Try to execute both gtk-update-icon-cache and
gtk4-update-icon-cache after checking whether the
commands are available.
This attempts to match what gtk-icon-cache.bbclass is doing.
This fixes running update_gtk_icon_cache during do_rootfs
for an image that contains only GTK4 related packages.
(From OE-Core rev: 1e01ef59a5864f1261e9c0ca76fcccee2eb7e7f3)
Signed-off-by: Zoltán Böszörményi <zboszor@gmail.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This only occurs in debian package builds when populating the sdk
and is a work around that seems to work. Eventually we should look
at why this is failing (I have ideas, it's somewhere in
lib/oe/package_management/deb/sdk.py), but for now, do this so we can
fix the core issue with nativesdk-intercepts.
(From OE-Core rev: a411123a95114233c5efd762dbcc8eb513030aab)
Signed-off-by: Eilís 'pidge' Ní Fhlannagáin <pidge@baylibre.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If you build an image with lots of manpages in, then each package will
run mandb inside qemu-user at rootfs time. This is a slow operation
and should be done once when all of the packages have installed using an
intercept instead.
The call to mandb has been changed too. mandb doesn't actually allow
the configuration file to be read from stdin so that was being ignored,
instead write the file to a temporary file and use that.
This means we then don't need to tell it where to search explicitly, and
it writes the indexes to the correct paths so we don't need to move
files afterwards either.
Sadly we do still need to run mandb inside qemu-user, as the underlying
database is a gdbm file and they are byte-order dependent.
For my test case of core-image-base with api-documentation
DISTRO_FEATURES and doc-pkgs IMAGE_FEATURES enabled, the performance
gain is significant:
core-image-base do_rootfs -1303.1s -73.6% 1771.6s -> 468.5s
(From OE-Core rev: fbd8a57aa307bfda70a08cb78af3c97f05c39a3a)
Signed-off-by: Ross Burton <ross.burton@arm.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
It duplicates udevadm in systemd recipe to make it could run
update_udev_hwdb with multilib enabled. Since systemd last update,
it deploys a shared library libsystemd-shared-250.so in /lib/systemd/.
The library will be overwritten when multilib enabled. Then if both
udev-hwdb and lib32-udev-hwdb are installed, it fails to run the
multilib version postinstall intercept update_udev_hwdb:
| /path/to/build/tmp-glibc/work/intel_x86_64-wrs-linux/wrlinux-image-small/1.0-r1/rootfs/usr/libexec/lib32-udevadm:
error while loading shared libraries: libsystemd-shared-250.so: wrong ELF class: ELFCLASS64
`udevadm hwdb --update` just concatenates .hwdb files in dirs
/etc/udev/hwdb.d and /lib/udev/hwdb.d. The output file hwdb.bin is
identical with the one created by lib32-udevadm. So do NOT duplicate
lib32-udevadm in systemd and eudev. And update intercept script
update_udev_hwdb that re-run udevadm with same arch qemuwrapper if run
${binprefix}qemuwrapper failed.
(From OE-Core rev: 3dba872a42c2be7d0865a30118984ab013850292)
Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Steps to reproduce:
echo "IMAGE_INSTALL_append = \" udev-hwdb lib32-udev-hwdb\"" >> conf/local.conf
When install both udev-hwdb and lib32-udev-hwdb as above,
there comes below do_populate_sdk error:
$ bitbake core-image-sato -c populate_sdk
ERROR: Task (/path/core-image-sato.bb:do_populate_sdk) failed with exit code '134'
NOTE: Tasks Summary: Attempted 5554 tasks of which 0 didn't need to be rerun and 1 failed.
$ cat /path/tmp/work/qemux86_64-poky-linux/core-image-sato/1.0-r5/pseudo/pseudo.log
[snip]
inode mismatch: '/path/tmp/work/qemux86_64-poky-linux/core-image-sato/1.0-r5/sdk/image/usr/local/oecore-x86_64/sysroots/core2-64-poky-linux/lib/udev/hwdb.bin' ino 427383040 in db, 427383042 in request.
