We may set the PATH variable such as in /etc/environment for ssh session,
so make PATH retain the previous existing value to fix the gap.
(From OE-Core rev: 4340cd7da25c69424915f29c2bb9531fc33617e6)
Signed-off-by: Mingli Yu <mingli.yu@windriver.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We encountered a newgrp regression in shadow 4.17.3:
root@qemux86-64:~# groupadd g1
root@qemux86-64:~# useradd t1
root@qemux86-64:~# gpasswd g1
Changing the password for group g1
New Password:
Re-enter new password:
root@qemux86-64:~# sudo -u t1 newgrp g1
Password:
Invalid password.
root@qemux86-64:~#
In versions prior to shadow 4.17.3, shadow used an internal
implementation to support shadow group because it could not correctly
detect whether glibc supports shadow group in a cross-compilation
environment. In 4.17.3, it can correctly check whether glibc supports
shadow group even in a cross-compilation environment[1]. If supported,
shadow will use it instead of its own internal implementation.
Shadow group support in glibc requires adding a gshadow entry in
nsswitch.conf.
After the patch:
root@qemux86-64:~# groupadd g1
root@qemux86-64:~# useradd t1
root@qemux86-64:~# gpasswd g1
Changing the password for group g1
New Password:
Re-enter new password:
root@qemux86-64:~# sudo -u t1 newgrp g1
Password:
t1@qemux86-64:/home/root$
[1] da6b9cff02
(From OE-Core rev: 0cb122f17cf264ef904880351db1c0bb325fe5a6)
Signed-off-by: Yi Zhao <yi.zhao@windriver.com>
Signed-off-by: Mathieu Dubois-Briand <mathieu.dubois-briand@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Remove /bin/sh from bash RPROVIDES as this has a side-effect which
confuses rpm package manager when also busybox provides /bin/sh and
base-files depend on /bin/sh . The problem is broken down below.
First, bash depends on base-files and bash pkg_postinst must run
after base-files was installed, because it requires /etc/shells
provided by base-files to be in place.
Second, base-files depends on /bin/sh, which is provided by either
bash or busybox in this case. This is the actual problem here, if
bash is selected as /bin/sh provider, then there is cyclic dependency
between bash and base-files, and that confuses dnf which may install
the packages in the wrong order, bash first and base-files second .
To make this worse, if busybox is also /bin/sh provider, it can and
does happen that some systems pick busybox as the /bin/sh provider,
while others pick bash as the /bin/sh provider, and that cyclic
dependency does not always appear.
Attempt to break this dependency, remove pre-inst script from the
base-files recipe, which removes its dependency on /bin/sh and
allows it to be installed very early, and always before bash.
(From OE-Core rev: e71b64a9b22c7db316e92e78a4bce8b9f994a4ae)
Signed-off-by: Marek Vasut <marex@denx.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* Drop VOLATILE_TMP_DIR, use FILESYSTEM_PERMS_TABLES instead. By default,
FILESYSTEM_PERMS_TABLES ?= "files/fs-perms.txt \
files/fs-perms-volatile-log.txt \
files/fs-perms-volatile-tmp.txt"
it contains 'files/fs-perms-volatile-tmp.txt', which means volatile tmp
is enabled. User can disable volatile tmp by remove
'files/fs-perms-volatile-tmp.txt' from FILESYSTEM_PERMS_TABLES.
* If volatile tmp is disabled, both /tmp and /var/tmp are persistent
(From OE-Core rev: 8d1ae67b89c45f78162e070228086c7ef88c3264)
Signed-off-by: Changqing Li <changqing.li@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Drop VOLATILE_LOG_DIR, use FILESYSTEM_PERMS_TABLES instead. By default,
it contains 'files/fs-perms-volatile-log.txt', which means volatile log
is enabled. User can disable volatile log by remove
'files/fs-perms-volatile-log.txt' from FILESYSTEM_PERMS_TABLES.
(From OE-Core rev: 91128c6517066715f2afe6b46aa3206c7cf3653e)
Signed-off-by: Changqing Li <changqing.li@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Where recipes use S = ${WORKDIR}, change them to set UNPACKDIR to
a subdir of WORKDIR and make S point at this instead.
I've chosen not to force S into any standard UNPACKDIR we may pick in
future just so the S = UNPACKDIR case is clearly visible by the
directory naming under WORKDIR as that should aid usability.
(From OE-Core rev: d9328e3b0b062f0621de3d114584b44f384a1c02)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Where recipes use S = ${WORKDIR}, ensure they are referencing ${S} correctly
to access files as soon we want to stop doing this in WORKDIR at which point
they would break unless corrected.
