bash has a concept of "loadables", which are "plugins" that can be
loaded at runtime by bash to add new builtin. For example:
# type whoami
whoami is hashed (/usr/bin/whoami)
# whoami
root
# enable -f /usr/lib/bash/whoami whoami
# type whoami
whoami is a shell builtin
# whoami
root
# enable -d whoami
# type whoami
whoami is hashed (/usr/bin/whoami)
# whoami
root
bash comes with a set of example loadables, installed in
/usr/lib/bash/. They take 312 KB on ARM32, and are by default not
used, and provide builtins that are for the most part already
available as external commands in Busybox/coreutils:
Makefile.inc finfo mkfifo realpath sync
accept head mktemp rm tee
basename id mypid rmdir truefalse
csv ln pathchk seq tty
cut loadables.h print setpgid uname
dirname logname printenv sleep unlink
fdflags mkdir push strftime whoami
So instead of having them unconditionally installed, add an option to
enable/disable their installation (their build apparently cannot be
disabled via a configure option).
Normally, we try to keep backward compatibility by preserving the
existing behavior. In this case, this would have meant making this
option "default y". But this also breaks our principle of "being
minimal by default", and in this case, it feels preferable to be
"minimal by default" than preserving existing behavior.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Nicolas Cavallari <nicolas.cavallari@green-communications.fr>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
SHOBJ_STATUS=unsupported was added in commit
4a2af11cba to work around a limitation
of the configure script that forgot to set this variable in
static-linking configurations.
It turns out that this issue has been fixed upstream as of bash 5.0:
https://git.savannah.gnu.org/cgit/bash.git/diff/configure.ac?id=d233b485e83c3a784b803fb894280773f16f2deb
(see hunk @@ -1151,6 +1179,9 @@)
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
We want bash to be installed as /bin/bash. For ages, Buildroot has
been doing this by overriding exec_prefix at install time. First of
all, it would be preferred to do this at configure time. But also,
overriding exec_prefix not only changes where "bash" goes, but also
where the pkgconfig file goes. Due to this, bash.pc goes into
/lib/pkgconfig/, and doesn't get removed by target-finalize.
Since all we want is to have 'bash' as /bin/bash, simply pass
--bindir=/bin at configure time. This allows to use the default target
installation logic for autotools-package. We keep a post-install
target hook to remove /bin/bashbug.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Nicolas Cavallari <nicolas.cavallari@green-communications.fr>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
This patch adds CPE ID information for a significant number of
packages.
Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Arnout reported an issue with the upcoming top-level parallel build.
The file /etc/shells can be modified by several packages (shell providers)
Signed-off-by: Romain Naour <romain.naour@smile.fr>
Cc: Arnout Vandecappelle <arnout@mind.be>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Commit 4d279697af missed one comment
from Yann E. Morin about the double-dollar.
Signed-off-by: Romain Naour <romain.naour@smile.fr>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Commit 02e5cef1 (system: add option to choose what /bin/sh points to)
removed the symlink creation, but forgot to remove the corresponding
comment, which is now more puzzling than amusing...
Remove the comment.
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Romain Naour <romain.naour@gmail.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
When bash is selected, /bin/bash is not added to /etc/shells
(see man shells). So, login tools like dropbear reject the ssh
connexions for users using bash as shell in /etc/passwd.
buildroot authpriv.warn dropbear[853]: User 'kubu' has invalid shell, rejected
Reported-by: Jeremy Rosen <jeremy.rosen@smile.fr>
Signed-off-by: Romain Naour <romain.naour@smile.fr>
Cc: "Yann E. MORIN" <yann.morin.1998@free.fr>
Acked-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
In addition, remove upstreamed patches and add a hash for the license
file.
Signed-off-by: Adam Duskett <aduskett@gmail.com>
[Thomas: indicate pgp signature file address in hash file, as
suggested by Baruch.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
When building bash statically, configure does not initialize the
SHOBJ_STATUS variable. The SHOBJ_STATUS make variable in
examples/loadables/Makefile.in must resolve to either 'supported' or
'unsupported'. Otherwise, the install-$(SHOBJ_STATUS) target does not
exist.
