Adding -f*-prefix-map to LDFLAGS caused the following issue:
QA Issue: netsnmp-agent.pc failed sanity test (tmpdir)
Fix by filtering out -f*-prefix-map from *.pc files.
[YOCTO #14481]
Signed-off-by: Tony Battersby <tonyb@cybernetics.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
(cherry picked from commit 5e042ac207)
Signed-off-by: Armin Kuster <akuster808@gmail.com>
Due to the sed commands in do_install_append() that removed
${STAGING_DIR_HOST} and it being empty when building for native, it was
impossible to add support for building this as native using a bbappend.
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
(cherry picked from commit 74d58bc6e8)
Signed-off-by: Armin Kuster <akuster808@gmail.com>
* Remove the explicit dependency on libnl as the libnl PACKAGECONFIG
depends on it as necessary.
* Add a PACKAGECONFIG for systemd to replace modifying EXTRA_OECONF
directly.
* Sort the PACKAGECONFIGs.
* Some whitespace clean up.
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
(cherry picked from commit 411c981ef0)
Signed-off-by: Armin Kuster <akuster808@gmail.com>
The openssl already added in DEPENDS and the openssl related library
will be in recipe-sysroot. So it's meanlingless to add the configure
option "--with-openssl=${STAGING_EXECPREFIXDIR}" as the below help message.
$ cd /prj/net-snmp-5.9/
$ ./configure --help
[snip]
--with-openssl=PATH Look for openssl in PATH/lib,
or PATH may be "internal" to build with
minimal copied OpenSSL code for USM only
[snip]
And there is also a side effect after add the above openssl configuration
as the build path is added for NSC_LDFLAGS in /usr/bin/net-snmp-config.
NSC_LDFLAGS="-L/prj/tmp/work/corei7-64-wrs-linux/net-snmp/5.9-r0/recipe-sysroot/usr/lib64 -Wl,-O1 -Wl,--hash-style=gnu -Wl,--as-needed -Wl,-z,relro,-z,now"
To improve reproducibility for netsnmp as below.
$ sed -i -e 's@${STAGING_DIR_HOST}@@g' -i ${D}${bindir}/net-snmp-config
The NSC_LDFLAGS in net-snmp-config will be changed to below:
NSC_LDFLAGS="-L/usr/lib64 -Wl,-O1 -Wl,--hash-style=gnu -Wl,--as-needed -Wl,-z,relro,-z,now"
But it will result in other packages which depend on net-snmp such as
corosync, quagga and etc uses the build host library and introduce
below do_configure error.
ERROR: QA Issue: This autoconf log indicates errors, it looked at host include and/or library paths while determining system capabilities.
Rerun configure task after fixing this. [configure-unsafe]
So remove the useless configuration to fix the issue.
Signed-off-by: Mingli Yu <mingli.yu@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
New autconf detects that NSC_LDFLAGS are hardcoded to use -L/usr/lib
therefore edit these variables during build so that they have
cross-compile friendly values when net-snmp-config is used during build
of dependent packages
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Add a patch to fix the gap between 32bit and 64bit system when
the configure option "--with-openssl=${STAGING_EXECPREFIXDIR}"
passed in.
Signed-off-by: Mingli Yu <mingli.yu@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Support for smux is always enabled by the recipe, but it can be a
security risk since it makes the snmpd daemon listen on TCP port 199.
This makes it contrallable via PACKAGECONFIG, so that it can be easily
disabled from the distro or local config. The mechanism makes it easy
to add control for other MIB modules via PACKAGECONFIG later if need
be.
For compatibility smux is added to the default PACKAGECONFIG, so there
is no change in the default build configuration.
Signed-off-by: Diego Santa Cruz <Diego.SantaCruz@spinetix.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Source: net-snmp.org
MR: 104509
Type: Security Fix
Disposition: Backport from https://github.com/net-snmp/net-snmp/commit/5f881d3bf24599b90d67a45cae7a3eb099cd71c9
ChangeID: 206d822029d48d904864f23fd1b1af69dffc26c8
Description:
Fixes CVE-2019-20892 which affect net-snmp <= 5.8pre1
Had to fix up some file do to later code restructioning.
"int refcnt;" addition was done in include/net-snmp/library/snmpusm.h
Signed-off-by: Armin Kuster <akuster@mvista.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Refreshed patches for 5.8 due to the following:
ERROR: net-snmp-5.8-r0 do_patch: Command Error: 'quilt --quiltrc .../net-snmp/5.8-r0/recipe-sysroot-native/etc/quiltrc push' exited with 0 Output:
Applying patch 0001-Add-pkg-config-support-for-building-applications-and.patch
patching file configure
...
Hunk #1 succeeded at 32248 with fuzz 2 (offset 1826 lines).
Hunk #2 FAILED at 31447.
1 out of 2 hunks FAILED -- rejects in file configure
...
Patch 0001-Add-pkg-config-support-for-building-applications-and.patch does not apply (enforce with -f)
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
net-snmp/net-snmp-config.h:
- encodes type sizes
- encodes pathing into the libdir
net-snmp-config:
- encodes build configuration data and lib pathing.