(From OE-Core rev: f25dd633fffe6560f191526d1869e657e129bad9)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Add nss-resolve plugin to the glibc Name Service Switch (NSS) with
systemd-resolved DISTRO_FEATURE so that systemd-resolved is used in DNS
name resolution.
This enables the resolution of Multicast DNS and Link-Local Multicast
Name Resolution names, depending on the selected options.
(From OE-Core rev: 81da1d6eecee9fd036121298abba6fdcffc3969d)
Signed-off-by: Eero Aaltonen <eero.aaltonen@vaisala.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
With a profile.d configuration in place that sets the EDITOR variable,
the automatic terminal 'resize' logic would not trigger. Which then
would possibly lead to a 80x24 fallback on the debug serial console.
This can simply be avoided by setting a flag variable when the shell
is first opened, then processing all profile.d includes, trigger the
'resize' depending on the flag and shell-level and finally only set
EDITOR to some default if it is still unset.
(From OE-Core rev: dcdb30c83eb77fb2d5ea04f9b7fd7371da633a34)
Signed-off-by: Johannes Schneider <johannes.schneider@leica-geosystems.com>
Signed-off-by: Luca Ceresoli <luca.ceresoli@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We've been removing PR values from recipes at upgrade time for a while. In general
anyone maintaining a binary distro would end up having to curate these themselves
so the values in OE-Core aren't really that useful anymore. In many ways it makes
sense to clear out the remaining ones (which are mostly for 'config' recipes that
are unlikely to increase in PV) and leave a clean slate for anyone implementing
a binary distro config.
References are left in meta-selftest since the tests there do involve them and
their removal upon upgrade.
(From OE-Core rev: d4c346e8ab8f3cae25d1b01c7331ed9f6d4f96ef)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
An automated conversion using scripts/contrib/convert-spdx-licenses.py to
convert to use the standard SPDX license identifiers. Two recipes in meta-selftest
were not converted as they're that way specifically for testing. A change in
linux-firmware was also skipped and may need a more manual tweak.
(From OE-Core rev: ceda3238cdbf1beb216ae9ddb242470d5dfc25e0)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This is the result of automated script conversion:
scripts/contrib/convert-overrides.py <oe-core directory>
converting the metadata to use ":" as the override character instead of "_".
(From OE-Core rev: 42344347be29f0997cc2f7636d9603b1fe1875ae)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Using an editor or any kind of command line that wraps beyond the
column width of the session on a serial port is quite problematic
unless you are using an 80x24 session.
The original /etc/profile tried to use the resize binary if it was
available. The problem is that you only get the resize binary if
xterm, or busybox is installed.
This updated /etc/profile will add a resize function available to the
shell when no xterm or busybox resize binary is found. More care is
taken in this new version to test that terminal is interactive. The
EDITOR and SHLVL environment variables are checked to prevent resize
from running necessarily.
The function definitions are not indented intentionally to keep them
to the 80 column width.
(From OE-Core rev: 3743892996172c8595a1cbe884c4a0e6ef50dcda)
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Make sure that the (newer) /dev/pts/ptmx is accessible by users. This
is useful e.g. when running containers which symlink /dev/ptmx to
/dev/pts/ptmx on start. The default mode (000) does not allow to
create ptys inside the container.
Using 666 when symlinking /dev/ptmx is also recommended by the kernel
documentation when /dev/ptmx is symlinked:
https://www.kernel.org/doc/Documentation/filesystems/devpts.txt
Also buildroot uses ptmxmode=0666. The patch introducing the change
explains related use cases why this is necessary a bit more in depth:
8196b299ba (diff-2d4604b9e565eb19fa52ce31f282f06c)
(From OE-Core rev: c999bc5ddd9beb5274f77b885e2ac71205e42266)
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Doesn't seem to be referenced by anything in oe-core any more.
(From OE-Core rev: 526188096bc37af3140c9210ece92ed94c3bdfe0)
Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The function do_install_basefilesissue is meant to deal with
/etc/issue* files and the following setting could actually be
overridden.
BASEFILESISSUEINSTALL ?= "do_install_basefilesissue"
So move the hostname operations out of this function.
(From OE-Core rev: f17bcc46341db5ee91bb26389ccaebc68c49b97e)
Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
mount can generally figure out the filesystem type from the superblock;
otherwise, /proc/filesystems is a more correct fallback than
/etc/filesystems and presumably always available.
