package/mosquitto: bump to v2.0.8

Mosquitto 2.0.8 is bugfix release. See the announcement:
https://mosquitto.org/blog/2021/02/version-2-0-8-released/

Signed-off-by: Titouan Christophe <titouanchristophe@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/openssh: security bump to version 8.4p1

Fixes CVE-2020-15778: scp in OpenSSH through 8.3p1 allows command injection in
the scp.c toremote function, as demonstrated by backtick characters in the
destination argument. NOTE: the vendor reportedly has stated that they
intentionally omit validation of "anomalous argument transfers" because that
could "stand a great chance of breaking existing workflows."

https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2020-15778

Signed-off-by: Christian Stewart <christian@paral.in>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/haproxy: bump to version 2.2.9

https://www.mail-archive.com/haproxy@formilux.org/msg39744.html

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Update for 2021.02-rc3

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/python-aiohttp: security bump to version 3.7.4

Fixes the following security issue:

CVE-2021-21330: Open redirect vulnerability in aiohttp
(normalize_path_middleware middleware)

Beast Glatisant and Jelmer Vernooij reported that python-aiohttp, a async
HTTP client/server framework, is prone to an open redirect vulnerability.  A
maliciously crafted link to an aiohttp-based web-server could redirect the
browser to a different website.

For more details, see the advisory:
https://github.com/aio-libs/aiohttp/security/advisories/GHSA-v6wp-4m6f-gcjg

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/libglib2: security bump to version 2.66.7

- Fix CVE-2021-27218: An issue was discovered in GNOME GLib before
  2.66.7 and 2.67.x before 2.67.4. If g_byte_array_new_take() was called
  with a buffer of 4GB or more on a 64-bit platform, the length would be
  truncated modulo 2**32, causing unintended length truncation.
- Fix CVE-2021-27219: An issue was discovered in GNOME GLib before
  2.66.6 and 2.67.x before 2.67.3. The function g_bytes_new has an
  integer overflow on 64-bit platforms due to an implicit cast from 64
  bits to 32 bits. The overflow could potentially lead to memory
  corruption.

https://gitlab.gnome.org/GNOME/glib/-/blob/2.66.7/NEWS

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/openntpd: add OPENNTPD_CPE_ID_VENDOR

cpe:2.3:a:openntpd:openntpd is a valid CPE identifier for this package:

https://nvd.nist.gov/products/cpe/search/results?namingFormat=2.3&keyword=cpe%3A2.3%3Aa%3Aopenntpd%3Aopenntpd

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/openldap: add upstream security fix for CVE-2021-27212

In OpenLDAP through 2.4.57 and 2.5.x through 2.5.1alpha, an assertion
failure in slapd can occur in the issuerAndThisUpdateCheck function via a
crafted packet, resulting in a denial of service (daemon exit) via a short
timestamp.  This is related to schema_init.c and checkTime.

For more details, see the bugtracker:
https://bugs.openldap.org/show_bug.cgi?id=9454

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/screen: add security fix for CVE-2021-26937

encoding.c in GNU Screen through 4.8.0 allows remote attackers to cause a
denial of service (invalid write access and application crash) or possibly
have unspecified other impact via a crafted UTF-8 character sequence.

For more details, see the oss-security discussion:
https://www.openwall.com/lists/oss-security/2021/02/09/3

So far no fix has been added to upstream git, and a number of early proposed
fixes caused regressions, so pull the security fix from the screen 4.8.0-5
Debian package.

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/imagemagick: disable remaining config options (heic, jxl, openjp2)

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/imagemagick: add optional libraw support

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/imagemagick: add optional zstd support

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/imagemagick: add optional libzip support

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/imagemagick: security bump to version 7.0.10-62

Fixes the following security issue:

CVE-2021-20176: A divide-by-zero flaw was found in ImageMagick 6.9.11-57 and
7.0.10-57 in gem.c.  This flaw allows an attacker who submits a crafted file
that is processed by ImageMagick to trigger undefined behavior through a
division by zero.  The highest threat from this vulnerability is to system
availability.

For more details, see the bugtracker:
https://github.com/ImageMagick/ImageMagick/issues/3077

- bump version to 7.0.10-62
- update license file hash (copyright year update)

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
[Peter: mention security fix]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/readline: disable bracketed paste by default

As of readline 8.1, "bracketed paste" is enabled by default. However,
the feature causes control characters to appear in captured (telnet)
session output. This can throw off pattern matching if the output is to
be processed by scripts.

Let's keep the previous default of leaving this feature disabled and
provide a configuration option for users to enable it.

Signed-off-by: Markus Mayer <mmayer@broadcom.com>
[yann.morin.1998@free.fr:
  - explicit enable/disable
  - no indentation in conditional block
  - rewrap help text
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/qemu: disable tests

tests/fp/fp-bench.c use fenv.h that is not always provided
by the libc (uClibc).

To workaround this issue, add an new meson option to
disable tests while building Qemu.