Signed-off-by: Jeremy A. Puhlman <jpuhlman@mvista.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
* Remove perl-lib since it had been removed by oe-core:
commit 68552c353255188de3d5b42135360a30e7eac535
Author: Alexander Kanavin <alex.kanavin@gmail.com>
Date: Sun Dec 2 12:46:37 2018 +0100
perl: remove the previous version of the recipe
Now the files are in perl pacakge.
* Fix perl paths when perl is enabled.
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Replace source zip ball with tarball for net-snmp to avoid zip bomb issue.
Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
A bit of an unholy mixture of MIT, BSD 3-clause,
and too old to really know BSD-style,
with a wide variety of copyright holders.
I'm open to better suggestions on how to handle this.
Signed-off-by: Douglas Royds <douglas.royds@taitradio.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
| scapi.c: In function 'sc_encrypt':
| scapi.c:1256:5: error: 'pad_size' undeclared (first use in this function); did you mean 'dysize'?
| pad_size = pai->pad_size;
| ^~~~~~~~
| dysize
pad_size is defined only without --disable-des
[snip]
int pad, plast, pad_size = 0;
but used when disable-des
[snip]
QUITFUN(SNMPERR_GENERR, sc_encrypt_quit);
}
pad_size = pai->pad_size;
memset(my_iv, 0, sizeof(my_iv));
if (USM_CREATE_USER_PRIV_DES == (pai->type & USM_PRIV_MASK_ALG)) {
/*
fix by move it into #ifndef NETSNMP_DISABLE_DES
Signed-off-by: Changqing Li <changqing.li@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Inherit ptest for net-snmp to create ${PN}-ptest. Update run-ptest as
well to avoid only could be run in the same directory.
Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Readability.
The existing patterns allowed each pattern to be matched multiple times (with no
intevening spaces), but the "g" modifier achieves this anyway.
Signed-off-by: Douglas Royds <douglas.royds@taitradio.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Don't check for /etc/printcap on the build machine when cross-compiling.
Use AC_CHECK_FILE to set the cached variable ac_cv_file__etc_printcap instead.
When cross-compiling, this variable should be set in the environment to "yes" or
"no" as appropriate for the target platform.
I have taken the simple expedient of setting ac_cv_file__etc_printcap=no.
If this proves to be a problem, we can easily add a new variable, HAS_PRINTCAP.
Upstream-Status: Submitted https://sourceforge.net/p/net-snmp/patches/1385/
Signed-off-by: Douglas Royds <douglas.royds@taitradio.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
PKG_CONFIG_PATH and PKG_CONFIG_LIBDIR point into the net-snmp recipe-sysroot.
Careful not to trim trailing quotes from the CFLAGS
Signed-off-by: Douglas Royds <douglas.royds@taitradio.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
By default `net-snmp-libs` contains all compiled libs.
This commit splits `net-snmp-libs` into subpackages for each library.
This allows for smaller resulting image due to finer packaging.
Signed-off-by: Alexander Filippov <a.filippov@yadro.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
If "/usr/local/ssl/include" directory exists on the host machine,
net-snmp will also search the host openssl headers:
x86_64-wrs-linux-libtool: compile: x86_64-wrs-linux-gcc ...
-I/usr/local/ssl/include
Fix this by selecting the proper sysroot headers using
--with-openssl=${STAGING_EXECPREFIXDIR}
Signed-off-by: Changqing Li <changqing.li@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
WARNING: net-snmp-5.7.3-r0 do_patch:
Some of the context lines in patches were ignored. This can lead to incorrectly applied patches.
The context lines in the patches can be updated with devtool:
devtool modify <recipe>
devtool finish --force-patch-refresh <recipe> <layer_path>
Then the updated patches and the source tree (in devtool's workspace)
should be reviewed to make sure the patches apply in the correct place
and don't introduce duplicate lines (which can, and does happen
when some of the context is ignored). Further information:
http://lists.openembedded.org/pipermail/openembedded-core/2018-March/148675.htmlhttps://bugzilla.yoctoproject.org/show_bug.cgi?id=10450
Details:
Applying patch 0001-BUG-a2584-Fix-snmptrap-to-use-clientaddr-from-snmp.c.patch
patching file snmplib/transports/snmpUDPIPv6Domain.c
Hunk #1 succeeded at 286 with fuzz 2 (offset 30 lines).
Now at patch 0001-BUG-a2584-Fix-snmptrap-to-use-clientaddr-from-snmp.c.patch
Signed-off-by: Armin Kuster <akuster808@gmail.com>
options which involve paths into workspace are usually filtered out
with the sed command, however currently it does half the job and the
resulting config file gets partial replacements leaving the compiler
options with bad syntax e.g.