This leads to the following build failure:
make[2]: Entering directory '.../output/build/bash-4.4/examples/loadables'
make[2]: *** No rule to make target 'install-', needed by 'install'. Stop.
make[2]: Leaving directory '.../output/build/bash-4.4/examples/loadables'
Makefile:793: recipe for target 'install' failed
with the following defconfig:
BR2_arm=y
BR2_STATIC_LIBS=y
BR2_PACKAGE_BUSYBOX_SHOW_OTHERS=y
BR2_PACKAGE_BASH=y
Note that this doesn't terminate the build so it's not detected as an
error by the autobuilders.
Set SHOBJ_STATUS in the configure environment as a workaround.
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Bash's malloc relies on sbrk which is implemented as a fail-only stub in
musl. Presently, it is disabled when configured for static
libs. Instead, default to using libc malloc.
Fixes:
# bash
bash: xmalloc: locale.c:81: cannot allocate 18 bytes (0 bytes allocated)
Signed-off-by: Daniel Sabogal <dsabogalcc@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
We want to use SPDX identifier for license string as much as possible.
SPDX short identifier for GPLv3/GPLv3+ is GPL-3.0/GPL-3.0+.
This change is done using following command.
find . -name "*.mk" | xargs sed -ri '/LICENSE( )?[\+:]?=/s/\<GPLv3\>/GPL-3.0/g'
Signed-off-by: Rahul Bedarkar <rahulbedarkar89@gmail.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
They don't seem to be causing any issues after bumping to the 4.3.x
series and several tests.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Tested-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
When Bash attempts to find the current working directory, it uses a C
library call `getcwd` to resolve it. When cross-compiling, the
configuration process cannot determine if the target system's C library
can support an "unfixed" path length. Therefore, Bash will fallback to a
size of `PATH_MAX` for determining the current working directory. When
using OverlayFS (and possible other file systems), this becomes an issue
since file paths can commonly exceed standard `PATH_MAX` length. This
typically results in the following error appearing:
error retrieving current directory: [...]
Common C library `getcwd` calls can default to a higher limit (usually
the system's page size). The current configurable C libraries (as of at
least 2015.08) support a zero (0) size buffer length. Most use the
system's page size; musl, being an exception, which defaults to
`PATH_MAX` (as Bash was doing). Since these C libraries support
allocating buffer space with a zero (0) provided size, the following
configuration change allows Bash to support getting a larger-length'ed
working directory on target's that support it.
Signed-off-by: James Knight <james.knight@rockwellcollins.com>
Acked-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Since a while, the semantic of BR2_PREFER_STATIC_LIB has been changed
from "prefer static libraries when possible" to "use only static
libraries". The former semantic didn't make much sense, since the user
had absolutely no control/idea of which package would use static
libraries, and which packages would not. Therefore, for quite some
time, we have been starting to enforce that BR2_PREFER_STATIC_LIB
should really build everything with static libraries.
As a consequence, this patch renames BR2_PREFER_STATIC_LIB to
BR2_STATIC_LIBS, and adjust the Config.in option accordingly.
This also helps preparing the addition of other options to select
shared, shared+static or just static.
Note that we have verified that this commit can be reproduced by
simply doing a global rename of BR2_PREFER_STATIC_LIB to
BR2_STATIC_LIBS plus adding BR2_PREFER_STATIC_LIB to Config.in.legacy.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Yay!
The release tarball allows us to drop sizable in-tree patches, so switch
to it.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
bash fails to link for static builds with uClibc toolchains due to
getenv redefinitions. This is caused because bash is unable to check if
getenv is already defined when cross-compiling, so it defaults to 'yes':
configure:14438: WARNING: cannot check getenv redefinition if cross
compiling -- defaulting to yes
We can avoid this redefinition by passing bash_cv_getenv_redef=no to the
configure script.