Fixes:
http://autobuild.buildroot.net/results/53f5d8baa994d599b9da013ee643b82353366ec3/build-end.log

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/botan: avoid empty -l

Add upstream patch to fix upstream commit
af63fe89228172e5a395f7e6491fae3bfa9da4b1 which was added to buildroot in
commit d71de4143d7a8554929f2a1e9731f83a4cf85fd3

Fixes:
 - http://autobuild.buildroot.org/results/801007860b7787b28b2b2e3611b59350034a3694

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/libuwsc: disable example

BUILD_EXAMPLE=OFF is already passed by cmake-package

Fixes:
 - http://autobuild.buildroot.org/results/f5256d5a3a86112f008506f1910d0600c491a2a0

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/brltty: fix build with gcc < 5

Fix build of brltty in version 6.2 with gcc < 5

Fixes:
 - http://autobuild.buildroot.org/results/b758c6ffc7a14b24d5482e65ba6f90bc046ebd01

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
[yann.morin.1998@free.fr: do an actual backport]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/babeltrace2: link with libatomic if needed

Fix build of babeltrace2 in version 2.0.3 with Bootlin SPARC uclibc
toolchain added with commit 1348c569d0cb7f67eca30f170b782aa8b51cc259

Fixes:
 - http://autobuild.buildroot.org/results/31770bf70f9ce4e3be8fb310d084b214820c6829

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/elfutils: link with libatomic if needed

Fix build of elfutils 0.181 with Bootlin SPARC uclibc toolchain added
with commit 1348c569d0cb7f67eca30f170b782aa8b51cc259

Fixes:
 - http://autobuild.buildroot.org/results/31ce9e3861c6229a7869a15d322f5d2f5bfc6165

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/intel-mediasdk: disable samples and tutorials

Disable samples and tutorials which are enabled by default and fail to
build with gcc 10 without upstream commit:
https://github.com/Intel-Media-SDK/MediaSDK/commit/c7d40371eb0c2042261fe1f91a364f69a1457235

Fixes:
 - http://autobuild.buildroot.org/results/9ee28e5dc0b2ba854766d9bc82b95c28be2722d3

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/nodejs: security bump to version v12.21.0

Fixes the following security issues:

CVE-2021-22883: HTTP2 'unknownProtocol' cause Denial of Service by resource exhaustion

Affected Node.js versions are vulnerable to denial of service attacks when
too many connection attempts with an 'unknownProtocol' are established.
This leads to a leak of file descriptors.  If a file descriptor limit is
configured on the system, then the server is unable to accept new
connections and prevent the process also from opening, e.g.  a file.  If no
file descriptor limit is configured, then this lead to an excessive memory
usage and cause the system to run out of memory.

CVE-2021-22884: DNS rebinding in --inspect

Affected Node.js versions are vulnerable to denial of service attacks when
the whitelist includes “localhost6”.  When “localhost6” is not present in
/etc/hosts, it is just an ordinary domain that is resolved via DNS, i.e.,
over network.  If the attacker controls the victim's DNS server or can spoof
its responses, the DNS rebinding protection can be bypassed by using the
“localhost6” domain.  As long as the attacker uses the “localhost6” domain,
they can still apply the attack described in CVE-2018-7160.

For more details, see the advisory:
https://nodejs.org/en/blog/vulnerability/february-2021-security-releases/

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/ply: build needs flex and bison

Building needs flex and bison installed on the host system.

Fixes:
http://autobuild.buildroot.net/results/7cfe75725f4746367f2870ee9545f31ba56f6ec1

Signed-off-by: Andreas Klinger <ak@it-klinger.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/screen: add SCREEN_CPE_ID_VENDOR

cpe:2.3:a:gnu:screen is a valid CPE identifier for this package:

  https://nvd.nist.gov/products/cpe/search/results?namingFormat=2.3&keyword=cpe%3A2.3%3Aa%3Agnu%3Ascreen

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/xterm: add XTERM_CPE_ID_VENDOR

cpe:2.3:a:invisible-island:xterm is a valid CPE identifier for this
package:

  https://nvd.nist.gov/products/cpe/search/results?namingFormat=2.3&keyword=cpe%3A2.3%3Aa%3Ainvisible-island%3Axterm

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/python3: security bump to version 3.9.2

Fixes the following security issue:

- CVE-2021-23336: urllib.parse.parse_qsl(): Web cache poisoning - `; ` as a
  query args separator
  https://bugs.python.org/issue42967

And fixes a number of issues. For details, see the changelog:
https://docs.python.org/release/3.9.2/whatsnew/changelog.html

Drop the now upstreamed security patch and update the license hash for a
change of copyright year:

-2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019, 2020 Python Software Foundation;
+2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019, 2020, 2021 Python Software Foundation;

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

support/download: drop sub-second precision in tarball creation

Some download backends, like svn, will provide timestamps with a
sub-second precision, e.g.