(From OE-Core rev: 4199676383ce50b81c05a4d2b2610c9cdb49342e)
Signed-off-by: Jonas Bonn <jonas@norrbonn.se>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If DISTRO_VERSION is defined to some undefined variable, e.g.,
DISTRO_VERSION = "${FOO_VERSION}", and /bin/sh is dash, then
do_install() would fail with an error such as:
run.do_install.2945:193: run.do_install.2945: Syntax error:
Unterminated quoted string
This was due to unexpanded Python code making it into the shell code,
confusing the shell parser.
Reported-by: srinivasan <srinivasan.rns@gmail.com>
(From OE-Core rev: fdd0b9e26ef11fa287692848ea3add64ba623433)
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Move /etc/hosts to base-files, and also add entry to it according
to hostname setting. This fixes the problem of commands like
`hostname -f' failing due to lack of such entry.
(From OE-Core rev: 31e6d08f20a804fccb958e19045d8f9e9806071a)
Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The kernel mounts /proc and /sys with the mode 555. Fedora explicitly
sets this value in its filesystem setup package. Debian doesn't seem
to set it explictly. Having them be 755 causes permission issues on
upgrades inside a container where the guest does not have the
permission to change the modes of the mount points.
So, just bite the bullet and force them to be 555.
(From OE-Core rev: 7e311b0c7222fa9127a96945c9ded7bee5e40eb3)
Signed-off-by: Dan McGregor <dan.mcgregor@usask.ca>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The current handling of resize is incorrect. Using `resize > /dev/null
2>&1 && resize > /dev/null' will cause the second resize command to not
execute because 'resize > /dev/null 2>&1' will fail for resize utility
from busybox.
What we really should do is just to check whether ${bindir}/resize
is executable and execute it if so. Using '-x' is sufficient.
(From OE-Core rev: 41688279cba3a5afc4fdc65fd245b9bb6ada695e)
Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Since the existence of "command" in itself is not guaranteed, using it
to determine if other executables exist is moot. Instead just run the
executables and let the shell determine if they exist. By piping stderr
to /dev/null we avoid unnecessary warnings in case they do not exist.
(From OE-Core rev: ba8510deb73e07f17be051fa15a91731ec10382f)
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Respect VOLATILE_LOG_DIR variable. In this way, if the user overrides
this variable to be any valid boolean false value, /var/log on the final
image would reside on persistent storage.
[YOCTO #6132]
(From OE-Core rev: fc19114ca409fc3178ae5cbb44db875b12eeedfc)
Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When using "su - myuser" to change from root to a non-privileged user,
"mesg n" from the default .profile fails with "mesg: error: tty device
is not owned by group `tty' or "mesg: cannot open /dev/ttyS0:
Permission denied", depending on whether mesg comes from busybox or
util-linux.
This does not happen during a normal login because permissions on
/dev/tty* get changed while doing that, something that isn't possible
with plain "su -".
As the error can't be avoided and failures of mesg probably aren't
particularly important, now error messages get dumped to /dev/null.
[YOCTO #11127]
(From OE-Core rev: 4511a524f906c97db7c7674ad34e7bc8630649e3)
Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
It is preferred to use `[ "$FOO" ] || ...` instead of
`[ -z "$FOO" ] && ...` as the latter leaves $? set to 1.
(From OE-Core rev: d76c68505c36dbf383a989f3c2458abc765e2c19)
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This avoids the following error when logging in to a host that does
not have the tty command:
-sh: tty: not found
Reported-by: Cristian Ionescu-Idbohrn <cristian.ionescu-idbohrn@axis.com>
(From OE-Core rev: e77cdb761169e404556487ac650dc562000da406)
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We don't want to run resize on non serial consoles. There's
been an earlier attempt (6557787), so this builds upon that.
The problem we're seeing is that if there is text buffered in
the virtual console (like from a desperate user trying to
enter login details), resize will get stuck while calling
ioctl(tty, TCSETAW);
Since serial consoles are named (not just numbered), this
change limits resize's reach even further to run only on
/dev/tty[A-z] (thus avoiding /dev/tty[0-9]).
(From OE-Core rev: 474ef7c95722aa68ee5dfbae2920d7c3d436d717)
Signed-off-by: Daniel Díaz <daniel.diaz@linaro.org>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This is causing a problem in multilib where base-files and lib64/32-base-files
clash because they may have different dates. Also, if the package is coming
from sstate it has an incorrect date anyway.
(From OE-Core rev: 846c86ffc4574feafd89a028e50f7719075633f2)
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
getVar() now defaults to expanding by default, thus remove the True
option from getVar() calls with a regex search and replace.