Related:
http://lists.gnu.org/archive/html/bug-bash/2012-03/msg00052.html
Fixes:
http://autobuild.buildroot.net/results/a20/a2007e6dbcfe53e7cd837ae642869ee26376826a/
Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Reviewed-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Not all our shells do install a pointer to /bin/sh. Besides, between
those that do and multiple ones are enabled, the last one to install
wins the the symlink.
Add a new config choice in the system sub-menu that allows the user to
explicitly select the shell to provide /bin/sh. If busybox is not
enabled, default to using dash, a POSIX shell.
Remove the symlink creation from bash.mk at the same time.
Note: for every shell, we select them, except busybox, on which we
depend, on the assumption that we do not want to force busybox in case
the user decided not to enable it.
[Peter: remove redundant !BR2_PACKAGE_BUSYBOX dependency]
Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Cc: Gustavo Zacarias <gustavo@zacarias.com.ar>
Cc: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Remove the old symlink backup since we don't do uninstalls now.
Make the busybox dependency conditional shorter.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
To be consistent with the recent change of FOO_MAKE_OPT into FOO_MAKE_OPTS,
make the same change for FOO_CONF_OPT.
Sed command used:
find * -type f | xargs sed -i 's#_CONF_OPT\>#&S#g'
Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>
Reviewed-by: "Yann E. MORIN" <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Always prefix the readline directory otherwise autodetection may pick up
the host readline when headers are available and target == host (or are
link-compatible). Fixes:
http://autobuild.buildroot.org/results/44a/44a962623be757143a3d6d37fc6cd14964094f79/
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Switch to external readline support to potentially reduce target size.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
There are some bash features that can not be detected automatically during
configure stage while cross-compiling. This commit forces them on.
Signed-off-by: Arkady Gilinsky <arcadyg@gmail.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Thanks to the pkgparentdir and pkgname functions, we can rewrite the
AUTOTARGETS macro in a way that avoids the need for each package to
repeat its name and the directory in which it is present.
[Peter: pkgdir->pkgparentdir]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Make sure bash builds after busybox when it's enabled to override
busybox's /bin/sh symlink with bash.
Signed-off-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
The cleanup of $(TARGET_DIR)/usr/share/man, $(TARGET_DIR)/usr/man,
$(TARGET_DIR)/usr/share/info, $(TARGET_DIR)/usr/info,
$(TARGET_DIR)/usr/share/doc and $(TARGET_DIR)/usr/doc is already done
globally in the main Makefile. Therefore, there's no need to handle
that on a per-package basis.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Patch bash32-011 modifies the configure.in script. Therefor we need to
run autoconf before launching the configure script. Otherwise, once
the configure script called, the makefile's dependencies over
configure.in will relaunch autoconf without options/environnement
variables we usually set when calling configure.
[Peter: use our own host-autoconf instead of what's on the host]
Signed-off-by: Lionel Landwerlin <llandwerlin@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
So "/opt/whatever_toolchain/gcc --sysroot $(STAGING_DIR)" isn't
stripped to "/opt/whatever_toolchain/gcc", making gcc to look at
$(ORIGINAL_TOOLCHAIN_DIRECTORY)/usr/include instead of
$(STAGING_DIR)/usr/include. For more CC is already filled when
running the configure script.
This problem should only be visible with an external toolchain.
Signed-off-by: Lionel Landwerlin <llandwerlin@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
We have been passing -q to ./configure when using 'make -s' for
packages using Makefile.autotools.in for some time. Do the same
for packages using autotools, but not using the
Makefile.autotools.in infrastructure, taking care to not do it
for packages with hand written configure scripts.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
A C library will have been built by the toolchain makefiles, so there is no
need for packages to explicitly depend on uclibc.
Signed-off-by: Will Newton <will.newton@gmail.com>
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
-Force some autoconf options so that job control
is turned on.
(Otherwise these checks fail when cross compiling.
And when job control is disabled, bash can't be
the default shell)