    $ svn info --show-item last-changed-date [...]
    2021-02-19T20:22:34.889717Z

However, the PAX headers do not accept sub-second precision, leading to
failure to download from subversion:

    tar: Time stamp is out of allowed range
    tar: Exiting with failure status due to previous errors
    make[1]: *** [package/pkg-generic.mk:148: [...]/build/subversion-1886712/.stamp_downloaded] Error 1

Fix that by massaging the timestamp to drop the sub-second part. We
do that in the generic helper, rather than the svn backend, so that
all callers to the generic helper benefit from this, as this is more
an internal details of the tarball limitations, than of the backends
themselves.

Reported-by: Roosen Henri <Henri.Roosen@ginzinger.com>
Signed-off-by: Vincent Fazio <vfazio@xes-inc.com>
[yann.morin.1998@free.fr:
  - add Henri as reporter
  - move it out of the svn backend, and to the generic helper
  - reword the commit log accordingly
  - use an explicit time format rather than -Iseconds
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/bind: security bump to version 9.11.28

Fixes the following security issue:

- CVE-2020-8625: When tkey-gssapi-keytab or tkey-gssapi-credential was
  configured, a specially crafted GSS-TSIG query could cause a buffer
  overflow in the ISC implementation of SPNEGO (a protocol enabling
  negotiation of the security mechanism to use for GSSAPI authentication).
  This flaw could be exploited to crash named.  Theoretically, it also
  enabled remote code execution, but achieving the latter is very difficult
  in real-world conditions

For details, see the advisory:
https://kb.isc.org/docs/cve-2020-8625

In addition, 9.11.26-27 fixed a number of issues, see the release notes for
details:
https://downloads.isc.org/isc/bind9/9.11.28/RELEASE-NOTES-bind-9.11.28.html

Drop now upstreamed patches, update the GPG key for the 2021-2022 variant
and update the COPYRIGHT hash for a change of year:

-Copyright (C) 1996-2020  Internet Systems Consortium, Inc. ("ISC")
+Copyright (C) 1996-2021  Internet Systems Consortium, Inc. ("ISC")

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/fakeroot: fix glibc detection on patch for new wrappers

Commit f45925a951318e9e53bead80b363e004301adc6f add the patch:

0003-libfakeroot.c-add-wrappers-for-new-glibc-2.33-symbol.patch

which allowed fakeroot to be compiled with GLIBC 2.33 or above.
However, this introduce a bug for building with a non-GLIBC based
toolchain as a GLIBC macro - __GLIBC_PREREQ - is used on the same line
as the detection of GLIBC.

Fix this by backporting the fix to this incorrect macro from upstream
commit:

https://salsa.debian.org/clint/fakeroot/-/commit/8090dffdad8fda86dccd47ce7a7db8840bdf7d7b

CC: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Ryan Barnett <ryanbarnett3@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/unbound: bump to version 1.13.1

This release contains a number of bug fixes. There is added support
for the EDNS Padding option (RFC7830 and RFC8467), and the EDNS NSID
option (RFC 5001). Unbound control has added commands to enable and
disable rpz processing. Reply callbacks have a start time passed to
them that can be used to calculate time, these are callbacks for
response processing. With the option serve-original-ttl the TTL served
in responses is the original, not counted down, value, for when in
front of authority service.

https://github.com/NLnetLabs/unbound/releases/tag/release-1.13.1

Signed-off-by: Stefan Ott <stefan@ott.net>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/irqbalance: fix irqbalance/irqbalance-ui socket communication

Add patch to fix irqbalance/irqbalance-ui socket communication by
fixing uint64_t printf format usage.

Fixes:

  $ irqbalance-ui
  Invalid data sent.  Unexpected token: (null)TYPE

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
[yann.morin.1998@free.fr:
  - do an actual backport as upstream applied the patch
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/open62541: fix library version definition

Manually specified version must start with letter 'v',
otherwise, the generated version macro will be zero
in the <build_dir>/src_generated/open62541/config.h file:
  #define UA_OPEN62541_VER_MAJOR 0
  #define UA_OPEN62541_VER_MINOR 0
  #define UA_OPEN62541_VER_PATCH 0

Reference from the following link:
https://open62541.org/doc/current/building.html

Signed-off-by: Scott Fan <fancp2007@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

support/scripts/boot-qemu-image.py: properly catch timeout

As reported on IRC by sephthir, the gitlab test of the defconfig
qemu_sparc_ss10_defconfig doesn't error out while the system
is not working properly.

This is because we explicitly wait for the timeout as an expected
condition, but do not check for it. Indeed, pexpect.expect() returns
the index of the matching condition in the list of expected conditions,
but we just ignore the return code, so we are not able to differentiate
between a successful login (or prompt) from a timeout.

By default, pexepect.expect() raises the pexpect.TIMEOUT exception on a
timeout, and we are already prepared to catch and handle that exception.
But because pexpect.TIMEOUT is passed as an expected condition, the
exception is not raised.

Remove pexpect.TIMEOUT from the list of expected conditions, so that the
exception is properly raised again, and so that we can catch it.

The qemu_sparc_ss10_defconfig is already fixed by
4d16e6f5324f0285f51bfbb5a3503584f3b3ad12.