[snip]
It is because both udev-hwdb and lib32-udev-hwdb will generate
${SDK_OUTPUT}/${SDKTARGETSYSROOT}/lib/udev/hwdb.bin during do_populate_sdk
and it triggers pseudo error.
So clean hwdb.bin before generate hwdb.bin to avoid conflict to
fix the above do_populate_sdk error.
(From OE-Core rev: c7472925feb53ce92c1799feba2b7a9104e3f38f)
Signed-off-by: Mingli Yu <mingli.yu@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This is a pretty big change to the way pseudo operates when used in OpenEmbedded.
Normally, pseudo monitors and logs (adds to its database) any file created or
modified whilst in a fakeroot environment. There are large numbers of files
we simply don't care about the permissions of whilst in that fakeroot context,
for example ${S}, ${B}, ${T}, ${SSTATE_DIR}, the central sstate control
directories,
This change uses new functionality in pseudo to ignore these directory trees,
resulting in a cleaner database with less chance of "stray" mismatches if files
are modified outside pseudo context. It also should reduce some overhead from
pseudo as the interprocess round trip to the server is avoided.
There is a possible complication where some existing recipe may break, for
example, we found a recipe which was writing to "${B}/install" for
"make install" in do_install and since we listed ${B} as not to be tracked,
there were errors trying to chown root for files in this location.
This patch fixes a few corner cases in OE-Core when used with this new
ignore list:
* The archiver directory matched a "${WORKDIR}/deploy*" pattern so was renamed
to something else since that directory does need its root permissions
* The ${S} and ${B} ignoring is conditional on them being different to ${WORKDIR}
* package_write_* task output (the debs/rpms/ipks) are now owned by the build
user so we don't want the file ownership information in the hashequiv outhash
calculation even if they are built under pseudo.
* The fontcache postinstall intercept is run under qemu outside of pseudo context
so delete files it may delete up front where pseudo can see this.
* SSTATE_DIR is in PSEUDO_PATHS_IGNORE, which is in FAKEROOTENV which is cached
by bitbake. We therefore need to trigger reparsing if this changes, which means
SSTATE_DIR can be in BB_HASHBASE_WHITELIST but not BB_HASHCONFIG_WHITELIST.
Rework the variables to handle this. This otherwise breaks some of our sstate
tests in oe-selftest.
* Ignore the temp directory wic uses for rebuilding rootfs.
(From OE-Core rev: ad8f5532ffaead9a5ad13e1034fe9e5e1b7979f4)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
systemd supports a distribution hwdb.bin in /usr/lib/udev/hwdb.bin,
which is used if /etc/udev/hwdb.bin is not present. When generating the
install time hwdb, for systemd, ensure that we put it in /usr/lib/udev,
which then ensures that at boot time we do not regenerate it, unless the
system is marked for update.
This allows fragments dropped into /etc/udev/hwdb.d to be processed
correctly, but without requiring a first boot time build:
root@qemumips:~# systemctl status systemd-hwdb-update.service
* systemd-hwdb-update.service - Rebuild Hardware Database
Loaded: loaded (/usr/lib/systemd/system/systemd-hwdb-update.service; static; vendor preset: disabled)
Active: inactive (dead)
Condition: start condition failed at Wed 2020-03-04 15:18:11 UTC; 44s ago
|- ConditionPathExists=|!/usr/lib/udev/hwdb.bin was not met
|- ConditionPathExists=|/etc/udev/hwdb.bin was not met
`- ConditionDirectoryNotEmpty=|/etc/udev/hwdb.d was not met
Docs: man:hwdb(7)
man:systemd-hwdb(8)
(From OE-Core rev: 78ff610e571e4d739326520c391a37ee1f1290ae)
Signed-off-by: Alex Kiernan <alex.kiernan@gmail.com>
Acked-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When opening files by file-browsers on fresh images, user has to choose the
application to open from the pool af ALL known applications even those not
designed to open the file selected. By inheriting this classs in recipes the
assosiations in /usr/share/applications/mimeinfo.cache are build by calling
update-desktop-database.