Search made with the following regex: getVar ?\(( ?[^,()]*), True\)
(From OE-Core rev: 7c552996597faaee2fbee185b250c0ee30ea3b5f)
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Previous attempts to constrain execution of `resize` to only TTYs did
not properly handle situations when `tty` would return the string "not a
tty". The symptom is "/etc/profile: line 34: test: too many arguments".
Fix this by utilizing the exit code of `tty`. Also use `case` instead of
`cut` to eliminate a subshell.
(From OE-Core rev: e67637e4472ff3a1e2801b84ee3d69d4e14b9efc)
Signed-off-by: Richard Tollerton <rich.tollerton@ni.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If no /etc/localtime (or /etc/TZ for uclibc) is found, then the libc
will default to UTC, so setting UTC as a fallback default via the TZ
environment variable is redundant.
Since having the TZ environment variable set causes /etc/localtime
to be ignored, it can cause confusion if /etc/localtime is added
interactively after /etc/profile has been run.
(From OE-Core rev: 98b6420952cbf73ddd1318f36c68d575c330eb71)
Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
tty can return "not a tt" which results in warnings when /etc/profile
is executed.
(From OE-Core rev: eed586dd238efe859442b21b425f04e262bcdb2b)
Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
We don't need/wan't to run resize on an ssh connection. It's useless and
it breaks the Eclipse SSH debug connection. So, we added a check.
YOCTO #9362
(From OE-Core rev: 655778769f50d3aff74d7a436d28ac31b6aebb11)
Signed-off-by: bavery <brian.avery@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Add some safety checks when sourcing files in /etc/profile.d/, in particular:
- source only *.sh files, not every file. This is the practice in use in both
Fedora and Debian/Ubuntu (see
https://help.ubuntu.com/community/EnvironmentVariables#A.2Fetc.2Fprofile.d.2F.2A.sh);
- check the input is actually a file and is readable. This check is especially
important if profile.d is empty, as "*.sh" will get expanded only if
profile.d is not empty. Previously if profile.d was present but empty,
"/etc/profile.d/*" was sourced causing errors on login and breaking stuff, for
example X startup.
(From OE-Core rev: 8961bc4b71723477a3b4a837a1d9c25c1b860b9e)
Signed-off-by: Diego Rondini <diego.ml@zoho.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Having skel in the staging area lets other recipes access these files.
For ex. recipes inheriting useradd, if passing "-m" to USERADD_PARAM,
.bashrc and .profile are put automatically in $HOME, as expected.
(From OE-Core rev: 784f4c80204a09b85b7e83c9bd91cd48f001384f)
Signed-off-by: Ioan-Adrian Ratiu <adrian.ratiu@ni.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The usb filesystem was removed in Linux 3.5.
(From OE-Core rev: c24f0fa366cd96a1d67564dd760b0640fa9f7b13)
Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Commit ea647cd9ee moved the locations
of /run and /var/lock to match the FHS 3 draft specifications.
However, the install doesn't remove the existing directories.
As a result, upgrading a system may result in /run as a softlink
to /var/run and /var/run as a softlink to /run, creating a circular
link.
During pre-install, check for the existence of the old softlinks and
remove them so the new directories can be installed.
(From OE-Core rev: 272805f064848f14306b1ed2648a7760017ee318)
Signed-off-by: Bryan Evenson <bevenson@melinkcorp.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Change the hostname logic the default value is ${MACHINE}, and explain how to
change it.
(From OE-Core rev: f2588c55fb089371f239dde39efe9d63d8614062)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Setting "EDITOR=/bin/vi" breaks on non-busybox systems, as
vim will populate /usr/bin/vi instead, and you get stuff like:
op3:~/poky/meta-builder$ git commit -s
error: cannot run /bin/vi: No such file or directory
error: unable to start editor '/bin/vi'
Please supply the message using either -m or -F option.
op3:~/poky/meta-builder$ which vi
/usr/bin/vi
op3:~/poky/meta-builder$
Since we've already specified a proper path above in the profile,
we've no need to call out where in the path vi lives, and hence
this will work with busybox and a full vim install w/o busybox.
(From OE-Core rev: dcdaa651f3180e4561d7d08f56bfb29f35f2085f)
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Use CONFFILES to mark editable files as such,
/etc/motd
/etc/nsswitch.conf
/etc/profile
If there is no %config micro before the file name in the spec file,
this file will be overwritten after updating package.
This will make our settings lost.
(From OE-Core rev: 2ffc174e9506a00081457145f0fcf3bef5846f4a)
Signed-off-by: Jian Liu <jian.liu@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