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Jugurtha BELKALEM <jugurtha.belkalem@smile.fr>
[yann.morin.1998@free.fr: reword commit log]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/irqbalance: fix sysv startup script (add mkdir /run/irqbalance)

- add mkdir -p /run/irqbalance to sysv startup script needed to
  create socket /run/irqbalance/irqbalance<pid>.sock

Fixes:

  - Bug 13541 [1]

  daemon.warn /usr/sbin/irqbalance: Daemon couldn't be bound to the file-based socket.

[1] https://bugs.busybox.net/show_bug.cgi?id=13541

Reported-by: Alfredo Pons Menargues <alfredo.pons@gmail.com>
Signed-off-by: Peter Seiderer <ps.report@gmx.net>
[yann.morin.1998@free.fr: only create in start case]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/irqbalance: fix systemd startup script (add RuntimeDirectory)

- add RuntimeDirectory=irqbalance to create /run/irqbalanace needed to
  create socket /run/irqbalance/irqbalance<pid>.sock

Fixes:

  - Bug 13541 [1]

  /usr/sbin/irqbalance[158]: Daemon couldn't be bound to the file-based socket.

[1] https://bugs.busybox.net/show_bug.cgi?id=13541

Reported-by: Alfredo Pons Menargues <alfredo.pons@gmail.com>
Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

DEVELOPERS: remove Scott Fan

Signed-off-by: Scott Fan <fancp2007@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

utils/scanpypi: use python3 explicitly

scanpypi is python3 compatible. In addition, it executes the setup.py
of Python modules to extract the relevant information. Since these are
more and more commonly using python3 constructs, using "python" to run
scanpypi causes problems on systems that have python2 installed as
python, when trying to parse setup.py scripts with python3 constructs.

Fixes part of #13516.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/taglib: drop config options to enable MP4/ASF support

Both options where removed in git commit dd846904cbc1ef3ee628d77f0c9df88ef8967816
back in year 2011.

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
[yann.morin.1998@free.fr: drop the legacy handling]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/rust: disable ninja

Ninja has recently be enabled as the default build system to build
llvm fork for rust compiler [1]. But we can still use Make if
"ninja = false" is provided in config.toml.

Ninja support can be enabled by a following patch.

[1] https://github.com/rust-lang/rust/commit/30b7dac745b1555cd96f41977f7d24435cbe7fa2

Fixes:
https://gitlab.com/buildroot.org/buildroot/-/jobs/1019386205

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Eric Le Bihan <eric.le.bihan.dev@free.fr>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/cegui: use plain assignemnt for first _CONF_OPTS

Commit 689b9c1a7cf5 (package/cegui: disable xerces support) added
an unconditional assignment to _CONF_OPTS before all the conditional
ones, but used the append-assignment instead of the traditional plain
assignment.

Fix that by removing the append-assignment.

Use that opportunity to also move the first item of this multi-line
assignment, to its own line.

Signed-off-by: Bartosz Bilas <b.bilas@grinn-global.com>
[yann.morin.1998@free.fr:
  - reference the exact commit that introduce the issue
  - also move the first item to its own line
]
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/python-django: security bump to version 3.0.13

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>

package/botan: fix build with -latomic

Static build with toolchains needing -latomic (e.g sparc) is broken
since version 2.17.0 and
https://github.com/randombit/botan/commit/88af81b88976d9a1293280f68df597220ab42767

Fixes:
 - http://autobuild.buildroot.org/results/5c03ee53a34a3cdb409cffcda76e5cc2c723778b

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/libselinux: fix build with musl 1.2.2

Fixes:
 - http://autobuild.buildroot.org/results/34b010e76d65cf1d79ef53207cbc00a86674e17a

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/libusb: apply upstream patch to fix descriptor parsing

v1.0.24 of libusb has a bug in the Linux backend where it fails to
enumerate any device with more than one configuration.  Backport the
upstream patch which fixes this as otherwise libusb based applications
are unable to communicate with any devices advertising more than one
configuration.

Signed-off-by: John Keeping <john@metanate.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

docs/website: update for 2020.02.11

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Update for 2020.02.11

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
(cherry picked from commit 08e03785d3812c085c438a6040ccedc3e9f5809d)
[Peter: drop Makefile changes]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

docs/website: update for 2020.11.3

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Update for 2020.11.3

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
(cherry picked from commit 610e67b1fc4ac44e0c4a7ba437c917ad6d63f481)
[Peter: drop Makefile changes]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Update for 2021.02-rc2

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/perl-extutils-pkgconfig: set PATH to BR_PATH

Set PATH to BR_PATH to allow perl-extutils-pkgconfig to find pkg-config
binary

Fixes:
 - http://autobuild.buildroot.org/results/d87787fbf2a8cb9bbaa3b59d1e8004ad1459536a

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/libopenssl: security bump to version 1.1.1j

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>

support/scripts/pkg-stats: add ignored_cves to json output

Add the list of <pkg>_IGNORE_CVES to the json output to show that we have a
known cause (available patch or the CVE is not valid for our package
configuration) that a affected CVE is not reported.

