-0/output-1/per-package/apache/host/x86_64-buildroot-linux-musl/sysroot/usr/bin/../../usr/include/apr-1 -I/usr/lfs/hdd_v1/rc-buildroot-test/scripts/instance-0/output-1/per-package/apache/host/x86_64-buildroot-linux-musl/sysroot/usr/bin/../../../../x86_64-buildroot-lin
ux-musl/sysroot/usr/include -c -o mod_apparmor.lo mod_apparmor.c && touch mod_apparmor.slo
mod_apparmor.c:28:10: fatal error: sys/apparmor.h: No such file or directory
#include <sys/apparmor.h>
^~~~~~~~~~~~~~~~
The issue is that sys/appamor.h is not installed in the apache
per-package directory which is mangled by
APACHE_FIX_STAGING_APACHE_CONFIG, i.e.
/usr/lfs/hdd_v1/rc-buildroot-test/scripts/instance-0/output-1/per-package/apache/host/x86_64-buildroot-linux-musl/sysroot/usr/include
So implement the same workaround made on apache to replace those wrong
apache paths by apparmor paths in apxs binary and its configuration file
(i.e. config_vars.mk) as suggested by Thomas Petazzoni and Yann E. Morin
during review of the first iteration of this patch
Fixes:
- http://autobuild.buildroot.org/results/ef1fcd57e0c09a2806bf2272bb21df6d3300b45b
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
(cherry picked from commit 3c836e5420)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
libgnutls: Fixed "no_renegotiation" alert handling at incorrect timing.
The server sending a "no_renegotiation" alert in an unexpected timing,
followed by an invalid second handshake was able to cause a TLS 1.3
client to crash via a null-pointer dereference. The crash happens in the
application's error handling path, where the gnutls_deinit function is
called after detecting a handshake failure (#1071).
[GNUTLS-SA-2020-09-04, CVSS: medium]
https://lists.gnupg.org/pipermail/gnutls-help/2020-September/004669.html
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
(cherry picked from commit fb3b23220b)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
In commit c2009e9f75
("package/linux-headers: license files hashes only valid for latest
version"), we introduced BR2_KERNEL_HEADERS_LATEST, which should only
be set for the most recent kernel headers versions.
Indeed, the COPYING file of Linux has changed before/after Linux 5.6,
causing its hash file to be different. Since linux-headers uses
linux/linux.hash as the hash file, and this hash file contains the
COPYING hash of Linux >= 5.6, we cannot use that hash for Linux
versions older than 5.6.
When newer versions of the headers than 5.4 were added, this
BR2_KERNEL_HEADERS_LATEST was not moved as it should have been. We fix
this, which fixes a legal-info failure happening when Linux kernel
headers 5.4 are used:
>>> linux-headers 5.4.61 Patching
>>> linux-headers 5.4.61 Collecting legal info
ERROR: COPYING has wrong sha256 hash:
ERROR: expected: fb5a425bd3b3cd6071a3a9aff9909a859e7c1158d54d32e07658398cd67eb6a0
ERROR: got : ee5808b032a67f587d3541099d46de34f5bec8cd5976114ba07f1299ee6001ff
ERROR: Incomplete download, or man-in-the-middle (MITM) attack
make: *** [package/linux-headers/linux-headers.mk:158: linux-headers-legal-info] Error 1
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Fixes the following security issue:
CVE-2020-16845: Go before 1.13.15 and 14.x before 1.14.7 can have an
infinite read loop in ReadUvarint and ReadVarint in encoding/binary via
invalid inputs
https://github.com/golang/go/issues/40620
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Fixes the following security issues:
CVE-2020-25125: Importing an OpenPGP key having a preference list for AEAD
algorithms will lead to an array overflow and thus often to a crash or other
undefined behaviour (affected: 2.2.21 / 2.2.22)
For more details, see the announcement:
https://lists.gnupg.org/pipermail/gnupg-announce/2020q3/000448.html
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
(cherry picked from commit 918a9fb455)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Cherry-pick a patch from upstream project that fixes the build when used
with a uClibc based toolchain.
Signed-off-by: Paul Cercueil <paul@crapouillou.net>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
(cherry picked from commit 0e3febb75f)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
The ControlValue structure is currently defined with a 16-bit hole
(causing unaligned access to the numElements_ field, though that's a
separate topic).
This structure has a static assertion to ensure that its size does not
change without due care, as it forms part of our ABI and is used in
Serialisation between the pipeline handlers and IPA components.
The m68k architecture is the only target which fails this assertion,
which is likely because it can pack the structure more efficiently,
producing a different binary size.
This is likely an area we will tackle before stabilising our ABI, but
until then, disable m68k builds as libcamera is not expected to be
supported on this target.
Fixes;
- http://autobuild.buildroot.net/results/9dce26e94299a2c61bba60cbc7803926e2f85e29/
Signed-off-by: Kieran Bingham <kieran.bingham@ideasonboard.com>
[yann.morin.1998@free.fr, suggestions from Thomas:
- introduce BR2_PACKAGE_LIBCAMERA_ARCH_SUPPORTS
- propagate that to the comment
- add autobuilder reference
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
(cherry picked from commit 66526e3518)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Extract from bug report:
"In usb_modeswitch Makefile dispatcher-script, dispatcher-dynlink and
dispatcher-statlink are .PHONY targets. The result is that sources are
compiled also when install targets are called.
USB_MODESWITCH_INSTALL_TARGET_CMDS calls $(MAKE) which is a call to
parallel make eg. make -j9. So the install phase can install empty
usb_modeswitch binary (happened once) if the compiler have just cleared
the binary and install command installs it before compiler writes the
binary. USB_MODESWITCH_INSTALL_TARGET_CMDS should call $(MAKE1)."
Instead of disabling parellel install, use install-common target instead
of install-{dyn,stat}link targets. Indeed, the dynamic or static
usb_modeswitch_dispatcher binary will be built by
all-with-{dyn,stat}link-dispatcher targets, there is no need to rebuild
it during the install step
Fixes:
- https://bugs.buildroot.org/show_bug.cgi?id=12911
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
(cherry picked from commit 345c68f04f)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