-fdebug-prefix-map option is left with -fdebug-prefix-map=
which is not correct syntax, the effect of this is seen in other
recipes which then invoke net-snmp-config script and add the
flags obtained from this script into its own configure scripts
and then tries to execute tests, these tests fail because of
bad compiler options e.g. keepalived where configure tests fails like
| configure: error: *** incorrect CFLAGS from net-snmp-config
this is because of wrong compiler options that it got from
recipe-sysroot/usr/bin/crossscripts/net-snmp-config
This patch tries to be specific about which options should
be skimmed and removes the options completely
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Armin Kuster <akuster808@gmail.com>
Signed-off-by: Joe MacDonald <joe_macdonald@mentor.com>
The previous path value to -fdebug-prefix-map
is null which may result in other package such as
quagga do_config error as below:
====================================================
add DISTRO_FEATURES_append = " snmp" to conf/local.conf
test@buildserver@ bitbake quagga
| checking for i586-poky-linux-net-snmp-config... no
| checking for net-snmp-config... $Prj/tmp/work/i586-poky-linux/quagga/1.2.1-r0/recipe-sysroot/usr/bin/crossscripts/net-snmp-config
| checking whether we can link to Net-SNMP... no
| configure: error: --enable-snmp given but not usable
| NOTE: The following config.log files may provide further information.
| NOTE: $Prj/tmp/work/i586-poky-linux/quagga/1.2.1-r0/build/config.log
| ERROR: configure failed
| WARNING: $Prj/tmp/work/i586-poky-linux/quagga/1.2.1-r0/temp/run.do_configure.80493:1 exit 1 from 'exit 1'
| ERROR: Function failed: do_configure (log file is located at $Prj/tmp/work/i586-poky-linux/quagga/1.2.1-r0/temp/log.do_configure.80493)
====================================================
Signed-off-by: Yu Mingli <mingli.yu@windriver.com>
Signed-off-by: Armin Kuster <akuster808@gmail.com>
Signed-off-by: Joe MacDonald <joe_macdonald@mentor.com>
net-snmp enables 3des support by default and fails to build with distro
feature openssl-no-weak-ciphers:
| ../../net-snmp-5.7.3/snmplib/scapi.c:82:25: fatal error: openssl/des.h: No such file or directory
| #include <openssl/des.h>
To fix the issue:
* add a patch to include des.h only if it's found in openssl
* disable des when openssl-no-weak-ciphers is enabled
Signed-off-by: Jackie Huang <jackie.huang@windriver.com>
Signed-off-by: Joe MacDonald <joe_macdonald@mentor.com>
Signed-off-by: Armin Kuster <akuster808@gmail.com>
Signed-off-by: Joe MacDonald <joe_macdonald@mentor.com>
The recipe for net-snmp has snmpd and snmptrapd in seperate packages, so one or the other
or both could be installed. In a common case where only snmpd is installed, the startup
script will fail to run because the snmptrapd executable does not exist.
This patch simply qualifies the test by first checking to see if the executable is to
be used.
-Bill
Signed-off-by: Bill Randle <bill.randle@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Joe MacDonald <joe_macdonald@mentor.com>
Store the incremented engineBoots value on SIGHUP.
And don't reset engineBoots to 1 when oldEngineIDLength is 0.
For the first run, the oldEngineIDLength is 0.
When we say first run of the daemon, we talk about the
first run ever on the machine, not only first run of every boot.
Signed-off-by: Marian Florea <marian.florea@windriver.com>
Reviewed-by: Wenkuan Wang <Wenkuan.Wang@windriver.com>
Reviewed-by: Zhaolong Zhang <Zhaolong.Zhang@windriver.com>
Signed-off-by: Li Zhou <li.zhou@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Joe MacDonald <joe_macdonald@mentor.com>
The patch solves two issues:
1. Supported cross compile for the perl embedded and perl modules.
2. Solved runtime depend issue.
Signed-off-by: Wenlin Kang <wenlin.kang@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Joe MacDonald <joe_macdonald@mentor.com>
Backport a succeeding commit from net-snmp upstream to fix the issue
introduced by commit
<BUG#a2584: Fix snmptrap to use clientaddr from snmp.conf>.
The missing return will cause crash when binding to a non-exist IPv6
address.
Signed-off-by: Li Zhou <li.zhou@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Joe MacDonald <joe_macdonald@mentor.com>
Under IPv6 IP-multihomed environment, the socket does not bind to the
clientaddr indicated in snmp.conf when sending snmptrap and it might
choose a random one.
Backport the patch from net-snmp upstream to fix it.
Signed-off-by: Li Zhou <li.zhou@windriver.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Joe MacDonald <joe_macdonald@mentor.com>
Tony Battersby
Peter Kjellerstedt
zhengruoqin
Mingli Yu
Khem Raj
Martin Jansa
Diego Santa Cruz
Ovidiu Panait
Mingde (Matthew) Zeng
Armin Kuster
Patrick Williams
Jeremy Puhlman
Robert Yang
Kai Kang
Hongxu Jia
douglas.royds
Changqing Li
Alexander Filippov
Andrej Valek
Yi Zhao
Armin Kuster
Slater, Joseph
Jackie Huang
Bill Randle
Li Zhou
Wenlin Kang