Signed-off-by: Heiko Thiery <heiko.thiery@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/openblas: fix detection of gfortran compiler

The compiler detection since openblas 0.3.8 added support for gcc 10, but
this broke detection of compilers created with crosstool-ng, or other
toolchains that have a package version containing a version like x.y.z where
at least one of x, y or z have more than one digit, for example
"Crosstool-NG 1.24.0".

See the reported issue for more details [1].

Backport the upstream patch that fixes it.

[1] https://github.com/xianyi/OpenBLAS/issues/3099

Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/openblas: allow disabling multithreading

Buildroot would automatically enable multithreading in OpenBLAS if the
architecture supports it. However, one may want to avoid OpenBLAS creating
threads itself and configure single-threaded operation. To accommodate this
use case, add a config option for multithreading.

When multithreading is disabled but OpenBLAS functions are called in the
same application by multiple threads, then locking is mandatory. The
USE_LOCKING flag was added in version 0.3.7 with following release note:

    a new option USE_LOCKING was added to ensure thread safety when OpenBLAS
    itself is built without multithreading but will be called from multiple
    threads.

However, if one knows that OpenBLAS will only be called from single-threaded
applications, then passing USE_LOCKING is not necessary, so make it a config
option too.

When multithreading is enabled, locking is implicitly enabled inside
openblas, so only provide the locking option when multithreading is
disabled.

Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/flashrom: fix build on riscv

Retrieve an upstream patch to fix build with riscv as it fails to
retrieve architecture due to "Use sigaction with SA_RESTART instead"
being caught before riscv:

exec: export LC_ALL=C ; { /home/fabrice/buildroot/output/host/bin/riscv32-linux-gcc -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -E archtest.c 2>/dev/null | grep -v ^# | grep ' | cut -f 2 -d' ; }
Use sigaction with SA_RESTART instead
riscv

Fixes:
 - http://autobuild.buildroot.org/results/61ac6c9bfcd3bd9306aa49faf47b9f16e5abe846

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/gdk-pixbuf: fix static build

Fix static build failure which is raised since the switch to
meson-package in commit a7b51ed3013c919b293deb95299e33363fb9df70

Fixes:
 - http://autobuild.buildroot.org/results/6cd54c497f5d19342ec94ece713547b887e4c02d

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
[Arnout: add link to upstream MR]
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/wpewebkit: bump version to 2.30.5

Bugfix release, fixing a number of issues:

- Fix RunLoop objects leaked in worker threads.
- Fix JavaScriptCore AArch64 LLInt build with JIT disabled.
- Use Internet Explorer quirk for Google Docs.

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Acked-by: Adrian Perez de Castro <aperez@igalia.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/webkitgtk: security bump to version 2.30.5

Fixes the following security issue:

- CVE-2020-13558: Processing maliciously crafted web content may lead to
  arbitrary code execution.  Description: A use after free issue in the
  AudioSourceProviderGStreamer class was addressed with improved memory
  management

For more details, see the advisory:
https://webkitgtk.org/security/WSA-2021-0001.html

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Acked-by: Adrian Perez de Castro <aperez@igalia.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/perl-gd: provide gd options

Now that gdlib-config is gone, provide the GD options otherwise perl-gd
will assume that everything is available:

$features = 'GD_GIF GD_GIFANIM GD_OPENPOLYGON GD_ZLIB GD_PNG GD_FREETYPE GD_FONTCONFIG GD_JPEG GD_XPM GD_TIFF GD_WEBP';

Also, while at it, also make some of the dependencies as optional as
suggested by François Perrad

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Tested-by: Francois Perrad <francois.perrad@gadz.org> (with
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/perl-gd: needs perl-extutils-pkgconfig

Commit 3a291be2e89bc64388c10dae50233c751a86733d forgot to add
perl-extutils-pkgconfig dependency

Fixes:
 - http://autobuild.buildroot.org/results/e590f1990180eae21512b23b884755e105a4c588

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Acked-by: Francois Perrad <francois.perrad@gadz.org>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/perl-extutils-pkgconfig: new package

host-perl-extutils-pkgconfig is needed by perl-gd to find gd in version
2.3.0

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Acked-by: Francois Perrad <francois.perrad@gadz.org>
[Arnout:
 - remove Config.in - it's host-only
 - add DEVELOPERS entry
 - use HOST_PERL_EXTUTILS_PKGCONFIG_DEPENDENCIES]
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/tzdate: use classic 'fat' format for uClibc/glibc compatibility

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 7868289fd5348 (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:

https://github.com/eggert/tz/commit/6ba6f2117b95eab345a7ed9159cef939e30c4cd3

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>

package/uboot-tools: depend on u-boot when selected

Currently, the envimage creation logic only depends on u-boot when the
user does not specify a custom envimage source via
BR2_PACKAGE_HOST_UBOOT_TOOLS_ENVIMAGE_SOURCE. This assumes that the
user-provided envimage source is not coming from the u-boot source
tree.