(From OE-Core rev: 20208d6763e725cea211f933ec1c8e32733a10af)
Signed-off-by: Andreas Müller <schnitzeltony@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* add a short descriptions of class' use case
* remove checks for update-mime-database - it can be considered available:
* at build time by PACKAGE_WRITE_DEPS
* at package upgrade by RDEPENDS chain
pkg -> shared-mime-info-data -> shared-mime-info
* simplify (accelerate?) xml file extension detection
* run update-mime-database once only at image creation to avoid expensive
redundant operations
* allow shared-mime-info to inherit mime.bbclass by avoiding circular
dependencies
(From OE-Core rev: 6467b7b98c9a55e27d1ab9f253ec48da2a722e77)
Signed-off-by: Andreas Müller <schnitzeltony@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The intercept is called update_icon_cache which is vague: rename to
update_gtk_icon_cache to make it clearer what it is for, and add a comment
explaining what class caused it to be used.
(From OE-Core rev: 3158adbe684890adc56af11e19af872e90e09d41)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This adds SPDX license headers in place of the wide assortment of things
currently in our script headers. We default to GPL-2.0-only except for the
oeqa code where it was clearly submitted and marked as MIT on the most part
or some scripts which had the "or later" GPL versioning.
The patch also drops other obsolete bits of file headers where they were
encoountered such as editor modelines, obsolete maintainer information or
the phrase "All rights reserved" which is now obsolete and not required in
copyright headers (in this case its actually confusing for licensing as all
rights were not reserved).
More work is needed for OE-Core but this takes care of the bulk of the scripts
and meta/lib directories.
The top level LICENSE files are tweaked to match the new structure and the
SPDX naming.
(From OE-Core rev: f8c9c511b5f1b7dbd45b77f345cb6c048ae6763e)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When multilib is enabled and both udev-hwdb and ${MLPREFIX}udev-hwdb are
installed to image, it fails to run one of their postinstall scripts
that they both call ${base_bindir}/udevadm with same user mode qemu.
Duplicate udevadm and add postinst-intercept update_udev_hwdb to fix the
failures.
(From OE-Core rev: d6b737a31a8842cdc770b05e28503c81a691d10d)
Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Postinstall script update_gtk_immodules_cache calls
${bindir}/gtk-query-immodules-${version}. When multilib is enabled, both
packages foo and lib32-foo call ${bindir}/gtk-query-immodules-${version}
and one of them will fail to run obviously.
Duplicate install files gtk-query-immodules-${version} to ${libexecdir}
with ${MLPREFIX}. And update update_gtk_immodules_cache calls proper
binary.
(From OE-Core rev: cad28bd74438d8edbdb02a2d56169e2a55260ca8)
Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Packages which inherit fontcache.bbclass call postinstall script
update_font_cache. And in update_font_cache, it calls ${bindir}/fc-cache
by qemuwrapper. When multilib is enabled, both packages foo and lib32-foo
will call ${bindir}/fc-cache and one of them will fail to run obviously.
Duplicate install file fc-cache to ${libexecdir} with ${MLPREFIX} and
call proper fc-cache in update_font_cache.
(From OE-Core rev: 53d8625732b0c8416e367d5eef43863ec2065433)
Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Previously only one global variant of the script was created,
which caused numerous issues:
1) multilib was not properly supported due to multilib variants each
needing their own version of the qemu binary and library paths to be set
2) nativesdk was not properly supported for the same reason
This patch also moves setting LD_LIBRARY_PATH directly into the
recipe, as passing it down from other recipes did not work when
said recipes were allarch, and adjusts calls to qemuwrapper from
postinst-intercepts, so that its correct variant is selected.
Also, the various qemu fallbacks in qemuwrapper script are all removed,
as they are no longer necessary.
(From OE-Core rev: d10fd6ae3fe46290c6e3a5250878966d9f12ca3f)
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This has the following benefits:
- consistent with how the other caches are created into target rootfs
- only runs once per package manager transaction, instead of once per every immodule package
- correctly postpones to first boot if qemu is not working; from postinst itself
this would've required special arrangements to avoid what is now a do_rootfs failure.