But especially given the fact that the envimage creation logic used to
be part of the u-boot package, this is a realistic scenario: users may
have provided a value of BR2_PACKAGE_HOST_UBOOT_TOOLS_ENVIMAGE_SOURCE
based on $(UBOOT_DIR), e.g.:

    $(UBOOT_DIR)/board/foo-vendor/bar-board/env.txt

Therefore, always add the u-boot dependency if u-boot is selected, for
either case of custom or default envimage source.

Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
[Thomas: re-organize code a bit.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

support/scripts/cpedb.py: remove import pickle

pickle is no longer used since 09a71e6a75636

Fixes:
support/scripts/cpedb.py:7:1: F401 'pickle' imported but unused

Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/gstreamer1/gst1-python: needs gst1-plugins-base

gst1-plugins-base is a mandatory dependency since at least version
1.9.90 and
https://github.com/GStreamer/gst-python/commit/16f971226df1980b58ebde330123debaaf3b53d0

Fixes:
 - http://autobuild.buildroot.org/results/48b22c66c3a610d70931b9adfd6e5082bb3ff3d1

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/lcms2: disable tiff

tiff is only used by tificc sample and upstream rejected the patch to
fix the static build failure because "adding pkg-config dependency for a
sample is an overkill": https://github.com/mm2/Little-CMS/pull/244

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

Revert "package/lcms2: fix static building with tiff"

This reverts commit 7e4f054d2347708c9e22fe84c1d5f374d5b343cd.

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/collectd: add 'synproxy' option

The synproxy plugin exists since 5.8.0 and is enabled by default in
collectd.

Add an option in Buildroot, disabled by default.

Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/collectd: add 'logparser' option

The logparser plugin is new since 5.11.0 and enabled by default in
collectd.

Add an option in Buildroot, disabled by default.

Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/collectd: add 'mdevents' option

The mdevents plugin is new since 5.12.0 and enabled by default in
collectd.

Add an option in Buildroot, disabled by default.

Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/collectd: add 'infiniband' option

The infiniband plugin is new since 5.12.0 and enabled by default in
collectd.

Add an option in Buildroot, disabled by default.

Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/orc: fix powerpc build with headers < 4.11

Autobuilder failures are raised with bootlin toolchains but it affects
orc since version 0.4.30

Fixes:
 - http://autobuild.buildroot.org/results/0821e96cba3e455edd47b87485501d892fc7ac6a

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/ebtables: install symlink to ebtables-legacy

Since the upgrade of ebtables from 2.0.10-4 to 2.0.11, there no longer is an
'ebtables' binary. It has been renamed to 'ebtables-legacy' and moved from
'/sbin' to '/usr/sbin'. This change is part of the upstream change to
integrate the functionality of ebtables (and arptables) in the iptables
package, using the nf_tables kernel backend [1].

Unfortunately, the renaming (and move) of the original 'ebtables' binary
breaks existing scripts that are calling 'ebtables' or '/sbin/ebtables'.
Therefore, add a symlink from the original path to 'ebtables-legacy'.

However, do not provide this symlink if BR2_PACKAGE_IPTABLES_NFTABLES is
enabled. In this case, the iptables package will build the new equivalent
of ebtables -- a symlink to ebtables-legacy would cause conflicts.

[1] https://wiki.nftables.org/wiki-nftables/index.php/Legacy_xtables_tools

Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Acked-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/tcpdump: remove duplicated binary

Since tcpdump 4.99.0, the 'tcpdump' binary is no longer installed in
/usr/sbin but in /usr/bin. This change invalidates the Buildroot hook
'TCPDUMP_REMOVE_DUPLICATED_BINARY', causing a fairly large rootfs size
increase as a result.

Update the path inside this hook.

Signed-off-by: Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/mongoose: add MONGOOSE_CPE_ID_VENDOR

cpe:2.3:a:cesanta:mongoose is a valid CPE identifier for this package:

  https://nvd.nist.gov/products/cpe/search/results?namingFormat=2.3&keyword=cpe%3A2.3%3Aa%3Acesanta%3Amongoose

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/mongoose: security bump to version 7.1

- Fix CVE-2021-26528: The mg_http_serve_file function in Cesanta
  Mongoose HTTP server 7.0 is vulnerable to remote OOB write attack via
  connection request after exhausting memory pool.
- Fix CVE-2021-26529: The mg_tls_init function in Cesanta Mongoose HTTPS
  server 7.0 and 6.7-6.18 (compiled with mbedTLS support) is vulnerable
  to remote OOB write attack via connection request after exhausting
  memory pool.
- Fix CVE-2021-26530: The mg_tls_init function in Cesanta Mongoose HTTPS
  server 7.0 (compiled with OpenSSL support) is vulnerable to remote OOB
  write attack via connection request after exhausting memory pool.

https://github.com/cesanta/mongoose/releases/tag/7.1

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/mongoose: fix activation of openssl/mbedtls