(From OE-Core rev: cca3c084b6c9bf600d7306e3fe12c4f236b78656)
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This allows postinsts scripts to explicitly request a deferral to first boot
(by calling 'postinst_intercept delay_to_first_boot') instead of 'exit 1' which
should be used only to indicate actual script failures.
(From OE-Core rev: 853ac4a061e47c70b61e255c3cf6109d2058d2cc)
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Only change the ownership of ${libdir}/gio/modules/giomodule.cache if
it exists.
(From OE-Core rev: df2e1a8fbadffac0f1781a0d07e050356a007327)
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>
update_gio_module_cache intercept creates file:
$D${libdir}/gio/modules/giomodule.cache
Change ownership of this file to root:root to avoid user contamination
by host.
(From OE-Core rev: 9a23af37ad11a7176248ade88511f34fe6dd97bb)
Signed-off-by: Abdur Rehman <abdur_rehman@mentor.com>
Signed-off-by: Christopher Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The new class uses gio-querymodules tool on postinst and postrm.
This regenerates the module cache which is useful to avoid loading
modules that are not needed at runtime: If a Gio module is not
listed in the cache file it will always get loaded.
* Add a postinst-intercept 'gio-module-cache': it runs
gio-querymodules using qemuwrapper. This is required because the tool
actually loads the modules to generate the cache.
* Add a gio-module-cache class that adds postinstall and postrm
scripts. In the sysroot population case use the new intercept.
* Inherit the new class in glib-2.0, glib-networking and gconf.
Fixes [YOCTO #9241].
(From OE-Core rev: 168cf5502a2dda78348cc9679e37bed0c69c0208)
Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If a postinst has a problem (say, qemu crashes) and set -e isn't in operation,
the only mention of the problem is a single line in the rootfs log that doesn't
trigger any warnings.
(From OE-Core rev: 072800f89a136bb5da44627f25599d3060cca0a1)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
By default fc-cache will scan both system and user directories, which means it
attempts to scan $HOME/.fonts. As this is the build host's idea of $HOME this
generally doesn't exist, and causes fc-cache to exit with a failure.
Solve this by passing --system-only so that fc-cache will only scan system
directories, as is appropriate for a rootfs-time invocation.
(From OE-Core rev: 541315d6c56df6448f64c262f99d43d5c1e9400b)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The intercept runs against the native sysroot so we need to pass it
the native libdir instead of the target libdir, as otherwise it will
use target paths (such as lib64) in the native sysroot.
(From OE-Core rev: 0fe84007176c98644b1917966c98501beb7e0ce2)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Instead of expecting that the calling postinst has exported
STAGING_LIBDIR_NATIVE (which will get set to the sysroot at package build time
and may not be correct if sstate is used), use the new STAGING_DIR_NATIVE that
is exported by rootfs.py.
[ YOCTO #8547 ]
(From OE-Core rev: 9fa5ff645eb683533af482925079dff5c9151831)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* this can be useful for passing extra parameters, pass
-v by default to see what's going on in do_rootfs
* we need to use this for extra parameter we implemented
in fontconfig:
--ignore-mtime always use cache file regardless of font directory mtime
because the checksum of fontcache generated in do_rootfs
doesn't match with /usr/share/fonts directory as seen on
target device causing fontconfig to re-create the cache
when fontconfig is used for first time or worse create
new cache in every user's home directory when /usr/
filesystem is read only and cache cannot be updated.
Running FC_DEBUG=16 fc-cache -v on such device shows:
FcCacheTimeValid dir "/usr/share/fonts" cache checksum 1441207803 dir checksum 1441206149
* my guess is that the checksum is different, because pseudo
(which is unloaded when running qemuwrapper) or because some
influence of running the rootfs under qemu.
(From OE-Core rev: e07fdc5f047f309a0e99331f430c14d708ed2161)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If for example gdk-pixbuf and lib32-gdk-pixbuf are in an image then only one
${bindir}/gdk-pixbuf-query-loaders will be installed, so only one variant will
actually be usable.
Solve this by moving gdk-pixbuf-query-loaders into ${libdir} as it's intimately
tied to the library and rarely directly invoked by the user, and update the
callers to use the right path.