MG_ENABLE_SSL and MG_SSL_IF have been dropped since version 7.0 and
https://github.com/cesanta/mongoose/commit/f2fba1d2004c5ddf2fc0a7ca8dc75b5f78feed85

So use the new MG_ENABLE_OPENSSL and MG_ENABLE_MBEDTLS variables

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

DEVELOPERS: drop Rahul Jain, user no longer exists

<rahul.jain@imgtec.com>: host mxa-00376f01.gslb.pphosted.com[185.132.180.163]
    said: 550 5.1.1 User Unknown (in reply to RCPT TO command)

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

DEVELOPERS: drop Guillaume Gardet, domain no longer exists

The oliseo.fr domain no longer responds to SMTP requests:

smtplib.SMTPRecipientsRefused: {'Guillaume Gardet <guillaume.gardet@oliseo.fr>': (550, b'5.1.2 <guillaume.gardet@oliseo.fr>: Recipient address rejected: Domain not found')}

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/fakeroot: add upstream patches to fix glibc 2.33 compatibility

Glibc 2.33 removed `_STAT_VER`. On host machines, 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 package maintainer threads, e.g.:
https://bugs.archlinux.org/task/69572
https://bugzilla.redhat.com/show_bug.cgi?id=1889862#c13

A patch series was prepared by Ilya Lipnitskiy which included two other
patches not related to the glibc 2.33 compatibility issue and submitted 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.

Tested on my Arch Linux machine, building a UBIFS/OverlayFS-based root
filesystem for an i.MX6ULL target board.

Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Tested-by: Bartosz Bilas <b.bilas@grinn-global.com>
[Peter: drop patch numbering (PATCH x/y) as pointed out by check-package]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/libgpg-error: fix build without threads

Fix build without threads of libgpg-error in version >= 1.40

Fixes:
 - http://autobuild.buildroot.org/results/3344c96e5627a9327b0eabe0b27f34490bbabc0d

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/ne10: disable unit tests and examples

Unit tests fail to build with gcc 10 on:

[100%] Linking C executable NE10_dsp_unit_test_smoke
/home/buildroot/autobuild/instance-2/output-1/host/opt/ext-toolchain/bin/../lib/gcc/aarch64-none-linux-gnu/10.2.1/../../../../aarch64-none-linux-gnu/bin/ld: CMakeFiles/NE10_dsp_unit_test_static.dir/__/modules/dsp/test/test_suite_fft_float32.c.o:(.bss+0x0): multiple definition of `seatest_simple_test_result'; CMakeFiles/NE10_dsp_unit_test_static.dir/__/modules/dsp/test/test_main.c.o:(.bss+0x0): first defined here

So just disable them and, while at it, also disable examples which are
also enabled by default

Fixes:
 - http://autobuild.buildroot.org/results/c658d52668825c26a15d6ac3ca538472cad5cd78

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/jasper: security bump version to 2.0.25

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>

support/scripts/cpedb.py: drop CPE XML database caching

Currently, the CPE XML database is parsed into a Python dict, which is
then pickled into a local file, to speed up the processing of further
invocations.

However, it turns out that since the initial implementation, we have
switched the XML parsing from the out of tree xmltodict module to the
standard ElementTree one, which has made the parsing much faster. The
pickle caching only saves 6 seconds, on something that takes more than
13 minutes total.

In addition, this pickle caching consumes a significant amount of RAM,
causing the Python process to be OOM-killed on a server with 4 GB of
RAM.

So let's just drop this caching entirely.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/ply: fix dependencies of comment for dependencies

Commits ca1afcb2171f (package/ply: needs headers >= 4.14) and
debe9eb13ebd (package/ply: needs dynamic library) added restrictions
on the availability of ply. The first forgot to add a comment, and
the second mis-handled the dependency on the headers version.

Indeed, we want the comment to show the requirement on the headers
version (since that is not a hardware dependency).

Fix this comment to include the headers version, and fix the condition
accordingly.

Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Cc: Andreas Klinger <ak@it-klinger.de>

package/dnsmasq: bump version to 2.84

Bugfix release, fixing a regression introduced in 2.83.  For more details,
see the announcement:

http://lists.thekelleys.org.uk/pipermail/dnsmasq-discuss/2021q1/014640.html

Signed-off-by: Peter Seiderer <ps.report@gmx.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
(cherry picked from commit 8fcdd2023ee6bc2efd96a3b43fec103f2afa0e2f)
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/mpd: fix build of GenParseName

Fix build of GenParseName which has been added in version 0.22:
https://github.com/MusicPlayerDaemon/MPD/commit/fa45a8adfa44f6bc815ae7428770112c15c76d73

Fixes:
 - http://autobuild.buildroot.org/results/871e1362c44e5b68a149e6a5dd3caf99ea0d904a

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/libgeos: fix build failure due to Gcc Bug 68485

Package libgeos manifests Microblaze Gcc Bug 68485 resulting in a build
failure due to an Internal Compiler Error.

As done for other packages in Buildroot work around this Gcc Bug by
setting optimization to -O0 if BR2_TOOLCHAIN_HAS_GCC_BUG_68485=y.

Fixes:
http://autobuild.buildroot.net/results/0da/0daa6b259aea5381fad86d01e6dd026b1c8ad073/

Signed-off-by: Giulio Benetti <giulio.benetti@benettiengineering.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/wireshark: security bump to version 3.4.3

The following vulnerabilities have been fixed:
 - wnpa-sec-2021-01 USB HID dissector memory leak. Bug 17124.
   CVE-2021-22173.
 - wnpa-sec-2021-02 USB HID dissector crash. Bug 17165. CVE-2021-22174.

https://www.wireshark.org/docs/relnotes/wireshark-3.4.3.html

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

support/testing: TestATFAllwinner needs python3 for u-boot

New U-Boot version needs Python 3.x for pylibfdt.

Fixes:
- https://gitlab.com/buildroot.org/buildroot/-/jobs/1019385909

Signed-off-by: Sergey Matyukevich <geomatsi@gmail.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/lcms2: fix static building with tiff

Strangely enough, we have only one build failure on the autobuilder even
if lcms2 never used pkg-config to retrieve static dependencies of tiff
(which also depends on xz since 2016)

Fixes:
 - http://autobuild.buildroot.org/results/07c5ca780bcdbfcd7cad6502345f1553ce17bdc3

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/ply: needs dynamic library

ply builds and installs a library. Some objects that go in that library
are tagged with a 'section' attribute (excerpt):

    __attribute__((section("providers")))

Later on, it references the bounds of that section, with the canonical
__start and __stop markers, which will eventually be created by the
linker:

    extern struct provider __start_providers;
    extern struct provider __stop_providers;

Sections only exists in an ELF file, and a static library id not an ELF.
So, when creating a static library, the markers are not created. Thus,
when linking the final executable, the link fails because of missing
symbols:

    .../powerpc-buildroot-linux-uclibc/bin/ld: ../libply/.libs/libply.a(libply_la-provider.o): in function `provider_get':
    provider.c:(.text+0xe): undefined reference to `__start_providers'
    .../powerpc-buildroot-linux-uclibc/bin/ld: provider.c:(.text+0x12): undefined reference to `__stop_providers'
    .../powerpc-buildroot-linux-uclibc/bin/ld: provider.c:(.text+0x2a): undefined reference to `__start_providers'
    .../powerpc-buildroot-linux-uclibc/bin/ld: provider.c:(.text+0x32): undefined reference to `__stop_providers'

So, conceptually, ply can not build in static-only.

Fixes:
 - http://autobuild.buildroot.net/results/3a586241d37614b644ff6c4674ae28df2b22fdf8

Signed-off-by: Andreas Klinger <ak@it-klinger.de>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>

package/fetchmail: set FETCHMAIL_CPE_ID_VENDOR

cpe:2.3:a:fetchmail:fetchmail is a valid CPE identifier for this
package:

  https://nvd.nist.gov/products/cpe/search/results?namingFormat=2.3&keyword=cpe%3A2.3%3Aa%3Afetchmail%3Afetchmail

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/fail2ban: add FAIL2BAN_CPE_ID_VENDOR

cpe:2.3:a:fail2ban:fail2ban is a valid CPE identifier for this package:

  https://nvd.nist.gov/products/cpe/search/results?namingFormat=2.3&keyword=cpe%3A2.3%3Aa%3Afail2ban%3Afail2ban

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/file: set FILE_CPE_ID_VALID

cpe:2.3:a:file_project:file is a valid CPE identifier for this package:

  https://nvd.nist.gov/products/cpe/search/results?namingFormat=2.3&keyword=cpe%3A2.3%3Aa%3Afile_project%3Afile

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/x11vnc: set X11VNC_CPE_ID_VALID

cpe:2.3:a:x11vnc_project:x11vnc is a valid CPE identifier for this package:

  https://nvd.nist.gov/products/cpe/search/results?namingFormat=2.3&keyword=cpe%3A2.3%3Aa%3Ax11vnc_project%3Ax11vnc

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/axel: set AXEL_CPE_ID_VALID

cpe:2.3:a:axel_project:axel is a valid CPE identifier for this package:

  https://nvd.nist.gov/products/cpe/search/results?namingFormat=2.3&keyword=cpe%3A2.3%3Aa%3Aaxel_project%3Aaxel

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/postgresql: security bump version to 13.2

Release notes:
https://www.postgresql.org/about/news/postgresql-132-126-1111-1016-9621-and-9525-released-2165/

Fixes CVE-2021-3393 & CVE-2021-20229.

Updated license hash due to copyright year bump:
https://git.postgresql.org/gitweb/?p=postgresql.git;a=commitdiff;h=c09f6882d6f78bde26fcc1e1a3da11c274de596a

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

docs/website: mention Bootlin Buildroot training courses

Our documentation page already mentions the open-source and freely
available training materials from Bootlin on Buildroot.

It turns out that we now have online training courses accessible to
public registration, which makes them accessible to a wider
audience. It probably makes sense to mention them alongside the
training materials.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>