(From OE-Core rev: 69df75f268e2b3d5874f05e2b5a6125f6d990a03)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* trying to pass foo="a b" through postinst_intercept ends
with the actual script header to containing:
b
foo=a
which fails because "b" command doesn't exist.
(From OE-Core rev: c66d7d85b7225be8c838449324d506565dd0081d)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The file ownership of the cache files in /var/cache/fontconfig needs to
be set to root:root otherwise it inherits the user and group id of the
build user.
[YOCTO #7411]
(From OE-Core rev: 23393c6685860d0b1c459874d35395360d0b6d3c)
Signed-off-by: Jonathan Liu <net147@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Pass GDK_PIXBUF_FATAL_LOADER to the sstate postinst and intercept so that any
problems are flagged as errors instead of being silently ignored.
(From OE-Core rev: 0a97b1571821848af11d8651c7145ed9592f9e31)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The redirection was necessary with the old bash code because the log
checking routine was searching for error strings in the log and abort
the build in case failures occured. With the new python code,
redirecting the intercept stderr is no longer necessary. This also makes
the intercept hooks easier to debug.
(From OE-Core rev: e2c3c59b26d84cd4052a953adafcd4d456264c26)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When using multilib, the hooks for lib32/lib64 must be different because
the libdir/base_libdir point to different locations. Postinstalls
calling postint_intercept script must pass the mlprefix in the 3rd
argument.
(From OE-Core rev: 2c5c6e3ffcd561c25a34603922b622449f677a34)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Postinstalls that use qemu are throwing a segmentation fault when
building for qemux86-64 on a 64bit host (it might also happen for
qemux86 if building on a 32bit host but I didn't test). It looks like
qemu looks for ld.so.cache which is not found because it is generated
after rootfs_(rpm|ipk|deb)_do_rootfs is called and then it tries to load
libraries from the default paths (which are the host's). In order to
avoid this, pass the LD_LIBRARY_PATH explicitly to the target's dynamic
loader.
(From OE-Core rev: 48e8b613b3f5c7b1d917bf3147606d44072ce49e)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The intercept scripts fail to run on 32 bit hosts. Apparently, the
current approach worked on 64 bit hosts due to the larger virtual address
space (probably). On 32 bit hosts, however, calling the target binary like:
qemu-arm ld-linux.so --library-path /lib:/usr/lib arm_binary
fails with:
arm_binary: error while loading shared libraries: arm_binary: failed to
map segment from shared object: Operation not permitted
When run like this, qemu-arm fails to map the arm_binary executable in
memory because it's hitting the lower limit of
/proc/sys/vm/mmap_min_addr. That's because it loads the
ld-linux.so binary successfully, taking into account mmap_min_addr, runs
it, and then ld-linux.so will map the arm_binary at a fixed address but this
will fail because it is below mmap_min_addr. The qemu's guest base probing,
apparently, doesn't work fine when a program runs inside other.
One way around this would be to set mmap_min_addr to 0 (on recent
distributions is set to 65536 to avoid "kernel NULL pointer dereference"
defects) but this approach is not safe.
The other way is to call the binary directly but providing qemu with a
prefix (-L option) in order to find the elf interpreter correctly. This
way, both the target binary and dynamic loader are mapped into memory
under qemu's control and, only after, the dynamic loader is started.
[YOCTO #4179]
(From OE-Core rev: 78f91e08c8a7b0f0c831a087f7c89e2c76047e7a)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The scripts/postinst-intercepts will contain all postinstall hooks that
we need to run after all packages have been installed.
If one wants to install such a postinst hook, all it needs to do is put
the hook in this directory and, from the package postinstall scriptlet,
call:
postinst_intercept <hook_name> <package_name> <var1=...> ...
This will, practically, add the package_name in the list of packages
that need the hook to run and, also, set any variables that would be
needed in the hook. For example, variables like ${libdir}, ${bindir},
etc. that might depend on distribution can be passed on to the script in
this way.
(From OE-Core rev: 0ef538d75c2f3921a2fcbe6ca1deed5525b276cc)
Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
